From dddac6a7b445de95515f64fdf82fe5dc36c02f26 Mon Sep 17 00:00:00 2001 From: Alan Jenkins Date: Wed, 29 Jul 2009 21:07:55 +0200 Subject: PM / Hibernate: Replace bdget call with simple atomic_inc of i_count Create bdgrab(). This function copies an existing reference to a block_device. It is safe to call from any context. Hibernation code wishes to copy a reference to the active swap device. Right now it calls bdget() under a spinlock, but this is wrong because bdget() can sleep. It doesn't need a full bdget() because we already hold a reference to active swap devices (and the spinlock protects against swapoff). Fixes http://bugzilla.kernel.org/show_bug.cgi?id=13827 Signed-off-by: Alan Jenkins Signed-off-by: Rafael J. Wysocki --- mm/swapfile.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'mm') diff --git a/mm/swapfile.c b/mm/swapfile.c index d1ade1a48ee..8ffdc0d23c5 100644 --- a/mm/swapfile.c +++ b/mm/swapfile.c @@ -753,7 +753,7 @@ int swap_type_of(dev_t device, sector_t offset, struct block_device **bdev_p) if (!bdev) { if (bdev_p) - *bdev_p = bdget(sis->bdev->bd_dev); + *bdev_p = bdgrab(sis->bdev); spin_unlock(&swap_lock); return i; @@ -765,7 +765,7 @@ int swap_type_of(dev_t device, sector_t offset, struct block_device **bdev_p) struct swap_extent, list); if (se->start_block == offset) { if (bdev_p) - *bdev_p = bdget(sis->bdev->bd_dev); + *bdev_p = bdgrab(sis->bdev); spin_unlock(&swap_lock); bdput(bdev); -- cgit v1.2.3 From f5886c7f96f2542382d3a983c5f13e03d7fc5259 Mon Sep 17 00:00:00 2001 From: Catalin Marinas Date: Wed, 29 Jul 2009 16:26:57 +0100 Subject: kmemleak: Protect the seq start/next/stop sequence by rcu_read_lock() Objects passed to kmemleak_seq_next() have an incremented reference count (hence not freed) but they may point via object_list.next to other freed objects. To avoid this, the whole start/next/stop sequence must be protected by rcu_read_lock(). Signed-off-by: Catalin Marinas Signed-off-by: Linus Torvalds --- mm/kmemleak.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'mm') diff --git a/mm/kmemleak.c b/mm/kmemleak.c index 5aabd41ffb8..487267310a8 100644 --- a/mm/kmemleak.c +++ b/mm/kmemleak.c @@ -1217,7 +1217,6 @@ static void *kmemleak_seq_start(struct seq_file *seq, loff_t *pos) } object = NULL; out: - rcu_read_unlock(); return object; } @@ -1233,13 +1232,11 @@ static void *kmemleak_seq_next(struct seq_file *seq, void *v, loff_t *pos) ++(*pos); - rcu_read_lock(); list_for_each_continue_rcu(n, &object_list) { next_obj = list_entry(n, struct kmemleak_object, object_list); if (get_object(next_obj)) break; } - rcu_read_unlock(); put_object(prev_obj); return next_obj; @@ -1255,6 +1252,7 @@ static void kmemleak_seq_stop(struct seq_file *seq, void *v) * kmemleak_seq_start may return ERR_PTR if the scan_mutex * waiting was interrupted, so only release it if !IS_ERR. */ + rcu_read_unlock(); mutex_unlock(&scan_mutex); if (v) put_object(v); -- cgit v1.2.3 From e084b2d95e48b31aa45f9c49ffc6cdae8bdb21d4 Mon Sep 17 00:00:00 2001 From: Mel Gorman Date: Wed, 29 Jul 2009 15:02:04 -0700 Subject: page-allocator: preserve PFN ordering when __GFP_COLD is set Fix a post-2.6.24 performace regression caused by 3dfa5721f12c3d5a441448086bee156887daa961 ("page-allocator: preserve PFN ordering when __GFP_COLD is set"). Narayanan reports "The regression is around 15%. There is no disk controller as our setup is based on Samsung OneNAND used as a memory mapped device on a OMAP2430 based board." The page allocator tries to preserve contiguous PFN ordering when returning pages such that repeated callers to the allocator have a strong chance of getting physically contiguous pages, particularly when external fragmentation is low. However, of the bulk of the allocations have __GFP_COLD set as they are due to aio_read() for example, then the PFNs are in reverse PFN order. This can cause performance degration when used with IO controllers that could have merged the requests. This patch attempts to preserve the contiguous ordering of PFNs for users of __GFP_COLD. Signed-off-by: Mel Gorman Reported-by: Narayananu Gopalakrishnan Tested-by: Narayanan Gopalakrishnan Cc: KAMEZAWA Hiroyuki Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/page_alloc.