- 13 Feb, 2015 40 commits
-
-
Grazvydas Ignotas authored
For whatever reason, generic_access_phys() only remaps one page, but actually allows to access arbitrary size. It's quite easy to trigger large reads, like printing out large structure with gdb, which leads to a crash. Fix it by remapping correct size. Fixes: 28b2ee20 ("access_process_vm device memory infrastructure") Signed-off-by:
Grazvydas Ignotas <notasas@gmail.com> Cc: Rik van Riel <riel@redhat.com> Cc: <stable@vger.kernel.org> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Rasmus Villemoes authored
The only caller of cpuset_init_current_mems_allowed is the __init annotated build_all_zonelists_init, so we can also make the former __init. Signed-off-by:
Rasmus Villemoes <linux@rasmusvillemoes.dk> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Rik van Riel <riel@redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: Vishnu Pratap Singh <vishnu.ps@samsung.com> Cc: Pintu Kumar <pintu.k@samsung.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Tim Chen <tim.c.chen@linux.intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Li Zefan <lizefan@huawei.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by:
-
Rasmus Villemoes authored
mminit_loglevel is only referenced from __init and __meminit functions, so we can mark it __meminitdata. Signed-off-by:
-
Rasmus Villemoes authored
The only caller of mminit_verify_zonelist is build_all_zonelists_init, which is annotated with __init, so it should be safe to also mark the former as __init, saving ~400 bytes of .text. Signed-off-by:
-
Rasmus Villemoes authored
Pulling the code protected by if (system_state == SYSTEM_BOOTING) into its own helper allows us to shrink .text a little. This relies on build_all_zonelists already having a __ref annotation. Add a comment explaining why so one doesn't have to track it down through git log. The real saving comes in 3/5, ("mm/mm_init.c: Mark mminit_verify_zonelist as __init"), where we save about 400 bytes Signed-off-by: -
Rasmus Villemoes authored
All users of mminit_dprintk pass a compile-time constant as level, so this just makes gcc emit a single printk call instead of two. Signed-off-by:
-
Kirill A. Shutemov authored
mm->nr_pmds doesn't make sense on !MMU configurations Signed-off-by:
Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Guenter Roeck <linux@roeck-us.net> Signed-off-by:
-
Vladimir Davydov authored
Move memcg_socket_limit_enabled decrement to tcp_destroy_cgroup (called from memcg_destroy_kmem -> mem_cgroup_sockets_destroy) and zap a bunch of wrapper functions. Although this patch moves static keys decrement from __mem_cgroup_free to mem_cgroup_css_free, it does not introduce any functional changes, because the keys are incremented on setting the limit (tcp or kmem), which can only happen after successful mem_cgroup_css_online. Signed-off-by:
Vladimir Davydov <vdavydov@parallels.com> Cc: Glauber Costa <glommer@parallels.com> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujtisu.com> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: David S. Miller <davem@davemloft.net> Cc: Johannes Weiner <hannes@cmpxchg.org> Acked-by:
Michal Hocko <mhocko@suse.cz> Signed-off-by:
-
Joonsoo Kim authored
Currently, freepage isolation in one pageblock doesn't consider how many freepages we isolate. When I traced flow of compaction, compaction sometimes isolates more than 256 freepages to migrate just 32 pages. In this patch, freepage isolation is stopped at the point that we have more isolated freepage than isolated page for migration. This results in slowing down free page scanner and make compaction success rate higher. stress-highalloc test in mmtests with non movable order 7 allocation shows increase of compaction success rate. Compaction success rate (Compaction success * 100 / Compaction stalls, %) 27.13 : 31.82 pfn where both scanners meets on compaction complete (separate test due to enormous tracepoint buffer) (zone_start=4096, zone_end=1048576) 586034 : 654378 In fact, I didn't fully understand why this patch results in such good result. There was a guess that not used freepages are released to pcp list and on next compaction trial we won't isolate them again so compaction success rate would decrease. To