linux-toradex.git/mm/page_alloc.c, branch v3.4.39

mm: fix pageblock bitmap allocation

2013-02-28T14:58:58+00:00

commit 7c45512df987c5619db041b5c9b80d281e26d3db upstream.

Commit c060f943d092 ("mm: use aligned zone start for pfn_to_bitidx
calculation") fixed out calculation of the index into the pageblock
bitmap when a !SPARSEMEM zome was not aligned to pageblock_nr_pages.

However, the _allocation_ of that bitmap had never taken this alignment
requirement into accout, so depending on the exact size and alignment of
the zone, the use of that index could then access past the allocation,
resulting in some very subtle memory corruption.

This was reported (and bisected) by Ingo Molnar: one of his random
config builds would hang with certain very specific kernel command line
options.

In the meantime, commit c060f943d092 has been marked for stable, so this
fix needs to be back-ported to the stable kernels that backported the
commit to use the right alignment.

Bisected-and-tested-by: Ingo Molnar 
Acked-by: Mel Gorman 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

mm: use aligned zone start for pfn_to_bitidx calculation

2013-01-17T16:50:43+00:00

commit c060f943d0929f3e429c5d9522290584f6281d6e upstream.

The current calculation in pfn_to_bitidx assumes that (pfn -
zone->zone_start_pfn) >> pageblock_order will return the same bit for
all pfn in a pageblock.  If zone_start_pfn is not aligned to
pageblock_nr_pages, this may not always be correct.

Consider the following with pageblock order = 10, zone start 2MB:

  pfn     | pfn - zone start | (pfn - zone start) >> page block order
  ----------------------------------------------------------------
  0x26000 | 0x25e00	   |  0x97
  0x26100 | 0x25f00	   |  0x97
  0x26200 | 0x26000	   |  0x98
  0x26300 | 0x26100	   |  0x98

This means that calling {get,set}_pageblock_migratetype on a single page
will not set the migratetype for the full block.  Fix this by rounding
down zone_start_pfn when doing the bitidx calculation.

For our use case, the effects of this bug were mostly tied to the fact
that CMA allocations would either take a long time or fail to happen.
Depending on the driver using CMA, this could result in anything from
visual glitches to application failures.

Signed-off-by: Laura Abbott 
Acked-by: Mel Gorman 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

mm/page_alloc: fix the page address of higher page's buddy calculation

2012-10-02T17:30:05+00:00

commit 0ba8f2d59304dfe69b59c034de723ad80f7ab9ac upstream.

The heuristic method for buddy has been introduced since commit
43506fad21ca ("mm/page_alloc.c: simplify calculation of combined index
of adjacent buddy lists").  But the page address of higher page's buddy
was wrongly calculated, which will lead page_is_buddy to fail for ever.
IOW, the heuristic method would be disabled with the wrong page address
of higher page's buddy.

Calculating the page address of higher page's buddy should be based
higher_page with the offset between index of higher page and index of
higher page's buddy.

Signed-off-by: Haifeng Li 
Signed-off-by: Gavin Shan 
Reviewed-by: Michal Hocko 
Cc: KyongHo Cho 
Cc: Mel Gorman 
Cc: Minchan Kim 
Cc: Johannes Weiner 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

mm: raise MemFree by reverting percpu_pagelist_fraction to 0

2012-05-11T16:23:39+00:00

Why is there less MemFree than there used to be?  It perturbed a test,
so I've just been bisecting linux-next, and now find the offender went
upstream yesterday.

Commit 93278814d359 "mm: fix division by 0 in percpu_pagelist_fraction()"
mistakenly initialized percpu_pagelist_fraction to the sysctl's minimum 8,
which leaves 1/8th of memory on percpu lists (on each cpu??); but most of
us expect it to be left unset at 0 (and it's not then used as a divisor).

  MemTotal: 8061476kB  8061476kB  8061476kB  8061476kB  8061476kB  8061476kB
  Repetitive test with percpu_pagelist_fraction 8:
  MemFree:  6948420kB  6237172kB  6949696kB  6840692kB  6949048kB  6862984kB
  Same test with percpu_pagelist_fraction back to 0:
  MemFree:  7945000kB  7944908kB  7948568kB  7949060kB  7948796kB  7948812kB

Signed-off-by: Hugh Dickins 
[ We really should fix the crazy sysctl interface too, but that's a
  separate thing - Linus ]
Signed-off-by: Linus Torvalds

mm: fix division by 0 in percpu_pagelist_fraction()

2012-05-10T22:06:44+00:00

percpu_pagelist_fraction_sysctl_handler() has only considered -EINVAL as
a possible error from proc_dointvec_minmax().

If any other error is returned, it would proceed to divide by zero since
percpu_pagelist_fraction wasn't getting initialized at any point.  For
example, writing 0 bytes into the proc file would trigger the issue.

Signed-off-by: Sasha Levin 
Reviewed-by: Minchan Kim 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm: only IPI CPUs to drain local pages if they exist

2012-03-29T00:14:35+00:00

Calculate a cpumask of CPUs with per-cpu pages in any zone and only send
an IPI requesting CPUs to drain these pages to the buddy allocator if they
actually have pages when asked to flush.

This patch saves 85%+ of IPIs asking to drain per-cpu pages in case of
severe memory pressure that leads to OOM since in these cases multiple,
possibly concurrent, allocation requests end up in the direct reclaim code
path so when the per-cpu pages end up reclaimed on first allocation
failure for most of the proceeding allocation attempts until the memory
pressure is off (possibly via the OOM killer) there are no per-cpu pages
on most CPUs (and there can easily be hundreds of them).

