linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Qi Zheng <zhengqi.arch@bytedance.com>
To: akpm@linux-foundation.org, david@fromorbit.com, tkhai@ya.ru,
	vbabka@suse.cz, roman.gushchin@linux.dev, djwong@kernel.org,
	brauner@kernel.org, paulmck@kernel.org, tytso@mit.edu,
	steven.price@arm.com, cel@kernel.org, senozhatsky@chromium.org,
	yujie.liu@intel.com, gregkh@linuxfoundation.org,
	muchun.song@linux.dev
Cc: linux-kernel@vger.kernel.org, linux-mm@kvack.org, x86@kernel.org,
	kvm@vger.kernel.org, xen-devel@lists.xenproject.org,
	linux-erofs@lists.ozlabs.org,
	linux-f2fs-devel@lists.sourceforge.net, cluster-devel@redhat.com,
	linux-nfs@vger.kernel.org, linux-mtd@lists.infradead.org,
	rcu@vger.kernel.org, netdev@vger.kernel.org,
	dri-devel@lists.freedesktop.org, linux-arm-msm@vger.kernel.org,
	dm-devel@redhat.com, linux-raid@vger.kernel.org,
	linux-bcache@vger.kernel.org,
	virtualization@lists.linux-foundation.org,
	linux-fsdevel@vger.kernel.org, linux-ext4@vger.kernel.org,
	linux-xfs@vger.kernel.org, linux-btrfs@vger.kernel.org,
	Qi Zheng <zhengqi.arch@bytedance.com>
Subject: [PATCH v3 47/49] mm: shrinker: make memcg slab shrink lockless
Date: Thu, 27 Jul 2023 16:05:00 +0800	[thread overview]
Message-ID: <20230727080502.77895-48-zhengqi.arch@bytedance.com> (raw)
In-Reply-To: <20230727080502.77895-1-zhengqi.arch@bytedance.com>

Like global slab shrink, this commit also uses refcount+RCU method to make
memcg slab shrink lockless.

Use the following script to do slab shrink stress test:

```

DIR="/root/shrinker/memcg/mnt"

do_create()
{
    mkdir -p /sys/fs/cgroup/memory/test
    echo 4G > /sys/fs/cgroup/memory/test/memory.limit_in_bytes
    for i in `seq 0 $1`;
    do
        mkdir -p /sys/fs/cgroup/memory/test/$i;
        echo $$ > /sys/fs/cgroup/memory/test/$i/cgroup.procs;
        mkdir -p $DIR/$i;
    done
}

do_mount()
{
    for i in `seq $1 $2`;
    do
        mount -t tmpfs $i $DIR/$i;
    done
}

do_touch()
{
    for i in `seq $1 $2`;
    do
        echo $$ > /sys/fs/cgroup/memory/test/$i/cgroup.procs;
        dd if=/dev/zero of=$DIR/$i/file$i bs=1M count=1 &
    done
}

case "$1" in
  touch)
    do_touch $2 $3
    ;;
  test)
    do_create 4000
    do_mount 0 4000
    do_touch 0 3000
    ;;
  *)
    exit 1
    ;;
esac
```

Save the above script, then run test and touch commands. Then we can use
the following perf command to view hotspots:

perf top -U -F 999

1) Before applying this patchset:

  40.44%  [kernel]            [k] down_read_trylock
  17.59%  [kernel]            [k] up_read
  13.64%  [kernel]            [k] pv_native_safe_halt
  11.90%  [kernel]            [k] shrink_slab
   8.21%  [kernel]            [k] idr_find
   2.71%  [kernel]            [k] _find_next_bit
   1.36%  [kernel]            [k] shrink_node
   0.81%  [kernel]            [k] shrink_lruvec
   0.80%  [kernel]            [k] __radix_tree_lookup
   0.50%  [kernel]            [k] do_shrink_slab
   0.21%  [kernel]            [k] list_lru_count_one
   0.16%  [kernel]            [k] mem_cgroup_iter

2) After applying this patchset:

  60.17%  [kernel]           [k] shrink_slab
  20.42%  [kernel]           [k] pv_native_safe_halt
   3.03%  [kernel]           [k] do_shrink_slab
   2.73%  [kernel]           [k] shrink_node
   2.27%  [kernel]           [k] shrink_lruvec
   2.00%  [kernel]           [k] __rcu_read_unlock
   1.92%  [kernel]           [k] mem_cgroup_iter
   0.98%  [kernel]           [k] __rcu_read_lock
   0.91%  [kernel]           [k] osq_lock
   0.63%  [kernel]           [k] mem_cgroup_calculate_protection
   0.55%  [kernel]           [k] shrinker_put
   0.46%  [kernel]           [k] list_lru_count_one

We can see that the first perf hotspot becomes shrink_slab, which is what
we expect.

Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
---
 mm/shrinker.c | 80 ++++++++++++++++++++++++++++++++++-----------------
 1 file changed, 54 insertions(+), 26 deletions(-)

diff --git a/mm/shrinker.c b/mm/shrinker.c
index d318f5621862..fee6f62904fb 100644
--- a/mm/shrinker.c
+++ b/mm/shrinker.c
@@ -107,6 +107,12 @@ static struct shrinker_info *shrinker_info_protected(struct mem_cgroup *memcg,
 					 lockdep_is_held(&shrinker_rwsem));
 }
 
+static struct shrinker_info *shrinker_info_rcu(struct mem_cgroup *memcg,
+					       int nid)
+{
+	return rcu_dereference(memcg->nodeinfo[nid]->shrinker_info);
+}
+
 static int expand_one_shrinker_info(struct mem_cgroup *memcg, int new_size,
 				    int old_size, int new_nr_max)
 {
@@ -198,7 +204,7 @@ void set_shrinker_bit(struct mem_cgroup *memcg, int nid, int shrinker_id)
 		struct shrinker_info_unit *unit;
 
 		rcu_read_lock();
-		info = rcu_dereference(memcg->nodeinfo[nid]->shrinker_info);
+		info = shrinker_info_rcu(memcg, nid);
 		unit = info->unit[shriner_id_to_index(shrinker_id)];
 		if (!WARN_ON_ONCE(shrinker_id >= info->map_nr_max)) {
 			/* Pairs with smp mb in shrink_slab() */
@@ -211,7 +217,7 @@ void set_shrinker_bit(struct mem_cgroup *memcg, int nid, int shrinker_id)
 
 static DEFINE_IDR(shrinker_idr);
 
-static int prealloc_memcg_shrinker(struct shrinker *shrinker)
+static int shrinker_memcg_alloc(struct shrinker *shrinker)
 {
 	int id, ret = -ENOMEM;
 
@@ -219,7 +225,6 @@ static int prealloc_memcg_shrinker(struct shrinker *shrinker)
 		return -ENOSYS;
 
 	down_write(&shrinker_rwsem);
-	/* This may call shrinker, so it must use down_read_trylock() */
 	id = idr_alloc(&shrinker_idr, shrinker, 0, 0, GFP_KERNEL);
 	if (id < 0)
 		goto unlock;
@@ -237,7 +242,7 @@ static int prealloc_memcg_shrinker(struct shrinker *shrinker)
 	return ret;
 }
 
-static void unregister_memcg_shrinker(struct shrinker *shrinker)
+static void shrinker_memcg_remove(struct shrinker *shrinker)
 {
 	int id = shrinker->id;
 
@@ -253,10 +258,15 @@ static long xchg_nr_deferred_memcg(int nid, struct shrinker *shrinker,
 {
 	struct shrinker_info *info;
 	struct shrinker_info_unit *unit;
+	long nr_deferred;
 
-	info = shrinker_info_protected(memcg, nid);
+	rcu_read_lock();
+	info = shrinker_info_rcu(memcg, nid);
 	unit = info->unit[shriner_id_to_index(shrinker->id)];
-	return atomic_long_xchg(&unit->nr_deferred[shriner_id_to_offset(shrinker->id)], 0);
+	nr_deferred = atomic_long_xchg(&unit->nr_deferred[shriner_id_to_offset(shrinker->id)], 0);
+	rcu_read_unlock();
+
+	return nr_deferred;
 }
 
 static long add_nr_deferred_memcg(long nr, int nid, struct shrinker *shrinker,
@@ -264,10 +274,16 @@ static long add_nr_deferred_memcg(long nr, int nid, struct shrinker *shrinker,
 {
 	struct shrinker_info *info;
 	struct shrinker_info_unit *unit;
+	long nr_deferred;
 
