- 15 Apr, 2015 40 commits
-
-
Minchan Kim authored
Recently, we started to use zram heavily and some of issues popped. 1) external fragmentation I got a report from Juneho Choi that fork failed although there are plenty of free pages in the system. His investigation revealed zram is one of the culprit to make heavy fragmentation so there was no more contiguous 16K page for pgd to fork in the ARM. 2) non-movable pages Other problem of zram now is that inherently, user want to use zram as swap in small memory system so they use zRAM with CMA to use memory efficiently. However, unfortunately, it doesn't work well because zRAM cannot use CMA's movable pages unless it doesn't support compaction. I got several reports about that OOM happened with zram although there are lots of swap space and free space in CMA area. 3) internal fragmentation zRAM has started support memory limitation feature to limit memory usage and I sent a patchset(https://lkml.org/lkml/2014/9/21/148) for VM to be harmonized with zram-swap to stop anonymous page reclaim if zram consumed memory up to the limit although there are free space on the swap. One problem for that direction is zram has no way to know any hole in memory space zsmalloc allocated by internal fragmentation so zram would regard swap is full although there are free space in zsmalloc. For solving the issue, zram want to trigger compaction of zsmalloc before it decides full or not. This patchset is first step to support above issues. For that, it adds indirect layer between handle and object location and supports manual compaction to solve 3th problem first of all. After this patchset got merged, next step is to make VM aware of zsmalloc compaction so that generic compaction will move zsmalloced-pages automatically in runtime. In my imaginary experiment(ie, high compress ratio data with heavy swap in/out on 8G zram-swap), data is as follows, Before = zram allocated object : 60212066 bytes zram total used: 140103680 bytes ratio: 42.98 percent MemFree: 840192 kB Compaction After = frag ratio after compaction zram allocated object : 60212066 bytes zram total used: 76185600 bytes ratio: 79.03 percent MemFree: 901932 kB Juneho reported below in his real platform with small aging. So, I think the benefit would be bigger in real aging system for a long time. - frag_ratio increased 3% (ie, higher is better) - memfree increased about 6MB - In buddy info, Normal 2^3: 4, 2^2: 1: 2^1 increased, Highmem: 2^1 21 increased frag ratio after swap fragment used : 156677 kbytes total: 166092 kbytes frag_ratio : 94 meminfo before compaction MemFree: 83724 kB Node 0, zone Normal 13642 1364 57 10 61 17 9 5 4 0 0 Node 0, zone HighMem 425 29 1 0 0 0 0 0 0 0 0 num_migrated : 23630 compaction done frag ratio after compaction used : 156673 kbytes total: 160564 kbytes frag_ratio : 97 meminfo after compaction MemFree: 89060 kB Node 0, zone Normal 14076 1544 67 14 61 17 9 5 4 0 0 Node 0, zone HighMem 863 50 1 0 0 0 0 0 0 0 0 This patchset adds more logics(about 480 lines) in zsmalloc but when I tested heavy swapin/out program, the regression for swapin/out speed is marginal because most of overheads were caused by compress/decompress and other MM reclaim stuff. This patch (of 7): Currently, handle of zsmalloc encodes object's location directly so it makes support of migration hard. This patch decouples handle and object via adding indirect layer. For that, it allocates handle dynamically and returns it to user. The handle is the address allocated by slab allocation so it's unique and we could keep object's location in the memory space allocated for handle. With it, we can change object's position without changing handle itself. Signed-off-by:
Minchan Kim <minchan@kernel.org> Cc: Juneho Choi <juno.choi@lge.com> Cc: Gunho Lee <gunho.lee@lge.com> Cc: Luigi Semenzato <semenzato@google.com> Cc: Dan Streetman <ddstreet@ieee.org> Cc: Seth Jennings <sjennings@variantweb.net> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchan@redhat.com> Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Mel Gorman <mel@csn.ul.ie> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Andrew Morton authored
mm/compaction.c:250:13: warning: 'suitable_migration_target' defined but not used [-Wunused-function] Reported-by:
Fengguang Wu <fengguang.wu@gmail.com> Cc: Vlastimil Babka <vbabka@suse.cz> Signed-off-by:
-
Boaz Harrosh authored
The original dax patchset split the ext2/4_file_operations because of the two NULL splice_read/splice_write in the dax case. In the vfs if splice_read/splice_write are NULL we then call default_splice_read/write. What we do here is make generic_file_splice_read aware of IS_DAX() so the original ext2/4_file_operations can be used as is. For write it appears that iter_file_splice_write is just fine. It uses the regular f_op->write(file,..) or new_sync_write(file, ...). Signed-off-by:
Boaz Harrosh <boaz@plexistor.com> Reviewed-by:
Jan Kara <jack@suse.cz> Cc: Dave Chinner <dchinner@redhat.com> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Signed-off-by:
-
Boaz Harrosh authored
From: Yigal Korman <yigal@plexistor.com> [v1] Without this patch, c/mtime is not updated correctly when mmap'ed page is first read from and then written to. A new xfstest is submitted for testing this (generic/080) [v2] Jan Kara has pointed out that if we add the sb_start/end_pagefault pair in the new pfn_mkwrite we are then fixing another bug where: A user could start writing to the page while filesystem is frozen. Signed-off-by:
Yigal Korman <yigal@plexistor.com> Signed-off-by:
