linux.git/mm/vmstat.c, branch v4.5-rc2

vmstat: Remove BUG_ON from vmstat_update

2016-01-24T16:55:52+00:00

If we detect that there is nothing to do just set the flag and do not
check if it was already set before.  Races really do not matter.  If the
flag is set by any code then the shepherd will start dealing with the
situation and reenable the vmstat workers when necessary again.

Since commit 0eb77e988032 ("vmstat: make vmstat_updater deferrable again
and shut down on idle") quiet_vmstat might update cpu_stat_off and mark
a particular cpu to be handled by vmstat_shepherd.  This might trigger a
VM_BUG_ON in vmstat_update because the work item might have been
sleeping during the idle period and see the cpu_stat_off updated after
the wake up.  The VM_BUG_ON is therefore misleading and no more
appropriate.  Moreover it doesn't really suite any protection from real
bugs because vmstat_shepherd will simply reschedule the vmstat_work
anytime it sees a particular cpu set or vmstat_update would do the same
from the worker context directly.  Even when the two would race the
result wouldn't be incorrect as the counters update is fully idempotent.

Reported-by: Sasha Levin 
Signed-off-by: Christoph Lameter 
Acked-by: Michal Hocko 
Cc: Johannes Weiner 
Cc: Tetsuo Handa 
Cc: Andrew Morton 
Signed-off-by: Linus Torvalds

mm: support madvise(MADV_FREE)

2016-01-16T01:56:32+00:00

Linux doesn't have an ability to free pages lazy while other OS already
have been supported that named by madvise(MADV_FREE).

The gain is clear that kernel can discard freed pages rather than
swapping out or OOM if memory pressure happens.

Without memory pressure, freed pages would be reused by userspace
without another additional overhead(ex, page fault + allocation +
zeroing).

Jason Evans said:

: Facebook has been using MAP_UNINITIALIZED
: (https://lkml.org/lkml/2012/1/18/308) in some of its applications for
: several years, but there are operational costs to maintaining this
: out-of-tree in our kernel and in jemalloc, and we are anxious to retire it
: in favor of MADV_FREE.  When we first enabled MAP_UNINITIALIZED it
: increased throughput for much of our workload by ~5%, and although the
: benefit has decreased using newer hardware and kernels, there is still
: enough benefit that we cannot reasonably retire it without a replacement.
:
: Aside from Facebook operations, there are numerous broadly used
: applications that would benefit from MADV_FREE.  The ones that immediately
: come to mind are redis, varnish, and MariaDB.  I don't have much insight
: into Android internals and development process, but I would hope to see
: MADV_FREE support eventually end up there as well to benefit applications
: linked with the integrated jemalloc.
:
: jemalloc will use MADV_FREE once it becomes available in the Linux kernel.
: In fact, jemalloc already uses MADV_FREE or equivalent everywhere it's
: available: *BSD, OS X, Windows, and Solaris -- every platform except Linux
: (and AIX, but I'm not sure it even compiles on AIX).  The lack of
: MADV_FREE on Linux forced me down a long series of increasingly
: sophisticated heuristics for madvise() volume reduction, and even so this
: remains a common performance issue for people using jemalloc on Linux.
: Please integrate MADV_FREE; many people will benefit substantially.

How it works:

When madvise syscall is called, VM clears dirty bit of ptes of the
range.  If memory pressure happens, VM checks dirty bit of page table
and if it found still "clean", it means it's a "lazyfree pages" so VM
could discard the page instead of swapping out.  Once there was store
operation for the page before VM peek a page to reclaim, dirty bit is
set so VM can swap out the page instead of discarding.

One thing we should notice is that basically, MADV_FREE relies on dirty
bit in page table entry to decide whether VM allows to discard the page
or not.  IOW, if page table entry includes marked dirty bit, VM
shouldn't discard the page.

However, as a example, if swap-in by read fault happens, page table
entry doesn't have dirty bit so MADV_FREE could discard the page
wrongly.

For avoiding the problem, MADV_FREE did more checks with PageDirty and
PageSwapCache.  It worked out because swapped-in page lives on swap
cache and since it is evicted from the swap cache, the page has PG_dirty
flag.  So both page flags check effectively prevent wrong discarding by
MADV_FREE.

