Bug 77181
| Summary: | radeon -- GPU lockup when hibernating or waking up | ||
|---|---|---|---|
| Product: | Drivers | Reporter: | Mantas Mikulėnas (grawity) |
| Component: | Video(DRI - non Intel) | Assignee: | drivers_video-dri |
| Status: | NEW --- | ||
| Severity: | normal | CC: | alexdeucher, deviatov, mister.freeman, reuben_p, szg00000 |
| Priority: | P1 | ||
| Hardware: | x86-64 | ||
| OS: | Linux | ||
| Kernel Version: | 3.15.0-rc8 | Subsystem: | |
| Regression: | No | Bisected commit-id: | |
|
Description
Mantas Mikulėnas
2014-06-02 07:28:49 UTC
Same problem here (mobility radeon 4550, kernel 3.14.14-gentoo) Is this a regression? If so, can you bisect? (In reply to Alex Deucher from comment #2) I am sorry, but I don't think I understand you. What do you mean by "regression"? What am I trying to bisect? (In reply to Rostislav Devyatov from comment #3) > (In reply to Alex Deucher from comment #2) > > I am sorry, but I don't think I understand you. What do you mean by > "regression"? What am I trying to bisect? Did it used to work previously? If so what kernel? If you know what kernel used to work, you can use git to bisect the changes and identify what change broke it. (In reply to Alex Deucher from comment #4) > Did it used to work previously? If so what kernel? If you know what kernel > used to work, you can use git to bisect the changes and identify what change > broke it. Previously, in kernel 3.12.21-r1-gentoo, I had the same problem (the screen is sometimes messy after hibernation, and the same "fence wait failed" error in syslog). Before that, I ran deblobbed kernels 3.8.13-gentoo and 3.4.9-gentoo, and there were also problems with the screen, but a bit different. The screen just became black (sometimes it happened after hibernation, sometimes just during the normal work), but the mouse pointer was there, and the error in syslog was different: it started with "kernel BUG" at some location. I don't have the syslog from 3.8.13, but in 3.4.9 it was at mm/slub.c:3474 . On my system, both problems (the one I have now and the one I had previously) happen rather irregularly. Also, after a "normal" wake-up from hibernate, when the screen does not get distorted, I see the following message in dmesg: [drm:rv770_dpm_set_power_state] *ERROR* rv770_set_sw_state failed (In reply to Alex Deucher from comment #2) > Is this a regression? If so, can you bisect? Narrowed it down to the UVD merge for 3.10 (f18353eee757). ec5891fbe1b0 ("drm/radeon: UVD bringup v8") was the first commit with the "fence wait failed" errors: [ +0.010612] sd 0:0:0:0: [sda] Starting disk [ +9.661530] radeon 0000:01:00.0: GPU lockup CP stall for more than 10000msec [ +0.000003] radeon 0000:01:00.0: GPU lockup (waiting for 0x0000000000000004 last fence id 0x0000000000000002) [ +0.000003] [drm:r600_uvd_ib_test] *ERROR* radeon: fence wait failed (-35). [ +0.000002] [drm:radeon_ib_ring_tests] *ERROR* radeon: failed testing IB on ring 5 (-35). [ +1.614706] PM: restore of devices complete after 11819.193 msecs Seven commits in between [f2ba57b5eab8, ef0e6e657cfe] were also bad, but with a different error message: [ +0.010618] sd 0:0:0:0: [sda] Starting disk [ +0.560078] [drm:r600_uvd_init] *ERROR* UVD not responding, trying to reset the VCPU!!! [ +1.020296] [drm:r600_uvd_init] *ERROR* UVD not responding, trying to reset the VCPU!!! [ +1.020318] [drm:r600_uvd_init] *ERROR* UVD not responding, trying to reset the VCPU!!! [ +1.020299] [drm:r600_uvd_init] *ERROR* UVD not responding, trying to reset the VCPU!!! [ +1.020297] [drm:r600_uvd_init] *ERROR* UVD not responding, trying to reset the VCPU!!! [ +1.020298] [drm:r600_uvd_init] *ERROR* UVD not responding, trying to reset the VCPU!!! [ +1.020298] [drm:r600_uvd_init] *ERROR* UVD not responding, trying to reset the VCPU!!! [ +1.020323] [drm:r600_uvd_init] *ERROR* UVD not responding, trying to reset the VCPU!!! [ +1.020297] [drm:r600_uvd_init] *ERROR* UVD not responding, trying to reset the VCPU!!! [ +1.020297] [drm:r600_uvd_init] *ERROR* UVD not responding, trying to reset the VCPU!!! [ +0.020007] [drm:r600_uvd_init] *ERROR* UVD not responding, giving up!!! [ +0.000001] [drm:evergreen_startup] *ERROR* radeon: error initializing UVD (-1). [ +0.000026] [drm] ib test on ring 0 succeeded in 0 usecs [ +0.000025] [drm] ib test on ring 3 succeeded in 1 usecs [ +1.614533] PM: restore of devices complete after 11878.332 msecs [ +0.055061] PM: Image restored successfully. 4474f3a91f95 was the last good commit: [ +0.010691] sd 0:0:0:0: [sda] Starting disk [ +1.154120] PM: restore of devices complete after 1645.354 msecs I have the same problem with my radeon HD4650 PCie ( I use the radeon driver ), with dpm actived, the hibernation doesn't work, it fails with an error message in the console related to my radeon HD4650 "fence ring error", then at reboot fsck starts, which reveals that the hibernation has failed |