dlan17/linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
873,808 Commits 4 Branches 0 Tags
d2cd795c4ece1a24fda170c35eeb4f17d9826cbb
Commit Graph
100 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
Mel Gorman 3e8fc0075e mm, meminit: recalculate pcpu batch and high limits after init completes 
Deferred memory initialisation updates zone->managed_pages during the
initialisation phase but before that finishes, the per-cpu page
allocator (pcpu) calculates the number of pages allocated/freed in
batches as well as the maximum number of pages allowed on a per-cpu
list.  As zone->managed_pages is not up to date yet, the pcpu
initialisation calculates inappropriately low batch and high values.

This increases zone lock contention quite severely in some cases with
the degree of severity depending on how many CPUs share a local zone and
the size of the zone.  A private report indicated that kernel build
times were excessive with extremely high system CPU usage.  A perf
profile indicated that a large chunk of time was lost on zone->lock
contention.

This patch recalculates the pcpu batch and high values after deferred
initialisation completes for every populated zone in the system.  It was
tested on a 2-socket AMD EPYC 2 machine using a kernel compilation
workload -- allmodconfig and all available CPUs.

mmtests configuration: config-workload-kernbench-max Configuration was
modified to build on a fresh XFS partition.

kernbench
                                5.4.0-rc3              5.4.0-rc3
                                  vanilla           resetpcpu-v2
Amean     user-256    13249.50 (   0.00%)    16401.31 * -23.79%*
Amean     syst-256    14760.30 (   0.00%)     4448.39 *  69.86%*
Amean     elsp-256      162.42 (   0.00%)      119.13 *  26.65%*
Stddev    user-256       42.97 (   0.00%)       19.15 (  55.43%)
Stddev    syst-256      336.87 (   0.00%)        6.71 (  98.01%)
Stddev    elsp-256        2.46 (   0.00%)        0.39 (  84.03%)

                   5.4.0-rc3    5.4.0-rc3
                     vanilla resetpcpu-v2
Duration User       39766.24     49221.79
Duration System     44298.10     13361.67
Duration Elapsed      519.11       388.87

The patch reduces system CPU usage by 69.86% and total build time by
26.65%.  The variance of system CPU usage is also much reduced.

Before, this was the breakdown of batch and high values over all zones
was:

    256               batch: 1
    256               batch: 63
    512               batch: 7
    256               high:  0
    256               high:  378
    512               high:  42

512 pcpu pagesets had a batch limit of 7 and a high limit of 42.  After
the patch:

    256               batch: 1
    768               batch: 63
    256               high:  0
    768               high:  378

[mgorman@techsingularity.net: fix merge/linkage snafu]
  Link: http://lkml.kernel.org/r/20191023084705.GD3016@techsingularity.netLink: http://lkml.kernel.org/r/20191021094808.28824-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Qian Cai <cai@lca.pw>
Cc: <stable@vger.kernel.org>	[4.1+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-11-06 08:28:58 -08:00
Mel Gorman 28360f3987 mm, vmscan: do not special-case slab reclaim when watermarks are boosted 
Dave Chinner reported a problem pointing a finger at commit 1c30844d2d
("mm: reclaim small amounts of memory when an external fragmentation
event occurs").

The report is extensive:

  https://lore.kernel.org/linux-mm/20190807091858.2857-1-david@fromorbit.com/

and it's worth recording the most relevant parts (colorful language and
typos included).

	When running a simple, steady state 4kB file creation test to
	simulate extracting tarballs larger than memory full of small
	files into the filesystem, I noticed that once memory fills up
	the cache balance goes to hell.

	The workload is creating one dirty cached inode for every dirty
	page, both of which should require a single IO each to clean and
	reclaim, and creation of inodes is throttled by the rate at which
	dirty writeback runs at (via balance dirty pages). Hence the ingest
	rate of new cached inodes and page cache pages is identical and
	steady. As a result, memory reclaim should quickly find a steady
	balance between page cache and inode caches.

	The moment memory fills, the page cache is reclaimed at a much
	faster rate than the inode cache, and evidence suggests that
	the inode cache shrinker is not being called when large batches
	of pages are being reclaimed. In roughly the same time period
	that it takes to fill memory with 50% pages and 50% slab caches,
	memory reclaim reduces the page cache down to just dirty pages
	and slab caches fill the entirety of memory.

	The LRU is largely full of dirty pages, and we're getting spikes
	of random writeback from memory reclaim so it's all going to shit.
	Behaviour never recovers, the page cache remains pinned at just
	dirty pages, and nothing I could tune would make any difference.
	vfs_cache_pressure makes no difference - I would set it so high
	it should trim the entire inode caches in a single pass, yet it
	didn't do anything. It was clear from tracing and live telemetry
	that the shrinkers were pretty much not running except when
	there was absolutely no memory free at all, and then they did
	the minimum necessary to free memory to make progress.

	So I went looking at the code, trying to find places where pages
	got reclaimed and the shrinkers weren't called. There's only one
	- kswapd doing boosted reclaim as per commit 1c30844d2d ("mm:
	reclaim small amounts of memory when an external fragmentation
	event occurs").

The watermark boosting introduced by the commit is triggered in response
to an allocation "fragmentation event".  The boosting was not intended
to target THP specifically and triggers even if THP is disabled.
However, with Dave's perfectly reasonable workload, fragmentation events
can be very common given the ratio of slab to page cache allocations so
boosting remains active for long periods of time.

As high-order allocations might use compaction and compaction cannot
move slab pages the decision was made in the commit to special-case
kswapd when watermarks are boosted -- kswapd avoids reclaiming slab as
reclaiming slab does not directly help compaction.

As Dave notes, this decision means that slab can be artificially
protected for long periods of time and messes up the balance with slab
and page caches.

Removing the special casing can still indirectly help avoid
fragmentation by avoiding fragmentation-causing events due to slab
allocation as pages from a slab pageblock will have some slab objects
freed.  Furthermore, with the special casing, reclaim behaviour is
unpredictable as kswapd sometimes examines slab and sometimes does not
in a manner that is tricky to tune or analyse.

This patch removes the special casing.  The downside is that this is not
a universal performance win.  Some benchmarks that depend on the
residency of data when rereading metadata may see a regression when slab
reclaim is restored to its original behaviour.  Similarly, some
benchmarks that only read-once or write-once may perform better when
page reclaim is too aggressive.  The primary upside is that slab
shrinker is less surprising (arguably more sane but that's a matter of
opinion), behaves consistently regardless of the fragmentation state of
the system and properly obeys VM sysctls.

A fsmark benchmark configuration was constructed similar to what Dave
reported and is codified by the mmtest configuration
config-io-fsmark-small-file-stream.  It was evaluated on a 1-socket
machine to avoid dealing with NUMA-related issues and the timing of
reclaim.  The storage was an SSD Samsung Evo and a fresh trimmed XFS
filesystem was used for the test data.

This is not an exact replication of Dave's setup.  The configuration
scales its parameters depending on the memory size of the SUT to behave
similarly across machines.  The parameters mean the first sample
reported by fs_mark is using 50% of RAM which will barely be throttled
and look like a big outlier.  Dave used fake NUMA to have multiple
kswapd instances which I didn't replicate.  Finally, the number of
iterations differ from Dave's test as the target disk was not large
enough.  While not identical, it should be representative.

  fsmark
                                     5.3.0-rc3              5.3.0-rc3
                                       vanilla          shrinker-v1r1
  Min       1-files/sec     4444.80 (   0.00%)     4765.60 (   7.22%)
  1st-qrtle 1-files/sec     5005.10 (   0.00%)     5091.70 (   1.73%)
  2nd-qrtle 1-files/sec     4917.80 (   0.00%)     4855.60 (  -1.26%)
  3rd-qrtle 1-files/sec     4667.40 (   0.00%)     4831.20 (   3.51%)
  Max-1     1-files/sec    11421.50 (   0.00%)     9999.30 ( -12.45%)
  Max-5     1-files/sec    11421.50 (   0.00%)     9999.30 ( -12.45%)
  Max-10    1-files/sec    11421.50 (   0.00%)     9999.30 ( -12.45%)
  Max-90    1-files/sec     4649.60 (   0.00%)     4780.70 (   2.82%)
  Max-95    1-files/sec     4491.00 (   0.00%)     4768.20 (   6.17%)
  Max-99    1-files/sec     4491.00 (   0.00%)     4768.20 (   6.17%)
  Max       1-files/sec    11421.50 (   0.00%)     9999.30 ( -12.45%)
  Hmean     1-files/sec     5004.75 (   0.00%)     5075.96 (   1.42%)
  Stddev    1-files/sec     1778.70 (   0.00%)     1369.66 (  23.00%)
  CoeffVar  1-files/sec       33.70 (   0.00%)       26.05 (  22.71%)
  BHmean-99 1-files/sec     5053.72 (   0.00%)     5101.52 (   0.95%)
  BHmean-95 1-files/sec     5053.72 (   0.00%)     5101.52 (   0.95%)
  BHmean-90 1-files/sec     5107.05 (   0.00%)     5131.41 (   0.48%)
  BHmean-75 1-files/sec     5208.45 (   0.00%)     5206.68 (  -0.03%)
  BHmean-50 1-files/sec     5405.53 (   0.00%)     5381.62 (  -0.44%)
  BHmean-25 1-files/sec     6179.75 (   0.00%)     6095.14 (  -1.37%)

                     5.3.0-rc3   5.3.0-rc3
                       vanillashrinker-v1r1
  Duration User         501.82      497.29
  Duration System      4401.44     4424.08
  Duration Elapsed     8124.76     8358.05

This is showing a slight skew for the max result representing a large
outlier for the 1st, 2nd and 3rd quartile are similar indicating that
the bulk of the results show little difference.  Note that an earlier
version of the fsmark configuration showed a regression but that
included more samples taken while memory was still filling.

Note that the elapsed time is higher.  Part of this is that the
configuration included time to delete all the test files when the test
completes -- the test automation handles the possibility of testing
fsmark with multiple thread counts.  Without the patch, many of these
objects would be memory resident which is part of what the patch is
addressing.

There are other important observations that justify the patch.

1. With the vanilla kernel, the number of dirty pages in the system is
   very low for much of the test. With this patch, dirty pages is
   generally kept at 10% which matches vm.dirty_background_ratio which
   is normal expected historical behaviour.

2. With the vanilla kernel, the ratio of Slab/Pagecache is close to
   0.95 for much of the test i.e. Slab is being left alone and
   dominating memory consumption. With the patch applied, the ratio
   varies between 0.35 and 0.45 with the bulk of the measured ratios
   roughly half way between those values. This is a different balance to
   what Dave reported but it was at least consistent.

3. Slabs are scanned throughout the entire test with the patch applied.
   The vanille kernel has periods with no scan activity and then
   relatively massive spikes.

4. Without the patch, kswapd scan rates are very variable. With the
   patch, the scan rates remain quite steady.

4. Overall vmstats are closer to normal expectations

	                                5.3.0-rc3      5.3.0-rc3
	                                  vanilla  shrinker-v1r1
    Ops Direct pages scanned             99388.00      328410.00
    Ops Kswapd pages scanned          45382917.00    33451026.00
    Ops Kswapd pages reclaimed        30869570.00    25239655.00
    Ops Direct pages reclaimed           74131.00        5830.00
    Ops Kswapd efficiency %                 68.02          75.45
    Ops Kswapd velocity                   5585.75        4002.25
    Ops Page reclaim immediate         1179721.00      430927.00
    Ops Slabs scanned                 62367361.00    73581394.00
    Ops Direct inode steals               2103.00        1002.00
    Ops Kswapd inode steals             570180.00     5183206.00

	o Vanilla kernel is hitting direct reclaim more frequently,
	  not very much in absolute terms but the fact the patch
	  reduces it is interesting
	o "Page reclaim immediate" in the vanilla kernel indicates
	  dirty pages are being encountered at the tail of the LRU.
	  This is generally bad and means in this case that the LRU
	  is not long enough for dirty pages to be cleaned by the
	  background flush in time. This is much reduced by the
	  patch.
	o With the patch, kswapd is reclaiming 10 times more slab
	  pages than with the vanilla kernel. This is indicative
	  of the watermark boosting over-protecting slab

A more complete set of tests were run that were part of the basis for
introducing boosting and while there are some differences, they are well
within tolerances.

Bottom line, the special casing kswapd to avoid slab behaviour is
unpredictable and can lead to abnormal results for normal workloads.

This patch restores the expected behaviour that slab and page cache is
balanced consistently for a workload with a steady allocation ratio of
slab/pagecache pages.  It also means that if there are workloads that
favour the preservation of slab over pagecache that it can be tuned via
vm.vfs_cache_pressure where as the vanilla kernel effectively ignores
the parameter when boosting is active.

Link: http://lkml.kernel.org/r/20190808182946.GM2739@techsingularity.net
Fixes: 1c30844d2d ("mm: reclaim small amounts of memory when an external fragmentation event occurs")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: <stable@vger.kernel.org>	[5.0+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-08-13 16:06:53 -07:00
Mel Gorman 670105a256 mm: compaction: avoid 100% CPU usage during compaction when a task is killed 
"howaboutsynergy" reported via kernel buzilla number 204165 that
compact_zone_order was consuming 100% CPU during a stress test for
prolonged periods of time.  Specifically the following command, which
should exit in 10 seconds, was taking an excessive time to finish while
the CPU was pegged at 100%.

  stress -m 220 --vm-bytes 1000000000 --timeout 10

Tracing indicated a pattern as follows

          stress-3923  [007]   519.106208: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0
          stress-3923  [007]   519.106212: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0
          stress-3923  [007]   519.106216: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0
          stress-3923  [007]   519.106219: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0
          stress-3923  [007]   519.106223: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0
          stress-3923  [007]   519.106227: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0
          stress-3923  [007]   519.106231: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0
          stress-3923  [007]   519.106235: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0
          stress-3923  [007]   519.106238: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0
          stress-3923  [007]   519.106242: mm_compaction_isolate_migratepages: range=(0x70bb80 ~ 0x70bb80) nr_scanned=0 nr_taken=0

Note that compaction is entered in rapid succession while scanning and
isolating nothing.  The problem is that when a task that is compacting
receives a fatal signal, it retries indefinitely instead of exiting
while making no progress as a fatal signal is pending.

It's not easy to trigger this condition although enabling zswap helps on
the basis that the timing is altered.  A very small window has to be hit
for the problem to occur (signal delivered while compacting and
isolating a PFN for migration that is not aligned to SWAP_CLUSTER_MAX).

This was reproduced locally -- 16G single socket system, 8G swap, 30%
zswap configured, vm-bytes 22000000000 using Colin Kings stress-ng
implementation from github running in a loop until the problem hits).
Tracing recorded the problem occurring almost 200K times in a short
window.  With this patch, the problem hit 4 times but the task existed
normally instead of consuming CPU.

This problem has existed for some time but it was made worse by commit
cf66f0700c ("mm, compaction: do not consider a need to reschedule as
contention").  Before that commit, if the same condition was hit then
locks would be quickly contended and compaction would exit that way.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=204165
Link: http://lkml.kernel.org/r/20190718085708.GE24383@techsingularity.net
Fixes: cf66f0700c ("mm, compaction: do not consider a need to reschedule as contention")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org>	[5.1+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-08-03 07:02:00 -07:00
Mel Gorman 60fce36afa mm/compaction.c: correct zone boundary handling when isolating pages from a pageblock 
syzbot reported the following error from a tree with a head commit of
baf76f0c58 ("slip: make slhc_free() silently accept an error pointer")

  BUG: unable to handle kernel paging request at ffffea0003348000
  #PF error: [normal kernel read fault]
  PGD 12c3f9067 P4D 12c3f9067 PUD 12c3f8067 PMD 0
  Oops: 0000 [#1] PREEMPT SMP KASAN
  CPU: 1 PID: 28916 Comm: syz-executor.2 Not tainted 5.1.0-rc6+ #89
  Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
  RIP: 0010:constant_test_bit arch/x86/include/asm/bitops.h:314 [inline]
  RIP: 0010:PageCompound include/linux/page-flags.h:186 [inline]
  RIP: 0010:isolate_freepages_block+0x1c0/0xd40 mm/compaction.c:579
  Code: 01 d8 ff 4d 85 ed 0f 84 ef 07 00 00 e8 29 00 d8 ff 4c 89 e0 83 85 38 ff
  ff ff 01 48 c1 e8 03 42 80 3c 38 00 0f 85 31 0a 00 00 <4d> 8b 2c 24 31 ff 49
  c1 ed 10 41 83 e5 01 44 89 ee e8 3a 01 d8 ff
  RSP: 0018:ffff88802b31eab8 EFLAGS: 00010246
  RAX: 1ffffd4000669000 RBX: 00000000000cd200 RCX: ffffc9000a235000
  RDX: 000000000001ca5e RSI: ffffffff81988cc7 RDI: 0000000000000001
  RBP: ffff88802b31ebd8 R08: ffff88805af700c0 R09: 0000000000000000
  R10: 0000000000000000 R11: 0000000000000000 R12: ffffea0003348000
  R13: 0000000000000000 R14: ffff88802b31f030 R15: dffffc0000000000
  FS:  00007f61648dc700(0000) GS:ffff8880ae900000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: ffffea0003348000 CR3: 0000000037c64000 CR4: 00000000001426e0
  Call Trace:
   fast_isolate_around mm/compaction.c:1243 [inline]
   fast_isolate_freepages mm/compaction.c:1418 [inline]
   isolate_freepages mm/compaction.c:1438 [inline]
   compaction_alloc+0x1aee/0x22e0 mm/compaction.c:1550

There is no reproducer and it is difficult to hit -- 1 crash every few
days.  The issue is very similar to the fix in commit 6b0868c820
("mm/compaction.c: correct zone boundary handling when resetting pageblock
skip hints").  When isolating free pages around a target pageblock, the
boundary handling is off by one and can stray into the next pageblock.
Triggering the syzbot error requires that the end of pageblock is section
or zone aligned, and that the next section is unpopulated.

A more subtle consequence of the bug is that pageblocks were being
improperly used as migration targets which potentially hurts fragmentation
avoidance in the long-term one page at a time.

A debugging patch revealed that it's definitely possible to stray outside
of a pageblock which is not intended.  While syzbot cannot be used to
verify this patch, it was confirmed that the debugging warning no longer
triggers with this patch applied.  It has also been confirmed that the THP
allocation stress tests are not degraded by this patch.

Link: http://lkml.kernel.org/r/20190510182124.GI18914@techsingularity.net
Fixes: e332f741a8 ("mm, compaction: be selective about what pageblocks to clear skip hints")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: syzbot+d84c80f9fe26a0f7a734@syzkaller.appspotmail.com
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Qian Cai <cai@lca.pw>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org> # v5.1+
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-05-18 15:52:26 -07:00
Mel Gorman ee8ab0eeb4 mm, page_alloc: always use a captured page regardless of compaction result 
During the development of commit 5e1f0f098b ("mm, compaction: capture
a page under direct compaction"), a paranoid check was added to ensure
that if a captured page was available after compaction that it was
consistent with the final state of compaction.  The intent was to catch
serious programming bugs such as using a stale page pointer and causing
corruption problems.

However, it is possible to get a captured page even if compaction was
unsuccessful if an interrupt triggered and happened to free pages in
interrupt context that got merged into a suitable high-order page.  It's
highly unlikely but Li Wang did report the following warning on s390
occuring when testing OOM handling.  Note that the warning is slightly
edited for clarity.

  WARNING: CPU: 0 PID: 9783 at mm/page_alloc.c:3777 __alloc_pages_direct_compact+0x182/0x190
  Modules linked in: rpcsec_gss_krb5 auth_rpcgss nfsv4 dns_resolver nfs
    lockd grace fscache sunrpc pkey ghash_s390 prng xts aes_s390
    des_s390 des_generic sha512_s390 zcrypt_cex4 zcrypt vmur binfmt_misc
    ip_tables xfs libcrc32c dasd_fba_mod qeth_l2 dasd_eckd_mod dasd_mod
    qeth qdio lcs ctcm ccwgroup fsm dm_mirror dm_region_hash dm_log
    dm_mod
  CPU: 0 PID: 9783 Comm: copy.sh Kdump: loaded Not tainted 5.1.0-rc 5 #1

This patch simply removes the check entirely instead of trying to be
clever about pages freed from interrupt context.  If a serious
programming error was introduced, it is highly likely to be caught by
prep_new_page() instead.

Link: http://lkml.kernel.org/r/20190419085133.GH18914@techsingularity.net
Fixes: 5e1f0f098b ("mm, compaction: capture a page under direct compaction")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: Li Wang <liwang@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-04-26 09:18:05 -07:00
Mel Gorman 24512228b7 mm: do not boost watermarks to avoid fragmentation for the DISCONTIG memory model 
Mikulas Patocka reported that commit 1c30844d2d ("mm: reclaim small
amounts of memory when an external fragmentation event occurs") "broke"
memory management on parisc.

The machine is not NUMA but the DISCONTIG model creates three pgdats
even though it's a UMA machine for the following ranges

        0) Start 0x0000000000000000 End 0x000000003fffffff Size   1024 MB
        1) Start 0x0000000100000000 End 0x00000001bfdfffff Size   3070 MB
        2) Start 0x0000004040000000 End 0x00000040ffffffff Size   3072 MB

Mikulas reported:

	With the patch 1c30844d2, the kernel will incorrectly reclaim the
	first zone when it fills up, ignoring the fact that there are two
	completely free zones. Basiscally, it limits cache size to 1GiB.

	For example, if I run:
	# dd if=/dev/sda of=/dev/null bs=1M count=2048

	- with the proper kernel, there should be "Buffers - 2GiB"
	when this command finishes. With the patch 1c30844d2, buffers
	will consume just 1GiB or slightly more, because the kernel was
	incorrectly reclaiming them.

The page allocator and reclaim makes assumptions that pgdats really
represent NUMA nodes and zones represent ranges and makes decisions on
that basis.  Watermark boosting for small pgdats leads to unexpected
results even though this would have behaved reasonably on SPARSEMEM.

DISCONTIG is essentially deprecated and even parisc plans to move to
SPARSEMEM so there is no need to be fancy, this patch simply disables
watermark boosting by default on DISCONTIGMEM.

Link: http://lkml.kernel.org/r/20190419094335.GJ18914@techsingularity.net
Fixes: 1c30844d2d ("mm: reclaim small amounts of memory when an external fragmentation event occurs")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: Mikulas Patocka <mpatocka@redhat.com>
Tested-by: Mikulas Patocka <mpatocka@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: James Bottomley <James.Bottomley@hansenpartnership.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-04-26 09:18:05 -07:00
Mel Gorman 6b0868c820 mm/compaction.c: correct zone boundary handling when resetting pageblock skip hints 
Mikhail Gavrilo reported the following bug being triggered in a Fedora
kernel based on 5.1-rc1 but it is relevant to a vanilla kernel.

 kernel: page dumped because: VM_BUG_ON_PAGE(PagePoisoned(p))
 kernel: ------------[ cut here ]------------
 kernel: kernel BUG at include/linux/mm.h:1021!
 kernel: invalid opcode: 0000 [#1] SMP NOPTI
 kernel: CPU: 6 PID: 116 Comm: kswapd0 Tainted: G         C        5.1.0-0.rc1.git1.3.fc31.x86_64 #1
 kernel: Hardware name: System manufacturer System Product Name/ROG STRIX X470-I GAMING, BIOS 1201 12/07/2018
 kernel: RIP: 0010:__reset_isolation_pfn+0x244/0x2b0
 kernel: Code: fe 06 e8 0f 8e fc ff 44 0f b6 4c 24 04 48 85 c0 0f 85 dc fe ff ff e9 68 fe ff ff 48 c7 c6 58 b7 2e 8c 4c 89 ff e8 0c 75 00 00 <0f> 0b 48 c7 c6 58 b7 2e 8c e8 fe 74 00 00 0f 0b 48 89 fa 41 b8 01
 kernel: RSP: 0018:ffff9e2d03f0fde8 EFLAGS: 00010246
 kernel: RAX: 0000000000000034 RBX: 000000000081f380 RCX: ffff8cffbddd6c20
 kernel: RDX: 0000000000000000 RSI: 0000000000000006 RDI: ffff8cffbddd6c20
 kernel: RBP: 0000000000000001 R08: 0000009898b94613 R09: 0000000000000000
 kernel: R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000100000
 kernel: R13: 0000000000100000 R14: 0000000000000001 R15: ffffca7de07ce000
 kernel: FS:  0000000000000000(0000) GS:ffff8cffbdc00000(0000) knlGS:0000000000000000
 kernel: CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 kernel: CR2: 00007fc1670e9000 CR3: 00000007f5276000 CR4: 00000000003406e0
 kernel: Call Trace:
 kernel:  __reset_isolation_suitable+0x62/0x120
 kernel:  reset_isolation_suitable+0x3b/0x40
 kernel:  kswapd+0x147/0x540
 kernel:  ? finish_wait+0x90/0x90
 kernel:  kthread+0x108/0x140
 kernel:  ? balance_pgdat+0x560/0x560
 kernel:  ? kthread_park+0x90/0x90
 kernel:  ret_from_fork+0x27/0x50

He bisected it down to e332f741a8 ("mm, compaction: be selective about
what pageblocks to clear skip hints").  The problem is that the patch in
question was sloppy with respect to the handling of zone boundaries.  In
some instances, it was possible for PFNs outside of a zone to be examined
and if those were not properly initialised or poisoned then it would
trigger the VM_BUG_ON.  This patch corrects the zone boundary issues when
resetting pageblock skip hints and Mikhail reported that the bug did not
trigger after 30 hours of testing.

Link: http://lkml.kernel.org/r/20190327085424.GL3189@techsingularity.net
Fixes: e332f741a8 ("mm, compaction: be selective about what pageblocks to clear skip hints")
Reported-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Cc: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Qian Cai <cai@lca.pw>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
 2019-04-04 11:56:01 +01:00
Mel Gorman 0e9f02450d sched/fair: Do not re-read ->h_load_next during hierarchical load calculation 
A NULL pointer dereference bug was reported on a distribution kernel but
the same issue should be present on mainline kernel. It occured on s390
but should not be arch-specific.  A partial oops looks like:

  Unable to handle kernel pointer dereference in virtual kernel address space
  ...
  Call Trace:
    ...
    try_to_wake_up+0xfc/0x450
    vhost_poll_wakeup+0x3a/0x50 [vhost]
    __wake_up_common+0xbc/0x178
    __wake_up_common_lock+0x9e/0x160
    __wake_up_sync_key+0x4e/0x60
    sock_def_readable+0x5e/0x98

The bug hits any time between 1 hour to 3 days. The dereference occurs
in update_cfs_rq_h_load when accumulating h_load. The problem is that
cfq_rq->h_load_next is not protected by any locking and can be updated
by parallel calls to task_h_load. Depending on the compiler, code may be
generated that re-reads cfq_rq->h_load_next after the check for NULL and
then oops when reading se->avg.load_avg. The dissassembly showed that it
was possible to reread h_load_next after the check for NULL.

While this does not appear to be an issue for later compilers, it's still
an accident if the correct code is generated. Full locking in this path
would have high overhead so this patch uses READ_ONCE to read h_load_next
only once and check for NULL before dereferencing. It was confirmed that
there were no further oops after 10 days of testing.

As Peter pointed out, it is also necessary to use WRITE_ONCE() to avoid any
potential problems with store tearing.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Valentin Schneider <valentin.schneider@arm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: <stable@vger.kernel.org>
Fixes: 685207963b ("sched: Move h_load calculation to task_h_load()")
Link: https://lkml.kernel.org/r/20190319123610.nsivgf3mjbjjesxb@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2019-04-03 09:50:22 +02:00
Mel Gorman 5e1f0f098b mm, compaction: capture a page under direct compaction 
Compaction is inherently race-prone as a suitable page freed during
compaction can be allocated by any parallel task.  This patch uses a
capture_control structure to isolate a page immediately when it is freed
by a direct compactor in the slow path of the page allocator.  The
intent is to avoid redundant scanning.

