openwrt/staging/blogic.git
12 years agomm: hugetlb: fix pgoff computation when unmapping page from vma
Hillf Danton [Tue, 10 Jan 2012 23:08:19 +0000 (15:08 -0800)]
mm: hugetlb: fix pgoff computation when unmapping page from vma

The computation for pgoff is incorrect, at least with

(vma->vm_pgoff >> PAGE_SHIFT)

involved.  It is fixed with the available method if HPAGE_SIZE is
concerned in page cache lookup.

[akpm@linux-foundation.org: use vma_hugecache_offset() directly, per Michal]
Signed-off-by: Hillf Danton <dhillf@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm/vmscan.c: consider swap space when deciding whether to continue reclaim
Minchan Kim [Tue, 10 Jan 2012 23:08:18 +0000 (15:08 -0800)]
mm/vmscan.c: consider swap space when deciding whether to continue reclaim

It's pointless to continue reclaiming when we have no swap space and lots
of anon pages in the inactive list.

Without this patch, it is possible when swap is disabled to continue
trying to reclaim when there are only anonymous pages in the system even
though that will not make any progress.

Signed-off-by: Minchan Kim <minchan@kernel.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: bootmem: try harder to free pages in bulk
Johannes Weiner [Tue, 10 Jan 2012 23:08:15 +0000 (15:08 -0800)]
mm: bootmem: try harder to free pages in bulk

The loop that frees pages to the page allocator while bootstrapping tries
to free higher-order blocks only when the starting address is aligned to
that block size.  Otherwise it will free all pages on that node
one-by-one.

Change it to free individual pages up to the first aligned block and then
try higher-order frees from there.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: bootmem: drop superfluous range check when freeing pages in bulk
Johannes Weiner [Tue, 10 Jan 2012 23:08:13 +0000 (15:08 -0800)]
mm: bootmem: drop superfluous range check when freeing pages in bulk

The area node_bootmem_map represents is aligned to BITS_PER_LONG, and all
bits in any aligned word of that map valid.  When the represented area
extends beyond the end of the node, the non-existant pages will be marked
as reserved.

As a result, when freeing a page block, doing an explicit range check for
whether that block is within the node's range is redundant as the bitmap
is consulted anyway to see whether all pages in the block are unreserved.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: page_alloc: generalize order handling in __free_pages_bootmem()
Johannes Weiner [Tue, 10 Jan 2012 23:08:10 +0000 (15:08 -0800)]
mm: page_alloc: generalize order handling in __free_pages_bootmem()

__free_pages_bootmem() used to special-case higher-order frees to save
individual page checking with free_pages_bulk().

Nowadays, both zero order and non-zero order frees use free_pages(), which
checks each individual page anyway, and so there is little point in making
the distinction anymore.  The higher-order loop will work just fine for
zero order pages.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agotracepoint: add tracepoints for debugging oom_score_adj
KAMEZAWA Hiroyuki [Tue, 10 Jan 2012 23:08:09 +0000 (15:08 -0800)]
tracepoint: add tracepoints for debugging oom_score_adj

oom_score_adj is used for guarding processes from OOM-Killer.  One of
problem is that it's inherited at fork().  When a daemon set oom_score_adj
and make children, it's hard to know where the value is set.

This patch adds some tracepoints useful for debugging. This patch adds
3 trace points.
  - creating new task
  - renaming a task (exec)
  - set oom_score_adj

To debug, users need to enable some trace pointer. Maybe filtering is useful as

# EVENT=/sys/kernel/debug/tracing/events/task/
# echo "oom_score_adj != 0" > $EVENT/task_newtask/filter
# echo "oom_score_adj != 0" > $EVENT/task_rename/filter
# echo 1 > $EVENT/enable
# EVENT=/sys/kernel/debug/tracing/events/oom/
# echo 1 > $EVENT/enable

output will be like this.
# grep oom /sys/kernel/debug/tracing/trace
bash-7699  [007] d..3  5140.744510: oom_score_adj_update: pid=7699 comm=bash oom_score_adj=-1000
bash-7699  [007] ...1  5151.818022: task_newtask: pid=7729 comm=bash clone_flags=1200011 oom_score_adj=-1000
ls-7729  [003] ...2  5151.818504: task_rename: pid=7729 oldcomm=bash newcomm=ls oom_score_adj=-1000
bash-7699  [002] ...1  5175.701468: task_newtask: pid=7730 comm=bash clone_flags=1200011 oom_score_adj=-1000
grep-7730  [007] ...2  5175.701993: task_rename: pid=7730 oldcomm=bash newcomm=grep oom_score_adj=-1000

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: simplify find_vma_prev()
KOSAKI Motohiro [Tue, 10 Jan 2012 23:08:07 +0000 (15:08 -0800)]
mm: simplify find_vma_prev()

commit 297c5eee37 ("mm: make the vma list be doubly linked") added the
vm_prev member to vm_area_struct.  We can simplify find_vma_prev() by
using it.  Also, this change helps to improve page fault performance
because it has stronger locality of reference.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomremap: enforce rmap src/dst vma ordering in case of vma_merge() succeeding in copy_vma()
Andrea Arcangeli [Tue, 10 Jan 2012 23:08:05 +0000 (15:08 -0800)]
mremap: enforce rmap src/dst vma ordering in case of vma_merge() succeeding in copy_vma()

migrate was doing an rmap_walk with speculative lock-less access on
pagetables.  That could lead it to not serializing properly against mremap
PT locks.  But a second problem remains in the order of vmas in the
same_anon_vma list used by the rmap_walk.

If vma_merge succeeds in copy_vma, the src vma could be placed after the
dst vma in the same_anon_vma list.  That could still lead to migrate
missing some pte.

This patch adds an anon_vma_moveto_tail() function to force the dst vma at
the end of the list before mremap starts to solve the problem.

If the mremap is very large and there are a lots of parents or childs
sharing the anon_vma root lock, this should still scale better than taking
the anon_vma root lock around every pte copy practically for the whole
duration of mremap.

Update: Hugh noticed special care is needed in the error path where
move_page_tables goes in the reverse direction, a second
anon_vma_moveto_tail() call is needed in the error path.

This program exercises the anon_vma_moveto_tail:

===

int main()
{
static struct timeval oldstamp, newstamp;
long diffsec;
char *p, *p2, *p3, *p4;
if (posix_memalign((void **)&p, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p2, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p3, 2*1024*1024, SIZE))
perror("memalign"), exit(1);

memset(p, 0xff, SIZE);
printf("%p\n", p);
memset(p2, 0xff, SIZE);
memset(p3, 0x77, 4096);
if (memcmp(p, p2, SIZE))
printf("error\n");
p4 = mremap(p+SIZE/2, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p3);
if (p4 != p3)
perror("mremap"), exit(1);
p4 = mremap(p4, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p+SIZE/2);
if (p4 != p+SIZE/2)
perror("mremap"), exit(1);
if (memcmp(p, p2, SIZE))
printf("error\n");
printf("ok\n");

return 0;
}
===

$ perf probe -a anon_vma_moveto_tail
Add new event:
  probe:anon_vma_moveto_tail (on anon_vma_moveto_tail)

You can now use it on all perf tools, such as:

        perf record -e probe:anon_vma_moveto_tail -aR sleep 1

$ perf record -e probe:anon_vma_moveto_tail -aR ./anon_vma_moveto_tail
0x7f2ca2800000
ok
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.043 MB perf.data (~1860 samples) ]
$ perf report --stdio
   100.00%  anon_vma_moveto  [kernel.kallsyms]  [k] anon_vma_moveto_tail

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Nai Xia <nai.xia@gmail.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Cc: Pawel Sikora <pluto@agmk.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: fix off-by-two in __zone_watermark_ok()
Michal Hocko [Tue, 10 Jan 2012 23:08:02 +0000 (15:08 -0800)]
mm: fix off-by-two in __zone_watermark_ok()

Commit 88f5acf88ae6 ("mm: page allocator: adjust the per-cpu counter
threshold when memory is low") changed the form how free_pages is
calculated but it forgot that we used to do free_pages - ((1 << order) -
1) so we ended up with off-by-two when calculating free_pages.

Reported-by: Wang Sheng-Hui <shhuiw@gmail.com>
Signed-off-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agobootmem: micro optimize freeing pages in bulk
Uwe Kleine-König [Tue, 10 Jan 2012 23:08:00 +0000 (15:08 -0800)]
bootmem: micro optimize freeing pages in bulk

The first entry of bdata->node_bootmem_map holds the data for
bdata->node_min_pfn up to bdata->node_min_pfn + BITS_PER_LONG - 1.  So the
test for freeing all pages of a single map entry can be slightly relaxed.

Moreover use DIV_ROUND_UP in another place instead of open coding it.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Cc: Johannes Weiner <hannes@saeurebad.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: compaction: push isolate search base of compact control one pfn ahead
Hillf Danton [Tue, 10 Jan 2012 23:07:59 +0000 (15:07 -0800)]
mm: compaction: push isolate search base of compact control one pfn ahead

After isolated the current pfn will no longer be scanned and isolated if
the next round is necessary, so push the isolate_migratepages search base
of the given compact_control one step ahead.

Signed-off-by: Hillf Danton <dhillf@gmail.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agobtrfs: pass __GFP_WRITE for buffered write page allocations
Johannes Weiner [Tue, 10 Jan 2012 23:07:55 +0000 (15:07 -0800)]
btrfs: pass __GFP_WRITE for buffered write page allocations

Tell the page allocator that pages allocated for a buffered write are
expected to become dirty soon.

Signed-off-by: Johannes Weiner <jweiner@redhat.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Chris Mason <chris.mason@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: filemap: pass __GFP_WRITE from grab_cache_page_write_begin()
Johannes Weiner [Tue, 10 Jan 2012 23:07:53 +0000 (15:07 -0800)]
mm: filemap: pass __GFP_WRITE from grab_cache_page_write_begin()

Tell the page allocator that pages allocated through
grab_cache_page_write_begin() are expected to become dirty soon.