-
Boaz Harrosh authored
This will allow FS that uses VM_PFNMAP | VM_MIXEDMAP (no page structs) to get notified when access is a write to a read-only PFN. This can happen if we mmap() a file then first mmap-read from it to page-in a read-only PFN, than we mmap-write to the same page. We need this functionality to fix a DAX bug, where in the scenario above we fail to set ctime/mtime though we modified the file. An xfstest is attached to this patchset that shows the failure and the fix. (A DAX patch will follow) This functionality is extra important for us, because upon dirtying of a pmem page we also want to RDMA the page to a remote cluster node. We define a new pfn_mkwrite and do not reuse page_mkwrite because 1 - The name ;-) 2 - But mainly because it would take a very long and tedious audit of all page_mkwrite functions of VM_MIXEDMAP/VM_PFNMAP users. To make sure they do not now CRASH. For example current DAX code (which this is for) would crash. If we would want to reuse page_mkwrite, We will need to first patch all users, so to not-crash-on-no-page. Then enable this patch. But even if I did that I would not sleep so well at night. Adding a new vector is the safest thing to do, and is not that expensive. an extra pointer at a static function vector per driver. Also the new vector is better for performance, because else we Will call all current Kernel vectors, so to: check-ha-no-page-do-nothing and return. No need to call it from do_shared_fault because do_wp_page is called to change pte permissions anyway. Signed-off-by:Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Matthew Wilcox <matthew.r.wilcox@intel.com> Cc: Jan Kara <jack@suse.cz> Cc: Hugh Dickins <hughd@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Dave Chinner <david@fromorbit.com> Signed-off-by:
-
Konstantin Khlebnikov authored
A lot of filesystems use generic_file_mmap() and filemap_fault(), f_op->mmap and vm_ops->fault aren't enough to identify filesystem. This prints file name, vm_ops->fault, f_op->mmap and a_ops->readpage (which is almost always implemented and filesystem-specific). Example: [ 23.676410] BUG: Bad page map in process sh pte:1b7e6025 pmd:19bbd067 [ 23.676887] page:ffffea00006df980 count:4 mapcount:1 mapping:ffff8800196426c0 index:0x97 [ 23.677481] flags: 0x10000000000000c(referenced|uptodate) [ 23.677896] page dumped because: bad pte [ 23.678205] addr:00007f52fcb17000 vm_flags:00000075 anon_vma: (null) mapping:ffff8800196426c0 index:97 [ 23.678922] file:libc-2.19.so fault:filemap_fault mmap:generic_file_readonly_mmap readpage:v9fs_vfs_readpage [akpm@linux-foundation.org: use pr_alert, per Kirill] Signed-off-by:
Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Cc: Sasha Levin <sasha.levin@oracle.com> Acked-by:
Kirill A. Shutemov <kirill@shutemov.name> Signed-off-by:
-
Andrey Ryabinin authored
Mempools keep allocated objects in reserved for situations when ordinary allocation may not be possible to satisfy. These objects shouldn't be accessed before they leave the pool. This patch poison elements when get into the pool and unpoison when they leave it. This will let KASan to detect use-after-free of mempool's elements. Signed-off-by:
Andrey Ryabinin <a.ryabinin@samsung.com> Tested-by:
David Rientjes <rientjes@google.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Dmitry Chernenkov <drcheren@gmail.com> Cc: Dmitry Vyukov <dvyukov@google.com> Cc: Alexander Potapenko <glider@google.com> Signed-off-by:
-
Andrew Morton authored
Like EXPORT_SYMBOL(): the positioning communicates that the macro pertains to the immediately preceding function. Cc: Dmitry Safonov <d.safonov@partner.samsung.com> Cc: Michal Nazarewicz <mina86@mina86.com> Cc: Stefan Strogin <stefan.strogin@gmail.com> Cc: Marek Szyprowski <m.szyprowski@samsung.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Pintu Kumar <pintu.k@samsung.com> Cc: Weijie Yang <weijie.yang@samsung.com> Cc: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com> Cc: Vyacheslav Tyrtov <v.tyrtov@samsung.com> Cc: Aleksei Mateosian <a.mateosian@samsung.com> Signed-off-by:
-
Dmitry Safonov authored
Here are two functions that provide interface to compute/get used size and size of biggest free chunk in cma region. Add that information to debugfs. [akpm@linux-foundation.org: move debug code from cma.c into cma_debug.c] [stefan.strogin@gmail.com: move code from cma_get_used() and cma_get_maxchunk() to cma_used_get() and cma_maxchunk_get()] Signed-off-by:
Dmitry Safonov <d.safonov@partner.samsung.com> Signed-off-by:
Stefan Strogin <stefan.strogin@gmail.com> Acked-by:
Michal Nazarewicz <mina86@mina86.com> Cc: Marek Szyprowski <m.szyprowski@samsung.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Pintu Kumar <pintu.k@samsung.com> Cc: Weijie Yang <weijie.yang@samsung.com> Cc: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com> Cc: Vyacheslav Tyrtov <v.tyrtov@samsung.com> Cc: Aleksei Mateosian <a.mateosian@samsung.com> Signed-off-by:
-
Kirill A. Shutemov authored
Few trivial cleanups: - no need to call set_recommended_min_free_kbytes() from late_initcall() -- start_khugepaged() calls it; - no need to call set_recommended_min_free_kbytes() from start_khugepaged() if khugepaged is not started; - there isn't much point in running start_khugepaged() if we've just set transparent_hugepage_flags to zero; - start_khugepaged() is misnamed -- it also used to stop the thread; Signed-off-by:
-
Kirill A. Shutemov authored
