Created attachment 121201 [details]
dmesg1

It does not happen always, but usually after a few hibernate/resume cycles (2 or 3, sometimes earlier). The following messages can be found in dmesg (see attachment for full output).

[ 1001.941216] [drm] ring test on 0 succeeded in 0 usecs
[ 1001.941331] [drm] ring test on 1 succeeded in 0 usecs
[ 1001.941370] [drm] ring test on 2 succeeded in 0 usecs
[ 1001.941565] [drm] ring test on 3 succeeded in 0 usecs
[ 1001.941571] [drm] ring test on 4 succeeded in 0 usecs
[ 1001.967662] [drm] ring test on 5 succeeded in 0 usecs
[ 1001.967696] [drm] UVD initialized successfully.
[ 1002.249117] [drm:radeon_vce_ring_test] *ERROR* radeon: ring 6 test failed]
[ 1002.249139] [drm] ib test on ring 0 succeeded in 0 usecs
[ 1002.749193] [drm] ib test on ring 1 succeeded in 0 usecs
[ 1002.749221] [drm] ib test on ring 2 succeeded in 0 usecs
[ 1002.749304] [drm] ib test on ring 3 succeeded in 0 usecs
[ 1002.749354] [drm] ib test on ring 4 succeeded in 0 usecs
[ 1003.269160] [drm] ib test on ring 5 succeeded
[ 1013.788806] radeon 0000:01:00.0: ring 7 stalled for more than 10000msec
[ 1013.788809] radeon 0000:01:00.0: GPU lockup (current fence id 0x0000000000000008 last fence id 0x000000000000000a on ring 7)
[ 1013.788868] [drm:radeon_vce_ib_test] *ERROR* radeon: fence wait failed (-35).
[ 1013.788874] [drm:radeon_ib_ring_tests] *ERROR* radeon: failed testing IB on ring 7 (-35).

After that, if the machine hibernates and resumes once again, additional errors will be printed and the X session becomes unusable (only the cursor blinking on the terminal). I will attach the dmesg of this too, though it might be less interesting.