Re: BUG: Bad page map in process udevd (anon_vma: (null)) in 2.6.38-rc4

[Linus Torvalds <torvalds@linux-foundation.org>]

On Thu, Feb 17, 2011 at 7:16 PM, Eric W. Biederman
<ebiederm@xmission.com> wrote:
>
> Interesting.  I just got this with DEBUG_PAGEALLOC
> It looks like something in DEBUG_PAGEALLOC is interfering with taking a
> successful crashdump.

Hmm. I don't see why, but we don't care. Just the IP and the Code:
section is plenty good enough.

> BUG: unable to handle kernel paging request at ffff8801adf8d760
> IP: [<ffffffff8140c7ca>] unregister_netdevice_queue+0x3a/0xb0

Yup. That's the "list_move()". The disassembly is exactly what I'd
expect from __list_del():

  16:	48 8b 93 a0 00 00 00 	mov    0xa0(%rbx),%rdx
  1d:	48 8b 83 a8 00 00 00 	mov    0xa8(%rbx),%rax
  24:	48 8d bb a0 00 00 00 	lea    0xa0(%rbx),%rdi
  2b:*	48 89 42 08          	mov    %rax,0x8(%rdx)     <-- trapping instruction
  2f:	48 89 10             	mov    %rdx,(%rax)

So I think we can consider this confirmed: it really is the stale
queue left over on the stack (introduced by commit 443457242beb). With
CONFIG_DEBUG_PAGEALLOC, you get a page fault when it tries to update
the now stale pointers.

The patch from Eric Dumazet (which adds a few more cases to my patch
and hopefully catches them all) almost certainly fixes this rather
nasty memory corruption.

                           Linus

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>