Most-used page->mapping helper -- page_mapping() -- has already uninlined. Let's uninline also page_rmapping() and page_anon_vma(). It saves us depending on configuration around 400 bytes in text: text data bss dec hex filename 660318 99254 410000 1169572 11d8a4 mm/built-in.o-before 659854 99254 410000 1169108 11d6d4 mm/built-in.o I also tried to make code a bit more clean. [akpm@linux-foundation.org: coding-style fixes] Signed-off-by:
-
Stefan Strogin authored
Add trace events for cma_alloc() and cma_release(). The cma_alloc tracepoint is used both for successful and failed allocations, in case of allocation failure pfn=-1UL is stored and printed. Signed-off-by:
-
Borislav Petkov authored
Flip the flag test so that it is the simplest. No functional change, just a small readability improvement: No code changed: # arch/x86/kernel/sys_x86_64.o: text data bss dec hex filename 1551 24 0 1575 627 sys_x86_64.o.before 1551 24 0 1575 627 sys_x86_64.o.after md5: 70708d1b1ad35cc891118a69dc1a63f9 sys_x86_64.o.before.asm 70708d1b1ad35cc891118a69dc1a63f9 sys_x86_64.o.after.asm Signed-off-by:
Borislav Petkov <bp@suse.de> Signed-off-by:
-
Alexander Kuleshov authored
memblock_reserve() calls memblock_reserve_region() which prints debugging information if 'memblock=debug' was passed on the command line. This patch adds the same behaviour, but for memblock_add function(). [akpm@linux-foundation.org: s/memblock_memory/memblock_add/ in message] Signed-off-by:
Alexander Kuleshov <kuleshovmail@gmail.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Philipp Hachtmann <phacht@linux.vnet.ibm.com> Cc: Fabian Frederick <fabf@skynet.be> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Emil Medve <Emilian.Medve@freescale.com> Cc: Akinobu Mita <akinobu.mita@gmail.com> Cc: Tang Chen <tangchen@cn.fujitsu.com> Cc: Tony Luck <tony.luck@intel.com> Signed-off-by:
-
Naoya Horiguchi authored
Now we have an easy access to hugepages' activeness, so existing helpers to get the information can be cleaned up. [akpm@linux-foundation.org: s/PageHugeActive/page_huge_active/] Signed-off-by:
Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Hugh Dickins <hughd@google.com> Reviewed-by:
Michal Hocko <mhocko@suse.cz> Cc: Mel Gorman <mgorman@suse.de> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: David Rientjes <rientjes@google.com> Signed-off-by:
-
Naoya Horiguchi authored
We are not safe from calling isolate_huge_page() on a hugepage concurrently, which can make the victim hugepage in invalid state and results in BUG_ON(). The root problem of this is that we don't have any information on struct page (so easily accessible) about hugepages' activeness. Note that hugepages' activeness means just being linked to hstate->hugepage_activelist, which is not the same as normal pages' activeness represented by PageActive flag. Normal pages are isolated by isolate_lru_page() which prechecks PageLRU before isolation, so let's do similarly for hugetlb with a new paeg_huge_active(). set/clear_page_huge_active() should be called within hugetlb_lock. But hugetlb_cow() and hugetlb_no_page() don't do this, being justified because in these functions set_page_huge_active() is called right after the hugepage is allocated and no other thread tries to isolate it. [akpm@linux-foundation.org: s/PageHugeActive/page_huge_active/, make it return bool] [fengguang.wu@intel.com: set_page_huge_active() can be static] Signed-off-by:
Fengguang Wu <fengguang.wu@intel.com> Signed-off-by:
-
Naoya Horiguchi authored
__put_compound_page() calls __page_cache_release() to do some freeing work, but it's obviously for thps, not for hugetlb. We don't care because PageLRU is always cleared and page->mem_cgroup is always NULL for hugetlb. But it's not correct and has potential risks, so let's make it conditional. Signed-off-by:
-
Rasmus Villemoes authored
The creators of the C language gave us the while keyword. Let's use that instead of synthesizing it from if+goto. Made possible by 6597d783 ("mm/mmap.c: replace find_vma_prepare() with clearer find_vma_links()"). [akpm@linux-foundation.org: fix 80-col overflows] Signed-off-by:
Rasmus Villemoes <linux@rasmusvillemoes.dk> Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com> Cc: Sasha Levin <sasha.levin@oracle.com> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Roman Gushchin <klamm@yandex-team.ru> Cc: Hugh Dickins <hughd@google.com> Signed-off-by:
-
David Rientjes authored
When MAP_HUGETLB memory is unmapped, the length must be hugepage aligned, otherwise it fails with -EINVAL. All tests currently behave correctly, but it's better to explcitly test the return value for completeness and document the requirement, especially if users copy map_hugetlb.c as a sample implementation. Signed-off-by:
Michael Ellerman <mpe@ellerman.id.au> Signed-off-by:
-
David Rientjes authored
munmap(2) of hugetlb memory requires a length that is hugepage aligned, otherwise it may fail. Add this to the documentation. This also cleans up the documentation and separates it into logical units: one part refers to MAP_HUGETLB and another part refers to requirements for shared memory segments. Signed-off-by:
Hugh Dickins <hughd@google.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Joern Engel <joern@logfs.org> Cc: Jianguo Wu <wujianguo@huawei.com> Cc: Eric B Munson <emunson@akamai.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by:
-
Kirill A. Shutemov authored
set_recommended_min_free_kbytes() adjusts zone water marks to be suitable for khugepaged. We avoid doing this if khugepaged is disabled, but don't catch the case when khugepaged is failed to start. Let's address this by checking khugepaged_thread instead of khugepaged_enabled() in set_recommended_min_free_kbytes(). It's NULL if the kernel thread is stopped or failed to start. Signed-off-by:
-
Kirill A. Shutemov authored
We miss error-handling in few cases hugepage_init(). Let's fix that. Signed-off-by:
-
David Rientjes authored
Mempools keep elements in a reserved pool for contexts in which allocation may not be possible. When an element is allocated from the reserved pool, its memory contents is the same as when it was added to the reserved pool. Because of this, elements lack any free poisoning to detect use-after-free errors. This patch adds free poisoning for elements backed by the slab allocator. This is possible because the mempool layer knows the object size of each element. When an element is added to the reserved pool, it is poisoned with POISON_FREE. When it is removed from the reserved pool, the contents are checked for POISON_FREE. If there is a mismatch, a warning is emitted to the kernel log. This is only effective for configs with CONFIG_DEBUG_SLAB or CONFIG_SLUB_DEBUG_ON. [fabio.estevam@freescale.com: use '%zu' for printing 'size_t' variable] [arnd@arndb.de: add missing include] Signed-off-by:
Fabio Estevam <fabio.estevam@freescale.com> Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
David Rientjes authored
All occurrences of mempools based on slab caches with object constructors have been removed from the tree, so disallow creating them. We can only dereference mem->ctor in mm/mempool.c without including mm/slab.h in include/linux/mempool.h. So simply note the restriction, just like the comment restricting usage of __GFP_ZERO, and warn on kernels with CONFIG_DEBUG_VM() if such a mempool is allocated from. We don't want to incur this check on every element allocation, so use VM_BUG_ON(). Signed-off-by:
-
David Rientjes authored
Mempools based on slab caches with object constructors are risky because element allocation can happen either from the slab cache itself, meaning the constructor is properly called before returning, or from the mempool reserve pool, meaning the constructor is not called before returning, depending on the allocation context. For this reason, we should disallow creating mempools based on slab caches that have object constructors. Callers of mempool_alloc() will be responsible for properly initializing the returned element. Then, it doesn't matter if the element came from the slab cache or the mempool reserved pool. The only occurrence of a mempool being based on a slab cache with an object constructor in the tree is in fs/jfs/jfs_metapage.c. Remove it and properly initialize the element in alloc_metapage(). At the same time, META_free is never used, so remove it as well. Signed-off-by:
Dave Kleikamp <dave.kleikamp@oracle.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Mikulas Patocka <mpatocka@redhat.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by:
-
Jason Low authored
We converted some of the usages of ACCESS_ONCE to READ_ONCE in the mm/ tree since it doesn't work reliably on non-scalar types. This patch removes the rest of the usages of ACCESS_ONCE, and use the new READ_ONCE API for the read accesses. This makes things cleaner, instead of using separate/multiple sets of APIs. Signed-off-by:
Jason Low <jason.low2@hp.com> Acked-by:
Davidlohr Bueso <dave@stgolabs.net> Acked-by:
Rik van Riel <riel@redhat.com> Reviewed-by:
Christian Borntraeger <borntraeger@de.ibm.com> Signed-off-by:
-
Jason Low authored
Commit 38c5ce93 ("mm/gup: Replace ACCESS_ONCE with READ_ONCE") converted ACCESS_ONCE usage in gup_pmd_range() to READ_ONCE, since ACCESS_ONCE doesn't work reliably on non-scalar types. This patch also fixes the other ACCESS_ONCE usages in gup_pte_range() and __get_user_pages_fast() in mm/gup.c Signed-off-by:
-
Derek authored
As suggested by Kirill the "goto"s in vma_to_resize aren't necessary, just change them to explicit return. Signed-off-by:
Derek Che <crquan@ymail.com> Suggested-by:
-
Derek authored
Recently I straced bash behavior in this dd zero pipe to read test, in part of testing under vm.overcommit_memory=2 (OVERCOMMIT_NEVER mode): # dd if=/dev/zero | read x The bash sub shell is calling mremap to reallocate more and more memory untill it finally failed -ENOMEM (I expect), or to be killed by system OOM killer (which should not happen under OVERCOMMIT_NEVER mode); But the mremap system call actually failed of -EFAULT, which is a surprise to me, I think it's supposed to be -ENOMEM? then I wrote this piece of C code testing confirmed it: https://gist.github.com/crquan/326bde37e1ddda8effe5 $ ./remap allocated one page @0x7f686bf71000, (PAGE_SIZE: 4096) grabbed 7680512000 bytes of memory (1875125 pages) @ 00007f6690993000. mremap failed Bad address (14). The -EFAULT comes from the branch of security_vm_enough_memory_mm failure, underlyingly it calls __vm_enough_memory which returns only 0 for success or -ENOMEM; So why vma_to_resize needs to return -EFAULT in this case? this sounds like a mistake to me. Some more digging into git history: 1) Before commit 119f657c ("RLIMIT_AS checking fix") in May 1 2005 (pre 2.6.12 days) it was returning -ENOMEM for this failure; 2) but commit 119f657c ("untangling do_mremap(), part 1") changed it accidentally, to what ever is preserved in local ret, which happened to be -EFAULT, in a previous assignment; 3) then in commit 54f5de70 code refactoring, it's explicitly returning -EFAULT, should be wrong. Signed-off-by: -
Roman Pen authored
