The patch titled Subject: memcg: return -EINTR at bypassing try_charge() has been added to the -mm tree. Its filename is memcg-simplify-lru-handling-by-new-rule-memcg-return-eintr-at-bypassing-try_charge.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/ ------------------------------------------------------ From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx> Subject: memcg: return -EINTR at bypassing try_charge() This patch is a fix for memcg-simplify-lru-handling-by-new-rule.patch When running testprogram and stop it by Ctrl-C, add_lru/del_lru will find pc->mem_cgroup is NULL and get panic. The reason is bypass code in try_charge(). At try_charge(), it checks the thread is fatal or not as.. fatal_signal_pending() or TIF_MEMDIE. In this case, __try_charge() returns 0(success) with setting *ptr as NULL. Now, lruvec are deteremined by pc->mem_cgroup. So, it's better to reset pc->mem_cgroup as root_mem_cgroup. This patch does following change in try_charge() 1. return -EINTR at bypassing. 2. set *ptr = root_mem_cgroup at bypassing. By this change, in page fault / radix-tree-insert path, the page will be charged against root_mem_cgroup and the thread's operations will go ahead without trouble. In other path, migration or move_account etc..., -EINTR will stop the operation. (may need some cleanup later..) After this change, pc->mem_cgroup will have valid pointer if the page is used. Changelog: v2 -> v3 - handle !mm case in another way. - removed redundant commments - fixed move_parent bug of uninitialized pointer Changelog: v1 -> v2 - returns -EINTR at bypassing. - change error code handling at callers. - changed the name of patch. Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx> Cc: Hugh Dickins <hughd@xxxxxxxxxx> Cc: Johannes Weiner <hannes@xxxxxxxxxxx> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx> Cc: Michal Hocko <mhocko@xxxxxxx> Cc: Miklos Szeredi <mszeredi@xxxxxxx> Cc: Ying Han <yinghan@xxxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- mm/memcontrol.c | 45 ++++++++++++++++++++++++++++++++++----------- 1 file changed, 34 insertions(+), 11 deletions(-) diff -puN mm/memcontrol.c~memcg-simplify-lru-handling-by-new-rule-memcg-return-eintr-at-bypassing-try_charge mm/memcontrol.c --- a/mm/memcontrol.c~memcg-simplify-lru-handling-by-new-rule-memcg-return-eintr-at-bypassing-try_charge +++ a/mm/memcontrol.c @@ -2186,6 +2186,24 @@ static int mem_cgroup_do_charge(struct m } /* + * __mem_cgroup_try_charge() does + * 1. detect memcg to be charged against from passed *mm and *ptr, + * 2. update res_counter + * 3. call memory reclaim if necessary. + * + * In some special case, if the task is fatal, fatal_signal_pending() or + * TIF_MEMDIE, this functoion returns -EINTR with filling *ptr as + * root_mem_cgroup. There are 2 reasons for this. 1st is that + * fatal threads should quit as soon as possible without any hazards. + * 2nd is that all page should have valid pc->mem_cgroup if it will be + * used. If mm is NULL and the caller doesn't pass valid memcg pointer, + * that's treated as charge to root_mem_cgroup. + * + * So, try_charge will return + * 0 ... at success. filling *ptr with a valid memcg pointer. + * -ENOMEM ... charge failure because of resource limits. + * -EINTR ... if thread is fatal. *ptr is filled with root_mem_cgroup. + * * Unlike exported interface, "oom" parameter is added. if oom==true, * oom-killer can be invoked. */ @@ -2216,7 +2234,7 @@ static int __mem_cgroup_try_charge(struc * set, if so charge the init_mm (happens for pagecache usage). */ if (!