From b6bf9abb0aa44e53ffe9c1e6e1d32568f5b25e4a Mon Sep 17 00:00:00 2001 From: Dan Schatzberg Date: Fri, 14 Jan 2022 14:05:35 -0800 Subject: mm/memcg: add oom_group_kill memory event Our container agent wants to know when a container exits if it was OOM killed or not to report to the user. We use memory.oom.group = 1 to ensure that OOM kills within the container's cgroup kill everything. Existing memory.events are insufficient for knowing if this triggered: 1) Our current approach reads memory.events oom_kill and reports the container was killed if the value is non-zero. This is erroneous in some cases where containers create their children cgroups with memory.oom.group=1 as such OOM kills will get counted against the container cgroup's oom_kill counter despite not actually OOM killing the entire container. 2) Reading memory.events.local will fail to identify OOM kills in leaf cgroups (that don't set memory.oom.group) within the container cgroup. This patch adds a new oom_group_kill event when memory.oom.group triggers to allow userspace to cleanly identify when an entire cgroup is oom killed. [schatzberg.dan@gmail.com: changes from Johannes and Chris] Link: https://lkml.kernel.org/r/20211213162511.2492267-1-schatzberg.dan@gmail.com Link: https://lkml.kernel.org/r/20211203162426.3375036-1-schatzberg.dan@gmail.com Signed-off-by: Dan Schatzberg Reviewed-by: Roman Gushchin Acked-by: Johannes Weiner Acked-by: Chris Down Reviewed-by: Shakeel Butt Acked-by: Michal Hocko Cc: Tejun Heo Cc: Zefan Li Cc: Jonathan Corbet Cc: Vladimir Davydov Cc: Matthew Wilcox (Oracle) Cc: Muchun Song Cc: Alex Shi Cc: Wei Yang Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/oom_kill.c | 1 + 1 file changed, 1 insertion(+) (limited to 'mm/oom_kill.c') diff --git a/mm/oom_kill.c b/mm/oom_kill.c index 1ddabefcfb5a..e52ce0b1465d 100644 --- a/mm/oom_kill.c +++ b/mm/oom_kill.c @@ -994,6 +994,7 @@ static void oom_kill_process(struct oom_control *oc, const char *message) * If necessary, kill all tasks in the selected memory cgroup. */ if (oom_group) { + memcg_memory_event(oom_group, MEMCG_OOM_GROUP_KILL); mem_cgroup_print_oom_group(oom_group); mem_cgroup_scan_tasks(oom_group, oom_kill_memcg_member, (void *)message); -- cgit v1.2.3-59-g8ed1b From ba535c1caf3ee78aa7719e9e4b07a0dc1d153b9e Mon Sep 17 00:00:00 2001 From: Suren Baghdasaryan Date: Fri, 14 Jan 2022 14:06:22 -0800 Subject: mm/oom_kill: allow process_mrelease to run under mmap_lock protection With exit_mmap holding mmap_write_lock during free_pgtables call, process_mrelease does not need to elevate mm->mm_users in order to prevent exit_mmap from destrying pagetables while __oom_reap_task_mm is walking the VMA tree. The change prevents process_mrelease from calling the last mmput, which can lead to waiting for IO completion in exit_aio. Link: https://lkml.kernel.org/r/20211209191325.3069345-3-surenb@google.com Signed-off-by: Suren Baghdasaryan Acked-by: Michal Hocko Reviewed-by: Jason Gunthorpe Cc: Andrea Arcangeli Cc: Andy Lutomirski Cc: Christian Brauner Cc: Christian Brauner Cc: Christoph Hellwig Cc: David Hildenbrand Cc: David Rientjes Cc: Florian Weimer Cc: Jan Engelhardt Cc: Jann Horn Cc: Johannes Weiner Cc: Kirill A. Shutemov Cc: Matthew Wilcox Cc: Minchan Kim Cc: Oleg Nesterov Cc: Rik van Riel Cc: Roman Gushchin Cc: Shakeel Butt Cc: Tim Murray Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/oom_kill.c | 27 +++++++++++++++------------ 1 file changed, 15 insertions(+), 12 deletions(-) (limited to 'mm/oom_kill.c') diff --git a/mm/oom_kill.c b/mm/oom_kill.c index e52ce0b1465d..3390316c8a32 100644 --- a/mm/oom_kill.c +++ b/mm/oom_kill.c @@ -1170,15 +1170,15 @@ SYSCALL_DEFINE2(process_mrelease, int, pidfd, unsigned int, flags) goto put_task; } - if (mmget_not_zero(p->mm)) { - mm = p->mm; - if (task_will_free_mem(p)) - reap = true; - else { - /* Error only if the work has not been done already */ - if (!test_bit(MMF_OOM_SKIP, &mm->flags)) - ret = -EINVAL; - } + mm = p->mm; + mmgrab(mm); + + if (task_will_free_mem(p)) + reap = true; + else { + /* Error only if the work has not been done already */ + if (!test_bit(MMF_OOM_SKIP, &mm->flags)) + ret = -EINVAL; } task_unlock(p); @@ -1189,13 +1189,16 @@ SYSCALL_DEFINE2(process_mrelease, int, pidfd, unsigned int, flags) ret = -EINTR; goto drop_mm; } - if (!__oom_reap_task_mm(mm)) + /* + * Check MMF_OOM_SKIP again under mmap_read_lock protection to ensure + * possible change in exit_mmap is seen + */ + if (!test_bit(MMF_OOM_SKIP, &mm->flags) && !__oom_reap_task_mm(mm)) ret = -EAGAIN; mmap_read_unlock(mm); drop_mm: - if (mm) - mmput(mm); + mmdrop(mm); put_task: put_task_struct(task); return ret; -- cgit v1.2.3-59-g8ed1b From f530243a172d2ff03f88d0056f838928d6445c6d Mon Sep 17 00:00:00 2001 From: Jann Horn Date: Fri, 14 Jan 2022 14:08:27 -0800 Subject: mm, oom: OOM sysrq should always kill a process The OOM kill sysrq (alt+sysrq+F) should allow the user to kill the process with the highest OOM badness with a single execution. However, at the moment, the OOM kill can bail out if an OOM notifier (e.g. the i915 one) says that it reclaimed a tiny amount of memory from somewhere. That's probably not what the user wants, so skip the bailout if the OOM was triggered via sysrq. Link: https://lkml.kernel.org/r/20220106102605.635656-1-jannh@google.com Signed-off-by: Jann Horn Acked-by: Michal Hocko Acked-by: David Rientjes Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/oom_kill.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'mm/oom_kill.c') diff --git a/mm/oom_kill.c b/mm/oom_kill.c index 3390316c8a32..3934ff500878 100644 --- a/mm/oom_kill.c +++ b/mm/oom_kill.c @@ -1058,7 +1058,7 @@ bool out_of_memory(struct oom_control *oc) if (!is_memcg_oom(oc)) { blocking_notifier_call_chain(&oom_notify_list, 0, &freed); - if (freed > 0) + if (freed > 0 && !is_sysrq_oom(oc)) /* Got some memory back in the last second. */ return true; } -- cgit v1.2.3-59-g8ed1b