GPU lockup problem

[morgon]

Hi,

I am running testing and have recently switched from Gnome-shell to i3.

It may be purely coincidental and may have nothing to do with i3, but currently I have a lot of problems when my PC comes out from hibernation (I hibernate via systemctl hibernate).

What happens is that the desktop is completely frozen, I can move the mouse around but no key work.

The system as such however comes up and I can log on via the virtual terminals, and when I do that I usually see message like "GPU lockup" via dmsg, which I believe (however I don't really know) may the the root of the problem.

This is an old PC, my graphics is an integrated Radeon HD 4250, and I even though I had occasional hiccups with hibernation, it was never as bad now when basically 3 out of 4 attempts to come back from hibernation fail and I have to reboot.

Does anyone have any idea about what could be the problem here and how to attack it?

Many thanks!

[debiman]

so it worked earlier, i assume?

how exactly did you "switch"?

has there also been a system upgrade at the same time?

have you made sure that all the daemons & services started by gnome shell login are still started & working properly?

what exactly do you see in dmesg? please copy paste text here. no screenshots please.

[milomak]

what happens if you use another de?

[morgon]

As I said I had the occasional hiccup before but never really investigated.

I switched by installing i3 (how else?) which gives me the option to run it in gdm.
I still have gnome-shell installed. There has been no system upgrade.

At the moment I cannot post dmesg output , I'll have to wait for the next time it happens...

[morgon]

This is what I see in dmesg:

[79241.965753] radeon 0000:01:05.0: GPU lockup (current fence id 0x00000000002a0f71 last fence id 0x00000000002a0f98 on ring 0)
[79242.477758] radeon 0000:01:05.0: ring 0 stalled for more than 18296msec
[79242.477773] radeon 0000:01:05.0: GPU lockup (current fence id 0x00000000002a0f71 last fence id 0x00000000002a0f98 on ring 0)
[79242.989763] radeon 0000:01:05.0: ring 0 stalled for more than 18808msec
[79242.989778] radeon 0000:01:05.0: GPU lockup (current fence id 0x00000000002a0f71 last fence id 0x00000000002a0f98 on ring 0)
[79243.501745] radeon 0000:01:05.0: ring 0 stalled for more than 19320msec
[79243.501760] radeon 0000:01:05.0: GPU lockup (current fence id 0x00000000002a0f71 last fence id 0x00000000002a0f98 on ring 0)
[79244.013761] radeon 0000:01:05.0: ring 0 stalled for more than 19832msec
[79244.013777] radeon 0000:01:05.0: GPU lockup (current fence id 0x00000000002a0f71 last fence id 0x00000000002a0f98 on ring 0)
[79244.525776] radeon 0000:01:05.0: ring 0 stalled for more than 20344msec
[79244.525791] radeon 0000:01:05.0: GPU lockup (current fence id 0x00000000002a0f71 last fence id 0x00000000002a0f98 on ring 0)
[79245.037715] radeon 0000:01:05.0: ring 0 stalled for more than 20856msec
[79245.037730] radeon 0000:01:05.0: GPU lockup (current fence id 0x00000000002a0f71 last fence id 0x00000000002a0f98 on ring 0)
[79245.409434] radeon 0000:01:05.0: failed to get a new IB (-35)
[79245.409484] [drm:radeon_cs_ioctl [radeon]] *ERROR* Failed to get ib !
[79245.416478] radeon 0000:01:05.0: Saved 1273 dwords of commands on ring 0.
[79245.416484] radeon 0000:01:05.0: GPU softreset: 0x00000009
[79245.416485] radeon 0000:01:05.0: R_008010_GRBM_STATUS = 0xE5702030
[79245.416486] radeon 0000:01:05.0: R_008014_GRBM_STATUS2 = 0x00110103
[79245.416487] radeon 0000:01:05.0: R_000E50_SRBM_STATUS = 0x20001040
[79245.416488] radeon 0000:01:05.0: R_008674_CP_STALLED_STAT1 = 0x00000000
[79245.416490] radeon 0000:01:05.0: R_008678_CP_STALLED_STAT2 = 0x00008002
[79245.416491] radeon 0000:01:05.0: R_00867C_CP_BUSY_STAT = 0x00008086
[79245.416492] radeon 0000:01:05.0: R_008680_CP_STAT = 0x80018645
[79245.416493] radeon 0000:01:05.0: R_00D034_DMA_STATUS_REG = 0x44C83D57
[79245.467669] radeon 0000:01:05.0: R_008020_GRBM_SOFT_RESET=0x00007FEF
[79245.467721] radeon 0000:01:05.0: SRBM_SOFT_RESET=0x00000100
[79245.469808] radeon 0000:01:05.0: R_008010_GRBM_STATUS = 0xA0003030
[79245.469809] radeon 0000:01:05.0: R_008014_GRBM_STATUS2 = 0x00000003
[79245.469811] radeon 0000:01:05.0: R_000E50_SRBM_STATUS = 0x20008040
[79245.469812] radeon 0000:01:05.0: R_008674_CP_STALLED_STAT1 = 0x00000000
[79245.469813] radeon 0000:01:05.0: R_008678_CP_STALLED_STAT2 = 0x00000000
[79245.469814] radeon 0000:01:05.0: R_00867C_CP_BUSY_STAT = 0x00000000
[79245.469816] radeon 0000:01:05.0: R_008680_CP_STAT = 0x80100000
[79245.469817] radeon 0000:01:05.0: R_00D034_DMA_STATUS_REG = 0x44C83D57
[79245.469821] radeon 0000:01:05.0: GPU reset succeeded, trying to resume
[79245.487984] [drm] PCIE GART of 512M enabled (table at 0x00000000C0146000).
[79245.488023] radeon 0000:01:05.0: WB enabled
[79245.488026] radeon 0000:01:05.0: fence driver on ring 0 use gpu addr 0x00000000a0000c00 and cpu addr 0x0000000068a37c81
[79245.490278] radeon 0000:01:05.0: fence driver on ring 5 use gpu addr 0x00000000c0056038 and cpu addr 0x00000000e1c3d270
[79245.522037] [drm] ring test on 0 succeeded in 1 usecs
[79245.696708] [drm] ring test on 5 succeeded in 1 usecs
[79245.696717] [drm] UVD initialized successfully.
[79246.733839] [drm:r600_ib_test [radeon]] *ERROR* radeon: fence wait timed out.
[79246.733913] [drm:radeon_ib_ring_tests [radeon]] *ERROR* radeon: failed testing IB on GFX ring (-110).
[79257.005818] /dev/vmmon[12482]: HostIFReadUptimeWork: detected settimeofday: fixed uptimeBase old 18445200754668021482 new 18445200754666720245 attempts 1

[vbrummond]

Without being an expert looks like that is a bug with the kernel. The gpu did restart properly, so you likely can "fix it" by restarting the gdm service. This will restart your desktop though which isn't much different from a reboot. You likely have to wait for it to be fixed in the drm drivers.