9 years agoMerge branch 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 3 Feb 2011 16:52:05 +0000]
Merge branch 'perf-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  perf: Fix reading in perf_event_read()
  watchdog: Don't change watchdog state on read of sysctl
  watchdog: Fix sysctl consistency
  watchdog: Fix broken nowatchdog logic
  perf: Fix Pentium4 raw event validation
  perf: Fix alloc_callchain_buffers()

9 years agoperf: Fix reading in perf_event_read()
Peter Zijlstra [Wed, 26 Jan 2011 14:38:35 +0000]
perf: Fix reading in perf_event_read()

It is quite possible for the event to have been disabled between
perf_event_read() sending the IPI and the CPU servicing the IPI and
calling __perf_event_read(), hence revalidate the state.

Reported-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

9 years agoRevert "exofs: Set i_mapping->backing_dev_info anyway"
Boaz Harrosh [Wed, 2 Feb 2011 19:02:12 +0000]
Revert "exofs: Set i_mapping->backing_dev_info anyway"

This reverts commit 115e19c53501edc11f730191f7f047736815ae3d.

Apparently setting inode->bdi to one's own sb->s_bdi stops VFS from
sending *read-aheads*.  This problem was bisected to this commit.  A
revert fixes it.  I'll investigate farther why is this happening for the
next Kernel, but for now a revert.

I'm sending to stable@kernel.org as well, since it exists also in
2.6.37.  2.6.36 is good and does not have this patch.

CC: Stable Tree <stable@kernel.org>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoMerge branch 'media_fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Thu, 3 Feb 2011 01:52:19 +0000]
Merge branch 'media_fixes' of git://git./linux/kernel/git/mchehab/linux-2.6

* 'media_fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6:
  [media] fix saa7111 non-detection
  [media] rc/streamzap: fix reporting response times
  [media] mceusb: really fix remaining keybounce issues
  [media] rc: use time unit conversion macros correctly
  [media] rc/ir-lirc-codec: add back debug spew
  [media] ir-kbd-i2c: improve remote behavior with z8 behind usb
  [media] lirc_zilog: z8 on usb doesn't like back-to-back i2c_master_send
  [media] hdpvr: fix up i2c device registration
  [media] rc/mce: add mappings for missing keys
  [media] gspca - zc3xx: Discard the partial frames
  [media] gspca - zc3xx: Fix bad images with the sensor hv7131r
  [media] gspca - zc3xx: Bad delay when given by a table

9 years agoMerge branch 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6
Linus Torvalds [Thu, 3 Feb 2011 01:51:31 +0000]
Merge branch 'for-linus' of git://git390.marist.edu/linux-2.6

* 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6:
  [S390] reset default for CONFIG_CHSC_SCH
  [S390] qdio: prevent compile warning under CONFIG_32BIT
  [S390] use asm-generic/cacheflush.h
  [S390] tlb: fix build error caused by THP
  [S390] missing sacf in uaccess
  [S390] pgtable_list corruption
  [S390] dasd: prevent panic with unresumed devices

9 years agofs: make block fiemap mapping length at least blocksize long
Josef Bacik [Tue, 1 Feb 2011 23:52:47 +0000]
fs: make block fiemap mapping length at least blocksize long

Some filesystems don't deal well with being asked to map less than
blocksize blocks (GFS2 for example).  Since we are always mapping at least
blocksize sections anyway, just make sure len is at least as big as a
blocksize so we don't trip up any filesystems.  Thanks,

Signed-off-by: Josef Bacik <josef@redhat.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agovfs: sparse: add __FMODE_EXEC
Namhyung Kim [Tue, 1 Feb 2011 23:52:46 +0000]
vfs: sparse: add __FMODE_EXEC

FMODE_EXEC is a constant type of fmode_t but was used with normal integer
constants.  This results in following warnings from sparse.  Fix it using
new macro __FMODE_EXEC.

 fs/exec.c:116:58: warning: restricted fmode_t degrades to integer
 fs/exec.c:689:58: warning: restricted fmode_t degrades to integer
 fs/fcntl.c:777:9: warning: restricted fmode_t degrades to integer

Signed-off-by: Namhyung Kim <namhyung@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agovfs: sparse: remove a warning on OPEN_FMODE()
Namhyung Kim [Tue, 1 Feb 2011 23:52:46 +0000]
vfs: sparse: remove a warning on OPEN_FMODE()

AND-ing FMODE_* constant with normal integer results in following
sparse warnings. Fix it.

 fs/open.c:662:21: warning: restricted fmode_t degrades to integer
 fs/anon_inodes.c:123:34: warning: restricted fmode_t degrades to integer

Signed-off-by: Namhyung Kim <namhyung@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomemcg: fix event counting breakage from recent THP update
KAMEZAWA Hiroyuki [Tue, 1 Feb 2011 23:52:45 +0000]
memcg: fix event counting breakage from recent THP update

Changes in e401f1761 ("memcg: modify accounting function for supporting
THP better") adds nr_pages to support multiple page size in
memory_cgroup_charge_statistics.

But counting the number of event nees abs(nr_pages) for increasing
counters.  This patch fixes event counting.

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomemcg: never OOM when charging huge pages
Johannes Weiner [Tue, 1 Feb 2011 23:52:44 +0000]
memcg: never OOM when charging huge pages

Huge page coverage should obviously have less priority than the continued
execution of a process.

Never kill a process when charging it a huge page fails.  Instead, give up
after the first failed reclaim attempt and fall back to regular pages.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomemcg: prevent endless loop when charging huge pages to near-limit group
Johannes Weiner [Tue, 1 Feb 2011 23:52:43 +0000]
memcg: prevent endless loop when charging huge pages to near-limit group

If reclaim after a failed charging was unsuccessful, the limits are
checked again, just in case they settled by means of other tasks.

This is all fine as long as every charge is of size PAGE_SIZE, because in
that case, being below the limit means having at least PAGE_SIZE bytes
available.

But with transparent huge pages, we may end up in an endless loop where
charging and reclaim fail, but we keep going because the limits are not
yet exceeded, although not allowing for a huge page.

Fix this up by explicitely checking for enough room, not just whether we
are within limits.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomemcg: prevent endless loop when charging huge pages
Johannes Weiner [Tue, 1 Feb 2011 23:52:42 +0000]
memcg: prevent endless loop when charging huge pages

The charging code can encounter a charge size that is bigger than a
regular page in two situations: one is a batched charge to fill the
per-cpu stocks, the other is a huge page charge.

This code is distributed over two functions, however, and only the outer
one is aware of huge pages.  In case the charging fails, the inner
function will tell the outer function to retry if the charge size is
bigger than regular pages--assuming batched charging is the only case.
And the outer function will retry forever charging a huge page.

This patch makes sure the inner function can distinguish between batch
charging and a single huge page charge.  It will only signal another
attempt if batch charging failed, and go into regular reclaim when it is
called on behalf of a huge page.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agothp: fix unsuitable behavior for hwpoisoned tail page
Jin Dongming [Tue, 1 Feb 2011 23:52:41 +0000]
thp: fix unsuitable behavior for hwpoisoned tail page

When a tail page of THP is poisoned, memory-failure will do nothing except
setting PG_hwpoison, while the expected behavior is that the process, who
is using the poisoned tail page, should be killed.

The above problem is caused by lru check of the poisoned tail page of THP.
Because PG_lru flag is only set on the head page of THP, the check always
consider the poisoned tail page as NON lru page.

So the lru check for the tail page of THP should be avoided, as like as
hugetlb.

