Bug#513537: linux-image-2.6.26-1-openvz-amd64 hanging

[Tom Rathborne <tomr@tinyplanet.ca>]

Hi Vitaliy,

On Tue, Mar 10, 2009 at 12:58:48PM +0300, Vitaliy Gusev wrote:
> On Saturday 07 March 2009 12:13:14 am Tom Rathborne wrote:
> > Hi Vitaliy,
> >
> > On Fri, Mar 06, 2009 at 10:04:13PM +0300, Vitaliy Gusev wrote:
> > > Tom wrote:
> > > > Is there anything else I can do to help you debug this?
> > >
> > > Warn message reffers to the shmem_free_blocks():
> > >
> > >      static void shmem_free_blocks(struct inode *inode, long pages)
> > >      {
> > > 	struct shmem_sb_info *sbinfo = SHMEM_SB(inode->i_sb);
> > > 	if (sbinfo->max_blocks) {
> > > 		spin_lock(&sbinfo->stat_lock);
> > >                ^^^^
> > >               Here is a place where cpu1 is in a loop .
> > >
> > > It seems like someone already held &sbinfo->stat_lock.
> >
> > Ok, that makes sense.
> >
> > > Can you do "sysrq-p" , "sysrq-t", "sysrq-w"  in the host ?
> >
> > Done, attached!
> >
> > I have not rebooted the machine yet, in case you need some other
> > kernel debug info. Let me know!
> 
> Hmm,  i saw only trace for cpu 3, cpu 0 and cpu 1. Where is a trace for cpu2 ?

I re-checked the log. There was no trace for cpu2!

> Also it will be nice if you compile your kernel with spin lock debug
> and do sysrq-d.

Ok ... I will have to learn the Debian Way to recompile my kernel!
I will get back to you with a spin lock trace from sysrq-d.

Thanks,

Tom

-- 
-- Tom Rathborne
<Bootsy> Tommer needs a revision that takes care of itself.