David Jeffery [Wed, 21 Nov 2012 07:39:54 +0000 (02:39 -0500)]
[SCSI] qla2xxx: Test and clear FCPORT_UPDATE_NEEDED atomically.
When the qla2xxx driver loses access to multiple, remote ports, there is a race
condition which can occur which will keep the request stuck on a scsi request
queue indefinitely.
This bad state occurred do to a race condition with how the FCPORT_UPDATE_NEEDED
bit is set in qla2x00_schedule_rport_del(), and how it is cleared in
qla2x00_do_dpc(). The problem port has its drport pointer set, but it has never
been processed by the driver to inform the fc transport that the port has been
lost. qla2x00_schedule_rport_del() sets drport, and then sets the
FCPORT_UPDATE_NEEDED bit. In qla2x00_do_dpc(), the port lists are walked and
any drport pointer is handled and the fc transport informed of the port loss,
then the FCPORT_UPDATE_NEEDED bit is cleared. This leaves a race where the
dpc thread is processing one port removal, another port removal is marked
with a call to qla2x00_schedule_rport_del(), and the dpc thread clears the
bit for both removals, even though only the first removal was actually
handled. Until another event occurs to set FCPORT_UPDATE_NEEDED, the later
port removal is never finished and qla2xxx stays in a bad state which causes
requests to become stuck on request queues.
This patch updates the driver to test and clear FCPORT_UPDATE_NEEDED
atomically. This ensures the port state changes are processed and not lost.
Signed-off-by: David Jeffery <djeffery@redhat.com> Signed-off-by: Chad Dupuis <chad.dupuis@qlogic.com> Cc: stable@vger.kernel.org Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Armen Baloyan [Wed, 21 Nov 2012 07:39:53 +0000 (02:39 -0500)]
[SCSI] qla2xxx: Properly set result field of bsg_job reply structure for success and failure.
FC transport on receiving bsg_job submission failure, calls bsg_job->job_done()
and sets the bsg_job->reply->result the returned value. In contrast, when the
success code (0) is returned fc transport doesn't call bsg_job->job_done() and
doesn't populate bsg_job->reply->result.
Signed-off-by: Steve Hodgson <steve@purestorage.com> Signed-off-by: Armen Baloyan <armen.baloyan@qlogic.com> Signed-off-by: Saurav Kashyap <saurav.kashyap@qlogic.com> Cc: <stable@vger.kernel.org> #3.7 Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Aaron Lu [Fri, 9 Nov 2012 07:27:55 +0000 (15:27 +0800)]
[SCSI] sd: update sd to use the new pm callbacks
Update sd driver to use the callbacks defined in dev_pm_ops.
sd_freeze is NULL, the bus level callback has taken care of quiescing
the device so there should be nothing needs to be done here.
Consequently, sd_thaw is not needed here either.
suspend, poweroff and runtime suspend share the same routine sd_suspend,
which will sync flush and then stop the drive, this is the same as before.
resume, restore and runtime resume share the same routine sd_resume,
which will start the drive by putting it into active power state, this
is also the same as before.
Signed-off-by: Aaron Lu <aaron.lu@intel.com> Acked-by: Alan Stern <stern@rowland.harvard.edu> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Aaron Lu [Fri, 9 Nov 2012 07:27:54 +0000 (15:27 +0800)]
[SCSI] scsi_pm: use callbacks from dev_pm_ops for scsi devices
Use of pm_message_t is deprecated and device driver is not supposed
to use that. This patch migrates the SCSI bus level pm callbacks
to call device's pm callbacks defined in its driver's dev_pm_ops.
This is achieved by finding out which device pm callback should be used
in bus callback function, and then pass that callback function pointer
as a param to the scsi_bus_{suspend,resume}_common routine, which will
further pass that callback to scsi_dev_type_{suspend,resume} after
proper handling.
The special case for freeze in scsi_bus_suspend_common is not necessary
since there is no high level SCSI driver has implemented freeze, so no
need to runtime resume the device if it is in runtime suspended state
for system freeze, just return like the system suspend/hibernate case.
Since only sd has implemented drv->suspend/drv->resume, and I'll update
sd driver to use the new callbacks in the following patch, there is no
need to fallback to call drv->suspend/drv->resume if dev_pm_ops is NULL.
Signed-off-by: Aaron Lu <aaron.lu@intel.com> Acked-by: Alan Stern <stern@rowland.harvard.edu> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Aaron Lu <aaron.lu@intel.com> Acked-by: Alan Stern <stern@rowland.harvard.edu> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Aaron Lu <aaron.lu@intel.com> Acked-by: Alan Stern <stern@rowland.harvard.edu> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Aaron Lu [Fri, 9 Nov 2012 07:27:51 +0000 (15:27 +0800)]
[SCSI] sd: put to stopped power state when runtime suspend
When device is runtime suspended, put it to stopped power state to save
some power.
This will also make the behaviour consistent with what the scsi_pm.c
thinks about sd as the comment says:
sd treats runtime suspend, system suspend and system hibernate identical.
With this patch, it is now identical.
