From 0ae5e89c60c9eb87da36a2614836bc434b0ec2ad Mon Sep 17 00:00:00 2001 From: Ying Han Date: Thu, 26 May 2011 16:25:25 -0700 Subject: memcg: count the soft_limit reclaim in global background reclaim The global kswapd scans per-zone LRU and reclaims pages regardless of the cgroup. It breaks memory isolation since one cgroup can end up reclaiming pages from another cgroup. Instead we should rely on memcg-aware target reclaim including per-memcg kswapd and soft_limit hierarchical reclaim under memory pressure. In the global background reclaim, we do soft reclaim before scanning the per-zone LRU. However, the return value is ignored. This patch is the first step to skip shrink_zone() if soft_limit reclaim does enough work. This is part of the effort which tries to reduce reclaiming pages in global LRU in memcg. The per-memcg background reclaim patchset further enhances the per-cgroup targetting reclaim, which I should have V4 posted shortly. Try running multiple memory intensive workloads within seperate memcgs. Watch the counters of soft_steal in memory.stat. $ cat /dev/cgroup/A/memory.stat | grep 'soft' soft_steal 240000 soft_scan 240000 total_soft_steal 240000 total_soft_scan 240000 This patch: In the global background reclaim, we do soft reclaim before scanning the per-zone LRU. However, the return value is ignored. We would like to skip shrink_zone() if soft_limit reclaim does enough work. Also, we need to make the memory pressure balanced across per-memcg zones, like the logic vm-core. This patch is the first step where we start with counting the nr_scanned and nr_reclaimed from soft_limit reclaim into the global scan_control. Signed-off-by: Ying Han Cc: KOSAKI Motohiro Cc: Minchan Kim Cc: Rik van Riel Cc: Mel Gorman Cc: KAMEZAWA Hiroyuki Cc: Balbir Singh Acked-by: Daisuke Nishimura Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/memcontrol.h | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'include/linux/memcontrol.h') diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index 5e9840f5098..0629121f2c0 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -144,7 +144,8 @@ static inline void mem_cgroup_dec_page_stat(struct page *page, } unsigned long mem_cgroup_soft_limit_reclaim(struct zone *zone, int order, - gfp_t gfp_mask); + gfp_t gfp_mask, + unsigned long *total_scanned); u64 mem_cgroup_get_limit(struct mem_cgroup *mem); #ifdef CONFIG_TRANSPARENT_HUGEPAGE @@ -338,7 +339,8 @@ static inline void mem_cgroup_dec_page_stat(struct page *page, static inline unsigned long mem_cgroup_soft_limit_reclaim(struct zone *zone, int order, - gfp_t gfp_mask) + gfp_t gfp_mask, + unsigned long *total_scanned) { return 0; } -- cgit v1.2.3-70-g09d2 From 889976dbcb1218119fdd950fb7819084e37d7d37 Mon Sep 17 00:00:00 2001 From: Ying Han Date: Thu, 26 May 2011 16:25:33 -0700 Subject: memcg: reclaim memory from nodes in round-robin order Presently, memory cgroup's direct reclaim frees memory from the current node. But this has some troubles. Usually when a set of threads works in a cooperative way, they tend to operate on the same node. So if they hit limits under memcg they will reclaim memory from themselves, damaging the active working set. For example, assume 2 node system which has Node 0 and Node 1 and a memcg which has 1G limit. After some work, file cache remains and the usages are Node 0: 1M Node 1: 998M. and run an application on Node 0, it will eat its foot before freeing unnecessary file caches. This patch adds round-robin for NUMA and adds