This patch adds !PageTransCompound() before lru check for THP, because of
the check (!PageHuge() && !PageTransCompound()) the whole branch could be
optimized away at build time when both hugetlbfs and THP are set with "N"
(or in archs not supporting either of those).

[akpm@linux-foundation.org: fix unrelated typo in shake_page() comment]
Signed-off-by: Jin Dongming <jin.dongming@np.css.fujitsu.com>
Reviewed-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agothp: fix the wrong reported address of hwpoisoned hugepages
Jin Dongming [Tue, 1 Feb 2011 23:52:40 +0000]
thp: fix the wrong reported address of hwpoisoned hugepages

When the tail page of THP is poisoned, the head page will be poisoned too.
 And the wrong address, address of head page, will be sent with sigbus
always.

So when the poisoned page is used by Guest OS which is running on KVM,
after the address changing(hva->gpa) by qemu, the unexpected process on
Guest OS will be killed by sigbus.

What we expected is that the process using the poisoned tail page could be
killed on Guest OS, but not that the process using the healthy head page
is killed.

Since it is not good to poison the healthy page, avoid poisoning other
than the page which is really poisoned.
  (While we poison all pages in a huge page in case of hugetlb,
   we can do this for THP thanks to split_huge_page().)

Here we fix two parts:
  1. Isolate the poisoned page only to make sure
     the reported address is the address of poisoned page.
  2. make the poisoned page work as the poisoned regular page.

[akpm@linux-foundation.org: fix spello in comment]
Signed-off-by: Jin Dongming <jin.dongming@np.css.fujitsu.com>
Reviewed-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agothp: fix splitting of hwpoisoned hugepages
Jin Dongming [Tue, 1 Feb 2011 23:52:39 +0000]
thp: fix splitting of hwpoisoned hugepages

The poisoned THP is now split with split_huge_page() in
collect_procs_anon().  If kmalloc() is failed in collect_procs(),
split_huge_page() could not be called.  And the work after
split_huge_page() for collecting the processes using poisoned page will
not be done, too.  So the processes using the poisoned page could not be
killed.

The condition becomes worse when CONFIG_DEBUG_VM == "Y".  Because the
poisoned THP could not be split, system panic will be caused by
VM_BUG_ON(PageTransHuge(page)) in try_to_unmap().

This patch does:
  1. move split_huge_page() to the place before collect_procs().
     This can be sure the failure of splitting THP is caused by itself.
  2. when splitting THP is failed, stop the operations after it.
     This can avoid unexpected system panic or non sense works.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Jin Dongming <jin.dongming@np.css.fujitsu.com>
Reviewed-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoMAINTAINERS: fixup Simtec support email entries
Ben Dooks [Tue, 1 Feb 2011 23:52:38 +0000]
MAINTAINERS: fixup Simtec support email entries

The support@simtec.co.uk address is for direct customer support only, the
EB2410ITX and EB110ATX entries should direct to the Simtec Linux Team
address of linux@simtec.co.uk

Also add correct email address for Vincent Sanders

[akpm@linux-foundation.org: fix Vincent's address]
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Cc: Vincent Sanders <vince@simtec.co.uk>
Cc: Simtec Support <support@simtec.co.uk>
Cc: Simtec Linux Team <linux@simtec.co.uk>
Cc: Jack Stone <jwjstone@fastmail.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoMAINTAINERS: fixup file entries for "SIMTEC EB2410ITX (BAST)"
Ben Dooks [Tue, 1 Feb 2011 23:52:37 +0000]
MAINTAINERS: fixup file entries for "SIMTEC EB2410ITX (BAST)"

Add the correct files for the Simtec BAST machine, ensuring the IDE and
IRQ routing are added, and move to the machine specific file instead of
trying to catch all of arch/arm/mach-s3c2410

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Cc: Simtec Linux Team <linux@simtec.co.uk>
Cc: Simtec Support <support@simtec.co.uk>
Cc: Jack Stone <jwjstone@fastmail.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoMAINTAINERS: move s3c2410 drivers to ARM/SAMSUNG ARM
Ben Dooks [Tue, 1 Feb 2011 23:52:36 +0000]
MAINTAINERS: move s3c2410 drivers to ARM/SAMSUNG ARM

There are currently two entries under the "SIMTEC EB2410ITX (BAST)"
machine entry for drivers/*/*s3c2410*, which is catching everything
s3c2410 driver related.

This entry is for a specific S3C2410 based machine, so move these two file
entries to the "ARM/SAMSUNG ARM ARCHITECTURES" entry, where it will reach
a wider audience of interested parties.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Cc: Simtec Linux Team <linux@simtec.co.uk>
Acked-by: Kukjin Kim <kgene.kim@samsung.com>
Cc: Jack Stone <jwjstone@fastmail.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoepoll: epoll_wait() should not use timespec_add_ns()
Eric Dumazet [Tue, 1 Feb 2011 23:52:35 +0000]
epoll: epoll_wait() should not use timespec_add_ns()

commit 95aac7b1cd224f ("epoll: make epoll_wait() use the hrtimer range
feature") added a performance regression because it uses timespec_add_ns()
with potential very large 'ns' values.

[akpm@linux-foundation.org: s/epoll_set_mstimeout/ep_set_mstimeout/, per Davide]
Reported-by: Simon Kirby <sim@hostway.ca>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Shawn Bohrer <shawn.bohrer@gmail.com>
Acked-by: Davide Libenzi <davidel@xmailserver.org>
Cc: <stable@kernel.org> [2.6.37.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm/migration: fix page corruption during hugepage migration
Minchan Kim [Tue, 1 Feb 2011 23:52:33 +0000]
mm/migration: fix page corruption during hugepage migration

If migrate_huge_page by memory-failure fails , it calls put_page in itself
to decrease page reference and caller of migrate_huge_page also calls
putback_lru_pages.  It can do double free of page so it can make page
corruption on page holder.

In addtion, clean of pages on caller is consistent behavior with
migrate_pages by cf608ac19c ("mm: compaction: fix COMPACTPAGEFAILED
counting").

Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomm: when migrate_pages returns 0, all pages must have been released
Andrea Arcangeli [Tue, 1 Feb 2011 23:52:32 +0000]
mm: when migrate_pages returns 0, all pages must have been released

In some cases migrate_pages could return zero while still leaving a few
pages in the pagelist (and some caller wouldn't notice it has to call
putback_lru_pages after commit cf608ac19c9 ("mm: compaction: fix
COMPACTPAGEFAILED counting")).

Add one missing putback_lru_pages not added by commit cf608ac19c95 ("mm:
compaction: fix COMPACTPAGEFAILED counting").

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Cc: Christoph Lameter <cl@linux.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomemsw: deprecate noswapaccount kernel parameter and schedule it for removal
Michal Hocko [Tue, 1 Feb 2011 23:52:31 +0000]
memsw: deprecate noswapaccount kernel parameter and schedule it for removal

noswapaccount couldn't be used to control memsw for both on/off cases so
we have added swapaccount[=0|1] parameter.  This way we can turn the
feature in two ways noswapaccount resp.  swapaccount=0.  We have kept the
original noswapaccount but I think we should remove it after some time as
it just makes more command line parameters without any advantages and also
the code to handle parameters is uglier if we want both parameters.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Requested-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agomemsw: handle swapaccount kernel parameter correctly
Michal Hocko [Tue, 1 Feb 2011 23:52:30 +0000]
memsw: handle swapaccount kernel parameter correctly

__setup based kernel command line parameters handlers which are handled in
obsolete_checksetup are provided with the parameter value including =
(more precisely everything right after the parameter name).