And sd_shutdown will also do nothing when it finds the device has been
runtime suspended, if we do not spin down the disk in runtime suspend
by putting it into stopped power state, the disk will be shut down
incorrectly.
And the the same problem can be solved for runtime power off after
runtime suspended case by this change.
With the current runtime scheme for disk, it will only be runtime
suspended when no process opens the disk, so this shouldn't happen a
lot, which makes it acceptable to spin down the disk when runtime
suspended. If some day a more aggressive runtime scheme is used, like
the 'request based runtime pm for disk' that Alan Stern and Lin Ming
has been working, we can introduce some policy to control this. But for
now, make it simple and correct by spinning down the disk.
Signed-off-by: Aaron Lu <aaron.lu@intel.com> Acked-by: Alan Stern <stern@rowland.harvard.edu> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Xi Wang [Fri, 16 Nov 2012 19:40:03 +0000 (14:40 -0500)]
[SCSI] mvsas: fix undefined bit shift
The macro bit(n) is defined as ((u32)1 << n), and thus it doesn't work
with n >= 32, such as in mvs_94xx_assign_reg_set():
if (i >= 32) {
mvi->sata_reg_set |= bit(i);
...
}
The shift ((u32)1 << n) with n >= 32 also leads to undefined behavior.
The result varies depending on the architecture.
This patch changes bit(n) to do a 64-bit shift. It also simplifies
mv_ffc64() using __ffs64(), since invoking ffz() with ~0 is undefined.
Signed-off-by: Xi Wang <xi.wang@gmail.com> Acked-by: Xiangliang Yu <yuxiangl@marvell.com> Cc: stable@vger.kernel.org Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Sasha Levin [Thu, 15 Nov 2012 20:51:46 +0000 (15:51 -0500)]
[SCSI] prevent stack buffer overflow in host_reset
store_host_reset() has tried to re-invent the wheel to compare sysfs strings.
Unfortunately it did so poorly and never bothered to check the input from
userspace before overwriting stack with it, so something simple as:
This patch contains updates to firmware/hardware header files shared
between csiostor and cxgb4/cxgb4vf, and the resulting changes to the
cxgb4/cxgb4vf source files.
Signed-off-by: Naresh Kumar Inna <naresh@chelsio.com> Cc: David Miller <davem@davemloft.net> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Jason J. Herne [Wed, 14 Nov 2012 22:03:22 +0000 (17:03 -0500)]
[SCSI] sd: Use SCSI read/write(16) with > 32-bit LBA drives
Force large capacity (> 0xFFFFFFFF blocks) drives to use READ/WRITE(16) instead
of READ/WRITE(10). Some(most/all?) USB enclosures do not like READ(10) commands
when a large capacity drive is installed. This issue was reported and discussed
here: http://marc.info/?l=linux-usb&m=135247705222324
Signed-off-by: Jason J. Herne <hernejj@gmail.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Eric Northup [Thu, 8 Nov 2012 09:55:50 +0000 (01:55 -0800)]
[SCSI] virtio_scsi: fix memory leak on full queue condition.
virtscsi_queuecommand was leaking memory when the virtio queue was full.
Tested: Guest operates correctly even with very small queue sizes, validated
we're not leaking kmalloc-192 sized allocations anymore.
Signed-off-by: Eric Northup <digitaleric@google.com> Acked-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
James Smart [Wed, 31 Oct 2012 18:44:33 +0000 (14:44 -0400)]
[SCSI] lpfc 8.3.36: Fixed boot from san failure
Fixed boot from san failure when SLI4 FC device presented on the same PCI bus
The request_firmware interface can induce delays while looking
for firmware files, even if no fw file is present. In some situations
the delays exceeded scan_wait timeouts, resulting in situations in which
the boot device had not been discovered in time. Boot Device does not
need to be on a lpfc device.
Change request_firmware use to be module paramater driven. Default is to
not attempt firmware download on boot. Add sysfs parameter to invoke
firmware update.
Signed-off-by: James Smart <james.smart@emulex.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Eddie Wai [Tue, 16 Oct 2012 00:31:37 +0000 (17:31 -0700)]
[SCSI] bnx2i: removed the individual PCI DEVICE ID checking
Removed the individual PCI DEVICE ID checking inside bnx2i. The device
type can easily be read from the corresponding cnic->flags. This will
free bnx2i from having to get updated for every new device ID that gets
added.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com> Acked-by: Michael Chan <mchan@broadcom.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:16:04 +0000 (04:46 +0530)]
[SCSI] be2iscsi: Bump the driver version
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Driver periodically checks adapter state,is up fine or not.
Based on the value updates the internal structures of driver.
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:15:40 +0000 (04:45 +0530)]
[SCSI] be2iscsi: Fix for MBX timeout issue
The MBX timeout value set to 100 and if adapter doesn;t
return response in that time driver will return from waiting
for completion with an error to the caller. In the earlier code
driver use to wait until MBX response comes from adapter.
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:15:27 +0000 (04:45 +0530)]
[SCSI] be2iscsi: Update the copyright information
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:15:06 +0000 (04:45 +0530)]
[SCSI] be2iscsi: Fix issue of displaying adapter family.
