- 09 Oct, 2012 40 commits
-
-
Sachin Kamat authored
Signed-off-by:
Sachin Kamat <sachin.kamat@linaro.org> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Mel Gorman authored
Compaction caches if a pageblock was scanned and no pages were isolated so that the pageblocks can be skipped in the future to reduce scanning. This information is not cleared by the page allocator based on activity due to the impact it would have to the page allocator fast paths. Hence there is a requirement that something clear the cache or pageblocks will be skipped forever. Currently the cache is cleared if there were a number of recent allocation failures and it has not been cleared within the last 5 seconds. Time-based decisions like this are terrible as they have no relationship to VM activity and is basically a big hammer. Unfortunately, accurate heuristics would add cost to some hot paths so this patch implements a rough heuristic. There are two cases where the cache is cleared. 1. If a !kswapd process completes a compaction cycle (migrate and free scanner meet), the zone is marked compact_blockskip_flush. When kswapd goes to sleep, it will clear the cache. This is expected to be the common case where the cache is cleared. It does not really matter if kswapd happens to be asleep or going to sleep when the flag is set as it will be woken on the next allocation request. 2. If there have been multiple failures recently and compaction just finished being deferred then a process will clear the cache and start a full scan. This situation happens if there are multiple high-order allocation requests under heavy memory pressure. The clearing of the PG_migrate_skip bits and other scans is inherently racy but the race is harmless. For allocations that can fail such as THP, they will simply fail. For requests that cannot fail, they will retry the allocation. Tests indicated that scanning rates were roughly similar to when the time-based heuristic was used and the allocation success rates were similar. Signed-off-by:
Mel Gorman <mgorman@suse.de> Cc: Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Cc: Rafael Aquini <aquini@redhat.com> Signed-off-by:
-
Mel Gorman authored
This is almost entirely based on Rik's previous patches and discussions with him about how this might be implemented. Order > 0 compaction stops when enough free pages of the correct page order have been coalesced. When doing subsequent higher order allocations, it is possible for compaction to be invoked many times. However, the compaction code always starts out looking for things to compact at the start of the zone, and for free pages to compact things to at the end of the zone. This can cause quadratic behaviour, with isolate_freepages starting at the end of the zone each time, even though previous invocations of the compaction code already filled up all free memory on that end of the zone. This can cause isolate_freepages to take enormous amounts of CPU with certain workloads on larger memory systems. This patch caches where the migration and free scanner should start from on subsequent compaction invocations using the pageblock-skip information. When compaction starts it begins from the cached restart points and will update the cached restart points until a page is isolated or a pageblock is skipped that would have been scanned by synchronous compaction. Signed-off-by:
Rik van Riel <riel@redhat.com> Cc: Richard Davies <richard@arachsys.com> Cc: Shaohua Li <shli@kernel.org> Cc: Avi Kivity <avi@redhat.com> Acked-by:
Rafael Aquini <aquini@redhat.com> Cc: Fengguang Wu <fengguang.wu@intel.com> Signed-off-by:
-
Mel Gorman authored
When compaction was implemented it was known that scanning could potentially be excessive. The ideal was that a counter be maintained for each pageblock but maintaining this information would incur a severe penalty due to a shared writable cache line. It has reached the point where the scanning costs are a serious problem, particularly on long-lived systems where a large process starts and allocates a large number of THPs at the same time. Instead of using a shared counter, this patch adds another bit to the pageblock flags called PG_migrate_skip. If a pageblock is scanned by either migrate or free scanner and 0 pages were isolated, the pageblock is marked to be skipped in the future. When scanning, this bit is checked before any scanning takes place and the block skipped if set. The main difficulty with a patch like this is "when to ignore the cached information?" If it's ignored too often, the scanning rates will still be excessive. If the information is too stale then allocations will fail that might have otherwise