Kernel messages produced during runtime PM can cause a never-ending cycle
because user space utilities (e.g. journald or rsyslog) write the messages
back to storage, causing runtime resume, more messages, and so on.
Messages that tell of things that are expected to happen, are arguably
unnecessary, so suppress them.
UFS driver messages are changes to from dev_err() to dev_dbg() which means
they will not display unless activated by dynamic debug of building with
-DDEBUG.
sdev->silence_suspend is set to skip messages from sd_suspend_common()
"Synchronizing SCSI cache", "Stopping disk" and scsi_report_sense()
"Power-on or device reset occurred" message (Note, that message appears
when the LUN is accessed after runtime PM, not during runtime PM)
Example messages from Ubuntu 21.10:
$ dmesg | tail
[ 1620.380071] ufshcd 0000:00:12.5: ufshcd_print_pwr_info:[RX, TX]: gear=[1, 1], lane[1, 1], pwr[SLOWAUTO_MODE, SLOWAUTO_MODE], rate = 0
[ 1620.408825] ufshcd 0000:00:12.5: ufshcd_print_pwr_info:[RX, TX]: gear=[4, 4], lane[2, 2], pwr[FAST MODE, FAST MODE], rate = 2
[ 1620.409020] ufshcd 0000:00:12.5: ufshcd_find_max_sup_active_icc_level: Regulator capability was not set, actvIccLevel=0
[ 1620.409524] sd 0:0:0:0: Power-on or device reset occurred
[ 1622.938794] sd 0:0:0:0: [sda] Synchronizing SCSI cache
[ 1622.939184] ufs_device_wlun 0:0:0:49488: Power-on or device reset occurred
[ 1625.183175] ufshcd 0000:00:12.5: ufshcd_print_pwr_info:[RX, TX]: gear=[1, 1], lane[1, 1], pwr[SLOWAUTO_MODE, SLOWAUTO_MODE], rate = 0
[ 1625.208041] ufshcd 0000:00:12.5: ufshcd_print_pwr_info:[RX, TX]: gear=[4, 4], lane[2, 2], pwr[FAST MODE, FAST MODE], rate = 2
[ 1625.208311] ufshcd 0000:00:12.5: ufshcd_find_max_sup_active_icc_level: Regulator capability was not set, actvIccLevel=0
[ 1625.209035] sd 0:0:0:0: Power-on or device reset occurred
Note for stable: depends on patch "scsi: core: sd: Add silence_suspend flag
to suppress some PM messages".
Link: https://lore.kernel.org/r/20220228113652.970857-3-adrian.hunter@intel.com
Cc: stable@vger.kernel.org
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Kernel messages produced during runtime PM can cause a never-ending cycle
because user space utilities (e.g. journald or rsyslog) write the messages
back to storage, causing runtime resume, more messages, and so on.
Messages that tell of things that are expected to happen are arguably
unnecessary, so add a flag to suppress them. This flag is used by the UFS
driver.
Link: https://lore.kernel.org/r/20220228113652.970857-2-adrian.hunter@intel.com
Cc: stable@vger.kernel.org
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
We currently allocate a workqueue per host and only use it for removing the
target. For the session per host case we could be using this workqueue to
be able to do recoveries (block, unblock, timeout handling) in parallel. To
also allow offload drivers to do their session recoveries in parallel, this
drops the per host workqueue and replaces it with a per session one.
Link: https://lore.kernel.org/r/20220226230435.38733-5-michael.christie@oracle.com
Reviewed-by: Lee Duncan <lduncan@suse.com>
Reviewed-by: Chris Leech <cleech@redhat.com>
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
When the iSCSI class was added upstream, blocking a queue was fast because
it just set some flag bits and didn't handle I/O that was in the process of
being sent to the driver. That's no longer the case so blocking a queue is
expensive and we can end up with a backlog of blocks by the time we have
relogged in and are trying to start the queues.
For the session unblock case, this has try to cancel the block and recovery
work in case they are still queued so we can avoid unneeded queue
manipulations. For removal, we also now try to cancel all the recovery
related works since a couple lines down we will set the session and device
state so running those functions are not necessary.
Link: https://lore.kernel.org/r/20220226230435.38733-3-michael.christie@oracle.com
Reviewed-by: Lee Duncan <lduncan@suse.com>
Reviewed-by: Chris Leech <cleech@redhat.com>
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
If the user sets the iscsi_eh_timer_workq/iscsi_eh workqueue's max_active
to greater than 1, the recovery_work could be running when
__iscsi_unblock_session() runs. The cancel_delayed_work() will then not
wait for the running work and we can race where we end up with the wrong
session state and scsi_device state set.
