- 07 Apr, 2020 40 commits
-
-
Baoquan He authored
And tell check_pfn_span() gating the porper alignment and size of hot added memory region. And also move the code comments from inside section_deactivate() to being above it. The code comments are reasonable for the whole function, and the moving makes code cleaner. Signed-off-by:
Baoquan He <bhe@redhat.com> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Reviewed-by:
David Hildenbrand <david@redhat.com> Acked-by:
Michal Hocko <mhocko@suse.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com> Cc: Wei Yang <richard.weiyang@gmail.com> Link: http://lkml.kernel.org/r/20200312124414.439-5-bhe@redhat.comSigned-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Baoquan He authored
Currently, to support subsection aligned memory region adding for pmem, subsection map is added to track which subsection is present. However, config ZONE_DEVICE depends on SPARSEMEM_VMEMMAP. It means subsection map only makes sense when SPARSEMEM_VMEMMAP enabled. For the classic sparse, it's meaningless. Even worse, it may confuse people when checking code related to the classic sparse. About the classic sparse which doesn't support subsection hotplug, Dan said it's more because the effort and maintenance burden outweighs the benefit. Besides, the current 64 bit ARCHes all enable SPARSEMEM_VMEMMAP_ENABLE by default. Combining the above reasons, no need to provide subsection map and the relevant handling for the classic sparse. Let's remove them. Signed-off-by:
-
Baoquan He authored
Factor out the code which clear subsection map of one memory region from section_deactivate() into clear_subsection_map(). And also add helper function is_subsection_map_empty() to check if the current subsection map is empty or not. Signed-off-by:
Pankaj Gupta <pankaj.gupta.linux@gmail.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Michal Hocko <mhocko@suse.com> Cc: Wei Yang <richard.weiyang@gmail.com> Link: http://lkml.kernel.org/r/20200312124414.439-3-bhe@redhat.comSigned-off-by:
-
Baoquan He authored
Patch series "mm/hotplug: Only use subsection map for VMEMMAP", v4. Memory sub-section hotplug was added to fix the issue that nvdimm could be mapped at non-section aligned starting address. A subsection map is added into struct mem_section_usage to implement it. However, config ZONE_DEVICE depends on SPARSEMEM_VMEMMAP. It means subsection map only makes sense when SPARSEMEM_VMEMMAP enabled. For the classic sparse, subsection map is meaningless and confusing. About the classic sparse which doesn't support subsection hotplug, Dan said it's more because the effort and maintenance burden outweighs the benefit. Besides, the current 64 bit ARCHes all enable SPARSEMEM_VMEMMAP_ENABLE by default. This patch (of 5): Factor out the code that fills the subsection map from section_activate() into fill_subsection_map(), this makes section_activate() cleaner and easier to follow. Signed-off-by:
Wei Yang <richard.weiyang@gmail.com> Reviewed-by:
-
David Hildenbrand authored
Let's drop the basically unused section stuff and simplify. The logic now matches the logic in __remove_pages(). Signed-off-by:
-
David Hildenbrand authored
In commit 52fb87c8 ("mm/memory_hotplug: cleanup __remove_pages()"), we cleaned up __remove_pages(), and introduced a shorter variant to calculate the number of pages to the next section boundary. Turns out we can make this calculation easier to read. We always want to have the number of pages (> 0) to the next section boundary, starting from the current pfn. We'll clean up __remove_pages() in a follow-up patch and directly make use of this computation. Suggested-by:
Segher Boessenkool <segher@kernel.crashing.org> Signed-off-by:
-
Baoquan He authored
In commit 357b4da5 ("x86: respect memory size limiting via mem= parameter") a global varialbe max_mem_size is added to store the value parsed from 'mem= ', then checked when memory region is added. This truly stops those DIMMs from being added into system memory during boot-time. However, it also limits the later memory hotplug functionality. Any DIMM can't be hotplugged any more if its region is beyond the max_mem_size. We will get errors like: [ 216.387164] acpi PNP0C80:02: add_memory failed [ 216.389301] acpi PNP0C80:02: acpi_memory_enable_device() error [ 216.392187] acpi PNP0C80:02: Enumeration failure This will cause issue in a known use case where 'mem=' is added to the hypervisor. The memory that lies after 'mem=' boundary will be assigned to KVM guests. After commit 357b4da5 merged, memory can't be extended dynamically if system memory on hypervisor is not sufficient. So fix it by also checking if it's during boot-time restricting to add memory. Otherwise, skip the restriction. And also add this use case to document of 'mem=' kernel parameter. Fixes: 357b4da5 ("x86: respect memory size limiting via mem= parameter") Signed-off-by:
Juergen Gross <jgross@suse.com> Acked-by:
-
David Hildenbrand authored
Since commit c5e79ef5 ("mm/memory_hotplug.c: don't allow to online/offline memory blocks with holes") we disallow to offline any memory with holes. As all boot memory is online and hotplugged memory cannot contain holes, we never online memory with holes. This present check can be dropped. Signed-off-by:
-
David Hildenbrand authored
pages_correctly_probed() is a leftover from ancient times. It dates back to commit 3947be19 ("[PATCH] memory hotplug: sysfs and add/remove functions"), where Pg_reserved checks were added as a sfety net: /* * The probe routines leave the pages reserved, just * as the bootmem code does. Make sure they're still * that way. */ The checks were refactored quite a bit over the years, especially in commit b77eab70 ("mm/memory_hotplug: optimize probe routine"), where checks for present, valid, and online sections were added. Hotplugged memory is added via add_memory(), which will create the full memmap for the hotplugged memory, and mark all sections valid and present. Only full memory blocks are onlined/offlined, so we also cannot have an inconsistency in that regard (especially, memory blocks with some sections being online and some being offline). 1. Boot memory always starts online. Since commit c5e79ef5 ("mm/memory_hotplug.c: don't allow to online/offline memory blocks with holes") we disallow to offline any memory with holes. Therefore, we never online memory with holes. Present and validity checks are superfluous. 2. Only complete memory blocks are onlined/offlined (and especially, the state - online or offline - is stored for whole memory blocks). Besides the core, only arch/powerpc/platforms/powernv/memtrace.c manually calls offline_pages() and fiddels with memory block states. But it also only offlines complete memory blocks. 3. To make any of these conditions trigger, something would have to be terribly messed up in the core. (e.g., online/offline only some sections of a memory block). 4. Memory unplug properly makes sure that all sysfs attributes were removed (and therefore, that all threads left the sysfs handlers). We don't have to worry about zombie devices at this point. 5. The valid_section_nr(section_nr) check is actually dead code, as it would never have been reached due to the WARN_ON_ONCE(!pfn_valid(pfn)). No wonder we haven't seen any of these errors in a long time (or even ever, according to my search). Let's just get rid of them. Now, all checks that could hinder onlining and offlining are completely contained in online_pages()/offline_pages(). Signed-off-by:
-
David Hildenbrand authored
Patch series "mm: drop superfluous section checks when onlining/offlining". Let's drop some superfluous section checks on the onlining/offlining path. This patch (of 3): Since commit c5e79ef5 ("mm/memory_hotplug.c: don't allow to online/offline memory blocks with holes") we have a generic check in offline_pages() that disallows offlining memory blocks with holes. Memory blocks with missing sections are just another variant of these type of blocks. We can stop checking (and especially storing) present sections. A proper error message is now printed why offlining failed. section_count was initially introduced in commit 07681215 ("Driver core: Add section count to memory_block struct") in order to detect when it is okay to remove a memory block. It was used in commit 26bbe7ef ("drivers/base/memory.c: prohibit offlining of memory blocks with missing sections") to disallow offlining memory blocks with missing sections. As we refactored creation/removal of memory devices and have a proper check for holes in place, we can drop the section_count. This also removes a leftover comment regarding the mem_sysfs_mutex, which was removed in commit 848e19ad ("drivers/base/memory.c: drop the mem_sysfs_mutex"). Signed-off-by:
-
Peter Xu authored
Add uffd tests for write protection. Instead of introducing new tests for it, let's simply squashing uffd-wp tests into existing uffd-missing test cases. Changes are: (1) Bouncing tests We do the write-protection in two ways during the bouncing test: - By using UFFDIO_COPY_MODE_WP when resolving MISSING pages: then we'll make sure for each bounce process every single page will be at least fault twice: once for MISSING, once for WP. - By direct call UFFDIO_WRITEPROTECT on existing faulted memories: To further torture the explicit page protection procedures of uffd-wp, we split each bounce procedure into two halves (in the background thread): the first half will be MISSING+WP for each page as explained above. After the first half, we write protect the faulted region in the background thread to make sure at least half of the pages will be write protected again which is the first half to test the new UFFDIO_WRITEPROTECT call. Then we continue with the 2nd half, which will contain both MISSING and WP faulting tests for the 2nd half and WP-only faults from the 1st half. (2) Event/Signal test Mostly previous tests but will do MISSING+WP for each page. For sigbus-mode test we'll need to provide standalone path to handle the write protection faults. For all tests, do statistics as well for uffd-wp pages. Signed-off-by:Peter Xu <peterx@redhat.com> Signed-off-by:
-
Peter Xu authored
Introduce uffd_stats structure for statistics of the self test, at the same time refactor the code to always pass in the uffd_stats for either read() or poll() typed fault handling threads instead of using two different ways to return the statistic results. No functional change. With the new structure, it's very easy to introduce new statistics. Signed-off-by:
