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>
Shakeel Butt reported premature oom on kernel with
"cgroup_disable=memory" since mem_cgroup_is_root() returns false even
though memcg is actually NULL. The drop_caches is also broken.
It is because commit aeed1d325d ("mm/vmscan.c: generalize
shrink_slab() calls in shrink_node()") removed the !memcg check before
!mem_cgroup_is_root(). And, surprisingly root memcg is allocated even
though memory cgroup is disabled by kernel boot parameter.
Add mem_cgroup_disabled() check to make reclaimer work as expected.
Link: http://lkml.kernel.org/r/1563385526-20805-1-git-send-email-yang.shi@linux.alibaba.com
Fixes: aeed1d325d ("mm/vmscan.c: generalize shrink_slab() calls in shrink_node()")
Signed-off-by: Yang Shi <yang.shi@linux.alibaba.com>
Reported-by: Shakeel Butt <shakeelb@google.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Jan Hadrava <had@kam.mff.cuni.cz>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Roman Gushchin <guro@fb.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Qian Cai <cai@lca.pw>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: <stable@vger.kernel.org> [4.19+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Six sites are presently altering current->reclaim_state. There is a
risk that one function stomps on a caller's value. Use a helper
function to catch such errors.
Cc: Yafang Shao <laoar.shao@gmail.com>
Cc: Kirill Tkhai <ktkhai@virtuozzo.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
There are six different reclaim paths by now:
- kswapd reclaim path
- node reclaim path
- hibernate preallocate memory reclaim path
- direct reclaim path
- memcg reclaim path
- memcg softlimit reclaim path
The slab caches reclaimed in these paths are only calculated in the
above three paths.
There're some drawbacks if we don't calculate the reclaimed slab caches.
- The sc->nr_reclaimed isn't correct if there're some slab caches
relcaimed in this path.
- The slab caches may be reclaimed thoroughly if there're lots of
reclaimable slab caches and few page caches.
Let's take an easy example for this case. If one memcg is full of
slab caches and the limit of it is 512M, in other words there're
approximately 512M slab caches in this memcg. Then the limit of the
memcg is reached and the memcg reclaim begins, and then in this memcg
reclaim path it will continuesly reclaim the slab caches until the
sc->priority drops to 0. After this reclaim stops, you will find
there're few slab caches left, which is less than 20M in my test
case. While after this patch applied the number is greater than 300M
and the sc->priority only drops to 3.
Link: http://lkml.kernel.org/r/1561112086-6169-3-git-send-email-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Kirill Tkhai <ktkhai@virtuozzo.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "mm/vmscan: calculate reclaimed slab in all reclaim paths".
This patchset is to fix the issues in doing shrink slab.
There're six different reclaim paths by now,
- kswapd reclaim path
- node reclaim path
- hibernate preallocate memory reclaim path
- direct reclaim path
- memcg reclaim path
- memcg softlimit reclaim path
The slab caches reclaimed in these paths are only calculated in the
above three paths. The issues are detailed explained in patch #2. We
should calculate the reclaimed slab caches in every reclaim path. In
order to do it, the struct reclaim_state is placed into the struct
shrink_control.
In node reclaim path, there'is another issue about shrinking slab, which
is adressed in "mm/vmscan: shrink slab in node reclaim"
(https://lore.kernel.org/linux-mm/1559874946-22960-1-git-send-email-laoar.shao@gmail.com/).
This patch (of 2):
The struct reclaim_state is used to record how many slab caches are
reclaimed in one reclaim path. The struct shrink_control is used to
control one reclaim path. So we'd better put reclaim_state into
shrink_control.
[laoar.shao@gmail.com: remove reclaim_state assignment from __perform_reclaim()]
Link: http://lkml.kernel.org/r/1561381582-13697-1-git-send-email-laoar.shao@gmail.com
Link: http://lkml.kernel.org/r/1561112086-6169-2-git-send-email-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit bd4c82c22c ("mm, THP, swap: delay splitting THP after swapped
out"), THP can be swapped out in a whole. But, nr_reclaimed and some
other vm counters still get inc'ed by one even though a whole THP (512
pages) gets swapped out.
This doesn't make too much sense to memory reclaim.
For example, direct reclaim may just need reclaim SWAP_CLUSTER_MAX
pages, reclaiming one THP could fulfill it. But, if nr_reclaimed is not
increased correctly, direct reclaim may just waste time to reclaim more
pages, SWAP_CLUSTER_MAX * 512 pages in worst case.
And, it may cause pgsteal_{kswapd|direct} is greater than
pgscan_{kswapd|direct}, like the below:
pgsteal_kswapd 122933
pgsteal_direct 26600225
pgscan_kswapd 174153
pgscan_direct 14678312
nr_reclaimed and nr_scanned must be fixed in parallel otherwise it would
break some page reclaim logic, e.g.
vmpressure: this looks at the scanned/reclaimed ratio so it won't change
semantics as long as scanned & reclaimed are fixed in parallel.
compaction/reclaim: compaction wants a certain number of physical pages
freed up before going back to compacting.
kswapd priority raising: kswapd raises priority if we scan fewer pages
than the reclaim target (which itself is obviously expressed in order-0
pages). As a result, kswapd can falsely raise its aggressiveness even
when it's making great progress.
Other than nr_scanned and nr_reclaimed, some other counters, e.g.
pgactivate, nr_skipped, nr_ref_keep and nr_unmap_fail need to be fixed too
since they are user visible via cgroup, /proc/vmstat or trace points,
otherwise they would be underreported.
When isolating pages from LRUs, nr_taken has been accounted in base page,
but nr_scanned and nr_skipped are still accounted in THP. It doesn't make
too much sense too since this may cause trace point underreport the
numbers as well.
So accounting those counters in base page instead of accounting THP as one
page.
nr_dirty, nr_unqueued_dirty, nr_congested and nr_writeback are used by
file cache, so they are not impacted by THP swap.
This change may result in lower steal/scan ratio in some cases since THP
may get split during page reclaim, then a part of tail pages get reclaimed
instead of the whole 512 pages, but nr_scanned is accounted by 512,
particularly for direct reclaim. But, this should be not a significant
issue.
Link: http://lkml.kernel.org/r/1559025859-72759-2-git-send-email-yang.shi@linux.alibaba.com
Signed-off-by: Yang Shi <yang.shi@linux.alibaba.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Hillf Danton <hdanton@sina.com>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 9092c71bb7 ("mm: use sc->priority for slab shrink targets") has
broken up the relationship between sc->nr_scanned and slab pressure.
The sc->nr_scanned can't double slab pressure anymore. So, it sounds no
sense to still keep sc->nr_scanned inc'ed. Actually, it would prevent
from adding pressure on slab shrink since excessive sc->nr_scanned would
prevent from scan->priority raise.
The bonnie test doesn't show this would change the behavior of slab
shrinkers.
w/ w/o
/sec %CP /sec %CP
Sequential delete: 3960.6 94.6 3997.6 96.2
Random delete: 2518 63.8 2561.6 64.6
The slight increase of "/sec" without the patch would be caused by the
slight increase of CPU usage.
