The patch titled
Subject: mm: shrinker: make global slab shrink lockless
has been added to the -mm mm-unstable branch. Its filename is
mm-shrinker-make-global-slab-shrink-lockless.patch
This patch will shortly appear at
https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches/mm-shrinker-make-global-slab-shrink-lockless.patch
This patch will later appear in the mm-unstable branch at
git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/process/submit-checklist.rst when testing your code ***
The -mm tree is included into linux-next via the mm-everything
branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
and is updated there every 2-3 working days
------------------------------------------------------
From: Qi Zheng <zhengqi.arch@xxxxxxxxxxxxx>
Subject: mm: shrinker: make global slab shrink lockless
Date: Mon, 11 Sep 2023 17:44:41 +0800
The shrinker_rwsem is a global read-write lock in shrinkers subsystem,
which protects most operations such as slab shrink, registration and
unregistration of shrinkers, etc. This can easily cause problems in the
following cases.
1) When the memory pressure is high and there are many filesystems
mounted or unmounted at the same time, slab shrink will be affected
(down_read_trylock() failed).
Such as the real workload mentioned by Kirill Tkhai:
```
One of the real workloads from my experience is start
of an overcommitted node containing many starting
containers after node crash (or many resuming containers
after reboot for kernel update). In these cases memory
pressure is huge, and the node goes round in long reclaim.
```
2) If a shrinker is blocked (such as the case mentioned
in [1]) and a writer comes in (such as mount a fs),
then this writer will be blocked and cause all
subsequent shrinker-related operations to be blocked.
Even if there is no competitor when shrinking slab, there may still be a
problem. The down_read_trylock() may become a perf hotspot with frequent
calls to shrink_slab(). Because of the poor multicore scalability of
atomic operations, this can lead to a significant drop in IPC
(instructions per cycle).
We used to implement the lockless slab shrink with SRCU [2], but then
kernel test robot reported -88.8% regression in
stress-ng.ramfs.ops_per_sec test case [3], so we reverted it [4].
This commit uses the refcount+RCU method [5] proposed by Dave Chinner
to re-implement the lockless global slab shrink. The memcg slab shrink is
handled in the subsequent patch.
For now, all shrinker instances are converted to dynamically allocated and
will be freed by call_rcu(). So we can use rcu_read_{lock,unlock}() to
ensure that the shrinker instance is valid.
And the shrinker instance will not be run again after unregistration. So
the structure that records the pointer of shrinker instance can be safely
freed without waiting for the RCU read-side critical section.
In this way, while we implement the lockless slab shrink, we don't need to
be blocked in unregister_shrinker().
The following are the test results:
stress-ng --timeout 60 --times --verify --metrics-brief --ramfs 9 &
1) Before applying this patchset:
setting to a 60 second run per stressor
dispatching hogs: 9 ramfs
stressor bogo ops real time usr time sys time bogo ops/s bogo ops/s
(secs) (secs) (secs) (real time) (usr+sys time)
ramfs 473062 60.00 8.00 279.13 7884.12 1647.59
for a 60.01s run time:
1440.34s available CPU time
7.99s user time ( 0.55%)
279.13s system time ( 19.38%)
287.12s total time ( 19.93%)
load average: 7.12 2.99 1.15
successful run completed in 60.01s (1 min, 0.01 secs)
2) After applying this patchset:
setting to a 60 second run per stressor
dispatching hogs: 9 ramfs
stressor bogo ops real time usr time sys time bogo ops/s bogo ops/s
(secs) (secs) (secs) (real time) (usr+sys time)
ramfs 477165 60.00 8.13 281.34 7952.55 1648.40
for a 60.01s run time:
1440.33s available CPU time
8.12s user time ( 0.56%)
281.34s system time ( 19.53%)
289.46s total time ( 20.10%)
load average: 6.98 3.03 1.19
successful run completed in 60.01s (1 min, 0.01 secs)
We can see that the ops/s has hardly changed.
