Mesa 7.6 Planned For Release Next Week

[whizse]

I was wondering about that too. It wasn't mentioned in the article, but there is clearly R600 code in the 7.6 branch. However, I don't know what the state of that code is or how well it gels with what's happening in kernel land.

"I guess the question is that the current r600 code in the 7.6 branch
is much less stable than the code in master. But for 7.6 I guess we
can just say r600 isn't stable yet and not built it by default"

From: http://www.nabble.com/Re%3A-More-on-...p25478675.html

[bridgman]

I'm not sure whether Alex meant stable in terms of "it's not changing every day" or "it's not reliable". Most of the changes since the 7.6 branch seem to be "improvements" rather than "bad bug fixes"... at first glance it seems like there are only a few fixes we would need (primarily the ones before Sep 12) to make 6xx/7xx support in 7.6 useful.

The r600 driver would still need to be disabled in any distro release which automatically enabled Compiz if a suitable driver was found, because of the text corruption issue. In those cases users would need a post-release Mesa update (once the text corruption bug is found and fixed) to enable 3D hardware acceleration on 6xx/7xx.

[DeepDayze]

Will the new Mesa support the R100 chips? I have an IBM T42 that has an ATI Radeon Mobility 7500 graphics chip.

[nanonyme]

In those cases users would need a post-release Mesa update (once the text corruption bug is found and fixed) to enable 3D hardware acceleration on 6xx/7xx.

Oh, DRI1 is still having it? It's not in DRI2, me thinks.
Edit: I've been reading that comment many many times and it seems I read it wrong. Never mind.

[bridgman]

Yep, TFP seems to work in DRI2 over GEM/TTM; it's just the DRI1 path that shows corruption.

Then again the idea of having a common memory manager was to make things like TextureFromPixmap simpler and more efficient, so I guess it's not a surprise

[nanonyme]

Yep, TFP seems to work in DRI2; it's just the DRI1 path that has problems.

Including which TFP requires indirect code path in DRI1. >.<

[bridgman]

BTW Alex has pushed the low risk r6xx/7xx fixes from master to the 7.6 branch, so 6xx/7xx support in 7.6 should be roughly the same as master except for the work-in-process draw_prims changes pushed in the last few days.

[russell_h]

BTW Alex has pushed the low risk r6xx/7xx fixes from master to the 7.6 branch, so 6xx/7xx support in 7.6 should be roughly the same as master except for the work-in-process draw_prims changes pushed in the last few days.

Is the KMS support for 6xx/7xx going to be in 7.6?

[bridgman]

I believe it should work for both KMS/DRI2 and non-KMS/DRI1 cases, as long as appropriate kernel/drm code is used.

[whizse]

Fuck off Ian!!!

Pissing off graphics driver developers doesn't seem like the smart thing to do. Your games might suddenly start to show beastiality porn at the most awkward of times...