In original implementation of vm_map_ram made by Nick Piggin there were two bitmaps: alloc_map and dirty_map. None of them were used as supposed to be: finding a suitable free hole for next allocation in block. vm_map_ram allocates space sequentially in block and on free call marks pages as dirty, so freed space can't be reused anymore. Actually it would be very interesting to know the real meaning of those bitmaps, maybe implementation was incomplete, etc. But long time ago Zhang Yanfei removed alloc_map by these two commits: mm/vmalloc.c: remove dead code in vb_alloc 3fcd76e8 mm/vmalloc.c: remove alloc_map from vmap_block b8e748b6 In this patch I replaced dirty_map with two range variables: dirty min and max. These variables store minimum and maximum position of dirty space in a block, since we need only to know the dirty range, not exact position of dirty pages. Why it was made? Several reasons: at first glance it seems that vm_map_ram allocator concerns about fragmentation thus it uses bitmaps for finding free hole, but it is not true. To avoid complexity seems it is better to use something simple, like min or max range values. Secondly, code also becomes simpler, without iteration over bitmap, just comparing values in min and max macros. Thirdly, bitmap occupies up to 1024 bits (4MB is a max size of a block). Here I replaced the whole bitmap with two longs. Finally vm_unmap_aliases should be slightly faster and the whole vmap_block structure occupies less memory. Signed-off-by:Roman Pen <r.peniaev@gmail.com> Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com> Cc: Eric Dumazet <edumazet@google.com> Acked-by:
Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: David Rientjes <rientjes@google.com> Cc: WANG Chao <chaowang@redhat.com> Cc: Fabian Frederick <fabf@skynet.be> Cc: Christoph Lameter <cl@linux.com> Cc: Gioh Kim <gioh.kim@lge.com> Cc: Rob Jones <rob.jones@codethink.co.uk> Signed-off-by:
-
Roman Pen authored
Previous implementation allocates new vmap block and repeats search of a free block from the very beginning, iterating over the CPU free list. Why it can be better?? 1. Allocation can happen on one CPU, but search can be done on another CPU. In worst case we preallocate amount of vmap blocks which is equal to CPU number on the system. 2. In previous patch I added newly allocated block to the tail of free list to avoid soon exhaustion of virtual space and give a chance to occupy blocks which were allocated long time ago. Thus to find newly allocated block all the search sequence should be repeated, seems it is not efficient. In this patch newly allocated block is occupied right away, address of virtual space is returned to the caller, so there is no any need to repeat the search sequence, allocation job is done. Signed-off-by:
-
Roman Pen authored
Recently I came across high fragmentation of vm_map_ram allocator: vmap_block has free space, but still new blocks continue to appear. Further investigation showed that certain mapping/unmapping sequences can exhaust vmalloc space. On small 32bit systems that's not a big problem, cause purging will be called soon on a first allocation failure (alloc_vmap_area), but on 64bit machines, e.g. x86_64 has 45 bits of vmalloc space, that can be a disaster. 1) I came up with a simple allocation sequence, which exhausts virtual space very quickly: while (iters) { /* Map/unmap big chunk */ vaddr = vm_map_ram(pages, 16, -1, PAGE_KERNEL); vm_unmap_ram(vaddr, 16); /* Map/unmap small chunks. * * -1 for hole, which should be left at the end of each block * to keep it partially used, with some free space available */ for (i = 0; i < (VMAP_BBMAP_BITS - 16) / 8 - 1; i++) { vaddr = vm_map_ram(pages, 8, -1, PAGE_KERNEL); vm_unmap_ram(vaddr, 8); } } The idea behind is simple: 1. We have to map a big chunk, e.g. 16 pages. 2. Then we have to occupy the remaining space with smaller chunks, i.e. 8 pages. At the end small hole should remain to keep block in free list, but do not let big chunk to occupy remaining space. 3. Goto 1 - allocation request of 16 pages can't be completed (only 8 slots are left free in the block in the #2 step), new block will be allocated, all further requests will lay into newly allocated block. To have some measurement numbers for all further tests I setup ftrace and enabled 4 basic calls in a function profile: echo vm_map_ram > /sys/kernel/debug/tracing/set_ftrace_filter; echo alloc_vmap_area >> /sys/kernel/debug/tracing/set_ftrace_filter; echo vm_unmap_ram >> /sys/kernel/debug/tracing/set_ftrace_filter; echo free_vmap_block >> /sys/kernel/debug/tracing/set_ftrace_filter; So for this scenario I got these results: BEFORE (all new blocks are put to the head of a free list) # cat /sys/kernel/debug/tracing/trace_stat/function0 Function Hit Time Avg s^2 -------- --- ---- --- --- vm_map_ram 126000 30683.30 us 0.243 us 30819.36 us vm_unmap_ram 126000 22003.24 us 0.174 us 340.886 us alloc_vmap_area 1000 4132.065 us 4.132 us 0.903 us AFTER (all new blocks are put to the tail of a free list) # cat /sys/kernel/debug/tracing/trace_stat/function0 Function Hit Time Avg s^2 -------- --- ---- --- --- vm_map_ram 126000 28713.13 us 0.227 us 24944.70 us vm_unmap_ram 126000 20403.96 us 0.161 us 1429.872 us alloc_vmap_area 993 3916.795 us 3.944 us 29.370 us free_vmap_block 992 654.157 us 0.659 us 1.273 us SUMMARY: The most interesting numbers in those tables are numbers of block allocations and deallocations: alloc_vmap_area and free_vmap_block calls, which show that before the change blocks were not freed, and virtual space and physical memory (vmap_block structure allocations, etc) were consumed. Average time which were spent in vm_map_ram/vm_unmap_ram became slightly better. That can be explained with a reasonable amount of blocks in a free list, which we need to iterate to find a suitable free block. 