However, a problem in above logic is that swapped-in page has PG_dirty
still after they are removed from swap cache so VM cannot consider the
page as freeable any more even if madvise_free is called in future.

Look at below example for detail.

    ptr = malloc();
    memset(ptr);
    ..
    ..
    .. heavy memory pressure so all of pages are swapped out
    ..
    ..
    var = *ptr; -> a page swapped-in and could be removed from
                   swapcache. Then, page table doesn't mark
                   dirty bit and page descriptor includes PG_dirty
    ..
    ..
    madvise_free(ptr); -> It doesn't clear PG_dirty of the page.
    ..
    ..
    ..
    .. heavy memory pressure again.
    .. In this time, VM cannot discard the page because the page
    .. has *PG_dirty*

To solve the problem, this patch clears PG_dirty if only the page is
owned exclusively by current process when madvise is called because
PG_dirty represents ptes's dirtiness in several processes so we could
clear it only if we own it exclusively.

Firstly, heavy users would be general allocators(ex, jemalloc, tcmalloc
and hope glibc supports it) and jemalloc/tcmalloc already have supported
the feature for other OS(ex, FreeBSD)

  barrios@blaptop:~/benchmark/ebizzy$ lscpu
  Architecture:          x86_64
  CPU op-mode(s):        32-bit, 64-bit
  Byte Order:            Little Endian
  CPU(s):                12
  On-line CPU(s) list:   0-11
  Thread(s) per core:    1
  Core(s) per socket:    1
  Socket(s):             12
  NUMA node(s):          1
  Vendor ID:             GenuineIntel
  CPU family:            6
  Model:                 2
  Stepping:              3
  CPU MHz:               3200.185
  BogoMIPS:              6400.53
  Virtualization:        VT-x
  Hypervisor vendor:     KVM
  Virtualization type:   full
  L1d cache:             32K
  L1i cache:             32K
  L2 cache:              4096K
  NUMA node0 CPU(s):     0-11
  ebizzy benchmark(./ebizzy -S 10 -n 512)

  Higher avg is better.

   vanilla-jemalloc             MADV_free-jemalloc

  1 thread
  records: 10                   records: 10
  avg:   2961.90                avg:  12069.70
  std:     71.96(2.43%)         std:    186.68(1.55%)
  max:   3070.00                max:  12385.00
  min:   2796.00                min:  11746.00

  2 thread
  records: 10                   records: 10
  avg:   5020.00                avg:  17827.00
  std:    264.87(5.28%)         std:    358.52(2.01%)
  max:   5244.00                max:  18760.00
  min:   4251.00                min:  17382.00

  4 thread
  records: 10                   records: 10
  avg:   8988.80                avg:  27930.80
  std:   1175.33(13.08%)        std:   3317.33(11.88%)
  max:   9508.00                max:  30879.00
  min:   5477.00                min:  21024.00

  8 thread
  records: 10                   records: 10
  avg:  13036.50                avg:  33739.40
  std:    170.67(1.31%)         std:   5146.22(15.25%)
  max:  13371.00                max:  40572.00
  min:  12785.00                min:  24088.00

  16 thread
  records: 10                   records: 10
  avg:  11092.40                avg:  31424.20
  std:    710.60(6.41%)         std:   3763.89(11.98%)
  max:  12446.00                max:  36635.00
  min:   9949.00                min:  25669.00

  32 thread
  records: 10                   records: 10
  avg:  11067.00                avg:  34495.80
  std:    971.06(8.77%)         std:   2721.36(7.89%)
  max:  12010.00                max:  38598.00
  min:   9002.00                min:  30636.00

In summary, MADV_FREE is about much faster than MADV_DONTNEED.

This patch (of 12):

Add core MADV_FREE implementation.

