From bc43f75cd9815833b27831600ccade672edb5e43 Mon Sep 17 00:00:00 2001 From: Johannes Weiner Date: Thu, 30 Apr 2009 15:08:08 -0700 Subject: mm: fix pageref leak in do_swap_page() By the time the memory cgroup code is notified about a swapin we already hold a reference on the fault page. If the cgroup callback fails make sure to unlock AND release the page reference which was taken by lookup_swap_cach(), or we leak the reference. Signed-off-by: Johannes Weiner Cc: Balbir Singh Reviewed-by: Minchan Kim Acked-by: KAMEZAWA Hiroyuki Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/memory.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'mm') diff --git a/mm/memory.c b/mm/memory.c index cf6873e91c6..6a4ef0fd071 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -2458,8 +2458,7 @@ static int do_swap_page(struct mm_struct *mm, struct vm_area_struct *vma, if (mem_cgroup_try_charge_swapin(mm, page, GFP_KERNEL, &ptr)) { ret = VM_FAULT_OOM; - unlock_page(page); - goto out; + goto out_page; } /* @@ -2521,6 +2520,7 @@ out: out_nomap: mem_cgroup_cancel_charge_swapin(ptr); pte_unmap_unlock(page_table, ptl); +out_page: unlock_page(page); page_cache_release(page); return ret; -- cgit v1.2.3-18-g5258 From c0bd3f63ce01a1757dbce6373122a05fbf99ced7 Mon Sep 17 00:00:00 2001 From: Daisuke Nishimura Date: Thu, 30 Apr 2009 15:08:11 -0700 Subject: memcg: fix try_get_mem_cgroup_from_swapcache() This is a bugfix for commit 3c776e64660028236313f0e54f3a9945764422df ("memcg: charge swapcache to proper memcg"). Used bit of swapcache is solid under page lock, but considering move_account, pc->mem_cgroup is not. We need lock_page_cgroup() anyway. Signed-off-by: Daisuke Nishimura Acked-by: KAMEZAWA Hiroyuki Cc: Balbir Singh Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/memcontrol.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'mm') diff --git a/mm/memcontrol.c b/mm/memcontrol.c index e44fb0fbb80..575203ae210 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -1024,9 +1024,7 @@ static struct mem_cgroup *try_get_mem_cgroup_from_swapcache(struct page *page) return NULL; pc = lookup_page_cgroup(page); - /* - * Used bit of swapcache is solid under page lock. - */ + lock_page_cgroup(pc); if (PageCgroupUsed(pc)) { mem = pc->mem_cgroup; if (mem && !css_tryget(&mem->css)) @@ -1040,6 +1038,7 @@ static struct mem_cgroup *try_get_mem_cgroup_from_swapcache(struct page *page) mem = NULL; rcu_read_unlock(); } + unlock_page_cgroup(pc); return mem; } -- cgit v1.2.3-18-g5258 From b827e496c893de0c0f142abfaeb8730a2fd6b37f Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Thu, 30 Apr 2009 15:08:16 -0700 Subject: mm: close page_mkwrite races Change page_mkwrite to allow implementations to return with the page locked, and also change it's callers (in page fault paths) to hold the lock until the page is marked dirty. This allows the filesystem to have full control of page dirtying events coming from the VM. Rather than simply hold the page locked over the page_mkwrite call, we call page_mkwrite with the page unlocked and allow callers to return with it locked, so filesystems can avoid LOR conditions with page lock. The problem with the current scheme is this: a filesystem that wants to associate some metadata with a page as long as the page is dirty, will perform this manipulation in its ->page_mkwrite. It currently then must return with the page unlocked and may not hold any other locks (according to existing page_mkwrite convention). In this window, the VM could write out the page, clearing page-dirty. The filesystem has no good way to detect that a dirty pte is about to be attached, so it will happily write out the page, at which point, the filesystem may manipulate the metadata to reflect that the page is no longer dirty. It is not always possible to perform the required metadata manipulation in ->set_page_dirty, because that function cannot block or fail. The filesystem may need to allocate some data structure, for example. And the VM cannot mark the pte dirty before page_mkwrite, because page_mkwrite is allowed to fail, so we must not allow any window where the page could be written to if page_mkwrite does fail. This solution of holding the page locked over the 3 critical operations (page_mkwrite, setting the pte dirty, and finally setting the page dirty) closes out races nicely, preventing page cleaning for writeout being initiated in that window. This provides the filesystem with a strong synchronisation against the VM here. - Sage needs this race closed for ceph filesystem. - Trond for NFS (http://bugzilla.kernel.org/show_bug.cgi?id=12913). - I need it for fsblock. - I suspect other filesystems may need it too (eg. btrfs). - I have converted buffer.c to the new locking. Even simple block allocation under dirty pages might be susceptible to i_size changing under partial page at the end of file (we also have a buffer.c-side problem here, but it cannot be fixed properly without this patch). - Other filesystems (eg. NFS, maybe btrfs) will need to change their page_mkwrite functions themselves. [ This also moves page_mkwrite another step closer to fault, which should eventually allow page_mkwrite to be moved into ->fault, and thus avoiding a filesystem calldown and page lock/unlock cycle in __do_fault. ] [akpm@linux-foundation.org: fix derefs of NULL ->mapping] Cc: Sage Weil Cc: Trond Myklebust Signed-off-by: Nick Piggin Cc: Valdis Kletnieks Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/memory.c | 108 ++++++++++++++++++++++++++++++++++++++++++------------------ 1 file changed, 76 insertions(+), 32 deletions(-) (limited to 'mm') diff --git a/mm/memory.c b/mm/memory.c index 6a4ef0fd071..4126dd16778 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -1971,6 +1971,15 @@ static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma, ret = tmp; goto unwritable_page; } + if (unlikely(!(tmp & VM_FAULT_LOCKED))) { + lock_page(old_page); + if (!old_page->mapping) { + ret = 0; /* retry the fault */ + unlock_page(old_page); + goto unwritable_page; + } + } else + VM_BUG_ON(!PageLocked(old_page)); /* * Since we dropped the lock we need to revalidate @@ -1980,9 +1989,11 @@ static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma, */ page_table = pte_offset_map_lock(mm, pmd, address, &ptl); - page_cache_release(old_page); - if (!pte_same(*page_table, orig_pte)) + if (!pte_same(*page_table, orig_pte)) { + unlock_page(old_page); + page_cache_release(old_page); goto unlock; + } page_mkwrite = 1; } @@ -2094,9 +2105,6 @@ gotten: unlock: pte_unmap_unlock(page_table, ptl); if (dirty_page) { - if (vma->vm_file) - file_update_time(vma->vm_file); - /* * Yes, Virginia, this is actually required to prevent a race * with clear_page_dirty_for_io() from clearing the page dirty @@ -2105,16 +2113,41 @@ unlock: * * do_no_page is protected similarly. */ - wait_on_page_locked(dirty_page); - set_page_dirty_balance(dirty_page, page_mkwrite); + if (!page_mkwrite) { + wait_on_page_locked(dirty_page); + set_page_dirty_balance(dirty_page, page_mkwrite); + } put_page(dirty_page); + if (page_mkwrite) { + struct address_space *mapping = dirty_page->mapping; + + set_page_dirty(dirty_page); + unlock_page(dirty_page); + page_cache_release(dirty_page); + if (mapping) { + /* + * Some device drivers do not set page.mapping + * but still dirty their pages + */ + balance_dirty_pages_ratelimited(mapping); + } + } + + /* file_update_time outside page_lock */ + if (vma->vm_file) + file_update_time(vma->vm_file); } return ret; oom_free_new: page_cache_release(new_page); oom: - if (old_page) + if (old_page) { + if (page_mkwrite) { + unlock_page(old_page); + page_cache_release(old_page); + } page_cache_release(old_page); + } return VM_FAULT_OOM; unwritable_page: @@ -2664,27 +2697,22 @@ static int __do_fault(struct mm_struct *mm, struct vm_area_struct *vma, int tmp; unlock_page(page); - vmf.flags |= FAULT_FLAG_MKWRITE; + vmf.flags = FAULT_FLAG_WRITE|FAULT_FLAG_MKWRITE; tmp = vma->vm_ops->page_mkwrite(vma, &vmf); if (unlikely(tmp & (VM_FAULT_ERROR | VM_FAULT_NOPAGE))) { ret = tmp; - anon = 1; /* no anon but release vmf.page */ - goto out_unlocked; - } - lock_page(page); - /* - * XXX: this is not quite right (racy vs - * invalidate) to unlock and relock the page - * like this, however a better fix requires - * reworking page_mkwrite locking API, which - * is better done later. - */ - if (!page->mapping) { - ret = 0; - anon = 1; /* no anon but release vmf.page */ - goto out; + goto unwritable_page; } + if (unlikely(!