This means that the current implementation of swapaccount[=1|0] doesn't
work at all because if there is a value for the parameter then we are
testing for "0" resp.  "1" but we are getting "=0" resp.  "=1" and if
there is no parameter value we are getting an empty string rather than
NULL.

The original noswapccount parameter, which doesn't care about the value,
works correctly.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoMerge branch 'next' of git://git.monstr.eu/linux-2.6-microblaze
Linus Torvalds [Tue, 1 Feb 2011 23:23:58 +0000]
Merge branch 'next' of git://git.monstr.eu/linux-2.6-microblaze

* 'next' of git://git.monstr.eu/linux-2.6-microblaze:
  microblaze: Fix ASM optimized code for LE
  microblaze: Fix unaligned issue on MMU system with BS=0 DIV=1
  microblaze: Fix DTB passing from bootloader

9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Tue, 1 Feb 2011 23:22:40 +0000]
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6:
  cifs: fix length checks in checkSMB
  [CIFS] Update cifs minor version
  cifs: No need to check crypto blockcipher allocation
  cifs: clean up some compiler warnings
  cifs: make CIFS depend on CRYPTO_MD4
  cifs: force a reconnect if there are too many MIDs in flight
  cifs: don't pop a printk when sending on a socket is interrupted
  cifs: simplify SMB header check routine
  cifs: send an NT_CANCEL request when a process is signalled
  cifs: handle cancelled requests better
  cifs: fix two compiler warning about uninitialized vars

9 years agomlock: operate on any regions with protection != PROT_NONE
Michel Lespinasse [Tue, 1 Feb 2011 01:03:41 +0000]
mlock: operate on any regions with protection != PROT_NONE

As Tao Ma noticed, change 5ecfda0 breaks blktrace. This is because
blktrace mmaps a file with PROT_WRITE permissions but without PROT_READ,
so my attempt to not unnecessarity break COW during mlock ended up
causing mlock to fail with a permission problem.

I am proposing to let mlock ignore vma protection in all cases except
PROT_NONE. In particular, mlock should not fail for PROT_WRITE regions
(as in the blktrace case, which broke at 5ecfda0) or for PROT_EXEC
regions (which seem to me like they were always broken).

Signed-off-by: Michel Lespinasse <walken@google.com>
Acked-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agogpu/stub: fix acpi_video build error, fix stub kconfig dependencies
Randy Dunlap [Tue, 1 Feb 2011 19:17:28 +0000]
gpu/stub: fix acpi_video build error, fix stub kconfig dependencies

The comments under "config STUB_POULSBO" are close to correct,
but they are not being followed.  This patch updates them to reflect
the requirements for THERMAL.

This build error is caused by STUB_POULSBO selecting ACPI_VIDEO
when ACPI_VIDEO's config requirements are not met.

  ERROR: "thermal_cooling_device_register" [drivers/acpi/video.ko] undefined!
  ERROR: "thermal_cooling_device_unregister" [drivers/acpi/video.ko] undefined!

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoLinux 2.6.38-rc3
Linus Torvalds [Tue, 1 Feb 2011 03:05:49 +0000]
Linux 2.6.38-rc3

9 years agoMerge branch 'bugfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6
Linus Torvalds [Mon, 31 Jan 2011 23:41:02 +0000]
Merge branch 'bugfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6

* 'bugfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6:
  NFS: NFSv4 readdir loses entries
  NFS: Micro-optimize nfs4_decode_dirent()
  NFS: Fix an NFS client lockdep issue
  NFS construct consistent co_ownerid for v4.1
  NFS: nfs_wcc_update_inode() should set nfsi->attr_gencount
  NFS improve pnfs_put_deviceid_cache debug print
  NFS fix cb_sequence error processing
  NFS do not find client in NFSv4 pg_authenticate
  NLM: Fix "kernel BUG at fs/lockd/host.c:417!" or ".../host.c:283!"
  NFS: Prevent memory allocation failure in nfsacl_encode()
  NFS: nfsacl_{encode,decode} should return signed integer
  NFS: Fix "kernel BUG at fs/nfs/nfs3xdr.c:1338!"
  NFS: Fix "kernel BUG at fs/aio.c:554!"
  NFS4: Avoid potential NULL pointer dereference in decode_and_add_ds().
  NFS: fix handling of malloc failure during nfs_flush_multi()

9 years agocifs: fix length checks in checkSMB
Jeff Layton [Mon, 31 Jan 2011 14:14:17 +0000]
cifs: fix length checks in checkSMB

The cERROR message in checkSMB when the calculated length doesn't match
the RFC1001 length is incorrect in many cases. It always says that the
RFC1001 length is bigger than the SMB, even when it's actually the
reverse.

Fix the error message to say the reverse of what it does now when the
SMB length goes beyond the end of the received data. Also, clarify the
error message when the RFC length is too big. Finally, clarify the
comments to show that the 512 byte limit on extra data at the end of
the packet is arbitrary.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>

9 years agoMerge branch 'fixes' of master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Mon, 31 Jan 2011 22:30:31 +0000]
Merge branch 'fixes' of /home/rmk/linux-2.6-arm

* 'fixes' of master.kernel.org:/home/rmk/linux-2.6-arm:
  ARM: smp_on_up: allow non-ARM SMP processors
  ARM: io: ensure inb/outb() et.al. are properly ordered on ARMv6+
  ARM: initrd: disable initrd if passed address overlaps reserved region
  ARM: footbridge: fix debug macros
  ARM: mmci: round down the bytes transferred on error
  ARM: mmci: complete the transaction on error
  ARM: 6642/1: mmci: calculate remaining bytes at error correctly

9 years agoMerge branch 'omap-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 31 Jan 2011 22:27:55 +0000]
Merge branch 'omap-fixes-for-linus' of git://git./linux/kernel/git/tmlind/linux-omap-2.6

* 'omap-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap-2.6:
  arch/arm/mach-omap2/dma.c: Convert IS_ERR result to PTR_ERR
  arm: omap2: mux: fix compile warning
  omap1: Simplify use of omap_irq_flags
  omap2+: Fix unused variable warning for omap_irq_base

9 years agokernel.h: fix kernel-doc warning
Randy Dunlap [Mon, 31 Jan 2011 17:29:14 +0000]
kernel.h: fix kernel-doc warning

Fix kernel-doc warning in kernel.h from commit 7ef88ad56145
("BUILD_BUG_ON: make it handle more cases"):

  Warning(include/linux/kernel.h:605): No description found for parameter 'condition'
  Warning(include/linux/kernel.h:605): Excess function parameter 'cond' description in 'BUILD_BUG_ON'

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoMerge branch 'for-linus' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Mon, 31 Jan 2011 22:15:40 +0000]
Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs

* 'for-linus' of git://oss.sgi.com/xfs/xfs:
  xfs: xfs_bmap_add_extent_delay_real should init br_startblock
  xfs: fix dquot shaker deadlock
  xfs: handle CIl transaction commit failures correctly
  xfs: limit extsize to size of AGs and/or MAXEXTLEN
  xfs: prevent extsize alignment from exceeding maximum extent size
  xfs: limit extent length for allocation to AG size
  xfs: speculative delayed allocation uses rounddown_power_of_2 badly
  xfs: fix efi item leak on forced shutdown
  xfs: fix log ticket leak on forced shutdown.