2) Another scenario is a random allocation: while (iters) { /* Randomly take number from a range [1..32/64] */ nr = rand(1, VMAP_MAX_ALLOC); vaddr = vm_map_ram(pages, nr, -1, PAGE_KERNEL); vm_unmap_ram(vaddr, nr); } I chose mersenne twister PRNG to generate persistent random state to guarantee that both runs have the same random sequence. For each vm_map_ram call random number from [1..32/64] was taken to represent amount of pages which I do map. I did 10'000 vm_map_ram calls and got these two tables: BEFORE (all new blocks are put to the head of a free list) # cat /sys/kernel/debug/tracing/trace_stat/function0 Function Hit Time Avg s^2 -------- --- ---- --- --- vm_map_ram 10000 10170.01 us 1.017 us 993.609 us vm_unmap_ram 10000 5321.823 us 0.532 us 59.789 us alloc_vmap_area 420 2150.239 us 5.119 us 3.307 us free_vmap_block 37 159.587 us 4.313 us 134.344 us AFTER (all new blocks are put to the tail of a free list) # cat /sys/kernel/debug/tracing/trace_stat/function0 Function Hit Time Avg s^2 -------- --- ---- --- --- vm_map_ram 10000 7745.637 us 0.774 us 395.229 us vm_unmap_ram 10000 5460.573 us 0.546 us 67.187 us alloc_vmap_area 414 2201.650 us 5.317 us 5.591 us free_vmap_block 412 574.421 us 1.394 us 15.138 us SUMMARY: 'BEFORE' table shows, that 420 blocks were allocated and only 37 were freed. Remained 383 blocks are still in a free list, consuming virtual space and physical memory. 'AFTER' table shows, that 414 blocks were allocated and 412 were really freed. 2 blocks remained in a free list. So fragmentation was dramatically reduced. Why? Because when we put newly allocated block to the head, all further requests will occupy new block, regardless remained space in other blocks. In this scenario all requests come randomly. Eventually remained free space will be less than requested size, free list will be iterated and it is possible that nothing will be found there - finally new block will be created. So exhaustion in random scenario happens for the maximum possible allocation size: 32 pages for 32-bit system and 64 pages for 64-bit system. Also average cost of vm_map_ram was reduced from 1.017 us to 0.774 us. Again this can be explained by iteration through smaller list of free blocks. 3) Next simple scenario is a sequential allocation, when the allocation order is increased for each block. This scenario forces allocator to reach maximum amount of partially free blocks in a free list: while (iters) { /* Populate free list with blocks with remaining space */ for (order = 0; order <= ilog2(VMAP_MAX_ALLOC); order++) { nr = VMAP_BBMAP_BITS / (1 << order); /* Leave a hole */ nr -= 1; for (i = 0; i < nr; i++) { vaddr = vm_map_ram(pages, (1 << order), -1, PAGE_KERNEL); vm_unmap_ram(vaddr, (1 << order)); } /* Completely occupy blocks from a free list */ for (order = 0; order <= ilog2(VMAP_MAX_ALLOC); order++) { vaddr = vm_map_ram(pages, (1 << order), -1, PAGE_KERNEL); vm_unmap_ram(vaddr, (1 << order)); } } Results which I got: BEFORE (all new blocks are put to the head of a free list) # cat /sys/kernel/debug/tracing/trace_stat/function0 Function Hit Time Avg s^2 -------- --- ---- --- --- vm_map_ram 2032000 399545.2 us 0.196 us 467123.7 us vm_unmap_ram 2032000 363225.7 us 0.178 us 111405.9 us alloc_vmap_area 7001 30627.76 us 4.374 us 495.755 us free_vmap_block 6993 7011.685 us 1.002 us 159.090 us AFTER (all new blocks are put to the tail of a free list) # cat /sys/kernel/debug/tracing/trace_stat/function0 Function Hit Time Avg s^2 -------- --- ---- --- --- vm_map_ram 2032000 394259.7 us 0.194 us 589395.9 us vm_unmap_ram 2032000 292500.7 us 0.143 us 94181.08 us alloc_vmap_area 7000 31103.11 us 4.443 us 703.225 us free_vmap_block 7000 6750.844 us 0.964 us 119.112 us SUMMARY: No surprises here, almost all numbers are the same. Fixing this fragmentation problem I also did some improvements in a allocation logic of a new vmap block: occupy block immediately and get rid of extra search in a free list. Also I replaced dirty bitmap with min/max dirty range values to make the logic simpler and slightly faster, since two longs comparison costs less, than loop thru bitmap. This patchset raises several questions: Q: Think the problem you comments is already known so that I wrote comments about it as "it could consume lots of address space through fragmentation". Could you tell me about your situation and reason why it should be avoided? Gioh Kim A: Indeed, there was a commit 36437638 which adds explicit comment about fragmentation. But fragmentation which is described in this comment caused by mixing of long-lived and short-lived objects, when a whole block is pinned in memory because some page slots are still in use. But here I am talking about blocks which are free, nobody uses them, and allocator keeps them alive forever, continuously allocating new blocks. Q: I think that if you put newly allocated block to the tail of a free list, below example would results in enormous performance degradation. new block: 1MB (256 pages) while (iters--) { vm_map_ram(3 or something else not dividable for 256) * 85 vm_unmap_ram(3) * 85 } On every iteration, it needs newly allocated block and it is put to the tail of a free list so finding it consumes large amount of time. Joonsoo Kim A: Second patch in current patchset gets rid of extra search in a free list, so