linux/drivers/scsi/scsi_debug.c

7857 lines
225 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-or-later
/*
* vvvvvvvvvvvvvvvvvvvvvvv Original vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
* Copyright (C) 1992 Eric Youngdale
* Simulate a host adapter with 2 disks attached. Do a lot of checking
* to make sure that we are not getting blocks mixed up, and PANIC if
* anything out of the ordinary is seen.
* ^^^^^^^^^^^^^^^^^^^^^^^ Original ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
*
* Copyright (C) 2001 - 2021 Douglas Gilbert
*
* For documentation see http://sg.danny.cz/sg/scsi_debug.html
*/
#define pr_fmt(fmt) KBUILD_MODNAME ":%s: " fmt, __func__
#include <linux/module.h>
#include <linux/align.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/jiffies.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include <linux/types.h>
#include <linux/string.h>
#include <linux/fs.h>
#include <linux/init.h>
#include <linux/proc_fs.h>
#include <linux/vmalloc.h>
#include <linux/moduleparam.h>
#include <linux/scatterlist.h>
#include <linux/blkdev.h>
#include <linux/crc-t10dif.h>
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
#include <linux/spinlock.h>
#include <linux/interrupt.h>
#include <linux/atomic.h>
#include <linux/hrtimer.h>
#include <linux/uuid.h>
#include <linux/t10-pi.h>
#include <linux/msdos_partition.h>
#include <linux/random.h>
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
#include <linux/xarray.h>
#include <linux/prefetch.h>
#include <net/checksum.h>
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
#include <asm/unaligned.h>
#include <scsi/scsi.h>
#include <scsi/scsi_cmnd.h>
#include <scsi/scsi_device.h>
#include <scsi/scsi_host.h>
#include <scsi/scsicam.h>
#include <scsi/scsi_eh.h>
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
#include <scsi/scsi_tcq.h>
#include <scsi/scsi_dbg.h>
#include "sd.h"
#include "scsi_logging.h"
/* make sure inq_product_rev string corresponds to this version */
#define SDEBUG_VERSION "0191" /* format to fit INQUIRY revision field */
static const char *sdebug_version_date = "20210520";
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
#define MY_NAME "scsi_debug"
/* Additional Sense Code (ASC) */
#define NO_ADDITIONAL_SENSE 0x0
#define LOGICAL_UNIT_NOT_READY 0x4
#define LOGICAL_UNIT_COMMUNICATION_FAILURE 0x8
#define UNRECOVERED_READ_ERR 0x11
#define PARAMETER_LIST_LENGTH_ERR 0x1a
#define INVALID_OPCODE 0x20
#define LBA_OUT_OF_RANGE 0x21
#define INVALID_FIELD_IN_CDB 0x24
#define INVALID_FIELD_IN_PARAM_LIST 0x26
#define WRITE_PROTECTED 0x27
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
#define UA_RESET_ASC 0x29
#define UA_CHANGED_ASC 0x2a
#define TARGET_CHANGED_ASC 0x3f
#define LUNS_CHANGED_ASCQ 0x0e
#define INSUFF_RES_ASC 0x55
#define INSUFF_RES_ASCQ 0x3
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
#define POWER_ON_RESET_ASCQ 0x0
#define POWER_ON_OCCURRED_ASCQ 0x1
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
#define BUS_RESET_ASCQ 0x2 /* scsi bus reset occurred */
#define MODE_CHANGED_ASCQ 0x1 /* mode parameters changed */
#define CAPACITY_CHANGED_ASCQ 0x9
#define SAVING_PARAMS_UNSUP 0x39
#define TRANSPORT_PROBLEM 0x4b
#define THRESHOLD_EXCEEDED 0x5d
#define LOW_POWER_COND_ON 0x5e
#define MISCOMPARE_VERIFY_ASC 0x1d
#define MICROCODE_CHANGED_ASCQ 0x1 /* with TARGET_CHANGED_ASC */
#define MICROCODE_CHANGED_WO_RESET_ASCQ 0x16
#define WRITE_ERROR_ASC 0xc
#define UNALIGNED_WRITE_ASCQ 0x4
#define WRITE_BOUNDARY_ASCQ 0x5
#define READ_INVDATA_ASCQ 0x6
#define READ_BOUNDARY_ASCQ 0x7
#define ATTEMPT_ACCESS_GAP 0x9
#define INSUFF_ZONE_ASCQ 0xe
/* Additional Sense Code Qualifier (ASCQ) */
#define ACK_NAK_TO 0x3
/* Default values for driver parameters */
#define DEF_NUM_HOST 1
#define DEF_NUM_TGTS 1
#define DEF_MAX_LUNS 1
/* With these defaults, this driver will make 1 host with 1 target
* (id 0) containing 1 logical unit (lun 0). That is 1 device.
*/
#define DEF_ATO 1
#define DEF_CDB_LEN 10
#define DEF_JDELAY 1 /* if > 0 unit is a jiffy */
#define DEF_DEV_SIZE_PRE_INIT 0
#define DEF_DEV_SIZE_MB 8
#define DEF_ZBC_DEV_SIZE_MB 128
#define DEF_DIF 0
#define DEF_DIX 0
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
#define DEF_PER_HOST_STORE false
#define DEF_D_SENSE 0
#define DEF_EVERY_NTH 0
#define DEF_FAKE_RW 0
#define DEF_GUARD 0
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
#define DEF_HOST_LOCK 0
#define DEF_LBPU 0
#define DEF_LBPWS 0
#define DEF_LBPWS10 0
#define DEF_LBPRZ 1
#define DEF_LOWEST_ALIGNED 0
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
#define DEF_NDELAY 0 /* if > 0 unit is a nanosecond */
#define DEF_NO_LUN_0 0
#define DEF_NUM_PARTS 0
#define DEF_OPTS 0
#define DEF_OPT_BLKS 1024
#define DEF_PHYSBLK_EXP 0
#define DEF_OPT_XFERLEN_EXP 0
#define DEF_PTYPE TYPE_DISK
#define DEF_RANDOM false
#define DEF_REMOVABLE false
#define DEF_SCSI_LEVEL 7 /* INQUIRY, byte2 [6->SPC-4; 7->SPC-5] */
#define DEF_SECTOR_SIZE 512
#define DEF_UNMAP_ALIGNMENT 0
#define DEF_UNMAP_GRANULARITY 1
#define DEF_UNMAP_MAX_BLOCKS 0xFFFFFFFF
#define DEF_UNMAP_MAX_DESC 256
#define DEF_VIRTUAL_GB 0
#define DEF_VPD_USE_HOSTNO 1
#define DEF_WRITESAME_LENGTH 0xFFFF
#define DEF_STRICT 0
#define DEF_STATISTICS false
#define DEF_SUBMIT_QUEUES 1
#define DEF_TUR_MS_TO_READY 0
#define DEF_UUID_CTL 0
#define JDELAY_OVERRIDDEN -9999
/* Default parameters for ZBC drives */
#define DEF_ZBC_ZONE_SIZE_MB 128
#define DEF_ZBC_MAX_OPEN_ZONES 8
#define DEF_ZBC_NR_CONV_ZONES 1
#define SDEBUG_LUN_0_VAL 0
/* bit mask values for sdebug_opts */
#define SDEBUG_OPT_NOISE 1
#define SDEBUG_OPT_MEDIUM_ERR 2
#define SDEBUG_OPT_TIMEOUT 4
#define SDEBUG_OPT_RECOVERED_ERR 8
#define SDEBUG_OPT_TRANSPORT_ERR 16
#define SDEBUG_OPT_DIF_ERR 32
#define SDEBUG_OPT_DIX_ERR 64
#define SDEBUG_OPT_MAC_TIMEOUT 128
#define SDEBUG_OPT_SHORT_TRANSFER 0x100
#define SDEBUG_OPT_Q_NOISE 0x200
#define SDEBUG_OPT_ALL_TSF 0x400 /* ignore */
#define SDEBUG_OPT_RARE_TSF 0x800
#define SDEBUG_OPT_N_WCE 0x1000
#define SDEBUG_OPT_RESET_NOISE 0x2000
#define SDEBUG_OPT_NO_CDB_NOISE 0x4000
#define SDEBUG_OPT_HOST_BUSY 0x8000
#define SDEBUG_OPT_CMD_ABORT 0x10000
#define SDEBUG_OPT_ALL_NOISE (SDEBUG_OPT_NOISE | SDEBUG_OPT_Q_NOISE | \
SDEBUG_OPT_RESET_NOISE)
#define SDEBUG_OPT_ALL_INJECTING (SDEBUG_OPT_RECOVERED_ERR | \
SDEBUG_OPT_TRANSPORT_ERR | \
SDEBUG_OPT_DIF_ERR | SDEBUG_OPT_DIX_ERR | \
SDEBUG_OPT_SHORT_TRANSFER | \
SDEBUG_OPT_HOST_BUSY | \
SDEBUG_OPT_CMD_ABORT)
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
#define SDEBUG_OPT_RECOV_DIF_DIX (SDEBUG_OPT_RECOVERED_ERR | \
SDEBUG_OPT_DIF_ERR | SDEBUG_OPT_DIX_ERR)
/* As indicated in SAM-5 and SPC-4 Unit Attentions (UAs) are returned in
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
* priority order. In the subset implemented here lower numbers have higher
* priority. The UA numbers should be a sequence starting from 0 with
* SDEBUG_NUM_UAS being 1 higher than the highest numbered UA. */
#define SDEBUG_UA_POR 0 /* Power on, reset, or bus device reset */
#define SDEBUG_UA_POOCCUR 1 /* Power on occurred */
#define SDEBUG_UA_BUS_RESET 2
#define SDEBUG_UA_MODE_CHANGED 3
#define SDEBUG_UA_CAPACITY_CHANGED 4
#define SDEBUG_UA_LUNS_CHANGED 5
#define SDEBUG_UA_MICROCODE_CHANGED 6 /* simulate firmware change */
#define SDEBUG_UA_MICROCODE_CHANGED_WO_RESET 7
#define SDEBUG_NUM_UAS 8
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
/* when 1==SDEBUG_OPT_MEDIUM_ERR, a medium error is simulated at this
* sector on read commands: */
#define OPT_MEDIUM_ERR_ADDR 0x1234 /* that's sector 4660 in decimal */
#define OPT_MEDIUM_ERR_NUM 10 /* number of consecutive medium errs */
/* SDEBUG_CANQUEUE is the maximum number of commands that can be queued
* (for response) per submit queue at one time. Can be reduced by max_queue
* option. Command responses are not queued when jdelay=0 and ndelay=0. The
* per-device DEF_CMD_PER_LUN can be changed via sysfs:
* /sys/class/scsi_device/<h:c:t:l>/device/queue_depth
* but cannot exceed SDEBUG_CANQUEUE .
*/
#define SDEBUG_CANQUEUE_WORDS 3 /* a WORD is bits in a long */
#define SDEBUG_CANQUEUE (SDEBUG_CANQUEUE_WORDS * BITS_PER_LONG)
#define DEF_CMD_PER_LUN SDEBUG_CANQUEUE
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
/* UA - Unit Attention; SA - Service Action; SSU - Start Stop Unit */
#define F_D_IN 1 /* Data-in command (e.g. READ) */
#define F_D_OUT 2 /* Data-out command (e.g. WRITE) */
#define F_D_OUT_MAYBE 4 /* WRITE SAME, NDOB bit */
#define F_D_UNKN 8
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
#define F_RL_WLUN_OK 0x10 /* allowed with REPORT LUNS W-LUN */
#define F_SKIP_UA 0x20 /* bypass UAs (e.g. INQUIRY command) */
#define F_DELAY_OVERR 0x40 /* for commands like INQUIRY */
#define F_SA_LOW 0x80 /* SA is in cdb byte 1, bits 4 to 0 */
#define F_SA_HIGH 0x100 /* SA is in cdb bytes 8 and 9 */
#define F_INV_OP 0x200 /* invalid opcode (not supported) */
#define F_FAKE_RW 0x400 /* bypass resp_*() when fake_rw set */
#define F_M_ACCESS 0x800 /* media access, reacts to SSU state */
#define F_SSU_DELAY 0x1000 /* SSU command delay (long-ish) */
#define F_SYNC_DELAY 0x2000 /* SYNCHRONIZE CACHE delay */
/* Useful combinations of the above flags */
#define FF_RESPOND (F_RL_WLUN_OK | F_SKIP_UA | F_DELAY_OVERR)
#define FF_MEDIA_IO (F_M_ACCESS | F_FAKE_RW)
#define FF_SA (F_SA_HIGH | F_SA_LOW)
#define F_LONG_DELAY (F_SSU_DELAY | F_SYNC_DELAY)
#define SDEBUG_MAX_PARTS 4
#define SDEBUG_MAX_CMD_LEN 32
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
#define SDEB_XA_NOT_IN_USE XA_MARK_1
static struct kmem_cache *queued_cmd_cache;
#define TO_QUEUED_CMD(scmd) ((void *)(scmd)->host_scribble)
#define ASSIGN_QUEUED_CMD(scmnd, qc) { (scmnd)->host_scribble = (void *) qc; }
/* Zone types (zbcr05 table 25) */
enum sdebug_z_type {
ZBC_ZTYPE_CNV = 0x1,
ZBC_ZTYPE_SWR = 0x2,
ZBC_ZTYPE_SWP = 0x3,
/* ZBC_ZTYPE_SOBR = 0x4, */
ZBC_ZTYPE_GAP = 0x5,
};
/* enumeration names taken from table 26, zbcr05 */
enum sdebug_z_cond {
ZBC_NOT_WRITE_POINTER = 0x0,
ZC1_EMPTY = 0x1,
ZC2_IMPLICIT_OPEN = 0x2,
ZC3_EXPLICIT_OPEN = 0x3,
ZC4_CLOSED = 0x4,
ZC6_READ_ONLY = 0xd,
ZC5_FULL = 0xe,
ZC7_OFFLINE = 0xf,
};
struct sdeb_zone_state { /* ZBC: per zone state */
enum sdebug_z_type z_type;
enum sdebug_z_cond z_cond;
bool z_non_seq_resource;
unsigned int z_size;
sector_t z_start;
sector_t z_wp;
};
struct sdebug_dev_info {
struct list_head dev_list;
unsigned int channel;
unsigned int target;
u64 lun;
uuid_t lu_name;
struct sdebug_host_info *sdbg_host;
unsigned long uas_bm[1];
atomic_t stopped; /* 1: by SSU, 2: device start */
bool used;
/* For ZBC devices */
enum blk_zoned_model zmodel;
unsigned int zcap;
unsigned int zsize;
unsigned int zsize_shift;
unsigned int nr_zones;
unsigned int nr_conv_zones;
unsigned int nr_seq_zones;
unsigned int nr_imp_open;
unsigned int nr_exp_open;
unsigned int nr_closed;
unsigned int max_open;
ktime_t create_ts; /* time since bootup that this device was created */
struct sdeb_zone_state *zstate;
};
struct sdebug_host_info {
struct list_head host_list;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
int si_idx; /* sdeb_store_info (per host) xarray index */
struct Scsi_Host *shost;
struct device dev;
struct list_head dev_info_list;
};
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
/* There is an xarray of pointers to this struct's objects, one per host */
struct sdeb_store_info {
rwlock_t macc_lck; /* for atomic media access on this store */
u8 *storep; /* user data storage (ram) */
struct t10_pi_tuple *dif_storep; /* protection info */
void *map_storep; /* provisioning map */
};
#define dev_to_sdebug_host(d) \
container_of(d, struct sdebug_host_info, dev)
#define shost_to_sdebug_host(shost) \
dev_to_sdebug_host(shost->dma_dev)
enum sdeb_defer_type {SDEB_DEFER_NONE = 0, SDEB_DEFER_HRT = 1,
SDEB_DEFER_WQ = 2, SDEB_DEFER_POLL = 3};
struct sdebug_defer {
struct hrtimer hrt;
struct execute_work ew;
ktime_t cmpl_ts;/* time since boot to complete this cmd */
int issuing_cpu;
bool aborted; /* true when blk_abort_request() already called */
enum sdeb_defer_type defer_t;
};
struct sdebug_queued_cmd {
/* corresponding bit set in in_use_bm[] in owning struct sdebug_queue
* instance indicates this slot is in use.
*/
struct sdebug_defer sd_dp;
struct scsi_cmnd *scmd;
};
struct sdebug_scsi_cmd {
spinlock_t lock;
};
static atomic_t sdebug_cmnd_count; /* number of incoming commands */
static atomic_t sdebug_completions; /* count of deferred completions */
static atomic_t sdebug_miss_cpus; /* submission + completion cpus differ */
static atomic_t sdebug_a_tsf; /* 'almost task set full' counter */
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
static atomic_t sdeb_inject_pending;
static atomic_t sdeb_mq_poll_count; /* bumped when mq_poll returns > 0 */
struct opcode_info_t {
u8 num_attached; /* 0 if this is it (i.e. a leaf); use 0xff */
/* for terminating element */
u8 opcode; /* if num_attached > 0, preferred */
u16 sa; /* service action */
u32 flags; /* OR-ed set of SDEB_F_* */
int (*pfp)(struct scsi_cmnd *, struct sdebug_dev_info *);
const struct opcode_info_t *arrp; /* num_attached elements or NULL */
u8 len_mask[16]; /* len_mask[0]-->cdb_len, then mask for cdb */
/* 1 to min(cdb_len, 15); ignore cdb[15...] */
};
/* SCSI opcodes (first byte of cdb) of interest mapped onto these indexes */
enum sdeb_opcode_index {
SDEB_I_INVALID_OPCODE = 0,
SDEB_I_INQUIRY = 1,
SDEB_I_REPORT_LUNS = 2,
SDEB_I_REQUEST_SENSE = 3,
SDEB_I_TEST_UNIT_READY = 4,
SDEB_I_MODE_SENSE = 5, /* 6, 10 */
SDEB_I_MODE_SELECT = 6, /* 6, 10 */
SDEB_I_LOG_SENSE = 7,
SDEB_I_READ_CAPACITY = 8, /* 10; 16 is in SA_IN(16) */
SDEB_I_READ = 9, /* 6, 10, 12, 16 */
SDEB_I_WRITE = 10, /* 6, 10, 12, 16 */
SDEB_I_START_STOP = 11,
SDEB_I_SERV_ACT_IN_16 = 12, /* add ...SERV_ACT_IN_12 if needed */
SDEB_I_SERV_ACT_OUT_16 = 13, /* add ...SERV_ACT_OUT_12 if needed */
SDEB_I_MAINT_IN = 14,
SDEB_I_MAINT_OUT = 15,
SDEB_I_VERIFY = 16, /* VERIFY(10), VERIFY(16) */
SDEB_I_VARIABLE_LEN = 17, /* READ(32), WRITE(32), WR_SCAT(32) */
SDEB_I_RESERVE = 18, /* 6, 10 */
SDEB_I_RELEASE = 19, /* 6, 10 */
SDEB_I_ALLOW_REMOVAL = 20, /* PREVENT ALLOW MEDIUM REMOVAL */
SDEB_I_REZERO_UNIT = 21, /* REWIND in SSC */
SDEB_I_ATA_PT = 22, /* 12, 16 */
SDEB_I_SEND_DIAG = 23,
SDEB_I_UNMAP = 24,
scsi: scsi_debug: Fix a recently introduced regression A recent commit removed an element from opcode_info_arr[] but did not modify opcode_ind_arr[] nor was SDEB_I_XDWRITEREAD removed. Remove SDEB_I_XDWRITEREAD and bring the two arrays again in sync. This patch avoids that the following is reported: BUG: KASAN: null-ptr-deref in scsi_debug_queuecommand+0x60f/0xc90 [scsi_debug] Read of size 1 at addr 0000000000000001 by task iscsi-test-cu/683 CPU: 3 PID: 683 Comm: iscsi-test-cu Not tainted 5.0.0-rc5-dbg+ #1 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014 Call Trace: dump_stack+0x86/0xca kasan_report.cold.3+0x5/0x3e __asan_load1+0x47/0x50 scsi_debug_queuecommand+0x60f/0xc90 [scsi_debug] scsi_queue_rq+0xc17/0x12e0 blk_mq_dispatch_rq_list+0x5fc/0xb10 blk_mq_sched_dispatch_requests+0x2f7/0x300 __blk_mq_run_hw_queue+0xd6/0x180 __blk_mq_delay_run_hw_queue+0x25c/0x290 blk_mq_run_hw_queue+0x119/0x1b0 blk_mq_sched_insert_request+0x274/0x350 blk_execute_rq_nowait+0x78/0x90 blk_execute_rq+0xcc/0x140 sg_io+0x30f/0x700 scsi_cmd_ioctl+0x4d4/0x540 scsi_cmd_blk_ioctl+0x7b/0x8b sd_ioctl+0xba/0x150 blkdev_ioctl+0x6e1/0xea0 block_ioctl+0x79/0x90 do_vfs_ioctl+0x12b/0x9b0 ksys_ioctl+0x41/0x80 __x64_sys_ioctl+0x43/0x50 do_syscall_64+0x71/0x210 entry_SYSCALL_64_after_hwframe+0x49/0xbe Cc: Christoph Hellwig <hch@lst.de> Cc: Douglas Gilbert <dgilbert@interlog.com> Fixes: ae3d56d81507 ("scsi: remove bidirectional command support") Signed-off-by: Bart Van Assche <bvanassche@acm.org> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2019-02-08 21:21:27 +00:00
SDEB_I_WRITE_BUFFER = 25,
SDEB_I_WRITE_SAME = 26, /* 10, 16 */
SDEB_I_SYNC_CACHE = 27, /* 10, 16 */
SDEB_I_COMP_WRITE = 28,
SDEB_I_PRE_FETCH = 29, /* 10, 16 */
SDEB_I_ZONE_OUT = 30, /* 0x94+SA; includes no data xfer */
SDEB_I_ZONE_IN = 31, /* 0x95+SA; all have data-in */
SDEB_I_LAST_ELEM_P1 = 32, /* keep this last (previous + 1) */
};
static const unsigned char opcode_ind_arr[256] = {
/* 0x0; 0x0->0x1f: 6 byte cdbs */
SDEB_I_TEST_UNIT_READY, SDEB_I_REZERO_UNIT, 0, SDEB_I_REQUEST_SENSE,
0, 0, 0, 0,
SDEB_I_READ, 0, SDEB_I_WRITE, 0, 0, 0, 0, 0,
0, 0, SDEB_I_INQUIRY, 0, 0, SDEB_I_MODE_SELECT, SDEB_I_RESERVE,
SDEB_I_RELEASE,
0, 0, SDEB_I_MODE_SENSE, SDEB_I_START_STOP, 0, SDEB_I_SEND_DIAG,
SDEB_I_ALLOW_REMOVAL, 0,
/* 0x20; 0x20->0x3f: 10 byte cdbs */
0, 0, 0, 0, 0, SDEB_I_READ_CAPACITY, 0, 0,
SDEB_I_READ, 0, SDEB_I_WRITE, 0, 0, 0, 0, SDEB_I_VERIFY,
0, 0, 0, 0, SDEB_I_PRE_FETCH, SDEB_I_SYNC_CACHE, 0, 0,
0, 0, 0, SDEB_I_WRITE_BUFFER, 0, 0, 0, 0,
/* 0x40; 0x40->0x5f: 10 byte cdbs */
0, SDEB_I_WRITE_SAME, SDEB_I_UNMAP, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, SDEB_I_LOG_SENSE, 0, 0,
scsi: scsi_debug: Fix a recently introduced regression A recent commit removed an element from opcode_info_arr[] but did not modify opcode_ind_arr[] nor was SDEB_I_XDWRITEREAD removed. Remove SDEB_I_XDWRITEREAD and bring the two arrays again in sync. This patch avoids that the following is reported: BUG: KASAN: null-ptr-deref in scsi_debug_queuecommand+0x60f/0xc90 [scsi_debug] Read of size 1 at addr 0000000000000001 by task iscsi-test-cu/683 CPU: 3 PID: 683 Comm: iscsi-test-cu Not tainted 5.0.0-rc5-dbg+ #1 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014 Call Trace: dump_stack+0x86/0xca kasan_report.cold.3+0x5/0x3e __asan_load1+0x47/0x50 scsi_debug_queuecommand+0x60f/0xc90 [scsi_debug] scsi_queue_rq+0xc17/0x12e0 blk_mq_dispatch_rq_list+0x5fc/0xb10 blk_mq_sched_dispatch_requests+0x2f7/0x300 __blk_mq_run_hw_queue+0xd6/0x180 __blk_mq_delay_run_hw_queue+0x25c/0x290 blk_mq_run_hw_queue+0x119/0x1b0 blk_mq_sched_insert_request+0x274/0x350 blk_execute_rq_nowait+0x78/0x90 blk_execute_rq+0xcc/0x140 sg_io+0x30f/0x700 scsi_cmd_ioctl+0x4d4/0x540 scsi_cmd_blk_ioctl+0x7b/0x8b sd_ioctl+0xba/0x150 blkdev_ioctl+0x6e1/0xea0 block_ioctl+0x79/0x90 do_vfs_ioctl+0x12b/0x9b0 ksys_ioctl+0x41/0x80 __x64_sys_ioctl+0x43/0x50 do_syscall_64+0x71/0x210 entry_SYSCALL_64_after_hwframe+0x49/0xbe Cc: Christoph Hellwig <hch@lst.de> Cc: Douglas Gilbert <dgilbert@interlog.com> Fixes: ae3d56d81507 ("scsi: remove bidirectional command support") Signed-off-by: Bart Van Assche <bvanassche@acm.org> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2019-02-08 21:21:27 +00:00
0, 0, 0, 0, 0, SDEB_I_MODE_SELECT, SDEB_I_RESERVE,
SDEB_I_RELEASE,
0, 0, SDEB_I_MODE_SENSE, 0, 0, 0, 0, 0,
/* 0x60; 0x60->0x7d are reserved, 0x7e is "extended cdb" */
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, SDEB_I_VARIABLE_LEN,
/* 0x80; 0x80->0x9f: 16 byte cdbs */
0, 0, 0, 0, 0, SDEB_I_ATA_PT, 0, 0,
SDEB_I_READ, SDEB_I_COMP_WRITE, SDEB_I_WRITE, 0,
0, 0, 0, SDEB_I_VERIFY,
SDEB_I_PRE_FETCH, SDEB_I_SYNC_CACHE, 0, SDEB_I_WRITE_SAME,
SDEB_I_ZONE_OUT, SDEB_I_ZONE_IN, 0, 0,
0, 0, 0, 0, 0, 0, SDEB_I_SERV_ACT_IN_16, SDEB_I_SERV_ACT_OUT_16,
/* 0xa0; 0xa0->0xbf: 12 byte cdbs */
SDEB_I_REPORT_LUNS, SDEB_I_ATA_PT, 0, SDEB_I_MAINT_IN,
SDEB_I_MAINT_OUT, 0, 0, 0,
SDEB_I_READ, 0 /* SDEB_I_SERV_ACT_OUT_12 */, SDEB_I_WRITE,
0 /* SDEB_I_SERV_ACT_IN_12 */, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
/* 0xc0; 0xc0->0xff: vendor specific */
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
};
/*
* The following "response" functions return the SCSI mid-level's 4 byte
* tuple-in-an-int. To handle commands with an IMMED bit, for a faster
* command completion, they can mask their return value with
* SDEG_RES_IMMED_MASK .
*/
#define SDEG_RES_IMMED_MASK 0x40000000
static int resp_inquiry(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_report_luns(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_requests(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_mode_sense(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_mode_select(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_log_sense(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_readcap(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_read_dt0(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_write_dt0(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_write_scat(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_start_stop(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_readcap16(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_get_lba_status(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_report_tgtpgs(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_unmap(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_rsup_opcodes(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_rsup_tmfs(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_verify(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_write_same_10(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_write_same_16(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_comp_write(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_write_buffer(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_sync_cache(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_pre_fetch(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_report_zones(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_open_zone(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_close_zone(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_finish_zone(struct scsi_cmnd *, struct sdebug_dev_info *);
static int resp_rwp_zone(struct scsi_cmnd *, struct sdebug_dev_info *);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static int sdebug_do_add_host(bool mk_new_store);
static int sdebug_add_host_helper(int per_host_idx);
static void sdebug_do_remove_host(bool the_end);
static int sdebug_add_store(void);
static void sdebug_erase_store(int idx, struct sdeb_store_info *sip);
static void sdebug_erase_all_stores(bool apart_from_first);
static void sdebug_free_queued_cmd(struct sdebug_queued_cmd *sqcp);
/*
* The following are overflow arrays for cdbs that "hit" the same index in
* the opcode_info_arr array. The most time sensitive (or commonly used) cdb
* should be placed in opcode_info_arr[], the others should be placed here.
*/
static const struct opcode_info_t msense_iarr[] = {
{0, 0x1a, 0, F_D_IN, NULL, NULL,
{6, 0xe8, 0xff, 0xff, 0xff, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
};
static const struct opcode_info_t mselect_iarr[] = {
{0, 0x15, 0, F_D_OUT, NULL, NULL,
{6, 0xf1, 0, 0, 0xff, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
};
static const struct opcode_info_t read_iarr[] = {
{0, 0x28, 0, F_D_IN | FF_MEDIA_IO, resp_read_dt0, NULL,/* READ(10) */
{10, 0xff, 0xff, 0xff, 0xff, 0xff, 0x3f, 0xff, 0xff, 0xc7, 0, 0,
0, 0, 0, 0} },
{0, 0x8, 0, F_D_IN | FF_MEDIA_IO, resp_read_dt0, NULL, /* READ(6) */
{6, 0xff, 0xff, 0xff, 0xff, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
{0, 0xa8, 0, F_D_IN | FF_MEDIA_IO, resp_read_dt0, NULL,/* READ(12) */
{12, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xbf,
0xc7, 0, 0, 0, 0} },
};
static const struct opcode_info_t write_iarr[] = {
{0, 0x2a, 0, F_D_OUT | FF_MEDIA_IO, resp_write_dt0, /* WRITE(10) */
NULL, {10, 0xfb, 0xff, 0xff, 0xff, 0xff, 0x3f, 0xff, 0xff, 0xc7,
0, 0, 0, 0, 0, 0} },
{0, 0xa, 0, F_D_OUT | FF_MEDIA_IO, resp_write_dt0, /* WRITE(6) */
NULL, {6, 0xff, 0xff, 0xff, 0xff, 0xc7, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0} },
{0, 0xaa, 0, F_D_OUT | FF_MEDIA_IO, resp_write_dt0, /* WRITE(12) */
NULL, {12, 0xfb, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xbf, 0xc7, 0, 0, 0, 0} },
};
static const struct opcode_info_t verify_iarr[] = {
{0, 0x2f, 0, F_D_OUT_MAYBE | FF_MEDIA_IO, resp_verify,/* VERIFY(10) */
NULL, {10, 0xf7, 0xff, 0xff, 0xff, 0xff, 0xbf, 0xff, 0xff, 0xc7,
0, 0, 0, 0, 0, 0} },
};
static const struct opcode_info_t sa_in_16_iarr[] = {
{0, 0x9e, 0x12, F_SA_LOW | F_D_IN, resp_get_lba_status, NULL,
{16, 0x12, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0, 0xc7} }, /* GET LBA STATUS(16) */
};
static const struct opcode_info_t vl_iarr[] = { /* VARIABLE LENGTH */
{0, 0x7f, 0xb, F_SA_HIGH | F_D_OUT | FF_MEDIA_IO, resp_write_dt0,
NULL, {32, 0xc7, 0, 0, 0, 0, 0x3f, 0x18, 0x0, 0xb, 0xfa,
0, 0xff, 0xff, 0xff, 0xff} }, /* WRITE(32) */
{0, 0x7f, 0x11, F_SA_HIGH | F_D_OUT | FF_MEDIA_IO, resp_write_scat,
NULL, {32, 0xc7, 0, 0, 0, 0, 0x3f, 0x18, 0x0, 0x11, 0xf8,
0, 0xff, 0xff, 0x0, 0x0} }, /* WRITE SCATTERED(32) */
};
static const struct opcode_info_t maint_in_iarr[] = { /* MAINT IN */
{0, 0xa3, 0xc, F_SA_LOW | F_D_IN, resp_rsup_opcodes, NULL,
{12, 0xc, 0x87, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0,
0xc7, 0, 0, 0, 0} }, /* REPORT SUPPORTED OPERATION CODES */
{0, 0xa3, 0xd, F_SA_LOW | F_D_IN, resp_rsup_tmfs, NULL,
{12, 0xd, 0x80, 0, 0, 0, 0xff, 0xff, 0xff, 0xff, 0, 0xc7, 0, 0,
0, 0} }, /* REPORTED SUPPORTED TASK MANAGEMENT FUNCTIONS */
};
static const struct opcode_info_t write_same_iarr[] = {
{0, 0x93, 0, F_D_OUT_MAYBE | FF_MEDIA_IO, resp_write_same_16, NULL,
{16, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0x3f, 0xc7} }, /* WRITE SAME(16) */
};
static const struct opcode_info_t reserve_iarr[] = {
{0, 0x16, 0, F_D_OUT, NULL, NULL, /* RESERVE(6) */
{6, 0x1f, 0xff, 0xff, 0xff, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
};
static const struct opcode_info_t release_iarr[] = {
{0, 0x17, 0, F_D_OUT, NULL, NULL, /* RELEASE(6) */
{6, 0x1f, 0xff, 0, 0, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
};
static const struct opcode_info_t sync_cache_iarr[] = {
{0, 0x91, 0, F_SYNC_DELAY | F_M_ACCESS, resp_sync_cache, NULL,
{16, 0x6, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0xff, 0x3f, 0xc7} }, /* SYNC_CACHE (16) */
};
static const struct opcode_info_t pre_fetch_iarr[] = {
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
{0, 0x90, 0, F_SYNC_DELAY | FF_MEDIA_IO, resp_pre_fetch, NULL,
{16, 0x2, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0xff, 0x3f, 0xc7} }, /* PRE-FETCH (16) */
};
static const struct opcode_info_t zone_out_iarr[] = { /* ZONE OUT(16) */
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
{0, 0x94, 0x1, F_SA_LOW | F_M_ACCESS, resp_close_zone, NULL,
{16, 0x1, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0, 0, 0xff, 0xff, 0x1, 0xc7} }, /* CLOSE ZONE */
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
{0, 0x94, 0x2, F_SA_LOW | F_M_ACCESS, resp_finish_zone, NULL,
{16, 0x2, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0, 0, 0xff, 0xff, 0x1, 0xc7} }, /* FINISH ZONE */
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
{0, 0x94, 0x4, F_SA_LOW | F_M_ACCESS, resp_rwp_zone, NULL,
{16, 0x4, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0, 0, 0xff, 0xff, 0x1, 0xc7} }, /* RESET WRITE POINTER */
};
static const struct opcode_info_t zone_in_iarr[] = { /* ZONE IN(16) */
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
{0, 0x95, 0x6, F_SA_LOW | F_D_IN | F_M_ACCESS, NULL, NULL,
{16, 0x6, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0xff, 0x3f, 0xc7} }, /* REPORT ZONES */
};
/* This array is accessed via SDEB_I_* values. Make sure all are mapped,
* plus the terminating elements for logic that scans this table such as
* REPORT SUPPORTED OPERATION CODES. */
static const struct opcode_info_t opcode_info_arr[SDEB_I_LAST_ELEM_P1 + 1] = {
/* 0 */
{0, 0, 0, F_INV_OP | FF_RESPOND, NULL, NULL, /* unknown opcodes */
{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
{0, 0x12, 0, FF_RESPOND | F_D_IN, resp_inquiry, NULL, /* INQUIRY */
{6, 0xe3, 0xff, 0xff, 0xff, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
{0, 0xa0, 0, FF_RESPOND | F_D_IN, resp_report_luns, NULL,
{12, 0xe3, 0xff, 0, 0, 0, 0xff, 0xff, 0xff, 0xff, 0, 0xc7, 0, 0,
0, 0} }, /* REPORT LUNS */
{0, 0x3, 0, FF_RESPOND | F_D_IN, resp_requests, NULL,
{6, 0xe1, 0, 0, 0xff, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
{0, 0x0, 0, F_M_ACCESS | F_RL_WLUN_OK, NULL, NULL,/* TEST UNIT READY */
{6, 0, 0, 0, 0, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
/* 5 */
{ARRAY_SIZE(msense_iarr), 0x5a, 0, F_D_IN, /* MODE SENSE(10) */
resp_mode_sense, msense_iarr, {10, 0xf8, 0xff, 0xff, 0, 0, 0,
0xff, 0xff, 0xc7, 0, 0, 0, 0, 0, 0} },
{ARRAY_SIZE(mselect_iarr), 0x55, 0, F_D_OUT, /* MODE SELECT(10) */
resp_mode_select, mselect_iarr, {10, 0xf1, 0, 0, 0, 0, 0, 0xff,
0xff, 0xc7, 0, 0, 0, 0, 0, 0} },
{0, 0x4d, 0, F_D_IN, resp_log_sense, NULL, /* LOG SENSE */
{10, 0xe3, 0xff, 0xff, 0, 0xff, 0xff, 0xff, 0xff, 0xc7, 0, 0, 0,
0, 0, 0} },
{0, 0x25, 0, F_D_IN, resp_readcap, NULL, /* READ CAPACITY(10) */
{10, 0xe1, 0xff, 0xff, 0xff, 0xff, 0, 0, 0x1, 0xc7, 0, 0, 0, 0,
0, 0} },
{ARRAY_SIZE(read_iarr), 0x88, 0, F_D_IN | FF_MEDIA_IO, /* READ(16) */
resp_read_dt0, read_iarr, {16, 0xfe, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xc7} },
/* 10 */
{ARRAY_SIZE(write_iarr), 0x8a, 0, F_D_OUT | FF_MEDIA_IO,
resp_write_dt0, write_iarr, /* WRITE(16) */
{16, 0xfa, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0xff, 0xff, 0xc7} },
{0, 0x1b, 0, F_SSU_DELAY, resp_start_stop, NULL,/* START STOP UNIT */
{6, 0x1, 0, 0xf, 0xf7, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
{ARRAY_SIZE(sa_in_16_iarr), 0x9e, 0x10, F_SA_LOW | F_D_IN,
resp_readcap16, sa_in_16_iarr, /* SA_IN(16), READ CAPACITY(16) */
{16, 0x10, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0xff, 0x1, 0xc7} },
{0, 0x9f, 0x12, F_SA_LOW | F_D_OUT | FF_MEDIA_IO, resp_write_scat,
NULL, {16, 0x12, 0xf9, 0x0, 0xff, 0xff, 0, 0, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0xff, 0xc7} }, /* SA_OUT(16), WRITE SCAT(16) */
{ARRAY_SIZE(maint_in_iarr), 0xa3, 0xa, F_SA_LOW | F_D_IN,
resp_report_tgtpgs, /* MAINT IN, REPORT TARGET PORT GROUPS */
maint_in_iarr, {12, 0xea, 0, 0, 0, 0, 0xff, 0xff, 0xff,
0xff, 0, 0xc7, 0, 0, 0, 0} },
/* 15 */
{0, 0, 0, F_INV_OP | FF_RESPOND, NULL, NULL, /* MAINT OUT */
{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
{ARRAY_SIZE(verify_iarr), 0x8f, 0,
F_D_OUT_MAYBE | FF_MEDIA_IO, resp_verify, /* VERIFY(16) */
verify_iarr, {16, 0xf6, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x3f, 0xc7} },
{ARRAY_SIZE(vl_iarr), 0x7f, 0x9, F_SA_HIGH | F_D_IN | FF_MEDIA_IO,
resp_read_dt0, vl_iarr, /* VARIABLE LENGTH, READ(32) */
{32, 0xc7, 0, 0, 0, 0, 0x3f, 0x18, 0x0, 0x9, 0xfe, 0, 0xff, 0xff,
0xff, 0xff} },
{ARRAY_SIZE(reserve_iarr), 0x56, 0, F_D_OUT,
NULL, reserve_iarr, /* RESERVE(10) <no response function> */
{10, 0xff, 0xff, 0xff, 0, 0, 0, 0xff, 0xff, 0xc7, 0, 0, 0, 0, 0,
0} },
{ARRAY_SIZE(release_iarr), 0x57, 0, F_D_OUT,
NULL, release_iarr, /* RELEASE(10) <no response function> */
{10, 0x13, 0xff, 0xff, 0, 0, 0, 0xff, 0xff, 0xc7, 0, 0, 0, 0, 0,
0} },
/* 20 */
{0, 0x1e, 0, 0, NULL, NULL, /* ALLOW REMOVAL */
{6, 0, 0, 0, 0x3, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
{0, 0x1, 0, 0, resp_start_stop, NULL, /* REWIND ?? */
{6, 0x1, 0, 0, 0, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
{0, 0, 0, F_INV_OP | FF_RESPOND, NULL, NULL, /* ATA_PT */
{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
{0, 0x1d, F_D_OUT, 0, NULL, NULL, /* SEND DIAGNOSTIC */
{6, 0xf7, 0, 0xff, 0xff, 0xc7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
{0, 0x42, 0, F_D_OUT | FF_MEDIA_IO, resp_unmap, NULL, /* UNMAP */
{10, 0x1, 0, 0, 0, 0, 0x3f, 0xff, 0xff, 0xc7, 0, 0, 0, 0, 0, 0} },
/* 25 */
{0, 0x3b, 0, F_D_OUT_MAYBE, resp_write_buffer, NULL,
{10, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xc7, 0, 0,
0, 0, 0, 0} }, /* WRITE_BUFFER */
{ARRAY_SIZE(write_same_iarr), 0x41, 0, F_D_OUT_MAYBE | FF_MEDIA_IO,
resp_write_same_10, write_same_iarr, /* WRITE SAME(10) */
{10, 0xff, 0xff, 0xff, 0xff, 0xff, 0x3f, 0xff, 0xff, 0xc7, 0,
0, 0, 0, 0, 0} },
{ARRAY_SIZE(sync_cache_iarr), 0x35, 0, F_SYNC_DELAY | F_M_ACCESS,
resp_sync_cache, sync_cache_iarr,
{10, 0x7, 0xff, 0xff, 0xff, 0xff, 0x3f, 0xff, 0xff, 0xc7, 0, 0,
0, 0, 0, 0} }, /* SYNC_CACHE (10) */
{0, 0x89, 0, F_D_OUT | FF_MEDIA_IO, resp_comp_write, NULL,
{16, 0xf8, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0, 0,
0, 0xff, 0x3f, 0xc7} }, /* COMPARE AND WRITE */
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
{ARRAY_SIZE(pre_fetch_iarr), 0x34, 0, F_SYNC_DELAY | FF_MEDIA_IO,
resp_pre_fetch, pre_fetch_iarr,
{10, 0x2, 0xff, 0xff, 0xff, 0xff, 0x3f, 0xff, 0xff, 0xc7, 0, 0,
0, 0, 0, 0} }, /* PRE-FETCH (10) */
/* 30 */
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
{ARRAY_SIZE(zone_out_iarr), 0x94, 0x3, F_SA_LOW | F_M_ACCESS,
resp_open_zone, zone_out_iarr, /* ZONE_OUT(16), OPEN ZONE) */
{16, 0x3 /* SA */, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0x0, 0x0, 0xff, 0xff, 0x1, 0xc7} },
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
{ARRAY_SIZE(zone_in_iarr), 0x95, 0x0, F_SA_LOW | F_M_ACCESS,
resp_report_zones, zone_in_iarr, /* ZONE_IN(16), REPORT ZONES) */
{16, 0x0 /* SA */, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xbf, 0xc7} },
/* sentinel */
{0xff, 0, 0, 0, NULL, NULL, /* terminating element */
{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0} },
};
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
static int sdebug_num_hosts;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static int sdebug_add_host = DEF_NUM_HOST; /* in sysfs this is relative */
static int sdebug_ato = DEF_ATO;
static int sdebug_cdb_len = DEF_CDB_LEN;
static int sdebug_jdelay = DEF_JDELAY; /* if > 0 then unit is jiffies */
static int sdebug_dev_size_mb = DEF_DEV_SIZE_PRE_INIT;
static int sdebug_dif = DEF_DIF;
static int sdebug_dix = DEF_DIX;
static int sdebug_dsense = DEF_D_SENSE;
static int sdebug_every_nth = DEF_EVERY_NTH;
static int sdebug_fake_rw = DEF_FAKE_RW;
static unsigned int sdebug_guard = DEF_GUARD;
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
static int sdebug_host_max_queue; /* per host */
static int sdebug_lowest_aligned = DEF_LOWEST_ALIGNED;
static int sdebug_max_luns = DEF_MAX_LUNS;
static int sdebug_max_queue = SDEBUG_CANQUEUE; /* per submit queue */
static unsigned int sdebug_medium_error_start = OPT_MEDIUM_ERR_ADDR;
static int sdebug_medium_error_count = OPT_MEDIUM_ERR_NUM;
static int sdebug_ndelay = DEF_NDELAY; /* if > 0 then unit is nanoseconds */
static int sdebug_no_lun_0 = DEF_NO_LUN_0;
static int sdebug_no_uld;
static int sdebug_num_parts = DEF_NUM_PARTS;
static int sdebug_num_tgts = DEF_NUM_TGTS; /* targets per host */
static int sdebug_opt_blks = DEF_OPT_BLKS;
static int sdebug_opts = DEF_OPTS;
static int sdebug_physblk_exp = DEF_PHYSBLK_EXP;
static int sdebug_opt_xferlen_exp = DEF_OPT_XFERLEN_EXP;
static int sdebug_ptype = DEF_PTYPE; /* SCSI peripheral device type */
static int sdebug_scsi_level = DEF_SCSI_LEVEL;
static int sdebug_sector_size = DEF_SECTOR_SIZE;
static int sdeb_tur_ms_to_ready = DEF_TUR_MS_TO_READY;
static int sdebug_virtual_gb = DEF_VIRTUAL_GB;
static int sdebug_vpd_use_hostno = DEF_VPD_USE_HOSTNO;
static unsigned int sdebug_lbpu = DEF_LBPU;
static unsigned int sdebug_lbpws = DEF_LBPWS;
static unsigned int sdebug_lbpws10 = DEF_LBPWS10;
static unsigned int sdebug_lbprz = DEF_LBPRZ;
static unsigned int sdebug_unmap_alignment = DEF_UNMAP_ALIGNMENT;
static unsigned int sdebug_unmap_granularity = DEF_UNMAP_GRANULARITY;
static unsigned int sdebug_unmap_max_blocks = DEF_UNMAP_MAX_BLOCKS;
static unsigned int sdebug_unmap_max_desc = DEF_UNMAP_MAX_DESC;
static unsigned int sdebug_write_same_length = DEF_WRITESAME_LENGTH;
static int sdebug_uuid_ctl = DEF_UUID_CTL;
static bool sdebug_random = DEF_RANDOM;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static bool sdebug_per_host_store = DEF_PER_HOST_STORE;
static bool sdebug_removable = DEF_REMOVABLE;
static bool sdebug_clustering;
static bool sdebug_host_lock = DEF_HOST_LOCK;
static bool sdebug_strict = DEF_STRICT;
static bool sdebug_any_injecting_opt;
static bool sdebug_no_rwlock;
static bool sdebug_verbose;
static bool have_dif_prot;
static bool write_since_sync;
static bool sdebug_statistics = DEF_STATISTICS;
static bool sdebug_wp;
/* Following enum: 0: no zbc, def; 1: host aware; 2: host managed */
static enum blk_zoned_model sdeb_zbc_model = BLK_ZONED_NONE;
static char *sdeb_zbc_model_s;
enum sam_lun_addr_method {SAM_LUN_AM_PERIPHERAL = 0x0,
SAM_LUN_AM_FLAT = 0x1,
SAM_LUN_AM_LOGICAL_UNIT = 0x2,
SAM_LUN_AM_EXTENDED = 0x3};
static enum sam_lun_addr_method sdebug_lun_am = SAM_LUN_AM_PERIPHERAL;
static int sdebug_lun_am_i = (int)SAM_LUN_AM_PERIPHERAL;
static unsigned int sdebug_store_sectors;
static sector_t sdebug_capacity; /* in sectors */
/* old BIOS stuff, kernel may get rid of them but some mode sense pages
may still need them */
static int sdebug_heads; /* heads per disk */
static int sdebug_cylinders_per; /* cylinders per surface */
static int sdebug_sectors_per; /* sectors per cylinder */
static LIST_HEAD(sdebug_host_list);
static DEFINE_MUTEX(sdebug_host_list_mutex);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static struct xarray per_store_arr;
static struct xarray *per_store_ap = &per_store_arr;
static int sdeb_first_idx = -1; /* invalid index ==> none created */
static int sdeb_most_recent_idx = -1;
static DEFINE_RWLOCK(sdeb_fake_rw_lck); /* need a RW lock when fake_rw=1 */
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
static unsigned long map_size;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
static int num_aborts;
static int num_dev_resets;
static int num_target_resets;
static int num_bus_resets;
static int num_host_resets;
static int dix_writes;
static int dix_reads;
static int dif_errors;
/* ZBC global data */
static bool sdeb_zbc_in_use; /* true for host-aware and host-managed disks */
static int sdeb_zbc_zone_cap_mb;
static int sdeb_zbc_zone_size_mb;
static int sdeb_zbc_max_open = DEF_ZBC_MAX_OPEN_ZONES;
static int sdeb_zbc_nr_conv = DEF_ZBC_NR_CONV_ZONES;
static int submit_queues = DEF_SUBMIT_QUEUES; /* > 1 for multi-queue (mq) */
static int poll_queues; /* iouring iopoll interface.*/
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
static char sdebug_proc_name[] = MY_NAME;
static const char *my_name = MY_NAME;
static struct bus_type pseudo_lld_bus;
static struct device_driver sdebug_driverfs_driver = {
.name = sdebug_proc_name,
.bus = &pseudo_lld_bus,
};
static const int check_condition_result =
SAM_STAT_CHECK_CONDITION;
static const int illegal_condition_result =
(DID_ABORT << 16) | SAM_STAT_CHECK_CONDITION;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
static const int device_qfull_result =
(DID_ABORT << 16) | SAM_STAT_TASK_SET_FULL;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
static const int condition_met_result = SAM_STAT_CONDITION_MET;
/* Only do the extra work involved in logical block provisioning if one or
* more of the lbpu, lbpws or lbpws10 parameters are given and we are doing
* real reads and writes (i.e. not skipping them for speed).
