[lkml]   [2010]   [Aug]   [27]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [PATCH] mm: fix hang on anon_vma->root->lock
On Fri, 27 Aug 2010, Hugh Dickins wrote:

> >> I do not see a second check (*after* taking the lock) in the patch
> if (page_mapped(page))
> return anon_vma;

As far as I can tell you would have to recheck the mapping pointer and the
pointer to the root too after taking the lock because only taking the lock
stabilitzes the object. Any other data you may have obtained before
acquiring the lock may have changed.

> >> and the way the lock is taken can be a problem in itself.
> No, that's what we rely upon SLAB_DESTROY_BY_RCU for.

SLAB_DESTROY_BY_RCU does not guarantee that the object stays the same nor
does it prevent any fields from changing. Going through a pointer with
only SLAB_DESTROY_BY_RCU means that you can only rely on the atomicity
guarantee for pointer updates. You get a valid pointer but pointer changes
are not prevented by SLAB_DESTROY_BY_RCU.

The only guarantee of that would be through other synchronization
techniques. If you believe that the page lock provides sufficient
synchronization that then this approach may be ok.

 \ /
  Last update: 2010-08-28 01:09    [W:0.079 / U:13.956 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site