Direct rendering disabled on Radeon 9800 Pro/Power G5

[Sergey V. Udaltsov]

Binary package hint: xserver-xorg-video-ati

After today's upgrade of Jaunty, I lost DRI on my Mac. In Xorg.0.log I see:
[ 0.647058] (II) RADEON(0): [DRI] installation complete
[ 0.650806] (II) RADEON(0): [drm] removed 1 reserved context for kernel
[ 0.650889] (II) RADEON(0): [drm] unmapping 8192 bytes of SAREA 0x19fff000 at 0xf7fdb000
[ 0.650983] (II) RADEON(0): [drm] Closed DRM master.
[ 0.651009] (WW) RADEON(0): Direct rendering disabled

About the hardware:

$ lspci | grep Rade
0000:f0:10.0 VGA compatible controller: ATI Technologies Inc Radeon R350 [Radeon 9800 Pro]

dmesg snippet:
[drm:radeon_do_init_cp] *ERROR* could not find ioremap agp regions!

[Sergey V. Udaltsov (sergey-udaltsov)]

Uups, sorry, wrong log file.

[Bryce Harrington (bryce)]

Actually what I need is the PCI ID's, which can be gotten via:

lspci -vvnn | head -n2

lspci -vvnn | grep -A2 "VGA "

I gather your video card is AGP? Have you tested setting different VGAModes? See

  https://wiki.ubuntu.com/X/Quirks#ATI Radeon Driver Quirks

(am I remembering correctly that we went through this once already for your hardware?)

[Sergey V. Udaltsov (sergey-udaltsov)]

Yes we did:)

$ lspci -vvnn | head -n2
0000:f0:0b.0 Host bridge [0600]: Apple Computer Inc. U3H AGP Bridge [106b:0059]
 Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-

lspci -vvnn | grep -A2 "VGA "
0000:f0:10.0 VGA compatible controller [0300]: ATI Technologies Inc Radeon R350 [Radeon 9800 Pro] [1002:4e48]
 Subsystem: ATI Technologies Inc Radeon R350 [Radeon 9800 Pro] [1002:4e48]
 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping+ SERR- FastB2B- DisINTx-

Yes it is AGP. But I did not try VGAModes yet - because it worked just fine, till yesterday. ATM, the Device section for the adapter is pretty much clean.

Or to do you mean AGPMode?

[Bryce Harrington (bryce)]

On Fri, Mar 20, 2009 at 08:06:42AM -0000, Sergey V. Udaltsov wrote:
> Yes we did:)
>
> $ lspci -vvnn | head -n2
> 0000:f0:0b.0 Host bridge [0600]: Apple Computer Inc. U3H AGP Bridge [106b:0059]
> Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
>
> lspci -vvnn | grep -A2 "VGA "
> 0000:f0:10.0 VGA compatible controller [0300]: ATI Technologies Inc Radeon R350 [Radeon 9800 Pro] [1002:4e48]
> Subsystem: ATI Technologies Inc Radeon R350 [Radeon 9800 Pro] [1002:4e48]
> Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping+ SERR- FastB2B- DisINTx-
>
> Yes it is AGP. But I did not try VGAModes yet - because it worked just
> fine, till yesterday. ATM, the Device section for the adapter is pretty
> much clean.
>
> Or to do you mean AGPMode?

Sorry, yeah AGPMode. Long day.

Well, DRI problems and AGP cards often seem to go hand in hand, so
quirking that would be my first guess.

-ati has received a good bit of DRM lovin' in the last few days which
should mean everything works better, but of course regressions are
always possible. If the AGPMode fiddling doesn't work we can bump this
upstream and see what Alex says.

Bryce

[Sergey V. Udaltsov (sergey-udaltsov)]

ok, I will try to play with agpmode tonight.

PS A bit surprised about those kind of changes at the final stages of the release cycle...

[Sergey V. Udaltsov (sergey-udaltsov)]

Tried AGP Modes 2 and 4 - no changes. There is one suspicious line in Xorg.0.log:

(II) RADEON(0): [dri] Found DRI library version 1.3.0 and kernel module version 1.29.0

Can this version mismatch cause troubles?

[Sergey V. Udaltsov (sergey-udaltsov)]

Another interesting effect found after that update.

x server does not report keyboard plug-in event any more (when I turn monitor on along with its built-in USB hub).

[Herbert V. Riedel (hvr)]

...any progress on this one? I got the same issue here...

0000:00:10.0 VGA compatible controller: ATI Technologies Inc RV350 [Mobility Radeon 9600 M10]

[Tormod Volden (tormodvolden)]

Can you please attach dmesg output also?

[Sergey V. Udaltsov (sergey-udaltsov)]

dmesg

[Tormod Volden (tormodvolden)]

Thanks. The most relevant part:

[ 50.926040] [drm] Initialized drm 1.1.0 20060810
[ 50.964331] [drm] Initialized radeon 1.29.0 20080528 on minor 0
[ 51.271279] ioctl32(Xorg:3851): Unknown cmd fd(0) cmd(80086408){t:'d';sz:8} arg(ffe808b8) on /var/log/Xorg.0.log
[ 51.271297] ioctl32(Xorg:3851): Unknown cmd fd(0) cmd(80086408){t:'d';sz:8} arg(ffe808b8) on /var/log/Xorg.0.log
[ 51.342143] agpgart: BIOS bug. AGP bridge claims to only support x4 rateFixing up support for x2 & x1
[ 51.342171] agpgart-uninorth 0000:f0:0b.0: putting AGP V2 device into 4x mode
[ 51.342184] radeonfb 0000:f0:10.0: putting AGP V2 device into 4x mode
[ 51.583610] [drm:radeon_do_init_cp] *ERROR* could not find ioremap agp regions!

BTW, there is also a kernel oops in there later, something with nautilus after plugging in a USB hub with compact flash cards and other devices it seems like.