Link: http://lkml.kernel.org/r/1559025859-72759-1-git-send-email-yang.shi@linux.alibaba.com
Signed-off-by: Yang Shi <yang.shi@linux.alibaba.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Hillf Danton <hdanton@sina.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When file refaults are detected and there are many inactive file pages,
the system never reclaim anonymous pages, the file pages are dropped
aggressively when there are still a lot of cold anonymous pages and
system thrashes. This issue impacts the performance of applications
with large executable, e.g. chrome.
With this patch, when file refault is detected, inactive_list_is_low()
always returns true for file pages in get_scan_count() to enable
scanning anonymous pages.
The problem can be reproduced by the following test program.
---8<---
void fallocate_file(const char *filename, off_t size)
{
struct stat st;
int fd;
if (!stat(filename, &st) && st.st_size >= size)
return;
fd = open(filename, O_WRONLY | O_CREAT, 0600);
if (fd < 0) {
perror("create file");
exit(1);
}
if (posix_fallocate(fd, 0, size)) {
perror("fallocate");
exit(1);
}
close(fd);
}
long *alloc_anon(long size)
{
long *start = malloc(size);
memset(start, 1, size);
return start;
}
long access_file(const char *filename, long size, long rounds)
{
int fd, i;
volatile char *start1, *end1, *start2;
const int page_size = getpagesize();
long sum = 0;
fd = open(filename, O_RDONLY);
if (fd == -1) {
perror("open");
exit(1);
}
/*
* Some applications, e.g. chrome, use a lot of executable file
* pages, map some of the pages with PROT_EXEC flag to simulate
* the behavior.
*/
start1 = mmap(NULL, size / 2, PROT_READ | PROT_EXEC, MAP_SHARED,
fd, 0);
if (start1 == MAP_FAILED) {
perror("mmap");
exit(1);
}
end1 = start1 + size / 2;
start2 = mmap(NULL, size / 2, PROT_READ, MAP_SHARED, fd, size / 2);
if (start2 == MAP_FAILED) {
perror("mmap");
exit(1);
}
for (i = 0; i < rounds; ++i) {
struct timeval before, after;
volatile char *ptr1 = start1, *ptr2 = start2;
gettimeofday(&before, NULL);
for (; ptr1 < end1; ptr1 += page_size, ptr2 += page_size)
sum += *ptr1 + *ptr2;
gettimeofday(&after, NULL);
printf("File access time, round %d: %f (sec)
", i,
(after.tv_sec - before.tv_sec) +
(after.tv_usec - before.tv_usec) / 1000000.0);
}
return sum;
}
int main(int argc, char *argv[])
{
const long MB = 1024 * 1024;
long anon_mb, file_mb, file_rounds;
const char filename[] = "large";
long *ret1;
long ret2;
if (argc != 4) {
printf("usage: thrash ANON_MB FILE_MB FILE_ROUNDS
");
exit(0);
}
anon_mb = atoi(argv[1]);
file_mb = atoi(argv[2]);
file_rounds = atoi(argv[3]);
fallocate_file(filename, file_mb * MB);
printf("Allocate %ld MB anonymous pages
", anon_mb);
ret1 = alloc_anon(anon_mb * MB);
printf("Access %ld MB file pages
", file_mb);
ret2 = access_file(filename, file_mb * MB, file_rounds);
printf("Print result to prevent optimization: %ld
",
*ret1 + ret2);
return 0;
}
---8<---
Running the test program on 2GB RAM VM with kernel 5.2.0-rc5, the program
fills ram with 2048 MB memory, access a 200 MB file for 10 times. Without
this patch, the file cache is dropped aggresively and every access to the
file is from disk.
$ ./thrash 2048 200 10
Allocate 2048 MB anonymous pages
Access 200 MB file pages
File access time, round 0: 2.489316 (sec)
File access time, round 1: 2.581277 (sec)
File access time, round 2: 2.487624 (sec)
File access time, round 3: 2.449100 (sec)
File access time, round 4: 2.420423 (sec)
File access time, round 5: 2.343411 (sec)
File access time, round 6: 2.454833 (sec)
File access time, round 7: 2.483398 (sec)
File access time, round 8: 2.572701 (sec)
File access time, round 9: 2.493014 (sec)
With this patch, these file pages can be cached.
$ ./thrash 2048 200 10
Allocate 2048 MB anonymous pages
Access 200 MB file pages
File access time, round 0: 2.475189 (sec)
File access time, round 1: 2.440777 (sec)
File access time, round 2: 2.411671 (sec)
File access time, round 3: 1.955267 (sec)
File access time, round 4: 0.029924 (sec)
File access time, round 5: 0.000808 (sec)
File access time, round 6: 0.000771 (sec)
File access time, round 7: 0.000746 (sec)
File access time, round 8: 0.000738 (sec)
File access time, round 9: 0.000747 (sec)
Checked the swap out stats during the test [1], 19006 pages swapped out
with this patch, 3418 pages swapped out without this patch. There are
more swap out, but I think it's within reasonable range when file backed
data set doesn't fit into the memory.
$ ./thrash 2000 100 2100 5 1 # ANON_MB FILE_EXEC FILE_NOEXEC ROUNDS
PROCESSES Allocate 2000 MB anonymous pages active_anon: 1613644,
inactive_anon: 348656, active_file: 892, inactive_file: 1384 (kB)
pswpout: 7972443, pgpgin: 478615246 Access 100 MB executable file pages
Access 2100 MB regular file pages File access time, round 0: 12.165,
(sec) active_anon: 1433788, inactive_anon: 478116, active_file: 17896,
inactive_file: 24328 (kB) File access time, round 1: 11.493, (sec)
active_anon: 1430576, inactive_anon: 477144, active_file: 25440,
inactive_file: 26172 (kB) File access time, round 2: 11.455, (sec)
active_anon: 1427436, inactive_anon: 476060, active_file: 21112,
inactive_file: 28808 (kB) File access time, round 3: 11.454, (sec)
active_anon: 1420444, inactive_anon: 473632, active_file: 23216,
inactive_file: 35036 (kB) File access time, round 4: 11.479, (sec)
active_anon: 1413964, inactive_anon: 471460, active_file: 31728,
inactive_file: 32224 (kB) pswpout: 7991449 (+ 19006), pgpgin: 489924366
(+ 11309120)
With 4 processes accessing non-overlapping parts of a large file, 30316
pages swapped out with this patch, 5152 pages swapped out without this
patch. The swapout number is small comparing to pgpgin.
[1]: https://github.com/vovo/testing/blob/master/mem_thrash.c
Link: http://lkml.kernel.org/r/20190701081038.GA83398@google.com
Fixes: e986850598 ("mm,vmscan: only evict file pages when we have plenty")
Fixes: 7c5bd705d8 ("mm: memcg: only evict file pages when we have plenty")
Signed-off-by: Kuo-Hsin Yang <vovoy@chromium.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Sonny Rao <sonnyrao@chromium.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Rik van Riel <riel@redhat.com>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: <stable@vger.kernel.org> [4.12+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In production we have noticed hard lockups on large machines running
large jobs due to kswaps hoarding lru lock within isolate_lru_pages when
sc->reclaim_idx is 0 which is a small zone. The lru was couple hundred
GiBs and the condition (page_zonenum(page) > sc->reclaim_idx) in
isolate_lru_pages() was basically skipping GiBs of pages while holding
the LRU spinlock with interrupt disabled.