*/
static inline bool scsi_debug_lbp(void)
{
return 0 == sdebug_fake_rw &&
(sdebug_lbpu || sdebug_lbpws || sdebug_lbpws10);
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static void *lba2fake_store(struct sdeb_store_info *sip,
unsigned long long lba)
{
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
struct sdeb_store_info *lsip = sip;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
lba = do_div(lba, sdebug_store_sectors);
if (!sip || !sip->storep) {
WARN_ON_ONCE(true);
lsip = xa_load(per_store_ap, 0); /* should never be NULL */
}
return lsip->storep + lba * sdebug_sector_size;
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static struct t10_pi_tuple *dif_store(struct sdeb_store_info *sip,
sector_t sector)
{
sector = sector_div(sector, sdebug_store_sectors);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
return sip->dif_storep + sector;
}
static void sdebug_max_tgts_luns(void)
{
struct sdebug_host_info *sdbg_host;
struct Scsi_Host *hpnt;
mutex_lock(&sdebug_host_list_mutex);
list_for_each_entry(sdbg_host, &sdebug_host_list, host_list) {
hpnt = sdbg_host->shost;
if ((hpnt->this_id >= 0) &&
(sdebug_num_tgts > hpnt->this_id))
hpnt->max_id = sdebug_num_tgts + 1;
else
hpnt->max_id = sdebug_num_tgts;
/* sdebug_max_luns; */
hpnt->max_lun = SCSI_W_LUN_REPORT_LUNS + 1;
}
mutex_unlock(&sdebug_host_list_mutex);
}
enum sdeb_cmd_data {SDEB_IN_DATA = 0, SDEB_IN_CDB = 1};
/* Set in_bit to -1 to indicate no bit position of invalid field */
static void mk_sense_invalid_fld(struct scsi_cmnd *scp,
enum sdeb_cmd_data c_d,
int in_byte, int in_bit)
{
unsigned char *sbuff;
u8 sks[4];
int sl, asc;
sbuff = scp->sense_buffer;
if (!sbuff) {
sdev_printk(KERN_ERR, scp->device,
"%s: sense_buffer is NULL\n", __func__);
return;
}
asc = c_d ? INVALID_FIELD_IN_CDB : INVALID_FIELD_IN_PARAM_LIST;
memset(sbuff, 0, SCSI_SENSE_BUFFERSIZE);
scsi_build_sense(scp, sdebug_dsense, ILLEGAL_REQUEST, asc, 0);
memset(sks, 0, sizeof(sks));
sks[0] = 0x80;
if (c_d)
sks[0] |= 0x40;
if (in_bit >= 0) {
sks[0] |= 0x8;
sks[0] |= 0x7 & in_bit;
}
put_unaligned_be16(in_byte, sks + 1);
if (sdebug_dsense) {
sl = sbuff[7] + 8;
sbuff[7] = sl;
sbuff[sl] = 0x2;
sbuff[sl + 1] = 0x6;
memcpy(sbuff + sl + 4, sks, 3);
} else
memcpy(sbuff + 15, sks, 3);
if (sdebug_verbose)
sdev_printk(KERN_INFO, scp->device, "%s: [sense_key,asc,ascq"
"]: [0x5,0x%x,0x0] %c byte=%d, bit=%d\n",
my_name, asc, c_d ? 'C' : 'D', in_byte, in_bit);
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
static void mk_sense_buffer(struct scsi_cmnd *scp, int key, int asc, int asq)
{
if (!scp->sense_buffer) {
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
sdev_printk(KERN_ERR, scp->device,
"%s: sense_buffer is NULL\n", __func__);
return;
}
memset(scp->sense_buffer, 0, SCSI_SENSE_BUFFERSIZE);
scsi_build_sense(scp, sdebug_dsense, key, asc, asq);
if (sdebug_verbose)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
sdev_printk(KERN_INFO, scp->device,
"%s: [sense_key,asc,ascq]: [0x%x,0x%x,0x%x]\n",
my_name, key, asc, asq);
}
static void mk_sense_invalid_opcode(struct scsi_cmnd *scp)
{
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_OPCODE, 0);
}
scsi: ata: Use unsigned int for cmd's type in ioctls in scsi_host_template Clang warns several times in the scsi subsystem (trimmed for brevity): drivers/scsi/hpsa.c:6209:7: warning: overflow converting case value to switch condition type (2147762695 to 18446744071562347015) [-Wswitch] case CCISS_GETBUSTYPES: ^ drivers/scsi/hpsa.c:6208:7: warning: overflow converting case value to switch condition type (2147762694 to 18446744071562347014) [-Wswitch] case CCISS_GETHEARTBEAT: ^ The root cause is that the _IOC macro can generate really large numbers, which don't fit into type 'int', which is used for the cmd parameter in the ioctls in scsi_host_template. My research into how GCC and Clang are handling this at a low level didn't prove fruitful. However, looking at the rest of the kernel tree, all ioctls use an 'unsigned int' for the cmd parameter, which will fit all of the _IOC values in the scsi/ata subsystems. Make that change because none of the ioctls expect a negative value for any command, it brings the ioctls inline with the reset of the kernel, and it removes ambiguity, which is never good when dealing with compilers. Link: https://github.com/ClangBuiltLinux/linux/issues/85 Link: https://github.com/ClangBuiltLinux/linux/issues/154 Link: https://github.com/ClangBuiltLinux/linux/issues/157 Signed-off-by: Nathan Chancellor <natechancellor@gmail.com> Acked-by: Bradley Grove <bgrove@attotech.com> Acked-by: Don Brace <don.brace@microsemi.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> Tested-by: Nick Desaulniers <ndesaulniers@google.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2019-02-07 16:07:20 +00:00
static int scsi_debug_ioctl(struct scsi_device *dev, unsigned int cmd,
void __user *arg)
{
if (sdebug_verbose) {
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
if (0x1261 == cmd)
sdev_printk(KERN_INFO, dev,
"%s: BLKFLSBUF [0x1261]\n", __func__);
else if (0x5331 == cmd)
sdev_printk(KERN_INFO, dev,
"%s: CDROM_GET_CAPABILITY [0x5331]\n",
__func__);
else
sdev_printk(KERN_INFO, dev, "%s: cmd=0x%x\n",
__func__, cmd);
}
return -EINVAL;
/* return -ENOTTY; // correct return but upsets fdisk */
}
static void config_cdb_len(struct scsi_device *sdev)
{
switch (sdebug_cdb_len) {
case 6: /* suggest 6 byte READ, WRITE and MODE SENSE/SELECT */
sdev->use_10_for_rw = false;
sdev->use_16_for_rw = false;
sdev->use_10_for_ms = false;
break;
case 10: /* suggest 10 byte RWs and 6 byte MODE SENSE/SELECT */
sdev->use_10_for_rw = true;
sdev->use_16_for_rw = false;
sdev->use_10_for_ms = false;
break;
case 12: /* suggest 10 byte RWs and 10 byte MODE SENSE/SELECT */
sdev->use_10_for_rw = true;
sdev->use_16_for_rw = false;
sdev->use_10_for_ms = true;
break;
case 16:
sdev->use_10_for_rw = false;
sdev->use_16_for_rw = true;
sdev->use_10_for_ms = true;
break;
case 32: /* No knobs to suggest this so same as 16 for now */
sdev->use_10_for_rw = false;
sdev->use_16_for_rw = true;
sdev->use_10_for_ms = true;
break;
default:
pr_warn("unexpected cdb_len=%d, force to 10\n",
sdebug_cdb_len);
sdev->use_10_for_rw = true;
sdev->use_16_for_rw = false;
sdev->use_10_for_ms = false;
sdebug_cdb_len = 10;
break;
}
}
static void all_config_cdb_len(void)
{
struct sdebug_host_info *sdbg_host;
struct Scsi_Host *shost;
struct scsi_device *sdev;
mutex_lock(&sdebug_host_list_mutex);
list_for_each_entry(sdbg_host, &sdebug_host_list, host_list) {
shost = sdbg_host->shost;
shost_for_each_device(sdev, shost) {
config_cdb_len(sdev);
}
}
mutex_unlock(&sdebug_host_list_mutex);
}
static void clear_luns_changed_on_target(struct sdebug_dev_info *devip)
{
struct sdebug_host_info *sdhp = devip->sdbg_host;
struct sdebug_dev_info *dp;
list_for_each_entry(dp, &sdhp->dev_info_list, dev_list) {
if ((devip->sdbg_host == dp->sdbg_host) &&
(devip->target == dp->target)) {
clear_bit(SDEBUG_UA_LUNS_CHANGED, dp->uas_bm);
}
}
}
static int make_ua(struct scsi_cmnd *scp, struct sdebug_dev_info *devip)
{
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
int k;
k = find_first_bit(devip->uas_bm, SDEBUG_NUM_UAS);
if (k != SDEBUG_NUM_UAS) {
const char *cp = NULL;
switch (k) {
case SDEBUG_UA_POR:
mk_sense_buffer(scp, UNIT_ATTENTION, UA_RESET_ASC,
POWER_ON_RESET_ASCQ);
if (sdebug_verbose)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
cp = "power on reset";
break;
case SDEBUG_UA_POOCCUR:
mk_sense_buffer(scp, UNIT_ATTENTION, UA_RESET_ASC,
POWER_ON_OCCURRED_ASCQ);
if (sdebug_verbose)
cp = "power on occurred";
break;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
case SDEBUG_UA_BUS_RESET:
mk_sense_buffer(scp, UNIT_ATTENTION, UA_RESET_ASC,
BUS_RESET_ASCQ);
if (sdebug_verbose)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
cp = "bus reset";
break;
case SDEBUG_UA_MODE_CHANGED:
mk_sense_buffer(scp, UNIT_ATTENTION, UA_CHANGED_ASC,
MODE_CHANGED_ASCQ);
if (sdebug_verbose)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
cp = "mode parameters changed";
break;
case SDEBUG_UA_CAPACITY_CHANGED:
mk_sense_buffer(scp, UNIT_ATTENTION, UA_CHANGED_ASC,
CAPACITY_CHANGED_ASCQ);
if (sdebug_verbose)
cp = "capacity data changed";
break;
case SDEBUG_UA_MICROCODE_CHANGED:
mk_sense_buffer(scp, UNIT_ATTENTION,
TARGET_CHANGED_ASC,
MICROCODE_CHANGED_ASCQ);
if (sdebug_verbose)
cp = "microcode has been changed";
break;
case SDEBUG_UA_MICROCODE_CHANGED_WO_RESET:
mk_sense_buffer(scp, UNIT_ATTENTION,
TARGET_CHANGED_ASC,
MICROCODE_CHANGED_WO_RESET_ASCQ);
if (sdebug_verbose)
cp = "microcode has been changed without reset";
break;
case SDEBUG_UA_LUNS_CHANGED:
/*
* SPC-3 behavior is to report a UNIT ATTENTION with
* ASC/ASCQ REPORTED LUNS DATA HAS CHANGED on every LUN
* on the target, until a REPORT LUNS command is
* received. SPC-4 behavior is to report it only once.
* NOTE: sdebug_scsi_level does not use the same
* values as struct scsi_device->scsi_level.
*/
if (sdebug_scsi_level >= 6) /* SPC-4 and above */
clear_luns_changed_on_target(devip);
mk_sense_buffer(scp, UNIT_ATTENTION,
TARGET_CHANGED_ASC,
LUNS_CHANGED_ASCQ);
if (sdebug_verbose)
cp = "reported luns data has changed";
break;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
default:
pr_warn("unexpected unit attention code=%d\n", k);
if (sdebug_verbose)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
cp = "unknown";
break;
}
clear_bit(k, devip->uas_bm);
if (sdebug_verbose)
sdev_printk(KERN_INFO, scp->device,
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
"%s reports: Unit attention: %s\n",
my_name, cp);
return check_condition_result;
}
return 0;
}
/* Build SCSI "data-in" buffer. Returns 0 if ok else (DID_ERROR << 16). */
static int fill_from_dev_buffer(struct scsi_cmnd *scp, unsigned char *arr,
int arr_len)
{
int act_len;
struct scsi_data_buffer *sdb = &scp->sdb;
if (!sdb->length)
return 0;
if (scp->sc_data_direction != DMA_FROM_DEVICE)
return DID_ERROR << 16;
act_len = sg_copy_from_buffer(sdb->table.sgl, sdb->table.nents,
arr, arr_len);
scsi_set_resid(scp, scsi_bufflen(scp) - act_len);
return 0;
}
/* Partial build of SCSI "data-in" buffer. Returns 0 if ok else
* (DID_ERROR << 16). Can write to offset in data-in buffer. If multiple
* calls, not required to write in ascending offset order. Assumes resid
* set to scsi_bufflen() prior to any calls.
*/
static int p_fill_from_dev_buffer(struct scsi_cmnd *scp, const void *arr,
int arr_len, unsigned int off_dst)
{
unsigned int act_len, n;
struct scsi_data_buffer *sdb = &scp->sdb;
off_t skip = off_dst;
if (sdb->length <= off_dst)
return 0;
if (scp->sc_data_direction != DMA_FROM_DEVICE)
return DID_ERROR << 16;
act_len = sg_pcopy_from_buffer(sdb->table.sgl, sdb->table.nents,
arr, arr_len, skip);
pr_debug("%s: off_dst=%u, scsi_bufflen=%u, act_len=%u, resid=%d\n",
__func__, off_dst, scsi_bufflen(scp), act_len,
scsi_get_resid(scp));
n = scsi_bufflen(scp) - (off_dst + act_len);
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
scsi_set_resid(scp, min_t(u32, scsi_get_resid(scp), n));
return 0;
}
/* Fetches from SCSI "data-out" buffer. Returns number of bytes fetched into
* 'arr' or -1 if error.
*/
static int fetch_to_dev_buffer(struct scsi_cmnd *scp, unsigned char *arr,
int arr_len)
{
if (!scsi_bufflen(scp))
return 0;
if (scp->sc_data_direction != DMA_TO_DEVICE)
return -1;
return scsi_sg_copy_to_buffer(scp, arr, arr_len);
}
static char sdebug_inq_vendor_id[9] = "Linux ";
static char sdebug_inq_product_id[17] = "scsi_debug ";
static char sdebug_inq_product_rev[5] = SDEBUG_VERSION;
/* Use some locally assigned NAAs for SAS addresses. */
static const u64 naa3_comp_a = 0x3222222000000000ULL;
static const u64 naa3_comp_b = 0x3333333000000000ULL;
static const u64 naa3_comp_c = 0x3111111000000000ULL;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
/* Device identification VPD page. Returns number of bytes placed in arr */
static int inquiry_vpd_83(unsigned char *arr, int port_group_id,
int target_dev_id, int dev_id_num,
const char *dev_id_str, int dev_id_str_len,
const uuid_t *lu_name)
{
int num, port_a;
char b[32];
port_a = target_dev_id + 1;
/* T10 vendor identifier field format (faked) */
arr[0] = 0x2; /* ASCII */
arr[1] = 0x1;
arr[2] = 0x0;
memcpy(&arr[4], sdebug_inq_vendor_id, 8);
memcpy(&arr[12], sdebug_inq_product_id, 16);
memcpy(&arr[28], dev_id_str, dev_id_str_len);
num = 8 + 16 + dev_id_str_len;
arr[3] = num;
num += 4;
if (dev_id_num >= 0) {
if (sdebug_uuid_ctl) {
/* Locally assigned UUID */
arr[num++] = 0x1; /* binary (not necessarily sas) */
arr[num++] = 0xa; /* PIV=0, lu, naa */
arr[num++] = 0x0;
arr[num++] = 0x12;
arr[num++] = 0x10; /* uuid type=1, locally assigned */
arr[num++] = 0x0;
memcpy(arr + num, lu_name, 16);
num += 16;
} else {
/* NAA-3, Logical unit identifier (binary) */
arr[num++] = 0x1; /* binary (not necessarily sas) */
arr[num++] = 0x3; /* PIV=0, lu, naa */
arr[num++] = 0x0;
arr[num++] = 0x8;
put_unaligned_be64(naa3_comp_b + dev_id_num, arr + num);
num += 8;
}
/* Target relative port number */
arr[num++] = 0x61; /* proto=sas, binary */
arr[num++] = 0x94; /* PIV=1, target port, rel port */
arr[num++] = 0x0; /* reserved */
arr[num++] = 0x4; /* length */
arr[num++] = 0x0; /* reserved */
arr[num++] = 0x0; /* reserved */
arr[num++] = 0x0;
arr[num++] = 0x1; /* relative port A */
}
/* NAA-3, Target port identifier */
arr[num++] = 0x61; /* proto=sas, binary */
arr[num++] = 0x93; /* piv=1, target port, naa */
arr[num++] = 0x0;
arr[num++] = 0x8;
put_unaligned_be64(naa3_comp_a + port_a, arr + num);
num += 8;
/* NAA-3, Target port group identifier */
arr[num++] = 0x61; /* proto=sas, binary */
arr[num++] = 0x95; /* piv=1, target port group id */
arr[num++] = 0x0;
arr[num++] = 0x4;
arr[num++] = 0;
arr[num++] = 0;
put_unaligned_be16(port_group_id, arr + num);
num += 2;
/* NAA-3, Target device identifier */
arr[num++] = 0x61; /* proto=sas, binary */
arr[num++] = 0xa3; /* piv=1, target device, naa */
arr[num++] = 0x0;
arr[num++] = 0x8;
put_unaligned_be64(naa3_comp_a + target_dev_id, arr + num);
num += 8;
/* SCSI name string: Target device identifier */
arr[num++] = 0x63; /* proto=sas, UTF-8 */
arr[num++] = 0xa8; /* piv=1, target device, SCSI name string */
arr[num++] = 0x0;
arr[num++] = 24;
memcpy(arr + num, "naa.32222220", 12);
num += 12;
snprintf(b, sizeof(b), "%08X", target_dev_id);
memcpy(arr + num, b, 8);
num += 8;
memset(arr + num, 0, 4);
num += 4;
return num;
}
static unsigned char vpd84_data[] = {
/* from 4th byte */ 0x22,0x22,0x22,0x0,0xbb,0x0,
0x22,0x22,0x22,0x0,0xbb,0x1,
0x22,0x22,0x22,0x0,0xbb,0x2,
};
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
/* Software interface identification VPD page */
static int inquiry_vpd_84(unsigned char *arr)
{
memcpy(arr, vpd84_data, sizeof(vpd84_data));
return sizeof(vpd84_data);
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
/* Management network addresses VPD page */
static int inquiry_vpd_85(unsigned char *arr)
{
int num = 0;
const char *na1 = "https://www.kernel.org/config";
const char *na2 = "http://www.kernel.org/log";
int plen, olen;
arr[num++] = 0x1; /* lu, storage config */
arr[num++] = 0x0; /* reserved */
arr[num++] = 0x0;
olen = strlen(na1);
plen = olen + 1;
if (plen % 4)
plen = ((plen / 4) + 1) * 4;
arr[num++] = plen; /* length, null termianted, padded */
memcpy(arr + num, na1, olen);
memset(arr + num + olen, 0, plen - olen);
num += plen;
arr[num++] = 0x4; /* lu, logging */
arr[num++] = 0x0; /* reserved */
arr[num++] = 0x0;
olen = strlen(na2);
plen = olen + 1;
if (plen % 4)
plen = ((plen / 4) + 1) * 4;
arr[num++] = plen; /* length, null terminated, padded */
memcpy(arr + num, na2, olen);
memset(arr + num + olen, 0, plen - olen);
num += plen;
return num;
}
/* SCSI ports VPD page */
static int inquiry_vpd_88(unsigned char *arr, int target_dev_id)
{
int num = 0;
int port_a, port_b;
port_a = target_dev_id + 1;
port_b = port_a + 1;
arr[num++] = 0x0; /* reserved */
arr[num++] = 0x0; /* reserved */
arr[num++] = 0x0;
arr[num++] = 0x1; /* relative port 1 (primary) */
memset(arr + num, 0, 6);
num += 6;
arr[num++] = 0x0;
arr[num++] = 12; /* length tp descriptor */
/* naa-5 target port identifier (A) */
arr[num++] = 0x61; /* proto=sas, binary */
arr[num++] = 0x93; /* PIV=1, target port, NAA */
arr[num++] = 0x0; /* reserved */
arr[num++] = 0x8; /* length */
put_unaligned_be64(naa3_comp_a + port_a, arr + num);
num += 8;
arr[num++] = 0x0; /* reserved */
arr[num++] = 0x0; /* reserved */
arr[num++] = 0x0;
arr[num++] = 0x2; /* relative port 2 (secondary) */
memset(arr + num, 0, 6);
num += 6;
arr[num++] = 0x0;
arr[num++] = 12; /* length tp descriptor */
/* naa-5 target port identifier (B) */
arr[num++] = 0x61; /* proto=sas, binary */
arr[num++] = 0x93; /* PIV=1, target port, NAA */
arr[num++] = 0x0; /* reserved */
arr[num++] = 0x8; /* length */
put_unaligned_be64(naa3_comp_a + port_b, arr + num);
num += 8;
return num;
}
static unsigned char vpd89_data[] = {
/* from 4th byte */ 0,0,0,0,
'l','i','n','u','x',' ',' ',' ',
'S','A','T',' ','s','c','s','i','_','d','e','b','u','g',' ',' ',
'1','2','3','4',
0x34,0,0,0,1,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,
0xec,0,0,0,
0x5a,0xc,0xff,0x3f,0x37,0xc8,0x10,0,0,0,0,0,0x3f,0,0,0,
0,0,0,0,0x58,0x58,0x58,0x58,0x58,0x58,0x58,0x58,0x20,0x20,0x20,0x20,
0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0,0,0,0x40,0x4,0,0x2e,0x33,
0x38,0x31,0x20,0x20,0x20,0x20,0x54,0x53,0x38,0x33,0x30,0x30,0x33,0x31,
0x53,0x41,
0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,
0x20,0x20,
0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,0x20,
0x10,0x80,
0,0,0,0x2f,0,0,0,0x2,0,0x2,0x7,0,0xff,0xff,0x1,0,
0x3f,0,0xc1,0xff,0x3e,0,0x10,0x1,0xb0,0xf8,0x50,0x9,0,0,0x7,0,
0x3,0,0x78,0,0x78,0,0xf0,0,0x78,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0x2,0,0,0,0,0,0,0,
0x7e,0,0x1b,0,0x6b,0x34,0x1,0x7d,0x3,0x40,0x69,0x34,0x1,0x3c,0x3,0x40,
0x7f,0x40,0,0,0,0,0xfe,0xfe,0,0,0,0,0,0xfe,0,0,
0,0,0,0,0,0,0,0,0xb0,0xf8,0x50,0x9,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0x1,0,0xb0,0xf8,0x50,0x9,0xb0,0xf8,0x50,0x9,0x20,0x20,0x2,0,0xb6,0x42,
0,0x80,0x8a,0,0x6,0x3c,0xa,0x3c,0xff,0xff,0xc6,0x7,0,0x1,0,0x8,
0xf0,0xf,0,0x10,0x2,0,0x30,0,0,0,0,0,0,0,0x6,0xfe,
0,0,0x2,0,0x50,0,0x8a,0,0x4f,0x95,0,0,0x21,0,0xb,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0xa5,0x51,
};
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
/* ATA Information VPD page */
static int inquiry_vpd_89(unsigned char *arr)
{
memcpy(arr, vpd89_data, sizeof(vpd89_data));
return sizeof(vpd89_data);
}
static unsigned char vpdb0_data[] = {
/* from 4th byte */ 0,0,0,4, 0,0,0x4,0, 0,0,0,64,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
};
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
/* Block limits VPD page (SBC-3) */
static int inquiry_vpd_b0(unsigned char *arr)
{
unsigned int gran;
memcpy(arr, vpdb0_data, sizeof(vpdb0_data));
/* Optimal transfer length granularity */
if (sdebug_opt_xferlen_exp != 0 &&
sdebug_physblk_exp < sdebug_opt_xferlen_exp)
gran = 1 << sdebug_opt_xferlen_exp;
else
gran = 1 << sdebug_physblk_exp;
put_unaligned_be16(gran, arr + 2);
/* Maximum Transfer Length */
if (sdebug_store_sectors > 0x400)
put_unaligned_be32(sdebug_store_sectors, arr + 4);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
/* Optimal Transfer Length */
put_unaligned_be32(sdebug_opt_blks, &arr[8]);
if (sdebug_lbpu) {
/* Maximum Unmap LBA Count */
put_unaligned_be32(sdebug_unmap_max_blocks, &arr[16]);
/* Maximum Unmap Block Descriptor Count */
put_unaligned_be32(sdebug_unmap_max_desc, &arr[20]);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
}
/* Unmap Granularity Alignment */
if (sdebug_unmap_alignment) {
put_unaligned_be32(sdebug_unmap_alignment, &arr[28]);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
arr[28] |= 0x80; /* UGAVALID */
}
/* Optimal Unmap Granularity */
put_unaligned_be32(sdebug_unmap_granularity, &arr[24]);
/* Maximum WRITE SAME Length */
put_unaligned_be64(sdebug_write_same_length, &arr[32]);
return 0x3c; /* Mandatory page length for Logical Block Provisioning */
}
/* Block device characteristics VPD page (SBC-3) */
static int inquiry_vpd_b1(struct sdebug_dev_info *devip, unsigned char *arr)
{
memset(arr, 0, 0x3c);
arr[0] = 0;
arr[1] = 1; /* non rotating medium (e.g. solid state) */
arr[2] = 0;
arr[3] = 5; /* less than 1.8" */
if (devip->zmodel == BLK_ZONED_HA)
arr[4] = 1 << 4; /* zoned field = 01b */
return 0x3c;
}
/* Logical block provisioning VPD page (SBC-4) */
static int inquiry_vpd_b2(unsigned char *arr)
{
memset(arr, 0, 0x4);
arr[0] = 0; /* threshold exponent */
if (sdebug_lbpu)
arr[1] = 1 << 7;
if (sdebug_lbpws)
arr[1] |= 1 << 6;
if (sdebug_lbpws10)
arr[1] |= 1 << 5;
if (sdebug_lbprz && scsi_debug_lbp())
arr[1] |= (sdebug_lbprz & 0x7) << 2; /* sbc4r07 and later */
/* anc_sup=0; dp=0 (no provisioning group descriptor) */
/* minimum_percentage=0; provisioning_type=0 (unknown) */
/* threshold_percentage=0 */
return 0x4;
}
/* Zoned block device characteristics VPD page (ZBC mandatory) */
static int inquiry_vpd_b6(struct sdebug_dev_info *devip, unsigned char *arr)
{
memset(arr, 0, 0x3c);
arr[0] = 0x1; /* set URSWRZ (unrestricted read in seq. wr req zone) */
/*
* Set Optimal number of open sequential write preferred zones and
* Optimal number of non-sequentially written sequential write
* preferred zones fields to 'not reported' (0xffffffff). Leave other
* fields set to zero, apart from Max. number of open swrz_s field.
*/
put_unaligned_be32(0xffffffff, &arr[4]);
put_unaligned_be32(0xffffffff, &arr[8]);
if (sdeb_zbc_model == BLK_ZONED_HM && devip->max_open)
put_unaligned_be32(devip->max_open, &arr[12]);
else
put_unaligned_be32(0xffffffff, &arr[12]);
if (devip->zcap < devip->zsize) {
arr[19] = ZBC_CONSTANT_ZONE_START_OFFSET;
put_unaligned_be64(devip->zsize, &arr[20]);
} else {
arr[19] = 0;
}
return 0x3c;
}
#define SDEBUG_LONG_INQ_SZ 96
#define SDEBUG_MAX_INQ_ARR_SZ 584
static int resp_inquiry(struct scsi_cmnd *scp, struct sdebug_dev_info *devip)
{
unsigned char pq_pdt;
unsigned char *arr;
unsigned char *cmd = scp->cmnd;
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
u32 alloc_len, n;
int ret;
bool have_wlun, is_disk, is_zbc, is_disk_zbc;
alloc_len = get_unaligned_be16(cmd + 3);
arr = kzalloc(SDEBUG_MAX_INQ_ARR_SZ, GFP_ATOMIC);
if (! arr)
return DID_REQUEUE << 16;
is_disk = (sdebug_ptype == TYPE_DISK);
is_zbc = (devip->zmodel != BLK_ZONED_NONE);
is_disk_zbc = (is_disk || is_zbc);
have_wlun = scsi_is_wlun(scp->device->lun);
if (have_wlun)
pq_pdt = TYPE_WLUN; /* present, wlun */
else if (sdebug_no_lun_0 && (devip->lun == SDEBUG_LUN_0_VAL))
pq_pdt = 0x7f; /* not present, PQ=3, PDT=0x1f */
else
pq_pdt = (sdebug_ptype & 0x1f);
arr[0] = pq_pdt;
if (0x2 & cmd[1]) { /* CMDDT bit set */
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 1, 1);
kfree(arr);
return check_condition_result;
} else if (0x1 & cmd[1]) { /* EVPD bit set */
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
int lu_id_num, port_group_id, target_dev_id;
u32 len;
char lu_id_str[6];
int host_no = devip->sdbg_host->shost->host_no;
port_group_id = (((host_no + 1) & 0x7f) << 8) +
(devip->channel & 0x7f);
if (sdebug_vpd_use_hostno == 0)
host_no = 0;
lu_id_num = have_wlun ? -1 : (((host_no + 1) * 2000) +
(devip->target * 1000) + devip->lun);
target_dev_id = ((host_no + 1) * 2000) +
(devip->target * 1000) - 3;
len = scnprintf(lu_id_str, 6, "%d", lu_id_num);
if (0 == cmd[2]) { /* supported vital product data pages */
arr[1] = cmd[2]; /*sanity */
n = 4;
arr[n++] = 0x0; /* this page */
arr[n++] = 0x80; /* unit serial number */
arr[n++] = 0x83; /* device identification */
arr[n++] = 0x84; /* software interface ident. */
arr[n++] = 0x85; /* management network addresses */
arr[n++] = 0x86; /* extended inquiry */
arr[n++] = 0x87; /* mode page policy */
arr[n++] = 0x88; /* SCSI ports */
if (is_disk_zbc) { /* SBC or ZBC */
arr[n++] = 0x89; /* ATA information */
arr[n++] = 0xb0; /* Block limits */
arr[n++] = 0xb1; /* Block characteristics */
if (is_disk)
arr[n++] = 0xb2; /* LB Provisioning */
if (is_zbc)
arr[n++] = 0xb6; /* ZB dev. char. */
}
arr[3] = n - 4; /* number of supported VPD pages */
} else if (0x80 == cmd[2]) { /* unit serial number */
arr[1] = cmd[2]; /*sanity */
arr[3] = len;
memcpy(&arr[4], lu_id_str, len);
} else if (0x83 == cmd[2]) { /* device identification */
arr[1] = cmd[2]; /*sanity */
arr[3] = inquiry_vpd_83(&arr[4], port_group_id,
target_dev_id, lu_id_num,
lu_id_str, len,
&devip->lu_name);
} else if (0x84 == cmd[2]) { /* Software interface ident. */
arr[1] = cmd[2]; /*sanity */
arr[3] = inquiry_vpd_84(&arr[4]);
} else if (0x85 == cmd[2]) { /* Management network addresses */
arr[1] = cmd[2]; /*sanity */
arr[3] = inquiry_vpd_85(&arr[4]);
} else if (0x86 == cmd[2]) { /* extended inquiry */
arr[1] = cmd[2]; /*sanity */
arr[3] = 0x3c; /* number of following entries */
if (sdebug_dif == T10_PI_TYPE3_PROTECTION)
arr[4] = 0x4; /* SPT: GRD_CHK:1 */
else if (have_dif_prot)
arr[4] = 0x5; /* SPT: GRD_CHK:1, REF_CHK:1 */
else
arr[4] = 0x0; /* no protection stuff */
arr[5] = 0x7; /* head of q, ordered + simple q's */
} else if (0x87 == cmd[2]) { /* mode page policy */
arr[1] = cmd[2]; /*sanity */
arr[3] = 0x8; /* number of following entries */
arr[4] = 0x2; /* disconnect-reconnect mp */
arr[6] = 0x80; /* mlus, shared */
arr[8] = 0x18; /* protocol specific lu */
arr[10] = 0x82; /* mlus, per initiator port */
} else if (0x88 == cmd[2]) { /* SCSI Ports */
arr[1] = cmd[2]; /*sanity */
arr[3] = inquiry_vpd_88(&arr[4], target_dev_id);
} else if (is_disk_zbc && 0x89 == cmd[2]) { /* ATA info */
arr[1] = cmd[2]; /*sanity */
n = inquiry_vpd_89(&arr[4]);
put_unaligned_be16(n, arr + 2);
} else if (is_disk_zbc && 0xb0 == cmd[2]) { /* Block limits */
arr[1] = cmd[2]; /*sanity */
arr[3] = inquiry_vpd_b0(&arr[4]);
} else if (is_disk_zbc && 0xb1 == cmd[2]) { /* Block char. */
arr[1] = cmd[2]; /*sanity */
arr[3] = inquiry_vpd_b1(devip, &arr[4]);
} else if (is_disk && 0xb2 == cmd[2]) { /* LB Prov. */
arr[1] = cmd[2]; /*sanity */
arr[3] = inquiry_vpd_b2(&arr[4]);
} else if (is_zbc && cmd[2] == 0xb6) { /* ZB dev. charact. */
arr[1] = cmd[2]; /*sanity */
arr[3] = inquiry_vpd_b6(devip, &arr[4]);
} else {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 2, -1);
kfree(arr);
return check_condition_result;
}
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
len = min_t(u32, get_unaligned_be16(arr + 2) + 4, alloc_len);
ret = fill_from_dev_buffer(scp, arr,
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
min_t(u32, len, SDEBUG_MAX_INQ_ARR_SZ));
kfree(arr);
return ret;
}
/* drops through here for a standard inquiry */
arr[1] = sdebug_removable ? 0x80 : 0; /* Removable disk */
arr[2] = sdebug_scsi_level;
arr[3] = 2; /* response_data_format==2 */
arr[4] = SDEBUG_LONG_INQ_SZ - 5;
arr[5] = (int)have_dif_prot; /* PROTECT bit */
if (sdebug_vpd_use_hostno == 0)
arr[5] |= 0x10; /* claim: implicit TPGS */
arr[6] = 0x10; /* claim: MultiP */
/* arr[6] |= 0x40; ... claim: EncServ (enclosure services) */
arr[7] = 0xa; /* claim: LINKED + CMDQUE */
memcpy(&arr[8], sdebug_inq_vendor_id, 8);
memcpy(&arr[16], sdebug_inq_product_id, 16);
memcpy(&arr[32], sdebug_inq_product_rev, 4);
/* Use Vendor Specific area to place driver date in ASCII hex */
memcpy(&arr[36], sdebug_version_date, 8);
/* version descriptors (2 bytes each) follow */
put_unaligned_be16(0xc0, arr + 58); /* SAM-6 no version claimed */
put_unaligned_be16(0x5c0, arr + 60); /* SPC-5 no version claimed */
n = 62;
if (is_disk) { /* SBC-4 no version claimed */
put_unaligned_be16(0x600, arr + n);
n += 2;
} else if (sdebug_ptype == TYPE_TAPE) { /* SSC-4 rev 3 */
put_unaligned_be16(0x525, arr + n);
n += 2;
} else if (is_zbc) { /* ZBC BSR INCITS 536 revision 05 */
put_unaligned_be16(0x624, arr + n);
n += 2;
}
put_unaligned_be16(0x2100, arr + n); /* SPL-4 no version claimed */
ret = fill_from_dev_buffer(scp, arr,
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
min_t(u32, alloc_len, SDEBUG_LONG_INQ_SZ));
kfree(arr);
return ret;
}
/* See resp_iec_m_pg() for how this data is manipulated */
static unsigned char iec_m_pg[] = {0x1c, 0xa, 0x08, 0, 0, 0, 0, 0,
0, 0, 0x0, 0x0};
static int resp_requests(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
unsigned char *cmd = scp->cmnd;
unsigned char arr[SCSI_SENSE_BUFFERSIZE]; /* assume >= 18 bytes */
bool dsense = !!(cmd[1] & 1);
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
u32 alloc_len = cmd[4];
u32 len = 18;
int stopped_state = atomic_read(&devip->stopped);
memset(arr, 0, sizeof(arr));
if (stopped_state > 0) { /* some "pollable" data [spc6r02: 5.12.2] */
if (dsense) {
arr[0] = 0x72;
arr[1] = NOT_READY;
arr[2] = LOGICAL_UNIT_NOT_READY;
arr[3] = (stopped_state == 2) ? 0x1 : 0x2;
len = 8;
} else {
arr[0] = 0x70;
arr[2] = NOT_READY; /* NO_SENSE in sense_key */
arr[7] = 0xa; /* 18 byte sense buffer */
arr[12] = LOGICAL_UNIT_NOT_READY;
arr[13] = (stopped_state == 2) ? 0x1 : 0x2;
}
} else if ((iec_m_pg[2] & 0x4) && (6 == (iec_m_pg[3] & 0xf))) {
/* Information exceptions control mode page: TEST=1, MRIE=6 */
if (dsense) {
arr[0] = 0x72;
arr[1] = 0x0; /* NO_SENSE in sense_key */
arr[2] = THRESHOLD_EXCEEDED;
arr[3] = 0xff; /* Failure prediction(false) */
len = 8;
} else {
arr[0] = 0x70;
arr[2] = 0x0; /* NO_SENSE in sense_key */
arr[7] = 0xa; /* 18 byte sense buffer */
arr[12] = THRESHOLD_EXCEEDED;
arr[13] = 0xff; /* Failure prediction(false) */
}
} else { /* nothing to report */
if (dsense) {
len = 8;
memset(arr, 0, len);
arr[0] = 0x72;
} else {
memset(arr, 0, len);
arr[0] = 0x70;
arr[7] = 0xa;
}
}
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
return fill_from_dev_buffer(scp, arr, min_t(u32, len, alloc_len));
}
static int resp_start_stop(struct scsi_cmnd *scp, struct sdebug_dev_info *devip)
{
unsigned char *cmd = scp->cmnd;
int power_cond, want_stop, stopped_state;
bool changing;
power_cond = (cmd[4] & 0xf0) >> 4;
if (power_cond) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 4, 7);
return check_condition_result;
}
want_stop = !(cmd[4] & 1);
stopped_state = atomic_read(&devip->stopped);
if (stopped_state == 2) {
ktime_t now_ts = ktime_get_boottime();
if (ktime_to_ns(now_ts) > ktime_to_ns(devip->create_ts)) {
u64 diff_ns = ktime_to_ns(ktime_sub(now_ts, devip->create_ts));
if (diff_ns >= ((u64)sdeb_tur_ms_to_ready * 1000000)) {
/* tur_ms_to_ready timer extinguished */
atomic_set(&devip->stopped, 0);
stopped_state = 0;
}
}
if (stopped_state == 2) {
if (want_stop) {
stopped_state = 1; /* dummy up success */
} else { /* Disallow tur_ms_to_ready delay to be overridden */
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 4, 0 /* START bit */);
return check_condition_result;
}
}
}
changing = (stopped_state != want_stop);
if (changing)
atomic_xchg(&devip->stopped, want_stop);
if (!changing || (cmd[1] & 0x1)) /* state unchanged or IMMED bit set in cdb */
return SDEG_RES_IMMED_MASK;
else
return 0;
}
static sector_t get_sdebug_capacity(void)
{
static const unsigned int gibibyte = 1073741824;
if (sdebug_virtual_gb > 0)
return (sector_t)sdebug_virtual_gb *
(gibibyte / sdebug_sector_size);
else
return sdebug_store_sectors;
}
#define SDEBUG_READCAP_ARR_SZ 8
static int resp_readcap(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
unsigned char arr[SDEBUG_READCAP_ARR_SZ];
unsigned int capac;
/* following just in case virtual_gb changed */
sdebug_capacity = get_sdebug_capacity();
memset(arr, 0, SDEBUG_READCAP_ARR_SZ);
if (sdebug_capacity < 0xffffffff) {
capac = (unsigned int)sdebug_capacity - 1;
put_unaligned_be32(capac, arr + 0);
} else
put_unaligned_be32(0xffffffff, arr + 0);
put_unaligned_be16(sdebug_sector_size, arr + 6);
return fill_from_dev_buffer(scp, arr, SDEBUG_READCAP_ARR_SZ);
}
#define SDEBUG_READCAP16_ARR_SZ 32
static int resp_readcap16(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
unsigned char *cmd = scp->cmnd;
unsigned char arr[SDEBUG_READCAP16_ARR_SZ];
scsi: scsi_debug: Fix out-of-bound read in resp_readcap16() The following warning was observed running syzkaller: [ 3813.830724] sg_write: data in/out 65466/242 bytes for SCSI command 0x9e-- guessing data in; [ 3813.830724] program syz-executor not setting count and/or reply_len properly [ 3813.836956] ================================================================== [ 3813.839465] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x157/0x1e0 [ 3813.841773] Read of size 4096 at addr ffff8883cf80f540 by task syz-executor/1549 [ 3813.846612] Call Trace: [ 3813.846995] dump_stack+0x108/0x15f [ 3813.847524] print_address_description+0xa5/0x372 [ 3813.848243] kasan_report.cold+0x236/0x2a8 [ 3813.849439] check_memory_region+0x240/0x270 [ 3813.850094] memcpy+0x30/0x80 [ 3813.850553] sg_copy_buffer+0x157/0x1e0 [ 3813.853032] sg_copy_from_buffer+0x13/0x20 [ 3813.853660] fill_from_dev_buffer+0x135/0x370 [ 3813.854329] resp_readcap16+0x1ac/0x280 [ 3813.856917] schedule_resp+0x41f/0x1630 [ 3813.858203] scsi_debug_queuecommand+0xb32/0x17e0 [ 3813.862699] scsi_dispatch_cmd+0x330/0x950 [ 3813.863329] scsi_request_fn+0xd8e/0x1710 [ 3813.863946] __blk_run_queue+0x10b/0x230 [ 3813.864544] blk_execute_rq_nowait+0x1d8/0x400 [ 3813.865220] sg_common_write.isra.0+0xe61/0x2420 [ 3813.871637] sg_write+0x6c8/0xef0 [ 3813.878853] __vfs_write+0xe4/0x800 [ 3813.883487] vfs_write+0x17b/0x530 [ 3813.884008] ksys_write+0x103/0x270 [ 3813.886268] __x64_sys_write+0x77/0xc0 [ 3813.886841] do_syscall_64+0x106/0x360 [ 3813.887415] entry_SYSCALL_64_after_hwframe+0x44/0xa9 This issue can be reproduced with the following syzkaller log: r0 = openat(0xffffffffffffff9c, &(0x7f0000000040)='./file0\x00', 0x26e1, 0x0) r1 = syz_open_procfs(0xffffffffffffffff, &(0x7f0000000000)='fd/3\x00') open_by_handle_at(r1, &(0x7f00000003c0)=ANY=[@ANYRESHEX], 0x602000) r2 = syz_open_dev$sg(&(0x7f0000000000), 0x0, 0x40782) write$binfmt_aout(r2, &(0x7f0000000340)=ANY=[@ANYBLOB="00000000deff000000000000000000000000000000000000000000000000000047f007af9e107a41ec395f1bded7be24277a1501ff6196a83366f4e6362bc0ff2b247f68a972989b094b2da4fb3607fcf611a22dd04310d28c75039d"], 0x126) In resp_readcap16() we get "int alloc_len" value -1104926854, and then pass the huge arr_len to fill_from_dev_buffer(), but arr is only 32 bytes. This leads to OOB in sg_copy_buffer(). To solve this issue, define alloc_len as u32. Link: https://lore.kernel.org/r/20211013033913.2551004-2-yebin10@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Ye Bin <yebin10@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-10-13 03:39:12 +00:00
u32 alloc_len;
alloc_len = get_unaligned_be32(cmd + 10);
/* following just in case virtual_gb changed */
sdebug_capacity = get_sdebug_capacity();
memset(arr, 0, SDEBUG_READCAP16_ARR_SZ);
put_unaligned_be64((u64)(sdebug_capacity - 1), arr + 0);
put_unaligned_be32(sdebug_sector_size, arr + 8);
arr[13] = sdebug_physblk_exp & 0xf;
arr[14] = (sdebug_lowest_aligned >> 8) & 0x3f;
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
if (scsi_debug_lbp()) {
arr[14] |= 0x80; /* LBPME */
/* from sbc4r07, this LBPRZ field is 1 bit, but the LBPRZ in
* the LB Provisioning VPD page is 3 bits. Note that lbprz=2
* in the wider field maps to 0 in this field.