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) (limited to 'mm') diff --git a/mm/page_alloc.c b/mm/page_alloc.c index caa92689aac..ae28c22a7fd 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -882,7 +882,7 @@ retry_reserve: */ static int rmqueue_bulk(struct zone *zone, unsigned int order, unsigned long count, struct list_head *list, - int migratetype) + int migratetype, int cold) { int i; @@ -901,7 +901,10 @@ static int rmqueue_bulk(struct zone *zone, unsigned int order, * merge IO requests if the physical pages are ordered * properly. */ - list_add(&page->lru, list); + if (likely(cold == 0)) + list_add(&page->lru, list); + else + list_add_tail(&page->lru, list); set_page_private(page, migratetype); list = &page->lru; } @@ -1119,7 +1122,8 @@ again: local_irq_save(flags); if (!pcp->count) { pcp->count = rmqueue_bulk(zone, 0, - pcp->batch, &pcp->list, migratetype); + pcp->batch, &pcp->list, + migratetype, cold); if (unlikely(!pcp->count)) goto failed; } @@ -1138,7 +1142,8 @@ again: /* Allocate more to the pcp list if necessary */ if (unlikely(&page->lru == &pcp->list)) { pcp->count += rmqueue_bulk(zone, 0, - pcp->batch, &pcp->list, migratetype); + pcp->batch, &pcp->list, + migratetype, cold); page = list_entry(pcp->list.next, struct page, lru); } -- cgit v1.2.3 From 6583bb64fc370842b32a87c67750c26f6d559af0 Mon Sep 17 00:00:00 2001 From: David Rientjes Date: Wed, 29 Jul 2009 15:02:06 -0700 Subject: mm: avoid endless looping for oom killed tasks If a task is oom killed and still cannot find memory when trying with no watermarks, it's better to fail the allocation attempt than to loop endlessly. Direct reclaim has already failed and the oom killer will be a no-op since current has yet to die, so there is no other alternative for allocations that are not __GFP_NOFAIL. Acked-by: Mel Gorman Signed-off-by: David Rientjes Acked-by: Hugh Dickins Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/page_alloc.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'mm') diff --git a/mm/page_alloc.c b/mm/page_alloc.c index ae28c22a7fd..2dbb2fc6883 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -1794,6 +1794,10 @@ rebalance: if (p->flags & PF_MEMALLOC) goto nopage; + /* Avoid allocations with no watermarks from looping endlessly */ + if (test_thread_flag(TIF_MEMDIE) && !(gfp_mask & __GFP_NOFAIL)) + goto nopage; + /* Try direct reclaim and then allocating */ page = __alloc_pages_direct_reclaim(gfp_mask, order, zonelist, high_zoneidx, -- cgit v1.2.3 From e4c6f8bed01f9f9a5c607bd689bf67e7b8a36bd8 Mon Sep 17 00:00:00 2001 From: Eric Sandeen Date: Wed, 29 Jul 2009 15:02:16 -0700 Subject: hugetlbfs: fix i_blocks accounting As reported in Red Hat bz #509671, i_blocks for files on hugetlbfs get accounting wrong when doing something like: $ > foo $ date > foo date: write error: Invalid argument $ /usr/bin/stat foo File: `foo' Size: 0 Blocks: 18446744073709547520 IO Block: 2097152 regular ... This is because hugetlb_unreserve_pages() is unconditionally removing blocks_per_huge_page(h) on each call rather than using the freed amount. If there were 0 blocks, it goes negative, resulting in the above. This is a regression from commit a5516438959d90b071ff0a484ce4f3f523dc3152 ("hugetlb: modular state for hugetlb page size") which did: - inode->i_blocks -= BLOCKS_PER_HUGEPAGE * freed; + inode->i_blocks -= blocks_per_huge_page(h); so just put back the freed multiplier, and it's all happy again. Signed-off-by: Eric Sandeen Acked-by: Andi Kleen Cc: William Lee Irwin III Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/hugetlb.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'mm') diff --git a/mm/hugetlb.c b/mm/hugetlb.c index d0351e31f47..cafdcee154e 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -2370,7 +2370,7 @@ void hugetlb_unreserve_pages(struct inode *inode, long offset, long freed) long chg = region_truncate(&inode->i_mapping->private_list, offset); spin_lock(&inode->i_lock); - inode->i_blocks -= blocks_per_huge_page(h); + inode->i_blocks -= (blocks_per_huge_page(h) * freed); spin_unlock(&inode->i_lock); hugetlb_put_quota(inode->i_mapping, (chg - freed)); -- cgit v1.2.3 From 887032670d47366a8c8f25396ea7c14b7b2cc620 Mon Sep 17 00:00:00 2001 From: KAMEZAWA Hiroyuki Date: Wed, 29 Jul 2009 15:04:06 -0700 Subject: cgroup avoid permanent sleep at rmdir After commit ec64f51545fffbc4cb968f0cea56341a4b07e85a ("cgroup: fix frequent -EBUSY at rmdir"), cgroup's rmdir (especially against memcg) doesn't return -EBUSY by temporary ref counts. That commit expects all refs after pre_destroy() is temporary but...it wasn't. Then, rmdir can wait permanently. This patch tries to fix that and change followings. - set CGRP_WAIT_ON_RMDIR flag before pre_destroy(). - clear CGRP_WAIT_ON_RMDIR flag when the subsys finds racy case. if there are sleeping ones, wakes them up. - rmdir() sleeps only when CGRP_WAIT_ON_RMDIR flag is set. Tested-by: Daisuke Nishimura Reported-by: Daisuke Nishimura Reviewed-by: Paul Menage Acked-by: Balbir Sigh Signed-off-by: KAMEZAWA Hiroyuki Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/memcontrol.c | 23 ++++++++++++++++++++--- 1 file changed, 20 insertions(+), 3 deletions(-) (limited to 'mm') diff --git a/mm/memcontrol.c b/mm/memcontrol.c index e717964cb5a..fd4529d86de 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -1207,6 +1207,12 @@ static int mem_cgroup_move_account(struct page_cgroup *pc, ret = 0; out: unlock_page_cgroup(pc); + /* + * We charges against "to" which may not have any tasks. Then, "to" + * can be under rmdir(). But in current implementation, caller of + * this function is just force_empty() and it's garanteed that + * "to" is never removed. So, we don't check rmdir status here. + */ return ret; } @@ -1428,6 +1434,7 @@ __mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr, return; if (!ptr) return; + cgroup_exclude_rmdir(&ptr->css); pc = lookup_page_cgroup(page); mem_cgroup_lru_del_before_commit_swapcache(page); __mem_cgroup_commit_charge(ptr, pc, ctype); @@ -1457,8 +1464,12 @@ __mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr, } rcu_read_unlock(); } - /* add this page(page_cgroup) to the LRU we want. */ - + /* + * At swapin, we may charge account against cgroup which has no tasks. + * So, rmdir()->pre_destroy() can be called while we do this charge. + * In that case, we need to call pre_destroy() again. check it here. + */ + cgroup_release_and_wakeup_rmdir(&ptr->css); } void mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr) @@ -1664,7 +1675,7 @@ void mem_cgroup_end_migration(struct mem_cgroup *mem, if (!mem) return; - + cgroup_exclude_rmdir(&mem->css); /* at migration success, oldpage->mapping is NULL. */ if (oldpage->mapping) { target = oldpage; @@ -1704,6 +1715,12 @@ void mem_cgroup_end_migration(struct mem_cgroup *mem, */ if (ctype == MEM_CGROUP_CHARGE_TYPE_MAPPED) mem_cgroup_uncharge_page(target); + /* + * At migration, we may charge account against cgroup which has no tasks + * So, rmdir()->pre_destroy() can be called while we do this charge. + * In that case, we need to call pre_destroy() again. check it here. + */ + cgroup_release_and_wakeup_rmdir(&mem->css); } /* -- cgit v1.2.3 From 1fc28b70fe2dbf87e061b6ce5091a1f8e4e5d4e7 Mon Sep 17 00:00:00 2001 From: Mel Gorman Date: Wed, 29 Jul 2009 15:04:08 -0700 Subject: page-allocator: allow too high-order warning messages to be suppressed with __GFP_NOWARN The page allocator warns once when an order >= MAX_ORDER is specified. This is to catch callers of the allocator that are always falling back to their worst-case when it was not expected. However, there are cases where the caller is behaving correctly but cannot suppress the warning. This patch allows the warning to be suppressed by the callers by specifying __GFP_NOWARN. Signed-off-by: Mel Gorman Acked-by: David Rientjes Cc: Arnaldo Carvalho de Melo Cc: "David S. Miller" Cc: "Rafael J. Wysocki" Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/page_alloc.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'mm') diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 2dbb2fc6883..d052abbe306 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -1745,8 +1745,10 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order, * be using allocators in order of preference for an area that is * too large. */ - if (WARN_ON_ONCE(order >= MAX_ORDER)) + if (order >= MAX_ORDER) { + WARN_ON_ONCE(!(gfp_mask & __GFP_NOWARN)); return NULL; + } /* * GFP_THISNODE (meaning __GFP_THISNODE, __GFP_NORETRY and -- cgit v1.2.3