On further inspection, it seems like there are two issues:
(1) If kswapd on the return from balance_pgdat() could not sleep (i.e.
node is still unbalanced), the classzone_idx is unintentionally set
to 0 and the whole reclaim cycle of kswapd will try to reclaim only
the lowest and smallest zone while traversing the whole memory.
(2) Fundamentally isolate_lru_pages() is really bad when the
allocation has woken kswapd for a smaller zone on a very large machine
running very large jobs. It can hoard the LRU spinlock while skipping
over 100s of GiBs of pages.
This patch only fixes (1). (2) needs a more fundamental solution. To
fix (1), in the kswapd context, if pgdat->kswapd_classzone_idx is
invalid use the classzone_idx of the previous kswapd loop otherwise use
the one the waker has requested.
Link: http://lkml.kernel.org/r/20190701201847.251028-1-shakeelb@google.com
Fixes: e716f2eb24 ("mm, vmscan: prevent kswapd sleeping prematurely due to mismatched classzone_idx")
Signed-off-by: Shakeel Butt <shakeelb@google.com>
Reviewed-by: Yang Shi <yang.shi@linux.alibaba.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hdanton@sina.com>
Cc: Roman Gushchin <guro@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
There was the below bug report from Wu Fangsuo.
On the CMA allocation path, isolate_migratepages_range() could isolate
unevictable LRU pages and reclaim_clean_page_from_list() can try to
reclaim them if they are clean file-backed pages.
page:ffffffbf02f33b40 count:86 mapcount:84 mapping:ffffffc08fa7a810 index:0x24
flags: 0x19040c(referenced|uptodate|arch_1|mappedtodisk|unevictable|mlocked)
raw: 000000000019040c ffffffc08fa7a810 0000000000000024 0000005600000053
raw: ffffffc009b05b20 ffffffc009b05b20 0000000000000000 ffffffc09bf3ee80
page dumped because: VM_BUG_ON_PAGE(PageLRU(page) || PageUnevictable(page))
page->mem_cgroup:ffffffc09bf3ee80
------------[ cut here ]------------
kernel BUG at /home/build/farmland/adroid9.0/kernel/linux/mm/vmscan.c:1350!
Internal error: Oops - BUG: 0 [#1] PREEMPT SMP
Modules linked in:
CPU: 0 PID: 7125 Comm: syz-executor Tainted: G S 4.14.81 #3
Hardware name: ASR AQUILAC EVB (DT)
task: ffffffc00a54cd00 task.stack: ffffffc009b00000
PC is at shrink_page_list+0x1998/0x3240
LR is at shrink_page_list+0x1998/0x3240
pc : [<ffffff90083a2158>] lr : [<ffffff90083a2158>] pstate: 60400045
sp : ffffffc009b05940
..
shrink_page_list+0x1998/0x3240
reclaim_clean_pages_from_list+0x3c0/0x4f0
alloc_contig_range+0x3bc/0x650
cma_alloc+0x214/0x668
ion_cma_allocate+0x98/0x1d8
ion_alloc+0x200/0x7e0
ion_ioctl+0x18c/0x378
do_vfs_ioctl+0x17c/0x1780
SyS_ioctl+0xac/0xc0
Wu found it's due to commit ad6b67041a ("mm: remove SWAP_MLOCK in
ttu"). Before that, unevictable pages go to cull_mlocked so that we
can't reach the VM_BUG_ON_PAGE line.
To fix the issue, this patch filters out unevictable LRU pages from the
reclaim_clean_pages_from_list in CMA.
Link: http://lkml.kernel.org/r/20190524071114.74202-1-minchan@kernel.org
Fixes: ad6b67041a ("mm: remove SWAP_MLOCK in ttu")
Signed-off-by: Minchan Kim <minchan@kernel.org>
Reported-by: Wu Fangsuo <fangsuowu@asrmicro.com>
Debugged-by: Wu Fangsuo <fangsuowu@asrmicro.com>
Tested-by: Wu Fangsuo <fangsuowu@asrmicro.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Pankaj Suryawanshi <pankaj.suryawanshi@einfochips.com>
Cc: <stable@vger.kernel.org> [4.12+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Johannes pointed out that after commit 886cf1901d ("mm: move
recent_rotated pages calculation to shrink_inactive_list()") we lost all
zone_reclaim_stat::recent_rotated history.
This fixes it.
Link: http://lkml.kernel.org/r/155905972210.26456.11178359431724024112.stgit@localhost.localdomain
Fixes: 886cf1901d ("mm: move recent_rotated pages calculation to shrink_inactive_list()")
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reported-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Daniel Jordan <daniel.m.jordan@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "mm: memcontrol: memory.stat cost & correctness".
The cgroup memory.stat file holds recursive statistics for the entire
subtree. The current implementation does this tree walk on-demand
whenever the file is read. This is giving us problems in production.
1. The cost of aggregating the statistics on-demand is high. A lot of
system service cgroups are mostly idle and their stats don't change
between reads, yet we always have to check them. There are also always
some lazily-dying cgroups sitting around that are pinned by a handful
of remaining page cache; the same applies to them.
In an application that periodically monitors memory.stat in our
fleet, we have seen the aggregation consume up to 5% CPU time.
2. When cgroups die and disappear from the cgroup tree, so do their
accumulated vm events. The result is that the event counters at
higher-level cgroups can go backwards and confuse some of our
automation, let alone people looking at the graphs over time.
To address both issues, this patch series changes the stat
implementation to spill counts upwards when the counters change.
The upward spilling is batched using the existing per-cpu cache. In a
sparse file stress test with 5 level cgroup nesting, the additional cost
of the flushing was negligible (a little under 1% of CPU at 100% CPU
utilization, compared to the 5% of reading memory.stat during regular
operation).
This patch (of 4):
memcg_page_state(), lruvec_page_state(), memcg_sum_events() are
currently returning the state of the local memcg or lruvec, not the
recursive state.
In practice there is a demand for both versions, although the callers
that want the recursive counts currently sum them up by hand.
Per default, cgroups are considered recursive entities and generally we
expect more users of the recursive counters, with the local counts being
special cases. To reflect that in the name, add a _local suffix to the
current implementations.
The following patch will re-incarnate these functions with recursive
semantics, but with an O(1) implementation.
[hannes@cmpxchg.org: fix bisection hole]
Link: http://lkml.kernel.org/r/20190417160347.GC23013@cmpxchg.org
Link: http://lkml.kernel.org/r/20190412151507.2769-2-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We can use __count_memcg_events() directly because this callsite is alreay
protected by spin_lock_irq().
Link: http://lkml.kernel.org/r/1556093494-30798-1-git-send-email-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This merges together duplicated patterns of code. Also, replace
count_memcg_events() with its irq-careless namesake, because they are
already called in interrupts disabled context.
Link: http://lkml.kernel.org/r/2ece1df4-2989-bc9b-6172-61e9fdde5bfd@virtuozzo.com
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
There are three tracepoints using this template, which are
mm_vmscan_direct_reclaim_begin,
mm_vmscan_memcg_reclaim_begin,
mm_vmscan_memcg_softlimit_reclaim_begin.
Regarding mm_vmscan_direct_reclaim_begin,
sc.may_writepage is !laptop_mode, that's a static setting, and
reclaim_idx is derived from gfp_mask which is already show in this
tracepoint.
Regarding mm_vmscan_memcg_reclaim_begin,
may_writepage is !laptop_mode too, and reclaim_idx is (MAX_NR_ZONES-1),
which are both static value.
mm_vmscan_memcg_softlimit_reclaim_begin is the same with
mm_vmscan_memcg_reclaim_begin.
So we can drop them all.
Link: http://lkml.kernel.org/r/1553736322-32235-1-git-send-email-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.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>
Instead of adding up the zone counters, use lruvec_page_state() to get the
node state directly. This is a bit cheaper and more stream-lined.
Link: http://lkml.kernel.org/r/20190228163020.24100-3-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The page alloc fast path it may perform node reclaim, which may cause a
latency spike. We should add tracepoint for this event, and also measure
the latency it causes.
So bellow two tracepoints are introduced,
mm_vmscan_node_reclaim_begin
mm_vmscan_node_reclaim_end
Link: http://lkml.kernel.org/r/1551421452-5385-1-git-send-email-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Souptick Joarder <jrdr.linux@gmail.com>
Cc: <shaoyafang@didiglobal.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This combines two similar functions move_active_pages_to_lru() and
putback_inactive_pages() into single move_pages_to_lru(). This remove
duplicate code and makes object file size smaller.
Before:
text data bss dec hex filename
57082 4732 128 61942 f1f6 mm/vmscan.o
After:
text data bss dec hex filename
55112 4600 128 59840 e9c0 mm/vmscan.o
Note, that now we are checking for !page_evictable() coming from
shrink_active_list(), which shouldn't change any behavior since that path
works with evictable pages only.
Link: http://lkml.kernel.org/r/155290129627.31489.8321971028677203248.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We may use input argument list as output argument too. This makes the
function more similar to putback_inactive_pages().
Link: http://lkml.kernel.org/r/155290129079.31489.16180612694090502942.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "mm: Generalize putback functions"]
putback_inactive_pages() and move_active_pages_to_lru() are almost
similar, so this patchset merges them ina single function.
This patch (of 4):
The patch moves the calculation from putback_inactive_pages() to
shrink_inactive_list(). This makes putback_inactive_pages() looking more
similar to move_active_pages_to_lru().
To do that, we account activated pages in reclaim_stat::nr_activate.
Since a page may change its LRU type from anon to file cache inside
shrink_page_list() (see ClearPageSwapBacked()), we have to account pages
for the both types. So, nr_activate becomes an array.
Previously we used nr_activate to account PGACTIVATE events, but now we
account them into pgactivate variable (since they are about number of
pages in general, not about sum of hpage_nr_pages).
Link: http://lkml.kernel.org/r/155290127956.31489.3393586616054413298.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
-----BEGIN PGP SIGNATURE-----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=WZfC
-----END PGP SIGNATURE-----
Merge tag 'printk-for-5.2' of git://git.kernel.org/pub/scm/linux/kernel/git/pmladek/printk
Pull printk updates from Petr Mladek:
- Allow state reset of printk_once() calls.
- Prevent crashes when dereferencing invalid pointers in vsprintf().
Only the first byte is checked for simplicity.
- Make vsprintf warnings consistent and inlined.
- Treewide conversion of obsolete %pf, %pF to %ps, %pF printf
modifiers.
- Some clean up of vsprintf and test_printf code.
* tag 'printk-for-5.2' of git://git.kernel.org/pub/scm/linux/kernel/git/pmladek/printk:
lib/vsprintf: Make function pointer_string static
vsprintf: Limit the length of inlined error messages
vsprintf: Avoid confusion between invalid address and value
vsprintf: Prevent crash when dereferencing invalid pointers
vsprintf: Consolidate handling of unknown pointer specifiers
vsprintf: Factor out %pO handler as kobject_string()
vsprintf: Factor out %pV handler as va_format()
vsprintf: Factor out %p[iI] handler as ip_addr_string()
vsprintf: Do not check address of well-known strings
vsprintf: Consistent %pK handling for kptr_restrict == 0
vsprintf: Shuffle restricted_pointer()
printk: Tie printk_once / printk_deferred_once into .data.once for reset
treewide: Switch printk users from %pf and %pF to %ps and %pS, respectively
lib/test_printf: Switch to bitmap_zalloc()
During !CONFIG_CGROUP reclaim, we expand the inactive list size if it's
thrashing on the node that is about to be reclaimed. But when cgroups
are enabled, we suddenly ignore the node scope and use the cgroup scope
only. The result is that pressure bleeds between NUMA nodes depending
on whether cgroups are merely compiled into Linux. This behavioral
difference is unexpected and undesirable.
When the refault adaptivity of the inactive list was first introduced,
there were no statistics at the lruvec level - the intersection of node
and memcg - so it was better than nothing.
But now that we have that infrastructure, use lruvec_page_state() to
make the list balancing decision always NUMA aware.
[hannes@cmpxchg.org: fix bisection hole]
Link: http://lkml.kernel.org/r/20190417155241.GB23013@cmpxchg.org
Link: http://lkml.kernel.org/r/20190412144438.2645-1-hannes@cmpxchg.org
Fixes: 2a2e48854d ("mm: vmscan: fix IO/refault regression in cache workingset transition")
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Cc: Roman Gushchin <guro@fb.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We have common pattern to access lru_lock from a page pointer:
zone_lru_lock(page_zone(page))
Which is silly, because it unfolds to this:
&NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]->zone_pgdat->lru_lock
while we can simply do
&NODE_DATA(page_to_nid(page))->lru_lock
Remove zone_lru_lock() function, since it's only complicate things. Use
'page_pgdat(page)->lru_lock' pattern instead.
[aryabinin@virtuozzo.com: a slightly better version of __split_huge_page()]
Link: http://lkml.kernel.org/r/20190301121651.7741-1-aryabinin@virtuozzo.com
Link: http://lkml.kernel.org/r/20190228083329.31892-2-aryabinin@virtuozzo.com
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: William Kucharski <william.kucharski@oracle.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
workingset_eviction() doesn't use and never did use the @mapping
argument. Remove it.
Link: http://lkml.kernel.org/r/20190228083329.31892-1-aryabinin@virtuozzo.com
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Rik van Riel <riel@surriel.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: William Kucharski <william.kucharski@oracle.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
On path shrink_inactive_list() ---> shrink_page_list() we allocate stack
variables for the statistics twice. This is completely useless, and
this just consumes stack much more, then we really need.
