radeon problems with latest -current update

adamk75 · 01-11-2011, 06:05 PM

I updated -current today and have run into a serious problem.

This is KDE: http://68.45.22.62/X-KDE-current.png

xfce shows the same basic problem. Upon starting X, dmesg fills with:

Code:

[  205.345807] radeon 0000:07:00.0: r600_cs_track_validate_cb:203 cb pitch (64) invalid
[  205.345811] radeon 0000:07:00.0: r600_packet3_check:1108 invalid cmd stream 569
[  205.345814] [drm:radeon_cs_ioctl] *ERROR* Invalid command stream !
[  205.345980] radeon 0000:07:00.0: r600_cs_track_validate_cb:203 cb pitch (64) invalid
[  205.345983] radeon 0000:07:00.0: r600_packet3_check:1108 invalid cmd stream 569
[  205.345986] [drm:radeon_cs_ioctl] *ERROR* Invalid command stream !
[  205.346201] radeon 0000:07:00.0: r600_cs_track_validate_cb:203 cb pitch (64) invalid
[  205.346207] radeon 0000:07:00.0: r600_packet3_check:1108 invalid cmd stream 569
[  205.346211] [drm:radeon_cs_ioctl] *ERROR* Invalid command stream !
[  265.398988] radeon 0000:07:00.0: r600_cs_track_validate_cb:203 cb pitch (256) invalid
[  265.398993] radeon 0000:07:00.0: r600_packet3_check:1108 invalid cmd stream 2888

Prior to starting X, these are my drm messages:

Code:

[   23.351119] [drm] Initialized drm 1.1.0 20060810
[   23.451285] [drm] radeon kernel modesetting enabled.
[   23.456355] radeon 0000:07:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
[   23.465929] [drm] initializing kernel modesetting (RV770 0x1002:0x9442).
[   23.472873] [drm] register mmio base: 0xFCDF0000
[   23.477489] [drm] register mmio size: 65536
[   23.481825] ATOM BIOS: Wekiva
[   23.484817] [drm] Clocks initialized !
[   23.488577] radeon 0000:07:00.0: VRAM: 1024M 0x00000000 - 0x3FFFFFFF (1024M used)
[   23.496052] radeon 0000:07:00.0: GTT: 512M 0x40000000 - 0x5FFFFFFF
[   23.502247] mtrr: type mismatch for d0000000,10000000 old: write-back new: write-combining
[   23.510502] [drm] Detected VRAM RAM=1024M, BAR=256M
[   23.515376] [drm] RAM width 256bits DDR
[   23.519277] [TTM] Zone  kernel: Available graphics memory: 2021622 kiB.
[   23.525884] [TTM] Initializing pool allocator.
[   23.530347] [drm] radeon: 1024M of VRAM memory ready
[   23.535307] [drm] radeon: 512M of GTT memory ready.
[   23.540242] radeon 0000:07:00.0: radeon: using MSI.
[   23.545149] [drm] radeon: irq initialized.
[   23.549242] [drm] GART: num cpu pages 131072, num gpu pages 131072
[   23.556621] [drm] Loading RV770 Microcode
[   23.694387] [drm] ring test succeeded in 1 usecs
[   23.699196] [drm] radeon: ib pool ready.
[   23.703224] [drm] ib test succeeded in 0 usecs
[   23.707665] [drm] Enabling audio support
[   23.711552] [drm] Default TV standard: NTSC
[   23.715927] [drm] Default TV standard: NTSC
[   23.720162] [drm] Default TV standard: NTSC
[   23.724441] [drm] Radeon Display Connectors
[   23.728627] [drm] Connector 0:
[   23.731681] [drm]   DVI-I
[   23.734301] [drm]   HPD1
[   23.736833] [drm]   DDC: 0x7e60 0x7e60 0x7e64 0x7e64 0x7e68 0x7e68 0x7e6c 0x7e6c
[   23.744218] [drm]   Encoders:
[   23.747184] [drm]     DFP1: INTERNAL_UNIPHY
[   23.751364] [drm]     CRT2: INTERNAL_KLDSCP_DAC2
[   23.755978] [drm] Connector 1:
[   23.759030] [drm]   DIN
[   23.761475] [drm]   Encoders:
[   23.764440] [drm]     TV1: INTERNAL_KLDSCP_DAC2
[   23.768966] [drm] Connector 2:
[   23.772019] [drm]   DVI-I
[   23.774638] [drm]   HPD2
[   23.777170] [drm]   DDC: 0x7e20 0x7e20 0x7e24 0x7e24 0x7e28 0x7e28 0x7e2c 0x7e2c
[   23.784555] [drm]   Encoders:
[   23.787521] [drm]     CRT1: INTERNAL_KLDSCP_DAC1
[   23.792132] [drm]     DFP2: INTERNAL_KLDSCP_LVTMA
[   23.903145] [drm] Internal thermal controller with fan control
[   23.908988] [drm] radeon: power management initialized
[   24.058837] [drm] fb mappable at 0xD0142000
[   24.063021] [drm] vram apper at 0xD0000000
[   24.067114] [drm] size 7258112
[   24.070165] [drm] fb depth is 24
[   24.073392] [drm]    pitch is 6912
[   24.486626] Console: switching to colour frame buffer device 160x64
[   24.497310] fb0: radeondrmfb frame buffer device
[   24.497312] drm: registered panic notifier
[   24.506025] [drm] Initialized radeon 2.5.0 20080528 for 0000:07:00.0 on minor 0