                                     5.0.0-rc1              5.0.0-rc1
                               selective-v3r17          capture-v3r19
Amean     fault-both-1         0.00 (   0.00%)        0.00 *   0.00%*
Amean     fault-both-3      2582.11 (   0.00%)     2563.68 (   0.71%)
Amean     fault-both-5      4500.26 (   0.00%)     4233.52 (   5.93%)
Amean     fault-both-7      5819.53 (   0.00%)     6333.65 (  -8.83%)
Amean     fault-both-12     9321.18 (   0.00%)     9759.38 (  -4.70%)
Amean     fault-both-18     9782.76 (   0.00%)    10338.76 (  -5.68%)
Amean     fault-both-24    15272.81 (   0.00%)    13379.55 *  12.40%*
Amean     fault-both-30    15121.34 (   0.00%)    16158.25 (  -6.86%)
Amean     fault-both-32    18466.67 (   0.00%)    18971.21 (  -2.73%)

Latency is only moderately affected but the devil is in the details.  A
closer examination indicates that base page fault latency is reduced but
latency of huge pages is increased as it takes creater care to succeed.
Part of the "problem" is that allocation success rates are close to 100%
even when under pressure and compaction gets harder

                                5.0.0-rc1              5.0.0-rc1
                          selective-v3r17          capture-v3r19
Percentage huge-3        96.70 (   0.00%)       98.23 (   1.58%)
Percentage huge-5        96.99 (   0.00%)       95.30 (  -1.75%)
Percentage huge-7        94.19 (   0.00%)       97.24 (   3.24%)
Percentage huge-12       94.95 (   0.00%)       97.35 (   2.53%)
Percentage huge-18       96.74 (   0.00%)       97.30 (   0.58%)
Percentage huge-24       97.07 (   0.00%)       97.55 (   0.50%)
Percentage huge-30       95.69 (   0.00%)       98.50 (   2.95%)
Percentage huge-32       96.70 (   0.00%)       99.27 (   2.65%)

And scan rates are reduced as expected by 6% for the migration scanner
and 29% for the free scanner indicating that there is less redundant
work.

Compaction migrate scanned    20815362    19573286
Compaction free scanned       16352612    11510663

[mgorman@techsingularity.net: remove redundant check]
  Link: http://lkml.kernel.org/r/20190201143853.GH9565@techsingularity.net
Link: http://lkml.kernel.org/r/20190118175136.31341-23-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:17 -08:00
Mel Gorman e332f741a8 mm, compaction: be selective about what pageblocks to clear skip hints 
Pageblock hints are cleared when compaction restarts or kswapd makes
enough progress that it can sleep but it's over-eager in that the bit is
cleared for migration sources with no LRU pages and migration targets
with no free pages.  As pageblock skip hint flushes are relatively rare
and out-of-band with respect to kswapd, this patch makes a few more
expensive checks to see if it's appropriate to even clear the bit.
Every pageblock that is not cleared will avoid 512 pages being scanned
unnecessarily on x86-64.

The impact is variable with different workloads showing small
differences in latency, success rates and scan rates.  This is expected
as clearing the hints is not that common but doing a small amount of
work out-of-band to avoid a large amount of work in-band later is
generally a good thing.

Link: http://lkml.kernel.org/r/20190118175136.31341-22-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Qian Cai <cai@lca.pw>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
[cai@lca.pw: no stuck in __reset_isolation_pfn()]
  Link: http://lkml.kernel.org/r/20190206034732.75687-1-cai@lca.pw
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:17 -08:00
Mel Gorman 4fca9730c5 mm, compaction: sample pageblocks for free pages 
Once fast searching finishes, there is a possibility that the linear
scanner is scanning full blocks found by the fast scanner earlier.  This
patch uses an adaptive stride to sample pageblocks for free pages.  The
more consecutive full pageblocks encountered, the larger the stride
until a pageblock with free pages is found.  The scanners might meet
slightly sooner but it is an acceptable risk given that the search of
the free lists may still encounter the pages and adjust the cached PFN
of the free scanner accordingly.

                                     5.0.0-rc1              5.0.0-rc1
                              roundrobin-v3r17       samplefree-v3r17
Amean     fault-both-1         0.00 (   0.00%)        0.00 *   0.00%*
Amean     fault-both-3      2752.37 (   0.00%)     2729.95 (   0.81%)
Amean     fault-both-5      4341.69 (   0.00%)     4397.80 (  -1.29%)
Amean     fault-both-7      6308.75 (   0.00%)     6097.61 (   3.35%)
Amean     fault-both-12    10241.81 (   0.00%)     9407.15 (   8.15%)
Amean     fault-both-18    13736.09 (   0.00%)    10857.63 *  20.96%*
Amean     fault-both-24    16853.95 (   0.00%)    13323.24 *  20.95%*
Amean     fault-both-30    15862.61 (   0.00%)    17345.44 (  -9.35%)
Amean     fault-both-32    18450.85 (   0.00%)    16892.00 (   8.45%)

The latency is mildly improved offseting some overhead from earlier
patches that are prerequisites for the rest of the series.  However, a
major impact is on the free scan rate with an 82% reduction.

                                5.0.0-rc1      5.0.0-rc1
                         roundrobin-v3r17 samplefree-v3r17
Compaction migrate scanned    21607271            20116887
Compaction free scanned       95336406            16668703

It's also the first time in the series where the number of pages scanned
by the migration scanner is greater than the free scanner due to the
increased search efficiency.

Link: http://lkml.kernel.org/r/20190118175136.31341-21-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:17 -08:00
Mel Gorman dbe2d4e4f1 mm, compaction: round-robin the order while searching the free lists for a target 
As compaction proceeds and creates high-order blocks, the free list
search gets less efficient as the larger blocks are used as compaction
targets.  Eventually, the larger blocks will be behind the migration
scanner for partially migrated pageblocks and the search fails.  This
patch round-robins what orders are searched so that larger blocks can be
ignored and find smaller blocks that can be used as migration targets.

The overall impact was small on 1-socket but it avoids corner cases
where the migration/free scanners meet prematurely or situations where
many of the pageblocks encountered by the free scanner are almost full
instead of being properly packed.  Previous testing had indicated that
without this patch there were occasional large spikes in the free
scanner without this patch.

[dan.carpenter@oracle.com: fix static checker warning]
Link: http://lkml.kernel.org/r/20190118175136.31341-20-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:17 -08:00
Mel Gorman d097a6f635 mm, compaction: reduce premature advancement of the migration target scanner 
The fast isolation of free pages allows the cached PFN of the free
scanner to advance faster than necessary depending on the contents of
the free list.  The key is that fast_isolate_freepages() can update
zone->compact_cached_free_pfn via isolate_freepages_block().  When the
fast search fails, the linear scan can start from a point that has
skipped valid migration targets, particularly pageblocks with just
low-order free pages.  This can cause the migration source/target
scanners to meet prematurely causing a reset.

This patch starts by avoiding an update of the pageblock skip
information and cached PFN from isolate_freepages_block() and puts the
responsibility of updating that information in the callers.  The fast
scanner will update the cached PFN if and only if it finds a block that
is higher than the existing cached PFN and sets the skip if the
pageblock is full or nearly full.  The linear scanner will update
skipped information and the cached PFN only when a block is completely
scanned.  The total impact is that the free scanner advances more slowly
as it is primarily driven by the linear scanner instead of the fast
search.

                                     5.0.0-rc1              5.0.0-rc1
                               noresched-v3r17         slowfree-v3r17
Amean     fault-both-3      2965.68 (   0.00%)     3036.75 (  -2.40%)
Amean     fault-both-5      3995.90 (   0.00%)     4522.24 * -13.17%*
Amean     fault-both-7      5842.12 (   0.00%)     6365.35 (  -8.96%)
Amean     fault-both-12     9550.87 (   0.00%)    10340.93 (  -8.27%)
Amean     fault-both-18    13304.72 (   0.00%)    14732.46 ( -10.73%)
Amean     fault-both-24    14618.59 (   0.00%)    16288.96 ( -11.43%)
Amean     fault-both-30    16650.96 (   0.00%)    16346.21 (   1.83%)
Amean     fault-both-32    17145.15 (   0.00%)    19317.49 ( -12.67%)

The impact to latency is higher than the last version but it appears to
be due to a slight increase in the free scan rates which is a potential
side-effect of the patch.  However, this is necessary for later patches
that are more careful about how pageblocks are treated as earlier
iterations of those patches hit corner cases where the restarts were
punishing and very visible.

Link: http://lkml.kernel.org/r/20190118175136.31341-19-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:17 -08:00
Mel Gorman cf66f0700c mm, compaction: do not consider a need to reschedule as contention 
Scanning on large machines can take a considerable length of time and
eventually need to be rescheduled.  This is treated as an abort event
but that's not appropriate as the attempt is likely to be retried after
making numerous checks and taking another cycle through the page
allocator.  This patch will check the need to reschedule if necessary
but continue the scanning.

The main benefit is reduced scanning when compaction is taking a long
time or the machine is over-saturated.  It also avoids an unnecessary
exit of compaction that ends up being retried by the page allocator in
the outer loop.

                                     5.0.0-rc1              5.0.0-rc1
                              synccached-v3r16        noresched-v3r17
Amean     fault-both-1         0.00 (   0.00%)        0.00 *   0.00%*
Amean     fault-both-3      2958.27 (   0.00%)     2965.68 (  -0.25%)
Amean     fault-both-5      4091.90 (   0.00%)     3995.90 (   2.35%)
Amean     fault-both-7      5803.05 (   0.00%)     5842.12 (  -0.67%)
Amean     fault-both-12     9481.06 (   0.00%)     9550.87 (  -0.74%)
Amean     fault-both-18    14141.51 (   0.00%)    13304.72 (   5.92%)
Amean     fault-both-24    16438.00 (   0.00%)    14618.59 (  11.07%)
Amean     fault-both-30    17531.72 (   0.00%)    16650.96 (   5.02%)
Amean     fault-both-32    17101.96 (   0.00%)    17145.15 (  -0.25%)

Link: http://lkml.kernel.org/r/20190118175136.31341-18-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:17 -08:00
Mel Gorman cb810ad294 mm, compaction: rework compact_should_abort as compact_check_resched 
With incremental changes, compact_should_abort no longer makes any
documented sense.  Rename to compact_check_resched and update the
associated comments.  There is no benefit other than reducing redundant
code and making the intent slightly clearer.  It could potentially be
merged with earlier patches but it just makes the review slightly
harder.

Link: http://lkml.kernel.org/r/20190118175136.31341-17-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:17 -08:00
Mel Gorman 8854c55f54 mm, compaction: keep cached migration PFNs synced for unusable pageblocks 
Migrate has separate cached PFNs for ASYNC and SYNC* migration on the
basis that some migrations will fail in ASYNC mode.  However, if the
cached PFNs match at the start of scanning and pageblocks are skipped
due to having no isolation candidates, then the sync state does not
matter.  This patch keeps matching cached PFNs in sync until a pageblock
with isolation candidates is found.

The actual benefit is marginal given that the sync scanner following the
async scanner will often skip a number of pageblocks but it's useless
work.  Any benefit depends heavily on whether the scanners restarted
recently.

Link: http://lkml.kernel.org/r/20190118175136.31341-16-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:17 -08:00
Mel Gorman 9bebefd590 mm, compaction: check early for huge pages encountered by the migration scanner 
When scanning for sources or targets, PageCompound is checked for huge
pages as they can be skipped quickly but it happens relatively late
after a lot of setup and checking.  This patch short-cuts the check to
make it earlier.  It might still change when the lock is acquired but
this has less overhead overall.  The free scanner advances but the
migration scanner does not.  Typically the free scanner encounters more
movable blocks that change state over the lifetime of the system and
also tends to scan more aggressively as it's actively filling its
portion of the physical address space with data.  This could change in
the future but for the moment, this worked better in practice and
incurred fewer scan restarts.

The impact on latency and allocation success rates is marginal but the
free scan rates are reduced by 15% and system CPU usage is reduced by
3.3%.  The 2-socket results are not materially different.

Link: http://lkml.kernel.org/r/20190118175136.31341-15-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman cb2dcaf023 mm, compaction: finish pageblock scanning on contention 
Async migration aborts on spinlock contention but contention can be high
when there are multiple compaction attempts and kswapd is active.  The
consequence is that the migration scanners move forward uselessly while
still contending on locks for longer while leaving suitable migration
sources behind.

This patch will acquire the lock but track when contention occurs.  When
it does, the current pageblock will finish as compaction may succeed for
that block and then abort.  This will have a variable impact on latency
as in some cases useless scanning is avoided (reduces latency) but a
lock will be contended (increase latency) or a single contended
pageblock is scanned that would otherwise have been skipped (increase
latency).

                                     5.0.0-rc1              5.0.0-rc1
                                norescan-v3r16    finishcontend-v3r16
Amean     fault-both-1         0.00 (   0.00%)        0.00 *   0.00%*
Amean     fault-both-3      3002.07 (   0.00%)     3153.17 (  -5.03%)
Amean     fault-both-5      4684.47 (   0.00%)     4280.52 (   8.62%)
Amean     fault-both-7      6815.54 (   0.00%)     5811.50 *  14.73%*
Amean     fault-both-12    10864.02 (   0.00%)     9276.85 (  14.61%)
Amean     fault-both-18    12247.52 (   0.00%)    11032.67 (   9.92%)
Amean     fault-both-24    15683.99 (   0.00%)    14285.70 (   8.92%)
Amean     fault-both-30    18620.02 (   0.00%)    16293.76 *  12.49%*
Amean     fault-both-32    19250.28 (   0.00%)    16721.02 *  13.14%*

                                5.0.0-rc1              5.0.0-rc1
                           norescan-v3r16    finishcontend-v3r16
Percentage huge-1         0.00 (   0.00%)        0.00 (   0.00%)
Percentage huge-3        95.00 (   0.00%)       96.82 (   1.92%)
Percentage huge-5        94.22 (   0.00%)       95.40 (   1.26%)
Percentage huge-7        92.35 (   0.00%)       95.92 (   3.86%)
Percentage huge-12       91.90 (   0.00%)       96.73 (   5.25%)
Percentage huge-18       89.58 (   0.00%)       96.77 (   8.03%)
Percentage huge-24       90.03 (   0.00%)       96.05 (   6.69%)
Percentage huge-30       89.14 (   0.00%)       96.81 (   8.60%)
Percentage huge-32       90.58 (   0.00%)       97.41 (   7.54%)

There is a variable impact that is mostly good on latency while allocation
success rates are slightly higher.  System CPU usage is reduced by about
10% but scan rate impact is mixed

Compaction migrate scanned    27997659.00    20148867
Compaction free scanned      120782791.00   118324914

Migration scan rates are reduced 28% which is expected as a pageblock is
used by the async scanner instead of skipped.  The impact on the free
scanner is known to be variable.  Overall the primary justification for
this patch is that completing scanning of a pageblock is very important
for later patches.

[yuehaibing@huawei.com: fix unused variable warning]
Link: http://lkml.kernel.org/r/20190118175136.31341-14-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: YueHaibing <yuehaibing@huawei.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman 804d3121ba mm, compaction: avoid rescanning the same pageblock multiple times 
Pageblocks are marked for skip when no pages are isolated after a scan.
However, it's possible to hit corner cases where the migration scanner
gets stuck near the boundary between the source and target scanner.  Due
to pages being migrated in blocks of COMPACT_CLUSTER_MAX, pages that are
migrated can be reallocated before the pageblock is complete.  The
pageblock is not necessarily skipped so it can be rescanned multiple
times.  Similarly, a pageblock with some dirty/writeback pages may fail
to migrate and be rescanned until writeback completes which is wasteful.

This patch tracks if a pageblock is being rescanned.  If so, then the
entire pageblock will be migrated as one operation.  This narrows the
race window during which pages can be reallocated during migration.
Secondly, if there are pages that cannot be isolated then the pageblock
will still be fully scanned and marked for skipping.  On the second
rescan, the pageblock skip is set and the migration scanner makes
progress.

                                     5.0.0-rc1              5.0.0-rc1
                                findfree-v3r16         norescan-v3r16
Amean     fault-both-1         0.00 (   0.00%)        0.00 *   0.00%*
Amean     fault-both-3      3200.68 (   0.00%)     3002.07 (   6.21%)
Amean     fault-both-5      4847.75 (   0.00%)     4684.47 (   3.37%)
Amean     fault-both-7      6658.92 (   0.00%)     6815.54 (  -2.35%)
Amean     fault-both-12    11077.62 (   0.00%)    10864.02 (   1.93%)
Amean     fault-both-18    12403.97 (   0.00%)    12247.52 (   1.26%)
Amean     fault-both-24    15607.10 (   0.00%)    15683.99 (  -0.49%)
Amean     fault-both-30    18752.27 (   0.00%)    18620.02 (   0.71%)
Amean     fault-both-32    21207.54 (   0.00%)    19250.28 *   9.23%*

                                5.0.0-rc1              5.0.0-rc1
                           findfree-v3r16         norescan-v3r16
Percentage huge-3        96.86 (   0.00%)       95.00 (  -1.91%)
Percentage huge-5        93.72 (   0.00%)       94.22 (   0.53%)
Percentage huge-7        94.31 (   0.00%)       92.35 (  -2.08%)
Percentage huge-12       92.66 (   0.00%)       91.90 (  -0.82%)
Percentage huge-18       91.51 (   0.00%)       89.58 (  -2.11%)
Percentage huge-24       90.50 (   0.00%)       90.03 (  -0.52%)
Percentage huge-30       91.57 (   0.00%)       89.14 (  -2.65%)
Percentage huge-32       91.00 (   0.00%)       90.58 (  -0.46%)

Negligible difference but this was likely a case when the specific
corner case was not hit.  A previous run of the same patch based on an
earlier iteration of the series showed large differences where migration
rates could be halved when the corner case was hit.

The specific corner case where migration scan rates go through the roof
was due to a dirty/writeback pageblock located at the boundary of the
migration/free scanner did not happen in this case.  When it does
happen, the scan rates multipled by massive margins.

Link: http://lkml.kernel.org/r/20190118175136.31341-13-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman 5a811889de mm, compaction: use free lists to quickly locate a migration target 
Similar to the migration scanner, this patch uses the free lists to
quickly locate a migration target.  The search is different in that
lower orders will be searched for a suitable high PFN if necessary but
the search is still bound.  This is justified on the grounds that the
free scanner typically scans linearly much more than the migration
scanner.

If a free page is found, it is isolated and compaction continues if
enough pages were isolated.  For SYNC* scanning, the full pageblock is
scanned for any remaining free pages so that is can be marked for
skipping in the near future.

1-socket thpfioscale
                                     5.0.0-rc1              5.0.0-rc1
                                 isolmig-v3r15         findfree-v3r16
Amean     fault-both-3      3024.41 (   0.00%)     3200.68 (  -5.83%)
Amean     fault-both-5      4749.30 (   0.00%)     4847.75 (  -2.07%)
Amean     fault-both-7      6454.95 (   0.00%)     6658.92 (  -3.16%)
Amean     fault-both-12    10324.83 (   0.00%)    11077.62 (  -7.29%)
Amean     fault-both-18    12896.82 (   0.00%)    12403.97 (   3.82%)
Amean     fault-both-24    13470.60 (   0.00%)    15607.10 * -15.86%*
Amean     fault-both-30    17143.99 (   0.00%)    18752.27 (  -9.38%)
Amean     fault-both-32    17743.91 (   0.00%)    21207.54 * -19.52%*

The impact on latency is variable but the search is optimistic and
sensitive to the exact system state.  Success rates are similar but the
major impact is to the rate of scanning

                                5.0.0-rc1      5.0.0-rc1
                            isolmig-v3r15 findfree-v3r16
Compaction migrate scanned    25646769          29507205
Compaction free scanned      201558184         100359571

The free scan rates are reduced by 50%.  The 2-socket reductions for the
free scanner are more dramatic which is a likely reflection that the
machine has more memory.

[dan.carpenter@oracle.com: fix static checker warning]
[vbabka@suse.cz: correct number of pages scanned for lower orders]
Link: http://lkml.kernel.org/r/20190118175136.31341-12-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman e380bebe47 mm, compaction: keep migration source private to a single compaction instance 
Due to either a fast search of the free list or a linear scan, it is
possible for multiple compaction instances to pick the same pageblock
for migration.  This is lucky for one scanner and increased scanning for
all the others.  It also allows a race between requests on which first
allocates the resulting free block.

This patch tests and updates the pageblock skip for the migration
scanner carefully.  When isolating a block, it will check and skip if
the block is already in use.  Once the zone lock is acquired, it will be
rechecked so that only one scanner can set the pageblock skip for
exclusive use.  Any scanner contending will continue with a linear scan.
The skip bit is still set if no pages can be isolated in a range.  While
this may result in redundant scanning, it avoids unnecessarily acquiring
the zone lock when there are no suitable migration sources.

1-socket thpscale
Amean     fault-both-1         0.00 (   0.00%)        0.00 *   0.00%*
Amean     fault-both-3      3390.40 (   0.00%)     3024.41 (  10.80%)
Amean     fault-both-5      5082.28 (   0.00%)     4749.30 (   6.55%)
Amean     fault-both-7      7012.51 (   0.00%)     6454.95 (   7.95%)
Amean     fault-both-12    11346.63 (   0.00%)    10324.83 (   9.01%)
Amean     fault-both-18    15324.19 (   0.00%)    12896.82 *  15.84%*
Amean     fault-both-24    16088.50 (   0.00%)    13470.60 *  16.27%*
Amean     fault-both-30    18723.42 (   0.00%)    17143.99 (   8.44%)
Amean     fault-both-32    18612.01 (   0.00%)    17743.91 (   4.66%)

                                5.0.0-rc1              5.0.0-rc1
                            findmig-v3r15          isolmig-v3r15
Percentage huge-3        89.83 (   0.00%)       92.96 (   3.48%)
Percentage huge-5        91.96 (   0.00%)       93.26 (   1.41%)
Percentage huge-7        92.85 (   0.00%)       93.63 (   0.84%)
Percentage huge-12       92.74 (   0.00%)       92.80 (   0.07%)
Percentage huge-18       91.71 (   0.00%)       91.62 (  -0.10%)
Percentage huge-24       92.13 (   0.00%)       91.50 (  -0.69%)
Percentage huge-30       93.79 (   0.00%)       92.73 (  -1.13%)
Percentage huge-32       91.27 (   0.00%)       91.94 (   0.74%)

This shows a reasonable reduction in latency as multiple compaction
scanners do not operate on the same blocks with a similar allocation
success rate.

Compaction migrate scanned    41093126    25646769

Migration scan rates are reduced by 38%.

Link: http://lkml.kernel.org/r/20190118175136.31341-11-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman 70b44595ea mm, compaction: use free lists to quickly locate a migration source 
The migration scanner is a linear scan of a zone with a potentiall large
search space.  Furthermore, many pageblocks are unusable such as those
filled with reserved pages or partially filled with pages that cannot
migrate.  These still get scanned in the common case of allocating a THP
and the cost accumulates.

The patch uses a partial search of the free lists to locate a migration
source candidate that is marked as MOVABLE when allocating a THP.  It
prefers picking a block with a larger number of free pages already on
the basis that there are fewer pages to migrate to free the entire
block.  The lowest PFN found during searches is tracked as the basis of
the start for the linear search after the first search of the free list
fails.  After the search, the free list is shuffled so that the next
search will not encounter the same page.  If the search fails then the
subsequent searches will be shorter and the linear scanner is used.

If this search fails, or if the request is for a small or
unmovable/reclaimable allocation then the linear scanner is still used.
It is somewhat pointless to use the list search in those cases.  Small
free pages must be used for the search and there is no guarantee that
movable pages are located within that block that are contiguous.

                                     5.0.0-rc1              5.0.0-rc1
                                 noboost-v3r10          findmig-v3r15
Amean     fault-both-3      3771.41 (   0.00%)     3390.40 (  10.10%)
Amean     fault-both-5      5409.05 (   0.00%)     5082.28 (   6.04%)
Amean     fault-both-7      7040.74 (   0.00%)     7012.51 (   0.40%)
Amean     fault-both-12    11887.35 (   0.00%)    11346.63 (   4.55%)
Amean     fault-both-18    16718.19 (   0.00%)    15324.19 (   8.34%)
Amean     fault-both-24    21157.19 (   0.00%)    16088.50 *  23.96%*
Amean     fault-both-30    21175.92 (   0.00%)    18723.42 *  11.58%*
Amean     fault-both-32    21339.03 (   0.00%)    18612.01 *  12.78%*

                                5.0.0-rc1              5.0.0-rc1
                            noboost-v3r10          findmig-v3r15
Percentage huge-3        86.50 (   0.00%)       89.83 (   3.85%)
Percentage huge-5        92.52 (   0.00%)       91.96 (  -0.61%)
Percentage huge-7        92.44 (   0.00%)       92.85 (   0.44%)
Percentage huge-12       92.98 (   0.00%)       92.74 (  -0.25%)
Percentage huge-18       91.70 (   0.00%)       91.71 (   0.02%)
Percentage huge-24       91.59 (   0.00%)       92.13 (   0.60%)
Percentage huge-30       90.14 (   0.00%)       93.79 (   4.04%)
Percentage huge-32       90.03 (   0.00%)       91.27 (   1.37%)

This shows an improvement in allocation latencies with similar
allocation success rates.  While not presented, there was a 31%
reduction in migration scanning and a 8% reduction on system CPU usage.
A 2-socket machine showed similar benefits.

[mgorman@techsingularity.net: several fixes]
  Link: http://lkml.kernel.org/r/20190204120111.GL9565@techsingularity.net
[vbabka@suse.cz: migrate block that was found-fast, some optimisations]
Link: http://lkml.kernel.org/r/20190118175136.31341-10-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <Vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman fd1444b272 mm, compaction: ignore the fragmentation avoidance boost for isolation and compaction 
When pageblocks get fragmented, watermarks are artifically boosted to
reclaim pages to avoid further fragmentation events.  However,
compaction is often either fragmentation-neutral or moving movable pages
away from unmovable/reclaimable pages.  As the true watermarks are
preserved, allow compaction to ignore the boost factor.

The expected impact is very slight as the main benefit is that
compaction is slightly more likely to succeed when the system has been
fragmented very recently.  On both 1-socket and 2-socket machines for
THP-intensive allocation during fragmentation the success rate was
increased by less than 1% which is marginal.  However, detailed tracing
indicated that failure of migration due to a premature ENOMEM triggered
by watermark checks were eliminated.

Link: http://lkml.kernel.org/r/20190118175136.31341-9-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman efe771c760 mm, compaction: always finish scanning of a full pageblock 
When compaction is finishing, it uses a flag to ensure the pageblock is
complete but it makes sense to always complete migration of a pageblock.
Minimally, skip information is based on a pageblock and partially
scanned pageblocks may incur more scanning in the future.  The pageblock
skip handling also becomes more strict later in the series and the hint
is more useful if a complete pageblock was always scanned.

The potentially impacts latency as more scanning is done but it's not a
consistent win or loss as the scanning is not always a high percentage
of the pageblock and sometimes it is offset by future reductions in
scanning.  Hence, the results are not presented this time due to a
misleading mix of gains/losses without any clear pattern.  However, full
scanning of the pageblock is important for later patches.

Link: http://lkml.kernel.org/r/20190118175136.31341-8-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman 806031bb5e mm, migrate: immediately fail migration of a page with no migration handler 
Pages with no migration handler use a fallback handler which sometimes
works and sometimes persistently retries.  A historical example was
blockdev pages but there are others such as odd refcounting when
page->private is used.  These are retried multiple times which is
wasteful during compaction so this patch will fail migration faster
unless the caller specifies MIGRATE_SYNC.

This is not expected to help THP allocation success rates but it did
reduce latencies very slightly in some cases.

