Re: [Bugme-new] [Bug 33502] New: Caught 64-bit read from uninitialized memory in __alloc_skb

[Vegard Nossum <vegardno@ifi.uio.no>]

 On Tue, 10 May 2011 13:19:35 +0300, Pekka Enberg 
 <penberg@cs.helsinki.fi> wrote:
> On 5/10/11 1:17 PM, Eric Dumazet wrote:
>> Le mardi 10 mai 2011 à 13:03 +0300, Pekka Enberg a écrit :
>>
>>> Can't we fix the issue by putting kmemcheck_mark_initialized() to
>>> set_freepointer()?
>>
>> This would solve kmemcheck problem, not DEBUG_PAGEALLOC
>
> Oh, right. Christoph? We need to support DEBUG_PAGEALLOC with SLUB.

 Hi,

 Disclaimer: I don't know how the algorithm is supposed to work, so 
 please excuse me if this makes no sense at all. But here goes:

 Presumably the problem is that the page can get freed, and that with 
 DEBUG_PAGEALLOC, the page will therefore not be present and subsequently 
 trigger a page fault when doing this cmpxchg() on the possibly freed 
 object.

 Regardless of DEBUG_PAGEALLOC or kmemcheck, what happens if the page 
 gets freed, then allocated again for a completely different purpose in 
 another part of the kernel, and new user of the page by chance writes 
 the same "tid" number that the cmpxchg() is expecting?

 I guess I'm asking: Are we relying on the fact that this has a 
 negligible probability of happening? Or did I just misunderstand what 
 the algorithm does, and it is in fact provable (in theory) that nothing 
 wrong can happen?

 Thanks,


 Vegard