The patch kills duplicate stack variables from shrink_page_list(), and
this reduce stack usage and object file size significantly:
Stack usage:
Before: vmscan.c:1122:22:shrink_page_list 648 static
After: vmscan.c:1122:22:shrink_page_list 616 static
Size of vmscan.o:
text data bss dec hex filename
Before: 56866 4720 128 61714 f112 mm/vmscan.o
After: 56770 4720 128 61618 f0b2 mm/vmscan.o
Link: http://lkml.kernel.org/r/154894900030.5211.12104993874109647641.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In current implementation, both kswapd and direct reclaim has to iterate
all mem cgroups. It is not a problem before offline mem cgroups could
be iterated. But, currently with iterating offline mem cgroups, it
could be very time consuming. In our workloads, we saw over 400K mem
cgroups accumulated in some cases, only a few hundred are online memcgs.
Although kswapd could help out to reduce the number of memcgs, direct
reclaim still get hit with iterating a number of offline memcgs in some
cases. We experienced the responsiveness problems due to this
occassionally.
A simple test with pref shows it may take around 220ms to iterate 8K
memcgs in direct reclaim:
dd 13873 [011] 578.542919: vmscan:mm_vmscan_direct_reclaim_begin
dd 13873 [011] 578.758689: vmscan:mm_vmscan_direct_reclaim_end
So for 400K, it may take around 11 seconds to iterate all memcgs.
Here just break the iteration once it reclaims enough pages as what
memcg direct reclaim does. This may hurt the fairness among memcgs.
But the cached iterator cookie could help to achieve the fairness more
or less.
Link: http://lkml.kernel.org/r/1548799877-10949-1-git-send-email-yang.shi@linux.alibaba.com
Signed-off-by: Yang Shi <yang.shi@linux.alibaba.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
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>
We may simply check for sc->may_unmap in isolate_lru_pages() instead of
doing that in both of its callers.
Link: http://lkml.kernel.org/r/154748280735.29962.15867846875217618569.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.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>
No functional change.
Link: http://lkml.kernel.org/r/20190118235123.27843-1-richard.weiyang@gmail.com
Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
Reviewed-by: Pekka Enberg <penberg@kernel.org>
Acked-by: Mike Rapoport <rppt@linux.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This reverts commit 172b06c32b ("mm: slowly shrink slabs with a
relatively small number of objects").
This change changes the agressiveness of shrinker reclaim, causing small
cache and low priority reclaim to greatly increase scanning pressure on
small caches. As a result, light memory pressure has a disproportionate
affect on small caches, and causes large caches to be reclaimed much
faster than previously.
As a result, it greatly perturbs the delicate balance of the VFS caches
(dentry/inode vs file page cache) such that the inode/dentry caches are
reclaimed much, much faster than the page cache and this drives us into
several other caching imbalance related problems.
As such, this is a bad change and needs to be reverted.
[ Needs some massaging to retain the later seekless shrinker
modifications.]
Link: http://lkml.kernel.org/r/20190130041707.27750-3-david@fromorbit.com
Fixes: 172b06c32b ("mm: slowly shrink slabs with a relatively small number of objects")
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Cc: Wolfgang Walter <linux@stwm.de>
Cc: Roman Gushchin <guro@fb.com>
Cc: Spock <dairinin@gmail.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db9165 ("sched/wait: Break up long wake
list walk") 11a19c7b09 ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f521 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
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>
The i915 driver uses shmemfs to allocate backing storage for gem
objects. These shmemfs pages can be pinned (increased ref count) by
shmem_read_mapping_page_gfp(). When a lot of pages are pinned, vmscan
wastes a lot of time scanning these pinned pages. In some extreme case,
all pages in the inactive anon lru are pinned, and only the inactive
anon lru is scanned due to inactive_ratio, the system cannot swap and
invokes the oom-killer. Mark these pinned pages as unevictable to speed
up vmscan.
Export pagevec API check_move_unevictable_pages().
This patch was inspired by Chris Wilson's change [1].
[1]: https://patchwork.kernel.org/patch/9768741/
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Kuo-Hsin Yang <vovoy@chromium.org>
Acked-by: Michal Hocko <mhocko@suse.com> # mm part
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Dave Hansen <dave.hansen@intel.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20181106132324.17390-1-chris@chris-wilson.co.uk
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Pull XArray conversion from Matthew Wilcox:
"The XArray provides an improved interface to the radix tree data
structure, providing locking as part of the API, specifying GFP flags
at allocation time, eliminating preloading, less re-walking the tree,
more efficient iterations and not exposing RCU-protected pointers to
its users.
This patch set
1. Introduces the XArray implementation
2. Converts the pagecache to use it
3. Converts memremap to use it
The page cache is the most complex and important user of the radix
tree, so converting it was most important. Converting the memremap
code removes the only other user of the multiorder code, which allows
us to remove the radix tree code that supported it.
I have 40+ followup patches to convert many other users of the radix
tree over to the XArray, but I'd like to get this part in first. The
other conversions haven't been in linux-next and aren't suitable for
applying yet, but you can see them in the xarray-conv branch if you're
interested"
* 'xarray' of git://git.infradead.org/users/willy/linux-dax: (90 commits)
radix tree: Remove multiorder support
radix tree test: Convert multiorder tests to XArray
radix tree tests: Convert item_delete_rcu to XArray
radix tree tests: Convert item_kill_tree to XArray
radix tree tests: Move item_insert_order
radix tree test suite: Remove multiorder benchmarking
radix tree test suite: Remove __item_insert
memremap: Convert to XArray
xarray: Add range store functionality
xarray: Move multiorder_check to in-kernel tests
xarray: Move multiorder_shrink to kernel tests
xarray: Move multiorder account test in-kernel
radix tree test suite: Convert iteration test to XArray
radix tree test suite: Convert tag_tagged_items to XArray
radix tree: Remove radix_tree_clear_tags
radix tree: Remove radix_tree_maybe_preload_order
radix tree: Remove split/join code
radix tree: Remove radix_tree_update_node_t
page cache: Finish XArray conversion
dax: Convert page fault handlers to XArray
...
The page cache and most shrinkable slab caches hold data that has been
read from disk, but there are some caches that only cache CPU work, such
as the dentry and inode caches of procfs and sysfs, as well as the subset
of radix tree nodes that track non-resident page cache.
Currently, all these are shrunk at the same rate: using DEFAULT_SEEKS for
the shrinker's seeks setting tells the reclaim algorithm that for every
two page cache pages scanned it should scan one slab object.
This is a bogus setting. A virtual inode that required no IO to create is
not twice as valuable as a page cache page; shadow cache entries with
eviction distances beyond the size of memory aren't either.
In most cases, the behavior in practice is still fine. Such virtual
caches don't tend to grow and assert themselves aggressively, and usually
get picked up before they cause problems. But there are scenarios where
that's not true.
Our database workloads suffer from two of those. For one, their file
workingset is several times bigger than available memory, which has the
kernel aggressively create shadow page cache entries for the non-resident
parts of it. The workingset code does tell the VM that most of these are
expendable, but the VM ends up balancing them 2:1 to cache pages as per
the seeks setting. This is a huge waste of memory.