[1]. https://lore.kernel.org/lkml/20191129214541.3110-1-ptikhomirov@xxxxxxxxxxxxx/
[2]. https://lore.kernel.org/lkml/20230313112819.38938-1-zhengqi.arch@xxxxxxxxxxxxx/
[3]. https://lore.kernel.org/lkml/202305230837.db2c233f-yujie.liu@xxxxxxxxx/
[4]. https://lore.kernel.org/all/20230609081518.3039120-1-qi.zheng@xxxxxxxxx/
[5]. https://lore.kernel.org/lkml/ZIJhou1d55d4H1s0@xxxxxxxxxxxxxxxxxxx/
Link: https://lkml.kernel.org/r/20230911094444.68966-43-zhengqi.arch@xxxxxxxxxxxxx
Signed-off-by: Qi Zheng <zhengqi.arch@xxxxxxxxxxxxx>
Cc: Abhinav Kumar <quic_abhinavk@xxxxxxxxxxx>
Cc: Alasdair Kergon <agk@xxxxxxxxxx>
Cc: Alexander Viro <viro@xxxxxxxxxxxxxxxxxx>
Cc: Alyssa Rosenzweig <alyssa.rosenzweig@xxxxxxxxxxxxx>
Cc: Andreas Dilger <adilger.kernel@xxxxxxxxx>
Cc: Andreas Gruenbacher <agruenba@xxxxxxxxxx>
Cc: Anna Schumaker <anna@xxxxxxxxxx>
Cc: Arnd Bergmann <arnd@xxxxxxxx>
Cc: Bob Peterson <rpeterso@xxxxxxxxxx>
Cc: Borislav Petkov <bp@xxxxxxxxx>
Cc: Carlos Llamas <cmllamas@xxxxxxxxxx>
Cc: Chandan Babu R <chandan.babu@xxxxxxxxxx>
Cc: Chao Yu <chao@xxxxxxxxxx>
Cc: Chris Mason <clm@xxxxxx>
Cc: Christian Brauner <brauner@xxxxxxxxxx>
Cc: Christian Koenig <christian.koenig@xxxxxxx>
Cc: Chuck Lever <cel@xxxxxxxxxx>
Cc: Coly Li <colyli@xxxxxxx>
Cc: Dai Ngo <Dai.Ngo@xxxxxxxxxx>
Cc: Daniel Vetter <daniel@xxxxxxxx>
Cc: Daniel Vetter <daniel.vetter@xxxxxxxx>
Cc: "Darrick J. Wong" <djwong@xxxxxxxxxx>
Cc: Dave Chinner <david@xxxxxxxxxxxxx>
Cc: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>
Cc: David Airlie <airlied@xxxxxxxxx>
Cc: David Hildenbrand <david@xxxxxxxxxx>
Cc: David Sterba <dsterba@xxxxxxxx>
Cc: Dmitry Baryshkov <dmitry.baryshkov@xxxxxxxxxx>
Cc: Gao Xiang <hsiangkao@xxxxxxxxxxxxxxxxx>
Cc: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
Cc: Huang Rui <ray.huang@xxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: Jaegeuk Kim <jaegeuk@xxxxxxxxxx>
Cc: Jani Nikula <jani.nikula@xxxxxxxxxxxxxxx>
Cc: Jan Kara <jack@xxxxxxx>
Cc: Jason Wang <jasowang@xxxxxxxxxx>
Cc: Jeff Layton <jlayton@xxxxxxxxxx>
Cc: Jeffle Xu <jefflexu@xxxxxxxxxxxxxxxxx>
Cc: Joel Fernandes (Google) <joel@xxxxxxxxxxxxxxxxx>
Cc: Joonas Lahtinen <joonas.lahtinen@xxxxxxxxxxxxxxx>
Cc: Josef Bacik <josef@xxxxxxxxxxxxxx>
Cc: Juergen Gross <jgross@xxxxxxxx>
Cc: Kent Overstreet <kent.overstreet@xxxxxxxxx>
Cc: Kirill Tkhai <tkhai@xxxxx>
Cc: Marijn Suijten <marijn.suijten@xxxxxxxxxxxxxx>
Cc: "Michael S. Tsirkin" <mst@xxxxxxxxxx>
Cc: Mike Snitzer <snitzer@xxxxxxxxxx>
Cc: Minchan Kim <minchan@xxxxxxxxxx>
Cc: Muchun Song <muchun.song@xxxxxxxxx>
Cc: Muchun Song <songmuchun@xxxxxxxxxxxxx>
Cc: Nadav Amit <namit@xxxxxxxxxx>
Cc: Neil Brown <neilb@xxxxxxx>
Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@xxxxxxxx>
Cc: Olga Kornievskaia <kolga@xxxxxxxxxx>
Cc: Paul E. McKenney <paulmck@xxxxxxxxxx>
Cc: Richard Weinberger <richard@xxxxxx>
Cc: Rob Clark <robdclark@xxxxxxxxx>
Cc: Rob Herring <robh@xxxxxxxxxx>
