- 03 Sep, 2021 40 commits
-
-
Suren Baghdasaryan authored
Split off from prev patch in the series that implements the syscall. Link: https://lkml.kernel.org/r/20210809185259.405936-2-surenb@google.comSigned-off-by:
Suren Baghdasaryan <surenb@google.com> Acked-by:
Geert Uytterhoeven <geert@linux-m68k.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: David Hildenbrand <david@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Florian Weimer <fweimer@redhat.com> Cc: Jan Engelhardt <jengelh@inai.de> Cc: Jann Horn <jannh@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Michal Hocko <mhocko@suse.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Rik van Riel <riel@surriel.com> Cc: Roman Gushchin <guro@fb.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Tim Murray <timmurray@google.com> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Suren Baghdasaryan authored
In modern systems it's not unusual to have a system component monitoring memory conditions of the system and tasked with keeping system memory pressure under control. One way to accomplish that is to kill non-essential processes to free up memory for more important ones. Examples of this are Facebook's OOM killer daemon called oomd and Android's low memory killer daemon called lmkd. For such system component it's important to be able to free memory quickly and efficiently. Unfortunately the time process takes to free up its memory after receiving a SIGKILL might vary based on the state of the process (uninterruptible sleep), size and OPP level of the core the process is running. A mechanism to free resources of the target process in a more predictable way would improve system's ability to control its memory pressure. Introduce process_mrelease system call that releases memory of a dying process from the context of the caller. This way the memory is freed in a more controllable way with CPU affinity and priority of the caller. The workload of freeing the memory will also be charged to the caller. The operation is allowed only on a dying process. After previous discussions [1, 2, 3] the decision was made [4] to introduce a dedicated system call to cover this use case. The API is as follows, int process_mrelease(int pidfd, unsigned int flags); DESCRIPTION The process_mrelease() system call is used to free the memory of an exiting process. The pidfd selects the process referred to by the PID file descriptor. (See pidfd_open(2) for further information) The flags argument is reserved for future use; currently, this argument must be specified as 0. RETURN VALUE On success, process_mrelease() returns 0. On error, -1 is returned and errno is set to indicate the error. ERRORS EBADF pidfd is not a valid PID file descriptor. EAGAIN Failed to release part of the address space. EINTR The call was interrupted by a signal; see signal(7). EINVAL flags is not 0. EINVAL The memory of the task cannot be released because the process is not exiting, the address space is shared with another live process or there is a core dump in progress. ENOSYS This system call is not supported, for example, without MMU support built into Linux. ESRCH The target process does not exist (i.e., it has terminated and been waited on). [1] https://lore.kernel.org/lkml/20190411014353.113252-3-surenb@google.com/ [2] https://lore.kernel.org/linux-api/20201113173448.1863419-1-surenb@google.com/ [3] https://lore.kernel.org/linux-api/20201124053943.1684874-3-surenb@google.com/ [4] https://lore.kernel.org/linux-api/20201223075712.GA4719@lst.de/ Link: https://lkml.kernel.org/r/20210809185259.405936-1-surenb@google.comSigned-off-by:Shakeel Butt <shakeelb@google.com> Acked-by:
David Hildenbrand <david@redhat.com> Acked-by:
Michal Hocko <mhocko@suse.com> Acked-by:
Christian Brauner <christian.brauner@ubuntu.com> Cc: David Rientjes <rientjes@google.com> Cc: Matthew Wilcox (Oracle) <willy@infradead.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Roman Gushchin <guro@fb.com> Cc: Rik van Riel <riel@surriel.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Christoph Hellwig <hch@infradead.org> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Jann Horn <jannh@google.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Christian Brauner <christian.brauner@ubuntu.com> Cc: Florian Weimer <fweimer@redhat.com> Cc: Jan Engelhardt <jengelh@inai.de> Cc: Tim Murray <timmurray@google.com> Signed-off-by:
-
Mike Rapoport authored
There are a lot of uses of memblock_find_in_range() along with memblock_reserve() from the times memblock allocation APIs did not exist. memblock_find_in_range() is the very core of memblock allocations, so any future changes to its internal behaviour would mandate updates of all the users outside memblock. Replace the calls to memblock_find_in_range() with an equivalent calls to memblock_phys_alloc() and memblock_phys_alloc_range() and make memblock_find_in_range() private method of memblock. This simplifies the callers, ensures that (unlikely) errors in memblock_reserve() are handled and improves maintainability of memblock_find_in_range(). Link: https://lkml.kernel.org/r/20210816122622.30279-1-rppt@kernel.orgSigned-off-by:
Mike Rapoport <rppt@linux.ibm.com> Reviewed-by: Catalin Marinas <catalin.marinas@arm.com> [arm64] Acked-by:
Kirill A. Shutemov <kirill.shtuemov@linux.intel.com> Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> [ACPI] Acked-by:
Russell King (Oracle) <rmk+kernel@armlinux.org.uk> Acked-by: Nick Kossifidis <mick@ics.forth.gr> [riscv] Tested-by:
Guenter Roeck <linux@roeck-us.net> Acked-by:
Rob Herring <robh@kernel.org> Signed-off-by:
-
Vasily Averin authored
Obsoleted in_intrrupt() include task context with disabled BH, it's better to use in_task() instead. Link: https://lkml.kernel.org/r/984ee771-4834-21da-801f-c15c18ddf4d1@virtuozzo.comSigned-off-by:
Vasily Averin <vvs@virtuozzo.com> Signed-off-by:
-
Feng Tang authored
As they all do the same thing: sanity check and save nodemask info, create one mpol_new_nodemask() to reduce redundancy. Link: https://lkml.kernel.org/r/1627970362-61305-6-git-send-email-feng.tang@intel.comSigned-off-by:
Feng Tang <feng.tang@intel.com> Acked-by:
-
Ben Widawsky authored
Adds a new mode to the existing mempolicy modes, MPOL_PREFERRED_MANY. MPOL_PREFERRED_MANY will be adequately documented in the internal admin-guide with this patch. Eventually, the man pages for mbind(2), get_mempolicy(2), set_mempolicy(2) and numactl(8) will also have text about this mode. Those shall contain the canonical reference. NUMA systems continue to become more prevalent. New technologies like PMEM make finer grain control over memory access patterns increasingly desirable. MPOL_PREFERRED_MANY allows userspace to specify a set of nodes that will be tried first when performing allocations. If those allocations fail, all remaining nodes will be tried. It's a straight forward API which solves many of the presumptive needs of system administrators wanting to optimize workloads on such machines. The mode will work either per VMA, or per thread. [Michal Hocko: refine kernel doc for MPOL_PREFERRED_MANY] Link: https://lore.kernel.org/r/20200630212517.308045-13-ben.widawsky@intel.com Link: https://lkml.kernel.org/r/1627970362-61305-5-git-send-email-feng.tang@intel.comSigned-off-by:
Ben Widawsky <ben.widawsky@intel.com> Signed-off-by:
-
Ben Widawsky authored
Implement the missing huge page allocation functionality while obeying the preferred node semantics. This is similar to the implementation for general page allocation, as it uses a fallback mechanism to try multiple preferred nodes first, and then all other nodes. To avoid adding too many "#ifdef CONFIG_NUMA" check, add a helper function in mempolicy.h to check whether a mempolicy is MPOL_PREFERRED_MANY. [akpm@linux-foundation.org: fix compiling issue when merging with other hugetlb patch] [Thanks to 0day bot for catching the !CONFIG_NUMA compiling issue] [mhocko@suse.com: suggest to remove the #ifdef CONFIG_NUMA check] [ben.widawsky@intel.com: add helpers to avoid ifdefs] Link: https://lore.kernel.org/r/20200630212517.308045-12-ben.widawsky@intel.com Link: https://lkml.kernel.org/r/1627970362-61305-4-git-send-email-feng.tang@intel.com Link: https://lkml.kernel.org/r/20210809024430.GA46432@shbuild999.sh.intel.com [nathan@kernel.org: initialize page to NULL in alloc_buddy_huge_page_with_mpol()] Link: https://lkml.kernel.org/r/20210810200632.3812797-1-nathan@kernel.org Link: https://lore.kernel.org/r/20200630212517.308045-12-ben.widawsky@intel.com Link: https://lkml.kernel.org/r/1627970362-61305-4-git-send-email-feng.tang@intel.com Link: https://lkml.kernel.org/r/20210809024430.GA46432@shbuild999.sh.intel.comSigned-off-by:
Nathan Chancellor <nathan@kernel.org> Co-developed-by:
-
Feng Tang authored
The semantics of MPOL_PREFERRED_MANY is similar to MPOL_PREFERRED, that it will first try to allocate memory from the preferred node(s), and fallback to all nodes in system when first try fails. Add a dedicated function alloc_pages_preferred_many() for it just like for 'interleave' policy, which will be used by 2 general memoory allocation APIs: alloc_pages() and alloc_pages_vma() Link: https://lore.kernel.org/r/20200630212517.308045-9-ben.widawsky@intel.com Link: https://lkml.kernel.org/r/1627970362-61305-3-git-send-email-feng.tang@intel.comSuggested-by:
-
Dave Hansen authored
Patch series "Introduce multi-preference mempolicy", v7. This patch series introduces the concept of the MPOL_PREFERRED_MANY mempolicy. This mempolicy mode can be used with either the set_mempolicy(2) or mbind(2) interfaces. Like the MPOL_PREFERRED interface, it allows an application to set a preference for nodes which will fulfil memory allocation requests. Unlike the MPOL_PREFERRED mode, it takes a set of nodes. Like the MPOL_BIND interface, it works over a set of nodes. Unlike MPOL_BIND, it will not cause a SIGSEGV or invoke the OOM killer if those preferred nodes are not available. Along with these patches are patches for libnuma, numactl, numademo, and memhog. They still need some polish, but can be found here: https://gitlab.com/bwidawsk/numactl/-/tree/prefer-many It allows new usage: `numactl -P 0,3,4` The goal of the new mode is to enable some use-cases when using tiered memory usage models which I've lovingly named. 1a. The Hare - The interconnect is fast enough to meet bandwidth and latency requirements allowing preference to be given to all nodes with "fast" memory. 1b. The Indiscriminate Hare - An application knows it wants fast memory (or perhaps slow memory), but doesn't care which node it runs on. The application can prefer a set of nodes and then xpu bind to the local node (cpu, accelerator, etc). This reverses the nodes are chosen today where the kernel attempts to use local memory to the CPU whenever possible. This will attempt to use the local accelerator to the memory. 2. The Tortoise - The administrator (or the application itself) is aware it only needs slow memory, and so can prefer that. Much of this is almost achievable with the bind interface, but the bind interface suffers from an inability to fallback to another set of nodes if binding fails to all nodes in the nodemask. Like MPOL_BIND a nodemask is given. Inherently this removes ordering from the preference. > /* Set first two nodes as preferred in an 8 node system. */ > const unsigned long nodes = 0x3 > set_mempolicy(MPOL_PREFER_MANY, &nodes, 8); > /* Mimic interleave policy, but have fallback *. > const unsigned long nodes = 0xaa > set_mempolicy(MPOL_PREFER_MANY, &nodes, 8); Some internal discussion took place around the interface. There are two alternatives which we have discussed, plus one I stuck in: 1. Ordered list of nodes. Currently it's believed that the added complexity is nod needed for expected usecases. 2. A flag for bind to allow falling back to other nodes. This confuses the notion of binding and is less flexible than the current solution. 3. Create flags or new modes that helps with some ordering. This offers both a friendlier API as well as a solution for more customized usage. It's unknown if it's worth the complexity to support this. Here is sample code for how this might work: > // Prefer specific nodes for some something wacky > set_mempolicy(MPOL_PREFER_MANY, 0x17c, 1024); > > // Default > set_mempolicy(MPOL_PREFER_MANY | MPOL_F_PREFER_ORDER_SOCKET, NULL, 0); > // which is the same as > set_mempolicy(MPOL_DEFAULT, NULL, 0); > > // The Hare > set_mempolicy(MPOL_PREFER_MANY | MPOL_F_PREFER_ORDER_TYPE, NULL, 0); > > // The Tortoise > set_mempolicy(MPOL_PREFER_MANY | MPOL_F_PREFER_ORDER_TYPE_REV, NULL, 0); > > // Prefer the fast memory of the first two sockets > set_mempolicy(MPOL_PREFER_MANY | MPOL_F_PREFER_ORDER_TYPE, -1, 2); > This patch (of 5): The NUMA APIs currently allow passing in a "preferred node" as a single bit set in a nodemask. If more than one bit it set, bits after the first are ignored. This single node is generally OK for location-based NUMA where memory being allocated will eventually be operated on by a single CPU. However, in systems with multiple memory types, folks want to target a *type* of memory instead of a location. For instance, someone might want some high-bandwidth memory but do not care about the CPU next to which it is allocated. Or, they want a cheap, high capacity allocation and want to target all NUMA nodes which have persistent memory in volatile mode. In both of these cases, the application wants to target a *set* of nodes, but does not want strict MPOL_BIND behavior as that could lead to OOM killer or SIGSEGV. So add MPOL_PREFERRED_MANY policy to support the multiple preferred nodes requirement. This is not a pie-in-the-sky dream for an API. This was a response to a specific ask of more than one group at Intel. Specifically: 1. There are existing libraries that target memory types such as https://github.com/memkind/memkind. These are known to suffer from SIGSEGV's when memory is low on targeted memory "kinds" that span more than one node. The MCDRAM on a Xeon Phi in "Cluster on Die" mode is an example of this. 2. Volatile-use persistent memory users want to have a memory policy which is targeted at either "cheap and slow" (PMEM) or "expensive and fast" (DRAM). However, they do not want to experience allocation failures when the targeted type is unavailable. 