-	info = shrinker_info_protected(memcg, nid);
+	rcu_read_lock();
+	info = shrinker_info_rcu(memcg, nid);
 	unit = info->unit[shriner_id_to_index(shrinker->id)];
-	return atomic_long_add_return(nr, &unit->nr_deferred[shriner_id_to_offset(shrinker->id)]);
+	nr_deferred =
+		atomic_long_add_return(nr, &unit->nr_deferred[shriner_id_to_offset(shrinker->id)]);
+	rcu_read_unlock();
+
+	return nr_deferred;
 }
 
 void reparent_shrinker_deferred(struct mem_cgroup *memcg)
@@ -299,12 +315,12 @@ void reparent_shrinker_deferred(struct mem_cgroup *memcg)
 	up_read(&shrinker_rwsem);
 }
 #else
-static int prealloc_memcg_shrinker(struct shrinker *shrinker)
+static int shrinker_memcg_alloc(struct shrinker *shrinker)
 {
 	return -ENOSYS;
 }
 
-static void unregister_memcg_shrinker(struct shrinker *shrinker)
+static void shrinker_memcg_remove(struct shrinker *shrinker)
 {
 }
 
@@ -464,18 +480,23 @@ static unsigned long shrink_slab_memcg(gfp_t gfp_mask, int nid,
 	if (!mem_cgroup_online(memcg))
 		return 0;
 
-	if (!down_read_trylock(&shrinker_rwsem))
-		return 0;
-
-	info = shrinker_info_protected(memcg, nid);
+again:
+	rcu_read_lock();
+	info = shrinker_info_rcu(memcg, nid);
 	if (unlikely(!info))
 		goto unlock;
 
-	for (; index < shriner_id_to_index(info->map_nr_max); index++) {
+	if (index < shriner_id_to_index(info->map_nr_max)) {
 		struct shrinker_info_unit *unit;
 
 		unit = info->unit[index];
 
+		/*
+		 * The shrinker_info_unit will not be freed, so we can
+		 * safely release the RCU lock here.
+		 */
+		rcu_read_unlock();
+
 		for_each_set_bit(offset, unit->map, SHRINKER_UNIT_BITS) {
 			struct shrink_control sc = {
 				.gfp_mask = gfp_mask,
@@ -485,12 +506,14 @@ static unsigned long shrink_slab_memcg(gfp_t gfp_mask, int nid,
 			struct shrinker *shrinker;
 			int shrinker_id = calc_shrinker_id(index, offset);
 
+			rcu_read_lock();
 			shrinker = idr_find(&shrinker_idr, shrinker_id);
-			if (unlikely(!shrinker || !(shrinker->flags & SHRINKER_REGISTERED))) {
-				if (!shrinker)
-					clear_bit(offset, unit->map);
+			if (unlikely(!shrinker || !shrinker_try_get(shrinker))) {
+				clear_bit(offset, unit->map);
+				rcu_read_unlock();
 				continue;
 			}
+			rcu_read_unlock();
 
 			/* Call non-slab shrinkers even though kmem is disabled */
 			if (!memcg_kmem_online() &&
@@ -523,15 +546,20 @@ static unsigned long shrink_slab_memcg(gfp_t gfp_mask, int nid,
 					set_shrinker_bit(memcg, nid, shrinker_id);
 			}
 			freed += ret;
-
-			if (rwsem_is_contended(&shrinker_rwsem)) {
-				freed = freed ? : 1;
-				goto unlock;
-			}
+			shrinker_put(shrinker);
 		}
+
+		/*
+		 * We have already exited the read-side of rcu critical section
+		 * before calling do_shrink_slab(), the shrinker_info may be
+		 * released in expand_one_shrinker_info(), so reacquire the
+		 * shrinker_info.
+		 */
+		index++;
+		goto again;
 	}
 unlock:
-	up_read(&shrinker_rwsem);
+	rcu_read_unlock();
 	return freed;
 }
 #else /* !CONFIG_MEMCG */
@@ -638,7 +666,7 @@ struct shrinker *shrinker_alloc(unsigned int flags, const char *fmt, ...)
 	shrinker->flags = flags | SHRINKER_ALLOCATED;
 