(tmp & VM_FAULT_LOCKED))) { + lock_page(page); + if (!page->mapping) { + ret = 0; /* retry the fault */ + unlock_page(page); + goto unwritable_page; + } + } else + VM_BUG_ON(!PageLocked(page)); page_mkwrite = 1; } } @@ -2736,19 +2764,35 @@ static int __do_fault(struct mm_struct *mm, struct vm_area_struct *vma, pte_unmap_unlock(page_table, ptl); out: - unlock_page(vmf.page); -out_unlocked: - if (anon) - page_cache_release(vmf.page); - else if (dirty_page) { - if (vma->vm_file) - file_update_time(vma->vm_file); + if (dirty_page) { + struct address_space *mapping = page->mapping; - set_page_dirty_balance(dirty_page, page_mkwrite); + if (set_page_dirty(dirty_page)) + page_mkwrite = 1; + unlock_page(dirty_page); put_page(dirty_page); + if (page_mkwrite && mapping) { + /* + * Some device drivers do not set page.mapping but still + * dirty their pages + */ + balance_dirty_pages_ratelimited(mapping); + } + + /* file_update_time outside page_lock */ + if (vma->vm_file) + file_update_time(vma->vm_file); + } else { + unlock_page(vmf.page); + if (anon) + page_cache_release(vmf.page); } return ret; + +unwritable_page: + page_cache_release(page); + return ret; } static int do_linear_fault(struct mm_struct *mm, struct vm_area_struct *vma, -- cgit v1.2.3-18-g5258 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 --- mm/memcontrol.c | 33 ++++++++++++--------------------- mm/shmem.c | 8 ++++++-- 2 files changed, 18 insertions(+), 23 deletions(-) (limited to 'mm') diff --git a/mm/memcontrol.c b/mm/memcontrol.c index 575203ae210..01c2d8f1468 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 f9cb20ebb99..b25f95ce3db 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 v1.2.3-18-g5258 From 00a62ce91e554198ef28234c91c36f850f5a3bc9 Mon Sep 17 00:00:00 2001 From: KOSAKI Motohiro Date: Thu, 30 Apr 2009 15:08:51 -0700 Subject: mm: fix Committed_AS underflow on large NR_CPUS environment The Committed_AS field can underflow in certain situations: > # while true; do cat /proc/meminfo | grep _AS; sleep 1; done | uniq -c > 1 Committed_AS: 18446744073709323392 kB > 11 Committed_AS: 18446744073709455488 kB > 6 Committed_AS: 35136 kB > 5 Committed_AS: 18446744073709454400 kB > 7 Committed_AS: 35904 kB > 3 Committed_AS: 18446744073709453248 kB > 2 Committed_AS: 34752 kB > 9 Committed_AS: 18446744073709453248 kB > 8 Committed_AS: 34752 kB > 3 Committed_AS: 18446744073709320960 kB > 7 Committed_AS: 18446744073709454080 kB > 3 Committed_AS: 18446744073709320960 kB > 5 Committed_AS: 18446744073709454080 kB > 6 Committed_AS: 18446744073709320960 kB Because NR_CPUS can be greater than 1000 and meminfo_proc_show() does not check for underflow. But NR_CPUS proportional isn't good calculation. In general, possibility of lock contention is proportional to the number of online cpus, not theorical maximum cpus (NR_CPUS). The current kernel has generic percpu-counter stuff. using it is right way. it makes code simplify and percpu_counter_read_positive() don't make underflow issue. Reported-by: Dave Hansen Signed-off-by: KOSAKI Motohiro Cc: Eric B Munson Cc: Mel Gorman Cc: Christoph Lameter Cc: [All kernel versions] Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/mmap.c | 12 ++++++------ mm/nommu.c | 13 +++++++------ mm/swap.c | 46 ---------------------------------------------- 3 files changed, 13 insertions(+), 58 deletions(-) (limited to 'mm') diff --git a/mm/mmap.c b/mm/mmap.c index 3303d1ba8e8..6b7b1a95944 100644 --- a/mm/mmap.c +++ b/mm/mmap.c @@ -85,7 +85,7 @@ EXPORT_SYMBOL(vm_get_page_prot); int sysctl_overcommit_memory = OVERCOMMIT_GUESS; /* heuristic overcommit */ int sysctl_overcommit_ratio = 50; /* default is 50% */ int sysctl_max_map_count __read_mostly = DEFAULT_MAX_MAP_COUNT; -atomic_long_t vm_committed_space = ATOMIC_LONG_INIT(0); +struct percpu_counter vm_committed_as; /* * Check that a process has enough memory to allocate a new virtual @@ -179,11 +179,7 @@ int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin) if (mm) allowed -= mm->total_vm / 32; - /* - * cast `allowed' as a signed long because vm_committed_space - * sometimes has a