Fix issue of displaying adapter family through the sysfs
entry for each Scsi_Host created for the adapter.
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
The completion events returned by adapter differs based on the
adapter. This fix checks for the adapter type and process the
completion event.
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:14:35 +0000 (04:44 +0530)]
[SCSI] be2iscsi: Fix session update context with V2 version.
For updating session context on adapter, V2 version is to
be used with the latest adapter. This fix checks for the adapter type
and uses correct version of session context.
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:14:23 +0000 (04:44 +0530)]
[SCSI] be2iscsi: Fix support for V2 version of WRB.
Latest adapters use the V2 version of WRB. This fix checks for the
adapter type and uses appropriate version of WRB.
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:13:44 +0000 (04:43 +0530)]
[SCSI] be2iscsi: Fix support for handling CQ_CREATE V2 version.
For latest adapters, V2 version of cq_create MBX_CMD is to be used. When
driver is loaded depending on the adapter type appropriate cq_create command
will be called.
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:13:32 +0000 (04:43 +0530)]
[SCSI] be2iscsi: Fix max EQ supported by the driver.
Fix the max EQ created when driver is loaded. Max EQ for
for new adapters will be 64.
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:13:20 +0000 (04:43 +0530)]
[SCSI] be2iscsi: Fix driver support for Skyhawk-R adapter.
Fix support for Skyhawk-R adapter by populating the pci_id_table
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:13:03 +0000 (04:43 +0530)]
[SCSI] be2iscsi: Fix return value and typo.
Fix return value and typo in the message displayed.
Fix the goto label when wrb_hanlde allocation fails.
Fix the error message display in beiscsi_alloc_pdu
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:12:49 +0000 (04:42 +0530)]
[SCSI] be2iscsi: Fix kernel panic in blk_iopoll disable mode.
Kernel used to panic while running IO is disable mode, as there
was an issue with getting the correct EQ on which completion has come.
Fix done is create workqueue per hba and work item for each EQ created.
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:12:37 +0000 (04:42 +0530)]
[SCSI] be2iscsi: Issue an function level reset when driver is loaded
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:12:25 +0000 (04:42 +0530)]
[SCSI] be2iscsi: Display driver name and version in device attribute
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:12:13 +0000 (04:42 +0530)]
[SCSI] be2iscsi: Fix max supported EQ count to 8.
The maximum EQ that can be created for a function is 8. Check the
CPU online count and create only 8 EQ if CPU_Count >= 8
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:12:00 +0000 (04:42 +0530)]
[SCSI] be2iscsi: Fix memory leak in control path of driver
In contorl path of the driver the task was mapped using
pci_map_single which was not unmapped when the completion
for the task had come.
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
John Soni Jose [Fri, 19 Oct 2012 23:11:45 +0000 (04:41 +0530)]
[SCSI] be2iscsi: Display Completion Event string instead of Opcode
Display the event string along with the opcode and CID on which
an event has occured.
Signed-off-by: John Soni Jose <sony.john-n@emulex.com> Signed-off-by: Jayamohan Kallickal <jayamohan.kallickal@emulex.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Boaz Harrosh [Wed, 24 Oct 2012 21:51:41 +0000 (14:51 -0700)]
[SCSI] osd_uld: Add osdname & systemid sysfs at scsi_osd class
This patch adds the support for the following two read-only sysfs attributes
to scsi_osd class members : osdname & systemid
These attributes will show up as below in sysfs class hierarchy:
/sys/class/scsi_osd/osdX/osdname
/sys/class/scsi_osd/osdX/systemid
The osdname & systemid are OSD device attributes which uniquely
identify a device on the network, while it's IP and certainly
it's /dev/osdX device path might change.
Userspace utilities (e.g. mkfs.exofs) can parse these attributes to
identify the correct OSD in safer and faster way.
(Today osd apps open each device in the system and send a
attributes query for these, in order to access the user
requested device)
Joel D. Diaz [Wed, 10 Oct 2012 08:36:11 +0000 (10:36 +0200)]
[SCSI] sd: Reshuffle init_sd to avoid crash
scsi_register_driver will register a prep_fn() function, which
in turn migh need to use the sd_cdp_pool for DIF.
Which hasn't been initialised at this point, leading to
a crash. So reshuffle the init_sd() and exit_sd() paths
to have the driver registered last.
Signed-off-by: Joel D. Diaz <joeldiaz@us.ibm.com> Signed-off-by: Hannes Reinecke <hare@suse.de> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Linus Torvalds [Mon, 26 Nov 2012 01:57:01 +0000 (17:57 -0800)]
Merge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
Pull powerpc EEH bugfixes from Benjamin Herrenschmidt.
Two one-liner fixes for the new EEH code.
* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc:
powerpc/eeh: Do not invalidate PE properly
powerpc/pseries: Fix oops with MSIs when missing EEH PEs
Linus Torvalds [Mon, 26 Nov 2012 01:55:04 +0000 (17:55 -0800)]
Merge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus
Pull MIPS fixes from Ralf Baechle:
"Three issues fixed accross the field:
- Some functions that were recently outlined as part of a preemption
fix were causing problems with function tracing.