9 years agovirtio: update MAINTAINERS
Michael S. Tsirkin [Mon, 31 Jan 2011 06:16:20 +0000]
virtio: update MAINTAINERS

Patches should keep coming through Rusty but it helps if I'm Cc'd as
well.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoMerge branch 'usb-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Mon, 31 Jan 2011 22:07:40 +0000]
Merge branch 'usb-linus' of git://git./linux/kernel/git/gregkh/usb-2.6

* 'usb-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6: (42 commits)
  usb: gadget: composite: avoid access beyond array max length
  USB: serial: handle Data Carrier Detect changes
  USB: gadget: Fix endpoint representation in ci13xxx_udc
  USB: gadget: Fix error path in ci13xxx_udc gadget probe function
  usb: pch_udc: Fix the worning log issue at gadget driver remove
  USB: serial: Updated support for ICOM devices
  USB: ehci-mxc: add work-around for efika mx/sb bug
  USB: unbreak ehci-mxc on otg port of i.MX27
  drivers: update to pl2303 usb-serial to support Motorola cables
  USB: adding USB support for Cinterion's HC2x, EU3 and PH8 products
  USB serial: add missing .usb_driver field in serial drivers
  USB: ehci-fsl: Fix 'have_sysif_regs' detection
  USB: g_printer: fix bug in module parameter definitions
  USB: g_printer: fix bug in unregistration
  USB: uss720: remove duplicate USB device
  MAINTAINERS: add ueagle-atm entry
  USB: EHCI: fix DMA deallocation bug
  USB: pch_udc: support new device ML7213 IOH
  usb: pch_udc: Fixed issue which does not work with g_serial
  usb: set ep_dev async suspend should be later than device_initialize
  ...

9 years agoMerge branch 'staging-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Mon, 31 Jan 2011 22:06:31 +0000]
Merge branch 'staging-linus' of git://git./linux/kernel/git/gregkh/staging-2.6

* 'staging-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6: (26 commits)
  staging: r8712u: Add new device IDs
  staging: brcm80211: fix suspend/resume issue in brcmsmac
  staging: brcm80211: remove assert to avoid panic since 2.6.37 kernel
  Staging: iio: Aditional fixpoint formatted output bugfix
  staging: usbip: vhci: use urb->dev->portnum to find port
  staging: usbip: vhci: handle EAGAIN from SO_RCVTIMEO
  staging: usbip: vhci: friendly log messages for connection errors
  staging: usbip: vhci: refuse to enqueue for dead connections
  staging: usbip: vhci: give back URBs from in-flight unlink requests
  staging: usbip: vhci: update reference count for usb_device
  staging: usbip: stub: update refcounts for devices and interfaces
  staging: tidspbridge: replace mbox callback with notifier_call
  staging: comedi: ni_labpc: Use shared IRQ for PCMCIA card
  Staging: speakup: &&/|| confusion in silent_store()
  iio: Fixpoint formatted output bugfix
  staging: rt2860: Fix incorrect netif_stop_queue usage warning
  staging: r8712u: Fix memory leak in firmware loading
  staging: tidspbridge: configure full L1 MMU range
  staging: rt2870sta: Add ID for Linksys WUSB100v2
  Staging: xgfib: put parenthesis in the right place
  ...

9 years agoMerge branch 'tty-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Mon, 31 Jan 2011 22:05:19 +0000]
Merge branch 'tty-linus' of git://git./linux/kernel/git/gregkh/tty-2.6

* 'tty-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty-2.6:
  tty/serial: fix apbuart build
  n_hdlc: fix read and write locking
  serial: unbreak billionton CF card
  tty: use for_each_console() and WARN() on sysfs failures
  vt: fix issue when fbcon wants to takeover a second time.

Fix up trivial conflict in drivers/tty/tty_io.c

9 years ago[CIFS] Update cifs minor version
Steve French [Mon, 31 Jan 2011 21:56:35 +0000]
[CIFS] Update cifs minor version

Signed-off-by: Steve French <sfrench@us.ibm.com>

9 years agoARM: smp_on_up: allow non-ARM SMP processors
Russell King [Sun, 30 Jan 2011 16:40:20 +0000]
ARM: smp_on_up: allow non-ARM SMP processors

Allow non-ARM SMP processors to use the SMP_ON_UP feature.  CPUs
supporting SMP must have the new CPU ID format, so check for this first.
Then check for ARM11MPCore, which fails the MPIDR check.  Lastly check
the MPIDR reports multiprocessing extensions and that the CPU is part of
a multiprocessing system.

Cc: <stable@kernel.org>
Reported-and-Tested-by: Stephen Boyd <sboyd@codeaurora.org>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

9 years agocifs: No need to check crypto blockcipher allocation
Shirish Pargaonkar [Sat, 29 Jan 2011 19:54:58 +0000]
cifs: No need to check crypto blockcipher allocation

Missed one change as per earlier suggestion.

Signed-off-by: Shirish Pargaonkar <shirishpargaonkar@gmail.com>
Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>

9 years ago[media] fix saa7111 non-detection
Russell King [Sun, 30 Jan 2011 13:03:30 +0000]
[media] fix saa7111 non-detection

One saa7111 device is reporting a different ID:

saa7115 0-0024: chip found @ 0x48 (ID 0f7111d0e111111) does not match a known saa711x chip.

As this is for sure a saa7111, change the detection code to also
cover this device.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years agocifs: clean up some compiler warnings
Jeff Layton [Mon, 31 Jan 2011 12:24:46 +0000]
cifs: clean up some compiler warnings

New compiler warnings that I noticed when building a patchset based
on recent Fedora kernel:

fs/cifs/cifssmb.c: In function 'CIFSSMBSetFileSize':
fs/cifs/cifssmb.c:4813:8: warning: variable 'data_offset' set but not used
[-Wunused-but-set-variable]

fs/cifs/file.c: In function 'cifs_open':
fs/cifs/file.c:349:24: warning: variable 'pCifsInode' set but not used
[-Wunused-but-set-variable]
fs/cifs/file.c: In function 'cifs_partialpagewrite':
fs/cifs/file.c:1149:23: warning: variable 'cifs_sb' set but not used
[-Wunused-but-set-variable]
fs/cifs/file.c: In function 'cifs_iovec_write':
fs/cifs/file.c:1740:9: warning: passing argument 6 of 'CIFSSMBWrite2' from
incompatible pointer type [enabled by default]
fs/cifs/cifsproto.h:337:12: note: expected 'unsigned int *' but argument is
of type 'size_t *'

fs/cifs/readdir.c: In function 'cifs_readdir':
fs/cifs/readdir.c:767:23: warning: variable 'cifs_sb' set but not used
[-Wunused-but-set-variable]

fs/cifs/cifs_dfs_ref.c: In function 'cifs_dfs_d_automount':
fs/cifs/cifs_dfs_ref.c:342:2: warning: 'rc' may be used uninitialized in
this function [-Wuninitialized]
fs/cifs/cifs_dfs_ref.c:278:6: note: 'rc' was declared here

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Pavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: Steve French <sfrench@us.ibm.com>

9 years agocifs: make CIFS depend on CRYPTO_MD4
Jeff Layton [Mon, 31 Jan 2011 13:41:36 +0000]
cifs: make CIFS depend on CRYPTO_MD4

Recently CIFS was changed to use the kernel crypto API for MD4 hashes,
but the Kconfig dependencies were not changed to reflect this.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reported-and-Tested-by: Suresh Jayaraman <sjayaraman@suse.de>
Reviewed-by: Shirish Pargaonkar <shirishpargaonkar@gmail.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>

9 years ago[media] rc/streamzap: fix reporting response times
Jarod Wilson [Thu, 27 Jan 2011 18:08:35 +0000]
[media] rc/streamzap: fix reporting response times

The streamzap driver has relatively low sampling resolution, and any
delays in reporting events seem to cause some minor problems for the
likes of irw when using the lirc bridge driver, resulting in a single
keypress registering as multiple independent ones, rather than as a
single press with repeats. If we call ir_raw_event_handle() more
frequently and reset the rawir kfifo at end-of-signal, the behavior
improves quite a bit.

Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years ago[media] mceusb: really fix remaining keybounce issues
Jarod Wilson [Wed, 26 Jan 2011 15:20:09 +0000]
[media] mceusb: really fix remaining keybounce issues

Make sure rawir struct is zeroed out before populating it for each
ir_raw_event_store_with_filter() call, and when we see a trailing 0x80
packet (end-of-data), issue an ir_raw_event_reset() call.

Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years ago[media] rc: use time unit conversion macros correctly
Jarod Wilson [Tue, 18 Jan 2011 20:31:24 +0000]
[media] rc: use time unit conversion macros correctly

Due to my own stupidity, some of the wrong time unit conversion macros
were being used inside some of the IR drivers I've been working on. Fix
that, and convert over some additional places to also use the macros.

Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years ago[media] rc/ir-lirc-codec: add back debug spew
Jarod Wilson [Tue, 18 Jan 2011 05:33:08 +0000]
[media] rc/ir-lirc-codec: add back debug spew

Some occasionally useful debug spew disappeared as part of a feature
update a while back, and I'm finding myself in need of it again to help
diagnose some issues.

Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years ago[media] ir-kbd-i2c: improve remote behavior with z8 behind usb
Jarod Wilson [Thu, 20 Jan 2011 21:31:18 +0000]
[media] ir-kbd-i2c: improve remote behavior with z8 behind usb

Add the same "are you ready?" i2c_master_send() poll command to
get_key_haup_xvr found in lirc_zilog, which is apparently seen in
the Windows driver for the PVR-150 w/a z8. This stabilizes what is
received from both the HD-PVR and HVR-1950, even with their polling
intervals at the default of 100, thus the removal of the custom
260ms polling_interval in pvrusb2-i2c-core.c.

Acked-by: Andy Walls <awalls@md.metrocast.net>
Acked-by: Mike Isely <isely@isely.net>
Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years ago[media] lirc_zilog: z8 on usb doesn't like back-to-back i2c_master_send
Jarod Wilson [Wed, 19 Jan 2011 21:49:19 +0000]
[media] lirc_zilog: z8 on usb doesn't like back-to-back i2c_master_send

Both the HD-PVR and HVR-1950, driven by the hdpvr and pvrusb2 drivers
respectively, have a zilog z8 chip exposed via i2c. These are both
usb-connected devices, and on both of them, back-to-back i2c_master_send
calls that work fine with a z8 on a pci card fail with a -EIO, as the
chip isn't yet ready from the prior command. To cope with that, add a
delay and retry loop where necessary.

Acked-by: Andy Walls <awalls@md.metrocast.net>
Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years ago[media] hdpvr: fix up i2c device registration
Jarod Wilson [Wed, 19 Jan 2011 21:10:14 +0000]
[media] hdpvr: fix up i2c device registration

We have to actually call i2c_new_device() once for each of the rx and tx
addresses. Also improve error-handling and device remove i2c cleanup.

Reviewed-by: Andy Walls <awalls@md.metrocast.net>
Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years ago[media] rc/mce: add mappings for missing keys
Jarod Wilson [Tue, 18 Jan 2011 05:27:45 +0000]
[media] rc/mce: add mappings for missing keys

Per http://mediacenterguides.com/book/export/html/31 and investigation
by Erin, we were missing these last three mappings to complete the mce
key table. Lets remedy that.

Reported-by: Erin Simonds <fisslefink@gmail.com>
Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years ago[media] gspca - zc3xx: Discard the partial frames
Jean-François Moine [Mon, 24 Jan 2011 19:31:58 +0000]
[media] gspca - zc3xx: Discard the partial frames

In some cases, some frames may not end with the JPEG end of frame.
Being not complete, they are now discarded.

Signed-off-by: Jean-François Moine <moinejf@free.fr>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years ago[media] gspca - zc3xx: Fix bad images with the sensor hv7131r
Jean-François Moine [Mon, 24 Jan 2011 19:22:11 +0000]
[media] gspca - zc3xx: Fix bad images with the sensor hv7131r

The problem was introduced by the commit 2af0b4c60cc0daf0.
Some registers were no more initialized.

Tested-by: <Giovanni Scafora giovanni@archlinux.org>
Tested-by: <Sergey Manucharian sm@ingeniware.com>
Signed-off-by: Jean-François Moine <moinejf@free.fr>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years ago[media] gspca - zc3xx: Bad delay when given by a table
Jean-François Moine [Mon, 24 Jan 2011 18:15:34 +0000]
[media] gspca - zc3xx: Bad delay when given by a table

Signed-off-by: Jean-François Moine <moinejf@free.fr>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

9 years agowatchdog: Don't change watchdog state on read of sysctl
Marcin Slusarz [Fri, 28 Jan 2011 16:00:33 +0000]
watchdog: Don't change watchdog state on read of sysctl

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
[ add {}'s to fix a warning ]
Signed-off-by: Don Zickus <dzickus@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: <stable@kernel.org>
LKML-Reference: <1296230433-6261-3-git-send-email-dzickus@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

9 years agowatchdog: Fix sysctl consistency
Marcin Slusarz [Fri, 28 Jan 2011 16:00:32 +0000]
watchdog: Fix sysctl consistency

If it was not possible to enable watchdog for any cpu, switch
watchdog_enabled back to 0, because it's visible via
kernel.watchdog sysctl.

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Signed-off-by: Don Zickus <dzickus@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: <stable@kernel.org>
LKML-Reference: <1296230433-6261-2-git-send-email-dzickus@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

9 years agowatchdog: Fix broken nowatchdog logic
Marcin Slusarz [Fri, 28 Jan 2011 16:00:31 +0000]
watchdog: Fix broken nowatchdog logic

Passing nowatchdog to kernel disables 2 things: creation of
watchdog threads AND initialization of percpu watchdog_hrtimer.
As hrtimers are initialized only at boot it's not possible to
enable watchdog later - for me all watchdog threads started to
eat 100% of CPU time, but they could just crash.

Additionally, even if these threads would start properly,
watchdog_disable_all_cpus was guarded by no_watchdog check, so
you couldn't disable watchdog.

To fix this, remove no_watchdog variable and use already
existing watchdog_enabled variable.

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
[ removed another no_watchdog instance ]
Signed-off-by: Don Zickus <dzickus@redhat.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: <stable@kernel.org>
LKML-Reference: <1296230433-6261-1-git-send-email-dzickus@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

9 years agoARM: io: ensure inb/outb() et.al. are properly ordered on ARMv6+
Russell King [Sun, 30 Jan 2011 11:29:40 +0000]
ARM: io: ensure inb/outb() et.al. are properly ordered on ARMv6+

Ensure that the ISA/PCI IO space accessors are properly ordered on
ARMv6+ architectures.  These should always be ordered with respect to
all other accesses.

This also fixes __iormb() and __iowmb() not being visible to ioread/
iowrite if a platform defines its own MMIO accessors.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

9 years agoARM: initrd: disable initrd if passed address overlaps reserved region
Russell King [Sun, 30 Jan 2011 11:21:05 +0000]
ARM: initrd: disable initrd if passed address overlaps reserved region

Disable the initrd if the passed address already overlaps the reserved
region.  This avoids oopses on Netwinders when NeTTrom tells the kernel
that an initrd is located at mem+4MB, but this overlaps the BSS,
resulting in the kernels in-use BSS being freed.

This should be applied to v2.6.37-stable.