Signed-off-by: Johannes Weiner <jweiner@redhat.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Chris Mason <chris.mason@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: try to distribute dirty pages fairly across zones
Johannes Weiner [Tue, 10 Jan 2012 23:07:49 +0000 (15:07 -0800)]
mm: try to distribute dirty pages fairly across zones

The maximum number of dirty pages that exist in the system at any time is
determined by a number of pages considered dirtyable and a user-configured
percentage of those, or an absolute number in bytes.

This number of dirtyable pages is the sum of memory provided by all the
zones in the system minus their lowmem reserves and high watermarks, so
that the system can retain a healthy number of free pages without having
to reclaim dirty pages.

But there is a flaw in that we have a zoned page allocator which does not
care about the global state but rather the state of individual memory
zones.  And right now there is nothing that prevents one zone from filling
up with dirty pages while other zones are spared, which frequently leads
to situations where kswapd, in order to restore the watermark of free
pages, does indeed have to write pages from that zone's LRU list.  This
can interfere so badly with IO from the flusher threads that major
filesystems (btrfs, xfs, ext4) mostly ignore write requests from reclaim
already, taking away the VM's only possibility to keep such a zone
balanced, aside from hoping the flushers will soon clean pages from that
zone.

Enter per-zone dirty limits.  They are to a zone's dirtyable memory what
the global limit is to the global amount of dirtyable memory, and try to
make sure that no single zone receives more than its fair share of the
globally allowed dirty pages in the first place.  As the number of pages
considered dirtyable excludes the zones' lowmem reserves and high
watermarks, the maximum number of dirty pages in a zone is such that the
zone can always be balanced without requiring page cleaning.

As this is a placement decision in the page allocator and pages are
dirtied only after the allocation, this patch allows allocators to pass
__GFP_WRITE when they know in advance that the page will be written to and
become dirty soon.  The page allocator will then attempt to allocate from
the first zone of the zonelist - which on NUMA is determined by the task's
NUMA memory policy - that has not exceeded its dirty limit.

At first glance, it would appear that the diversion to lower zones can
increase pressure on them, but this is not the case.  With a full high
zone, allocations will be diverted to lower zones eventually, so it is
more of a shift in timing of the lower zone allocations.  Workloads that
previously could fit their dirty pages completely in the higher zone may
be forced to allocate from lower zones, but the amount of pages that
"spill over" are limited themselves by the lower zones' dirty constraints,
and thus unlikely to become a problem.

For now, the problem of unfair dirty page distribution remains for NUMA
configurations where the zones allowed for allocation are in sum not big
enough to trigger the global dirty limits, wake up the flusher threads and
remedy the situation.  Because of this, an allocation that could not
succeed on any of the considered zones is allowed to ignore the dirty
limits before going into direct reclaim or even failing the allocation,
until a future patch changes the global dirty throttling and flusher
thread activation so that they take individual zone states into account.

Test results

15M DMA + 3246M DMA32 + 504 Normal = 3765M memory
40% dirty ratio
16G USB thumb drive
10 runs of dd if=/dev/zero of=disk/zeroes bs=32k count=$((10 << 15))

seconds nr_vmscan_write
        (stddev)        min|     median|        max
xfs
vanilla:  549.747( 3.492)      0.000|      0.000|      0.000
patched:  550.996( 3.802)      0.000|      0.000|      0.000

fuse-ntfs
vanilla: 1183.094(53.178)  54349.000|  59341.000|  65163.000
patched:  558.049(17.914)      0.000|      0.000|     43.000

btrfs
vanilla:  573.679(14.015) 156657.000| 460178.000| 606926.000
patched:  563.365(11.368)      0.000|      0.000|   1362.000

ext4
vanilla:  561.197(15.782)      0.000|2725438.000|4143837.000
patched:  568.806(17.496)      0.000|      0.000|      0.000

Signed-off-by: Johannes Weiner <jweiner@redhat.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Tested-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Chris Mason <chris.mason@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: writeback: cleanups in preparation for per-zone dirty limits
Johannes Weiner [Tue, 10 Jan 2012 23:07:44 +0000 (15:07 -0800)]
mm: writeback: cleanups in preparation for per-zone dirty limits

The next patch will introduce per-zone dirty limiting functions in
addition to the traditional global dirty limiting.

Rename determine_dirtyable_memory() to global_dirtyable_memory() before
adding the zone-specific version, and fix up its documentation.

Also, move the functions to determine the dirtyable memory and the
function to calculate the dirty limit based on that together so that their
relationship is more apparent and that they can be commented on as a
group.

Signed-off-by: Johannes Weiner <jweiner@redhat.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Acked-by: Mel Gorman <mel@suse.de>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Chris Mason <chris.mason@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: exclude reserved pages from dirtyable memory
Johannes Weiner [Tue, 10 Jan 2012 23:07:42 +0000 (15:07 -0800)]
mm: exclude reserved pages from dirtyable memory

Per-zone dirty limits try to distribute page cache pages allocated for
writing across zones in proportion to the individual zone sizes, to reduce
the likelihood of reclaim having to write back individual pages from the
LRU lists in order to make progress.

This patch:

The amount of dirtyable pages should not include the full number of free
pages: there is a number of reserved pages that the page allocator and
kswapd always try to keep free.

The closer (reclaimable pages - dirty pages) is to the number of reserved
pages, the more likely it becomes for reclaim to run into dirty pages:

       +----------+ ---
       |   anon   |  |
       +----------+  |
       |          |  |
       |          |  -- dirty limit new    -- flusher new
       |   file   |  |                     |
       |          |  |                     |
       |          |  -- dirty limit old    -- flusher old
       |          |                        |
       +----------+                       --- reclaim
       | reserved |
       +----------+
       |  kernel  |
       +----------+

This patch introduces a per-zone dirty reserve that takes both the lowmem
reserve as well as the high watermark of the zone into account, and a
global sum of those per-zone values that is subtracted from the global
amount of dirtyable pages.  The lowmem reserve is unavailable to page
cache allocations and kswapd tries to keep the high watermark free.  We
don't want to end up in a situation where reclaim has to clean pages in
order to balance zones.

Not treating reserved pages as dirtyable on a global level is only a
conceptual fix.  In reality, dirty pages are not distributed equally
across zones and reclaim runs into dirty pages on a regular basis.

But it is important to get this right before tackling the problem on a
per-zone level, where the distance between reclaim and the dirty pages is
mostly much smaller in absolute numbers.

[akpm@linux-foundation.org: fix highmem build]
Signed-off-by: Johannes Weiner <jweiner@redhat.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Chris Mason <chris.mason@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agovmscan: add task name to warn_scan_unevictable() messages
KOSAKI Motohiro [Tue, 10 Jan 2012 23:07:40 +0000 (15:07 -0800)]
vmscan: add task name to warn_scan_unevictable() messages

If we need to know a usecase, caller program name is critical important.
Show it.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
David Rientjes <rientjes@google.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-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>
12 years agomm, debug: test for online nid when allocating on single node
David Rientjes [Tue, 10 Jan 2012 23:07:38 +0000 (15:07 -0800)]
mm, debug: test for online nid when allocating on single node

Calling alloc_pages_exact_node() means the allocation only passes the
zonelist of a single node into the page allocator.  If that node isn't
online, it's zonelist may never have been initialized causing a strange
oops that may not immediately be clear.

I recently debugged an issue where node 0 wasn't online and an allocator
was passing 0 to alloc_pages_exact_node() and it resulted in a NULL
pointer on zonelist->_zoneref.  If CONFIG_DEBUG_VM is enabled, though, it
would be nice to catch this a bit earlier.

Signed-off-by: David Rientjes <rientjes@google.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agofadvise: only initiate writeback for specified range with FADV_DONTNEED
Shawn Bohrer [Tue, 10 Jan 2012 23:07:35 +0000 (15:07 -0800)]
fadvise: only initiate writeback for specified range with FADV_DONTNEED

Previously POSIX_FADV_DONTNEED would start writeback for the entire file
when the bdi was not write congested.  This negatively impacts performance
if the file contains dirty pages outside of the requested range.  This
change uses __filemap_fdatawrite_range() to only initiate writeback for
the requested range.

Signed-off-by: Shawn Bohrer <sbohrer@rgmadvisors.com>
Acked-by: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoslub: min order when debug_guardpage_minorder > 0
Stanislaw Gruszka [Tue, 10 Jan 2012 23:07:32 +0000 (15:07 -0800)]
slub: min order when debug_guardpage_minorder > 0

Disable slub debug facilities and allocate slabs at minimal order when
debug_guardpage_minorder > 0 to increase probability to catch random
memory corruption by cpu exception.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Christoph Lameter <cl@linux.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Stanislaw Gruszka <sgruszka@redhat.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoPM/Hibernate: do not count debug pages as savable
Stanislaw Gruszka [Tue, 10 Jan 2012 23:07:31 +0000 (15:07 -0800)]
PM/Hibernate: do not count debug pages as savable

When debugging with CONFIG_DEBUG_PAGEALLOC and debug_guardpage_minorder >
0, we have lot of free pages that are not marked so.  Snapshot code
account them as savable, what cause hibernate memory preallocation
failure.

It is pretty hard to make hibernate allocation succeed with
debug_guardpage_minorder=1.  This change at least make it possible when
system has relatively big amount of RAM.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Acked-by: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: more intensive memory corruption debugging
Stanislaw Gruszka [Tue, 10 Jan 2012 23:07:28 +0000 (15:07 -0800)]
mm: more intensive memory corruption debugging

With CONFIG_DEBUG_PAGEALLOC configured, the CPU will generate an exception
on access (read,write) to an unallocated page, which permits us to catch
code which corrupts memory.  However the kernel is trying to maximise
memory usage, hence there are usually few free pages in the system and
buggy code usually corrupts some crucial data.

This patch changes the buddy allocator to keep more free/protected pages
and to interlace free/protected and allocated pages to increase the
probability of catching corruption.

When the kernel is compiled with CONFIG_DEBUG_PAGEALLOC,
debug_guardpage_minorder defines the minimum order used by the page
allocator to grant a request.  The requested size will be returned with
the remaining pages used as guard pages.

