Re: WAPBL patch for testing

[Juergen Hannken-Illjes <hannken%eis.cs.tu-bs.de@localhost>]

On Sat, Nov 01, 2008 at 09:09:05AM +0100, Joerg Sonnenberger wrote:
> On Sat, Nov 01, 2008 at 07:27:30AM +0100, Juergen Hannken-Illjes wrote:
> > This would bring back a deadlock (see Rev. 1.12 of genfs_io.c).  We have
> > to take the wapbl LOCK before we lock pages.
> 
> How should this deadlock happen? It has been proven that the changes
> *did* introduce a deadlock and one that is very easy to trigger.

I have seen this deadlock on a dual-cpu machine while running three
parallel builds.

One thread has a wapbl transaction, calls VOP_GETPAGES and wants
to busy a page  while  another thread calls VOP_PUTPAGES on the
same vnode, takes pages busy and wants to start a wapbl transaction.

-- 
Juergen Hannken-Illjes - hannken%eis.cs.tu-bs.de@localhost - TU Braunschweig 
(Germany)