Mike Rapoport <rppt@linux.vnet.ibm.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Bobby Powers <bobbypowers@gmail.com> Cc: Brian Geffon <bgeffon@google.com> Cc: David Hildenbrand <david@redhat.com> Cc: Denis Plotnikov <dplotnikov@virtuozzo.com> Cc: "Dr . David Alan Gilbert" <dgilbert@redhat.com> Cc: Hugh Dickins <hughd@google.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: "Kirill A . Shutemov" <kirill@shutemov.name> Cc: Martin Cracauer <cracauer@cons.org> Cc: Marty McFadden <mcfadden8@llnl.gov> Cc: Maya Gokhale <gokhale2@llnl.gov> Cc: Mel Gorman <mgorman@suse.de> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Pavel Emelyanov <xemul@openvz.org> Cc: Rik van Riel <riel@redhat.com> Cc: Shaohua Li <shli@fb.com> Link: http://lkml.kernel.org/r/20200220163112.11409-19-peterx@redhat.comSigned-off-by:
-
Peter Xu authored
Only declare _UFFDIO_WRITEPROTECT if the user specified UFFDIO_REGISTER_MODE_WP and if all the checks passed. Then when the user registers regions with shmem/hugetlbfs we won't expose the new ioctl to them. Even with complete anonymous memory range, we'll only expose the new WP ioctl bit if the register mode has MODE_WP. Signed-off-by:
-
Martin Cracauer authored
Add documentation about the write protection support. [peterx@redhat.com: rewrite in rst format; fixups here and there] Signed-off-by:
Martin Cracauer <cracauer@cons.org> Signed-off-by:
Andrea Arcangeli <aarcange@redhat.com> Signed-off-by:
Jerome Glisse <jglisse@redhat.com> Reviewed-by:
-
Peter Xu authored
It does not make sense to try to wake up any waiting thread when we're write-protecting a memory region. Only wake up when resolving a write protected page fault. Signed-off-by:
-
Shaohua Li authored
Now it's safe to enable write protection in userfaultfd API Signed-off-by:
Shaohua Li <shli@fb.com> Signed-off-by:
-
Andrea Arcangeli authored
Introduce the new uffd-wp APIs for userspace. Firstly, we'll allow to do UFFDIO_REGISTER with write protection tracking using the new UFFDIO_REGISTER_MODE_WP flag. Note that this flag can co-exist with the existing UFFDIO_REGISTER_MODE_MISSING, in which case the userspace program can not only resolve missing page faults, and at the same time tracking page data changes along the way. Secondly, we introduced the new UFFDIO_WRITEPROTECT API to do page level write protection tracking. Note that we will need to register the memory region with UFFDIO_REGISTER_MODE_WP before that. [peterx@redhat.com: write up the commit message] [peterx@redhat.com: remove useless block, write commit message, check against VM_MAYWRITE rather than VM_WRITE when register] Signed-off-by:
-
Shaohua Li authored
Add API to enable/disable writeprotect a vma range. Unlike mprotect, this doesn't split/merge vmas. [peterx@redhat.com: - use the helper to find VMA; - return -ENOENT if not found to match mcopy case; - use the new MM_CP_UFFD_WP* flags for change_protection - check against mmap_changing for failures - replace find_dst_vma with vma_find_uffd] Signed-off-by:
-
Peter Xu authored
Don't collapse the huge PMD if there is any userfault write protected small PTEs. The problem is that the write protection is in small page granularity and there's no way to keep all these write protection information if the small pages are going to be merged into a huge PMD. The same thing needs to be considered for swap entries and migration entries. So do the check as well disregarding khugepaged_max_ptes_swap. Signed-off-by:
-
Peter Xu authored
For either swap and page migration, we all use the bit 2 of the entry to identify whether this entry is uffd write-protected. It plays a similar role as the existing soft dirty bit in swap entries but only for keeping the uffd-wp tracking for a specific PTE/PMD. Something special here is that when we want to recover the uffd-wp bit from a swap/migration entry to the PTE bit we'll also need to take care of the _PAGE_RW bit and make sure it's cleared, otherwise even with the _PAGE_UFFD_WP bit we can't trap it at all. In change_pte_range() we do nothing for uffd if the PTE is a swap entry. That can lead to data mismatch if the page that we are going to write protect is swapped out when sending the UFFDIO_WRITEPROTECT. This patch also applies/removes the uffd-wp bit even for the swap entries. Signed-off-by:
-
Peter Xu authored
Adding these missing helpers for uffd-wp operations with pmd swap/migration entries. Signed-off-by:
-
Peter Xu authored
UFFD_EVENT_FORK support for uffd-wp should be already there, except that we should clean the uffd-wp bit if uffd fork event is not enabled. Detect that to avoid _PAGE_UFFD_WP being set even if the VMA is not being tracked by VM_UFFD_WP. Do this for both small PTEs and huge PMDs. Signed-off-by:
-
Peter Xu authored
Firstly, introduce two new flags MM_CP_UFFD_WP[_RESOLVE] for change_protection() when used with uffd-wp and make sure the two new flags are exclusively used. Then, - For MM_CP_UFFD_WP: apply the _PAGE_UFFD_WP bit and remove _PAGE_RW when a range of memory is write protected by uffd - For MM_CP_UFFD_WP_RESOLVE: remove the _PAGE_UFFD_WP bit and recover _PAGE_RW when write protection is resolved from userspace And use this new interface in mwriteprotect_range() to replace the old MM_CP_DIRTY_ACCT. Do this change for both PTEs and huge PMDs. Then we can start to identify which PTE/PMD is write protected by general (e.g., COW or soft dirty tracking), and which is for userfaultfd-wp. Since we should keep the _PAGE_UFFD_WP when doing pte_modify(), add it into _PAGE_CHG_MASK as well. Meanwhile, since we have this new bit, we can be even more strict when detecting uffd-wp page faults in either do_wp_page() or wp_huge_pmd(). After we're with _PAGE_UFFD_WP, a special case is when a page is both protected by the general COW logic and also userfault-wp. Here the userfault-wp will have higher priority and will be handled first. Only after the uffd-wp bit is cleared on the PTE/PMD will we continue to handle the general COW. These are the steps on what will happen with such a page: 1. CPU accesses write protected shared page (so both protected by general COW and uffd-wp), blocked by uffd-wp first because in do_wp_page we'll handle uffd-wp first, so it has higher priority than general COW. 2. Uffd service thread receives the request, do UFFDIO_WRITEPROTECT to remove the uffd-wp bit upon the PTE/PMD. However here we still keep the write bit cleared. Notify the blocked CPU. 3. The blocked CPU resumes the page fault process with a fault retry, during retry it'll notice it was not with the uffd-wp bit this time but it is still write protected by general COW, then it'll go though the COW path in the fault handler, copy the page, apply write bit where necessary, and retry again. 4. The CPU will be able to access this page with write bit set. Suggested-by: -
Peter Xu authored
change_protection() was used by either the NUMA or mprotect() code, there's one parameter for each of the callers (dirty_accountable and prot_numa). Further, these parameters are passed along the calls: - change_protection_range() - change_p4d_range() - change_pud_range() - change_pmd_range() - ... Now we introduce a flag for change_protect() and all these helpers to replace these parameters. Then we can avoid passing multiple parameters multiple times along the way. More importantly, it'll greatly simplify the work if we want to introduce any new parameters to change_protection(). In the follow up patches, a new parameter for userfaultfd write protection will be introduced. No functional change at all. Signed-off-by:
-
Andrea Arcangeli authored
This allows UFFDIO_COPY to map pages write-protected. [peterx@redhat.com: switch to VM_WARN_ON_ONCE in mfill_atomic_pte; add brackets around "dst_vma->vm_flags & VM_WRITE"; fix wordings in comments and commit messages] Signed-off-by:
-
Andrea Arcangeli authored
Implement helpers methods to invoke userfaultfd wp faults more selectively: not only when a wp fault triggers on a vma with vma->vm_flags VM_UFFD_WP set, but only if the _PAGE_UFFD_WP bit is set in the pagetable too. Signed-off-by:
-
Andrea Arcangeli authored
Accurate userfaultfd WP tracking is possible by tracking exactly which virtual memory ranges were writeprotected by userland. We can't relay only on the RW bit of the mapped pagetable because that information is destroyed by fork() or KSM or swap. If we were to relay on that, we'd need to stay on the safe side and generate false positive wp faults for every swapped out page. [peterx@redhat.com: append _PAGE_UFD_WP to _PAGE_CHG_MASK] Signed-off-by:
-
Andrea Arcangeli authored
There are several cases write protection fault happens. It could be a write to zero page, swaped page or userfault write protected page. When the fault happens, there is no way to know if userfault write protect the page before. Here we just blindly issue a userfault notification for vma with VM_UFFD_WP regardless if app write protects it yet. Application should be ready to handle such wp fault. In the swapin case, always swapin as readonly. This will cause false positive userfaults. We need to decide later if to eliminate them with a flag like soft-dirty in the swap entry (see _PAGE_SWP_SOFT_DIRTY). hugetlbfs wouldn't need to worry about swapouts but and tmpfs would be handled by a swap entry bit like anonymous memory. The main problem with no easy solution to eliminate the false positives, will be if/when userfaultfd is extended to real filesystem pagecache. When the pagecache is freed by reclaim we can't leave the radix tree pinned if the inode and in turn the radix tree is reclaimed as well. The estimation is that full accuracy and lack of false positives could be easily provided only to anonymous memory (as long as there's no fork or as long as MADV_DONTFORK is used on the userfaultfd anonymous range) tmpfs and hugetlbfs, it's most certainly worth to achieve it but in a later incremental patch. [peterx@redhat.com: don't conditionally drop FAULT_FLAG_WRITE in do_swap_page] Signed-off-by:
-
Shaohua Li authored
Patch series "userfaultfd: write protection support", v6. Overview ======== The uffd-wp work was initialized by Shaohua Li [1], and later continued by Andrea [2]. This series is based upon Andrea's latest userfaultfd tree, and it is a continuous works from both Shaohua and Andrea. Many of the follow up ideas come from Andrea too. Besides the old MISSING register mode of userfaultfd, the new uffd-wp support provides another alternative register mode called UFFDIO_REGISTER_MODE_WP that can be used to listen to not only missing page faults but also write protection page faults, or even they can be registered together. At the same time, the new feature also provides a new userfaultfd ioctl called UFFDIO_WRITEPROTECT which allows the userspace to write protect a range or memory or fixup write permission of faulted pages. Please refer to the document patch "userfaultfd: wp: UFFDIO_REGISTER_MODE_WP documentation update" for more information on the new interface and what it can do. The major workflow of an uffd-wp program should be: 1. Register a memory region with WP mode using UFFDIO_REGISTER_MODE_WP 2. Write protect part of the whole registered region using UFFDIO_WRITEPROTECT, passing in UFFDIO_WRITEPROTECT_MODE_WP to show that we want to write protect the range. 3. Start a working thread that modifies the protected pages, meanwhile listening to UFFD messages. 4. When a write is detected upon the protected range, page fault happens, a UFFD message will be generated and reported to the page fault handling thread 5. The page fault handler thread resolves the page fault using the new UFFDIO_WRITEPROTECT ioctl, but this time passing in !UFFDIO_WRITEPROTECT_MODE_WP instead showing that we want to recover the write permission. Before this operation, the fault handler thread can do anything it wants, e.g., dumps the page to a persistent storage. 6. The worker thread will continue running with the correctly applied write permission from step 5. Currently there are already two projects that are based on this new userfaultfd feature. QEMU Live Snapshot: The project provides a way to allow the QEMU hypervisor to take snapshot of VMs without stopping the VM [3]. LLNL umap library: The project provides a mmap-like interface and "allow to have an application specific buffer of pages cached from a large file, i.e. out-of-core execution using memory map" [4][5]. Before posting the patchset, this series was smoke tested against QEMU live snapshot and the LLNL umap library (by doing parallel quicksort using 128 sorting threads + 80 uffd servicing threads). My sincere thanks to Marty Mcfadden and Denis Plotnikov for the help along the way. TODO ==== - hugetlbfs/shmem support - performance - more architectures - cooperate with mprotect()-allowed processes (???) - ... References ========== [1] https://lwn.net/Articles/666187/ [2] https://git.kernel.org/pub/scm/linux/kernel/git/andrea/aa.git/log/?h=userfault [3] https://github.com/denis-plotnikov/qemu/commits/background-snapshot-kvm [4] https://github.com/LLNL/umap [5] https://llnl-umap.readthedocs.io/en/develop/ [6] https://git.kernel.org/pub/scm/linux/kernel/git/andrea/aa.git/commit/?h=userfault&id=b245ecf6cf59156966f3da6e6b674f6695a5ffa5 [7] https://lkml.org/lkml/2018/11/21/370 [8] https://lkml.org/lkml/2018/12/30/64 This patch (of 19): Add helper for writeprotect check. Will use it later. Signed-off-by: -
David Hildenbrand authored
Commit 71994620 ("virtio_balloon: replace oom notifier with shrinker") changed the behavior when deflation happens automatically. Instead of deflating when called by the OOM handler, the shrinker is used. However, the balloon is not simply some other slab cache that should be shrunk when under memory pressure. The shrinker does not have a concept of priorities yet, so this behavior cannot be configured. Eventually once that is in place, we might want to switch back after doing proper testing. There was a report that this results in undesired side effects when inflating the balloon to shrink the page cache. [1] "When inflating the balloon against page cache (i.e. no free memory remains) vmscan.c will both shrink page cache, but also invoke the shrinkers -- including the balloon's shrinker. So the balloon driver allocates memory which requires reclaim, vmscan gets this memory by shrinking the balloon, and then the driver adds the memory back to the balloon. Basically a busy no-op." The name "deflate on OOM" makes it pretty clear when deflation should happen - after other approaches to reclaim memory failed, not while reclaiming. This allows to minimize the footprint of a guest - memory will only be taken out of the balloon when really needed. Keep using the shrinker for VIRTIO_BALLOON_F_FREE_PAGE_HINT, because this has no such side effects. Always register the shrinker with VIRTIO_BALLOON_F_FREE_PAGE_HINT now. We are always allowed to reuse free pages that are still to be processed by the guest. The hypervisor takes care of identifying and resolving possible races between processing a hinting request and the guest reusing a page. In contrast to pre commit 71994620 ("virtio_balloon: replace oom notifier with shrinker"), don't add a module parameter to configure the number of pages to deflate on OOM. Can be re-added if really needed. Also, pay attention that leak_balloon() returns the number of 4k pages - convert it properly in virtio_balloon_oom_notify(). Testing done by Tyler for future reference: Test setup: VM with 16 CPU, 64GB RAM. Running Debian 10. We have a 42 GB file full of random bytes that we continually cat to /dev/null. This fills the page cache as the file is read. Meanwhile, we trigger the balloon to inflate, with a target size of 53 GB. This setup causes the balloon inflation to pressure the page cache as the page cache is also trying to grow. Afterwards we shrink the balloon back to zero (so total deflate == total inflate). Without this patch (kernel 4.19.0-5): Inflation never reaches the target until we stop the "cat file > /dev/null" process. Total inflation time was 542 seconds. The longest period that made no net forward progress was 315 seconds. Result of "grep balloon /proc/vmstat" after the test: balloon_inflate 154828377 balloon_deflate 154828377 With this patch (kernel 5.6.0-rc4+): Total inflation duration was 63 seconds. No deflate-queue activity occurs when pressuring the page-cache. Result of "grep balloon /proc/vmstat" after the test: balloon_inflate 12968539 balloon_deflate 12968539 Conclusion: This patch fixes the issue. In the test it reduced inflate/deflate activity by 12x, and reduced inflation time by 8.6x. But more importantly, if we hadn't killed the "cat file > /dev/null" process then, without the patch, the inflation process would never reach the target. [1] https://www.spinics.net/lists/linux-virtualization/msg40863.html Link: http://lkml.kernel.org/r/20200311135523.18512-2-david@redhat.com Fixes: 71994620 ("virtio_balloon: replace oom notifier with shrinker") Signed-off-by:
Tyler Sanderson <tysand@google.com> Tested-by:
David Rientjes <rientjes@google.com> Acked-by:
Michael S. Tsirkin <mst@redhat.com> Cc: Wei Wang <wei.w.wang@intel.com> Cc: Alexander Duyck <alexander.h.duyck@linux.intel.com> Cc: David Rientjes <rientjes@google.com> Cc: Nadav Amit <namit@vmware.com> Cc: Michal Hocko <mhocko@kernel.org> Signed-off-by:
-
Alexander Duyck authored
Add documentation for free page reporting. Currently the only consumer is virtio-balloon, however it is possible that other drivers might make use of this so it is best to add a bit of documetation explaining at a high level how to use the API. Signed-off-by:
Alexander Duyck <alexander.h.duyck@linux.intel.com> Signed-off-by:
-
Alexander Duyck authored
In order to keep ourselves from reporting pages that are just going to be reused again in the case of heavy churn we can put a limit on how many total pages we will process per pass. Doing this will allow the worker thread to go into idle much more quickly so that we avoid competing with other threads that might be allocating or freeing pages. The logic added here will limit the worker thread to no more than one sixteenth of the total free pages in a given area per list. Once that limit is reached it will update the state so that at the end of the pass we will reschedule the worker to try again in 2 seconds when the memory churn has hopefully settled down. Again this optimization doesn't show much of a benefit in the standard case as the memory churn is minmal. However with page allocator shuffling enabled the gain is quite noticeable. Below are the results with a THP enabled version of the will-it-scale page_fault1 test showing the improvement in iterations for 16 processes or threads. Without: tasks processes processes_idle threads threads_idle 16 8283274.75 0.17 5594261.00 38.15 With: tasks processes processes_idle threads threads_idle 16 8767010.50 0.21 5791312.75 36.98 Signed-off-by:
Mel Gorman <mgorman@techsingularity.net> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: David Hildenbrand <david@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Luiz Capitulino <lcapitulino@redhat.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Nitesh Narayan Lal <nitesh@redhat.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Pankaj Gupta <pagupta@redhat.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Rik van Riel <riel@surriel.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Wei Wang <wei.w.wang@intel.com> Cc: Yang Zhang <yang.zhang.wz@gmail.com> Cc: wei qi <weiqi4@huawei.com> Link: http://lkml.kernel.org/r/20200211224719.29318.72113.stgit@localhost.localdomainSigned-off-by:
-
Alexander Duyck authored
Rather than walking over the same pages again and again to get to the pages that have yet to be reported we can save ourselves a significant amount of time by simply rotating the list so that when we have a full list of reported pages the head of the list is pointing to the next non-reported page. Doing this should save us some significant time when processing each free list. This doesn't gain us much in the standard case as all of the non-reported pages should be near the top of the list already. However in the case of page shuffling this results in a noticeable improvement. Below are the will-it-scale page_fault1 w/ THP numbers for 16 tasks with and without this patch. Without: tasks processes processes_idle threads threads_idle 16 8093776.25 0.17 5393242.00 38.20 With: tasks processes processes_idle threads threads_idle 16 8283274.75 0.17 5594261.00 38.15 Signed-off-by:
-
Alexander Duyck authored
Add support for the page reporting feature provided by virtio-balloon. Reporting differs from the regular balloon functionality in that is is much less durable than a standard memory balloon. Instead of creating a list of pages that cannot be accessed the pages are only inaccessible while they are being indicated to the virtio interface. Once the interface has acknowledged them they are placed back into their respective free lists and are once again accessible by the guest system. Unlike a standard balloon we don't inflate and deflate the pages. Instead we perform the reporting, and once the reporting is completed it is assumed that the page has been dropped from the guest and will be faulted back in the next time the page is accessed. Signed-off-by:
-
Alexander Duyck authored
Currently the page poisoning setting wasn't being enabled unless free page hinting was enabled. However we will need the page poisoning tracking logic as well for free page reporting. As such pull it out and make it a separate bit of config in the probe function. In addition we need to add support for the more recent init_on_free feature which expects a behavior similar to page poisoning in that we expect the page to be pre-zeroed. Signed-off-by:
-
Alexander Duyck authored
In order to pave the way for free page reporting in virtualized environments we will need a way to get pages out of the free lists and identify those pages after they have been returned. To accomplish this, this patch adds the concept of a Reported Buddy, which is essentially meant to just be the Uptodate flag used in conjunction with the Buddy page type. To prevent the reported pages from leaking outside of the buddy lists I added a check to clear the PageReported bit in the del_page_from_free_list function. As a result any reported page that is split, merged, or allocated will have the flag cleared prior to the PageBuddy value being cleared. The process for reporting pages is fairly simple. Once we free a page that meets the minimum order for page reporting we will schedule a worker thread to start 2s or more in the future. That worker thread will begin working from the lowest supported page reporting order up to MAX_ORDER - 1 pulling unreported pages from the free list and storing them in the scatterlist. When processing each individual free list it is necessary for the worker thread to release the zone lock when it needs to stop and report the full scatterlist of pages. To reduce the work of the next iteration the worker thread will rotate the free list so that the first unreported page in the free list becomes the first entry in the list. It will then call a reporting function providing information on how many entries are in the scatterlist. Once the function completes it will return the pages to the free area from which they were allocated and start over pulling more pages from the free areas until there are no longer enough pages to report on to keep the worker busy, or we have processed as many pages as were contained in the free area when we started processing the list. The worker thread will work in a round-robin fashion making its way though each zone requesting reporting, and through each reportable free list within that zone. Once all free areas within the zone have been processed it will check to see if there have been any requests for reporting while it was processing. If so it will reschedule the worker thread to start up again in roughly 2s and exit. Signed-off-by:
-
Alexander Duyck authored
There are cases where we would benefit from avoiding having to go through the allocation and free cycle to return an isolated page. Examples for this might include page poisoning in which we isolate a page and then put it back in the free list without ever having actually allocated it. This will enable us to also avoid notifiers for the future free page reporting which will need to avoid retriggering page reporting when returning pages that have been reported on. Signed-off-by:
-
Alexander Duyck authored
In order to enable the use of the zone from the list manipulator functions I will need access to the zone pointer. As it turns out most of the accessors were always just being directly passed &zone->free_area[order] anyway so it would make sense to just fold that into the function itself and pass the zone and order as arguments instead of the free area. In order to be able to reference the zone we need to move the declaration of the functions down so that we have the zone defined before we define the list manipulation functions. Since the functions are only used in the file mm/page_alloc.c we can just move them there to reduce noise in the header. Signed-off-by:
Dan Williams <dan.j.williams@intel.com> Reviewed-by:
Pankaj Gupta <pagupta@redhat.com> Acked-by:
-
Alexander Duyck authored
Patch series "mm / virtio: Provide support for free page reporting", v17. This series provides an asynchronous means of reporting free guest pages to a hypervisor so that the memory associated with those pages can be dropped and reused by other processes and/or guests on the host. Using this it is possible to avoid unnecessary I/O to disk and greatly improve performance in the case of memory overcommit on the host. When enabled we will be performing a scan of free memory every 2 seconds while pages of sufficiently high order are being freed. In each pass at least one sixteenth of each free list will be reported. By doing this we avoid racing against other threads that may be causing a high amount of memory churn. The lowest page order currently scanned when reporting pages is pageblock_order so that this feature will not interfere with the use of Transparent Huge Pages in the case of virtualization. Currently this is only in use by virtio-balloon however there is the hope that at some point in the future other hypervisors might be able to make use of it. In the virtio-balloon/QEMU implementation the hypervisor is currently using MADV_DONTNEED to indicate to the host kernel that the page is currently free. It will be zeroed and faulted back into the guest the next time the page is accessed. To track if a page is reported or not the Uptodate flag was repurposed and used as a Reported flag for Buddy pages. We walk though the free list isolating pages and adding them to the scatterlist until we either encounter the end of the list or have processed at least one sixteenth of the pages that were listed in nr_free prior to us starting. If we fill the scatterlist before we reach the end of the list we rotate the list so that the first unreported page we encounter is moved to the head of the list as that is where we will resume after we have freed the reported pages back into the tail of the list. Below are the results from various benchmarks. I primarily focused on two tests. The first is the will-it-scale/page_fault2 test, and the other is a modified version of will-it-scale/page_fault1 that was enabled to use THP. I did this as it allows for better visibility into different parts of the memory subsystem. The guest is running with 32G for RAM on one node of a E5-2630 v3. The host has had some features such as CPU turbo disabled in the BIOS. Test page_fault1 (THP) page_fault2 Name tasks Process Iter STDEV Process Iter STDEV Baseline 1 1012402.50 0.14% 361855.25 0.81% 16 8827457.25 0.09% 3282347.00 0.34% Patches Applied 1 1007897.00 0.23% 361887.00 0.26% 16 8784741.75 0.39% 3240669.25 0.48% Patches Enabled 1 1010227.50 0.39% 359749.25 0.56% 16 8756219.00 0.24% 3226608.75 0.97% Patches Enabled 1 1050982.00 4.26% 357966.25 0.14% page shuffle 16 8672601.25 0.49% 3223177.75 0.40% Patches enabled 1 1003238.00 0.22% 360211.00 0.22% shuffle w/ RFC 16 8767010.50 0.32% 3199874.00 0.71% The results above are for a baseline with a linux-next-20191219 kernel, that kernel with this patch set applied but page reporting disabled in virtio-balloon, the patches applied and page reporting fully enabled, the patches enabled with page shuffling enabled, and the patches applied with page shuffling enabled and an RFC patch that makes used of MADV_FREE in QEMU. These results include the deviation seen between the average value reported here versus the high and/or low value. I observed that during the test memory usage for the first three tests never dropped whereas with the patches fully enabled the VM would drop to using only a few GB of the host's memory when switching from memhog to page fault tests. Any of the overhead visible with this patch set enabled seems due to page faults caused by accessing the reported pages and the host zeroing the page before giving it back to the guest. This overhead is much more visible when using THP than with standard 4K pages. In addition page shuffling seemed to increase the amount of faults generated due to an increase in memory churn. The overehad is reduced when using MADV_FREE as we can avoid the extra zeroing of the pages when they are reintroduced to the host, as can be seen when the RFC is applied with shuffling enabled. The overall guest size is kept fairly small to only a few GB while the test is running. If the host memory were oversubscribed this patch set should result in a performance improvement as swapping memory in the host can be avoided. A brief history on the background of free page reporting can be found at: https://lore.kernel.org/lkml/29f43d5796feed0dec8e8bb98b187d9dac03b900.camel@linux.intel.com/ This patch (of 9): Move the head/tail adding logic out of the shuffle code and into the __free_one_page function since ultimately that is where it is really needed anyway. By doing this we should be able to reduce the overhead and can consolidate all of the list addition bits in one spot. Signed-off-by: -
Huang Ying authored
Some comments for MADV_FREE is revised and added to help people understand the MADV_FREE code, especially the page flag, PG_swapbacked. This makes page_is_file_cache() isn't consistent with its comments. So the function is renamed to page_is_file_lru() to make them consistent again. All these are put in one patch as one logical change. Suggested-by:
Johannes Weiner <hannes@cmpxchg.org> Suggested-by:
"Huang, Ying" <ying.huang@intel.com> Signed-off-by:
Michal Hocko <mhocko@kernel.org> Acked-by:
Vlastimil Babka <vbabka@suse.cz> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan@kernel.org> Cc: Hugh Dickins <hughd@google.com> Cc: Rik van Riel <riel@surriel.com> Link: http://lkml.kernel.org/r/20200317100342.2730705-1-ying.huang@intel.comSigned-off-by:
-