prevent this effect, I tested with adding pcp drain code on release_freepages(), but, it has no good effect. Anyway, this patch reduces waste time to isolate unneeded freepages so seems reasonable. Vlastimil said: : I briefly tried it on top of the pivot-changing series and with order-9 : allocations it reduced free page scanned counter by almost 10%. No effect : on success rates (maybe because pivot changing already took care of the : scanners meeting problem) but the scanning reduction is good on its own. : : It also explains why e14c720e ("mm, compaction: remember position : within pageblock in free pages scanner") had less than expected : improvements. It would only actually stop within pageblock in case of : async compaction detecting contention. I guess that's also why the : infinite loop problem fixed by 1d5bfe1f affected so relatively few : people. Signed-off-by:
Joonsoo Kim <iamjoonsoo.kim@lge.com> Acked-by:
Vlastimil Babka <vbabka@suse.cz> Tested-by:
Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Cc: Mel Gorman <mgorman@suse.de> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by:
-
Joonsoo Kim authored
What we want to check here is whether there is highorder freepage in buddy list of other migratetype in order to steal it without fragmentation. But, current code just checks cc->order which means allocation request order. So, this is wrong. Without this fix, non-movable synchronous compaction below pageblock order would not stopped until compaction is complete, because migratetype of most pageblocks are movable and high order freepage made by compaction is usually on movable type buddy list. There is some report related to this bug. See below link. http://www.spinics.net/lists/linux-mm/msg81666.html Although the issued system still has load spike comes from compaction, this makes that system completely stable and responsive according to his report. stress-highalloc test in mmtests with non movable order 7 allocation doesn't show any notable difference in allocation success rate, but, it shows more compaction success rate. Compaction success rate (Compaction success * 100 / Compaction stalls, %) 18.47 : 28.94 Fixes: 1fb3f8ca ("mm: compaction: capture a suitable high-order page immediately when it is made available") Signed-off-by:
-
Vladimir Davydov authored
To speed up further allocations SLUB may store empty slabs in per cpu/node partial lists instead of freeing them immediately. This prevents per memcg caches destruction, because kmem caches created for a memory cgroup are only destroyed after the last page charged to the cgroup is freed. To fix this issue, this patch resurrects approach first proposed in [1]. It forbids SLUB to cache empty slabs after the memory cgroup that the cache belongs to was destroyed. It is achieved by setting kmem_cache's cpu_partial and min_partial constants to 0 and tuning put_cpu_partial() so that it would drop frozen empty slabs immediately if cpu_partial = 0. The runtime overhead is minimal. From all the hot functions, we only touch relatively cold put_cpu_partial(): we make it call unfreeze_partials() after freezing a slab that belongs to an offline memory cgroup. Since slab freezing exists to avoid moving slabs from/to a partial list on free/alloc, and there can't be allocations from dead caches, it shouldn't cause any overhead. We do have to disable preemption for put_cpu_partial() to achieve that though. The original patch was accepted well and even merged to the mm tree. However, I decided to withdraw it due to changes happening to the memcg core at that time. I had an idea of introducing per-memcg shrinkers for kmem caches, but now, as memcg has finally settled down, I do not see it as an option, because SLUB shrinker would be too costly to call since SLUB does not keep free slabs on a separate list. Besides, we currently do not even call per-memcg shrinkers for offline memcgs. Overall, it would introduce much more complexity to both SLUB and memcg than this small patch. Regarding to SLAB, there's no problem with it, because it shrinks per-cpu/node caches periodically. Thanks to list_lru reparenting, we no longer keep entries for offline cgroups in per-memcg arrays (such as memcg_cache_params->memcg_caches), so we do not have to bother if a per-memcg cache will be shrunk a bit later than it could be. [1] http://thread.gmane.org/gmane.linux.kernel.mm/118649/focus=118650Signed-off-by:
-
Vladimir Davydov authored
It is supposed to return 0 if the cache has no remaining objects and 1 otherwise, while currently it always returns 0. Fix it. Signed-off-by:
Christoph Lameter <cl@linux.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: David Rientjes <rientjes@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by:
-
Vladimir Davydov authored
SLUB's version of __kmem_cache_shrink() not only removes empty slabs, but also tries to rearrange the partial lists to place slabs filled up most to the head to cope with fragmentation. To achieve that, it allocates a temporary array of lists used to sort slabs by the number of objects in use. If the allocation fails, the whole procedure is aborted. This is unacceptable for the kernel memory accounting extension of the memory cgroup, where we want to make sure that kmem_cache_shrink() successfully discarded empty slabs. Although the allocation failure is utterly unlikely with the current page allocator implementation, which retries GFP_KERNEL allocations of order <= 2 infinitely, it is better not to rely on that. This patch therefore makes __kmem_cache_shrink() allocate the array on stack instead of calling kmalloc, which may fail. The array size is chosen to be equal to 32, because most SLUB caches store not more than 32 objects per slab page. Slab pages with <= 32 free objects are sorted using the array by the number of objects in use and promoted to the head of the partial list, while slab pages with > 32 free objects are left in the end of the list without any ordering imposed on them. Signed-off-by:
Pekka Enberg <penberg@kernel.org> Cc: David Rientjes <rientjes@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Huang Ying <ying.huang@intel.com> Signed-off-by:
-
Vladimir Davydov authored
Now, the only reason to keep kmemcg_id till css free is list_lru, which uses it to distribute elements between per-memcg lists. However, it can be easily sorted out - we only need to change kmemcg_id of an offline cgroup to its parent's id, making further list_lru_add()'s add elements to the parent's list, and then move all elements from the offline cgroup's list to the one of its parent. It will work, because a racing list_lru_del() does not need to know the list it is deleting the element from. It can decrement the wrong nr_items counter though, but the ongoing reparenting will fix it. After list_lru reparenting is done we are free to release kmemcg_id saving a valuable slot in a per-memcg array for new cgroups. Signed-off-by:
-
Vladimir Davydov authored
Currently, the isolate callback passed to the list_lru_walk family of functions is supposed to just delete an item from the list upon returning LRU_REMOVED or LRU_REMOVED_RETRY, while nr_items counter is fixed by __list_lru_walk_one after the callback returns. Since the callback is allowed to drop the lock after removing an item (it has to return LRU_REMOVED_RETRY then), the nr_items can be less than the actual number of elements on the list even if we check them under the lock. This makes it difficult to move items from one list_lru_one to another, which is required for per-memcg list_lru reparenting - we can't just splice the lists, we have to move entries one by one. This patch therefore introduces helpers that must be used by callback functions to isolate items instead of raw list_del/list_move. These are list_lru_isolate and list_lru_isolate_move. They not only remove the entry from the list, but also fix the nr_items counter, making sure nr_items always reflects the actual number of elements on the list if checked under the appropriate lock. Signed-off-by:
-
Vladimir Davydov authored
We need to look up a kmem_cache in ->memcg_params.memcg_caches arrays only on allocations, so there is no need to have the array entries set until css free - we can clear them on css offline. This will allow us to reuse array entries more efficiently and avoid costly array relocations. Signed-off-by:
-
Vladimir Davydov authored