negative value - */ - if (atomic_long_read(&vm_committed_space) < (long)allowed) + if (percpu_counter_read_positive(&vm_committed_as) < allowed) return 0; error: vm_unacct_memory(pages); @@ -2481,4 +2477,8 @@ void mm_drop_all_locks(struct mm_struct *mm) */ void __init mmap_init(void) { + int ret; + + ret = percpu_counter_init(&vm_committed_as, 0); + VM_BUG_ON(ret); } diff --git a/mm/nommu.c b/mm/nommu.c index 72eda4aee2c..809998aa7b5 100644 --- a/mm/nommu.c +++ b/mm/nommu.c @@ -62,7 +62,7 @@ void *high_memory; struct page *mem_map; unsigned long max_mapnr; unsigned long num_physpages; -atomic_long_t vm_committed_space = ATOMIC_LONG_INIT(0); +struct percpu_counter vm_committed_as; int sysctl_overcommit_memory = OVERCOMMIT_GUESS; /* heuristic overcommit */ int sysctl_overcommit_ratio = 50; /* default is 50% */ int sysctl_max_map_count = DEFAULT_MAX_MAP_COUNT; @@ -463,6 +463,10 @@ SYSCALL_DEFINE1(brk, unsigned long, brk) */ void __init mmap_init(void) { + int ret; + + ret = percpu_counter_init(&vm_committed_as, 0); + VM_BUG_ON(ret); vm_region_jar = KMEM_CACHE(vm_region, SLAB_PANIC); } @@ -1847,12 +1851,9 @@ int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin) if (mm) allowed -= mm->total_vm / 32; - /* - * cast `allowed' as a signed long because vm_committed_space - * sometimes has a negative value - */ - if (atomic_long_read(&vm_committed_space) < (long)allowed) + if (percpu_counter_read_positive(&vm_committed_as) < allowed) return 0; + error: vm_unacct_memory(pages); diff --git a/mm/swap.c b/mm/swap.c index bede23ce64e..cb29ae5d33a 100644 --- a/mm/swap.c +++ b/mm/swap.c @@ -491,49 +491,6 @@ unsigned pagevec_lookup_tag(struct pagevec *pvec, struct address_space *mapping, EXPORT_SYMBOL(pagevec_lookup_tag); -#ifdef CONFIG_SMP -/* - * We tolerate a little inaccuracy to avoid ping-ponging the counter between - * CPUs - */ -#define ACCT_THRESHOLD max(16, NR_CPUS * 2) - -static DEFINE_PER_CPU(long, committed_space); - -void vm_acct_memory(long pages) -{ - long *local; - - preempt_disable(); - local = &__get_cpu_var(committed_space); - *local += pages; - if (*local > ACCT_THRESHOLD || *local < -ACCT_THRESHOLD) { - atomic_long_add(*local, &vm_committed_space); - *local = 0; - } - preempt_enable(); -} - -#ifdef CONFIG_HOTPLUG_CPU - -/* Drop the CPU's cached committed space back into the central pool. */ -static int cpu_swap_callback(struct notifier_block *nfb, - unsigned long action, - void *hcpu) -{ - long *committed; - - committed = &per_cpu(committed_space, (long)hcpu); - if (action == CPU_DEAD || action == CPU_DEAD_FROZEN) { - atomic_long_add(*committed, &vm_committed_space); - *committed = 0; - drain_cpu_pagevecs((long)hcpu); - } - return NOTIFY_OK; -} -#endif /* CONFIG_HOTPLUG_CPU */ -#endif /* CONFIG_SMP */ - /* * Perform any setup for the swap system */ @@ -554,7 +511,4 @@ void __init swap_setup(void) * Right now other parts of the system means that we * _really_ don't want to cluster much more */ -#ifdef CONFIG_HOTPLUG_CPU - hotcpu_notifier(cpu_swap_callback, 0); -#endif } -- cgit v1.2.3-18-g5258 From 8713e01295140f674a41f2199b0f7ca99dfb69d5 Mon Sep 17 00:00:00 2001 From: Andrew Morton Date: Thu, 30 Apr 2009 15:08:55 -0700 Subject: vmscan: avoid multiplication overflow in shrink_zone() Local variable `scan' can overflow on zones which are larger than (2G * 4k) / 100 = 80GB. Making it 64-bit on 64-bit will fix that up. Cc: KOSAKI Motohiro Cc: Wu Fengguang Cc: Peter Zijlstra Cc: Rik van Riel Cc: Lee Schermerhorn Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/vmscan.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'mm') diff --git a/mm/vmscan.c b/mm/vmscan.c index eac9577941f..5fa3eda1f03 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -1471,7 +1471,7 @@ static void shrink_zone(int priority, struct zone *zone, for_each_evictable_lru(l) { int file = is_file_lru(l); - int scan; + unsigned long scan; scan = zone_nr_pages(zone, sc, l); if (priority) { -- cgit v1.2.3-18-g5258 From a425a638c858fd10370b573bde81df3ba500e271 Mon Sep 17 00:00:00 2001 From: Mel Gorman Date: Tue, 5 May 2009 16:37:17 +0100 Subject: Ignore madvise(MADV_WILLNEED) for hugetlbfs-backed regions madvise(MADV_WILLNEED) forces page cache readahead on a range of memory backed by a file. The assumption is made that the page required is order-0 and "normal" page cache. On hugetlbfs, this assumption is not true and order-0 pages are allocated and inserted into the hugetlbfs page cache. This leaks hugetlbfs page reservations and can cause BUGs to trigger related to corrupted page tables. This patch causes MADV_WILLNEED to be ignored for hugetlbfs-backed regions. Signed-off-by: Mel Gorman Cc: stable@kernel.org Signed-off-by: Linus Torvalds --- mm/madvise.