The default value of debug_guardpage_minorder is zero: no change from
current behaviour.

[akpm@linux-foundation.org: tweak documentation, s/flg/flag/]
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agokernel.h: add BUILD_BUG() macro
David Daney [Tue, 10 Jan 2012 23:07:25 +0000 (15:07 -0800)]
kernel.h: add BUILD_BUG() macro

We can place this in definitions that we expect the compiler to remove by
dead code elimination.  If this assertion fails, we get a nice error
message at build time.

The GCC function attribute error("message") was added in version 4.3, so
we define a new macro __linktime_error(message) to expand to this for
GCC-4.3 and later.  This will give us an error diagnostic from the
compiler on the line that fails.  For other compilers
__linktime_error(message) expands to nothing, and we have to be content
with a link time error, but at least we will still get a build error.

BUILD_BUG() expands to the undefined function __build_bug_failed() and
will fail at link time if the compiler ever emits code for it.  On GCC-4.3
and later, attribute((error())) is used so that the failure will be noted
at compile time instead.

Signed-off-by: David Daney <david.daney@cavium.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: DM <dm.n9107@gmail.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm/hugetlb.c: fix virtual address handling in hugetlb fault
KAMEZAWA Hiroyuki [Tue, 10 Jan 2012 23:07:22 +0000 (15:07 -0800)]
mm/hugetlb.c: fix virtual address handling in hugetlb fault

handle_mm_fault() passes 'faulted' address to hugetlb_fault().  This
address is not aligned to a hugepage boundary.

Most of the functions for hugetlb pages are aware of that and calculate an
alignment themselves.  However some functions such as
copy_user_huge_page() and clear_huge_page() don't handle alignment by
themselves.

This patch make hugeltb_fault() fix the alignment and pass an aligned
addresss (to address of a faulted hugepage) to functions.

[akpm@linux-foundation.org: use &=]
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agohugetlb: clarify hugetlb_instantiation_mutex usage
Michal Hocko [Tue, 10 Jan 2012 23:07:21 +0000 (15:07 -0800)]
hugetlb: clarify hugetlb_instantiation_mutex usage

Let's make it clear that we cannot race with other fault handlers due to
hugetlb (global) mutex.  Also make it clear that we want to keep pte_same
checks anayway to have a transition from the global mutex easier.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Cc: Hillf Danton <dhillf@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agohugetlb: detect race upon page allocation failure during COW
Hillf Danton [Tue, 10 Jan 2012 23:07:20 +0000 (15:07 -0800)]
hugetlb: detect race upon page allocation failure during COW

Currently we are not rechecking pte_same in hugetlb_cow after we take ptl
lock again in the page allocation failure code path and simply retry
again.  This is not an issue at the moment because hugetlb fault path is
protected by hugetlb_instantiation_mutex so we cannot race.

The original page is locked and so we cannot race even with the page
migration.

Let's add the pte_same check anyway as we want to be consistent with the
other check later in this function and be safe if we ever remove the
mutex.

[mhocko@suse.cz: reworded the changelog]
Signed-off-by: Hillf Danton <dhillf@gmail.com>
Signed-off-by: Michal Hocko <mhocko@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: account reaped page cache on inode cache pruning
Konstantin Khlebnikov [Tue, 10 Jan 2012 23:07:18 +0000 (15:07 -0800)]
mm: account reaped page cache on inode cache pruning

Inode cache pruning indirectly reclaims page-cache by invalidating mapping
pages.  Let's account them into reclaim-state to notice this progress in
memory reclaimer.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Dave Chinner <david@fromorbit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: avoid livelock on !__GFP_FS allocations
Mel Gorman [Tue, 10 Jan 2012 23:07:15 +0000 (15:07 -0800)]
mm: avoid livelock on !__GFP_FS allocations

Colin Cross reported;

  Under the following conditions, __alloc_pages_slowpath can loop forever:
  gfp_mask & __GFP_WAIT is true
  gfp_mask & __GFP_FS is false
  reclaim and compaction make no progress
  order <= PAGE_ALLOC_COSTLY_ORDER

  These conditions happen very often during suspend and resume,
  when pm_restrict_gfp_mask() effectively converts all GFP_KERNEL
  allocations into __GFP_WAIT.

  The oom killer is not run because gfp_mask & __GFP_FS is false,
  but should_alloc_retry will always return true when order is less
  than PAGE_ALLOC_COSTLY_ORDER.

In his fix, he avoided retrying the allocation if reclaim made no progress
and __GFP_FS was not set.  The problem is that this would result in
GFP_NOIO allocations failing that previously succeeded which would be very
unfortunate.

The big difference between GFP_NOIO and suspend converting GFP_KERNEL to
behave like GFP_NOIO is that normally flushers will be cleaning pages and
kswapd reclaims pages allowing GFP_NOIO to succeed after a short delay.
The same does not necessarily apply during suspend as the storage device
may be suspended.

This patch special cases the suspend case to fail the page allocation if
reclaim cannot make progress and adds some documentation on how
gfp_allowed_mask is currently used.  Failing allocations like this may
cause suspend to abort but that is better than a livelock.

[mgorman@suse.de: Rework fix to be suspend specific]
[rientjes@google.com: Move suspended device check to should_alloc_retry]
Reported-by: Colin Cross <ccross@android.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: reduce the amount of work done when updating min_free_kbytes
Mel Gorman [Tue, 10 Jan 2012 23:07:14 +0000 (15:07 -0800)]
mm: reduce the amount of work done when updating min_free_kbytes

When min_free_kbytes is updated, some pageblocks are marked
MIGRATE_RESERVE.  Ordinarily, this work is unnoticable as it happens early
in boot but on large machines with 1TB of memory, this has been reported
to delay boot times, probably due to the NUMA distances involved.

The bulk of the work is due to calling calling pageblock_is_reserved() an
unnecessary amount of times and accessing far more struct page metadata
than is necessary.  This patch significantly reduces the amount of work
done by setup_zone_migrate_reserve() improving boot times on 1TB machines.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: migrate: one less atomic operation
Jacobo Giralt [Tue, 10 Jan 2012 23:07:11 +0000 (15:07 -0800)]
mm: migrate: one less atomic operation

migrate_page_move_mapping() drops a reference from the old page after
unfreezing its counter.  Both operations can be merged into a single
atomic operation by directly unfreezing to one less reference.

The same applies to migrate_huge_page_move_mapping().

Signed-off-by: Jacobo Giralt <jacobo.giralt@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm-tracepoint: fix documentation and examples
Konstantin Khlebnikov [Tue, 10 Jan 2012 23:07:10 +0000 (15:07 -0800)]
mm-tracepoint: fix documentation and examples

We renamed the page-free mm tracepoints.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm-tracepoint: rename page-free events
Konstantin Khlebnikov [Tue, 10 Jan 2012 23:07:09 +0000 (15:07 -0800)]
mm-tracepoint: rename page-free events

Rename mm_page_free_direct into mm_page_free and mm_pagevec_free into
mm_page_free_batched

Since v2.6.33-5426-gc475dab the kernel triggers mm_page_free_direct for
all freed pages, not only for directly freed.  So, let's name it properly.
 For pages freed via page-list we also trigger mm_page_free_batched event.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: remove unused pagevec_free
Konstantin Khlebnikov [Tue, 10 Jan 2012 23:07:06 +0000 (15:07 -0800)]
mm: remove unused pagevec_free

It not exported and now nobody uses it.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Acked-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: add free_hot_cold_page_list() helper
Konstantin Khlebnikov [Tue, 10 Jan 2012 23:07:04 +0000 (15:07 -0800)]
mm: add free_hot_cold_page_list() helper

This patch adds helper free_hot_cold_page_list() to free list of 0-order
pages.  It frees pages directly from list without temporary page-vector.
It also calls trace_mm_pagevec_free() to simulate pagevec_free()
behaviour.

bloat-o-meter:

add/remove: 1/1 grow/shrink: 1/3 up/down: 267/-295 (-28)
function                                     old     new   delta
free_hot_cold_page_list                        -     264    +264
get_page_from_freelist                      2129    2132      +3
__pagevec_free                               243     239      -4
split_free_page                              380     373      -7
release_pages                                606     510     -96
free_page_list                               188       -    -188

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Minchan Kim <minchan.kim@gmail.com>
Acked-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agovmscan: activate executable pages after first usage
Konstantin Khlebnikov [Tue, 10 Jan 2012 23:07:03 +0000 (15:07 -0800)]
vmscan: activate executable pages after first usage

Logic added in commit 8cab4754d24a0 ("vmscan: make mapped executable pages
the first class citizen") was noticeably weakened in commit
645747462435d84 ("vmscan: detect mapped file pages used only once").

Currently these pages can become "first class citizens" only after second
usage.  After this patch page_check_references() will activate they after
first usage, and executable code gets yet better chance to stay in memory.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agovmscan: promote shared file mapped pages
Konstantin Khlebnikov [Tue, 10 Jan 2012 23:06:59 +0000 (15:06 -0800)]
vmscan: promote shared file mapped pages

Commit 645747462435 ("vmscan: detect mapped file pages used only once")
greatly decreases lifetime of single-used mapped file pages.
Unfortunately it also decreases life time of all shared mapped file
pages.  Because after commit bf3f3bc5e7347 ("mm: don't mark_page_accessed
in fault path") page-fault handler does not mark page active or even
referenced.

Thus page_check_references() activates file page only if it was used twice
while it stays in inactive list, meanwhile it activates anon pages after
first access.  Inactive list can be small enough, this way reclaimer can
accidentally throw away any widely used page if it wasn't used twice in
short period.

After this patch page_check_references() also activate file mapped page at
first inactive list scan if this page is already used multiple times via
several ptes.