equal pressure to each node. When using cpuset's spread memory feature, this will work very well. But yes, a better algorithm is needed. [akpm@linux-foundation.org: comment editing] [kamezawa.hiroyu@jp.fujitsu.com: fix time comparisons] Signed-off-by: Ying Han Signed-off-by: KAMEZAWA Hiroyuki Cc: Balbir Singh Cc: KOSAKI Motohiro Cc: Daisuke Nishimura Cc: Mel Gorman Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/memcontrol.h | 1 + mm/memcontrol.c | 102 ++++++++++++++++++++++++++++++++++++++++++--- mm/vmscan.c | 10 ++++- 3 files changed, 106 insertions(+), 7 deletions(-) (limited to 'include/linux/memcontrol.h') diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index 0629121f2c0..16052117131 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -106,6 +106,7 @@ extern void mem_cgroup_end_migration(struct mem_cgroup *mem, */ int mem_cgroup_inactive_anon_is_low(struct mem_cgroup *memcg); int mem_cgroup_inactive_file_is_low(struct mem_cgroup *memcg); +int mem_cgroup_select_victim_node(struct mem_cgroup *memcg); unsigned long mem_cgroup_zone_nr_pages(struct mem_cgroup *memcg, struct zone *zone, enum lru_list lru); diff --git a/mm/memcontrol.c b/mm/memcontrol.c index fc62c714f3b..1520efd1c7c 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -231,6 +231,11 @@ struct mem_cgroup { * reclaimed from. */ int last_scanned_child; + int last_scanned_node; +#if MAX_NUMNODES > 1 + nodemask_t scan_nodes; + unsigned long next_scan_node_update; +#endif /* * Should the accounting and control be hierarchical, per subtree? */ @@ -624,18 +629,27 @@ static void mem_cgroup_charge_statistics(struct mem_cgroup *mem, preempt_enable(); } +static unsigned long +mem_cgroup_get_zonestat_node(struct mem_cgroup *mem, int nid, enum lru_list idx) +{ + struct mem_cgroup_per_zone *mz; + u64 total = 0; + int zid; + + for (zid = 0; zid < MAX_NR_ZONES; zid++) { + mz = mem_cgroup_zoneinfo(mem, nid, zid); + total += MEM_CGROUP_ZSTAT(mz, idx); + } + return total; +} static unsigned long mem_cgroup_get_local_zonestat(struct mem_cgroup *mem, enum lru_list idx) { - int nid, zid; - struct mem_cgroup_per_zone *mz; + int nid; u64 total = 0; for_each_online_node(nid) - for (zid = 0; zid < MAX_NR_ZONES; zid++) { - mz = mem_cgroup_zoneinfo(mem, nid, zid); - total += MEM_CGROUP_ZSTAT(mz, idx); - } + total += mem_cgroup_get_zonestat_node(mem, nid, idx); return total; } @@ -1418,6 +1432,81 @@ mem_cgroup_select_victim(struct mem_cgroup *root_mem) return ret; } +#if MAX_NUMNODES > 1 + +/* + * Always updating the nodemask is not very good - even if we have an empty + * list or the wrong list here, we can start from some node and traverse all + * nodes based on the zonelist. So update the list loosely once per 10 secs. + * + */ +static void mem_cgroup_may_update_nodemask(struct mem_cgroup *mem) +{ + int nid; + + if (time_after(mem->next_scan_node_update, jiffies)) + return; + + mem->next_scan_node_update = jiffies + 10*HZ; + /* make a nodemask where this memcg uses memory from */ + mem->scan_nodes = node_states[N_HIGH_MEMORY]; + + for_each_node_mask(nid, node_states[N_HIGH_MEMORY]) { + + if (mem_cgroup_get_zonestat_node(mem, nid, LRU_INACTIVE_FILE) || + mem_cgroup_get_zonestat_node(mem, nid, LRU_ACTIVE_FILE)) + continue; + + if (total_swap_pages && + (mem_cgroup_get_zonestat_node(mem, nid, LRU_INACTIVE_ANON) || + mem_cgroup_get_zonestat_node(mem, nid, LRU_ACTIVE_ANON))) + continue; + node_clear(nid, mem->scan_nodes); + } +} + +/* + * Selecting a node where we start reclaim from. Because