Cc: Rodrigo Vivi <rodrigo.vivi@xxxxxxxxx>
Cc: Roman Gushchin <roman.gushchin@xxxxxxxxx>
Cc: Sean Paul <sean@xxxxxxxxxx>
Cc: Sergey Senozhatsky <senozhatsky@xxxxxxxxxxxx>
Cc: Song Liu <song@xxxxxxxxxx>
Cc: Stefano Stabellini <sstabellini@xxxxxxxxxx>
Cc: Steven Price <steven.price@xxxxxxx>
Cc: "Theodore Ts'o" <tytso@xxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Tomeu Vizoso <tomeu.vizoso@xxxxxxxxxxxxx>
Cc: Tom Talpey <tom@xxxxxxxxxx>
Cc: Trond Myklebust <trond.myklebust@xxxxxxxxxxxxxxx>
Cc: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxxxxxxxx>
Cc: Vlastimil Babka <vbabka@xxxxxxx>
Cc: Xuan Zhuo <xuanzhuo@xxxxxxxxxxxxxxxxx>
Cc: Yue Hu <huyue2@xxxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
include/linux/shrinker.h | 24 +++++++++
mm/shrinker.c | 89 ++++++++++++++++++++++++++++---------
2 files changed, 92 insertions(+), 21 deletions(-)
--- a/include/linux/shrinker.h~mm-shrinker-make-global-slab-shrink-lockless
+++ a/include/linux/shrinker.h
@@ -4,6 +4,8 @@
#include <linux/atomic.h>
#include <linux/types.h>
+#include <linux/refcount.h>
+#include <linux/completion.h>
#define SHRINKER_UNIT_BITS BITS_PER_LONG
@@ -87,6 +89,17 @@ struct shrinker {
int seeks; /* seeks to recreate an obj */
unsigned flags;
+ /*
+ * The reference count of this shrinker. Registered shrinker have an
+ * initial refcount of 1, then the lookup operations are now allowed
+ * to use it via shrinker_try_get(). Later in the unregistration step,
+ * the initial refcount will be discarded, and will free the shrinker
+ * asynchronously via RCU after its refcount reaches 0.
+ */
+ refcount_t refcount;
+ struct completion done; /* use to wait for refcount to reach 0 */
+ struct rcu_head rcu;
+
void *private_data;
/* These are for internal use */
@@ -120,6 +133,17 @@ struct shrinker *shrinker_alloc(unsigned
void shrinker_register(struct shrinker *shrinker);
void shrinker_free(struct shrinker *shrinker);
+static inline bool shrinker_try_get(struct shrinker *shrinker)
+{
+ return refcount_inc_not_zero(&shrinker->refcount);
+}
+
+static inline void shrinker_put(struct shrinker *shrinker)
+{
+ if (refcount_dec_and_test(&shrinker->refcount))
+ complete(&shrinker->done);
+}
+
#ifdef CONFIG_SHRINKER_DEBUG
extern int __printf(2, 3) shrinker_debugfs_rename(struct shrinker *shrinker,
const char *fmt, ...);
--- a/mm/shrinker.c~mm-shrinker-make-global-slab-shrink-lockless
+++ a/mm/shrinker.c
@@ -2,6 +2,7 @@
#include <linux/memcontrol.h>
#include <linux/rwsem.h>
#include <linux/shrinker.h>
+#include <linux/rculist.h>
#include <trace/events/vmscan.h>
#include "internal.h"
@@ -576,33 +577,50 @@ unsigned long shrink_slab(gfp_t gfp_mask
if (!mem_cgroup_disabled() && !mem_cgroup_is_root(memcg))
return shrink_slab_memcg(gfp_mask, nid, memcg, priority);
- if (!down_read_trylock(&shrinker_rwsem))
- goto out;
-
- list_for_each_entry(shrinker, &shrinker_list, list) {
+ /*
+ * lockless algorithm of global shrink.