1-socket thpfioscale
                                        4.20.0                 4.20.0
                              noreserved-v2r15         failfast-v2r15
Amean     fault-both-1         0.00 (   0.00%)        0.00 *   0.00%*
Amean     fault-both-3      3839.67 (   0.00%)     3833.72 (   0.15%)
Amean     fault-both-5      5177.47 (   0.00%)     4967.15 (   4.06%)
Amean     fault-both-7      7245.03 (   0.00%)     7139.19 (   1.46%)
Amean     fault-both-12    11534.89 (   0.00%)    11326.30 (   1.81%)
Amean     fault-both-18    16241.10 (   0.00%)    16270.70 (  -0.18%)
Amean     fault-both-24    19075.91 (   0.00%)    19839.65 (  -4.00%)
Amean     fault-both-30    22712.11 (   0.00%)    21707.05 (   4.43%)
Amean     fault-both-32    21692.92 (   0.00%)    21968.16 (  -1.27%)

The 2-socket results are not materially different.  Scan rates are
similar as expected.

Link: http://lkml.kernel.org/r/20190118175136.31341-7-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman 4469ab9847 mm, compaction: rename map_pages to split_map_pages 
It's non-obvious that high-order free pages are split into order-0 pages
from the function name.  Fix it.

Link: http://lkml.kernel.org/r/20190118175136.31341-6-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman 40cacbcb32 mm, compaction: remove unnecessary zone parameter in some instances 
A zone parameter is passed into a number of top-level compaction
functions despite the fact that it's already in compact_control.  This
is harmless but it did need an audit to check if zone actually ever
changes meaningfully.  This patches removes the parameter in a number of
top-level functions.  The change could be much deeper but this was
enough to briefly clarify the flow.

No functional change.

Link: http://lkml.kernel.org/r/20190118175136.31341-5-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman 566e54e113 mm, compaction: remove last_migrated_pfn from compact_control 
The last_migrated_pfn field is a bit dubious as to whether it really
helps but either way, the information from it can be inferred without
increasing the size of compact_control so remove the field.

Link: http://lkml.kernel.org/r/20190118175136.31341-4-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman c5943b9c53 mm, compaction: rearrange compact_control 
compact_control spans two cache lines with write-intensive lines on
both.  Rearrange so the most write-intensive fields are in the same
cache line.  This has a negligible impact on the overall performance of
compaction and is more a tidying exercise than anything.

Link: http://lkml.kernel.org/r/20190118175136.31341-3-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: David Rientjes <rientjes@google.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman c5fbd937b6 mm, compaction: shrink compact_control 
Patch series "Increase success rates and reduce latency of compaction", v3.

This series reduces scan rates and success rates of compaction,
primarily by using the free lists to shorten scans, better controlling
of skip information and whether multiple scanners can target the same
block and capturing pageblocks before being stolen by parallel requests.
The series is based on mmotm from January 9th, 2019 with the previous
compaction series reverted.

I'm mostly using thpscale to measure the impact of the series.  The
benchmark creates a large file, maps it, faults it, punches holes in the
mapping so that the virtual address space is fragmented and then tries
to allocate THP.  It re-executes for different numbers of threads.  From
a fragmentation perspective, the workload is relatively benign but it
does stress compaction.

The overall impact on latencies for a 1-socket machine is

				      baseline		      patches
Amean     fault-both-3      3832.09 (   0.00%)     2748.56 *  28.28%*
Amean     fault-both-5      4933.06 (   0.00%)     4255.52 (  13.73%)
Amean     fault-both-7      7017.75 (   0.00%)     6586.93 (   6.14%)
Amean     fault-both-12    11610.51 (   0.00%)     9162.34 *  21.09%*
Amean     fault-both-18    17055.85 (   0.00%)    11530.06 *  32.40%*
Amean     fault-both-24    19306.27 (   0.00%)    17956.13 (   6.99%)
Amean     fault-both-30    22516.49 (   0.00%)    15686.47 *  30.33%*
Amean     fault-both-32    23442.93 (   0.00%)    16564.83 *  29.34%*

The allocation success rates are much improved

			 	 baseline		 patches
Percentage huge-3        85.99 (   0.00%)       97.96 (  13.92%)
Percentage huge-5        88.27 (   0.00%)       96.87 (   9.74%)
Percentage huge-7        85.87 (   0.00%)       94.53 (  10.09%)
Percentage huge-12       82.38 (   0.00%)       98.44 (  19.49%)
Percentage huge-18       83.29 (   0.00%)       99.14 (  19.04%)
Percentage huge-24       81.41 (   0.00%)       97.35 (  19.57%)
Percentage huge-30       80.98 (   0.00%)       98.05 (  21.08%)
Percentage huge-32       80.53 (   0.00%)       97.06 (  20.53%)

That's a nearly perfect allocation success rate.

The biggest impact is on the scan rates

Compaction migrate scanned    55893379    19341254
Compaction free scanned      474739990    11903963

The number of pages scanned for migration was reduced by 65% and the
free scanner was reduced by 97.5%.  So much less work in exchange for
lower latency and better success rates.

The series was also evaluated using a workload that heavily fragments
memory but the benefits there are also significant, albeit not
presented.

It was commented that we should be rethinking scanning entirely and to a
large extent I agree.  However, to achieve that you need a lot of this
series in place first so it's best to make the linear scanners as best
as possible before ripping them out.

This patch (of 22):

The isolate and migrate scanners should never isolate more than a
pageblock of pages so unsigned int is sufficient saving 8 bytes on a
64-bit build.

Link: http://lkml.kernel.org/r/20190118175136.31341-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-03-05 21:07:16 -08:00
Mel Gorman 94b3334cbe mm, page_alloc: fix a division by zero error when boosting watermarks v2 
Yury Norov reported that an arm64 KVM instance could not boot since
after v5.0-rc1 and could addressed by reverting the patches

  1c30844d2d ("mm: reclaim small amounts of memory when an external
  73444bc4d8 ("mm, page_alloc: do not wake kswapd with zone lock held")

The problem is that a division by zero error is possible if boosting
occurs very early in boot if the system has very little memory.  This
patch avoids the division by zero error.

Link: http://lkml.kernel.org/r/20190213143012.GT9565@techsingularity.net
Fixes: 1c30844d2d ("mm: reclaim small amounts of memory when an external fragmentation event occurs")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: Yury Norov <yury.norov@gmail.com>
Tested-by: Yury Norov <yury.norov@gmail.com>
Tested-by: Will Deacon <will.deacon@arm.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-02-21 09:01:00 -08:00
Mel Gorman 73444bc4d8 mm, page_alloc: do not wake kswapd with zone lock held 
syzbot reported the following regression in the latest merge window and
it was confirmed by Qian Cai that a similar bug was visible from a
different context.

  ======================================================
  WARNING: possible circular locking dependency detected
  4.20.0+ #297 Not tainted
  ------------------------------------------------------
  syz-executor0/8529 is trying to acquire lock:
  000000005e7fb829 (&pgdat->kswapd_wait){....}, at:
  __wake_up_common_lock+0x19e/0x330 kernel/sched/wait.c:120

  but task is already holding lock:
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: spin_lock
  include/linux/spinlock.h:329 [inline]
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue_bulk
  mm/page_alloc.c:2548 [inline]
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: __rmqueue_pcplist
  mm/page_alloc.c:3021 [inline]
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue_pcplist
  mm/page_alloc.c:3050 [inline]
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at: rmqueue
  mm/page_alloc.c:3072 [inline]
  000000009bb7bae0 (&(&zone->lock)->rlock){-.-.}, at:
  get_page_from_freelist+0x1bae/0x52a0 mm/page_alloc.c:3491

It appears to be a false positive in that the only way the lock ordering
should be inverted is if kswapd is waking itself and the wakeup
allocates debugging objects which should already be allocated if it's
kswapd doing the waking.  Nevertheless, the possibility exists and so
it's best to avoid the problem.

This patch flags a zone as needing a kswapd using the, surprisingly,
unused zone flag field.  The flag is read without the lock held to do
the wakeup.  It's possible that the flag setting context is not the same
as the flag clearing context or for small races to occur.  However, each
race possibility is harmless and there is no visible degredation in
fragmentation treatment.

While zone->flag could have continued to be unused, there is potential
for moving some existing fields into the flags field instead.
Particularly read-mostly ones like zone->initialized and
zone->contiguous.

Link: http://lkml.kernel.org/r/20190103225712.GJ31517@techsingularity.net
Fixes: 1c30844d2d ("mm: reclaim small amounts of memory when an external fragmentation event occurs")
Reported-by: syzbot+93d94a001cfbce9e60e1@syzkaller.appspotmail.com
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Tested-by: Qian Cai <cai@lca.pw>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2019-01-08 17:15:11 -08:00
Mel Gorman 1c30844d2d mm: reclaim small amounts of memory when an external fragmentation event occurs 
An external fragmentation event was previously described as

    When the page allocator fragments memory, it records the event using
    the mm_page_alloc_extfrag event. If the fallback_order is smaller
    than a pageblock order (order-9 on 64-bit x86) then it's considered
    an event that will cause external fragmentation issues in the future.

The kernel reduces the probability of such events by increasing the
watermark sizes by calling set_recommended_min_free_kbytes early in the
lifetime of the system.  This works reasonably well in general but if
there are enough sparsely populated pageblocks then the problem can still
occur as enough memory is free overall and kswapd stays asleep.

This patch introduces a watermark_boost_factor sysctl that allows a zone
watermark to be temporarily boosted when an external fragmentation causing
events occurs.  The boosting will stall allocations that would decrease
free memory below the boosted low watermark and kswapd is woken if the
calling context allows to reclaim an amount of memory relative to the size
of the high watermark and the watermark_boost_factor until the boost is
cleared.  When kswapd finishes, it wakes kcompactd at the pageblock order
to clean some of the pageblocks that may have been affected by the
fragmentation event.  kswapd avoids any writeback, slab shrinkage and swap
from reclaim context during this operation to avoid excessive system
disruption in the name of fragmentation avoidance.  Care is taken so that
kswapd will do normal reclaim work if the system is really low on memory.

This was evaluated using the same workloads as "mm, page_alloc: Spread
allocations across zones before introducing fragmentation".

1-socket Skylake machine
config-global-dhp__workload_thpfioscale XFS (no special madvise)
4 fio threads, 1 THP allocating thread
--------------------------------------

4.20-rc3 extfrag events < order 9:   804694
4.20-rc3+patch:                      408912 (49% reduction)
4.20-rc3+patch1-4:                    18421 (98% reduction)

                                   4.20.0-rc3             4.20.0-rc3
                                 lowzone-v5r8             boost-v5r8
Amean     fault-base-1      653.58 (   0.00%)      652.71 (   0.13%)
Amean     fault-huge-1        0.00 (   0.00%)      178.93 * -99.00%*

                              4.20.0-rc3             4.20.0-rc3
                            lowzone-v5r8             boost-v5r8
Percentage huge-1        0.00 (   0.00%)        5.12 ( 100.00%)

Note that external fragmentation causing events are massively reduced by
this path whether in comparison to the previous kernel or the vanilla
kernel.  The fault latency for huge pages appears to be increased but that
is only because THP allocations were successful with the patch applied.

1-socket Skylake machine
global-dhp__workload_thpfioscale-madvhugepage-xfs (MADV_HUGEPAGE)
-----------------------------------------------------------------

4.20-rc3 extfrag events < order 9:  291392
4.20-rc3+patch:                     191187 (34% reduction)
4.20-rc3+patch1-4:                   13464 (95% reduction)

thpfioscale Fault Latencies
                                   4.20.0-rc3             4.20.0-rc3
                                 lowzone-v5r8             boost-v5r8
Min       fault-base-1      912.00 (   0.00%)      905.00 (   0.77%)
Min       fault-huge-1      127.00 (   0.00%)      135.00 (  -6.30%)
Amean     fault-base-1     1467.55 (   0.00%)     1481.67 (  -0.96%)
Amean     fault-huge-1     1127.11 (   0.00%)     1063.88 *   5.61%*

                              4.20.0-rc3             4.20.0-rc3
                            lowzone-v5r8             boost-v5r8
Percentage huge-1       77.64 (   0.00%)       83.46 (   7.49%)

As before, massive reduction in external fragmentation events, some jitter
on latencies and an increase in THP allocation success rates.

2-socket Haswell machine
config-global-dhp__workload_thpfioscale XFS (no special madvise)
4 fio threads, 5 THP allocating threads
----------------------------------------------------------------

4.20-rc3 extfrag events < order 9:  215698
4.20-rc3+patch:                     200210 (7% reduction)
4.20-rc3+patch1-4:                   14263 (93% reduction)

                                   4.20.0-rc3             4.20.0-rc3
                                 lowzone-v5r8             boost-v5r8
Amean     fault-base-5     1346.45 (   0.00%)     1306.87 (   2.94%)
Amean     fault-huge-5     3418.60 (   0.00%)     1348.94 (  60.54%)

                              4.20.0-rc3             4.20.0-rc3
                            lowzone-v5r8             boost-v5r8
Percentage huge-5        0.78 (   0.00%)        7.91 ( 910.64%)

There is a 93% reduction in fragmentation causing events, there is a big
reduction in the huge page fault latency and allocation success rate is
higher.

2-socket Haswell machine
global-dhp__workload_thpfioscale-madvhugepage-xfs (MADV_HUGEPAGE)
-----------------------------------------------------------------

4.20-rc3 extfrag events < order 9: 166352
4.20-rc3+patch:                    147463 (11% reduction)
4.20-rc3+patch1-4:                  11095 (93% reduction)

thpfioscale Fault Latencies
                                   4.20.0-rc3             4.20.0-rc3
                                 lowzone-v5r8             boost-v5r8
Amean     fault-base-5     6217.43 (   0.00%)     7419.67 * -19.34%*
Amean     fault-huge-5     3163.33 (   0.00%)     3263.80 (  -3.18%)

                              4.20.0-rc3             4.20.0-rc3
                            lowzone-v5r8             boost-v5r8
Percentage huge-5       95.14 (   0.00%)       87.98 (  -7.53%)

There is a large reduction in fragmentation events with some jitter around
the latencies and success rates.  As before, the high THP allocation
success rate does mean the system is under a lot of pressure.  However, as
the fragmentation events are reduced, it would be expected that the
long-term allocation success rate would be higher.

Link: http://lkml.kernel.org/r/20181123114528.28802-5-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Zi Yan <zi.yan@cs.rutgers.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2018-12-28 12:11:48 -08:00
Mel Gorman 0a79cdad5e mm: use alloc_flags to record if kswapd can wake 
This is a preparation patch that copies the GFP flag __GFP_KSWAPD_RECLAIM
into alloc_flags.  This is a preparation patch only that avoids having to
pass gfp_mask through a long callchain in a future patch.

Note that the setting in the fast path happens in alloc_flags_nofragment()
and it may be claimed that this has nothing to do with ALLOC_NO_FRAGMENT.
That's true in this patch but is not true later so it's done now for
easier review to show where the flag needs to be recorded.

No functional change.

[mgorman@techsingularity.net: ALLOC_KSWAPD flag needs to be applied in the !CONFIG_ZONE_DMA32 case]
  Link: http://lkml.kernel.org/r/20181126143503.GO23260@techsingularity.net
Link: http://lkml.kernel.org/r/20181123114528.28802-4-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Zi Yan <zi.yan@cs.rutgers.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2018-12-28 12:11:48 -08:00
Mel Gorman a921444382 mm: move zone watermark accesses behind an accessor 
This is a preparation patch only, no functional change.

Link: http://lkml.kernel.org/r/20181123114528.28802-3-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Zi Yan <zi.yan@cs.rutgers.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2018-12-28 12:11:48 -08:00
Mel Gorman 6bb154504f mm, page_alloc: spread allocations across zones before introducing fragmentation 
Patch series "Fragmentation avoidance improvements", v5.

It has been noted before that fragmentation avoidance (aka
anti-fragmentation) is not perfect. Given sufficient time or an adverse
workload, memory gets fragmented and the long-term success of high-order
allocations degrades. This series defines an adverse workload, a definition
of external fragmentation events (including serious) ones and a series
that reduces the level of those fragmentation events.

The details of the workload and the consequences are described in more
detail in the changelogs. However, from patch 1, this is a high-level
summary of the adverse workload. The exact details are found in the
mmtests implementation.

The broad details of the workload are as follows;

1. Create an XFS filesystem (not specified in the configuration but done
   as part of the testing for this patch)
2. Start 4 fio threads that write a number of 64K files inefficiently.
   Inefficiently means that files are created on first access and not
   created in advance (fio parameterr create_on_open=1) and fallocate
   is not used (fallocate=none). With multiple IO issuers this creates
   a mix of slab and page cache allocations over time. The total size
   of the files is 150% physical memory so that the slabs and page cache
   pages get mixed
3. Warm up a number of fio read-only threads accessing the same files
   created in step 2. This part runs for the same length of time it
   took to create the files. It'll fault back in old data and further
   interleave slab and page cache allocations. As it's now low on
   memory due to step 2, fragmentation occurs as pageblocks get
   stolen.
4. While step 3 is still running, start a process that tries to allocate
   75% of memory as huge pages with a number of threads. The number of
   threads is based on a (NR_CPUS_SOCKET - NR_FIO_THREADS)/4 to avoid THP
   threads contending with fio, any other threads or forcing cross-NUMA
   scheduling. Note that the test has not been used on a machine with less
   than 8 cores. The benchmark records whether huge pages were allocated
   and what the fault latency was in microseconds
5. Measure the number of events potentially causing external fragmentation,
   the fault latency and the huge page allocation success rate.
6. Cleanup

Overall the series reduces external fragmentation causing events by over 94%
on 1 and 2 socket machines, which in turn impacts high-order allocation
success rates over the long term. There are differences in latencies and
high-order allocation success rates. Latencies are a mixed bag as they
are vulnerable to exact system state and whether allocations succeeded
so they are treated as a secondary metric.

Patch 1 uses lower zones if they are populated and have free memory
	instead of fragmenting a higher zone. It's special cased to
	handle a Normal->DMA32 fallback with the reasons explained
	in the changelog.

Patch 2-4 boosts watermarks temporarily when an external fragmentation
	event occurs. kswapd wakes to reclaim a small amount of old memory
	and then wakes kcompactd on completion to recover the system
	slightly. This introduces some overhead in the slowpath. The level
	of boosting can be tuned or disabled depending on the tolerance
	for fragmentation vs allocation latency.

Patch 5 stalls some movable allocation requests to let kswapd from patch 4
	make some progress. The duration of the stalls is very low but it
	is possible to tune the system to avoid fragmentation events if
	larger stalls can be tolerated.

The bulk of the improvement in fragmentation avoidance is from patches
1-4 but patch 5 can deal with a rare corner case and provides the option
of tuning a system for THP allocation success rates in exchange for
some stalls to control fragmentation.

This patch (of 5):

The page allocator zone lists are iterated based on the watermarks of each
zone which does not take anti-fragmentation into account.  On x86, node 0
may have multiple zones while other nodes have one zone.  A consequence is
that tasks running on node 0 may fragment ZONE_NORMAL even though
ZONE_DMA32 has plenty of free memory.  This patch special cases the
allocator fast path such that it'll try an allocation from a lower local
zone before fragmenting a higher zone.  In this case, stealing of
pageblocks or orders larger than a pageblock are still allowed in the fast
path as they are uninteresting from a fragmentation point of view.

This was evaluated using a benchmark designed to fragment memory before
attempting THP allocations.  It's implemented in mmtests as the following
configurations

configs/config-global-dhp__workload_thpfioscale
configs/config-global-dhp__workload_thpfioscale-defrag
configs/config-global-dhp__workload_thpfioscale-madvhugepage

e.g. from mmtests
./run-mmtests.sh --run-monitor --config configs/config-global-dhp__workload_thpfioscale test-run-1

The broad details of the workload are as follows;

1. Create an XFS filesystem (not specified in the configuration but done
   as part of the testing for this patch).
2. Start 4 fio threads that write a number of 64K files inefficiently.
   Inefficiently means that files are created on first access and not
   created in advance (fio parameter create_on_open=1) and fallocate
   is not used (fallocate=none). With multiple IO issuers this creates
   a mix of slab and page cache allocations over time. The total size
   of the files is 150% physical memory so that the slabs and page cache
   pages get mixed.
3. Warm up a number of fio read-only processes accessing the same files
   created in step 2. This part runs for the same length of time it
   took to create the files. It'll refault old data and further
   interleave slab and page cache allocations. As it's now low on
   memory due to step 2, fragmentation occurs as pageblocks get
   stolen.
4. While step 3 is still running, start a process that tries to allocate
   75% of memory as huge pages with a number of threads. The number of
   threads is based on a (NR_CPUS_SOCKET - NR_FIO_THREADS)/4 to avoid THP
   threads contending with fio, any other threads or forcing cross-NUMA
   scheduling. Note that the test has not been used on a machine with less
   than 8 cores. The benchmark records whether huge pages were allocated
   and what the fault latency was in microseconds.
5. Measure the number of events potentially causing external fragmentation,
   the fault latency and the huge page allocation success rate.
6. Cleanup the test files.

Note that due to the use of IO and page cache that this benchmark is not
suitable for running on large machines where the time to fragment memory
may be excessive.  Also note that while this is one mix that generates
fragmentation that it's not the only mix that generates fragmentation.
Differences in workload that are more slab-intensive or whether SLUB is
used with high-order pages may yield different results.

When the page allocator fragments memory, it records the event using the
mm_page_alloc_extfrag ftrace event.  If the fallback_order is smaller than
a pageblock order (order-9 on 64-bit x86) then it's considered to be an
"external fragmentation event" that may cause issues in the future.
Hence, the primary metric here is the number of external fragmentation
events that occur with order < 9.  The secondary metric is allocation
latency and huge page allocation success rates but note that differences
in latencies and what the success rate also can affect the number of
external fragmentation event which is why it's a secondary metric.

1-socket Skylake machine
config-global-dhp__workload_thpfioscale XFS (no special madvise)
4 fio threads, 1 THP allocating thread
--------------------------------------

4.20-rc3 extfrag events < order 9:   804694
4.20-rc3+patch:                      408912 (49% reduction)

thpfioscale Fault Latencies
                                   4.20.0-rc3             4.20.0-rc3
                                      vanilla           lowzone-v5r8
Amean     fault-base-1      662.92 (   0.00%)      653.58 *   1.41%*
Amean     fault-huge-1        0.00 (   0.00%)        0.00 (   0.00%)

                              4.20.0-rc3             4.20.0-rc3
                                 vanilla           lowzone-v5r8
Percentage huge-1        0.00 (   0.00%)        0.00 (   0.00%)

Fault latencies are slightly reduced while allocation success rates remain
at zero as this configuration does not make any special effort to allocate
THP and fio is heavily active at the time and either filling memory or
keeping pages resident.  However, a 49% reduction of serious fragmentation
events reduces the changes of external fragmentation being a problem in
the future.

Vlastimil asked during review for a breakdown of the allocation types
that are falling back.

vanilla
   3816 MIGRATE_UNMOVABLE
 800845 MIGRATE_MOVABLE
     33 MIGRATE_UNRECLAIMABLE

patch
    735 MIGRATE_UNMOVABLE
 408135 MIGRATE_MOVABLE
     42 MIGRATE_UNRECLAIMABLE

The majority of the fallbacks are due to movable allocations and this is
consistent for the workload throughout the series so will not be presented
again as the primary source of fallbacks are movable allocations.

Movable fallbacks are sometimes considered "ok" to fallback because they
can be migrated.  The problem is that they can fill an
unmovable/reclaimable pageblock causing those allocations to fallback
later and polluting pageblocks with pages that cannot move.  If there is a
movable fallback, it is pretty much guaranteed to affect an
unmovable/reclaimable pageblock and while it might not be enough to
actually cause a unmovable/reclaimable fallback in the future, we cannot
know that in advance so the patch takes the only option available to it.
Hence, it's important to control them.  This point is also consistent
throughout the series and will not be repeated.

1-socket Skylake machine
global-dhp__workload_thpfioscale-madvhugepage-xfs (MADV_HUGEPAGE)
-----------------------------------------------------------------

4.20-rc3 extfrag events < order 9:  291392
4.20-rc3+patch:                     191187 (34% reduction)

thpfioscale Fault Latencies
                                   4.20.0-rc3             4.20.0-rc3
                                      vanilla           lowzone-v5r8
Amean     fault-base-1     1495.14 (   0.00%)     1467.55 (   1.85%)
Amean     fault-huge-1     1098.48 (   0.00%)     1127.11 (  -2.61%)

thpfioscale Percentage Faults Huge
                              4.20.0-rc3             4.20.0-rc3
                                 vanilla           lowzone-v5r8
Percentage huge-1       78.57 (   0.00%)       77.64 (  -1.18%)

Fragmentation events were reduced quite a bit although this is known
to be a little variable. The latencies and allocation success rates
are similar but they were already quite high.

2-socket Haswell machine
config-global-dhp__workload_thpfioscale XFS (no special madvise)
4 fio threads, 5 THP allocating threads
----------------------------------------------------------------

4.20-rc3 extfrag events < order 9:  215698
4.20-rc3+patch:                     200210 (7% reduction)

thpfioscale Fault Latencies
                                   4.20.0-rc3             4.20.0-rc3
                                      vanilla           lowzone-v5r8
Amean     fault-base-5     1350.05 (   0.00%)     1346.45 (   0.27%)
Amean     fault-huge-5     4181.01 (   0.00%)     3418.60 (  18.24%)

                              4.20.0-rc3             4.20.0-rc3
                                 vanilla           lowzone-v5r8
Percentage huge-5        1.15 (   0.00%)        0.78 ( -31.88%)

The reduction of external fragmentation events is slight and this is
partially due to the removal of __GFP_THISNODE in commit ac5b2c1891
("mm: thp: relax __GFP_THISNODE for MADV_HUGEPAGE mappings") as THP
allocations can now spill over to remote nodes instead of fragmenting
local memory.

2-socket Haswell machine
global-dhp__workload_thpfioscale-madvhugepage-xfs (MADV_HUGEPAGE)
-----------------------------------------------------------------

4.20-rc3 extfrag events < order 9: 166352
4.20-rc3+patch:                    147463 (11% reduction)

thpfioscale Fault Latencies
                                   4.20.0-rc3             4.20.0-rc3
                                      vanilla           lowzone-v5r8
Amean     fault-base-5     6138.97 (   0.00%)     6217.43 (  -1.28%)
Amean     fault-huge-5     2294.28 (   0.00%)     3163.33 * -37.88%*

thpfioscale Percentage Faults Huge
                              4.20.0-rc3             4.20.0-rc3
                                 vanilla           lowzone-v5r8
Percentage huge-5       96.82 (   0.00%)       95.14 (  -1.74%)

There was a slight reduction in external fragmentation events although the
latencies were higher.  The allocation success rate is high enough that
the system is struggling and there is quite a lot of parallel reclaim and
compaction activity.  There is also a certain degree of luck on whether
processes start on node 0 or not for this patch but the relevance is
reduced later in the series.

Overall, the patch reduces the number of external fragmentation causing
events so the success of THP over long periods of time would be improved
for this adverse workload.

Link: http://lkml.kernel.org/r/20181123114528.28802-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Zi Yan <zi.yan@cs.rutgers.edu>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2018-12-28 12:11:48 -08:00
Mel Gorman 37355bdc5a sched/numa: Migrate pages to local nodes quicker early in the lifetime of a task 
Automatic NUMA Balancing uses a multi-stage pass to decide whether a page
should migrate to a local node. This filter avoids excessive ping-ponging
if a page is shared or used by threads that migrate cross-node frequently.

Threads inherit both page tables and the preferred node ID from the
parent. This means that threads can trigger hinting faults earlier than
a new task which delays scanning for a number of seconds. As it can be
load balanced very early in its lifetime there can be an unnecessary delay
before it starts migrating thread-local data. This patch migrates private
pages faster early in the lifetime of a thread using the sequence counter
as an identifier of new tasks.

With this patch applied, STREAM performance is the same as 4.17 even though
processes are not spread cross-node prematurely. Other workloads showed
a mix of minor gains and losses. This is somewhat expected most workloads
are not very sensitive to the starting conditions of a process.