*/
if (sdebug_lbprz & 1) /* precisely what the draft requires */
arr[14] |= 0x40;
}
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
/*
* Since the scsi_debug READ CAPACITY implementation always reports the
* total disk capacity, set RC BASIS = 1 for host-managed ZBC devices.
*/
if (devip->zmodel == BLK_ZONED_HM)
arr[12] |= 1 << 4;
arr[15] = sdebug_lowest_aligned & 0xff;
if (have_dif_prot) {
arr[12] = (sdebug_dif - 1) << 1; /* P_TYPE */
arr[12] |= 1; /* PROT_EN */
}
return fill_from_dev_buffer(scp, arr,
scsi: scsi_debug: Fix out-of-bound read in resp_readcap16() The following warning was observed running syzkaller: [ 3813.830724] sg_write: data in/out 65466/242 bytes for SCSI command 0x9e-- guessing data in; [ 3813.830724] program syz-executor not setting count and/or reply_len properly [ 3813.836956] ================================================================== [ 3813.839465] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x157/0x1e0 [ 3813.841773] Read of size 4096 at addr ffff8883cf80f540 by task syz-executor/1549 [ 3813.846612] Call Trace: [ 3813.846995] dump_stack+0x108/0x15f [ 3813.847524] print_address_description+0xa5/0x372 [ 3813.848243] kasan_report.cold+0x236/0x2a8 [ 3813.849439] check_memory_region+0x240/0x270 [ 3813.850094] memcpy+0x30/0x80 [ 3813.850553] sg_copy_buffer+0x157/0x1e0 [ 3813.853032] sg_copy_from_buffer+0x13/0x20 [ 3813.853660] fill_from_dev_buffer+0x135/0x370 [ 3813.854329] resp_readcap16+0x1ac/0x280 [ 3813.856917] schedule_resp+0x41f/0x1630 [ 3813.858203] scsi_debug_queuecommand+0xb32/0x17e0 [ 3813.862699] scsi_dispatch_cmd+0x330/0x950 [ 3813.863329] scsi_request_fn+0xd8e/0x1710 [ 3813.863946] __blk_run_queue+0x10b/0x230 [ 3813.864544] blk_execute_rq_nowait+0x1d8/0x400 [ 3813.865220] sg_common_write.isra.0+0xe61/0x2420 [ 3813.871637] sg_write+0x6c8/0xef0 [ 3813.878853] __vfs_write+0xe4/0x800 [ 3813.883487] vfs_write+0x17b/0x530 [ 3813.884008] ksys_write+0x103/0x270 [ 3813.886268] __x64_sys_write+0x77/0xc0 [ 3813.886841] do_syscall_64+0x106/0x360 [ 3813.887415] entry_SYSCALL_64_after_hwframe+0x44/0xa9 This issue can be reproduced with the following syzkaller log: r0 = openat(0xffffffffffffff9c, &(0x7f0000000040)='./file0\x00', 0x26e1, 0x0) r1 = syz_open_procfs(0xffffffffffffffff, &(0x7f0000000000)='fd/3\x00') open_by_handle_at(r1, &(0x7f00000003c0)=ANY=[@ANYRESHEX], 0x602000) r2 = syz_open_dev$sg(&(0x7f0000000000), 0x0, 0x40782) write$binfmt_aout(r2, &(0x7f0000000340)=ANY=[@ANYBLOB="00000000deff000000000000000000000000000000000000000000000000000047f007af9e107a41ec395f1bded7be24277a1501ff6196a83366f4e6362bc0ff2b247f68a972989b094b2da4fb3607fcf611a22dd04310d28c75039d"], 0x126) In resp_readcap16() we get "int alloc_len" value -1104926854, and then pass the huge arr_len to fill_from_dev_buffer(), but arr is only 32 bytes. This leads to OOB in sg_copy_buffer(). To solve this issue, define alloc_len as u32. Link: https://lore.kernel.org/r/20211013033913.2551004-2-yebin10@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Ye Bin <yebin10@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-10-13 03:39:12 +00:00
min_t(u32, alloc_len, SDEBUG_READCAP16_ARR_SZ));
}
#define SDEBUG_MAX_TGTPGS_ARR_SZ 1412
static int resp_report_tgtpgs(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
unsigned char *cmd = scp->cmnd;
unsigned char *arr;
int host_no = devip->sdbg_host->shost->host_no;
int port_group_a, port_group_b, port_a, port_b;
scsi: scsi_debug: Fix out-of-bound read in resp_report_tgtpgs() The following issue was observed running syzkaller: BUG: KASAN: slab-out-of-bounds in memcpy include/linux/string.h:377 [inline] BUG: KASAN: slab-out-of-bounds in sg_copy_buffer+0x150/0x1c0 lib/scatterlist.c:831 Read of size 2132 at addr ffff8880aea95dc8 by task syz-executor.0/9815 CPU: 0 PID: 9815 Comm: syz-executor.0 Not tainted 4.19.202-00874-gfc0fe04215a9 #2 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 Call Trace: __dump_stack lib/dump_stack.c:77 [inline] dump_stack+0xe4/0x14a lib/dump_stack.c:118 print_address_description+0x73/0x280 mm/kasan/report.c:253 kasan_report_error mm/kasan/report.c:352 [inline] kasan_report+0x272/0x370 mm/kasan/report.c:410 memcpy+0x1f/0x50 mm/kasan/kasan.c:302 memcpy include/linux/string.h:377 [inline] sg_copy_buffer+0x150/0x1c0 lib/scatterlist.c:831 fill_from_dev_buffer+0x14f/0x340 drivers/scsi/scsi_debug.c:1021 resp_report_tgtpgs+0x5aa/0x770 drivers/scsi/scsi_debug.c:1772 schedule_resp+0x464/0x12f0 drivers/scsi/scsi_debug.c:4429 scsi_debug_queuecommand+0x467/0x1390 drivers/scsi/scsi_debug.c:5835 scsi_dispatch_cmd+0x3fc/0x9b0 drivers/scsi/scsi_lib.c:1896 scsi_request_fn+0x1042/0x1810 drivers/scsi/scsi_lib.c:2034 __blk_run_queue_uncond block/blk-core.c:464 [inline] __blk_run_queue+0x1a4/0x380 block/blk-core.c:484 blk_execute_rq_nowait+0x1c2/0x2d0 block/blk-exec.c:78 sg_common_write.isra.19+0xd74/0x1dc0 drivers/scsi/sg.c:847 sg_write.part.23+0x6e0/0xd00 drivers/scsi/sg.c:716 sg_write+0x64/0xa0 drivers/scsi/sg.c:622 __vfs_write+0xed/0x690 fs/read_write.c:485 kill_bdev:block_device:00000000e138492c vfs_write+0x184/0x4c0 fs/read_write.c:549 ksys_write+0x107/0x240 fs/read_write.c:599 do_syscall_64+0xc2/0x560 arch/x86/entry/common.c:293 entry_SYSCALL_64_after_hwframe+0x49/0xbe We get 'alen' from command its type is int. If userspace passes a large length we will get a negative 'alen'. Switch n, alen, and rlen to u32. Link: https://lore.kernel.org/r/20211013033913.2551004-3-yebin10@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Ye Bin <yebin10@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-10-13 03:39:13 +00:00
u32 alen, n, rlen;
int ret;
alen = get_unaligned_be32(cmd + 6);
arr = kzalloc(SDEBUG_MAX_TGTPGS_ARR_SZ, GFP_ATOMIC);
if (! arr)
return DID_REQUEUE << 16;
/*
* EVPD page 0x88 states we have two ports, one
* real and a fake port with no device connected.
* So we create two port groups with one port each
* and set the group with port B to unavailable.
*/
port_a = 0x1; /* relative port A */
port_b = 0x2; /* relative port B */
port_group_a = (((host_no + 1) & 0x7f) << 8) +
(devip->channel & 0x7f);
port_group_b = (((host_no + 1) & 0x7f) << 8) +
(devip->channel & 0x7f) + 0x80;
/*
* The asymmetric access state is cycled according to the host_id.
*/
n = 4;
if (sdebug_vpd_use_hostno == 0) {
arr[n++] = host_no % 3; /* Asymm access state */
arr[n++] = 0x0F; /* claim: all states are supported */
} else {
arr[n++] = 0x0; /* Active/Optimized path */
arr[n++] = 0x01; /* only support active/optimized paths */
}
put_unaligned_be16(port_group_a, arr + n);
n += 2;
arr[n++] = 0; /* Reserved */
arr[n++] = 0; /* Status code */
arr[n++] = 0; /* Vendor unique */
arr[n++] = 0x1; /* One port per group */
arr[n++] = 0; /* Reserved */
arr[n++] = 0; /* Reserved */
put_unaligned_be16(port_a, arr + n);
n += 2;
arr[n++] = 3; /* Port unavailable */
arr[n++] = 0x08; /* claim: only unavailalbe paths are supported */
put_unaligned_be16(port_group_b, arr + n);
n += 2;
arr[n++] = 0; /* Reserved */
arr[n++] = 0; /* Status code */
arr[n++] = 0; /* Vendor unique */
arr[n++] = 0x1; /* One port per group */
arr[n++] = 0; /* Reserved */
arr[n++] = 0; /* Reserved */
put_unaligned_be16(port_b, arr + n);
n += 2;
rlen = n - 4;
put_unaligned_be32(rlen, arr + 0);
/*
* Return the smallest value of either
* - The allocated length
* - The constructed command length
* - The maximum array size
*/
scsi: scsi_debug: Fix out-of-bound read in resp_report_tgtpgs() The following issue was observed running syzkaller: BUG: KASAN: slab-out-of-bounds in memcpy include/linux/string.h:377 [inline] BUG: KASAN: slab-out-of-bounds in sg_copy_buffer+0x150/0x1c0 lib/scatterlist.c:831 Read of size 2132 at addr ffff8880aea95dc8 by task syz-executor.0/9815 CPU: 0 PID: 9815 Comm: syz-executor.0 Not tainted 4.19.202-00874-gfc0fe04215a9 #2 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 Call Trace: __dump_stack lib/dump_stack.c:77 [inline] dump_stack+0xe4/0x14a lib/dump_stack.c:118 print_address_description+0x73/0x280 mm/kasan/report.c:253 kasan_report_error mm/kasan/report.c:352 [inline] kasan_report+0x272/0x370 mm/kasan/report.c:410 memcpy+0x1f/0x50 mm/kasan/kasan.c:302 memcpy include/linux/string.h:377 [inline] sg_copy_buffer+0x150/0x1c0 lib/scatterlist.c:831 fill_from_dev_buffer+0x14f/0x340 drivers/scsi/scsi_debug.c:1021 resp_report_tgtpgs+0x5aa/0x770 drivers/scsi/scsi_debug.c:1772 schedule_resp+0x464/0x12f0 drivers/scsi/scsi_debug.c:4429 scsi_debug_queuecommand+0x467/0x1390 drivers/scsi/scsi_debug.c:5835 scsi_dispatch_cmd+0x3fc/0x9b0 drivers/scsi/scsi_lib.c:1896 scsi_request_fn+0x1042/0x1810 drivers/scsi/scsi_lib.c:2034 __blk_run_queue_uncond block/blk-core.c:464 [inline] __blk_run_queue+0x1a4/0x380 block/blk-core.c:484 blk_execute_rq_nowait+0x1c2/0x2d0 block/blk-exec.c:78 sg_common_write.isra.19+0xd74/0x1dc0 drivers/scsi/sg.c:847 sg_write.part.23+0x6e0/0xd00 drivers/scsi/sg.c:716 sg_write+0x64/0xa0 drivers/scsi/sg.c:622 __vfs_write+0xed/0x690 fs/read_write.c:485 kill_bdev:block_device:00000000e138492c vfs_write+0x184/0x4c0 fs/read_write.c:549 ksys_write+0x107/0x240 fs/read_write.c:599 do_syscall_64+0xc2/0x560 arch/x86/entry/common.c:293 entry_SYSCALL_64_after_hwframe+0x49/0xbe We get 'alen' from command its type is int. If userspace passes a large length we will get a negative 'alen'. Switch n, alen, and rlen to u32. Link: https://lore.kernel.org/r/20211013033913.2551004-3-yebin10@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Ye Bin <yebin10@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-10-13 03:39:13 +00:00
rlen = min(alen, n);
ret = fill_from_dev_buffer(scp, arr,
scsi: scsi_debug: Fix out-of-bound read in resp_report_tgtpgs() The following issue was observed running syzkaller: BUG: KASAN: slab-out-of-bounds in memcpy include/linux/string.h:377 [inline] BUG: KASAN: slab-out-of-bounds in sg_copy_buffer+0x150/0x1c0 lib/scatterlist.c:831 Read of size 2132 at addr ffff8880aea95dc8 by task syz-executor.0/9815 CPU: 0 PID: 9815 Comm: syz-executor.0 Not tainted 4.19.202-00874-gfc0fe04215a9 #2 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014 Call Trace: __dump_stack lib/dump_stack.c:77 [inline] dump_stack+0xe4/0x14a lib/dump_stack.c:118 print_address_description+0x73/0x280 mm/kasan/report.c:253 kasan_report_error mm/kasan/report.c:352 [inline] kasan_report+0x272/0x370 mm/kasan/report.c:410 memcpy+0x1f/0x50 mm/kasan/kasan.c:302 memcpy include/linux/string.h:377 [inline] sg_copy_buffer+0x150/0x1c0 lib/scatterlist.c:831 fill_from_dev_buffer+0x14f/0x340 drivers/scsi/scsi_debug.c:1021 resp_report_tgtpgs+0x5aa/0x770 drivers/scsi/scsi_debug.c:1772 schedule_resp+0x464/0x12f0 drivers/scsi/scsi_debug.c:4429 scsi_debug_queuecommand+0x467/0x1390 drivers/scsi/scsi_debug.c:5835 scsi_dispatch_cmd+0x3fc/0x9b0 drivers/scsi/scsi_lib.c:1896 scsi_request_fn+0x1042/0x1810 drivers/scsi/scsi_lib.c:2034 __blk_run_queue_uncond block/blk-core.c:464 [inline] __blk_run_queue+0x1a4/0x380 block/blk-core.c:484 blk_execute_rq_nowait+0x1c2/0x2d0 block/blk-exec.c:78 sg_common_write.isra.19+0xd74/0x1dc0 drivers/scsi/sg.c:847 sg_write.part.23+0x6e0/0xd00 drivers/scsi/sg.c:716 sg_write+0x64/0xa0 drivers/scsi/sg.c:622 __vfs_write+0xed/0x690 fs/read_write.c:485 kill_bdev:block_device:00000000e138492c vfs_write+0x184/0x4c0 fs/read_write.c:549 ksys_write+0x107/0x240 fs/read_write.c:599 do_syscall_64+0xc2/0x560 arch/x86/entry/common.c:293 entry_SYSCALL_64_after_hwframe+0x49/0xbe We get 'alen' from command its type is int. If userspace passes a large length we will get a negative 'alen'. Switch n, alen, and rlen to u32. Link: https://lore.kernel.org/r/20211013033913.2551004-3-yebin10@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Ye Bin <yebin10@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-10-13 03:39:13 +00:00
min_t(u32, rlen, SDEBUG_MAX_TGTPGS_ARR_SZ));
kfree(arr);
return ret;
}
static int resp_rsup_opcodes(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
bool rctd;
u8 reporting_opts, req_opcode, sdeb_i, supp;
u16 req_sa, u;
u32 alloc_len, a_len;
int k, offset, len, errsts, count, bump, na;
const struct opcode_info_t *oip;
const struct opcode_info_t *r_oip;
u8 *arr;
u8 *cmd = scp->cmnd;
rctd = !!(cmd[2] & 0x80);
reporting_opts = cmd[2] & 0x7;
req_opcode = cmd[3];
req_sa = get_unaligned_be16(cmd + 4);
alloc_len = get_unaligned_be32(cmd + 6);
if (alloc_len < 4 || alloc_len > 0xffff) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 6, -1);
return check_condition_result;
}
if (alloc_len > 8192)
a_len = 8192;
else
a_len = alloc_len;
scsi_debug: use atomic allocation in resp_rsup_opcodes resp_rsup_opcodes() may get called from atomic context and would need to use GFP_ATOMIC for allocations: [ 1237.913419] BUG: sleeping function called from invalid context at mm/slub.c:1262 [ 1237.914865] in_atomic(): 1, irqs_disabled(): 0, pid: 7556, name: trinity-c311 [ 1237.916142] 3 locks held by trinity-c311/7556: [ 1237.916981] #0: (sb_writers#5){.+.+.+}, at: do_readv_writev (include/linux/fs.h:2346 fs/read_write.c:844) [ 1237.919713] #1: (&of->mutex){+.+.+.}, at: kernfs_fop_write (fs/kernfs/file.c:297) [ 1237.922626] Mutex: counter: -1 owner: trinity-c311 [ 1237.924044] #2: (s_active#51){.+.+.+}, at: kernfs_fop_write (fs/kernfs/file.c:297) [ 1237.925960] Preemption disabled blk_execute_rq_nowait (block/blk-exec.c:95) [ 1237.927416] [ 1237.927680] CPU: 24 PID: 7556 Comm: trinity-c311 Not tainted 3.19.0-rc4-next-20150116-sasha-00054-g4ad498c-dirty #1744 [ 1237.929603] ffff8804fc9d8000 ffff8804d9bc3548 ffffffff9d439fb2 0000000000000000 [ 1237.931097] 0000000000000000 ffff8804d9bc3588 ffffffff9a18389a ffff8804d9bc3598 [ 1237.932466] ffffffff9a1b1715 ffffffffa15935d8 ffffffff9e6f8cb1 00000000000004ee [ 1237.933984] Call Trace: [ 1237.934434] dump_stack (lib/dump_stack.c:52) [ 1237.935323] ___might_sleep (kernel/sched/core.c:7339) [ 1237.936259] ? mark_held_locks (kernel/locking/lockdep.c:2549) [ 1237.937293] __might_sleep (kernel/sched/core.c:7305) [ 1237.938272] __kmalloc (mm/slub.c:1262 mm/slub.c:2419 mm/slub.c:2491 mm/slub.c:3291) [ 1237.939137] ? resp_rsup_opcodes (include/linux/slab.h:435 drivers/scsi/scsi_debug.c:1689) [ 1237.940173] resp_rsup_opcodes (include/linux/slab.h:435 drivers/scsi/scsi_debug.c:1689) [ 1237.941211] ? add_host_store (drivers/scsi/scsi_debug.c:1584) [ 1237.942261] scsi_debug_queuecommand (drivers/scsi/scsi_debug.c:5276) [ 1237.943404] ? blk_rq_map_sg (block/blk-merge.c:254) [ 1237.944398] ? scsi_init_sgtable (drivers/scsi/scsi_lib.c:1095) [ 1237.945402] sdebug_queuecommand_lock_or_not (drivers/scsi/scsi_debug.c:5300) [ 1237.946735] scsi_dispatch_cmd (drivers/scsi/scsi_lib.c:1706) [ 1237.947720] scsi_queue_rq (drivers/scsi/scsi_lib.c:1996) [ 1237.948687] __blk_mq_run_hw_queue (block/blk-mq.c:816) [ 1237.949796] blk_mq_run_hw_queue (block/blk-mq.c:896) [ 1237.950903] ? _raw_spin_unlock (./arch/x86/include/asm/preempt.h:95 include/linux/spinlock_api_smp.h:154 kernel/locking/spinlock.c:183) [ 1237.951862] blk_mq_insert_request (block/blk-mq.c:1037) [ 1237.952876] blk_execute_rq_nowait (block/blk-exec.c:95) [ 1237.953981] ? lockdep_init_map (kernel/locking/lockdep.c:3034) [ 1237.954967] blk_execute_rq (block/blk-exec.c:131) [ 1237.955929] ? blk_rq_bio_prep (block/blk-core.c:2835) [ 1237.956913] scsi_execute (drivers/scsi/scsi_lib.c:252) [ 1237.957821] scsi_execute_req_flags (drivers/scsi/scsi_lib.c:281) [ 1237.958968] scsi_report_opcode (drivers/scsi/scsi.c:956) [ 1237.960009] sd_revalidate_disk (drivers/scsi/sd.c:2707 drivers/scsi/sd.c:2792) [ 1237.961139] revalidate_disk (fs/block_dev.c:1081) [ 1237.962223] sd_rescan (drivers/scsi/sd.c:1532) [ 1237.963142] scsi_rescan_device (drivers/scsi/scsi_scan.c:1579) [ 1237.964165] store_rescan_field (drivers/scsi/scsi_sysfs.c:672) [ 1237.965254] dev_attr_store (drivers/base/core.c:138) [ 1237.966319] sysfs_kf_write (fs/sysfs/file.c:131) [ 1237.967289] kernfs_fop_write (fs/kernfs/file.c:311) [ 1237.968274] do_readv_writev (fs/read_write.c:722 fs/read_write.c:854) [ 1237.969295] ? __acct_update_integrals (kernel/tsacct.c:145) [ 1237.970452] ? kernfs_fop_open (fs/kernfs/file.c:271) [ 1237.971505] ? _raw_spin_unlock (./arch/x86/include/asm/preempt.h:95 include/linux/spinlock_api_smp.h:154 kernel/locking/spinlock.c:183) [ 1237.972512] ? context_tracking_user_exit (include/linux/vtime.h:89 include/linux/jump_label.h:114 include/trace/events/context_tracking.h:47 kernel/context_tracking.c:140) [ 1237.973668] ? trace_hardirqs_on_caller (kernel/locking/lockdep.c:2578 kernel/locking/lockdep.c:2625) [ 1237.974882] ? trace_hardirqs_on (kernel/locking/lockdep.c:2633) [ 1237.975850] vfs_writev (fs/read_write.c:893) [ 1237.976691] SyS_writev (fs/read_write.c:926 fs/read_write.c:917) [ 1237.977538] system_call_fastpath (arch/x86/kernel/entry_64.S:423) Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2015-01-17 22:47:37 +00:00
arr = kzalloc((a_len < 256) ? 320 : a_len + 64, GFP_ATOMIC);
if (NULL == arr) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INSUFF_RES_ASC,
INSUFF_RES_ASCQ);
return check_condition_result;
}
switch (reporting_opts) {
case 0: /* all commands */
/* count number of commands */
for (count = 0, oip = opcode_info_arr;
oip->num_attached != 0xff; ++oip) {
if (F_INV_OP & oip->flags)
continue;
count += (oip->num_attached + 1);
}
bump = rctd ? 20 : 8;
put_unaligned_be32(count * bump, arr);
for (offset = 4, oip = opcode_info_arr;
oip->num_attached != 0xff && offset < a_len; ++oip) {
if (F_INV_OP & oip->flags)
continue;
na = oip->num_attached;
arr[offset] = oip->opcode;
put_unaligned_be16(oip->sa, arr + offset + 2);
if (rctd)
arr[offset + 5] |= 0x2;
if (FF_SA & oip->flags)
arr[offset + 5] |= 0x1;
put_unaligned_be16(oip->len_mask[0], arr + offset + 6);
if (rctd)
put_unaligned_be16(0xa, arr + offset + 8);
r_oip = oip;
for (k = 0, oip = oip->arrp; k < na; ++k, ++oip) {
if (F_INV_OP & oip->flags)
continue;
offset += bump;
arr[offset] = oip->opcode;
put_unaligned_be16(oip->sa, arr + offset + 2);
if (rctd)
arr[offset + 5] |= 0x2;
if (FF_SA & oip->flags)
arr[offset + 5] |= 0x1;
put_unaligned_be16(oip->len_mask[0],
arr + offset + 6);
if (rctd)
put_unaligned_be16(0xa,
arr + offset + 8);
}
oip = r_oip;
offset += bump;
}
break;
case 1: /* one command: opcode only */
case 2: /* one command: opcode plus service action */
case 3: /* one command: if sa==0 then opcode only else opcode+sa */
sdeb_i = opcode_ind_arr[req_opcode];
oip = &opcode_info_arr[sdeb_i];
if (F_INV_OP & oip->flags) {
supp = 1;
offset = 4;
} else {
if (1 == reporting_opts) {
if (FF_SA & oip->flags) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB,
2, 2);
kfree(arr);
return check_condition_result;
}
req_sa = 0;
} else if (2 == reporting_opts &&
0 == (FF_SA & oip->flags)) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 4, -1);
kfree(arr); /* point at requested sa */
return check_condition_result;
}
if (0 == (FF_SA & oip->flags) &&
req_opcode == oip->opcode)
supp = 3;
else if (0 == (FF_SA & oip->flags)) {
na = oip->num_attached;
for (k = 0, oip = oip->arrp; k < na;
++k, ++oip) {
if (req_opcode == oip->opcode)
break;
}
supp = (k >= na) ? 1 : 3;
} else if (req_sa != oip->sa) {
na = oip->num_attached;
for (k = 0, oip = oip->arrp; k < na;
++k, ++oip) {
if (req_sa == oip->sa)
break;
}
supp = (k >= na) ? 1 : 3;
} else
supp = 3;
if (3 == supp) {
u = oip->len_mask[0];
put_unaligned_be16(u, arr + 2);
arr[4] = oip->opcode;
for (k = 1; k < u; ++k)
arr[4 + k] = (k < 16) ?
oip->len_mask[k] : 0xff;
offset = 4 + u;
} else
offset = 4;
}
arr[1] = (rctd ? 0x80 : 0) | supp;
if (rctd) {
put_unaligned_be16(0xa, arr + offset);
offset += 12;
}
break;
default:
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 2, 2);
kfree(arr);
return check_condition_result;
}
offset = (offset < a_len) ? offset : a_len;
len = (offset < alloc_len) ? offset : alloc_len;
errsts = fill_from_dev_buffer(scp, arr, len);
kfree(arr);
return errsts;
}
static int resp_rsup_tmfs(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
bool repd;
u32 alloc_len, len;
u8 arr[16];
u8 *cmd = scp->cmnd;
memset(arr, 0, sizeof(arr));
repd = !!(cmd[2] & 0x80);
alloc_len = get_unaligned_be32(cmd + 6);
if (alloc_len < 4) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 6, -1);
return check_condition_result;
}
arr[0] = 0xc8; /* ATS | ATSS | LURS */
arr[1] = 0x1; /* ITNRS */
if (repd) {
arr[3] = 0xc;
len = 16;
} else
len = 4;
len = (len < alloc_len) ? len : alloc_len;
return fill_from_dev_buffer(scp, arr, len);
}
/* <<Following mode page info copied from ST318451LW>> */
static int resp_err_recov_pg(unsigned char *p, int pcontrol, int target)
{ /* Read-Write Error Recovery page for mode_sense */
unsigned char err_recov_pg[] = {0x1, 0xa, 0xc0, 11, 240, 0, 0, 0,
5, 0, 0xff, 0xff};
memcpy(p, err_recov_pg, sizeof(err_recov_pg));
if (1 == pcontrol)
memset(p + 2, 0, sizeof(err_recov_pg) - 2);
return sizeof(err_recov_pg);
}
static int resp_disconnect_pg(unsigned char *p, int pcontrol, int target)
{ /* Disconnect-Reconnect page for mode_sense */
unsigned char disconnect_pg[] = {0x2, 0xe, 128, 128, 0, 10, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0};
memcpy(p, disconnect_pg, sizeof(disconnect_pg));
if (1 == pcontrol)
memset(p + 2, 0, sizeof(disconnect_pg) - 2);
return sizeof(disconnect_pg);
}
static int resp_format_pg(unsigned char *p, int pcontrol, int target)
{ /* Format device page for mode_sense */
unsigned char format_pg[] = {0x3, 0x16, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0x40, 0, 0, 0};
memcpy(p, format_pg, sizeof(format_pg));
put_unaligned_be16(sdebug_sectors_per, p + 10);
put_unaligned_be16(sdebug_sector_size, p + 12);
if (sdebug_removable)
p[20] |= 0x20; /* should agree with INQUIRY */
if (1 == pcontrol)
memset(p + 2, 0, sizeof(format_pg) - 2);
return sizeof(format_pg);
}
static unsigned char caching_pg[] = {0x8, 18, 0x14, 0, 0xff, 0xff, 0, 0,
0xff, 0xff, 0xff, 0xff, 0x80, 0x14, 0, 0,
0, 0, 0, 0};
static int resp_caching_pg(unsigned char *p, int pcontrol, int target)
{ /* Caching page for mode_sense */
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
unsigned char ch_caching_pg[] = {/* 0x8, 18, */ 0x4, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
unsigned char d_caching_pg[] = {0x8, 18, 0x14, 0, 0xff, 0xff, 0, 0,
0xff, 0xff, 0xff, 0xff, 0x80, 0x14, 0, 0, 0, 0, 0, 0};
if (SDEBUG_OPT_N_WCE & sdebug_opts)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
caching_pg[2] &= ~0x4; /* set WCE=0 (default WCE=1) */
memcpy(p, caching_pg, sizeof(caching_pg));
if (1 == pcontrol)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
memcpy(p + 2, ch_caching_pg, sizeof(ch_caching_pg));
else if (2 == pcontrol)
memcpy(p, d_caching_pg, sizeof(d_caching_pg));
return sizeof(caching_pg);
}
static unsigned char ctrl_m_pg[] = {0xa, 10, 2, 0, 0, 0, 0, 0,
0, 0, 0x2, 0x4b};
static int resp_ctrl_m_pg(unsigned char *p, int pcontrol, int target)
{ /* Control mode page for mode_sense */
unsigned char ch_ctrl_m_pg[] = {/* 0xa, 10, */ 0x6, 0, 0, 0, 0, 0,
0, 0, 0, 0};
unsigned char d_ctrl_m_pg[] = {0xa, 10, 2, 0, 0, 0, 0, 0,
0, 0, 0x2, 0x4b};
if (sdebug_dsense)
ctrl_m_pg[2] |= 0x4;
else
ctrl_m_pg[2] &= ~0x4;
if (sdebug_ato)
ctrl_m_pg[5] |= 0x80; /* ATO=1 */
memcpy(p, ctrl_m_pg, sizeof(ctrl_m_pg));
if (1 == pcontrol)
memcpy(p + 2, ch_ctrl_m_pg, sizeof(ch_ctrl_m_pg));
else if (2 == pcontrol)
memcpy(p, d_ctrl_m_pg, sizeof(d_ctrl_m_pg));
return sizeof(ctrl_m_pg);
}
static int resp_iec_m_pg(unsigned char *p, int pcontrol, int target)
{ /* Informational Exceptions control mode page for mode_sense */
unsigned char ch_iec_m_pg[] = {/* 0x1c, 0xa, */ 0x4, 0xf, 0, 0, 0, 0,
0, 0, 0x0, 0x0};
unsigned char d_iec_m_pg[] = {0x1c, 0xa, 0x08, 0, 0, 0, 0, 0,
0, 0, 0x0, 0x0};
memcpy(p, iec_m_pg, sizeof(iec_m_pg));
if (1 == pcontrol)
memcpy(p + 2, ch_iec_m_pg, sizeof(ch_iec_m_pg));
else if (2 == pcontrol)
memcpy(p, d_iec_m_pg, sizeof(d_iec_m_pg));
return sizeof(iec_m_pg);
}
static int resp_sas_sf_m_pg(unsigned char *p, int pcontrol, int target)
{ /* SAS SSP mode page - short format for mode_sense */
unsigned char sas_sf_m_pg[] = {0x19, 0x6,
0x6, 0x0, 0x7, 0xd0, 0x0, 0x0};
memcpy(p, sas_sf_m_pg, sizeof(sas_sf_m_pg));
if (1 == pcontrol)
memset(p + 2, 0, sizeof(sas_sf_m_pg) - 2);
return sizeof(sas_sf_m_pg);
}
static int resp_sas_pcd_m_spg(unsigned char *p, int pcontrol, int target,
int target_dev_id)
{ /* SAS phy control and discover mode page for mode_sense */
unsigned char sas_pcd_m_pg[] = {0x59, 0x1, 0, 0x64, 0, 0x6, 0, 2,
0, 0, 0, 0, 0x10, 0x9, 0x8, 0x0,
0, 0, 0, 0, 0, 0, 0, 0, /* insert SAS addr */
0, 0, 0, 0, 0, 0, 0, 0, /* insert SAS addr */
0x2, 0, 0, 0, 0, 0, 0, 0,
0x88, 0x99, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
0, 1, 0, 0, 0x10, 0x9, 0x8, 0x0,
0, 0, 0, 0, 0, 0, 0, 0, /* insert SAS addr */
0, 0, 0, 0, 0, 0, 0, 0, /* insert SAS addr */
0x3, 0, 0, 0, 0, 0, 0, 0,
0x88, 0x99, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
};
int port_a, port_b;
put_unaligned_be64(naa3_comp_a, sas_pcd_m_pg + 16);
put_unaligned_be64(naa3_comp_c + 1, sas_pcd_m_pg + 24);
put_unaligned_be64(naa3_comp_a, sas_pcd_m_pg + 64);
put_unaligned_be64(naa3_comp_c + 1, sas_pcd_m_pg + 72);
port_a = target_dev_id + 1;
port_b = port_a + 1;
memcpy(p, sas_pcd_m_pg, sizeof(sas_pcd_m_pg));
put_unaligned_be32(port_a, p + 20);
put_unaligned_be32(port_b, p + 48 + 20);
if (1 == pcontrol)
memset(p + 4, 0, sizeof(sas_pcd_m_pg) - 4);
return sizeof(sas_pcd_m_pg);
}
static int resp_sas_sha_m_spg(unsigned char *p, int pcontrol)
{ /* SAS SSP shared protocol specific port mode subpage */
unsigned char sas_sha_m_pg[] = {0x59, 0x2, 0, 0xc, 0, 0x6, 0x10, 0,
0, 0, 0, 0, 0, 0, 0, 0,
};
memcpy(p, sas_sha_m_pg, sizeof(sas_sha_m_pg));
if (1 == pcontrol)
memset(p + 4, 0, sizeof(sas_sha_m_pg) - 4);
return sizeof(sas_sha_m_pg);
}
#define SDEBUG_MAX_MSENSE_SZ 256
static int resp_mode_sense(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
int pcontrol, pcode, subpcode, bd_len;
unsigned char dev_spec;
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
u32 alloc_len, offset, len;
int target_dev_id;
int target = scp->device->id;
unsigned char *ap;
unsigned char arr[SDEBUG_MAX_MSENSE_SZ];
unsigned char *cmd = scp->cmnd;
bool dbd, llbaa, msense_6, is_disk, is_zbc, bad_pcode;
dbd = !!(cmd[1] & 0x8); /* disable block descriptors */
pcontrol = (cmd[2] & 0xc0) >> 6;
pcode = cmd[2] & 0x3f;
subpcode = cmd[3];
msense_6 = (MODE_SENSE == cmd[0]);
llbaa = msense_6 ? false : !!(cmd[1] & 0x10);
is_disk = (sdebug_ptype == TYPE_DISK);
is_zbc = (devip->zmodel != BLK_ZONED_NONE);
if ((is_disk || is_zbc) && !dbd)
bd_len = llbaa ? 16 : 8;
else
bd_len = 0;
alloc_len = msense_6 ? cmd[4] : get_unaligned_be16(cmd + 7);
memset(arr, 0, SDEBUG_MAX_MSENSE_SZ);
if (0x3 == pcontrol) { /* Saving values not supported */
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
mk_sense_buffer(scp, ILLEGAL_REQUEST, SAVING_PARAMS_UNSUP, 0);
return check_condition_result;
}
target_dev_id = ((devip->sdbg_host->shost->host_no + 1) * 2000) +
(devip->target * 1000) - 3;
/* for disks+zbc set DPOFUA bit and clear write protect (WP) bit */
if (is_disk || is_zbc) {
dev_spec = 0x10; /* =0x90 if WP=1 implies read-only */
if (sdebug_wp)
dev_spec |= 0x80;
} else
dev_spec = 0x0;
if (msense_6) {
arr[2] = dev_spec;
arr[3] = bd_len;
offset = 4;
} else {
arr[3] = dev_spec;
if (16 == bd_len)
arr[4] = 0x1; /* set LONGLBA bit */
arr[7] = bd_len; /* assume 255 or less */
offset = 8;
}
ap = arr + offset;
if ((bd_len > 0) && (!sdebug_capacity))
sdebug_capacity = get_sdebug_capacity();
if (8 == bd_len) {
if (sdebug_capacity > 0xfffffffe)
put_unaligned_be32(0xffffffff, ap + 0);
else
put_unaligned_be32(sdebug_capacity, ap + 0);
put_unaligned_be16(sdebug_sector_size, ap + 6);
offset += bd_len;
ap = arr + offset;
} else if (16 == bd_len) {
put_unaligned_be64((u64)sdebug_capacity, ap + 0);
put_unaligned_be32(sdebug_sector_size, ap + 12);
offset += bd_len;
ap = arr + offset;
}
if ((subpcode > 0x0) && (subpcode < 0xff) && (0x19 != pcode)) {
/* TODO: Control Extension page */
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 3, -1);
return check_condition_result;
}
bad_pcode = false;
switch (pcode) {
case 0x1: /* Read-Write error recovery page, direct access */
len = resp_err_recov_pg(ap, pcontrol, target);
offset += len;
break;
case 0x2: /* Disconnect-Reconnect page, all devices */
len = resp_disconnect_pg(ap, pcontrol, target);
offset += len;
break;
case 0x3: /* Format device page, direct access */
if (is_disk) {
len = resp_format_pg(ap, pcontrol, target);
offset += len;
} else
bad_pcode = true;
break;
case 0x8: /* Caching page, direct access */
if (is_disk || is_zbc) {
len = resp_caching_pg(ap, pcontrol, target);
offset += len;
} else
bad_pcode = true;
break;
case 0xa: /* Control Mode page, all devices */
len = resp_ctrl_m_pg(ap, pcontrol, target);
offset += len;
break;
case 0x19: /* if spc==1 then sas phy, control+discover */
if ((subpcode > 0x2) && (subpcode < 0xff)) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 3, -1);
return check_condition_result;
}
len = 0;
if ((0x0 == subpcode) || (0xff == subpcode))
len += resp_sas_sf_m_pg(ap + len, pcontrol, target);
if ((0x1 == subpcode) || (0xff == subpcode))
len += resp_sas_pcd_m_spg(ap + len, pcontrol, target,
target_dev_id);
if ((0x2 == subpcode) || (0xff == subpcode))
len += resp_sas_sha_m_spg(ap + len, pcontrol);
offset += len;
break;
case 0x1c: /* Informational Exceptions Mode page, all devices */
len = resp_iec_m_pg(ap, pcontrol, target);
offset += len;
break;
case 0x3f: /* Read all Mode pages */
if ((0 == subpcode) || (0xff == subpcode)) {
len = resp_err_recov_pg(ap, pcontrol, target);
len += resp_disconnect_pg(ap + len, pcontrol, target);
if (is_disk) {
len += resp_format_pg(ap + len, pcontrol,
target);
len += resp_caching_pg(ap + len, pcontrol,
target);
} else if (is_zbc) {
len += resp_caching_pg(ap + len, pcontrol,
target);
}
len += resp_ctrl_m_pg(ap + len, pcontrol, target);
len += resp_sas_sf_m_pg(ap + len, pcontrol, target);
if (0xff == subpcode) {
len += resp_sas_pcd_m_spg(ap + len, pcontrol,
target, target_dev_id);
len += resp_sas_sha_m_spg(ap + len, pcontrol);
}
len += resp_iec_m_pg(ap + len, pcontrol, target);
offset += len;
} else {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 3, -1);
return check_condition_result;
}
break;
default:
bad_pcode = true;
break;
}
if (bad_pcode) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 2, 5);
return check_condition_result;
}
if (msense_6)
arr[0] = offset - 1;
else
put_unaligned_be16((offset - 2), arr + 0);
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
return fill_from_dev_buffer(scp, arr, min_t(u32, alloc_len, offset));
}
#define SDEBUG_MAX_MSELECT_SZ 512
static int resp_mode_select(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
int pf, sp, ps, md_len, bd_len, off, spf, pg_len;
int param_len, res, mpage;
unsigned char arr[SDEBUG_MAX_MSELECT_SZ];
unsigned char *cmd = scp->cmnd;
int mselect6 = (MODE_SELECT == cmd[0]);
memset(arr, 0, sizeof(arr));
pf = cmd[1] & 0x10;
sp = cmd[1] & 0x1;
param_len = mselect6 ? cmd[4] : get_unaligned_be16(cmd + 7);
if ((0 == pf) || sp || (param_len > SDEBUG_MAX_MSELECT_SZ)) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, mselect6 ? 4 : 7, -1);
return check_condition_result;
}
res = fetch_to_dev_buffer(scp, arr, param_len);
if (-1 == res)
return DID_ERROR << 16;
else if (sdebug_verbose && (res < param_len))
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
sdev_printk(KERN_INFO, scp->device,
"%s: cdb indicated=%d, IO sent=%d bytes\n",
__func__, param_len, res);
md_len = mselect6 ? (arr[0] + 1) : (get_unaligned_be16(arr + 0) + 2);
bd_len = mselect6 ? arr[3] : get_unaligned_be16(arr + 6);
scsi: scsi_debug: Sanity check block descriptor length in resp_mode_select() In resp_mode_select() sanity check the block descriptor len to avoid UAF. BUG: KASAN: use-after-free in resp_mode_select+0xa4c/0xb40 drivers/scsi/scsi_debug.c:2509 Read of size 1 at addr ffff888026670f50 by task scsicmd/15032 CPU: 1 PID: 15032 Comm: scsicmd Not tainted 5.15.0-01d0625 #15 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Call Trace: <TASK> dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:107 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:257 kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:443 __asan_report_load1_noabort+0x14/0x20 mm/kasan/report_generic.c:306 resp_mode_select+0xa4c/0xb40 drivers/scsi/scsi_debug.c:2509 schedule_resp+0x4af/0x1a10 drivers/scsi/scsi_debug.c:5483 scsi_debug_queuecommand+0x8c9/0x1e70 drivers/scsi/scsi_debug.c:7537 scsi_queue_rq+0x16b4/0x2d10 drivers/scsi/scsi_lib.c:1521 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1640 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1762 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1839 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:63 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:837 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:775 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:941 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1166 __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:52 do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:50 entry_SYSCALL_64_after_hwframe+0x44/0xae arch/x86/entry/entry_64.S:113 Link: https://lore.kernel.org/r/1637262208-28850-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-18 19:03:28 +00:00
off = bd_len + (mselect6 ? 4 : 8);
if (md_len > 2 || off >= res) {
mk_sense_invalid_fld(scp, SDEB_IN_DATA, 0, -1);
return check_condition_result;
}
mpage = arr[off] & 0x3f;
ps = !!(arr[off] & 0x80);
if (ps) {
mk_sense_invalid_fld(scp, SDEB_IN_DATA, off, 7);
return check_condition_result;
}
spf = !!(arr[off] & 0x40);
pg_len = spf ? (get_unaligned_be16(arr + off + 2) + 4) :
(arr[off + 1] + 2);
if ((pg_len + off) > param_len) {
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
mk_sense_buffer(scp, ILLEGAL_REQUEST,
PARAMETER_LIST_LENGTH_ERR, 0);
return check_condition_result;
}
switch (mpage) {
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
case 0x8: /* Caching Mode page */
if (caching_pg[1] == arr[off + 1]) {
memcpy(caching_pg + 2, arr + off + 2,
sizeof(caching_pg) - 2);
goto set_mode_changed_ua;
}
break;
case 0xa: /* Control Mode page */
if (ctrl_m_pg[1] == arr[off + 1]) {
memcpy(ctrl_m_pg + 2, arr + off + 2,
sizeof(ctrl_m_pg) - 2);
if (ctrl_m_pg[4] & 0x8)
sdebug_wp = true;
else
sdebug_wp = false;
sdebug_dsense = !!(ctrl_m_pg[2] & 0x4);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
goto set_mode_changed_ua;
}
break;
case 0x1c: /* Informational Exceptions Mode page */
if (iec_m_pg[1] == arr[off + 1]) {
memcpy(iec_m_pg + 2, arr + off + 2,
sizeof(iec_m_pg) - 2);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
goto set_mode_changed_ua;
}
break;
default:
break;
}
mk_sense_invalid_fld(scp, SDEB_IN_DATA, off, 5);
return check_condition_result;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
set_mode_changed_ua:
set_bit(SDEBUG_UA_MODE_CHANGED, devip->uas_bm);
return 0;
}
static int resp_temp_l_pg(unsigned char *arr)
{
unsigned char temp_l_pg[] = {0x0, 0x0, 0x3, 0x2, 0x0, 38,
0x0, 0x1, 0x3, 0x2, 0x0, 65,
};
memcpy(arr, temp_l_pg, sizeof(temp_l_pg));
return sizeof(temp_l_pg);
}
static int resp_ie_l_pg(unsigned char *arr)
{
unsigned char ie_l_pg[] = {0x0, 0x0, 0x3, 0x3, 0x0, 0x0, 38,
};
memcpy(arr, ie_l_pg, sizeof(ie_l_pg));
if (iec_m_pg[2] & 0x4) { /* TEST bit set */
arr[4] = THRESHOLD_EXCEEDED;
arr[5] = 0xff;
}
return sizeof(ie_l_pg);
}
static int resp_env_rep_l_spg(unsigned char *arr)
{
unsigned char env_rep_l_spg[] = {0x0, 0x0, 0x23, 0x8,
0x0, 40, 72, 0xff, 45, 18, 0, 0,
0x1, 0x0, 0x23, 0x8,
0x0, 55, 72, 35, 55, 45, 0, 0,
};
memcpy(arr, env_rep_l_spg, sizeof(env_rep_l_spg));
return sizeof(env_rep_l_spg);
}
#define SDEBUG_MAX_LSENSE_SZ 512
static int resp_log_sense(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
int ppc, sp, pcode, subpcode;
u32 alloc_len, len, n;
unsigned char arr[SDEBUG_MAX_LSENSE_SZ];
unsigned char *cmd = scp->cmnd;
memset(arr, 0, sizeof(arr));
ppc = cmd[1] & 0x2;
sp = cmd[1] & 0x1;
if (ppc || sp) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 1, ppc ? 1 : 0);
return check_condition_result;
}
pcode = cmd[2] & 0x3f;
subpcode = cmd[3] & 0xff;
alloc_len = get_unaligned_be16(cmd + 7);
arr[0] = pcode;
if (0 == subpcode) {
switch (pcode) {
case 0x0: /* Supported log pages log page */
n = 4;
arr[n++] = 0x0; /* this page */
arr[n++] = 0xd; /* Temperature */
arr[n++] = 0x2f; /* Informational exceptions */
arr[3] = n - 4;
break;
case 0xd: /* Temperature log page */
arr[3] = resp_temp_l_pg(arr + 4);
break;
case 0x2f: /* Informational exceptions log page */
arr[3] = resp_ie_l_pg(arr + 4);
break;
default:
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 2, 5);
return check_condition_result;
}
} else if (0xff == subpcode) {
arr[0] |= 0x40;
arr[1] = subpcode;
switch (pcode) {
case 0x0: /* Supported log pages and subpages log page */
n = 4;
arr[n++] = 0x0;
arr[n++] = 0x0; /* 0,0 page */
arr[n++] = 0x0;
arr[n++] = 0xff; /* this page */
arr[n++] = 0xd;
arr[n++] = 0x0; /* Temperature */
arr[n++] = 0xd;
arr[n++] = 0x1; /* Environment reporting */
arr[n++] = 0xd;
arr[n++] = 0xff; /* all 0xd subpages */
arr[n++] = 0x2f;
arr[n++] = 0x0; /* Informational exceptions */
arr[n++] = 0x2f;
arr[n++] = 0xff; /* all 0x2f subpages */
arr[3] = n - 4;
break;
case 0xd: /* Temperature subpages */
n = 4;
arr[n++] = 0xd;
arr[n++] = 0x0; /* Temperature */
arr[n++] = 0xd;
arr[n++] = 0x1; /* Environment reporting */
arr[n++] = 0xd;
arr[n++] = 0xff; /* these subpages */
arr[3] = n - 4;
break;
case 0x2f: /* Informational exceptions subpages */
n = 4;
arr[n++] = 0x2f;
arr[n++] = 0x0; /* Informational exceptions */
arr[n++] = 0x2f;
arr[n++] = 0xff; /* these subpages */
arr[3] = n - 4;
break;
default:
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 2, 5);
return check_condition_result;
}
} else if (subpcode > 0) {
arr[0] |= 0x40;
arr[1] = subpcode;
if (pcode == 0xd && subpcode == 1)
arr[3] = resp_env_rep_l_spg(arr + 4);
else {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 2, 5);
return check_condition_result;
}
} else {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 3, -1);
return check_condition_result;
}
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
len = min_t(u32, get_unaligned_be16(arr + 2) + 4, alloc_len);
return fill_from_dev_buffer(scp, arr,
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
min_t(u32, len, SDEBUG_MAX_INQ_ARR_SZ));
}
static inline bool sdebug_dev_is_zoned(struct sdebug_dev_info *devip)
{
return devip->nr_zones != 0;
}
static struct sdeb_zone_state *zbc_zone(struct sdebug_dev_info *devip,
unsigned long long lba)
{
u32 zno = lba >> devip->zsize_shift;
struct sdeb_zone_state *zsp;
if (devip->zcap == devip->zsize || zno < devip->nr_conv_zones)
return &devip->zstate[zno];
/*
* If the zone capacity is less than the zone size, adjust for gap
* zones.
