xf86-video-ati 6.13.0 X.Org Driver Released

[Melcar]

6.13 just hit the xorg-edgers ppa. Haven't used it much, but my laptop no longer freezes when using KMS power management.

[wswartzendruber]

HDMI audio works on my 4350! Amazing.

Now I don't suppose anyone's gotten 5.1 LPCM to work.

[yotambien]

Hi.

Does anyone know if this release fixes random lockups that have been observed on several machines? It looked like a difficult issue to solve (as I read in Gentoo forums). I have a Mobility 3670 and have had to disable KMS to prevent lockups. I still have 3D acceleration though.

Apparently not. I have random lockups with this driver since I tried it from Debian's experimental repository (6.12.192-1 I think). I get the same with 6.13, KMS or not. I had to revert back to 6.12-something.

[bridgman]

Remember that there are probably 50 different root causes for lockups; some specific to individual systems, some specific to more common things like GPU or bus, and some general ones. In general the developers can only fix problems which show up on *their* systems, although sometimes they notice things that "don't look right" in the code.

I imagine 6.13 fixes a bunch of these issues but not all of them. It's probably not a "yes or no" answer.

[yotambien]

Remember that there are probably 50 different root causes for lockups; some specific to individual systems, some specific to more common things like GPU or bus, and some general ones. In general the developers can only fix problems which show up on *their* systems, although sometimes they notice things that "don't look right" in the code.

I imagine 6.13 fixes a bunch of these issues but not all of them. It's probably not a "yes or no" answer.

I get these problems with 6.13 (6.12.19* in Debian was pulled from the 6.13 branch or whatever), so it's not that it fixes the lockups: it introduces them.

I understand it must be hard to debug these issues. For starters, I wouldn't even know how to send a meaningful bug report. The screen randomly locks up after some minutes of use and I have to reboot it the hard way. CPU doesn't seem to rev up if the fan noise is to be trusted. I haven't tried to SSH into the system, but how to get a backtrace if the whole thing is messed up? And, well, a backtrace of what? If somebody has a suggestion to retrieve the pertinent information I can try it, but at the moment I definitely don't have time to git-bisect anything.

[bridgman]

Ahh, so it's a regression. Sorry, I missed that first time.

If changing just the X driver makes the difference between lockups and no lockups I think that's worth getting into a bug report. I took a quick look through xorg->driver/radeon and didn't see an obvious bug ticket to piggyback on, so probably starting a new bug report would be OK. The main thing to make clear is that the problems don't happen on an earlier version (specify) but do happen on the most recent code (specify).

What would be hugely useful would be if you could bisect (ie binary search) between the working and non-working versions to find which commit made it stop working. I know that's not easy with intermittents but even a couple of tests might be able to narrow the problem down to a month rather than a year (or whatever) of changes.

[VinzC]

Apparently not. I have random lockups with this driver since I tried it from Debian's experimental repository (6.12.192-1 I think). I get the same with 6.13, KMS or not. I had to revert back to 6.12-something.

Doesn't sound good. I guess the only way is to try since it happens randomly and might be solved here and there.

[yotambien]

Doesn't sound good. I guess the only way is to try since it happens randomly and might be solved here and there.

Do try it. As bridgman said, it can be caused by anything and you may not experience any problems. Actually, it's only the two of us in this thread who mentioned this issue, so apparently not many people have troubles with the driver.

What would be hugely useful would be if you could bisect [...]

It probably would, but I guess you also missed the part where I mentioned that I've got no time to bisect anything at the moment : )

[bridgman]

...but I guess you also missed the part where I mentioned that I've got no time to bisect anything at the moment : )

Yes, apparently I did. I was quite surprised when I went back and looked at your previous post