what we need is just + * reducing usage counter, start from anywhere is O,K. Considering + * memory reclaim from current node, there are pros. and cons. + * + * Freeing memory from current node means freeing memory from a node which + * we'll use or we've used. So, it may make LRU bad. And if several threads + * hit limits, it will see a contention on a node. But freeing from remote + * node means more costs for memory reclaim because of memory latency. + * + * Now, we use round-robin. Better algorithm is welcomed. + */ +int mem_cgroup_select_victim_node(struct mem_cgroup *mem) +{ + int node; + + mem_cgroup_may_update_nodemask(mem); + node = mem->last_scanned_node; + + node = next_node(node, mem->scan_nodes); + if (node == MAX_NUMNODES) + node = first_node(mem->scan_nodes); + /* + * We call this when we hit limit, not when pages are added to LRU. + * No LRU may hold pages because all pages are UNEVICTABLE or + * memcg is too small and all pages are not on LRU. In that case, + * we use curret node. + */ + if (unlikely(node == MAX_NUMNODES)) + node = numa_node_id(); + + mem->last_scanned_node = node; + return node; +} + +#else +int mem_cgroup_select_victim_node(struct mem_cgroup *mem) +{ + return 0; +} +#endif + /* * Scan the hierarchy if needed to reclaim memory. We remember the last child * we reclaimed from, so that we don't end up penalizing one child extensively @@ -4606,6 +4695,7 @@ mem_cgroup_create(struct cgroup_subsys *ss, struct cgroup *cont) res_counter_init(&mem->memsw, NULL); } mem->last_scanned_child = 0; + mem->last_scanned_node = MAX_NUMNODES; INIT_LIST_HEAD(&mem->oom_notify); if (parent) diff --git a/mm/vmscan.c b/mm/vmscan.c index 884ae08c16c..b0875871820 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -2226,6 +2226,7 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont, { struct zonelist *zonelist; unsigned long nr_reclaimed; + int nid; struct scan_control sc = { .may_writepage = !laptop_mode, .may_unmap = 1, @@ -2242,7 +2243,14 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont, .gfp_mask = sc.gfp_mask, }; - zonelist = NODE_DATA(numa_node_id())->node_zonelists; + /* + * Unlike direct reclaim via alloc_pages(), memcg's reclaim doesn't + * take care of from where we get pages. So the node where we start the + * scan does not need to be the current node. + */ + nid = mem_cgroup_select_victim_node(mem_cont); + + zonelist = NODE_DATA(nid)->node_zonelists; trace_mm_vmscan_memcg_reclaim_begin(0, sc.may_writepage, -- cgit v1.2.3-70-g09d2 From 1bac180bd29e03989f50054af97b53b8d37a364a Mon Sep 17 00:00:00 2001 From: Ying Han Date: Thu, 26 May 2011 16:25:36 -0700 Subject: memcg: rename mem_cgroup_zone_nr_pages() to mem_cgroup_zone_nr_lru_pages() The caller of the function has been renamed to zone_nr_lru_pages(), and this is just fixing up in the memcg code. The current name is easily to be mis-read as zone's total number of pages. Signed-off-by: Ying Han Acked-by: Johannes Weiner Acked-by: KAMEZAWA Hiroyuki Reviewed-by: Minchan Kim Cc: Balbir Singh Cc: Daisuke Nishimura Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/memcontrol.h | 10 +++++----- mm/memcontrol.c | 6 +++--- mm/vmscan.c | 2 +- 3 files changed, 9 insertions(+), 9 deletions(-) (limited to 'include/linux/memcontrol.h') diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index 16052117131..ac1e5d20916 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -107,9 +107,9 @@ extern void mem_cgroup_end_migration(struct mem_cgroup *mem, int