*/
zno = 2 * zno - devip->nr_conv_zones;
WARN_ONCE(zno >= devip->nr_zones, "%u > %u\n", zno, devip->nr_zones);
zsp = &devip->zstate[zno];
if (lba >= zsp->z_start + zsp->z_size)
zsp++;
WARN_ON_ONCE(lba >= zsp->z_start + zsp->z_size);
return zsp;
}
static inline bool zbc_zone_is_conv(struct sdeb_zone_state *zsp)
{
return zsp->z_type == ZBC_ZTYPE_CNV;
}
static inline bool zbc_zone_is_gap(struct sdeb_zone_state *zsp)
{
return zsp->z_type == ZBC_ZTYPE_GAP;
}
static inline bool zbc_zone_is_seq(struct sdeb_zone_state *zsp)
{
return !zbc_zone_is_conv(zsp) && !zbc_zone_is_gap(zsp);
}
static void zbc_close_zone(struct sdebug_dev_info *devip,
struct sdeb_zone_state *zsp)
{
enum sdebug_z_cond zc;
if (!zbc_zone_is_seq(zsp))
return;
zc = zsp->z_cond;
if (!(zc == ZC2_IMPLICIT_OPEN || zc == ZC3_EXPLICIT_OPEN))
return;
if (zc == ZC2_IMPLICIT_OPEN)
devip->nr_imp_open--;
else
devip->nr_exp_open--;
if (zsp->z_wp == zsp->z_start) {
zsp->z_cond = ZC1_EMPTY;
} else {
zsp->z_cond = ZC4_CLOSED;
devip->nr_closed++;
}
}
static void zbc_close_imp_open_zone(struct sdebug_dev_info *devip)
{
struct sdeb_zone_state *zsp = &devip->zstate[0];
unsigned int i;
for (i = 0; i < devip->nr_zones; i++, zsp++) {
if (zsp->z_cond == ZC2_IMPLICIT_OPEN) {
zbc_close_zone(devip, zsp);
return;
}
}
}
static void zbc_open_zone(struct sdebug_dev_info *devip,
struct sdeb_zone_state *zsp, bool explicit)
{
enum sdebug_z_cond zc;
if (!zbc_zone_is_seq(zsp))
return;
zc = zsp->z_cond;
if ((explicit && zc == ZC3_EXPLICIT_OPEN) ||
(!explicit && zc == ZC2_IMPLICIT_OPEN))
return;
/* Close an implicit open zone if necessary */
if (explicit && zsp->z_cond == ZC2_IMPLICIT_OPEN)
zbc_close_zone(devip, zsp);
else if (devip->max_open &&
devip->nr_imp_open + devip->nr_exp_open >= devip->max_open)
zbc_close_imp_open_zone(devip);
if (zsp->z_cond == ZC4_CLOSED)
devip->nr_closed--;
if (explicit) {
zsp->z_cond = ZC3_EXPLICIT_OPEN;
devip->nr_exp_open++;
} else {
zsp->z_cond = ZC2_IMPLICIT_OPEN;
devip->nr_imp_open++;
}
}
static inline void zbc_set_zone_full(struct sdebug_dev_info *devip,
struct sdeb_zone_state *zsp)
{
switch (zsp->z_cond) {
case ZC2_IMPLICIT_OPEN:
devip->nr_imp_open--;
break;
case ZC3_EXPLICIT_OPEN:
devip->nr_exp_open--;
break;
default:
WARN_ONCE(true, "Invalid zone %llu condition %x\n",
zsp->z_start, zsp->z_cond);
break;
}
zsp->z_cond = ZC5_FULL;
}
static void zbc_inc_wp(struct sdebug_dev_info *devip,
unsigned long long lba, unsigned int num)
{
struct sdeb_zone_state *zsp = zbc_zone(devip, lba);
unsigned long long n, end, zend = zsp->z_start + zsp->z_size;
if (!zbc_zone_is_seq(zsp))
return;
if (zsp->z_type == ZBC_ZTYPE_SWR) {
zsp->z_wp += num;
if (zsp->z_wp >= zend)
zbc_set_zone_full(devip, zsp);
return;
}
while (num) {
if (lba != zsp->z_wp)
zsp->z_non_seq_resource = true;
end = lba + num;
if (end >= zend) {
n = zend - lba;
zsp->z_wp = zend;
} else if (end > zsp->z_wp) {
n = num;
zsp->z_wp = end;
} else {
n = num;
}
if (zsp->z_wp >= zend)
zbc_set_zone_full(devip, zsp);
num -= n;
lba += n;
if (num) {
zsp++;
zend = zsp->z_start + zsp->z_size;
}
}
}
static int check_zbc_access_params(struct scsi_cmnd *scp,
unsigned long long lba, unsigned int num, bool write)
{
struct scsi_device *sdp = scp->device;
struct sdebug_dev_info *devip = (struct sdebug_dev_info *)sdp->hostdata;
struct sdeb_zone_state *zsp = zbc_zone(devip, lba);
struct sdeb_zone_state *zsp_end = zbc_zone(devip, lba + num - 1);
if (!write) {
if (devip->zmodel == BLK_ZONED_HA)
return 0;
/* For host-managed, reads cannot cross zone types boundaries */
if (zsp->z_type != zsp_end->z_type) {
mk_sense_buffer(scp, ILLEGAL_REQUEST,
LBA_OUT_OF_RANGE,
READ_INVDATA_ASCQ);
return check_condition_result;
}
return 0;
}
/* Writing into a gap zone is not allowed */
if (zbc_zone_is_gap(zsp)) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, LBA_OUT_OF_RANGE,
ATTEMPT_ACCESS_GAP);
return check_condition_result;
}
/* No restrictions for writes within conventional zones */
if (zbc_zone_is_conv(zsp)) {
if (!zbc_zone_is_conv(zsp_end)) {
mk_sense_buffer(scp, ILLEGAL_REQUEST,
LBA_OUT_OF_RANGE,
WRITE_BOUNDARY_ASCQ);
return check_condition_result;
}
return 0;
}
if (zsp->z_type == ZBC_ZTYPE_SWR) {
/* Writes cannot cross sequential zone boundaries */
if (zsp_end != zsp) {
mk_sense_buffer(scp, ILLEGAL_REQUEST,
LBA_OUT_OF_RANGE,
WRITE_BOUNDARY_ASCQ);
return check_condition_result;
}
/* Cannot write full zones */
if (zsp->z_cond == ZC5_FULL) {
mk_sense_buffer(scp, ILLEGAL_REQUEST,
INVALID_FIELD_IN_CDB, 0);
return check_condition_result;
}
/* Writes must be aligned to the zone WP */
if (lba != zsp->z_wp) {
mk_sense_buffer(scp, ILLEGAL_REQUEST,
LBA_OUT_OF_RANGE,
UNALIGNED_WRITE_ASCQ);
return check_condition_result;
}
}
/* Handle implicit open of closed and empty zones */
if (zsp->z_cond == ZC1_EMPTY || zsp->z_cond == ZC4_CLOSED) {
if (devip->max_open &&
devip->nr_exp_open >= devip->max_open) {
mk_sense_buffer(scp, DATA_PROTECT,
INSUFF_RES_ASC,
INSUFF_ZONE_ASCQ);
return check_condition_result;
}
zbc_open_zone(devip, zsp, false);
}
return 0;
}
static inline int check_device_access_params
(struct scsi_cmnd *scp, unsigned long long lba,
unsigned int num, bool write)
{
struct scsi_device *sdp = scp->device;
struct sdebug_dev_info *devip = (struct sdebug_dev_info *)sdp->hostdata;
if (lba + num > sdebug_capacity) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, LBA_OUT_OF_RANGE, 0);
return check_condition_result;
}
/* transfer length excessive (tie in to block limits VPD page) */
if (num > sdebug_store_sectors) {
/* needs work to find which cdb byte 'num' comes from */
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_FIELD_IN_CDB, 0);
return check_condition_result;
}
if (write && unlikely(sdebug_wp)) {
mk_sense_buffer(scp, DATA_PROTECT, WRITE_PROTECTED, 0x2);
return check_condition_result;
}
if (sdebug_dev_is_zoned(devip))
return check_zbc_access_params(scp, lba, num, write);
return 0;
}
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
/*
* Note: if BUG_ON() fires it usually indicates a problem with the parser
* tables. Perhaps a missing F_FAKE_RW or FF_MEDIA_IO flag. Response functions
* that access any of the "stores" in struct sdeb_store_info should call this
* function with bug_if_fake_rw set to true.
*/
static inline struct sdeb_store_info *devip2sip(struct sdebug_dev_info *devip,
bool bug_if_fake_rw)
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
{
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
if (sdebug_fake_rw) {
BUG_ON(bug_if_fake_rw); /* See note above */
return NULL;
}
return xa_load(per_store_ap, devip->sdbg_host->si_idx);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
}
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
/* Returns number of bytes copied or -1 if error. */
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static int do_device_access(struct sdeb_store_info *sip, struct scsi_cmnd *scp,
u32 sg_skip, u64 lba, u32 num, bool do_write)
{
int ret;
u64 block, rest = 0;
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
enum dma_data_direction dir;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
struct scsi_data_buffer *sdb = &scp->sdb;
u8 *fsp;
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
if (do_write) {
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
dir = DMA_TO_DEVICE;
write_since_sync = true;
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
} else {
dir = DMA_FROM_DEVICE;
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
if (!sdb->length || !sip)
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
return 0;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
if (scp->sc_data_direction != dir)
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
return -1;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
fsp = sip->storep;
block = do_div(lba, sdebug_store_sectors);
if (block + num > sdebug_store_sectors)
rest = block + num - sdebug_store_sectors;
ret = sg_copy_buffer(sdb->table.sgl, sdb->table.nents,
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
fsp + (block * sdebug_sector_size),
(num - rest) * sdebug_sector_size, sg_skip, do_write);
if (ret != (num - rest) * sdebug_sector_size)
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
return ret;
if (rest) {
ret += sg_copy_buffer(sdb->table.sgl, sdb->table.nents,
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
fsp, rest * sdebug_sector_size,
sg_skip + ((num - rest) * sdebug_sector_size),
do_write);
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
}
return ret;
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
/* Returns number of bytes copied or -1 if error. */
static int do_dout_fetch(struct scsi_cmnd *scp, u32 num, u8 *doutp)
{
struct scsi_data_buffer *sdb = &scp->sdb;
if (!sdb->length)
return 0;
if (scp->sc_data_direction != DMA_TO_DEVICE)
return -1;
return sg_copy_buffer(sdb->table.sgl, sdb->table.nents, doutp,
num * sdebug_sector_size, 0, true);
}
/* If sip->storep+lba compares equal to arr(num), then copy top half of
* arr into sip->storep+lba and return true. If comparison fails then
* return false. */
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static bool comp_write_worker(struct sdeb_store_info *sip, u64 lba, u32 num,
const u8 *arr, bool compare_only)
{
bool res;
u64 block, rest = 0;
u32 store_blks = sdebug_store_sectors;
u32 lb_size = sdebug_sector_size;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
u8 *fsp = sip->storep;
block = do_div(lba, store_blks);
if (block + num > store_blks)
rest = block + num - store_blks;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
res = !memcmp(fsp + (block * lb_size), arr, (num - rest) * lb_size);
if (!res)
return res;
if (rest)
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
res = memcmp(fsp, arr + ((num - rest) * lb_size),
rest * lb_size);
if (!res)
return res;
if (compare_only)
return true;
arr += num * lb_size;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
memcpy(fsp + (block * lb_size), arr, (num - rest) * lb_size);
if (rest)
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
memcpy(fsp, arr + ((num - rest) * lb_size), rest * lb_size);
return res;
}
static __be16 dif_compute_csum(const void *buf, int len)
{
__be16 csum;
if (sdebug_guard)
csum = (__force __be16)ip_compute_csum(buf, len);
else
csum = cpu_to_be16(crc_t10dif(buf, len));
return csum;
}
static int dif_verify(struct t10_pi_tuple *sdt, const void *data,
sector_t sector, u32 ei_lba)
{
__be16 csum = dif_compute_csum(data, sdebug_sector_size);
if (sdt->guard_tag != csum) {
pr_err("GUARD check failed on sector %lu rcvd 0x%04x, data 0x%04x\n",
(unsigned long)sector,
be16_to_cpu(sdt->guard_tag),
be16_to_cpu(csum));
return 0x01;
}
if (sdebug_dif == T10_PI_TYPE1_PROTECTION &&
be32_to_cpu(sdt->ref_tag) != (sector & 0xffffffff)) {
pr_err("REF check failed on sector %lu\n",
(unsigned long)sector);
return 0x03;
}
if (sdebug_dif == T10_PI_TYPE2_PROTECTION &&
be32_to_cpu(sdt->ref_tag) != ei_lba) {
pr_err("REF check failed on sector %lu\n",
(unsigned long)sector);
return 0x03;
}
return 0;
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static void dif_copy_prot(struct scsi_cmnd *scp, sector_t sector,
unsigned int sectors, bool read)
{
size_t resid;
void *paddr;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
struct sdeb_store_info *sip = devip2sip((struct sdebug_dev_info *)
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
scp->device->hostdata, true);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
struct t10_pi_tuple *dif_storep = sip->dif_storep;
const void *dif_store_end = dif_storep + sdebug_store_sectors;
struct sg_mapping_iter miter;
/* Bytes of protection data to copy into sgl */
resid = sectors * sizeof(*dif_storep);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
sg_miter_start(&miter, scsi_prot_sglist(scp),
scsi_prot_sg_count(scp), SG_MITER_ATOMIC |
(read ? SG_MITER_TO_SG : SG_MITER_FROM_SG));
while (sg_miter_next(&miter) && resid > 0) {
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
size_t len = min_t(size_t, miter.length, resid);
void *start = dif_store(sip, sector);
size_t rest = 0;
if (dif_store_end < start + len)
rest = start + len - dif_store_end;
paddr = miter.addr;
if (read)
memcpy(paddr, start, len - rest);
else
memcpy(start, paddr, len - rest);
if (rest) {
if (read)
memcpy(paddr + len - rest, dif_storep, rest);
else
memcpy(dif_storep, paddr + len - rest, rest);
}
sector += len / sizeof(*dif_storep);
resid -= len;
}
sg_miter_stop(&miter);
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static int prot_verify_read(struct scsi_cmnd *scp, sector_t start_sec,
unsigned int sectors, u32 ei_lba)
{
int ret = 0;
unsigned int i;
sector_t sector;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
struct sdeb_store_info *sip = devip2sip((struct sdebug_dev_info *)
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
scp->device->hostdata, true);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
struct t10_pi_tuple *sdt;
for (i = 0; i < sectors; i++, ei_lba++) {
sector = start_sec + i;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
sdt = dif_store(sip, sector);
if (sdt->app_tag == cpu_to_be16(0xffff))
continue;
/*
* Because scsi_debug acts as both initiator and
* target we proceed to verify the PI even if
* RDPROTECT=3. This is done so the "initiator" knows
* which type of error to return. Otherwise we would
* have to iterate over the PI twice.
*/
if (scp->cmnd[1] >> 5) { /* RDPROTECT */
ret = dif_verify(sdt, lba2fake_store(sip, sector),
sector, ei_lba);
if (ret) {
dif_errors++;
break;
}
}
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
dif_copy_prot(scp, start_sec, sectors, true);
dix_reads++;
return ret;
}
static inline void
sdeb_read_lock(struct sdeb_store_info *sip)
{
if (sdebug_no_rwlock) {
if (sip)
__acquire(&sip->macc_lck);
else
__acquire(&sdeb_fake_rw_lck);
} else {
if (sip)
read_lock(&sip->macc_lck);
else
read_lock(&sdeb_fake_rw_lck);
}
}
static inline void
sdeb_read_unlock(struct sdeb_store_info *sip)
{
if (sdebug_no_rwlock) {
if (sip)
__release(&sip->macc_lck);
else
__release(&sdeb_fake_rw_lck);
} else {
if (sip)
read_unlock(&sip->macc_lck);
else
read_unlock(&sdeb_fake_rw_lck);
}
}
static inline void
sdeb_write_lock(struct sdeb_store_info *sip)
{
if (sdebug_no_rwlock) {
if (sip)
__acquire(&sip->macc_lck);
else
__acquire(&sdeb_fake_rw_lck);
} else {
if (sip)
write_lock(&sip->macc_lck);
else
write_lock(&sdeb_fake_rw_lck);
}
}
static inline void
sdeb_write_unlock(struct sdeb_store_info *sip)
{
if (sdebug_no_rwlock) {
if (sip)
__release(&sip->macc_lck);
else
__release(&sdeb_fake_rw_lck);
} else {
if (sip)
write_unlock(&sip->macc_lck);
else
write_unlock(&sdeb_fake_rw_lck);
}
}
static int resp_read_dt0(struct scsi_cmnd *scp, struct sdebug_dev_info *devip)
{
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
bool check_prot;
u32 num;
u32 ei_lba;
int ret;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
u64 lba;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, true);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
u8 *cmd = scp->cmnd;
switch (cmd[0]) {
case READ_16:
ei_lba = 0;
lba = get_unaligned_be64(cmd + 2);
num = get_unaligned_be32(cmd + 10);
check_prot = true;
break;
case READ_10:
ei_lba = 0;
lba = get_unaligned_be32(cmd + 2);
num = get_unaligned_be16(cmd + 7);
check_prot = true;
break;
case READ_6:
ei_lba = 0;
lba = (u32)cmd[3] | (u32)cmd[2] << 8 |
(u32)(cmd[1] & 0x1f) << 16;
num = (0 == cmd[4]) ? 256 : cmd[4];
check_prot = true;
break;
case READ_12:
ei_lba = 0;
lba = get_unaligned_be32(cmd + 2);
num = get_unaligned_be32(cmd + 6);
check_prot = true;
break;
case XDWRITEREAD_10:
ei_lba = 0;
lba = get_unaligned_be32(cmd + 2);
num = get_unaligned_be16(cmd + 7);
check_prot = false;
break;
default: /* assume READ(32) */
lba = get_unaligned_be64(cmd + 12);
ei_lba = get_unaligned_be32(cmd + 20);
num = get_unaligned_be32(cmd + 28);
check_prot = false;
break;
}
if (unlikely(have_dif_prot && check_prot)) {
if (sdebug_dif == T10_PI_TYPE2_PROTECTION &&
(cmd[1] & 0xe0)) {
mk_sense_invalid_opcode(scp);
return check_condition_result;
}
if ((sdebug_dif == T10_PI_TYPE1_PROTECTION ||
sdebug_dif == T10_PI_TYPE3_PROTECTION) &&
(cmd[1] & 0xe0) == 0)
sdev_printk(KERN_ERR, scp->device, "Unprotected RD "
"to DIF device\n");
}
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
if (unlikely((sdebug_opts & SDEBUG_OPT_SHORT_TRANSFER) &&
atomic_read(&sdeb_inject_pending))) {
num /= 2;
atomic_set(&sdeb_inject_pending, 0);
}
ret = check_device_access_params(scp, lba, num, false);
if (ret)
return ret;
if (unlikely((SDEBUG_OPT_MEDIUM_ERR & sdebug_opts) &&
(lba <= (sdebug_medium_error_start + sdebug_medium_error_count - 1)) &&
((lba + num) > sdebug_medium_error_start))) {
/* claim unrecoverable read error */
mk_sense_buffer(scp, MEDIUM_ERROR, UNRECOVERED_READ_ERR, 0);
/* set info field and valid bit for fixed descriptor */
if (0x70 == (scp->sense_buffer[0] & 0x7f)) {
scp->sense_buffer[0] |= 0x80; /* Valid bit */
ret = (lba < OPT_MEDIUM_ERR_ADDR)
? OPT_MEDIUM_ERR_ADDR : (int)lba;
put_unaligned_be32(ret, scp->sense_buffer + 3);
}
scsi_set_resid(scp, scsi_bufflen(scp));
return check_condition_result;
}
sdeb_read_lock(sip);
/* DIX + T10 DIF */
if (unlikely(sdebug_dix && scsi_prot_sg_count(scp))) {
switch (prot_verify_read(scp, lba, num, ei_lba)) {
case 1: /* Guard tag error */
if (cmd[1] >> 5 != 3) { /* RDPROTECT != 3 */
sdeb_read_unlock(sip);
mk_sense_buffer(scp, ABORTED_COMMAND, 0x10, 1);
return check_condition_result;
} else if (scp->prot_flags & SCSI_PROT_GUARD_CHECK) {
sdeb_read_unlock(sip);
mk_sense_buffer(scp, ILLEGAL_REQUEST, 0x10, 1);
return illegal_condition_result;
}
break;
case 3: /* Reference tag error */
if (cmd[1] >> 5 != 3) { /* RDPROTECT != 3 */
sdeb_read_unlock(sip);
mk_sense_buffer(scp, ABORTED_COMMAND, 0x10, 3);
return check_condition_result;
} else if (scp->prot_flags & SCSI_PROT_REF_CHECK) {
sdeb_read_unlock(sip);
mk_sense_buffer(scp, ILLEGAL_REQUEST, 0x10, 3);
return illegal_condition_result;
}
break;
}
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
ret = do_device_access(sip, scp, 0, lba, num, false);
sdeb_read_unlock(sip);
if (unlikely(ret == -1))
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
return DID_ERROR << 16;
scsi_set_resid(scp, scsi_bufflen(scp) - ret);
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
if (unlikely((sdebug_opts & SDEBUG_OPT_RECOV_DIF_DIX) &&
atomic_read(&sdeb_inject_pending))) {
if (sdebug_opts & SDEBUG_OPT_RECOVERED_ERR) {
mk_sense_buffer(scp, RECOVERED_ERROR, THRESHOLD_EXCEEDED, 0);
atomic_set(&sdeb_inject_pending, 0);
return check_condition_result;
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
} else if (sdebug_opts & SDEBUG_OPT_DIF_ERR) {
/* Logical block guard check failed */
mk_sense_buffer(scp, ABORTED_COMMAND, 0x10, 1);
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
atomic_set(&sdeb_inject_pending, 0);
return illegal_condition_result;
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
} else if (SDEBUG_OPT_DIX_ERR & sdebug_opts) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, 0x10, 1);
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
atomic_set(&sdeb_inject_pending, 0);
return illegal_condition_result;
}
}
scsi_debug: fix do_device_access() with wrap around range do_device_access() is a function that abstracts copying SG list from/to ramdisk storage (fake_storep). It must deal with the ranges exceeding actual fake_storep size, because such ranges are valid if virtual_gb is set greater than zero, and they should be treated as fake_storep is repeatedly mirrored up to virtual size. Unfortunately, it can't deal with the range which wraps around the end of fake_storep. A wrap around range is copied by two sg_copy_{from,to}_buffer() calls, but sg_copy_{from,to}_buffer() can't copy from/to in the middle of SG list, therefore the second call can't copy correctly. This fixes it by using sg_pcopy_{from,to}_buffer() that can copy from/to the middle of SG list. This also simplifies the assignment of sdb->resid in fill_from_dev_buffer(). Because fill_from_dev_buffer() is now only called once per command execution cycle. So it is not necessary to take care to decrease sdb->resid if fill_from_dev_buffer() is called more than once. Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Horia Geanta <horia.geanta@freescale.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-07-08 23:01:57 +00:00
return 0;
}
static int prot_verify_write(struct scsi_cmnd *SCpnt, sector_t start_sec,
unsigned int sectors, u32 ei_lba)
{
int ret;
struct t10_pi_tuple *sdt;
void *daddr;
sector_t sector = start_sec;
int ppage_offset;
int dpage_offset;
struct sg_mapping_iter diter;
struct sg_mapping_iter piter;
BUG_ON(scsi_sg_count(SCpnt) == 0);
BUG_ON(scsi_prot_sg_count(SCpnt) == 0);
sg_miter_start(&piter, scsi_prot_sglist(SCpnt),
scsi_prot_sg_count(SCpnt),
SG_MITER_ATOMIC | SG_MITER_FROM_SG);
sg_miter_start(&diter, scsi_sglist(SCpnt), scsi_sg_count(SCpnt),
SG_MITER_ATOMIC | SG_MITER_FROM_SG);
/* For each protection page */
while (sg_miter_next(&piter)) {
dpage_offset = 0;
if (WARN_ON(!sg_miter_next(&diter))) {
ret = 0x01;
goto out;
}
for (ppage_offset = 0; ppage_offset < piter.length;
ppage_offset += sizeof(struct t10_pi_tuple)) {
/* If we're at the end of the current
* data page advance to the next one
*/
if (dpage_offset >= diter.length) {
if (WARN_ON(!sg_miter_next(&diter))) {
ret = 0x01;
goto out;
}
dpage_offset = 0;
}
sdt = piter.addr + ppage_offset;
daddr = diter.addr + dpage_offset;
if (SCpnt->cmnd[1] >> 5 != 3) { /* WRPROTECT */
ret = dif_verify(sdt, daddr, sector, ei_lba);
if (ret)
goto out;
}
sector++;
ei_lba++;
dpage_offset += sdebug_sector_size;
}
diter.consumed = dpage_offset;
sg_miter_stop(&diter);
}
sg_miter_stop(&piter);
dif_copy_prot(SCpnt, start_sec, sectors, false);
dix_writes++;
return 0;
out:
dif_errors++;
sg_miter_stop(&diter);
sg_miter_stop(&piter);
return ret;
}
static unsigned long lba_to_map_index(sector_t lba)
{
if (sdebug_unmap_alignment)
lba += sdebug_unmap_granularity - sdebug_unmap_alignment;
sector_div(lba, sdebug_unmap_granularity);
return lba;
}
static sector_t map_index_to_lba(unsigned long index)
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
{
sector_t lba = index * sdebug_unmap_granularity;
if (sdebug_unmap_alignment)
lba -= sdebug_unmap_granularity - sdebug_unmap_alignment;
return lba;
}
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static unsigned int map_state(struct sdeb_store_info *sip, sector_t lba,
unsigned int *num)
{
sector_t end;
unsigned int mapped;
unsigned long index;
unsigned long next;
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
index = lba_to_map_index(lba);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
mapped = test_bit(index, sip->map_storep);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
if (mapped)
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
next = find_next_zero_bit(sip->map_storep, map_size, index);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
else
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
next = find_next_bit(sip->map_storep, map_size, index);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
end = min_t(sector_t, sdebug_store_sectors, map_index_to_lba(next));
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
*num = end - lba;
return mapped;
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static void map_region(struct sdeb_store_info *sip, sector_t lba,
unsigned int len)
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
{
sector_t end = lba + len;
while (lba < end) {
unsigned long index = lba_to_map_index(lba);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
if (index < map_size)
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
set_bit(index, sip->map_storep);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
lba = map_index_to_lba(index + 1);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
}
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static void unmap_region(struct sdeb_store_info *sip, sector_t lba,
unsigned int len)
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
{
sector_t end = lba + len;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
u8 *fsp = sip->storep;
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
while (lba < end) {
unsigned long index = lba_to_map_index(lba);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
if (lba == map_index_to_lba(index) &&
lba + sdebug_unmap_granularity <= end &&
index < map_size) {
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
clear_bit(index, sip->map_storep);
if (sdebug_lbprz) { /* for LBPRZ=2 return 0xff_s */
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
memset(fsp + lba * sdebug_sector_size,
(sdebug_lbprz & 1) ? 0 : 0xff,
sdebug_sector_size *
sdebug_unmap_granularity);
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
if (sip->dif_storep) {
memset(sip->dif_storep + lba, 0xff,
sizeof(*sip->dif_storep) *
sdebug_unmap_granularity);
}
}
lba = map_index_to_lba(index + 1);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
}
}
static int resp_write_dt0(struct scsi_cmnd *scp, struct sdebug_dev_info *devip)
{
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
bool check_prot;
u32 num;
u32 ei_lba;
int ret;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
u64 lba;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, true);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
u8 *cmd = scp->cmnd;
switch (cmd[0]) {
case WRITE_16:
ei_lba = 0;
lba = get_unaligned_be64(cmd + 2);
num = get_unaligned_be32(cmd + 10);
check_prot = true;
break;
case WRITE_10:
ei_lba = 0;
lba = get_unaligned_be32(cmd + 2);
num = get_unaligned_be16(cmd + 7);
check_prot = true;
break;
case WRITE_6:
ei_lba = 0;
lba = (u32)cmd[3] | (u32)cmd[2] << 8 |
(u32)(cmd[1] & 0x1f) << 16;
num = (0 == cmd[4]) ? 256 : cmd[4];
check_prot = true;
break;
case WRITE_12:
ei_lba = 0;
lba = get_unaligned_be32(cmd + 2);
num = get_unaligned_be32(cmd + 6);
check_prot = true;
break;
case 0x53: /* XDWRITEREAD(10) */
ei_lba = 0;
lba = get_unaligned_be32(cmd + 2);
num = get_unaligned_be16(cmd + 7);
check_prot = false;
break;
default: /* assume WRITE(32) */
lba = get_unaligned_be64(cmd + 12);
ei_lba = get_unaligned_be32(cmd + 20);
num = get_unaligned_be32(cmd + 28);
check_prot = false;
break;
}
if (unlikely(have_dif_prot && check_prot)) {
if (sdebug_dif == T10_PI_TYPE2_PROTECTION &&
(cmd[1] & 0xe0)) {
mk_sense_invalid_opcode(scp);
return check_condition_result;
}
if ((sdebug_dif == T10_PI_TYPE1_PROTECTION ||
sdebug_dif == T10_PI_TYPE3_PROTECTION) &&
(cmd[1] & 0xe0) == 0)
sdev_printk(KERN_ERR, scp->device, "Unprotected WR "
"to DIF device\n");
}
sdeb_write_lock(sip);
ret = check_device_access_params(scp, lba, num, true);
if (ret) {
sdeb_write_unlock(sip);
return ret;
}
/* DIX + T10 DIF */
if (unlikely(sdebug_dix && scsi_prot_sg_count(scp))) {
switch (prot_verify_write(scp, lba, num, ei_lba)) {
case 1: /* Guard tag error */
if (scp->prot_flags & SCSI_PROT_GUARD_CHECK) {
sdeb_write_unlock(sip);
mk_sense_buffer(scp, ILLEGAL_REQUEST, 0x10, 1);
return illegal_condition_result;
} else if (scp->cmnd[1] >> 5 != 3) { /* WRPROTECT != 3 */
sdeb_write_unlock(sip);
mk_sense_buffer(scp, ABORTED_COMMAND, 0x10, 1);
return check_condition_result;
}
break;
case 3: /* Reference tag error */
if (scp->prot_flags & SCSI_PROT_REF_CHECK) {
sdeb_write_unlock(sip);
mk_sense_buffer(scp, ILLEGAL_REQUEST, 0x10, 3);
return illegal_condition_result;
} else if (scp->cmnd[1] >> 5 != 3) { /* WRPROTECT != 3 */
sdeb_write_unlock(sip);
mk_sense_buffer(scp, ABORTED_COMMAND, 0x10, 3);
return check_condition_result;
}
break;
}
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
ret = do_device_access(sip, scp, 0, lba, num, true);
if (unlikely(scsi_debug_lbp()))
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
map_region(sip, lba, num);
/* If ZBC zone then bump its write pointer */
if (sdebug_dev_is_zoned(devip))
zbc_inc_wp(devip, lba, num);
sdeb_write_unlock(sip);
if (unlikely(-1 == ret))
return DID_ERROR << 16;
else if (unlikely(sdebug_verbose &&
(ret < (num * sdebug_sector_size))))
sdev_printk(KERN_INFO, scp->device,
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
"%s: write: cdb indicated=%u, IO sent=%d bytes\n",
my_name, num * sdebug_sector_size, ret);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
if (unlikely((sdebug_opts & SDEBUG_OPT_RECOV_DIF_DIX) &&
atomic_read(&sdeb_inject_pending))) {
if (sdebug_opts & SDEBUG_OPT_RECOVERED_ERR) {
mk_sense_buffer(scp, RECOVERED_ERROR, THRESHOLD_EXCEEDED, 0);
atomic_set(&sdeb_inject_pending, 0);
return check_condition_result;
} else if (sdebug_opts & SDEBUG_OPT_DIF_ERR) {
/* Logical block guard check failed */
mk_sense_buffer(scp, ABORTED_COMMAND, 0x10, 1);
atomic_set(&sdeb_inject_pending, 0);
return illegal_condition_result;
} else if (sdebug_opts & SDEBUG_OPT_DIX_ERR) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, 0x10, 1);
atomic_set(&sdeb_inject_pending, 0);
return illegal_condition_result;
}
}
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
return 0;
}
/*
* T10 has only specified WRITE SCATTERED(16) and WRITE SCATTERED(32).