Currently, we use mem_cgroup->kmemcg_id to guarantee kmem_cache->name uniqueness. This is correct, because kmemcg_id is only released on css free after destroying all per memcg caches. However, I am going to change that and release kmemcg_id on css offline, because it is not wise to keep it for so long, wasting valuable entries of memcg_cache_params->memcg_caches arrays. Therefore, to preserve cache name uniqueness, let us switch to css->id. Signed-off-by:
-
Vladimir Davydov authored
Currently, we release css->id in css_release_work_fn, right before calling css_free callback, so that when css_free is called, the id may have already been reused for a new cgroup. I am going to use css->id to create unique names for per memcg kmem caches. Since kmem caches are destroyed only on css_free, I need css->id to be freed after css_free was called to avoid name clashes. This patch therefore moves css->id removal to css_free_work_fn. To prevent css_from_id from returning a pointer to a stale css, it makes css_release_work_fn replace the css ptr at css_idr:css->id with NULL. Signed-off-by:
Tejun Heo <tj@kernel.org> Cc: Christoph Lameter <cl@linux.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: David Rientjes <rientjes@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Dave Chinner <david@fromorbit.com> Signed-off-by:
-
Vladimir Davydov authored
Sometimes, we need to iterate over all memcg copies of a particular root kmem cache. Currently, we use memcg_cache_params->memcg_caches array for that, because it contains all existing memcg caches. However, it's a bad practice to keep all caches, including those that belong to offline cgroups, in this array, because it will be growing beyond any bounds then. I'm going to wipe away dead caches from it to save space. To still be able to perform iterations over all memcg caches of the same kind, let us link them into a list. Signed-off-by:
-
Vladimir Davydov authored
Currently, kmem_cache stores a pointer to struct memcg_cache_params instead of embedding it. The rationale is to save memory when kmem accounting is disabled. However, the memcg_cache_params has shrivelled drastically since it was first introduced: * Initially: struct memcg_cache_params { bool is_root_cache; union { struct kmem_cache *memcg_caches[0]; struct { struct mem_cgroup *memcg; struct list_head list; struct kmem_cache *root_cache; bool dead; atomic_t nr_pages; struct work_struct destroy; }; }; }; * Now: struct memcg_cache_params { bool is_root_cache; union { struct { struct rcu_head rcu_head; struct kmem_cache *memcg_caches[0]; }; struct { struct mem_cgroup *memcg; struct kmem_cache *root_cache; }; }; }; So the memory saving does not seem to be a clear win anymore. OTOH, keeping a pointer to memcg_cache_params struct instead of embedding it results in touching one more cache line on kmem alloc/free hot paths. Besides, it makes linking kmem caches in a list chained by a field of struct memcg_cache_params really painful due to a level of indirection, while I want to make them linked in the following patch. That said, let us embed it. Signed-off-by: -
Vladimir Davydov authored
In super_cache_scan() we divide the number of objects of particular type by the total number of objects in order to distribute pressure among As a result, in some corner cases we can get nr_to_scan=0 even if there are some objects to reclaim, e.g. dentries=1, inodes=1, fs_objects=1, nr_to_scan=1/3=0. This is unacceptable for per memcg kmem accounting, because this means that some objects may never get reclaimed after memcg death, preventing it from being freed. This patch therefore assures that super_cache_scan() will scan at least one object of each type if any. [akpm@linux-foundation.org: add comment] Signed-off-by:
-
Vladimir Davydov authored
Now, to make any list_lru-based shrinker memcg aware we should only initialize its list_lru as memcg aware. Let's do it for the general FS shrinker (super_block::s_shrink). There are other FS-specific shrinkers that use list_lru for storing objects, such as XFS and GFS2 dquot cache shrinkers, but since they reclaim objects that are shared among different cgroups, there is no point making them memcg aware. It's a big question whether we should account them to memcg at all. Signed-off-by:
-
Vladimir Davydov authored