I found this while trying to fix degragation in rhel6 (~2.6.32) from rhel5
(~2.6.18).  There a complete mess with >100 web/mail/spam/ftp containers,
they share all their files but there a lot of anonymous pages: ~500mb
shared file mapped memory and 15-20Gb non-shared anonymous memory.  In
this situation major-pagefaults are very costly, because all containers
share the same page.  In my load kernel created a disproportionate
pressure on the file memory, compared with the anonymous, they equaled
only if I raise swappiness up to 150 =)

These patches actually wasn't helped a lot in my problem, but I saw
noticable (10-20 times) reduce in count and average time of
major-pagefault in file-mapped areas.

Actually both patches are fixes for commit v2.6.33-5448-g6457474, because
it was aimed at one scenario (singly used pages), but it breaks the logic
in other scenarios (shared and/or executable pages)

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Acked-by: Pekka Enberg <penberg@kernel.org>
Acked-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm/page-writeback.c: make determine_dirtyable_memory static again
Johannes Weiner [Tue, 10 Jan 2012 23:06:57 +0000 (15:06 -0800)]
mm/page-writeback.c: make determine_dirtyable_memory static again

The tracing ring-buffer used this function briefly, but not anymore.
Make it local to the writeback code again.

Also, move the function so that no forward declaration needs to be
reintroduced.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Tue, 10 Jan 2012 01:37:37 +0000 (17:37 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  vfs: new helper - d_make_root()
  dcache: use a dispose list in select_parent
  ceph: d_alloc_root() may fail
  ext4: fix failure exits
  isofs: inode leak on mount failure

12 years agovfs: new helper - d_make_root()
Al Viro [Sun, 8 Jan 2012 21:49:21 +0000 (16:49 -0500)]
vfs: new helper - d_make_root()

d_alloc_root() with iput() in case of allocation failure...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agodcache: use a dispose list in select_parent
Dave Chinner [Tue, 23 Aug 2011 08:56:24 +0000 (18:56 +1000)]
dcache: use a dispose list in select_parent

select_parent currently abuses the dentry cache LRU to provide
cleanup features for child dentries that need to be freed. It moves
them to the tail of the LRU, then tells shrink_dcache_parent() to
calls __shrink_dcache_sb to unconditionally move them to a dispose
list (as DCACHE_REFERENCED is ignored). __shrink_dcache_sb() has to
relock the dentries to move them off the LRU onto the dispose list,
but otherwise does not touch the dentries that select_parent() moved
to the tail of the LRU. It then passses the dispose list to
shrink_dentry_list() which tries to free the dentries.

IOWs, the use of __shrink_dcache_sb() is superfluous - we can build
exactly the same list of dentries for disposal directly in
select_parent() and call shrink_dentry_list() instead of calling
__shrink_dcache_sb() to do that. This means that we avoid long holds
on the lru lock walking the LRU moving dentries to the dispose list
We also avoid the need to relock each dentry just to move it off the
LRU, reducing the numebr of times we lock each dentry to dispose of
them in shrink_dcache_parent() from 3 to 2 times.

Further, we remove one of the two callers of __shrink_dcache_sb().
This also means that __shrink_dcache_sb can be moved into back into
prune_dcache_sb() and we no longer have to handle referenced
dentries conditionally, simplifying the code.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-next
Linus Torvalds [Mon, 9 Jan 2012 22:47:06 +0000 (14:47 -0800)]
Merge git://git./linux/kernel/git/davem/sparc-next

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-next:
  sparc32: remove unused file: include/asm/pgtsun4.h
  sparc32: fix PAGE_SIZE definition
  sparc32: enable different preemptions models
  sparc32: support atomic64_t
  apbuart: fix section mismatch warning
  sparc32: drop useless preprocessor conditional in atomic_32.h
  sparc32: drop unused atomic24 support

12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Mon, 9 Jan 2012 22:46:52 +0000 (14:46 -0800)]
Merge git://git./linux/kernel/git/davem/net

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
  igmp: Avoid zero delay when receiving odd mixture of IGMP queries
  netdev: make net_device_ops const
  bcm63xx: make ethtool_ops const
  usbnet: make ethtool_ops const
  net: Fix build with INET disabled.
  net: introduce netif_addr_lock_nested() and call if when appropriate
  net: correct lock name in dev_[uc/mc]_sync documentations.
  net: sk_update_clone is only used in net/core/sock.c
  8139cp: fix missing napi_gro_flush.
  pktgen: set correct max and min in pktgen_setup_inject()
  smsc911x: Unconditionally include linux/smscphy.h in smsc911x.h
  asix: fix infinite loop in rx_fixup()
  net: Default UDP and UNIX diag to 'n'.
  r6040: fix typo in use of MCR0 register bits
  net: fix sock_clone reference mismatch with tcp memcontrol

12 years agoMerge tag 'clk' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 9 Jan 2012 22:44:15 +0000 (14:44 -0800)]
Merge tag 'clk' of git://git./linux/kernel/git/arm/arm-soc

clock management changes for i.MX

Another simple series related to clock management, this time only for
imx.

* tag 'clk' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  ARM: mxs: select HAVE_CLK_PREPARE for clock
  clk: add config option HAVE_CLK_PREPARE into Kconfig
  ASoC: mxs-saif: convert to clk_prepare/clk_unprepare
  video: mxsfb: convert to clk_prepare/clk_unprepare
  serial: mxs-auart: convert to clk_prepare/clk_unprepare
  net: flexcan: convert to clk_prepare/clk_unprepare
  mtd: gpmi-lib: convert to clk_prepare/clk_unprepare
  mmc: mxs-mmc: convert to clk_prepare/clk_unprepare
  dma: mxs-dma: convert to clk_prepare/clk_unprepare
  net: fec: add clk_prepare/clk_unprepare
  ARM: mxs: convert platform code to clk_prepare/clk_unprepare
  clk: add helper functions clk_prepare_enable and clk_disable_unprepare

Fix up trivial conflicts in drivers/net/ethernet/freescale/fec.c due to
commit 0ebafefcaa7a ("net: fec: add clk_prepare/clk_unprepare") clashing
trivially with commit e163cc97f9ac ("net/fec: fix the .remove code").

12 years agoMerge tag 'timer' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 9 Jan 2012 22:40:48 +0000 (14:40 -0800)]
Merge tag 'timer' of git://git./linux/kernel/git/arm/arm-soc

timer changes for msm

A very simple series. We used to have more churn in the timer
area, so this is kept separate. Will probably put this into the
drivers series next time.

* tag 'timer' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  msm: timer: Use clockevents_config_and_register()
  msm: timer: Setup interrupt after registering clockevent
  msm: timer: Remove SoC specific #ifdefs
  msm: timer: Remove msm_clocks[] and simplify code
  msm: timer: Fix ONESHOT mode interrupts
  msm: timer: Use GPT for clockevents and DGT for clocksource
  msm: timer: Cleanup #includes and #defines
  msm: timer: Tighten #ifdef for local timer support

12 years agoMerge tag 'pm' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 9 Jan 2012 22:39:59 +0000 (14:39 -0800)]
Merge tag 'pm' of git://git./linux/kernel/git/arm/arm-soc

power management changes for omap and imx

A significant part of the changes for these two platforms went into
power management, so they are split out into a separate branch.

* tag 'pm' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (65 commits)
  ARM: imx6: remove __CPUINIT annotation from v7_invalidate_l1
  ARM: imx6: fix v7_invalidate_l1 by adding I-Cache invalidation
  ARM: imx6q: resume PL310 only when CACHE_L2X0 defined
  ARM: imx6q: build pm code only when CONFIG_PM selected
  ARM: mx5: use generic irq chip pm interface for pm functions on
  ARM: omap: pass minimal SoC/board data for UART from dt
  arm/dts: Add minimal device tree support for omap2420 and omap2430
  omap-serial: Add minimal device tree support
  omap-serial: Use default clock speed (48Mhz) if not specified
  omap-serial: Get rid of all pdev->id usage
  ARM: OMAP2+: hwmod: Add a new flag to handle hwmods left enabled at init
  ARM: OMAP4: PRM: use PRCM interrupt handler
  ARM: OMAP3: pm: use prcm chain handler
  ARM: OMAP: hwmod: add support for selecting mpu_irq for each wakeup pad
  ARM: OMAP2+: mux: add support for PAD wakeup interrupts
  ARM: OMAP: PRCM: add suspend prepare / finish support
  ARM: OMAP: PRCM: add support for chain interrupt handler
  ARM: OMAP3/4: PRM: add functions to read pending IRQs, PRM barrier
  ARM: OMAP2+: hwmod: Add API to enable IO ring wakeup
  ARM: OMAP2+: mux: add wakeup-capable hwmod mux entries to dynamic list
  ...

12 years agoMerge tag 'drivers' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 9 Jan 2012 22:39:22 +0000 (14:39 -0800)]
Merge tag 'drivers' of git://git./linux/kernel/git/arm/arm-soc

Driver specific changes

Again, a lot of platforms have changes in here: pxa, samsung, omap,
at91, imx, ...

* tag 'drivers' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (54 commits)
  ARM: sa1100: clean up of the clock support
  ARM: pxa: add dummy clock for sa1100-rtc
  RTC: sa1100: support sa1100, pxa and mmp soc families
  RTC: sa1100: remove redundant code of setting alarm
  RTC: sa1100: Clean out ost register
  Input: zylonite-wm97xx - replace IRQ_GPIO() with gpio_to_irq()
  pcmcia: pxa: replace IRQ_GPIO() with gpio_to_irq()
  ARM: EXYNOS: Modified files for SPI consolidation work
  ARM: S5P64X0: Enable SDHCI support
  ARM: S5P64X0: Add lookup of sdhci-s3c clocks using generic names
  ARM: S5P64X0: Add HSMMC setup for host Controller
  ARM: EXYNOS: Add USB OHCI support to ORIGEN board
  USB: Add Samsung Exynos OHCI diver
  ARM: EXYNOS: Add USB OHCI support to SMDKV310 board
  ARM: EXYNOS: Add USB OHCI device
  net: macb: fix build break with !CONFIG_OF
  i2c: tegra: Support DVC controller in device tree
  i2c: tegra: Add __devinit/exit to probe/remove
  net/at91_ether: use gpio_is_valid for phy IRQ line
  ARM: at91/net: add macb ethernet controller in 9g45/9g20 DT
  ...