* No READ GATHERED yet (requires bidi or long cdb holding gather list).
*/
static int resp_write_scat(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
u8 *cmd = scp->cmnd;
u8 *lrdp = NULL;
u8 *up;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, true);
u8 wrprotect;
u16 lbdof, num_lrd, k;
u32 num, num_by, bt_len, lbdof_blen, sg_off, cum_lb;
u32 lb_size = sdebug_sector_size;
u32 ei_lba;
u64 lba;
int ret, res;
bool is_16;
static const u32 lrd_size = 32; /* + parameter list header size */
if (cmd[0] == VARIABLE_LENGTH_CMD) {
is_16 = false;
wrprotect = (cmd[10] >> 5) & 0x7;
lbdof = get_unaligned_be16(cmd + 12);
num_lrd = get_unaligned_be16(cmd + 16);
bt_len = get_unaligned_be32(cmd + 28);
} else { /* that leaves WRITE SCATTERED(16) */
is_16 = true;
wrprotect = (cmd[2] >> 5) & 0x7;
lbdof = get_unaligned_be16(cmd + 4);
num_lrd = get_unaligned_be16(cmd + 8);
bt_len = get_unaligned_be32(cmd + 10);
if (unlikely(have_dif_prot)) {
if (sdebug_dif == T10_PI_TYPE2_PROTECTION &&
wrprotect) {
mk_sense_invalid_opcode(scp);
return illegal_condition_result;
}
if ((sdebug_dif == T10_PI_TYPE1_PROTECTION ||
sdebug_dif == T10_PI_TYPE3_PROTECTION) &&
wrprotect == 0)
sdev_printk(KERN_ERR, scp->device,
"Unprotected WR to DIF device\n");
}
}
if ((num_lrd == 0) || (bt_len == 0))
return 0; /* T10 says these do-nothings are not errors */
if (lbdof == 0) {
if (sdebug_verbose)
sdev_printk(KERN_INFO, scp->device,
"%s: %s: LB Data Offset field bad\n",
my_name, __func__);
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_FIELD_IN_CDB, 0);
return illegal_condition_result;
}
lbdof_blen = lbdof * lb_size;
if ((lrd_size + (num_lrd * lrd_size)) > lbdof_blen) {
if (sdebug_verbose)
sdev_printk(KERN_INFO, scp->device,
"%s: %s: LBA range descriptors don't fit\n",
my_name, __func__);
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_FIELD_IN_CDB, 0);
return illegal_condition_result;
}
lrdp = kzalloc(lbdof_blen, GFP_ATOMIC | __GFP_NOWARN);
if (lrdp == NULL)
return SCSI_MLQUEUE_HOST_BUSY;
if (sdebug_verbose)
sdev_printk(KERN_INFO, scp->device,
"%s: %s: Fetch header+scatter_list, lbdof_blen=%u\n",
my_name, __func__, lbdof_blen);
res = fetch_to_dev_buffer(scp, lrdp, lbdof_blen);
if (res == -1) {
ret = DID_ERROR << 16;
goto err_out;
}
sdeb_write_lock(sip);
sg_off = lbdof_blen;
/* Spec says Buffer xfer Length field in number of LBs in dout */
cum_lb = 0;
for (k = 0, up = lrdp + lrd_size; k < num_lrd; ++k, up += lrd_size) {
lba = get_unaligned_be64(up + 0);
num = get_unaligned_be32(up + 8);
if (sdebug_verbose)
sdev_printk(KERN_INFO, scp->device,
"%s: %s: k=%d LBA=0x%llx num=%u sg_off=%u\n",
my_name, __func__, k, lba, num, sg_off);
if (num == 0)
continue;
ret = check_device_access_params(scp, lba, num, true);
if (ret)
goto err_out_unlock;
num_by = num * lb_size;
ei_lba = is_16 ? 0 : get_unaligned_be32(up + 12);
if ((cum_lb + num) > bt_len) {
if (sdebug_verbose)
sdev_printk(KERN_INFO, scp->device,
"%s: %s: sum of blocks > data provided\n",
my_name, __func__);
mk_sense_buffer(scp, ILLEGAL_REQUEST, WRITE_ERROR_ASC,
0);
ret = illegal_condition_result;
goto err_out_unlock;
}
/* DIX + T10 DIF */
if (unlikely(sdebug_dix && scsi_prot_sg_count(scp))) {
int prot_ret = prot_verify_write(scp, lba, num,
ei_lba);
if (prot_ret) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, 0x10,
prot_ret);
ret = illegal_condition_result;
goto err_out_unlock;
}
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
ret = do_device_access(sip, scp, sg_off, lba, num, true);
/* If ZBC zone then bump its write pointer */
if (sdebug_dev_is_zoned(devip))
zbc_inc_wp(devip, lba, num);
if (unlikely(scsi_debug_lbp()))
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
map_region(sip, lba, num);
if (unlikely(-1 == ret)) {
ret = DID_ERROR << 16;
goto err_out_unlock;
} else if (unlikely(sdebug_verbose && (ret < num_by)))
sdev_printk(KERN_INFO, scp->device,
"%s: write: cdb indicated=%u, IO sent=%d bytes\n",
my_name, num_by, ret);
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
if (unlikely((sdebug_opts & SDEBUG_OPT_RECOV_DIF_DIX) &&
atomic_read(&sdeb_inject_pending))) {
if (sdebug_opts & SDEBUG_OPT_RECOVERED_ERR) {
mk_sense_buffer(scp, RECOVERED_ERROR, THRESHOLD_EXCEEDED, 0);
atomic_set(&sdeb_inject_pending, 0);
ret = check_condition_result;
goto err_out_unlock;
} else if (sdebug_opts & SDEBUG_OPT_DIF_ERR) {
/* Logical block guard check failed */
mk_sense_buffer(scp, ABORTED_COMMAND, 0x10, 1);
atomic_set(&sdeb_inject_pending, 0);
ret = illegal_condition_result;
goto err_out_unlock;
} else if (sdebug_opts & SDEBUG_OPT_DIX_ERR) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, 0x10, 1);
atomic_set(&sdeb_inject_pending, 0);
ret = illegal_condition_result;
goto err_out_unlock;
}
}
sg_off += num_by;
cum_lb += num;
}
ret = 0;
err_out_unlock:
sdeb_write_unlock(sip);
err_out:
kfree(lrdp);
return ret;
}
static int resp_write_same(struct scsi_cmnd *scp, u64 lba, u32 num,
u32 ei_lba, bool unmap, bool ndob)
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
{
struct scsi_device *sdp = scp->device;
struct sdebug_dev_info *devip = (struct sdebug_dev_info *)sdp->hostdata;
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
unsigned long long i;
u64 block, lbaa;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
u32 lb_size = sdebug_sector_size;
int ret;
struct sdeb_store_info *sip = devip2sip((struct sdebug_dev_info *)
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
scp->device->hostdata, true);
u8 *fs1p;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
u8 *fsp;
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
sdeb_write_lock(sip);
ret = check_device_access_params(scp, lba, num, true);
if (ret) {
sdeb_write_unlock(sip);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
return ret;
}
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
if (unmap && scsi_debug_lbp()) {
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
unmap_region(sip, lba, num);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
goto out;
}
lbaa = lba;
block = do_div(lbaa, sdebug_store_sectors);
/* if ndob then zero 1 logical block, else fetch 1 logical block */
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
fsp = sip->storep;
fs1p = fsp + (block * lb_size);
if (ndob) {
memset(fs1p, 0, lb_size);
ret = 0;
} else
ret = fetch_to_dev_buffer(scp, fs1p, lb_size);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
if (-1 == ret) {
sdeb_write_unlock(sip);
return DID_ERROR << 16;
} else if (sdebug_verbose && !ndob && (ret < lb_size))
sdev_printk(KERN_INFO, scp->device,
"%s: %s: lb size=%u, IO sent=%d bytes\n",
my_name, "write same", lb_size, ret);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
/* Copy first sector to remaining blocks */
for (i = 1 ; i < num ; i++) {
lbaa = lba + i;
block = do_div(lbaa, sdebug_store_sectors);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
memmove(fsp + (block * lb_size), fs1p, lb_size);
}
if (scsi_debug_lbp())
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
map_region(sip, lba, num);
/* If ZBC zone then bump its write pointer */
if (sdebug_dev_is_zoned(devip))
zbc_inc_wp(devip, lba, num);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
out:
sdeb_write_unlock(sip);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
return 0;
}
static int resp_write_same_10(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
u8 *cmd = scp->cmnd;
u32 lba;
u16 num;
u32 ei_lba = 0;
bool unmap = false;
if (cmd[1] & 0x8) {
if (sdebug_lbpws10 == 0) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 1, 3);
return check_condition_result;
} else
unmap = true;
}
lba = get_unaligned_be32(cmd + 2);
num = get_unaligned_be16(cmd + 7);
if (num > sdebug_write_same_length) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 7, -1);
return check_condition_result;
}
return resp_write_same(scp, lba, num, ei_lba, unmap, false);
}
static int resp_write_same_16(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
u8 *cmd = scp->cmnd;
u64 lba;
u32 num;
u32 ei_lba = 0;
bool unmap = false;
bool ndob = false;
if (cmd[1] & 0x8) { /* UNMAP */
if (sdebug_lbpws == 0) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 1, 3);
return check_condition_result;
} else
unmap = true;
}
if (cmd[1] & 0x1) /* NDOB (no data-out buffer, assumes zeroes) */
ndob = true;
lba = get_unaligned_be64(cmd + 2);
num = get_unaligned_be32(cmd + 10);
if (num > sdebug_write_same_length) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 10, -1);
return check_condition_result;
}
return resp_write_same(scp, lba, num, ei_lba, unmap, ndob);
}
/* Note the mode field is in the same position as the (lower) service action
* field. For the Report supported operation codes command, SPC-4 suggests
* each mode of this command should be reported separately; for future. */
static int resp_write_buffer(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
u8 *cmd = scp->cmnd;
struct scsi_device *sdp = scp->device;
struct sdebug_dev_info *dp;
u8 mode;
mode = cmd[1] & 0x1f;
switch (mode) {
case 0x4: /* download microcode (MC) and activate (ACT) */
/* set UAs on this device only */
set_bit(SDEBUG_UA_BUS_RESET, devip->uas_bm);
set_bit(SDEBUG_UA_MICROCODE_CHANGED, devip->uas_bm);
break;
case 0x5: /* download MC, save and ACT */
set_bit(SDEBUG_UA_MICROCODE_CHANGED_WO_RESET, devip->uas_bm);
break;
case 0x6: /* download MC with offsets and ACT */
/* set UAs on most devices (LUs) in this target */
list_for_each_entry(dp,
&devip->sdbg_host->dev_info_list,
dev_list)
if (dp->target == sdp->id) {
set_bit(SDEBUG_UA_BUS_RESET, dp->uas_bm);
if (devip != dp)
set_bit(SDEBUG_UA_MICROCODE_CHANGED,
dp->uas_bm);
}
break;
case 0x7: /* download MC with offsets, save, and ACT */
/* set UA on all devices (LUs) in this target */
list_for_each_entry(dp,
&devip->sdbg_host->dev_info_list,
dev_list)
if (dp->target == sdp->id)
set_bit(SDEBUG_UA_MICROCODE_CHANGED_WO_RESET,
dp->uas_bm);
break;
default:
/* do nothing for this command for other mode values */
break;
}
return 0;
}
static int resp_comp_write(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
u8 *cmd = scp->cmnd;
u8 *arr;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, true);
u64 lba;
u32 dnum;
u32 lb_size = sdebug_sector_size;
u8 num;
int ret;
int retval = 0;
lba = get_unaligned_be64(cmd + 2);
num = cmd[13]; /* 1 to a maximum of 255 logical blocks */
if (0 == num)
return 0; /* degenerate case, not an error */
if (sdebug_dif == T10_PI_TYPE2_PROTECTION &&
(cmd[1] & 0xe0)) {
mk_sense_invalid_opcode(scp);
return check_condition_result;
}
if ((sdebug_dif == T10_PI_TYPE1_PROTECTION ||
sdebug_dif == T10_PI_TYPE3_PROTECTION) &&
(cmd[1] & 0xe0) == 0)
sdev_printk(KERN_ERR, scp->device, "Unprotected WR "
"to DIF device\n");
ret = check_device_access_params(scp, lba, num, false);
if (ret)
return ret;
dnum = 2 * num;
treewide: kzalloc() -> kcalloc() The kzalloc() function has a 2-factor argument form, kcalloc(). This patch replaces cases of: kzalloc(a * b, gfp) with: kcalloc(a * b, gfp) as well as handling cases of: kzalloc(a * b * c, gfp) with: kzalloc(array3_size(a, b, c), gfp) as it's slightly less ugly than: kzalloc_array(array_size(a, b), c, gfp) This does, however, attempt to ignore constant size factors like: kzalloc(4 * 1024, gfp) though any constants defined via macros get caught up in the conversion. Any factors with a sizeof() of "unsigned char", "char", and "u8" were dropped, since they're redundant. The Coccinelle script used for this was: // Fix redundant parens around sizeof(). @@ type TYPE; expression THING, E; @@ ( kzalloc( - (sizeof(TYPE)) * E + sizeof(TYPE) * E , ...) | kzalloc( - (sizeof(THING)) * E + sizeof(THING) * E , ...) ) // Drop single-byte sizes and redundant parens. @@ expression COUNT; typedef u8; typedef __u8; @@ ( kzalloc( - sizeof(u8) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(__u8) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(char) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(unsigned char) * (COUNT) + COUNT , ...) | kzalloc( - sizeof(u8) * COUNT + COUNT , ...) | kzalloc( - sizeof(__u8) * COUNT + COUNT , ...) | kzalloc( - sizeof(char) * COUNT + COUNT , ...) | kzalloc( - sizeof(unsigned char) * COUNT + COUNT , ...) ) // 2-factor product with sizeof(type/expression) and identifier or constant. @@ type TYPE; expression THING; identifier COUNT_ID; constant COUNT_CONST; @@ ( - kzalloc + kcalloc ( - sizeof(TYPE) * (COUNT_ID) + COUNT_ID, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * COUNT_ID + COUNT_ID, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * (COUNT_CONST) + COUNT_CONST, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * COUNT_CONST + COUNT_CONST, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * (COUNT_ID) + COUNT_ID, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * COUNT_ID + COUNT_ID, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * (COUNT_CONST) + COUNT_CONST, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * COUNT_CONST + COUNT_CONST, sizeof(THING) , ...) ) // 2-factor product, only identifiers. @@ identifier SIZE, COUNT; @@ - kzalloc + kcalloc ( - SIZE * COUNT + COUNT, SIZE , ...) // 3-factor product with 1 sizeof(type) or sizeof(expression), with // redundant parens removed. @@ expression THING; identifier STRIDE, COUNT; type TYPE; @@ ( kzalloc( - sizeof(TYPE) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(TYPE) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(TYPE) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(TYPE) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kzalloc( - sizeof(THING) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kzalloc( - sizeof(THING) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kzalloc( - sizeof(THING) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kzalloc( - sizeof(THING) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) ) // 3-factor product with 2 sizeof(variable), with redundant parens removed. @@ expression THING1, THING2; identifier COUNT; type TYPE1, TYPE2; @@ ( kzalloc( - sizeof(TYPE1) * sizeof(TYPE2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kzalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kzalloc( - sizeof(THING1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kzalloc( - sizeof(THING1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kzalloc( - sizeof(TYPE1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) | kzalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) ) // 3-factor product, only identifiers, with redundant parens removed. @@ identifier STRIDE, SIZE, COUNT; @@ ( kzalloc( - (COUNT) * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - (COUNT) * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - (COUNT) * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - (COUNT) * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kzalloc( - COUNT * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) ) // Any remaining multi-factor products, first at least 3-factor products, // when they're not all constants... @@ expression E1, E2, E3; constant C1, C2, C3; @@ ( kzalloc(C1 * C2 * C3, ...) | kzalloc( - (E1) * E2 * E3 + array3_size(E1, E2, E3) , ...) | kzalloc( - (E1) * (E2) * E3 + array3_size(E1, E2, E3) , ...) | kzalloc( - (E1) * (E2) * (E3) + array3_size(E1, E2, E3) , ...) | kzalloc( - E1 * E2 * E3 + array3_size(E1, E2, E3) , ...) ) // And then all remaining 2 factors products when they're not all constants, // keeping sizeof() as the second factor argument. @@ expression THING, E1, E2; type TYPE; constant C1, C2, C3; @@ ( kzalloc(sizeof(THING) * C2, ...) | kzalloc(sizeof(TYPE) * C2, ...) | kzalloc(C1 * C2 * C3, ...) | kzalloc(C1 * C2, ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * (E2) + E2, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(TYPE) * E2 + E2, sizeof(TYPE) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * (E2) + E2, sizeof(THING) , ...) | - kzalloc + kcalloc ( - sizeof(THING) * E2 + E2, sizeof(THING) , ...) | - kzalloc + kcalloc ( - (E1) * E2 + E1, E2 , ...) | - kzalloc + kcalloc ( - (E1) * (E2) + E1, E2 , ...) | - kzalloc + kcalloc ( - E1 * E2 + E1, E2 , ...) ) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-12 21:03:40 +00:00
arr = kcalloc(lb_size, dnum, GFP_ATOMIC);
if (NULL == arr) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INSUFF_RES_ASC,
INSUFF_RES_ASCQ);
return check_condition_result;
}
sdeb_write_lock(sip);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
ret = do_dout_fetch(scp, dnum, arr);
if (ret == -1) {
retval = DID_ERROR << 16;
goto cleanup;
} else if (sdebug_verbose && (ret < (dnum * lb_size)))
sdev_printk(KERN_INFO, scp->device, "%s: compare_write: cdb "
"indicated=%u, IO sent=%d bytes\n", my_name,
dnum * lb_size, ret);
if (!comp_write_worker(sip, lba, num, arr, false)) {
mk_sense_buffer(scp, MISCOMPARE, MISCOMPARE_VERIFY_ASC, 0);
retval = check_condition_result;
goto cleanup;
}
if (scsi_debug_lbp())
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
map_region(sip, lba, num);
cleanup:
sdeb_write_unlock(sip);
kfree(arr);
return retval;
}
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
struct unmap_block_desc {
__be64 lba;
__be32 blocks;
__be32 __reserved;
};
static int resp_unmap(struct scsi_cmnd *scp, struct sdebug_dev_info *devip)
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
{
unsigned char *buf;
struct unmap_block_desc *desc;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, true);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
unsigned int i, payload_len, descriptors;
int ret;
if (!scsi_debug_lbp())
return 0; /* fib and say its done */
payload_len = get_unaligned_be16(scp->cmnd + 7);
BUG_ON(scsi_bufflen(scp) != payload_len);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
descriptors = (payload_len - 8) / 16;
if (descriptors > sdebug_unmap_max_desc) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 7, -1);
return check_condition_result;
}
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
buf = kzalloc(scsi_bufflen(scp), GFP_ATOMIC);
if (!buf) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INSUFF_RES_ASC,
INSUFF_RES_ASCQ);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
return check_condition_result;
}
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
scsi_sg_copy_to_buffer(scp, buf, scsi_bufflen(scp));
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
BUG_ON(get_unaligned_be16(&buf[0]) != payload_len - 2);
BUG_ON(get_unaligned_be16(&buf[2]) != descriptors * 16);
desc = (void *)&buf[8];
sdeb_write_lock(sip);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
for (i = 0 ; i < descriptors ; i++) {
unsigned long long lba = get_unaligned_be64(&desc[i].lba);
unsigned int num = get_unaligned_be32(&desc[i].blocks);
ret = check_device_access_params(scp, lba, num, true);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
if (ret)
goto out;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
unmap_region(sip, lba, num);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
}
ret = 0;
out:
sdeb_write_unlock(sip);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
kfree(buf);
return ret;
}
#define SDEBUG_GET_LBA_STATUS_LEN 32
static int resp_get_lba_status(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
{
u8 *cmd = scp->cmnd;
u64 lba;
u32 alloc_len, mapped, num;
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
int ret;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
u8 arr[SDEBUG_GET_LBA_STATUS_LEN];
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
lba = get_unaligned_be64(cmd + 2);
alloc_len = get_unaligned_be32(cmd + 10);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
if (alloc_len < 24)
return 0;
ret = check_device_access_params(scp, lba, 1, false);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
if (ret)
return ret;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
if (scsi_debug_lbp()) {
struct sdeb_store_info *sip = devip2sip(devip, true);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
mapped = map_state(sip, lba, &num);
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
} else {
mapped = 1;
/* following just in case virtual_gb changed */
sdebug_capacity = get_sdebug_capacity();
if (sdebug_capacity - lba <= 0xffffffff)
num = sdebug_capacity - lba;
else
num = 0xffffffff;
}
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
memset(arr, 0, SDEBUG_GET_LBA_STATUS_LEN);
put_unaligned_be32(20, arr); /* Parameter Data Length */
put_unaligned_be64(lba, arr + 8); /* LBA */
put_unaligned_be32(num, arr + 16); /* Number of blocks */
arr[20] = !mapped; /* prov_stat=0: mapped; 1: dealloc */
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
return fill_from_dev_buffer(scp, arr, SDEBUG_GET_LBA_STATUS_LEN);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
}
static int resp_sync_cache(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
int res = 0;
u64 lba;
u32 num_blocks;
u8 *cmd = scp->cmnd;
if (cmd[0] == SYNCHRONIZE_CACHE) { /* 10 byte cdb */
lba = get_unaligned_be32(cmd + 2);
num_blocks = get_unaligned_be16(cmd + 7);
} else { /* SYNCHRONIZE_CACHE(16) */
lba = get_unaligned_be64(cmd + 2);
num_blocks = get_unaligned_be32(cmd + 10);
}
if (lba + num_blocks > sdebug_capacity) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, LBA_OUT_OF_RANGE, 0);
return check_condition_result;
}
if (!write_since_sync || (cmd[1] & 0x2))
res = SDEG_RES_IMMED_MASK;
else /* delay if write_since_sync and IMMED clear */
write_since_sync = false;
return res;
}
/*
* Assuming the LBA+num_blocks is not out-of-range, this function will return
* CONDITION MET if the specified blocks will/have fitted in the cache, and
* a GOOD status otherwise. Model a disk with a big cache and yield
* CONDITION MET. Actually tries to bring range in main memory into the
* cache associated with the CPU(s).
*/
static int resp_pre_fetch(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
int res = 0;
u64 lba;
u64 block, rest = 0;
u32 nblks;
u8 *cmd = scp->cmnd;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, true);
u8 *fsp = sip->storep;
if (cmd[0] == PRE_FETCH) { /* 10 byte cdb */
lba = get_unaligned_be32(cmd + 2);
nblks = get_unaligned_be16(cmd + 7);
} else { /* PRE-FETCH(16) */
lba = get_unaligned_be64(cmd + 2);
nblks = get_unaligned_be32(cmd + 10);
}
if (lba + nblks > sdebug_capacity) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, LBA_OUT_OF_RANGE, 0);
return check_condition_result;
}
if (!fsp)
goto fini;
/* PRE-FETCH spec says nothing about LBP or PI so skip them */
block = do_div(lba, sdebug_store_sectors);
if (block + nblks > sdebug_store_sectors)
rest = block + nblks - sdebug_store_sectors;
/* Try to bring the PRE-FETCH range into CPU's cache */
sdeb_read_lock(sip);
prefetch_range(fsp + (sdebug_sector_size * block),
(nblks - rest) * sdebug_sector_size);
if (rest)
prefetch_range(fsp, rest * sdebug_sector_size);
sdeb_read_unlock(sip);
fini:
if (cmd[1] & 0x2)
res = SDEG_RES_IMMED_MASK;
return res | condition_met_result;
}
#define RL_BUCKET_ELEMS 8
/* Even though each pseudo target has a REPORT LUNS "well known logical unit"
* (W-LUN), the normal Linux scanning logic does not associate it with a
* device (e.g. /dev/sg7). The following magic will make that association:
* "cd /sys/class/scsi_host/host<n> ; echo '- - 49409' > scan"
* where <n> is a host number. If there are multiple targets in a host then
* the above will associate a W-LUN to each target. To only get a W-LUN
* for target 2, then use "echo '- 2 49409' > scan" .
*/
static int resp_report_luns(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
unsigned char *cmd = scp->cmnd;
unsigned int alloc_len;
unsigned char select_report;
u64 lun;
struct scsi_lun *lun_p;
u8 arr[RL_BUCKET_ELEMS * sizeof(struct scsi_lun)];
unsigned int lun_cnt; /* normal LUN count (max: 256) */
unsigned int wlun_cnt; /* report luns W-LUN count */
unsigned int tlun_cnt; /* total LUN count */
unsigned int rlen; /* response length (in bytes) */
int k, j, n, res;
unsigned int off_rsp = 0;
const int sz_lun = sizeof(struct scsi_lun);
clear_luns_changed_on_target(devip);
select_report = cmd[2];
alloc_len = get_unaligned_be32(cmd + 6);
if (alloc_len < 4) {
pr_err("alloc len too small %d\n", alloc_len);
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 6, -1);
return check_condition_result;
}
switch (select_report) {
case 0: /* all LUNs apart from W-LUNs */
lun_cnt = sdebug_max_luns;
wlun_cnt = 0;
break;
case 1: /* only W-LUNs */
lun_cnt = 0;
wlun_cnt = 1;
break;
case 2: /* all LUNs */
lun_cnt = sdebug_max_luns;
wlun_cnt = 1;
break;
case 0x10: /* only administrative LUs */
case 0x11: /* see SPC-5 */
case 0x12: /* only subsiduary LUs owned by referenced LU */
default:
pr_debug("select report invalid %d\n", select_report);
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 2, -1);
return check_condition_result;
}
if (sdebug_no_lun_0 && (lun_cnt > 0))
--lun_cnt;
tlun_cnt = lun_cnt + wlun_cnt;
rlen = tlun_cnt * sz_lun; /* excluding 8 byte header */
scsi_set_resid(scp, scsi_bufflen(scp));
pr_debug("select_report %d luns = %d wluns = %d no_lun0 %d\n",
select_report, lun_cnt, wlun_cnt, sdebug_no_lun_0);
/* loops rely on sizeof response header same as sizeof lun (both 8) */
lun = sdebug_no_lun_0 ? 1 : 0;
for (k = 0, j = 0, res = 0; true; ++k, j = 0) {
memset(arr, 0, sizeof(arr));
lun_p = (struct scsi_lun *)&arr[0];
if (k == 0) {
put_unaligned_be32(rlen, &arr[0]);
++lun_p;
j = 1;
}
for ( ; j < RL_BUCKET_ELEMS; ++j, ++lun_p) {
if ((k * RL_BUCKET_ELEMS) + j > lun_cnt)
break;
int_to_scsilun(lun++, lun_p);
if (lun > 1 && sdebug_lun_am == SAM_LUN_AM_FLAT)
lun_p->scsi_lun[0] |= 0x40;
}
if (j < RL_BUCKET_ELEMS)
break;
n = j * sz_lun;
res = p_fill_from_dev_buffer(scp, arr, n, off_rsp);
if (res)
return res;
off_rsp += n;
}
if (wlun_cnt) {
int_to_scsilun(SCSI_W_LUN_REPORT_LUNS, lun_p);
++j;
}
if (j > 0)
res = p_fill_from_dev_buffer(scp, arr, j * sz_lun, off_rsp);
return res;
}
static int resp_verify(struct scsi_cmnd *scp, struct sdebug_dev_info *devip)
{
bool is_bytchk3 = false;
u8 bytchk;
int ret, j;
u32 vnum, a_num, off;
const u32 lb_size = sdebug_sector_size;
u64 lba;
u8 *arr;
u8 *cmd = scp->cmnd;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, true);
bytchk = (cmd[1] >> 1) & 0x3;
if (bytchk == 0) {
return 0; /* always claim internal verify okay */
} else if (bytchk == 2) {
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 2, 2);
return check_condition_result;
} else if (bytchk == 3) {
is_bytchk3 = true; /* 1 block sent, compared repeatedly */
}
switch (cmd[0]) {
case VERIFY_16:
lba = get_unaligned_be64(cmd + 2);
vnum = get_unaligned_be32(cmd + 10);
break;
case VERIFY: /* is VERIFY(10) */
lba = get_unaligned_be32(cmd + 2);
vnum = get_unaligned_be16(cmd + 7);
break;
default:
mk_sense_invalid_opcode(scp);
return check_condition_result;
}
scsi: scsi_debug: Don't call kcalloc() if size arg is zero If the size arg to kcalloc() is zero, it returns ZERO_SIZE_PTR. Because of that, for a following NULL pointer check to work on the returned pointer, kcalloc() must not be called with the size arg equal to zero. Return early without error before the kcalloc() call if size arg is zero. BUG: KASAN: null-ptr-deref in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: null-ptr-deref in sg_copy_buffer+0x138/0x240 lib/scatterlist.c:974 Write of size 4 at addr 0000000000000010 by task syz-executor.1/22789 CPU: 1 PID: 22789 Comm: syz-executor.1 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 __kasan_report mm/kasan/report.c:446 [inline] kasan_report.cold.14+0x112/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x3b/0x60 mm/kasan/shadow.c:66 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x138/0x240 lib/scatterlist.c:974 do_dout_fetch drivers/scsi/scsi_debug.c:2954 [inline] do_dout_fetch drivers/scsi/scsi_debug.c:2946 [inline] resp_verify+0x49e/0x930 drivers/scsi/scsi_debug.c:4276 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 blk_execute_rq+0xdb/0x360 block/blk-exec.c:102 sg_scsi_ioctl drivers/scsi/scsi_ioctl.c:621 [inline] scsi_ioctl+0x8bb/0x15c0 drivers/scsi/scsi_ioctl.c:930 sg_ioctl_common+0x172d/0x2710 drivers/scsi/sg.c:1112 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636056397-13151-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-04 20:06:37 +00:00
if (vnum == 0)
return 0; /* not an error */
a_num = is_bytchk3 ? 1 : vnum;
/* Treat following check like one for read (i.e. no write) access */
ret = check_device_access_params(scp, lba, a_num, false);
if (ret)
return ret;
arr = kcalloc(lb_size, vnum, GFP_ATOMIC | __GFP_NOWARN);
if (!arr) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INSUFF_RES_ASC,
INSUFF_RES_ASCQ);
return check_condition_result;
}
/* Not changing store, so only need read access */
sdeb_read_lock(sip);
ret = do_dout_fetch(scp, a_num, arr);
if (ret == -1) {
ret = DID_ERROR << 16;
goto cleanup;
} else if (sdebug_verbose && (ret < (a_num * lb_size))) {
sdev_printk(KERN_INFO, scp->device,
"%s: %s: cdb indicated=%u, IO sent=%d bytes\n",
my_name, __func__, a_num * lb_size, ret);
}
if (is_bytchk3) {
for (j = 1, off = lb_size; j < vnum; ++j, off += lb_size)
memcpy(arr + off, arr, lb_size);
}
ret = 0;
if (!comp_write_worker(sip, lba, vnum, arr, true)) {
mk_sense_buffer(scp, MISCOMPARE, MISCOMPARE_VERIFY_ASC, 0);
ret = check_condition_result;
goto cleanup;
}
cleanup:
sdeb_read_unlock(sip);
kfree(arr);
return ret;
}
#define RZONES_DESC_HD 64
/* Report zones depending on start LBA and reporting options */
static int resp_report_zones(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
unsigned int rep_max_zones, nrz = 0;
int ret = 0;
u32 alloc_len, rep_opts, rep_len;
bool partial;
u64 lba, zs_lba;
u8 *arr = NULL, *desc;
u8 *cmd = scp->cmnd;
struct sdeb_zone_state *zsp = NULL;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, false);
if (!sdebug_dev_is_zoned(devip)) {
mk_sense_invalid_opcode(scp);
return check_condition_result;
}
zs_lba = get_unaligned_be64(cmd + 2);
alloc_len = get_unaligned_be32(cmd + 10);
scsi: scsi_debug: Don't call kcalloc() if size arg is zero If the size arg to kcalloc() is zero, it returns ZERO_SIZE_PTR. Because of that, for a following NULL pointer check to work on the returned pointer, kcalloc() must not be called with the size arg equal to zero. Return early without error before the kcalloc() call if size arg is zero. BUG: KASAN: null-ptr-deref in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: null-ptr-deref in sg_copy_buffer+0x138/0x240 lib/scatterlist.c:974 Write of size 4 at addr 0000000000000010 by task syz-executor.1/22789 CPU: 1 PID: 22789 Comm: syz-executor.1 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 __kasan_report mm/kasan/report.c:446 [inline] kasan_report.cold.14+0x112/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x3b/0x60 mm/kasan/shadow.c:66 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x138/0x240 lib/scatterlist.c:974 do_dout_fetch drivers/scsi/scsi_debug.c:2954 [inline] do_dout_fetch drivers/scsi/scsi_debug.c:2946 [inline] resp_verify+0x49e/0x930 drivers/scsi/scsi_debug.c:4276 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 blk_execute_rq+0xdb/0x360 block/blk-exec.c:102 sg_scsi_ioctl drivers/scsi/scsi_ioctl.c:621 [inline] scsi_ioctl+0x8bb/0x15c0 drivers/scsi/scsi_ioctl.c:930 sg_ioctl_common+0x172d/0x2710 drivers/scsi/sg.c:1112 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636056397-13151-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-04 20:06:37 +00:00
if (alloc_len == 0)
return 0; /* not an error */
rep_opts = cmd[14] & 0x3f;
partial = cmd[14] & 0x80;
if (zs_lba >= sdebug_capacity) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, LBA_OUT_OF_RANGE, 0);
return check_condition_result;
}
rep_max_zones = (alloc_len - 64) >> ilog2(RZONES_DESC_HD);
arr = kzalloc(alloc_len, GFP_ATOMIC | __GFP_NOWARN);
if (!arr) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INSUFF_RES_ASC,
INSUFF_RES_ASCQ);
return check_condition_result;
}
sdeb_read_lock(sip);
desc = arr + 64;
for (lba = zs_lba; lba < sdebug_capacity;
lba = zsp->z_start + zsp->z_size) {
if (WARN_ONCE(zbc_zone(devip, lba) == zsp, "lba = %llu\n", lba))
break;
zsp = zbc_zone(devip, lba);
switch (rep_opts) {
case 0x00:
/* All zones */
break;
case 0x01:
/* Empty zones */
if (zsp->z_cond != ZC1_EMPTY)
continue;
break;
case 0x02:
/* Implicit open zones */
if (zsp->z_cond != ZC2_IMPLICIT_OPEN)
continue;
break;
case 0x03:
/* Explicit open zones */
if (zsp->z_cond != ZC3_EXPLICIT_OPEN)
continue;
break;
case 0x04:
/* Closed zones */
if (zsp->z_cond != ZC4_CLOSED)
continue;
break;
case 0x05:
/* Full zones */
if (zsp->z_cond != ZC5_FULL)
continue;
break;
case 0x06:
case 0x07:
case 0x10:
/*
* Read-only, offline, reset WP recommended are
* not emulated: no zones to report;
*/
continue;
case 0x11:
/* non-seq-resource set */
if (!zsp->z_non_seq_resource)
continue;
break;
case 0x3e:
/* All zones except gap zones. */
if (zbc_zone_is_gap(zsp))
continue;
break;
case 0x3f:
/* Not write pointer (conventional) zones */
if (zbc_zone_is_seq(zsp))
continue;
break;
default:
mk_sense_buffer(scp, ILLEGAL_REQUEST,
INVALID_FIELD_IN_CDB, 0);
ret = check_condition_result;
goto fini;
}
if (nrz < rep_max_zones) {
/* Fill zone descriptor */
desc[0] = zsp->z_type;
desc[1] = zsp->z_cond << 4;
if (zsp->z_non_seq_resource)
desc[1] |= 1 << 1;
put_unaligned_be64((u64)zsp->z_size, desc + 8);
put_unaligned_be64((u64)zsp->z_start, desc + 16);
put_unaligned_be64((u64)zsp->z_wp, desc + 24);
desc += 64;
}
if (partial && nrz >= rep_max_zones)
break;
nrz++;
}
/* Report header */
/* Zone list length. */
put_unaligned_be32(nrz * RZONES_DESC_HD, arr + 0);
/* Maximum LBA */
put_unaligned_be64(sdebug_capacity - 1, arr + 8);
/* Zone starting LBA granularity. */
if (devip->zcap < devip->zsize)
put_unaligned_be64(devip->zsize, arr + 16);
rep_len = (unsigned long)desc - (unsigned long)arr;
scsi: scsi_debug: Fix type in min_t to avoid stack OOB Change min_t() to use type "u32" instead of type "int" to avoid stack out of bounds. With min_t() type "int" the values get sign extended and the larger value gets used causing stack out of bounds. BUG: KASAN: stack-out-of-bounds in memcpy include/linux/fortify-string.h:191 [inline] BUG: KASAN: stack-out-of-bounds in sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 Read of size 127 at addr ffff888072607128 by task syz-executor.7/18707 CPU: 1 PID: 18707 Comm: syz-executor.7 Not tainted 5.15.0-syzk #1 Hardware name: Red Hat KVM, BIOS 1.13.0-2 Call Trace: __dump_stack lib/dump_stack.c:88 [inline] dump_stack_lvl+0x89/0xb5 lib/dump_stack.c:106 print_address_description.constprop.9+0x28/0x160 mm/kasan/report.c:256 __kasan_report mm/kasan/report.c:442 [inline] kasan_report.cold.14+0x7d/0x117 mm/kasan/report.c:459 check_region_inline mm/kasan/generic.c:183 [inline] kasan_check_range+0x1a3/0x210 mm/kasan/generic.c:189 memcpy+0x23/0x60 mm/kasan/shadow.c:65 memcpy include/linux/fortify-string.h:191 [inline] sg_copy_buffer+0x1de/0x240 lib/scatterlist.c:976 sg_copy_from_buffer+0x33/0x40 lib/scatterlist.c:1000 fill_from_dev_buffer.part.34+0x82/0x130 drivers/scsi/scsi_debug.c:1162 fill_from_dev_buffer drivers/scsi/scsi_debug.c:1888 [inline] resp_readcap16+0x365/0x3b0 drivers/scsi/scsi_debug.c:1887 schedule_resp+0x4d8/0x1a70 drivers/scsi/scsi_debug.c:5478 scsi_debug_queuecommand+0x8c9/0x1ec0 drivers/scsi/scsi_debug.c:7533 scsi_dispatch_cmd drivers/scsi/scsi_lib.c:1520 [inline] scsi_queue_rq+0x16b0/0x2d40 drivers/scsi/scsi_lib.c:1699 blk_mq_dispatch_rq_list+0xb9b/0x2700 block/blk-mq.c:1639 __blk_mq_sched_dispatch_requests+0x28f/0x590 block/blk-mq-sched.c:325 blk_mq_sched_dispatch_requests+0x105/0x190 block/blk-mq-sched.c:358 __blk_mq_run_hw_queue+0xe5/0x150 block/blk-mq.c:1761 __blk_mq_delay_run_hw_queue+0x4f8/0x5c0 block/blk-mq.c:1838 blk_mq_run_hw_queue+0x18d/0x350 block/blk-mq.c:1891 blk_mq_sched_insert_request+0x3db/0x4e0 block/blk-mq-sched.c:474 blk_execute_rq_nowait+0x16b/0x1c0 block/blk-exec.c:62 sg_common_write.isra.18+0xeb3/0x2000 drivers/scsi/sg.c:836 sg_new_write.isra.19+0x570/0x8c0 drivers/scsi/sg.c:774 sg_ioctl_common+0x14d6/0x2710 drivers/scsi/sg.c:939 sg_ioctl+0xa2/0x180 drivers/scsi/sg.c:1165 vfs_ioctl fs/ioctl.c:51 [inline] __do_sys_ioctl fs/ioctl.c:874 [inline] __se_sys_ioctl fs/ioctl.c:860 [inline] __x64_sys_ioctl+0x19d/0x220 fs/ioctl.c:860 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3a/0x80 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/1636484247-21254-1-git-send-email-george.kennedy@oracle.com Reported-by: syzkaller <syzkaller@googlegroups.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: George Kennedy <george.kennedy@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2021-11-09 18:57:27 +00:00
ret = fill_from_dev_buffer(scp, arr, min_t(u32, alloc_len, rep_len));
fini:
sdeb_read_unlock(sip);
kfree(arr);
return ret;
}
/* Logic transplanted from tcmu-runner, file_zbc.c */
static void zbc_open_all(struct sdebug_dev_info *devip)
{
struct sdeb_zone_state *zsp = &devip->zstate[0];
unsigned int i;
for (i = 0; i < devip->nr_zones; i++, zsp++) {
if (zsp->z_cond == ZC4_CLOSED)
zbc_open_zone(devip, &devip->zstate[i], true);
}
}
static int resp_open_zone(struct scsi_cmnd *scp, struct sdebug_dev_info *devip)
{
int res = 0;
u64 z_id;
enum sdebug_z_cond zc;
u8 *cmd = scp->cmnd;
struct sdeb_zone_state *zsp;
bool all = cmd[14] & 0x01;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, false);
if (!sdebug_dev_is_zoned(devip)) {
mk_sense_invalid_opcode(scp);
return check_condition_result;
}
sdeb_write_lock(sip);
if (all) {
/* Check if all closed zones can be open */
if (devip->max_open &&
devip->nr_exp_open + devip->nr_closed > devip->max_open) {
mk_sense_buffer(scp, DATA_PROTECT, INSUFF_RES_ASC,
INSUFF_ZONE_ASCQ);
res = check_condition_result;
goto fini;
}
/* Open all closed zones */
zbc_open_all(devip);
goto fini;
}
/* Open the specified zone */
z_id = get_unaligned_be64(cmd + 2);
if (z_id >= sdebug_capacity) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, LBA_OUT_OF_RANGE, 0);
res = check_condition_result;
goto fini;
}
zsp = zbc_zone(devip, z_id);
if (z_id != zsp->z_start) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_FIELD_IN_CDB, 0);
res = check_condition_result;
goto fini;
}
if (zbc_zone_is_conv(zsp)) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_FIELD_IN_CDB, 0);
res = check_condition_result;
goto fini;
}
zc = zsp->z_cond;
if (zc == ZC3_EXPLICIT_OPEN || zc == ZC5_FULL)
goto fini;
if (devip->max_open && devip->nr_exp_open >= devip->max_open) {
mk_sense_buffer(scp, DATA_PROTECT, INSUFF_RES_ASC,
INSUFF_ZONE_ASCQ);
res = check_condition_result;
goto fini;
}
zbc_open_zone(devip, zsp, true);
fini:
sdeb_write_unlock(sip);
return res;
}
static void zbc_close_all(struct sdebug_dev_info *devip)
{
unsigned int i;
for (i = 0; i < devip->nr_zones; i++)
zbc_close_zone(devip, &devip->zstate[i]);
}
static int resp_close_zone(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
int res = 0;
u64 z_id;
u8 *cmd = scp->cmnd;
struct sdeb_zone_state *zsp;
bool all = cmd[14] & 0x01;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, false);
if (!sdebug_dev_is_zoned(devip)) {
mk_sense_invalid_opcode(scp);
return check_condition_result;
}
sdeb_write_lock(sip);
if (all) {
zbc_close_all(devip);
goto fini;
}
/* Close specified zone */
z_id = get_unaligned_be64(cmd + 2);
if (z_id >= sdebug_capacity) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, LBA_OUT_OF_RANGE, 0);
res = check_condition_result;
goto fini;
}
zsp = zbc_zone(devip, z_id);
if (z_id != zsp->z_start) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_FIELD_IN_CDB, 0);
res = check_condition_result;
goto fini;
}
if (zbc_zone_is_conv(zsp)) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_FIELD_IN_CDB, 0);
res = check_condition_result;
goto fini;
}
zbc_close_zone(devip, zsp);
fini:
sdeb_write_unlock(sip);
return res;
}
static void zbc_finish_zone(struct sdebug_dev_info *devip,
struct sdeb_zone_state *zsp, bool empty)
{
enum sdebug_z_cond zc = zsp->z_cond;
if (zc == ZC4_CLOSED || zc == ZC2_IMPLICIT_OPEN ||
zc == ZC3_EXPLICIT_OPEN || (empty && zc == ZC1_EMPTY)) {
if (zc == ZC2_IMPLICIT_OPEN || zc == ZC3_EXPLICIT_OPEN)
zbc_close_zone(devip, zsp);
if (zsp->z_cond == ZC4_CLOSED)
devip->nr_closed--;
zsp->z_wp = zsp->z_start + zsp->z_size;
zsp->z_cond = ZC5_FULL;
}
}
static void zbc_finish_all(struct sdebug_dev_info *devip)
{
unsigned int i;
for (i = 0; i < devip->nr_zones; i++)
zbc_finish_zone(devip, &devip->zstate[i], false);
}
static int resp_finish_zone(struct scsi_cmnd *scp,
struct sdebug_dev_info *devip)
{
struct sdeb_zone_state *zsp;
int res = 0;
u64 z_id;
u8 *cmd = scp->cmnd;
bool all = cmd[14] & 0x01;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, false);
if (!sdebug_dev_is_zoned(devip)) {
mk_sense_invalid_opcode(scp);
return check_condition_result;
}
sdeb_write_lock(sip);
if (all) {
zbc_finish_all(devip);
goto fini;
}
/* Finish the specified zone */
z_id = get_unaligned_be64(cmd + 2);
if (z_id >= sdebug_capacity) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, LBA_OUT_OF_RANGE, 0);
res = check_condition_result;
goto fini;
}
zsp = zbc_zone(devip, z_id);
if (z_id != zsp->z_start) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_FIELD_IN_CDB, 0);
res = check_condition_result;
goto fini;
}
if (zbc_zone_is_conv(zsp)) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_FIELD_IN_CDB, 0);
res = check_condition_result;
goto fini;
}
zbc_finish_zone(devip, zsp, true);
fini:
sdeb_write_unlock(sip);
return res;
}
static void zbc_rwp_zone(struct sdebug_dev_info *devip,
struct sdeb_zone_state *zsp)
{
enum sdebug_z_cond zc;
struct sdeb_store_info *sip = devip2sip(devip, false);
if (!zbc_zone_is_seq(zsp))
return;
zc = zsp->z_cond;
if (zc == ZC2_IMPLICIT_OPEN || zc == ZC3_EXPLICIT_OPEN)
zbc_close_zone(devip, zsp);
if (zsp->z_cond == ZC4_CLOSED)
devip->nr_closed--;
if (zsp->z_wp > zsp->z_start)
memset(sip->storep + zsp->z_start * sdebug_sector_size, 0,
(zsp->z_wp - zsp->z_start) * sdebug_sector_size);
zsp->z_non_seq_resource = false;
zsp->z_wp = zsp->z_start;
zsp->z_cond = ZC1_EMPTY;
}
static void zbc_rwp_all(struct sdebug_dev_info *devip)
{
unsigned int i;
for (i = 0; i < devip->nr_zones; i++)
zbc_rwp_zone(devip, &devip->zstate[i]);
}
static int resp_rwp_zone(struct scsi_cmnd *scp, struct sdebug_dev_info *devip)
{
struct sdeb_zone_state *zsp;
int res = 0;
u64 z_id;
u8 *cmd = scp->cmnd;
bool all = cmd[14] & 0x01;
scsi: scsi_debug: Parser tables and code interaction This patch is in response to a static analyser report from Dan Carpenter titled: "[bug report] scsi: scsi_debug: Add per_host_store option". This code may not clear the static analyzer reports, but may shed light on why they occur. Amongst other things this driver has a table driven SCSI command parser which also involves some C code. There are some invariants between the table entries and the corresponding C code (i.e. the resp_*() functions) that, if broken, may lead to a NULL dereference. And the report is valid, at least in the case of the PRE-FETCH command. Alas, that is not one of the cases that the static analyzer reported. In this particular corner case: when the fake_rw flag is set and the table entry for a "store"-accessing command does not have the required F_FAKE_RW flag set, do the following. Call BUG_ON() in the devip2sip() very close to a comment block explaining why it was called and how to fix it. checkpatch.pl complains about the BUG_ON() but there is no reasonable remedial action that can be taken at run time. This change allows the code reported by the static analyzer to be simplified. Comments were also added to the table flags (e.g. F_FAKE_RW) so developers who add commands might be more inclined to use them (properly). Link: https://lore.kernel.org/r/20200513013943.25285-1-dgilbert@interlog.com Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-05-13 01:39:43 +00:00
struct sdeb_store_info *sip = devip2sip(devip, false);
if (!sdebug_dev_is_zoned(devip)) {
mk_sense_invalid_opcode(scp);
return check_condition_result;
}
sdeb_write_lock(sip);
if (all) {
zbc_rwp_all(devip);
goto fini;
}
z_id = get_unaligned_be64(cmd + 2);
if (z_id >= sdebug_capacity) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, LBA_OUT_OF_RANGE, 0);
res = check_condition_result;
goto fini;
}
zsp = zbc_zone(devip, z_id);
if (z_id != zsp->z_start) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_FIELD_IN_CDB, 0);
res = check_condition_result;
goto fini;
}
if (zbc_zone_is_conv(zsp)) {
mk_sense_buffer(scp, ILLEGAL_REQUEST, INVALID_FIELD_IN_CDB, 0);
res = check_condition_result;
goto fini;
}
zbc_rwp_zone(devip, zsp);
fini:
sdeb_write_unlock(sip);
return res;
}
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
static u32 get_tag(struct scsi_cmnd *cmnd)
{
return blk_mq_unique_tag(scsi_cmd_to_rq(cmnd));
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
}
/* Queued (deferred) command completions converge here. */
static void sdebug_q_cmd_complete(struct sdebug_defer *sd_dp)
{
struct sdebug_queued_cmd *sqcp = container_of(sd_dp, struct sdebug_queued_cmd, sd_dp);
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
unsigned long flags;
struct scsi_cmnd *scp = sqcp->scmd;
struct sdebug_scsi_cmd *sdsc;
bool aborted;
if (sdebug_statistics) {
atomic_inc(&sdebug_completions);
if (raw_smp_processor_id() != sd_dp->issuing_cpu)
atomic_inc(&sdebug_miss_cpus);
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
if (!scp) {
pr_err("scmd=NULL\n");
goto out;
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
sdsc = scsi_cmd_priv(scp);
spin_lock_irqsave(&sdsc->lock, flags);
aborted = sd_dp->aborted;
if (unlikely(aborted))
sd_dp->aborted = false;
ASSIGN_QUEUED_CMD(scp, NULL);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
spin_unlock_irqrestore(&sdsc->lock, flags);
if (aborted) {
pr_info("bypassing scsi_done() due to aborted cmd, kicking-off EH\n");
blk_abort_request(scsi_cmd_to_rq(scp));
goto out;
}
scsi_done(scp); /* callback to mid level */
out:
sdebug_free_queued_cmd(sqcp);
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
/* When high resolution timer goes off this function is called. */
static enum hrtimer_restart sdebug_q_cmd_hrt_complete(struct hrtimer *timer)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
{
struct sdebug_defer *sd_dp = container_of(timer, struct sdebug_defer,
hrt);
sdebug_q_cmd_complete(sd_dp);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
return HRTIMER_NORESTART;
}
/* When work queue schedules work, it calls this function. */
static void sdebug_q_cmd_wq_complete(struct work_struct *work)
{
struct sdebug_defer *sd_dp = container_of(work, struct sdebug_defer,
ew.work);
sdebug_q_cmd_complete(sd_dp);
}
static bool got_shared_uuid;
static uuid_t shared_uuid;
static int sdebug_device_create_zones(struct sdebug_dev_info *devip)
{
struct sdeb_zone_state *zsp;
sector_t capacity = get_sdebug_capacity();
sector_t conv_capacity;
sector_t zstart = 0;
unsigned int i;
/*
* Set the zone size: if sdeb_zbc_zone_size_mb is not set, figure out
* a zone size allowing for at least 4 zones on the device. Otherwise,
* use the specified zone size checking that at least 2 zones can be
* created for the device.
