Am 30.10.2013 um 12:31 hat Fam Zheng geschrieben:
> On Wed, 10/30 12:15, Kevin Wolf wrote:
> > Am 30.10.2013 um 10:42 hat Fam Zheng geschrieben:
> > > Case 030 occasionally fails because of block job compltes too fast to be
> > > captured by script, and 'unexpected qmp event' of job completion causes
> > > the test failure.
> > > 
> > > Simply fill in some data to the test image to make this false alarm less
> > > likely to happen.
> > > 
> > > (For other benefits to prefill data to test image, see also commit
> > > ab68cdfaa).
> > > 
> > > Signed-off-by: Fam Zheng <[email protected]>
> > > 
> > > ---
> > > v2: subsequent to "qemu-iotests: fix 030 for faster machines".
> > > 
> > > Signed-off-by: Fam Zheng <[email protected]>
> > 
> > Did you try using blkdebug to make it deterministic instead of just
> > making failure less likely?
> 
> Yes. blkdebug relies on actual IO, but using images with no data doesn't
> trigger any IO and just completes. Prefilling is the prerequisite to use that.
> This alone already works very well for me, but blkdebug is good to have, too.

That's actually a good point.

> Suspending before starting the block job is obviously the way to do, but
> resuming before cancel_and_wait is not deterministic, either, in the same way.
> We'll need to add cancel_and_resume_and_wait().
> 
> Do you like me to respin or use a follow up patch?

A follow-up patch works for me.

Kevin

Reply via email to