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'mm') diff --git a/mm/madvise.c b/mm/madvise.c index b9ce574827c..36d6ea2b634 100644 --- a/mm/madvise.c +++ b/mm/madvise.c @@ -112,6 +112,14 @@ static long madvise_willneed(struct vm_area_struct * vma, if (!file) return -EBADF; + /* + * Page cache readahead assumes page cache pages are order-0 which + * is not the case for hugetlbfs. Do not give a bad return value + * but ignore the advice. + */ + if (vma->vm_flags & VM_HUGETLB) + return 0; + if (file->f_mapping->a_ops->get_xip_mem) { /* no bad return value, but ignore advice */ return 0; -- cgit v1.2.3-18-g5258 From 184101bf143ac96d62b3dcc17e7b3550f98d3350 Mon Sep 17 00:00:00 2001 From: David Rientjes Date: Wed, 6 May 2009 16:02:55 -0700 Subject: oom: prevent livelock when oom_kill_allocating_task is set When /proc/sys/vm/oom_kill_allocating_task is set for large systems that want to avoid the lengthy tasklist scan, it's possible to livelock if current is ineligible for oom kill. This normally happens when it is set to OOM_DISABLE, but is also possible if any threads are sharing the same ->mm with a different tgid. So change __out_of_memory() to fall back to the full task-list scan if it was unable to kill `current'. Cc: Nick Piggin Signed-off-by: David Rientjes Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/oom_kill.c | 44 +++++++++++++++++++++----------------------- 1 file changed, 21 insertions(+), 23 deletions(-) (limited to 'mm') diff --git a/mm/oom_kill.c b/mm/oom_kill.c index 2f3166e308d..92bcf1db16b 100644 --- a/mm/oom_kill.c +++ b/mm/oom_kill.c @@ -514,34 +514,32 @@ void clear_zonelist_oom(struct zonelist *zonelist, gfp_t gfp_mask) */ static void __out_of_memory(gfp_t gfp_mask, int order) { - if (sysctl_oom_kill_allocating_task) { - oom_kill_process(current, gfp_mask, order, 0, NULL, - "Out of memory (oom_kill_allocating_task)"); - - } else { - unsigned long points; - struct task_struct *p; - -retry: - /* - * Rambo mode: Shoot down a process and hope it solves whatever - * issues we may have. - */ - p = select_bad_process(&points, NULL); + struct task_struct *p; + unsigned long points; - if (PTR_ERR(p) == -1UL) + if (sysctl_oom_kill_allocating_task) + if (!oom_kill_process(current, gfp_mask, order, 0, NULL, + "Out of memory (oom_kill_allocating_task)")) return; +retry: + /* + * Rambo mode: Shoot down a process and hope it solves whatever + * issues we may have. + */ + p = select_bad_process(&points, NULL); - /* Found nothing?!?! Either we hang forever, or we panic. */ - if (!p) { - read_unlock(&tasklist_lock); - panic("Out of memory and no killable processes...\n"); - } + if (PTR_ERR(p) == -1UL) + return; - if (oom_kill_process(p, gfp_mask, order, points, NULL, - "Out of memory")) - goto retry; + /* Found nothing?!?! Either we hang forever, or we panic. */ + if (!p) { + read_unlock(&tasklist_lock); + panic("Out of memory and no killable processes...\n"); } + + if (oom_kill_process(p, gfp_mask, order, points, NULL, + "Out of memory")) + goto retry; } /* -- cgit v1.2.3-18-g5258 From 2498ce42d3a4d1a498f1df4884da960087547db7 Mon Sep 17 00:00:00 2001 From: Ralph Wuerthner Date: Wed, 6 May 2009 16:02:59 -0700 Subject: alloc_vmap_area: fix memory leak If alloc_vmap_area() fails the allocated struct vmap_area has to be freed. Signed-off-by: Ralph Wuerthner Reviewed-by: Christoph Lameter Reviewed-by: Minchan Kim Cc: Nick Piggin Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/vmalloc.c | 1 + 1 file changed, 1 insertion(+) (limited to 'mm') diff --git a/mm/vmalloc.c b/mm/vmalloc.c index fab19876b4d..083716ea38c 100644 --- a/mm/vmalloc.c +++ b/mm/vmalloc.c @@ -402,6 +402,7 @@ overflow: printk(KERN_WARNING "vmap allocation for size %lu failed: " "use vmalloc= to increase size.