Cc: <stable@kernel.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

9 years agoARM: footbridge: fix debug macros
Russell King [Fri, 28 Jan 2011 20:57:57 +0000]
ARM: footbridge: fix debug macros

0ea1293 (arm: return both physical and virtual addresses from addruart)
changed the way the 'addruart' worked, making it return both the virt
and phys addresses.  Unfortunately, for footbridge, these were reversed.
Fix that.  Tested on Netwinder.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

9 years agoARM: mmci: round down the bytes transferred on error
Russell King [Sun, 30 Jan 2011 21:06:53 +0000]
ARM: mmci: round down the bytes transferred on error

We should not report incomplete blocks on error.  Return the number of
bytes successfully transferred, rounded down to the nearest block.

Acked-by: Linus Walleij <linus.walleij@stericsson.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

9 years agoARM: mmci: complete the transaction on error
Russell King [Sun, 30 Jan 2011 21:03:50 +0000]
ARM: mmci: complete the transaction on error

When we encounter an error, make sure we complete the transaction
otherwise we'll leave the request dangling.

Acked-by: Linus Walleij <linus.walleij@stericsson.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

9 years ago[S390] reset default for CONFIG_CHSC_SCH
Sebastian Ott [Mon, 31 Jan 2011 10:30:09 +0000]
[S390] reset default for CONFIG_CHSC_SCH

6f9a3c33 "[S390] cleanup s390 Kconfig" accidentally changed
the default for CONFIG_CHSC_SCH. Reset it to m.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

9 years ago[S390] qdio: prevent compile warning under CONFIG_32BIT
Jan Glauber [Mon, 31 Jan 2011 10:30:08 +0000]
[S390] qdio: prevent compile warning under CONFIG_32BIT

Prevent the following compiler warning if compiling a 31 bit kernel:

drivers/s390/cio/qdio_main.c: In function ‘get_outbound_buffer_frontier’:
drivers/s390/cio/qdio_main.c:646:16: warning: ‘state’ may be used uninitialized in this function
CC lib/radix-tree.o
CC drivers/s390/scsi/zfcp_cfdc.o
drivers/s390/cio/qdio_main.c: In function ‘qdio_inbound_q_moved’:
drivers/s390/cio/qdio_main.c:479:16: warning: ‘state’ may be used uninitialized in this function
drivers/s390/cio/qdio_main.c:479:16: note: ‘state’ was declared here

Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

9 years ago[S390] use asm-generic/cacheflush.h
Akinobu Mita [Mon, 31 Jan 2011 10:30:07 +0000]
[S390] use asm-generic/cacheflush.h

The implementation of the cache flushing interfaces on the s390
is identical with the default implementation in asm-generic.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

9 years ago[S390] tlb: fix build error caused by THP
Heiko Carstens [Mon, 31 Jan 2011 10:30:06 +0000]
[S390] tlb: fix build error caused by THP

Fix this build error with !CONFIG_SWAP caused by tranparent huge pages support:

In file included from mm/pgtable-generic.c:9:0:
/linux-2.6/arch/s390/include/asm/tlb.h: In function 'tlb_remove_page':
/linux-2.6/arch/s390/include/asm/tlb.h:92:2: error: implicit declaration of function 'page_cache_release'

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

9 years ago[S390] missing sacf in uaccess
Martin Schwidefsky [Mon, 31 Jan 2011 10:30:05 +0000]
[S390] missing sacf in uaccess

The uaccess functions copy_in_user_std and clear_user_std fail to
switch back from secondary space mode to primary space mode with sacf
in case of an unresolvable page fault. We need to make sure that the
switch back to primary mode is done in all cases, otherwise the code
following the uaccess inline assembly will crash.

Reported-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

9 years ago[S390] pgtable_list corruption
Martin Schwidefsky [Mon, 31 Jan 2011 10:30:04 +0000]
[S390] pgtable_list corruption

After page_table_free_rcu removed a page from the pgtable_list
page_table_free better not add it again. Otherwise a page_table_alloc
can reuse a page table fragment that is still in the rcu process.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

9 years ago[S390] dasd: prevent panic with unresumed devices
Stefan Haberland [Mon, 31 Jan 2011 10:30:03 +0000]
[S390] dasd: prevent panic with unresumed devices

If a device is not resumed correctly the system crashes when this
device is set offline. This may happen if it gets disconnected
during suspend.
Check if the device is already removed from alias handling and skip
these steps to prevent the kernel panic.

Signed-off-by: Stefan Haberland <stefan.haberland@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

9 years agocifs: force a reconnect if there are too many MIDs in flight
Jeff Layton [Sat, 29 Jan 2011 12:02:28 +0000]
cifs: force a reconnect if there are too many MIDs in flight

Currently, we allow the pending_mid_q to grow without bound with
SIGKILL'ed processes. This could eventually be a DoS'able problem. An
unprivileged user could a process that does a long-running call and then
SIGKILL it.

If he can also intercept the NT_CANCEL calls or the replies from the
server, then the pending_mid_q could grow very large, possibly even to
2^16 entries which might leave GetNextMid in an infinite loop. Fix this
by imposing a hard limit of 32k calls per server. If we cross that
limit, set the tcpStatus to CifsNeedReconnect to force cifsd to
eventually reconnect the socket and clean out the pending_mid_q.

While we're at it, clean up the function a bit and eliminate an
unnecessary NULL pointer check.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Shirish Pargaonkar <shirishpargaonkar@gmail.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>

9 years agocifs: don't pop a printk when sending on a socket is interrupted
Jeff Layton [Fri, 28 Jan 2011 20:05:43 +0000]
cifs: don't pop a printk when sending on a socket is interrupted

If we kill the process while it's sending on a socket then the
kernel_sendmsg will return -EINTR. This is normal. No need to spam the
ring buffer with this info.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>

9 years agocifs: simplify SMB header check routine
Jeff Layton [Fri, 28 Jan 2011 20:05:42 +0000]
cifs: simplify SMB header check routine

...just cleanup. There should be no behavior change.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Pavel Shilovsky <piastryyy@gmail.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>

9 years agocifs: send an NT_CANCEL request when a process is signalled
Jeff Layton [Fri, 28 Jan 2011 12:08:28 +0000]
cifs: send an NT_CANCEL request when a process is signalled

Use the new send_nt_cancel function to send an NT_CANCEL when the
process is delivered a fatal signal. This is a "best effort" enterprise
however, so don't bother to check the return code. There's nothing we
can reasonably do if it fails anyway.

Reviewed-by: Pavel Shilovsky <piastryyy@gmail.com>
Reviewed-by: Suresh Jayaraman <sjayaraman@suse.de>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>

9 years agocifs: handle cancelled requests better
Jeff Layton [Fri, 28 Jan 2011 12:08:28 +0000]
cifs: handle cancelled requests better

Currently, when a request is cancelled via signal, we delete the mid
immediately. If the request was already transmitted however, the client
is still likely to receive a response. When it does, it won't recognize
it however and will pop a printk.

It's also a little dangerous to just delete the mid entry like this. We
may end up reusing that mid. If we do then we could potentially get the
response from the first request confused with the later one.

Prevent the reuse of mids by marking them as cancelled and keeping them
on the pending_mid_q list. If the reply comes in, we'll delete it from
the list then. If it never comes, then we'll delete it at reconnect
or when cifsd comes down.