There are several FS shrinkers, including super_block::s_shrink, that keep reclaimable objects in the list_lru structure. Hence to turn them to memcg-aware shrinkers, it is enough to make list_lru per-memcg. This patch does the trick. It adds an array of lru lists to the list_lru_node structure (per-node part of the list_lru), one for each kmem-active memcg, and dispatches every item addition or removal to the list corresponding to the memcg which the item is accounted to. So now the list_lru structure is not just per node, but per node and per memcg. Not all list_lrus need this feature, so this patch also adds a new method, list_lru_init_memcg, which initializes a list_lru as memcg aware. Otherwise (i.e. if initialized with old list_lru_init), the list_lru won't have per memcg lists. Just like per memcg caches arrays, the arrays of per-memcg lists are indexed by memcg_cache_id, so we must grow them whenever memcg_nr_cache_ids is increased. So we introduce a callback, memcg_update_all_list_lrus, invoked by memcg_alloc_cache_id if the id space is full. The locking is implemented in a manner similar to lruvecs, i.e. we have one lock per node that protects all lists (both global and per cgroup) on the node. Signed-off-by:
-
Vladimir Davydov authored
To make list_lru memcg aware, we need all list_lrus to be kept on a list protected by a mutex, so that we could sleep while walking over the list. Therefore after this change list_lru_destroy may sleep. Fortunately, there is only one user that calls it from an atomic context - it's put_super - and we can easily fix it by calling list_lru_destroy before put_super in destroy_locked_super - anyway we don't longer need lrus by that time. Another point that should be noted is that list_lru_destroy is allowed to be called on an uninitialized zeroed-out object, in which case it is a no-op. Before this patch this was guaranteed by kfree, but now we need an explicit check there. Signed-off-by:
-
Vladimir Davydov authored
The active_nodes mask allows us to skip empty nodes when walking over list_lru items from all nodes in list_lru_count/walk. However, these functions are never called from hot paths, so it doesn't seem we need such kind of optimization there. OTOH, removing the mask will make it easier to make list_lru per-memcg. Signed-off-by:
-
Vladimir Davydov authored
We need a stable value of memcg_nr_cache_ids in kmem_cache_create() (memcg_alloc_cache_params() wants it for root caches), where we only hold the slab_mutex and no memcg-related locks. As a result, we have to update memcg_nr_cache_ids under the slab_mutex, which we can only take on the slab's side (see memcg_update_array_size). This looks awkward and will become even worse when per-memcg list_lru is introduced, which also wants stable access to memcg_nr_cache_ids. To get rid of this dependency between the memcg_nr_cache_ids and the slab_mutex, this patch introduces a special rwsem. The rwsem is held for writing during memcg_caches arrays relocation and memcg_nr_cache_ids updates. Therefore one can take it for reading to get a stable access to memcg_caches arrays and/or memcg_nr_cache_ids. Currently the semaphore is taken for reading only from kmem_cache_create, right before taking the slab_mutex, so right now there's no much point in using rwsem instead of mutex. However, once list_lru is made per-memcg it will allow list_lru initializations to proceed concurrently. Signed-off-by:
-
Vladimir Davydov authored
memcg_limited_groups_array_size, which defines the size of memcg_caches arrays, sounds rather cumbersome. Also it doesn't point anyhow that it's related to kmem/caches stuff. So let's rename it to memcg_nr_cache_ids. It's concise and points us directly to memcg_cache_id. Also, rename kmem_limited_groups to memcg_cache_ida. Signed-off-by:
-
Vladimir Davydov authored
This patch adds SHRINKER_MEMCG_AWARE flag. If a shrinker has this flag set, it will be called per memory cgroup. The memory cgroup to scan objects from is passed in shrink_control->memcg. If the memory cgroup is NULL, a memcg aware shrinker is supposed to scan objects from the global list. Unaware shrinkers are only called on global pressure with memcg=NULL. Signed-off-by:
-
Vladimir Davydov authored