3. Allocate-then-run. Generally, we let the process scheduler decide on which physical CPU to run a task. That location provides a default allocation policy, and memory availability is not generally considered when placing tasks. For situations where memory is valuable and constrained, some users want to allocate memory first, *then* allocate close compute resources to the allocation. This is the reverse of the normal (CPU) model. Accelerators such as GPUs that operate on core-mm-managed memory are interested in this model. A check is added in sanitize_mpol_flags() to not permit 'prefer_many' policy to be used for now, and will be removed in later patch after all implementations for 'prefer_many' are ready, as suggested by Michal Hocko. [mhocko@kernel.org: suggest to refine policy_node/policy_nodemask handling] Link: https://lkml.kernel.org/r/1627970362-61305-1-git-send-email-feng.tang@intel.com Link: https://lore.kernel.org/r/20200630212517.308045-4-ben.widawsky@intel.com Link: https://lkml.kernel.org/r/1627970362-61305-2-git-send-email-feng.tang@intel.comCo-developed-by:
Dave Hansen <dave.hansen@linux.intel.com> Signed-off-by:
-
Baolin Wang authored
The caller of mpol_misplaced() already use NUMA_NO_NODE to check whether current page node is misplaced, thus using NUMA_NO_NODE in mpol_misplaced() instead of magic number is more readable. Link: https://lkml.kernel.org/r/1b77c0ce21183fa86f4db250b115cf5e27396528.1627558356.git.baolin.wang@linux.alibaba.comSigned-off-by:
Baolin Wang <baolin.wang@linux.alibaba.com> Reviewed-by:
-
Charan Teja Reddy authored
The proactive compaction[1] gets triggered for every 500msec and run compaction on the node for COMPACTION_HPAGE_ORDER (usually order-9) pages based on the value set to sysctl.compaction_proactiveness. Triggering the compaction for every 500msec in search of COMPACTION_HPAGE_ORDER pages is not needed for all applications, especially on the embedded system usecases which may have few MB's of RAM. Enabling the proactive compaction in its state will endup in running almost always on such systems. Other side, proactive compaction can still be very much useful for getting a set of higher order pages in some controllable manner(controlled by using the sysctl.compaction_proactiveness). So, on systems where enabling the proactive compaction always may proove not required, can trigger the same from user space on write to its sysctl interface. As an example, say app launcher decide to launch the memory heavy application which can be launched fast if it gets more higher order pages thus launcher can prepare the system in advance by triggering the proactive compaction from userspace. This triggering of proactive compaction is done on a write to sysctl.compaction_proactiveness by user. [1]https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit?id=facdaa917c4d5a376d09d25865f5a863f906234a [akpm@linux-foundation.org: tweak vm.rst, per Mike] Link: https://lkml.kernel.org/r/1627653207-12317-1-git-send-email-charante@codeaurora.orgSigned-off-by:
Charan Teja Reddy <charante@codeaurora.org> Acked-by:
Vlastimil Babka <vbabka@suse.cz> Acked-by:
Rafael Aquini <aquini@redhat.com> Cc: Mike Rapoport <rppt@kernel.org> Cc: Luis Chamberlain <mcgrof@kernel.org> Cc: Kees Cook <keescook@chromium.org> Cc: Iurii Zaikin <yzaikin@google.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: Nitin Gupta <nigupta@nvidia.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Khalid Aziz <khalid.aziz@oracle.com> Cc: David Rientjes <rientjes@google.com> Cc: Vinayak Menon <vinmenon@codeaurora.org> Signed-off-by:
-
Charan Teja Reddy authored
Vlastimil Babka figured out that when fragmentation score didn't go down across the proactive compaction i.e. when no progress is made, next wake up for proactive compaction is deferred for 1 << COMPACT_MAX_DEFER_SHIFT, i.e. 64 times, with each wakeup interval of HPAGE_FRAG_CHECK_INTERVAL_MSEC(=500). In each of this wakeup, it just decrement 'proactive_defer' counter and goes sleep i.e. it is getting woken to just decrement a counter. The same deferral time can also achieved by simply doing the HPAGE_FRAG_CHECK_INTERVAL_MSEC << COMPACT_MAX_DEFER_SHIFT thus unnecessary wakeup of kcompact thread is avoided thus also removes the need of 'proactive_defer' thread counter. [akpm@linux-foundation.org: tweak comment] Link: https://lore.kernel.org/linux-fsdevel/88abfdb6-2c13-b5a6-5b46-742d12d1c910@suse.cz/ Link: https://lkml.kernel.org/r/1626869599-25412-1-git-send-email-charante@codeaurora.orgSigned-off-by:
Khalid Aziz <khalid.aziz@oracle.com> Acked-by:
David Rientjes <rientjes@google.com> Cc: Nitin Gupta <nigupta@nvidia.com> Cc: Vinayak Menon <vinmenon@codeaurora.org> Signed-off-by:
-
Vlastimil Babka authored
drop_slab_node() is called as part of echo 2>/proc/sys/vm/drop_caches operation. It iterates over all memcgs and calls shrink_slab() which in turn iterates over all slab shrinkers. Freed objects are counted and as long as the total number of freed objects from all memcgs and shrinkers is higher than 10, drop_slab_node() loops for another full memcgs*shrinkers iteration. This arbitrary constant threshold of 10 can result in effectively an infinite loop on a system with large number of memcgs and/or parallel activity that allocates new objects. This has been reported previously by Chunxin Zang [1] and recently by our customer. The previous report [1] has resulted in commit 069c411d ("mm/vmscan: fix infinite loop in drop_slab_node") which added a check for signals allowing the user to terminate the command writing to drop_caches. At the time it was also considered to make the threshold grow with each iteration to guarantee termination, but such patch hasn't been formally proposed yet. This patch implements the dynamically growing threshold. At first iteration it's enough to free one object to continue, and this threshold effectively doubles with each iteration. Our customer's feedback was positive. There is always a risk that this change will result on some system in a previously terminating drop_caches operation to terminate sooner and free fewer objects. Ideally the semantics would guarantee freeing all freeable objects that existed at the moment of starting the operation, while not looping forever for newly allocated objects, but that's not feasible to track. In the less ideal solution based on thresholds, arguably the termination guarantee is more important than the exhaustiveness guarantee. If there are reports of large regression wrt being exhaustive, we can tune how fast the threshold grows. [1] https://lore.kernel.org/lkml/20200909152047.27905-1-zangchunxin@bytedance.com/T/#u [vbabka@suse.cz: avoid undefined shift behaviour] Link: https://lkml.kernel.org/r/2f034e6f-a753-550a-f374-e4e23899d3d5@suse.cz Link: https://lkml.kernel.org/r/20210818152239.25502-1-vbabka@suse.czSigned-off-by:
Chunxin Zang <zangchunxin@bytedance.com> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Chris Down <chris@chrisdown.name> Cc: Michal Hocko <mhocko@kernel.org> Cc: Matthew Wilcox <willy@infradead.org> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Signed-off-by:
-
Miaohe Lin authored
We could add 'else' to remove the somewhat odd check_pending label to make code core succinct. Link: https://lkml.kernel.org/r/20210717065911.61497-5-linmiaohe@huawei.comSigned-off-by:
Miaohe Lin <linmiaohe@huawei.com> Acked-by:
-
Miaohe Lin authored
The return value of kswapd_run() is unused now. Clean it up. Link: https://lkml.kernel.org/r/20210717065911.61497-4-linmiaohe@huawei.comSigned-off-by:
-
Miaohe Lin authored
The priority field of sc is used to control how many pages we should scan at once while we always traverse the list to shrink the pages in these functions. So these settings are unneeded and misleading. Link: https://lkml.kernel.org/r/20210717065911.61497-3-linmiaohe@huawei.comSigned-off-by:
-
Miaohe Lin authored
Patch series "Cleanups for vmscan", v2. This series contains cleanups to remove unneeded return value, misleading setting and so on. Also this remove the PageDirty check after MADV_FREE pages are page_ref_freezed. More details can be found in the respective changelogs. This patch (of 4): If the MADV_FREE pages are redirtied before they could be reclaimed, put the pages back to anonymous LRU list by setting SwapBacked flag and the pages will be reclaimed in normal swapout way. But as Yu Zhao pointed out, "The page has only one reference left, which is from the isolation. After the caller puts the page back on lru and drops the reference, the page will be freed anyway. It doesn't matter which lru it goes." So we don't bother checking PageDirty here. [Yu Zhao's comment is also quoted in the code.] Link: https://lkml.kernel.org/r/20210717065911.61497-1-linmiaohe@huawei.com Link: https://lkml.kernel.org/r/20210717065911.61497-2-linmiaohe@huawei.comSigned-off-by:
Yu Zhao <yuzhao@google.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Michal Hocko <mhocko@suse.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Alex Shi <alexs@kernel.org> Cc: Alistair Popple <apopple@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Minchan Kim <minchan@kernel.org> Cc: David Hildenbrand <david@redhat.com> Cc: Shaohua Li <shli@fb.com> Cc: Hillf Danton <hillf.zj@alibaba-inc.com> Cc: John Hubbard <jhubbard@nvidia.com> Signed-off-by:
-
Hui Su authored
We can get memcg directly form vmpr instead of vmpr->memcg->css->memcg, so add a new func helper vmpressure_to_memcg(). And no code will use vmpressure_to_css(), so delete it. Link: https://lkml.kernel.org/r/20210630112146.455103-1-suhui@zeku.comSigned-off-by:
Hui Su <suhui@zeku.com> Acked-by:
Chris Down <chris@chrisdown.name> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Signed-off-by:
-
Huang Ying authored
Some method is obviously needed to enable reclaim-based migration. Just like traditional autonuma, there will be some workloads that will benefit like workloads with more "static" configurations where hot pages stay hot and cold pages stay cold. If pages come and go from the hot and cold sets, the benefits of this approach will be more limited. The benefits are truly workload-based and *not* hardware-based. We do not believe that there is a viable threshold where certain hardware configurations should have this mechanism enabled while others do not. To be conservative, earlier work defaulted to disable reclaim- based migration and did not include a mechanism to enable it. This proposes add a new sysfs file /sys/kernel/mm/numa/demotion_enabled as a method to enable it. We are open to any alternative that allows end users to enable this mechanism or disable it if workload harm is detected (just like traditional autonuma). Once this is enabled page demotion may move data to a NUMA node that does not fall into the cpuset of the allocating process. This could be construed to violate the guarantees of cpusets. However, since this is an opt-in mechanism, the assumption is that anyone enabling it is content to relax the guarantees. Link: https://lkml.kernel.org/r/20210721063926.3024591-9-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210715055145.195411-10-ying.huang@intel.comSigned-off-by:
Huang Ying <ying.huang@intel.com> Originally-by:
-
Dave Hansen authored
Global reclaim aims to reduce the amount of memory used on a given node or set of nodes. Migrating pages to another node serves this purpose. memcg reclaim is different. Its goal is to reduce the total memory consumption of the entire memcg, across all nodes. Migration does not assist memcg reclaim because it just moves page contents between nodes rather than actually reducing memory consumption. Link: https://lkml.kernel.org/r/20210715055145.195411-9-ying.huang@intel.comSigned-off-by:
Yang Shi <yang.shi@linux.alibaba.com> Reviewed-by:
Yang Shi <shy828301@gmail.com> Reviewed-by:
Zi Yan <ziy@nvidia.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Wei Xu <weixugc@google.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: David Rientjes <rientjes@google.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: David Hildenbrand <david@redhat.com> Cc: Greg Thelen <gthelen@google.com> Cc: Keith Busch <kbusch@kernel.org> Signed-off-by:
-
Keith Busch authored
Reclaim anonymous pages if a migration path is available now that demotion provides a non-swap recourse for reclaiming anon pages. Note that this check is subtly different from the can_age_anon_pages() checks. This mechanism checks whether a specific page in a specific context can actually be reclaimed, given current swap space and cgroup limits. can_age_anon_pages() is a much simpler and more preliminary check which just says whether there is a possibility of future reclaim. [kbusch@kernel.org: v11] Link: https://lkml.kernel.org/r/20210715055145.195411-8-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210721063926.3024591-7-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210715055145.195411-8-ying.huang@intel.com Cc: Keith Busch <kbusch@kernel.org> Signed-off-by:
-
Dave Hansen authored