Anyone else with the same problem? I'll try updating to a newer kernel and see if the problem goes away.

Adam

Darth Vader · 01-11-2011, 06:13 PM

Strange! Looks like an old libDRM library speak gibberish with the kernel's DRM part!!?!!? Try an rebuild of libdrm and mesa.

adamk75 · 01-11-2011, 06:17 PM

I did reinstall libdrm and mesa, but not rebuilding from source. Guess I can give that a shot.

Adam

adamk75 · 01-11-2011, 07:52 PM

I'm running 2.6.36.3 now, built using the config for 2.6.35.10-huge, and now everything is fine. Looks like 2.6.35.10 has a bug in the radeon drm.

Adam

tpreitzel · 01-11-2011, 08:06 PM

If the 2.6.35.10 kernel is broken with the Radeon, I sure hope that the next release of Slackware is delayed until it's fixed. Since the compiler has just been upgraded, why not switch to a more reliable kernel, like 2.6.38 ...

Why? I need my Hauppauge remote to work and major new parts of the open source Radeon driver will be included. Personally, I can wait another month or two for a new Slackware release. I actually want a release which works in all respects ... I don't blame Pat and his team, but I AM growing a bit tired of dealing with incomplete and broken kernels... In the meantime, I'll just stick with 2.6.35.7 ....

adamk75 · 01-11-2011, 08:09 PM

Of course, it could just be an issue with 2.6.35.10 + HD4850 + my particular motherboard. I'd be interested in knowing if others are having the same problem.

Adam

tpreitzel · 01-11-2011, 08:15 PM

Quote:

Originally Posted by adamk75

Of course, it could just be an issue with 2.6.35.10 + HD4850 + my particular motherboard. I'd be interested in knowing if others are having the same problem.

Adam

Since I have an HD2600, I'm tempted to take a chance... Understand, though, that the current Catalyst 10.12 driver doesn't work with the current release of SDLMESS * so if the Radeon driver is broken when I upgrade the kernel to 2.6.35.10, I'm between a rock and a hard spot.... I think I'll wait until ATI/AMD releases a new Catalyst driver (should be less than two weeks) and test it with SDLMESS before upgrading to 2.6.35.10 ...

* sdlvideo_init errors ...

Chuck56 · 01-12-2011, 06:41 AM

ATI HD4200 motherboard here and I'm not getting those lines in dmesg. The radeon and drm modules are loaded but I have no access to KDE dektop effects at all. It's been like that for a while on -current. I do notice a Xorg.0.log entry calling out libdrm-2.6.0 or newer required. Maybe I'll give that a spin.

