KisLayerTest in hellgrind

[Boudewijn Rempt]

On Wednesday 01 December 2010, Dmitry Kazakov wrote:
> On Wed, Dec 1, 2010 at 9:30 PM, Boudewijn Rempt <boud at valdyas.org> wrote:
> 
> > On Wednesday 01 December 2010, Dmitry Kazakov wrote:
> > > > We have random failure of the KisLayerTest, and I am wondering if it is
> > not
> > > > associated with some of the random crash we have in krita. Following
> > Mark's
> > > > suggestion on IRC I ran helgrind on KisLayerTest (running it on krita
> > is
> > > > impossible), and I get the following errors. I hope someone (ie Dmitry
> > :D)
> > > > can
> > > > figure out something out of it.
> > > >
> > >
> > > Hm.. i will try to reproduce this crash.
> >
> > It's more than a crash: helgrind clearly shows we have some issues in the
> > threading code for the projection calculation. Whether or not you can
> > actually reproduce the crash isn't that much of an issue, as long as we can
> > make the layer test run cleanly under helgrind :-)
> >
> 
> The problem is, hellgring output is almost useless here, as tiles engine
> uses a lockless approach for the data access. But the hellgrind is intended
> to work with the structures protected with mutexes and etc, so it'll think
> our code a complete solid issue ;). So it doesn't proof much. There is a
> couple of interesting lines, though.

The helgrind manual suggests that we can add some directives to exclude that code from the helgrind checking, though I'm not going to claim I've figured it out yet.

> 
> >  > But it'll happen in the beginning
> > > of the next week. I have to do some stuff for the university right now.
> >
> > I think that means that we should wait with the RC until you've had time to
> > look at the issues. To me, this is clearly a release blocking issue.
> >
> 
> If it is reproducible (esp. on different systems), this is surely an issue
> =) Just don't rely on hellgrind output much.

I get the same output.
-- 
Boudewijn Rempt | http://www.valdyas.org
Ceterum censeo lapsum particulorum probae delendum esse