\n", size); + kfree(va); return ERR_PTR(-EBUSY); } -- cgit v1.2.3-18-g5258 From 9155203a5de94278525647b16733f0c315f3b786 Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 6 May 2009 16:03:02 -0700 Subject: mm: use roundown_pow_of_two() in zone_batchsize() Use roundown_pow_of_two(N) in zone_batchsize() rather than (1 << (fls(N)-1)) as they are equivalent, and with the former it is easier to see what is going on. Signed-off-by: David Howells Tested-by: Lanttor Guo Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/page_alloc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'mm') diff --git a/mm/page_alloc.c b/mm/page_alloc.c index e2f26991fff..8add7daf98b 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -2706,7 +2706,7 @@ static int zone_batchsize(struct zone *zone) * of pages of one half of the possible page colors * and the other with pages of the other colors. */ - batch = (1 << (fls(batch + batch/2)-1)) - 1; + batch = rounddown_pow_of_two(batch + batch/2) - 1; return batch; } -- cgit v1.2.3-18-g5258 From 3a6be87fd1e5cdbbc3b6a14d02a3efa9ecba1d3f Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 6 May 2009 16:03:03 -0700 Subject: nommu: clamp zone_batchsize() to 0 under NOMMU conditions Clamp zone_batchsize() to 0 under NOMMU conditions to stop free_hot_cold_page() from queueing and batching frees. The problem is that under NOMMU conditions it is really important to be able to allocate large contiguous chunks of memory, but when munmap() or exit_mmap() releases big stretches of memory, return of these to the buddy allocator can be deferred, and when it does finally happen, it can be in small chunks. Whilst the fragmentation this incurs isn't so much of a problem under MMU conditions as userspace VM is glued together from individual pages with the aid of the MMU, it is a real problem if there isn't an MMU. By clamping the page freeing queue size to 0, pages are returned to the allocator immediately, and the buddy detector is more likely to be able to glue them together into large chunks immediately, and fragmentation is less likely to occur. By disabling batching of frees, and by turning off the trimming of excess space during boot, Coldfire can manage to boot. Reported-by: Lanttor Guo Signed-off-by: David Howells Tested-by: Lanttor Guo Cc: Greg Ungerer Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/page_alloc.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) (limited to 'mm') diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 8add7daf98b..fe753ecf2aa 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -2681,6 +2681,7 @@ static void __meminit zone_init_free_lists(struct zone *zone) static int zone_batchsize(struct zone *zone) { +#ifdef CONFIG_MMU int batch; /* @@ -2709,6 +2710,23 @@ static int zone_batchsize(struct zone *zone) batch = rounddown_pow_of_two(batch + batch/2) - 1; return batch; + +#else + /* The deferral and batching of frees should be suppressed under NOMMU + * conditions. + * + * The problem is that NOMMU needs to be able to allocate large chunks + * of contiguous memory as there's no hardware page translation to + * assemble apparent contiguous memory from discontiguous pages. + * + * Queueing large contiguous runs of pages for batching, however, + * causes the pages to actually be freed in smaller chunks. As there + * can be a significant delay between the individual batches being + * recycled, this leads to the once large chunks of space being + * fragmented and becoming unavailable for high-order allocations. + */ + return 0; +#endif } static void setup_pageset(struct per_cpu_pageset *p, unsigned long batch) -- cgit v1.2.3-18-g5258 From fc4d5c292b68ef02514d2072dcbf82d090c34875 Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 6 May 2009 16:03:05 -0700 Subject: nommu: make the initial mmap allocation excess behaviour Kconfig configurable NOMMU mmap() has an option controlled by a sysctl variable that determines whether the allocations made by do_mmap_private() should have the excess space trimmed off and returned to the allocator. Make the initial setting of this variable a Kconfig configuration