- The recently merged in-kernel MPI library uses very outdated
headers that contain MIPS-specific code which won't build on with
gcc 4.4 or newer.
- The MIPS non-NUMA memory initialization was making only a very
half-baked attempt at merging adjacent memory ranges. This kept
the code simple enough but is now causing issues with kexec."
* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus:
MPI: Fix compilation on MIPS with GCC 4.4 and newer
MIPS: Fix crash that occurs when function tracing is enabled
MIPS: Merge overlapping bootmem ranges
Gavin Shan [Thu, 22 Nov 2012 21:58:26 +0000 (21:58 +0000)]
powerpc/eeh: Do not invalidate PE properly
While the EEH does recovery on the specific PE that has PCI errors,
the PCI devices belonging to the PE will be removed and the PE will
be marked as invalid since we still need the information stored in
the PE. We only invalidate the PE when it doesn't have associated
EEH devices and valid child PEs. However, the code used to check
that is wrong. The patch fixes that.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Linus Torvalds [Sat, 24 Nov 2012 06:03:14 +0000 (20:03 -1000)]
Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 arch fixes from Peter Anvin:
"Here is a collection of fixes for 3.7-rc7. This is a superset of
tglx' earlier pull request."
* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86-64: Fix ordering of CFI directives and recent ASM_CLAC additions
x86, microcode, AMD: Add support for family 16h processors
x86-32: Export kernel_stack_pointer() for modules
x86-32: Fix invalid stack address while in softirq
x86, efi: Fix processor-specific memcpy() build error
x86: remove dummy long from EFI stub
x86, mm: Correct vmflag test for checking VM_HUGETLB
x86, amd: Disable way access filter on Piledriver CPUs
x86/mce: Do not change worker's running cpu in cmci_rediscover().
x86/ce4100: Fix PCI configuration register access for devices without interrupts
x86/ce4100: Fix reboot by forcing the reboot method to be KBD
x86/ce4100: Fix pm_poweroff
MAINTAINERS: Update email address for Robert Richter
x86, microcode_amd: Change email addresses, MAINTAINERS entry
MAINTAINERS: Change Boris' email address
EDAC: Change Boris' email address
x86, AMD: Change Boris' email address
Linus Torvalds [Sat, 24 Nov 2012 01:12:17 +0000 (15:12 -1000)]
Merge tag 'for-linus-20121123' of git://git.infradead.org/mtd-2.6
Pull MTD fixes from David Woodhouse:
"The most important part of this is that it fixes a regression in
Samsung NAND chip detection, introduced by some rework which went into
3.7. The initial fix wasn't quite complete, so it's in two parts. In
fact the first part is committed twice (Artem committed his own copy
of the same patch) and I've merged Artem's tree into mine which
already had that fix.
I'd have recommitted that to make it somewhat cleaner, but figured by
this point in the release cycle it was better to merge *exactly* the
commits which have been in linux-next.
If I'd recommitted, I'd also omit the sparse warning fix. But it's
there, and it's harmless — just marking one function as 'static' in
onenand code.
This also includes a couple more fixes for stable: an AB-BA deadlock
in JFFS2, and an invalid range check in slram."
* tag 'for-linus-20121123' of git://git.infradead.org/mtd-2.6:
mtd: nand: fix Samsung SLC detection regression
mtd: nand: fix Samsung SLC NAND identification regression
jffs2: Fix lock acquisition order bug in jffs2_write_begin
mtd: onenand: Make flexonenand_set_boundary static
mtd: slram: invalid checking of absolute end address
mtd: ofpart: Fix incorrect NULL check in parse_ofoldpart_partitions()
mtd: nand: fix Samsung SLC NAND identification regression
Linus Torvalds [Fri, 23 Nov 2012 22:36:06 +0000 (12:36 -1000)]
Merge tag 'devicetree-for-linus' of git://git.secretlab.ca/git/linux-2.6
Pull device tree regression fix from Grant Likely:
"Simple build regression fix for DT device drivers on Sparc. An
earlier change had masked out the of_iomap() helper on SPARC."
* tag 'devicetree-for-linus' of git://git.secretlab.ca/git/linux-2.6:
of/address: sparc: Declare of_iomap as an extern function for sparc again
Linus Torvalds [Fri, 23 Nov 2012 22:16:43 +0000 (12:16 -1000)]
Merge tag 'pm-for-3.7-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull power management update from Rafael Wysocki:
"Fix for an incorrect error condition check in device PM QoS code that
may lead to an Oops from Guennadi Liakhovetski."