 	if (flags & SHRINKER_MEMCG_AWARE) {
-		err = prealloc_memcg_shrinker(shrinker);
+		err = shrinker_memcg_alloc(shrinker);
 		if (err == -ENOSYS)
 			shrinker->flags &= ~SHRINKER_MEMCG_AWARE;
 		else if (err == 0)
@@ -731,7 +759,7 @@ void shrinker_free(struct shrinker *shrinker)
 	}
 
 	if (shrinker->flags & SHRINKER_MEMCG_AWARE)
-		unregister_memcg_shrinker(shrinker);
+		shrinker_memcg_remove(shrinker);
 	up_write(&shrinker_rwsem);
 
 	if (debugfs_entry)
-- 
2.30.2



  parent reply	other threads:[~2023-07-27  8:15 UTC|newest]

Thread overview: 67+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-07-27  8:04 [PATCH v3 00/49] use refcount+RCU method to implement lockless slab shrink Qi Zheng
2023-07-27  8:04 ` [PATCH v3 01/49] binder: fix memory leak in binder_init() Qi Zheng
2023-07-27  8:04 ` [PATCH v3 02/49] mm: move some shrinker-related function declarations to mm/internal.h Qi Zheng
2023-07-27  8:04 ` [PATCH v3 03/49] mm: vmscan: move shrinker-related code into a separate file Qi Zheng
2023-07-27  8:04 ` [PATCH v3 04/49] mm: shrinker: remove redundant shrinker_rwsem in debugfs operations Qi Zheng
2023-07-28  8:13   ` Simon Horman
2023-07-28  8:19     ` Qi Zheng
2023-07-27  8:04 ` [PATCH v3 05/49] mm: shrinker: add infrastructure for dynamically allocating shrinker Qi Zheng
2023-07-28 12:17   ` Simon Horman
2023-07-29  8:47     ` Qi Zheng
2023-07-27  8:04 ` [PATCH v3 06/49] kvm: mmu: dynamically allocate the x86-mmu shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 07/49] binder: dynamically allocate the android-binder shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 08/49] drm/ttm: dynamically allocate the drm-ttm_pool shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 09/49] xenbus/backend: dynamically allocate the xen-backend shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 10/49] erofs: dynamically allocate the erofs-shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 11/49] f2fs: dynamically allocate the f2fs-shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 12/49] gfs2: dynamically allocate the gfs2-glock shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 13/49] gfs2: dynamically allocate the gfs2-qd shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 14/49] NFSv4.2: dynamically allocate the nfs-xattr shrinkers Qi Zheng
2023-07-27  8:04 ` [PATCH v3 15/49] nfs: dynamically allocate the nfs-acl shrinker Qi Zheng
2023-07-27  9:06   ` Qi Zheng
2023-07-27  8:04 ` [PATCH v3 16/49] nfsd: dynamically allocate the nfsd-filecache shrinker Qi Zheng
2023-07-27  9:08   ` Qi Zheng
2023-07-27  8:04 ` [PATCH v3 17/49] quota: dynamically allocate the dquota-cache shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 18/49] ubifs: dynamically allocate the ubifs-slab shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 19/49] rcu: dynamically allocate the rcu-lazy shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 20/49] rcu: dynamically allocate the rcu-kfree shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 21/49] mm: thp: dynamically allocate the thp-related shrinkers Qi Zheng
2023-07-27  8:04 ` [PATCH v3 22/49] sunrpc: dynamically allocate the sunrpc_cred shrinker Qi Zheng
2023-07-27  9:10   ` Qi Zheng
2023-07-27  8:04 ` [PATCH v3 23/49] mm: workingset: dynamically allocate the mm-shadow shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 24/49] drm/i915: dynamically allocate the i915_gem_mm shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 25/49] drm/msm: dynamically allocate the drm-msm_gem shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 26/49] drm/panfrost: dynamically allocate the drm-panfrost shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 27/49] dm: dynamically allocate the dm-bufio shrinker Qi Zheng
2023-07-27  9:13   ` Qi Zheng
2023-07-27  8:04 ` [PATCH v3 28/49] dm zoned: dynamically allocate the dm-zoned-meta shrinker Qi Zheng
2023-07-27  8:30   ` Damien Le Moal
2023-07-27  8:55     ` Qi Zheng
2023-07-27 10:20       ` Damien Le Moal
2023-07-27 10:32         ` Qi Zheng
2023-07-27 22:59         ` Dave Chinner
2023-07-27 23:48           ` Damien Le Moal
2023-07-27  8:04 ` [PATCH v3 29/49] md/raid5: dynamically allocate the md-raid5 shrinker Qi Zheng
2023-07-27  9:15   ` Qi Zheng
2023-07-27  8:04 ` [PATCH v3 30/49] bcache: dynamically allocate the md-bcache shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 31/49] vmw_balloon: dynamically allocate the vmw-balloon shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 32/49] virtio_balloon: dynamically allocate the virtio-balloon shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 33/49] mbcache: dynamically allocate the mbcache shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 34/49] ext4: dynamically allocate the ext4-es shrinker Qi Zheng
2023-07-27  9:17   ` Qi Zheng
2023-07-27  8:04 ` [PATCH v3 35/49] jbd2,ext4: dynamically allocate the jbd2-journal shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 36/49] nfsd: dynamically allocate the nfsd-client shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 37/49] nfsd: dynamically allocate the nfsd-reply shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 38/49] xfs: dynamically allocate the xfs-buf shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 39/49] xfs: dynamically allocate the xfs-inodegc shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 40/49] xfs: dynamically allocate the xfs-qm shrinker Qi Zheng
2023-07-27  9:19   ` Qi Zheng
2023-07-27  8:04 ` [PATCH v3 41/49] zsmalloc: dynamically allocate the mm-zspool shrinker Qi Zheng
2023-07-27  8:04 ` [PATCH v3 42/49] fs: super: dynamically allocate the s_shrink Qi Zheng
2023-07-27  8:04 ` [PATCH v3 43/49] mm: shrinker: remove old APIs Qi Zheng
2023-07-27  8:04 ` [PATCH v3 44/49] drm/ttm: introduce pool_shrink_rwsem Qi Zheng
2023-07-27  8:04 ` [PATCH v3 45/49] mm: shrinker: add a secondary array for shrinker_info::{map, nr_deferred} Qi Zheng
2023-07-27  8:04 ` [PATCH v3 46/49] mm: shrinker: make global slab shrink lockless Qi Zheng
2023-07-27  8:05 ` Qi Zheng [this message]
2023-07-27  8:05 ` [PATCH v3 48/49] mm: shrinker: hold write lock to reparent shrinker nr_deferred Qi Zheng
2023-07-27  8:05 ` [PATCH v3 49/49] mm: shrinker: convert shrinker_rwsem to mutex Qi Zheng