mem_cgroup_inactive_anon_is_low(struct mem_cgroup *memcg); int mem_cgroup_inactive_file_is_low(struct mem_cgroup *memcg); int mem_cgroup_select_victim_node(struct mem_cgroup *memcg); -unsigned long mem_cgroup_zone_nr_pages(struct mem_cgroup *memcg, - struct zone *zone, - enum lru_list lru); +unsigned long mem_cgroup_zone_nr_lru_pages(struct mem_cgroup *memcg, + struct zone *zone, + enum lru_list lru); struct zone_reclaim_stat *mem_cgroup_get_reclaim_stat(struct mem_cgroup *memcg, struct zone *zone); struct zone_reclaim_stat* @@ -304,8 +304,8 @@ mem_cgroup_inactive_file_is_low(struct mem_cgroup *memcg) } static inline unsigned long -mem_cgroup_zone_nr_pages(struct mem_cgroup *memcg, struct zone *zone, - enum lru_list lru) +mem_cgroup_zone_nr_lru_pages(struct mem_cgroup *memcg, struct zone *zone, + enum lru_list lru) { return 0; } diff --git a/mm/memcontrol.c b/mm/memcontrol.c index bcb0a0bee1f..cc48a6854f7 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -1078,9 +1078,9 @@ int mem_cgroup_inactive_file_is_low(struct mem_cgroup *memcg) return (active > inactive); } -unsigned long mem_cgroup_zone_nr_pages(struct mem_cgroup *memcg, - struct zone *zone, - enum lru_list lru) +unsigned long mem_cgroup_zone_nr_lru_pages(struct mem_cgroup *memcg, + struct zone *zone, + enum lru_list lru) { int nid = zone_to_nid(zone); int zid = zone_idx(zone); diff --git a/mm/vmscan.c b/mm/vmscan.c index 2e8fbacd874..faa0a088f9c 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -173,7 +173,7 @@ static unsigned long zone_nr_lru_pages(struct zone *zone, struct scan_control *sc, enum lru_list lru) { if (!scanning_global_lru(sc)) - return mem_cgroup_zone_nr_pages(sc->mem_cgroup, zone, lru); + return mem_cgroup_zone_nr_lru_pages(sc->mem_cgroup, zone, lru); return zone_page_state(zone, NR_LRU_BASE + lru); } -- cgit v1.2.3-70-g09d2 From 456f998ec817ebfa254464be4f089542fa390645 Mon Sep 17 00:00:00 2001 From: Ying Han Date: Thu, 26 May 2011 16:25:38 -0700 Subject: memcg: add the pagefault count into memcg stats Two new stats in per-memcg memory.stat which tracks the number of page faults and number of major page faults. "pgfault" "pgmajfault" They are different from "pgpgin"/"pgpgout" stat which count number of pages charged/discharged to the cgroup and have no meaning of reading/ writing page to disk. It is valuable to track the two stats for both measuring application's performance as well as the efficiency of the kernel page reclaim path. Counting pagefaults per process is useful, but we also need the aggregated value since processes are monitored and controlled in cgroup basis in memcg. Functional test: check the total number of pgfault/pgmajfault of all memcgs and compare with global vmstat value: $ cat /proc/vmstat | grep fault pgfault 1070751 pgmajfault 553 $ cat /dev/cgroup/memory.stat | grep fault pgfault 1071138 pgmajfault 553 total_pgfault 1071142 total_pgmajfault 553 $ cat /dev/cgroup/A/memory.stat | grep fault pgfault 199 pgmajfault 0 total_pgfault 199 total_pgmajfault 0 Performance test: run page fault test(pft) wit 16 thread on faulting in 15G anon pages in 16G container. There is no regression noticed on the "flt/cpu/s" Sample output from pft: TAG pft:anon-sys-default: Gb Thr CLine User System Wall flt/cpu/s fault/wsec 15 16 1 0.67s 233.41s 14.76s 16798.546 266356.260 +-------------------------------------------------------------------------+ N Min Max Median Avg Stddev x 10 16682.962 17344.027 16913.524 16928.812 166.5362 + 10 16695.568 16923.896 16820.604 16824.652 