succeeded. In this patch o CMA always ignores the information o If the migrate and free scanner meet then the cached information will be discarded if it's at least 5 seconds since the last time the cache was discarded o If there are a large number of allocation failures, discard the cache. The time-based heuristic is very clumsy but there are few choices for a better event. Depending solely on multiple allocation failures still allows excessive scanning when THP allocations are failing in quick succession due to memory pressure. Waiting until memory pressure is relieved would cause compaction to continually fail instead of using reclaim/compaction to try allocate the page. The time-based mechanism is clumsy but a better option is not obvious. Signed-off-by:
-
Mel Gorman authored
This reverts commit 7db8889a ("mm: have order > 0 compaction start off where it left") and commit de74f1cc ("mm: have order > 0 compaction start near a pageblock with free pages"). These patches were a good idea and tests confirmed that they massively reduced the amount of scanning but the implementation is complex and tricky to understand. A later patch will cache what pageblocks should be skipped and reimplements the concept of compact_cached_free_pfn on top for both migration and free scanners. Signed-off-by:
Minchan Kim <minchan@kernel.org> Signed-off-by:
-
Mel Gorman authored
Compaction's free scanner acquires the zone->lock when checking for PageBuddy pages and isolating them. It does this even if there are no PageBuddy pages in the range. This patch defers acquiring the zone lock for as long as possible. In the event there are no free pages in the pageblock then the lock will not be acquired at all which reduces contention on zone->lock. Signed-off-by:
Peter Ujfalusi <peter.ujfalusi@ti.com> Signed-off-by:
-
Mel Gorman authored
Richard Davies and Shaohua Li have both reported lock contention problems in compaction on the zone and LRU locks as well as significant amounts of time being spent in compaction. This series aims to reduce lock contention and scanning rates to reduce that CPU usage. Richard reported at https://lkml.org/lkml/2012/9/21/91 that this series made a big different to a problem he reported in August: http://marc.info/?l=kvm&m=134511507015614&w=2 Patch 1 defers acquiring the zone->lru_lock as long as possible. Patch 2 defers acquiring the zone->lock as lock as possible. Patch 3 reverts Rik's "skip-free" patches as the core concept gets reimplemented later and the remaining patches are easier to understand if this is reverted first. Patch 4 adds a pageblock-skip bit to the pageblock flags to cache what pageblocks should be skipped by the migrate and free scanners. This drastically reduces the amount of scanning compaction has to do. Patch 5 reimplements something similar to Rik's idea except it uses the pageblock-skip information to decide where the scanners should restart from and does not need to wrap around. I tested this on 3.6-rc6 + linux-next/akpm. Kernels tested were akpm-20120920 3.6-rc6 + linux-next/akpm as of Septeber 20th, 2012 lesslock Patches 1-6 revert Patches 1-7 cachefail Patches 1-8 skipuseless Patches 1-9 Stress high-order allocation tests looked ok. Success rates are more or less the same with the full series applied but there is an expectation that there is less opportunity to race with other allocation requests if there is less scanning. The time to complete the tests did not vary that much and are uninteresting as were the vmstat statistics so I will not present them here. Using ftrace I recorded how much scanning was done by compaction and got this 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 3.6.0-rc6 akpm-20120920 lockless revert-v2r2 cachefail skipuseless Total free scanned 360753976 515414028 565479007 17103281 18916589 Total free isolated 2852429 3597369 4048601 670493 727840 Total free efficiency 0.0079% 0.0070% 0.0072% 0.0392% 0.0385% Total migrate scanned 247728664 822729112 1004645830 17946827 14118903 Total migrate isolated 2555324 3245937 3437501 616359 658616 Total migrate efficiency 0.0103% 0.0039% 0.0034% 0.0343% 0.0466% The efficiency is worthless because of the nature of the test and the number of failures. The really interesting point as far as this patch series is concerned is the number of pages scanned. Note that reverting Rik's patches massively increases the number of pages scanned indicating that those patches really did make a difference to CPU usage. However, caching what pageblocks should be skipped has a much higher impact. With patches 1-8 applied, free page and migrate page scanning are both reduced by 95% in comparison to the akpm kernel. If