Reviewed-by: Pavel Shilovsky <piastryyy@gmail.com>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>

9 years agoMerge branch 'master' of /pub/scm/linux/kernel/git/torvalds/linux-2.6
Steve French [Mon, 31 Jan 2011 04:17:03 +0000]
Merge branch 'master' of /linux/kernel/git/torvalds/linux-2.6

9 years agocifs: fix two compiler warning about uninitialized vars
Jeff Layton [Sat, 29 Jan 2011 12:03:02 +0000]
cifs: fix two compiler warning about uninitialized vars

fs/cifs/link.c: In function ‘symlink_hash’:
fs/cifs/link.c:58:3: warning: ‘rc’ may be used uninitialized in this
function [-Wuninitialized]

fs/cifs/smbencrypt.c: In function ‘mdfour’:
fs/cifs/smbencrypt.c:61:3: warning: ‘rc’ may be used uninitialized in this
function [-Wuninitialized]

Reviewed-by: Shirish Pargaonkar <shirishpargaonkar@gmail.com>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>

9 years agoMerge branch 'for-linus' of git://git.infradead.org/ubi-2.6
Linus Torvalds [Mon, 31 Jan 2011 03:04:51 +0000]
Merge branch 'for-linus' of git://git.infradead.org/ubi-2.6

* 'for-linus' of git://git.infradead.org/ubi-2.6:
  Revert "UBI: use mtd->writebufsize to set minimal I/O unit size"

9 years agoInput: rc-keymap - return KEY_RESERVED for unknown mappings
Dmitry Torokhov [Sat, 29 Jan 2011 07:33:29 +0000]
Input: rc-keymap - return KEY_RESERVED for unknown mappings

Do not respond with -EINVAL to EVIOCGKEYCODE for not-yet-mapped
scancodes, but rather return KEY_RESERVED.

This fixes breakage with Ubuntu's input-kbd utility that stopped
returning full keymaps for remote controls.

Tested-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Tested-by: Mark Lord <kernel@teksavvy.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Mon, 31 Jan 2011 03:02:34 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: synaptics - retry failed resets when reconnecting
  Input: synaptics - fix reconnect logic on MT devices
  Input: tegra-kbc - fix keymap entry for LeftMeta key
  Input: tegra-kbc - fix build error

9 years agoFix prlimit64 for suid/sgid processes
Kacper Kornet [Fri, 28 Jan 2011 23:21:04 +0000]
Fix prlimit64 for suid/sgid processes

Since check_prlimit_permission always fails in the case of SUID/GUID
processes, such processes are not able to read or set their own limits.
This commit changes this by assuming that process can always read/change
its own limits.

Signed-off-by: Kacper Kornet <kornet@camk.edu.pl>
Acked-by: Jiri Slaby <jslaby@suse.cz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agointel_scu_ipc: remove duplicated #include
Huang Weiyi [Fri, 28 Jan 2011 14:05:25 +0000]
intel_scu_ipc: remove duplicated #include

Remove duplicated #include('s) in
  drivers/platform/x86/intel_scu_ipc.c

Signed-off-by: Huang Weiyi <weiyi.huang@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoNTFS: Fix invalid pointer dereference in ntfs_mft_record_alloc().
Anton Altaparmakov [Fri, 28 Jan 2011 20:45:28 +0000]
NTFS: Fix invalid pointer dereference in ntfs_mft_record_alloc().

In ntfs_mft_record_alloc() when mapping the new extent mft record with
map_extent_mft_record() we overwrite @m with the return value and on
error, we then try to use the old @m but that is no longer there as @m
now contains an error code instead so we crash when dereferencing the
error code as if it were a pointer.

The simple fix is to use a temporary variable to store the return value
thus preserving the original @m for later use.  This is a backport from
the commercial Tuxera-NTFS driver and is well tested...

Thanks go to Julia Lawall for pointing this out (whilst I had fixed it
in the commercial driver I had failed to fix it in the Linux kernel).

Signed-off-by: Anton Altaparmakov <anton@tuxera.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Mon, 31 Jan 2011 02:56:27 +0000]
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6:
  cifs: More crypto cleanup (try #2)
  CIFS: Add strictcache mount option
  CIFS: Implement cifs_strict_writev (try #4)
  [CIFS] Replace cifs md5 hashing functions with kernel crypto APIs

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas...
Linus Torvalds [Mon, 31 Jan 2011 02:55:38 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/cmarinas/linux-2.6-cm

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas/linux-2.6-cm:
  kmemleak: Allow kmemleak metadata allocations to fail
  kmemleak: remove memset by using kzalloc

9 years agoMerge branch 'zerolen' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/misc-2.6
Linus Torvalds [Mon, 31 Jan 2011 02:54:54 +0000]
Merge branch 'zerolen' of git://git./linux/kernel/git/jgarzik/misc-2.6

* 'zerolen' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/misc-2.6:
  Delete zero-length drivers/staging/vme/bridges/Module.symvers

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Mon, 31 Jan 2011 02:53:12 +0000]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: HDA: Fix automute on Thinkpad L412/L512
  ALSA: HDA: Fix dmesg output of HDMI supported bits
  ALSA: fix invalid hardware.h include in ac97c for AVR32 architecture
  ASoC: correct link specifications for corgi, poodle and spitz
  ASoC: Samsung: Fix outdated cpu_dai_name for s3c24xx i2s
  ASoC: Fix codec device id format used by some dai_links
  ALSA: azt3328 -  fix broken AZF_FMT_XLATE macro
  ALSA: Xonar, CS43xx: Don't overrun static array
  ASoC: Handle low measured DC offsets for wm_hubs devices
  ASoC: da8xx/omap-l1xx: match codec_name with i2c ids
  ASoC: WM8994: fix wrong value in tristate function
  ASoC: WM8995: Fix incorrect use of snd_soc_update_bits()

9 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Mon, 31 Jan 2011 02:51:28 +0000]
Merge branch 'upstream-linus' of git://git./linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  libata: set queue DMA alignment to sector size for ATAPI too
  libata: DVR-212D can't do SETXFER DVD-RW DVR-212D
  ahci: add HFLAG_YES_FBS and apply it to 88SE9128
  pata_hpt37x: inherit prereset() method for HPT374
  ahci: AHCI mode SATA patch for Intel DH89xxCC DeviceIDs
  pata_hpt37x: fold 'if' statement into 'switch'
  pata_hpt{37x|3x2n}: use pr_*(DRV_NAME ...) instead of printk(KERN_* ...)
  pata_hpt{366|37x}: use pr_warning(...) instead of printk(KERN_WARNING ...)
  pata_mpc52xx: inherit from ata_bmdma_port_ops

9 years agoMerge branch 'for-linus' of git://android.git.kernel.org/kernel/tegra
Linus Torvalds [Mon, 31 Jan 2011 02:49:26 +0000]
Merge branch 'for-linus' of git://android.git./kernel/tegra

* 'for-linus' of git://android.git.kernel.org/kernel/tegra:
  ARM: tegra: clock: Add forward reference to struct clk
  ARM: tegra: irq: Rename gic pointers to avoid conflicts
  arm/tegra: Fix tegra irq_data conversion

9 years agoRevert "UBI: use mtd->writebufsize to set minimal I/O unit size"
Artem Bityutskiy [Sat, 29 Jan 2011 16:27:13 +0000]
Revert "UBI: use mtd->writebufsize to set minimal I/O unit size"

This reverts commit a121f643993474548fe98144514c50dd4f3dbe76.

Unfortunately, this commit breaks UBIFS backward compatibility and
makes new UBIFS refuse older UBIFS-formatted media:

UBIFS error: validate_sb: min. I/O unit mismatch: 8 in superblock, 64 real

Thus, we have to revert this patch and work on a better solution.

