Hi,
I'm frequently updating Xorg for my oS 12.3 from http://download.opensuse.org/repositories/X11:/XOrg/openSUSE_12.3/ and my kernel from the respective kernel repo.
I got GPU resets occasionally, but since the most recent updates, all "3D applications" (such as Steam, etc.) are unusable due to frequent resets [1]. Actually I had to restart the whole machine most of the time.
[2]: graphics card info [3]: installed packages [4]: lsmod info [5]: kernel info
Is this more likely a problem on my machine or a bug? Something I can do but wait for updates? I'd happily test new stuff...
Regards,
--D.B. Daniel Bischof suse-bugz@bischof.org
--- [1] --- kernel: [ 1444.670046] radeon 0000:01:00.0: GPU lockup CP stall for more than 10000msec kernel: [ 1444.670051] radeon 0000:01:00.0: GPU lockup (waiting for 0x0000000000014637) kernel: [ 1444.670055] radeon 0000:01:00.0: failed to get a new IB (-35) kernel: [ 1444.670057] [drm:radeon_cs_ib_chunk] *ERROR* Failed to get ib ! kernel: [ 1444.677358] radeon 0000:01:00.0: Saved 791 dwords of commands on ring 0. kernel: [ 1444.677369] radeon 0000:01:00.0: GPU softreset: 0x00000009 kernel: [ 1444.677371] radeon 0000:01:00.0: GRBM_STATUS = 0xE77309A0 kernel: [ 1444.677373] radeon 0000:01:00.0: GRBM_STATUS_SE0 = 0xFC000001 kernel: [ 1444.677375] radeon 0000:01:00.0: GRBM_STATUS_SE1 = 0x00000007 kernel: [ 1444.677377] radeon 0000:01:00.0: SRBM_STATUS = 0x200000C0 kernel: [ 1444.677379] radeon 0000:01:00.0: SRBM_STATUS2 = 0x00000000 kernel: [ 1444.677381] radeon 0000:01:00.0: R_008674_CP_STALLED_STAT1 = 0x01000000 kernel: [ 1444.677383] radeon 0000:01:00.0: R_008678_CP_STALLED_STAT2 = 0x00011000 kernel: [ 1444.677385] radeon 0000:01:00.0: R_00867C_CP_BUSY_STAT = 0x00028506 kernel: [ 1444.677387] radeon 0000:01:00.0: R_008680_CP_STAT = 0x80838647 kernel: [ 1444.677388] radeon 0000:01:00.0: R_00D034_DMA_STATUS_REG = 0x44C83D57 kernel: [ 1444.692959] radeon 0000:01:00.0: GRBM_SOFT_RESET=0x00007F6B kernel: [ 1444.693013] radeon 0000:01:00.0: SRBM_SOFT_RESET=0x00000100 kernel: [ 1444.694172] radeon 0000:01:00.0: GRBM_STATUS = 0x00003828 kernel: [ 1444.694174] radeon 0000:01:00.0: GRBM_STATUS_SE0 = 0x00000007 kernel: [ 1444.694176] radeon 0000:01:00.0: GRBM_STATUS_SE1 = 0x00000007 kernel: [ 1444.694178] radeon 0000:01:00.0: SRBM_STATUS = 0x200000C0 kernel: [ 1444.694180] radeon 0000:01:00.0: SRBM_STATUS2 = 0x00000000 kernel: [ 1444.694181] radeon 0000:01:00.0: R_008674_CP_STALLED_STAT1 = 0x00000000 kernel: [ 1444.694183] radeon 0000:01:00.0: R_008678_CP_STALLED_STAT2 = 0x00000000 kernel: [ 1444.694185] radeon 0000:01:00.0: R_00867C_CP_BUSY_STAT = 0x00000000 kernel: [ 1444.694187] radeon 0000:01:00.0: R_008680_CP_STAT = 0x00000000 kernel: [ 1444.694189] radeon 0000:01:00.0: R_00D034_DMA_STATUS_REG = 0x44C83D57 kernel: [ 1444.694197] radeon 0000:01:00.0: GPU reset