                         4.19.0-rc5             4.19.0-rc5                 4.17.0
                         numab-v1r1       fastmigrate-v1r1                vanilla
MB/sec copy     43298.52 (   0.00%)    47335.46 (   9.32%)    47219.24 (   9.06%)
MB/sec scale    30115.06 (   0.00%)    32568.12 (   8.15%)    32527.56 (   8.01%)
MB/sec add      32825.12 (   0.00%)    36078.94 (   9.91%)    35928.02 (   9.45%)
MB/sec triad    32549.52 (   0.00%)    35935.94 (  10.40%)    35969.88 (  10.51%)

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reviewed-by: Rik van Riel <riel@surriel.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Jirka Hladky <jhladky@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Linux-MM <linux-mm@kvack.org>
Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20181001100525.29789-3-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-10-02 11:31:33 +02:00
Mel Gorman efaffc5e40 mm, sched/numa: Remove rate-limiting of automatic NUMA balancing migration 
Rate limiting of page migrations due to automatic NUMA balancing was
introduced to mitigate the worst-case scenario of migrating at high
frequency due to false sharing or slowly ping-ponging between nodes.
Since then, a lot of effort was spent on correctly identifying these
pages and avoiding unnecessary migrations and the safety net may no longer
be required.

Jirka Hladky reported a regression in 4.17 due to a scheduler patch that
avoids spreading STREAM tasks wide prematurely. However, once the task
was properly placed, it delayed migrating the memory due to rate limiting.
Increasing the limit fixed the problem for him.

Currently, the limit is hard-coded and does not account for the real
capabilities of the hardware. Even if an estimate was attempted, it would
not properly account for the number of memory controllers and it could
not account for the amount of bandwidth used for normal accesses. Rather
than fudging, this patch simply eliminates the rate limiting.

However, Jirka reports that a STREAM configuration using multiple
processes achieved similar performance to 4.16. In local tests, this patch
improved performance of STREAM relative to the baseline but it is somewhat
machine-dependent. Most workloads show little or not performance difference
implying that there is not a heavily reliance on the throttling mechanism
and it is safe to remove.

STREAM on 2-socket machine
                         4.19.0-rc5             4.19.0-rc5
                         numab-v1r1       noratelimit-v1r1
MB/sec copy     43298.52 (   0.00%)    44673.38 (   3.18%)
MB/sec scale    30115.06 (   0.00%)    31293.06 (   3.91%)
MB/sec add      32825.12 (   0.00%)    34883.62 (   6.27%)
MB/sec triad    32549.52 (   0.00%)    34906.60 (   7.24%

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reviewed-by: Rik van Riel <riel@surriel.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Jirka Hladky <jhladky@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Linux-MM <linux-mm@kvack.org>
Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20181001100525.29789-2-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-10-02 11:31:14 +02:00
Mel Gorman 05cbdf4f5c sched/numa: Limit the conditions where scan period is reset 
migrate_task_rq_fair() resets the scan rate for NUMA balancing on every
cross-node migration. In the event of excessive load balancing due to
saturation, this may result in the scan rate being pegged at maximum and
further overloading the machine.

This patch only resets the scan if NUMA balancing is active, a preferred
node has been selected and the task is being migrated from the preferred
node as these are the most harmful. For example, a migration to the preferred
node does not justify a faster scan rate. Similarly, a migration between two
nodes that are not preferred is probably bouncing due to over-saturation of
the machine.  In that case, scanning faster and trapping more NUMA faults
will further overload the machine.

Specjbb2005 results (8 warehouses)
Higher bops are better

2 Socket - 2  Node Haswell - X86
JVMS  Prev    Current  %Change
4     203370  205332   0.964744
1     328431  319785   -2.63252

2 Socket - 4 Node Power8 - PowerNV
JVMS  Prev    Current  %Change
1     206070  206585   0.249915

2 Socket - 2  Node Power9 - PowerNV
JVMS  Prev    Current  %Change
4     188386  189162   0.41192
1     201566  213760   6.04963

4 Socket - 4  Node Power7 - PowerVM
JVMS  Prev     Current  %Change
8     59157.4  58736.8  -0.710985
1     105495   105419   -0.0720413

Some events stats before and after applying the patch.

perf stats 8th warehouse Multi JVM 2 Socket - 2  Node Haswell - X86
Event                     Before          After
cs                        13,825,492      14,285,708
migrations                1,152,509       1,180,621
faults                    371,948         339,114
cache-misses              55,654,206,041  55,205,631,894
sched:sched_move_numa     1,856           843
sched:sched_stick_numa    4               6
sched:sched_swap_numa     428             219
migrate:mm_migrate_pages  898             365

vmstat 8th warehouse Multi JVM 2 Socket - 2  Node Haswell - X86
Event                   Before  After
numa_hint_faults        57146   26907
numa_hint_faults_local  51612   24279
numa_hit                238164  239771
numa_huge_pte_updates   16      0
numa_interleave         63      68
numa_local              238085  239688
numa_other              79      83
numa_pages_migrated     883     363
numa_pte_updates        67540   27415

perf stats 8th warehouse Single JVM 2 Socket - 2  Node Haswell - X86
Event                     Before          After
cs                        3,288,525       3,202,779
migrations                38,652          37,186
faults                    111,678         106,076
cache-misses              12,111,197,376  12,024,873,744
sched:sched_move_numa     900             931
sched:sched_stick_numa    0               0
sched:sched_swap_numa     5               1
migrate:mm_migrate_pages  714             637

vmstat 8th warehouse Single JVM 2 Socket - 2  Node Haswell - X86
Event                   Before  After
numa_hint_faults        18572   17409
numa_hint_faults_local  14850   14367
numa_hit                73197   73953
numa_huge_pte_updates   11      20
numa_interleave         25      25
numa_local              73138   73892
numa_other              59      61
numa_pages_migrated     712     668
numa_pte_updates        24021   27276

perf stats 8th warehouse Multi JVM 2 Socket - 2  Node Power9 - PowerNV
Event                     Before       After
cs                        8,451,543    8,474,013
migrations                202,804      254,934
faults                    310,024      320,506
cache-misses              253,522,507  110,580,458
sched:sched_move_numa     213          725
sched:sched_stick_numa    0            0
sched:sched_swap_numa     2            7
migrate:mm_migrate_pages  88           145

vmstat 8th warehouse Multi JVM 2 Socket - 2  Node Power9 - PowerNV
Event                   Before  After
numa_hint_faults        11830   22797
numa_hint_faults_local  11301   21539
numa_hit                90038   89308
numa_huge_pte_updates   0       0
numa_interleave         855     865
numa_local              89796   88955
numa_other              242     353
numa_pages_migrated     88      149
numa_pte_updates        12039   22930

perf stats 8th warehouse Single JVM 2 Socket - 2  Node Power9 - PowerNV
Event                     Before     After
cs                        2,049,153  2,195,628
migrations                11,405     11,179
faults                    162,309    149,656
cache-misses              7,203,343  8,117,515
sched:sched_move_numa     22         49
sched:sched_stick_numa    0          0
sched:sched_swap_numa     0          0
migrate:mm_migrate_pages  1          5

vmstat 8th warehouse Single JVM 2 Socket - 2  Node Power9 - PowerNV
Event                   Before  After
numa_hint_faults        1693    3577
numa_hint_faults_local  1669    3476
numa_hit                25177   26142
numa_huge_pte_updates   0       0
numa_interleave         194     358
numa_local              24993   26042
numa_other              184     100
numa_pages_migrated     1       5
numa_pte_updates        1577    3587

perf stats 8th warehouse Multi JVM 4 Socket - 4  Node Power7 - PowerVM
Event                     Before           After
cs                        94,515,937       100,602,296
migrations                4,203,554        4,135,630
faults                    832,697          789,256
cache-misses              226,248,698,331  226,160,621,058
sched:sched_move_numa     1,730            1,366
sched:sched_stick_numa    14               16
sched:sched_swap_numa     432              374
migrate:mm_migrate_pages  1,398            1,350

vmstat 8th warehouse Multi JVM 4 Socket - 4  Node Power7 - PowerVM
Event                   Before  After
numa_hint_faults        80079   47857
numa_hint_faults_local  68620   39768
numa_hit                241187  240165
numa_huge_pte_updates   0       0
numa_interleave         0       0
numa_local              241186  240165
numa_other              1       0
numa_pages_migrated     1347    1224
numa_pte_updates        80729   48354

perf stats 8th warehouse Single JVM 4 Socket - 4  Node Power7 - PowerVM
Event                     Before          After
cs                        63,704,961      58,515,496
migrations                573,404         564,845
faults                    230,878         245,807
cache-misses              76,568,222,781  73,603,757,976
sched:sched_move_numa     509             996
sched:sched_stick_numa    31              10
sched:sched_swap_numa     182             193
migrate:mm_migrate_pages  541             646

vmstat 8th warehouse Single JVM 4 Socket - 4  Node Power7 - PowerVM
Event                   Before  After
numa_hint_faults        8501    13422
numa_hint_faults_local  2960    5619
numa_hit                35526   36118
numa_huge_pte_updates   0       0
numa_interleave         0       0
numa_local              35526   36116
numa_other              0       2
numa_pages_migrated     539     616
numa_pte_updates        8433    13374

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Jirka Hladky <jhladky@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1537552141-27815-5-git-send-email-srikar@linux.vnet.ibm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-10-02 09:42:24 +02:00
Mel Gorman 37a4094e82 mremap: remove LATENCY_LIMIT from mremap to reduce the number of TLB shootdowns 
Commit 5d1904204c ("mremap: fix race between mremap() and page
cleanning") fixed races between mremap and other operations for both
file-backed and anonymous mappings.  The file-backed was the most
critical as it allowed the possibility that data could be changed on a
physical page after page_mkclean returned which could trigger data loss
or data integrity issues.

A customer reported that the cost of the TLBs for anonymous regressions
was excessive and resulting in a 30-50% drop in performance overall
since this commit on a microbenchmark.  Unfortunately I neither have
access to the test-case nor can I describe what it does other than
saying that mremap operations dominate heavily.

This patch removes the LATENCY_LIMIT to handle TLB flushes on a PMD
boundary instead of every 64 pages to reduce the number of TLB
shootdowns by a factor of 8 in the ideal case.  LATENCY_LIMIT was almost
certainly used originally to limit the PTL hold times but the latency
savings are likely offset by the cost of IPIs in many cases.  This patch
is not reported to completely restore performance but gets it within an
acceptable percentage.  The given metric here is simply described as
"higher is better".

Baseline that was known good
002:  Metric:       91.05
004:  Metric:      109.45
008:  Metric:       73.08
016:  Metric:       58.14
032:  Metric:       61.09
064:  Metric:       57.76
128:  Metric:       55.43

Current
001:  Metric:       54.98
002:  Metric:       56.56
004:  Metric:       41.22
008:  Metric:       35.96
016:  Metric:       36.45
032:  Metric:       35.71
064:  Metric:       35.73
128:  Metric:       34.96

With patch
001:  Metric:       61.43
002:  Metric:       81.64
004:  Metric:       67.92
008:  Metric:       51.67
016:  Metric:       50.47
032:  Metric:       52.29
064:  Metric:       50.01
128:  Metric:       49.04

So for low threads, it's not restored but for larger number of threads,
it's closer to the "known good" baseline.

Using a different mremap-intensive workload that is not representative
of the real workload there is little difference observed outside of
noise in the headline metrics However, the TLB shootdowns are reduced by
11% on average and at the peak, TLB shootdowns were reduced by 21%.
Interrupts were sampled every second while the workload ran to get those
figures.  It's known that the figures will vary as the
non-representative load is non-deterministic.

An alternative patch was posted that should have significantly reduced
the TLB flushes but unfortunately it does not perform as well as this
version on the customer test case.  If revisited, the two patches can
stack on top of each other.

Link: http://lkml.kernel.org/r/20180606183803.k7qaw2xnbvzshv34@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Nadav Amit <nadav.amit@gmail.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Aaron Lu <aaron.lu@intel.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>
 2018-06-15 07:55:24 +09:00
Mel Gorman 1378447598 sched/numa: Stagger NUMA balancing scan periods for new threads 
Threads share an address space and each can change the protections of the
same address space to trap NUMA faults. This is redundant and potentially
counter-productive as any thread doing the update will suffice. Potentially
only one thread is required but that thread may be idle or it may not have
any locality concerns and pick an unsuitable scan rate.

This patch uses independent scan period but they are staggered based on
the number of address space users when the thread is created.  The intent
is that threads will avoid scanning at the same time and have a chance
to adapt their scan rate later if necessary. This reduces the total scan
activity early in the lifetime of the threads.

The different in headline performance across a range of machines and
workloads is marginal but the system CPU usage is reduced as well as overall
scan activity.  The following is the time reported by NAS Parallel Benchmark
using unbound openmp threads and a D size class:

			      4.17.0-rc1             4.17.0-rc1
				 vanilla           stagger-v1r1
	Time bt.D      442.77 (   0.00%)      419.70 (   5.21%)
	Time cg.D      171.90 (   0.00%)      180.85 (  -5.21%)
	Time ep.D       33.10 (   0.00%)       32.90 (   0.60%)
	Time is.D        9.59 (   0.00%)        9.42 (   1.77%)
	Time lu.D      306.75 (   0.00%)      304.65 (   0.68%)
	Time mg.D       54.56 (   0.00%)       52.38 (   4.00%)
	Time sp.D     1020.03 (   0.00%)      903.77 (  11.40%)
	Time ua.D      400.58 (   0.00%)      386.49 (   3.52%)

Note it's not a universal win but we have no prior knowledge of which
thread matters but the number of threads created often exceeds the size
of the node when the threads are not bound. However, there is a reducation
of overall system CPU usage:

				    4.17.0-rc1             4.17.0-rc1
				       vanilla           stagger-v1r1
	sys-time-bt.D         48.78 (   0.00%)       48.22 (   1.15%)
	sys-time-cg.D         25.31 (   0.00%)       26.63 (  -5.22%)
	sys-time-ep.D          1.65 (   0.00%)        0.62 (  62.42%)
	sys-time-is.D         40.05 (   0.00%)       24.45 (  38.95%)
	sys-time-lu.D         37.55 (   0.00%)       29.02 (  22.72%)
	sys-time-mg.D         47.52 (   0.00%)       34.92 (  26.52%)
	sys-time-sp.D        119.01 (   0.00%)      109.05 (   8.37%)
	sys-time-ua.D         51.52 (   0.00%)       45.13 (  12.40%)

NUMA scan activity is also reduced:

	NUMA alloc local               1042828     1342670
	NUMA base PTE updates        140481138    93577468
	NUMA huge PMD updates           272171      180766
	NUMA page range updates      279832690   186129660
	NUMA hint faults               1395972     1193897
	NUMA hint local faults          877925      855053
	NUMA hint local percent             62          71
	NUMA pages migrated           12057909     9158023

Similar observations are made for other thread-intensive workloads. System
CPU usage is lower even though the headline gains in performance tend to be
small. For example, specjbb 2005 shows almost no difference in performance
but scan activity is reduced by a third on a 4-socket box. I didn't find
a workload (thread intensive or otherwise) that suffered badly.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Link: http://lkml.kernel.org/r/20180504154109.mvrha2qo5wdl65vr@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-05-14 09:12:24 +02:00
Mel Gorman 789ba28013 Revert "sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine()" 
This reverts commit 7347fc87df.

Srikar Dronamra pointed out that while the commit in question did show
a performance improvement on ppc64, it did so at the cost of disabling
active CPU migration by automatic NUMA balancing which was not the intent.
The issue was that a serious flaw in the logic failed to ever active balance
if SD_WAKE_AFFINE was disabled on scheduler domains. Even when it's enabled,
the logic is still bizarre and against the original intent.

Investigation showed that fixing the patch in either the way he suggested,
using the correct comparison for jiffies values or introducing a new
numa_migrate_deferred variable in task_struct all perform similarly to a
revert with a mix of gains and losses depending on the workload, machine
and socket count.

The original intent of the commit was to handle a problem whereby
wake_affine, idle balancing and automatic NUMA balancing disagree on the
appropriate placement for a task. This was particularly true for cases where
a single task was a massive waker of tasks but where wake_wide logic did
not apply.  This was particularly noticeable when a futex (a barrier) woke
all worker threads and tried pulling the wakees to the waker nodes. In that
specific case, it could be handled by tuning MPI or openMP appropriately,
but the behavior is not illogical and was worth attempting to fix. However,
the approach was wrong. Given that we're at rc4 and a fix is not obvious,
it's better to play safe, revert this commit and retry later.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: efault@gmx.de
Cc: ggherdovich@suse.cz
Cc: hpa@zytor.com
Cc: matt@codeblueprint.co.uk
Cc: mpe@ellerman.id.au
Link: http://lkml.kernel.org/r/20180509163115.6fnnyeg4vdm2ct4v@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-05-12 08:37:56 +02:00
Mel Gorman 09a913a7a9 sched/numa: avoid trapping faults and attempting migration of file-backed dirty pages 
change_pte_range is called from task work context to mark PTEs for
receiving NUMA faulting hints.  If the marked pages are dirty then
migration may fail.  Some filesystems cannot migrate dirty pages without
blocking so are skipped in MIGRATE_ASYNC mode which just wastes CPU.
Even when they can, it can be a waste of cycles when the pages are
shared forcing higher scan rates.  This patch avoids marking shared
dirty pages for hinting faults but also will skip a migration if the
page was dirtied after the scanner updated a clean page.

This is most noticeable running the NASA Parallel Benchmark when backed
by btrfs, the default root filesystem for some distributions, but also
noticeable when using XFS.

The following are results from a 4-socket machine running a 4.16-rc4
kernel with some scheduler patches that are pending for the next merge
window.

                        4.16.0-rc4             4.16.0-rc4
                 schedtip-20180309          nodirty-v1
  Time cg.D      459.07 (   0.00%)      444.21 (   3.24%)
  Time ep.D       76.96 (   0.00%)       77.69 (  -0.95%)
  Time is.D       25.55 (   0.00%)       27.85 (  -9.00%)
  Time lu.D      601.58 (   0.00%)      596.87 (   0.78%)
  Time mg.D      107.73 (   0.00%)      108.22 (  -0.45%)

is.D regresses slightly in terms of absolute time but note that that
particular load varies quite a bit from run to run.  The more relevant
observation is the total system CPU usage.

            4.16.0-rc4  4.16.0-rc4
          schedtip-20180309 nodirty-v1
  User        71471.91    70627.04
  System      11078.96     8256.13
  Elapsed       661.66      632.74

That is a substantial drop in system CPU usage and overall the workload
completes faster.  The NUMA balancing statistics are also interesting

  NUMA base PTE updates        111407972   139848884
  NUMA huge PMD updates           206506      264869
  NUMA page range updates      217139044   275461812
  NUMA hint faults               4300924     3719784
  NUMA hint local faults         3012539     3416618
  NUMA hint local percent             70          91
  NUMA pages migrated            1517487     1358420

While more PTEs are scanned due to changes in what faults are gathered,
it's clear that a far higher percentage of faults are local as the bulk
of the remote hits were dirty pages that, in this case with btrfs, had
no chance of migrating.

The following is a comparison when using XFS as that is a more realistic
filesystem choice for a data partition

                        4.16.0-rc4             4.16.0-rc4
                 schedtip-20180309          nodirty-v1r47
  Time cg.D      485.28 (   0.00%)      442.62 (   8.79%)
  Time ep.D       77.68 (   0.00%)       77.54 (   0.18%)
  Time is.D       26.44 (   0.00%)       24.79 (   6.24%)
  Time lu.D      597.46 (   0.00%)      597.11 (   0.06%)
  Time mg.D      142.65 (   0.00%)      105.83 (  25.81%)

That is a reasonable gain on two relatively long-lived workloads.  While
not presented, there is also a substantial drop in system CPu usage and
the NUMA balancing stats show similar improvements in locality as btrfs
did.

Link: http://lkml.kernel.org/r/20180326094334.zserdec62gwmmfqf@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reviewed-by: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2018-04-11 10:28:31 -07:00
Mel Gorman 7347fc87df sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine() 
If wake_affine() pulls a task to another node for any reason and the node is
no longer preferred then temporarily stop automatic NUMA balancing pulling
the task back. Otherwise, tasks with a strong waker/wakee relationship
may constantly fight automatic NUMA balancing over where a task should
be placed.

Once again netperf is interesting here. The performance barely changes
but automatic NUMA balancing is interesting:

 Hmean     send-64         354.67 (   0.00%)      352.15 (  -0.71%)
 Hmean     send-128        702.91 (   0.00%)      693.84 (  -1.29%)
 Hmean     send-256       1350.07 (   0.00%)     1344.19 (  -0.44%)
 Hmean     send-1024      5124.38 (   0.00%)     4941.24 (  -3.57%)
 Hmean     send-2048      9687.44 (   0.00%)     9624.45 (  -0.65%)
 Hmean     send-3312     14577.64 (   0.00%)    14514.35 (  -0.43%)
 Hmean     send-4096     16393.62 (   0.00%)    16488.30 (   0.58%)
 Hmean     send-8192     26877.26 (   0.00%)    26431.63 (  -1.66%)
 Hmean     send-16384    38683.43 (   0.00%)    38264.91 (  -1.08%)
 Hmean     recv-64         354.67 (   0.00%)      352.15 (  -0.71%)
 Hmean     recv-128        702.91 (   0.00%)      693.84 (  -1.29%)
 Hmean     recv-256       1350.07 (   0.00%)     1344.19 (  -0.44%)
 Hmean     recv-1024      5124.38 (   0.00%)     4941.24 (  -3.57%)
 Hmean     recv-2048      9687.43 (   0.00%)     9624.45 (  -0.65%)
 Hmean     recv-3312     14577.59 (   0.00%)    14514.35 (  -0.43%)
 Hmean     recv-4096     16393.55 (   0.00%)    16488.20 (   0.58%)
 Hmean     recv-8192     26876.96 (   0.00%)    26431.29 (  -1.66%)
 Hmean     recv-16384    38682.41 (   0.00%)    38263.94 (  -1.08%)

 NUMA alloc hit                 1465986     1423090
 NUMA alloc miss                      0           0
 NUMA interleave hit                  0           0
 NUMA alloc local               1465897     1423003
 NUMA base PTE updates             1473        1420
 NUMA huge PMD updates                0           0
 NUMA page range updates           1473        1420
 NUMA hint faults                  1383        1312
 NUMA hint local faults             451         124
 NUMA hint local percent             32           9

There is a slight degrading in performance but there are slightly fewer
NUMA faults. There is a large drop in the percentage of local faults but
the bulk of migrations for netperf are in small shared libraries so it's
reflecting the fact that automatic NUMA balancing has backed off. This is
a case where despite wake_affine() and automatic NUMA balancing fighting
for placement that there is a marginal benefit to rescheduling to local
data quickly. However, it should be noted that wake_affine() and automatic
NUMA balancing fighting each other constantly is undesirable.

However, the benefit in other cases is large. This is the result for NAS
with the D class sizing on a 4-socket machine:

 nas-mpi
                           4.15.0                 4.15.0
                     sdnuma-v1r23       delayretry-v1r23
 Time cg.D      557.00 (   0.00%)      431.82 (  22.47%)
 Time ep.D       77.83 (   0.00%)       79.01 (  -1.52%)
 Time is.D       26.46 (   0.00%)       26.64 (  -0.68%)
 Time lu.D      727.14 (   0.00%)      597.94 (  17.77%)
 Time mg.D      191.35 (   0.00%)      146.85 (  23.26%)

               4.15.0      4.15.0
         sdnuma-v1r23delayretry-v1r23
 User        75665.20    70413.30
 System      20321.59     8861.67
 Elapsed       766.13      634.92

 Minor Faults                  16528502     7127941
 Major Faults                      4553        5068
 NUMA alloc local               6963197     6749135
 NUMA base PTE updates        366409093   107491434
 NUMA huge PMD updates           687556      198880
 NUMA page range updates      718437765   209317994
 NUMA hint faults              13643410     4601187
 NUMA hint local faults         9212593     3063996
 NUMA hint local percent             67          66

Note the massive reduction in system CPU usage even though the percentage
of local faults is barely affected. There is a massive reduction in the
number of PTE updates showing that automatic NUMA balancing has backed off.
A critical observation is also that there is a massive reduction in minor
faults which is due to far fewer NUMA hinting faults being trapped.

There were questions on NAS OMP and how it behaved related to threads
being bound to CPUs. First, there are more gains than losses with this
patch applied and a reduction in system CPU usage:

nas-omp
                      4.16.0-rc1             4.16.0-rc1
                     sdnuma-v2r1        delayretry-v2r1
Time bt.D      436.71 (   0.00%)      430.05 (   1.53%)
Time cg.D      201.02 (   0.00%)      180.87 (  10.02%)
Time ep.D       32.84 (   0.00%)       32.68 (   0.49%)
Time is.D        9.63 (   0.00%)        9.64 (  -0.10%)
Time lu.D      331.20 (   0.00%)      304.80 (   7.97%)
Time mg.D       54.87 (   0.00%)       52.72 (   3.92%)
Time sp.D     1108.78 (   0.00%)      917.10 (  17.29%)
Time ua.D      378.81 (   0.00%)      398.83 (  -5.28%)

          4.16.0-rc1  4.16.0-rc1
         sdnuma-v2r1delayretry-v2r1
User       305633.08   296751.91
System        451.75      357.80
Elapsed      2595.73     2368.13

However, it does not close the gap between binding and being unbound. There
is negligible difference between the performance of the baseline and a
patched kernel when threads are bound so it is not presented here:

                      4.16.0-rc1             4.16.0-rc1
                 delayretry-bind     delayretry-unbound
Time bt.D      385.02 (   0.00%)      430.05 ( -11.70%)
Time cg.D      144.02 (   0.00%)      180.87 ( -25.59%)
Time ep.D       32.85 (   0.00%)       32.68 (   0.52%)
Time is.D       10.52 (   0.00%)        9.64 (   8.37%)
Time lu.D      285.31 (   0.00%)      304.80 (  -6.83%)
Time mg.D       43.21 (   0.00%)       52.72 ( -22.01%)
Time sp.D      820.24 (   0.00%)      917.10 ( -11.81%)
Time ua.D      337.09 (   0.00%)      398.83 ( -18.32%)

          4.16.0-rc1  4.16.0-rc1
        delayretry-binddelayretry-unbound
User       277731.25   296751.91
System        261.29      357.80
Elapsed      2100.55     2368.13

Unfortunately, while performance is improved by the patch, there is still
quite a long way to go before it's equivalent to hard binding.

Other workloads like hackbench, tbench, dbench and schbench are barely
affected. dbench shows a mix of gains and losses depending on the machine
although in general, the results are more stable.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Giovanni Gherdovich <ggherdovich@suse.cz>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180213133730.24064-7-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-02-21 08:49:45 +01:00
Mel Gorman 2c83362734 sched/fair: Consider SD_NUMA when selecting the most idle group to schedule on 
find_idlest_group() compares a local group with each other group to select
the one that is most idle. When comparing groups in different NUMA domains,
a very slight imbalance is enough to select a remote NUMA node even if the
runnable load on both groups is 0 or close to 0. This ignores the cost of
remote accesses entirely and is a problem when selecting the CPU for a
newly forked task to run on.  This is problematic when a forking server
is almost guaranteed to run on a remote node incurring numerous remote
accesses and potentially causing automatic NUMA balancing to try migrate
the task back or migrate the data to another node. Similar weirdness is
observed if a basic shell command pipes output to another as each process
in the pipeline is likely to start on different nodes and then get adjusted
later by wake_affine().

This patch adds imbalance to remote domains when considering whether to
select CPUs from remote domains. If the local domain is selected, imbalance
will still be used to try select a CPU from a lower scheduler domain's group
instead of stacking tasks on the same CPU.