the basic concept of Rik's patches are implemened on top then scanning then the free scanner barely changed but migrate scanning was further reduced. That said, tests on 3.6-rc5 indicated that the last patch had greater impact than what was measured here so it is a bit variable. One way or the other, this series has a large impact on the amount of scanning compaction does when there is a storm of THP allocations. This patch: Compaction's migrate scanner acquires the zone->lru_lock when scanning a range of pages looking for LRU pages to acquire. It does this even if there are no LRU pages in the range. If multiple processes are compacting then this can cause severe locking contention. To make matters worse commit b2eef8c0 ("mm: compaction: minimise the time IRQs are disabled while isolating pages for migration") releases the lru_lock every SWAP_CLUSTER_MAX pages that are scanned. This patch makes two changes to how the migrate scanner acquires the LRU lock. First, it only releases the LRU lock every SWAP_CLUSTER_MAX pages if the lock is contended. This reduces the number of times it unnecessarily disables and re-enables IRQs. The second is that it defers acquiring the LRU lock for as long as possible. If there are no LRU pages or the only LRU pages are transhuge then the LRU lock will not be acquired at all which reduces contention on zone->lru_lock. [minchan@kernel.org: augment comment] [akpm@linux-foundation.org: tweak comment text] Signed-off-by:
-
Mel Gorman authored
Parameters were added without documentation, tut tut. Signed-off-by:
-
Mel Gorman authored
Commit c67fe375 ("mm: compaction: Abort async compaction if locks are contended or taking too long") addressed a lock contention problem in compaction by introducing compact_checklock_irqsave() that effecively aborting async compaction in the event of compaction. To preserve existing behaviour it also moved a fatal_signal_pending() check into compact_checklock_irqsave() but that is very misleading. It "hides" the check within a locking function but has nothing to do with locking as such. It just happens to work in a desirable fashion. This patch moves the fatal_signal_pending() check to isolate_migratepages_range() where it belongs. Arguably the same check should also happen when isolating pages for freeing but it's overkill. Signed-off-by:
-
Shaohua Li authored
isolate_migratepages_range() might isolate no pages if for example when zone->lru_lock is contended and running asynchronous compaction. In this case, we should abort compaction, otherwise, compact_zone will run a useless loop and make zone->lru_lock is even contended. An additional check is added to ensure that cc.migratepages and cc.freepages get properly drained whan compaction is aborted. [minchan@kernel.org: Putback pages isolated for migration if aborting] [akpm@linux-foundation.org: compact_zone_order requires non-NULL arg contended] [akpm@linux-foundation.org: make compact_zone_order() require non-NULL arg `contended'] [minchan@kernel.org: Putback pages isolated for migration if aborting] Signed-off-by:
Andrea Arcangeli <aarcange@redhat.com> Signed-off-by:
Shaohua Li <shli@fusionio.com> Signed-off-by:
-
Wanpeng Li authored
Commit 0ee332c1 ("memblock: Kill early_node_map[]") removed early_node_map[]. Clean up the comments to comply with that change. Signed-off-by:
Wanpeng Li <liwanp@linux.vnet.ibm.com> Cc: Michal Hocko <mhocko@suse.cz> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Gavin Shan <shangw@linux.vnet.ibm.com> Cc: Yinghai Lu <yinghai@kernel.org> Signed-off-by:
-
Wanpeng Li authored
Use the existing interface function to set the NUMA node ID (NID) for the regions, either memory or reserved region. Signed-off-by:
-
Shaohua Li authored
.fault now can retry. The retry can break state machine of .fault. In filemap_fault, if page is miss, ra->mmap_miss is increased. In the second try, since the page is in page cache now, ra->mmap_miss is decreased. And these are done in one fault, so we can't detect random mmap file access. Add a new flag to indicate .fault is tried once. In the second try, skip ra->mmap_miss decreasing. The filemap_fault state machine is ok with it. I only tested x86, didn't test other archs, but looks the change for other archs is obvious, but who knows :) Signed-off-by:
Shaohua Li <shaohua.li@fusionio.com> Cc: Rik van Riel <riel@redhat.com> Cc: Wu Fengguang <fengguang.wu@intel.com> Signed-off-by:
-
Shaohua Li authored
The x86 implementation of atomic_dec_if_positive is quite generic, so make it available to all architectures. This is needed for "swap: add a simple detector for inappropriate swapin readahead". [akpm@linux-foundation.org: do the "#define foo foo" trick in the conventional manner] Signed-off-by:
-
Minchan Kim authored
If race between allocation and isolation in memory-hotplug offline happens, some pages could be in MIGRATE_MOVABLE of free_list although the pageblock's migratetype of the page is MIGRATE_ISOLATE. The race could be detected by get_freepage_migratetype in __test_page_isolated_in_pageblock. If it is detected, now EBUSY gets bubbled all the way up and the hotplug operations fails. But better idea is instead of returning and failing memory-hotremove, move the free page to the correct list at the time it is detected. It could enhance memory-hotremove operation success ratio although the race is really rare. Suggested by Mel Gorman. [akpm@linux-foundation.org: small cleanup] Signed-off-by:
Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com> Acked-by:
-
Minchan Kim authored
Like below, memory-hotplug makes race between page-isolation and page-allocation so it can hit BUG_ON in __offline_isolated_pages. CPU A CPU B start_isolate_page_range set_migratetype_isolate spin_lock_irqsave(zone->lock) free_hot_cold_page(Page A) /* without zone->lock */ migratetype = get_pageblock_migratetype(Page A); /* * Page could be moved into MIGRATE_MOVABLE * of per_cpu_pages */ list_add_tail(&page->lru, &pcp->lists[migratetype]); set_pageblock_isolate move_freepages_block drain_all_pages /* Page A could be in MIGRATE_MOVABLE of free_list. */ check_pages_isolated __test_page_isolated_in_pageblock /* * We can't catch freed page which * is free_list[MIGRATE_MOVABLE] */ if (PageBuddy(page A)) pfn += 1 << page_order(page A); /* So, Page A could be allocated */ __offline_isolated_pages /* * BUG_ON hit or offline page * which is used by someone */ BUG_ON(!PageBuddy(page A)); This patch checks page's migratetype in freelist in __test_page_isolated_in_pageblock. So now __test_page_isolated_in_pageblock can check the page caused by above race and can fail of memory offlining. Signed-off-by:
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Reviewed-by:
-
Minchan Kim authored
The page allocator caches the pageblock information in page->private while it is in the PCP freelists but this is overwritten with the order of the page when freed to the buddy allocator. This patch stores the migratetype of the page in the page->index field so that it is available at all times when the page remain in free_list. This patch adds a new call site in __free_pages_ok so it might be overhead a bit but it's for high order allocation. So I believe damage isn't hurt. Signed-off-by:
-
Minchan Kim authored
The page allocator uses set_page_private and page_private for handling migratetype when it frees page. Let's replace them with [set|get] _freepage_migratetype to make it more clear. Signed-off-by:
-
Bartlomiej Zolnierkiewicz authored
* Add ALLOC_CMA alloc flag and pass it to [__]zone_watermark_ok() (from Minchan Kim). * During watermark check decrease available free pages number by free CMA pages number if necessary (unmovable allocations cannot use pages from CMA areas). Signed-off-by:
Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com> Signed-off-by:
Kyungmin Park <kyungmin.park@samsung.com> Cc: Marek Szyprowski <m.szyprowski@samsung.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mel Gorman <mgorman@suse.de> Cc: Hugh Dickins <hughd@google.com> Signed-off-by:
-
Bartlomiej Zolnierkiewicz authored
Add NR_FREE_CMA_PAGES counter to be later used for checking watermark in __zone_watermark_ok(). For simplicity and to avoid #ifdef hell make this counter always available (not only when CONFIG_CMA=y). [akpm@linux-foundation.org: use conventional migratetype naming] Signed-off-by:
-
Bartlomiej Zolnierkiewicz authored
Isolated free pages shouldn't be accounted to NR_FREE_PAGES counter. Fix it by properly decreasing/increasing NR_FREE_PAGES counter in set_migratetype_isolate()/unset_migratetype_isolate() and removing counter adjustment for isolated pages from free_one_page() and split_free_page(). Signed-off-by:
-
Bartlomiej Zolnierkiewicz authored
page->private gets re-used in __free_one_page() to store page order (so trace_mm_page_pcpu_drain() may print order instead of migratetype) thus migratetype value must be cached locally. Fixes regression introduced in commit a7016235 ("mm: fix migratetype bug which slowed swapping"). This caused incorrect data to be attached to the mm_page_pcpu_drain trace event. [akpm@linux-foundation.org: add comment] Cc: Marek Szyprowski <m.szyprowski@samsung.com> Cc: Michal Nazarewicz <mina86@mina86.com> Acked-by:
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Signed-off-by:
-
Minchan Kim authored