option. The reason there can be excess space is that the allocator only allocates in power-of-2 size chunks, but mmap()'s can be made in sizes that aren't a power of 2. There are two alternatives: (1) Keep the excess as dead space. The dead space then remains unused for the lifetime of the mapping. Mappings of shared objects such as libc, ld.so or busybox's text segment may retain their dead space forever. (2) Return the excess to the allocator. This means that the dead space is limited to less than a page per mapping, but it means that for a transient process, there's more chance of fragmentation as the excess space may be reused fairly quickly. During the boot process, a lot of transient processes are created, and this can cause a lot of fragmentation as the pagecache and various slabs grow greatly during this time. By turning off the trimming of excess space during boot and disabling batching of frees, Coldfire can manage to boot. A better way of doing things might be to have /sbin/init turn this option off. By that point libc, ld.so and init - which are all long-duration processes - have all been loaded and trimmed. Reported-by: Lanttor Guo Signed-off-by: David Howells Tested-by: Lanttor Guo Cc: Greg Ungerer Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- mm/Kconfig | 28 ++++++++++++++++++++++++++++ mm/nommu.c | 2 +- 2 files changed, 29 insertions(+), 1 deletion(-) (limited to 'mm') diff --git a/mm/Kconfig b/mm/Kconfig index 57971d2ab84..c2b57d81e15 100644 --- a/mm/Kconfig +++ b/mm/Kconfig @@ -225,3 +225,31 @@ config HAVE_MLOCKED_PAGE_BIT config MMU_NOTIFIER bool + +config NOMMU_INITIAL_TRIM_EXCESS + int "Turn on mmap() excess space trimming before booting" + depends on !MMU + default 1 + help + The NOMMU mmap() frequently needs to allocate large contiguous chunks + of memory on which to store mappings, but it can only ask the system + allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently + more than it requires. To deal with this, mmap() is able to trim off + the excess and return it to the allocator. + + If trimming is enabled, the excess is trimmed off and returned to the + system allocator, which can cause extra fragmentation, particularly + if there are a lot of transient processes. + + If trimming is disabled, the excess is kept, but not used, which for + long-term mappings means that the space is wasted. + + Trimming can be dynamically controlled through a sysctl option + (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of + excess pages there must be before trimming should occur, or zero if + no trimming is to occur. + + This option specifies the initial value of this option. The default + of 1 says that all excess pages should be trimmed. + + See Documentation/nommu-mmap.txt for more information. diff --git a/mm/nommu.c b/mm/nommu.c index 809998aa7b5..67cd1a487ee 100644 --- a/mm/nommu.c +++ b/mm/nommu.c @@ -66,7 +66,7 @@ struct percpu_counter vm_committed_as; int sysctl_overcommit_memory = OVERCOMMIT_GUESS; /* heuristic overcommit */ int sysctl_overcommit_ratio = 50; /* default is 50% */ int sysctl_max_map_count = DEFAULT_MAX_MAP_COUNT; -int sysctl_nr_trim_pages = 1; /* page trimming behaviour */ +int sysctl_nr_trim_pages = CONFIG_NOMMU_INITIAL_TRIM_EXCESS; int heap_stack_gap = 0; atomic_long_t mmap_pages_allocated; -- cgit v1.2.3-18-g5258 From 8c9ed899b44c19e81859fbb0e9d659fe2f8630fc Mon Sep 17 00:00:00 2001 From: David Howells Date: Thu, 7 May 2009 11:41:37 +0100 Subject: NOMMU: Don't check vm_region::vm_start is page aligned in add_nommu_region() Don't check vm_region::vm_start is page aligned in add_nommu_region() because the region may reflect some non-page-aligned mapped file, such as could be obtained from RomFS XIP. Signed-off-by: David Howells Acked-by: Greg Ungerer Signed-off-by: Linus Torvalds --- mm/nommu.