Quote:

[ 34.849] (II) VESA: driver for VESA chipsets: vesa
[ 34.849] (++) using VT number 7

[ 34.859] (II) [KMS] Kernel modesetting enabled.
[ 34.859] (WW) Falling back to old probe method for vesa
[ 34.860] (II) RADEON(0): Creating default Display subsection in Screen section
"Default Screen Section" for depth/fbbpp 24/32
[ 34.860] (==) RADEON(0): Depth 24, (--) framebuffer bpp 32
[ 34.860] (II) RADEON(0): Pixel depth = 24 bits stored in 4 bytes (32 bpp pixmaps)
[ 34.860] (==) RADEON(0): Default visual is TrueColor
[ 34.860] (==) RADEON(0): RGB weight 888
[ 34.860] (II) RADEON(0): Using 8 bits per RGB (8 bit DAC)
[ 34.860] (--) RADEON(0): Chipset: "ATI Radeon HD 4200" (ChipID = 0x9710)
[ 34.860] (II) RADEON(0): PCI card detected
[ 34.860] drmOpenDevice: node name is /dev/dri/card0
[ 34.860] drmOpenDevice: open result is 9, (OK)
[ 34.860] drmOpenByBusid: Searching for BusID pci:0000:01:05.0
[ 34.860] drmOpenDevice: node name is /dev/dri/card0
[ 34.860] drmOpenDevice: open result is 9, (OK)
[ 34.860] drmOpenByBusid: drmOpenMinor returns 9
[ 34.860] drmOpenByBusid: Interface 1.4 failed, trying 1.1
[ 34.860] drmOpenByBusid: drmGetBusid reports pci:0000:01:05.0
[ 34.860] (II) RADEON(0): R6xx+ KMS Color Tiling requires radeon drm 2.6.0 or newer
[ 34.860] (II) RADEON(0): KMS Color Tiling: disabled
[ 34.888] (II) RADEON(0): Output VGA-0 has no monitor section
[ 34.942] (II) RADEON(0): Output DVI-0 has no monitor section
[ 34.956] (II) RADEON(0): EDID for output VGA-0
[ 35.010] (II) RADEON(0): EDID for output DVI-0

stonefly · 01-12-2011, 09:01 AM

Quote:

Originally Posted by adamk75

Of course, it could just be an issue with 2.6.35.10 + HD4850 + my particular motherboard. I'd be interested in knowing if others are having the same problem.

Same problem here with HD4350.
Downgrading kernel back to 2.6.35.7 helped.

disturbed1 · 01-27-2011, 06:07 PM

Quote:

Originally Posted by adamk75

Of course, it could just be an issue with 2.6.35.10 + HD4850 + my particular motherboard. I'd be interested in knowing if others are having the same problem.

Adam

Same here with an hd3470. Was fine in 2.6.35.7 - not so in 2.6.35.10. Built 2.6.37, radeon works for anything 2d. 3d is a mess, maybe I'll look into upgrading libdrm.
My x1600 agp runs acceptably on 2.6.35.7. Braid, Aquaria, WoP ...

Yes 2.6.35.10 is broken with radeon. Bug reports abound all over the place. Sadly this was fixed in both 2.6.36 and 2.6.37, but never back ported to the long term supported kernel. Maybe in .11.
I've attached the patch from Fedora that fixes this. https://bugzilla.redhat.com/show_bug.cgi?id=664206
IMO, considering you would have to patch .10, and recompile, it might be worth while to just use a working kernel (2.6.35.7), or upgrade and compile a newer non 2.6.35.x series kernel.

EdGr · 01-27-2011, 07:14 PM

I've stayed on the 2.6.35.7 kernel until 2.6.35.10 gets sorted out.

Here is another report of the same bug:

https://bugs.launchpad.net/ubuntu/+s...ux/+bug/703553

Ed

disturbed1 · 02-07-2011, 12:45 AM

2.6.35.11 was released with lots of radeon related fixes
http://www.kernel.org/pub/linux/kern...eLog-2.6.35.11