These workloads also deal with tens of thousands of open files and use
/proc for introspection, which ends up growing the proc_inode_cache to
absurdly large sizes - again at the cost of valuable cache space, which
isn't a reasonable trade-off, given that proc inodes can be re-created
without involving the disk.
This patch implements a "zero-seek" setting for shrinkers that results in
a target ratio of 0:1 between their objects and IO-backed caches. This
allows such virtual caches to grow when memory is available (they do
cache/avoid CPU work after all), but effectively disables them as soon as
IO-backed objects are under pressure.
It then switches the shrinkers for procfs and sysfs metadata, as well as
excess page cache shadow nodes, to the new zero-seek setting.
Link: http://lkml.kernel.org/r/20181009184732.762-5-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reported-by: Domas Mituzas <dmituzas@fb.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Rik van Riel <riel@surriel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When systems are overcommitted and resources become contended, it's hard
to tell exactly the impact this has on workload productivity, or how close
the system is to lockups and OOM kills. In particular, when machines work
multiple jobs concurrently, the impact of overcommit in terms of latency
and throughput on the individual job can be enormous.
In order to maximize hardware utilization without sacrificing individual
job health or risk complete machine lockups, this patch implements a way
to quantify resource pressure in the system.
A kernel built with CONFIG_PSI=y creates files in /proc/pressure/ that
expose the percentage of time the system is stalled on CPU, memory, or IO,
respectively. Stall states are aggregate versions of the per-task delay
accounting delays:
cpu: some tasks are runnable but not executing on a CPU
memory: tasks are reclaiming, or waiting for swapin or thrashing cache
io: tasks are waiting for io completions
These percentages of walltime can be thought of as pressure percentages,
and they give a general sense of system health and productivity loss
incurred by resource overcommit. They can also indicate when the system
is approaching lockup scenarios and OOMs.
To do this, psi keeps track of the task states associated with each CPU
and samples the time they spend in stall states. Every 2 seconds, the
samples are averaged across CPUs - weighted by the CPUs' non-idle time to
eliminate artifacts from unused CPUs - and translated into percentages of
walltime. A running average of those percentages is maintained over 10s,
1m, and 5m periods (similar to the loadaverage).
[hannes@cmpxchg.org: doc fixlet, per Randy]
Link: http://lkml.kernel.org/r/20180828205625.GA14030@cmpxchg.org
[hannes@cmpxchg.org: code optimization]
Link: http://lkml.kernel.org/r/20180907175015.GA8479@cmpxchg.org
[hannes@cmpxchg.org: rename psi_clock() to psi_update_work(), per Peter]
Link: http://lkml.kernel.org/r/20180907145404.GB11088@cmpxchg.org
[hannes@cmpxchg.org: fix build]
Link: http://lkml.kernel.org/r/20180913014222.GA2370@cmpxchg.org
Link: http://lkml.kernel.org/r/20180828172258.3185-9-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Daniel Drake <drake@endlessm.com>
Tested-by: Suren Baghdasaryan <surenb@google.com>
Cc: Christopher Lameter <cl@linux.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Johannes Weiner <jweiner@fb.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Enderborg <peter.enderborg@sony.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vinayak Menon <vinmenon@codeaurora.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Refaults happen during transitions between workingsets as well as in-place
thrashing. Knowing the difference between the two has a range of
applications, including measuring the impact of memory shortage on the
system performance, as well as the ability to smarter balance pressure
between the filesystem cache and the swap-backed workingset.
During workingset transitions, inactive cache refaults and pushes out
established active cache. When that active cache isn't stale, however,
and also ends up refaulting, that's bonafide thrashing.
Introduce a new page flag that tells on eviction whether the page has been
active or not in its lifetime. This bit is then stored in the shadow
entry, to classify refaults as transitioning or thrashing.
How many page->flags does this leave us with on 32-bit?
20 bits are always page flags
21 if you have an MMU
23 with the zone bits for DMA, Normal, HighMem, Movable
29 with the sparsemem section bits
30 if PAE is enabled
31 with this patch.
So on 32-bit PAE, that leaves 1 bit for distinguishing two NUMA nodes. If
that's not enough, the system can switch to discontigmem and re-gain the 6
or 7 sparsemem section bits.
Link: http://lkml.kernel.org/r/20180828172258.3185-3-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Daniel Drake <drake@endlessm.com>
Tested-by: Suren Baghdasaryan <surenb@google.com>
Cc: Christopher Lameter <cl@linux.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Johannes Weiner <jweiner@fb.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Enderborg <peter.enderborg@sony.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vinayak Menon <vinmenon@codeaurora.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
I've noticed, that dying memory cgroups are often pinned in memory by a
single pagecache page. Even under moderate memory pressure they sometimes
stayed in such state for a long time. That looked strange.
My investigation showed that the problem is caused by applying the LRU
pressure balancing math:
scan = div64_u64(scan * fraction[lru], denominator),
where
denominator = fraction[anon] + fraction[file] + 1.
Because fraction[lru] is always less than denominator, if the initial scan
size is 1, the result is always 0.
This means the last page is not scanned and has
no chances to be reclaimed.
Fix this by rounding up the result of the division.
In practice this change significantly improves the speed of dying cgroups
reclaim.
[guro@fb.com: prevent double calculation of DIV64_U64_ROUND_UP() arguments]
Link: http://lkml.kernel.org/r/20180829213311.GA13501@castle
Link: http://lkml.kernel.org/r/20180827162621.30187-3-guro@fb.com
Signed-off-by: Roman Gushchin <guro@fb.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Both callers of __delete_from_swap_cache have the swp_entry_t already,
so pass that in to make constructing the XA_STATE easier.
Signed-off-by: Matthew Wilcox <willy@infradead.org>
do_shrink_slab() returns unsigned long value, and the placing into int
variable cuts high bytes off. Then we compare ret and 0xfffffffe (since
SHRINK_EMPTY is converted to ret type).
Thus a large number of objects returned by do_shrink_slab() may be
interpreted as SHRINK_EMPTY, if low bytes of their value are equal to
0xfffffffe. Fix that by declaration ret as unsigned long in these
functions.
Link: http://lkml.kernel.org/r/153813407177.17544.14888305435570723973.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reported-by: Cyrill Gorcunov <gorcunov@openvz.org>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Shakeel Butt <shakeelb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9092c71bb7 ("mm: use sc->priority for slab shrink targets") changed the
way that the target slab pressure is calculated and made it
priority-based:
delta = freeable >> priority;
delta *= 4;
do_div(delta, shrinker->seeks);
The problem is that on a default priority (which is 12) no pressure is
applied at all, if the number of potentially reclaimable objects is less
than 4096 (1<<12).
This causes the last objects on slab caches of no longer used cgroups to
(almost) never get reclaimed. It's obviously a waste of memory.
It can be especially painful, if these stale objects are holding a
reference to a dying cgroup. Slab LRU lists are reparented on memcg
offlining, but corresponding objects are still holding a reference to the
dying cgroup. If we don't scan these objects, the dying cgroup can't go
away. Most likely, the parent cgroup hasn't any directly charged objects,
only remaining objects from dying children cgroups. So it can easily hold
a reference to hundreds of dying cgroups.
