- 24 Feb, 2021 40 commits
-
-
Andrey Konovalov authored
Add a test for kmem_cache_alloc/free_bulk to make sure there are no false-positives when these functions are used. Link: https://linux-review.googlesource.com/id/I2a8bf797aecf81baeac61380c567308f319e263d Link: https://lkml.kernel.org/r/418122ebe4600771ac81e9ca6eab6740cf8dcfa1.1610733117.git.andreyknvl@google.comSigned-off-by:
Andrey Konovalov <andreyknvl@google.com> Reviewed-by:
Marco Elver <elver@google.com> Reviewed-by:
Alexander Potapenko <glider@google.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: Branislav Rankov <Branislav.Rankov@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Dmitry Vyukov <dvyukov@google.com> Cc: Evgenii Stepanov <eugenis@google.com> Cc: Kevin Brodsky <kevin.brodsky@arm.com> Cc: Peter Collingbourne <pcc@google.com> Cc: Vincenzo Frascino <vincenzo.frascino@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Andrey Konovalov authored
The currently existing page allocator tests rely on kmalloc fallback with large sizes that is only present for SLUB. Add proper tests that use alloc/free_pages(). Link: https://linux-review.googlesource.com/id/Ia173d5a1b215fe6b2548d814ef0f4433cf983570 Link: https://lkml.kernel.org/r/a2648930e55ff75b8e700f2e0d905c2b55a67483.1610733117.git.andreyknvl@google.comSigned-off-by:
-
Andrey Konovalov authored
The currently existing kasan_check_read/write() annotations are intended to be used for kernel modules that have KASAN compiler instrumentation disabled. Thus, they are only relevant for the software KASAN modes that rely on compiler instrumentation. However there's another use case for these annotations: ksize() checks that the object passed to it is indeed accessible before unpoisoning the whole object. This is currently done via __kasan_check_read(), which is compiled away for the hardware tag-based mode that doesn't rely on compiler instrumentation. This leads to KASAN missing detecting some memory corruptions. Provide another annotation called kasan_check_byte() that is available for all KASAN modes. As the implementation rename and reuse kasan_check_invalid_free(). Use this new annotation in ksize(). To avoid having ksize() as the top frame in the reported stack trace pass _RET_IP_ to __kasan_check_byte(). Also add a new ksize_uaf() test that checks that a use-after-free is detected via ksize() itself, and via plain accesses that happen later. Link: https://linux-review.googlesource.com/id/Iaabf771881d0f9ce1b969f2a62938e99d3308ec5 Link: https://lkml.kernel.org/r/f32ad74a60b28d8402482a38476f02bb7600f620.1610733117.git.andreyknvl@google.comSigned-off-by:
-
Andrey Konovalov authored
Generic mm functions that call KASAN annotations that might report a bug pass _RET_IP_ to them as an argument. This allows KASAN to include the name of the function that called the mm function in its report's header. Now that KASAN has inline wrappers for all of its annotations, move _RET_IP_ to those wrappers to simplify annotation call sites. Link: https://linux-review.googlesource.com/id/I8fb3c06d49671305ee184175a39591bc26647a67 Link: https://lkml.kernel.org/r/5c1490eddf20b436b8c4eeea83fce47687d5e4a4.1610733117.git.andreyknvl@google.comSigned-off-by:
-
Andrey Konovalov authored
Since the hardware tag-based KASAN mode might not have a redzone that comes after an allocated object (when kasan.mode=prod is enabled), the kasan_bitops_tags() test ends up corrupting the next object in memory. Change the test so it always accesses the redzone that lies within the allocated object's boundaries. Link: https://linux-review.googlesource.com/id/I67f51d1ee48f0a8d0fe2658c2a39e4879fe0832a Link: https://lkml.kernel.org/r/7d452ce4ae35bb1988d2c9244dfea56cf2cc9315.1610733117.git.andreyknvl@google.comSigned-off-by:
-
Andrey Konovalov authored
In the kmalloc_uaf2() test, the pointers to the two allocated memory blocks might happen to be the same, and the test will fail. With the software tag-based mode, the probability of the that is 1/254, so it's hard to observe the failure. For the hardware tag-based mode though, the probablity is 1/14, which is quite noticable. Allow up to 16 attempts at generating different tags for the tag-based modes. Link: https://linux-review.googlesource.com/id/Ibfa458ef2804ff465d8eb07434a300bf36388d55 Link: https://lkml.kernel.org/r/9cd5cf2f633dcbf55cab801cd26845d2b075cec7.1610733117.git.andreyknvl@google.comSigned-off-by:
-
Andrey Konovalov authored
It might not be obvious to the compiler that the expression must be executed between writing and reading to fail_data. In this case, the compiler might reorder or optimize away some of the accesses, and the tests will fail. Add compiler barriers around the expression in KUNIT_EXPECT_KASAN_FAIL and use READ/WRITE_ONCE() for accessing fail_data fields. Link: https://linux-review.googlesource.com/id/I046079f48641a1d36fe627fc8827a9249102fd50 Link: https://lkml.kernel.org/r/6f11596f367d8ae8f71d800351e9a5d91eda19f6.1610733117.git.andreyknvl@google.comSigned-off-by:
-
Andrey Konovalov authored
Rename CONFIG_TEST_KASAN_MODULE to CONFIG_KASAN_MODULE_TEST. This naming is more consistent with the existing CONFIG_KASAN_KUNIT_TEST. Link: https://linux-review.googlesource.com/id/Id347dfa5fe8788b7a1a189863e039f409da0ae5f Link: https://lkml.kernel.org/r/f08250246683981bcf8a094fbba7c361995624d2.1610733117.git.andreyknvl@google.comSigned-off-by:
-
Andrey Konovalov authored
On a high level, this patch allows running KUnit KASAN tests with the hardware tag-based KASAN mode. Internally, this change reenables tag checking at the end of each KASAN test that triggers a tag fault and leads to tag checking being disabled. Also simplify is_write calculation in report_tag_fault. With this patch KASAN tests are still failing for the hardware tag-based mode; fixes come in the next few patches. [andreyknvl@google.com: export HW_TAGS symbols for KUnit tests] Link: https://lkml.kernel.org/r/e7eeb252da408b08f0c81b950a55fb852f92000b.1613155970.git.andreyknvl@google.com Link: https://linux-review.googlesource.com/id/Id94dc9eccd33b23cda4950be408c27f879e474c8 Link: https://lkml.kernel.org/r/51b23112cf3fd62b8f8e9df81026fa2b15870501.1610733117.git.andreyknvl@google.comSigned-off-by:
Catalin Marinas <catalin.marinas@arm.com> Reviewed-by:
Vincenzo Frascino <vincenzo.frascino@arm.com> Cc: Alexander Potapenko <glider@google.com> Cc: Andrey Ryabinin <aryabinin@virtuozzo.com> Cc: Branislav Rankov <Branislav.Rankov@arm.com> Cc: Dmitry Vyukov <dvyukov@google.com> Cc: Evgenii Stepanov <eugenis@google.com> Cc: Kevin Brodsky <kevin.brodsky@arm.com> Cc: Marco Elver <elver@google.com> Cc: Peter Collingbourne <pcc@google.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by:
-
Andrey Konovalov authored
Add 3 new tests for tag-based KASAN modes: 1. Check that match-all pointer tag is not assigned randomly. 2. Check that 0xff works as a match-all pointer tag. 3. Check that there are no match-all memory tags. Note, that test #3 causes a significant number (255) of KASAN reports to be printed during execution for the SW_TAGS mode. [arnd@arndb.de: export kasan_poison] Link: https://lkml.kernel.org/r/20210125112831.2156212-1-arnd@kernel.org [akpm@linux-foundation.org: s/EXPORT_SYMBOL_GPL/EXPORT_SYMBOL/, per Andrey] Link: https://linux-review.googlesource.com/id/I78f1375efafa162b37f3abcb2c5bc2f3955dfd8e Link: https://lkml.kernel.org/r/da841a5408e2204bf25f3b23f70540a65844e8a4.1610733117.git.andreyknvl@google.comSigned-off-by:
Arnd Bergmann <arnd@arndb.de> Reviewed-by:
-
Andrey Konovalov authored
Some KASAN tests require specific kernel configs to be enabled. Instead of copy-pasting the checks for these configs add a few helper macros and use them. Link: https://linux-review.googlesource.com/id/I237484a7fddfedf4a4aae9cc61ecbcdbe85a0a63 Link: https://lkml.kernel.org/r/6a0fcdb9676b7e869cfc415893ede12d916c246c.1610733117.git.andreyknvl@google.comSigned-off-by:
-
Andrey Konovalov authored
Clarify and update comments in KASAN tests. Link: https://linux-review.googlesource.com/id/I6c816c51fa1e0eb7aa3dead6bda1f339d2af46c8 Link: https://lkml.kernel.org/r/ba6db104d53ae0e3796f80ef395f6873c1c1282f.1610733117.git.andreyknvl@google.comSigned-off-by:
-
Andrey Konovalov authored
Mention in the documentation that enabling CONFIG_KASAN_HW_TAGS always results in in-kernel TBI (Top Byte Ignore) being enabled. Also do a few minor documentation cleanups. Link: https://linux-review.googlesource.com/id/Iba2a6697e3c6304cb53f89ec61dedc77fa29e3ae Link: https://lkml.kernel.org/r/3b4ea6875bb14d312092ad14ac55cb456c83c08e.1610733117.git.andreyknvl@google.comSigned-off-by:
-
Andrey Konovalov authored
Patch series "kasan: HW_TAGS tests support and fixes", v4. This patchset adds support for running KASAN-KUnit tests with the hardware tag-based mode and also contains a few fixes. This patch (of 15): There's a number of internal KASAN functions that are used across multiple source code files and therefore aren't marked as static inline. To avoid littering the kernel function names list with generic function names, prefix all such KASAN functions with kasan_. As a part of this change: - Rename internal (un)poison_range() to kasan_(un)poison() (no _range) to avoid name collision with a public kasan_unpoison_range(). - Rename check_memory_region() to kasan_check_range(), as it's a more fitting name. Link: https://lkml.kernel.org/r/cover.1610733117.git.andreyknvl@google.com Link: https://linux-review.googlesource.com/id/I719cc93483d4ba288a634dba80ee6b7f2809cd26 Link: https://lkml.kernel.org/r/13777aedf8d3ebbf35891136e1f2287e2f34aaba.1610733117.git.andreyknvl@google.comSigned-off-by:
-
Yang Li authored
Fix below warnings reported by coccicheck: fs/proc/vmcore.c:1503:2-7: WARNING: NULL check before some freeing functions is not needed. Link: https://lkml.kernel.org/r/1611216753-44598-1-git-send-email-abaci-bugfix@linux.alibaba.comSigned-off-by:
Yang Li <abaci-bugfix@linux.alibaba.com> Reported-by:
Abaci Robot <abaci@linux.alibaba.com> Acked-by:
Baoquan He <bhe@redhat.com> Cc: Dave Young <dyoung@redhat.com> Cc: Vivek Goyal <vgoyal@redhat.com> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: "Uladzislau Rezki (Sony)" <urezki@gmail.com> Signed-off-by:
-
sh_def@163.com authored
Replace '&next->lru != list' with list_entry_is_head(). No functional change. Link: https://lkml.kernel.org/r/20201222182735.GA1257912@ubuntu-A520I-ACSigned-off-by:
sh <sh_def@163.com> Reviewed-by:
David Hildenbrand <david@redhat.com> Signed-off-by:
-
Li Xinhai authored
mremap with MREMAP_DONTUNMAP can move all page table entries to new vma, which means all pages allocated for the old vma are not relevant to it anymore, and the relevant anon_vma links needs to be unlinked, in nature the old vma is much like been freshly created and have no pages been fault in. But we should not do unlink, if the new vma has effectively merged with the old one. [lixinhai.lxh@gmail.com: v2] Link: https://lkml.kernel.org/r/20210127083917.309264-2-lixinhai.lxh@gmail.com Link: https://lkml.kernel.org/r/20210119075126.3513154-2-lixinhai.lxh@gmail.comSigned-off-by:
Li Xinhai <lixinhai.lxh@gmail.com> Cc: Brian Geffon <bgeffon@google.com> Cc: Lokesh Gidra <lokeshgidra@google.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Andrea Arcangeli <aarcange@redhat.com> Signed-off-by:
-
Li Xinhai authored
In case the vma will continue to be used after unlink its relevant anon_vma, we need to reset the vma->anon_vma pointer to NULL. So, later when fault happen within this vma again, a new anon_vma will be prepared. By this way, the vma will only be checked for reverse mapping of pages which been fault in after the unlink_anon_vmas call. Currently, the mremap with MREMAP_DONTUNMAP scenario will continue use the vma after moved its page table entries to a new vma. For other scenarios, the vma itself will be freed after call unlink_anon_vmas. Link: https://lkml.kernel.org/r/20210119075126.3513154-1-lixinhai.lxh@gmail.comSigned-off-by:
-
Tianjia Zhang authored
Obviously, the error variable detection of the if statement is for the mprotect callback function, so it is also put into the scope of calling callbck. This is a cleanup which makes this site consistent with the rest of this function's error handling. Link: https://lkml.kernel.org/r/20210118133310.98375-1-tianjia.zhang@linux.alibaba.comSigned-off-by:
Tianjia Zhang <tianjia.zhang@linux.alibaba.com> Reported-by:
Jia Zhang <zhang.jia@linux.alibaba.com> Reviewed-by:
-
Miaohe Lin authored
If all pte entry is none in 'non-create' case, we would break the loop with pte unchanged. Then the wrong pte - 1 would be passed to pte_unmap_unlock. This is a theoretical issue which may not be a real bug. So it's not worth cc stable. Link: https://lkml.kernel.org/r/20210205081925.59809-1-linmiaohe@huawei.com Fixes: aee16b3c ("Add apply_to_page_range() which applies a function to a pte range") Signed-off-by:
Miaohe Lin <linmiaohe@huawei.com> Reviewed-by:
-
Miaohe Lin authored
The developer will have trouble figuring out why the BUG actually triggered when there is a complex expression in the VM_BUG_ON. Because we can only identify the condition triggered BUG via line number provided by VM_BUG_ON. Optimize this by spliting such a complex expression into two simple conditions. Link: https://lkml.kernel.org/r/20210203084137.25522-1-linmiaohe@huawei.comSigned-off-by:
-
Miaohe Lin authored
The condition (A && !C && !D) || !A is equivalent to !A || (A && !C && !D) and can be further simplified to !A || (!C && !D). Link: https://lkml.kernel.org/r/20210201114319.34720-1-linmiaohe@huawei.comSigned-off-by:
-
Miaohe Lin authored
Since commit 42e4089c ("x86/speculation/l1tf: Disallow non privileged high MMIO PROT_NONE mappings"), when the first pfn modify is not allowed, we would break the loop with pte unchanged. Then the wrong pte - 1 would be passed to pte_unmap_unlock. Andi said: "While the fix is correct, I'm not sure if it actually is a real bug. Is there any architecture that would do something else than unlocking the underlying page? If it's just the underlying page then it should be always the same page, so no bug" Link: https://lkml.kernel.org/r/20210109080118.20885-1-linmiaohe@huawei.com Fixes: 42e4089c ("x86/speculation/l1tf: Disallow non privileged high MMIO PROT_NONE mappings") Signed-off-by:
Hongxiang Lou <louhongxiang@huawei.com> Signed-off-by:
-
Adrian Huang authored
The local variable 'retval' is assigned just for once in __do_sys_brk(), and the function returns the value of the local variable right after the assignment. Remove unnecessary assignment and local variable declaration. Link: https://lkml.kernel.org/r/20201222103249.30683-1-adrianhuang0701@gmail.comSigned-off-by:
Adrian Huang <ahuang12@lenovo.com> Acked-by:
Souptick Joarder <jrdr.linux@gmail.com> Reviewed-by:
-
Muchun Song authored