Drop clean cache pages instead of migration during alloc_contig_range() to minimise allocation latency by reducing the amount of migration that is necessary. It's useful for CMA because latency of migration is more important than evicting the background process's working set. In addition, as pages are reclaimed then fewer free pages for migration targets are required so it avoids memory reclaiming to get free pages, which is a contributory factor to increased latency. I measured elapsed time of __alloc_contig_migrate_range() which migrates 10M in 40M movable zone in QEMU machine. Before - 146ms, After - 7ms [akpm@linux-foundation.org: fix nommu build] Signed-off-by:
Michal Nazarewicz <mina86@mina86.com> Cc: Rik van Riel <riel@redhat.com> Tested-by:
-
Andrea Arcangeli authored
The variable must be static especially given the variable name. s/RCU/SRCU/ over a few comments. Signed-off-by:
-
Xishi Qiu authored
online_pages() does build_all_zonelists() and zone_pcp_update(), I think offline_pages() should do it too. When the zone has no memory to allocate, remove it from other nodes' zonelists. zone_batchsize() depends on zone's present pages, if zone's present pages are changed, zone's pcp should be updated. Signed-off-by:
Xishi Qiu <qiuxishi@huawei.com> Cc: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com> Signed-off-by:
-
Michel Lespinasse authored
During mremap(), the destination VMA is generally placed after the original vma in rmap traversal order: in move_vma(), we always have new_pgoff >= vma->vm_pgoff, and as a result new_vma->vm_pgoff >= vma->vm_pgoff unless vma_merge() merged the new vma with an adjacent one. When the destination VMA is placed after the original in rmap traversal order, we can avoid taking the rmap locks in move_ptes(). Essentially, this reintroduces the optimization that had been disabled in "mm anon rmap: remove anon_vma_moveto_tail". The difference is that we don't try to impose the rmap traversal order; instead we just rely on things being in the desired order in the common case and fall back to taking locks in the uncommon case. Also we skip the i_mmap_mutex in addition to the anon_vma lock: in both cases, the vmas are traversed in increasing vm_pgoff order with ties resolved in tree insertion order. Signed-off-by:
Michel Lespinasse <walken@google.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Rik van Riel <riel@redhat.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Daniel Santos <daniel.santos@pobox.com> Cc: Hugh Dickins <hughd@google.com> Signed-off-by:
-
Michel Lespinasse authored
anon_vma_clone() expects new_vma->vm_{start,end,pgoff} to be correctly set so that the new vma can be indexed on the anon interval tree. copy_vma() was failing to do that, which broke mremap(). Signed-off-by:Sasha Levin <levinsasha928@gmail.com> Signed-off-by:
-
Michel Lespinasse authored
Add a CONFIG_DEBUG_VM_RB build option for the previously existing DEBUG_MM_RB code. Now that Andi Kleen modified it to avoid using recursive algorithms, we can expose it a bit more. Also extend this code to validate_mm() after stack expansion, and to check that the vma's start and last pgoffs have not changed since the nodes were inserted on the anon vma interval tree (as it is important that the nodes be reindexed after each such update). Signed-off-by:
-
Michel Lespinasse authored
In file and anon rmap, we use interval trees to find potentially relevant vmas and then call vma_address() to find the virtual address the given page might be found at in these vmas. vma_address() used to include a check that the returned address falls within the limits of the vma, but this check isn't necessary now that we always use interval trees in rmap: the interval tree just doesn't return any vmas which this check would find to be irrelevant. As a result, we can replace the use of -EFAULT error code (which then needed to be checked in every call site) with a VM_BUG_ON(). Signed-off-by:
-
Michel Lespinasse authored
When a large VMA (anon or private file mapping) is first touched, which will populate its anon_vma field, and then split into many regions through the use of mprotect(), the original anon_vma ends up linking all of the vmas on a linked list. This can cause rmap to become inefficient, as we have to walk potentially thousands of irrelevent vmas before finding the one a given anon page might fall into. By replacing the same_anon_vma linked list with an interval tree (where each avc's interval is determined by its vma's start and last pgoffs), we can make rmap efficient for this use case again. While the change is large, all of its pieces are fairly simple. Most places that were walking the