Change the location where the rpc_xprt structure is allocated so each
transport implementation can allocate a private area from the same
chunk of memory.
Note also that xprt->ops->destroy, rather than xprt_destroy, is now
responsible for freeing rpc_xprt when the transport is destroyed.
Test plan:
Connectathon.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Move the xid field in the rpc_xprt structure to be in the same cache line
as the reserve_lock, since these are used at the same time.
Test plan:
None.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Remove use of the Big Kernel Lock around indirect calls to
nfs3_proc_readlink and nfs4_proc_readlink, both of which
basically call rpc_call_sync.
Signed-off-by: Frank Filz <ffilz@us.ibm.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Remove use of the Big Kernel Lock around calls to rpc_call_sync.
Signed-off-by: Frank Filz <ffilz@us.ibm.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Remove use of the Big Kernel Lock around calls to rpc_execute.
Signed-off-by: Frank Filz <ffilz@us.ibm.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
All internal RPC client operations should no longer depend on the BKL,
however lockd and NFS callbacks may still require it.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Currently nfs_sync_inode_wait() will fail to loop correctly when we call
nfs_sync_inode_wait with the FLUSH_INVALIDATE argument.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
We must always call ->read_done() before we truncate the page data, or
decide to flag an error. The reasons are that
in NFSv2, ->read_done() is where the eof flag gets set.
in NFSv3/v4 ->read_done() handles EJUKEBOX-type errors, and
v4 state recovery.
However, we need to mark the pages as uptodate before we deal with short
read errors, since we may need to modify the nfs_read_data arguments.
We therefore split the current nfs_readpage_result() into two parts:
nfs_readpage_result(), which calls ->read_done() etc, and
nfs_readpage_retry(), which subsequently handles short reads.
Note: Removing the code that retries in case of a short read also fixes a
bug in nfs_direct_read_result(), which used to return a corrupted number of
bytes.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
When trying to open a file with the O_EXCL flag over NFS on a server that does
not support exclusive mode, the file does not open. The reason,
rpc_call_sync returns a errno number, and not the nfs error number. I fixed
it by changing the status check in nfs3proc.c. Either this is how it should
be fixed, or rpc_call_sync should be fixed to return the NFS error.
Signed-off-by: Andy Ryan <genanr@allantgroup.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Use RCU to ensure that we can safely call rpc_finish_wakeup after we've
called __rpc_do_wake_up_task. If not, there is a theoretical race, in which
the rpc_task finishes executing, and gets freed first.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The sunrpc scheduler contains a race condition that can let an RPC
task end up being neither running nor on any wait queue. The race takes
place between rpc_make_runnable (called from rpc_wake_up_task) and
__rpc_execute under the following condition:
First __rpc_execute calls tk_action which puts the task on some wait
queue. The task is dequeued by another process before __rpc_execute
continues its execution. While executing rpc_make_runnable exactly after
setting the task `running' bit and before clearing the `queued' bit
__rpc_execute picks up execution, clears `running' and subsequently
both functions fall through, both under the false assumption somebody
else took the job.
Swapping rpc_test_and_set_running with rpc_clear_queued in
rpc_make_runnable fixes that hole. This introduces another possible
race condition that can be handled by checking for `queued' after
setting the `running' bit.
Bug noticed on a 4-way x86_64 system under XEN with an NFSv4 server
on the same physical machine, apparently one of the few ways to hit
this race condition at all.
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Christophe Saout <christophe@saout.de>
Signed-off-by: Trond Myklebust <trond.myklebust@fys.uio.no>
Simple patch to add the new PCIe version of the 29320 card.
Signed-off: Mark Salyzyn <Mark_Salyzyn@adaptec.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
The original wait loop may be much longer than intended time.
Use more accurate timer_after for it. Also adjust wait value to
avoid unnecessary long waiting.
Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Add support for st_vsc1 type device (st_vsc is ok because it does not
require extra buffer).
Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
- add comments for various devices
- remove unused device ids(0xf350, 0x4301, 0x8301, 0x8302)
- add new device id(0xe350)
- fix vendor id of st_vsc
- modify Kconfig help info
Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Firmware of new version may adjust default queue length. It is
backward compatible.
Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
During hard reset, an all-1 value from PCI_COMMAND should be invalid.
Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
This command needs information from both firmware and driver. First copy
information from firmware to buffer, then fill in driver information.
Signed-off-by: Ed Lin <ed.lin@promise.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
megaraid's MMIO RD*/WR* macros directly call readl() and writel() with
an 'unsigned long' argument. This throws a warning, but is otherwise OK
because the 'unsigned long' is really the result of ioremap(). This
setup is also OK because the variable can hold an ioremap cookie /or/ a
PCI I/O port (PIO).
However, to fix the warning thrown when readl() and writel() are passed
an unsigned long cookie, I introduce 'void __iomem *mmio_base', holding
the same value as 'base'. This will silence the warnings, and also
cause an oops whenever these MMIO-only functions are ever accidentally
passed an I/O address.
Signed-off-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
drivers/scsi/scsi_tgt_if.c: In function 'tgt_uspace_send_event':
drivers/scsi/scsi_tgt_if.c:88: warning: implicit declaration of function 'flush_dcache_page'
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
CONFIG_LBD and CONFIG_LSF are spread into asm/types.h for no particularly
good reason.
Centralising the definition in linux/types.h means that arch maintainers
don't need to bother adding it, as well as fixing the problem with
x86-64 users being asked to make a decision that has absolutely no
effect.
The H8/300 porters seem particularly confused since I'm not aware of any
microcontrollers that need to support 2TB filesystems.
Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
ip_summed changes last summer had missed that one. As the result,
we have ip_summed interpreted as CHECKSUM_PARTIAL now. IOW,
->csum is interpreted as offset of checksum in the packet. net/core/*
will both read and modify the value as that offset, with obvious
reasons. At the very least it's a remote memory corruptor.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
* master.kernel.org:/pub/scm/linux/kernel/git/paulus/powerpc: (194 commits)
[POWERPC] Add missing EXPORTS for mpc52xx support
[POWERPC] Remove obsolete PPC_52xx and update CLASSIC32 comment
[POWERPC] ps3: add a default zImage target
[POWERPC] Add of_platform_bus support to mpc52xx psc uart driver
[POWERPC] typo fix and whitespace cleanup on mpc52xx-uart driver
[POWERPC] Fix debug printks for 32-bit resources in the PCI code
[POWERPC] Replace kmalloc+memset with kzalloc
[POWERPC] Linkstation / kurobox support
[POWERPC] Add the e300c3 core to the CPU table.
[POWERPC] ppc: m48t35 add missing bracket
[POWERPC] iSeries: don't build head_64.o unnecessarily
[POWERPC] iSeries: stop dt_mod.o being rebuilt unnecessarily
[POWERPC] Fix cputable.h for combined build
[POWERPC] Allow CONFIG_BOOTX_TEXT on iSeries
[POWERPC] Allow xmon to build on legacy iSeries
[POWERPC] Change ppc64_defconfig to use AUTOFS_V4 not V3
[POWERPC] Tell firmware we can handle POWER6 compatible mode
[POWERPC] Clean images in arch/powerpc/boot
[POWERPC] Fix OF pci flags parsing
[POWERPC] defconfig for lite5200 board
...
This is an update to the AT91 CompactFlash driver.
We replace the hard-coded "chip select 4" with the chip-select value
passed via platform_data. The configuration of the EBI memory
controller to enable Compact Flash access is now also handled in the
platform setup code and not in the driver.
Signed-off-by: Andrew Victor <andrew@sanpeople.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
More fallout of the post 2.6.19-rc1 IRQ changes...
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
This is an update to the AT91RM9200 CompactFlash driver.
The changes include:
- Use the I/O memory address passed via the platform_device resources
instead of constant global values.
- The IRQ should not be used as a random'ness source.
- Return errors if ioremap() or request_mem_region() fails.
Signed-off-by: Andrew Victor <andrew@sanpeople.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
struct pcmcia_device *p_dev->conf.ConfigBase and .Present are set in almost
all PCMICA driver right at the beginning, using the same calls but slightly
different implementations. Unfiy this in the PCMCIA core.
Includes a small bugfix ("drivers/net/pcmcia/xirc2ps_cs.c: remove unused
label") from and Signed-off-by Adrian Bunk <bunk@stusta.de>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
As we read out the product information strings (VERS_1) from the PCMCIA device
in the PCMCIA core, and device drivers can access those reliably in struct
pcmcia_device's fields prod_id[], remove additional product information string
detection logic from PCMCIA device drivers.
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
As we read out the manufactor and card_id from the PCMCIA device in the
PCMCIA core, and device drivers can access those reliably in struct
pcmcia_device's fields manf_id and card_id, remove additional (and partly
broken) manf_id and card_id detection logic from PCMCIA device drivers.
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Add IDs for Elan serial PCMCIA devices. In addition, move the
pdaudio_cf driver from matching based on manf_id and card_id to
the more specific prod_id1 and prod_id2 to avoid false
positives.
Signed-off-by: Tony Olech <tony.olech@elandigitalsystems.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Some Elan serial cards allow for four (independent) multifunction
subdevices. Teach the PCMCIA core to deal with such devices.
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