This replaces the cancel_delayed_work() with the sync version.
Link: https://lore.kernel.org/r/20220226230435.38733-2-michael.christie@oracle.com
Reviewed-by: Lee Duncan <lduncan@suse.com>
Reviewed-by: Chris Leech <cleech@redhat.com>
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Sparse throws a warning about context imbalance ("different lock contexts
for basic block") in sas_form_port() as it gets confused with the fact that
a port is locked within one of the two search loops and unlocked afterward
outside of the search loops once the phy is added to the port. Since this
code is not easy to follow, improve it by factoring out the code adding the
phy to the port once the port is locked into the helper function
sas_form_port_add_phy(). This helper can then be called directly within the
port search loops, avoiding confusion and clearing the sparse warning.
Link: https://lore.kernel.org/r/20220228094857.557329-1-damien.lemoal@opensource.wdc.com
Reviewed-by: John Garry <john.garry@huawei.com>
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Now that each scsi_request is backed by a scsi_cmnd, there is no need to
indirect the CDB storage. Change all submitters of SCSI passthrough
requests to store the CDB information directly in the scsi_cmnd, and while
doing so allocate the full 32 bytes that cover all Linux supported SCSI
hosts instead of requiring dynamic allocation for > 16 byte CDBs. On
64-bit systems this does not change the size of the scsi_cmnd at all, while
on 32-bit systems it slightly increases it for now, but that increase will
be made up by the removal of the remaining scsi_request fields.
Link: https://lore.kernel.org/r/20220224175552.988286-4-hch@lst.de
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: John Garry <john.garry@huawei.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Replace the big fat memset that requires saving and restoring various
fields with just initializing those fields that need initialization.
All the clearing to 0 is moved to scsi_prepare_cmd() as scsi_ioctl_reset()
alreadly uses kzalloc() to allocate a pre-zeroed command.
This is still conservative and can probably be optimized further.
Link: https://lore.kernel.org/r/20220224175552.988286-3-hch@lst.de
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: John Garry <john.garry@huawei.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
In case of SSP underflow allow the response frame IU to be examined for
setting the response stat value rather than always setting
SAS_DATA_UNDERRUN.
This will mean that we call sas_ssp_task_response() in those scenarios and
may send sense data to upper layer.
Such a condition would be for bad blocks were we just reporting an
underflow error to upper layer, but now the sense data will tell
immediately that the media is faulty.
Link: https://lore.kernel.org/r/1645703489-87194-7-git-send-email-john.garry@huawei.com
Signed-off-by: Xingui Yang <yangxingui@huawei.com>
Signed-off-by: Qi Liu <liuqi115@huawei.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
If the driver probe fails to request the channel IRQ or fatal IRQ, the
driver will free the IRQ vectors before freeing the IRQs in free_irq(),
and this will cause a kernel BUG like this:
------------[ cut here ]------------
kernel BUG at drivers/pci/msi.c:369!
Internal error: Oops - BUG: 0 [#1] PREEMPT SMP
Call trace:
free_msi_irqs+0x118/0x13c
pci_disable_msi+0xfc/0x120
pci_free_irq_vectors+0x24/0x3c
hisi_sas_v3_probe+0x360/0x9d0 [hisi_sas_v3_hw]
local_pci_probe+0x44/0xb0
work_for_cpu_fn+0x20/0x34
process_one_work+0x1d0/0x340
worker_thread+0x2e0/0x460
kthread+0x180/0x190
ret_from_fork+0x10/0x20
---[ end trace b88990335b610c11 ]---
So we use devm_add_action() to control the order in which we free the
vectors.
Link: https://lore.kernel.org/r/1645703489-87194-4-git-send-email-john.garry@huawei.com
Signed-off-by: Qi Liu <liuqi115@huawei.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The time of phyup not only depends on the controller but also the type of
disk connected. As an example, from experience, for some SATA disks the
amount of time from reset/power-on to receive the D2H FIS for phyup can
take upto and more than 10s sometimes. According to the specification of
some SATA disks such as ST14000NM0018, the max time from power-on to ready
is 30s.
Based on this the current timeout of phyup at 2s which is not enough. So
set the value as HISI_SAS_WAIT_PHYUP_TIMEOUT (30s) in
hisi_sas_control_phy().
For v3 hw there is a pre-existing workaround for a HW bug, being that we
issue a link reset when the OOB occurs but the phyup does not. The current
phyup timeout is HISI_SAS_WAIT_PHYUP_TIMEOUT. So if this does occur from
when issuing a phy enable or similar via hisi_sas_control_phy(), the
subsequent HW workaround linkreset processing calls hisi_sas_control_phy(),
but this will pend the original phy reset timing out, so it is safe.
