Hi Paul,
On Thu, 16 Dec 2010 17:08:14 +1100 Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote:
>
> On Wed, 15 Dec 2010 21:29:58 -0800 "Paul E. McKenney" <paulmck@xxxxxxxxxxxxxxxxxx> wrote:
> >
> > RCU problems would normally take longer to run the system out of memory,
> > but who knows?
> >
> > I did a push into -rcu in the suspect time frame, so have pulled it. I am
> > sure that kernel.org will push this change to its mirrors at some point.
> > Just in case tree-by-tree bisecting is faster than commit-by-commit
> > bisecting.
>
> I have bisected it down to the rcu tree, so the three commits that were
> added yesterday are the suspects. I am still bisecting. If will just
> revert those three commits from linux-next today in the hope that Andrew
> will end up with a working tree.
Bisect finished:
4e40200dab0e673b019979b5b8f5e5d1b25885c2 is first bad commit
commit 4e40200dab0e673b019979b5b8f5e5d1b25885c2
Author: Paul E. McKenney <paulmck@xxxxxxxxxxxxxxxxxx>
Date: Fri Dec 10 15:02:47 2010 -0800
rcu: fine-tune grace-period begin/end checks
Use the CPU's bit in rnp->qsmask to determine whether or not the CPU
should try to report a quiescent state. Handle overflow in the check
for rdp->gpnum having fallen behind.
Signed-off-by: Paul E. McKenney <paulmck@xxxxxxxxxxxxxxxxxx>
So far 4 of my 6 boot tests that failed yesterday have succeeded today
(with those last three rcu commits reverted) - the others are still
building.
--
Cheers,
Stephen Rothwell sfr@xxxxxxxxxxxxxxxx
http://www.canb.auug.org.au/~sfr/
Attachment:
pgpjJ2eq63jpr.pgp
Description: PGP signature