This also has the side effect of shortening the average latency of direct
reclaim by 1 or more order of magnitude since waiting for all the CPUs to
ACK the IPI takes a long time.

Tested by running "hackbench 400" on a 8 CPU x86 VM and observing the
difference between the number of direct reclaim attempts that end up in
drain_all_pages() and those were more then 1/2 of the online CPU had any
per-cpu page in them, using the vmstat counters introduced in the next
patch in the series and using proc/interrupts.

In the test sceanrio, this was seen to save around 3600 global
IPIs after trigerring an OOM on a concurrent workload:

$ cat /proc/vmstat | tail -n 2
pcp_global_drain 0
pcp_global_ipi_saved 0

$ cat /proc/interrupts | grep CAL
CAL:          1          2          1          2
          2          2          2          2   Function call interrupts

$ hackbench 400
[OOM messages snipped]

$ cat /proc/vmstat | tail -n 2
pcp_global_drain 3647
pcp_global_ipi_saved 3642

$ cat /proc/interrupts | grep CAL
CAL:          6         13          6          3
          3          3         1 2          7   Function call interrupts

Please note that if the global drain is removed from the direct reclaim
path as a patch from Mel Gorman currently suggests this should be replaced
with an on_each_cpu_cond invocation.

Signed-off-by: Gilad Ben-Yossef 
Acked-by: Mel Gorman 
Cc: KOSAKI Motohiro 
Acked-by: Christoph Lameter 
Acked-by: Peter Zijlstra 
Cc: Pekka Enberg 
Cc: Rik van Riel 
Cc: Andi Kleen 
Acked-by: Michal Nazarewicz 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm, coredump: fail allocations when coredumping instead of oom killing

2012-03-29T00:14:35+00:00

The size of coredump files is limited by RLIMIT_CORE, however, allocating
large amounts of memory results in three negative consequences:

 - the coredumping process may be chosen for oom kill and quickly deplete
   all memory reserves in oom conditions preventing further progress from
   being made or tasks from exiting,

 - the coredumping process may cause other processes to be oom killed
   without fault of their own as the result of a SIGSEGV, for example, in
   the coredumping process, or

 - the coredumping process may result in a livelock while writing to the
   dump file if it needs memory to allocate while other threads are in
   the exit path waiting on the coredumper to complete.

This is fixed by implying __GFP_NORETRY in the page allocator for
coredumping processes when reclaim has failed so the allocations fail and
the process continues to exit.

Signed-off-by: David Rientjes 
Cc: Mel Gorman 
Cc: KAMEZAWA Hiroyuki 
Cc: Minchan Kim 
Cc: Oleg Nesterov 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

page_alloc: remove unused find_zone_movable_pfns_for_nodes() argument

2012-03-22T00:55:00+00:00

find_zone_movable_pfns_for_nodes() does not use its argument.

Signed-off-by: Kautuk Consul 
Cc: David Rientjes 
Cc: Mel Gorman 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

page_alloc.c: remove add_from_early_node_map()

2012-03-22T00:55:00+00:00

add_from_early_node_map() is unused.

Signed-off-by: Kautuk Consul 
Acked-by: David Rientjes 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

cpuset: mm: reduce large amounts of memory barrier related damage v3

2012-03-22T00:54:59+00:00

Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.

[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths.  This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32.  The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.

For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.

This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side.  This is much cheaper on some architectures, including x86.  The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.

While updating the nodemask, a check is made to see if a false failure
is a risk.  If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.

In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%.  The
actual results were

                             3.3.0-rc3          3.3.0-rc3
                             rc3-vanilla        nobarrier-v2r1
    Clients   1 UserTime       0.07 (  0.00%)   0.08 (-14.19%)
    Clients   2 UserTime       0.07 (  0.00%)   0.07 (  2.72%)
    Clients   4 UserTime       0.08 (  0.00%)   0.07 (  3.29%)
    Clients   1 SysTime        0.70 (  0.00%)   0.65 (  6.65%)
    Clients   2 SysTime        0.85 (  0.00%)   0.82 (  3.65%)
    Clients   4 SysTime        1.41 (  0.00%)   1.41 (  0.32%)
    Clients   1 WallTime       0.77 (  0.00%)   0.74 (  4.19%)
    Clients   2 WallTime       0.47 (  0.00%)   0.45 (  3.73%)
    Clients   4 WallTime       0.38 (  0.00%)   0.37 (  1.58%)
    Clients   1 Flt/sec/cpu  497620.28 (  0.00%) 520294.53 (  4.56%)
    Clients   2 Flt/sec/cpu  414639.05 (  0.00%) 429882.01 (  3.68%)
    Clients   4 Flt/sec/cpu  257959.16 (  0.00%) 258761.48 (  0.31%)
    Clients   1 Flt/sec      495161.39 (  0.00%) 517292.87 (  4.47%)
    Clients   2 Flt/sec      820325.95 (  0.00%) 850289.77 (  3.65%)
    Clients   4 Flt/sec      1020068.93 (  0.00%) 1022674.06 (  0.26%)
    MMTests Statistics: duration
    Sys Time Running Test (seconds)             135.68    132.17
    User+Sys Time Running Test (seconds)         164.2    160.13
    Total Elapsed Time (seconds)                123.46    120.87

The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected).  The
actual number of page faults is noticeably improved.

For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.

To test the actual bug the commit fixed I opened two terminals.  The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data.  In a second window, the nodemask of the
cpuset was continually randomised in a loop.

Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.

Signed-off-by: Mel Gorman 
Cc: Miao Xie 
Cc: David Rientjes 
Cc: Peter Zijlstra 
Cc: Christoph Lameter 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds