Re: [regression] Crash in wb_clear_pending()

[Jens Axboe <axboe@kernel.dk>]

On 2010-07-06 08:47, Ingo Molnar wrote:
> 
> * Linus Torvalds <torvalds@linux-foundation.org> wrote:
> 
>> On Mon, Jul 5, 2010 at 2:19 PM, Bill Davidsen <davidsen@tmr.com> wrote:
>>>
>>> Absent a small fix, and given that the big fix has a lot more testing than
>>> any new patch might, in this case the quickie might be undesirable.
>>> Particularly since posters here seem sure that code will be replaced in the
>>> next version anyway, and lightly tested patch to obsolete code is actually
>>> less conservative.
>>
>> I have to agree. Especially as the "big patch" just removes the fragile code 
>> that caused the problem in the first place. So in this case I do suspect 
>> that the bigger patch ends up being the safer one.
> 
> Yeah, i agree - especially since the smaller patch is still pretty large (not 
> a oneliner), plus it does not appear that the precise failure mode is fully 
> understood either.

http://lkml.org/lkml/2010/7/5/232

It's about as straight forward as it can be :-)
It definitely fixes _a_ bug, but whether it's only that bug is not certain.
As long as Linus is fine with the larger fix, then I have no issues going
in that direction.

>> But I obviously don't actually see the problem, so it would be good to get 
>> confirmation that Christoph's patch actually fixes things first. Ingo, does 
>> the one in this thread apply for you?
> 
> Yes, the three larger patches survived overnight testing with 300+ iterations 
> and i did some other tests as well, which passed too. These are the patches i 
> applied:
> 
>  a73dd720  writeback: remove writeback_inodes_wbc
>  9f98c0fa  writeback: split writeback_inodes_wb
>  79338d2a  writeback: simplify the write back thread queue

Great, I'll upstream these bits today. Thanks Ingo.

-- 
Jens Axboe