A variety of workloads and machines were tested and as expected, there is no
difference on UMA. The difference on NUMA can be dramatic. This is a comparison
of elapsed times running the git regression test suite. It's fork-intensive with
short-lived processes:

                                  4.15.0                 4.15.0
                            noexit-v1r23           sdnuma-v1r23
 Elapsed min          1706.06 (   0.00%)     1435.94 (  15.83%)
 Elapsed mean         1709.53 (   0.00%)     1436.98 (  15.94%)
 Elapsed stddev          2.16 (   0.00%)        1.01 (  53.38%)
 Elapsed coeffvar        0.13 (   0.00%)        0.07 (  44.54%)
 Elapsed max          1711.59 (   0.00%)     1438.01 (  15.98%)

               4.15.0      4.15.0
         noexit-v1r23 sdnuma-v1r23
 User         5434.12     5188.41
 System       4878.77     3467.09
 Elapsed     10259.06     8624.21

That shows a considerable reduction in elapsed times. It's important to
note that automatic NUMA balancing does not affect this load as processes
are too short-lived.

There is also a noticable impact on hackbench such as this example using
processes and pipes:

 hackbench-process-pipes
                               4.15.0                 4.15.0
                         noexit-v1r23           sdnuma-v1r23
 Amean     1        1.0973 (   0.00%)      0.9393 (  14.40%)
 Amean     4        1.3427 (   0.00%)      1.3730 (  -2.26%)
 Amean     7        1.4233 (   0.00%)      1.6670 ( -17.12%)
 Amean     12       3.0250 (   0.00%)      3.3013 (  -9.13%)
 Amean     21       9.0860 (   0.00%)      9.5343 (  -4.93%)
 Amean     30      14.6547 (   0.00%)     13.2433 (   9.63%)
 Amean     48      22.5447 (   0.00%)     20.4303 (   9.38%)
 Amean     79      29.2010 (   0.00%)     26.7853 (   8.27%)
 Amean     110     36.7443 (   0.00%)     35.8453 (   2.45%)
 Amean     141     45.8533 (   0.00%)     42.6223 (   7.05%)
 Amean     172     55.1317 (   0.00%)     50.6473 (   8.13%)
 Amean     203     64.4420 (   0.00%)     58.3957 (   9.38%)
 Amean     234     73.2293 (   0.00%)     67.1047 (   8.36%)
 Amean     265     80.5220 (   0.00%)     75.7330 (   5.95%)
 Amean     296     88.7567 (   0.00%)     82.1533 (   7.44%)

It's not a universal win as there are occasions when spreading wide and
quickly is a benefit but it's more of a win than it is a loss. For other
workloads, there is little difference but netperf is interesting. Without
the patch, the server and client starts on different nodes but quickly get
migrated due to wake_affine. Hence, the difference is overall performance
is marginal but detectable:

                                      4.15.0                 4.15.0
                                noexit-v1r23           sdnuma-v1r23
 Hmean     send-64         349.09 (   0.00%)      354.67 (   1.60%)
 Hmean     send-128        699.16 (   0.00%)      702.91 (   0.54%)
 Hmean     send-256       1316.34 (   0.00%)     1350.07 (   2.56%)
 Hmean     send-1024      5063.99 (   0.00%)     5124.38 (   1.19%)
 Hmean     send-2048      9705.19 (   0.00%)     9687.44 (  -0.18%)
 Hmean     send-3312     14359.48 (   0.00%)    14577.64 (   1.52%)
 Hmean     send-4096     16324.20 (   0.00%)    16393.62 (   0.43%)
 Hmean     send-8192     26112.61 (   0.00%)    26877.26 (   2.93%)
 Hmean     send-16384    37208.44 (   0.00%)    38683.43 (   3.96%)
 Hmean     recv-64         349.09 (   0.00%)      354.67 (   1.60%)
 Hmean     recv-128        699.16 (   0.00%)      702.91 (   0.54%)
 Hmean     recv-256       1316.34 (   0.00%)     1350.07 (   2.56%)
 Hmean     recv-1024      5063.99 (   0.00%)     5124.38 (   1.19%)
 Hmean     recv-2048      9705.16 (   0.00%)     9687.43 (  -0.18%)
 Hmean     recv-3312     14359.42 (   0.00%)    14577.59 (   1.52%)
 Hmean     recv-4096     16323.98 (   0.00%)    16393.55 (   0.43%)
 Hmean     recv-8192     26111.85 (   0.00%)    26876.96 (   2.93%)
 Hmean     recv-16384    37206.99 (   0.00%)    38682.41 (   3.97%)

However, what is very interesting is how automatic NUMA balancing behaves.
Each netperf instance runs long enough for balancing to activate:

 NUMA base PTE updates             4620        1473
 NUMA huge PMD updates                0           0
 NUMA page range updates           4620        1473
 NUMA hint faults                  4301        1383
 NUMA hint local faults            1309         451
 NUMA hint local percent             30          32
 NUMA pages migrated               1335         491
 AutoNUMA cost                      21%          6%

There is an unfortunate number of remote faults although tracing indicated
that the vast majority are in shared libraries. However, the tendency to
start tasks on the same node if there is capacity means that there were
far fewer PTE updates and faults incurred overall.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Giovanni Gherdovich <ggherdovich@suse.cz>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180213133730.24064-6-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-02-21 08:49:43 +01:00
Mel Gorman 082f764a2f sched/fair: Do not migrate on wake_affine_weight() if weights are equal 
wake_affine_weight() will consider migrating a task to, or near, the current
CPU if there is a load imbalance. If the CPUs share LLC then either CPU
is valid as a search-for-idle-sibling target and equally appropriate for
stacking two tasks on one CPU if an idle sibling is unavailable. If they do
not share cache then a cross-node migration potentially impacts locality
so while they are equal from a CPU capacity point of view, they are not
equal in terms of memory locality. In either case, it's more appropriate
to migrate only if there is a difference in their effective load.

This patch modifies wake_affine_weight() to only consider migrating a task
if there is a load imbalance for normal wakeups but will allow potential
stacking if the loads are equal and it's a sync wakeup.

For the most part, the different in performance is marginal. For example,
on a 4-socket server running netperf UDP_STREAM on localhost the differences
are as follows:

                                      4.15.0                 4.15.0
                                       16rc0          noequal-v1r23
 Hmean     send-64         355.47 (   0.00%)      349.50 (  -1.68%)
 Hmean     send-128        697.98 (   0.00%)      693.35 (  -0.66%)
 Hmean     send-256       1328.02 (   0.00%)     1318.77 (  -0.70%)
 Hmean     send-1024      5051.83 (   0.00%)     5051.11 (  -0.01%)
 Hmean     send-2048      9637.02 (   0.00%)     9601.34 (  -0.37%)
 Hmean     send-3312     14355.37 (   0.00%)    14414.51 (   0.41%)
 Hmean     send-4096     16464.97 (   0.00%)    16301.37 (  -0.99%)
 Hmean     send-8192     26722.42 (   0.00%)    26428.95 (  -1.10%)
 Hmean     send-16384    38137.81 (   0.00%)    38046.11 (  -0.24%)
 Hmean     recv-64         355.47 (   0.00%)      349.50 (  -1.68%)
 Hmean     recv-128        697.98 (   0.00%)      693.35 (  -0.66%)
 Hmean     recv-256       1328.02 (   0.00%)     1318.77 (  -0.70%)
 Hmean     recv-1024      5051.83 (   0.00%)     5051.11 (  -0.01%)
 Hmean     recv-2048      9636.95 (   0.00%)     9601.30 (  -0.37%)
 Hmean     recv-3312     14355.32 (   0.00%)    14414.48 (   0.41%)
 Hmean     recv-4096     16464.74 (   0.00%)    16301.16 (  -0.99%)
 Hmean     recv-8192     26721.63 (   0.00%)    26428.17 (  -1.10%)
 Hmean     recv-16384    38136.00 (   0.00%)    38044.88 (  -0.24%)
 Stddev    send-64           7.30 (   0.00%)        4.75 (  34.96%)
 Stddev    send-128         15.15 (   0.00%)       22.38 ( -47.66%)
 Stddev    send-256         13.99 (   0.00%)       19.14 ( -36.81%)
 Stddev    send-1024       105.73 (   0.00%)       67.38 (  36.27%)
 Stddev    send-2048       294.57 (   0.00%)      223.88 (  24.00%)
 Stddev    send-3312       302.28 (   0.00%)      271.74 (  10.10%)
 Stddev    send-4096       195.92 (   0.00%)      121.10 (  38.19%)
 Stddev    send-8192       399.71 (   0.00%)      563.77 ( -41.04%)
 Stddev    send-16384     1163.47 (   0.00%)     1103.68 (   5.14%)
 Stddev    recv-64           7.30 (   0.00%)        4.75 (  34.96%)
 Stddev    recv-128         15.15 (   0.00%)       22.38 ( -47.66%)
 Stddev    recv-256         13.99 (   0.00%)       19.14 ( -36.81%)
 Stddev    recv-1024       105.73 (   0.00%)       67.38 (  36.27%)
 Stddev    recv-2048       294.59 (   0.00%)      223.89 (  24.00%)
 Stddev    recv-3312       302.24 (   0.00%)      271.75 (  10.09%)
 Stddev    recv-4096       196.03 (   0.00%)      121.14 (  38.20%)
 Stddev    recv-8192       399.86 (   0.00%)      563.65 ( -40.96%)
 Stddev    recv-16384     1163.79 (   0.00%)     1103.86 (   5.15%)

The difference in overall performance is marginal but note that most
measurements are less variable. There were similar observations for other
netperf comparisons. hackbench with sockets or threads with processes or
threads showed minor difference with some reduction of migration. tbench
showed only marginal differences that were within the noise. dbench,
regardless of filesystem, showed minor differences all of which are
within noise. Multiple machines, both UMA and NUMA were tested without
any regressions showing up.

The biggest risk with a patch like this is affecting wakeup latencies.
However, the schbench load from Facebook which is very sensitive to wakeup
latency showed a mixed result with mostly improvements in wakeup latency:

                                      4.15.0                 4.15.0
                                       16rc0          noequal-v1r23
 Lat 50.00th-qrtle-1        38.00 (   0.00%)       38.00 (   0.00%)
 Lat 75.00th-qrtle-1        49.00 (   0.00%)       41.00 (  16.33%)
 Lat 90.00th-qrtle-1        52.00 (   0.00%)       50.00 (   3.85%)
 Lat 95.00th-qrtle-1        54.00 (   0.00%)       51.00 (   5.56%)
 Lat 99.00th-qrtle-1        63.00 (   0.00%)       60.00 (   4.76%)
 Lat 99.50th-qrtle-1        66.00 (   0.00%)       61.00 (   7.58%)
 Lat 99.90th-qrtle-1        78.00 (   0.00%)       65.00 (  16.67%)
 Lat 50.00th-qrtle-2        38.00 (   0.00%)       38.00 (   0.00%)
 Lat 75.00th-qrtle-2        42.00 (   0.00%)       43.00 (  -2.38%)
 Lat 90.00th-qrtle-2        46.00 (   0.00%)       48.00 (  -4.35%)
 Lat 95.00th-qrtle-2        49.00 (   0.00%)       50.00 (  -2.04%)
 Lat 99.00th-qrtle-2        55.00 (   0.00%)       57.00 (  -3.64%)
 Lat 99.50th-qrtle-2        58.00 (   0.00%)       60.00 (  -3.45%)
 Lat 99.90th-qrtle-2        65.00 (   0.00%)       68.00 (  -4.62%)
 Lat 50.00th-qrtle-4        41.00 (   0.00%)       41.00 (   0.00%)
 Lat 75.00th-qrtle-4        45.00 (   0.00%)       46.00 (  -2.22%)
 Lat 90.00th-qrtle-4        50.00 (   0.00%)       50.00 (   0.00%)
 Lat 95.00th-qrtle-4        54.00 (   0.00%)       53.00 (   1.85%)
 Lat 99.00th-qrtle-4        61.00 (   0.00%)       61.00 (   0.00%)
 Lat 99.50th-qrtle-4        65.00 (   0.00%)       64.00 (   1.54%)
 Lat 99.90th-qrtle-4        76.00 (   0.00%)       82.00 (  -7.89%)
 Lat 50.00th-qrtle-8        48.00 (   0.00%)       46.00 (   4.17%)
 Lat 75.00th-qrtle-8        55.00 (   0.00%)       54.00 (   1.82%)
 Lat 90.00th-qrtle-8        60.00 (   0.00%)       59.00 (   1.67%)
 Lat 95.00th-qrtle-8        63.00 (   0.00%)       63.00 (   0.00%)
 Lat 99.00th-qrtle-8        71.00 (   0.00%)       69.00 (   2.82%)
 Lat 99.50th-qrtle-8        74.00 (   0.00%)       73.00 (   1.35%)
 Lat 99.90th-qrtle-8        98.00 (   0.00%)       90.00 (   8.16%)
 Lat 50.00th-qrtle-16       56.00 (   0.00%)       55.00 (   1.79%)
 Lat 75.00th-qrtle-16       68.00 (   0.00%)       67.00 (   1.47%)
 Lat 90.00th-qrtle-16       77.00 (   0.00%)       78.00 (  -1.30%)
 Lat 95.00th-qrtle-16       82.00 (   0.00%)       84.00 (  -2.44%)
 Lat 99.00th-qrtle-16       90.00 (   0.00%)       93.00 (  -3.33%)
 Lat 99.50th-qrtle-16       93.00 (   0.00%)       97.00 (  -4.30%)
 Lat 99.90th-qrtle-16      110.00 (   0.00%)      110.00 (   0.00%)
 Lat 50.00th-qrtle-32       68.00 (   0.00%)       62.00 (   8.82%)
 Lat 75.00th-qrtle-32       90.00 (   0.00%)       83.00 (   7.78%)
 Lat 90.00th-qrtle-32      110.00 (   0.00%)      100.00 (   9.09%)
 Lat 95.00th-qrtle-32      122.00 (   0.00%)      111.00 (   9.02%)
 Lat 99.00th-qrtle-32      145.00 (   0.00%)      133.00 (   8.28%)
 Lat 99.50th-qrtle-32      154.00 (   0.00%)      143.00 (   7.14%)
 Lat 99.90th-qrtle-32     2316.00 (   0.00%)      515.00 (  77.76%)
 Lat 50.00th-qrtle-35       69.00 (   0.00%)       72.00 (  -4.35%)
 Lat 75.00th-qrtle-35       92.00 (   0.00%)       95.00 (  -3.26%)
 Lat 90.00th-qrtle-35      111.00 (   0.00%)      114.00 (  -2.70%)
 Lat 95.00th-qrtle-35      122.00 (   0.00%)      124.00 (  -1.64%)
 Lat 99.00th-qrtle-35      142.00 (   0.00%)      144.00 (  -1.41%)
 Lat 99.50th-qrtle-35      150.00 (   0.00%)      154.00 (  -2.67%)
 Lat 99.90th-qrtle-35     6104.00 (   0.00%)     5640.00 (   7.60%)

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Giovanni Gherdovich <ggherdovich@suse.cz>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180213133730.24064-4-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-02-21 08:49:08 +01:00
Mel Gorman eeb6039863 sched/fair: Defer calculation of 'prev_eff_load' in wake_affine_weight() until needed 
On sync wakeups, the previous CPU effective load may not be used so delay
the calculation until it's needed.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Giovanni Gherdovich <ggherdovich@suse.cz>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180213133730.24064-3-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-02-21 08:49:07 +01:00
Mel Gorman 7ebb66a12f sched/fair: Avoid an unnecessary lookup of current CPU ID during wake_affine 
The only caller of wake_affine() knows the CPU ID. Pass it in instead of
rechecking it.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Giovanni Gherdovich <ggherdovich@suse.cz>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180213133730.24064-2-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-02-21 08:49:07 +01:00
Mel Gorman 32e839dda3 sched/fair: Use a recently used CPU as an idle candidate and the basis for SIS 
The select_idle_sibling() (SIS) rewrite in commit:

  10e2f1acd0 ("sched/core: Rewrite and improve select_idle_siblings()")

... replaced a domain iteration with a search that broadly speaking
does a wrapped walk of the scheduler domain sharing a last-level-cache.

While this had a number of improvements, one consequence is that two tasks
that share a waker/wakee relationship push each other around a socket. Even
though two tasks may be active, all cores are evenly used. This is great from
a search perspective and spreads a load across individual cores, but it has
adverse consequences for cpufreq. As each CPU has relatively low utilisation,
cpufreq may decide the utilisation is too low to used a higher P-state and
overall computation throughput suffers.

While individual cpufreq and cpuidle drivers may compensate by artifically
boosting P-state (at c0) or avoiding lower C-states (during idle), it does
not help if hardware-based cpufreq (e.g. HWP) is used.

This patch tracks a recently used CPU based on what CPU a task was running
on when it last was a waker a CPU it was recently using when a task is a
wakee. During SIS, the recently used CPU is used as a target if it's still
allowed by the task and is idle.

The benefit may be non-obvious so consider an example of two tasks
communicating back and forth. Task A may be an application doing IO where
task B is a kworker or kthread like journald. Task A may issue IO, wake
B and B wakes up A on completion.  With the existing scheme this may look
like the following (potentially different IDs if SMT is in use but similar
principal applies).

 A (cpu 0)	wake	B (wakes on cpu 1)
 B (cpu 1)	wake	A (wakes on cpu 2)
 A (cpu 2)	wake	B (wakes on cpu 3)
 etc.

A careful reader may wonder why CPU 0 was not idle when B wakes A the
first time and it's simply due to the fact that A can be rescheduled to
another CPU and the pattern is that prev == target when B tries to wakeup A
and the information about CPU 0 has been lost.

With this patch, the pattern is more likely to be:

 A (cpu 0)	wake	B (wakes on cpu 1)
 B (cpu 1)	wake	A (wakes on cpu 0)
 A (cpu 0)	wake	B (wakes on cpu 1)
 etc

i.e. two communicating casts are more likely to use just two cores instead
of all available cores sharing a LLC.

The most dramatic speedup was noticed on dbench using the XFS filesystem on
UMA as clients interact heavily with workqueues in that configuration. Note
that a similar speedup is not observed on ext4 as the wakeup pattern
is different:

                          4.15.0-rc9             4.15.0-rc9
                           waprev-v1        biasancestor-v1
 Hmean      1      287.54 (   0.00%)      817.01 ( 184.14%)
 Hmean      2     1268.12 (   0.00%)     1781.24 (  40.46%)
 Hmean      4     1739.68 (   0.00%)     1594.47 (  -8.35%)
 Hmean      8     2464.12 (   0.00%)     2479.56 (   0.63%)
 Hmean     64     1455.57 (   0.00%)     1434.68 (  -1.44%)

The results can be less dramatic on NUMA where automatic balancing interferes
with the test. It's also known that network benchmarks running on localhost
also benefit quite a bit from this patch (roughly 10% on netperf RR for UDP
and TCP depending on the machine). Hackbench also seens small improvements
(6-11% depending on machine and thread count). The facebook schbench was also
tested but in most cases showed little or no different to wakeup latencies.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180130104555.4125-5-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-02-06 10:20:37 +01:00
Mel Gorman 806486c377 sched/fair: Do not migrate if the prev_cpu is idle 
wake_affine_idle() prefers to move a task to the current CPU if the
wakeup is due to an interrupt. The expectation is that the interrupt
data is cache hot and relevant to the waking task as well as avoiding
a search. However, there is no way to determine if there was cache hot
data on the previous CPU that may exceed the interrupt data. Furthermore,
round-robin delivery of interrupts can migrate tasks around a socket where
each CPU is under-utilised.  This can interact badly with cpufreq which
makes decisions based on per-cpu data. It has been observed on machines
with HWP that p-states are not boosted to their maximum levels even though
the workload is latency and throughput sensitive.

This patch uses the previous CPU for the task if it's idle and cache-affine
with the current CPU even if the current CPU is idle due to the wakup
being related to the interrupt. This reduces migrations at the cost of
the interrupt data not being cache hot when the task wakes.

A variety of workloads were tested on various machines and no adverse
impact was noticed that was outside noise. dbench on ext4 on UMA showed
roughly 10% reduction in the number of CPU migrations and it is a case
where interrupts are frequent for IO competions. In most cases, the
difference in performance is quite small but variability is often
reduced. For example, this is the result for pgbench running on a UMA
machine with different numbers of clients.

                          4.15.0-rc9             4.15.0-rc9
                            baseline              waprev-v1
 Hmean     1     22096.28 (   0.00%)    22734.86 (   2.89%)
 Hmean     4     74633.42 (   0.00%)    75496.77 (   1.16%)
 Hmean     7    115017.50 (   0.00%)   113030.81 (  -1.73%)
 Hmean     12   126209.63 (   0.00%)   126613.40 (   0.32%)
 Hmean     16   131886.91 (   0.00%)   130844.35 (  -0.79%)
 Stddev    1       636.38 (   0.00%)      417.11 (  34.46%)
 Stddev    4       614.64 (   0.00%)      583.24 (   5.11%)
 Stddev    7       542.46 (   0.00%)      435.45 (  19.73%)
 Stddev    12      173.93 (   0.00%)      171.50 (   1.40%)
 Stddev    16      671.42 (   0.00%)      680.30 (  -1.32%)
 CoeffVar  1         2.88 (   0.00%)        1.83 (  36.26%)

Note that the different in performance is marginal but for low utilisation,
there is less variability.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180130104555.4125-4-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-02-06 10:20:36 +01:00
Mel Gorman 3b76c4a339 sched/fair: Restructure wake_affine*() to return a CPU id 
This is a preparation patch that has wake_affine*() return a CPU ID instead of
a boolean. The intent is to allow the wake_affine() helpers to be avoided
if a decision is already made. This patch has no functional change.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180130104555.4125-3-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-02-06 10:20:35 +01:00
Mel Gorman 89a55f56fd sched/fair: Remove unnecessary parameters from wake_affine_idle() 
wake_affine_idle() takes parameters it never uses so clean it up.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20180130104555.4125-2-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-02-06 10:20:35 +01:00
Mel Gorman 69d763fc6d mm: pin address_space before dereferencing it while isolating an LRU page 
Minchan Kim asked the following question -- what locks protects
address_space destroying when race happens between inode trauncation and
__isolate_lru_page? Jan Kara clarified by describing the race as follows

CPU1                                            CPU2

truncate(inode)                                 __isolate_lru_page()
  ...
  truncate_inode_page(mapping, page);
    delete_from_page_cache(page)
      spin_lock_irqsave(&mapping->tree_lock, flags);
        __delete_from_page_cache(page, NULL)
          page_cache_tree_delete(..)
            ...                                   mapping = page_mapping(page);
            page->mapping = NULL;
            ...
      spin_unlock_irqrestore(&mapping->tree_lock, flags);
      page_cache_free_page(mapping, page)
        put_page(page)
          if (put_page_testzero(page)) -> false
- inode now has no pages and can be freed including embedded address_space

                                                  if (mapping && !mapping->a_ops->migratepage)
- we've dereferenced mapping which is potentially already free.

The race is theoretically possible but unlikely.  Before the
delete_from_page_cache, truncate_cleanup_page is called so the page is
likely to be !PageDirty or PageWriteback which gets skipped by the only
caller that checks the mappping in __isolate_lru_page.  Even if the race
occurs, a substantial amount of work has to happen during a tiny window
with no preemption but it could potentially be done using a virtual
machine to artifically slow one CPU or halt it during the critical
window.

This patch should eliminate the race with truncation by try-locking the
page before derefencing mapping and aborting if the lock was not
acquired.  There was a suggestion from Huang Ying to use RCU as a
side-effect to prevent mapping being freed.  However, I do not like the
solution as it's an unconventional means of preserving a mapping and
it's not a context where rcu_read_lock is obviously protecting rcu data.

Link: http://lkml.kernel.org/r/20180104102512.2qos3h5vqzeisrek@techsingularity.net
Fixes: c824493528 ("mm: compaction: make isolate_lru_page() filter-aware again")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2018-01-31 17:18:39 -08:00
Mel Gorman 7332dec055 sched/fair: Only immediately migrate tasks due to interrupts if prev and target CPUs share cache 
If waking from an idle CPU due to an interrupt then it's possible that
the waker task will be pulled to wake on the current CPU. Unfortunately,
depending on the type of interrupt and IRQ configuration, there may not
be a strong relationship between the CPU an interrupt was delivered on
and the CPU a task was running on. For example, the interrupts could all
be delivered to CPUs on one particular node due to the machine topology
or IRQ affinity configuration. Another example is an interrupt for an IO
completion which can be delivered to any CPU where there is no guarantee
the data is either cache hot or even local.

This patch was motivated by the observation that an IO workload was
being pulled cross-node on a frequent basis when IO completed.  From a
wakeup latency perspective, it's still useful to know that an idle CPU is
immediately available for use but lets only consider an automatic migration
if the CPUs share cache to limit damage due to NUMA migrations. Migrations
may still occur if wake_affine_weight determines it's appropriate.

These are the throughput results for dbench running on ext4 comparing
4.15-rc3 and this patch on a 2-socket machine where interrupts due to IO
completions can happen on any CPU.

                          4.15.0-rc3             4.15.0-rc3
                             vanilla            lessmigrate
Hmean     1        854.64 (   0.00%)      865.01 (   1.21%)
Hmean     2       1229.60 (   0.00%)     1274.44 (   3.65%)
Hmean     4       1591.81 (   0.00%)     1628.08 (   2.28%)
Hmean     8       1845.04 (   0.00%)     1831.80 (  -0.72%)
Hmean     16      2038.61 (   0.00%)     2091.44 (   2.59%)
Hmean     32      2327.19 (   0.00%)     2430.29 (   4.43%)
Hmean     64      2570.61 (   0.00%)     2568.54 (  -0.08%)
Hmean     128     2481.89 (   0.00%)     2499.28 (   0.70%)
Stddev    1         14.31 (   0.00%)        5.35 (  62.65%)
Stddev    2         21.29 (   0.00%)       11.09 (  47.92%)
Stddev    4          7.22 (   0.00%)        6.80 (   5.92%)
Stddev    8         26.70 (   0.00%)        9.41 (  64.76%)
Stddev    16        22.40 (   0.00%)       20.01 (  10.70%)
Stddev    32        45.13 (   0.00%)       44.74 (   0.85%)
Stddev    64        93.10 (   0.00%)       93.18 (  -0.09%)
Stddev    128      184.28 (   0.00%)      177.85 (   3.49%)

Note the small increase in throughput for low thread counts but also
note that the standard deviation for each sample during the test run is
lower. The throughput figures for dbench can be misleading so the benchmark
is actually modified to time the latency of the processing of one load
file with many samples taken. The difference in latency is

                           4.15.0-rc3             4.15.0-rc3
                              vanilla            lessmigrate
Amean      1         21.71 (   0.00%)       21.47 (   1.08%)
Amean      2         30.89 (   0.00%)       29.58 (   4.26%)
Amean      4         47.54 (   0.00%)       46.61 (   1.97%)
Amean      8         82.71 (   0.00%)       82.81 (  -0.12%)
Amean      16       149.45 (   0.00%)      145.01 (   2.97%)
Amean      32       265.49 (   0.00%)      248.43 (   6.42%)
Amean      64       463.23 (   0.00%)      463.55 (  -0.07%)
Amean      128      933.97 (   0.00%)      935.50 (  -0.16%)
Stddev     1          1.58 (   0.00%)        1.54 (   2.26%)
Stddev     2          2.84 (   0.00%)        2.95 (  -4.15%)
Stddev     4          6.78 (   0.00%)        6.85 (  -0.99%)
Stddev     8         16.85 (   0.00%)       16.37 (   2.85%)
Stddev     16        41.59 (   0.00%)       41.04 (   1.32%)
Stddev     32       111.05 (   0.00%)      105.11 (   5.35%)
Stddev     64       285.94 (   0.00%)      288.01 (  -0.72%)
Stddev     128      803.39 (   0.00%)      809.73 (  -0.79%)

It's a small improvement which is not surprising given that migrations that
migrate to a different node as not that common. However, it is noticeable
in the CPU migration statistics which are reduced by 24%.