We are going to make FS shrinkers memcg-aware. To achieve that, we will have to pass the memcg to scan to the nr_cached_objects and free_cached_objects VFS methods, which currently take only the NUMA node to scan. Since the shrink_control structure already holds the node, and the memcg to scan will be added to it when we introduce memcg-aware vmscan, let us consolidate the methods' arguments in this structure to keep things clean. Signed-off-by:
Dave Chinner <david@fromorbit.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Michal Hocko <mhocko@suse.cz> Cc: Greg Thelen <gthelen@google.com> Cc: Glauber Costa <glommer@gmail.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Christoph Lameter <cl@linux.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: David Rientjes <rientjes@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by:
-
Vladimir Davydov authored
Kmem accounting of memcg is unusable now, because it lacks slab shrinker support. That means when we hit the limit we will get ENOMEM w/o any chance to recover. What we should do then is to call shrink_slab, which would reclaim old inode/dentry caches from this cgroup. This is what this patch set is intended to do. Basically, it does two things. First, it introduces the notion of per-memcg slab shrinker. A shrinker that wants to reclaim objects per cgroup should mark itself as SHRINKER_MEMCG_AWARE. Then it will be passed the memory cgroup to scan from in shrink_control->memcg. For such shrinkers shrink_slab iterates over the whole cgroup subtree under the target cgroup and calls the shrinker for each kmem-active memory cgroup. Secondly, this patch set makes the list_lru structure per-memcg. It's done transparently to list_lru users - everything they have to do is to tell list_lru_init that they want memcg-aware list_lru. Then the list_lru will automatically distribute objects among per-memcg lists basing on which cgroup the object is accounted to. This way to make FS shrinkers (icache, dcache) memcg-aware we only need to make them use memcg-aware list_lru, and this is what this patch set does. As before, this patch set only enables per-memcg kmem reclaim when the pressure goes from memory.limit, not from memory.kmem.limit. Handling memory.kmem.limit is going to be tricky due to GFP_NOFS allocations, and it is still unclear whether we will have this knob in the unified hierarchy. This patch (of 9): NUMA aware slab shrinkers use the list_lru structure to distribute objects coming from different NUMA nodes to different lists. Whenever such a shrinker needs to count or scan objects from a particular node, it issues commands like this: count = list_lru_count_node(lru, sc->nid); freed = list_lru_walk_node(lru, sc->nid, isolate_func, isolate_arg, &sc->nr_to_scan); where sc is an instance of the shrink_control structure passed to it from vmscan. To simplify this, let's add special list_lru functions to be used by shrinkers, list_lru_shrink_count() and list_lru_shrink_walk(), which consolidate the nid and nr_to_scan arguments in the shrink_control structure. This will also allow us to avoid patching shrinkers that use list_lru when we make shrink_slab() per-memcg - all we will have to do is extend the shrink_control structure to include the target memcg and make list_lru_shrink_{count,walk} handle this appropriately. Signed-off-by: -
Mel Gorman authored
If a PTE or PMD is already marked NUMA when scanning to mark entries for NUMA hinting then it is not necessary to update the entry and incur a TLB flush penalty. Avoid the avoidhead where possible. Signed-off-by:
Mel Gorman <mgorman@suse.de> Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Dave Jones <davej@redhat.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Kirill Shutemov <kirill.shutemov@linux.intel.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Rik van Riel <riel@redhat.com> Cc: Sasha Levin <sasha.levin@oracle.com> Signed-off-by:
-
Mel Gorman authored
pte_protnone_numa is only safe to use after VMA checks for PROT_NONE are complete. Treating a real PROT_NONE PTE as a NUMA hinting fault is going to result in strangeness so add a check for it. BUG_ON looks like overkill but if this is hit then it's a serious bug that could result in corruption so do not even try recovering. It would have been more comprehensive to check VMA flags in pte_protnone_numa but it would have made the API ugly just for a debugging check. Signed-off-by:
-
Mel Gorman authored
Commit b38af472 ("x86,mm: fix pte_special versus pte_numa") adjusted the pte_special check to take into account that a special pte had SPECIAL and neither PRESENT nor PROTNONE. Now that NUMA hinting PTEs are no longer modifying _PAGE_PRESENT it should be safe to restore the original pte_special behaviour. Signed-off-by:
-
Mel Gorman authored
Faults on the huge zero page are pointless and there is a BUG_ON to catch them during fault time. This patch reintroduces a check that avoids marking the zero page PAGE_NONE. Signed-off-by:
-
Mel Gorman authored
This patch removes the NUMA PTE bits and associated helpers. As a side-effect it increases the maximum possible swap space on x86-64. One potential source of problems is races between the marking of PTEs PROT_NONE, NUMA hinting faults and migration. It must be guaranteed that a PTE being protected is not faulted in parallel, seen as a pte_none and corrupting memory. The base case is safe but transhuge has problems in the past due to an different migration mechanism and a dependance on page lock to serialise migrations and warrants a closer look. task_work hinting update parallel fault ------------------------ -------------- change_pmd_range change_huge_pmd __pmd_trans_huge_lock pmdp_get_and_clear __handle_mm_fault pmd_none do_huge_pmd_anonymous_page read? pmd_lock blocks until hinting complete, fail !pmd_none test write? __do_huge_pmd_anonymous_page acquires pmd_lock, checks pmd_none pmd_modify set_pmd_at task_work hinting update parallel migration ------------------------ ------------------ change_pmd_range change_huge_pmd __pmd_trans_huge_lock pmdp_get_and_clear __handle_mm_fault do_huge_pmd_numa_page migrate_misplaced_transhuge_page pmd_lock waits for updates to complete, recheck pmd_same pmd_modify set_pmd_at Both of those are safe and the case where a transhuge page is inserted during a protection update is unchanged. The case where two processes try migrating at the same time is unchanged by this series so should still be ok. I could not find a case where we are accidentally depending on the PTE not being cleared and flushed. If one is missed, it'll manifest as corruption problems that start triggering shortly after this series is merged and only happen when NUMA balancing is enabled. Signed-off-by:Sasha Levin <sasha.levin@oracle.com> Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Dave Jones <davej@redhat.com> Cc: Hugh Dickins <hughd@google.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Kirill Shutemov <kirill.shutemov@linux.intel.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Rik van Riel <riel@redhat.com> Cc: Mark Brown <broonie@kernel.org> Cc: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by:
-
Mel Gorman authored
With PROT_NONE, the traditional page table manipulation functions are sufficient. [andre.przywara@arm.com: fix compiler warning in pmdp_invalidate()] [akpm@linux-foundation.org: fix build with STRICT_MM_TYPECHECKS] Signed-off-by:
Aneesh Kumar <aneesh.kumar@linux.vnet.ibm.com> Tested-by:
-
Mel Gorman authored
ppc64 should not be depending on DSISR_PROTFAULT and it's unexpected if they are triggered. This patch adds warnings just in case they are being accidentally depended upon. Signed-off-by:
-
Mel Gorman authored
Convert existing users of pte_numa and friends to the new helper. Note that the kernel is broken after this patch is applied until the other page table modifiers are also altered. This patch layout is to make review easier. Signed-off-by:
Benjamin Herrenschmidt <benh@kernel.crashing.org> Tested-by:
-
Mel Gorman authored
This is a preparatory patch that introduces protnone helpers for automatic NUMA balancing. Signed-off-by:
-
Mel Gorman authored
