Bug 780072

Summary:	radeonhd should not be used
Product:	[openSUSE] openSUSE 12.2	Reporter:	Jiri Slaby <jslaby>
Component:	X.Org	Assignee:	E-mail List <xorg-maintainer-bugs>
Status:	RESOLVED WONTFIX	QA Contact:	E-mail List <xorg-maintainer-bugs>
Severity:	Normal
Priority:	P3 - Medium	CC:	eich
Version:	Final
Target Milestone:	---
Hardware:	Other
OS:	Other
Whiteboard:
Found By:	---	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---
Attachments:	xorg.log -- radeon removed xorg.log -- radeon installed (and failing)

Description Jiri Slaby 2012-09-12 19:44:45 UTC

It looks like 12.2 likes and suggests to install radeonhd driver and X.org tries fglrx, radeonhd and radeon drivers in that order on my hardware. This is bogus as radeonhd does not play well with kernel modesetting, it does not support it at all acrually. So this breaks switching terminals for example.

The fix here would be to try the drivers in this order: fglrx, radeon, radeonhd.

Or remove radeonhd completely.

Comment 1 Stefan Dirsch 2012-09-13 08:37:29 UTC

Well, in your suggested order radeonhd would never been loaded, since KMS is our default meanwhile. With KMS enabled radeonhd bails out immediately, radeon takes over. And even in case you prefer to use UMS with radeon driver, this is no option any longer. It has been dropped from the driver recently.

Your bugreport sounds more like you have KMS disabled.

Nevertheless we can drop radeonhd driver for future products. Users then are going to lose an option for GPUs families R5xx-R7xx, in case KMS is broken.

Any comments?

Comment 2 Jiri Slaby 2012-09-13 11:36:01 UTC

(In reply to comment #1)
> Well, in your suggested order radeonhd would never been loaded, since KMS is
> our default meanwhile. With KMS enabled radeonhd bails out immediately, radeon
> takes over. And even in case you prefer to use UMS with radeon driver, this is
> no option any longer. It has been dropped from the driver recently.
> 
> Your bugreport sounds more like you have KMS disabled.

I don't know. I will check. But Xorg told me something like your kernel uses KMS and that won't work with radeonhd. But even after I resumed kernel from hibernation. I.e. after chvt 1; chvt 7. Radeonhd controlled my graphics at that time. Since I removed and blacklisted xorg radeonhd rpm, the problem went away.

> Nevertheless we can drop radeonhd driver for future products. Users then are
> going to lose an option for GPUs families R5xx-R7xx, in case KMS is broken.
> 
> Any comments?

I will provide more detailed info on what was going on.

Comment 3 Jiri Slaby 2012-09-13 11:39:04 UTC

And all this started happening after 12.1 -> 12.2 upgrade. But in 12.1 I had xorg.conf specifically mentioning driver "radeon". 12.2 seems like it ignores xorg.conf, so I removed it completely as it contained crap anyway (and I had thought I removed it already).

Comment 4 Jiri Slaby 2012-09-13 12:14:49 UTC

(In reply to comment #1)
> Well, in your suggested order radeonhd would never been loaded, since KMS is
> our default meanwhile. With KMS enabled radeonhd bails out immediately, radeon
> takes over. And even in case you prefer to use UMS with radeon driver, this is
> no option any longer. It has been dropped from the driver recently.
> 
> Your bugreport sounds more like you have KMS disabled.

Ah, you are right. Radeonhd tries to control the graphics, but fails and bails out. But it interferes somehow, because radeon then does not work after resume. I have to remove the rpm to have working graphics after resume. If I don't do that, I see only what is on the terminal one (login prompt) and mouse cursor from X. I will attach two Xorg.logs. When radeonhd is installed and when not.

Comment 5 Jiri Slaby 2012-09-13 12:15:14 UTC

Created attachment 505593 [details]
xorg.log -- radeon removed

Comment 6 Jiri Slaby 2012-09-13 12:15:51 UTC

Created attachment 505594 [details]
xorg.log -- radeon installed (and failing)

Comment 7 Stefan Dirsch 2012-09-13 12:58:33 UTC

Hmm. Ok. I don't understand that, since the KMS check is the first thing, which is done in the driver and then radeonhd driver bails out immediately ...

Comment 8 Jiri Slaby 2012-09-13 13:16:03 UTC

(In reply to comment #7)
> Hmm. Ok. I don't understand that, since the KMS check is the first thing, which
> is done in the driver and then radeonhd driver bails out immediately ...

Only an idea. What does the number in parenths in the log mean:
RADEON(0)
vs.
RADEONHD(0)
RADEON(1)

Does not RADEON mind it is a number 1 and not 0?

Comment 9 Stefan Dirsch 2012-09-13 14:29:40 UTC

I can't find "RADEON(1)" in any of your logfiles.

Comment 10 Jiri Slaby 2012-09-13 14:34:24 UTC

(In reply to comment #9)
> I can't find "RADEON(1)" in any of your logfiles.

Bah, I attached the ones which I compared with diff. I removed first 13 chars (timestamp) and did s/RADEON(1)/RADEON(0)/. Sorry for that.

Comment 11 Jiri Slaby 2013-01-09 11:57:41 UTC

I don't have the hardware anymore, so I don't care...