succeeded, trying to resume kernel: [ 1444.713085] [drm] PCIE gen 2 link speeds already enabled kernel: [ 1444.717080] [drm] PCIE GART of 512M enabled (table at 0x0000000000273000). kernel: [ 1444.717326] radeon 0000:01:00.0: WB enabled kernel: [ 1444.717329] radeon 0000:01:00.0: fence driver on ring 0 use gpu addr 0x0000000040000c00 and cpu addr 0xffff880035189c00 kernel: [ 1444.717332] radeon 0000:01:00.0: fence driver on ring 3 use gpu addr 0x0000000040000c0c and cpu addr 0xffff880035189c0c kernel: [ 1444.723341] radeon 0000:01:00.0: fence driver on ring 5 use gpu addr 0x0000000000072118 and cpu addr 0xffffc900104b2118 kernel: [ 1444.739913] [drm] ring test on 0 succeeded in 2 usecs kernel: [ 1444.740105] [drm] ring test on 3 succeeded in 1 usecs kernel: [ 1444.917260] [drm] ring test on 5 succeeded in 1 usecs kernel: [ 1444.917265] [drm] UVD initialized successfully. kernel: [ 1444.997272] [drm] ib test on ring 0 succeeded in 0 usecs kernel: [ 1444.997335] [drm] ib test on ring 3 succeeded in 0 usecs kernel: [ 1445.149847] [drm] ib test on ring 5 succeeded ---
--- [2] --- $ /sbin/lspci | grep VGA 01:00.0 VGA compatible controller: Advanced Micro Devices [AMD] nee ATI Turks [Radeon HD 6670] ---
--- [3] --- $ rpm -qa | grep -E -i 'mesa|xorg' | sort DirectFB-Mesa-1.6.2-3.1.1.x86_64 Mesa-32bit-9.2.0-300.1.x86_64 Mesa-9.2.0-300.1.x86_64 Mesa-demo-x-8.1.0-29.5.x86_64 Mesa-devel-9.2.0-300.1.x86_64 Mesa-libEGL-devel-9.2.0-300.1.x86_64 Mesa-libEGL1-32bit-9.2.0-300.1.x86_64 Mesa-libEGL1-9.2.0-300.1.x86_64 Mesa-libGL-devel-9.2.0-300.1.x86_64 Mesa-libGL1-32bit-9.2.0-300.1.x86_64 Mesa-libGL1-9.2.0-300.1.x86_64 Mesa-libGLESv1_CM-devel-9.2.0-300.1.x86_64 Mesa-libGLESv1_CM1-9.2.0-300.1.x86_64 Mesa-libGLESv2-2-9.2.0-300.1.x86_64 Mesa-libGLESv2-devel-9.2.0-300.1.x86_64 Mesa-libIndirectGL-devel-9.2.0-294.1.x86_64 Mesa-libIndirectGL1-9.2.0-294.1.x86_64 Mesa-libglapi-devel-9.2.0-300.1.x86_64 Mesa-libglapi0-32bit-9.2.0-300.1.x86_64 Mesa-libglapi0-9.2.0-300.1.x86_64 libOSMesa-devel-9.2.0-300.1.x86_64 libOSMesa9-32bit-9.2.0-300.1.x86_64 libOSMesa9-9.2.0-300.1.x86_64 xorg-cf-files-1.0.5-18.2.noarch xorg-scripts-1.0.1-7.1.noarch xorg-sgml-doctools-1.11-9.1.noarch xorg-x11-7.6_1-179.1.noarch xorg-x11-Xvnc-7.6_1.0.1-44.3.x86_64 xorg-x11-devel-7.6-122.1.noarch xorg-x11-driver-input-7.6_1-111.1.noarch xorg-x11-driver-video-7.6_1-213.1.x86_64 xorg-x11-driver-video-nouveau-1.0.9-66.2.x86_64 xorg-x11-essentials-7.6_1-179.1.noarch xorg-x11-fonts-7.6-98.1.noarch xorg-x11-fonts-core-7.6-98.1.noarch xorg-x11-libX11-ccache-7.6-32.3.noarch xorg-x11-proto-devel-7.6-152.1.x86_64 xorg-x11-server-7.6_1.14.3-251.5.x86_64 xorg-x11-util-devel-7.6_1-37.1.noarch ---
--- [4] --- $ /sbin/lsmod | grep -E -i radeon radeon 1313747 2 ttm 92139 1 radeon drm_kms_helper 52938 1 radeon drm 317493 4 radeon,ttm,drm_kms_helper i2c_algo_bit 13413 1 radeon ---