Link: https://lore.kernel.org/r/1645703489-87194-3-git-send-email-john.garry@huawei.com
Signed-off-by: Xiang Chen <chenxiang66@hisilicon.com>
Signed-off-by: John Garry <john.garry@huawei.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Firmware expects host driver to clear scratchpad rsvd 0 register after
non-fatal error is found.
This is done when firmware raises fatal error interrupt and indicates
non-fatal error. At this point firmware updates scratchpad rsvd 0 register
with non-fatal error value. Here host has to clear the register after
reading it during non-fatal errors.
Rename:
- MSGU_HOST_SCRATCH_PAD_6 to MSGU_SCRATCH_PAD_RSVD_0
- MSGU_HOST_SCRATCH_PAD_7 to MSGU_SCRATCH_PAD_RSVD_1
Link: https://lore.kernel.org/r/20220222092618.108198-1-Ajish.Koshy@microchip.com
Reviewed-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Acked-by: Jack Wang <jinpu.wang@ionos.com>
Signed-off-by: Ajish Koshy <Ajish.Koshy@microchip.com>
Signed-off-by: Viswas G <Viswas.G@microchip.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Just a few lines below this kzalloc() we have a mutex_lock() which can
sleep.
Moreover, the only way to call this function is when a delayed work is
schedule. And delayed work can sleep:
INIT_DELAYED_WORK(&fw_event->work, mptsas_firmware_event_work);
--> mptsas_firmware_event_work()
--> mptsas_send_link_status_event()
--> mptsas_expander_add()
So there is really no good reason to use GFP_ATOMIC here. Change it to
GFP_KERNEL to give more opportunities to the kernel.
Link: https://lore.kernel.org/r/eccb2179ce800529851ed4fabc9d3f95fbbf7d7f.1644906731.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
The main part of the pm8001_task_exec() function uses a do {} while(0) loop
that is useless and only makes the code harder to read. Remove this
loop. The unnecessary local variable t is also removed.
Additionally, avoid repeatedly declaring "struct task_status_struct *ts" to
handle error cases by declaring this variable for the entire function
scope. This allows simplifying the error cases, and together with the
addition of blank lines make the code more readable.
Finally, handling of the running_req counter is fixed to avoid decrementing
it without a corresponding incrementation in the case of an invalid task
protocol.
Link: https://lore.kernel.org/r/20220220031810.738362-29-damien.lemoal@opensource.wdc.com
Reviewed-by: John Garry <john.garry@huawei.com>
Reviewed-by: Jack Wang <jinpu.wang@ionos.com>
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
There is no need to pass a pointer to a struct inbound_queue_table to
pm8001_mpi_build_cmd(). Passing the start index in the inbound queue table
of the adapter is enough. This change allows avoiding the declaration of a
struct inbound_queue_table pointer (circularQ variables) in many functions,
simplifying the code.
While at it, blank lines are added i(e.g. after local variable
declarations) to make the code more readable.
Link: https://lore.kernel.org/r/20220220031810.738362-28-damien.lemoal@opensource.wdc.com
Reviewed-by: Jack Wang <jinpu.wang@ionos.com>
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Introduce the pm8001_ccb_alloc() and pm8001_ccb_free() helpers to replace
the typical code patterns:
res = pm8001_tag_alloc(pm8001_ha, &ccb_tag);
if (res)
...
ccb = &pm8001_ha->ccb_info[ccb_tag];
ccb->device = pm8001_ha_dev;
ccb->ccb_tag = ccb_tag;
ccb->task = task;
ccb->n_elem = 0;
and
ccb->task = NULL;
ccb->ccb_tag = PM8001_INVALID_TAG;
pm8001_tag_free(pm8001_ha, tag);
With the simpler function calls:
ccb = pm8001_ccb_alloc(pm8001_ha, pm8001_ha_dev, task);
if (!ccb)
...
and
pm8001_ccb_free(pm8001_ha, ccb);
The pm8001_ccb_alloc() helper ensures that all fields of the ccb info
structure for the newly allocated tag are all initialized, except the
buf_prd field. The pm8001_ccb_free() helper clears the initialized fields
and the ccb tag to ensure that iteration over the adapter ccb_info array
detects ccbs that are in use.
All call site of the pm8001_tag_alloc() function that use a ccb info
associated with an allocated tag are converted to use the new helpers.
Link: https://lore.kernel.org/r/20220220031810.738362-27-damien.lemoal@opensource.wdc.com
Reviewed-by: John Garry <john.garry@huawei.com>
Reviewed-by: Jack Wang <jinpu.wang@ionos.com>
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
