diff mbox series

[100/146] mm, oom: OOM sysrq should always kill a process

Message ID	20220114220827.OSbTBO1rc%akpm@linux-foundation.org (mailing list archive)
State	New
Headers	show Return-Path: <owner-linux-mm@kvack.org> Date: Fri, 14 Jan 2022 14:08:27 -0800 From: Andrew Morton <akpm@linux-foundation.org> To: akpm@linux-foundation.org, jannh@google.com, linux-mm@kvack.org, mhocko@suse.com, mm-commits@vger.kernel.org, rientjes@google.com, torvalds@linux-foundation.org Subject: [patch 100/146] mm, oom: OOM sysrq should always kill a process Message-ID: <20220114220827.OSbTBO1rc%akpm@linux-foundation.org> In-Reply-To: <20220114140222.6b14f0061194d3200000c52d@linux-foundation.org> User-Agent: s-nail v14.8.16 Sender: owner-linux-mm@kvack.org Precedence: bulk
Series	[001/146] kthread: add the helper function kthread_run_on_cpu() \| expand [001/146] kthread: add the helper function kthread_run_on_cpu() [002/146] RDMA/siw: make use of the helper function kthread_run_on_cpu() [003/146] ring-buffer: make use of the helper function kthread_run_on_cpu() [004/146] rcutorture: make use of the helper function kthread_run_on_cpu() [005/146] trace/osnoise: make use of the helper function kthread_run_on_cpu() [006/146] trace/hwlat: make use of the helper function kthread_run_on_cpu() [007/146] ia64: module: use swap() to make code cleaner [008/146] arch/ia64/kernel/setup.c: use swap() to make code cleaner [009/146] ia64: fix typo in a comment [010/146] ia64: topology: use default_groups in kobj_type [011/146] scripts/spelling.txt: add "oveflow" [012/146] fs/ntfs/attrib.c: fix one kernel-doc comment [013/146] squashfs: provide backing_dev_info in order to disable read-ahead [014/146] ocfs2: use BUG_ON instead of if condition followed by BUG. [015/146] ocfs2: clearly handle ocfs2_grab_pages_for_write() return value [016/146] ocfs2: use default_groups in kobj_type [017/146] ocfs2: remove redundant assignment to pointer root_bh [018/146] ocfs2: cluster: use default_groups in kobj_type [019/146] ocfs2: remove redundant assignment to variable free_space [020/146] fs/ioctl: remove unnecessary __user annotation [021/146] mm/slab_common: use WARN() if cache still has objects on destroy [022/146] mm: slab: make slab iterator functions static [023/146] kmemleak: fix kmemleak false positive report with HW tag-based kasan enable [024/146] mm: kmemleak: alloc gray object for reserved region with direct map [025/146] mm: defer kmemleak object creation of module_alloc() [026/146] mm/page_alloc: split prep_compound_page into head and tail subparts [027/146] mm/page_alloc: refactor memmap_init_zone_device() page init [028/146] mm/memremap: add ZONE_DEVICE support for compound pages [029/146] device-dax: use ALIGN() for determining pgoff [030/146] device-dax: use struct_size() [031/146] device-dax: ensure dev_dax->pgmap is valid for dynamic devices [032/146] device-dax: factor out page mapping initialization [033/146] device-dax: set mapping prior to vmf_insert_pfn{,_pmd,pud}() [034/146] device-dax: remove pfn from __dev_dax_{pte,pmd,pud}_fault() [035/146] device-dax: compound devmap support [036/146] kasan: test: add globals left-out-of-bounds test [037/146] kasan: add ability to detect double-kmem_cache_destroy() [038/146] kasan: test: add test case for double-kmem_cache_destroy() [039/146] kasan: fix quarantine conflicting with init_on_free [040/146] mm,fs: split dump_mapping() out from dump_page() [041/146] mm/debug_vm_pgtable: update comments regarding migration swap entries [042/146] mm/truncate.c: remove unneeded variable [043/146] gup: avoid multiple user access locking/unlocking in fault_in_{read/write}able [044/146] mm/gup.c: stricter check on THP migration entry during follow_pmd_mask [045/146] mm: shmem: don't truncate page if memory failure happens [046/146] shmem: fix a race between shmem_unused_huge_shrink and shmem_evict_inode [047/146] mm/frontswap.c: use non-atomic '__set_bit()' when possible [048/146] mm: memcontrol: make cgroup_memory_nokmem static [049/146] mm/page_counter: remove an incorrect call to propagate_protected_usage() [050/146] mm/memcg: add oom_group_kill memory event [051/146] memcg: better bounds on the memcg stats updates [052/146] mm/memcg: use struct_size() helper in kzalloc() [053/146] memcg: add per-memcg vmalloc stat [054/146] tools/testing/selftests/vm/userfaultfd.c: use swap() to make code cleaner [055/146] mm: remove redundant check about FAULT_FLAG_ALLOW_RETRY bit [056/146] mm: rearrange madvise code to allow for reuse [057/146] mm: add a field to store names for private anonymous memory [058/146] mm: add anonymous vma name refcounting [059/146] mm: move anon_vma declarations to linux/mm_inline.h [060/146] mm: move tlb_flush_pending inline helpers to mm_inline.h [061/146] mm: protect free_pgtables with mmap_lock write lock in exit_mmap [062/146] mm: document locking restrictions for vm_operations_struct::close [063/146] mm/oom_kill: allow process_mrelease to run under mmap_lock protection [064/146] docs/vm: add vmalloced-kernel-stacks document [065/146] mm: change page type prior to adding page table entry [066/146] mm: ptep_clear() page table helper [067/146] mm: page table check [068/146] x86: mm: add x86_64 support for page table check [069/146] mm: remove last argument of reuse_swap_page() [070/146] mm: remove the total_mapcount argument from page_trans_huge_map_swapcount() [071/146] mm: remove the total_mapcount argument from page_trans_huge_mapcount() [072/146] mm/dmapool.c: revert "make dma pool to use kmalloc_node" [073/146] mm/vmalloc: alloc GFP_NO{FS,IO} for vmalloc [074/146] mm/vmalloc: add support for __GFP_NOFAIL [075/146] mm/vmalloc: be more explicit about supported gfp flags. [076/146] mm: allow !GFP_KERNEL allocations for kvmalloc [077/146] mm: make slab and vmalloc allocators __GFP_NOLOCKDEP aware [078/146] mm: introduce memalloc_retry_wait() [079/146] mm/pagealloc: sysctl: change watermark_scale_factor max limit to 30% [080/146] mm: fix boolreturn.cocci warning [081/146] mm: page_alloc: fix building error on -Werror=array-compare [082/146] mm: drop node from alloc_pages_vma [083/146] include/linux/gfp.h: further document GFP_DMA32 [084/146] mm/page_alloc.c: modify the comment section for alloc_contig_pages() [085/146] mm_zone: add function to check if managed dma zone exists [086/146] dma/pool: create dma atomic pool only if dma zone has managed pages [087/146] mm/page_alloc.c: do not warn allocation failure on zone DMA if no managed pages [088/146] hugetlb: add hugetlb..numa_stat file [089/146] mm, hugepages: make memory size variable in hugepage-mremap selftest [090/146] mm/vmstat: add events for THP max_ptes_ exceeds [091/146] selftests/vm: make charge_reserved_hugetlb.sh work with existing cgroup setting [092/146] selftests/uffd: allow EINTR/EAGAIN [093/146] userfaultfd/selftests: clean up hugetlb allocation code [094/146] vmscan: make drop_slab_node static [095/146] mm/page_isolation: unset migratetype directly for non Buddy page [096/146] mm/mempolicy: use policy_node helper with MPOL_PREFERRED_MANY [097/146] mm/mempolicy: add set_mempolicy_home_node syscall [098/146] mm/mempolicy: wire up syscall set_mempolicy_home_node [099/146] mm/mempolicy: fix all kernel-doc warnings [100/146] mm, oom: OOM sysrq should always kill a process [101/146] hugetlbfs: fix off-by-one error in hugetlb_vmdelete_list() [102/146] mm: migrate: fix the return value of migrate_pages() [103/146] mm: migrate: correct the hugetlb migration stats [104/146] mm: compaction: fix the migration stats in trace_mm_compaction_migratepages() [105/146] mm: migrate: support multiple target nodes demotion [106/146] mm: migrate: add more comments for selecting target node randomly [107/146] mm/migrate: move node demotion code to near its user [108/146] mm/migrate: remove redundant variables used in a for-loop [109/146] mm/thp: drop unused trace events hugepage_[invalidate\|splitting] [110/146] mm: ksm: fix use-after-free kasan report in ksm_might_need_to_copy [111/146] mm/hwpoison: mf_mutex for soft offline and unpoison [112/146] mm/hwpoison: remove MF_MSG_BUDDY_2ND and MF_MSG_POISONED_HUGE [113/146] mm/hwpoison: fix unpoison_memory() [114/146] mm: memcg/percpu: account extra objcg space to memory cgroups [115/146] mm/rmap: fix potential batched TLB flush race [116/146] zpool: remove the list of pools_head [117/146] zram: use ATTRIBUTE_GROUPS [118/146] mm: fix some comment errors [119/146] mm: make some vars and functions static or __init [120/146] mm/hmm.c: allow VM_MIXEDMAP to work with hmm_range_fault [121/146] mm/damon: unified access_check function naming rules [122/146] mm/damon: add 'age' of region tracepoint support [123/146] mm/damon/core: use abs() instead of diff_of() [124/146] mm/damon: remove some unneeded function definitions in damon.h [125/146] mm/damon/vaddr: remove swap_ranges() and replace it with swap() [126/146] mm/damon/schemes: add the validity judgment of thresholds [127/146] mm/damon: move damon_rand() definition into damon.h [128/146] mm/damon: modify damon_rand() macro to static inline function [129/146] mm/damon: convert macro functions to static inline functions [131/146] Docs/admin-guide/mm/damon/usage: remove redundant information [132/146] Docs/admin-guide/mm/damon/usage: mention tracepoint at the beginning [133/146] Docs/admin-guide/mm/damon/usage: update for kdamond_pid and (mk\|rm)_contexts [134/146] mm/damon: remove a mistakenly added comment for a future feature [135/146] mm/damon/schemes: account scheme actions that successfully applied [136/146] mm/damon/schemes: account how many times quota limit has exceeded [137/146] mm/damon/reclaim: provide reclamation statistics [138/146] Docs/admin-guide/mm/damon/reclaim: document statistics parameters [139/146] mm/damon/dbgfs: support all DAMOS stats [140/146] Docs/admin-guide/mm/damon/usage: update for schemes statistics [141/146] mm/damon: add access checking for hugetlb pages [142/146] mm/damon: move the implementation of damon_insert_region to damon.h [143/146] mm/damon/dbgfs: remove an unnecessary variable [144/146] mm/damon/vaddr: use pr_debug() for damon_va_three_regions() failure logging [145/146] mm/damon/vaddr: hide kernel pointer from damon_va_three_regions() failure log [146/146] mm/damon: hide kernel pointer from tracepoint event

Commit Message

Andrew Morton Jan. 14, 2022, 10:08 p.m. UTC

From: Jann Horn <jannh@google.com>
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 <jannh@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 mm/oom_kill.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff mbox series

Patch

--- a/mm/oom_kill.c~mm-oom-oom-sysrq-should-always-kill-a-process
+++ a/mm/oom_kill.c
@@ -1058,7 +1058,7 @@  bool out_of_memory(struct oom_control *o
 
 	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;
 	}