* tag 'pm-for-3.7-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
PM / QoS: fix wrong error-checking condition
Linus Torvalds [Fri, 23 Nov 2012 22:11:13 +0000 (12:11 -1000)]
Merge tag 'md-3.7-fixes' of git://neil.brown.name/md
Pull md fixes from NeilBrown:
"Several bug fixes for md in 3.7:
- raid5 discard has problems
- raid10 replacement devices have problems
- bad block lock seqlock usage has problems
- dm-raid doesn't free everything"
* tag 'md-3.7-fixes' of git://neil.brown.name/md:
md/raid10: decrement correct pending counter when writing to replacement.
md/raid10: close race that lose writes lost when replacement completes.
md/raid5: Make sure we clear R5_Discard when discard is finished.
md/raid5: move resolving of reconstruct_state earlier in stripe_handle.
md/raid5: round discard alignment up to power of 2.
md: make sure everything is freed when dm-raid stops an array.
md: Avoid write invalid address if read_seqretry returned true.
md: Reassigned the parameters if read_seqretry returned true in func md_is_badblock.
Linus Torvalds [Fri, 23 Nov 2012 22:06:05 +0000 (12:06 -1000)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-block
Pull block layer fixes from Jens Axboe:
"Distilled down version of bug fixes for 3.7. The patches have been
well tested. If you notice that commit dates are from today, it's
because I pulled less important bits out and shuffled them into the
3.8 mix. Apart from that, no changes, base still the same.
It contains:
- Fix for aoe, don't run request_fn while it's plugged.
- Fix for a regression in floppy since 3.6, which causes problems if
no floppy is found.
- Stable fix for blk_exec(), don't touch a request after it has been
sent to the scheduler (and the device as well).
- Five fixes for various nasties in mtip32xx."
* 'for-linus' of git://git.kernel.dk/linux-block:
block: Don't access request after it might be freed
mtip32xx: Fix padding issue
aoe: avoid running request handler on plugged queue
mtip32xx: fix potential NULL pointer dereference in mtip_timeout_function()
mtip32xx: fix shift larger than type warning
mtip32xx: Fix incorrect mask used for erase mode
mtip32xx: Fix to make lba address correct in big-endian systems
mtip32xx: fix potential crash on SEC_ERASE_UNIT
dm: fix deadlock with request based dm and queue request_fn recursion
floppy: destroy floppy workqueue before cleaning up the queue
Andreas Larsson [Fri, 23 Nov 2012 11:24:09 +0000 (12:24 +0100)]
of/address: sparc: Declare of_iomap as an extern function for sparc again
This bug-fix makes sure that of_iomap is defined extern for sparc so that the
sparc-specific implementation of_iomap is once again used when including
include/linux/of_address.h in a sparc context. OF_GPIO that is now available for
sparc relies on this.
The bug was inadvertently introduced in a850a75, "of/address: add empty static
inlines for !CONFIG_OF", that added a static dummy inline for of_iomap when
!CONFIG_OF_ADDRESS. However, CONFIG_OF_ADDRESS is never defined for sparc, but
there is a sparc-specific implementation /arch/sparc/kernel/of_device_common.c.
This fix takes the same approach as 0bce04b that solved the equivalent problem
for of_address_to_resource.
Signed-off-by: Andreas Larsson <andreas@gaisler.com> Acked-by: David Miller <davem@davemloft.net> Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Linus Torvalds [Fri, 23 Nov 2012 22:01:02 +0000 (12:01 -1000)]
Merge tag 'omapdss-for-3.7-rc' of git://gitorious.org/linux-omap-dss2/linux
Pull omapdss fixes from Tomi Valkeinen:
"Here are a few OMAPDSS fixes for the next -rc. I'm sending these
directly to you, and quite late, as the fbdev tree maintainer
(Florian) has been busy with his work and hasn't had time to manage
the fb patches."
* tag 'omapdss-for-3.7-rc' of git://gitorious.org/linux-omap-dss2/linux:
OMAPDSS: do not fail if dpll4_m4_ck is missing
OMAPFB: Fix possible null pointer dereferencing
OMAPDSS: HDMI: fix missing unlock on error in hdmi_dump_regs()
omapdss: dss: Fix clocks on OMAP363x
OMAPDSS: DSI: fix dsi_get_dsidev_from_id()
Linus Torvalds [Fri, 23 Nov 2012 21:59:26 +0000 (11:59 -1000)]
Merge branch 'i2c-embedded/for-current' of git://git.pengutronix.de/git/wsa/linux
Pull i2c fixes from Wolfram Sang:
"Bugfixes for the i2c subsystem.
Except for a few one-liners, there is mainly one revert because of an
overlooked dependency. Since there is no linux-next at the moment, I
did some extra testing, and all was fine for me."
* 'i2c-embedded/for-current' of git://git.pengutronix.de/git/wsa/linux:
i2c: mxs: Handle i2c DMA failure properly
i2c: s3c2410: Fix code to free gpios
i2c: omap: ensure writes to dev->buf_len are ordered
Revert "ARM: OMAP: convert I2C driver to PM QoS for MPU latency constraints"
i2c: at91: fix SMBus quick command
Linus Torvalds [Fri, 23 Nov 2012 21:58:28 +0000 (11:58 -1000)]
Merge tag 'sound-3.7' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Pull sound fixes from Takashi Iwai:
"The highlight of this update is the fixes for ASoC kirkwood by
Russell. In addition to that, a couple of regression fixes for
HD-audio due to the runtime PM support on 3.7, and other driver-
specific regression fixes like USB MIDI on non-standard USB audio
drivers."
