[Openchrome-devel] New openchrome driver snapshot

[James Simmons]

> > 	Sorry I have been away due to health problems but I'm in
> > better shape now. 
> 
> Good to have you back! Been missing you ;)

Thanks.

> > Since the last snap shot the cursor image issues
> > have been fixed. Also the EXA layer has been reworked. The rework has
> > helped to reduce the EXA corruption people are seeing. Still some
> > corrupt but its better now, especially with firefox. 
> 
> I think the EXA corruption is a bug in EXA.
> 
> I reported this a while ago here for the openchrome driver:
> http://wiki.openchrome.org/pipermail/openchrome-devel/2011-September/000573.html
> 
> The exact same thing happens with the current 0.2.906 that enabled EXA
> by default. (I just discovered that setting "MigrationHeuristics"
> to "greedy" solves it for both 0.2.906 and 0.2.999-pre20120606)

This helped alot. Doing some digging showed this to be a pretty common 
problem. I checked in a patch to always set MigrationHeuristics to greedy.
 
> There was a recent cairo change that exposed a bug in EXA that I
> suspect is related:
> https://bugs.freedesktop.org/show_bug.cgi?id=47266
> The v2 patch there does not solve it on openchrome.

:-( Well at some point the EXA has to be reworked.

> > The EXA layer
> > has has been reworked for the upcoming H5/H6 dma packet support which
> > I'm working on. I hope that will resolve alot of the corrupt issues
> > plus give a performance boost on the VX900/VX800 series hardware. The
> > latest snapshot can be downloaded at
> > 
> > http://www.infradead.org/~jsimmons/xf86-video-openchrome-0.2.999-pre20120606.tbz
> 
> Running it on my HP mini 2130 now. Appears to be usable! I'm gonna
> switch to this driver now.

Cool. I'm going to post a new update with the EXA fix and also a CLE266 
fix as well.

> What is the status for the kernel driver now? Would it be an idea to
> start think of adding it to "staging"?

Bene ignored for the xorg driver tho I have been keeping it up to date.
After the release I really like to get the LVDS to finally work. That and 
I would need to add proper fencing support to the TTM layer. Only one more 
bug to fix for the kms_branch before the big push to have it release.