[akpm@linux-foundation.org: small cleanups]
Signed-off-by: Minchan Kim 
Acked-by: Michal Hocko 
Acked-by: Hugh Dickins 
Cc: Mika Penttil 
Cc: Michael Kerrisk 
Cc: Johannes Weiner 
Cc: Rik van Riel 
Cc: Mel Gorman 
Cc: KOSAKI Motohiro 
Cc: Jason Evans 
Cc: Daniel Micay 
Cc: "Kirill A. Shutemov" 
Cc: Shaohua Li 
Cc: 
Cc: Andy Lutomirski 
Cc: "James E.J. Bottomley" 
Cc: "Kirill A. Shutemov" 
Cc: "Shaohua Li" 
Cc: Andrea Arcangeli 
Cc: Arnd Bergmann 
Cc: Benjamin Herrenschmidt 
Cc: Catalin Marinas 
Cc: Chen Gang 
Cc: Chris Zankel 
Cc: Darrick J. Wong 
Cc: David S. Miller 
Cc: Helge Deller 
Cc: Ivan Kokshaysky 
Cc: Matt Turner 
Cc: Max Filippov 
Cc: Ralf Baechle 
Cc: Richard Henderson 
Cc: Roland Dreier 
Cc: Russell King 
Cc: Shaohua Li 
Cc: Will Deacon 
Cc: Wu Fengguang 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm, vmstats: new THP splitting event

2016-01-16T01:56:32+00:00

The patch replaces THP_SPLIT with tree events: THP_SPLIT_PAGE,
THP_SPLIT_PAGE_FAILED and THP_SPLIT_PMD.  It reflects the fact that we
are going to be able split PMD without the compound page and that
split_huge_page() can fail.

Signed-off-by: Kirill A.  Shutemov 
Acked-by: Christoph Lameter 
Tested-by: Sasha Levin 
Tested-by: Aneesh Kumar K.V 
Acked-by: Jerome Marchand 
Acked-by: Vlastimil Babka 
Cc: Andrea Arcangeli 
Cc: Hugh Dickins 
Cc: Dave Hansen 
Cc: Mel Gorman 
Cc: Rik van Riel 
Cc: Naoya Horiguchi 
Cc: Steve Capper 
Cc: Johannes Weiner 
Cc: Michal Hocko 
Cc: David Rientjes 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

vmstat: make vmstat_updater deferrable again and shut down on idle

2016-01-15T00:00:49+00:00

Currently the vmstat updater is not deferrable as a result of commit
ba4877b9ca51 ("vmstat: do not use deferrable delayed work for
vmstat_update").  This in turn can cause multiple interruptions of the
applications because the vmstat updater may run at

Make vmstate_update deferrable again and provide a function that folds
the differentials when the processor is going to idle mode thus
addressing the issue of the above commit in a clean way.

Note that the shepherd thread will continue scanning the differentials
from another processor and will reenable the vmstat workers if it
detects any changes.

Fixes: ba4877b9ca51 ("vmstat: do not use deferrable delayed work for vmstat_update")
Signed-off-by: Christoph Lameter 
Cc: Michal Hocko 
Cc: Johannes Weiner 
Cc: Tetsuo Handa 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

vmstat: allocate vmstat_wq before it is used

2016-01-09T07:47:54+00:00

kernel test robot has reported the following crash:

  BUG: unable to handle kernel NULL pointer dereference at 00000100
  IP: [] __queue_work+0x26/0x390
  *pdpt = 0000000000000000 *pde = f000ff53f000ff53 *pde = f000ff53f000ff53
  Oops: 0000 [#1] PREEMPT PREEMPT SMP SMP
  CPU: 0 PID: 24 Comm: kworker/0:1 Not tainted 4.4.0-rc4-00139-g373ccbe #1
  Workqueue: events vmstat_shepherd
  task: cb684600 ti: cb7ba000 task.ti: cb7ba000
  EIP: 0060:[] EFLAGS: 00010046 CPU: 0
  EIP is at __queue_work+0x26/0x390
  EAX: 00000046 EBX: cbb37800 ECX: cbb37800 EDX: 00000000
  ESI: 00000000 EDI: 00000000 EBP: cb7bbe68 ESP: cb7bbe38
   DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
  CR0: 8005003b CR2: 00000100 CR3: 01fd5000 CR4: 000006b0
  Stack:
  Call Trace:
    __queue_delayed_work+0xa1/0x160
    queue_delayed_work_on+0x36/0x60
    vmstat_shepherd+0xad/0xf0
    process_one_work+0x1aa/0x4c0
    worker_thread+0x41/0x440
    kthread+0xb0/0xd0
    ret_from_kernel_thread+0x21/0x40

The reason is that start_shepherd_timer schedules the shepherd work item
which uses vmstat_wq (vmstat_shepherd) before setup_vmstat allocates
that workqueue so if the further initialization takes more than HZ we
might end up scheduling on a NULL vmstat_wq.  This is really unlikely
but not impossible.