* tag 'sound-3.7' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
ALSA: snd-usb: properly initialize the sync endpoint
ALSA: hda - Cirrus: Correctly clear line_out_pins when moving to speaker
ALSA: hda - Add support for Realtek ALC292
ASoC: kirkwood-i2s: more pause-mode fixes
ASoC: kirkwood-i2s: fix DMA underruns
ASoC: kirkwood-i2s: fix DCO lock detection
ASoC: kirkwood-dma: don't ignore other irq causes on error
ASoC: kirkwood-dma: fix use of virt_to_phys()
ALSA: hda - Limit runtime PM support only to known Intel chips
ALSA: hda - Fix recursive suspend/resume call
ALSA: ua101, usx2y: fix broken MIDI output
ASoC: arizona: Fix typo - Swap value in 48k_rates[] and 44k1_rates[]
ASoC: bells: Fix up git patch application failure
ASoC: cs4271: free allocated GPIO
Pull networkign fixes from David Miller:
"Networking bug fixes, Cacio e Pepe edition:
1) BNX2X accidently accesses chip rev specific registers without an
appropriate guard, fix from Ariel Elior.
2) When we removed the routing cache, we set ip_rt_max_size to ~0 just
to keep reporting a value to userspace via sysfs. But the ipv4
IPSEC layer was using this to tune itself which is completely bogus
to now do. Fix from Steffen Klassert.
3) Missing initialization in netfilter ipset code from Jozsef
Kadlecsik.
4) Check CTA_TIMEOUT_NAME length properly in netfilter cttimeout code,
fix from Florian Westphal.
5) After removing the routing cache, we inadvertantly are caching
multicast routes that end up looping back locally, we cannot do
that legitimately any more. Fix from Julian Anastasov.
6) Revert a race fix for 8139cp qemu/kvm that doesn't actually work
properly on real hardware. From Francois Romieu.
7) Fixup errors in example command lines in VXLAN device docs."
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
bnx2x: remove redundant warning log
vxlan: fix command usage in its doc
8139cp: revert "set ring address before enabling receiver"
ipv4: do not cache looped multicasts
netfilter: cttimeout: fix buffer overflow
netfilter: ipset: Fix range bug in hash:ip,port,net
xfrm: Fix the gc threshold value for ipv4
Linus Torvalds [Fri, 23 Nov 2012 21:54:22 +0000 (11:54 -1000)]
Merge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Pull one more ARM SoC fix from Olof Johansson:
"I missed one pull request from Samsung with one fix in the previous
batch. Here it is -- a dma driver fix for an early version of silicon
that they still support."
* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
ARM: EXYNOS: PL330 MDMA1 fix for revision 0 of Exynos4210 SOC
dev_pm_qos_add_request() can return 0, 1, or a negative error code,
therefore the correct error test is "if (error < 0)." Checking just for
non-zero return code leads to erroneous setting of the req->dev pointer
to NULL, which then leads to a repeated call to
dev_pm_qos_add_ancestor_request() in st1232_ts_irq_handler(). This in turn
leads to an Oops, when the I2C host adapter is unloaded and reloaded again
because of the inconsistent state of its QoS request list.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Cc: <stable@vger.kernel.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Ariel Elior [Thu, 22 Nov 2012 07:16:17 +0000 (07:16 +0000)]
bnx2x: remove redundant warning log
fix bug where a register which was only meant to be read in 578xx/57712
devices causes a bogus error message to be logged when read from other
devices.
Signed-off-by: Ariel Elior <ariele@broadcom.com> Signed-off-by: Eilon Greenstein <eilong@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Manuel Lauss [Thu, 22 Nov 2012 10:58:22 +0000 (11:58 +0100)]
MPI: Fix compilation on MIPS with GCC 4.4 and newer
Since 4.4 GCC on MIPS no longer recognizes the "h" constraint,
leading to this build failure:
CC lib/mpi/generic_mpih-mul1.o
lib/mpi/generic_mpih-mul1.c: In function 'mpihelp_mul_1':
lib/mpi/generic_mpih-mul1.c:50:3: error: impossible constraint in 'asm'
This patch updates MPI with the latest umul_ppm implementations for MIPS.
Al Cooper [Thu, 15 Nov 2012 23:16:14 +0000 (18:16 -0500)]
MIPS: Fix crash that occurs when function tracing is enabled
A recent patch changed some irq routines from inlines to functions.
These routines are called by the tracer code. Now that they're functions,
if they are compiled for function tracing they will call the tracer
and crash the system due to infinite recursion. The fix disables
tracing in these functions by using "notrace" in the function
definition.
Signed-off-by: Al Cooper <alcooperx@gmail.com> Reviewed-by: David Daney <david.daney@cavium.com>
Pathchwork: https://patchwork.linux-mips.org/patch/4564/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
but the two System RAM ranges should be one single range. This particular
case will result in kexec failure on Octeon systems if the kernel being
loaded by kexec is bigger than the already running kernel.