12 years agoMerge tag 'devel' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 9 Jan 2012 22:38:51 +0000 (14:38 -0800)]
Merge tag 'devel' of git://git./linux/kernel/git/arm/arm-soc

New feature development

This adds support for new features, and contains stuff from most
platforms. A number of these patches could have fit into other
branches, too, but were small enough not to cause too much
confusion here.

* tag 'devel' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (28 commits)
  mfd/db8500-prcmu: remove support for early silicon revisions
  ARM: ux500: fix the smp_twd clock calculation
  ARM: ux500: remove support for early silicon revisions
  ARM: ux500: update register files
  ARM: ux500: register DB5500 PMU dynamically
  ARM: ux500: update ASIC detection for U5500
  ARM: ux500: support DB8520
  ARM: picoxcell: implement watchdog restart
  ARM: OMAP3+: hwmod data: Add the default clockactivity for I2C
  ARM: OMAP3: hwmod data: disable multiblock reads on MMC1/2 on OMAP34xx/35xx <= ES2.1
  ARM: OMAP: USB: EHCI and OHCI hwmod structures for OMAP4
  ARM: OMAP: USB: EHCI and OHCI hwmod structures for OMAP3
  ARM: OMAP: hwmod data: Add support for AM35xx UART4/ttyO3
  ARM: Orion: Remove address map info from all platform data structures
  ARM: Orion: Get address map from plat-orion instead of via platform_data
  ARM: Orion: mbus_dram_info consolidation
  ARM: Orion: Consolidate the address map setup
  ARM: Kirkwood: Add configuration for MPP12 as GPIO
  ARM: Kirkwood: Recognize A1 revision of 6282 chip
  ARM: ux500: update the MOP500 GPIO assignments
  ...

12 years agoMerge tag 'boards' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 9 Jan 2012 22:37:41 +0000 (14:37 -0800)]
Merge tag 'boards' of git://git./linux/kernel/git/arm/arm-soc

Board-level changes

This adds and extends support for specific boards on a number of
ARM platforms:  omap, imx, samsung, tegra, ...

* tag 'boards' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (49 commits)
  Enable 32 bit flash support for iMX21ADS board
  ARM: mx31pdk: Add MC13783 RTC support
  iomux-mx25: configuration to support CSPI3 on CSI pins
  MX1:apf9328: Add i2c support
  mioa701: add newly available DoC G3 chip
  arm/tegra: remove __initdata annotation from pinmux tables
  arm/tegra: Use bus notifiers to trigger pinmux setup
  arm/tegra: Refactor board-*-pinmux.c to share code
  arm/tegra: Fix mistake in Trimslice's pinmux
  arm/tegra: Rework Seaboard-vs-Ventana pinmux table
  arm/tegra: Remove useless entries from ventana_pinmux[]
  arm/tegra: PCIe: Remove include of mach/pinmux.h
  arm/tegra: Harmony PCIe: Don't touch pinmux
  arm/tegra: Add AUXDATA for tegra-pinmux and tegra-gpio
  arm/tegra: Split Seaboard GPIO table to allow for Ventana
  ARM: imx6q: generate imx6q dtb files
  arm/imx6q: Rename Sabreauto to Armadillo2
  arm/imx6q-sabrelite: add enet phy ksz9021rn fixup
  arm/imx6: add imx6q sabrelite board support
  dts/imx: rename uart labels to consistent with hw spec
  ...

12 years agoMerge tag 'soc' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 9 Jan 2012 22:33:17 +0000 (14:33 -0800)]
Merge tag 'soc' of git://git./linux/kernel/git/arm/arm-soc

SoC-level changes for tegra and omap

This adds support for the new tegra30 SoC, as well as small
changes to support minor variations of existing omap SoCs.

* tag 'soc' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (26 commits)
  arm/tegra: Compile tegra_dt_init_irq only when CONFIG_OF
  arm/tegra: Make MACH_TEGRA_DT depend on ARCH_TEGRA_2x_SOC
  arm/tegra: Delete tegra_init_clock()
  arm/tegra: Fix section mismatch errors in tegra30 pinmux
  arm/tegra: Fix section mismatch errors in tegra20 pinmux
  arm/tegra: refresh defconfig for tegra30
  arm/tegra: add support for tegra30 based board cardhu
  arm/tegra: implement support for tegra30
  arm/tegra: pinmux tables and definitions for tegra30
  arm/tegra: add new fields to struct tegra_pingroup_desc
  arm/tegra: prepare pinmux code for multiple tegra variants
  arm/tegra: rename tegra20 pinmux files
  arm/tegra: generalize L2 cache initialization
  arm/tegra: use PMC reset
  arm/tegra: rename board-dt.c to board-dt-tegra20.c
  arm/tegra: prepare early init for multiple tegra variants
  arm/tegra: don't export clk_measure_input_freq
  arm/tegra: prepare clock code for multiple tegra variants
  arm/tegra: cleanup tegra20 support
  arm/tegra: clk_get should not be fatal
  ...

Fix up trivial conflict in arch/arm/mach-tegra/board-dt-tegra20.c

12 years agoMerge tag 'cleanup2' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 9 Jan 2012 22:30:28 +0000 (14:30 -0800)]
Merge tag 'cleanup2' of git://git./linux/kernel/git/arm/arm-soc

Cleanups for the Samsung platforms

Various cleanup changes that the device driver changes are built upon.
Since the samsung cleanups depend on the device tree series, which
depends on the first set of cleanups for tegra.

* tag 'cleanup2' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  ARM: EXYNOS: Use gpio_request_one
  ARM: S5PV210: Use gpio_request_one
  ARM: S3C64XX: Modified according to SPI consolidation work
  ARM: S5PV210: Modified files for SPI consolidation work
  ARM: S5P64X0: Modified files for SPI consolidation work
  ARM: S5PC100: Modified files for SPI consolidation work
  ARM: S3C64XX: Modified files for SPI consolidation work
  ARM: SAMSUNG: Consolidation of SPI platform devices to plat-samsung
  ARM: SAMSUNG: Remove SPI bus clocks from platform data
  ARM: S5PV210: Add SPI clkdev support
  ARM: S5P64X0: Add SPI clkdev support
  ARM: S5PC100: Add SPI clkdev support
  ARM: S3C64XX: Add SPI clkdev support
  spi/s3c64xx: Use bus clocks created using clkdev
  mmc: sdhci-s3c: Use generic clock names for sdhci bus clock options
  ARM: SAMSUNG: Add lookup of sdhci-s3c clocks using generic names
  ARM: SAMSUNG: Remove SDHCI bus clocks from platform data
  ARM: SAMSUNG: Use kmemdup rather than duplicating its implementation
  ARM: EXYNOS: remove exynos4_scu_enable()

12 years agoMerge tag 'dt' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 9 Jan 2012 22:28:38 +0000 (14:28 -0800)]
Merge tag 'dt' of git://git./linux/kernel/git/arm/arm-soc

Device tree conversions for samsung and tegra

Both platforms had some initial device tree support, but this adds
much more to actually make it usable.

* tag 'dt' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (45 commits)
  ARM: dts: Add intial dts file for EXYNOS4210 SoC, SMDKV310 and ORIGEN
  ARM: EXYNOS: Add Exynos4 device tree enabled board file
  rtc: rtc-s3c: Add device tree support
  input: samsung-keypad: Add device tree support
  ARM: S5PV210: Modify platform data for pl330 driver
  ARM: S5PC100: Modify platform data for pl330 driver
  ARM: S5P64x0: Modify platform data for pl330 driver
  ARM: EXYNOS: Add a alias for pdma clocks
  ARM: EXYNOS: Limit usage of pl330 device instance to non-dt build
  ARM: SAMSUNG: Add device tree support for pl330 dma engine wrappers
  DMA: PL330: Add device tree support
  ARM: EXYNOS: Modify platform data for pl330 driver
  DMA: PL330: Infer transfer direction from transfer request instead of platform data
  DMA: PL330: move filter function into driver
  serial: samsung: Fix build for non-Exynos4210 devices
  serial: samsung: add device tree support
  serial: samsung: merge probe() function from all SoC specific extensions
  serial: samsung: merge all SoC specific port reset functions
  ARM: SAMSUNG: register uart clocks to clock lookup list
  serial: samsung: remove all uses of get_clksrc and set_clksrc
  ...

Fix up fairly trivial conflicts in arch/arm/mach-s3c2440/clock.c and
drivers/tty/serial/Kconfig both due to just adding code close to
changes.

12 years agoMerge tag 'cleanup' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Mon, 9 Jan 2012 22:21:03 +0000 (14:21 -0800)]
Merge tag 'cleanup' of git://git./linux/kernel/git/arm/arm-soc

Cleanups on various subarchitectures

Cleanup patches for various ARM platforms and some of their associated
drivers, the bulk of these is for mach-91.

Arnd ended up pulling in the restart branch from Russell in order to
fix up some simple but annoying merge conflicts.

* tag 'cleanup' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (44 commits)
  arm/at91: fix build of stamp9g20
  ARM: u300: delete memory.h
  MAINTAINERS: add maintainer entry for Picochip picoxcell
  ARM: picoxcell: move io mappings to common.c
  ARM: picoxcell: don't reserve irq_descs
  ARM: picoxcell: remove mach/memory.h
  ARM: at91: delete the pcontrol_g20_defconfig
  arm/tegra: Remove code that's ifndef CONFIG_ARM_GIC
  arm/tegra: remove unused defines
  arm/tegra: fix variable formatting in makefile
  ARM: davinci: vpif: move code to driver core header from platform
  ARM: at91/gpio: fix display of number of irq setuped
  ARM: at91/gpio: drop PIN_BASE
  ARM: at91/udc: use gpio_is_valid to check the gpio
  ARM: at91/ohci: use gpio_is_valid to check the gpio
  ARM: at91/nand: use gpio_is_valid to check the gpio
  ARM: at91/mmc: use gpio_is_valid to check the gpio
  ARM: at91/ide: use gpio_is_valid to check the gpio
  ARM: at91/pata: use gpio_is_valid to check the gpio
  ARM: at91/soc: use gpio_is_valid to check the gpio
  ...