Fixes: 373ccbe59270 ("mm, vmstat: allow WQ concurrency to discover memory reclaim doesn't make any progress")
Reported-by: kernel test robot 
Signed-off-by: Michal Hocko 
Tested-by: Tetsuo Handa 
Cc: stable@vger.kernel.org
Cc: Andrew Morton 
Signed-off-by: Linus Torvalds

mm/vmstat: fix overflow in mod_zone_page_state()

2015-12-30T01:45:49+00:00

mod_zone_page_state() takes a "delta" integer argument.  delta contains
the number of pages that should be added or subtracted from a struct
zone's vm_stat field.

If a zone is larger than 8TB this will cause overflows.  E.g.  for a
zone with a size slightly larger than 8TB the line

    mod_zone_page_state(zone, NR_ALLOC_BATCH, zone->managed_pages);

in mm/page_alloc.c:free_area_init_core() will result in a negative
result for the NR_ALLOC_BATCH entry within the zone's vm_stat, since 8TB
contain 0x8xxxxxxx pages which will be sign extended to a negative
value.

Fix this by changing the delta argument to long type.

This could fix an early boot problem seen on s390, where we have a 9TB
system with only one node.  ZONE_DMA contains 2GB and ZONE_NORMAL the
rest.  The system is trying to allocate a GFP_DMA page but ZONE_DMA is
completely empty, so it tries to reclaim pages in an endless loop.

This was seen on a heavily patched 3.10 kernel.  One possible
explaination seem to be the overflows caused by mod_zone_page_state().
Unfortunately I did not have the chance to verify that this patch
actually fixes the problem, since I don't have access to the system
right now.  However the overflow problem does exist anyway.

Given the description that a system with slightly less than 8TB does
work, this seems to be a candidate for the observed problem.

Signed-off-by: Heiko Carstens 
Cc: Christoph Lameter 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm, vmstat: allow WQ concurrency to discover memory reclaim doesn't make any progress

2015-12-12T18:15:34+00:00

Tetsuo Handa has reported that the system might basically livelock in
OOM condition without triggering the OOM killer.

The issue is caused by internal dependency of the direct reclaim on
vmstat counter updates (via zone_reclaimable) which are performed from
the workqueue context.  If all the current workers get assigned to an
allocation request, though, they will be looping inside the allocator
trying to reclaim memory but zone_reclaimable can see stalled numbers so
it will consider a zone reclaimable even though it has been scanned way
too much.  WQ concurrency logic will not consider this situation as a
congested workqueue because it relies that worker would have to sleep in
such a situation.  This also means that it doesn't try to spawn new
workers or invoke the rescuer thread if the one is assigned to the
queue.

In order to fix this issue we need to do two things.  First we have to
let wq concurrency code know that we are in trouble so we have to do a
short sleep.  In order to prevent from issues handled by 0e093d99763e
("writeback: do not sleep on the congestion queue if there are no
congested BDIs or if significant congestion is not being encountered in
the current zone") we limit the sleep only to worker threads which are
the ones of the interest anyway.

The second thing to do is to create a dedicated workqueue for vmstat and
mark it WQ_MEM_RECLAIM to note it participates in the reclaim and to
have a spare worker thread for it.

Signed-off-by: Michal Hocko 
Reported-by: Tetsuo Handa 
Cc: Tejun Heo 
Cc: Cristopher Lameter 
Cc: Joonsoo Kim 
Cc: Arkadiusz Miskiewicz 
Cc: 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm: fix swapped Movable and Reclaimable in /proc/pagetypeinfo

2015-12-12T18:15:34+00:00

Commit 016c13daa5c9 ("mm, page_alloc: use masks and shifts when
converting GFP flags to migrate types") has swapped MIGRATE_MOVABLE and
MIGRATE_RECLAIMABLE in the enum definition.  However, migratetype_names
wasn't updated to reflect that.

As a result, the file /proc/pagetypeinfo shows the counts for Movable as
Reclaimable and vice versa.

Additionally, commit 0aaa29a56e4f ("mm, page_alloc: reserve pageblocks
for high-order atomic allocations on demand") introduced
MIGRATE_HIGHATOMIC, but did not add a letter to distinguish it into
show_migration_types(), so it doesn't appear in the listing of free
areas during page alloc failures or oom kills.