Roland Dreier [Thu, 22 Nov 2012 10:00:11 +0000 (02:00 -0800)]
block: Don't access request after it might be freed
After we've done __elv_add_request() and __blk_run_queue() in
blk_execute_rq_nowait(), the request might finish and be freed
immediately. Therefore checking if the type is REQ_TYPE_PM_RESUME
isn't safe afterwards, because if it isn't, rq might be gone.
Instead, check beforehand and stash the result in a temporary.
This fixes crashes in blk_execute_rq_nowait() I get occasionally when
running with lots of memory debugging options enabled -- I think this
race is usually harmless because the window for rq to be reallocated
is so small.
Signed-off-by: Roland Dreier <roland@purestorage.com> Cc: stable@kernel.org Signed-off-by: Jens Axboe <axboe@kernel.dk>
Ed Cashin [Fri, 9 Nov 2012 00:17:15 +0000 (19:17 -0500)]
aoe: avoid running request handler on plugged queue
Calling the request handler directly on a plugged queue defeats
the performance improvements provided by the plugging mechanism.
Use the __blk_run_queue function instead of calling the request
handler directly, so that we don't interfere with the block
layer's ability to plug the queue.
Signed-off-by: Ed Cashin <ecashin@coraid.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Selvan Mani [Wed, 7 Nov 2012 13:03:53 +0000 (06:03 -0700)]
mtip32xx: Fix to make lba address correct in big-endian systems
Earlier lba address was assigned directly to lba_low and lba_low_ex,
which would result in a different number (bytes reversed) in
big-endian systems. Now assigning lba address byte-by-byte to fis.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Selvan Mani <smani@micron.com> Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Selvan Mani [Wed, 7 Nov 2012 13:03:37 +0000 (06:03 -0700)]
mtip32xx: fix potential crash on SEC_ERASE_UNIT
The mtip driver lifted this code from elsewhere and then added a special
handling check for SEC_ERASE_UNIT. If the caller tries to do a security
erase but passes no output data for the command then outbuf is not
allocated and the driver duly explodes.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Alan Cox <alan@linux.intel.com> Signed-off-by: Selvan Mani <smani@micron.com> Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
Jens Axboe [Tue, 6 Nov 2012 11:24:26 +0000 (12:24 +0100)]
dm: fix deadlock with request based dm and queue request_fn recursion
Request based dm attempts to re-run the request queue off the
request completion path. If used with a driver that potentially does
end_io from its request_fn, we could deadlock trying to recurse
back into request dispatch. Fix this by punting the request queue
run to kblockd.
Tested to fix a quickly reproducible deadlock in such a scenario.
Jiri Kosina [Tue, 6 Nov 2012 10:47:13 +0000 (11:47 +0100)]
floppy: destroy floppy workqueue before cleaning up the queue
We need to first destroy the floppy_wq workqueue before cleaning up
the queue. Otherwise we might race with still pending work with the
workqueue, but all the block queue already gone. This might lead to
various oopses, such as
Linus Torvalds [Fri, 23 Nov 2012 07:45:34 +0000 (21:45 -1000)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Pull input updates from Dmitry Torokhov:
"This fixes recent regression where /dev/input/mice got assigned wrong
device node which messed up setups with static /dev, and a regression
in ads7846 GPIO debounce setup."
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
ARM - OMAP: ads7846: fix pendown debounce setting
Input: ads7846 - enable pendown GPIO debounce time setting
Input: mousedev - move /dev/input/mice to the correct minor
Input: MT - document new 'flags' argument of input_mt_init_slots()
Olof Johansson [Fri, 23 Nov 2012 04:43:20 +0000 (20:43 -0800)]
Merge branch 'v3.7-samsung-fixes-3' of git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung into fixes
From Kukjin Kim:
Here is Samsung fixes for v3.7 and it is for fixing of mdma1 address
for exynos4210 rev0 SoC.
* 'v3.7-samsung-fixes-3' of git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung:
ARM: EXYNOS: PL330 MDMA1 fix for revision 0 of Exynos4210 SOC
powerpc/pseries: Fix oops with MSIs when missing EEH PEs
The new EEH code introduced a small regression, if the EEH PEs
are missin (which happens currently in qemu for example), it
will deref a NULL pointer in the MSI code.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Julian Anastasov [Thu, 22 Nov 2012 21:04:14 +0000 (23:04 +0200)]
ipv4: do not cache looped multicasts
Starting from 3.6 we cache output routes for
multicasts only when using route to 224/4. For local receivers
we can set RTCF_LOCAL flag depending on the membership but
in such case we use maddr and saddr which are not caching
keys as before. Additionally, we can not use same place to
cache routes that differ in RTCF_LOCAL flag value.
Fix it by caching only RTCF_MULTICAST entries
without RTCF_LOCAL (send-only, no loopback). As a side effect,
we avoid unneeded lookup for fnhe when not caching because
multicasts are not redirected and they do not learn PMTU.
