]> git.ipfire.org Git - thirdparty/linux.git/commit
mm/rmap: fix potential batched TLB flush race
authorHuang Ying <ying.huang@intel.com>
Fri, 14 Jan 2022 22:09:16 +0000 (14:09 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sat, 15 Jan 2022 14:30:31 +0000 (16:30 +0200)
commit5ee2fa2f063649570c702164f47a558a3432dd9e
tree7398eb899135c608aca7895b70e4a81faa023354
parent8c57c07741bf28e7d867f1200aa80120b8ca663e
mm/rmap: fix potential batched TLB flush race

In theory, the following race is possible for batched TLB flushing.

  CPU0                               CPU1
  ----                               ----
  shrink_page_list()
                                     unmap
                                       zap_pte_range()
                                         flush_tlb_batched_pending()
                                           flush_tlb_mm()
    try_to_unmap()
      set_tlb_ubc_flush_pending()
        mm->tlb_flush_batched = true
                                           mm->tlb_flush_batched = false

After the TLB is flushed on CPU1 via flush_tlb_mm() and before
mm->tlb_flush_batched is set to false, some PTE is unmapped on CPU0 and
the TLB flushing is pended.  Then the pended TLB flushing will be lost.
Although both set_tlb_ubc_flush_pending() and
flush_tlb_batched_pending() are called with PTL locked, different PTL
instances may be used.

Because the race window is really small, and the lost TLB flushing will
cause problem only if a TLB entry is inserted before the unmapping in
the race window, the race is only theoretical.  But the fix is simple
and cheap too.

Syzbot has reported this too as follows:

    ==================================================================
    BUG: KCSAN: data-race in flush_tlb_batched_pending / try_to_unmap_one

    write to 0xffff8881072cfbbc of 1 bytes by task 17406 on cpu 1:
     flush_tlb_batched_pending+0x5f/0x80 mm/rmap.c:691
     madvise_free_pte_range+0xee/0x7d0 mm/madvise.c:594
     walk_pmd_range mm/pagewalk.c:128 [inline]
     walk_pud_range mm/pagewalk.c:205 [inline]
     walk_p4d_range mm/pagewalk.c:240 [inline]
     walk_pgd_range mm/pagewalk.c:277 [inline]
     __walk_page_range+0x981/0x1160 mm/pagewalk.c:379
     walk_page_range+0x131/0x300 mm/pagewalk.c:475
     madvise_free_single_vma mm/madvise.c:734 [inline]
     madvise_dontneed_free mm/madvise.c:822 [inline]
     madvise_vma mm/madvise.c:996 [inline]
     do_madvise+0xe4a/0x1140 mm/madvise.c:1202
     __do_sys_madvise mm/madvise.c:1228 [inline]
     __se_sys_madvise mm/madvise.c:1226 [inline]
     __x64_sys_madvise+0x5d/0x70 mm/madvise.c:1226
     do_syscall_x64 arch/x86/entry/common.c:50 [inline]
     do_syscall_64+0x44/0xd0 arch/x86/entry/common.c:80
     entry_SYSCALL_64_after_hwframe+0x44/0xae

    write to 0xffff8881072cfbbc of 1 bytes by task 71 on cpu 0:
     set_tlb_ubc_flush_pending mm/rmap.c:636 [inline]
     try_to_unmap_one+0x60e/0x1220 mm/rmap.c:1515
     rmap_walk_anon+0x2fb/0x470 mm/rmap.c:2301
     try_to_unmap+0xec/0x110
     shrink_page_list+0xe91/0x2620 mm/vmscan.c:1719
     shrink_inactive_list+0x3fb/0x730 mm/vmscan.c:2394
     shrink_list mm/vmscan.c:2621 [inline]
     shrink_lruvec+0x3c9/0x710 mm/vmscan.c:2940
     shrink_node_memcgs+0x23e/0x410 mm/vmscan.c:3129
     shrink_node+0x8f6/0x1190 mm/vmscan.c:3252
     kswapd_shrink_node mm/vmscan.c:4022 [inline]
     balance_pgdat+0x702/0xd30 mm/vmscan.c:4213
     kswapd+0x200/0x340 mm/vmscan.c:4473
     kthread+0x2c7/0x2e0 kernel/kthread.c:327
     ret_from_fork+0x1f/0x30

    value changed: 0x01 -> 0x00

    Reported by Kernel Concurrency Sanitizer on:
    CPU: 0 PID: 71 Comm: kswapd0 Not tainted 5.16.0-rc1-syzkaller #0
    Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
    ==================================================================

[akpm@linux-foundation.org: tweak comments]

Link: https://lkml.kernel.org/r/20211201021104.126469-1-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Reported-by: syzbot+aa5bebed695edaccf0df@syzkaller.appspotmail.com
Cc: Nadav Amit <namit@vmware.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Will Deacon <will@kernel.org>
Cc: Yu Zhao <yuzhao@google.com>
Cc: Marco Elver <elver@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
include/linux/mm_types.h
mm/rmap.c