same_anon_vma list were looking for a known pgoff, so they can just use the anon_vma_interval_tree_foreach() interval tree iterator instead. The exception here is ksm, where the page's index is not known. It would probably be possible to rework ksm so that the index would be known, but for now I have decided to keep things simple and just walk the entirety of the interval tree there. When updating vma's that already have an anon_vma assigned, we must take care to re-index the corresponding avc's on their interval tree. This is done through the use of anon_vma_interval_tree_pre_update_vma() and anon_vma_interval_tree_post_update_vma(), which remove the avc's from their interval tree before the update and re-insert them after the update. The anon_vma stays locked during the update, so there is no chance that rmap would miss the vmas that are being updated. Signed-off-by:
-
Michel Lespinasse authored
mremap() had a clever optimization where move_ptes() did not take the anon_vma lock to avoid a race with anon rmap users such as page migration. Instead, the avc's were ordered in such a way that the origin vma was always visited by rmap before the destination. This ordering and the use of page table locks rmap usage safe. However, we want to replace the use of linked lists in anon rmap with an interval tree, and this will make it harder to impose such ordering as the interval tree will always be sorted by the avc->vma->vm_pgoff value. For now, let's replace the anon_vma_moveto_tail() ordering function with proper anon_vma locking in move_ptes(). Once we have the anon interval tree in place, we will re-introduce an optimization to avoid taking these locks in the most common cases. Signed-off-by:
-
Michel Lespinasse authored
Update the generic interval tree code that was introduced in "mm: replace vma prio_tree with an interval tree". Changes: - fixed 'endpoing' typo noticed by Andrew Morton - replaced include/linux/interval_tree_tmpl.h, which was used as a template (including it automatically defined the interval tree functions) with include/linux/interval_tree_generic.h, which only defines a preprocessor macro INTERVAL_TREE_DEFINE(), which itself defines the interval tree functions when invoked. Now that is a very long macro which is unfortunate, but it does make the usage sites (lib/interval_tree.c and mm/interval_tree.c) a bit nicer than previously. - make use of RB_DECLARE_CALLBACKS() in the INTERVAL_TREE_DEFINE() macro, instead of duplicating that code in the interval tree template. - replaced vma_interval_tree_add(), which was actually handling the nonlinear and interval tree cases, with vma_interval_tree_insert_after() which handles only the interval tree case and has an API that is more consistent with the other interval tree handling functions. The nonlinear case is now handled explicitly in kernel/fork.c dup_mmap(). Signed-off-by:
-
Michel Lespinasse authored
Provide rb_insert_augmented() and rb_erase_augmented() through a new rbtree_augmented.h include file. rb_erase_augmented() is defined there as an __always_inline function, in order to allow inlining of augmented rbtree callbacks into it. Since this generates a relatively large function, each augmented rbtree user should make sure to have a single call site. Signed-off-by:
-
Michel Lespinasse authored
After both prio_tree users have been converted to use red-black trees, there is no need to keep around the prio tree library anymore. Signed-off-by:
-
Michel Lespinasse authored
kmemleak uses a tree where each node represents an allocated memory object in order to quickly find out what object a given address is part of. However, the objects don't overlap, so rbtrees are a better choice than prio tree for this use. They are both faster and have lower memory overhead. Tested by booting a kernel with kmemleak enabled, loading the kmemleak_test module, and looking for the expected messages. Signed-off-by:
Catalin Marinas <catalin.marinas@arm.com> Tested-by:
-
Michel Lespinasse authored
Implement an interval tree as a replacement for the VMA prio_tree. The algorithms are similar to lib/interval_tree.c; however that code can't be directly reused as the interval endpoints are not explicitly stored in the VMA. So instead, the common algorithm is moved into a template and the details (node type, how to get interval endpoints from the node, etc) are filled in using the C preprocessor. Once the interval tree functions are available, using them as a replacement to the VMA prio tree is a relatively simple, mechanical job. Signed-off-by:
-
Michel Lespinasse authored
