From 61609d01cbb3ab865c8cccaf85e6837c47096480 Mon Sep 17 00:00:00 2001 From: Yuri Tikhonov Date: Mon, 13 Apr 2009 14:40:11 -0700 Subject: shmem: fix division by zero Fix a division by zero which we have in shmem_truncate_range() and shmem_unuse_inode() when using big PAGE_SIZE values (e.g. 256kB on ppc44x). With 256kB PAGE_SIZE, the ENTRIES_PER_PAGEPAGE constant becomes too large (0x1.0000.0000) on a 32-bit kernel, so this patch just changes its type from 'unsigned long' to 'unsigned long long'. Hugh: reverted its unsigned long longs in shmem_truncate_range() and shmem_getpage(): the pagecache index cannot be more than an unsigned long, so the divisions by zero occurred in unreached code. It's a pity we need any ULL arithmetic here, but I found no pretty way to avoid it. Signed-off-by: Yuri Tikhonov Signed-off-by: Hugh Dickins Cc: Paul Mackerras Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/shmem.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'mm/shmem.c') diff --git a/mm/shmem.c b/mm/shmem.c index d94d2e9146bc..28024ce8b8fd 100644 --- a/mm/shmem.c +++ b/mm/shmem.c @@ -66,7 +66,7 @@ static struct vfsmount *shm_mnt; #include #define ENTRIES_PER_PAGE (PAGE_CACHE_SIZE/sizeof(unsigned long)) -#define ENTRIES_PER_PAGEPAGE (ENTRIES_PER_PAGE*ENTRIES_PER_PAGE) +#define ENTRIES_PER_PAGEPAGE ((unsigned long long)ENTRIES_PER_PAGE*ENTRIES_PER_PAGE) #define BLOCKS_PER_PAGE (PAGE_CACHE_SIZE/512) #define SHMEM_MAX_INDEX (SHMEM_NR_DIRECT + (ENTRIES_PER_PAGEPAGE/2) * (ENTRIES_PER_PAGE+1)) -- cgit v1.2.3 From caefba1740d8016e6dfe8fda84f85bdcb8f8c85d Mon Sep 17 00:00:00 2001 From: Hugh Dickins Date: Mon, 13 Apr 2009 14:40:12 -0700 Subject: shmem: respect MAX_LFS_FILESIZE SHMEM_MAX_BYTES was derived from the maximum size of its triple-indirect swap vector, forgetting to take the MAX_LFS_FILESIZE limit into account. Never mind 256kB pages, even 8kB pages on 32-bit kernels allowed files to grow slightly bigger than that supposed maximum. Fix this by using the min of both (at build time not run time). And it happens that this calculation is good as far as 8MB pages on 32-bit or 16MB pages on 64-bit: though SHMSWP_MAX_INDEX gets truncated before that, it's truncated to such large numbers that we don't need to care. [akpm@linux-foundation.org: it needs pagemap.h] [akpm@linux-foundation.org: fix sparc64 min() warnings] Signed-off-by: Hugh Dickins Cc: Yuri Tikhonov Cc: Paul Mackerras Cc: Randy Dunlap Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/shmem.c | 25 ++++++++++++++++++++----- 1 file changed, 20 insertions(+), 5 deletions(-) (limited to 'mm/shmem.c') diff --git a/mm/shmem.c b/mm/shmem.c index 28024ce8b8fd..f9cb20ebb990 100644 --- a/mm/shmem.c +++ b/mm/shmem.c @@ -24,6 +24,7 @@ #include #include #include +#include #include #include #include @@ -43,7 +44,6 @@ static struct vfsmount *shm_mnt; #include #include #include -#include #include #include #include @@ -65,13 +65,28 @@ static struct vfsmount *shm_mnt; #include #include +/* + * The maximum size of a shmem/tmpfs file is limited by the maximum size of + * its triple-indirect swap vector - see illustration at shmem_swp_entry(). + * + * With 4kB page size, maximum file size is just over 2TB on a 32-bit kernel, + * but one eighth of that on a 64-bit kernel. With 8kB page size, maximum + * file size is just over 4TB on a 64-bit kernel, but 16TB on a 32-bit kernel, + * MAX_LFS_FILESIZE being then more restrictive than swap vector layout. + * + * We use / and * instead of shifts in the definitions below, so that the swap + * vector can be tested with small even values (e.g. 20) for ENTRIES_PER_PAGE. + */ #define ENTRIES_PER_PAGE (PAGE_CACHE_SIZE/sizeof(unsigned long)) #define ENTRIES_PER_PAGEPAGE ((unsigned long long)ENTRIES_PER_PAGE*ENTRIES_PER_PAGE) -#define BLOCKS_PER_PAGE (PAGE_CACHE_SIZE/512) -#define SHMEM_MAX_INDEX (SHMEM_NR_DIRECT + (ENTRIES_PER_PAGEPAGE/2) * (ENTRIES_PER_PAGE+1)) -#define SHMEM_MAX_BYTES ((unsigned long long)SHMEM_MAX_INDEX << PAGE_CACHE_SHIFT) +#define SHMSWP_MAX_INDEX (SHMEM_NR_DIRECT + (ENTRIES_PER_PAGEPAGE/2) * (ENTRIES_PER_PAGE+1)) +#define SHMSWP_MAX_BYTES (SHMSWP_MAX_INDEX << PAGE_CACHE_SHIFT) +#define SHMEM_MAX_BYTES min_t(unsigned long long, SHMSWP_MAX_BYTES, MAX_LFS_FILESIZE) +#define SHMEM_MAX_INDEX ((unsigned long)((SHMEM_MAX_BYTES+1) >> PAGE_CACHE_SHIFT)) + +#define BLOCKS_PER_PAGE (PAGE_CACHE_SIZE/512) #define VM_ACCT(size) (PAGE_CACHE_ALIGN(size) >> PAGE_SHIFT) /* info->flags needs VM_flags to handle pagein/truncate races efficiently */ @@ -2581,7 +2596,7 @@ int shmem_unuse(swp_entry_t entry, struct page *page) #define shmem_get_inode(sb, mode, dev, flags) ramfs_get_inode(sb, mode, dev) #define shmem_acct_size(flags, size) 0 #define shmem_unacct_size(flags, size) do {} while (0) -#define SHMEM_MAX_BYTES LLONG_MAX +#define SHMEM_MAX_BYTES MAX_LFS_FILESIZE #endif /* CONFIG_SHMEM */ -- cgit v1.2.3 From ae3abae64f177586be55b04a7fb7047a34b21a3e Mon Sep 17 00:00:00 2001 From: Daisuke Nishimura Date: Thu, 30 Apr 2009 15:08:19 -0700 Subject: memcg: fix mem_cgroup_shrink_usage() Current mem_cgroup_shrink_usage() has two problems. 