There was a query for v1 of this patch about NAS so here are the results
for C-class using MPI for parallelisation on the same machine

nas-mpi
                      4.15.0-rc3             4.15.0-rc3
                         vanilla                  noirq
Time cg.C       24.25 (   0.00%)       23.17 (   4.45%)
Time ep.C        8.22 (   0.00%)        8.29 (  -0.85%)
Time ft.C       22.67 (   0.00%)       20.34 (  10.28%)
Time is.C        1.42 (   0.00%)        1.47 (  -3.52%)
Time lu.C       55.62 (   0.00%)       54.81 (   1.46%)
Time mg.C        7.93 (   0.00%)        7.91 (   0.25%)

          4.15.0-rc3  4.15.0-rc3
             vanilla  noirq-v1r1
User         3799.96     3748.34
System        672.10      626.15
Elapsed        91.91       79.49

lu.C sees a small gain, ft.C a large gain and ep.C and is.C see small
regressions but in terms of absolute time, the difference is small and
likely within run-to-run variance. System CPU usage is slightly reduced.

schbench from Facebook was also requested. This is a bit of a mixed bag but
it's important to note that this workload should not be heavily impacted
by wakeups from interrupt context.

                                 4.15.0-rc3             4.15.0-rc3
                                    vanilla             noirq-v1r1
Lat 50.00th-qrtle-1        41.00 (   0.00%)       41.00 (   0.00%)
Lat 75.00th-qrtle-1        42.00 (   0.00%)       42.00 (   0.00%)
Lat 90.00th-qrtle-1        43.00 (   0.00%)       44.00 (  -2.33%)
Lat 95.00th-qrtle-1        44.00 (   0.00%)       46.00 (  -4.55%)
Lat 99.00th-qrtle-1        57.00 (   0.00%)       58.00 (  -1.75%)
Lat 99.50th-qrtle-1        59.00 (   0.00%)       59.00 (   0.00%)
Lat 99.90th-qrtle-1        67.00 (   0.00%)       78.00 ( -16.42%)
Lat 50.00th-qrtle-2        40.00 (   0.00%)       51.00 ( -27.50%)
Lat 75.00th-qrtle-2        45.00 (   0.00%)       56.00 ( -24.44%)
Lat 90.00th-qrtle-2        53.00 (   0.00%)       59.00 ( -11.32%)
Lat 95.00th-qrtle-2        57.00 (   0.00%)       61.00 (  -7.02%)
Lat 99.00th-qrtle-2        67.00 (   0.00%)       71.00 (  -5.97%)
Lat 99.50th-qrtle-2        69.00 (   0.00%)       74.00 (  -7.25%)
Lat 99.90th-qrtle-2        83.00 (   0.00%)       77.00 (   7.23%)
Lat 50.00th-qrtle-4        51.00 (   0.00%)       51.00 (   0.00%)
Lat 75.00th-qrtle-4        57.00 (   0.00%)       56.00 (   1.75%)
Lat 90.00th-qrtle-4        60.00 (   0.00%)       59.00 (   1.67%)
Lat 95.00th-qrtle-4        62.00 (   0.00%)       62.00 (   0.00%)
Lat 99.00th-qrtle-4        73.00 (   0.00%)       72.00 (   1.37%)
Lat 99.50th-qrtle-4        76.00 (   0.00%)       74.00 (   2.63%)
Lat 99.90th-qrtle-4        85.00 (   0.00%)       78.00 (   8.24%)
Lat 50.00th-qrtle-8        54.00 (   0.00%)       58.00 (  -7.41%)
Lat 75.00th-qrtle-8        59.00 (   0.00%)       62.00 (  -5.08%)
Lat 90.00th-qrtle-8        65.00 (   0.00%)       66.00 (  -1.54%)
Lat 95.00th-qrtle-8        67.00 (   0.00%)       70.00 (  -4.48%)
Lat 99.00th-qrtle-8        78.00 (   0.00%)       79.00 (  -1.28%)
Lat 99.50th-qrtle-8        81.00 (   0.00%)       80.00 (   1.23%)
Lat 99.90th-qrtle-8       116.00 (   0.00%)       83.00 (  28.45%)
Lat 50.00th-qrtle-16       65.00 (   0.00%)       64.00 (   1.54%)
Lat 75.00th-qrtle-16       77.00 (   0.00%)       71.00 (   7.79%)
Lat 90.00th-qrtle-16       83.00 (   0.00%)       82.00 (   1.20%)
Lat 95.00th-qrtle-16       87.00 (   0.00%)       87.00 (   0.00%)
Lat 99.00th-qrtle-16       95.00 (   0.00%)       96.00 (  -1.05%)
Lat 99.50th-qrtle-16       99.00 (   0.00%)      103.00 (  -4.04%)
Lat 99.90th-qrtle-16      104.00 (   0.00%)      122.00 ( -17.31%)
Lat 50.00th-qrtle-32       71.00 (   0.00%)       73.00 (  -2.82%)
Lat 75.00th-qrtle-32       91.00 (   0.00%)       92.00 (  -1.10%)
Lat 90.00th-qrtle-32      108.00 (   0.00%)      107.00 (   0.93%)
Lat 95.00th-qrtle-32      118.00 (   0.00%)      115.00 (   2.54%)
Lat 99.00th-qrtle-32      134.00 (   0.00%)      129.00 (   3.73%)
Lat 99.50th-qrtle-32      138.00 (   0.00%)      133.00 (   3.62%)
Lat 99.90th-qrtle-32      149.00 (   0.00%)      146.00 (   2.01%)
Lat 50.00th-qrtle-39       83.00 (   0.00%)       81.00 (   2.41%)
Lat 75.00th-qrtle-39      105.00 (   0.00%)      102.00 (   2.86%)
Lat 90.00th-qrtle-39      120.00 (   0.00%)      119.00 (   0.83%)
Lat 95.00th-qrtle-39      129.00 (   0.00%)      128.00 (   0.78%)
Lat 99.00th-qrtle-39      153.00 (   0.00%)      149.00 (   2.61%)
Lat 99.50th-qrtle-39      166.00 (   0.00%)      156.00 (   6.02%)
Lat 99.90th-qrtle-39    12304.00 (   0.00%)    12848.00 (  -4.42%)

When heavily loaded (e.g. 99.50th-qrtle-39 indicates 39 threads), there
are small gains in many cases. Otherwise it depends on the quartile used
where it can be bad -- e.g. 75.00th-qrtle-2. However, even these results
are probably a co-incidence. For this workload, much depends on what node
the threads get placed on and their relative locality and not wakeups from
interrupt context. A larger component on how it behaves would be automatic
NUMA balancing where a fault incurred to measure locality would be a much
larger contributer to latency than the wakeup path.

This is the results from an almost identical machine that happened to run
the same test.  They only differ in terms of storage which is irrelevant
for this test.

                                 4.15.0-rc3             4.15.0-rc3
                                    vanilla             noirq-v1r1
Lat 50.00th-qrtle-1        41.00 (   0.00%)       41.00 (   0.00%)
Lat 75.00th-qrtle-1        42.00 (   0.00%)       42.00 (   0.00%)
Lat 90.00th-qrtle-1        44.00 (   0.00%)       43.00 (   2.27%)
Lat 95.00th-qrtle-1        53.00 (   0.00%)       45.00 (  15.09%)
Lat 99.00th-qrtle-1        59.00 (   0.00%)       58.00 (   1.69%)
Lat 99.50th-qrtle-1        60.00 (   0.00%)       59.00 (   1.67%)
Lat 99.90th-qrtle-1        86.00 (   0.00%)       61.00 (  29.07%)
Lat 50.00th-qrtle-2        52.00 (   0.00%)       41.00 (  21.15%)
Lat 75.00th-qrtle-2        57.00 (   0.00%)       46.00 (  19.30%)
Lat 90.00th-qrtle-2        60.00 (   0.00%)       53.00 (  11.67%)
Lat 95.00th-qrtle-2        62.00 (   0.00%)       57.00 (   8.06%)
Lat 99.00th-qrtle-2        73.00 (   0.00%)       68.00 (   6.85%)
Lat 99.50th-qrtle-2        74.00 (   0.00%)       71.00 (   4.05%)
Lat 99.90th-qrtle-2        90.00 (   0.00%)       75.00 (  16.67%)
Lat 50.00th-qrtle-4        57.00 (   0.00%)       52.00 (   8.77%)
Lat 75.00th-qrtle-4        60.00 (   0.00%)       58.00 (   3.33%)
Lat 90.00th-qrtle-4        62.00 (   0.00%)       62.00 (   0.00%)
Lat 95.00th-qrtle-4        65.00 (   0.00%)       65.00 (   0.00%)
Lat 99.00th-qrtle-4        76.00 (   0.00%)       75.00 (   1.32%)
Lat 99.50th-qrtle-4        77.00 (   0.00%)       77.00 (   0.00%)
Lat 99.90th-qrtle-4        87.00 (   0.00%)       81.00 (   6.90%)
Lat 50.00th-qrtle-8        59.00 (   0.00%)       57.00 (   3.39%)
Lat 75.00th-qrtle-8        63.00 (   0.00%)       62.00 (   1.59%)
Lat 90.00th-qrtle-8        66.00 (   0.00%)       67.00 (  -1.52%)
Lat 95.00th-qrtle-8        68.00 (   0.00%)       70.00 (  -2.94%)
Lat 99.00th-qrtle-8        79.00 (   0.00%)       80.00 (  -1.27%)
Lat 99.50th-qrtle-8        80.00 (   0.00%)       84.00 (  -5.00%)
Lat 99.90th-qrtle-8        84.00 (   0.00%)       90.00 (  -7.14%)
Lat 50.00th-qrtle-16       65.00 (   0.00%)       65.00 (   0.00%)
Lat 75.00th-qrtle-16       77.00 (   0.00%)       75.00 (   2.60%)
Lat 90.00th-qrtle-16       84.00 (   0.00%)       83.00 (   1.19%)
Lat 95.00th-qrtle-16       88.00 (   0.00%)       87.00 (   1.14%)
Lat 99.00th-qrtle-16       97.00 (   0.00%)       96.00 (   1.03%)
Lat 99.50th-qrtle-16      100.00 (   0.00%)      104.00 (  -4.00%)
Lat 99.90th-qrtle-16      110.00 (   0.00%)      126.00 ( -14.55%)
Lat 50.00th-qrtle-32       70.00 (   0.00%)       71.00 (  -1.43%)
Lat 75.00th-qrtle-32       92.00 (   0.00%)       94.00 (  -2.17%)
Lat 90.00th-qrtle-32      110.00 (   0.00%)      110.00 (   0.00%)
Lat 95.00th-qrtle-32      121.00 (   0.00%)      118.00 (   2.48%)
Lat 99.00th-qrtle-32      135.00 (   0.00%)      137.00 (  -1.48%)
Lat 99.50th-qrtle-32      140.00 (   0.00%)      146.00 (  -4.29%)
Lat 99.90th-qrtle-32      150.00 (   0.00%)      160.00 (  -6.67%)
Lat 50.00th-qrtle-39       80.00 (   0.00%)       71.00 (  11.25%)
Lat 75.00th-qrtle-39      102.00 (   0.00%)       91.00 (  10.78%)
Lat 90.00th-qrtle-39      118.00 (   0.00%)      108.00 (   8.47%)
Lat 95.00th-qrtle-39      128.00 (   0.00%)      117.00 (   8.59%)
Lat 99.00th-qrtle-39      149.00 (   0.00%)      133.00 (  10.74%)
Lat 99.50th-qrtle-39      160.00 (   0.00%)      139.00 (  13.12%)
Lat 99.90th-qrtle-39    13808.00 (   0.00%)     4920.00 (  64.37%)

Despite being nearly identical, it showed a variety of major gains so
I'm not convinced that heavy emphasis should be placed on this particular
workload in terms of evaluating this particular patch. Further evidence of
this is the fact that testing on a UMA machine showed small gains/losses
even though the patch should be a no-op on UMA.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/20171219085947.13136-2-mgorman@techsingularity.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
 2018-01-10 11:30:31 +01:00
Mel Gorman 7f0b5fb953 mm, pagevec: rename pagevec drained field 
According to Vlastimil Babka, the drained field in pagevec is
potentially misleading because it might be interpreted as draining this
pagevec instead of the percpu lru pagevecs.  Rename the field for
clarity.

Link: http://lkml.kernel.org/r/20171019093346.ylahzdpzmoriyf4v@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Suggested-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-11-15 18:21:06 -08:00
Mel Gorman 453f85d43f mm: remove __GFP_COLD 
As the page free path makes no distinction between cache hot and cold
pages, there is no real useful ordering of pages in the free list that
allocation requests can take advantage of.  Juding from the users of
__GFP_COLD, it is likely that a number of them are the result of copying
other sites instead of actually measuring the impact.  Remove the
__GFP_COLD parameter which simplifies a number of paths in the page
allocator.

This is potentially controversial but bear in mind that the size of the
per-cpu pagelists versus modern cache sizes means that the whole per-cpu
list can often fit in the L3 cache.  Hence, there is only a potential
benefit for microbenchmarks that alloc/free pages in a tight loop.  It's
even worse when THP is taken into account which has little or no chance
of getting a cache-hot page as the per-cpu list is bypassed and the
zeroing of multiple pages will thrash the cache anyway.

The truncate microbenchmarks are not shown as this patch affects the
allocation path and not the free path.  A page fault microbenchmark was
tested but it showed no sigificant difference which is not surprising
given that the __GFP_COLD branches are a miniscule percentage of the
fault path.

Link: http://lkml.kernel.org/r/20171018075952.10627-9-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-11-15 18:21:06 -08:00
Mel Gorman 2d4894b5d2 mm: remove cold parameter from free_hot_cold_page* 
Most callers users of free_hot_cold_page claim the pages being released
are cache hot.  The exception is the page reclaim paths where it is
likely that enough pages will be freed in the near future that the
per-cpu lists are going to be recycled and the cache hotness information
is lost.  As no one really cares about the hotness of pages being
released to the allocator, just ditch the parameter.

The APIs are renamed to indicate that it's no longer about hot/cold
pages.  It should also be less confusing as there are subtle differences
between them.  __free_pages drops a reference and frees a page when the
refcount reaches zero.  free_hot_cold_page handled pages whose refcount
was already zero which is non-obvious from the name.  free_unref_page
should be more obvious.

No performance impact is expected as the overhead is marginal.  The
parameter is removed simply because it is a bit stupid to have a useless
parameter copied everywhere.

[mgorman@techsingularity.net: add pages to head, not tail]
  Link: http://lkml.kernel.org/r/20171019154321.qtpzaeftoyyw4iey@techsingularity.net
Link: http://lkml.kernel.org/r/20171018075952.10627-8-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-11-15 18:21:06 -08:00
Mel Gorman c6f92f9fbe mm: remove cold parameter for release_pages 
All callers of release_pages claim the pages being released are cache
hot.  As no one cares about the hotness of pages being released to the
allocator, just ditch the parameter.

No performance impact is expected as the overhead is marginal.  The
parameter is removed simply because it is a bit stupid to have a useless
parameter copied everywhere.

Link: http://lkml.kernel.org/r/20171018075952.10627-7-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-11-15 18:21:06 -08:00
Mel Gorman 8667982014 mm, pagevec: remove cold parameter for pagevecs 
Every pagevec_init user claims the pages being released are hot even in
cases where it is unlikely the pages are hot.  As no one cares about the
hotness of pages being released to the allocator, just ditch the
parameter.

No performance impact is expected as the overhead is marginal.  The
parameter is removed simply because it is a bit stupid to have a useless
parameter copied everywhere.

Link: http://lkml.kernel.org/r/20171018075952.10627-6-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-11-15 18:21:06 -08:00
Mel Gorman d9ed0d08b6 mm: only drain per-cpu pagevecs once per pagevec usage 
When a pagevec is initialised on the stack, it is generally used
multiple times over a range of pages, looking up entries and then
releasing them.  On each pagevec_release, the per-cpu deferred LRU
pagevecs are drained on the grounds the page being released may be on
those queues and the pages may be cache hot.  In many cases only the
first drain is necessary as it's unlikely that the range of pages being
walked is racing against LRU addition.  Even if there is such a race,
the impact is marginal where as constantly redraining the lru pagevecs
costs.

This patch ensures that pagevec is only drained once in a given
lifecycle without increasing the cache footprint of the pagevec
structure.  Only sparsetruncate tiny is shown here as large files have
many exceptional entries and calls pagecache_release less frequently.

sparsetruncate (tiny)
                              4.14.0-rc4             4.14.0-rc4
                        batchshadow-v1r1          onedrain-v1r1
Min          Time      141.00 (   0.00%)      141.00 (   0.00%)
1st-qrtle    Time      142.00 (   0.00%)      142.00 (   0.00%)
2nd-qrtle    Time      142.00 (   0.00%)      142.00 (   0.00%)
3rd-qrtle    Time      143.00 (   0.00%)      143.00 (   0.00%)
Max-90%      Time      144.00 (   0.00%)      144.00 (   0.00%)
Max-95%      Time      146.00 (   0.00%)      145.00 (   0.68%)
Max-99%      Time      198.00 (   0.00%)      194.00 (   2.02%)
Max          Time      254.00 (   0.00%)      208.00 (  18.11%)
Amean        Time      145.12 (   0.00%)      144.30 (   0.56%)
Stddev       Time       12.74 (   0.00%)        9.62 (  24.49%)
Coeff        Time        8.78 (   0.00%)        6.67 (  24.06%)
Best99%Amean Time      144.29 (   0.00%)      143.82 (   0.32%)
Best95%Amean Time      142.68 (   0.00%)      142.31 (   0.26%)
Best90%Amean Time      142.52 (   0.00%)      142.19 (   0.24%)
Best75%Amean Time      142.26 (   0.00%)      141.98 (   0.20%)
Best50%Amean Time      141.90 (   0.00%)      141.71 (   0.13%)
Best25%Amean Time      141.80 (   0.00%)      141.43 (   0.26%)

The impact on bonnie is marginal and within the noise because a
significant percentage of the file being truncated has been reclaimed
and consists of shadow entries which reduce the hotness of the
pagevec_release path.

Link: http://lkml.kernel.org/r/20171018075952.10627-5-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-11-15 18:21:06 -08:00
Mel Gorman f218759918 mm, truncate: remove all exceptional entries from pagevec under one lock 
During truncate each entry in a pagevec is checked to see if it is an
exceptional entry and if so, the shadow entry is cleaned up.  This is
potentially expensive as multiple entries for a mapping locks/unlocks
the tree lock.  This batches the operation such that any exceptional
entries removed from a pagevec only acquire the mapping tree lock once.
The corner case where this is more expensive is where there is only one
exceptional entry but this is unlikely due to temporal locality and how
it affects LRU ordering.  Note that for truncations of small files
created recently, this patch should show no gain because it only batches
the handling of exceptional entries.

sparsetruncate (large)
                              4.14.0-rc4             4.14.0-rc4
                         pickhelper-v1r1       batchshadow-v1r1
Min          Time       38.00 (   0.00%)       27.00 (  28.95%)
1st-qrtle    Time       40.00 (   0.00%)       28.00 (  30.00%)
2nd-qrtle    Time       44.00 (   0.00%)       41.00 (   6.82%)
3rd-qrtle    Time      146.00 (   0.00%)      147.00 (  -0.68%)
Max-90%      Time      153.00 (   0.00%)      153.00 (   0.00%)
Max-95%      Time      155.00 (   0.00%)      156.00 (  -0.65%)
Max-99%      Time      181.00 (   0.00%)      171.00 (   5.52%)
Amean        Time       93.04 (   0.00%)       88.43 (   4.96%)
Best99%Amean Time       92.08 (   0.00%)       86.13 (   6.46%)
Best95%Amean Time       89.19 (   0.00%)       83.13 (   6.80%)
Best90%Amean Time       85.60 (   0.00%)       79.15 (   7.53%)
Best75%Amean Time       72.95 (   0.00%)       65.09 (  10.78%)
Best50%Amean Time       39.86 (   0.00%)       28.20 (  29.25%)
Best25%Amean Time       39.44 (   0.00%)       27.70 (  29.77%)

bonnie
                                      4.14.0-rc4             4.14.0-rc4
                                 pickhelper-v1r1       batchshadow-v1r1
Hmean     SeqCreate ops         71.92 (   0.00%)       76.78 (   6.76%)
Hmean     SeqCreate read        42.42 (   0.00%)       45.01 (   6.10%)
Hmean     SeqCreate del      26519.88 (   0.00%)    27191.87 (   2.53%)
Hmean     RandCreate ops        71.92 (   0.00%)       76.95 (   7.00%)
Hmean     RandCreate read       44.44 (   0.00%)       49.23 (  10.78%)
Hmean     RandCreate del     24948.62 (   0.00%)    24764.97 (  -0.74%)

Truncation of a large number of files shows a substantial gain with 99%
of files being truncated 6.46% faster.  bonnie shows a modest gain of
2.53%

[jack@suse.cz: fix truncate_exceptional_pvec_entries()]
  Link: http://lkml.kernel.org/r/20171108164226.26788-1-jack@suse.cz
Link: http://lkml.kernel.org/r/20171018075952.10627-4-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-11-15 18:21:06 -08:00
Mel Gorman c7df8ad291 mm, truncate: do not check mapping for every page being truncated 
During truncation, the mapping has already been checked for shmem and
dax so it's known that workingset_update_node is required.

This patch avoids the checks on mapping for each page being truncated.
In all other cases, a lookup helper is used to determine if
workingset_update_node() needs to be called.  The one danger is that the
API is slightly harder to use as calling workingset_update_node directly
without checking for dax or shmem mappings could lead to surprises.
However, the API rarely needs to be used and hopefully the comment is
enough to give people the hint.

sparsetruncate (tiny)
                              4.14.0-rc4             4.14.0-rc4
                             oneirq-v1r1        pickhelper-v1r1
Min          Time      141.00 (   0.00%)      140.00 (   0.71%)
1st-qrtle    Time      142.00 (   0.00%)      141.00 (   0.70%)
2nd-qrtle    Time      142.00 (   0.00%)      142.00 (   0.00%)
3rd-qrtle    Time      143.00 (   0.00%)      143.00 (   0.00%)
Max-90%      Time      144.00 (   0.00%)      144.00 (   0.00%)
Max-95%      Time      147.00 (   0.00%)      145.00 (   1.36%)
Max-99%      Time      195.00 (   0.00%)      191.00 (   2.05%)
Max          Time      230.00 (   0.00%)      205.00 (  10.87%)
Amean        Time      144.37 (   0.00%)      143.82 (   0.38%)
Stddev       Time       10.44 (   0.00%)        9.00 (  13.74%)
Coeff        Time        7.23 (   0.00%)        6.26 (  13.41%)
Best99%Amean Time      143.72 (   0.00%)      143.34 (   0.26%)
Best95%Amean Time      142.37 (   0.00%)      142.00 (   0.26%)
Best90%Amean Time      142.19 (   0.00%)      141.85 (   0.24%)
Best75%Amean Time      141.92 (   0.00%)      141.58 (   0.24%)
Best50%Amean Time      141.69 (   0.00%)      141.31 (   0.27%)
Best25%Amean Time      141.38 (   0.00%)      140.97 (   0.29%)

As you'd expect, the gain is marginal but it can be detected.  The
differences in bonnie are all within the noise which is not surprising
given the impact on the microbenchmark.

radix_tree_update_node_t is a callback for some radix operations that
optionally passes in a private field.  The only user of the callback is
workingset_update_node and as it no longer requires a mapping, the
private field is removed.

Link: http://lkml.kernel.org/r/20171018075952.10627-3-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-11-15 18:21:06 -08:00
Mel Gorman 9cca35d42e mm, page_alloc: enable/disable IRQs once when freeing a list of pages 
Patch series "Follow-up for speed up page cache truncation", v2.

This series is a follow-on for Jan Kara's series "Speed up page cache
truncation" series.  We both ended up looking at the same problem but
saw different problems based on the same data.  This series builds upon
his work.

A variety of workloads were compared on four separate machines but each
machine showed gains albeit at different levels.  Minimally, some of the
differences are due to NUMA where truncating data from a remote node is
slower than a local node.  The workloads checked were

o sparse truncate microbenchmark, tiny
o sparse truncate microbenchmark, large
o reaim-io disk workfile
o dbench4 (modified by mmtests to produce more stable results)
o filebench varmail configuration for small memory size
o bonnie, directory operations, working set size 2*RAM

reaim-io, dbench and filebench all showed minor gains.  Truncation does
not dominate those workloads but were tested to ensure no other
regressions.  They will not be reported further.

The sparse truncate microbench was written by Jan.  It creates a number
of files and then times how long it takes to truncate each one.  The
"tiny" configuraiton creates a number of files that easily fits in
memory and times how long it takes to truncate files with page cache.
The large configuration uses enough files to have data that is twice the
size of memory and so timings there include truncating page cache and
working set shadow entries in the radix tree.

Patches 1-4 are the most relevant parts of this series.  Patches 5-8 are
optional as they are deleting code that is essentially useless but has a
negligible performance impact.

The changelogs have more information on performance but just for bonnie
delete options, the main comparison is

bonnie
                                      4.14.0-rc5             4.14.0-rc5             4.14.0-rc5
                                          jan-v2                vanilla                 mel-v2
Hmean     SeqCreate ops         76.20 (   0.00%)       75.80 (  -0.53%)       76.80 (   0.79%)
Hmean     SeqCreate read        85.00 (   0.00%)       85.00 (   0.00%)       85.00 (   0.00%)
Hmean     SeqCreate del      13752.31 (   0.00%)    12090.23 ( -12.09%)    15304.84 (  11.29%)
Hmean     RandCreate ops        76.00 (   0.00%)       75.60 (  -0.53%)       77.00 (   1.32%)
Hmean     RandCreate read       96.80 (   0.00%)       96.80 (   0.00%)       97.00 (   0.21%)
Hmean     RandCreate del     13233.75 (   0.00%)    11525.35 ( -12.91%)    14446.61 (   9.16%)

Jan's series is the baseline and the vanilla kernel is 12% slower where
as this series on top gains another 11%.  This is from a different
machine than the data in the changelogs but the detailed data was not
collected as there was no substantial change in v2.

This patch (of 8):

Freeing a list of pages current enables/disables IRQs for each page
freed.  This patch splits freeing a list of pages into two operations --
preparing the pages for freeing and the actual freeing.  This is a
tradeoff - we're taking two passes of the list to free in exchange for
avoiding multiple enable/disable of IRQs.

sparsetruncate (tiny)
                              4.14.0-rc4             4.14.0-rc4
                           janbatch-v1r1            oneirq-v1r1
Min          Time      149.00 (   0.00%)      141.00 (   5.37%)
1st-qrtle    Time      150.00 (   0.00%)      142.00 (   5.33%)
2nd-qrtle    Time      151.00 (   0.00%)      142.00 (   5.96%)
3rd-qrtle    Time      151.00 (   0.00%)      143.00 (   5.30%)
Max-90%      Time      153.00 (   0.00%)      144.00 (   5.88%)
Max-95%      Time      155.00 (   0.00%)      147.00 (   5.16%)
Max-99%      Time      201.00 (   0.00%)      195.00 (   2.99%)
Max          Time      236.00 (   0.00%)      230.00 (   2.54%)
Amean        Time      152.65 (   0.00%)      144.37 (   5.43%)
Stddev       Time        9.78 (   0.00%)       10.44 (  -6.72%)
Coeff        Time        6.41 (   0.00%)        7.23 ( -12.84%)
Best99%Amean Time      152.07 (   0.00%)      143.72 (   5.50%)
Best95%Amean Time      150.75 (   0.00%)      142.37 (   5.56%)
Best90%Amean Time      150.59 (   0.00%)      142.19 (   5.58%)
Best75%Amean Time      150.36 (   0.00%)      141.92 (   5.61%)
Best50%Amean Time      150.04 (   0.00%)      141.69 (   5.56%)
Best25%Amean Time      149.85 (   0.00%)      141.38 (   5.65%)

With a tiny number of files, each file truncated has resident page cache
and it shows that time to truncate is roughtly 5-6% with some minor
jitter.

