Re: SMP VFS Part 2

On Thu, Dec 02, 2004 at 02:20:46AM -0500, Jeff Roberson wrote:
> On Thu, 2 Dec 2004, Peter Holm wrote:
> 
> > On Wed, Dec 01, 2004 at 06:52:19PM -0500, Jeff Roberson wrote:
> > > On Wed, 1 Dec 2004, Jeff Roberson wrote:
> > >
> > > > On Wed, 1 Dec 2004, Peter Holm wrote:
> > > >
> > > > >
> > > > > Got a new one for you:
> > > > > 1) panic: wrong b_bufobj 0 should be 0xc2679c10
> > > > > 2) panic: unexpected clean buffer 0xc6624bc8
> > > > >
> > > > > http://www.holm.cc/stress/log/jeff04.html
> > > >
> > > > I fixed your last panic.   The new patch is up at the same URL.  I'll look
> > > > at these now, thanks.  This tool is really working out well.
> > >
> > > Also fixed the unexpected clean buffer.  That was mostly just a bad
> > > assert.  I'm still baffled by wrong bufobj.
> > >
> >
> > OK. I just do not seem to be able to get past the double panic problem:
> >
> > $ grep -A 3 ppanic /tmp/tip11.log
> > ppanic: unexpected clean buffer 0xc6624bc8
> > cpuid = 0
> > KDB: enter: panic
> > [thread pid 54 tid 100042 ]
> > --
> > ppanic: unexpected clean buffer 0xc6667ec0
> > cpuid = 0
> > KDB: enter: panic
> > [thread pid 54 tid 100042 ]
> > --
> > ppanic: unexpected clean buffer 0xc66564e0
> > cpuid = 0
> > KDB: enter: panic
> > [thread pid 54 tid 100042 ]
> >
> > so I've just updated with your last patch.
> 
> Are you just running your stress program?  Anything special about your
> setup?  I see that these are valid bugs, but I'm not able to trigger them
> myself.
> 
> >
> > - Peter

After 05:46:18 of stress testing I get a "wrong b_bufobj 0 should be 0xc2e99484"
panic. There's not much to go on, as the panic doesn't enter the debugger.
Details at http://www.holm.cc/stress/log/jeff05.html

-- 
Peter Holm