Re: [PROBLEM] 2.6.22-rc2 panics on x86-64 with slub

[Jens Axboe <jens.axboe@oracle.com>]

On Mon, May 21 2007, Christoph Lameter wrote:
> On Mon, 21 May 2007, Hugh Dickins wrote:
> 
> > On Mon, 21 May 2007, Christoph Lameter wrote:
> > > On Mon, 21 May 2007, Hugh Dickins wrote:
> > > 
> > > > Yes, sounded the same to me too: I couldn't reproduce it or see anything
> > > > wrong in the code back then.  But Srihari's info about CONFIG_DEBUG_SLUB
> > > > off has helped a lot: I was then able to reproduce it on my x86_64, and
> > > > after a lot of staring at the code, the problem became obvious...
> > > 
> > > Right. The #ifdef CONFIG_SLUB_DEBUG is at the wrong location. The best fix
> > > is to moving the #ifdef otherwise the size is still wrong for the 
> > > ctor case.
> > 
> > ? My patch did handle the ctor case.
> 
> True. I was thinking about just checking the problem case that we had 
> here.
> 
> > > SLUB Debug: Fix object size calculation
> > > 
> > > The object size calculation is wrong if !CONFIG_SLUB_DEBUG because
> > > the #ifdef CONFIG_SLUB_DEBUG is now switching off the size adjustments
> > > for DESTROY_BY_RCU and ctor.
> > > 
> > > Signed-off-by: Christoph Lameter <clameter@sgi.com>
> > 
> > Yes, I think that should do it too.  The reason behind my repeating
> > the block was to handle the case where SLAB_POISON is passed to
> > kmem_cache_create, but CONFIG_SLUB_DEBUG is off.  But apparently
> > that case would hit the BUG_ON(flags & ~CREATE_MASK), therefore
> > your patch is simpler and better.  Quite a maze.
> 
> Would you ack my patch? I do not want to repeat the block.

I can test whatever you want tomorrow morning, it was 100% repeatable
here. So which one, your patch or Hughs?

-- 
Jens Axboe