+ *
+ * In the unregistration setp, the shrinker will be freed asynchronously
+ * via RCU after its refcount reaches 0. So both rcu_read_lock() and
+ * shrinker_try_get() can be used to ensure the existence of the shrinker.
+ *
+ * So in the global shrink:
+ * step 1: use rcu_read_lock() to guarantee existence of the shrinker
+ * and the validity of the shrinker_list walk.
+ * step 2: use shrinker_try_get() to try get the refcount, if successful,
+ * then the existence of the shrinker can also be guaranteed,
+ * so we can release the RCU lock to do do_shrink_slab() that
+ * may sleep.
+ * step 3: *MUST* to reacquire the RCU lock before calling shrinker_put(),
+ * which ensures that neither this shrinker nor the next shrinker
+ * will be freed in the next traversal operation.
+ * step 4: do shrinker_put() paired with step 2 to put the refcount,
+ * if the refcount reaches 0, then wake up the waiter in
+ * shrinker_free() by calling complete().
+ */
+ rcu_read_lock();
+ list_for_each_entry_rcu(shrinker, &shrinker_list, list) {
struct shrink_control sc = {
.gfp_mask = gfp_mask,
.nid = nid,
.memcg = memcg,
};
+ if (!shrinker_try_get(shrinker))
+ continue;
+
+ rcu_read_unlock();
+
ret = do_shrink_slab(&sc, shrinker, priority);
if (ret == SHRINK_EMPTY)
ret = 0;
freed += ret;
- /*
- * Bail out if someone want to register a new shrinker to
- * prevent the registration from being stalled for long periods
- * by parallel ongoing shrinking.
- */
- if (rwsem_is_contended(&shrinker_rwsem)) {
- freed = freed ? : 1;
- break;
- }
+
+ rcu_read_lock();
+ shrinker_put(shrinker);
}
- up_read(&shrinker_rwsem);
-out:
+ rcu_read_unlock();
cond_resched();
return freed;
}
@@ -671,13 +689,29 @@ void shrinker_register(struct shrinker *
}
down_write(&shrinker_rwsem);
- list_add_tail(&shrinker->list, &shrinker_list);
+ list_add_tail_rcu(&shrinker->list, &shrinker_list);
shrinker->flags |= SHRINKER_REGISTERED;
shrinker_debugfs_add(shrinker);
up_write(&shrinker_rwsem);
+
+ init_completion(&shrinker->done);
+ /*
+ * Now the shrinker is fully set up, take the first reference to it to
+ * indicate that lookup operations are now allowed to use it via
+ * shrinker_try_get().
+ */
+ refcount_set(&shrinker->refcount, 1);
}
EXPORT_SYMBOL_GPL(shrinker_register);
+static void shrinker_free_rcu_cb(struct rcu_head *head)
+{
+ struct shrinker *shrinker = container_of(head, struct shrinker, rcu);
+
+ kfree(shrinker->nr_deferred);
+ kfree(shrinker);
+}
+
void shrinker_free(struct shrinker *shrinker)
{
struct dentry *debugfs_entry = NULL;
@@ -686,9 +720,25 @@ void shrinker_free(struct shrinker *shri
if (!shrinker)
return;
+ if (shrinker->flags & SHRINKER_REGISTERED) {
+ /* drop the initial refcount */
+ shrinker_put(shrinker);
+ /*
+ * Wait for all lookups of the shrinker to complete, after that,
+ * no shrinker is running or will run again, then we can safely
+ * free it asynchronously via RCU and safely free the structure
+ * where the shrinker is located, such as super_block etc.
+ */
+ wait_for_completion(&shrinker->done);
+ }
+
down_write(&shrinker_rwsem);
if (shrinker->flags & SHRINKER_REGISTERED) {
- list_del(&shrinker->list);
+ /*
+ * Now we can safely remove it from the shrinker_list and then
+ * free it.