Reported-by: Holger Brunck <holger.brunck@keymile.com>
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

9 years agoInput: synaptics - retry failed resets when reconnecting
Alexandre Peixoto Ferreira [Sat, 29 Jan 2011 06:05:14 +0000]
Input: synaptics - retry failed resets when reconnecting

On some machines, like Dell Studio XPS 16 (1640), touchpad fails to
respond to the standard query after first reset but may start
responding later, so let's repeat reset sequence several (3) times.

Signed-off-by: Alexandre Peixoto Ferreira <alexandref75@gmail.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>

9 years agoInput: synaptics - fix reconnect logic on MT devices
Alexandre Peixoto Ferreira [Sat, 29 Jan 2011 06:05:14 +0000]
Input: synaptics - fix reconnect logic on MT devices

synaptics_set_advanced_gesture_mode() affect capabilities bits we should
perform comparison after calling this function, otherwise they will never
match and we will be forced to perform full reconnect.

Signed-off-by: Alexandre Peixoto Ferreira <alexandref75@gmail.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>

9 years agoInput: tegra-kbc - fix keymap entry for LeftMeta key
Rakesh Iyer [Sat, 29 Jan 2011 06:05:14 +0000]
Input: tegra-kbc - fix keymap entry for LeftMeta key

Correct key mapping for Left Meta key.

Signed-off-by: Rakesh Iyer <riyer@nvidia.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>

9 years agoInput: tegra-kbc - fix build error
Rakesh Iyer [Sat, 29 Jan 2011 06:05:14 +0000]
Input: tegra-kbc - fix build error

Fix build error introduced by variable name change.

Signed-off-by: Rakesh Iyer <riyer@nvidia.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>

9 years agoNFS: NFSv4 readdir loses entries
Chuck Lever [Fri, 28 Jan 2011 17:41:05 +0000]
NFS: NFSv4 readdir loses entries

On recent 2.6.38-rc kernels, connectathon basic test 6 fails on
NFSv4 mounts of OpenSolaris with something like:

> ./test6: readdir
>  ./test6: (/mnt/klimt/matisse.test) didn't read expected 'file.12' dir entry, pass 0
>  ./test6: (/mnt/klimt/matisse.test) didn't read expected 'file.82' dir entry, pass 0
>  ./test6: (/mnt/klimt/matisse.test) didn't read expected 'file.164' dir entry, pass 0
>  ./test6: (/mnt/klimt/matisse.test) Test failed with 3 errors
> basic tests failed
> Tests failed, leaving /mnt/klimt mounted
> [cel@matisse cthon04]$

I narrowed the problem down to nfs4_decode_dirent() reporting that the
decode buffer had overflowed while decoding the entries for those
missing files.

verify_attr_len() assumes both it's pointer arguments reside on the
same page.  When these arguments point to locations on two different
pages, verify_attr_len() can report false errors.  This can happen now
that a large NFSv4 readdir result can span pages.

We have reasonably good checking in nfs4_decode_dirent() anyway, so
it should be safe to simply remove the extra checking.

At a guess, this was introduced by commit 6650239a, "NFS: Don't use
vm_map_ram() in readdir".

Cc: stable@kernel.org [2.6.37]
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

9 years agoNFS: Micro-optimize nfs4_decode_dirent()
Chuck Lever [Fri, 28 Jan 2011 17:40:55 +0000]
NFS: Micro-optimize nfs4_decode_dirent()

Make the decoding of NFSv4 directory entries slightly more efficient
by:

  1.  Avoiding unnecessary byte swapping when checking XDR booleans,
      and

  2.  Not bumping "p" when its value will be immediately replaced by
      xdr_inline_decode()

This commit makes nfs4_decode_dirent() consistent with similar logic
in the other two decode_dirent() functions.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

9 years agoNFS: Fix an NFS client lockdep issue
Trond Myklebust [Thu, 27 Jan 2011 19:55:39 +0000]
NFS: Fix an NFS client lockdep issue

There is no reason to be freeing the delegation cred in the rcu callback,
and doing so is resulting in a lockdep complaint that rpc_credcache_lock
is being called from both softirq and non-softirq contexts.

Reported-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: stable@kernel.org

9 years agoxfs: xfs_bmap_add_extent_delay_real should init br_startblock
bpm@sgi.com [Wed, 19 Jan 2011 17:41:58 +0000]
xfs: xfs_bmap_add_extent_delay_real should init br_startblock

When filling in the middle of a previous delayed allocation in
xfs_bmap_add_extent_delay_real, set br_startblock of the new delay
extent to the right to nullstartblock instead of 0 before inserting
the extent into the ifork (xfs_iext_insert), rather than setting
br_startblock afterward.

Adding the extent into the ifork with br_startblock=0 can lead to
the extent being copied into the btree by xfs_bmap_extent_to_btree
if we happen to convert from extents format to btree format before
updating br_startblock with the correct value.  The unexpected
addition of this delay extent to the btree can cause subsequent
XFS_WANT_CORRUPTED_GOTO filesystem shutdown in several
xfs_bmap_add_extent_delay_real cases where we are converting a delay
extent to real and unexpectedly find an extent already inserted.
For example:

911         case BMAP_LEFT_FILLING:
912                 /*
913                  * Filling in the first part of a previous delayed allocation.
914                  * The left neighbor is not contiguous.
915                  */
916                 trace_xfs_bmap_pre_update(ip, idx, state, _THIS_IP_);
917                 xfs_bmbt_set_startoff(ep, new_endoff);
918                 temp = PREV.br_blockcount - new->br_blockcount;
919                 xfs_bmbt_set_blockcount(ep, temp);
920                 xfs_iext_insert(ip, idx, 1, new, state);
921                 ip->i_df.if_lastex = idx;
922                 ip->i_d.di_nextents++;
923                 if (cur == NULL)
924                         rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
925                 else {
926                         rval = XFS_ILOG_CORE;
927                         if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
928                                         new->br_startblock, new->br_blockcount,
929                                         &i)))
930                                 goto done;
931                         XFS_WANT_CORRUPTED_GOTO(i == 0, done);

With the bogus extent in the btree we shutdown the filesystem at
931.  The conversion from extents to btree format happens when the
number of extents in the inode increases above ip->i_df.if_ext_max.
xfs_bmap_extent_to_btree copies extents from the ifork into the
btree, ignoring all delalloc extents which are denoted by
br_startblock having some value of nullstartblock.

SGI-PV: 1013221

Signed-off-by: Ben Myers <bpm@sgi.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Alex Elder <aelder@sgi.com>

9 years agoxfs: fix dquot shaker deadlock
Dave Chinner [Fri, 28 Jan 2011 00:20:46 +0000]
xfs: fix dquot shaker deadlock

Commit 368e136 ("xfs: remove duplicate code from dquot reclaim") fails
to unlock the dquot freelist when the number of loop restarts is
exceeded in xfs_qm_dqreclaim_one(). This causes hangs in memory
reclaim.

Rework the loop control logic into an unwind stack that all the
different cases jump into. This means there is only one set of code
that processes the loop exit criteria, and simplifies the unlocking
of all the items from different points in the loop. It also fixes a
double increment of the restart counter from the qi_dqlist_lock
case.

Reported-by: Malcolm Scott <lkml@malc.org.uk>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Alex Elder <aelder@sgi.com>

9 years agoxfs: handle CIl transaction commit failures correctly
Dave Chinner [Thu, 27 Jan 2011 02:23:28 +0000]
xfs: handle CIl transaction commit failures correctly

Failure to commit a transaction into the CIL is not handled
correctly. This currently can only happen when racing with a
shutdown and requires an explicit shutdown check, so it rare and can
be avoided. Remove the shutdown check and make the CIL commit a void
function to indicate it will always succeed, thereby removing the
incorrectly handled failure case.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Alex Elder <aelder@sgi.com>