new block will be immediately occupied.. Also, the scenario above is impossible, cause vm_map_ram allocates virtual range in orders, i.e. 2^n. I.e. passing 3 to vm_map_ram you will allocate 4 slots in a block and 256 slots (capacity of a block) of course dividable on 4, so block will be completely occupied. But there is a worst case which we can achieve: each free block has a hole equal to order size. The maximum size of allocation is 64 pages for 64-bit system (if you try to map more, original alloc_vmap_area will be called). So the maximum order is 6. That means that worst case, before allocator makes a decision to allocate a new block, is to iterate 7 blocks: HEAD 1st block - has 1 page slot free (order 0) 2nd block - has 2 page slots free (order 1) 3rd block - has 4 page slots free (order 2) 4th block - has 8 page slots free (order 3) 5th block - has 16 page slots free (order 4) 6th block - has 32 page slots free (order 5) 7th block - has 64 page slots free (order 6) TAIL So the worst scenario on 64-bit system is that each CPU queue can have 7 blocks in a free list. This can happen only and only if you allocate blocks increasing the order. (as I did in the function written in the comment of the first patch) This is weird and rare case, but still it is possible. Afterwards you will get 7 blocks in a list. All further requests should be placed in a newly allocated block or some free slots should be found in a free list. Seems it does not look dramatically awful. This patch (of 3): If suitable block can't be found, new block is allocated and put into a head of a free list, so on next iteration this new block will be found first. That's bad, because old blocks in a free list will not get a chance to be fully used, thus fragmentation will grow. Let's consider this simple example: #1 We have one block in a free list which is partially used, and where only one page is free: HEAD |xxxxxxxxx-| TAIL ^ free space for 1 page, order 0 #2 New allocation request of order 1 (2 pages) comes, new block is allocated since we do not have free space to complete this request. New block is put into a head of a free list: HEAD |----------|xxxxxxxxx-| TAIL #3 Two pages were occupied in a new found block: HEAD |xx--------|xxxxxxxxx-| TAIL ^ two pages mapped here #4 New allocation request of order 0 (1 page) comes. Block, which was created on #2 step, is located at the beginning of a free list, so it will be found first: HEAD |xxX-------|xxxxxxxxx-| TAIL ^ ^ page mapped here, but better to use this hole It is obvious, that it is better to complete request of #4 step using the old block, where free space is left, because in other case fragmentation will be highly increased. But fragmentation is not only the case. The worst thing is that I can easily create scenario, when the whole vmalloc space is exhausted by blocks, which are not used, but already dirty and have several free pages. Let's consider this function which execution should be pinned to one CPU: static void exhaust_virtual_space(struct page *pages[16], int iters) { /* Firstly we have to map a big chunk, e.g. 16 pages. * Then we have to occupy the remaining space with smaller * chunks, i.e. 8 pages. At the end small hole should remain. * So at the end of our allocation sequence block looks like * this: * XX big chunk * |XXxxxxxxx-| x small chunk * - hole, which is enough for a small chunk, * but is not enough for a big chunk */ while (iters--) { int i; void *vaddr; /* Map/unmap big chunk */ vaddr = vm_map_ram(pages, 16, -1, PAGE_KERNEL); vm_unmap_ram(vaddr, 16); /* Map/unmap small chunks. * * -1 for hole, which should be left at the end of each block * to keep it partially used, with some free space available */ for (i = 0; i < (VMAP_BBMAP_BITS - 16) / 8 - 1; i++) { vaddr = vm_map_ram(pages, 8, -1, PAGE_KERNEL); vm_unmap_ram(vaddr, 8); } } } On every iteration new block (1MB of vm area in my case) will be allocated and then will be occupied, without attempt to resolve small allocation request using previously allocated blocks in a free list. In case of random allocation (size should be randomly taken from the range [1..64] in 64-bit case or [1..32] in 32-bit case) situation is the same: new blocks continue to appear if maximum possible allocation size (32 or 64) passed to the allocator, because all remaining blocks in a free list do not have enough free space to complete this allocation request. In summary if new blocks are put into the head of a free list eventually virtual space will be exhausted. In current patch I simply put newly allocated block to the tail of a free list, thus reduce fragmentation, giving a chance to resolve allocation request using older blocks with possible holes left. Signed-off-by: -
Mike Kravetz authored
Add min_size mount option to the hugetlbfs documentation. Also, add the missing pagesize option and mention that size can be specified as bytes or a percentage of huge page pool. Signed-off-by:
Mike Kravetz <mike.kravetz@oracle.com> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: Aneesh Kumar <aneesh.kumar@linux.vnet.ibm.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: David Rientjes <rientjes@google.com> Signed-off-by:
-
Mike Kravetz authored
Make 'min_size=<value>' be an option when mounting a hugetlbfs. This option takes the same value as the 'size' option. min_size can be specified without specifying size. If both are specified, min_size must be less that or equal to size else the mount will fail. If min_size is specified, then at mount time an attempt is made to reserve min_size pages. If the reservation fails, the mount fails. At umount time, the reserved pages are released. Signed-off-by:
