- 09 Nov, 2021 13 commits
-
-
Stephen Brennan authored
Problem Description: When running running ~128 parallel instances of TZ=/etc/localtime ps -fe >/dev/null on a 128CPU machine, the %sys utilization reaches 97%, and perf shows the following code path as being responsible for heavy contention on the d_lockref spinlock: walk_component() lookup_fast() d_revalidate() pid_revalidate() // returns -ECHILD unlazy_child() lockref_get_not_dead(&nd->path.dentry->d_lockref) <-- contention The reason is that pid_revalidate() is triggering a drop from RCU to ref path walk mode. All concurrent path lookups thus try to grab a reference to the dentry for /proc/, before re-executing pid_revalidate() and then stepping into the /proc/$pid directory. Thus there is huge spinlock contention. This patch allows pid_revalidate() to execute in RCU mode, meaning that the path lookup can successfully enter the /proc/$pid directory while still in RCU mode. Later on, the path lookup may still drop into ref mode, but the contention will be much reduced at this point. By applying this patch, %sys utilization falls to around 85% under the same workload, and the number of ps processes executed per unit time increases by 3x-4x. Although this particular workload is a bit contrived, we have seen some large collections of eager monitoring scripts which produced similarly high %sys time due to contention in the /proc directory. As a result this patch, Al noted that several procfs methods which were only called in ref-walk mode could now be called from RCU mode. To ensure that this patch is safe, I audited all the inode get_link and permission() implementations, as well as dentry d_revalidate() implementations, in fs/proc. The purpose here is to ensure that they either are safe to call in RCU (i.e. don't sleep) or correctly bail out of RCU mode if they don't support it. My analysis shows that all at-risk procfs methods are safe to call under RCU, and thus this patch is safe. Procfs RCU-walk Analysis: This analysis is up-to-date with 5.15-rc3. When called under RCU mode, these functions have arguments as follows: * get_link() receives a NULL dentry pointer when called in RCU mode. * permission() receives MAY_NOT_BLOCK in the mode parameter when called from RCU. * d_revalidate() receives LOOKUP_RCU in flags. For the following functions, either they are trivially RCU safe, or they explicitly bail at the beginning of the function when they run: proc_ns_get_link (bails out) proc_get_link (RCU safe) proc_pid_get_link (bails out) map_files_d_revalidate (bails out) map_misc_d_revalidate (bails out) proc_net_d_revalidate (RCU safe) proc_sys_revalidate (bails out, also not under /proc/$pid) tid_fd_revalidate (bails out) proc_sys_permission (not under /proc/$pid) The remainder of the functions require a bit more detail: * proc_fd_permission: RCU safe. All of the body of this function is under rcu_read_lock(), except generic_permission() which declares itself RCU safe in its documentation string. * proc_self_get_link uses GFP_ATOMIC in the RCU case, so it is RCU aware and otherwise looks safe. The same is true of proc_thread_self_get_link. * proc_map_files_get_link: calls ns_capable, which calls capable(), and thus calls into the audit code (see note #1 below). The remainder is just a call to the trivially safe proc_pid_get_link(). * proc_pid_permission: calls ptrace_may_access(), which appears RCU safe, although it does call into the "security_ptrace_access_check()" hook, which looks safe under smack and selinux. Just the audit code is of concern. Also uses get_task_struct() and put_task_struct(), see note #2 below. * proc_tid_comm_permission: Appears safe, though calls put_task_struct (see note #2 below). Note #1: Most of the concern of RCU safety has centered around the audit code. However, since b17ec22f ("selinux: slow_avc_audit has become non-blocking"), it's safe to call this code under RCU. So all of the above are safe by my estimation. Note #2: get_task_struct() and put_task_struct(): The majority of get_task_struct() is under RCU read lock, and in any case it is a simple increment. But put_task_struct() is complex, given that it could at some point free the task struct, and this process has many steps which I couldn't manually verify. However, several other places call put_task_struct() under RCU, so it appears safe to use here too (see kernel/hung_task.c:165 or rcu/tree-stall.h:296) Patch description: pid_revalidate() drops from RCU into REF lookup mode. When many threads are resolving paths within /proc in parallel, this can result in heavy spinlock contention on d_lockref as each thread tries to grab a reference to the /proc dentry (and drop it shortly thereafter). Investigation indicates that it is not necessary to drop RCU in pid_revalidate(), as no RCU data is modified and the function never sleeps. So, remove the LOOKUP_RCU check. Link: https://lkml.kernel.org/r/20211004175629.292270-2-stephen.s.brennan@oracle.comSigned-off-by:Stephen Brennan <stephen.s.brennan@oracle.com> Cc: Konrad Wilk <konrad.wilk@oracle.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Matthew Wilcox <willy@infradead.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
David Hildenbrand authored
Although virtio-mem currently supports reading unplugged memory in the hypervisor, this will change in the future, indicated to the device via a new feature flag. We similarly sanitized /proc/kcore access recently. [1] Let's register a vmcore callback, to allow vmcore code to check if a PFN belonging to a virtio-mem device is either currently plugged and should be dumped or is currently unplugged and should not be accessed, instead mapping the shared zeropage or returning zeroes when reading. This is important when not capturing /proc/vmcore via tools like "makedumpfile" that can identify logically unplugged virtio-mem memory via PG_offline in the memmap, but simply by e.g., copying the file. Distributions that support virtio-mem+kdump have to make sure that the virtio_mem module will be part of the kdump kernel or the kdump initrd; dracut was recently [2] extended to include virtio-mem in the generated initrd. As long as no special kdump kernels are used, this will automatically make sure that virtio-mem will be around in the kdump initrd and sanitize /proc/vmcore access -- with dracut. With this series, we'll send one virtio-mem state request for every ~2 MiB chunk of virtio-mem memory indicated in the vmcore that we intend to read/map. In the future, we might want to allow building virtio-mem for kdump mode only, even without CONFIG_MEMORY_HOTPLUG and friends: this way, we could support special stripped-down kdump kernels that have many other config options disabled; we'll tackle that once required. Further, we might want to try sensing bigger blocks (e.g., memory sections) first before falling back to device blocks on demand. Tested with Fedora rawhide, which contains a recent kexec-tools version (considering "System RAM (virtio_mem)" when creating the vmcore header) and a recent dracut version (including the virtio_mem module in the kdump initrd). Link: https://lkml.kernel.org/r/20210526093041.8800-1-david@redhat.com [1] Link: https://github.com/dracutdevs/dracut/pull/1157 [2] Link: https://lkml.kernel.org/r/20211005121430.30136-10-david@redhat.comSigned-off-by:
David Hildenbrand <david@redhat.com> Cc: Baoquan He <bhe@redhat.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com> Cc: Dave Young <dyoung@redhat.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Juergen Gross <jgross@suse.com> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mike Rapoport <rppt@kernel.org> Cc: Oscar Salvador <osalvador@suse.de> Cc: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com> Cc: Stefano Stabellini <sstabellini@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by:
-
David Hildenbrand authored
Let's prepare for a new virtio-mem kdump mode in which we don't actually hot(un)plug any memory but only observe the state of device blocks. Link: https://lkml.kernel.org/r/20211005121430.30136-9-david@redhat.comSigned-off-by:
-
David Hildenbrand authored
Let's prepare for a new virtio-mem kdump mode in which we don't actually hot(un)plug any memory but only observe the state of device blocks. Link: https://lkml.kernel.org/r/20211005121430.30136-8-david@redhat.comSigned-off-by:
-
David Hildenbrand authored
Let's prepare for a new virtio-mem kdump mode in which we don't actually hot(un)plug any memory but only observe the state of device blocks. Link: https://lkml.kernel.org/r/20211005121430.30136-7-david@redhat.comSigned-off-by:
-
David Hildenbrand authored
Let's support multiple registered callbacks, making sure that registering vmcore callbacks cannot fail. Make the callback return a bool instead of an int, handling how to deal with errors internally. Drop unused HAVE_OLDMEM_PFN_IS_RAM. We soon want to make use of this infrastructure from other drivers: virtio-mem, registering one callback for each virtio-mem device, to prevent reading unplugged virtio-mem memory. Handle it via a generic vmcore_cb structure, prepared for future extensions: for example, once we support virtio-mem on s390x where the vmcore is completely constructed in the second kernel, we want to detect and add plugged virtio-mem memory ranges to the vmcore in order for them to get dumped properly. Handle corner cases that are unexpected and shouldn't happen in sane setups: registering a callback after the vmcore has already been opened (warn only) and unregistering a callback after the vmcore has already been opened (warn and essentially read only zeroes from that point on). Link: https://lkml.kernel.org/r/20211005121430.30136-6-david@redhat.comSigned-off-by:
-
David Hildenbrand authored
The callback should deal with errors internally, it doesn't make sense to expose these via pfn_is_ram(). We'll rework the callbacks next. Right now we consider errors as if "it's RAM"; no functional change. Link: https://lkml.kernel.org/r/20211005121430.30136-5-david@redhat.comSigned-off-by:
-
David Hildenbrand authored
HVMOP_get_mem_type is not expected to fail, "This call failing is indication of something going quite wrong and it would be good to know about this." [1] Let's add a pr_warn_once(). Link: https://lkml.kernel.org/r/3b935aa0-6d85-0bcd-100e-15098add3c4c@oracle.com [1] Link: https://lkml.kernel.org/r/20211005121430.30136-4-david@redhat.comSigned-off-by:
Boris Ostrovsky <boris.ostrovsky@oracle.com> Cc: Baoquan He <bhe@redhat.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Dave Young <dyoung@redhat.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Juergen Gross <jgross@suse.com> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Mike Rapoport <rppt@kernel.org> Cc: Oscar Salvador <osalvador@suse.de> Cc: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com> Cc: Stefano Stabellini <sstabellini@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vivek Goyal <vgoyal@redhat.com> Signed-off-by:
-
David Hildenbrand authored
Let's simplify return handling. Link: https://lkml.kernel.org/r/20211005121430.30136-3-david@redhat.comSigned-off-by:
-
David Hildenbrand authored
After removing /dev/kmem, sanitizing /proc/kcore and handling /dev/mem, this series tackles the last sane way how a VM could accidentially access logically unplugged memory managed by a virtio-mem device: /proc/vmcore When dumping memory via "makedumpfile", PG_offline pages, used by virtio-mem to flag logically unplugged memory, are already properly excluded; however, especially when accessing/copying /proc/vmcore "the usual way", we can still end up reading logically unplugged memory part of a virtio-mem device. Patch #1-#3 are cleanups. Patch #4 extends the existing oldmem_pfn_is_ram mechanism. Patch #5-#7 are virtio-mem refactorings for patch #8, which implements the virtio-mem logic to query the state of device blocks. Patch #8: "Although virtio-mem currently supports reading unplugged memory in the hypervisor, this will change in the future, indicated to the device via a new feature flag. We similarly sanitized /proc/kcore access recently. [...] Distributions that support virtio-mem+kdump have to make sure that the virtio_mem module will be part of the kdump kernel or the kdump initrd; dracut was recently [2] extended to include virtio-mem in the generated initrd. As long as no special kdump kernels are used, this will automatically make sure that virtio-mem will be around in the kdump initrd and sanitize /proc/vmcore access -- with dracut" This is the last remaining bit to support VIRTIO_MEM_F_UNPLUGGED_INACCESSIBLE [3] in the Linux implementation of virtio-mem. Note: this is best-effort. We'll never be able to control what runs inside the second kernel, really, but we also don't have to care: we only care about sane setups where we don't want our VM getting zapped once we touch the wrong memory location while dumping. While we usually expect sane setups to use "makedumfile", nothing really speaks against just copying /proc/vmcore, especially in environments where HWpoisioning isn't typically expected. Also, we really don't want to put all our trust completely on the memmap, so sanitizing also makes sense when just using "makedumpfile". [1] https://lkml.kernel.org/r/20210526093041.8800-1-david@redhat.com [2] https://github.com/dracutdevs/dracut/pull/1157 [3] https://lists.oasis-open.org/archives/virtio-comment/202109/msg00021.html This patch (of 9): The callback is only used for the vmcore nowadays. Link: https://lkml.kernel.org/r/20211005121430.30136-1-david@redhat.com Link: https://lkml.kernel.org/r/20211005121430.30136-2-david@redhat.comSigned-off-by:
Boris Ostrovsky <boris.ostrvsky@oracle.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ingo Molnar <mingo@redhat.com> Cc: Borislav Petkov <bp@alien8.de> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Juergen Gross <jgross@suse.com> Cc: Stefano Stabellini <sstabellini@kernel.org> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Dave Young <dyoung@redhat.com> Cc: Baoquan He <bhe@redhat.com> Cc: Vivek Goyal <vgoyal@redhat.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Mike Rapoport <rppt@kernel.org> Cc: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com> Signed-off-by:
-
Florian Weimer authored
If a task exits concurrently, task_pid_nr_ns may return 0. [akpm@linux-foundation.org: coding style tweaks] [adobriyan@gmail.com: test that /proc/*/task doesn't contain "0"] Link: https://lkml.kernel.org/r/YV88AnVzHxPafQ9o@localhost.localdomain Link: https://lkml.kernel.org/r/8735pn5dx7.fsf@oldenburg.str.redhat.comSigned-off-by:
Florian Weimer <fweimer@redhat.com> Signed-off-by:
Alexey Dobriyan <adobriyan@gmail.com> Acked-by:
Christian Brauner <christian.brauner@ubuntu.com> Reviewed-by:
-
zhangyiru authored
Commit 21a3c273 ("mm, hugetlb: add thread name and pid to SHM_HUGETLB mlock rlimit warning") marked this as deprecated in 2012, but it is not deleted yet. Mike says he still sees that message in log files on occasion, so maybe we should preserve this warning. Also remove hugetlbfs related user_shm_unlock in ipc/shm.c and remove the user_shm_unlock after out. Link: https://lkml.kernel.org/r/20211103105857.25041-1-zhangyiru3@huawei.comSigned-off-by:
zhangyiru <zhangyiru3@huawei.com> Reviewed-by:
Mike Kravetz <mike.kravetz@oracle.com> Cc: Hugh Dickins <hughd@google.com> Cc: Liu Zixian <liuzixian4@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: wuxu.wu <wuxu.wu@huawei.com> Signed-off-by:
-
Johannes Weiner authored
Historically (pre-2.5), the inode shrinker used to reclaim only empty inodes and skip over those that still contained page cache. This caused problems on highmem hosts: struct inode could put fill lowmem zones before the cache was getting reclaimed in the highmem zones. To address this, the inode shrinker started to strip page cache to facilitate reclaiming lowmem. However, this comes with its own set of problems: the shrinkers may drop actively used page cache just because the inodes are not currently open or dirty - think working with a large git tree. It further doesn't respect cgroup memory protection settings and can cause priority inversions between containers. Nowadays, the page cache also holds non-resident info for evicted cache pages in order to detect refaults. We've come to rely heavily on this data inside reclaim for protecting the cache workingset and driving swap behavior. We also use it to quantify and report workload health through psi. The latter in turn is used for fleet health monitoring, as well as driving automated memory sizing of workloads and containers, proactive reclaim and memory offloading schemes. The consequences of dropping page cache prematurely is that we're seeing subtle and not-so-subtle failures in all of the above-mentioned scenarios, with the workload generally entering unexpected thrashing states while losing the ability to reliably detect it. To fix this on non-highmem systems at least, going back to rotating inodes on the LRU isn't feasible. We've tried (commit a76cf1a4 ("mm: don't reclaim inodes with many attached pages")) and failed (commit 69056ee6 ("Revert "mm: don't reclaim inodes with many attached pages"")). The issue is mostly that shrinker pools attract pressure based on their size, and when objects get skipped the shrinkers remember this as deferred reclaim work. This accumulates excessive pressure on the remaining inodes, and we can quickly eat into heavily used ones, or dirty ones that require IO to reclaim, when there potentially is plenty of cold, clean cache around still. Instead, this patch keeps populated inodes off the inode LRU in the first place - just like an open file or dirty state would. An otherwise clean and unused inode then gets queued when the last cache entry disappears. This solves the problem without reintroducing the reclaim issues, and generally is a bit more scalable than having to wade through potentially hundreds of thousands of busy inodes. Locking is a bit tricky because the locks protecting the inode state (i_lock) and the inode LRU (lru_list.lock) don't nest inside the irq-safe page cache lock (i_pages.xa_lock). Page cache deletions are serialized through i_lock, taken before the i_pages lock, to make sure depopulated inodes are queued reliably. Additions may race with deletions, but we'll check again in the shrinker. If additions race with the shrinker itself, we're protected by the i_lock: if find_inode() or iput() win, the shrinker will bail on the elevated i_count or I_REFERENCED; if the shrinker wins and goes ahead with the inode, it will set I_FREEING and inhibit further igets(), which will cause the other side to create a new instance of the inode instead. Link: https://lkml.kernel.org/r/20210614211904.14420-4-hannes@cmpxchg.orgSigned-off-by:
Johannes Weiner <hannes@cmpxchg.org> Cc: Roman Gushchin <guro@fb.com> Cc: Tejun Heo <tj@kernel.org> Cc: Dave Chinner <david@fromorbit.com> Signed-off-by:
-
- 06 Nov, 2021 27 commits
-
-
Changbin Du authored
Since the return value of 'before_terminate' callback is never used, we make it have no return value. Link: https://lkml.kernel.org/r/20211029005023.8895-1-changbin.du@gmail.comSigned-off-by:
Changbin Du <changbin.du@gmail.com> Reviewed-by:
SeongJae Park <sj@kernel.org> Signed-off-by:
-
Colin Ian King authored
There are a few spelling mistakes in the code. Fix these. Link: https://lkml.kernel.org/r/20211028184157.614544-1-colin.i.king@gmail.comSigned-off-by:
Colin Ian King <colin.i.king@gmail.com> Reviewed-by:
-
Changbin Du authored
A kernel thread can exit gracefully with kthread_stop(). So we don't need a new flag 'kdamond_stop'. And to make sure the task struct is not freed when accessing it, get reference to it before termination. Link: https://lkml.kernel.org/r/20211027130517.4404-1-changbin.du@gmail.comSigned-off-by:
-
SeongJae Park authored
Some descriptions of page flags in 'pagemap.rst' are written in assumption of none-rst, which respects every new line, as below: 7 - SLAB page is managed by the SLAB/SLOB/SLUB/SLQB kernel memory allocator When compound page is used, SLUB/SLQB will only set this flag on the head Because rst ignores the new line between the first sentence and second sentence, resulting html looks a little bit weird, as below. 7 - SLAB page is managed by the SLAB/SLOB/SLUB/SLQB kernel memory allocator When ^ compound page is used, SLUB/SLQB will only set this flag on the head page; SLOB will not flag it at all. This change makes it more natural and consistent with other parts in the rendered version. Link: https://lkml.kernel.org/r/20211022090311.3856-5-sj@kernel.orgSigned-off-by: -
SeongJae Park authored
Information in 'TL; DR' section of 'Getting Started' is duplicated in other parts of the doc. It is also asking readers to visit the access pattern visualizations gallery web site to show the results of example visualization commands, while the users of the commands can use terminal output. To make the doc simple, this removes the duplicated 'TL; DR' section and replaces the visualization example commands with versions using terminal outputs. Link: https://lkml.kernel.org/r/20211022090311.3856-4-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
The 'Getting Started' of DAMON is providing a link to DAMON's user interface document while saying about its user space tool's detailed usages. This fixes the link. Link: https://lkml.kernel.org/r/20211022090311.3856-3-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
Patch series "Fix trivial nits in Documentation/admin-guide/mm". This patchset fixes trivial nits in admin guide documents for DAMON and pagemap. This patch (of 4): Some of the example commands in DAMON getting started guide are outdated, missing sudo, or just wrong. This fixes those. Link: https://lkml.kernel.org/r/20211022090311.3856-2-sj@kernel.orgSigned-off-by:
-
Xin Hao authored
When the ctx->adaptive_targets list is empty, I did some test on monitor_on interface like this. # cat /sys/kernel/debug/damon/target_ids # # echo on > /sys/kernel/debug/damon/monitor_on # damon: kdamond (5390) starts Though the ctx->adaptive_targets list is empty, but the kthread_run still be called, and the kdamond.x thread still be created, this is meaningless. So there adds a judgment in 'dbgfs_monitor_on_write', if the ctx->adaptive_targets list is empty, return -EINVAL. Link: https://lkml.kernel.org/r/0a60a6e8ec9d71989e0848a4dc3311996ca3b5d4.1634720326.git.xhao@linux.alibaba.comSigned-off-by:Xin Hao <xhao@linux.alibaba.com> Reviewed-by:
-
Xin Hao authored
Patch series "mm/damon: Fix some small bugs", v4. This patch (of 2): In 'damon_va_apply_three_regions' there is no need to set variable 'i' to zero. Link: https://lkml.kernel.org/r/b7df8d3dad0943a37e01f60c441b1968b2b20354.1634720326.git.xhao@linux.alibaba.com Link: https://lkml.kernel.org/r/cover.1634720326.git.xhao@linux.alibaba.comSigned-off-by:
-
SeongJae Park authored
This adds an admin-guide document for DAMON-based Reclamation. Link: https://lkml.kernel.org/r/20211019150731.16699-16-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
This implements a new kernel subsystem that finds cold memory regions using DAMON and reclaims those immediately. It is intended to be used as proactive lightweigh reclamation logic for light memory pressure. For heavy memory pressure, it could be inactivated and fall back to the traditional page-scanning based reclamation. It's implemented on top of DAMON framework to use the DAMON-based Operation Schemes (DAMOS) feature. It utilizes all the DAMOS features including speed limit, prioritization, and watermarks. It could be enabled and tuned in boot time via the kernel boot parameter, and in run time via its module parameters ('/sys/module/damon_reclaim/parameters/') interface. [yangyingliang@huawei.com: fix error return code in damon_reclaim_turn()] Link: https://lkml.kernel.org/r/20211025124500.2758060-1-yangyingliang@huawei.com Link: https://lkml.kernel.org/r/20211019150731.16699-15-sj@kernel.orgSigned-off-by:Yang Yingliang <yangyingliang@huawei.com> Cc: Amit Shah <amit@kernel.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: David Hildenbrand <david@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: David Woodhouse <dwmw@amazon.com> Cc: Greg Thelen <gthelen@google.com> Cc: Jonathan Cameron <Jonathan.Cameron@huawei.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Leonard Foerster <foersleo@amazon.de> Cc: Marco Elver <elver@google.com> Cc: Markus Boehme <markubo@amazon.de> Cc: Shakeel Butt <shakeelb@google.com> Cc: Shuah Khan <shuah@kernel.org> Signed-off-by:
-
SeongJae Park authored
This updates DAMON selftests for 'schemes' debugfs file to reflect the changes in the format. Link: https://lkml.kernel.org/r/20211019150731.16699-14-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
This updates DAMON debugfs interface to support the watermarks based schemes activation. For this, now 'schemes' file receives five more values. Link: https://lkml.kernel.org/r/20211019150731.16699-13-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
DAMON-based operation schemes need to be manually turned on and off. In some use cases, however, the condition for turning a scheme on and off would depend on the system's situation. For example, schemes for proactive pages reclamation would need to be turned on when some memory pressure is detected, and turned off when the system has enough free memory. For easier control of schemes activation based on the system situation, this introduces a watermarks-based mechanism. The client can describe the watermark metric (e.g., amount of free memory in the system), watermark check interval, and three watermarks, namely high, mid, and low. If the scheme is deactivated, it only gets the metric and compare that to the three watermarks for every check interval. If the metric is higher than the high watermark, the scheme is deactivated. If the metric is between the mid watermark and the low watermark, the scheme is activated. If the metric is lower than the low watermark, the scheme is deactivated again. This is to allow users fall back to traditional page-granularity mechanisms. Link: https://lkml.kernel.org/r/20211019150731.16699-12-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
This updates the DAMON selftests for 'schemes' debugfs file, as the file format is updated. Link: https://lkml.kernel.org/r/20211019150731.16699-11-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
This allows DAMON debugfs interface users set the prioritization weights by putting three more numbers to the 'schemes' file. Link: https://lkml.kernel.org/r/20211019150731.16699-10-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
This makes the default monitoring primitives for virtual address spaces and the physical address sapce to support memory regions prioritization for 'PAGEOUT' DAMOS action. It calculates hotness of each region as weighted sum of 'nr_accesses' and 'age' of the region and get the priority score as reverse of the hotness, so that cold regions can be paged out first. Link: https://lkml.kernel.org/r/20211019150731.16699-9-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
This makes DAMON apply schemes to regions having higher priority first, if it cannot apply schemes to all regions due to the quotas. The prioritization function should be implemented in the monitoring primitives. Those would commonly calculate the priority of the region using attributes of regions, namely 'size', 'nr_accesses', and 'age'. For example, some primitive would calculate the priority of each region using a weighted sum of 'nr_accesses' and 'age' of the region. The optimal weights would depend on give environments, so this makes those customizable. Nevertheless, the score calculation functions are only encouraged to respect the weights, not mandated. Link: https://lkml.kernel.org/r/20211019150731.16699-8-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
This updates DAMON selftests to support updated schemes debugfs file format for the quotas. Link: https://lkml.kernel.org/r/20211019150731.16699-7-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
This makes the debugfs interface of DAMON support the scheme quotas by chaning the format of the input for the schemes file. Link: https://lkml.kernel.org/r/20211019150731.16699-6-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
The size quota feature of DAMOS is useful for IO resource-critical systems, but not so intuitive for CPU time-critical systems. Systems using zram or zswap-like swap device would be examples. To provide another intuitive ways for such systems, this implements time-based quota for DAMON-based Operation Schemes. If the quota is set, DAMOS tries to use only up to the user-defined quota of CPU time within a given time window. Link: https://lkml.kernel.org/r/20211019150731.16699-5-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
If DAMOS has stopped applying action in the middle of a group of memory regions due to its size quota, it starts the work again from the beginning of the address space in the next charge window. If there is a huge memory region at the beginning of the address space and it fulfills the scheme's target data access pattern always, the action will applied to only the region. This mitigates the case by skipping memory regions that charged in current charge window at the beginning of next charge window. Link: https://lkml.kernel.org/r/20211019150731.16699-4-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
There could be arbitrarily large memory regions fulfilling the target data access pattern of a DAMON-based operation scheme. In the case, applying the action of the scheme could incur too high overhead. To provide an intuitive way for avoiding it, this implements a feature called size quota. If the quota is set, DAMON tries to apply the action only up to the given amount of memory regions within a given time window. Link: https://lkml.kernel.org/r/20211019150731.16699-3-sj@kernel.orgSigned-off-by:
-
SeongJae Park authored
Introduction ============ This patchset 1) makes the engine for general data access pattern-oriented memory management (DAMOS) be more useful for production environments, and 2) implements a static kernel module for lightweight proactive reclamation using the engine. Proactive Reclamation --------------------- On general memory over-committed systems, proactively reclaiming cold pages helps saving memory and reducing latency spikes that incurred by the direct reclaim or the CPU consumption of kswapd, while incurring only minimal performance degradation[2]. A Free Pages Reporting[8] based memory over-commit virtualization system would be one more specific use case. In the system, the guest VMs reports their free memory to host, and the host reallocates the reported memory to other guests. As a result, the system's memory utilization can be maximized. However, the guests could be not so memory-frugal, because some kernel subsystems and user-space applications are designed to use as much memory as available. Then, guests would report only small amount of free memory to host, results in poor memory utilization. Running the proactive reclamation in such guests could help mitigating this problem. Google has also implemented this idea and using it in their data center. They further proposed upstreaming it in LSFMM'19, and "the general consensus was that, while this sort of proactive reclaim would be useful for a number of users, the cost of this particular solution was too high to consider merging it upstream"[3]. The cost mainly comes from the coldness tracking. Roughly speaking, the implementation periodically scans the 'Accessed' bit of each page. For the reason, the overhead linearly increases as the size of the memory and the scanning frequency grows. As a result, Google is known to dedicating one CPU for the work. That's a reasonable option to someone like Google, but it wouldn't be so to some others. DAMON and DAMOS: An engine for data access pattern-oriented memory management ----------------------------------------------------------------------------- DAMON[4] is a framework for general data access monitoring. Its adaptive monitoring overhead control feature minimizes its monitoring overhead. It also let the upper-bound of the overhead be configurable by clients, regardless of the size of the monitoring target memory. While monitoring 70 GiB memory of a production system every 5 milliseconds, it consumes less than 1% single CPU time. For this, it could sacrify some of the quality of the monitoring results. Nevertheless, the lower-bound of the quality is configurable, and it uses a best-effort algorithm for better quality. Our test results[5] show the quality is practical enough. From the production system monitoring, we were able to find a 4 KiB region in the 70 GiB memory that shows highest access frequency. We normally don't monitor the data access pattern just for fun but to improve something like memory management. Proactive reclamation is one such usage. For such general cases, DAMON provides a feature called DAMon-based Operation Schemes (DAMOS)[6]. It makes DAMON an engine for general data access pattern oriented memory management. Using this, clients can ask DAMON to find memory regions of specific data access pattern and apply some memory management action (e.g., page out, move to head of the LRU list, use huge page, ...). We call the request 'scheme'. Proactive Reclamation on top of DAMON/DAMOS ------------------------------------------- Therefore, by using DAMON for the cold pages detection, the proactive reclamation's monitoring overhead issue can be solved. Actually, we previously implemented a version of proactive reclamation using DAMOS and achieved noticeable improvements with our evaluation setup[5]. Nevertheless, it more for a proof-of-concept, rather than production uses. It supports only virtual address spaces of processes, and require additional tuning efforts for given workloads and the hardware. For the tuning, we introduced a simple auto-tuning user space tool[8]. Google is also known to using a ML-based similar approach for their fleets[2]. But, making it just works with intuitive knobs in the kernel would be helpful for general users. To this end, this patchset improves DAMOS to be ready for such production usages, and implements another version of the proactive reclamation, namely DAMON_RECLAIM, on top of it. DAMOS Improvements: Aggressiveness Control, Prioritization, and Watermarks -------------------------------------------------------------------------- First of all, the current version of DAMOS supports only virtual address spaces. This patchset makes it supports the physical address space for the page out action. Next major problem of the current version of DAMOS is the lack of the aggressiveness control, which can results in arbitrary overhead. For example, if huge memory regions having the data access pattern of interest are found, applying the requested action to all of the regions could incur significant overhead. It can be controlled by tuning the target data access pattern with manual or automated approaches[2,7]. But, some people would prefer the kernel to just work with only intuitive tuning or default values. For such cases, this patchset implements a safeguard, namely time/size quota. Using this, the clients can specify up to how much time can be used for applying the action, and/or up to how much memory regions the action can be applied within a user-specified time duration. A followup question is, to which memory regions should the action applied within the limits? We implement a simple regions prioritization mechanism for each action and make DAMOS to apply the action to high priority regions first. It also allows clients tune the prioritization mechanism to use different weights for size, access frequency, and age of memory regions. This means we could use not only LRU but also LFU or some fancy algorithms like CAR[9] with lightweight overhead. Though DAMON is lightweight, someone would want to remove even the cold pages monitoring overhead when it is unnecessary. Currently, it should manually turned on and off by clients, but some clients would simply want to turn it on and off based on some metrics like free memory ratio or memory fragmentation. For such cases, this patchset implements a watermarks-based automatic activation feature. It allows the clients configure the metric of their interest, and three watermarks of the metric. If the metric is higher than the high watermark or lower than the low watermark, the scheme is deactivated. If the metric is lower than the mid watermark but higher than the low watermark, the scheme is activated. DAMON-based Reclaim ------------------- Using the improved version of DAMOS, this patchset implements a static kernel module called 'damon_reclaim'. It finds memory regions that didn't accessed for specific time duration and page out. Consuming too much CPU for the paging out operations, or doing pageout too frequently can be critical for systems configuring their swap devices with software-defined in-memory block devices like zram/zswap or total number of writes limited devices like SSDs, respectively. To avoid the problems, the time/size quotas can be configured. Under the quotas, it pages out memory regions that didn't accessed longer first. Also, to remove the monitoring overhead under peaceful situation, and to fall back to the LRU-list based page granularity reclamation when it doesn't make progress, the three watermarks based activation mechanism is used, with the free memory ratio as the watermark metric. For convenient configurations, it provides several module parameters. Using these, sysadmins can enable/disable it, and tune its parameters including the coldness identification time threshold, the time/size quotas and the three watermarks. Evaluation ========== In short, DAMON_RECLAIM with 50ms/s time quota and regions prioritization on v5.15-rc5 Linux kernel with ZRAM swap device achieves 38.58% memory saving with only 1.94% runtime overhead. For this, DAMON_RECLAIM consumes only 4.97% of single CPU time. Setup ----- We evaluate DAMON_RECLAIM to show how each of the DAMOS improvements make effect. For this, we measure DAMON_RECLAIM's CPU consumption, entire system memory footprint, total number of major page faults, and runtime of 24 realistic workloads in PARSEC3 and SPLASH-2X benchmark suites on my QEMU/KVM based virtual machine. The virtual machine runs on an i3.metal AWS instance, has 130GiB memory, and runs a linux kernel built on latest -mm tree[1] plus this patchset. It also utilizes a 4 GiB ZRAM swap device. We repeats the measurement 5 times and use averages. [1] https://github.com/hnaz/linux-mm/tree/v5.15-rc5-mmots-2021-10-13-19-55 Detailed Results ---------------- The results are summarized in the below table. With coldness identification threshold of 5 seconds, DAMON_RECLAIM without the time quota-based speed limit achieves 47.21% memory saving, but incur 4.59% runtime slowdown to the workloads on average. For this, DAMON_RECLAIM consumes about 11.28% single CPU time. Applying time quotas of 200ms/s, 50ms/s, and 10ms/s without the regions prioritization reduces the slowdown to 4.89%, 2.65%, and 1.5%, respectively. Time quota of 200ms/s (20%) makes no real change compared to the quota unapplied version, because the quota unapplied version consumes only 11.28% CPU time. DAMON_RECLAIM's CPU utilization also similarly reduced: 11.24%, 5.51%, and 2.01% of single CPU time. That is, the overhead is proportional to the speed limit. Nevertheless, it also reduces the memory saving because it becomes less aggressive. In detail, the three variants show 48.76%, 37.83%, and 7.85% memory saving, respectively. Applying the regions prioritization (page out regions that not accessed longer first within the time quota) further reduces the performance degradation. Runtime slowdowns and total number of major page faults increase has been 4.89%/218,690% -> 4.39%/166,136% (200ms/s), 2.65%/111,886% -> 1.94%/59,053% (50ms/s), and 1.5%/34,973.40% -> 2.08%/8,781.75% (10ms/s). The runtime under 10ms/s time quota has increased with prioritization, but apparently that's under the margin of error. time quota prioritization memory_saving cpu_util slowdown pgmajfaults overhead N N 47.21% 11.28% 4.59% 194,802% 200ms/s N 48.76% 11.24% 4.89% 218,690% 50ms/s N 37.83% 5.51% 2.65% 111,886% 10ms/s N 7.85% 2.01% 1.5% 34,793.40% 200ms/s Y 50.08% 10.38% 4.39% 166,136% 50ms/s Y 38.58% 4.97% 1.94% 59,053% 10ms/s Y 3.63% 1.73% 2.08% 8,781.75% Baseline and Complete Git Trees =============================== The patches are based on the latest -mm tree (v5.15-rc5-mmots-2021-10-13-19-55). You can also clone the complete git tree from: $ git clone git://github.com/sjp38/linux -b damon_reclaim/patches/v1 The web is also available: https://git.kernel.org/pub/scm/linux/kernel/git/sj/linux.git/tag/?h=damon_reclaim/patches/v1 Sequence Of Patches =================== The first patch makes DAMOS support the physical address space for the page out action. Following five patches (patches 2-6) implement the time/size quotas. Next four patches (patches 7-10) implement the memory regions prioritization within the limit. Then, three following patches (patches 11-13) implement the watermarks-based schemes activation. Finally, the last two patches (patches 14-15) implement and document the DAMON-based reclamation using the advanced DAMOS. [1] https://www.kernel.org/doc/html/v5.15-rc1/vm/damon/index.html [2] https://research.google/pubs/pub48551/ [3] https://lwn.net/Articles/787611/ [4] https://damonitor.github.io [5] https://damonitor.github.io/doc/html/latest/vm/damon/eval.html [6] https://lore.kernel.org/linux-mm/20211001125604.29660-1-sj@kernel.org/ [7] https://github.com/awslabs/damoos [8] https://www.kernel.org/doc/html/latest/vm/free_page_reporting.html [9] https://www.usenix.org/conference/fast-04/car-clock-adaptive-replacement This patch (of 15): This makes the DAMON primitives for physical address space support the pageout action for DAMON-based Operation Schemes. With this commit, hence, users can easily implement system-level data access-aware reclamations using DAMOS. [sj@kernel.org: fix missing-prototype build warning] Link: https://lkml.kernel.org/r/20211025064220.13904-1-sj@kernel.org Link: https://lkml.kernel.org/r/20211019150731.16699-1-sj@kernel.org Link: https://lkml.kernel.org/r/20211019150731.16699-2-sj@kernel.orgSigned-off-by:
-
Rongwei Wang authored
In some functions, it's unnecessary to declare 'err' and 'ret' variables at the same time. This patch mainly to simplify the issue of such declarations by reusing one variable. Link: https://lkml.kernel.org/r/20211014073014.35754-1-sj@kernel.orgSigned-off-by:
Rongwei Wang <rongwei.wang@linux.alibaba.com> Signed-off-by:
-
Rikard Falkeborn authored
The only usage of these structs is to pass their addresses to walk_page_range(), which takes a pointer to const mm_walk_ops as argument. Make them const to allow the compiler to put them in read-only memory. Link: https://lkml.kernel.org/r/20211014075042.17174-2-rikard.falkeborn@gmail.comSigned-off-by:
Rikard Falkeborn <rikard.falkeborn@gmail.com> Reviewed-by:
Anshuman Khandual <anshuman.khandual@arm.com> Signed-off-by:
-
SeongJae Park authored
This updates the DAMON documents for the physical memory address space monitoring support. Link: https://lkml.kernel.org/r/20211012205711.29216-8-sj@kernel.orgSigned-off-by:
-