*ptr && !mm) - goto bypass; + *ptr = root_mem_cgroup; again: if (*ptr) { /* css should be a valid one */ memcg = *ptr; @@ -2317,8 +2335,8 @@ nomem: *ptr = NULL; return -ENOMEM; bypass: - *ptr = NULL; - return 0; + *ptr = root_mem_cgroup; + return -EINTR; } /* @@ -2584,7 +2602,7 @@ static int mem_cgroup_move_parent(struct parent = mem_cgroup_from_cont(pcg); ret = __mem_cgroup_try_charge(NULL, gfp_mask, nr_pages, &parent, false); - if (ret || !parent) + if (ret) goto put_back; if (nr_pages > 1) @@ -2631,9 +2649,8 @@ static int mem_cgroup_charge_common(stru pc = lookup_page_cgroup(page); ret = __mem_cgroup_try_charge(mm, gfp_mask, nr_pages, &memcg, oom); - if (ret || !memcg) + if (ret == -ENOMEM) return ret; - __mem_cgroup_commit_charge(memcg, page, nr_pages, pc, ctype); return 0; } @@ -2744,11 +2761,16 @@ int mem_cgroup_try_charge_swapin(struct *memcgp = memcg; ret = __mem_cgroup_try_charge(NULL, mask, 1, memcgp, true); css_put(&memcg->css); + if (ret == -EINTR) + ret = 0; return ret; charge_cur_mm: if (unlikely(!mm)) mm = &init_mm; - return __mem_cgroup_try_charge(mm, mask, 1, memcgp, true); + ret = __mem_cgroup_try_charge(mm, mask, 1, memcgp, true); + if (ret == -EINTR) + ret = 0; + return ret; } static void @@ -3205,7 +3227,7 @@ int mem_cgroup_prepare_migration(struct *memcgp = memcg; ret = __mem_cgroup_try_charge(NULL, gfp_mask, 1, memcgp, false); css_put(&memcg->css);/* drop extra refcnt */ - if (ret || *memcgp == NULL) { + if (ret) { if (PageAnon(page)) { lock_page_cgroup(pc); ClearPageCgroupMigration(pc); @@ -3215,6 +3237,7 @@ int mem_cgroup_prepare_migration(struct */ mem_cgroup_uncharge_page(page); } + /* we'll need to revisit this error code (we have -EINTR) */ return -ENOMEM; } /* @@ -3633,7 +3656,7 @@ static int mem_cgroup_force_empty_list(s pc = lookup_page_cgroup(page); ret = mem_cgroup_move_parent(page, pc, memcg, GFP_KERNEL); - if (ret == -ENOMEM) + if (ret == -ENOMEM || ret == -EINTR) break; if (ret == -EBUSY || ret == -EINVAL) { @@ -5091,9 +5114,9 @@ one_by_one: } ret = __mem_cgroup_try_charge(NULL, GFP_KERNEL, 1, &memcg, false); - if (ret || !memcg) + if (ret) /* mem_cgroup_clear_mc() will do uncharge later */ - return -ENOMEM; + return ret; mc.precharge++; } return ret; _ Subject: Subject: memcg: return -EINTR at bypassing try_charge() Patches currently in -mm which might be from kamezawa.hiroyu@xxxxxxxxxxxxxx are origin.patch linux-next.patch vmscan-promote-shared-file-mapped-pages.patch vmscan-activate-executable-pages-after-first-usage.patch mm-avoid-livelock-on-__gfp_fs-allocations-v2.patch mm-hugetlbc-fix-virtual-address-handling-in-hugetlb-fault.patch mm-hugetlbc-fix-virtual-address-handling-in-hugetlb-fault-fix.patch vmscan-add-task-name-to-warn_scan_unevictable-messages.patch mm-exclude-reserved-pages-from-dirtyable-memory.patch mm-exclude-reserved-pages-from-dirtyable-memory-fix.patch mm-writeback-cleanups-in-preparation-for-per-zone-dirty-limits.patch mm-try-to-distribute-dirty-pages-fairly-across-zones.patch mm-filemap-pass-__gfp_write-from-grab_cache_page_write_begin.patch