12 years agoMerge tag 'fixes-non-critical' of git://git.kernel.org/pub/scm/linux/kernel/git/arm...
Linus Torvalds [Mon, 9 Jan 2012 22:20:39 +0000 (14:20 -0800)]
Merge tag 'fixes-non-critical' of git://git./linux/kernel/git/arm/arm-soc

Non-critical bug fixes

Simple bug fixes that were not considered important enough for inclusion
into 3.2.

* tag 'fixes-non-critical' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  MAINTAINERS: update pxa and mmp
  ARM: pxa: Include linux/export.h in balloon3.c
  ARM: OMAP4: clock: Add CPU local timer clock node
  ARM: OMAP4: hwmod: Don't wait for the idle status if modulemode is not supported
  ARM: OMAP: AM3517/3505: fix crash on boot due to incorrect voltagedomain data
  ARM: OMAP: hwmod data: fix the panic on Nokia RM-680 during boot
  ARM: OMAP2+: DMA: Workaround for invalid destination position
  ARM: OMAP2+: DMA: Workaround for invalid source position

12 years agotracing/mm: Move include of trace/events/kmem.h out of header into slab.c
Steven Rostedt [Mon, 9 Jan 2012 22:15:42 +0000 (17:15 -0500)]
tracing/mm: Move include of trace/events/kmem.h out of header into slab.c

Including trace/events/*.h TRACE_EVENT() macro headers in other headers
can cause strange side effects if another trace/event/*.h header
includes that header.  Having trace/events/kmem.h inside slab_def.h
caused a compile error in sparc64 when changes were done to some header
files.  Moving the kmem.h trace header out of slab.h and into slab.c
fixes the problem.

Note, both slub.c and slob.c already include the trace/events/kmem.h
file. Only slab.c had it missing.

Link: http://lkml.kernel.org/r/20120105190405.1e3191fb5a43b2a0f1655e1f@canb.auug.org.au
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoigmp: Avoid zero delay when receiving odd mixture of IGMP queries
Ben Hutchings [Mon, 9 Jan 2012 22:06:46 +0000 (14:06 -0800)]
igmp: Avoid zero delay when receiving odd mixture of IGMP queries

Commit 5b7c84066733c5dfb0e4016d939757b38de189e4 ('ipv4: correct IGMP
behavior on v3 query during v2-compatibility mode') added yet another
case for query parsing, which can result in max_delay = 0.  Substitute
a value of 1, as in the usual v3 case.

Reported-by: Simon McVittie <smcv@debian.org>
References: http://bugs.debian.org/654876
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonetdev: make net_device_ops const
stephen hemminger [Thu, 5 Jan 2012 19:10:25 +0000 (19:10 +0000)]
netdev: make net_device_ops const

More drivers where net_device_ops should be const.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobcm63xx: make ethtool_ops const
stephen hemminger [Thu, 5 Jan 2012 19:10:24 +0000 (19:10 +0000)]
bcm63xx: make ethtool_ops const

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agousbnet: make ethtool_ops const
stephen hemminger [Thu, 5 Jan 2012 19:10:23 +0000 (19:10 +0000)]
usbnet: make ethtool_ops const

The ethtool_ops table of function pointers should be const.
Fix all the usb network drivers.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: Fix build with INET disabled.
David S. Miller [Mon, 9 Jan 2012 21:44:23 +0000 (13:44 -0800)]
net: Fix build with INET disabled.

> net/core/sock.c: In function 'sk_update_clone':
> net/core/sock.c:1278:3: error: implicit declaration of function 'sock_update_memcg'

Reported-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoceph: d_alloc_root() may fail
Al Viro [Mon, 9 Jan 2012 21:34:32 +0000 (16:34 -0500)]
ceph: d_alloc_root() may fail

... and ceph_init_dentry(NULL) will oops

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agoMerge branch 'for-3.3' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu
Linus Torvalds [Mon, 9 Jan 2012 21:08:28 +0000 (13:08 -0800)]
Merge branch 'for-3.3' of git://git./linux/kernel/git/tj/percpu

* 'for-3.3' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu:
  percpu: Remove irqsafe_cpu_xxx variants

Fix up conflict in arch/x86/include/asm/percpu.h due to clash with
cebef5beed3d ("x86: Fix and improve percpu_cmpxchg{8,16}b_double()")
which edited the (now removed) irqsafe_cpu_cmpxchg*_double code.

12 years agoMerge branch 'for-3.3' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup
Linus Torvalds [Mon, 9 Jan 2012 20:59:24 +0000 (12:59 -0800)]
Merge branch 'for-3.3' of git://git./linux/kernel/git/tj/cgroup

* 'for-3.3' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup: (21 commits)
  cgroup: fix to allow mounting a hierarchy by name
  cgroup: move assignement out of condition in cgroup_attach_proc()
  cgroup: Remove task_lock() from cgroup_post_fork()
  cgroup: add sparse annotation to cgroup_iter_start() and cgroup_iter_end()
  cgroup: mark cgroup_rmdir_waitq and cgroup_attach_proc() as static
  cgroup: only need to check oldcgrp==newgrp once
  cgroup: remove redundant get/put of task struct
  cgroup: remove redundant get/put of old css_set from migrate
  cgroup: Remove unnecessary task_lock before fetching css_set on migration
  cgroup: Drop task_lock(parent) on cgroup_fork()
  cgroups: remove redundant get/put of css_set from css_set_check_fetched()
  resource cgroups: remove bogus cast
  cgroup: kill subsys->can_attach_task(), pre_attach() and attach_task()
  cgroup, cpuset: don't use ss->pre_attach()
  cgroup: don't use subsys->can_attach_task() or ->attach_task()
  cgroup: introduce cgroup_taskset and use it in subsys->can_attach(), cancel_attach() and attach()
  cgroup: improve old cgroup handling in cgroup_attach_proc()
  cgroup: always lock threadgroup during migration
  threadgroup: extend threadgroup_lock() to cover exit and exec
  threadgroup: rename signal->threadgroup_fork_lock to ->group_rwsem
  ...

Fix up conflict in kernel/cgroup.c due to commit e0197aae59e5: "cgroups:
fix a css_set not found bug in cgroup_attach_proc" that already
mentioned that the bug is fixed (differently) in Tejun's cgroup
patchset. This one, in other words.

12 years agoext4: fix failure exits
Al Viro [Mon, 9 Jan 2012 20:53:24 +0000 (15:53 -0500)]
ext4: fix failure exits

a) leaking root dentry is bad
b) in case of failed ext4_mb_init() we don't want to do ext4_mb_release()
c) OTOH, in the same case we *do* want ext4_ext_release()

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs
Linus Torvalds [Mon, 9 Jan 2012 20:51:21 +0000 (12:51 -0800)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-fs

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
  ext2/3/4: delete unneeded includes of module.h
  ext{3,4}: Fix potential race when setversion ioctl updates inode
  udf: Mark LVID buffer as uptodate before marking it dirty
  ext3: Don't warn from writepage when readonly inode is spotted after error
  jbd: Remove j_barrier mutex
  reiserfs: Force inode evictions before umount to avoid crash
  reiserfs: Fix quota mount option parsing
  udf: Treat symlink component of type 2 as /
  udf: Fix deadlock when converting file from in-ICB one to normal one
  udf: Cleanup calling convention of inode_getblk()
  ext2: Fix error handling on inode bitmap corruption
  ext3: Fix error handling on inode bitmap corruption
  ext3: replace ll_rw_block with other functions
  ext3: NULL dereference in ext3_evict_inode()
  jbd: clear revoked flag on buffers before a new transaction started
  ext3: call ext3_mark_recovery_complete() when recovery is really needed

12 years agoMerge branch 'for-linus' of git://git.open-osd.org/linux-open-osd
Linus Torvalds [Mon, 9 Jan 2012 20:51:01 +0000 (12:51 -0800)]
Merge branch 'for-linus' of git://git.open-osd.org/linux-open-osd

* 'for-linus' of git://git.open-osd.org/linux-open-osd:
  ore: Must support none-PAGE-aligned IO
  ore: fix BUG_ON, too few sgs when reading
  ore: Fix crash in case of an IO error.
  ore: FIX breakage when MISC_FILESYSTEMS is not set

12 years agoMerge branch 'for-linus' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Mon, 9 Jan 2012 20:50:15 +0000 (12:50 -0800)]
Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs

* 'for-linus' of git://oss.sgi.com/xfs/xfs:
  xfs: fix endian conversion issue in discard code

12 years agonet: introduce netif_addr_lock_nested() and call if when appropriate
Jiri Pirko [Mon, 9 Jan 2012 06:36:54 +0000 (06:36 +0000)]
net: introduce netif_addr_lock_nested() and call if when appropriate

dev_uc_sync() and dev_mc_sync() are acquiring netif_addr_lock for
destination device of synchronization. Since netif_addr_lock is
already held at the time for source device, this triggers lockdep
deadlock warning.

There's no way this deadlock can happen so use spin_lock_nested() to
silence the warning.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: correct lock name in dev_[uc/mc]_sync documentations.
Jiri Pirko [Mon, 9 Jan 2012 06:18:34 +0000 (06:18 +0000)]
net: correct lock name in dev_[uc/mc]_sync documentations.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'staging-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Mon, 9 Jan 2012 20:18:17 +0000 (12:18 -0800)]
Merge branch 'staging-next' of git://git./linux/kernel/git/gregkh/staging

* 'staging-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging: (466 commits)
  net/hyperv: Add support for jumbo frame up to 64KB
  net/hyperv: Add NETVSP protocol version negotiation
  net/hyperv: Remove unnecessary kmap_atomic in netvsc driver
  staging/rtl8192e: Register against lib80211
  staging/rtl8192e: Convert to lib80211_crypt_info
  staging/rtl8192e: Convert to lib80211_crypt_data and lib80211_crypt_ops
  staging/rtl8192e: Add lib80211.h to rtllib.h
  staging/mei: add watchdog device registration wrappers
  drm/omap: GEM, deal with cache
  staging: vt6656: int.c, int.h: Change return of function to void
  staging: usbip: removed unused definitions from header
  staging: usbip: removed dead code from receive function
  staging:iio: Drop {mark,unmark}_in_use callbacks
  staging:iio: Drop buffer mark_param_change callback
  staging:iio: Drop the unused buffer enable() and is_enabled() callbacks
  staging:iio: Drop buffer busy flag
  staging:iio: Make sure a device is only opened once at a time
  staging:iio: Disallow modifying buffer size when buffer is enabled
  staging:iio: Disallow changing scan elements in all buffered modes
  staging:iio: Use iio_buffer_enabled instead of open coding it
  ...