SLUB currently account kmalloc() and kmalloc_node() allocations larger than order-1 page per-node. But it forget to update the per-memcg vmstats. So it can lead to inaccurate statistics of "slab_unreclaimable" which is from memory.stat. Fix it by using mod_lruvec_page_state instead of mod_node_page_state. Link: https://lkml.kernel.org/r/20210223092423.42420-1-songmuchun@bytedance.com Fixes: 6a486c0a ("mm, sl[ou]b: improve memory accounting") Signed-off-by:
Muchun Song <songmuchun@bytedance.com> Reviewed-by:
Shakeel Butt <shakeelb@google.com> Reviewed-by:
Roman Gushchin <guro@fb.com> Reviewed-by:
Michal Koutný <mkoutny@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Michal Hocko <mhocko@kernel.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Signed-off-by:
-
Muchun Song authored
We use a global percpu int_active_memcg variable to store the remote memcg when we are in the interrupt context. But get_active_memcg always return the current->active_memcg or root_mem_cgroup. The remote memcg (set in the interrupt context) is ignored. This is not what we want. So fix it. Link: https://lkml.kernel.org/r/20210223091101.42150-1-songmuchun@bytedance.com Fixes: 37d5985c ("mm: kmem: prepare remote memcg charging infra for interrupt contexts") Signed-off-by:
-
Muchun Song authored
When pages are swapped in, the VM may retain the swap copy to avoid repeated writes in the future. It's also retained if shared pages are faulted back in some processes, but not in others. During that time we have an in-memory copy of the page, as well as an on-swap copy. Cgroup1 and cgroup2 handle these overlapping lifetimes slightly differently due to the nature of how they account memory and swap: Cgroup1 has a unified memory+swap counter that tracks a data page regardless whether it's in-core or swapped out. On swapin, we transfer the charge from the swap entry to the newly allocated swapcache page, even though the swap entry might stick around for a while. That's why we have a mem_cgroup_uncharge_swap() call inside mem_cgroup_charge(). Cgroup2 tracks memory and swap as separate, independent resources and thus has split memory and swap counters. On swapin, we charge the newly allocated swapcache page as memory, while the swap slot in turn must remain charged to the swap counter as long as its allocated too. The cgroup2 logic was broken by commit 2d1c4980 ("mm: memcontrol: make swap tracking an integral part of memory control"), because it accidentally removed the do_memsw_account() check in the branch inside mem_cgroup_uncharge() that was supposed to tell the difference between the charge transfer in cgroup1 and the separate counters in cgroup2. As a result, cgroup2 currently undercounts retained swap to varying degrees: swap slots are cached up to 50% of the configured limit or total available swap space; partially faulted back shared pages are only limited by physical capacity. This in turn allows cgroups to significantly overconsume their alloted swap space. Add the do_memsw_account() check back to fix this problem. Link: https://lkml.kernel.org/r/20210217153237.92484-1-songmuchun@bytedance.com Fixes: 2d1c4980 ("mm: memcontrol: make swap tracking an integral part of memory control") Signed-off-by:
Johannes Weiner <hannes@cmpxchg.org> Reviewed-by:
Michal Hocko <mhocko@suse.com> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: <stable@vger.kernel.org> [5.8+] Signed-off-by:
-
Johannes Weiner authored
alloc_page_buffers() currently uses get_mem_cgroup_from_page() for charging the buffers to the page owner, which does an rcu-protected page->memcg lookup and acquires a reference. But buffer allocation has the page lock held throughout, which pins the page to the memcg and thereby the memcg - neither rcu nor holding an extra reference during the allocation are necessary. Use a raw page_memcg() instead. This was the last user of get_mem_cgroup_from_page(), delete it. Link: https://lkml.kernel.org/r/20210209190126.97842-1-hannes@cmpxchg.orgSigned-off-by:
-
Shakeel Butt authored
The list_lru file used to have local kvfree_rcu() which was renamed by commit e0feed08 ("mm/list_lru.c: Rename kvfree_rcu() to local variant") to introduce the globally visible kvfree_rcu(). Now we have global kvfree_rcu(), so remove the local kvfree_rcu_local() and just use the global one. Link: https://lkml.kernel.org/r/20210207152148.1285842-1-shakeelb@google.comSigned-off-by:
Uladzislau Rezki <urezki@gmail.com> Reviewed-by:
Kirill Tkhai <ktkhai@virtuozzo.com> Signed-off-by:
-
Muchun Song authored
The rule of list walk has gone since commit a9d5adee ("mm/memcontrol: allow to uncharge page without using page->lru field") So remove the strange comment and replace the loop with a list_for_each_entry(). There is only one caller of the uncharge_list(). So just fold it into mem_cgroup_uncharge_list() and remove it. Link: https://lkml.kernel.org/r/20210204163055.56080-1-songmuchun@bytedance.comSigned-off-by:
-
Yang Li authored
Fix below warnings reported by coccicheck: mm/memcontrol.c:451:3-9: WARNING: NULL check before some freeing functions is not needed. Link: https://lkml.kernel.org/r/1611216029-34397-1-git-send-email-abaci-bugfix@linux.alibaba.comSigned-off-by:
-
Feng Tang authored
When checking a memory cgroup related performance regression [1], from the perf c2c profiling data, we found high false sharing for accessing 'usage' and 'parent'. On 64 bit system, the 'usage' and 'parent' are close to each other, and easy to be in one cacheline (for cacheline size == 64+ B). 'usage' is usally written, while 'parent' is usually read as the cgroup's hierarchical counting nature. So move the 'parent' to the end of the structure to make sure they are in different cache lines. Following are some performance data with the patch, against v5.11-rc1. [ In the data, A means a platform with 2 sockets 48C/96T, B is a platform of 4 sockests 72C/144T, and if a %stddev will be shown bigger than 2%, P100/P50 means number of test tasks equals to 100%/50% of nr_cpu] will-it-scale/malloc1 --------------------- v5.11-rc1 v5.11-rc1+patch A-P100 15782 ± 2% -0.1% 15765 ± 3% will-it-scale.per_process_ops A-P50 21511 +8.9% 23432 will-it-scale.per_process_ops B-P100 9155 +2.2% 9357 will-it-scale.per_process_ops B-P50 10967 +7.1% 11751 ± 2% will-it-scale.per_process_ops will-it-scale/pagefault2 ------------------------ v5.11-rc1 v5.11-rc1+patch A-P100 79028 +3.0% 81411 will-it-scale.per_process_ops A-P50 183960 ± 2% +4.4% 192078 ± 2% will-it-scale.per_process_ops B-P100 85966 +9.9% 94467 ± 3% will-it-scale.per_process_ops B-P50 198195 +9.8% 217526 will-it-scale.per_process_ops fio (4k/1M is block size) ------------------------- v5.11-rc1 v5.11-rc1+patch A-P50-r-4k 16881 ± 2% +1.2% 17081 ± 2% fio.read_bw_MBps A-P50-w-4k 3931 +4.5% 4111 ± 2% fio.write_bw_MBps A-P50-r-1M 15178 -0.2% 15154 fio.read_bw_MBps A-P50-w-1M 3924 +0.1% 3929 fio.write_bw_MBps [1].https://lore.kernel.org/lkml/20201102091543.GM31092@shao2-debian/ Link: https://lkml.kernel.org/r/1611040814-33449-1-git-send-email-feng.tang@intel.comSigned-off-by:
Feng Tang <feng.tang@intel.com> Reviewed-by:
-
Roman Gushchin authored
I've noticed that __memcg_kmem_charge() and __memcg_kmem_uncharge() are not used anywhere except memcontrol.c. Yet they are not declared as non-static and are declared in memcontrol.h. This patch makes them static. Link: https://lkml.kernel.org/r/20210108020332.4096911-1-guro@fb.comSigned-off-by:
-
Shakeel Butt authored