btrfs-pass-__gfp_write-for-buffered-write-page-allocations.patch mm-simplify-find_vma_prev.patch tracepoint-add-tracepoints-for-debugging-oom_score_adj.patch mm-hugetlb-fix-pgoff-computation-when-unmapping-page-from-vma.patch mm-hugetlb-fix-pgoff-computation-when-unmapping-page-from-vma-fix.patch memcg-add-mem_cgroup_replace_page_cache-to-fix-lru-issue.patch mm-memcg-consolidate-hierarchy-iteration-primitives.patch mm-vmscan-distinguish-global-reclaim-from-global-lru-scanning.patch mm-vmscan-distinguish-between-memcg-triggering-reclaim-and-memcg-being-scanned.patch mm-memcg-per-priority-per-zone-hierarchy-scan-generations.patch mm-move-memcg-hierarchy-reclaim-to-generic-reclaim-code.patch mm-memcg-remove-optimization-of-keeping-the-root_mem_cgroup-lru-lists-empty.patch mm-vmscan-convert-global-reclaim-to-per-memcg-lru-lists.patch mm-collect-lru-list-heads-into-struct-lruvec.patch mm-make-per-memcg-lru-lists-exclusive.patch mm-memcg-remove-unused-node-section-info-from-pc-flags.patch mm-memcg-remove-unused-node-section-info-from-pc-flags-fix.patch memcg-make-mem_cgroup_split_huge_fixup-more-efficient.patch memcg-make-mem_cgroup_split_huge_fixup-more-efficient-fix.patch mm-memcg-shorten-preempt-disabled-section-around-event-checks.patch documentation-cgroups-memorytxt-fix-typo.patch memcg-fix-pgpgin-pgpgout-documentation.patch mm-oom_kill-remove-memcg-argument-from-oom_kill_task.patch mm-unify-remaining-mem_cont-mem-etc-variable-names-to-memcg.patch mm-memcg-clean-up-fault-accounting.patch mm-memcg-lookup_page_cgroup-almost-never-returns-null.patch mm-page_cgroup-check-page_cgroup-arrays-in-lookup_page_cgroup-only-when-necessary.patch mm-memcg-remove-unneeded-checks-from-newpage_charge.patch mm-memcg-remove-unneeded-checks-from-uncharge_page.patch page_cgroup-add-helper-function-to-get-swap_cgroup.patch page_cgroup-add-helper-function-to-get-swap_cgroup-cleanup.patch memcg-clean-up-soft_limit_tree-if-allocation-fails.patch oom-memcg-fix-exclusion-of-memcg-threads-after-they-have-detached-their-mm.patch memcg-simplify-page-cache-charging.patch memcg-simplify-corner-case-handling-of-lru.patch memcg-clear-pc-mem_cgorup-if-necessary.patch memcg-clear-pc-mem_cgorup-if-necessary-fix.patch memcg-clear-pc-mem_cgorup-if-necessary-fix-2.patch memcg-clear-pc-mem_cgorup-if-necessary-fix-2-fix.patch memcg-clear-pc-mem_cgorup-if-necessary-comments.patch memcg-simplify-lru-handling-by-new-rule.patch memcg-simplify-lru-handling-by-new-rule-fix.patch memcg-simplify-lru-handling-by-new-rule-memcg-return-eintr-at-bypassing-try_charge.patch memcg-simplify-lru-handling-by-new-rule-memcg-return-eintr-at-bypassing-try_charge-fix.patch memcg-cleanup-for_each_node_state.patch page_alloc-break-early-in-check_for_regular_memory.patch page_cgroup-drop-multi-config_memory_hotplug.patch vmscan-trace-add-file-info-to-trace_mm_vmscan_lru_isolate.patch c-r-introduce-checkpoint_restore-symbol.patch c-r-procfs-add-start_data-end_data-start_brk-members-to-proc-pid-stat-v4.patch c-r-procfs-add-start_data-end_data-start_brk-members-to-proc-pid-stat-v4-fix.patch c-r-prctl-add-pr_set_mm-codes-to-set-up-mm_struct-entries.patch c-r-prctl-add-pr_set_mm-codes-to-set-up-mm_struct-entries-fix.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