1. It doesn't call mem_cgroup_out_of_memory and doesn't update last_oom_jiffies, so pagefault_out_of_memory invokes global OOM. 2. Considering hierarchy, shrinking has to be done from the mem_over_limit, not from the memcg which the page would be charged to. mem_cgroup_try_charge_swapin() does all of these things properly, so we use it and call cancel_charge_swapin when it succeeded. The name of "shrink_usage" is not appropriate for this behavior, so we change it too. Signed-off-by: Daisuke Nishimura Acked-by: KAMEZAWA Hiroyuki Cc: Li Zefan Cc: Paul Menage Cc: Dhaval Giani Cc: Daisuke Nishimura Cc: YAMAMOTO Takashi Cc: KOSAKI Motohiro Cc: David Rientjes Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/memcontrol.h | 4 ++-- mm/memcontrol.c | 33 ++++++++++++--------------------- mm/shmem.c | 8 ++++++-- 3 files changed, 20 insertions(+), 25 deletions(-) (limited to 'mm/shmem.c') diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index a9e3b76aa884..25b9ca93d232 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -56,7 +56,7 @@ extern void mem_cgroup_move_lists(struct page *page, enum lru_list from, enum lru_list to); extern void mem_cgroup_uncharge_page(struct page *page); extern void mem_cgroup_uncharge_cache_page(struct page *page); -extern int mem_cgroup_shrink_usage(struct page *page, +extern int mem_cgroup_shmem_charge_fallback(struct page *page, struct mm_struct *mm, gfp_t gfp_mask); extern unsigned long mem_cgroup_isolate_pages(unsigned long nr_to_scan, @@ -155,7 +155,7 @@ static inline void mem_cgroup_uncharge_cache_page(struct page *page) { } -static inline int mem_cgroup_shrink_usage(struct page *page, +static inline int mem_cgroup_shmem_charge_fallback(struct page *page, struct mm_struct *mm, gfp_t gfp_mask) { return 0; diff --git a/mm/memcontrol.c b/mm/memcontrol.c index 575203ae2109..01c2d8f14685 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -1617,37 +1617,28 @@ void mem_cgroup_end_migration(struct mem_cgroup *mem, } /* - * A call to try to shrink memory usage under specified resource controller. - * This is typically used for page reclaiming for shmem for reducing side - * effect of page allocation from shmem, which is used by some mem_cgroup. + * A call to try to shrink memory usage on charge failure at shmem's swapin. + * Calling hierarchical_reclaim is not enough because we should update + * last_oom_jiffies to prevent pagefault_out_of_memory from invoking global OOM. + * Moreover considering hierarchy, we should reclaim from the mem_over_limit, + * not from the memcg which this page would be charged to. + * try_charge_swapin does all of these works properly. */ -int mem_cgroup_shrink_usage(struct page *page, +int mem_cgroup_shmem_charge_fallback(struct page *page, struct mm_struct *mm, gfp_t gfp_mask) { struct mem_cgroup *mem = NULL; - int progress = 0; - int retry = MEM_CGROUP_RECLAIM_RETRIES; + int ret; if (mem_cgroup_disabled()) return 0; - if (page) - mem = try_get_mem_cgroup_from_swapcache(page); - if (!mem && mm) - mem = try_get_mem_cgroup_from_mm(mm); - if (unlikely(!mem)) - return 0; - do { - progress = mem_cgroup_hierarchical_reclaim(mem, - gfp_mask, true, false); - progress += mem_cgroup_check_under_limit(mem); - } while (!progress && --retry); + ret = mem_cgroup_try_charge_swapin(mm, page, gfp_mask, &mem); + if (!ret) + mem_cgroup_cancel_charge_swapin(mem); /* it does !mem check */ - css_put(&mem->css); - if (!retry) - return -ENOMEM; - return 0; + return ret; } static DEFINE_MUTEX(set_limit_mutex); diff --git a/mm/shmem.c b/mm/shmem.c index f9cb20ebb990..b25f95ce3db7 100644 --- a/mm/shmem.c +++ b/mm/shmem.c @@ -1340,8 +1340,12 @@ repeat: shmem_swp_unmap(entry); spin_unlock(&info->lock); if (error == -ENOMEM) { - /* allow reclaim from this memory cgroup */ - error = mem_cgroup_shrink_usage(swappage, + /* + * reclaim from proper memory cgroup and + * call memcg's OOM if needed. + */ + error = mem_cgroup_shmem_charge_fallback( + swappage, current->mm, gfp); if (error) { -- cgit v1.2.3