This patch adds swapcache stat for the cgroup v2. The swapcache represents the memory that is accounted against both the memory and the swap limit of the cgroup. The main motivation behind exposing the swapcache stat is for enabling users to gracefully migrate from cgroup v1's memsw counter to cgroup v2's memory and swap counters. Cgroup v1's memsw limit allows users to limit the memory+swap usage of a workload but without control on the exact proportion of memory and swap. Cgroup v2 provides separate limits for memory and swap which enables more control on the exact usage of memory and swap individually for the workload. With some little subtleties, the v1's memsw limit can be switched with the sum of the v2's memory and swap limits. However the alternative for memsw usage is not yet available in cgroup v2. Exposing per-cgroup swapcache stat enables that alternative. Adding the memory usage and swap usage and subtracting the swapcache will approximate the memsw usage. This will help in the transparent migration of the workloads depending on memsw usage and limit to v2' memory and swap counters. The reasons these applications are still interested in this approximate memsw usage are: (1) these applications are not really interested in two separate memory and swap usage metrics. A single usage metric is more simple to use and reason about for them. (2) The memsw usage metric hides the underlying system's swap setup from the applications. Applications with multiple instances running in a datacenter with heterogeneous systems (some have swap and some don't) will keep seeing a consistent view of their usage. [akpm@linux-foundation.org: fix CONFIG_SWAP=n build] Link: https://lkml.kernel.org/r/20210108155813.2914586-3-shakeelb@google.comSigned-off-by:
-
Alex Shi authored
lock_page_lruvec() and its variants used rcu_read_lock() with the intention of safeguarding against the mem_cgroup being destroyed concurrently; but so long as they are called under the specified conditions (as they are), there is no way for the page's mem_cgroup to be destroyed. Delete the unnecessary rcu_read_lock() and _unlock(). Hugh Dickins polished the commit log. Thanks a lot! Link: https://lkml.kernel.org/r/1608614453-10739-2-git-send-email-alex.shi@linux.alibaba.comSigned-off-by:
Alex Shi <alex.shi@linux.alibaba.com> Acked-by:
Hugh Dickins <hughd@google.com> Cc: Hugh Dickins <hughd@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Michal Hocko <mhocko@kernel.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Signed-off-by:
-
Alex Shi authored
lock_page_lruvec() and its variants are safe to use under the same conditions as commit_charge(): add lock_page_memcg() to the comment. Polished with Hugh Dickins' suggestions, thanks! Link: https://lkml.kernel.org/r/1608614453-10739-1-git-send-email-alex.shi@linux.alibaba.comSigned-off-by:
-
Muchun Song authored
Although the ratio of the slab is one, we also should read the ratio from the related memory_stats instead of hard-coding. And the local variable of size is already the value of slab_unreclaimable. So we do not need to read again. To do this we need some code like below: if (unlikely(memory_stats[i].idx == NR_SLAB_UNRECLAIMABLE_B)) { - size = memcg_page_state(memcg, NR_SLAB_RECLAIMABLE_B) + - memcg_page_state(memcg, NR_SLAB_UNRECLAIMABLE_B); + VM_BUG_ON(i < 1); + VM_BUG_ON(memory_stats[i - 1].idx != NR_SLAB_RECLAIMABLE_B); + size += memcg_page_state(memcg, memory_stats[i - 1].idx) * + memory_stats[i - 1].ratio; It requires a series of VM_BUG_ONs or comments to ensure these two items are actually adjacent and in the right order. So it would probably be easier to implement this using a wrapper that has a big switch() for unit conversion. More details about this discussion can refer to: https://lore.kernel.org/patchwork/patch/1348611/ This would fix the ratio inconsistency and get rid of the order guarantee. Link: https://lkml.kernel.org/r/20201228164110.2838-8-songmuchun@bytedance.comSigned-off-by: -
Muchun Song authored
Currently we use struct per_cpu_nodestat to cache the vmstat counters, which leads to inaccurate statistics especially THP vmstat counters. In the systems with hundreds of processors it can be GBs of memory. For example, for a 96 CPUs system, the threshold is the maximum number of 125. And the per cpu counters can cache 23.4375 GB in total. The THP page is already a form of batched addition (it will add 512 worth of memory in one go) so skipping the batching seems like sensible. Although every THP stats update overflows the per-cpu counter, resorting to atomic global updates. But it can make the statistics more accuracy for the THP vmstat counters. So we convert the NR_FILE_PMDMAPPED account to pages. This patch is consistent with 8f182270 ("mm/swap.c: flush lru pvecs on compound page arrival"). Doing this also can make the unit of vmstat counters more unified. Finally, the unit of the vmstat counters are pages, kB and bytes. The B/KB suffix can tell us that the unit is bytes or kB. The rest which is without suffix are pages. Link: https://lkml.kernel.org/r/20201228164110.2838-7-songmuchun@bytedance.comSigned-off-by:
-
Muchun Song authored
Currently we use struct per_cpu_nodestat to cache the vmstat counters, which leads to inaccurate statistics especially THP vmstat counters. In the systems with hundreds of processors it can be GBs of memory. For example, for a 96 CPUs system, the threshold is the maximum number of 125. And the per cpu counters can cache 23.4375 GB in total. The THP page is already a form of batched addition (it will add 512 worth of memory in one go) so skipping the batching seems like sensible. Although every THP stats update overflows the per-cpu counter, resorting to atomic global updates. But it can make the statistics more accuracy for the THP vmstat counters. So we convert the NR_SHMEM_PMDMAPPED account to pages. This patch is consistent with 8f182270 ("mm/swap.c: flush lru pvecs on compound page arrival"). Doing this also can make the unit of vmstat counters more unified. Finally, the unit of the vmstat counters are pages, kB and bytes. The B/KB suffix can tell us that the unit is bytes or kB. The rest which is without suffix are pages. Link: https://lkml.kernel.org/r/20201228164110.2838-6-songmuchun@bytedance.comSigned-off-by:
-
Muchun Song authored
Currently we use struct per_cpu_nodestat to cache the vmstat counters, which leads to inaccurate statistics especially THP vmstat counters. In the systems with hundreds of processors it can be GBs of memory. For example, for a 96 CPUs system, the threshold is the maximum number of 125. And the per cpu counters can cache 23.4375 GB in total. The THP page is already a form of batched addition (it will add 512 worth of memory in one go) so skipping the batching seems like sensible. Although every THP stats update overflows the per-cpu counter, resorting to atomic global updates. But it can make the statistics more accuracy for the THP vmstat counters. So we convert the NR_SHMEM_THPS account to pages. This patch is consistent with 8f182270 ("mm/swap.c: flush lru pvecs on compound page arrival"). Doing this also can make the unit of vmstat counters more unified. Finally, the unit of the vmstat counters are pages, kB and bytes. The B/KB suffix can tell us that the unit is bytes or kB. The rest which is without suffix are pages. Link: https://lkml.kernel.org/r/20201228164110.2838-5-songmuchun@bytedance.comSigned-off-by:
-