                                      4.14.0-rc4             4.14.0-rc4
                                   janbatch-v1r1            oneirq-v1r1
Hmean     SeqCreate ops         65.27 (   0.00%)       81.86 (  25.43%)
Hmean     SeqCreate read        39.48 (   0.00%)       47.44 (  20.16%)
Hmean     SeqCreate del      24963.95 (   0.00%)    26319.99 (   5.43%)
Hmean     RandCreate ops        65.47 (   0.00%)       82.01 (  25.26%)
Hmean     RandCreate read       42.04 (   0.00%)       51.75 (  23.09%)
Hmean     RandCreate del     23377.66 (   0.00%)    23764.79 (   1.66%)

As expected, there is a small gain for the delete operation.

[mgorman@techsingularity.net: use page_private and set_page_private helpers]
  Link: http://lkml.kernel.org/r/20171018101547.mjycw7zreb66jzpa@techsingularity.net
Link: http://lkml.kernel.org/r/20171018075952.10627-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Jan Kara <jack@suse.cz>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-11-15 18:21:06 -08:00
Mel Gorman 4647706ebe mm: always flush VMA ranges affected by zap_page_range 
Nadav Amit report zap_page_range only specifies that the caller protect
the VMA list but does not specify whether it is held for read or write
with callers using either.  madvise holds mmap_sem for read meaning that
a parallel zap operation can unmap PTEs which are then potentially
skipped by madvise which potentially returns with stale TLB entries
present.  While the API could be extended, it would be a difficult API
to use.  This patch causes zap_page_range() to always consider flushing
the full affected range.  For small ranges or sparsely populated
mappings, this may result in one additional spurious TLB flush.  For
larger ranges, it is possible that the TLB has already been flushed and
the overhead is negligible.  Either way, this approach is safer overall
and avoids stale entries being present when madvise returns.

This can be illustrated with the following program provided by Nadav
Amit and slightly modified.  With the patch applied, it has an exit code
of 0 indicating a stale TLB entry did not leak to userspace.

---8<---

volatile int sync_step = 0;
volatile char *p;

static inline unsigned long rdtsc()
{
	unsigned long hi, lo;
	__asm__ __volatile__ ("rdtsc" : "=a"(lo), "=d"(hi));
	 return lo | (hi << 32);
}

static inline void wait_rdtsc(unsigned long cycles)
{
	unsigned long tsc = rdtsc();

	while (rdtsc() - tsc < cycles);
}

void *big_madvise_thread(void *ign)
{
	sync_step = 1;
	while (sync_step != 2);
	madvise((void*)p, PAGE_SIZE * N_PAGES, MADV_DONTNEED);
}

int main(void)
{
	pthread_t aux_thread;

	p = mmap(0, PAGE_SIZE * N_PAGES, PROT_READ|PROT_WRITE,
		 MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);

	memset((void*)p, 8, PAGE_SIZE * N_PAGES);

	pthread_create(&aux_thread, NULL, big_madvise_thread, NULL);
	while (sync_step != 1);

	*p = 8;		// Cache in TLB
	sync_step = 2;
	wait_rdtsc(100000);
	madvise((void*)p, PAGE_SIZE, MADV_DONTNEED);
	printf("data: %d (%s)\n", *p, (*p == 8 ? "stale, broken" : "cleared, fine"));
	return *p == 8 ? -1 : 0;
}
---8<---

Link: http://lkml.kernel.org/r/20170725101230.5v7gvnjmcnkzzql3@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@suse.de>
Reported-by: Nadav Amit <nadav.amit@gmail.com>
Cc: Andy Lutomirski <luto@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-09-06 17:27:26 -07:00
Mel Gorman e832bf48c8 audit: Reduce overhead using a coarse clock 
Commit 2115bb250f ("audit: Use timespec64 to represent audit timestamps")
noted that audit timestamps were not y2038 safe and used a 64-bit
timestamp. In itself, this makes sense but the conversion was from
CURRENT_TIME to ktime_get_real_ts64() which is a heavier call to record
an accurate timestamp which is required in some, but not all, cases. The
impact is that when auditd is running without any rules that all syscalls
have higher overhead. This is visible in the sysbench-thread benchmark as
a 11.5% performance hit. That benchmark is dumb as rocks but it's also
visible in redis as an 8-10% hit on all operations which is of greater
concern. It is somewhat stupid of audit to track syscalls without any
rules related to syscalls but that is how it behaves.

The overhead can be directly measured with perf comparing 4.9 with 4.12

4.9
     7.76%  sysbench         [kernel.vmlinux]    [k] __schedule
     7.62%  sysbench         [kernel.vmlinux]    [k] _raw_spin_lock
     7.37%  sysbench         libpthread-2.22.so  [.] __lll_lock_elision
     7.29%  sysbench         [kernel.vmlinux]    [.] syscall_return_via_sysret
     6.59%  sysbench         [kernel.vmlinux]    [k] native_sched_clock
     5.21%  sysbench         libc-2.22.so        [.] __sched_yield
     4.38%  sysbench         [kernel.vmlinux]    [k] entry_SYSCALL_64
     4.28%  sysbench         [kernel.vmlinux]    [k] do_syscall_64
     3.49%  sysbench         libpthread-2.22.so  [.] __lll_unlock_elision
     3.13%  sysbench         [kernel.vmlinux]    [k] __audit_syscall_exit
     2.87%  sysbench         [kernel.vmlinux]    [k] update_curr
     2.73%  sysbench         [kernel.vmlinux]    [k] pick_next_task_fair
     2.31%  sysbench         [kernel.vmlinux]    [k] syscall_trace_enter
     2.20%  sysbench         [kernel.vmlinux]    [k] __audit_syscall_entry
.....
     0.00%  swapper          [kernel.vmlinux]    [k] read_tsc

4.12
     7.84%  sysbench         [kernel.vmlinux]    [k] __schedule
     7.05%  sysbench         [kernel.vmlinux]    [k] _raw_spin_lock
     6.57%  sysbench         libpthread-2.22.so  [.] __lll_lock_elision
     6.50%  sysbench         [kernel.vmlinux]    [.] syscall_return_via_sysret
     5.95%  sysbench         [kernel.vmlinux]    [k] read_tsc
     5.71%  sysbench         [kernel.vmlinux]    [k] native_sched_clock
     4.78%  sysbench         libc-2.22.so        [.] __sched_yield
     4.30%  sysbench         [kernel.vmlinux]    [k] entry_SYSCALL_64
     3.94%  sysbench         [kernel.vmlinux]    [k] do_syscall_64
     3.37%  sysbench         libpthread-2.22.so  [.] __lll_unlock_elision
     3.32%  sysbench         [kernel.vmlinux]    [k] __audit_syscall_exit
     2.91%  sysbench         [kernel.vmlinux]    [k] __getnstimeofday64

Note the additional overhead from read_tsc which goes from 0% to 5.95%.
This is on a single-socket E3-1230 but similar overheads have been measured
on an older machine which the patch also eliminates.

The patch in question has no explanation as to why a fully-accurate timestamp
is required and is likely an oversight.  Using a coarser, but monotically
increasing, timestamp the overhead can be eliminated.  While it can be
worked around by configuring or disabling audit, it's tricky enough to
detect that a kernel fix is justified. With this patch, we see the following;

sysbenchthread
                              4.9.0                 4.12.0                 4.12.0
                            vanilla                vanilla            coarse-v1r1
Amean     1         1.49 (   0.00%)        1.66 ( -11.42%)        1.51 (  -1.34%)
Amean     3         1.48 (   0.00%)        1.65 ( -11.45%)        1.50 (  -0.96%)
Amean     5         1.49 (   0.00%)        1.67 ( -12.31%)        1.51 (  -1.83%)
Amean     7         1.49 (   0.00%)        1.66 ( -11.72%)        1.50 (  -0.67%)
Amean     12        1.48 (   0.00%)        1.65 ( -11.57%)        1.52 (  -2.89%)
Amean     16        1.49 (   0.00%)        1.65 ( -11.13%)        1.51 (  -1.73%)

The benchmark is reporting the time required for different thread counts to
lock/unlock a private mutex which, while dense, demonstrates the syscall
overhead. This is showing that 4.12 took a 11-12% hit but the overhead is
almost eliminated by the patch. While the variance is not reported here,
it's well within the noise with the patch applied.

Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Deepa Dinamani <deepa.kernel@gmail.com>
Signed-off-by: Paul Moore <paul@paul-moore.com>
 2017-09-05 09:46:54 -04:00
Mel Gorman c461ad6a63 mm, madvise: ensure poisoned pages are removed from per-cpu lists 
Wendy Wang reported off-list that a RAS HWPOISON-SOFT test case failed
and bisected it to the commit 479f854a20 ("mm, page_alloc: defer
debugging checks of pages allocated from the PCP").

The problem is that a page that was poisoned with madvise() is reused.
The commit removed a check that would trigger if DEBUG_VM was enabled
but re-enabling the check only fixes the problem as a side-effect by
printing a bad_page warning and recovering.

The root of the problem is that an madvise() can leave a poisoned page
on the per-cpu list.  This patch drains all per-cpu lists after pages
are poisoned so that they will not be reused.  Wendy reports that the
test case in question passes with this patch applied.  While this could
be done in a targeted fashion, it is over-complicated for such a rare
operation.

Link: http://lkml.kernel.org/r/20170828133414.7qro57jbepdcyz5x@techsingularity.net
Fixes: 479f854a20 ("mm, page_alloc: defer debugging checks of pages allocated from the PCP")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: Wang, Wendy <wendy.wang@intel.com>
Tested-by: Wang, Wendy <wendy.wang@intel.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: "Hansen, Dave" <dave.hansen@intel.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Naoya Horiguchi <nao.horiguchi@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-08-31 16:33:15 -07:00
Mel Gorman e716f2eb24 mm, vmscan: prevent kswapd sleeping prematurely due to mismatched classzone_idx 
kswapd is woken to reclaim a node based on a failed allocation request
from any eligible zone.  Once reclaiming in balance_pgdat(), it will
continue reclaiming until there is an eligible zone available for the
zone it was woken for.  kswapd tracks what zone it was recently woken
for in pgdat->kswapd_classzone_idx.  If it has not been woken recently,
this zone will be 0.

However, the decision on whether to sleep is made on
kswapd_classzone_idx which is 0 without a recent wakeup request and that
classzone does not account for lowmem reserves.  This allows kswapd to
sleep when a low small zone such as ZONE_DMA is balanced for a GFP_DMA
request even if a stream of allocations cannot use that zone.  While
kswapd may be woken again shortly in the near future there are two
consequences -- the pgdat bits that control congestion are cleared
prematurely and direct reclaim is more likely as kswapd slept
prematurely.

This patch flips kswapd_classzone_idx to default to MAX_NR_ZONES (an
invalid index) when there has been no recent wakeups.  If there are no
wakeups, it'll decide whether to sleep based on the highest possible
zone available (MAX_NR_ZONES - 1).  It then becomes critical that the
"pgdat balanced" decisions during reclaim and when deciding to sleep are
the same.  If there is a mismatch, kswapd can stay awake continually
trying to balance tiny zones.

simoop was used to evaluate it again.  Two of the preparation patches
regressed the workload so they are included as the second set of
results.  Otherwise this patch looks artifically excellent

                                         4.11.0-rc1            4.11.0-rc1            4.11.0-rc1
                                            vanilla              clear-v2          keepawake-v2
Amean    p50-Read             21670074.18 (  0.00%) 19786774.76 (  8.69%) 22668332.52 ( -4.61%)
Amean    p95-Read             25456267.64 (  0.00%) 24101956.27 (  5.32%) 26738688.00 ( -5.04%)
Amean    p99-Read             29369064.73 (  0.00%) 27691872.71 (  5.71%) 30991404.52 ( -5.52%)
Amean    p50-Write                1390.30 (  0.00%)     1011.91 ( 27.22%)      924.91 ( 33.47%)
Amean    p95-Write              412901.57 (  0.00%)    34874.98 ( 91.55%)     1362.62 ( 99.67%)
Amean    p99-Write             6668722.09 (  0.00%)   575449.60 ( 91.37%)    16854.04 ( 99.75%)
Amean    p50-Allocation          78714.31 (  0.00%)    84246.26 ( -7.03%)    74729.74 (  5.06%)
Amean    p95-Allocation         175533.51 (  0.00%)   400058.43 (-127.91%)   101609.74 ( 42.11%)
Amean    p99-Allocation         247003.02 (  0.00%) 10905600.00 (-4315.17%)   125765.57 ( 49.08%)

With this patch on top, write and allocation latencies are massively
improved.  The read latencies are slightly impaired but it's worth
noting that this is mostly due to the IO scheduler and not directly
related to reclaim.  The vmstats are a bit of a mix but the relevant
ones are as follows;

                            4.10.0-rc7  4.10.0-rc7  4.10.0-rc7
                          mmots-20170209 clear-v1r25keepawake-v1r25
Swap Ins                             0           0           0
Swap Outs                            0         608           0
Direct pages scanned           6910672     3132699     6357298
Kswapd pages scanned          57036946    82488665    56986286
Kswapd pages reclaimed        55993488    63474329    55939113
Direct pages reclaimed         6905990     2964843     6352115
Kswapd efficiency                  98%         76%         98%
Kswapd velocity              12494.375   17597.507   12488.065
Direct efficiency                  99%         94%         99%
Direct velocity               1513.835     668.306    1393.148
Page writes by reclaim           0.000 4410243.000       0.000
Page writes file                     0     4409635           0
Page writes anon                     0         608           0
Page reclaim immediate         1036792    14175203     1042571

                            4.11.0-rc1  4.11.0-rc1  4.11.0-rc1
                               vanilla  clear-v2  keepawake-v2
Swap Ins                             0          12           0
Swap Outs                            0         838           0
Direct pages scanned           6579706     3237270     6256811
Kswapd pages scanned          61853702    79961486    54837791
Kswapd pages reclaimed        60768764    60755788    53849586
Direct pages reclaimed         6579055     2987453     6256151
Kswapd efficiency                  98%         75%         98%
Page writes by reclaim           0.000 4389496.000       0.000
Page writes file                     0     4388658           0
Page writes anon                     0         838           0
Page reclaim immediate         1073573    14473009      982507

Swap-outs are equivalent to baseline.

Direct reclaim is reduced but not eliminated.  It's worth noting that
there are two periods of direct reclaim for this workload.  The first is
when it switches from preparing the files for the actual test itself.
It's a lot of file IO followed by a lot of allocs that reclaims heavily
for a brief window.  While direct reclaim is lower with clear-v2, it is
due to kswapd scanning aggressively and trying to reclaim the world
which is not the right thing to do.  With the patches applied, there is
still direct reclaim but the phase change from "creating work files" to
starting multiple threads that allocate a lot of anonymous memory faster
than kswapd can reclaim.

Scanning/reclaim efficiency is restored by this patch.

Page writes from reclaim context are back at 0 which is ideal.

Pages immediately reclaimed after IO completes is slightly improved but
it is expected this will vary slightly.

On UMA, there is almost no change so this is not expected to be a
universal win.

[mgorman@suse.de: fix ->kswapd_classzone_idx initialization]
  Link: http://lkml.kernel.org/r/20170406174538.5msrznj6nt6qpbx5@suse.de
Link: http://lkml.kernel.org/r/20170309075657.25121-4-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shantanu Goel <sgoel01@yahoo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-05-03 15:52:09 -07:00
Mel Gorman 631b6e083e mm, vmscan: only clear pgdat congested/dirty/writeback state when balanced 
A pgdat tracks if recent reclaim encountered too many dirty, writeback
or congested pages.  The flags control whether kswapd writes pages back
from reclaim context, tags pages for immediate reclaim when IO
completes, whether processes block on wait_iff_congested and whether
kswapd blocks when too many pages marked for immediate reclaim are
encountered.

The state is cleared in a check function with side-effects.  With the
patch "mm, vmscan: fix zone balance check in prepare_kswapd_sleep", the
timing of when the bits get cleared changed.  Due to the way the check
works, it'll clear the bits if ZONE_DMA is balanced for a GFP_DMA
allocation because it does not account for lowmem reserves properly.

For the simoop workload, kswapd is not stalling when it should due to
the premature clearing, writing pages from reclaim context like crazy
and generally being unhelpful.

This patch resets the pgdat bits related to page reclaim only when
kswapd is going to sleep.  The comparison with simoop is then

                                         4.11.0-rc1            4.11.0-rc1            4.11.0-rc1
                                            vanilla           fixcheck-v2              clear-v2
Amean    p50-Read             21670074.18 (  0.00%) 20464344.18 (  5.56%) 19786774.76 (  8.69%)
Amean    p95-Read             25456267.64 (  0.00%) 25721423.64 ( -1.04%) 24101956.27 (  5.32%)
Amean    p99-Read             29369064.73 (  0.00%) 30174230.76 ( -2.74%) 27691872.71 (  5.71%)
Amean    p50-Write                1390.30 (  0.00%)     1395.28 ( -0.36%)     1011.91 ( 27.22%)
Amean    p95-Write              412901.57 (  0.00%)    37737.74 ( 90.86%)    34874.98 ( 91.55%)
Amean    p99-Write             6668722.09 (  0.00%)   666489.04 ( 90.01%)   575449.60 ( 91.37%)
Amean    p50-Allocation          78714.31 (  0.00%)    86286.22 ( -9.62%)    84246.26 ( -7.03%)
Amean    p95-Allocation         175533.51 (  0.00%)   351812.27 (-100.42%)   400058.43 (-127.91%)
Amean    p99-Allocation         247003.02 (  0.00%)  6291171.56 (-2447.00%) 10905600.00 (-4315.17%)

Read latency is improved, write latency is mostly improved but
allocation latency is regressed.  kswapd is still reclaiming
inefficiently, pages are being written back from writeback context and a
host of other issues.  However, given the change, it needed to be
spelled out why the side-effect was moved.

Link: http://lkml.kernel.org/r/20170309075657.25121-3-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shantanu Goel <sgoel01@yahoo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-05-03 15:52:09 -07:00
Mel Gorman d34b0733b4 Revert "mm, page_alloc: only use per-cpu allocator for irq-safe requests" 
This reverts commit 374ad05ab6.

While the patch worked great for userspace allocations, the fact that
softirq loses the per-cpu allocator caused problems.  It needs to be
redone taking into account that a separate list is needed for hard/soft
IRQs or alternatively find a cheap way of detecting reentry due to an
interrupt.  Both are possible but sufficiently tricky that it shouldn't
be rushed.

Jesper had one method for allowing softirqs but reported that the cost
was high enough that it performed similarly to a plain revert.  His
figures for netperf TCP_STREAM were as follows

  Baseline v4.10.0  : 60316 Mbit/s
  Current 4.11.0-rc6: 47491 Mbit/s
  Jesper's patch    : 60662 Mbit/s
  This patch        : 60106 Mbit/s

As this is a regression, I wish to revert to noirq allocator for now and
go back to the drawing board.

Link: http://lkml.kernel.org/r/20170415145350.ixy7vtrzdzve57mh@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: Tariq Toukan <ttoukan.linux@gmail.com>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-04-20 15:30:18 -07:00
Mel Gorman bd233f538d mm, page_alloc: use static global work_struct for draining per-cpu pages 
As suggested by Vlastimil Babka and Tejun Heo, this patch uses a static
work_struct to co-ordinate the draining of per-cpu pages on the
workqueue.  Only one task can drain at a time but this is better than
the previous scheme that allowed multiple tasks to send IPIs at a time.

One consideration is whether parallel requests should synchronise
against each other.  This patch does not synchronise for a global drain
as the common case for such callers is expected to be multiple parallel
direct reclaimers competing for pages when the watermark is close to
min.  Draining the per-cpu list is unlikely to make much progress and
serialising the drain is of dubious merit.  Drains are synchonrised for
callers such as memory hotplug and CMA that care about the drain being
complete when the function returns.

Link: http://lkml.kernel.org/r/20170125083038.rzb5f43nptmk7aed@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Suggested-by: Tejun Heo <tj@kernel.org>
Suggested-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-02-24 17:46:54 -08:00
Mel Gorman 374ad05ab6 mm, page_alloc: only use per-cpu allocator for irq-safe requests 
Many workloads that allocate pages are not handling an interrupt at a
time.  As allocation requests may be from IRQ context, it's necessary to
disable/enable IRQs for every page allocation.  This cost is the bulk of
the free path but also a significant percentage of the allocation path.

This patch alters the locking and checks such that only irq-safe
allocation requests use the per-cpu allocator.  All others acquire the
irq-safe zone->lock and allocate from the buddy allocator.  It relies on
disabling preemption to safely access the per-cpu structures.  It could
be slightly modified to avoid soft IRQs using it but it's not clear it's
worthwhile.

This modification may slow allocations from IRQ context slightly but the
main gain from the per-cpu allocator is that it scales better for
allocations from multiple contexts.  There is an implicit assumption
that intensive allocations from IRQ contexts on multiple CPUs from a
single NUMA node are rare and that the fast majority of scaling issues
are encountered in !IRQ contexts such as page faulting.  It's worth
noting that this patch is not required for a bulk page allocator but it
significantly reduces the overhead.

The following is results from a page allocator micro-benchmark.  Only
order-0 is interesting as higher orders do not use the per-cpu allocator

                                          4.10.0-rc2                 4.10.0-rc2
                                             vanilla               irqsafe-v1r5
Amean    alloc-odr0-1               287.15 (  0.00%)           219.00 ( 23.73%)
Amean    alloc-odr0-2               221.23 (  0.00%)           183.23 ( 17.18%)
Amean    alloc-odr0-4               187.00 (  0.00%)           151.38 ( 19.05%)
Amean    alloc-odr0-8               167.54 (  0.00%)           132.77 ( 20.75%)
Amean    alloc-odr0-16              156.00 (  0.00%)           123.00 ( 21.15%)
Amean    alloc-odr0-32              149.00 (  0.00%)           118.31 ( 20.60%)
Amean    alloc-odr0-64              138.77 (  0.00%)           116.00 ( 16.41%)
Amean    alloc-odr0-128             145.00 (  0.00%)           118.00 ( 18.62%)
Amean    alloc-odr0-256             136.15 (  0.00%)           125.00 (  8.19%)
Amean    alloc-odr0-512             147.92 (  0.00%)           121.77 ( 17.68%)
Amean    alloc-odr0-1024            147.23 (  0.00%)           126.15 ( 14.32%)
Amean    alloc-odr0-2048            155.15 (  0.00%)           129.92 ( 16.26%)
Amean    alloc-odr0-4096            164.00 (  0.00%)           136.77 ( 16.60%)
Amean    alloc-odr0-8192            166.92 (  0.00%)           138.08 ( 17.28%)
Amean    alloc-odr0-16384           159.00 (  0.00%)           138.00 ( 13.21%)
Amean    free-odr0-1                165.00 (  0.00%)            89.00 ( 46.06%)
Amean    free-odr0-2                113.00 (  0.00%)            63.00 ( 44.25%)
Amean    free-odr0-4                 99.00 (  0.00%)            54.00 ( 45.45%)
Amean    free-odr0-8                 88.00 (  0.00%)            47.38 ( 46.15%)
Amean    free-odr0-16                83.00 (  0.00%)            46.00 ( 44.58%)
Amean    free-odr0-32                80.00 (  0.00%)            44.38 ( 44.52%)
Amean    free-odr0-64                72.62 (  0.00%)            43.00 ( 40.78%)
Amean    free-odr0-128               78.00 (  0.00%)            42.00 ( 46.15%)
Amean    free-odr0-256               80.46 (  0.00%)            57.00 ( 29.16%)
Amean    free-odr0-512               96.38 (  0.00%)            64.69 ( 32.88%)
Amean    free-odr0-1024             107.31 (  0.00%)            72.54 ( 32.40%)
Amean    free-odr0-2048             108.92 (  0.00%)            78.08 ( 28.32%)
Amean    free-odr0-4096             113.38 (  0.00%)            82.23 ( 27.48%)
Amean    free-odr0-8192             112.08 (  0.00%)            82.85 ( 26.08%)
Amean    free-odr0-16384            110.38 (  0.00%)            81.92 ( 25.78%)
Amean    total-odr0-1               452.15 (  0.00%)           308.00 ( 31.88%)
Amean    total-odr0-2               334.23 (  0.00%)           246.23 ( 26.33%)
Amean    total-odr0-4               286.00 (  0.00%)           205.38 ( 28.19%)
Amean    total-odr0-8               255.54 (  0.00%)           180.15 ( 29.50%)
Amean    total-odr0-16              239.00 (  0.00%)           169.00 ( 29.29%)
Amean    total-odr0-32              229.00 (  0.00%)           162.69 ( 28.96%)
Amean    total-odr0-64              211.38 (  0.00%)           159.00 ( 24.78%)
Amean    total-odr0-128             223.00 (  0.00%)           160.00 ( 28.25%)
Amean    total-odr0-256             216.62 (  0.00%)           182.00 ( 15.98%)
Amean    total-odr0-512             244.31 (  0.00%)           186.46 ( 23.68%)
Amean    total-odr0-1024            254.54 (  0.00%)           198.69 ( 21.94%)
Amean    total-odr0-2048            264.08 (  0.00%)           208.00 ( 21.24%)
Amean    total-odr0-4096            277.38 (  0.00%)           219.00 ( 21.05%)
Amean    total-odr0-8192            279.00 (  0.00%)           220.92 ( 20.82%)
Amean    total-odr0-16384           269.38 (  0.00%)           219.92 ( 18.36%)

This is the alloc, free and total overhead of allocating order-0 pages
in batches of 1 page up to 16384 pages.  Avoiding disabling/enabling
overhead massively reduces overhead.  Alloc overhead is roughly reduced
by 14-20% in most cases.  The free path is reduced by 26-46% and the
total reduction is significant.

Many users require zeroing of pages from the page allocator which is the
vast cost of allocation.  Hence, the impact on a basic page faulting
benchmark is not that significant

                              4.10.0-rc2            4.10.0-rc2
                                 vanilla          irqsafe-v1r5
Hmean    page_test   656632.98 (  0.00%)   675536.13 (  2.88%)
Hmean    brk_test   3845502.67 (  0.00%)  3867186.94 (  0.56%)
Stddev   page_test    10543.29 (  0.00%)     4104.07 ( 61.07%)
Stddev   brk_test     33472.36 (  0.00%)    15538.39 ( 53.58%)
CoeffVar page_test        1.61 (  0.00%)        0.61 ( 62.15%)
CoeffVar brk_test         0.87 (  0.00%)        0.40 ( 53.84%)
Max      page_test   666513.33 (  0.00%)   678640.00 (  1.82%)
Max      brk_test   3882800.00 (  0.00%)  3887008.66 (  0.11%)

This is from aim9 and the most notable outcome is that fault variability
is reduced by the patch.  The headline improvement is small as the
overall fault cost, zeroing, page table insertion etc dominate relative
to disabling/enabling IRQs in the per-cpu allocator.

Similarly, little benefit was seen on networking benchmarks both
localhost and between physical server/clients where other costs
dominate.  It's possible that this will only be noticable on very high
speed networks.