Automatic NUMA balancing depends on being able to protect PTEs to trap a fault and gather reference locality information. Very broadly speaking it would mark PTEs as not present and use another bit to distinguish between NUMA hinting faults and other types of faults. It was universally loved by everybody and caused no problems whatsoever. That last sentence might be a lie. This series is very heavily based on patches from Linus and Aneesh to replace the existing PTE/PMD NUMA helper functions with normal change protections. I did alter and add parts of it but I consider them relatively minor contributions. At their suggestion, acked-bys are in there but I've no problem converting them to Signed-off-by if requested. AFAIK, this has received no testing on ppc64 and I'm depending on Aneesh for that. I tested trinity under kvm-tool and passed and ran a few other basic tests. At the time of writing, only the short-lived tests have completed but testing of V2 indicated that long-term testing had no surprises. In most cases I'm leaving out detail as it's not that interesting. specjbb single JVM: There was negligible performance difference in the benchmark itself for short runs. However, system activity is higher and interrupts are much higher over time -- possibly TLB flushes. Migrations are also higher. Overall, this is more overhead but considering the problems faced with the old approach I think we just have to suck it up and find another way of reducing the overhead. specjbb multi JVM: Negligible performance difference to the actual benchmark but like the single JVM case, the system overhead is noticeably higher. Again, interrupts are a major factor. autonumabench: This was all over the place and about all that can be reasonably concluded is that it's different but not necessarily better or worse. autonumabench 3.18.0-rc5 3.18.0-rc5 mmotm-20141119 protnone-v3r3 User NUMA01 32380.24 ( 0.00%) 21642.92 ( 33.16%) User NUMA01_THEADLOCAL 22481.02 ( 0.00%) 22283.22 ( 0.88%) User NUMA02 3137.00 ( 0.00%) 3116.54 ( 0.65%) User NUMA02_SMT 1614.03 ( 0.00%) 1543.53 ( 4.37%) System NUMA01 322.97 ( 0.00%) 1465.89 (-353.88%) System NUMA01_THEADLOCAL 91.87 ( 0.00%) 49.32 ( 46.32%) System NUMA02 37.83 ( 0.00%) 14.61 ( 61.38%) System NUMA02_SMT 7.36 ( 0.00%) 7.45 ( -1.22%) Elapsed NUMA01 716.63 ( 0.00%) 599.29 ( 16.37%) Elapsed NUMA01_THEADLOCAL 553.98 ( 0.00%) 539.94 ( 2.53%) Elapsed NUMA02 83.85 ( 0.00%) 83.04 ( 0.97%) Elapsed NUMA02_SMT 86.57 ( 0.00%) 79.15 ( 8.57%) CPU NUMA01 4563.00 ( 0.00%) 3855.00 ( 15.52%) CPU NUMA01_THEADLOCAL 4074.00 ( 0.00%) 4136.00 ( -1.52%) CPU NUMA02 3785.00 ( 0.00%) 3770.00 ( 0.40%) CPU NUMA02_SMT 1872.00 ( 0.00%) 1959.00 ( -4.65%) System CPU usage of NUMA01 is worse but it's an adverse workload on this machine so I'm reluctant to conclude that it's a problem that matters. On the other workloads that are sensible on this machine, system CPU usage is great. Overall time to complete the benchmark is comparable 3.18.0-rc5 3.18.0-rc5 mmotm-20141119protnone-v3r3 User 59612.50 48586.44 System 460.22 1537.45 Elapsed 1442.20 1304.29 NUMA alloc hit 5075182 5743353 NUMA alloc miss 0 0 NUMA interleave hit 0 0 NUMA alloc local 5075174 5743339 NUMA base PTE updates 637061448 443106883 NUMA huge PMD updates 1243434 864747 NUMA page range updates 1273699656 885857347 NUMA hint faults 1658116 1214277 NUMA hint local faults 959487 754113 NUMA hint local percent 57 62 NUMA pages migrated 5467056 61676398 The NUMA pages migrated look terrible but when I looked at a graph of the activity over time I see that the massive spike in migration activity was during NUMA01. This correlates with high system CPU usage and could be simply down to bad luck but any modifications that affect that workload would be related to scan rates and migrations, not the protection mechanism. For all other workloads, migration activity was comparable. Overall, headline performance figures are comparable but the overhead is higher, mostly in interrupts. To some extent, higher overhead from this approach was anticipated but not to this degree. It's going to be necessary to reduce this again with a separate series in the future. It's still worth going ahead with this series though as it's likely to avoid constant headaches with Xen and is probably easier to maintain. This patch (of 10): A transhuge NUMA hinting fault may find the page is migrating and should wait until migration completes. The check is race-prone because the pmd is deferenced outside of the page lock and while the race is tiny, it'll be larger if the PMD is cleared while marking PMDs for hinting fault. This patch closes the race. Signed-off-by:
-