+ */
+ list_del_rcu(&shrinker->list);
debugfs_entry = shrinker_debugfs_detach(shrinker, &debugfs_id);
shrinker->flags &= ~SHRINKER_REGISTERED;
} else {
@@ -702,9 +752,6 @@ void shrinker_free(struct shrinker *shri
if (debugfs_entry)
shrinker_debugfs_remove(debugfs_entry, debugfs_id);
- kfree(shrinker->nr_deferred);
- shrinker->nr_deferred = NULL;
-
- kfree(shrinker);
+ call_rcu(&shrinker->rcu, shrinker_free_rcu_cb);
}
EXPORT_SYMBOL_GPL(shrinker_free);
_
Patches currently in -mm which might be from zhengqi.arch@xxxxxxxxxxxxx are
mm-move-some-shrinker-related-function-declarations-to-mm-internalh.patch
mm-vmscan-move-shrinker-related-code-into-a-separate-file.patch
mm-shrinker-remove-redundant-shrinker_rwsem-in-debugfs-operations.patch
drm-ttm-introduce-pool_shrink_rwsem.patch
mm-shrinker-add-infrastructure-for-dynamically-allocating-shrinker.patch
kvm-mmu-dynamically-allocate-the-x86-mmu-shrinker.patch
binder-dynamically-allocate-the-android-binder-shrinker.patch
drm-ttm-dynamically-allocate-the-drm-ttm_pool-shrinker.patch
xenbus-backend-dynamically-allocate-the-xen-backend-shrinker.patch
erofs-dynamically-allocate-the-erofs-shrinker.patch
f2fs-dynamically-allocate-the-f2fs-shrinker.patch
gfs2-dynamically-allocate-the-gfs2-glock-shrinker.patch
gfs2-dynamically-allocate-the-gfs2-qd-shrinker.patch
nfsv42-dynamically-allocate-the-nfs-xattr-shrinkers.patch
nfs-dynamically-allocate-the-nfs-acl-shrinker.patch
nfsd-dynamically-allocate-the-nfsd-filecache-shrinker.patch
quota-dynamically-allocate-the-dquota-cache-shrinker.patch
ubifs-dynamically-allocate-the-ubifs-slab-shrinker.patch
rcu-dynamically-allocate-the-rcu-lazy-shrinker.patch
rcu-dynamically-allocate-the-rcu-kfree-shrinker.patch
mm-thp-dynamically-allocate-the-thp-related-shrinkers.patch
sunrpc-dynamically-allocate-the-sunrpc_cred-shrinker.patch
mm-workingset-dynamically-allocate-the-mm-shadow-shrinker.patch
drm-i915-dynamically-allocate-the-i915_gem_mm-shrinker.patch
drm-msm-dynamically-allocate-the-drm-msm_gem-shrinker.patch
drm-panfrost-dynamically-allocate-the-drm-panfrost-shrinker.patch
dm-dynamically-allocate-the-dm-bufio-shrinker.patch
dm-zoned-dynamically-allocate-the-dm-zoned-meta-shrinker.patch
md-raid5-dynamically-allocate-the-md-raid5-shrinker.patch
bcache-dynamically-allocate-the-md-bcache-shrinker.patch
vmw_balloon-dynamically-allocate-the-vmw-balloon-shrinker.patch
virtio_balloon-dynamically-allocate-the-virtio-balloon-shrinker.patch
mbcache-dynamically-allocate-the-mbcache-shrinker.patch
ext4-dynamically-allocate-the-ext4-es-shrinker.patch
jbd2ext4-dynamically-allocate-the-jbd2-journal-shrinker.patch
nfsd-dynamically-allocate-the-nfsd-client-shrinker.patch
nfsd-dynamically-allocate-the-nfsd-reply-shrinker.patch
xfs-dynamically-allocate-the-xfs-buf-shrinker.patch
xfs-dynamically-allocate-the-xfs-inodegc-shrinker.patch
xfs-dynamically-allocate-the-xfs-qm-shrinker.patch
zsmalloc-dynamically-allocate-the-mm-zspool-shrinker.patch
fs-super-dynamically-allocate-the-s_shrink.patch
mm-shrinker-remove-old-apis.patch
mm-shrinker-add-a-secondary-array-for-shrinker_info-map-nr_deferred.patch
mm-shrinker-rename-preallocunregister_memcg_shrinker-to-shrinker_memcg_allocremove.patch
mm-shrinker-make-global-slab-shrink-lockless.patch
mm-shrinker-make-memcg-slab-shrink-lockless.patch
mm-shrinker-hold-write-lock-to-reparent-shrinker-nr_deferred.patch
mm-shrinker-convert-shrinker_rwsem-to-mutex.patch