Anonymous pages are kept on their own LRU(s). These lists could theoretically always be scanned and maintained. But, without swap, there is currently nothing the kernel can *do* with the results of a scanned, sorted LRU for anonymous pages. A check for '!total_swap_pages' currently serves as a valid check as to whether anonymous LRUs should be maintained. However, another method will be added shortly: page demotion. Abstract out the 'total_swap_pages' checks into a helper, give it a logically significant name, and check for the possibility of page demotion. [dave.hansen@linux.intel.com: v11] Link: https://lkml.kernel.org/r/20210715055145.195411-7-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210721063926.3024591-6-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210715055145.195411-7-ying.huang@intel.comSigned-off-by:
Greg Thelen <gthelen@google.com> Reviewed-by:
-
Yang Shi authored
Account the number of demoted pages. Add pgdemote_kswapd and pgdemote_direct VM counters showed in /proc/vmstat. [ daveh: - __count_vm_events() a bit, and made them look at the THP size directly rather than getting data from migrate_pages() ] Link: https://lkml.kernel.org/r/20210721063926.3024591-5-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210715055145.195411-6-ying.huang@intel.comSigned-off-by:Wei Xu <weixugc@google.com> Reviewed-by:
-
Dave Hansen authored
This is mostly derived from a patch from Yang Shi: https://lore.kernel.org/linux-mm/1560468577-101178-10-git-send-email-yang.shi@linux.alibaba.com/ Add code to the reclaim path (shrink_page_list()) to "demote" data to another NUMA node instead of discarding the data. This always avoids the cost of I/O needed to read the page back in and sometimes avoids the writeout cost when the page is dirty. A second pass through shrink_page_list() will be made if any demotions fail. This essentially falls back to normal reclaim behavior in the case that demotions fail. Previous versions of this patch may have simply failed to reclaim pages which were eligible for demotion but were unable to be demoted in practice. For some cases, for example, MADV_PAGEOUT, the pages are always discarded instead of demoted to follow the kernel API definition. Because MADV_PAGEOUT is defined as freeing specified pages regardless in which tier they are. Note: This just adds the start of infrastructure for migration. It is actually disabled next to the FIXME in migrate_demote_page_ok(). [dave.hansen@linux.intel.com: v11] Link: https://lkml.kernel.org/r/20210715055145.195411-5-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210721063926.3024591-4-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210715055145.195411-5-ying.huang@intel.comSigned-off-by:
Oscar Salvador <osalvador@suse.de> Reviewed-by:
-
Yang Shi authored
Under normal circumstances, migrate_pages() returns the number of pages migrated. In error conditions, it returns an error code. When returning an error code, there is no way to know how many pages were migrated or not migrated. Make migrate_pages() return how many pages are demoted successfully for all cases, including when encountering errors. Page reclaim behavior will depend on this in subsequent patches. Link: https://lkml.kernel.org/r/20210721063926.3024591-3-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210715055145.195411-4-ying.huang@intel.comSigned-off-by:
-
Dave Hansen authored
Reclaim-based migration is attempting to optimize data placement in memory based on the system topology. If the system changes, so must the migration ordering. The implementation is conceptually simple and entirely unoptimized. On any memory or CPU hotplug events, assume that a node was added or removed and recalculate all migration targets. This ensures that the node_demotion[] array is always ready to be used in case the new reclaim mode is enabled. This recalculation is far from optimal, most glaringly that it does not even attempt to figure out the hotplug event would have some *actual* effect on the demotion order. But, given the expected paucity of hotplug events, this should be fine. Link: https://lkml.kernel.org/r/20210721063926.3024591-2-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210715055145.195411-3-ying.huang@intel.comSigned-off-by:
-
Dave Hansen authored
Patch series "Migrate Pages in lieu of discard", v11. We're starting to see systems with more and more kinds of memory such as Intel's implementation of persistent memory. Let's say you have a system with some DRAM and some persistent memory. Today, once DRAM fills up, reclaim will start and some of the DRAM contents will be thrown out. Allocations will, at some point, start falling over to the slower persistent memory. That has two nasty properties. First, the newer allocations can end up in the slower persistent memory. Second, reclaimed data in DRAM are just discarded even if there are gobs of space in persistent memory that could be used. This patchset implements a solution to these problems. At the end of the reclaim process in shrink_page_list() just before the last page refcount is dropped, the page is migrated to persistent memory instead of being dropped. While I've talked about a DRAM/PMEM pairing, this approach would function in any environment where memory tiers exist. This is not perfect. It "strands" pages in slower memory and never brings them back to fast DRAM. Huang Ying has follow-on work which repurposes NUMA balancing to promote hot pages back to DRAM. This is also all based on an upstream mechanism that allows persistent memory to be onlined and used as if it were volatile: http://lkml.kernel.org/r/20190124231441.37A4A305@viggo.jf.intel.com With that, the DRAM and PMEM in each socket will be represented as 2 separate NUMA nodes, with the CPUs sit in the DRAM node. So the general inter-NUMA demotion mechanism introduced in the patchset can migrate the cold DRAM pages to the PMEM node. We have tested the patchset with the postgresql and pgbench. On a 2-socket server machine with DRAM and PMEM, the kernel with the patchset can improve the score of pgbench up to 22.1% compared with that of the DRAM only + disk case. This comes from the reduced disk read throughput (which reduces up to 70.8%). == Open Issues == * Memory policies and cpusets that, for instance, restrict allocations to DRAM can be demoted to PMEM whenever they opt in to this new mechanism. A cgroup-level API to opt-in or opt-out of these migrations will likely be required as a follow-on. * Could be more aggressive about where anon LRU scanning occurs since it no longer necessarily involves I/O. get_scan_count() for instance says: "If we have no swap space, do not bother scanning anon pages" This patch (of 9): Prepare for the kernel to auto-migrate pages to other memory nodes with a node migration table. This allows creating single migration target for each NUMA node to enable the kernel to do NUMA page migrations instead of simply discarding colder pages. A node with no target is a "terminal node", so reclaim acts normally