This patch fixes both problems.  The atomic reserves will show with a
letter 'H' in the free areas listings.

Fixes: 016c13daa5c9 ("mm, page_alloc: use masks and shifts when converting GFP flags to migrate types")
Fixes: 0aaa29a56e4f ("mm, page_alloc: reserve pageblocks for high-order atomic allocations on demand")
Signed-off-by: Vlastimil Babka 
Acked-by: Mel Gorman 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm, page_alloc: reserve pageblocks for high-order atomic allocations on demand

2015-11-07T01:50:42+00:00

High-order watermark checking exists for two reasons -- kswapd high-order
awareness and protection for high-order atomic requests.  Historically the
kernel depended on MIGRATE_RESERVE to preserve min_free_kbytes as
high-order free pages for as long as possible.  This patch introduces
MIGRATE_HIGHATOMIC that reserves pageblocks for high-order atomic
allocations on demand and avoids using those blocks for order-0
allocations.  This is more flexible and reliable than MIGRATE_RESERVE was.

A MIGRATE_HIGHORDER pageblock is created when an atomic high-order
allocation request steals a pageblock but limits the total number to 1% of
the zone.  Callers that speculatively abuse atomic allocations for
long-lived high-order allocations to access the reserve will quickly fail.
 Note that SLUB is currently not such an abuser as it reclaims at least
once.  It is possible that the pageblock stolen has few suitable
high-order pages and will need to steal again in the near future but there
would need to be strong justification to search all pageblocks for an
ideal candidate.

The pageblocks are unreserved if an allocation fails after a direct
reclaim attempt.

The watermark checks account for the reserved pageblocks when the
allocation request is not a high-order atomic allocation.

The reserved pageblocks can not be used for order-0 allocations.  This may
allow temporary wastage until a failed reclaim reassigns the pageblock.
This is deliberate as the intent of the reservation is to satisfy a
limited number of atomic high-order short-lived requests if the system
requires them.

The stutter benchmark was used to evaluate this but while it was running
there was a systemtap script that randomly allocated between 1 high-order
page and 12.5% of memory's worth of order-3 pages using GFP_ATOMIC.  This
is much larger than the potential reserve and it does not attempt to be
realistic.  It is intended to stress random high-order allocations from an
unknown source, show that there is a reduction in failures without
introducing an anomaly where atomic allocations are more reliable than
regular allocations.  The amount of memory reserved varied throughout the
workload as reserves were created and reclaimed under memory pressure.
The allocation failures once the workload warmed up were as follows;

4.2-rc5-vanilla		70%
4.2-rc5-atomic-reserve	56%

The failure rate was also measured while building multiple kernels.  The
failure rate was 14% but is 6% with this patch applied.

Overall, this is a small reduction but the reserves are small relative to
the number of allocation requests.  In early versions of the patch, the
failure rate reduced by a much larger amount but that required much larger
reserves and perversely made atomic allocations seem more reliable than
regular allocations.

[yalin.wang2010@gmail.com: fix redundant check and a memory leak]
Signed-off-by: Mel Gorman 
Acked-by: Vlastimil Babka 
Acked-by: Michal Hocko 
Acked-by: Johannes Weiner 
Cc: Christoph Lameter 
Cc: David Rientjes 
Cc: Vitaly Wool 
Cc: Rik van Riel 
Signed-off-by: yalin wang 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm, page_alloc: remove MIGRATE_RESERVE

2015-11-07T01:50:42+00:00

MIGRATE_RESERVE preserves an old property of the buddy allocator that
existed prior to fragmentation avoidance -- min_free_kbytes worth of pages
tended to remain contiguous until the only alternative was to fail the
allocation.  At the time it was discovered that high-order atomic
allocations relied on this property so MIGRATE_RESERVE was introduced.  A
later patch will introduce an alternative MIGRATE_HIGHATOMIC so this patch
deletes MIGRATE_RESERVE and supporting code so it'll be easier to review.
Note that this patch in isolation may look like a false regression if
someone was bisecting high-order atomic allocation failures.

Signed-off-by: Mel Gorman 
Acked-by: Vlastimil Babka 
Cc: Christoph Lameter 
Cc: David Rientjes 
Cc: Johannes Weiner 
Cc: Michal Hocko 
Cc: Vitaly Wool 
Cc: Rik van Riel 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds