The patch titled ksm: unmerge is an origin of OOMs has been added to the -mm tree. Its filename is ksm-unmerge-is-an-origin-of-ooms.patch 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/SubmitChecklist when testing your code *** See http://userweb.kernel.org/~akpm/stuff/added-to-mm.txt to find out what to do about this The current -mm tree may be found at http://userweb.kernel.org/~akpm/mmotm/ ------------------------------------------------------ Subject: ksm: unmerge is an origin of OOMs From: Hugh Dickins <hugh.dickins@xxxxxxxxxxxxx> Just as the swapoff system call allocates many pages of RAM to various processes, perhaps triggering OOM, so "echo 2 >/sys/kernel/mm/ksm/run" (unmerge) is liable to allocate many pages of RAM to various processes, perhaps triggering OOM; and each is normally run from a modest admin process (swapoff or shell), easily repeated until it succeeds. So treat unmerge_and_remove_all_rmap_items() in the same way that we treat try_to_unuse(): generalize PF_SWAPOFF to PF_OOM_ORIGIN, and bracket both with that, to ask the OOM killer to kill them first, to prevent them from spawning more and more OOM kills. Signed-off-by: Hugh Dickins <hugh.dickins@xxxxxxxxxxxxx> Acked-by: Izik Eidus <ieidus@xxxxxxxxxx> Cc: Andrea Arcangeli <aarcange@xxxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- include/linux/sched.h | 2 +- mm/ksm.c | 2 ++ mm/oom_kill.c | 2 +- mm/swapfile.c | 4 ++-- 4 files changed, 6 insertions(+), 4 deletions(-) diff -puN include/linux/sched.h~ksm-unmerge-is-an-origin-of-ooms include/linux/sched.h --- a/include/linux/sched.h~ksm-unmerge-is-an-origin-of-ooms +++ a/include/linux/sched.h @@ -1723,7 +1723,7 @@ extern cputime_t task_gtime(struct task_ #define PF_FROZEN 0x00010000 /* frozen for system suspend */ #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ #define PF_KSWAPD 0x00040000 /* I am kswapd */ -#define PF_SWAPOFF 0x00080000 /* I am in swapoff */ +#define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */ #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ diff -puN mm/ksm.c~ksm-unmerge-is-an-origin-of-ooms mm/ksm.c --- a/mm/ksm.c~ksm-unmerge-is-an-origin-of-ooms +++ a/mm/ksm.c @@ -1564,7 +1564,9 @@ static ssize_t run_store(struct kobject if (ksm_run != flags) { ksm_run = flags; if (flags & KSM_RUN_UNMERGE) { + current->flags |= PF_OOM_ORIGIN; err = unmerge_and_remove_all_rmap_items(); + current->flags &= ~PF_OOM_ORIGIN; if (err) { ksm_run = KSM_RUN_STOP; count = err; diff -puN mm/oom_kill.c~ksm-unmerge-is-an-origin-of-ooms mm/oom_kill.c --- a/mm/oom_kill.c~ksm-unmerge-is-an-origin-of-ooms +++ a/mm/oom_kill.c @@ -79,7 +79,7 @@ unsigned long badness(struct task_struct /* * swapoff can easily use up all memory, so kill those first. */ - if (p->flags & PF_SWAPOFF) + if (p->flags & PF_OOM_ORIGIN) return ULONG_MAX; /* diff -puN mm/swapfile.c~ksm-unmerge-is-an-origin-of-ooms mm/swapfile.c --- a/mm/swapfile.c~ksm-unmerge-is-an-origin-of-ooms +++ a/mm/swapfile.c @@ -1573,9 +1573,9 @@ SYSCALL_DEFINE1(swapoff, const char __us p->flags &= ~SWP_WRITEOK; spin_unlock(&swap_lock); - current->flags |= PF_SWAPOFF; + current->flags |= PF_OOM_ORIGIN; err = try_to_unuse(type); - current->flags &= ~PF_SWAPOFF; + current->flags &= ~PF_OOM_ORIGIN; if (err) { /* re-insert swap space back into swap_list */ _ Patches currently in -mm which might be from hugh.dickins@xxxxxxxxxxxxx are linux-next.patch acpi-fix-null-bug-for-hid-uid-string-2.patch vfs-optimize-touch_time-too-fix.patch fs-new-truncate-helpers.patch fs-use-new-truncate-helpers.patch fs-introduce-new-truncate-sequence.patch fs-convert-simple-fs-to-new-truncate.patch tmpfs-convert-to-use-the-new-truncate-convention.patch ext2-convert-to-use-the-new-truncate-convention.patch fat-convert-to-use-the-new-truncate-convention.patch btrfs-convert-to-use-the-new-truncate-convention.patch jfs-convert-to-use-the-new-truncate-convention.patch udf-convert-to-use-the-new-truncate-convention.patch minix-convert-to-use-the-new-truncate-convention.patch mm-oom-analysis-add-shmem-vmstat.patch ksm-add-mmu_notifier-set_pte_at_notify.patch ksm-first-tidy-up-madvise_vma.patch ksm-define-madv_mergeable-and-madv_unmergeable.patch ksm-the-mm-interface-to-ksm.patch ksm-no-debug-in-page_dup_rmap.patch ksm-identify-pageksm-pages.patch ksm-kernel-samepage-merging.patch ksm-prevent-mremap-move-poisoning.patch ksm-change-copyright-message.patch ksm-change-ksm-nice-level-to-be-5.patch ksm-rename-kernel_pages_allocated.patch ksm-move-pages_sharing-updates.patch ksm-pages_unshared-and-pages_volatile.patch ksm-break-cow-once-unshared.patch ksm-keep-quiet-while-list-empty.patch ksm-five-little-cleanups.patch ksm-fix-endless-loop-on-oom.patch ksm-distribute-remove_mm_from_lists.patch ksm-fix-oom-deadlock.patch ksm-fix-deadlock-with-munlock-in-exit_mmap.patch ksm-sysfs-and-defaults.patch ksm-add-some-documentation.patch ksm-remove-vm_mergeable_flags.patch ksm-clean-up-obsolete-references.patch ksm-unmerge-is-an-origin-of-ooms.patch ksm-mremap-use-err-from-ksm_madvise.patch mm-add_to_swap_cache-must-not-sleep.patch mm-add_to_swap_cache-does-not-return-eexist.patch mm-includecheck-fix-for-mm-shmemc.patch mm-introduce-page_lru_base_type-fix.patch mm-replace-various-uses-of-num_physpages-by-totalram_pages.patch hugetlbfs-allow-the-creation-of-files-suitable-for-map_private-on-the-vfs-internal-mount.patch hugetlb-add-map_hugetlb-for-mmaping-pseudo-anonymous-huge-page-regions.patch hugetlb-add-map_hugetlb-example.patch getrusage-fill-ru_maxrss-value.patch getrusage-fill-ru_maxrss-value-update.patch ramfs-move-ramfs_magic-to-include-linux-magich.patch memory-controller-soft-limit-organize-cgroups-v9-fix.patch prio_tree-debugging-patch.patch -- To unsubscribe from this list: send the line "unsubscribe mm-commits" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html