linux_dsm_epyc7002/mm
Alexander Duyck a2129f2479 mm: adjust shuffle code to allow for future coalescing
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: Alexander Duyck <alexander.h.duyck@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Yang Zhang <yang.zhang.wz@gmail.com>
Cc: Pankaj Gupta <pagupta@redhat.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Nitesh Narayan Lal <nitesh@redhat.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Luiz Capitulino <lcapitulino@redhat.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Wei Wang <wei.w.wang@intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: wei qi <weiqi4@huawei.com>
Link: http://lkml.kernel.org/r/20200211224602.29318.84523.stgit@localhost.localdomain
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-04-07 10:43:38 -07:00
..
kasan kasan: detect negative size in memory operation function 2020-04-02 09:35:30 -07:00
backing-dev.c memcg: fix a crash in wb_workfn when a device disappears 2020-01-31 10:30:36 -08:00
balloon_compaction.c
cleancache.c
cma_debug.c mm/cma_debug.c: use DEFINE_DEBUGFS_ATTRIBUTE to define debugfs fops 2019-12-01 12:59:09 -08:00
cma.c mm/cma.c: switch to bitmap_zalloc() for cma bitmap allocation 2019-12-01 12:59:09 -08:00
cma.h
compaction.c mm: code cleanup for MADV_FREE 2020-04-07 10:43:38 -07:00
debug_page_ref.c
debug.c mm: dump_page(): additional diagnostics for huge pinned pages 2020-04-02 09:35:27 -07:00
dmapool.c
early_ioremap.c mm/early_ioremap.c: use %pa to print resource_size_t variables 2020-01-31 10:30:38 -08:00
fadvise.c
failslab.c
filemap.c mm: allow VM_FAULT_RETRY for multiple times 2020-04-02 09:35:30 -07:00
frame_vector.c
frontswap.c
gup_benchmark.c mm/gup_benchmark: support pin_user_pages() and related calls 2020-04-02 09:35:27 -07:00
gup.c mm: code cleanup for MADV_FREE 2020-04-07 10:43:38 -07:00
highmem.c mm, x86/mm: Untangle address space layout definitions from basic pgtable type definitions 2019-12-10 10:12:55 +01:00
hmm.c mm/hmm: return error for non-vma snapshots 2020-03-30 16:58:36 -03:00
huge_memory.c mm: remove CONFIG_TRANSPARENT_HUGE_PAGECACHE 2020-04-07 10:43:38 -07:00
hugetlb_cgroup.c hugetlb_cgroup: add accounting for shared mappings 2020-04-02 09:35:32 -07:00
hugetlb.c mm/hugetlb: remove unnecessary memory fetch in PageHeadHuge() 2020-04-02 09:35:32 -07:00
hwpoison-inject.c mm/hwpoison-inject: use DEFINE_DEBUGFS_ATTRIBUTE to define debugfs fops 2019-12-01 12:59:09 -08:00
init-mm.c
internal.h mm,compaction,cma: add alloc_contig flag to compact_control 2020-04-02 09:35:31 -07:00
interval_tree.c
Kconfig mm: remove CONFIG_TRANSPARENT_HUGE_PAGECACHE 2020-04-07 10:43:38 -07:00
Kconfig.debug mm: add generic ptdump 2020-02-04 03:05:25 +00:00
khugepaged.c mm: code cleanup for MADV_FREE 2020-04-07 10:43:38 -07:00
kmemleak-test.c
kmemleak.c mm/kmemleak.c: use address-of operator on section symbols 2020-04-02 09:35:26 -07:00
ksm.c mm/ksm.c: update get_user_pages() argument in comment 2020-04-07 10:43:38 -07:00
list_lru.c mm: memcg/slab: use mem_cgroup_from_obj() 2020-04-02 09:35:28 -07:00
maccess.c
madvise.c mm: do not allow MADV_PAGEOUT for CoW pages 2020-03-21 18:56:06 -07:00
Makefile mm/Makefile: disable KCSAN for kmemleak 2020-04-02 09:35:26 -07:00
mapping_dirty_helpers.c mm/mapping_dirty_helpers: update huge page-table entry callbacks 2020-04-02 09:35:29 -07:00
memblock.c mm/memblock.c: remove redundant assignment to variable max_addr 2020-04-02 09:35:32 -07:00
memcontrol.c mm, memcg: bypass high reclaim iteration for cgroup hierarchy root 2020-04-07 10:43:37 -07:00
memfd.c
memory_hotplug.c mm: code cleanup for MADV_FREE 2020-04-07 10:43:38 -07:00
memory-failure.c mm: code cleanup for MADV_FREE 2020-04-07 10:43:38 -07:00
memory.c mm: remove CONFIG_TRANSPARENT_HUGE_PAGECACHE 2020-04-07 10:43:38 -07:00
mempolicy.c mm: code cleanup for MADV_FREE 2020-04-07 10:43:38 -07:00
mempool.c
memremap.c memremap: add an owner field to struct dev_pagemap 2020-03-26 14:33:37 -03:00
memtest.c
migrate.c mm: code cleanup for MADV_FREE 2020-04-07 10:43:38 -07:00
mincore.c mm: pagewalk: add 'depth' parameter to pte_hole 2020-02-04 03:05:25 +00:00
mlock.c
mm_init.c
mmap.c mm/vma: make vma_is_accessible() available for general use 2020-04-07 10:43:37 -07:00
mmu_context.c
mmu_gather.c asm-generic/tlb: provide MMU_GATHER_TABLE_FREE 2020-02-04 03:05:26 +00:00
mmu_notifier.c mm/mmu_notifier: silence PROVE_RCU_LIST warnings 2020-03-21 18:56:06 -07:00
mmzone.c
mprotect.c mm: code cleanup for MADV_FREE 2020-04-07 10:43:38 -07:00
mremap.c mm/mremap: add MREMAP_DONTUNMAP to mremap() 2020-04-02 09:35:30 -07:00
msync.c
nommu.c x86/mm: split vmalloc_sync_all() 2020-03-21 18:56:06 -07:00
oom_kill.c mm, oom: dump stack of victim when reaping failed 2020-01-31 10:30:38 -08:00
page_alloc.c mm: adjust shuffle code to allow for future coalescing 2020-04-07 10:43:38 -07:00
page_counter.c mm, memcg: prevent memory.min load/store tearing 2020-04-02 09:35:29 -07:00
page_ext.c mm/sparse: rename pfn_present() to pfn_in_present_section() 2020-04-02 09:35:30 -07:00
page_idle.c
page_io.c fs: Enable bmap() function to properly return errors 2020-02-03 08:05:37 -05:00
page_isolation.c mm/page_isolation: fix potential warning from user 2020-01-31 10:30:39 -08:00
page_owner.c
page_poison.c
page_vma_mapped.c mm/page_vma_mapped.c: explicitly compare pfn for normal, hugetlbfs and THP page 2020-01-31 10:30:38 -08:00
page-writeback.c mm/gup/writeback: add callbacks for inaccessible pages 2020-04-02 09:35:27 -07:00
pagewalk.c x86: mm: avoid allocating struct mm_struct on the stack 2020-02-04 03:05:25 +00:00
percpu-internal.h
percpu-km.c
percpu-stats.c percpu: update copyright emails to dennis@kernel.org 2020-04-01 10:09:12 -07:00
percpu-vm.c
percpu.c percpu: update copyright emails to dennis@kernel.org 2020-04-01 10:09:12 -07:00
pgtable-generic.c asm-generic/mm: stub out p{4,u}d_clear_bad() if __PAGETABLE_P{4,U}D_FOLDED 2019-12-01 06:29:19 -08:00
process_vm_access.c mm: docs: Fix a comment in process_vm_rw_core 2020-03-25 10:04:01 -05:00
ptdump.c x86: mm: avoid allocating struct mm_struct on the stack 2020-02-04 03:05:25 +00:00
readahead.c
rmap.c mm: remove CONFIG_TRANSPARENT_HUGE_PAGECACHE 2020-04-07 10:43:38 -07:00
rodata_test.c
shmem.c mm: remove CONFIG_TRANSPARENT_HUGE_PAGECACHE 2020-04-07 10:43:38 -07:00
shuffle.c mm: adjust shuffle code to allow for future coalescing 2020-04-07 10:43:38 -07:00
shuffle.h mm: adjust shuffle code to allow for future coalescing 2020-04-07 10:43:38 -07:00
slab_common.c mm, memcg: fix build error around the usage of kmem_caches 2020-04-02 09:35:28 -07:00
slab.c mm, debug_pagealloc: don't rely on static keys too early 2020-01-13 18:19:02 -08:00
slab.h mm: kmem: rename (__)memcg_kmem_(un)charge_memcg() to __memcg_kmem_(un)charge() 2020-04-02 09:35:28 -07:00
slob.c
slub.c slub: relocate freelist pointer to middle of object 2020-04-02 09:35:26 -07:00
sparse-vmemmap.c
sparse.c mm/sparse.c: allocate memmap preferring the given node 2020-04-02 09:35:30 -07:00
swap_cgroup.c
swap_slots.c mm/swap_slots.c: assign|reset cache slot by value directly 2020-04-02 09:35:27 -07:00
swap_state.c mm/swap_state.c: use the same way to count page in [add_to|delete_from]_swap_cache 2020-04-02 09:35:28 -07:00
swap.c mm: code cleanup for MADV_FREE 2020-04-07 10:43:38 -07:00
swapfile.c mm/swapfile: fix data races in try_to_unuse() 2020-04-02 09:35:27 -07:00
truncate.c
usercopy.c
userfaultfd.c hugetlbfs: use i_mmap_rwsem for more pmd sharing synchronization 2020-04-02 09:35:32 -07:00
util.c mm/mmap.c: rb_parent is not necessary in __vma_link_list() 2019-12-01 06:29:19 -08:00
vmacache.c
vmalloc.c mm/vmalloc: fix a typo in comment 2020-04-07 10:43:38 -07:00
vmpressure.c mm: vmpressure: use mem_cgroup_is_root API 2020-04-02 09:35:31 -07:00
vmscan.c mm: code cleanup for MADV_FREE 2020-04-07 10:43:38 -07:00
vmstat.c mm, thp: track fallbacks due to failed memcg charges separately 2020-04-07 10:43:38 -07:00
workingset.c mm: vmscan: detect file thrashing at the reclaim root 2019-12-01 12:59:07 -08:00
z3fold.c mm/z3fold.c: do not include rwlock.h directly 2020-03-06 07:06:09 -06:00
zbud.c
zpool.c
zsmalloc.c mm/zsmalloc.c: fix the migrated zspage statistics. 2020-01-04 13:55:09 -08:00
zswap.c zswap: potential NULL dereference on error in init_zswap() 2020-01-31 10:30:39 -08:00