there. The migration target does not fundamentally _need_ to be a single node, but this implementation starts there to limit complexity. When memory fills up on a node, memory contents can be automatically migrated to another node. The biggest problems are knowing when to migrate and to where the migration should be targeted. The most straightforward way to generate the "to where" list would be to follow the page allocator fallback lists. Those lists already tell us if memory is full where to look next. It would also be logical to move memory in that order. But, the allocator fallback lists have a fatal flaw: most nodes appear in all the lists. This would potentially lead to migration cycles (A->B, B->A, A->B, ...). Instead of using the allocator fallback lists directly, keep a separate node migration ordering. But, reuse the same data used to generate page allocator fallback in the first place: find_next_best_node(). This means that the firmware data used to populate node distances essentially dictates the ordering for now. It should also be architecture-neutral since all NUMA architectures have a working find_next_best_node(). RCU is used to allow lock-less read of node_demotion[] and prevent demotion cycles been observed. If multiple reads of node_demotion[] are performed, a single rcu_read_lock() must be held over all reads to ensure no cycles are observed. Details are as follows. === What does RCU provide? === Imagine a simple loop which walks down the demotion path looking for the last node: terminal_node = start_node; while (node_demotion[terminal_node] != NUMA_NO_NODE) { terminal_node = node_demotion[terminal_node]; } The initial values are: node_demotion[0] = 1; node_demotion[1] = NUMA_NO_NODE; and are updated to: node_demotion[0] = NUMA_NO_NODE; node_demotion[1] = 0; What guarantees that the cycle is not observed: node_demotion[0] = 1; node_demotion[1] = 0; and would loop forever? With RCU, a rcu_read_lock/unlock() can be placed around the loop. Since the write side does a synchronize_rcu(), the loop that observed the old contents is known to be complete before the synchronize_rcu() has completed. RCU, combined with disable_all_migrate_targets(), ensures that the old migration state is not visible by the time __set_migration_target_nodes() is called. === What does READ_ONCE() provide? === READ_ONCE() forbids the compiler from merging or reordering successive reads of node_demotion[]. This ensures that any updates are *eventually* observed. Consider the above loop again. The compiler could theoretically read the entirety of node_demotion[] into local storage (registers) and never go back to memory, and *permanently* observe bad values for node_demotion[]. Note: RCU does not provide any universal compiler-ordering guarantees: https://lore.kernel.org/lkml/20150921204327.GH4029@linux.vnet.ibm.com/ This code is unused for now. It will be called later in the series. Link: https://lkml.kernel.org/r/20210721063926.3024591-1-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210715055145.195411-1-ying.huang@intel.com Link: https://lkml.kernel.org/r/20210715055145.195411-2-ying.huang@intel.comSigned-off-by:
-
Nadav Amit authored
When userfaultfd copy-ioctl fails since the PTE already exists, an -EEXIST error is returned and the faulting thread is not woken. The current userfaultfd test does not wake the faulting thread in such case. The assumption is presumably that another thread set the PTE through copy/wp ioctl and would wake the faulting thread or that alternatively the fault handler would realize there is no need to "must_wait" and continue. This is not necessarily true. There is an assumption that the "must_wait" tests in handle_userfault() are sufficient to provide definitive answer whether the offending PTE is populated or not. However, userfaultfd_must_wait() test is lockless. Consequently, concurrent calls to ptep_modify_prot_start(), for instance, can clear the PTE and can cause userfaultfd_must_wait() to wrongly assume it is not populated and a wait is needed. There are therefore 3 options: (1) Change the tests to wake on copy failure. (2) Wake faulting thread unconditionally on zero/copy ioctls before returning -EEXIST. (3) Change the userfaultfd_must_wait() to hold locks. This patch took the first approach, but the others are valid solutions with different tradeoffs. Link: https://lkml.kernel.org/r/20210808020724.1022515-4-namit@vmware.comSigned-off-by:Nadav Amit <namit@vmware.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Peter Xu <peterx@redhat.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: Mike Rapoport <rppt@linux.vnet.ibm.com> Signed-off-by:
-
Nadav Amit authored
userfaultfd assumes that the enabled features are set once and never changed after UFFDIO_API ioctl succeeded. However, currently, UFFDIO_API can be called concurrently from two different threads, succeed on both threads and leave userfaultfd's features in non-deterministic state. Theoretically, other uffd operations (ioctl's and page-faults) can be dispatched while adversely affected by such changes of features. Moreover, the writes to ctx->state and ctx->features are not ordered, which can - theoretically, again - let userfaultfd_ioctl() think that userfaultfd API completed, while the features are still not initialized. To avoid races, it is arguably best to get rid of ctx->state. Since there are only 2 states, record the API initialization in ctx->features as the uppermost bit and remove ctx->state. Link: https://lkml.kernel.org/r/20210808020724.1022515-3-namit@vmware.com Fixes: 9cd75c3c ("userfaultfd: non-cooperative: add ability to report non-PF events from uffd descriptor") Signed-off-by:
-
Nadav Amit authored
Patch series "userfaultfd: minor bug fixes". Three unrelated bug fixes. The first two addresses possible issues (not too theoretical ones), but I did not encounter them in practice. The third patch addresses a test bug that causes the test to fail on my system. It has been sent before as part of a bigger RFC. This patch (of 3): mmap_changing is currently a boolean variable, which is set and cleared without any lock that protects against concurrent modifications. mmap_changing is supposed to mark whether userfaultfd page-faults handling should be retried since mappings are undergoing a change. However, concurrent calls, for instance to madvise(MADV_DONTNEED), might cause mmap_changing to be false, although the remove event was still not read (hence acknowledged) by the user. Change mmap_changing to atomic_t and increase/decrease appropriately. Add a debug assertion to see whether mmap_changing is negative. Link: https://lkml.kernel.org/r/20210808020724.1022515-1-namit@vmware.com Link: https://lkml.kernel.org/r/20210808020724.1022515-2-namit@vmware.com Fixes: df2cc96e ("userfaultfd: prevent non-cooperative events vs mcopy_atomic races") Signed-off-by:
-
Mike Kravetz authored
Guillaume Morin reported hitting the following WARNING followed by GPF or NULL pointer deference either in cgroups_destroy or in the kill_css path.: percpu ref (css_release) <= 0 (-1) after switching to atomic WARNING: CPU: 23 PID: 130 at lib/percpu-refcount.c:196 percpu_ref_switch_to_atomic_rcu+0x127/0x130 CPU: 23 PID: 130 Comm: ksoftirqd/23 Kdump: loaded Tainted: G O 5.10.60 #1 RIP: 0010:percpu_ref_switch_to_atomic_rcu+0x127/0x130 Call Trace: rcu_core+0x30f/0x530 rcu_core_si+0xe/0x10 __do_softirq+0x103/0x2a2 run_ksoftirqd+0x2b/0x40 smpboot_thread_fn+0x11a/0x170 kthread+0x10a/0x140 ret_from_fork+0x22/0x30 Upon further examination, it was discovered that the css structure was associated with hugetlb reservations. For private hugetlb mappings the vma points to a reserve map that contains a pointer to the css. At mmap time, reservations are set up and a reference to the css is taken. This reference is dropped in the vma close operation; hugetlb_vm_op_close. However, if a vma is split no additional reference to the css is taken yet hugetlb_vm_op_close will be called twice for the split vma resulting in an underflow. Fix by taking another reference in hugetlb_vm_op_open. Note that the reference is only taken for the owner of the reserve map. In the more common fork case, the pointer to the reserve map is cleared for non-owning vmas. Link: https://lkml.kernel.org/r/20210830215015.155224-1-mike.kravetz@oracle.com Fixes: e9fe92ae ("hugetlb_cgroup: add reservation accounting for private mappings") Signed-off-by:Mike Kravetz <mike.kravetz@oracle.com> Reported-by:
Guillaume Morin <guillaume@morinfr.org> Suggested-by:
-
Mike Kravetz authored
When removing a hugetlb page from the pool the ref count is set to one (as the free page has no ref count) and compound page destructor is set to NULL_COMPOUND_DTOR. Since a subsequent call to free the hugetlb page will call __free_pages for non-gigantic pages and free_gigantic_page for gigantic pages the destructor is not used. However, consider the following race with code taking a speculative reference on the page: Thread 0 Thread 1 -------- -------- remove_hugetlb_page set_page_refcounted(page); set_compound_page_dtor(page, NULL_COMPOUND_DTOR); get_page_unless_zero(page) __update_and_free_page __free_pages(page, huge_page_order(h)); /* Note that __free_pages() will simply drop the reference to the page. */ put_page(page) __put_compound_page() destroy_compound_page NULL_COMPOUND_DTOR BUG: kernel NULL pointer dereference, address: 0000000000000000 To address this race, set the dtor to the normal compound page dtor for non-gigantic pages. The dtor for gigantic pages does not matter as gigantic pages are changed from a compound page to 'just a group of pages' before freeing. Hence, the destructor is not used. Link: https://lkml.kernel.org/r/20210809184832.18342-4-mike.kravetz@oracle.comSigned-off-by:Muchun Song <songmuchun@bytedance.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: David Hildenbrand <david@redhat.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Naoya Horiguchi <naoya.horiguchi@linux.dev> Cc: Mina Almasry <almasrymina@google.com> Signed-off-by:
-
Mike Kravetz authored
When discussing the possibility of inflated page ref counts, Muuchun Song pointed out this potential issue [1]. It is true that any code could potentially take a reference on a compound page after allocation and before it is converted to and put into use as a hugetlb page. Specifically, this could be done by any users of get_page_unless_zero. There are three areas of concern within hugetlb code. 1) When adding pages to the pool. In this case, new pages are allocated added to the pool by calling put_page to invoke the hugetlb destructor (free_huge_page). If there is an inflated ref count on the page, it will not be immediately added to the free list. It will only be added to the free list when the temporary ref count is dropped. This is deemed acceptable and will not be addressed. 2) A page is allocated for immediate use normally as a surplus page or migration target. In this case, the user of the page will also hold a reference. There is no issue as this is just like normal page ref counting. 3) A page is allocated and MUST be added to the free list to satisfy a reservation. One such example is gather_surplus_pages as pointed out by Muchun in [1]. More specifically, this case covers callers of enqueue_huge_page where the page reference count must be zero. This patch covers this third case. Three routines call enqueue_huge_page when the page reference count could potentially be inflated. They are: gather_surplus_pages, alloc_and_dissolve_huge_page and add_hugetlb_page. add_hugetlb_page is called on error paths when a huge page can not be freed due to the inability to allocate vmemmap pages. In this case, the temporairly inflated ref count is not an issue. When the ref is dropped the appropriate action will be taken. Instead of VM_BUG_ON if the ref count does not drop to zero, simply return. In gather_surplus_pages and alloc_and_dissolve_huge_page the caller expects a page (or pages) to be put on the free lists. In this case we must ensure there are no temporary ref counts. We do this by calling put_page_testzero() earlier and not using pages without a zero ref count. The temporary page flag (HPageTemporary) is used in such cases so that as soon as the inflated ref count is dropped the page will be freed. [1] https://lore.kernel.org/linux-mm/CAMZfGtVMn3daKrJwZMaVOGOaJU+B4dS--x_oPmGQMD=c=QNGEg@mail.gmail.com/ Link: https://lkml.kernel.org/r/20210809184832.18342-3-mike.kravetz@oracle.comSigned-off-by:
-
Mike Kravetz authored
Code in prep_compound_gigantic_page waits for a rcu grace period if it notices a temporarily inflated ref count on a tail page. This was due to the identified potential race with speculative page cache references which could only last for a rcu grace period. This is overly complicated as this situation is VERY unlikely to ever happen. Instead, just quickly return an error. Also, only print a warning in prep_compound_gigantic_page instead of multiple callers. Link: https://lkml.kernel.org/r/20210809184832.18342-2-mike.kravetz@oracle.comSigned-off-by:
-
Michael Wang authored
In commit 510d25c9 ("mm/hwpoison: disable pcp for page_handle_poison()"), __page_handle_poison() was introduced, and if we mark: RET_A = dissolve_free_huge_page(); RET_B = take_page_off_buddy(); then __page_handle_poison was supposed to return TRUE When RET_A == 0 && RET_B == TRUE But since it failed to take care the case when RET_A is -EBUSY or -ENOMEM, and just return the ret as a bool which actually become TRUE, it break the original logic. The following result is a huge page in freelist but was referenced as poisoned, and lead into the final panic: kernel BUG at mm/internal.h:95! invalid opcode: 0000 [#1] SMP PTI skip... RIP: 0010:set_page_refcounted mm/internal.h:95 [inline] RIP: 0010:remove_hugetlb_page+0x23c/0x240 mm/hugetlb.c:1371 skip... Call Trace: remove_pool_huge_page+0xe4/0x110 mm/hugetlb.c:1892 return_unused_surplus_pages+0x8d/0x150 mm/hugetlb.c:2272 hugetlb_acct_memory.part.91+0x524/0x690 mm/hugetlb.c:4017 This patch replaces 'bool' with 'int' to handle RET_A correctly. Link: https://lkml.kernel.org/r/61782ac6-1e8a-4f6f-35e6-e94fce3b37f5@linux.alibaba.com Fixes: 510d25c9 ("mm/hwpoison: disable pcp for page_handle_poison()") Signed-off-by:
Michael Wang <yun.wang@linux.alibaba.com> Acked-by:
Naoya Horiguchi <naoya.horiguchi@nec.com> Reported-by:
Abaci <abaci@linux.alibaba.com> Cc: <stable@vger.kernel.org> [5.14+] Signed-off-by:
-
Yang Shi authored
Currently just very simple message is shown for unhandlable page, e.g. non-LRU page, like: soft_offline: 0x1469f2: unknown non LRU page type 5ffff0000000000 () It is not very helpful for further debug, calling dump_page() could show more useful information. Calling dump_page() in get_any_page() in order to not duplicate the call in a couple of different places. It may be called with pcp disabled and holding memory hotplug lock, it should be not a big deal since hwpoison handler is not called very often. [shy828301@gmail.com: remove redundant pr_info per Noaya Horiguchi] Link: https://lkml.kernel.org/r/20210824020946.195257-3-shy828301@gmail.com Link: https://lkml.kernel.org/r/20210819054116.266126-3-shy828301@gmail.comSigned-off-by:
Matthew Wilcox <willy@infradead.org> Acked-by:
-
Yang Shi authored
The hwpoison support for huge page, both hugetlb and THP, has been in kernel for a while, the statement in document is obsolete, correct it. Link: https://lkml.kernel.org/r/20210819054116.266126-2-shy828301@gmail.comSigned-off-by:
-
Yang Shi authored
In the current implementation of soft offline, if non-LRU page is met, all the slab caches will be dropped to free the page then offline. But if the page is not slab page all the effort is wasted in vain. Even though it is a slab page, it is not guaranteed the page could be freed at all. However the side effect and cost is quite high. It does not only drop the slab caches, but also may drop a significant amount of page caches which are associated with inode caches. It could make the most workingset gone in order to just offline a page. And the offline is not guaranteed to succeed at all, actually I really doubt the success rate for real life workload. Furthermore the worse consequence is the system may be locked up and unusable since the page cache release may incur huge amount of works queued for memcg release. Actually we ran into such unpleasant case in our production environment. Firstly, the workqueue of memory_failure_work_func is locked up as below: BUG: workqueue lockup - pool cpus=1 node=0 flags=0x0 nice=0 stuck for 53s! Showing busy workqueues and worker pools: workqueue events: flags=0x0 pwq 2: cpus=1 node=0 flags=0x0 nice=0 active=14/256 refcnt=15 in-flight: 409271:memory_failure_work_func pending: kfree_rcu_work, kfree_rcu_monitor, kfree_rcu_work, rht_deferred_worker, rht_deferred_worker, rht_deferred_worker, rht_deferred_worker, kfree_rcu_work, kfree_rcu_work, kfree_rcu_work, kfree_rcu_work, drain_local_stock, kfree_rcu_work workqueue mm_percpu_wq: flags=0x8 pwq 2: cpus=1 node=0 flags=0x0 nice=0 active=1/256 refcnt=2 pending: vmstat_update workqueue cgroup_destroy: flags=0x0 pwq 2: cpus=1 node=0 flags=0x0 nice=0 active=1/1 refcnt=12072 pending: css_release_work_fn There were over 12K css_release_work_fn queued, and this caused a few lockups due to the contention of worker pool lock with IRQ disabled, for example: NMI watchdog: Watchdog detected hard LOCKUP on cpu 1 Modules linked in: amd64_edac_mod edac_mce_amd crct10dif_pclmul crc32_pclmul ghash_clmulni_intel xt_DSCP iptable_mangle kvm_amd bpfilter vfat fat acpi_ipmi i2c_piix4 usb_storage ipmi_si k10temp i2c_core ipmi_devintf ipmi_msghandler acpi_cpufreq sch_fq_codel xfs libcrc32c crc32c_intel mlx5_core mlxfw nvme xhci_pci ptp nvme_core pps_core xhci_hcd CPU: 1 PID: 205500 Comm: kworker/1:0 Tainted: G L 5.10.32-t1.el7.twitter.x86_64 #1 Hardware name: TYAN F5AMT /z /S8026GM2NRE-CGN, BIOS V8.030 03/30/2021 Workqueue: events memory_failure_work_func RIP: 0010:queued_spin_lock_slowpath+0x41/0x1a0 Code: 41 f0 0f ba 2f 08 0f 92 c0 0f b6 c0 c1 e0 08 89 c2 8b 07 30 e4 09 d0 a9 00 01 ff ff 75 1b 85 c0 74 0e 8b 07 84 c0 74 08 f3 90 <8b> 07 84 c0 75 f8 b8 01 00 00 00 66 89 07 c3 f6 c4 01 75 04 c6 47 RSP: 0018:ffff9b2ac278f900 EFLAGS: 00000002 RAX: 0000000000480101 RBX: ffff8ce98ce71800 RCX: 0000000000000084 RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff8ce98ce6a140 RBP: 00000000000284c8 R08: ffffd7248dcb6808 R09: 0000000000000000 R10: 0000000000000003 R11: ffff9b2ac278f9b0 R12: 0000000000000001 R13: ffff8cb44dab9c00 R14: ffffffffbd1ce6a0 R15: ffff8cacaa37f068 FS: 0000000000000000(0000) GS:ffff8ce98ce40000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007fcf6e8cb000 CR3: 0000000a0c60a000 CR4: 0000000000350ee0 Call Trace: __queue_work+0xd6/0x3c0 queue_work_on+0x1c/0x30 uncharge_batch+0x10e/0x110 mem_cgroup_uncharge_list+0x6d/0x80 release_pages+0x37f/0x3f0 __pagevec_release+0x1c/0x50 __invalidate_mapping_pages+0x348/0x380 inode_lru_isolate+0x10a/0x160 __list_lru_walk_one+0x7b/0x170 list_lru_walk_one+0x4a/0x60 prune_icache_sb+0x37/0x50 super_cache_scan+0x123/0x1a0 do_shrink_slab+0x10c/0x2c0 shrink_slab+0x1f1/0x290 drop_slab_node+0x4d/0x70 soft_offline_page+0x1ac/0x5b0 memory_failure_work_func+0x6a/0x90 process_one_work+0x19e/0x340 worker_thread+0x30/0x360 kthread+0x116/0x130 The lockup made the machine is quite unusable. And it also made the most workingset gone, the reclaimabled slab caches were reduced from 12G to 300MB, the page caches were decreased from 17G to 4G. But the most disappointing thing is all the effort doesn't make the page offline, it just returns: soft_offline: 0x1469f2: unknown non LRU page type 5ffff0000000000 () It seems the aggressive behavior for non-LRU page didn't pay back, so it doesn't make too much sense to keep it considering the terrible side effect. Link: https://lkml.kernel.org/r/20210819054116.266126-1-shy828301@gmail.comSigned-off-by:David Mackey <tdmackey@twitter.com> Acked-by:
-
Miaohe Lin authored
Since commit cb731d6c ("vmscan: per memory cgroup slab shrinkers"), shrink_node_slabs is renamed to drop_slab_node. And doit argument is changed to forcekill since commit 6751ed65 ("x86/mce: Fix siginfo_t->si_addr value for non-recoverable memory faults"). Link: https://lkml.kernel.org/r/20210814105131.48814-5-linmiaohe@huawei.comSigned-off-by:
-
Miaohe Lin authored
It's unnecessary to pass in a struct page **hpagep because it's never modified. Changing to use *hpage to simplify the code. Link: https://lkml.kernel.org/r/20210814105131.48814-4-linmiaohe@huawei.comSigned-off-by:
-