Thanks to Maxime Bizon for showing the caching
problems in __mkroute_output for 3.6 kernels: different
RTCF_LOCAL flag in cache can lead to wrong ip_mc_output or
ip_output call and the visible problem is that traffic can
not reach local receivers via loopback.
Reported-by: Maxime Bizon <mbizon@freebox.fr> Tested-by: Maxime Bizon <mbizon@freebox.fr> Signed-off-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Thu, 22 Nov 2012 20:24:30 +0000 (15:24 -0500)]
Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec
Steffen Klassert says:
====================
This pull request is intended for 3.7 and contains a single patch to
fix the IPsec gc threshold value for ipv4.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Mack [Thu, 22 Nov 2012 19:27:59 +0000 (20:27 +0100)]
ALSA: snd-usb: properly initialize the sync endpoint
Jeffrey Barish reported an obvious bug in the pcm part of the usb-audio
driver which causes the code to not initialize the sync endpoint from
configure_endpoint().
Linus Torvalds [Thu, 22 Nov 2012 19:22:13 +0000 (09:22 -1000)]
Merge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Pull ARM SoC fixes from Olof Johansson:
"A few more fixes for final 3.7. Two dealing with pinmux setup on
OMAP, and one dealing with TV output on DaVinci. And one small
MAINTAINER update."
* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
ARM: davinci: dm644x: fix out range signal for ED
ARM: OMAP4: TWL: mux sys_drm_msecure as output for PMIC
ARM: OMAP3: igep0020: Set WIFI/BT GPIO pins in correct mux mode
ARM: OMAP: Add maintainer entry for IGEP machines
Linus Torvalds [Thu, 22 Nov 2012 19:16:29 +0000 (09:16 -1000)]
Merge tag 'parisc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/parisc-2.6
Pull PARISC fixes from James Bottomley:
"This is two bug fixes: one fixes a loophole where rt_sigprocmask()
with the wrong values panics the box (Denial of Service) and the other
fixes an aliasing problem with get_shared_area() which could cause
data corruption.
Signed-off-by: James Bottomley <JBottomley@Parallels.com>"
* tag 'parisc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/parisc-2.6:
[PARISC] fix user-triggerable panic on parisc
[PARISC] fix virtual aliasing issue in get_shared_area()
Linus Torvalds [Thu, 22 Nov 2012 19:14:54 +0000 (09:14 -1000)]
Merge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Pull SCSI fixes from James Bottomley:
"This is a set of four bug fixes.
The isci one is an obvious thinko (using request buffer instead of
response buffer) which causes a command to fail.
The three others are DIF/DIX updates which are required because
they're part of a series of ten patches, the other seven of which went
into the block layer during the merge window meaning our current
DIF/DIX implementation is broken without these three.
Signed-off-by: James Bottomley <JBottomley@Parallels.com>"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
[SCSI] sd: Implement support for WRITE SAME
[SCSI] sd: Permit merged discard requests
[SCSI] Add a report opcode helper
[SCSI] isci: copy fis 0x34 response into proper buffer
Linus Torvalds [Thu, 22 Nov 2012 19:14:24 +0000 (09:14 -1000)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Pull drm fixes from Dave Airlie.
Small fixes for (mostly Nouveau, some radeon) regressions.
* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
drm/nouveau: use the correct fence implementation for nv50
drm/radeon: add new SI pci id
radeon: add AGPMode 1 quirk for RV250
drm/radeon: properly track the crtc not_enabled case evergreen_mc_stop()
drm/nouveau/bios: fix DCB v1.5 parsing
drm/nouveau: add missing pll_calc calls
drm/nouveau: fix crash with noaccel=1
drm/nv40: allocate ctxprog with kmalloc
drm/nvc0/disp: fix thinko in vblank regression fix..
NeilBrown [Thu, 22 Nov 2012 04:12:09 +0000 (15:12 +1100)]
md/raid10: decrement correct pending counter when writing to replacement.
When a write to a replacement device completes, we carefully
and correctly found the rdev that the write actually went to
and the blithely called rdev_dec_pending on the primary rdev,
even if this write was to the replacement.
This means that any writes to an array while a replacement
was ongoing would cause the nr_pending count for the primary
device to go negative, so it could never be removed.
This bug has been present since replacement was introduced in
3.3, so it is suitable for any -stable kernel since then.
NeilBrown [Thu, 22 Nov 2012 03:42:49 +0000 (14:42 +1100)]
md/raid10: close race that lose writes lost when replacement completes.
When a replacement operation completes there is a small window
when the original device is marked 'faulty' and the replacement
still looks like a replacement. The faulty should be removed and
the replacement moved in place very quickly, bit it isn't instant.
So the code write out to the array must handle the possibility that
the only working device for some slot in the replacement - but it
doesn't. If the primary device is faulty it just gives up. This
can lead to corruption.
So make the code more robust: if either the primary or the
replacement is present and working, write to them. Only when
neither are present do we give up.
This bug has been present since replacement was introduced in
3.3, so it is suitable for any -stable kernel since then.
projects / linux-beck.git / log