DRI locks up the machine when starting X on a Radeon 9250

The problem is still present in Dapper as of 2006-03-26. Breezy doesn't have any problem running on the same machine, and I've found that mainline 2.6.15.6 works correctly.

Applying the patch from http://www.kernel.org/git/gitweb.cgi?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=392c14beaca2ee85a98d0c6b453501be67423a20 (which reverts a previous buggy DRM change) on a rebuilt Dapper 2.6.15-19 kernel fixes the problem.

Quoting the patch comment: "It fixed some machines, but seems to continually interact badly with some X versions. [...] So I think at this point, the best is that we keep the old bogus code that at least is consistent with the bug in the server."

This patch is included in the mainline 2.6.15 kernel. Why isn't applied in Dapper's one?

Usually this ATI card has a lot of problems with Ubuntu. Xgl does not work corectly and I can get rendering only with fglrx proprietary ati drivers... I think it's ATI who produces bad low end cards.

Can we change this bug's status to confirmed?

> Usually this ATI card has a lot of problems with Ubuntu.

The card is working perfectly with many other distributions (and even different OSs), including heavy 3D games, so it isn't an hardware problem, just a driver bug.
And I can't even get the fglrx driver shipped with Dapper to work, the system just locks up in a similar way (but I'm known to hate binary drivers anyway :)).

The RV280 is one of the better supported cards under Linux using only free drivers, it's a pity it doesn't work with Dapper just because of a patch altready accepted in mainline...

I tried to install Dapper yesterday (8 April 2006), after the installation, I booting into single user mode, and applied all updates up and until that day. X still wouldn't start, so this issue still isn't resolved.

I have this card:
0000:01:00.0 VGA compatible controller: ATI Technologies Inc RV280 [Radeon 9200 PRO] (rev 01)
0000:01:00.1 Display controller: ATI Technologies Inc: Unknown device 5940 (rev 01)

So I can also confirm this issue.

I have the same issue with:
 * Dapper with a current kernel (linux-image-2.6.15-21-amd64-k8) running on 64 bit
 * Radeon 9250 with 128 MB from Powercolor
 * Asus K8V Deluxe mainboard (K8T800, amd64...)

Everything works well until I activate DRI in the module section in the xorg.conf. Maybe there is something helpful in my Xorg.0.log after activating DRI... (-> http://draugr.de/Xorg.0.log). The fix posted by ath in the second comment didn't help here... but within the next few days I will try a drm snapshot...

After modprobing drm and radeon (the official ones) dmesg says nothing about the card (it should):
[...]

[ 957.997784] [drm] Initialized drm 1.0.0 20040925
[ 960.617552] GSI 21 sharing vector 0xD1 and IRQ 21
[ 960.617628] ACPI: PCI Interrupt 0000:01:00.0[A] -> GSI 16 (level, low) -> IRQ 209
[ 960.618018] [drm] Initialized radeon 1.19.0 20050911 on minor 0:

#### lspci -vv ####
[...]

0000:01:00.0 VGA compatible controller: ATI Technologies Inc RV280 [Radeon 9200 PRO] (rev 01) (prog-if 00 [VGA])
        Subsystem: C.P. Technology Co. Ltd: Unknown device 2094
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B-
        Status: Cap+ 66MHz+ UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
        Latency: 64 (2000ns min), Cache Line Size: 0x40 (256 bytes)
        Interrupt: pin A routed to IRQ 209
        Region 0: Memory at d0000000 (32-bit, prefetchable) [size=128M]
        Region 1: I/O ports at a000 [size=256]
        Region 2: Memory at fd600000 (32-bit, non-prefetchable) [size=64K]
        Expansion ROM at fd500000 [disabled] [size=128K]
        Capabilities: <available only to root>

0000:01:00.1 Display controller: ATI Technologies Inc: Unknown device 5940 (rev 01)
        Subsystem: C.P. Technology Co. Ltd: Unknown device 2095
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
        Status: Cap+ 66MHz+ UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
        Latency: 64 (2000ns min), Cache Line Size: 0x40 (256 bytes)
        Region 0: Memory at c8000000 (32-bit, prefetchable) [size=128M]
        Region 1: Memory at fd400000 (32-bit, non-prefetchable) [size=64K]
        Capabilities: <available only to root>

I got the same bug with both Debian and Dapper... Looking through the web I got this :
http://www.spinics.net/lists/xorg/msg14673.html

I tried version 6.5.8 version of xserver-xorg-video-ati package on my Debian etch+unstable (mixed) and DRI works since a few minutes from now.

This is not really a fix for Ubuntu, but maybe this would be useful to know this fix here (as I often fall here when I look for informations about this bug).

And many thanks to Benjamin Herrenschmidt and all people who do great work for our best enjoyment ! :)

I dowloaded and installed ubuntu 5.10 on DVD and had a problem with my PCI radeon 9250 just as described here. After poking around a little I tried adding the modules DBE and V41 to the list of loaded modules. Mostly because DBE was mentioned in the context of secondary VGA cards. Well wouldn't you know that X started working with DRI.

I am not sure this will work for others but it might be worth a try

I can confirm this (Asus A7V133 and HIS Radeon 9250). gl-117 causes a hang within a few seconds, and some OpenGL screensavers as well (they shouldn't be randomly chosen on Radeons by the screensaver package, until this bug is fixed). Disabling dri removes the hangs, but 3D is very slow, about 3fps in gl-117. Setting the BusType option to PCI in xorg.conf also seems to avoid the hangs, but 3D is slow, too - about 6fps in gl-117.

It looks like the problem is still in the current X.org: https://bugs.freedesktop.org/show_bug.cgi?id=5449 - the good news is that they are still trying to fix it. If the fix enters X.org CVS after Dapper release, will it become available as a Dapper update, evn if it's not, technically speaking, a security issue?

The new radeon driver (6.5.8.0) doesn't fix my problem, probably it's a different issue. X still hangs instantly at startup (right while it's changing the video mode from the usplash one).

The only way I've found to avoid the crash (short of disabling DRI) is applying the kernel patch.

Ben, please have a look at this patch

Sorry, I think I messed things up with that link to the DRI bugzilla. There are two different issues: one, a change in kernel's drm implementation that makes X hang when dri is enabled, which is what this bug is about. I can confirm this. Now I run kernel 2.6.16.13 from kernel.org, without the startup hang.

Where I messed things up is with that link: the other X hangs that I experience are only when running some OpenGL programs, and it takes from a few seconds to some hours. The end effect is the same, I have to reset the machine. Should I enter a separate bug report for that?

This is not a linux-meta bug, so rejecting task there. Other bugs remain open.

This fix is tagged as committed, but the latest Dapper package does not seem to reflect it. When will it be available so I can test (I have the same problem).

The latest kernel update for Dapper (2.6.15-23) fixes the problem (at least on this machine).

Thanks for the great work.

This fixes the issue for me too. Thankyou!