Patch 1 implements support for interval trees, on top of the augmented rbtree API. It also adds synthetic tests to compare the performance of interval trees vs prio trees. Short answers is that interval trees are slightly faster (~25%) on insert/erase, and much faster (~2.4 - 3x) on search. It is debatable how realistic the synthetic test is, and I have not made such measurements yet, but my impression is that interval trees would still come out faster. Patch 2 uses a preprocessor template to make the interval tree generic, and uses it as a replacement for the vma prio_tree. Patch 3 takes the other prio_tree user, kmemleak, and converts it to use a basic rbtree. We don't actually need the augmented rbtree support here because the intervals are always non-overlapping. Patch 4 removes the now-unused prio tree library. Patch 5 proposes an additional optimization to rb_erase_augmented, now providing it as an inline function so that the augmented callbacks can be inlined in. This provides an additional 5-10% performance improvement for the interval tree insert/erase benchmark. There is a maintainance cost as it exposes augmented rbtree users to some of the rbtree library internals; however I think this cost shouldn't be too high as I expect the augmented rbtree will always have much less users than the base rbtree. I should probably add a quick summary of why I think it makes sense to replace prio trees with augmented rbtree based interval trees now. One of the drivers is that we need augmented rbtrees for Rik's vma gap finding code, and once you have them, it just makes sense to use them for interval trees as well, as this is the simpler and more well known algorithm. prio trees, in comparison, seem *too* clever: they impose an additional 'heap' constraint on the tree, which they use to guarantee a faster worst-case complexity of O(k+log N) for stabbing queries in a well-balanced prio tree, vs O(k*log N) for interval trees (where k=number of matches, N=number of intervals). Now this sounds great, but in practice prio trees don't realize this theorical benefit. First, the additional constraint makes them harder to update, so that the kernel implementation has to simplify things by balancing them like a radix tree, which is not always ideal. Second, the fact that there are both index and heap properties makes both tree manipulation and search more complex, which results in a higher multiplicative time constant. As it turns out, the simple interval tree algorithm ends up running faster than the more clever prio tree. This patch: Add two test modules: - prio_tree_test measures the performance of lib/prio_tree.c, both for insertion/removal and for stabbing searches - interval_tree_test measures the performance of a library of equivalent functionality, built using the augmented rbtree support. In order to support the second test module, lib/interval_tree.c is introduced. It is kept separate from the interval_tree_test main file for two reasons: first we don't want to provide an unfair advantage over prio_tree_test by having everything in a single compilation unit, and second there is the possibility that the interval tree functionality could get some non-test users in kernel over time. Signed-off-by:
-
Michel Lespinasse authored
As proposed by Peter Zijlstra, this makes it easier to define the augmented rbtree callbacks. Signed-off-by:
-
Michel Lespinasse authored
convert arch/x86/mm/pat_rbtree.c to the proposed augmented rbtree api and remove the old augmented rbtree implementation. Signed-off-by:
-
Michel Lespinasse authored
Introduce new augmented rbtree APIs that allow minimal recalculation of augmented node information. A new callback is added to the rbtree insertion and erase rebalancing functions, to be called on each tree rotations. Such rotations preserve the subtree's root augmented value, but require recalculation of the one child that was previously located at the subtree root. In the insertion case, the handcoded search phase must be updated to maintain the augmented information on insertion, and then the rbtree coloring/rebalancing algorithms keep it up to date. In the erase case, things are more complicated since it is library code that manipulates the rbtree in order to remove internal nodes. This requires a couple additional callbacks to copy a subtree's augmented value when a new root is stitched in, and to recompute augmented values down the ancestry path when a node is removed from the tree. In order to preserve maximum speed for the non-augmented case, we provide two versions of each tree manipulation function. rb_insert_augmented() is the augmented equivalent of rb_insert_color(), and rb_erase_augmented() is the augmented equivalent of rb_erase(). Signed-off-by:
-