84.816568 No difference proven at 95.0% confidence [akpm@linux-foundation.org: fix build] [hughd@google.com: shmem fix] Signed-off-by: Ying Han Acked-by: KAMEZAWA Hiroyuki Cc: KOSAKI Motohiro Reviewed-by: Minchan Kim Cc: Daisuke Nishimura Acked-by: Balbir Singh Signed-off-by: Hugh Dickins Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/ncpfs/mmap.c | 2 ++ include/linux/memcontrol.h | 7 +++++++ mm/filemap.c | 1 + mm/memcontrol.c | 47 ++++++++++++++++++++++++++++++++++++++++++++++ mm/memory.c | 2 ++ mm/shmem.c | 11 ++++++----- 6 files changed, 65 insertions(+), 5 deletions(-) (limited to 'include/linux/memcontrol.h') diff --git a/fs/ncpfs/mmap.c b/fs/ncpfs/mmap.c index a7c07b44b10..e5d71b27a5b 100644 --- a/fs/ncpfs/mmap.c +++ b/fs/ncpfs/mmap.c @@ -16,6 +16,7 @@ #include #include #include +#include #include #include @@ -92,6 +93,7 @@ static int ncp_file_mmap_fault(struct vm_area_struct *area, * -- wli */ count_vm_event(PGMAJFAULT); + mem_cgroup_count_vm_event(area->vm_mm, PGMAJFAULT); return VM_FAULT_MAJOR; } diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index ac1e5d20916..9724a38ee69 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -20,6 +20,8 @@ #ifndef _LINUX_MEMCONTROL_H #define _LINUX_MEMCONTROL_H #include +#include + struct mem_cgroup; struct page_cgroup; struct page; @@ -149,6 +151,7 @@ unsigned long mem_cgroup_soft_limit_reclaim(struct zone *zone, int order, unsigned long *total_scanned); u64 mem_cgroup_get_limit(struct mem_cgroup *mem); +void mem_cgroup_count_vm_event(struct mm_struct *mm, enum vm_event_item idx); #ifdef CONFIG_TRANSPARENT_HUGEPAGE void mem_cgroup_split_huge_fixup(struct page *head, struct page *tail); #endif @@ -357,6 +360,10 @@ static inline void mem_cgroup_split_huge_fixup(struct page *head, { } +static inline +void mem_cgroup_count_vm_event(struct mm_struct *mm, enum vm_event_item idx) +{ +} #endif /* CONFIG_CGROUP_MEM_CONT */ #if !defined(CONFIG_CGROUP_MEM_RES_CTLR) || !defined(CONFIG_DEBUG_VM) diff --git a/mm/filemap.c b/mm/filemap.c index 7455ccd8bda..bcdc393b658 100644 --- a/mm/filemap.c +++ b/mm/filemap.c @@ -1661,6 +1661,7 @@ int filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf) /* No page in the page cache at all */ do_sync_mmap_readahead(vma, ra, file, offset); count_vm_event(PGMAJFAULT); + mem_cgroup_count_vm_event(vma->vm_mm, PGMAJFAULT); ret = VM_FAULT_MAJOR; retry_find: page = find_get_page(mapping, offset); diff --git a/mm/memcontrol.c b/mm/memcontrol.c index 4021fcd71b6..bd9052a5d3a 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -94,6 +94,8 @@ enum mem_cgroup_events_index { MEM_CGROUP_EVENTS_PGPGIN, /* # of pages paged in */ MEM_CGROUP_EVENTS_PGPGOUT, /* # of pages paged out */ MEM_CGROUP_EVENTS_COUNT, /* # of pages paged in/out */ + MEM_CGROUP_EVENTS_PGFAULT, /* # of page-faults */ + MEM_CGROUP_EVENTS_PGMAJFAULT, /* # of major page-faults */ MEM_CGROUP_EVENTS_NSTATS, }; /* @@ -590,6 +592,16 @@ static void mem_cgroup_swap_statistics(struct mem_cgroup *mem, this_cpu_add(mem->stat->count[MEM_CGROUP_STAT_SWAPOUT], val); } +void mem_cgroup_pgfault(struct mem_cgroup *mem, int val) +{ + this_cpu_add(mem->stat->events[MEM_CGROUP_EVENTS_PGFAULT], val); +} + +void mem_cgroup_pgmajfault(struct mem_cgroup *mem, int val) +{ + this_cpu_add(mem->stat->events[MEM_CGROUP_EVENTS_PGMAJFAULT], val); +} + static unsigned long mem_cgroup_read_events(struct mem_cgroup *mem, enum mem_cgroup_events_index idx) { @@ -827,6 +839,33 @@ static inline bool mem_cgroup_is_root(struct mem_cgroup *mem) return (mem == root_mem_cgroup); } +void mem_cgroup_count_vm_event(struct mm_struct *mm, enum vm_event_item idx) +{ + struct mem_cgroup *mem; + + if (!mm) + return; + + rcu_read_lock(); + mem = mem_cgroup_from_task(rcu_dereference(mm->owner)); + if (unlikely(!mem)) + goto out; + + switch (idx) { + case PGMAJFAULT: + mem_cgroup_pgmajfault(mem, 1); + break; + case PGFAULT: + mem_cgroup_pgfault(mem, 1); + break; + default: + BUG(); + } +out: + rcu_read_unlock(); +} +EXPORT_SYMBOL(mem_cgroup_count_vm_event); + /* * Following LRU functions are allowed to be used without PCG_LOCK. * Operations are called by routine of global LRU independently from memcg. @@ -3958,6 +3997,8 @@ enum { MCS_PGPGIN, MCS_PGPGOUT, MCS_SWAP, + MCS_PGFAULT, + MCS_PGMAJFAULT, MCS_INACTIVE_ANON, MCS_ACTIVE_ANON, MCS_INACTIVE_FILE, @@ -3980,6 +4021,8 @@ struct { {"pgpgin", "total_pgpgin"}, {"pgpgout", "total_pgpgout"}, {"swap", "total_swap"}, + {"pgfault", "total_pgfault"}, + {"pgmajfault", "total_pgmajfault"}, {"inactive_anon", "total_inactive_anon"}, {"active_anon", "total_active_anon"}, {"inactive_file", "total_inactive_file"}, @@ -4008,6 +4051,10 @@ mem_cgroup_get_local_stat(struct mem_cgroup *mem, struct mcs_total_stat *s) val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_SWAPOUT); s->stat[MCS_SWAP] += val * PAGE_SIZE; } + val = mem_cgroup_read_events(mem, MEM_CGROUP_EVENTS_PGFAULT); + s->stat[MCS_PGFAULT] += val; + val = mem_cgroup_read_events(mem, MEM_CGROUP_EVENTS_PGMAJFAULT); + s->stat[MCS_PGMAJFAULT] += val; /* per zone stat */ val = mem_cgroup_get_local_zonestat(mem, LRU_INACTIVE_ANON); diff --git a/mm/memory.c b/mm/memory.c index fc24f7d788b..6953d3926e0 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -2874,6 +2874,7 @@ static int do_swap_page(struct mm_struct *mm, struct vm_area_struct *vma, /* Had to read the page from swap area: Major fault */ ret = VM_FAULT_MAJOR; count_vm_event(PGMAJFAULT); + mem_cgroup_count_vm_event(mm, PGMAJFAULT); } else if (PageHWPoison(page)) { /* * hwpoisoned dirty swapcache pages are kept for killing @@ -3413,6 +3414,7 @@ int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, __set_current_state(TASK_RUNNING); count_vm_event(PGFAULT); + mem_cgroup_count_vm_event(mm, PGFAULT); /* do counter updates before entering really critical section. */ check_sync_rss_stat(current); diff --git a/mm/shmem.c b/mm/shmem.c index 69edb45a9f2..1acfb2687bf 100644 --- a/mm/shmem.c +++ b/mm/shmem.c @@ -1305,12 +1305,10 @@ repeat: swappage = lookup_swap_cache(swap); if (!swappage) { shmem_swp_unmap(entry); + spin_unlock(&info->lock); /* here we actually do the io */ - if (type && !(*type & VM_FAULT_MAJOR)) { - __count_vm_event(PGMAJFAULT); + if (type) *type |= VM_FAULT_MAJOR; - } - spin_unlock(&info->lock); swappage = shmem_swapin(swap, gfp, info, idx); if (!swappage) { spin_lock(&info->lock); @@ -1549,7 +1547,10 @@ static int shmem_fault(struct vm_area_struct *vma, struct vm_fault *vmf) error = shmem_getpage(inode, vmf->pgoff, &vmf->page, SGP_CACHE, &ret); if (error) return ((error == -ENOMEM) ? VM_FAULT_OOM : VM_FAULT_SIGBUS); - + if (ret & VM_FAULT_MAJOR) { + count_vm_event(PGMAJFAULT); + mem_cgroup_count_vm_event(vma->vm_mm, PGMAJFAULT); + } return ret | VM_FAULT_LOCKED; } -- cgit v1.2.3-70-g09d2 From a433658c30974fc87ba3ff52d7e4e6299762aa3d Mon Sep 17 00:00:00 2001 From: KOSAKI Motohiro Date: Wed, 15 Jun 2011 15:08:13 -0700 Subject: vmscan,memcg: memcg aware swap token Currently, memcg reclaim can disable swap token even if the swap token mm doesn't belong in its memory cgroup. It's slightly risky. If an admin creates very small mem-cgroup and silly guy runs contentious heavy memory pressure workload, every tasks are going to lose swap token and then system may become unresponsive. That's bad. This patch adds 'memcg' parameter into disable_swap_token(). and if the parameter doesn't match swap token, VM doesn't disable it. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: KOSAKI Motohiro Reviewed-by: KAMEZAWA Hiroyuki Reviewed-by: Rik van Riel Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/memcontrol.h | 6 ++++ include/linux/swap.h | 8 ++--- mm/memcontrol.c | 16 ++++----- mm/thrash.c | 87 ++++++++++++++++++++++++++++++++++++---------- mm/vmscan.c | 4 +-- 5 files changed, 85 insertions(+), 36 deletions(-) (limited to 'include/linux/memcontrol.h') diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index 9724a38ee69..50940da6adf 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -84,6 +84,7 @@ int task_in_mem_cgroup(struct task_struct *task, const struct mem_cgroup *mem); extern struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page); extern struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p); +extern struct mem_cgroup *try_get_mem_cgroup_from_mm(struct mm_struct *mm); static inline int mm_match_cgroup(const struct mm_struct *mm, const struct mem_cgroup *cgroup) @@ -246,6 +247,11 @@ static inline struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page) return NULL; } +static inline struct mem_cgroup *try_get_mem_cgroup_from_mm(struct mm_struct *mm) +{ + return NULL; +} + static inline int mm_match_cgroup(struct mm_struct *mm, struct mem_cgroup *mem) { return 1; diff --git a/include/linux/swap.h b/include/linux/swap.h index 384eb5fe530..e7056464703 100644 --- a/include/linux/swap.h +++ b/include/linux/swap.h @@ -358,6 +358,7 @@ struct backing_dev_info; extern struct mm_struct *swap_token_mm; extern void grab_swap_token(struct mm_struct *); extern void __put_swap_token(struct mm_struct *); +extern void disable_swap_token(struct mem_cgroup *memcg); static inline int has_swap_token(struct mm_struct *mm) { @@ -370,11 +371,6 @@ static inline void put_swap_token(struct mm_struct *mm) __put_swap_token(mm); } -static inline void disable_swap_token(void) -{ - put_swap_token(swap_token_mm); -} - #ifdef CONFIG_CGROUP_MEM_RES_CTLR extern void mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout); @@ -500,7 +496,7 @@ static inline int has_swap_token(struct mm_struct *mm) return 0; } -static inline void disable_swap_token(void) +static inline void disable_swap_token(struct mem_cgroup *memcg) { } diff --git a/mm/memcontrol.c b/mm/memcontrol.c index bd9052a5d3a..e37c44d88d4 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -735,7 +735,7 @@ struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p) struct mem_cgroup, css); } -static struct mem_cgroup *try_get_mem_cgroup_from_mm(struct mm_struct *mm) +struct mem_cgroup *try_get_mem_cgroup_from_mm(struct mm_struct *mm) { struct mem_cgroup *mem = NULL; @@ -5414,18 +5414,16 @@ static void mem_cgroup_move_task(struct cgroup_subsys *ss, struct cgroup *old_cont, struct task_struct *p) { - struct mm_struct *mm; + struct mm_struct *mm = get_task_mm(p); - if (!mc.to) - /* no need to move charge */ - return; - - mm = get_task_mm(p); if (mm) { - mem_cgroup_move_charge(mm); + if (mc.to) + mem_cgroup_move_charge(mm); + put_swap_token(mm); mmput(mm); } - mem_cgroup_clear_mc(); + if (mc.to) + mem_cgroup_clear_mc(); } #else /* !CONFIG_MMU */ static int mem_cgroup_can_attach(struct cgroup_subsys *ss, diff --git a/mm/thrash.c b/mm/thrash.c index 2372d4ed5dd..6cdf8651138 100644 --- a/mm/thrash.c +++ b/mm/thrash.c @@ -21,11 +21,31 @@ #include #include #include +#include static DEFINE_SPINLOCK(swap_token_lock); struct mm_struct *swap_token_mm; +struct mem_cgroup *swap_token_memcg; static unsigned int global_faults; +#ifdef CONFIG_CGROUP_MEM_RES_CTLR +static struct mem_cgroup *swap_token_memcg_from_mm(struct mm_struct *mm) +{ + struct mem_cgroup *memcg; + + memcg = try_get_mem_cgroup_from_mm(mm); + if (memcg) + css_put(mem_cgroup_css(memcg)); + + return memcg; +} +#else +static struct mem_cgroup *swap_token_memcg_from_mm(struct mm_struct *mm) +{ + return NULL; +} +#endif + void grab_swap_token(struct mm_struct *mm) { int current_interval; @@ -38,40 +58,69 @@ void grab_swap_token(struct mm_struct *mm) return; /* First come first served */ - if (swap_token_mm == NULL) { - mm->token_priority = mm->token_priority + 2; - swap_token_mm = mm; + if (!swap_token_mm) + goto replace_token; + + if (mm == swap_token_mm) { + mm->token_priority += 2; goto out; } - if (mm != swap_token_mm) { - if (current_interval < mm->last_interval) - mm->token_priority++; - else { - if (likely(mm->token_priority > 0)) - mm->token_priority--; - } - /* Check if we deserve the token */ - if (mm->token_priority > swap_token_mm->token_priority) { - mm->token_priority += 2; - swap_token_mm = mm; - } - } else { - /* Token holder came in again! */ - mm->token_priority += 2; + if (current_interval < mm->last_interval) + mm->token_priority++; + else { + if (likely(mm->token_priority > 0)) + mm->token_priority--; } + /* Check if we deserve the token */ + if (mm->token_priority > swap_token_mm->token_priority) + goto replace_token; + out: mm->faultstamp = global_faults; mm->last_interval = current_interval; spin_unlock(&swap_token_lock); + return; + +replace_token: + mm->token_priority += 2; + swap_token_mm = mm; + swap_token_memcg = swap_token_memcg_from_mm(mm); + goto out; } /* Called on process exit. */ void __put_swap_token(struct mm_struct *mm) { spin_lock(&swap_token_lock); - if (likely(mm == swap_token_mm)) + if (likely(mm == swap_token_mm)) { swap_token_mm = NULL; + swap_token_memcg = NULL; + } spin_unlock(&swap_token_lock); } + +static bool match_memcg(struct mem_cgroup *a, struct mem_cgroup *b) +{ + if (!a) + return true; + if (!b) + return true; + if (a == b) + return true; + return false; +} + +void disable_swap_token(struct mem_cgroup *memcg) +{ + /* memcg reclaim don't disable unrelated mm token. */ + if (match_memcg(memcg, swap_token_memcg)) { + spin_lock(&swap_token_lock); + if (match_memcg(memcg, swap_token_memcg)) { + swap_token_mm = NULL; + swap_token_memcg = NULL; + } + spin_unlock(&swap_token_lock); + } +} diff --git a/mm/vmscan.c b/mm/vmscan.c index faa0a088f9c..dbe6ea321df 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -2081,7 +2081,7 @@ static unsigned long do_try_to_free_pages(struct zonelist *zonelist, for (priority = DEF_PRIORITY; priority >= 0; priority--) { sc->nr_scanned = 0; if (!priority) - disable_swap_token(); + disable_swap_token(sc->mem_cgroup); total_scanned += shrink_zones(priority, zonelist, sc); /* * Don't shrink slabs when reclaiming memory from @@ -2407,7 +2407,7 @@ loop_again: /* The swap token gets in the way of swapout... */ if (!priority) - disable_swap_token(); + disable_swap_token(NULL); all_zones_ok = 1; balanced = 0; -- cgit v1.2.3-70-g09d2