*/
if (!sdeb_zbc_zone_size_mb) {
devip->zsize = (DEF_ZBC_ZONE_SIZE_MB * SZ_1M)
>> ilog2(sdebug_sector_size);
while (capacity < devip->zsize << 2 && devip->zsize >= 2)
devip->zsize >>= 1;
if (devip->zsize < 2) {
pr_err("Device capacity too small\n");
return -EINVAL;
}
} else {
if (!is_power_of_2(sdeb_zbc_zone_size_mb)) {
pr_err("Zone size is not a power of 2\n");
return -EINVAL;
}
devip->zsize = (sdeb_zbc_zone_size_mb * SZ_1M)
>> ilog2(sdebug_sector_size);
if (devip->zsize >= capacity) {
pr_err("Zone size too large for device capacity\n");
return -EINVAL;
}
}
devip->zsize_shift = ilog2(devip->zsize);
devip->nr_zones = (capacity + devip->zsize - 1) >> devip->zsize_shift;
if (sdeb_zbc_zone_cap_mb == 0) {
devip->zcap = devip->zsize;
} else {
devip->zcap = (sdeb_zbc_zone_cap_mb * SZ_1M) >>
ilog2(sdebug_sector_size);
if (devip->zcap > devip->zsize) {
pr_err("Zone capacity too large\n");
return -EINVAL;
}
}
conv_capacity = (sector_t)sdeb_zbc_nr_conv << devip->zsize_shift;
if (conv_capacity >= capacity) {
pr_err("Number of conventional zones too large\n");
return -EINVAL;
}
devip->nr_conv_zones = sdeb_zbc_nr_conv;
devip->nr_seq_zones = ALIGN(capacity - conv_capacity, devip->zsize) >>
devip->zsize_shift;
devip->nr_zones = devip->nr_conv_zones + devip->nr_seq_zones;
/* Add gap zones if zone capacity is smaller than the zone size */
if (devip->zcap < devip->zsize)
devip->nr_zones += devip->nr_seq_zones;
if (devip->zmodel == BLK_ZONED_HM) {
/* zbc_max_open_zones can be 0, meaning "not reported" */
if (sdeb_zbc_max_open >= devip->nr_zones - 1)
devip->max_open = (devip->nr_zones - 1) / 2;
else
devip->max_open = sdeb_zbc_max_open;
}
devip->zstate = kcalloc(devip->nr_zones,
sizeof(struct sdeb_zone_state), GFP_KERNEL);
if (!devip->zstate)
return -ENOMEM;
for (i = 0; i < devip->nr_zones; i++) {
zsp = &devip->zstate[i];
zsp->z_start = zstart;
if (i < devip->nr_conv_zones) {
zsp->z_type = ZBC_ZTYPE_CNV;
zsp->z_cond = ZBC_NOT_WRITE_POINTER;
zsp->z_wp = (sector_t)-1;
zsp->z_size =
min_t(u64, devip->zsize, capacity - zstart);
} else if ((zstart & (devip->zsize - 1)) == 0) {
if (devip->zmodel == BLK_ZONED_HM)
zsp->z_type = ZBC_ZTYPE_SWR;
else
zsp->z_type = ZBC_ZTYPE_SWP;
zsp->z_cond = ZC1_EMPTY;
zsp->z_wp = zsp->z_start;
zsp->z_size =
min_t(u64, devip->zcap, capacity - zstart);
} else {
zsp->z_type = ZBC_ZTYPE_GAP;
zsp->z_cond = ZBC_NOT_WRITE_POINTER;
zsp->z_wp = (sector_t)-1;
zsp->z_size = min_t(u64, devip->zsize - devip->zcap,
capacity - zstart);
}
WARN_ON_ONCE((int)zsp->z_size <= 0);
zstart += zsp->z_size;
}
return 0;
}
static struct sdebug_dev_info *sdebug_device_create(
struct sdebug_host_info *sdbg_host, gfp_t flags)
{
struct sdebug_dev_info *devip;
devip = kzalloc(sizeof(*devip), flags);
if (devip) {
if (sdebug_uuid_ctl == 1)
uuid_gen(&devip->lu_name);
else if (sdebug_uuid_ctl == 2) {
if (got_shared_uuid)
devip->lu_name = shared_uuid;
else {
uuid_gen(&shared_uuid);
got_shared_uuid = true;
devip->lu_name = shared_uuid;
}
}
devip->sdbg_host = sdbg_host;
if (sdeb_zbc_in_use) {
devip->zmodel = sdeb_zbc_model;
if (sdebug_device_create_zones(devip)) {
kfree(devip);
return NULL;
}
} else {
devip->zmodel = BLK_ZONED_NONE;
}
devip->create_ts = ktime_get_boottime();
atomic_set(&devip->stopped, (sdeb_tur_ms_to_ready > 0 ? 2 : 0));
list_add_tail(&devip->dev_list, &sdbg_host->dev_info_list);
}
return devip;
}
static struct sdebug_dev_info *find_build_dev_info(struct scsi_device *sdev)
{
struct sdebug_host_info *sdbg_host;
struct sdebug_dev_info *open_devip = NULL;
struct sdebug_dev_info *devip;
sdbg_host = shost_to_sdebug_host(sdev->host);
list_for_each_entry(devip, &sdbg_host->dev_info_list, dev_list) {
if ((devip->used) && (devip->channel == sdev->channel) &&
(devip->target == sdev->id) &&
(devip->lun == sdev->lun))
return devip;
else {
if ((!devip->used) && (!open_devip))
open_devip = devip;
}
}
if (!open_devip) { /* try and make a new one */
open_devip = sdebug_device_create(sdbg_host, GFP_ATOMIC);
if (!open_devip) {
pr_err("out of memory at line %d\n", __LINE__);
return NULL;
}
}
open_devip->channel = sdev->channel;
open_devip->target = sdev->id;
open_devip->lun = sdev->lun;
open_devip->sdbg_host = sdbg_host;
set_bit(SDEBUG_UA_POOCCUR, open_devip->uas_bm);
open_devip->used = true;
return open_devip;
}
static int scsi_debug_slave_alloc(struct scsi_device *sdp)
{
if (sdebug_verbose)
pr_info("slave_alloc <%u %u %u %llu>\n",
sdp->host->host_no, sdp->channel, sdp->id, sdp->lun);
return 0;
}
static int scsi_debug_slave_configure(struct scsi_device *sdp)
{
struct sdebug_dev_info *devip =
(struct sdebug_dev_info *)sdp->hostdata;
if (sdebug_verbose)
pr_info("slave_configure <%u %u %u %llu>\n",
sdp->host->host_no, sdp->channel, sdp->id, sdp->lun);
if (sdp->host->max_cmd_len != SDEBUG_MAX_CMD_LEN)
sdp->host->max_cmd_len = SDEBUG_MAX_CMD_LEN;
if (devip == NULL) {
devip = find_build_dev_info(sdp);
if (devip == NULL)
return 1; /* no resources, will be marked offline */
}
sdp->hostdata = devip;
if (sdebug_no_uld)
sdp->no_uld_attach = 1;
config_cdb_len(sdp);
return 0;
}
static void scsi_debug_slave_destroy(struct scsi_device *sdp)
{
struct sdebug_dev_info *devip =
(struct sdebug_dev_info *)sdp->hostdata;
if (sdebug_verbose)
pr_info("slave_destroy <%u %u %u %llu>\n",
sdp->host->host_no, sdp->channel, sdp->id, sdp->lun);
if (devip) {
/* make this slot available for re-use */
devip->used = false;
sdp->hostdata = NULL;
}
}
/* Returns true if we require the queued memory to be freed by the caller. */
static bool stop_qc_helper(struct sdebug_defer *sd_dp,
enum sdeb_defer_type defer_t)
{
if (defer_t == SDEB_DEFER_HRT) {
int res = hrtimer_try_to_cancel(&sd_dp->hrt);
switch (res) {
case 0: /* Not active, it must have already run */
case -1: /* -1 It's executing the CB */
return false;
case 1: /* Was active, we've now cancelled */
default:
return true;
}
} else if (defer_t == SDEB_DEFER_WQ) {
/* Cancel if pending */
if (cancel_work_sync(&sd_dp->ew.work))
return true;
/* Was not pending, so it must have run */
return false;
} else if (defer_t == SDEB_DEFER_POLL) {
return true;
}
return false;
}
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
static bool scsi_debug_stop_cmnd(struct scsi_cmnd *cmnd)
{
enum sdeb_defer_type l_defer_t;
struct sdebug_defer *sd_dp;
struct sdebug_scsi_cmd *sdsc = scsi_cmd_priv(cmnd);
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
struct sdebug_queued_cmd *sqcp = TO_QUEUED_CMD(cmnd);
lockdep_assert_held(&sdsc->lock);
if (!sqcp)
return false;
sd_dp = &sqcp->sd_dp;
l_defer_t = READ_ONCE(sd_dp->defer_t);
ASSIGN_QUEUED_CMD(cmnd, NULL);
if (stop_qc_helper(sd_dp, l_defer_t))
sdebug_free_queued_cmd(sqcp);
return true;
}
/*
* Called from scsi_debug_abort() only, which is for timed-out cmd.
*/
static bool scsi_debug_abort_cmnd(struct scsi_cmnd *cmnd)
{
struct sdebug_scsi_cmd *sdsc = scsi_cmd_priv(cmnd);
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
unsigned long flags;
bool res;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
spin_lock_irqsave(&sdsc->lock, flags);
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
res = scsi_debug_stop_cmnd(cmnd);
spin_unlock_irqrestore(&sdsc->lock, flags);
return res;
}
/*
* All we can do is set the cmnd as internally aborted and wait for it to
* finish. We cannot call scsi_done() as normal completion path may do that.
*/
static bool sdebug_stop_cmnd(struct request *rq, void *data)
{
scsi_debug_abort_cmnd(blk_mq_rq_to_pdu(rq));
return true;
}
/* Deletes (stops) timers or work queues of all queued commands */
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
static void stop_all_queued(void)
{
struct sdebug_host_info *sdhp;
mutex_lock(&sdebug_host_list_mutex);
list_for_each_entry(sdhp, &sdebug_host_list, host_list) {
struct Scsi_Host *shost = sdhp->shost;
blk_mq_tagset_busy_iter(&shost->tag_set, sdebug_stop_cmnd, NULL);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
}
mutex_unlock(&sdebug_host_list_mutex);
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
static int scsi_debug_abort(struct scsi_cmnd *SCpnt)
{
bool ok = scsi_debug_abort_cmnd(SCpnt);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
++num_aborts;
if (SDEBUG_OPT_ALL_NOISE & sdebug_opts)
sdev_printk(KERN_INFO, SCpnt->device,
"%s: command%s found\n", __func__,
ok ? "" : " not");
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
return SUCCESS;
}
static bool scsi_debug_stop_all_queued_iter(struct request *rq, void *data)
{
struct scsi_device *sdp = data;
struct scsi_cmnd *scmd = blk_mq_rq_to_pdu(rq);
if (scmd->device == sdp)
scsi_debug_abort_cmnd(scmd);
return true;
}
/* Deletes (stops) timers or work queues of all queued commands per sdev */
static void scsi_debug_stop_all_queued(struct scsi_device *sdp)
{
struct Scsi_Host *shost = sdp->host;
blk_mq_tagset_busy_iter(&shost->tag_set,
scsi_debug_stop_all_queued_iter, sdp);
}
static int scsi_debug_device_reset(struct scsi_cmnd *SCpnt)
{
struct scsi_device *sdp = SCpnt->device;
struct sdebug_dev_info *devip = sdp->hostdata;
++num_dev_resets;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
if (SDEBUG_OPT_ALL_NOISE & sdebug_opts)
sdev_printk(KERN_INFO, sdp, "%s\n", __func__);
scsi_debug_stop_all_queued(sdp);
if (devip)
set_bit(SDEBUG_UA_POR, devip->uas_bm);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
return SUCCESS;
}
static int scsi_debug_target_reset(struct scsi_cmnd *SCpnt)
{
struct scsi_device *sdp = SCpnt->device;
struct sdebug_host_info *sdbg_host = shost_to_sdebug_host(sdp->host);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
struct sdebug_dev_info *devip;
int k = 0;
++num_target_resets;
if (SDEBUG_OPT_ALL_NOISE & sdebug_opts)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
sdev_printk(KERN_INFO, sdp, "%s\n", __func__);
list_for_each_entry(devip, &sdbg_host->dev_info_list, dev_list) {
if (devip->target == sdp->id) {
set_bit(SDEBUG_UA_BUS_RESET, devip->uas_bm);
++k;
}
}
if (SDEBUG_OPT_RESET_NOISE & sdebug_opts)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
sdev_printk(KERN_INFO, sdp,
"%s: %d device(s) found in target\n", __func__, k);
return SUCCESS;
}
static int scsi_debug_bus_reset(struct scsi_cmnd *SCpnt)
{
struct scsi_device *sdp = SCpnt->device;
struct sdebug_host_info *sdbg_host = shost_to_sdebug_host(sdp->host);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
struct sdebug_dev_info *devip;
int k = 0;
++num_bus_resets;
if (SDEBUG_OPT_ALL_NOISE & sdebug_opts)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
sdev_printk(KERN_INFO, sdp, "%s\n", __func__);
list_for_each_entry(devip, &sdbg_host->dev_info_list, dev_list) {
set_bit(SDEBUG_UA_BUS_RESET, devip->uas_bm);
++k;
}
if (SDEBUG_OPT_RESET_NOISE & sdebug_opts)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
sdev_printk(KERN_INFO, sdp,
"%s: %d device(s) found in host\n", __func__, k);
return SUCCESS;
}
static int scsi_debug_host_reset(struct scsi_cmnd *SCpnt)
{
struct sdebug_host_info *sdbg_host;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
struct sdebug_dev_info *devip;
int k = 0;
++num_host_resets;
if (SDEBUG_OPT_ALL_NOISE & sdebug_opts)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
sdev_printk(KERN_INFO, SCpnt->device, "%s\n", __func__);
mutex_lock(&sdebug_host_list_mutex);
list_for_each_entry(sdbg_host, &sdebug_host_list, host_list) {
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
list_for_each_entry(devip, &sdbg_host->dev_info_list,
dev_list) {
set_bit(SDEBUG_UA_BUS_RESET, devip->uas_bm);
++k;
}
}
mutex_unlock(&sdebug_host_list_mutex);
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
stop_all_queued();
if (SDEBUG_OPT_RESET_NOISE & sdebug_opts)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
sdev_printk(KERN_INFO, SCpnt->device,
"%s: %d device(s) found\n", __func__, k);
return SUCCESS;
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static void sdebug_build_parts(unsigned char *ramp, unsigned long store_size)
{
struct msdos_partition *pp;
int starts[SDEBUG_MAX_PARTS + 2], max_part_secs;
int sectors_per_part, num_sectors, k;
int heads_by_sects, start_sec, end_sec;
/* assume partition table already zeroed */
if ((sdebug_num_parts < 1) || (store_size < 1048576))
return;
if (sdebug_num_parts > SDEBUG_MAX_PARTS) {
sdebug_num_parts = SDEBUG_MAX_PARTS;
pr_warn("reducing partitions to %d\n", SDEBUG_MAX_PARTS);
}
num_sectors = (int)get_sdebug_capacity();
sectors_per_part = (num_sectors - sdebug_sectors_per)
/ sdebug_num_parts;
heads_by_sects = sdebug_heads * sdebug_sectors_per;
starts[0] = sdebug_sectors_per;
max_part_secs = sectors_per_part;
for (k = 1; k < sdebug_num_parts; ++k) {
starts[k] = ((k * sectors_per_part) / heads_by_sects)
* heads_by_sects;
if (starts[k] - starts[k - 1] < max_part_secs)
max_part_secs = starts[k] - starts[k - 1];
}
starts[sdebug_num_parts] = num_sectors;
starts[sdebug_num_parts + 1] = 0;
ramp[510] = 0x55; /* magic partition markings */
ramp[511] = 0xAA;
pp = (struct msdos_partition *)(ramp + 0x1be);
for (k = 0; starts[k + 1]; ++k, ++pp) {
start_sec = starts[k];
end_sec = starts[k] + max_part_secs - 1;
pp->boot_ind = 0;
pp->cyl = start_sec / heads_by_sects;
pp->head = (start_sec - (pp->cyl * heads_by_sects))
/ sdebug_sectors_per;
pp->sector = (start_sec % sdebug_sectors_per) + 1;
pp->end_cyl = end_sec / heads_by_sects;
pp->end_head = (end_sec - (pp->end_cyl * heads_by_sects))
/ sdebug_sectors_per;
pp->end_sector = (end_sec % sdebug_sectors_per) + 1;
pp->start_sect = cpu_to_le32(start_sec);
pp->nr_sects = cpu_to_le32(end_sec - start_sec + 1);
pp->sys_ind = 0x83; /* plain Linux partition */
}
}
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
static void block_unblock_all_queues(bool block)
{
struct sdebug_host_info *sdhp;
lockdep_assert_held(&sdebug_host_list_mutex);
list_for_each_entry(sdhp, &sdebug_host_list, host_list) {
struct Scsi_Host *shost = sdhp->shost;
if (block)
scsi_block_requests(shost);
else
scsi_unblock_requests(shost);
}
}
/* Adjust (by rounding down) the sdebug_cmnd_count so abs(every_nth)-1
* commands will be processed normally before triggers occur.
*/
static void tweak_cmnd_count(void)
{
int count, modulo;
modulo = abs(sdebug_every_nth);
if (modulo < 2)
return;
mutex_lock(&sdebug_host_list_mutex);
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
block_unblock_all_queues(true);
count = atomic_read(&sdebug_cmnd_count);
atomic_set(&sdebug_cmnd_count, (count / modulo) * modulo);
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
block_unblock_all_queues(false);
mutex_unlock(&sdebug_host_list_mutex);
}
static void clear_queue_stats(void)
{
atomic_set(&sdebug_cmnd_count, 0);
atomic_set(&sdebug_completions, 0);
atomic_set(&sdebug_miss_cpus, 0);
atomic_set(&sdebug_a_tsf, 0);
}
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
static bool inject_on_this_cmd(void)
{
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
if (sdebug_every_nth == 0)
return false;
return (atomic_read(&sdebug_cmnd_count) % abs(sdebug_every_nth)) == 0;
}
#define INCLUSIVE_TIMING_MAX_NS 1000000 /* 1 millisecond */
void sdebug_free_queued_cmd(struct sdebug_queued_cmd *sqcp)
{
if (sqcp)
kmem_cache_free(queued_cmd_cache, sqcp);
}
static struct sdebug_queued_cmd *sdebug_alloc_queued_cmd(struct scsi_cmnd *scmd)
{
struct sdebug_queued_cmd *sqcp;
struct sdebug_defer *sd_dp;
sqcp = kmem_cache_zalloc(queued_cmd_cache, GFP_ATOMIC);
if (!sqcp)
return NULL;
sd_dp = &sqcp->sd_dp;
hrtimer_init(&sd_dp->hrt, CLOCK_MONOTONIC, HRTIMER_MODE_REL_PINNED);
sd_dp->hrt.function = sdebug_q_cmd_hrt_complete;
INIT_WORK(&sd_dp->ew.work, sdebug_q_cmd_wq_complete);
sqcp->scmd = scmd;
return sqcp;
}
/* Complete the processing of the thread that queued a SCSI command to this
* driver. It either completes the command by calling cmnd_done() or
* schedules a hr timer or work queue then returns 0. Returns
* SCSI_MLQUEUE_HOST_BUSY if temporarily out of resources.
*/
static int schedule_resp(struct scsi_cmnd *cmnd, struct sdebug_dev_info *devip,
int scsi_result,
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
int (*pfp)(struct scsi_cmnd *,
struct sdebug_dev_info *),
int delta_jiff, int ndelay)
{
struct request *rq = scsi_cmd_to_rq(cmnd);
bool polled = rq->cmd_flags & REQ_POLLED;
struct sdebug_scsi_cmd *sdsc = scsi_cmd_priv(cmnd);
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
unsigned long flags;
u64 ns_from_boot = 0;
struct sdebug_queued_cmd *sqcp;
struct scsi_device *sdp;
struct sdebug_defer *sd_dp;
if (unlikely(devip == NULL)) {
if (scsi_result == 0)
scsi_result = DID_NO_CONNECT << 16;
goto respond_in_thread;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
}
sdp = cmnd->device;
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
if (delta_jiff == 0)
goto respond_in_thread;
if (unlikely(sdebug_every_nth && (SDEBUG_OPT_RARE_TSF & sdebug_opts) &&
(scsi_result == 0))) {
int num_in_q = scsi_device_busy(sdp);
int qdepth = cmnd->device->queue_depth;
if ((num_in_q == qdepth) &&
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
(atomic_inc_return(&sdebug_a_tsf) >=
abs(sdebug_every_nth))) {
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
atomic_set(&sdebug_a_tsf, 0);
scsi_result = device_qfull_result;
if (unlikely(SDEBUG_OPT_Q_NOISE & sdebug_opts))
sdev_printk(KERN_INFO, sdp, "%s: num_in_q=%d +1, <inject> status: TASK SET FULL\n",
__func__, num_in_q);
}
}
sqcp = sdebug_alloc_queued_cmd(cmnd);
if (!sqcp) {
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
pr_err("%s no alloc\n", __func__);
return SCSI_MLQUEUE_HOST_BUSY;
}
sd_dp = &sqcp->sd_dp;
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
if (polled)
ns_from_boot = ktime_get_boottime_ns();
/* one of the resp_*() response functions is called here */
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
cmnd->result = pfp ? pfp(cmnd, devip) : 0;
if (cmnd->result & SDEG_RES_IMMED_MASK) {
cmnd->result &= ~SDEG_RES_IMMED_MASK;
delta_jiff = ndelay = 0;
}
if (cmnd->result == 0 && scsi_result != 0)
cmnd->result = scsi_result;
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
if (cmnd->result == 0 && unlikely(sdebug_opts & SDEBUG_OPT_TRANSPORT_ERR)) {
if (atomic_read(&sdeb_inject_pending)) {
mk_sense_buffer(cmnd, ABORTED_COMMAND, TRANSPORT_PROBLEM, ACK_NAK_TO);
atomic_set(&sdeb_inject_pending, 0);
cmnd->result = check_condition_result;
}
}
if (unlikely(sdebug_verbose && cmnd->result))
sdev_printk(KERN_INFO, sdp, "%s: non-zero result=0x%x\n",
__func__, cmnd->result);
if (delta_jiff > 0 || ndelay > 0) {
ktime_t kt;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
if (delta_jiff > 0) {
u64 ns = jiffies_to_nsecs(delta_jiff);
if (sdebug_random && ns < U32_MAX) {
ns = get_random_u32_below((u32)ns);
} else if (sdebug_random) {
ns >>= 12; /* scale to 4 usec precision */
if (ns < U32_MAX) /* over 4 hours max */
ns = get_random_u32_below((u32)ns);
ns <<= 12;
}
kt = ns_to_ktime(ns);
} else { /* ndelay has a 4.2 second max */
kt = sdebug_random ? get_random_u32_below((u32)ndelay) :
(u32)ndelay;
if (ndelay < INCLUSIVE_TIMING_MAX_NS) {
u64 d = ktime_get_boottime_ns() - ns_from_boot;
if (kt <= d) { /* elapsed duration >= kt */
/* call scsi_done() from this thread */
sdebug_free_queued_cmd(sqcp);
scsi_done(cmnd);
return 0;
}
/* otherwise reduce kt by elapsed time */
kt -= d;
}
}
if (sdebug_statistics)
sd_dp->issuing_cpu = raw_smp_processor_id();
if (polled) {
spin_lock_irqsave(&sdsc->lock, flags);
sd_dp->cmpl_ts = ktime_add(ns_to_ktime(ns_from_boot), kt);
ASSIGN_QUEUED_CMD(cmnd, sqcp);
WRITE_ONCE(sd_dp->defer_t, SDEB_DEFER_POLL);
spin_unlock_irqrestore(&sdsc->lock, flags);
} else {
/* schedule the invocation of scsi_done() for a later time */
spin_lock_irqsave(&sdsc->lock, flags);
ASSIGN_QUEUED_CMD(cmnd, sqcp);
WRITE_ONCE(sd_dp->defer_t, SDEB_DEFER_HRT);
hrtimer_start(&sd_dp->hrt, kt, HRTIMER_MODE_REL_PINNED);
/*
* The completion handler will try to grab sqcp->lock,
* so there is no chance that the completion handler
* will call scsi_done() until we release the lock
* here (so ok to keep referencing sdsc).
*/
spin_unlock_irqrestore(&sdsc->lock, flags);
}
} else { /* jdelay < 0, use work queue */
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
if (unlikely((sdebug_opts & SDEBUG_OPT_CMD_ABORT) &&
atomic_read(&sdeb_inject_pending))) {
sd_dp->aborted = true;
atomic_set(&sdeb_inject_pending, 0);
sdev_printk(KERN_INFO, sdp, "abort request tag=%#x\n",
blk_mq_unique_tag_to_tag(get_tag(cmnd)));
}
if (sdebug_statistics)
sd_dp->issuing_cpu = raw_smp_processor_id();
if (polled) {
spin_lock_irqsave(&sdsc->lock, flags);
ASSIGN_QUEUED_CMD(cmnd, sqcp);
sd_dp->cmpl_ts = ns_to_ktime(ns_from_boot);
WRITE_ONCE(sd_dp->defer_t, SDEB_DEFER_POLL);
spin_unlock_irqrestore(&sdsc->lock, flags);
} else {
spin_lock_irqsave(&sdsc->lock, flags);
ASSIGN_QUEUED_CMD(cmnd, sqcp);
WRITE_ONCE(sd_dp->defer_t, SDEB_DEFER_WQ);
schedule_work(&sd_dp->ew.work);
spin_unlock_irqrestore(&sdsc->lock, flags);
}
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
return 0;
respond_in_thread: /* call back to mid-layer using invocation thread */
cmnd->result = pfp != NULL ? pfp(cmnd, devip) : 0;
cmnd->result &= ~SDEG_RES_IMMED_MASK;
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
if (cmnd->result == 0 && scsi_result != 0)
cmnd->result = scsi_result;
scsi_done(cmnd);
return 0;
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
/* Note: The following macros create attribute files in the
/sys/module/scsi_debug/parameters directory. Unfortunately this
driver is unaware of a change and cannot trigger auxiliary actions
as it can when the corresponding attribute in the
/sys/bus/pseudo/drivers/scsi_debug directory is changed.
*/
module_param_named(add_host, sdebug_add_host, int, S_IRUGO | S_IWUSR);
module_param_named(ato, sdebug_ato, int, S_IRUGO);
module_param_named(cdb_len, sdebug_cdb_len, int, 0644);
module_param_named(clustering, sdebug_clustering, bool, S_IRUGO | S_IWUSR);
module_param_named(delay, sdebug_jdelay, int, S_IRUGO | S_IWUSR);
module_param_named(dev_size_mb, sdebug_dev_size_mb, int, S_IRUGO);
module_param_named(dif, sdebug_dif, int, S_IRUGO);
module_param_named(dix, sdebug_dix, int, S_IRUGO);
module_param_named(dsense, sdebug_dsense, int, S_IRUGO | S_IWUSR);
module_param_named(every_nth, sdebug_every_nth, int, S_IRUGO | S_IWUSR);
module_param_named(fake_rw, sdebug_fake_rw, int, S_IRUGO | S_IWUSR);
module_param_named(guard, sdebug_guard, uint, S_IRUGO);
module_param_named(host_lock, sdebug_host_lock, bool, S_IRUGO | S_IWUSR);
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
module_param_named(host_max_queue, sdebug_host_max_queue, int, S_IRUGO);
module_param_string(inq_product, sdebug_inq_product_id,
sizeof(sdebug_inq_product_id), S_IRUGO | S_IWUSR);
module_param_string(inq_rev, sdebug_inq_product_rev,
sizeof(sdebug_inq_product_rev), S_IRUGO | S_IWUSR);
module_param_string(inq_vendor, sdebug_inq_vendor_id,
sizeof(sdebug_inq_vendor_id), S_IRUGO | S_IWUSR);
module_param_named(lbprz, sdebug_lbprz, int, S_IRUGO);
module_param_named(lbpu, sdebug_lbpu, int, S_IRUGO);
module_param_named(lbpws, sdebug_lbpws, int, S_IRUGO);
module_param_named(lbpws10, sdebug_lbpws10, int, S_IRUGO);
module_param_named(lowest_aligned, sdebug_lowest_aligned, int, S_IRUGO);
module_param_named(lun_format, sdebug_lun_am_i, int, S_IRUGO | S_IWUSR);
module_param_named(max_luns, sdebug_max_luns, int, S_IRUGO | S_IWUSR);
module_param_named(max_queue, sdebug_max_queue, int, S_IRUGO | S_IWUSR);
module_param_named(medium_error_count, sdebug_medium_error_count, int,
S_IRUGO | S_IWUSR);
module_param_named(medium_error_start, sdebug_medium_error_start, int,
S_IRUGO | S_IWUSR);
module_param_named(ndelay, sdebug_ndelay, int, S_IRUGO | S_IWUSR);
module_param_named(no_lun_0, sdebug_no_lun_0, int, S_IRUGO | S_IWUSR);
module_param_named(no_rwlock, sdebug_no_rwlock, bool, S_IRUGO | S_IWUSR);
module_param_named(no_uld, sdebug_no_uld, int, S_IRUGO);
module_param_named(num_parts, sdebug_num_parts, int, S_IRUGO);
module_param_named(num_tgts, sdebug_num_tgts, int, S_IRUGO | S_IWUSR);
module_param_named(opt_blks, sdebug_opt_blks, int, S_IRUGO);
module_param_named(opt_xferlen_exp, sdebug_opt_xferlen_exp, int, S_IRUGO);
module_param_named(opts, sdebug_opts, int, S_IRUGO | S_IWUSR);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
module_param_named(per_host_store, sdebug_per_host_store, bool,
S_IRUGO | S_IWUSR);
module_param_named(physblk_exp, sdebug_physblk_exp, int, S_IRUGO);
module_param_named(ptype, sdebug_ptype, int, S_IRUGO | S_IWUSR);
module_param_named(random, sdebug_random, bool, S_IRUGO | S_IWUSR);
module_param_named(removable, sdebug_removable, bool, S_IRUGO | S_IWUSR);
module_param_named(scsi_level, sdebug_scsi_level, int, S_IRUGO);
module_param_named(sector_size, sdebug_sector_size, int, S_IRUGO);
module_param_named(statistics, sdebug_statistics, bool, S_IRUGO | S_IWUSR);
module_param_named(strict, sdebug_strict, bool, S_IRUGO | S_IWUSR);
module_param_named(submit_queues, submit_queues, int, S_IRUGO);
module_param_named(poll_queues, poll_queues, int, S_IRUGO);
module_param_named(tur_ms_to_ready, sdeb_tur_ms_to_ready, int, S_IRUGO);
module_param_named(unmap_alignment, sdebug_unmap_alignment, int, S_IRUGO);
module_param_named(unmap_granularity, sdebug_unmap_granularity, int, S_IRUGO);
module_param_named(unmap_max_blocks, sdebug_unmap_max_blocks, int, S_IRUGO);
module_param_named(unmap_max_desc, sdebug_unmap_max_desc, int, S_IRUGO);
module_param_named(uuid_ctl, sdebug_uuid_ctl, int, S_IRUGO);
module_param_named(virtual_gb, sdebug_virtual_gb, int, S_IRUGO | S_IWUSR);
module_param_named(vpd_use_hostno, sdebug_vpd_use_hostno, int,
S_IRUGO | S_IWUSR);
module_param_named(wp, sdebug_wp, bool, S_IRUGO | S_IWUSR);
module_param_named(write_same_length, sdebug_write_same_length, int,
S_IRUGO | S_IWUSR);
module_param_named(zbc, sdeb_zbc_model_s, charp, S_IRUGO);
module_param_named(zone_cap_mb, sdeb_zbc_zone_cap_mb, int, S_IRUGO);
module_param_named(zone_max_open, sdeb_zbc_max_open, int, S_IRUGO);
module_param_named(zone_nr_conv, sdeb_zbc_nr_conv, int, S_IRUGO);
module_param_named(zone_size_mb, sdeb_zbc_zone_size_mb, int, S_IRUGO);
MODULE_AUTHOR("Eric Youngdale + Douglas Gilbert");
MODULE_DESCRIPTION("SCSI debug adapter driver");
MODULE_LICENSE("GPL");
MODULE_VERSION(SDEBUG_VERSION);
MODULE_PARM_DESC(add_host, "add n hosts, in sysfs if negative remove host(s) (def=1)");
MODULE_PARM_DESC(ato, "application tag ownership: 0=disk 1=host (def=1)");
MODULE_PARM_DESC(cdb_len, "suggest CDB lengths to drivers (def=10)");
MODULE_PARM_DESC(clustering, "when set enables larger transfers (def=0)");
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
MODULE_PARM_DESC(delay, "response delay (def=1 jiffy); 0:imm, -1,-2:tiny");
MODULE_PARM_DESC(dev_size_mb, "size in MiB of ram shared by devs(def=8)");
MODULE_PARM_DESC(dif, "data integrity field type: 0-3 (def=0)");
MODULE_PARM_DESC(dix, "data integrity extensions mask (def=0)");
MODULE_PARM_DESC(dsense, "use descriptor sense format(def=0 -> fixed)");
MODULE_PARM_DESC(every_nth, "timeout every nth command(def=0)");
MODULE_PARM_DESC(fake_rw, "fake reads/writes instead of copying (def=0)");
MODULE_PARM_DESC(guard, "protection checksum: 0=crc, 1=ip (def=0)");
MODULE_PARM_DESC(host_lock, "host_lock is ignored (def=0)");
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
MODULE_PARM_DESC(host_max_queue,
"host max # of queued cmds (0 to max(def) [max_queue fixed equal for !0])");
MODULE_PARM_DESC(inq_product, "SCSI INQUIRY product string (def=\"scsi_debug\")");
MODULE_PARM_DESC(inq_rev, "SCSI INQUIRY revision string (def=\""
SDEBUG_VERSION "\")");
MODULE_PARM_DESC(inq_vendor, "SCSI INQUIRY vendor string (def=\"Linux\")");
MODULE_PARM_DESC(lbprz,
"on read unmapped LBs return 0 when 1 (def), return 0xff when 2");
MODULE_PARM_DESC(lbpu, "enable LBP, support UNMAP command (def=0)");
MODULE_PARM_DESC(lbpws, "enable LBP, support WRITE SAME(16) with UNMAP bit (def=0)");
MODULE_PARM_DESC(lbpws10, "enable LBP, support WRITE SAME(10) with UNMAP bit (def=0)");
MODULE_PARM_DESC(lowest_aligned, "lowest aligned lba (def=0)");
MODULE_PARM_DESC(lun_format, "LUN format: 0->peripheral (def); 1 --> flat address method");
MODULE_PARM_DESC(max_luns, "number of LUNs per target to simulate(def=1)");
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
MODULE_PARM_DESC(max_queue, "max number of queued commands (1 to max(def))");
MODULE_PARM_DESC(medium_error_count, "count of sectors to return follow on MEDIUM error");
MODULE_PARM_DESC(medium_error_start, "starting sector number to return MEDIUM error");
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
MODULE_PARM_DESC(ndelay, "response delay in nanoseconds (def=0 -> ignore)");
MODULE_PARM_DESC(no_lun_0, "no LU number 0 (def=0 -> have lun 0)");
MODULE_PARM_DESC(no_rwlock, "don't protect user data reads+writes (def=0)");
MODULE_PARM_DESC(no_uld, "stop ULD (e.g. sd driver) attaching (def=0))");
MODULE_PARM_DESC(num_parts, "number of partitions(def=0)");
MODULE_PARM_DESC(num_tgts, "number of targets per host to simulate(def=1)");
MODULE_PARM_DESC(opt_blks, "optimal transfer length in blocks (def=1024)");
MODULE_PARM_DESC(opt_xferlen_exp, "optimal transfer length granularity exponent (def=physblk_exp)");
MODULE_PARM_DESC(opts, "1->noise, 2->medium_err, 4->timeout, 8->recovered_err... (def=0)");
MODULE_PARM_DESC(per_host_store, "If set, next positive add_host will get new store (def=0)");
MODULE_PARM_DESC(physblk_exp, "physical block exponent (def=0)");
MODULE_PARM_DESC(poll_queues, "support for iouring iopoll queues (1 to max(submit_queues - 1))");
MODULE_PARM_DESC(ptype, "SCSI peripheral type(def=0[disk])");
MODULE_PARM_DESC(random, "If set, uniformly randomize command duration between 0 and delay_in_ns");
MODULE_PARM_DESC(removable, "claim to have removable media (def=0)");
MODULE_PARM_DESC(scsi_level, "SCSI level to simulate(def=7[SPC-5])");
MODULE_PARM_DESC(sector_size, "logical block size in bytes (def=512)");
MODULE_PARM_DESC(statistics, "collect statistics on commands, queues (def=0)");
MODULE_PARM_DESC(strict, "stricter checks: reserved field in cdb (def=0)");
MODULE_PARM_DESC(submit_queues, "support for block multi-queue (def=1)");
MODULE_PARM_DESC(tur_ms_to_ready, "TEST UNIT READY millisecs before initial good status (def=0)");
MODULE_PARM_DESC(unmap_alignment, "lowest aligned thin provisioning lba (def=0)");
MODULE_PARM_DESC(unmap_granularity, "thin provisioning granularity in blocks (def=1)");
MODULE_PARM_DESC(unmap_max_blocks, "max # of blocks can be unmapped in one cmd (def=0xffffffff)");
MODULE_PARM_DESC(unmap_max_desc, "max # of ranges that can be unmapped in one cmd (def=256)");
MODULE_PARM_DESC(uuid_ctl,
"1->use uuid for lu name, 0->don't, 2->all use same (def=0)");
MODULE_PARM_DESC(virtual_gb, "virtual gigabyte (GiB) size (def=0 -> use dev_size_mb)");
MODULE_PARM_DESC(vpd_use_hostno, "0 -> dev ids ignore hostno (def=1 -> unique dev ids)");
MODULE_PARM_DESC(wp, "Write Protect (def=0)");
MODULE_PARM_DESC(write_same_length, "Maximum blocks per WRITE SAME cmd (def=0xffff)");
MODULE_PARM_DESC(zbc, "'none' [0]; 'aware' [1]; 'managed' [2] (def=0). Can have 'host-' prefix");
MODULE_PARM_DESC(zone_cap_mb, "Zone capacity in MiB (def=zone size)");
MODULE_PARM_DESC(zone_max_open, "Maximum number of open zones; [0] for no limit (def=auto)");
MODULE_PARM_DESC(zone_nr_conv, "Number of conventional zones (def=1)");
MODULE_PARM_DESC(zone_size_mb, "Zone size in MiB (def=auto)");
#define SDEBUG_INFO_LEN 256
static char sdebug_info[SDEBUG_INFO_LEN];
static const char *scsi_debug_info(struct Scsi_Host *shp)
{
int k;
k = scnprintf(sdebug_info, SDEBUG_INFO_LEN, "%s: version %s [%s]\n",
my_name, SDEBUG_VERSION, sdebug_version_date);
if (k >= (SDEBUG_INFO_LEN - 1))
return sdebug_info;
scnprintf(sdebug_info + k, SDEBUG_INFO_LEN - k,
" dev_size_mb=%d, opts=0x%x, submit_queues=%d, %s=%d",
sdebug_dev_size_mb, sdebug_opts, submit_queues,
"statistics", (int)sdebug_statistics);
return sdebug_info;
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
/* 'echo <val> > /proc/scsi/scsi_debug/<host_id>' writes to opts */
static int scsi_debug_write_info(struct Scsi_Host *host, char *buffer,
int length)
{
char arr[16];
int opts;
int minLen = length > 15 ? 15 : length;
if (!capable(CAP_SYS_ADMIN) || !capable(CAP_SYS_RAWIO))
return -EACCES;
memcpy(arr, buffer, minLen);
arr[minLen] = '\0';
if (1 != sscanf(arr, "%d", &opts))
return -EINVAL;
sdebug_opts = opts;
sdebug_verbose = !!(SDEBUG_OPT_NOISE & opts);
sdebug_any_injecting_opt = !!(SDEBUG_OPT_ALL_INJECTING & opts);
if (sdebug_every_nth != 0)
tweak_cmnd_count();
return length;
}
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
struct sdebug_submit_queue_data {
int *first;
int *last;
int queue_num;
};
static bool sdebug_submit_queue_iter(struct request *rq, void *opaque)
{
struct sdebug_submit_queue_data *data = opaque;
u32 unique_tag = blk_mq_unique_tag(rq);
u16 hwq = blk_mq_unique_tag_to_hwq(unique_tag);
u16 tag = blk_mq_unique_tag_to_tag(unique_tag);
int queue_num = data->queue_num;
if (hwq != queue_num)
return true;
/* Rely on iter'ing in ascending tag order */
if (*data->first == -1)
*data->first = *data->last = tag;
else
*data->last = tag;
return true;
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
/* Output seen with 'cat /proc/scsi/scsi_debug/<host_id>'. It will be the
* same for each scsi_debug host (if more than one). Some of the counters
* output are not atomics so might be inaccurate in a busy system. */
static int scsi_debug_show_info(struct seq_file *m, struct Scsi_Host *host)
{
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
struct sdebug_host_info *sdhp;
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
int j;
seq_printf(m, "scsi_debug adapter driver, version %s [%s]\n",
SDEBUG_VERSION, sdebug_version_date);
seq_printf(m, "num_tgts=%d, %ssize=%d MB, opts=0x%x, every_nth=%d\n",
sdebug_num_tgts, "shared (ram) ", sdebug_dev_size_mb,
sdebug_opts, sdebug_every_nth);
seq_printf(m, "delay=%d, ndelay=%d, max_luns=%d, sector_size=%d %s\n",
sdebug_jdelay, sdebug_ndelay, sdebug_max_luns,
sdebug_sector_size, "bytes");
seq_printf(m, "cylinders=%d, heads=%d, sectors=%d, command aborts=%d\n",
sdebug_cylinders_per, sdebug_heads, sdebug_sectors_per,
num_aborts);
seq_printf(m, "RESETs: device=%d, target=%d, bus=%d, host=%d\n",
num_dev_resets, num_target_resets, num_bus_resets,
num_host_resets);
seq_printf(m, "dix_reads=%d, dix_writes=%d, dif_errors=%d\n",
dix_reads, dix_writes, dif_errors);
seq_printf(m, "usec_in_jiffy=%lu, statistics=%d\n", TICK_NSEC / 1000,
sdebug_statistics);
seq_printf(m, "cmnd_count=%d, completions=%d, %s=%d, a_tsf=%d, mq_polls=%d\n",
atomic_read(&sdebug_cmnd_count),
atomic_read(&sdebug_completions),
"miss_cpus", atomic_read(&sdebug_miss_cpus),
atomic_read(&sdebug_a_tsf),
atomic_read(&sdeb_mq_poll_count));
seq_printf(m, "submit_queues=%d\n", submit_queues);
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
for (j = 0; j < submit_queues; ++j) {
int f = -1, l = -1;
struct sdebug_submit_queue_data data = {
.queue_num = j,
.first = &f,
.last = &l,
};
seq_printf(m, " queue %d:\n", j);
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
blk_mq_tagset_busy_iter(&host->tag_set, sdebug_submit_queue_iter,
&data);
if (f >= 0) {
seq_printf(m, " in_use_bm BUSY: %s: %d,%d\n",
"first,last bits", f, l);
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
seq_printf(m, "this host_no=%d\n", host->host_no);
if (!xa_empty(per_store_ap)) {
bool niu;
int idx;
unsigned long l_idx;
struct sdeb_store_info *sip;
seq_puts(m, "\nhost list:\n");
j = 0;
list_for_each_entry(sdhp, &sdebug_host_list, host_list) {
idx = sdhp->si_idx;
seq_printf(m, " %d: host_no=%d, si_idx=%d\n", j,
sdhp->shost->host_no, idx);
++j;
}
seq_printf(m, "\nper_store array [most_recent_idx=%d]:\n",
sdeb_most_recent_idx);
j = 0;
xa_for_each(per_store_ap, l_idx, sip) {
niu = xa_get_mark(per_store_ap, l_idx,
SDEB_XA_NOT_IN_USE);
idx = (int)l_idx;
seq_printf(m, " %d: idx=%d%s\n", j, idx,
(niu ? " not_in_use" : ""));
++j;
}
}
return 0;
}
static ssize_t delay_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_jdelay);
}
/* Returns -EBUSY if jdelay is being changed and commands are queued. The unit
* of delay is jiffies.
