Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1369912 > unrolled thread

mm: BUG in khugepaged_scan_mm_slot

Back to article view | Back to linux.kernel

Contents

  mm: BUG in khugepaged_scan_mm_slot Dmitry Vyukov <dvyukov@google.com> - 2016-04-02 11:50 +0200
    Re: mm: BUG in khugepaged_scan_mm_slot "Kirill A. Shutemov" <kirill@shutemov.name> - 2016-04-04 14:10 +0200
      Re: mm: BUG in khugepaged_scan_mm_slot Andrea Arcangeli <aarcange@redhat.com> - 2016-04-04 19:20 +0200
    Re: mm: BUG in khugepaged_scan_mm_slot Vlastimil Babka <vbabka@suse.cz> - 2016-04-04 14:10 +0200

#1369912 — mm: BUG in khugepaged_scan_mm_slot

Hello,

The following program triggers a BUG in khugepaged_scan_mm_slot:


vma ffff880032698f90 start 0000000020c57000 end 0000000020c58000
next ffff88003269a1b8 prev ffff88003269ac18 mm ffff88005e274780
prot 35 anon_vma ffff88003182c000 vm_ops           (null)
pgoff fed00 file ffff8800324552c0 private_data           (null)
flags: 0x5144477(read|write|exec|mayread|maywrite|mayexec|pfnmap|io|dontexpand|account)
------------[ cut here ]------------
kernel BUG at mm/huge_memory.c:2313!
invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN
Modules linked in:
CPU: 2 PID: 1180 Comm: khugepaged Not tainted 4.5.0-rc7+ #337
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
task: ffff88003d910000 ti: ffff88003da70000 task.ti: ffff88003da70000
RIP: 0010:[<ffffffff8178bd07>]  [<ffffffff8178bd07>]
hugepage_vma_check+0x117/0x150
RSP: 0018:ffff88003da77bb0  EFLAGS: 00010286
RAX: 0000000000000001 RBX: ffff880032698f90 RCX: 0000000000000000
RDX: 0000000000000001 RSI: ffff88006d616d18 RDI: ffffed0007b4ef4c
RBP: ffff88003da77bc8 R08: 0000000000000001 R09: 0000000000000000
R10: 1ffff100064d31f2 R11: 0000000000000001 R12: 0000000000000001
R13: ffff880032698fe0 R14: 0000000000000806 R15: ffff880032698f90
FS:  0000000000000000(0000) GS:ffff88006d600000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 00007fe965ad9e78 CR3: 0000000007ae9000 CR4: 00000000000006e0
Stack:
 0000000020c00000 0000000000000000 ffffffff88937638 ffff88003da77e00
 ffffffff81790b33 0000000000000082 ffffffff87fc1580 0000000000000004
 ffff88003da77c38 ffff88003d910810 0000000000000000 ffff88003d910000
Call Trace:
 [<     inline     >] khugepaged_scan_mm_slot mm/huge_memory.c:2651
 [<     inline     >] khugepaged_do_scan mm/huge_memory.c:2755
 [<ffffffff81790b33>] khugepaged+0x993/0x48e0 mm/huge_memory.c:2790
 [<ffffffff813c195f>] kthread+0x23f/0x2d0 drivers/block/aoe/aoecmd.c:1303
 [<ffffffff866d1b2f>] ret_from_fork+0x3f/0x70 arch/x86/entry/entry_64.S:468
Code: 00 fc ff df 48 c1 ea 03 80 3c 02 00 75 2c 48 f7 43 50 88 44 44
10 41 bc 01 00 00 00 74 b4 e8 71 cf de ff 48 89 df e8 99 e5 f5 ff <0f>
0b 4c 89 ef e8 ff c0 fe ff e9 0a ff ff ff 4c 89 ef e8 f2 c0
RIP  [<ffffffff8178bd07>] hugepage_vma_check+0x117/0x150 mm/huge_memory.c:2313
 RSP <ffff88003da77bb0>
---[ end trace 61cae986a344948b ]---


The process itself hangs dead.