Jesper Dangaard Brouer independently tested this with a separate
microbenchmark from
  https://github.com/netoptimizer/prototype-kernel/tree/master/kernel/mm/bench

Micro-benchmarked with [1] page_bench02:
 modprobe page_bench02 page_order=0 run_flags=$((2#010)) loops=$((10**8)); \
  rmmod page_bench02 ; dmesg --notime | tail -n 4

Compared to baseline: 213 cycles(tsc) 53.417 ns
 - against this     : 184 cycles(tsc) 46.056 ns
 - Saving           : -29 cycles
 - Very close to expected 27 cycles saving [see below [2]]

Micro benchmarking via time_bench_sample[3], we get the cost of these
operations:

 time_bench: Type:for_loop                 Per elem: 0 cycles(tsc) 0.232 ns (step:0)
 time_bench: Type:spin_lock_unlock         Per elem: 33 cycles(tsc) 8.334 ns (step:0)
 time_bench: Type:spin_lock_unlock_irqsave Per elem: 62 cycles(tsc) 15.607 ns (step:0)
 time_bench: Type:irqsave_before_lock      Per elem: 57 cycles(tsc) 14.344 ns (step:0)
 time_bench: Type:spin_lock_unlock_irq     Per elem: 34 cycles(tsc) 8.560 ns (step:0)
 time_bench: Type:simple_irq_disable_before_lock Per elem: 37 cycles(tsc) 9.289 ns (step:0)
 time_bench: Type:local_BH_disable_enable  Per elem: 19 cycles(tsc) 4.920 ns (step:0)
 time_bench: Type:local_IRQ_disable_enable Per elem: 7 cycles(tsc) 1.864 ns (step:0)
 time_bench: Type:local_irq_save_restore   Per elem: 38 cycles(tsc) 9.665 ns (step:0)
 [Mel's patch removes a ^^^^^^^^^^^^^^^^]            ^^^^^^^^^ expected saving - preempt cost
 time_bench: Type:preempt_disable_enable   Per elem: 11 cycles(tsc) 2.794 ns (step:0)
 [adds a preempt  ^^^^^^^^^^^^^^^^^^^^^^]            ^^^^^^^^^ adds this cost
 time_bench: Type:funcion_call_cost        Per elem: 6 cycles(tsc) 1.689 ns (step:0)
 time_bench: Type:func_ptr_call_cost       Per elem: 11 cycles(tsc) 2.767 ns (step:0)
 time_bench: Type:page_alloc_put           Per elem: 211 cycles(tsc) 52.803 ns (step:0)

Thus, expected improvement is: 38-11 = 27 cycles.

[mgorman@techsingularity.net: s/preempt_enable_no_resched/preempt_enable/]
  Link: http://lkml.kernel.org/r/20170208143128.25ahymqlyspjcixu@techsingularity.net
Link: http://lkml.kernel.org/r/20170123153906.3122-5-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-02-24 17:46:54 -08:00
Mel Gorman 0ccce3b924 mm, page_alloc: drain per-cpu pages from workqueue context 
The per-cpu page allocator can be drained immediately via
drain_all_pages() which sends IPIs to every CPU.  In the next patch, the
per-cpu allocator will only be used for interrupt-safe allocations which
prevents draining it from IPI context.  This patch uses workqueues to
drain the per-cpu lists instead.

This is slower but no slowdown during intensive reclaim was measured and
the paths that use drain_all_pages() are not that sensitive to
performance.  This is particularly true as the path would only be
triggered when reclaim is failing.  It also makes a some sense to avoid
storming a machine with IPIs when it's under memory pressure.  Arguably,
it should be further adjusted so that only one caller at a time is
draining pages but it's beyond the scope of the current patch.

Link: http://lkml.kernel.org/r/20170123153906.3122-4-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-02-24 17:46:54 -08:00
Mel Gorman 9cd7555875 mm, page_alloc: split alloc_pages_nodemask() 
alloc_pages_nodemask does a number of preperation steps that determine
what zones can be used for the allocation depending on a variety of
factors.  This is fine but a hypothetical caller that wanted multiple
order-0 pages has to do the preparation steps multiple times.  This
patch structures __alloc_pages_nodemask such that it's relatively easy
to build a bulk order-0 page allocator.  There is no functional change.

Link: http://lkml.kernel.org/r/20170123153906.3122-3-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-02-24 17:46:54 -08:00
Mel Gorman 066b239355 mm, page_alloc: split buffered_rmqueue() 
Patch series "Use per-cpu allocator for !irq requests and prepare for a
bulk allocator", v5.

This series is motivated by a conversation led by Jesper Dangaard Brouer
at the last LSF/MM proposing a generic page pool for DMA-coherent pages.
Part of his motivation was due to the overhead of allocating multiple
order-0 that led some drivers to use high-order allocations and
splitting them.  This is very slow in some cases.

The first two patches in this series restructure the page allocator such
that it is relatively easy to introduce an order-0 bulk page allocator.
A patch exists to do that and has been handed over to Jesper until an
in-kernel users is created.  The third patch prevents the per-cpu
allocator being drained from IPI context as that can potentially corrupt
the list after patch four is merged.  The final patch alters the per-cpu
alloctor to make it exclusive to !irq requests.  This cuts
allocation/free overhead by roughly 30%.

Performance tests from both Jesper and me are included in the patch.

This patch (of 4):

buffered_rmqueue removes a page from a given zone and uses the per-cpu
list for order-0.  This is fine but a hypothetical caller that wanted
multiple order-0 pages has to disable/reenable interrupts multiple
times.  This patch structures buffere_rmqueue such that it's relatively
easy to build a bulk order-0 page allocator.  There is no functional
change.

[mgorman@techsingularity.net: failed per-cpu refill may blow up]
  Link: http://lkml.kernel.org/r/20170124112723.mshmgwq2ihxku2um@techsingularity.net
Link: http://lkml.kernel.org/r/20170123153906.3122-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2017-02-24 17:46:54 -08:00
Mel Gorman a6de734bc0 mm, page_alloc: keep pcp count and list contents in sync if struct page is corrupted 
Vlastimil Babka pointed out that commit 479f854a20 ("mm, page_alloc:
defer debugging checks of pages allocated from the PCP") will allow the
per-cpu list counter to be out of sync with the per-cpu list contents if
a struct page is corrupted.

The consequence is an infinite loop if the per-cpu lists get fully
drained by free_pcppages_bulk because all the lists are empty but the
count is positive.  The infinite loop occurs here

                do {
                        batch_free++;
                        if (++migratetype == MIGRATE_PCPTYPES)
                                migratetype = 0;
                        list = &pcp->lists[migratetype];
                } while (list_empty(list));

What the user sees is a bad page warning followed by a soft lockup with
interrupts disabled in free_pcppages_bulk().

This patch keeps the accounting in sync.

Fixes: 479f854a20 ("mm, page_alloc: defer debugging checks of pages allocated from the PCP")
Link: http://lkml.kernel.org/r/20161202112951.23346-2-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@suse.de>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: <stable@vger.kernel.org>	[4.7+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-12-12 18:55:08 -08:00
Mel Gorman 6aa303defb mm, vmscan: only allocate and reclaim from zones with pages managed by the buddy allocator 
Firmware Assisted Dump (FA_DUMP) on ppc64 reserves substantial amounts
of memory when booting a secondary kernel.  Srikar Dronamraju reported
that multiple nodes may have no memory managed by the buddy allocator
but still return true for populated_zone().

Commit 1d82de618d ("mm, vmscan: make kswapd reclaim in terms of
nodes") was reported to cause kswapd to spin at 100% CPU usage when
fadump was enabled.  The old code happened to deal with the situation of
a populated node with zero free pages by co-incidence but the current
code tries to reclaim populated zones without realising that is
impossible.

We cannot just convert populated_zone() as many existing users really
need to check for present_pages.  This patch introduces a managed_zone()
helper and uses it in the few cases where it is critical that the check
is made for managed pages -- zonelist construction and page reclaim.

Link: http://lkml.kernel.org/r/20160831195104.GB8119@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Tested-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-09-01 17:52:01 -07:00
Mel Gorman 2f95ff90b9 proc, meminfo: use correct helpers for calculating LRU sizes in meminfo 
meminfo_proc_show() and si_mem_available() are using the wrong helpers
for calculating the size of the LRUs.  The user-visible impact is that
there appears to be an abnormally high number of unevictable pages.

Link: http://lkml.kernel.org/r/20160805105805.GR2799@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Dave Chinner <david@fromorbit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-08-11 16:58:13 -07:00
Mel Gorman b4911ea2bc mm: initialise per_cpu_nodestats for all online pgdats at boot 
Paul Mackerras and Reza Arbab reported that machines with memoryless
nodes fail when vmstats are refreshed.  Paul reported an oops as follows

  Unable to handle kernel paging request for data at address 0xff7a10000
  Faulting instruction address: 0xc000000000270cd0
  Oops: Kernel access of bad area, sig: 11 [#1]
  SMP NR_CPUS=2048 NUMA PowerNV
  Modules linked in:
  CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.7.0-kvm+ #118
  task: c000000ff0680010 task.stack: c000000ff0704000
  NIP: c000000000270cd0 LR: c000000000270ce8 CTR: 0000000000000000
  REGS: c000000ff0707900 TRAP: 0300   Not tainted  (4.7.0-kvm+)
  MSR: 9000000102009033 <SF,HV,VEC,EE,ME,IR,DR,RI,LE,TM[E]>  CR: 846b6824  XER: 20000000
  CFAR: c000000000008768 DAR: 0000000ff7a10000 DSISR: 42000000 SOFTE: 1
  NIP refresh_zone_stat_thresholds+0x80/0x240
  LR refresh_zone_stat_thresholds+0x98/0x240
  Call Trace:
    refresh_zone_stat_thresholds+0xb8/0x240 (unreliable)

Both supplied potential fixes but one potentially misses checks and
another had redundant initialisations.  This version initialises
per_cpu_nodestats on a per-pgdat basis instead of on a per-zone basis.

Link: http://lkml.kernel.org/r/20160804092404.GI2799@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: Paul Mackerras <paulus@ozlabs.org>
Reported-by: Reza Arbab <arbab@linux.vnet.ibm.com>
Tested-by: Reza Arbab <arbab@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-08-04 20:02:09 -04:00
Mel Gorman d7f05528ee mm, vmscan: account for skipped pages as a partial scan 
Page reclaim determines whether a pgdat is unreclaimable by examining
how many pages have been scanned since a page was freed and comparing
that to the LRU sizes.  Skipped pages are not reclaim candidates but
contribute to scanned.  This can prematurely mark a pgdat as
unreclaimable and trigger an OOM kill.

This patch accounts for skipped pages as a partial scan so that an
unreclaimable pgdat will still be marked as such but by scaling the cost
of a skip, it'll avoid the pgdat being marked prematurely.

Link: http://lkml.kernel.org/r/1469110261-7365-6-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman f8d1a31163 mm: consider whether to decivate based on eligible zones inactive ratio 
Minchan Kim reported that with per-zone lru state it was possible to
identify that a normal zone with 8^M anonymous pages could trigger OOM
with non-atomic order-0 allocations as all pages in the zone were in the
active list.

   gfp_mask=0x26004c0(GFP_KERNEL|__GFP_REPEAT|__GFP_NOTRACK), order=0
   Call Trace:
     __alloc_pages_nodemask+0xe52/0xe60
     ? new_slab+0x39c/0x3b0
     new_slab+0x39c/0x3b0
     ___slab_alloc.constprop.87+0x6da/0x840
     ? __alloc_skb+0x3c/0x260
     ? enqueue_task_fair+0x73/0xbf0
     ? poll_select_copy_remaining+0x140/0x140
     __slab_alloc.isra.81.constprop.86+0x40/0x6d
     ? __alloc_skb+0x3c/0x260
     kmem_cache_alloc+0x22c/0x260
     ? __alloc_skb+0x3c/0x260
     __alloc_skb+0x3c/0x260
     alloc_skb_with_frags+0x4e/0x1a0
     sock_alloc_send_pskb+0x16a/0x1b0
     ? wait_for_unix_gc+0x31/0x90
     unix_stream_sendmsg+0x28d/0x340
     sock_sendmsg+0x2d/0x40
     sock_write_iter+0x6c/0xc0
     __vfs_write+0xc0/0x120
     vfs_write+0x9b/0x1a0
     ? __might_fault+0x49/0xa0
     SyS_write+0x44/0x90
     do_fast_syscall_32+0xa6/0x1e0

   Mem-Info:
   active_anon:101103 inactive_anon:102219 isolated_anon:0
    active_file:503 inactive_file:544 isolated_file:0
    unevictable:0 dirty:0 writeback:34 unstable:0
    slab_reclaimable:6298 slab_unreclaimable:74669
    mapped:863 shmem:0 pagetables:100998 bounce:0
    free:23573 free_pcp:1861 free_cma:0
   Node 0 active_anon:404412kB inactive_anon:409040kB active_file:2012kB inactive_file:2176kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:3452kB dirty:0kB writeback:136kB shmem:0kB writeback_tmp:0kB unstable:0kB pages_scanned:1320845 all_unreclaimable? yes
   DMA free:3296kB min:68kB low:84kB high:100kB active_anon:5540kB inactive_anon:0kB active_file:0kB inactive_file:0kB present:15992kB managed:15916kB mlocked:0kB slab_reclaimable:248kB slab_unreclaimable:2628kB kernel_stack:792kB pagetables:2316kB bounce:0kB free_pcp:0kB local_pcp:0kB free_cma:0kB
   lowmem_reserve[]: 0 809 1965 1965
   Normal free:3600kB min:3604kB low:4504kB high:5404kB active_anon:86304kB inactive_anon:0kB active_file:160kB inactive_file:376kB present:897016kB managed:858524kB mlocked:0kB slab_reclaimable:24944kB slab_unreclaimable:296048kB kernel_stack:163832kB pagetables:35892kB bounce:0kB free_pcp:3076kB local_pcp:656kB free_cma:0kB
   lowmem_reserve[]: 0 0 9247 9247
   HighMem free:86156kB min:512kB low:1796kB high:3080kB active_anon:312852kB inactive_anon:410024kB active_file:1924kB inactive_file:2012kB present:1183736kB managed:1183736kB mlocked:0kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:365784kB bounce:0kB free_pcp:3868kB local_pcp:720kB free_cma:0kB
   lowmem_reserve[]: 0 0 0 0
   DMA: 8*4kB (UM) 8*8kB (UM) 4*16kB (M) 2*32kB (UM) 2*64kB (UM) 1*128kB (M) 3*256kB (UME) 2*512kB (UE) 1*1024kB (E) 0*2048kB 0*4096kB = 3296kB
   Normal: 240*4kB (UME) 160*8kB (UME) 23*16kB (ME) 3*32kB (UE) 3*64kB (UME) 2*128kB (ME) 1*256kB (U) 0*512kB 0*1024kB 0*2048kB 0*4096kB = 3408kB
   HighMem: 10942*4kB (UM) 3102*8kB (UM) 866*16kB (UM) 76*32kB (UM) 11*64kB (UM) 4*128kB (UM) 1*256kB (M) 0*512kB 0*1024kB 0*2048kB 0*4096kB = 86344kB
   Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
   54409 total pagecache pages
   53215 pages in swap cache
   Swap cache stats: add 300982, delete 247765, find 157978/226539
   Free swap  = 3803244kB
   Total swap = 4192252kB
   524186 pages RAM
   295934 pages HighMem/MovableOnly
   9642 pages reserved
   0 pages cma reserved

The problem is due to the active deactivation logic in
inactive_list_is_low:

	Node 0 active_anon:404412kB inactive_anon:409040kB

IOW, (inactive_anon of node * inactive_ratio > active_anon of node) due
to highmem anonymous stat so VM never deactivates normal zone's
anonymous pages.

This patch is a modified version of Minchan's original solution but
based upon it.  The problem with Minchan's patch is that any low zone
with an imbalanced list could force a rotation.

In this patch, a zone-constrained global reclaim will rotate the list if
the inactive/active ratio of all eligible zones needs to be corrected.
It is possible that higher zone pages will be initially rotated
prematurely but this is the safer choice to maintain overall LRU age.

Link: http://lkml.kernel.org/r/20160722090929.GJ10438@techsingularity.net
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 5a1c84b404 mm: remove reclaim and compaction retry approximations 
If per-zone LRU accounting is available then there is no point
approximating whether reclaim and compaction should retry based on pgdat
statistics.  This is effectively a revert of "mm, vmstat: remove zone
and node double accounting by approximating retries" with the difference
that inactive/active stats are still available.  This preserves the
history of why the approximation was retried and why it had to be
reverted to handle OOM kills on 32-bit systems.

Link: http://lkml.kernel.org/r/1469110261-7365-4-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman bb4cc2bea6 mm, vmscan: remove highmem_file_pages 
With the reintroduction of per-zone LRU stats, highmem_file_pages is
redundant so remove it.

[mgorman@techsingularity.net: wrong stat is being accumulated in highmem_dirtyable_memory]
  Link: http://lkml.kernel.org/r/20160725092324.GM10438@techsingularity.netLink: http://lkml.kernel.org/r/1469110261-7365-3-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 785b99febb mm, vmscan: release/reacquire lru_lock on pgdat change 
With node-lru, the locking is based on the pgdat.  As Minchan pointed
out, there is an opportunity to reduce LRU lock release/acquire in
check_move_unevictable_pages by only changing lock on a pgdat change.

[mgorman@techsingularity.net: remove double initialisation]
  Link: http://lkml.kernel.org/r/20160719074835.GC10438@techsingularity.net
Link: http://lkml.kernel.org/r/1468853426-12858-3-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 22fecdf5e1 mm, vmscan: remove redundant check in shrink_zones() 
As pointed out by Minchan Kim, shrink_zones() checks for populated zones
in a zonelist but a zonelist can never contain unpopulated zones.  While
it's not related to the node-lru series, it can be cleaned up now.

Link: http://lkml.kernel.org/r/1468853426-12858-2-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Suggested-by: Minchan Kim <minchan@kernel.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 7ee36a14f0 mm, vmscan: Update all zone LRU sizes before updating memcg 
Minchan Kim reported setting the following warning on a 32-bit system
although it can affect 64-bit systems.

  WARNING: CPU: 4 PID: 1322 at mm/memcontrol.c:998 mem_cgroup_update_lru_size+0x103/0x110
  mem_cgroup_update_lru_size(f44b4000, 1, -7): zid 1 lru_size 1 but empty
  Modules linked in:
  CPU: 4 PID: 1322 Comm: cp Not tainted 4.7.0-rc4-mm1+ #143
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
  Call Trace:
    dump_stack+0x76/0xaf
    __warn+0xea/0x110
    ? mem_cgroup_update_lru_size+0x103/0x110
    warn_slowpath_fmt+0x3b/0x40
    mem_cgroup_update_lru_size+0x103/0x110
    isolate_lru_pages.isra.61+0x2e2/0x360
    shrink_active_list+0xac/0x2a0
    ? __delay+0xe/0x10
    shrink_node_memcg+0x53c/0x7a0
    shrink_node+0xab/0x2a0
    do_try_to_free_pages+0xc6/0x390
    try_to_free_pages+0x245/0x590

LRU list contents and counts are updated separately.  Counts are updated
before pages are added to the LRU and updated after pages are removed.
The warning above is from a check in mem_cgroup_update_lru_size that
ensures that list sizes of zero are empty.

The problem is that node-lru needs to account for highmem pages if
CONFIG_HIGHMEM is set.  One impact of the implementation is that the
sizes are updated in multiple passes when pages from multiple zones were
isolated.  This happens whether HIGHMEM is set or not.  When multiple
zones are isolated, it's possible for a debugging check in memcg to be
tripped.

This patch forces all the zone counts to be updated before the memcg
function is called.

Link: http://lkml.kernel.org/r/1468588165-12461-6-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Tested-by: Minchan Kim <minchan@kernel.org>
Reported-by: Minchan Kim <minchan@kernel.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 68eb0731c4 mm, pagevec: release/reacquire lru_lock on pgdat change 
With node-lru, the locking is based on the pgdat.  Previously it was
required that a pagevec drain released one zone lru_lock and acquired
another zone lru_lock on every zone change.  Now, it's only necessary if
the node changes.  The end-result is fewer lock release/acquires if the
pages are all on the same node but in different zones.

Link: http://lkml.kernel.org/r/1468588165-12461-4-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Minchan Kim <minchan@kernel.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman bca6759258 mm, vmstat: remove zone and node double accounting by approximating retries 
The number of LRU pages, dirty pages and writeback pages must be
accounted for on both zones and nodes because of the reclaim retry
logic, compaction retry logic and highmem calculations all depending on
per-zone stats.

Many lowmem allocations are immune from OOM kill due to a check in
__alloc_pages_may_oom for (ac->high_zoneidx < ZONE_NORMAL) since commit
03668b3ceb ("oom: avoid oom killer for lowmem allocations").  The
exception is costly high-order allocations or allocations that cannot
fail.  If the __alloc_pages_may_oom avoids OOM-kill for low-order lowmem
allocations then it would fall through to __alloc_pages_direct_compact.

This patch will blindly retry reclaim for zone-constrained allocations
in should_reclaim_retry up to MAX_RECLAIM_RETRIES.  This is not ideal
but without per-zone stats there are not many alternatives.  The impact
it that zone-constrained allocations may delay before considering the
OOM killer.

As there is no guarantee enough memory can ever be freed to satisfy
compaction, this patch avoids retrying compaction for zone-contrained
allocations.

In combination, that means that the per-node stats can be used when
deciding whether to continue reclaim using a rough approximation.  While
it is possible this will make the wrong decision on occasion, it will
not infinite loop as the number of reclaim attempts is capped by
MAX_RECLAIM_RETRIES.

The final step is calculating the number of dirtyable highmem pages.  As
those calculations only care about the global count of file pages in
highmem.  This patch uses a global counter used instead of per-zone
stats as it is sufficient.

In combination, this allows the per-zone LRU and dirty state counters to
be removed.

[mgorman@techsingularity.net: fix acct_highmem_file_pages()]
  Link: http://lkml.kernel.org/r/1468853426-12858-4-git-send-email-mgorman@techsingularity.netLink: http://lkml.kernel.org/r/1467970510-21195-35-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Suggested by: Michal Hocko <mhocko@kernel.org>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman e2ecc8a79e mm, vmstat: print node-based stats in zoneinfo file 
There are a number of stats that were previously accessible via zoneinfo
that are now invisible.  While it is possible to create a new file for
the node stats, this may be missed by users.  Instead this patch prints
the stats under the first populated zone in /proc/zoneinfo.

Link: http://lkml.kernel.org/r/1467970510-21195-34-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 7cc30fcfd2 mm: vmstat: account per-zone stalls and pages skipped during reclaim 
The vmstat allocstall was fairly useful in the general sense but
node-based LRUs change that.  It's important to know if a stall was for
an address-limited allocation request as this will require skipping
pages from other zones.  This patch adds pgstall_* counters to replace
allocstall.  The sum of the counters will equal the old allocstall so it
can be trivially recalculated.  A high number of address-limited
allocation requests may result in a lot of useless LRU scanning for
suitable pages.

As address-limited allocations require pages to be skipped, it's
important to know how much useless LRU scanning took place so this patch
adds pgskip* counters.  This yields the following model

1. The number of address-space limited stalls can be accounted for (pgstall)
2. The amount of useless work required to reclaim the data is accounted (pgskip)
3. The total number of scans is available from pgscan_kswapd and pgscan_direct
   so from that the ratio of useful to useless scans can be calculated.

[mgorman@techsingularity.net: s/pgstall/allocstall/]
  Link: http://lkml.kernel.org/r/1468404004-5085-3-git-send-email-mgorman@techsingularity.netLink: http://lkml.kernel.org/r/1467970510-21195-33-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 16709d1de1 mm: vmstat: replace __count_zone_vm_events with a zone id equivalent 
This is partially a preparation patch for more vmstat work but it also
has the slight advantage that __count_zid_vm_events is cheaper to
calculate than __count_zone_vm_events().

Link: http://lkml.kernel.org/r/1467970510-21195-32-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 3b8c0be43c mm: page_alloc: cache the last node whose dirty limit is reached 
If a page is about to be dirtied then the page allocator attempts to
limit the total number of dirty pages that exists in any given zone.
The call to node_dirty_ok is expensive so this patch records if the last
pgdat examined hit the dirty limits.  In some cases, this reduces the
number of calls to node_dirty_ok().

Link: http://lkml.kernel.org/r/1467970510-21195-31-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman e6cbd7f2ef mm, page_alloc: remove fair zone allocation policy 
The fair zone allocation policy interleaves allocation requests between
zones to avoid an age inversion problem whereby new pages are reclaimed
to balance a zone.  Reclaim is now node-based so this should no longer
be an issue and the fair zone allocation policy is not free.  This patch
removes it.

Link: http://lkml.kernel.org/r/1467970510-21195-30-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman e5146b12e2 mm, vmscan: add classzone information to tracepoints 
This is convenient when tracking down why the skip count is high because
it'll show what classzone kswapd woke up at and what zones are being
isolated.

Link: http://lkml.kernel.org/r/1467970510-21195-29-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 84c7a7771f mm, vmscan: Have kswapd reclaim from all zones if reclaiming and buffer_heads_over_limit 
The buffer_heads_over_limit limit in kswapd is inconsistent with direct
reclaim behaviour.  It may force an an attempt to reclaim from all zones
and then not reclaim at all because higher zones were balanced than
required by the original request.

This patch will causes kswapd to consider reclaiming from all zones if
buffer_heads_over_limit.  However, if there are eligible zones for the
allocation request that woke kswapd then no reclaim will occur even if
buffer_heads_over_limit.  This avoids kswapd over-reclaiming just
because buffer_heads_over_limit.

[mgorman@techsingularity.net: fix comment about buffer_heads_over_limit]
  Link: http://lkml.kernel.org/r/1468404004-5085-2-git-send-email-mgorman@techsingularity.net
Link: http://lkml.kernel.org/r/1467970510-21195-28-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman d9f21d426d mm, vmscan: avoid passing in `remaining' unnecessarily to prepare_kswapd_sleep() 
As pointed out by Minchan Kim, the first call to prepare_kswapd_sleep()
always passes in 0 for `remaining' and the second call can trivially
check the parameter in advance.

Suggested-by: Minchan Kim <minchan@kernel.org>
Link: http://lkml.kernel.org/r/1467970510-21195-27-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 4f588331bd mm, vmscan: avoid passing in classzone_idx unnecessarily to compaction_ready 
The scan_control structure has enough information available for
compaction_ready() to make a decision.  The classzone_idx manipulations
in shrink_zones() are no longer necessary as the highest populated zone
is no longer used to determine if shrink_slab should be called or not.

[mgorman@techsingularity.net remove redundant check in shrink_zones()]
  Link: http://lkml.kernel.org/r/1468588165-12461-3-git-send-email-mgorman@techsingularity.net
Link: http://lkml.kernel.org/r/1467970510-21195-26-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 970a39a363 mm, vmscan: avoid passing in classzone_idx unnecessarily to shrink_node 
shrink_node receives all information it needs about classzone_idx from
sc->reclaim_idx so remove the aliases.

Link: http://lkml.kernel.org/r/1467970510-21195-25-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman a5f5f91da6 mm: convert zone_reclaim to node_reclaim 
As reclaim is now per-node based, convert zone_reclaim to be
node_reclaim.  It is possible that a node will be reclaimed multiple
times if it has multiple zones but this is unavoidable without caching
all nodes traversed so far.  The documentation and interface to
userspace is the same from a configuration perspective and will will be
similar in behaviour unless the node-local allocation requests were also
limited to lower zones.

Link: http://lkml.kernel.org/r/1467970510-21195-24-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman 52e9f87ae8 mm, page_alloc: wake kswapd based on the highest eligible zone 
The ac_classzone_idx is used as the basis for waking kswapd and that is
based on the preferred zoneref.  If the preferred zoneref's first zone
is lower than what is available on other nodes, it's possible that
kswapd is woken on a zone with only higher, but still eligible, zones.
As classzone_idx is strictly adhered to now, it causes a problem because
eligible pages are skipped.

For example, node 0 has only DMA32 and node 1 has only NORMAL.  An
allocating context running on node 0 may wake kswapd on node 1 telling
it to skip all NORMAL pages.

Link: http://lkml.kernel.org/r/1467970510-21195-23-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00
Mel Gorman e1a556374a mm, vmscan: only wakeup kswapd once per node for the requested classzone 
kswapd is woken when zones are below the low watermark but the wakeup
decision is not taking the classzone into account.  Now that reclaim is
node-based, it is only required to wake kswapd once per node and only if
all zones are unbalanced for the requested classzone.

Note that one node might be checked multiple times if the zonelist is
ordered by node because there is no cheap way of tracking what nodes
have already been visited.  For zone-ordering, each node should be
checked only once.

Link: http://lkml.kernel.org/r/1467970510-21195-22-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 2016-07-28 16:07:41 -07:00