*/
static ssize_t delay_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int jdelay, res;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
if (count > 0 && sscanf(buf, "%d", &jdelay) == 1) {
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
res = count;
if (sdebug_jdelay != jdelay) {
struct sdebug_host_info *sdhp;
mutex_lock(&sdebug_host_list_mutex);
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
block_unblock_all_queues(true);
list_for_each_entry(sdhp, &sdebug_host_list, host_list) {
struct Scsi_Host *shost = sdhp->shost;
if (scsi_host_busy(shost)) {
res = -EBUSY; /* queued commands */
break;
}
}
if (res > 0) {
sdebug_jdelay = jdelay;
sdebug_ndelay = 0;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
}
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
block_unblock_all_queues(false);
mutex_unlock(&sdebug_host_list_mutex);
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
return res;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(delay);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
static ssize_t ndelay_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_ndelay);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
}
/* Returns -EBUSY if ndelay is being changed and commands are queued */
/* If > 0 and accepted then sdebug_jdelay is set to JDELAY_OVERRIDDEN */
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
static ssize_t ndelay_store(struct device_driver *ddp, const char *buf,
size_t count)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
{
int ndelay, res;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
if ((count > 0) && (1 == sscanf(buf, "%d", &ndelay)) &&
(ndelay >= 0) && (ndelay < (1000 * 1000 * 1000))) {
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
res = count;
if (sdebug_ndelay != ndelay) {
struct sdebug_host_info *sdhp;
mutex_lock(&sdebug_host_list_mutex);
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
block_unblock_all_queues(true);
list_for_each_entry(sdhp, &sdebug_host_list, host_list) {
struct Scsi_Host *shost = sdhp->shost;
if (scsi_host_busy(shost)) {
res = -EBUSY; /* queued commands */
break;
}
}
if (res > 0) {
sdebug_ndelay = ndelay;
sdebug_jdelay = ndelay ? JDELAY_OVERRIDDEN
: DEF_JDELAY;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
}
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
block_unblock_all_queues(false);
mutex_unlock(&sdebug_host_list_mutex);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
}
return res;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(ndelay);
static ssize_t opts_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "0x%x\n", sdebug_opts);
}
static ssize_t opts_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int opts;
char work[20];
if (sscanf(buf, "%10s", work) == 1) {
if (strncasecmp(work, "0x", 2) == 0) {
if (kstrtoint(work + 2, 16, &opts) == 0)
goto opts_done;
} else {
if (kstrtoint(work, 10, &opts) == 0)
goto opts_done;
}
}
return -EINVAL;
opts_done:
sdebug_opts = opts;
sdebug_verbose = !!(SDEBUG_OPT_NOISE & opts);
sdebug_any_injecting_opt = !!(SDEBUG_OPT_ALL_INJECTING & opts);
tweak_cmnd_count();
return count;
}
static DRIVER_ATTR_RW(opts);
static ssize_t ptype_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_ptype);
}
static ssize_t ptype_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
/* Cannot change from or to TYPE_ZBC with sysfs */
if (sdebug_ptype == TYPE_ZBC)
return -EINVAL;
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n >= 0)) {
if (n == TYPE_ZBC)
return -EINVAL;
sdebug_ptype = n;
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(ptype);
static ssize_t dsense_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_dsense);
}
static ssize_t dsense_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n >= 0)) {
sdebug_dsense = n;
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(dsense);
static ssize_t fake_rw_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_fake_rw);
}
static ssize_t fake_rw_store(struct device_driver *ddp, const char *buf,
size_t count)
{
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
int n, idx;
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n >= 0)) {
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
bool want_store = (n == 0);
struct sdebug_host_info *sdhp;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
n = (n > 0);
sdebug_fake_rw = (sdebug_fake_rw > 0);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
if (sdebug_fake_rw == n)
return count; /* not transitioning so do nothing */
if (want_store) { /* 1 --> 0 transition, set up store */
if (sdeb_first_idx < 0) {
idx = sdebug_add_store();
if (idx < 0)
return idx;
} else {
idx = sdeb_first_idx;
xa_clear_mark(per_store_ap, idx,
SDEB_XA_NOT_IN_USE);
}
/* make all hosts use same store */
list_for_each_entry(sdhp, &sdebug_host_list,
host_list) {
if (sdhp->si_idx != idx) {
xa_set_mark(per_store_ap, sdhp->si_idx,
SDEB_XA_NOT_IN_USE);
sdhp->si_idx = idx;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
}
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
sdeb_most_recent_idx = idx;
} else { /* 0 --> 1 transition is trigger for shrink */
sdebug_erase_all_stores(true /* apart from first */);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
sdebug_fake_rw = n;
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(fake_rw);
static ssize_t no_lun_0_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_no_lun_0);
}
static ssize_t no_lun_0_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n >= 0)) {
sdebug_no_lun_0 = n;
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(no_lun_0);
static ssize_t num_tgts_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_num_tgts);
}
static ssize_t num_tgts_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n >= 0)) {
sdebug_num_tgts = n;
sdebug_max_tgts_luns();
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(num_tgts);
static ssize_t dev_size_mb_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_dev_size_mb);
}
static DRIVER_ATTR_RO(dev_size_mb);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static ssize_t per_host_store_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_per_host_store);
}
static ssize_t per_host_store_store(struct device_driver *ddp, const char *buf,
size_t count)
{
bool v;
if (kstrtobool(buf, &v))
return -EINVAL;
sdebug_per_host_store = v;
return count;
}
static DRIVER_ATTR_RW(per_host_store);
static ssize_t num_parts_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_num_parts);
}
static DRIVER_ATTR_RO(num_parts);
static ssize_t every_nth_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_every_nth);
}
static ssize_t every_nth_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int nth;
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
char work[20];
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
if (sscanf(buf, "%10s", work) == 1) {
if (strncasecmp(work, "0x", 2) == 0) {
if (kstrtoint(work + 2, 16, &nth) == 0)
goto every_nth_done;
} else {
if (kstrtoint(work, 10, &nth) == 0)
goto every_nth_done;
}
}
return -EINVAL;
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
every_nth_done:
sdebug_every_nth = nth;
if (nth && !sdebug_statistics) {
pr_info("every_nth needs statistics=1, set it\n");
sdebug_statistics = true;
}
tweak_cmnd_count();
return count;
}
static DRIVER_ATTR_RW(every_nth);
static ssize_t lun_format_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", (int)sdebug_lun_am);
}
static ssize_t lun_format_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
bool changed;
if (kstrtoint(buf, 0, &n))
return -EINVAL;
if (n >= 0) {
if (n > (int)SAM_LUN_AM_FLAT) {
pr_warn("only LUN address methods 0 and 1 are supported\n");
return -EINVAL;
}
changed = ((int)sdebug_lun_am != n);
sdebug_lun_am = n;
if (changed && sdebug_scsi_level >= 5) { /* >= SPC-3 */
struct sdebug_host_info *sdhp;
struct sdebug_dev_info *dp;
mutex_lock(&sdebug_host_list_mutex);
list_for_each_entry(sdhp, &sdebug_host_list, host_list) {
list_for_each_entry(dp, &sdhp->dev_info_list, dev_list) {
set_bit(SDEBUG_UA_LUNS_CHANGED, dp->uas_bm);
}
}
mutex_unlock(&sdebug_host_list_mutex);
}
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(lun_format);
static ssize_t max_luns_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_max_luns);
}
static ssize_t max_luns_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
bool changed;
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n >= 0)) {
if (n > 256) {
pr_warn("max_luns can be no more than 256\n");
return -EINVAL;
}
changed = (sdebug_max_luns != n);
sdebug_max_luns = n;
sdebug_max_tgts_luns();
if (changed && (sdebug_scsi_level >= 5)) { /* >= SPC-3 */
struct sdebug_host_info *sdhp;
struct sdebug_dev_info *dp;
mutex_lock(&sdebug_host_list_mutex);
list_for_each_entry(sdhp, &sdebug_host_list,
host_list) {
list_for_each_entry(dp, &sdhp->dev_info_list,
dev_list) {
set_bit(SDEBUG_UA_LUNS_CHANGED,
dp->uas_bm);
}
}
mutex_unlock(&sdebug_host_list_mutex);
}
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(max_luns);
static ssize_t max_queue_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_max_queue);
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
/* N.B. max_queue can be changed while there are queued commands. In flight
* commands beyond the new max_queue will be completed. */
static ssize_t max_queue_store(struct device_driver *ddp, const char *buf,
size_t count)
{
scsi: scsi_debug: Only allow sdebug_max_queue be modified when no shosts The shost->can_queue value is initially used to set per-HW queue context tag depth in the block layer. This ensures that the shost is not sent too many commands which it can deal with. However lowering sdebug_max_queue separately means that we can easily overload the shost, as in the following example: $ cat /sys/bus/pseudo/drivers/scsi_debug/max_queue 192 $ cat /sys/class/scsi_host/host0/can_queue 192 $ echo 100 > /sys/bus/pseudo/drivers/scsi_debug/max_queue $ cat /sys/class/scsi_host/host0/can_queue 192 $ fio --filename=/dev/sda --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=1200 --numjobs=10 --time_based --group_reporting --name=iops-test-job --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error iops-test-job: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 10 processes [ 111.269885] scsi_io_completion_action: 400 callbacks suppressed [ 111.269885] blk_print_req_error: 400 callbacks suppressed [ 111.269889] I/O error, dev sda, sector 440 op 0x0:(READ) flags 0x1200000 phys_seg 1 prio class 2 [ 111.269892] sd 0:0:0:0: [sda] tag#132 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 111.269897] sd 0:0:0:0: [sda] tag#132 CDB: Read(10) 28 00 00 00 01 68 00 00 08 00 [ 111.277058] I/O error, dev sda, sector 360 op 0x0:(READ) flags 0x1200000 phys_seg 1 prio class 2 [...] Ensure that this cannot happen by allowing sdebug_max_queue be modified only when we have no shosts. As such, any shost->can_queue value will match sdebug_max_queue, and sdebug_max_queue cannot be modified separately. Since retired_max_queue is no longer set, remove support. Continue to apply the restriction that sdebug_host_max_queue cannot be modified when sdebug_host_max_queue is set. Adding support for that would mean extra code, and no one has complained about this restriction previously. A command like the following may be used to remove a shost: echo -1 > /sys/bus/pseudo/drivers/scsi_debug/add_host Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-11-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:09 +00:00
int n;
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n > 0) &&
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
(n <= SDEBUG_CANQUEUE) &&
(sdebug_host_max_queue == 0)) {
mutex_lock(&sdebug_host_list_mutex);
scsi: scsi_debug: Only allow sdebug_max_queue be modified when no shosts The shost->can_queue value is initially used to set per-HW queue context tag depth in the block layer. This ensures that the shost is not sent too many commands which it can deal with. However lowering sdebug_max_queue separately means that we can easily overload the shost, as in the following example: $ cat /sys/bus/pseudo/drivers/scsi_debug/max_queue 192 $ cat /sys/class/scsi_host/host0/can_queue 192 $ echo 100 > /sys/bus/pseudo/drivers/scsi_debug/max_queue $ cat /sys/class/scsi_host/host0/can_queue 192 $ fio --filename=/dev/sda --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=1200 --numjobs=10 --time_based --group_reporting --name=iops-test-job --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error iops-test-job: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 10 processes [ 111.269885] scsi_io_completion_action: 400 callbacks suppressed [ 111.269885] blk_print_req_error: 400 callbacks suppressed [ 111.269889] I/O error, dev sda, sector 440 op 0x0:(READ) flags 0x1200000 phys_seg 1 prio class 2 [ 111.269892] sd 0:0:0:0: [sda] tag#132 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 111.269897] sd 0:0:0:0: [sda] tag#132 CDB: Read(10) 28 00 00 00 01 68 00 00 08 00 [ 111.277058] I/O error, dev sda, sector 360 op 0x0:(READ) flags 0x1200000 phys_seg 1 prio class 2 [...] Ensure that this cannot happen by allowing sdebug_max_queue be modified only when we have no shosts. As such, any shost->can_queue value will match sdebug_max_queue, and sdebug_max_queue cannot be modified separately. Since retired_max_queue is no longer set, remove support. Continue to apply the restriction that sdebug_host_max_queue cannot be modified when sdebug_host_max_queue is set. Adding support for that would mean extra code, and no one has complained about this restriction previously. A command like the following may be used to remove a shost: echo -1 > /sys/bus/pseudo/drivers/scsi_debug/add_host Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-11-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:09 +00:00
/* We may only change sdebug_max_queue when we have no shosts */
if (list_empty(&sdebug_host_list))
sdebug_max_queue = n;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
else
scsi: scsi_debug: Only allow sdebug_max_queue be modified when no shosts The shost->can_queue value is initially used to set per-HW queue context tag depth in the block layer. This ensures that the shost is not sent too many commands which it can deal with. However lowering sdebug_max_queue separately means that we can easily overload the shost, as in the following example: $ cat /sys/bus/pseudo/drivers/scsi_debug/max_queue 192 $ cat /sys/class/scsi_host/host0/can_queue 192 $ echo 100 > /sys/bus/pseudo/drivers/scsi_debug/max_queue $ cat /sys/class/scsi_host/host0/can_queue 192 $ fio --filename=/dev/sda --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=1200 --numjobs=10 --time_based --group_reporting --name=iops-test-job --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error iops-test-job: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 10 processes [ 111.269885] scsi_io_completion_action: 400 callbacks suppressed [ 111.269885] blk_print_req_error: 400 callbacks suppressed [ 111.269889] I/O error, dev sda, sector 440 op 0x0:(READ) flags 0x1200000 phys_seg 1 prio class 2 [ 111.269892] sd 0:0:0:0: [sda] tag#132 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 111.269897] sd 0:0:0:0: [sda] tag#132 CDB: Read(10) 28 00 00 00 01 68 00 00 08 00 [ 111.277058] I/O error, dev sda, sector 360 op 0x0:(READ) flags 0x1200000 phys_seg 1 prio class 2 [...] Ensure that this cannot happen by allowing sdebug_max_queue be modified only when we have no shosts. As such, any shost->can_queue value will match sdebug_max_queue, and sdebug_max_queue cannot be modified separately. Since retired_max_queue is no longer set, remove support. Continue to apply the restriction that sdebug_host_max_queue cannot be modified when sdebug_host_max_queue is set. Adding support for that would mean extra code, and no one has complained about this restriction previously. A command like the following may be used to remove a shost: echo -1 > /sys/bus/pseudo/drivers/scsi_debug/add_host Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-11-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:09 +00:00
count = -EBUSY;
mutex_unlock(&sdebug_host_list_mutex);
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(max_queue);
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
static ssize_t host_max_queue_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_host_max_queue);
}
static ssize_t no_rwlock_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_no_rwlock);
}
static ssize_t no_rwlock_store(struct device_driver *ddp, const char *buf, size_t count)
{
bool v;
if (kstrtobool(buf, &v))
return -EINVAL;
sdebug_no_rwlock = v;
return count;
}
static DRIVER_ATTR_RW(no_rwlock);
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
/*
* Since this is used for .can_queue, and we get the hc_idx tag from the bitmap
* in range [0, sdebug_host_max_queue), we can't change it.
*/
static DRIVER_ATTR_RO(host_max_queue);
static ssize_t no_uld_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_no_uld);
}
static DRIVER_ATTR_RO(no_uld);
static ssize_t scsi_level_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_scsi_level);
}
static DRIVER_ATTR_RO(scsi_level);
static ssize_t virtual_gb_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_virtual_gb);
}
static ssize_t virtual_gb_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
bool changed;
/* Ignore capacity change for ZBC drives for now */
if (sdeb_zbc_in_use)
return -ENOTSUPP;
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n >= 0)) {
changed = (sdebug_virtual_gb != n);
sdebug_virtual_gb = n;
sdebug_capacity = get_sdebug_capacity();
if (changed) {
struct sdebug_host_info *sdhp;
struct sdebug_dev_info *dp;
mutex_lock(&sdebug_host_list_mutex);
list_for_each_entry(sdhp, &sdebug_host_list,
host_list) {
list_for_each_entry(dp, &sdhp->dev_info_list,
dev_list) {
set_bit(SDEBUG_UA_CAPACITY_CHANGED,
dp->uas_bm);
}
}
mutex_unlock(&sdebug_host_list_mutex);
}
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(virtual_gb);
static ssize_t add_host_show(struct device_driver *ddp, char *buf)
{
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
/* absolute number of hosts currently active is what is shown */
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_num_hosts);
}
static ssize_t add_host_store(struct device_driver *ddp, const char *buf,
size_t count)
{
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
bool found;
unsigned long idx;
struct sdeb_store_info *sip;
bool want_phs = (sdebug_fake_rw == 0) && sdebug_per_host_store;
int delta_hosts;
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
if (sscanf(buf, "%d", &delta_hosts) != 1)
return -EINVAL;
if (delta_hosts > 0) {
do {
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
found = false;
if (want_phs) {
xa_for_each_marked(per_store_ap, idx, sip,
SDEB_XA_NOT_IN_USE) {
sdeb_most_recent_idx = (int)idx;
found = true;
break;
}
if (found) /* re-use case */
sdebug_add_host_helper((int)idx);
else
sdebug_do_add_host(true);
} else {
sdebug_do_add_host(false);
scsi: scsi_debug: Address races following module load When scsi_debug is loaded as a module with many (simulated) hosts, targets, and devices (LUs), modprobe can take a long time to return. Only a small amount of this time is spent in the scsi_debug_init(); the rest is other parts of the kernel reacting to to the appearance of new storage devices. As soon as scsi_debug_init() has completed the user space may call 'rmmod scsi_debug' and this was found to cause race problems as outlined here: https://bugzilla.kernel.org/show_bug.cgi?id=212337 To reliably generate this race a sysfs parameter called rm_all_hosts was added and the code was strengthened in this area. The main change was to make the count of scsi_debug hosts present an atomic. Then it was found that the handling of the existing add_host parameter needed the same strengthening. Further: 'echo -9999 > /sys/bus/pseudo/drivers/scsi_debug/add_host has the same effect as rm_all_hosts so rm_all_hosts was not needed. To inhibit a race between two invocations of writes to add_host, a mutex was added. Also address a possible race when rmmod is called but LUs are still being added. The logic to remove (all) hosts is rather crude: it works backwards down a linked lists of hosts. Any pending requests are terminated with DID_NO_CONNECT as are any new requests. In the case where not all hosts are being removed, the ones that remain may have lost requests as just outlined. The lowest numbered host (id) hosts will remain. Cc: Bart Van Assche <bvanassche@acm.org> Link: https://lore.kernel.org/r/20220109012853.301953-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-01-09 01:28:45 +00:00
}
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
} while (--delta_hosts);
} else if (delta_hosts < 0) {
do {
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
sdebug_do_remove_host(false);
} while (++delta_hosts);
}
return count;
}
static DRIVER_ATTR_RW(add_host);
static ssize_t vpd_use_hostno_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_vpd_use_hostno);
}
static ssize_t vpd_use_hostno_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n >= 0)) {
sdebug_vpd_use_hostno = n;
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(vpd_use_hostno);
static ssize_t statistics_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", (int)sdebug_statistics);
}
static ssize_t statistics_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
if ((count > 0) && (sscanf(buf, "%d", &n) == 1) && (n >= 0)) {
if (n > 0)
sdebug_statistics = true;
else {
clear_queue_stats();
sdebug_statistics = false;
}
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(statistics);
static ssize_t sector_size_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%u\n", sdebug_sector_size);
}
static DRIVER_ATTR_RO(sector_size);
static ssize_t submit_queues_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", submit_queues);
}
static DRIVER_ATTR_RO(submit_queues);
static ssize_t dix_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_dix);
}
static DRIVER_ATTR_RO(dix);
static ssize_t dif_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_dif);
}
static DRIVER_ATTR_RO(dif);
static ssize_t guard_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%u\n", sdebug_guard);
}
static DRIVER_ATTR_RO(guard);
static ssize_t ato_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_ato);
}
static DRIVER_ATTR_RO(ato);
static ssize_t map_show(struct device_driver *ddp, char *buf)
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
{
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
ssize_t count = 0;
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
if (!scsi_debug_lbp())
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
return scnprintf(buf, PAGE_SIZE, "0-%u\n",
sdebug_store_sectors);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
if (sdebug_fake_rw == 0 && !xa_empty(per_store_ap)) {
struct sdeb_store_info *sip = xa_load(per_store_ap, 0);
if (sip)
count = scnprintf(buf, PAGE_SIZE - 1, "%*pbl",
(int)map_size, sip->map_storep);
}
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
buf[count++] = '\n';
buf[count] = '\0';
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
return count;
}
static DRIVER_ATTR_RO(map);
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
static ssize_t random_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_random);
}
static ssize_t random_store(struct device_driver *ddp, const char *buf,
size_t count)
{
bool v;
if (kstrtobool(buf, &v))
return -EINVAL;
sdebug_random = v;
return count;
}
static DRIVER_ATTR_RW(random);
static ssize_t removable_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_removable ? 1 : 0);
}
static ssize_t removable_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n >= 0)) {
sdebug_removable = (n > 0);
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(removable);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
static ssize_t host_lock_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", !!sdebug_host_lock);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
}
/* N.B. sdebug_host_lock does nothing, kept for backward compatibility */
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
static ssize_t host_lock_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n >= 0)) {
sdebug_host_lock = (n > 0);
return count;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
}
return -EINVAL;
}
static DRIVER_ATTR_RW(host_lock);
static ssize_t strict_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", !!sdebug_strict);
}
static ssize_t strict_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int n;
if ((count > 0) && (1 == sscanf(buf, "%d", &n)) && (n >= 0)) {
sdebug_strict = (n > 0);
return count;
}
return -EINVAL;
}
static DRIVER_ATTR_RW(strict);
static ssize_t uuid_ctl_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", !!sdebug_uuid_ctl);
}
static DRIVER_ATTR_RO(uuid_ctl);
static ssize_t cdb_len_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdebug_cdb_len);
}
static ssize_t cdb_len_store(struct device_driver *ddp, const char *buf,
size_t count)
{
int ret, n;
ret = kstrtoint(buf, 0, &n);
if (ret)
return ret;
sdebug_cdb_len = n;
all_config_cdb_len();
return count;
}
static DRIVER_ATTR_RW(cdb_len);
static const char * const zbc_model_strs_a[] = {
[BLK_ZONED_NONE] = "none",
[BLK_ZONED_HA] = "host-aware",
[BLK_ZONED_HM] = "host-managed",
};
static const char * const zbc_model_strs_b[] = {
[BLK_ZONED_NONE] = "no",
[BLK_ZONED_HA] = "aware",
[BLK_ZONED_HM] = "managed",
};
static const char * const zbc_model_strs_c[] = {
[BLK_ZONED_NONE] = "0",
[BLK_ZONED_HA] = "1",
[BLK_ZONED_HM] = "2",
};
static int sdeb_zbc_model_str(const char *cp)
{
int res = sysfs_match_string(zbc_model_strs_a, cp);
if (res < 0) {
res = sysfs_match_string(zbc_model_strs_b, cp);
if (res < 0) {
res = sysfs_match_string(zbc_model_strs_c, cp);
if (res < 0)
return -EINVAL;
}
}
return res;
}
static ssize_t zbc_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%s\n",
zbc_model_strs_a[sdeb_zbc_model]);
}
static DRIVER_ATTR_RO(zbc);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
static ssize_t tur_ms_to_ready_show(struct device_driver *ddp, char *buf)
{
return scnprintf(buf, PAGE_SIZE, "%d\n", sdeb_tur_ms_to_ready);
}
static DRIVER_ATTR_RO(tur_ms_to_ready);
/* Note: The following array creates attribute files in the
/sys/bus/pseudo/drivers/scsi_debug directory. The advantage of these
files (over those found in the /sys/module/scsi_debug/parameters
directory) is that auxiliary actions can be triggered when an attribute
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
is changed. For example see: add_host_store() above.
*/
static struct attribute *sdebug_drv_attrs[] = {
&driver_attr_delay.attr,
&driver_attr_opts.attr,
&driver_attr_ptype.attr,
&driver_attr_dsense.attr,
&driver_attr_fake_rw.attr,
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
&driver_attr_host_max_queue.attr,
&driver_attr_no_lun_0.attr,
&driver_attr_num_tgts.attr,
&driver_attr_dev_size_mb.attr,
&driver_attr_num_parts.attr,
&driver_attr_every_nth.attr,
&driver_attr_lun_format.attr,
&driver_attr_max_luns.attr,
&driver_attr_max_queue.attr,
&driver_attr_no_rwlock.attr,
&driver_attr_no_uld.attr,
&driver_attr_scsi_level.attr,
&driver_attr_virtual_gb.attr,
&driver_attr_add_host.attr,
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
&driver_attr_per_host_store.attr,
&driver_attr_vpd_use_hostno.attr,
&driver_attr_sector_size.attr,
&driver_attr_statistics.attr,
&driver_attr_submit_queues.attr,
&driver_attr_dix.attr,
&driver_attr_dif.attr,
&driver_attr_guard.attr,
&driver_attr_ato.attr,
&driver_attr_map.attr,
&driver_attr_random.attr,
&driver_attr_removable.attr,
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
&driver_attr_host_lock.attr,
&driver_attr_ndelay.attr,
&driver_attr_strict.attr,
&driver_attr_uuid_ctl.attr,
&driver_attr_cdb_len.attr,
&driver_attr_tur_ms_to_ready.attr,
&driver_attr_zbc.attr,
NULL,
};
ATTRIBUTE_GROUPS(sdebug_drv);
static struct device *pseudo_primary;
static int __init scsi_debug_init(void)
{
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
bool want_store = (sdebug_fake_rw == 0);
unsigned long sz;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
int k, ret, hosts_to_add;
int idx = -1;
if (sdebug_ndelay >= 1000 * 1000 * 1000) {
pr_warn("ndelay must be less than 1 second, ignored\n");
sdebug_ndelay = 0;
} else if (sdebug_ndelay > 0)
sdebug_jdelay = JDELAY_OVERRIDDEN;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
switch (sdebug_sector_size) {
case 512:
case 1024:
case 2048:
case 4096:
break;
default:
pr_err("invalid sector_size %d\n", sdebug_sector_size);
return -EINVAL;
}
switch (sdebug_dif) {
case T10_PI_TYPE0_PROTECTION:
break;
case T10_PI_TYPE1_PROTECTION:
case T10_PI_TYPE2_PROTECTION:
case T10_PI_TYPE3_PROTECTION:
have_dif_prot = true;
break;
default:
pr_err("dif must be 0, 1, 2 or 3\n");
return -EINVAL;
}
if (sdebug_num_tgts < 0) {
pr_err("num_tgts must be >= 0\n");
return -EINVAL;
}
if (sdebug_guard > 1) {
pr_err("guard must be 0 or 1\n");
return -EINVAL;
}
if (sdebug_ato > 1) {
pr_err("ato must be 0 or 1\n");
return -EINVAL;
}
if (sdebug_physblk_exp > 15) {
pr_err("invalid physblk_exp %u\n", sdebug_physblk_exp);
return -EINVAL;
}
sdebug_lun_am = sdebug_lun_am_i;
if (sdebug_lun_am > SAM_LUN_AM_FLAT) {
pr_warn("Invalid LUN format %u, using default\n", (int)sdebug_lun_am);
sdebug_lun_am = SAM_LUN_AM_PERIPHERAL;
}
if (sdebug_max_luns > 256) {
if (sdebug_max_luns > 16384) {
pr_warn("max_luns can be no more than 16384, use default\n");
sdebug_max_luns = DEF_MAX_LUNS;
}
sdebug_lun_am = SAM_LUN_AM_FLAT;
}
if (sdebug_lowest_aligned > 0x3fff) {
pr_err("lowest_aligned too big: %u\n", sdebug_lowest_aligned);
return -EINVAL;
}
if (submit_queues < 1) {
pr_err("submit_queues must be 1 or more\n");
return -EINVAL;
}
scsi: scsi_debug: Add check for sdebug_max_queue during module init sdebug_max_queue should not exceed SDEBUG_CANQUEUE, otherwise crashes like this can be triggered by passing an out-of-range value: Hardware name: Huawei D06 /D06, BIOS Hisilicon D06 UEFI RC0 - V1.16.01 03/15/2019 pstate: 20400009 (nzCv daif +PAN -UAO BTYPE=--) pc : schedule_resp+0x2a4/0xa70 [scsi_debug] lr : schedule_resp+0x52c/0xa70 [scsi_debug] sp : ffff800022ab36f0 x29: ffff800022ab36f0 x28: ffff0023a935a610 x27: ffff800008e0a648 x26: 0000000000000003 x25: ffff0023e84f3200 x24: 00000000003d0900 x23: 0000000000000000 x22: 0000000000000000 x21: ffff0023be60a320 x20: ffff0023be60b538 x19: ffff800008e13000 x18: 0000000000000000 x17: 0000000000000000 x16: 0000000000000000 x15: 0000000000000000 x14: 0000000000000000 x13: 0000000000000000 x12: 0000000000000000 x11: 0000000000000000 x10: 0000000000000000 x9 : 0000000000000001 x8 : 0000000000000000 x7 : 0000000000000000 x6 : 00000000000000c1 x5 : 0000020000200000 x4 : dead0000000000ff x3 : 0000000000000200 x2 : 0000000000000200 x1 : ffff800008e13d88 x0 : 0000000000000000 Call trace: schedule_resp+0x2a4/0xa70 [scsi_debug] scsi_debug_queuecommand+0x2c4/0x9e0 [scsi_debug] scsi_queue_rq+0x698/0x840 __blk_mq_try_issue_directly+0x108/0x228 blk_mq_request_issue_directly+0x58/0x98 blk_mq_try_issue_list_directly+0x5c/0xf0 blk_mq_sched_insert_requests+0x18c/0x200 blk_mq_flush_plug_list+0x11c/0x190 blk_flush_plug_list+0xdc/0x110 blk_finish_plug+0x38/0x210 blkdev_direct_IO+0x450/0x4d8 generic_file_read_iter+0x84/0x180 blkdev_read_iter+0x3c/0x50 aio_read+0xc0/0x170 io_submit_one+0x5c8/0xc98 __arm64_sys_io_submit+0x1b0/0x258 el0_svc_common.constprop.3+0x68/0x170 do_el0_svc+0x24/0x90 el0_sync_handler+0x13c/0x1a8 el0_sync+0x158/0x180 Code: 528847e0 72a001e0 6b00003f 540018cd (3941c340) In addition, it should not be less than 1. So add checks for these, and fail the module init for those cases. [mkp: changed if condition to match error message] Link: https://lore.kernel.org/r/1594297400-24756-2-git-send-email-john.garry@huawei.com Fixes: c483739430f1 ("scsi_debug: add multiple queue support") Reviewed-by: Ming Lei <ming.lei@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:19 +00:00
if ((sdebug_max_queue > SDEBUG_CANQUEUE) || (sdebug_max_queue < 1)) {
pr_err("max_queue must be in range [1, %d]\n", SDEBUG_CANQUEUE);
return -EINVAL;
}
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
if ((sdebug_host_max_queue > SDEBUG_CANQUEUE) ||
(sdebug_host_max_queue < 0)) {
pr_err("host_max_queue must be in range [0 %d]\n",
SDEBUG_CANQUEUE);
return -EINVAL;
}
if (sdebug_host_max_queue &&
(sdebug_max_queue != sdebug_host_max_queue)) {
sdebug_max_queue = sdebug_host_max_queue;
pr_warn("fixing max submit queue depth to host max queue depth, %d\n",
sdebug_max_queue);
}
/*
* check for host managed zoned block device specified with
* ptype=0x14 or zbc=XXX.