Fix up conflict in drivers/staging/iio/adc/ad799x_core.c (removal of
module_init due to using module_i2c_driver() helper, next to removal of
MODULE_ALIAS due to using MODULE_DEVICE_TABLE instead).

12 years agoMerge branch 'usb-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Mon, 9 Jan 2012 20:09:47 +0000 (12:09 -0800)]
Merge branch 'usb-next' of git://git./linux/kernel/git/gregkh/usb

* 'usb-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (232 commits)
  USB: Add USB-ID for Multiplex RC serial adapter to cp210x.c
  xhci: Clean up 32-bit build warnings.
  USB: update documentation for usbmon
  usb: usb-storage doesn't support dynamic id currently, the patch disables the feature to fix an oops
  drivers/usb/class/cdc-acm.c: clear dangling pointer
  drivers/usb/dwc3/dwc3-pci.c: introduce missing kfree
  drivers/usb/host/isp1760-if.c: introduce missing kfree
  usb: option: add ZD Incorporated HSPA modem
  usb: ch9: fix up MaxStreams helper
  USB: usb-skeleton.c: cleanup open_count
  USB: usb-skeleton.c: fix open/disconnect race
  xhci: Properly handle COMP_2ND_BW_ERR
  USB: remove dead code from suspend/resume path
  USB: add quirk for another camera
  drivers: usb: wusbcore: Fix dependency for USB_WUSB
  xhci: Better debugging for critical host errors.
  xhci: Be less verbose during URB cancellation.
  xhci: Remove debugging about ring structure allocation.
  xhci: Remove debugging about toggling cycle bits.
  xhci: Remove debugging for individual transfers.
  ...

12 years agoMerge branch 'tty-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Mon, 9 Jan 2012 20:09:24 +0000 (12:09 -0800)]
Merge branch 'tty-next' of git://git./linux/kernel/git/gregkh/tty

* 'tty-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty: (65 commits)
  tty: serial: imx: move del_timer_sync() to avoid potential deadlock
  imx: add polled io uart methods
  imx: Add save/restore functions for UART control regs
  serial/imx: let probing fail for the dt case without a valid alias
  serial/imx: propagate error from of_alias_get_id instead of using -ENODEV
  tty: serial: imx: Allow UART to be a source for wakeup
  serial: driver for m32 arch should not have DEC alpha errata
  serial/documentation: fix documented name of DCD cpp symbol
  atmel_serial: fix spinlock lockup in RS485 code
  tty: Fix memory leak in virtual console when enable unicode translation
  serial: use DIV_ROUND_CLOSEST instead of open coding it
  serial: add support for 400 and 800 v3 series Titan cards
  serial: bfin-uart: Remove ASYNC_CTS_FLOW flag for hardware automatic CTS.
  serial: bfin-uart: Enable hardware automatic CTS only when CTS pin is available.
  serial: make FSL errata depend on 8250_CONSOLE, not just 8250
  serial: add irq handler for Freescale 16550 errata.
  serial: manually inline serial8250_handle_port
  serial: make 8250 timeout use the specified IRQ handler
  serial: export the key functions for an 8250 IRQ handler
  serial: clean up parameter passing for 8250 Rx IRQ handling
  ...

12 years agoMerge branch 'char-misc-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Mon, 9 Jan 2012 20:08:59 +0000 (12:08 -0800)]
Merge branch 'char-misc-next' of git://git./linux/kernel/git/gregkh/char-misc

* 'char-misc-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  isl29020: Remove a redundant semi-colon from return statement
  BMP085: Remove redundant semi-colon from return statement
  drivers:misc: ti-st: DEBUG uart, baud rate mods
  drivers:misc: ti-st: flush UART upon fw failure
  drivers:misc: ti-st: protect registrations
  char_dev.c: fix up some whitespace errors
  s390: tape_class.h: remove kobj_map.h inclusion
  misc: ad525x_dpot: Add support for SPI module device table matching

12 years agoMerge branch 'samsung/cleanup' into next/boards
Arnd Bergmann [Mon, 9 Jan 2012 17:06:36 +0000 (17:06 +0000)]
Merge branch 'samsung/cleanup' into next/boards

Conflicts:
arch/arm/mach-imx/mach-imx6q.c
arch/arm/mach-omap2/board-ti8168evm.c
arch/arm/mach-s3c64xx/Kconfig
arch/arm/mach-tegra/Makefile
arch/arm/mach-tegra/board-dt-tegra20.c
arch/arm/mach-tegra/common.c

Lots of relatively simple conflicts between the board
changes and stuff from the arm tree. This pulls in
the resolution from the samsung/cleanup tree, so we
don't get conflicting merges.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Mon, 9 Jan 2012 16:31:22 +0000 (08:31 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  ext[34]: avoid i_nlink warnings triggered by drop_nlink/inc_nlink kludge in symlink()
  exofs: oops after late failure in mount
  devpts: fix double-free on mount failure
  ... and the same for gadgetfs
  functionfs: unfuck failure exits on mount

12 years agoMerge branch 'samsung/driver' into next/drivers
Arnd Bergmann [Mon, 9 Jan 2012 16:16:29 +0000 (16:16 +0000)]
Merge branch 'samsung/driver' into next/drivers

Conflicts:
arch/arm/mach-mxs/include/mach/common.h

Pull in previous samsung conflict merges and do a trivial
merge of an mxs double-add conflict.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
12 years agoMerge branch 'samsung/cleanup' into samsung/driver
Arnd Bergmann [Mon, 9 Jan 2012 16:14:07 +0000 (16:14 +0000)]
Merge branch 'samsung/cleanup' into samsung/driver

Conflicts:
arch/arm/mach-s5p64x0/cpu.c -> common.c

More changes to a file that got moved into common.c,
see previous conflict resolutions.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Mon, 9 Jan 2012 16:11:13 +0000 (08:11 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/s390/linux

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (31 commits)
  [S390] disassembler: mark exception causing instructions
  [S390] Enable exception traces by default
  [S390] return address of compat signals
  [S390] sysctl: get rid of dead declaration
  [S390] dasd: fix fixpoint divide exception in define_extent
  [S390] dasd: add sanity check to detect path connection error
  [S390] qdio: fix kernel panic for zfcp 31-bit
  [S390] Add s390x description to Documentation/kdump/kdump.txt
  [S390] Add VMCOREINFO_SYMBOL(high_memory) to vmcoreinfo
  [S390] dasd: fix expiration handling for recovery requests
  [S390] outstanding interrupts vs. smp_send_stop
  [S390] ipc: call generic sys_ipc demultiplexer
  [S390] zcrypt: Fix error return codes.
  [S390] zcrypt: Rework length parameter checking.
  [S390] cleanup trap handling
  [S390] Remove Kerntypes leftovers
  [S390] topology: increase poll frequency if change is anticipated
  [S390] entry[64].S improvements
  [S390] make arch/s390 subdirectories depend on config option
  [S390] kvm: move cmf host id constant out of lowcore
  ...

Fix up conflicts in arch/s390/kernel/{smp.c,topology.c} due to the
sysdev removal clashing with "topology: get rid of ifdefs" which moved
some of that code around.

12 years agoMerge branch 'samsung/cleanup' into next/cleanup2
Arnd Bergmann [Mon, 9 Jan 2012 16:06:31 +0000 (16:06 +0000)]
Merge branch 'samsung/cleanup' into next/cleanup2

12 years agoMerge branch 'samsung/dt' into samsung/cleanup
Arnd Bergmann [Mon, 9 Jan 2012 16:01:00 +0000 (16:01 +0000)]
Merge branch 'samsung/dt' into samsung/cleanup

Conflicts:
arch/arm/mach-s3c64xx/Makefile
arch/arm/mach-s5pc100/Makefile
arch/arm/mach-s5pv210/Makefile

Pull in previously resolved conflicts:

The Makefiles were reorganized in the "rmk/restart" series and modified
in the "samsung/cleanup series". This also pulls in the other conflict
resolutions from the restart series against the samsung/dt series.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
12 years agoisofs: inode leak on mount failure
Al Viro [Mon, 9 Jan 2012 15:48:11 +0000 (10:48 -0500)]
isofs: inode leak on mount failure

d_alloc_root() failure leaves root inode leaked...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agoext2/3/4: delete unneeded includes of module.h
Paul Gortmaker [Wed, 4 Jan 2012 20:59:47 +0000 (15:59 -0500)]
ext2/3/4: delete unneeded includes of module.h

Delete any instances of include module.h that were not strictly
required.  In the case of ext2, the declaration of MODULE_LICENSE
etc. were in inode.c but the module_init/exit were in super.c, so
relocate the MODULE_LICENCE/AUTHOR block to super.c which makes it
consistent with ext3 and ext4 at the same time.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agoext{3,4}: Fix potential race when setversion ioctl updates inode
Djalal Harouni [Tue, 3 Jan 2012 01:31:52 +0000 (02:31 +0100)]
ext{3,4}: Fix potential race when setversion ioctl updates inode

The EXT{3,4}_IOC_SETVERSION ioctl() updates i_ctime and i_generation
without i_mutex. This can lead to a race with the other operations that
update i_ctime. This is not a big issue but let's make the ioctl consistent
with how we handle e.g. other timestamp updates and use i_mutex to protect
inode changes.