If there are no big spikes in memory pressure, and new memory cgroups are
created and destroyed periodically, this causes the number of dying
cgroups grow steadily, causing a slow-ish and hard-to-detect memory
"leak". It's not a real leak, as the memory can be eventually reclaimed,
but it could not happen in a real life at all. I've seen hosts with a
steadily climbing number of dying cgroups, which doesn't show any signs of
a decline in months, despite the host is loaded with a production
workload.
It is an obvious waste of memory, and to prevent it, let's apply a minimal
pressure even on small shrinker lists. E.g. if there are freeable
objects, let's scan at least min(freeable, scan_batch) objects.
This fix significantly improves a chance of a dying cgroup to be
reclaimed, and together with some previous patches stops the steady growth
of the dying cgroups number on some of our hosts.
Link: http://lkml.kernel.org/r/20180905230759.12236-1-guro@fb.com
Fixes: 9092c71bb7 ("mm: use sc->priority for slab shrink targets")
Signed-off-by: Roman Gushchin <guro@fb.com>
Acked-by: Rik van Riel <riel@surriel.com>
Cc: Josef Bacik <jbacik@fb.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
page_freeze_refs/page_unfreeze_refs have already been relplaced by
page_ref_freeze/page_ref_unfreeze , but they are not modified in the
comments.
Link: http://lkml.kernel.org/r/1532590226-106038-1-git-send-email-jiang.biao2@zte.com.cn
Signed-off-by: Jiang Biao <jiang.biao2@zte.com.cn>
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>
There is a sad BUG introduced in patch adding SHRINKER_REGISTERING.
shrinker_idr business is only for memcg-aware shrinkers. Only such type
of shrinkers have id and they must be finaly installed via idr_replace()
in this function. For !memcg-aware shrinkers we never initialize
shrinker->id field.
But there are all types of shrinkers passed to idr_replace(), and every
!memcg-aware shrinker with random ID (most probably, its id is 0)
replaces memcg-aware shrinker pointed by the ID in IDR.
This patch fixes the problem.
Link: http://lkml.kernel.org/r/8ff8a793-8211-713a-4ed9-d6e52390c2fc@virtuozzo.com
Fixes: 7e010df53c "mm: use special value SHRINKER_REGISTERING instead of list_empty() check"
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reported-by: <syzbot+d5f648a1bfe15678786b@syzkaller.appspotmail.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <jbacik@fb.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: <syzkaller-bugs@googlegroups.com>
Cc: Huang Ying <ying.huang@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The patch introduces a special value SHRINKER_REGISTERING to use instead
of list_empty() to differ a registering shrinker from unregistered
shrinker. Why we need that at all?
Shrinker registration is split in two parts. The first one is
prealloc_shrinker(), which allocates shrinker memory and reserves ID in
shrinker_idr. This function can fail. The second is
register_shrinker_prepared(), and it finalizes the registration. This
function actually makes shrinker available to be used from
shrink_slab(), and it can't fail.
One shrinker may be based on more then one LRU lists. So, we never
clear the bit in memcg shrinker maps, when (one of) corresponding LRU
list becomes empty, since other LRU lists may be not empty. See
superblock shrinker for example: it is based on two LRU lists:
s_inode_lru and s_dentry_lru. We do not want to clear shrinker bit,
when there are no inodes in s_inode_lru, as s_dentry_lru may contain
dentries.
Instead of that, we use special algorithm to detect shrinkers having no
elements at all its LRU lists, and this is made in shrink_slab_memcg().
See the comment in this function for the details.
Also, in shrink_slab_memcg() we clear shrinker bit in the map, when we
meet unregistered shrinker (bit is set, while there is no a shrinker in
IDR). Otherwise, we would have done that at the moment of shrinker
unregistration for all memcgs (and this looks worse, since iteration
over all memcg may take much time). Also this would have imposed
restrictions on shrinker unregistration order for its users: they would
have had to guarantee, there are no new elements after
unregister_shrinker() (otherwise, a new added element would have set a
bit).
So, if we meet a set bit in map and no shrinker in IDR when we're
iterating over the map in shrink_slab_memcg(), this means the
corresponding shrinker is unregistered, and we must clear the bit.
Another case is shrinker registration. We want two things there:
1) do_shrink_slab() can be called only for completely registered
shrinkers;
2) shrinker internal lists may be populated in any order with
register_shrinker_prepared() (let's talk on the example with sb). Both
of:
a)list_lru_add(&inode->i_sb->s_inode_lru, &inode->i_lru); [cpu0]
memcg_set_shrinker_bit(); [cpu0]
...
register_shrinker_prepared(); [cpu1]
and
b)register_shrinker_prepared(); [cpu0]
...
list_lru_add(&inode->i_sb->s_inode_lru, &inode->i_lru); [cpu1]
memcg_set_shrinker_bit(); [cpu1]
are legitimate. We don't want to impose restriction here and to
force people to use only (b) variant. We don't want to force people to
care, there is no elements in LRU lists before the shrinker is
completely registered. Internal users of LRU lists and shrinker code
are two different subsystems, and they have to be closed in themselves
each other.
In (a) case we have the bit set before shrinker is completely
registered. We don't want do_shrink_slab() is called at this moment, so
we have to detect such the registering shrinkers.
Before this patch list_empty() (shrinker is not linked to the list)
check was used for that. So, in (a) there could be a bit set, but we
don't call do_shrink_slab() unless shrinker is linked to the list. It's
just an indicator, I just overloaded linking to the list.
This was not the best solution, since it's better not to touch the
shrinker memory from shrink_slab_memcg() before it's completely
registered (this also will be useful in the future to make shrink_slab()
completely lockless).
So, this patch introduces better way to detect registering shrinker,
which allows not to dereference shrinker memory. It's just a ~0UL
value, which we insert into the IDR during ID allocation. After
shrinker is ready to be used, we insert actual shrinker pointer in the
IDR, and it becomes available to shrink_slab_memcg().
We can't use NULL instead of this new value for this purpose as:
shrink_slab_memcg() already uses NULL to detect unregistered shrinkers,
and we don't want the function sees NULL and clears the bit, otherwise
(a) won't work.
This is the only thing the patch makes: the better way to detect
registering shrinker. Nothing else this patch makes.
Also this gives a better assembler, but it's minor side of the patch:
Before:
callq <idr_find>
mov %rax,%r15
test %rax,%rax
je <shrink_slab_memcg+0x1d5>
mov 0x20(%rax),%rax
lea 0x20(%r15),%rdx
cmp %rax,%rdx
je <shrink_slab_memcg+0xbd>
mov 0x8(%rsp),%edx
mov %r15,%rsi
lea 0x10(%rsp),%rdi
callq <do_shrink_slab>
After:
callq <idr_find>
mov %rax,%r15
lea -0x1(%rax),%rax
cmp $0xfffffffffffffffd,%rax
ja <shrink_slab_memcg+0x1cd>
mov 0x8(%rsp),%edx
mov %r15,%rsi
lea 0x10(%rsp),%rdi
callq ffffffff810cefd0 <do_shrink_slab>
[ktkhai@virtuozzo.com: add #ifdef CONFIG_MEMCG_KMEM around idr_replace()]
Link: http://lkml.kernel.org/r/758b8fec-7573-47eb-b26a-7b2847ae7b8c@virtuozzo.com
Link: http://lkml.kernel.org/r/153355467546.11522.4518015068123480218.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Josef Bacik <jbacik@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In case of shrink_slab_memcg() we do not zero nid, when shrinker is not
numa-aware. This is not a real problem, since currently all memcg-aware
shrinkers are numa-aware too (we have two: super_block shrinker and
workingset shrinker), but something may change in the future.