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230727080502.77895-48-zhengqi.arch@bytedance.com \
    --to=zhengqi.arch@bytedance.com \
    --cc=akpm@linux-foundation.org \
    --cc=brauner@kernel.org \
    --cc=cel@kernel.org \
    --cc=cluster-devel@redhat.com \
    --cc=david@fromorbit.com \
    --cc=djwong@kernel.org \
    --cc=dm-devel@redhat.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=gregkh@linuxfoundation.org \
    --cc=kvm@vger.kernel.org \
    --cc=linux-arm-msm@vger.kernel.org \
    --cc=linux-bcache@vger.kernel.org \
    --cc=linux-btrfs@vger.kernel.org \
    --cc=linux-erofs@lists.ozlabs.org \
    --cc=linux-ext4@vger.kernel.org \
    --cc=linux-f2fs-devel@lists.sourceforge.net \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linux-mtd@lists.infradead.org \
    --cc=linux-nfs@vger.kernel.org \
    --cc=linux-raid@vger.kernel.org \
    --cc=linux-xfs@vger.kernel.org \
    --cc=muchun.song@linux.dev \
    --cc=netdev@vger.kernel.org \
    --cc=paulmck@kernel.org \
    --cc=rcu@vger.kernel.org \
    --cc=roman.gushchin@linux.dev \
    --cc=senozhatsky@chromium.org \
    --cc=steven.price@arm.com \
    --cc=tkhai@ya.ru \
    --cc=tytso@mit.edu \
    --cc=vbabka@suse.cz \
    --cc=virtualization@lists.linux-foundation.org \
    --cc=x86@kernel.org \
    --cc=xen-devel@lists.xenproject.org \
    --cc=yujie.liu@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox