Skip to content

L
linux
Switch branch/tag
  1. 27 Sep, 2022 12 commits
    • Peter Xu's avatar
      mm: remember young/dirty bit for page migrations · 2e346877
      Peter Xu authored 
      When page migration happens, we always ignore the young/dirty bit settings
in the old pgtable, and marking the page as old in the new page table
using either pte_mkold() or pmd_mkold(), and keeping the pte clean.

That's fine from functional-wise, but that's not friendly to page reclaim
because the moving page can be actively accessed within the procedure. 
Not to mention hardware setting the young bit can bring quite some
overhead on some systems, e.g.  x86_64 needs a few hundreds nanoseconds to
set the bit.  The same slowdown problem to dirty bits when the memory is
first written after page migration happened.

Actually we can easily remember the A/D bit configuration and recover the
information after the page is migrated.  To achieve it, define a new set
of bits in the migration swap offset field to cache the A/D bits for old
pte.  Then when removing/recovering the migration entry, we can recover
the A/D bits even if the page changed.

One thing to mention is that here we used max_swapfile_size() to detect
how many swp offset bits we have, and we'll only enable this feature if we
know the swp offset is big enough to store both the PFN value and the A/D
bits.  Otherwise the A/D bits are dropped like before.

Link: https://lkml.kernel.org/r/20220811161331.37055-6-peterx@redhat.comSigned-off-by: default avatarPeter Xu <peterx@redhat.com>
Reviewed-by: default avatar"Huang, Ying" <ying.huang@intel.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: "Kirill A . Shutemov" <kirill@shutemov.name>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nadav Amit <nadav.amit@gmail.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      2e346877
    • Peter Xu's avatar
      mm/thp: carry over dirty bit when thp splits on pmd · 0ccf7f16
      Peter Xu authored 
      Carry over the dirty bit from pmd to pte when a huge pmd splits.  It
shouldn't be a correctness issue since when pmd_dirty() we'll have the
page marked dirty anyway, however having dirty bit carried over helps the
next initial writes of split ptes on some archs like x86.

Link: https://lkml.kernel.org/r/20220811161331.37055-5-peterx@redhat.comSigned-off-by: default avatarPeter Xu <peterx@redhat.com>
Reviewed-by: default avatarHuang Ying <ying.huang@intel.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: "Kirill A . Shutemov" <kirill@shutemov.name>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nadav Amit <nadav.amit@gmail.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      0ccf7f16
    • Peter Xu's avatar
      mm/swap: add swp_offset_pfn() to fetch PFN from swap entry · 0d206b5d
      Peter Xu authored 
      We've got a bunch of special swap entries that stores PFN inside the swap
offset fields.  To fetch the PFN, normally the user just calls
swp_offset() assuming that'll be the PFN.

Add a helper swp_offset_pfn() to fetch the PFN instead, fetching only the
max possible length of a PFN on the host, meanwhile doing proper check
with MAX_PHYSMEM_BITS to make sure the swap offsets can actually store the
PFNs properly always using the BUILD_BUG_ON() in is_pfn_swap_entry().

One reason to do so is we never tried to sanitize whether swap offset can
really fit for storing PFN.  At the meantime, this patch also prepares us
with the future possibility to store more information inside the swp
offset field, so assuming "swp_offset(entry)" to be the PFN will not stand
any more very soon.

Replace many of the swp_offset() callers to use swp_offset_pfn() where
proper.  Note that many of the existing users are not candidates for the
replacement, e.g.:

  (1) When the swap entry is not a pfn swap entry at all, or,
  (2) when we wanna keep the whole swp_offset but only change the swp type.

For the latter, it can happen when fork() triggered on a write-migration
swap entry pte, we may want to only change the migration type from
write->read but keep the rest, so it's not "fetching PFN" but "changing
swap type only".  They're left aside so that when there're more
information within the swp offset they'll be carried over naturally in
those cases.

Since at it, dropping hwpoison_entry_to_pfn() because that's exactly what
the new swp_offset_pfn() is about.

Link: https://lkml.kernel.org/r/20220811161331.37055-4-peterx@redhat.comSigned-off-by: default avatarPeter Xu <peterx@redhat.com>
Reviewed-by: default avatar"Huang, Ying" <ying.huang@intel.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: "Kirill A . Shutemov" <kirill@shutemov.name>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nadav Amit <nadav.amit@gmail.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      0d206b5d
    • Peter Xu's avatar
      mm/swap: comment all the ifdef in swapops.h · eba4d770
      Peter Xu authored 
      swapops.h contains quite a few layers of ifdef, some of the "else" and
"endif" doesn't get proper comment on the macro so it's hard to follow on
what are they referring to.  Add the comments.

Link: https://lkml.kernel.org/r/20220811161331.37055-3-peterx@redhat.comSigned-off-by: default avatarPeter Xu <peterx@redhat.com>
Suggested-by: default avatarNadav Amit <nadav.amit@gmail.com>
Reviewed-by: default avatarHuang Ying <ying.huang@intel.com>
Reviewed-by: default avatarAlistair Popple <apopple@nvidia.com>
Cc: Andi Kleen <andi.kleen@intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: "Kirill A . Shutemov" <kirill@shutemov.name>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      eba4d770
    • Peter Xu's avatar
      mm/x86: use SWP_TYPE_BITS in 3-level swap macros · 9c61d532
      Peter Xu authored 
      Patch series "mm: Remember a/d bits for migration entries", v4.


Problem
=======

When migrating a page, right now we always mark the migrated page as old &
clean.

However that could lead to at least two problems:

  (1) We lost the real hot/cold information while we could have persisted.
      That information shouldn't change even if the backing page is changed
      after the migration,

  (2) There can be always extra overhead on the immediate next access to
      any migrated page, because hardware MMU needs cycles to set the young
      bit again for reads, and dirty bits for write, as long as the
      hardware MMU supports these bits.

Many of the recent upstream works showed that (2) is not something trivial
and actually very measurable.  In my test case, reading 1G chunk of memory
- jumping in page size intervals - could take 99ms just because of the
extra setting on the young bit on a generic x86_64 system, comparing to
4ms if young set.

This issue is originally reported by Andrea Arcangeli.

Solution
========

To solve this problem, this patchset tries to remember the young/dirty
bits in the migration entries and carry them over when recovering the
ptes.

We have the chance to do so because in many systems the swap offset is not
really fully used.  Migration entries use swp offset to store PFN only,
while the PFN is normally not as large as swp offset and normally smaller.
It means we do have some free bits in swp offset that we can use to store
things like A/D bits, and that's how this series tried to approach this
problem.

max_swapfile_size() is used here to detect per-arch offset length in swp
entries.  We'll automatically remember the A/D bits when we find that we
have enough swp offset field to keep both the PFN and the extra bits.

Since max_swapfile_size() can be slow, the last two patches cache the
results for it and also swap_migration_ad_supported as a whole.

Known Issues / TODOs
====================

We still haven't taught madvise() to recognize the new A/D bits in
migration entries, namely MADV_COLD/MADV_FREE.  E.g.  when MADV_COLD upon
a migration entry.  It's not clear yet on whether we should clear the A
bit, or we should just drop the entry directly.

We didn't teach idle page tracking on the new migration entries, because
it'll need larger rework on the tree on rmap pgtable walk.  However it
should make it already better because before this patchset page will be
old page after migration, so the series will fix potential false negative
of idle page tracking when pages were migrated before observing.

The other thing is migration A/D bits will not start to working for
private device swap entries.  The code is there for completeness but since
private device swap entries do not yet have fields to store A/D bits, even
if we'll persistent A/D across present pte switching to migration entry,
we'll lose it again when the migration entry converted to private device
swap entry.

Tests
=====

After the patchset applied, the immediate read access test [1] of above 1G
chunk after migration can shrink from 99ms to 4ms.  The test is done by
moving 1G pages from node 0->1->0 then read it in page size jumps.  The
test is with Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz.

Similar effect can also be measured when writting the memory the 1st time
after migration.

After applying the patchset, both initial immediate read/write after page
migrated will perform similarly like before migration happened.

Patch Layout
============

Patch 1-2:  Cleanups from either previous versions or on swapops.h macros.

Patch 3-4:  Prepare for the introduction of migration A/D bits

Patch 5:    The core patch to remember young/dirty bit in swap offsets.

Patch 6-7:  Cache relevant fields to make migration_entry_supports_ad() fast.

[1] https://github.com/xzpeter/clibs/blob/master/misc/swap-young.c


This patch (of 7):

Replace all the magic "5" with the macro.

