On Wed, Oct 20, 2010 at 11:19:27AM -0400, Christoph Hellwig wrote:
> On Wed, Oct 20, 2010 at 07:24:09PM +1100, Nick Piggin wrote:
> > Hi, I keep hitting this bug running xfstests, and managed to
> > reproduce it several times on current vanilla upstream kernel
> > too, below is one trace.
> > Nothing special about the xfstests run, except that it is run
> > on brd device for both the test and the scratch mounts. No
> > special mount options or mkfs flags. It seems pretty reproducable
> > though, triggering every time I've run the tests like this.
> > I didn't narrow down the test that causes the bug because it
> > keeps running other tests and then eventually hanging. Any hints
> > to narrow it down?
> Care to check if it goes away if you remove the special casing of
> BLKFLSBUF in brd_ioctl?
Ah yes that stupid special case is probably causing other problems
even if not this crash. I'll try your suggestions when I start some more
tests running later today.