--- [5] --- $ cat /proc/version Linux version 3.11.2-1.g4c1fc5a-desktop (geeko@buildhost) (gcc version 4.7.2 20130108 [gcc-4_7-branch revision 195012] (SUSE Linux) ) #1 SMP PREEMPT Fri Sep 27 09:05:13 UTC 2013 (4c1fc5a) ---
Hi,
On Sun, 29 Sep 2013, Daniel Bischof wrote:
I'm frequently updating Xorg for my oS 12.3 from http://download.opensuse.org/repositories/X11:/XOrg/openSUSE_12.3/ and my kernel from the respective kernel repo.
I got GPU resets occasionally, but since the most recent updates, all "3D applications" (such as Steam, etc.) are unusable due to frequent resets [1]. Actually I had to restart the whole machine most of the time. [...]
I investigated this further:
- updated to most recent Xorg repo updates from today: No change - downgraded to 12.3 kernel (3.7.10-1.16-desktop), kept stuff from Xorg repo: GPU resets almost vanished, but occasional lags and 3D performance poor - installed latest AMD Catalyst proprietary driver: no more resets, no lags and great 3D performance (so my card appears to be ok)
I was kind of rejoicing that the Radeon OSS driver finally made a 3D performance jump, but apparently we're not quite there yet. Is there something I could help with (test packages, provide information, ...)?
Best regards,
--D.
Hi,
On Mon, 30 Sep 2013, Daniel Bischof wrote:
Hi,
On Sun, 29 Sep 2013, Daniel Bischof wrote:
I'm frequently updating Xorg for my oS 12.3 from http://download.opensuse.org/repositories/X11:/XOrg/openSUSE_12.3/ and my kernel from the respective kernel repo.
I got GPU resets occasionally, but since the most recent updates, all "3D applications" (such as Steam, etc.) are unusable due to frequent resets [1]. Actually I had to restart the whole machine most of the time. [...]
I investigated this further:
- updated to most recent Xorg repo updates from today: No change
- downgraded to 12.3 kernel (3.7.10-1.16-desktop), kept stuff from Xorg repo:
GPU resets almost vanished, but occasional lags and 3D performance poor
- installed latest AMD Catalyst proprietary driver: no more resets, no lags
and great 3D performance (so my card appears to be ok)
I was kind of rejoicing that the Radeon OSS driver finally made a 3D performance jump, but apparently we're not quite there yet. Is there something I could help with (test packages, provide information, ...)?
I could finally resolve this: Booting with "radeon.dpm=1" does the trick. No further settings in /sys/class/drm/card0/device or something else, just the boot option.
The effect is reproducible: Watching the WebGL fish tank [1] causes GPU resets after a couple of seconds. After having rebooted with DPM enabled, it works like a charm (and my other OpenGL stuff, too).
[1] http://people.mozilla.org/~jmuizelaar/fishie/fishie-gl.html
Best regards,
--D.B. --- Daniel Bischof suse-bugz@bischof.org
-
Daniel Bischof