Signed-off-by: Djalal Harouni <tixxdz@opendz.org>
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agoudf: Mark LVID buffer as uptodate before marking it dirty
Jan Kara [Fri, 23 Dec 2011 10:53:07 +0000 (11:53 +0100)]
udf: Mark LVID buffer as uptodate before marking it dirty

When we hit EIO while writing LVID, the buffer uptodate bit is cleared.
This then results in an anoying warning from mark_buffer_dirty() when we
write the buffer again. So just set uptodate flag unconditionally.

Reviewed-by: Namjae Jeon <linkinjeon@gmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agoext3: Don't warn from writepage when readonly inode is spotted after error
Jan Kara [Thu, 22 Dec 2011 15:49:05 +0000 (16:49 +0100)]
ext3: Don't warn from writepage when readonly inode is spotted after error

WARN_ON_ONCE(IS_RDONLY(inode)) tends to trip when filesystem hits error and is
remounted read-only. This unnecessarily scares users (well, they should be
scared because of filesystem error, but the stack trace distracts them from the
right source of their fear ;-). We could as well just remove the WARN_ON but
it's not hard to fix it to not trip on filesystem with errors and not use more
cycles in the common case so that's what we do.

CC: stable@kernel.org
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agojbd: Remove j_barrier mutex
Jan Kara [Thu, 22 Dec 2011 13:52:21 +0000 (14:52 +0100)]
jbd: Remove j_barrier mutex

j_barrier mutex is used for serializing different journal lock operations.  The
problem with it is that e.g. FIFREEZE ioctl results in process leaving kernel
with j_barrier mutex held which makes lockdep freak out. Also hibernation code
wants to freeze filesystem but it cannot do so because it then cannot hibernate
the system because of mutex being locked.

So we remove j_barrier mutex and use direct wait on j_barrier_count instead.
Since locking journal is a rare operation we don't have to care about fairness
or such things.

CC: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agoreiserfs: Force inode evictions before umount to avoid crash
Jeff Mahoney [Wed, 21 Dec 2011 20:18:43 +0000 (21:18 +0100)]
reiserfs: Force inode evictions before umount to avoid crash

This patch fixes a crash in reiserfs_delete_xattrs during umount.

When shrink_dcache_for_umount clears the dcache from
generic_shutdown_super, delayed evictions are forced to disk. If an
evicted inode has extended attributes associated with it, it will
need to walk the xattr tree to locate and remove them.

But since shrink_dcache_for_umount will BUG if it encounters active
dentries, the xattr tree must be released before it's called or it will
crash during every umount.

This patch forces the evictions to occur before generic_shutdown_super
by calling shrink_dcache_sb first. The additional evictions caused
by the removal of each associated xattr file and dir will be automatically
handled as they're added to the LRU list.

CC: reiserfs-devel@vger.kernel.org
CC: stable@kernel.org
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agoreiserfs: Fix quota mount option parsing
Jan Kara [Wed, 21 Dec 2011 16:35:34 +0000 (17:35 +0100)]
reiserfs: Fix quota mount option parsing

When jqfmt mount option is not specified on remount, we mistakenly clear
s_jquota_fmt value stored in superblock. Fix the problem.

CC: stable@kernel.org
CC: reiserfs-devel@vger.kernel.org
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agoudf: Treat symlink component of type 2 as /
Jan Kara [Mon, 12 Dec 2011 14:13:50 +0000 (15:13 +0100)]
udf: Treat symlink component of type 2 as /

Currently, we ignore symlink component of type 2. But mkisofs and other OS'
seem to treat it as / so do the same for compatibility.

Reported-by: "Gábor S." <otnaccess@hotmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agoudf: Fix deadlock when converting file from in-ICB one to normal one
Jan Kara [Sat, 10 Dec 2011 01:30:48 +0000 (02:30 +0100)]
udf: Fix deadlock when converting file from in-ICB one to normal one

During BKL removal in 2.6.38, conversion of files from in-ICB format to normal
format got broken. We call ->writepage with i_data_sem held but udf_get_block()
also acquires i_data_sem thus creating A-A deadlock.

We fix the problem by dropping i_data_sem before calling ->writepage() which is
safe since i_mutex still protects us against any changes in the file. Also fix
pagelock - i_data_sem lock inversion in udf_expand_file_adinicb() by dropping
i_data_sem before calling find_or_create_page().

CC: stable@kernel.org
Reported-by: Matthias Matiak <netzpython@mail-on.us>
Tested-by: Matthias Matiak <netzpython@mail-on.us>
Reviewed-by: Namjae Jeon <linkinjeon@gmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agoudf: Cleanup calling convention of inode_getblk()
Jan Kara [Sat, 10 Dec 2011 00:43:33 +0000 (01:43 +0100)]
udf: Cleanup calling convention of inode_getblk()

inode_getblk() always returned NULL and passed results in its parameters.
Make the function return something useful - found block number.

Signed-off-by: Jan Kara <jack@suse.cz>
12 years agoext2: Fix error handling on inode bitmap corruption
Jan Kara [Thu, 8 Dec 2011 23:08:58 +0000 (00:08 +0100)]
ext2: Fix error handling on inode bitmap corruption

When insert_inode_locked() fails in ext2_new_inode() it most likely means inode
bitmap got corrupted and we allocated again inode which is already in use. Also
doing unlock_new_inode() during error recovery is wrong since the inode does
not have I_NEW set. Fix the problem by informing about filesystem error and
jumping to fail: (instead of fail_drop:) which doesn't call unlock_new_inode().

Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agoext3: Fix error handling on inode bitmap corruption
Jan Kara [Thu, 8 Dec 2011 20:13:46 +0000 (21:13 +0100)]
ext3: Fix error handling on inode bitmap corruption

When insert_inode_locked() fails in ext3_new_inode() it most likely
means inode bitmap got corrupted and we allocated again inode which
is already in use. Also doing unlock_new_inode() during error recovery
is wrong since inode does not have I_NEW set. Fix the problem by jumping
to fail: (instead of fail_drop:) which declares filesystem error and
does not call unlock_new_inode().

Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agoext3: replace ll_rw_block with other functions
Zheng Liu [Mon, 5 Dec 2011 07:55:11 +0000 (15:55 +0800)]
ext3: replace ll_rw_block with other functions

ll_rw_block() is deprecated. Thus we replace it with other functions.

CC: Jan Kara <jack@suse.cz>
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com>
Signed-off-by: Jan Kara <jack@suse.cz>
12 years agonet: sk_update_clone is only used in net/core/sock.c
Stephen Rothwell [Mon, 9 Jan 2012 05:33:16 +0000 (16:33 +1100)]
net: sk_update_clone is only used in net/core/sock.c

so move it there.  Fixes build errors when CONFIG_INET is not defined:

In file included from include/linux/tcp.h:211:0,
                 from include/linux/ipv6.h:221,
                 from include/net/ipv6.h:16,
                 from include/linux/sunrpc/clnt.h:26,
                 from include/linux/nfs_fs.h:50,
                 from init/do_mounts.c:20:
include/net/sock.h: In function 'sk_update_clone':
include/net/sock.h:1109:3: error: implicit declaration of function 'sock_update_memcg' [-Werror=implicit-function-declaration]

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years ago8139cp: fix missing napi_gro_flush.
françois romieu [Sun, 8 Jan 2012 13:41:33 +0000 (13:41 +0000)]
8139cp: fix missing napi_gro_flush.

The driver uses __napi_complete and napi_gro_receive. Without it, the
driver hits the BUG_ON(n->gro_list) assertion hard in __napi_complete.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Tested-by: Marin Glibic <zhilla2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoext[34]: avoid i_nlink warnings triggered by drop_nlink/inc_nlink kludge in symlink()
Al Viro [Mon, 9 Jan 2012 00:50:23 +0000 (19:50 -0500)]
ext[34]: avoid i_nlink warnings triggered by drop_nlink/inc_nlink kludge in symlink()

Both ext3 and ext4 put the half-created symlink inode into the orphan list
for a while (see the comment in ext[34]_symlink() for gory details).  Then,
if everything went fine, they pull it out of the orphan list and bump the
link count back to 1.  The thing is, inc_nlink() is going to complain about
seeing somebody changing i_nlink from 0 to 1.  With a good reason, since
normally something like that is a bug.  Explicit set_nlink(inode, 1) does
the same thing as inc_nlink() here, but it does *not* complain - exactly
because it should be usable in strange situations like this one.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agoexofs: oops after late failure in mount
Al Viro [Mon, 9 Jan 2012 00:45:28 +0000 (19:45 -0500)]
exofs: oops after late failure in mount

We have already set ->s_root, so ->put_super() is going to be called.
Freeing ->s_fs_info is a bloody bad idea when it's going to be
dereferenced very shortly...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agodevpts: fix double-free on mount failure
Al Viro [Mon, 9 Jan 2012 00:40:27 +0000 (19:40 -0500)]
devpts: fix double-free on mount failure

devpts_kill_sb() is called even if devpts_fill_super() fails;
we should not do that kfree() in the latter, especially not
with ->s_fs_info left pointing to freed object.  Double kfree()
is a Bad Thing(tm)...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years ago... and the same for gadgetfs
Al Viro [Sun, 8 Jan 2012 20:59:45 +0000 (15:59 -0500)]
... and the same for gadgetfs

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
12 years agofunctionfs: unfuck failure exits on mount
Al Viro [Sun, 8 Jan 2012 20:38:27 +0000 (15:38 -0500)]
functionfs: unfuck failure exits on mount

* if you do dput() of root dentry, do *not* follow that with iput() of root
inode.
* while we are at it, don't do that dput() at all - you are leaving the pointer
in ->s_root and your ->kill_sb() will be very unhappy with that.  It will do
proper dput(), though, so the easiest way is to leave that to it entirely.
* freeing ->s_fs_info is also best left to ->kill_sb() (which will do it
anyway), especially since we leave the pointer in place.
* that xchg() in ->kill_sb() is not a bug per se, but it's a plain and simple
masturbation with fewer excuses than Onan had...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>