Link: https://lkml.kernel.org/r/20220811161331.37055-1-peterx@redhat.com
Link: https://lkml.kernel.org/r/20220811161331.37055-2-peterx@redhat.comSigned-off-by: default avatarPeter Xu <peterx@redhat.com>
Reviewed-by: default avatarDavid Hildenbrand <david@redhat.com>
Reviewed-by: default avatarHuang Ying <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: "Kirill A . Shutemov" <kirill@shutemov.name>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Andi Kleen <andi.kleen@intel.com>
Cc: Nadav Amit <nadav.amit@gmail.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      9c61d532
    • Miaohe Lin's avatar
      mm, hwpoison: cleanup some obsolete comments · 9cf28191
      Miaohe Lin authored 
      1.Remove meaningless comment in kill_proc(). That doesn't tell anything.
2.Fix the wrong function name get_hwpoison_unless_zero(). It should be
get_page_unless_zero().
3.The gate keeper for free hwpoison page has moved to check_new_page().
Update the corresponding comment.

Link: https://lkml.kernel.org/r/20220830123604.25763-7-linmiaohe@huawei.comSigned-off-by: default avatarMiaohe Lin <linmiaohe@huawei.com>
Acked-by: default avatarNaoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      9cf28191
    • Miaohe Lin's avatar
      mm, hwpoison: check PageTable() explicitly in hwpoison_user_mappings() · b680dae9
      Miaohe Lin authored 
      PageTable can't be handled by memory_failure(). Filter it out explicitly in
hwpoison_user_mappings(). This will also make code more consistent with the
relevant check in unpoison_memory().

Link: https://lkml.kernel.org/r/20220830123604.25763-6-linmiaohe@huawei.comSigned-off-by: default avatarMiaohe Lin <linmiaohe@huawei.com>
Acked-by: default avatarNaoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      b680dae9
    • Miaohe Lin's avatar
      mm, hwpoison: avoid unneeded page_mapped_in_vma() overhead in collect_procs_anon() · 36537a67
      Miaohe Lin authored 
      If vma->vm_mm != t->mm, there's no need to call page_mapped_in_vma() as
add_to_kill() won't be called in this case. Move up the mm check to avoid
possible unneeded calling to page_mapped_in_vma().

Link: https://lkml.kernel.org/r/20220830123604.25763-5-linmiaohe@huawei.comSigned-off-by: default avatarMiaohe Lin <linmiaohe@huawei.com>
Acked-by: default avatarNaoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      36537a67
    • Miaohe Lin's avatar
      mm, hwpoison: use num_poisoned_pages_sub() to decrease num_poisoned_pages · 21c9e90a
      Miaohe Lin authored 
      Use num_poisoned_pages_sub() to combine multiple atomic ops into one. Also
num_poisoned_pages_dec() can be killed as there's no caller now.

Link: https://lkml.kernel.org/r/20220830123604.25763-4-linmiaohe@huawei.comSigned-off-by: default avatarMiaohe Lin <linmiaohe@huawei.com>
Acked-by: default avatarNaoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      21c9e90a
    • Miaohe Lin's avatar
      mm, hwpoison: use __PageMovable() to detect non-lru movable pages · da294991
      Miaohe Lin authored 
      It's more recommended to use __PageMovable() to detect non-lru movable
pages. We can avoid bumping page refcnt via isolate_movable_page() for
the isolated lru pages. Also if pages become PageLRU just after they're
checked but before trying to isolate them, isolate_lru_page() will be
called to do the right work.

[linmiaohe@huawei.com: fixes per Naoya Horiguchi]
  Link: https://lkml.kernel.org/r/1f7ee86e-7d28-0d8c-e0de-b7a5a94519e8@huawei.com
Link: https://lkml.kernel.org/r/20220830123604.25763-3-linmiaohe@huawei.comSigned-off-by: default avatarMiaohe Lin <linmiaohe@huawei.com>
Cc: Naoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      da294991
    • Miaohe Lin's avatar
      mm, hwpoison: use ClearPageHWPoison() in memory_failure() · 2fe62e22
      Miaohe Lin authored 
      Patch series "A few cleanup patches for memory-failure".

his series contains a few cleanup patches to use __PageMovable() to detect
non-lru movable pages, use num_poisoned_pages_sub() to reduce multiple
atomic ops overheads and so on.  More details can be found in the
respective changelogs.


This patch (of 6):

Use ClearPageHWPoison() instead of TestClearPageHWPoison() to clear page
hwpoison flags to avoid unneeded full memory barrier overhead.

Link: https://lkml.kernel.org/r/20220830123604.25763-1-linmiaohe@huawei.com
Link: https://lkml.kernel.org/r/20220830123604.25763-2-linmiaohe@huawei.comSigned-off-by: default avatarMiaohe Lin <linmiaohe@huawei.com>
Acked-by: default avatarNaoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      2fe62e22
    • Yang Shi's avatar
      mm: MADV_COLLAPSE: refetch vm_end after reacquiring mmap_lock · 4d24de94
      Yang Shi authored 
      The syzbot reported the below problem:

BUG: Bad page map in process syz-executor198  pte:8000000071c00227 pmd:74b30067
addr:0000000020563000 vm_flags:08100077 anon_vma:ffff8880547d2200 mapping:0000000000000000 index:20563
file:(null) fault:0x0 mmap:0x0 read_folio:0x0
CPU: 1 PID: 3614 Comm: syz-executor198 Not tainted 6.0.0-rc3-next-20220901-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 08/26/2022
Call Trace:
 <TASK>
 __dump_stack lib/dump_stack.c:88 [inline]
 dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106
 print_bad_pte.cold+0x2a7/0x2d0 mm/memory.c:565
 vm_normal_page+0x10c/0x2a0 mm/memory.c:636
 hpage_collapse_scan_pmd+0x729/0x1da0 mm/khugepaged.c:1199
 madvise_collapse+0x481/0x910 mm/khugepaged.c:2433
 madvise_vma_behavior+0xd0a/0x1cc0 mm/madvise.c:1062
 madvise_walk_vmas+0x1c7/0x2b0 mm/madvise.c:1236
 do_madvise.part.0+0x24a/0x340 mm/madvise.c:1415
 do_madvise mm/madvise.c:1428 [inline]
 __do_sys_madvise mm/madvise.c:1428 [inline]
 __se_sys_madvise mm/madvise.c:1426 [inline]
 __x64_sys_madvise+0x113/0x150 mm/madvise.c:1426
 do_syscall_x64 arch/x86/entry/common.c:50 [inline]
 do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
 entry_SYSCALL_64_after_hwframe+0x63/0xcd
RIP: 0033:0x7f770ba87929
Code: 28 00 00 00 75 05 48 83 c4 28 c3 e8 11 15 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f770ba18308 EFLAGS: 00000246 ORIG_RAX: 000000000000001c
RAX: ffffffffffffffda RBX: 00007f770bb0f3f8 RCX: 00007f770ba87929
RDX: 0000000000000019 RSI: 0000000000600003 RDI: 0000000020000000
RBP: 00007f770bb0f3f0 R08: 00007f770ba18700 R09: 0000000000000000
R10: 00007f770ba18700 R11: 0000000000000246 R12: 00007f770bb0f3fc
R13: 00007ffc2d8b62ef R14: 00007f770ba18400 R15: 0000000000022000

Basically the test program does the below conceptually:
1. mmap 0x2000000 - 0x21000000 as anonymous region
2. mmap io_uring SQ stuff at 0x20563000 with MAP_FIXED, io_uring_mmap()
   actually remaps the pages with special PTEs
3. call MADV_COLLAPSE for 0x20000000 - 0x21000000

It actually triggered the below race:

             CPU A                                          CPU B
mmap 0x20000000 - 0x21000000 as anon
                                           madvise_collapse is called on this area
                                             Retrieve start and end address from the vma (NEVER updated later!)
                                             Collapsed the first 2M area and dropped mmap_lock
Acquire mmap_lock
mmap io_uring file at 0x20563000
Release mmap_lock
                                             Reacquire mmap_lock
                                             revalidate vma pass since 0x20200000 + 0x200000 > 0x20563000
                                             scan the next 2M (0x20200000 - 0x20400000), but due to whatever reason it didn't release mmap_lock
                                             scan the 3rd 2M area (start from 0x20400000)
                                               get into the vma created by io_uring

The hend should be updated after MADV_COLLAPSE reacquire mmap_lock since
the vma may be shrunk.  We don't have to worry about shink from the other
direction since it could be caught by hugepage_vma_revalidate().  Either
no valid vma is found or the vma doesn't fit anymore.

Link: https://lkml.kernel.org/r/20220914162220.787703-1-shy828301@gmail.com
Fixes: 7d8faaf1 ("mm/madvise: introduce MADV_COLLAPSE sync hugepage collapse")
Reported-by: syzbot+915f3e317adb0e85835f@syzkaller.appspotmail.com
Signed-off-by: default avatarYang Shi <shy828301@gmail.com>
Reviewed-by: default avatarZach O'Keefe <zokeefe@google.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
      4d24de94
  3. 26 Sep, 2022 12 commits
  5. 12 Sep, 2022 16 commits
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7