Link: http://lkml.kernel.org/r/153320759911.18959.8842396230157677671.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Josef Bacik <jbacik@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
To avoid further unneed calls of do_shrink_slab() for shrinkers, which
already do not have any charged objects in a memcg, their bits have to
be cleared.
This patch introduces a lockless mechanism to do that without races
without parallel list lru add. After do_shrink_slab() returns
SHRINK_EMPTY the first time, we clear the bit and call it once again.
Then we restore the bit, if the new return value is different.
Note, that single smp_mb__after_atomic() in shrink_slab_memcg() covers
two situations:
1)list_lru_add() shrink_slab_memcg
list_add_tail() for_each_set_bit() <--- read bit
do_shrink_slab() <--- missed list update (no barrier)
<MB> <MB>
set_bit() do_shrink_slab() <--- seen list update
This situation, when the first do_shrink_slab() sees set bit, but it
doesn't see list update (i.e., race with the first element queueing), is
rare. So we don't add <MB> before the first call of do_shrink_slab()
instead of this to do not slow down generic case. Also, it's need the
second call as seen in below in (2).
2)list_lru_add() shrink_slab_memcg()
list_add_tail() ...
set_bit() ...
... for_each_set_bit()
do_shrink_slab() do_shrink_slab()
clear_bit() ...
... ...
list_lru_add() ...
list_add_tail() clear_bit()
<MB> <MB>
set_bit() do_shrink_slab()
The barriers guarantee that the second do_shrink_slab() in the right
side task sees list update if really cleared the bit. This case is
drawn in the code comment.
[Results/performance of the patchset]
After the whole patchset applied the below test shows signify increase
of performance:
$echo 1 > /sys/fs/cgroup/memory/memory.use_hierarchy
$mkdir /sys/fs/cgroup/memory/ct
$echo 4000M > /sys/fs/cgroup/memory/ct/memory.kmem.limit_in_bytes
$for i in `seq 0 4000`; do mkdir /sys/fs/cgroup/memory/ct/$i;
echo $$ > /sys/fs/cgroup/memory/ct/$i/cgroup.procs;
mkdir -p s/$i; mount -t tmpfs $i s/$i;
touch s/$i/file; done
Then, 5 sequential calls of drop caches:
$time echo 3 > /proc/sys/vm/drop_caches
1)Before:
0.00user 13.78system 0:13.78elapsed 99%CPU
0.00user 5.59system 0:05.60elapsed 99%CPU
0.00user 5.48system 0:05.48elapsed 99%CPU
0.00user 8.35system 0:08.35elapsed 99%CPU
0.00user 8.34system 0:08.35elapsed 99%CPU
2)After
0.00user 1.10system 0:01.10elapsed 99%CPU
0.00user 0.00system 0:00.01elapsed 64%CPU
0.00user 0.01system 0:00.01elapsed 82%CPU
0.00user 0.00system 0:00.01elapsed 64%CPU
0.00user 0.01system 0:00.01elapsed 82%CPU
The results show the performance increases at least in 548 times.
Shakeel Butt tested this patchset with fork-bomb on his configuration:
> I created 255 memcgs, 255 ext4 mounts and made each memcg create a
> file containing few KiBs on corresponding mount. Then in a separate
> memcg of 200 MiB limit ran a fork-bomb.
>
> I ran the "perf record -ag -- sleep 60" and below are the results:
>
> Without the patch series:
> Samples: 4M of event 'cycles', Event count (approx.): 3279403076005
> + 36.40% fb.sh [kernel.kallsyms] [k] shrink_slab
> + 18.97% fb.sh [kernel.kallsyms] [k] list_lru_count_one
> + 6.75% fb.sh [kernel.kallsyms] [k] super_cache_count
> + 0.49% fb.sh [kernel.kallsyms] [k] down_read_trylock
> + 0.44% fb.sh [kernel.kallsyms] [k] mem_cgroup_iter
> + 0.27% fb.sh [kernel.kallsyms] [k] up_read
> + 0.21% fb.sh [kernel.kallsyms] [k] osq_lock
> + 0.13% fb.sh [kernel.kallsyms] [k] shmem_unused_huge_count
> + 0.08% fb.sh [kernel.kallsyms] [k] shrink_node_memcg
> + 0.08% fb.sh [kernel.kallsyms] [k] shrink_node
>
> With the patch series:
> Samples: 4M of event 'cycles', Event count (approx.): 2756866824946
> + 47.49% fb.sh [kernel.kallsyms] [k] down_read_trylock
> + 30.72% fb.sh [kernel.kallsyms] [k] up_read
> + 9.51% fb.sh [kernel.kallsyms] [k] mem_cgroup_iter
> + 1.69% fb.sh [kernel.kallsyms] [k] shrink_node_memcg
> + 1.35% fb.sh [kernel.kallsyms] [k] mem_cgroup_protected
> + 1.05% fb.sh [kernel.kallsyms] [k] queued_spin_lock_slowpath
> + 0.85% fb.sh [kernel.kallsyms] [k] _raw_spin_lock
> + 0.78% fb.sh [kernel.kallsyms] [k] lruvec_lru_size
> + 0.57% fb.sh [kernel.kallsyms] [k] shrink_node
> + 0.54% fb.sh [kernel.kallsyms] [k] queue_work_on
> + 0.46% fb.sh [kernel.kallsyms] [k] shrink_slab_memcg
[ktkhai@virtuozzo.com: v9]
Link: http://lkml.kernel.org/r/153112561772.4097.11011071937553113003.stgit@localhost.localdomain
Link: http://lkml.kernel.org/r/153063070859.1818.11870882950920963480.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Tested-by: Shakeel Butt <shakeelb@google.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <jbacik@fb.com>
Cc: Li RongQing <lirongqing@baidu.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Matthias Kaehlcke <mka@chromium.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Philippe Ombredanne <pombredanne@nexb.com>
Cc: Roman Gushchin <guro@fb.com>
Cc: Sahitya Tummala <stummala@codeaurora.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Waiman Long <longman@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Mel Gorman
Yang Shi
Andrew Morton
Yafang Shao
Kuo-Hsin Yang
Shakeel Butt
Minchan Kim
Kirill Tkhai
Johannes Weiner
Linus Torvalds
Sakari Ailus
Andrey Ryabinin
Alexey Dobriyan
Wei Yang
Dave Chinner
Hugh Dickins
Jani Nikula
Roman Gushchin
Matthew Wilcox
Jiang Biao