*/
if (sdebug_ptype == TYPE_ZBC) {
sdeb_zbc_model = BLK_ZONED_HM;
} else if (sdeb_zbc_model_s && *sdeb_zbc_model_s) {
k = sdeb_zbc_model_str(sdeb_zbc_model_s);
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
if (k < 0)
return k;
sdeb_zbc_model = k;
switch (sdeb_zbc_model) {
case BLK_ZONED_NONE:
case BLK_ZONED_HA:
sdebug_ptype = TYPE_DISK;
break;
case BLK_ZONED_HM:
sdebug_ptype = TYPE_ZBC;
break;
default:
pr_err("Invalid ZBC model\n");
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
return -EINVAL;
}
}
if (sdeb_zbc_model != BLK_ZONED_NONE) {
sdeb_zbc_in_use = true;
if (sdebug_dev_size_mb == DEF_DEV_SIZE_PRE_INIT)
sdebug_dev_size_mb = DEF_ZBC_DEV_SIZE_MB;
}
if (sdebug_dev_size_mb == DEF_DEV_SIZE_PRE_INIT)
sdebug_dev_size_mb = DEF_DEV_SIZE_MB;
if (sdebug_dev_size_mb < 1)
sdebug_dev_size_mb = 1; /* force minimum 1 MB ramdisk */
sz = (unsigned long)sdebug_dev_size_mb * 1048576;
sdebug_store_sectors = sz / sdebug_sector_size;
sdebug_capacity = get_sdebug_capacity();
/* play around with geometry, don't waste too much on track 0 */
sdebug_heads = 8;
sdebug_sectors_per = 32;
if (sdebug_dev_size_mb >= 256)
sdebug_heads = 64;
else if (sdebug_dev_size_mb >= 16)
sdebug_heads = 32;
sdebug_cylinders_per = (unsigned long)sdebug_capacity /
(sdebug_sectors_per * sdebug_heads);
if (sdebug_cylinders_per >= 1024) {
/* other LLDs do this; implies >= 1GB ram disk ... */
sdebug_heads = 255;
sdebug_sectors_per = 63;
sdebug_cylinders_per = (unsigned long)sdebug_capacity /
(sdebug_sectors_per * sdebug_heads);
}
if (scsi_debug_lbp()) {
sdebug_unmap_max_blocks =
clamp(sdebug_unmap_max_blocks, 0U, 0xffffffffU);
sdebug_unmap_max_desc =
clamp(sdebug_unmap_max_desc, 0U, 256U);
sdebug_unmap_granularity =
clamp(sdebug_unmap_granularity, 1U, 0xffffffffU);
if (sdebug_unmap_alignment &&
sdebug_unmap_granularity <=
sdebug_unmap_alignment) {
pr_err("ERR: unmap_granularity <= unmap_alignment\n");
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
return -EINVAL;
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
}
xa_init_flags(per_store_ap, XA_FLAGS_ALLOC | XA_FLAGS_LOCK_IRQ);
if (want_store) {
idx = sdebug_add_store();
scsi: scsi_debug: Drop sdebug_queue It's easy to get scsi_debug to error on throughput testing when we have multiple shosts: $ lsscsi [7:0:0:0] disk Linux scsi_debug 0191 [0:0:0:0] disk Linux scsi_debug 0191 $ fio --filename=/dev/sda --filename=/dev/sdb --direct=1 --rw=read --bs=4k --iodepth=256 --runtime=60 --numjobs=40 --time_based --name=jpg --eta-newline=1 --readonly --ioengine=io_uring --hipri --exitall_on_error jpg: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=io_uring, iodepth=256 ... fio-3.28 Starting 40 processes [ 27.521809] hrtimer: interrupt took 33067 ns [ 27.904660] sd 7:0:0:0: [sdb] tag#171 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s [ 27.904660] sd 0:0:0:0: [sda] tag#58 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s fio: io_u error [ 27.904667] sd 0:0:0:0: [sda] tag#58 CDB: Read(10) 28 00 00 00 27 00 00 01 18 00 on file /dev/sda[ 27.904670] sd 0:0:0:0: [sda] tag#62 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK cmd_age=0s The issue is related to how the driver manages submit queues and tags. A single array of submit queues - sdebug_q_arr - with its own set of tags is shared among all shosts. As such, for occasions when we have more than one shost it is possible to overload the submit queues and run out of tags. The struct sdebug_queue is to manage tags and hold the associated queued command entry pointer (for that tag). Since the tagset iters are now used for functions like sdebug_blk_mq_poll(), there is no need to manage these queues. Indeed, blk-mq already provides what we need for managing tags and queues. Drop sdebug_queue and all its usage in the driver. Signed-off-by: John Garry <john.g.garry@oracle.com> Link: https://lore.kernel.org/r/20230327074310.1862889-12-john.g.garry@oracle.com Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2023-03-27 07:43:10 +00:00
if (idx < 0)
return idx;
[SCSI] scsi_debug: Thin provisioning support This version fixes 64-bit modulo on 32-bit as well as inadvertent map updates when TP was disabled. Implement support for thin provisioning in scsi_debug. No actual memory de-allocation is taking place. The intent is to emulate a thinly provisioned storage device, not to be one. There are four new module options: - unmap_granularity specifies the granularity at which to track mapped blocks (specified in number of logical blocks). 2048 (1 MB) is a realistic value for disk arrays although some may have a finer granularity. - unmap_alignment specifies the first LBA which is naturally aligned on an unmap_granularity boundary. - unmap_max_desc specifies the maximum number of ranges that can be unmapped using one UNMAP command. If this is 0, only WRITE SAME is supported and UNMAP will cause a check condition. - unmap_max_blocks specifies the maximum number of blocks that can be unmapped using a single UNMAP command. Default is 0xffffffff. These parameters are reported in the new and extended block limits VPD. If unmap_granularity is specified the device is tagged as thin provisioning capable in READ CAPACITY(16). A bitmap is allocated to track whether blocks are mapped or not. A WRITE request will cause a block to be mapped. So will WRITE SAME unless the UNMAP bit is set. Blocks can be unmapped using either WRITE SAME or UNMAP. No accounting is done to track partial blocks. This means that only whole blocks will be marked free. This is how the array people tell me their firmwares work. GET LBA STATUS is also supported. This command reports whether a block is mapped or not, and how long the adjoining mapped/unmapped extent is. The block allocation bitmap can also be viewed from user space via: /sys/bus/pseudo/drivers/scsi_debug/map Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: James Bottomley <James.Bottomley@suse.de>
2009-10-15 18:45:27 +00:00
}
pseudo_primary = root_device_register("pseudo_0");
if (IS_ERR(pseudo_primary)) {
pr_warn("root_device_register() error\n");
ret = PTR_ERR(pseudo_primary);
goto free_vm;
}
ret = bus_register(&pseudo_lld_bus);
if (ret < 0) {
pr_warn("bus_register error: %d\n", ret);
goto dev_unreg;
}
ret = driver_register(&sdebug_driverfs_driver);
if (ret < 0) {
pr_warn("driver_register error: %d\n", ret);
goto bus_unreg;
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
hosts_to_add = sdebug_add_host;
sdebug_add_host = 0;
queued_cmd_cache = KMEM_CACHE(sdebug_queued_cmd, SLAB_HWCACHE_ALIGN);
if (!queued_cmd_cache) {
ret = -ENOMEM;
goto driver_unreg;
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
for (k = 0; k < hosts_to_add; k++) {
if (want_store && k == 0) {
ret = sdebug_add_host_helper(idx);
if (ret < 0) {
pr_err("add_host_helper k=%d, error=%d\n",
k, -ret);
break;
}
} else {
ret = sdebug_do_add_host(want_store &&
sdebug_per_host_store);
if (ret < 0) {
pr_err("add_host k=%d error=%d\n", k, -ret);
break;
}
}
}
if (sdebug_verbose)
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
pr_info("built %d host(s)\n", sdebug_num_hosts);
return 0;
driver_unreg:
driver_unregister(&sdebug_driverfs_driver);
bus_unreg:
bus_unregister(&pseudo_lld_bus);
dev_unreg:
root_device_unregister(pseudo_primary);
free_vm:
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
sdebug_erase_store(idx, NULL);
return ret;
}
static void __exit scsi_debug_exit(void)
{
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
int k = sdebug_num_hosts;
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
for (; k; k--)
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
sdebug_do_remove_host(true);
kmem_cache_destroy(queued_cmd_cache);
driver_unregister(&sdebug_driverfs_driver);
bus_unregister(&pseudo_lld_bus);
root_device_unregister(pseudo_primary);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
sdebug_erase_all_stores(false);
xa_destroy(per_store_ap);
}
device_initcall(scsi_debug_init);
module_exit(scsi_debug_exit);
static void sdebug_release_adapter(struct device *dev)
{
struct sdebug_host_info *sdbg_host;
sdbg_host = dev_to_sdebug_host(dev);
kfree(sdbg_host);
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
/* idx must be valid, if sip is NULL then it will be obtained using idx */
static void sdebug_erase_store(int idx, struct sdeb_store_info *sip)
{
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
if (idx < 0)
return;
if (!sip) {
if (xa_empty(per_store_ap))
return;
sip = xa_load(per_store_ap, idx);
if (!sip)
return;
}
vfree(sip->map_storep);
vfree(sip->dif_storep);
vfree(sip->storep);
xa_erase(per_store_ap, idx);
kfree(sip);
}
/* Assume apart_from_first==false only in shutdown case. */
static void sdebug_erase_all_stores(bool apart_from_first)
{
unsigned long idx;
struct sdeb_store_info *sip = NULL;
xa_for_each(per_store_ap, idx, sip) {
if (apart_from_first)
apart_from_first = false;
else
sdebug_erase_store(idx, sip);
}
if (apart_from_first)
sdeb_most_recent_idx = sdeb_first_idx;
}
/*
* Returns store xarray new element index (idx) if >=0 else negated errno.
* Limit the number of stores to 65536.
*/
static int sdebug_add_store(void)
{
int res;
u32 n_idx;
unsigned long iflags;
unsigned long sz = (unsigned long)sdebug_dev_size_mb * 1048576;
struct sdeb_store_info *sip = NULL;
struct xa_limit xal = { .max = 1 << 16, .min = 0 };
sip = kzalloc(sizeof(*sip), GFP_KERNEL);
if (!sip)
return -ENOMEM;
xa_lock_irqsave(per_store_ap, iflags);
res = __xa_alloc(per_store_ap, &n_idx, sip, xal, GFP_ATOMIC);
if (unlikely(res < 0)) {
xa_unlock_irqrestore(per_store_ap, iflags);
kfree(sip);
pr_warn("%s: xa_alloc() errno=%d\n", __func__, -res);
return res;
}
sdeb_most_recent_idx = n_idx;
if (sdeb_first_idx < 0)
sdeb_first_idx = n_idx;
xa_unlock_irqrestore(per_store_ap, iflags);
res = -ENOMEM;
sip->storep = vzalloc(sz);
if (!sip->storep) {
pr_err("user data oom\n");
goto err;
}
if (sdebug_num_parts > 0)
sdebug_build_parts(sip->storep, sz);
/* DIF/DIX: what T10 calls Protection Information (PI) */
if (sdebug_dix) {
int dif_size;
dif_size = sdebug_store_sectors * sizeof(struct t10_pi_tuple);
sip->dif_storep = vmalloc(dif_size);
pr_info("dif_storep %u bytes @ %pK\n", dif_size,
sip->dif_storep);
if (!sip->dif_storep) {
pr_err("DIX oom\n");
goto err;
}
memset(sip->dif_storep, 0xff, dif_size);
}
/* Logical Block Provisioning */
if (scsi_debug_lbp()) {
map_size = lba_to_map_index(sdebug_store_sectors - 1) + 1;
sip->map_storep = vmalloc(array_size(sizeof(long),
BITS_TO_LONGS(map_size)));
pr_info("%lu provisioning blocks\n", map_size);
if (!sip->map_storep) {
pr_err("LBP map oom\n");
goto err;
}
bitmap_zero(sip->map_storep, map_size);
/* Map first 1KB for partition table */
if (sdebug_num_parts)
map_region(sip, 0, 2);
}
rwlock_init(&sip->macc_lck);
return (int)n_idx;
err:
sdebug_erase_store((int)n_idx, sip);
pr_warn("%s: failed, errno=%d\n", __func__, -res);
return res;
}
static int sdebug_add_host_helper(int per_host_idx)
{
int k, devs_per_host, idx;
int error = -ENOMEM;
struct sdebug_host_info *sdbg_host;
struct sdebug_dev_info *sdbg_devinfo, *tmp;
sdbg_host = kzalloc(sizeof(*sdbg_host), GFP_KERNEL);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
if (!sdbg_host)
return -ENOMEM;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
idx = (per_host_idx < 0) ? sdeb_first_idx : per_host_idx;
if (xa_get_mark(per_store_ap, idx, SDEB_XA_NOT_IN_USE))
xa_clear_mark(per_store_ap, idx, SDEB_XA_NOT_IN_USE);
sdbg_host->si_idx = idx;
INIT_LIST_HEAD(&sdbg_host->dev_info_list);
devs_per_host = sdebug_num_tgts * sdebug_max_luns;
for (k = 0; k < devs_per_host; k++) {
sdbg_devinfo = sdebug_device_create(sdbg_host, GFP_KERNEL);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
if (!sdbg_devinfo)
goto clean;
}
mutex_lock(&sdebug_host_list_mutex);
list_add_tail(&sdbg_host->host_list, &sdebug_host_list);
mutex_unlock(&sdebug_host_list_mutex);
sdbg_host->dev.bus = &pseudo_lld_bus;
sdbg_host->dev.parent = pseudo_primary;
sdbg_host->dev.release = &sdebug_release_adapter;
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
dev_set_name(&sdbg_host->dev, "adapter%d", sdebug_num_hosts);
error = device_register(&sdbg_host->dev);
if (error) {
mutex_lock(&sdebug_host_list_mutex);
list_del(&sdbg_host->host_list);
mutex_unlock(&sdebug_host_list_mutex);
goto clean;
}
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
++sdebug_num_hosts;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
return 0;
clean:
list_for_each_entry_safe(sdbg_devinfo, tmp, &sdbg_host->dev_info_list,
dev_list) {
list_del(&sdbg_devinfo->dev_list);
kfree(sdbg_devinfo->zstate);
kfree(sdbg_devinfo);
}
if (sdbg_host->dev.release)
put_device(&sdbg_host->dev);
else
kfree(sdbg_host);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
pr_warn("%s: failed, errno=%d\n", __func__, -error);
return error;
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
static int sdebug_do_add_host(bool mk_new_store)
{
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
int ph_idx = sdeb_most_recent_idx;
if (mk_new_store) {
ph_idx = sdebug_add_store();
if (ph_idx < 0)
return ph_idx;
}
return sdebug_add_host_helper(ph_idx);
}
static void sdebug_do_remove_host(bool the_end)
{
int idx = -1;
struct sdebug_host_info *sdbg_host = NULL;
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
struct sdebug_host_info *sdbg_host2;
mutex_lock(&sdebug_host_list_mutex);
if (!list_empty(&sdebug_host_list)) {
sdbg_host = list_entry(sdebug_host_list.prev,
struct sdebug_host_info, host_list);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
idx = sdbg_host->si_idx;
}
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
if (!the_end && idx >= 0) {
bool unique = true;
list_for_each_entry(sdbg_host2, &sdebug_host_list, host_list) {
if (sdbg_host2 == sdbg_host)
continue;
if (idx == sdbg_host2->si_idx) {
unique = false;
break;
}
}
if (unique) {
xa_set_mark(per_store_ap, idx, SDEB_XA_NOT_IN_USE);
if (idx == sdeb_most_recent_idx)
--sdeb_most_recent_idx;
}
}
if (sdbg_host)
list_del(&sdbg_host->host_list);
mutex_unlock(&sdebug_host_list_mutex);
if (!sdbg_host)
return;
device_unregister(&sdbg_host->dev);
Revert "scsi: scsi_debug: Address races following module load" Revert the patch mentioned in the subject since it blocks I/O after module unload has started while this is a legitimate use case. For e.g. blktests test case srp/001 that patch causes a command timeout to be triggered for the following call stack: __schedule+0x4c3/0xd20 schedule+0x82/0x110 schedule_timeout+0x122/0x200 io_schedule_timeout+0x7b/0xc0 __wait_for_common+0x2bc/0x380 wait_for_completion_io_timeout+0x1d/0x20 blk_execute_rq+0x1db/0x200 __scsi_execute+0x1fb/0x310 sd_sync_cache+0x155/0x2c0 [sd_mod] sd_shutdown+0xbb/0x190 [sd_mod] sd_remove+0x5b/0x80 [sd_mod] device_remove+0x9a/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 __scsi_remove_device+0x168/0x1a0 scsi_forget_host+0xa8/0xb0 scsi_remove_host+0x9b/0x150 sdebug_driver_remove+0x3d/0x140 [scsi_debug] device_remove+0x6f/0xb0 device_release_driver_internal+0x2c5/0x360 device_release_driver+0x12/0x20 bus_remove_device+0x1aa/0x270 device_del+0x2d4/0x640 device_unregister+0x18/0x70 sdebug_do_remove_host+0x138/0x180 [scsi_debug] scsi_debug_exit+0x45/0xd5 [scsi_debug] __do_sys_delete_module.constprop.0+0x210/0x320 __x64_sys_delete_module+0x1f/0x30 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Link: https://lore.kernel.org/r/20220409043704.28573-1-bvanassche@acm.org Fixes: 2aad3cd85370 ("scsi: scsi_debug: Address races following module load") Cc: Douglas Gilbert <dgilbert@interlog.com> Cc: Yi Zhang <yi.zhang@redhat.com> Cc: Bob Pearson <rpearsonhpe@gmail.com> Reported-by: Yi Zhang <yi.zhang@redhat.com> Tested-by: Yi Zhang <yi.zhang@redhat.com> Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2022-04-09 04:37:03 +00:00
--sdebug_num_hosts;
}
static int sdebug_change_qdepth(struct scsi_device *sdev, int qdepth)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
{
struct sdebug_dev_info *devip = sdev->hostdata;
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
if (!devip)
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
return -ENODEV;
mutex_lock(&sdebug_host_list_mutex);
block_unblock_all_queues(true);
if (qdepth > SDEBUG_CANQUEUE) {
qdepth = SDEBUG_CANQUEUE;
pr_warn("%s: requested qdepth [%d] exceeds canqueue [%d], trim\n", __func__,
qdepth, SDEBUG_CANQUEUE);
}
if (qdepth < 1)
qdepth = 1;
if (qdepth != sdev->queue_depth)
scsi_change_queue_depth(sdev, qdepth);
block_unblock_all_queues(false);
mutex_unlock(&sdebug_host_list_mutex);
if (SDEBUG_OPT_Q_NOISE & sdebug_opts)
sdev_printk(KERN_INFO, sdev, "%s: qdepth=%d\n", __func__, qdepth);
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
return sdev->queue_depth;
}
static bool fake_timeout(struct scsi_cmnd *scp)
{
if (0 == (atomic_read(&sdebug_cmnd_count) % abs(sdebug_every_nth))) {
if (sdebug_every_nth < -1)
sdebug_every_nth = -1;
if (SDEBUG_OPT_TIMEOUT & sdebug_opts)
return true; /* ignore command causing timeout */
else if (SDEBUG_OPT_MAC_TIMEOUT & sdebug_opts &&
scsi_medium_access_command(scp))
return true; /* time out reads and writes */
}
return false;
}
/* Response to TUR or media access command when device stopped */
static int resp_not_ready(struct scsi_cmnd *scp, struct sdebug_dev_info *devip)
{
int stopped_state;
u64 diff_ns = 0;
ktime_t now_ts = ktime_get_boottime();
struct scsi_device *sdp = scp->device;
stopped_state = atomic_read(&devip->stopped);
if (stopped_state == 2) {
if (ktime_to_ns(now_ts) > ktime_to_ns(devip->create_ts)) {
diff_ns = ktime_to_ns(ktime_sub(now_ts, devip->create_ts));
if (diff_ns >= ((u64)sdeb_tur_ms_to_ready * 1000000)) {
/* tur_ms_to_ready timer extinguished */
atomic_set(&devip->stopped, 0);
return 0;
}
}
mk_sense_buffer(scp, NOT_READY, LOGICAL_UNIT_NOT_READY, 0x1);
if (sdebug_verbose)
sdev_printk(KERN_INFO, sdp,
"%s: Not ready: in process of becoming ready\n", my_name);
if (scp->cmnd[0] == TEST_UNIT_READY) {
u64 tur_nanosecs_to_ready = (u64)sdeb_tur_ms_to_ready * 1000000;
if (diff_ns <= tur_nanosecs_to_ready)
diff_ns = tur_nanosecs_to_ready - diff_ns;
else
diff_ns = tur_nanosecs_to_ready;
/* As per 20-061r2 approved for spc6 by T10 on 20200716 */
do_div(diff_ns, 1000000); /* diff_ns becomes milliseconds */
scsi_set_sense_information(scp->sense_buffer, SCSI_SENSE_BUFFERSIZE,
diff_ns);
return check_condition_result;
}
}
mk_sense_buffer(scp, NOT_READY, LOGICAL_UNIT_NOT_READY, 0x2);
if (sdebug_verbose)
sdev_printk(KERN_INFO, sdp, "%s: Not ready: initializing command required\n",
my_name);
return check_condition_result;
}
static void sdebug_map_queues(struct Scsi_Host *shost)
{
int i, qoff;
if (shost->nr_hw_queues == 1)
return;
for (i = 0, qoff = 0; i < HCTX_MAX_TYPES; i++) {
struct blk_mq_queue_map *map = &shost->tag_set.map[i];
map->nr_queues = 0;
if (i == HCTX_TYPE_DEFAULT)
map->nr_queues = submit_queues - poll_queues;
else if (i == HCTX_TYPE_POLL)
map->nr_queues = poll_queues;
if (!map->nr_queues) {
BUG_ON(i == HCTX_TYPE_DEFAULT);
continue;
}
map->queue_offset = qoff;
blk_mq_map_queues(map);
qoff += map->nr_queues;
}
}
struct sdebug_blk_mq_poll_data {
unsigned int queue_num;
int *num_entries;
};
/*
* We don't handle aborted commands here, but it does not seem possible to have
* aborted polled commands from schedule_resp()
*/
static bool sdebug_blk_mq_poll_iter(struct request *rq, void *opaque)
{
struct sdebug_blk_mq_poll_data *data = opaque;
struct scsi_cmnd *cmd = blk_mq_rq_to_pdu(rq);
struct sdebug_scsi_cmd *sdsc = scsi_cmd_priv(cmd);
struct sdebug_defer *sd_dp;
u32 unique_tag = blk_mq_unique_tag(rq);
u16 hwq = blk_mq_unique_tag_to_hwq(unique_tag);
struct sdebug_queued_cmd *sqcp;
unsigned long flags;
int queue_num = data->queue_num;
ktime_t time;
/* We're only interested in one queue for this iteration */
if (hwq != queue_num)
return true;
/* Subsequent checks would fail if this failed, but check anyway */
if (!test_bit(SCMD_STATE_INFLIGHT, &cmd->state))
return true;
time = ktime_get_boottime();
spin_lock_irqsave(&sdsc->lock, flags);
sqcp = TO_QUEUED_CMD(cmd);
if (!sqcp) {
spin_unlock_irqrestore(&sdsc->lock, flags);
return true;
}
sd_dp = &sqcp->sd_dp;
if (READ_ONCE(sd_dp->defer_t) != SDEB_DEFER_POLL) {
spin_unlock_irqrestore(&sdsc->lock, flags);
return true;
}
if (time < sd_dp->cmpl_ts) {
spin_unlock_irqrestore(&sdsc->lock, flags);
return true;
}
ASSIGN_QUEUED_CMD(cmd, NULL);
spin_unlock_irqrestore(&sdsc->lock, flags);
if (sdebug_statistics) {
atomic_inc(&sdebug_completions);
if (raw_smp_processor_id() != sd_dp->issuing_cpu)
atomic_inc(&sdebug_miss_cpus);
}
sdebug_free_queued_cmd(sqcp);
scsi_done(cmd); /* callback to mid level */
(*data->num_entries)++;
return true;
}
static int sdebug_blk_mq_poll(struct Scsi_Host *shost, unsigned int queue_num)
{
int num_entries = 0;
struct sdebug_blk_mq_poll_data data = {
.queue_num = queue_num,
.num_entries = &num_entries,
};
blk_mq_tagset_busy_iter(&shost->tag_set, sdebug_blk_mq_poll_iter,
&data);
if (num_entries > 0)
atomic_add(num_entries, &sdeb_mq_poll_count);
return num_entries;
}
static int scsi_debug_queuecommand(struct Scsi_Host *shost,
struct scsi_cmnd *scp)
{
u8 sdeb_i;
struct scsi_device *sdp = scp->device;
const struct opcode_info_t *oip;
const struct opcode_info_t *r_oip;
struct sdebug_dev_info *devip;
u8 *cmd = scp->cmnd;
int (*r_pfp)(struct scsi_cmnd *, struct sdebug_dev_info *);
int (*pfp)(struct scsi_cmnd *, struct sdebug_dev_info *) = NULL;
int k, na;
int errsts = 0;
u64 lun_index = sdp->lun & 0x3FFF;
u32 flags;
u16 sa;
u8 opcode = cmd[0];
bool has_wlun_rl;
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
bool inject_now;
scsi_set_resid(scp, 0);
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
if (sdebug_statistics) {
atomic_inc(&sdebug_cmnd_count);
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
inject_now = inject_on_this_cmd();
} else {
inject_now = false;
}
if (unlikely(sdebug_verbose &&
!(SDEBUG_OPT_NO_CDB_NOISE & sdebug_opts))) {
char b[120];
int n, len, sb;
len = scp->cmd_len;
sb = (int)sizeof(b);
if (len > 32)
strcpy(b, "too long, over 32 bytes");
else {
for (k = 0, n = 0; k < len && n < sb; ++k)
n += scnprintf(b + n, sb - n, "%02x ",
(u32)cmd[k]);
}
sdev_printk(KERN_INFO, sdp, "%s: tag=%#x, cmd %s\n", my_name,
blk_mq_unique_tag(scsi_cmd_to_rq(scp)), b);
}
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
if (unlikely(inject_now && (sdebug_opts & SDEBUG_OPT_HOST_BUSY)))
return SCSI_MLQUEUE_HOST_BUSY;
has_wlun_rl = (sdp->lun == SCSI_W_LUN_REPORT_LUNS);
if (unlikely(lun_index >= sdebug_max_luns && !has_wlun_rl))
goto err_out;
sdeb_i = opcode_ind_arr[opcode]; /* fully mapped */
oip = &opcode_info_arr[sdeb_i]; /* safe if table consistent */
devip = (struct sdebug_dev_info *)sdp->hostdata;
if (unlikely(!devip)) {
devip = find_build_dev_info(sdp);
if (NULL == devip)
goto err_out;
}
scsi: scsi_debug: every_nth triggered error injection This patch simplifies, or at least makes more consistent, the way setting the every_nth parameter injects errors. Here is a list of 'opts' flags and in which cases they inject errors when abs(every_nth)%command_count == 0 is reached: - OPT_RECOVERED_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIF_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_DIX_ERR: issued on READ(*)s, WRITE(*)s and WRITE_SCATTEREDs - OPT_SHORT_TRANSFER: issued on READ(*)s - OPT_TRANSPORT_ERR: issued on all commands - OPT_CMD_ABORT: issued on all commands The other uses of every_nth were not modified. Previously if, for example, OPT_SHORT_TRANSFER was armed then if (abs(every_nth) % command_count == 0) occurred during a command that was _not_ a READ, then no error injection occurred. This behaviour puzzled several testers. Now a global "inject_pending" flag is set and the _next_ READ will get hit and that flag is cleared. OPT_RECOVERED_ERR, OPT_DIF_ERR and OPT_DIX_ERR have similar behaviour. A downside of this is that there might be a hang-over pending injection that gets triggered by a following test. Also expand the every_nth runtime parameter so that it can take hex value (i.e. with a leading '0x') as well as a decimal value. Now both the 'opts' and the 'every_nth' runtime parameters can take hexadecimal values. Link: https://lore.kernel.org/r/20200712182927.72044-2-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-12 18:29:26 +00:00
if (unlikely(inject_now && !atomic_read(&sdeb_inject_pending)))
atomic_set(&sdeb_inject_pending, 1);
na = oip->num_attached;
r_pfp = oip->pfp;
if (na) { /* multiple commands with this opcode */
r_oip = oip;
if (FF_SA & r_oip->flags) {
if (F_SA_LOW & oip->flags)
sa = 0x1f & cmd[1];
else
sa = get_unaligned_be16(cmd + 8);
for (k = 0; k <= na; oip = r_oip->arrp + k++) {
if (opcode == oip->opcode && sa == oip->sa)
break;
}
} else { /* since no service action only check opcode */
for (k = 0; k <= na; oip = r_oip->arrp + k++) {
if (opcode == oip->opcode)
break;
}
}
if (k > na) {
if (F_SA_LOW & r_oip->flags)
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 1, 4);
else if (F_SA_HIGH & r_oip->flags)
mk_sense_invalid_fld(scp, SDEB_IN_CDB, 8, 7);
else
mk_sense_invalid_opcode(scp);
goto check_cond;
}
} /* else (when na==0) we assume the oip is a match */
flags = oip->flags;
if (unlikely(F_INV_OP & flags)) {
mk_sense_invalid_opcode(scp);
goto check_cond;
}
if (unlikely(has_wlun_rl && !(F_RL_WLUN_OK & flags))) {
if (sdebug_verbose)
sdev_printk(KERN_INFO, sdp, "%s: Opcode 0x%x not%s\n",
my_name, opcode, " supported for wlun");
mk_sense_invalid_opcode(scp);
goto check_cond;
}
if (unlikely(sdebug_strict)) { /* check cdb against mask */
u8 rem;
int j;
for (k = 1; k < oip->len_mask[0] && k < 16; ++k) {
rem = ~oip->len_mask[k] & cmd[k];
if (rem) {
for (j = 7; j >= 0; --j, rem <<= 1) {
if (0x80 & rem)
break;
}
mk_sense_invalid_fld(scp, SDEB_IN_CDB, k, j);
goto check_cond;
}
}
}
if (unlikely(!(F_SKIP_UA & flags) &&
find_first_bit(devip->uas_bm,
SDEBUG_NUM_UAS) != SDEBUG_NUM_UAS)) {
errsts = make_ua(scp, devip);
if (errsts)
goto check_cond;
}
if (unlikely(((F_M_ACCESS & flags) || scp->cmnd[0] == TEST_UNIT_READY) &&
atomic_read(&devip->stopped))) {
errsts = resp_not_ready(scp, devip);
if (errsts)
goto fini;
}
if (sdebug_fake_rw && (F_FAKE_RW & flags))
goto fini;
if (unlikely(sdebug_every_nth)) {
if (fake_timeout(scp))
return 0; /* ignore command: make trouble */
}
if (likely(oip->pfp))
pfp = oip->pfp; /* calls a resp_* function */
else
pfp = r_pfp; /* if leaf function ptr NULL, try the root's */
fini:
if (F_DELAY_OVERR & flags) /* cmds like INQUIRY respond asap */
return schedule_resp(scp, devip, errsts, pfp, 0, 0);
else if ((flags & F_LONG_DELAY) && (sdebug_jdelay > 0 ||
sdebug_ndelay > 10000)) {
/*
* Skip long delays if ndelay <= 10 microseconds. Otherwise
* for Start Stop Unit (SSU) want at least 1 second delay and
* if sdebug_jdelay>1 want a long delay of that many seconds.
* For Synchronize Cache want 1/20 of SSU's delay.
*/
int jdelay = (sdebug_jdelay < 2) ? 1 : sdebug_jdelay;
int denom = (flags & F_SYNC_DELAY) ? 20 : 1;
jdelay = mult_frac(USER_HZ * jdelay, HZ, denom * USER_HZ);
return schedule_resp(scp, devip, errsts, pfp, jdelay, 0);
} else
return schedule_resp(scp, devip, errsts, pfp, sdebug_jdelay,
sdebug_ndelay);
check_cond:
return schedule_resp(scp, devip, check_condition_result, NULL, 0, 0);
err_out:
return schedule_resp(scp, NULL, DID_NO_CONNECT << 16, NULL, 0, 0);
}
static int sdebug_init_cmd_priv(struct Scsi_Host *shost, struct scsi_cmnd *cmd)
{
struct sdebug_scsi_cmd *sdsc = scsi_cmd_priv(cmd);
spin_lock_init(&sdsc->lock);
return 0;
}
static struct scsi_host_template sdebug_driver_template = {
.show_info = scsi_debug_show_info,
.write_info = scsi_debug_write_info,
.proc_name = sdebug_proc_name,
.name = "SCSI DEBUG",
.info = scsi_debug_info,
.slave_alloc = scsi_debug_slave_alloc,
.slave_configure = scsi_debug_slave_configure,
.slave_destroy = scsi_debug_slave_destroy,
.ioctl = scsi_debug_ioctl,
.queuecommand = scsi_debug_queuecommand,
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
.change_queue_depth = sdebug_change_qdepth,
.map_queues = sdebug_map_queues,
.mq_poll = sdebug_blk_mq_poll,
.eh_abort_handler = scsi_debug_abort,
.eh_device_reset_handler = scsi_debug_device_reset,
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
.eh_target_reset_handler = scsi_debug_target_reset,
.eh_bus_reset_handler = scsi_debug_bus_reset,
.eh_host_reset_handler = scsi_debug_host_reset,
.can_queue = SDEBUG_CANQUEUE,
.this_id = 7,
.sg_tablesize = SG_MAX_SEGMENTS,
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
.cmd_per_lun = DEF_CMD_PER_LUN,
.max_sectors = -1U,
.max_segment_size = -1U,
.module = THIS_MODULE,
.track_queue_depth = 1,
.cmd_size = sizeof(struct sdebug_scsi_cmd),
.init_cmd_priv = sdebug_init_cmd_priv,
};
static int sdebug_driver_probe(struct device *dev)
{
int error = 0;
struct sdebug_host_info *sdbg_host;
struct Scsi_Host *hpnt;
int hprot;
sdbg_host = dev_to_sdebug_host(dev);
sdebug_driver_template.can_queue = sdebug_max_queue;
sdebug_driver_template.cmd_per_lun = sdebug_max_queue;
if (!sdebug_clustering)
sdebug_driver_template.dma_boundary = PAGE_SIZE - 1;
hpnt = scsi_host_alloc(&sdebug_driver_template, 0);
if (NULL == hpnt) {
pr_err("scsi_host_alloc failed\n");
error = -ENODEV;
return error;
}
if (submit_queues > nr_cpu_ids) {
pr_warn("%s: trim submit_queues (was %d) to nr_cpu_ids=%u\n",
my_name, submit_queues, nr_cpu_ids);
submit_queues = nr_cpu_ids;
}
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
/*
* Decide whether to tell scsi subsystem that we want mq. The
* following should give the same answer for each host.
scsi: scsi_debug: Support hostwide tags Many SCSI HBAs support a hostwide tagset, whereby each command submitted to the HW from all submission queues must have a unique tag identifier. Normally this unique tag will be in the range [0, max queue], where "max queue" is the depth of each of the submission queues. Add support for this hostwide tag feature, via module parameter "host_max_queue". A non-zero value means that the feature is enabled. In this case, the submission queues are not exposed to upper layer, i.e. from blk-mq prespective, the device has a single hw queue. There are 2 reasons for this: a. It is assumed that the host can support nr_hw_queues * can_queue commands, but this is not true for hostwide tags b. For nr_hw_queues != 0, the request tag is not unique over all HW queues, and some HBA drivers want to use this tag for the hostwide tag However, like many SCSI HBA drivers today - megaraid sas being an example - the full set of HW submission queues are still used in the LLDD driver. So instead of using a complicated "reply_map" to create a per-CPU submission queue mapping like megaraid_sas (as it depends on a PCI device + MSIs) - use a simple algorithm: hwq = cpu % queue count If the host_max_queue param is set non-zero, then the max queue depth is fixed at this value also. If and when hostwide shared tags are supported in blk-mq/scsi mid-layer, then the policy to set nr_hw_queues = 0 for hostwide tags can be revised. Link: https://lore.kernel.org/r/1594297400-24756-3-git-send-email-john.garry@huawei.com Acked-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-07-09 12:23:20 +00:00
*/
hpnt->nr_hw_queues = submit_queues;
if (sdebug_host_max_queue)
hpnt->host_tagset = 1;
/* poll queues are possible for nr_hw_queues > 1 */
if (hpnt->nr_hw_queues == 1 || (poll_queues < 1)) {
pr_warn("%s: trim poll_queues to 0. poll_q/nr_hw = (%d/%d)\n",
my_name, poll_queues, hpnt->nr_hw_queues);
poll_queues = 0;
}
/*
* Poll queues don't need interrupts, but we need at least one I/O queue
* left over for non-polled I/O.
* If condition not met, trim poll_queues to 1 (just for simplicity).
*/
if (poll_queues >= submit_queues) {
if (submit_queues < 3)
pr_warn("%s: trim poll_queues to 1\n", my_name);
else
pr_warn("%s: trim poll_queues to 1. Perhaps try poll_queues=%d\n",
my_name, submit_queues - 1);
poll_queues = 1;
}
if (poll_queues)
hpnt->nr_maps = 3;
sdbg_host->shost = hpnt;
if ((hpnt->this_id >= 0) && (sdebug_num_tgts > hpnt->this_id))
hpnt->max_id = sdebug_num_tgts + 1;
else
hpnt->max_id = sdebug_num_tgts;
/* = sdebug_max_luns; */
hpnt->max_lun = SCSI_W_LUN_REPORT_LUNS + 1;
hprot = 0;
switch (sdebug_dif) {
case T10_PI_TYPE1_PROTECTION:
hprot = SHOST_DIF_TYPE1_PROTECTION;
if (sdebug_dix)
hprot |= SHOST_DIX_TYPE1_PROTECTION;
break;
case T10_PI_TYPE2_PROTECTION:
hprot = SHOST_DIF_TYPE2_PROTECTION;
if (sdebug_dix)
hprot |= SHOST_DIX_TYPE2_PROTECTION;
break;
case T10_PI_TYPE3_PROTECTION:
hprot = SHOST_DIF_TYPE3_PROTECTION;
if (sdebug_dix)
hprot |= SHOST_DIX_TYPE3_PROTECTION;
break;
default:
if (sdebug_dix)
hprot |= SHOST_DIX_TYPE0_PROTECTION;
break;
}
scsi_host_set_prot(hpnt, hprot);
if (have_dif_prot || sdebug_dix)
pr_info("host protection%s%s%s%s%s%s%s\n",
(hprot & SHOST_DIF_TYPE1_PROTECTION) ? " DIF1" : "",
(hprot & SHOST_DIF_TYPE2_PROTECTION) ? " DIF2" : "",
(hprot & SHOST_DIF_TYPE3_PROTECTION) ? " DIF3" : "",
(hprot & SHOST_DIX_TYPE0_PROTECTION) ? " DIX0" : "",
(hprot & SHOST_DIX_TYPE1_PROTECTION) ? " DIX1" : "",
(hprot & SHOST_DIX_TYPE2_PROTECTION) ? " DIX2" : "",
(hprot & SHOST_DIX_TYPE3_PROTECTION) ? " DIX3" : "");
if (sdebug_guard == 1)
scsi_host_set_guard(hpnt, SHOST_DIX_GUARD_IP);
else
scsi_host_set_guard(hpnt, SHOST_DIX_GUARD_CRC);
sdebug_verbose = !!(SDEBUG_OPT_NOISE & sdebug_opts);
sdebug_any_injecting_opt = !!(SDEBUG_OPT_ALL_INJECTING & sdebug_opts);
if (sdebug_every_nth) /* need stats counters for every_nth */
sdebug_statistics = true;
error = scsi_add_host(hpnt, &sdbg_host->dev);
if (error) {
pr_err("scsi_add_host failed\n");
error = -ENODEV;
scsi_host_put(hpnt);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
} else {
scsi_scan_host(hpnt);
scsi: scsi_debug: Add per_host_store option The scsi_debug driver has always been restricted to using one ramdisk image (or none) for its storage. This means that thousands of scsi_debug devices can be created without exhausting the host machine's RAM. The downside is that all scsi_debug devices share the same ramdisk image. This option changes the way a following write to the add_host parameter (or an add_host in the module/driver invocation) operates. For each new host that is created while per_host_store is true, a new store (of dev-size_mb MiB) is created and associated with all the LUs that belong to that new host. The user (who will need root permissions) needs to take care not to exhaust all the machine's available RAM. One reason for doing this is to check that (partial) disk to disk copies based on scsi_debug devices have actually copied accurately. To test this the add_host=<n> parameter where <n> is 2 or greater can be used when the scsi_debug module is loaded. Let us assume that /dev/sdb and /dev/sg1 are the same scsi_debug device, while /dev/sdc and /dev/sg2 are the same scsi_debug device. With per_host_store=1 add_host=2 they will have different ramdisk images. Then the following pseudocode could be executed to check if the sgh_dd copy worked: dd if=/dev/urandom of=/dev/sdb sgh_dd if=/dev/sg1 of=/dev/sg2 [plus option(s) to test] cmp /dev/sdb /dev/sdc If the cmp fails then the copy has failed (or some other mechanism wrote to /dev/sdb or /dev/sdc in the interim). [mkp: use kstrtobool()] Link: https://lore.kernel.org/r/20200421151424.32668-3-dgilbert@interlog.com Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2020-04-21 15:14:18 +00:00
}
scsi_debug: support scsi-mq, queues and locks - add host_lock option whose default value is 0 which removes the host_lock around all queued commands - accept delay=-1 (_hi_) or -2 which use a tasklet to invoke the scsi_done callback into the mid-layer. The default is still delay=1 which uses a timer to delay 1 jiffy - wire .change_queue_depth and .change_queue_type functions to better simulate queueing in a modern LLD - add SCSI_DEBUG_OPT_Q_NOISE (0x200) mask to only produce debug output associated with queue full, plus from .change_queue_depth and .change_queue_type functions - add SCSI_DEBUG_OPT_ALL_TSF (0x400) mask which reports all queued_arr fulls at TASK_SET_FULL, otherwise SCSI_MLQUEUE_HOST_BUSY is returned - add SCSI_DEBUG_OPT_RARE_TSF (0x800) mask which works together with the every_nth option (> 0) to count occurrences of num_in_q==queue_depth. When every_nth is reached the victim (a command) yields TASK SET FULL - clean up many debug messages. - add ndelay=<nanosecs> option that uses high resolution timers; active if > 0 and then overrides delay= option - expand Unit Attention handling: POR, BUS_RESET and MODE PARAMETERS CHANGED - support .eh_target_reset_handler and drop .bios_param - add OPT_N_WCE mask so caching page yields WCE=0 - add OPT_RESET_NOISE mask to log aborts and resets - add OPT_NO_CDB_NOISE mask to not log each cdb - MODE SELECT support for changing caching page's WCE - name common ioctls in log - when fake_rw=1, do not vmalloc fake store; make UNMAP and WRITE SAME obey fake_rw - more logging and code improvements including better sense buffer handling With fio and four (pseudo) devices I have observed 1.2 M IOPS on my equipment. Rob Elliott who has done much testing and made numerous suggestions, has better IOPS results than mine. Signed-off-by: Douglas Gilbert <dgilbert@interlog.com> Reviewed-by: Robert Elliott <elliott@hp.com> Tested-by: Robert Elliott <elliott@hp.com> Signed-off-by: Christoph Hellwig <hch@lst.de>
2014-07-26 15:55:35 +00:00
return error;
}
bus: Make remove callback return void The driver core ignores the return value of this callback because there is only little it can do when a device disappears. This is the final bit of a long lasting cleanup quest where several buses were converted to also return void from their remove callback. Additionally some resource leaks were fixed that were caused by drivers returning an error code in the expectation that the driver won't go away. With struct bus_type::remove returning void it's prevented that newly implemented buses return an ignored error code and so don't anticipate wrong expectations for driver authors. Reviewed-by: Tom Rix <trix@redhat.com> (For fpga) Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org> Reviewed-by: Cornelia Huck <cohuck@redhat.com> (For drivers/s390 and drivers/vfio) Acked-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> (For ARM, Amba and related parts) Acked-by: Mark Brown <broonie@kernel.org> Acked-by: Chen-Yu Tsai <wens@csie.org> (for sunxi-rsb) Acked-by: Pali Rohár <pali@kernel.org> Acked-by: Mauro Carvalho Chehab <mchehab@kernel.org> (for media) Acked-by: Hans de Goede <hdegoede@redhat.com> (For drivers/platform) Acked-by: Alexandre Belloni <alexandre.belloni@bootlin.com> Acked-By: Vinod Koul <vkoul@kernel.org> Acked-by: Juergen Gross <jgross@suse.com> (For xen) Acked-by: Lee Jones <lee.jones@linaro.org> (For mfd) Acked-by: Johannes Thumshirn <jth@kernel.org> (For mcb) Acked-by: Johan Hovold <johan@kernel.org> Acked-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org> (For slimbus) Acked-by: Kirti Wankhede <kwankhede@nvidia.com> (For vfio) Acked-by: Maximilian Luz <luzmaximilian@gmail.com> Acked-by: Heikki Krogerus <heikki.krogerus@linux.intel.com> (For ulpi and typec) Acked-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com> (For ipack) Acked-by: Geoff Levand <geoff@infradead.org> (For ps3) Acked-by: Yehezkel Bernat <YehezkelShB@gmail.com> (For thunderbolt) Acked-by: Alexander Shishkin <alexander.shishkin@linux.intel.com> (For intel_th) Acked-by: Dominik Brodowski <linux@dominikbrodowski.net> (For pcmcia) Acked-by: Rafael J. Wysocki <rafael@kernel.org> (For ACPI) Acked-by: Bjorn Andersson <bjorn.andersson@linaro.org> (rpmsg and apr) Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com> (For intel-ish-hid) Acked-by: Dan Williams <dan.j.williams@intel.com> (For CXL, DAX, and NVDIMM) Acked-by: William Breathitt Gray <vilhelm.gray@gmail.com> (For isa) Acked-by: Stefan Richter <stefanr@s5r6.in-berlin.de> (For firewire) Acked-by: Benjamin Tissoires <benjamin.tissoires@redhat.com> (For hid) Acked-by: Thorsten Scherer <t.scherer@eckelmann.de> (For siox) Acked-by: Sven Van Asbroeck <TheSven73@gmail.com> (For anybuss) Acked-by: Ulf Hansson <ulf.hansson@linaro.org> (For MMC) Acked-by: Wolfram Sang <wsa@kernel.org> # for I2C Acked-by: Sudeep Holla <sudeep.holla@arm.com> Acked-by: Geert Uytterhoeven <geert@linux-m68k.org> Acked-by: Dmitry Torokhov <dmitry.torokhov@gmail.com> Acked-by: Finn Thain <fthain@linux-m68k.org> Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Link: https://lore.kernel.org/r/20210713193522.1770306-6-u.kleine-koenig@pengutronix.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-07-13 19:35:22 +00:00
static void sdebug_driver_remove(struct device *dev)
{
struct sdebug_host_info *sdbg_host;
struct sdebug_dev_info *sdbg_devinfo, *tmp;
sdbg_host = dev_to_sdebug_host(dev);
scsi_remove_host(sdbg_host->shost);
list_for_each_entry_safe(sdbg_devinfo, tmp, &sdbg_host->dev_info_list,
dev_list) {
list_del(&sdbg_devinfo->dev_list);
kfree(sdbg_devinfo->zstate);
kfree(sdbg_devinfo);
}
scsi_host_put(sdbg_host->shost);
}
static struct bus_type pseudo_lld_bus = {
.name = "pseudo",
.probe = sdebug_driver_probe,
.remove = sdebug_driver_remove,
.drv_groups = sdebug_drv_groups,
};