Skip to contentSkip to conversation

memcg, oom: no oom-kill for __GFP_RETRY_MAYFAIL

PATCH RESEND Superseded Shakeel Butt 2 participants 3 messages May 12, 2019 – May 13, 2019

Subsystems (7)
Memory management VFS cgroup memcg +3 more subsystems

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH RESEND v2 1/2 memcg, oom: no oom-kill for __GFP_RETRY_MAYFAIL

The documentation of __GFP_RETRY_MAYFAIL clearly mentioned that the OOM killer will not be triggered and indeed the page alloc does not invoke OOM killer for such allocations. However we do trigger memcg OOM killer for __GFP_RETRY_MAYFAIL. Fix that. This flag will used later to not trigger oom-killer in the charging path for fanotify and inotify event allocations.

---
Changelog since v1:
- commit message updated.

 mm/memcontrol.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
mm/memcontrol.c +1 -3
index 2535e54e7989..9548dfcae432 100644
@@ -2294,7 +2294,6 @@ static int try_charge(struct mem_cgroup *memcg, gfp_t gfp_mask,
 	unsigned long nr_reclaimed;
 	bool may_swap = true;
 	bool drained = false;
-	bool oomed = false;
 	enum oom_status oom_status;
 
 	if (mem_cgroup_is_root(memcg))
@@ -2381,7 +2380,7 @@ static int try_charge(struct mem_cgroup *memcg, gfp_t gfp_mask,
 	if (nr_retries--)
 		goto retry;
 
-	if (gfp_mask & __GFP_RETRY_MAYFAIL && oomed)
+	if (gfp_mask & __GFP_RETRY_MAYFAIL)
 		goto nomem;
 
 	if (gfp_mask & __GFP_NOFAIL)
@@ -2400,7 +2399,6 @@ static int try_charge(struct mem_cgroup *memcg, gfp_t gfp_mask,
 	switch (oom_status) {
 	case OOM_SUCCESS:
 		nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
-		oomed = true;
 		goto retry;
 	case OOM_FAILED:
 		goto force;
Signature · 1 line
2.21.0.1020.gf2820cf01a-goog
#1 lore raw

PATCH RESEND v2 2/2 memcg, fsnotify: no oom-kill for remote memcg charging

The commit d46eb14b735b ("fs: fsnotify: account fsnotify metadata to kmemcg") added remote memcg charging for fanotify and inotify event objects. The aim was to charge the memory to the listener who is interested in the events but without triggering the OOM killer. Otherwise there would be security concerns for the listener. At the time, oom-kill trigger was not in the charging path. A parallel work added the oom-kill back to charging path i.e. commit 29ef680ae7c2 ("memcg, oom: move out_of_memory back to the charge path"). So to not trigger oom-killer in the remote memcg, explicitly add __GFP_RETRY_MAYFAIL to the fanotigy and inotify event allocations.

---
Changelog since v1:
- Fixed usage of __GFP_RETRY_MAYFAIL flag.

 fs/notify/fanotify/fanotify.c        | 5 ++++-
 fs/notify/inotify/inotify_fsnotify.c | 7 +++++--
 2 files changed, 9 insertions(+), 3 deletions(-)
fs/notify/fanotify/fanotify.c +4 -1
index 6b9c27548997..f78fd4c8f12d 100644
@@ -288,10 +288,13 @@ struct fanotify_event *fanotify_alloc_event(struct fsnotify_group *group,
 	/*
 	 * For queues with unlimited length lost events are not expected and
 	 * can possibly have security implications. Avoid losing events when
-	 * memory is short.
+	 * memory is short. Also make sure to not trigger OOM killer in the
+	 * target memcg for the limited size queues.
 	 */
 	if (group->max_events == UINT_MAX)
 		gfp |= __GFP_NOFAIL;
+	else
+		gfp |= __GFP_RETRY_MAYFAIL;
 
 	/* Whoever is interested in the event, pays for the allocation. */
 	memalloc_use_memcg(group->memcg);
fs/notify/inotify/inotify_fsnotify.c +5 -2
index ff30abd6a49b..17c08daa1ba7 100644
@@ -99,9 +99,12 @@ int inotify_handle_event(struct fsnotify_group *group,
 	i_mark = container_of(inode_mark, struct inotify_inode_mark,
 			      fsn_mark);
 
-	/* Whoever is interested in the event, pays for the allocation. */
+	/*
+	 * Whoever is interested in the event, pays for the allocation. However
+	 * do not trigger the OOM killer in the target memcg.
+	 */
 	memalloc_use_memcg(group->memcg);
-	event = kmalloc(alloc_len, GFP_KERNEL_ACCOUNT);
+	event = kmalloc(alloc_len, GFP_KERNEL_ACCOUNT | __GFP_RETRY_MAYFAIL);
 	memalloc_unuse_memcg();
 
 	if (unlikely(!event)) {
Signature · 1 line
2.21.0.1020.gf2820cf01a-goog
maintainer
#2 lore raw

Re: [RESEND PATCH v2 2/2] memcg, fsnotify: no oom-kill for remote memcg charging

On Sun 12-05-19 09:09:27, Shakeel Butt wrote:
[...]

Quoted 24 lines
fs/notify/fanotify/fanotify.c +4 -1
index 6b9c27548997..f78fd4c8f12d 100644
@@ -288,10 +288,13 @@ struct fanotify_event *fanotify_alloc_event(struct fsnotify_group *group,
 	/*
 	 * For queues with unlimited length lost events are not expected and
 	 * can possibly have security implications. Avoid losing events when
-	 * memory is short.
+	 * memory is short. Also make sure to not trigger OOM killer in the
+	 * target memcg for the limited size queues.
 	 */
 	if (group->max_events == UINT_MAX)
 		gfp |= __GFP_NOFAIL;
+	else
+		gfp |= __GFP_RETRY_MAYFAIL;
 
 	/* Whoever is interested in the event, pays for the allocation. */
 	memalloc_use_memcg(group->memcg);
fs/notify/inotify/inotify_fsnotify.c +0 -0
index ff30abd6a49b..17c08daa1ba7 100644

fs/notify/inotify/inotify_fsnotify.c int inotify_handle_event(struct fsnotify_group *group, · line 104

Quoted 7 lines
fs/notify/inotify/inotify_fsnotify.c +3 -1
@@ -99,9 +99,12 @@ int inotify_handle_event(struct fsnotify_group *group,
 	i_mark = container_of(inode_mark, struct inotify_inode_mark,
 			      fsn_mark);
 
-	/* Whoever is interested in the event, pays for the allocation. */
+	/*
+	 * Whoever is interested in the event, pays for the allocation. However
+	 * do not trigger the OOM killer in the target memcg.

Both comments would be much more helpful if they mentioned _why_ we do not want to trigger the OOM iller.

Quoted 7 lines
fs/notify/inotify/inotify_fsnotify.c +2 -1
+	 */
 	memalloc_use_memcg(group->memcg);
-	event = kmalloc(alloc_len, GFP_KERNEL_ACCOUNT);
+	event = kmalloc(alloc_len, GFP_KERNEL_ACCOUNT | __GFP_RETRY_MAYFAIL);
 	memalloc_unuse_memcg();
 
 	if (unlikely(!event)) {
Signature · 2 lines
Michal Hocko
SUSE Labs