On Mon, 2005-01-10 at 19:03 -0500, Sonny Rao wrote:
> On Mon, Jan 10, 2005 at 01:39:34PM -0600, Dave Kleikamp wrote:
> > Does this patch help?
> > 
> 
> It doesn't appear to help.  It looks like pdflush is trying to
> writeback inodes and is blocking waiting for a tlock in txLockAlloc
> again, maybe this is holding everyone else from completing their
> transactions?  Also, you can clearly see one of the fstest threads in
> the diNewIAG path waiting on a tlock.

Yeah, that seems to be the trouble.  We're not doing a good enough job
of preventing tlock starvation.

I also noticed that several of the static functions called by diAlloc do
show up in this latest stack trace, so I believe I was mistaken about
the cause of the earlier deadlock.  I now think that the thread in
diAlloc was trying to grab the AG_LOCK, and never made it down into
diNewIAG.  I'm afraid there may still be two different problems that
still need to be figured out.

> Also, I'm doing this on a 2.6.8.1 kernel, would that make any difference?

I don't think so.

> I've attached the full backtrace output and txstats output.
> 
> Sonny
-- 
David Kleikamp
IBM Linux Technology Center

_______________________________________________
Jfs-discussion mailing list
[email protected]
http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jfs-discussion

Reply via email to