-
Mike Kravetz authored
The same routines that perform subpool maximum size accounting hugepage_subpool_get/put_pages() are modified to also perform minimum size accounting. When a delta value is passed to these routines, calculate how global reservations must be adjusted to maintain the subpool minimum size. The routines now return this global reserve count adjustment. This global reserve count adjustment is then passed to the global accounting routine hugetlb_acct_memory(). Signed-off-by:
-
Mike Kravetz authored
hugetlbfs allocates huge pages from the global pool as needed. Even if the global pool contains a sufficient number pages for the filesystem size at mount time, those global pages could be grabbed for some other use. As a result, filesystem huge page allocations may fail due to lack of pages. Applications such as a database want to use huge pages for performance reasons. hugetlbfs filesystem semantics with ownership and modes work well to manage access to a pool of huge pages. However, the application would like some reasonable assurance that allocations will not fail due to a lack of huge pages. At application startup time, the application would like to configure itself to use a specific number of huge pages. Before starting, the application can check to make sure that enough huge pages exist in the system global pools. However, there are no guarantees that those pages will be available when needed by the application. What the application wants is exclusive use of a subset of huge pages. Add a new hugetlbfs mount option 'min_size=<value>' to indicate that the specified number of pages will be available for use by the filesystem. At mount time, this number of huge pages will be reserved for exclusive use of the filesystem. If there is not a sufficient number of free pages, the mount will fail. As pages are allocated to and freeed from the filesystem, the number of reserved pages is adjusted so that the specified minimum is maintained. This patch (of 4): Add a field to the subpool structure to indicate the minimimum number of huge pages to always be used by this subpool. This minimum count includes allocated pages as well as reserved pages. If the minimum number of pages for the subpool have not been allocated, pages are reserved up to this minimum. An additional field (rsv_hpages) is used to track the number of pages reserved to meet this minimum size. The hstate pointer in the subpool is convenient to have when reserving and unreserving the pages. Signed-off-by:
-
Gioh Kim authored
When the compaction is activated via /proc/sys/vm/compact_memory it would better scan the whole zone. And some platforms, for instance ARM, have the start_pfn of a zone at zero. Therefore the first try to compact via /proc doesn't work. It needs to reset the compaction scanner position first. Signed-off-by:
Gioh Kim <gioh.kim@lge.com> Acked-by:
Vlastimil Babka <vbabka@suse.cz> Acked-by:
-
Michal Hocko authored
memcg currently uses hardcoded GFP_TRANSHUGE gfp flags for all THP charges. THP allocations, however, might be using different flags depending on /sys/kernel/mm/transparent_hugepage/{,khugepaged/}defrag and the current allocation context. The primary difference is that defrag configured to "madvise" value will clear __GFP_WAIT flag from the core gfp mask to make the allocation lighter for all mappings which are not backed by VM_HUGEPAGE vmas. If memcg charge path ignores this fact we will get light allocation but the a potential memcg reclaim would kill the whole point of the configuration. Fix the mismatch by providing the same gfp mask used for the allocation to the charge functions. This is quite easy for all paths except for hugepaged kernel thread with !CONFIG_NUMA which is doing a pre-allocation long before the allocated page is used in collapse_huge_page via khugepaged_alloc_page. To prevent from cluttering the whole code path from khugepaged_do_scan we simply return the current flags as per khugepaged_defrag() value which might have changed since the preallocation. If somebody changed the value of the knob we would charge differently but this shouldn't happen often and it is definitely not critical because it would only lead to a reduced success rate of one-off THP promotion. [akpm@linux-foundation.org: fix weird code layout while we're there] [rientjes@google.com: clean up around alloc_hugepage_gfpmask()] Signed-off-by: -
Minchan Kim authored
"deactivate_page" was created for file invalidation so it has too specific logic for file-backed pages. So, let's change the name of the function and date to a file-specific one and yield the generic name. Signed-off-by:
-
Eric B Munson authored
Documentation/vm/unevictable-lru.txt: document interaction between compaction and the unevictable LRU The memory compaction code uses the migration code to do most of the work in compaction. However, the compaction code interacts with the unevictable LRU differently than migration code and this difference should be noted in the documentation. [akpm@linux-foundation.org: identify /proc/sys/vm/compact_unevictable directly] Signed-off-by:
Eric B Munson <emunson@akamai.com> Cc: Michal Hocko <mhocko@suse.cz> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Christoph Lameter <cl@linux.com> Cc: David Rientjes <rientjes@google.com> Cc: Rik van Riel <riel@redhat.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mel Gorman <mgorman@suse.de> Signed-off-by:
-