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'mm') diff --git a/mm/nommu.c b/mm/nommu.c index 67cd1a487ee..b571ef70742 100644 --- a/mm/nommu.c +++ b/mm/nommu.c @@ -515,8 +515,6 @@ static void add_nommu_region(struct vm_region *region) validate_nommu_regions(); - BUG_ON(region->vm_start & ~PAGE_MASK); - parent = NULL; p = &nommu_region_tree.rb_node; while (*p) { -- cgit v1.2.3-18-g5258 From 0f181328287db30671e9997329cff71395d4af8b Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Wed, 13 May 2009 08:29:12 -0700 Subject: Revert "Ignore madvise(MADV_WILLNEED) for hugetlbfs-backed regions" This reverts commit a425a638c858fd10370b573bde81df3ba500e271. Now that the previous commit removed the "readpage" actor for hugetlb files, read-ahead will no longer mess up the mapping, and there's no longer any reason to treat hugetlbfs mappings specially. Tested-and-acked-by: Mel Gorman Signed-off-by: Linus Torvalds --- mm/madvise.c | 8 -------- 1 file changed, 8 deletions(-) (limited to 'mm') diff --git a/mm/madvise.c b/mm/madvise.c index 36d6ea2b634..b9ce574827c 100644 --- a/mm/madvise.c +++ b/mm/madvise.c @@ -112,14 +112,6 @@ static long madvise_willneed(struct vm_area_struct * vma, if (!file) return -EBADF; - /* - * Page cache readahead assumes page cache pages are order-0 which - * is not the case for hugetlbfs. Do not give a bad return value - * but ignore the advice. - */ - if (vma->vm_flags & VM_HUGETLB) - return 0; - if (file->f_mapping->a_ops->get_xip_mem) { /* no bad return value, but ignore advice */ return 0; -- cgit v1.2.3-18-g5258 From cd17cbfda004fe5f406c01b318c6378d9895896f Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Fri, 15 May 2009 11:32:24 +0200 Subject: Revert "mm: add /proc controls for pdflush threads" This reverts commit fafd688e4c0c34da0f3de909881117d374e4c7af. Work is progressing to switch away from pdflush as the process backing for flushing out dirty data. So it seems pointless to add more knobs to control pdflush threads. The original author of the patch did not have any specific use cases for adding the knobs, so we can easily revert this before 2.6.30 to avoid having to maintain this API forever. Signed-off-by: Jens Axboe --- mm/pdflush.c | 31 ++++++++++++------------------- 1 file changed, 12 insertions(+), 19 deletions(-) (limited to 'mm') diff --git a/mm/pdflush.c b/mm/pdflush.c index f2caf96993f..235ac440c44 100644 --- a/mm/pdflush.c +++ b/mm/pdflush.c @@ -57,14 +57,6 @@ static DEFINE_SPINLOCK(pdflush_lock); */ int nr_pdflush_threads = 0; -/* - * The max/min number of pdflush threads. R/W by sysctl at - * /proc/sys/vm/nr_pdflush_threads_max/min - */ -int nr_pdflush_threads_max __read_mostly = MAX_PDFLUSH_THREADS; -int nr_pdflush_threads_min __read_mostly = MIN_PDFLUSH_THREADS; - - /* * The time at which the pdflush thread pool last went empty */ @@ -76,7 +68,7 @@ static unsigned long last_empty_jifs; * Thread pool management algorithm: * * - The minimum and maximum number of pdflush instances are bound - * by nr_pdflush_threads_min and nr_pdflush_threads_max. + * by MIN_PDFLUSH_THREADS and MAX_PDFLUSH_THREADS. * * - If there have been no idle pdflush instances for 1 second, create * a new one. @@ -142,13 +134,14 @@ static int __pdflush(struct pdflush_work *my_work) * To throttle creation, we reset last_empty_jifs. */ if (time_after(jiffies, last_empty_jifs + 1 * HZ)) { - if (list_empty(&pdflush_list) && - nr_pdflush_threads < nr_pdflush_threads_max) { - last_empty_jifs = jiffies; - nr_pdflush_threads++; - spin_unlock_irq(&pdflush_lock); - start_one_pdflush_thread(); - spin_lock_irq(&pdflush_lock); + if (list_empty(&pdflush_list)) { + if (nr_pdflush_threads < MAX_PDFLUSH_THREADS) { + last_empty_jifs = jiffies; + nr_pdflush_threads++; + spin_unlock_irq(&pdflush_lock); + start_one_pdflush_thread(); + spin_lock_irq(&pdflush_lock); + } } } @@ -160,7 +153,7 @@ static int __pdflush(struct pdflush_work *my_work) */ if (list_empty(&pdflush_list)) continue; - if (nr_pdflush_threads <= nr_pdflush_threads_min) + if (nr_pdflush_threads <= MIN_PDFLUSH_THREADS) continue; pdf = list_entry(pdflush_list.prev, struct pdflush_work, list); if (time_after(jiffies, pdf->when_i_went_to_sleep + 1 * HZ)) { @@ -266,9 +259,9 @@ static int __init pdflush_init(void) * Pre-set nr_pdflush_threads... If we fail to create, * the count will be decremented. */ - nr_pdflush_threads = nr_pdflush_threads_min; + nr_pdflush_threads = MIN_PDFLUSH_THREADS; - for (i = 0; i < nr_pdflush_threads_min; i++) + for (i = 0; i < MIN_PDFLUSH_THREADS; i++) start_one_pdflush_thread(); return 0; } -- cgit v1.2.3-18-g5258