// autogenerated by syzkaller (http://github.com/google/syzkaller)
#include <pthread.h>
#include <stdint.h>
#include <string.h>
#include <sys/syscall.h>
#include <unistd.h>

#ifndef SYS_userfaultfd
#define SYS_userfaultfd 323
#endif

long r[29];

void* thr(void* arg)
{
  switch ((long)arg) {
  case 0:
    r[0] = syscall(SYS_mmap, 0x20000000ul, 0xc59000ul, 0x3ul, 0x32ul,
                   0xfffffffffffffffful, 0x0ul);
    break;
  case 1:
    r[1] = syscall(SYS_accept, 0x1869ful, 0x20c51ffful, 0x20c51000ul, 0,
                   0, 0);
    break;
  case 2:
    r[2] = syscall(SYS_fcntl, r[1], 0x9ul, 0, 0, 0, 0);
    break;
  case 3:
    r[3] = syscall(SYS_ioprio_set, 0x1ul, r[2], 0xfffffffffffffffful, 0,
                   0, 0);
    break;
  case 4:
    r[4] = syscall(SYS_userfaultfd, 0x0ul, 0, 0, 0, 0, 0);
    break;
  case 5:
    *(uint64_t*)0x20a49fe8 = (uint64_t)0xaa;
    *(uint64_t*)0x20a49ff0 = (uint64_t)0x0;
    *(uint64_t*)0x20a49ff8 = (uint64_t)0x0;
    r[8] =
        syscall(SYS_ioctl, r[4], 0xc018aa3ful, 0x20a49fe8ul, 0, 0, 0);
    break;
  case 6:
    *(uint64_t*)0x20c4c000 = (uint64_t)0x200cb000;
    *(uint64_t*)0x20c4c008 = (uint64_t)0x800000;
    *(uint64_t*)0x20c4c010 = (uint64_t)0x1;
    *(uint64_t*)0x20c4c018 = (uint64_t)0x0;
    r[13] =
        syscall(SYS_ioctl, r[4], 0xc020aa00ul, 0x20c4c000ul, 0, 0, 0);
    break;
  case 7:
    r[14] = syscall(SYS_readahead, 0xfffffffffffffffful, 0x40ul, 0x6ul,
                    0, 0, 0);
    break;
  case 8:
    r[15] = syscall(SYS_sched_getaffinity, 0x0ul, 0x8ul, 0x20472000ul,
                    0, 0, 0);
    break;
  case 9:
    r[16] = syscall(SYS_prctl, 0xful, 0x205d5ff8ul, 0, 0, 0, 0);
    break;
  case 10:
    r[18] =
        syscall(SYS_open, "/dev/hpet", 0x40ul, 0, 0, 0);
    break;
  case 11:
    *(uint32_t*)0x20c56000 = (uint32_t)0x7fff;
    r[20] = syscall(SYS_ioctl, r[18], 0x5420ul, 0x20c56000ul, 0, 0, 0);
    break;
  case 12:
    r[21] = syscall(SYS_mmap, 0x20c57000ul, 0x1000ul, 0x7ul, 0x812ul,
                    r[18], 0x0ul);
    break;
  case 13:
    *(uint8_t*)0x20c57ffe = (uint8_t)0x6;
    *(uint8_t*)0x20c57fff = (uint8_t)0x100000001;
    r[24] = syscall(SYS_ioctl, r[18], 0x541cul, 0x20c57ffeul, 0, 0, 0);
    break;
  case 14:
    memcpy((void*)0x20814ffd, "\x2e\x2f\x66\x69\x6c\x65\x30\x00", 8);
    r[26] = syscall(SYS_creat, 0x20814ffdul, 0x80ul, 0, 0, 0, 0);
    break;
  case 15:
    r[27] = syscall(SYS_pipe, 0x205a4ffful, 0, 0, 0, 0, 0);
    break;
  case 16:
    r[28] = syscall(SYS_accept4, r[18], 0x20b0112aul, 0x20c55ffeul,
                    0x80800ul, 0, 0);
    break;
  }
  return 0;
}

int main()
{
  long i;
  pthread_t th[17];

  memset(r, -1, sizeof(r));
  for (i = 0; i < 17; i++) {
    pthread_create(&th[i], 0, thr, (void*)i);
    usleep(10000);
  }
  usleep(100000);
  return 0;
}


For better reproducibility also do:

$ echo 0 > /sys/kernel/mm/transparent_hugepage/khugepaged/alloc_sleep_millisecs;
echo 0 > /sys/kernel/mm/transparent_hugepage/khugepaged/scan_sleep_millisecs


On commit 8e0f93cda48ed054e1216bab5c60017e1a5fc1e8.

[toc] | [next] | [standalone]

#1370582

On Mon, Apr 04, 2016 at 02:03:54PM +0200, Vlastimil Babka wrote:
> [+CC Andrea]
> 
> On 04/02/2016 11:48 AM, Dmitry Vyukov wrote:
> >Hello,
> >
> >The following program triggers a BUG in khugepaged_scan_mm_slot:
> >
> >
> >vma ffff880032698f90 start 0000000020c57000 end 0000000020c58000
> >next ffff88003269a1b8 prev ffff88003269ac18 mm ffff88005e274780
> >prot 35 anon_vma ffff88003182c000 vm_ops           (null)
> >pgoff fed00 file ffff8800324552c0 private_data           (null)
> >flags: 0x5144477(read|write|exec|mayread|maywrite|mayexec|pfnmap|io|dontexpand|account)
> >------------[ cut here ]------------
> >kernel BUG at mm/huge_memory.c:2313!
> >invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN
> 
> That's VM_BUG_ON_VMA(vma->vm_flags & VM_NO_THP, vma) in
> hugepage_vma_check().
> 
> #define VM_NO_THP (VM_SPECIAL | VM_HUGETLB | VM_SHARED | VM_MAYSHARE)
> 
> #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
> 
> Of those, we have VM_IO | VM_DONTEXPAND.
> 
> I don't know if it's valid for a vma with anon_vma to have such flags, if
> yes, we should probably modify hugepage_vma_check(). Called from
> khugepaged_scan_mm_slot() it should just return false out VM_NO_THP. Called
> from collapse_huge_page() it could keep the VM_BUG_ON. Or maybe just have
> VM_BUG_ON(!hugepage_vma_check()) there? Hmm actually no, there's a mmap_sem
> release for read and then acquire for write, so we can't rely on the check
> done earlier from khugepaged_scan_mm_slot().
> 
> So we should probably just change the VM_BUG_ON to another "return false"
> condition. Unless the VM_BUG_ON uncovered a real bug and the earlier
> conditions in hugepage_vma_check() should guarantee the VM_BUG_ON be false
> for any vma.

http://lkml.kernel.org/r/145961146490.28194.16019687861681349309.stgit@zurg

-- 
 Kirill A. Shutemov

[toc] | [prev] | [next] | [standalone]

#1370795

Hello,

On Mon, Apr 04, 2016 at 03:06:25PM +0300, Kirill A. Shutemov wrote:
> On Mon, Apr 04, 2016 at 02:03:54PM +0200, Vlastimil Babka wrote:
> > [+CC Andrea]
> > 
> > On 04/02/2016 11:48 AM, Dmitry Vyukov wrote:
> > >Hello,
> > >
> > >The following program triggers a BUG in khugepaged_scan_mm_slot:
> > >
> > >
> > >vma ffff880032698f90 start 0000000020c57000 end 0000000020c58000
> > >next ffff88003269a1b8 prev ffff88003269ac18 mm ffff88005e274780
> > >prot 35 anon_vma ffff88003182c000 vm_ops           (null)
> > >pgoff fed00 file ffff8800324552c0 private_data           (null)
> > >flags: 0x5144477(read|write|exec|mayread|maywrite|mayexec|pfnmap|io|dontexpand|account)
> > >------------[ cut here ]------------
> > >kernel BUG at mm/huge_memory.c:2313!
> > >invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN
> > 
> > That's VM_BUG_ON_VMA(vma->vm_flags & VM_NO_THP, vma) in
> > hugepage_vma_check().
> > 
> > #define VM_NO_THP (VM_SPECIAL | VM_HUGETLB | VM_SHARED | VM_MAYSHARE)
> > 
> > #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
> > 
> > Of those, we have VM_IO | VM_DONTEXPAND.
> > 
> > I don't know if it's valid for a vma with anon_vma to have such flags, if
> > yes, we should probably modify hugepage_vma_check(). Called from
> > khugepaged_scan_mm_slot() it should just return false out VM_NO_THP. Called
> > from collapse_huge_page() it could keep the VM_BUG_ON. Or maybe just have
> > VM_BUG_ON(!hugepage_vma_check()) there? Hmm actually no, there's a mmap_sem
> > release for read and then acquire for write, so we can't rely on the check
> > done earlier from khugepaged_scan_mm_slot().
> > 
> > So we should probably just change the VM_BUG_ON to another "return false"
> > condition. Unless the VM_BUG_ON uncovered a real bug and the earlier
> > conditions in hugepage_vma_check() should guarantee the VM_BUG_ON be false
> > for any vma.
> 
> http://lkml.kernel.org/r/145961146490.28194.16019687861681349309.stgit@zurg

That's not the only place that assumes vm_ops NULL means anonymous and
not VM_IO though, so I agree with Vlastimil we should think once more
about this fix, either that or extend it to other places.

I wonder if perhaps there was a mistake in checking vm_ops in the
first place and leaving the vm_ops check isn't the right fix. Wouldn't
it be more correct to apply a s/!vm_ops/!vm_file/ and not just there?
What problem would then we run into if we used !vm_file?

The assumption in this vm_ops check is that it was safer to a vm_file
check but clearly it isn't as some chardev is not setting vm_ops
(don't they need to vm_ops->close?). But all chardevs have vm_file
set, so if we could use that instead, we can retain the VM_BUG_ON or
better convert it to a graceful warn on that bails out.

Thanks,
Andrea

[toc] | [prev] | [next] | [standalone]

#1370583

[+CC Andrea]

On 04/02/2016 11:48 AM, Dmitry Vyukov wrote:
> Hello,
>
> The following program triggers a BUG in khugepaged_scan_mm_slot:
>
>
> vma ffff880032698f90 start 0000000020c57000 end 0000000020c58000
> next ffff88003269a1b8 prev ffff88003269ac18 mm ffff88005e274780
> prot 35 anon_vma ffff88003182c000 vm_ops           (null)
> pgoff fed00 file ffff8800324552c0 private_data           (null)
> flags: 0x5144477(read|write|exec|mayread|maywrite|mayexec|pfnmap|io|dontexpand|account)
> ------------[ cut here ]------------
> kernel BUG at mm/huge_memory.c:2313!
> invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN

That's VM_BUG_ON_VMA(vma->vm_flags & VM_NO_THP, vma) in 
hugepage_vma_check().

#define VM_NO_THP (VM_SPECIAL | VM_HUGETLB | VM_SHARED | VM_MAYSHARE)

#define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)

Of those, we have VM_IO | VM_DONTEXPAND.

I don't know if it's valid for a vma with anon_vma to have such flags, 
if yes, we should probably modify hugepage_vma_check(). Called from 
khugepaged_scan_mm_slot() it should just return false out VM_NO_THP. 
Called from collapse_huge_page() it could keep the VM_BUG_ON. Or maybe 
just have VM_BUG_ON(!hugepage_vma_check()) there? Hmm actually no, 
there's a mmap_sem release for read and then acquire for write, so we 
can't rely on the check done earlier from khugepaged_scan_mm_slot().

So we should probably just change the VM_BUG_ON to another "return 
false" condition. Unless the VM_BUG_ON uncovered a real bug and the 
earlier conditions in hugepage_vma_check() should guarantee the 
VM_BUG_ON be false for any vma.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web