Created attachment 163251 [details] a sudo journalctl retrieved dmesg I have recently updated to 3.19-rc4 from 3.19-rc3 and with rc3 I haven't experienced this. With rc4 what happened was that I went away from the keyboard while listening to vlc audio, came back after an hour, the screen was blank(as usual), audio was still working, but mouse movements and key presses had no effect in waking up from blank, the screen remained black(no backlight). The only thing that did work was SysRq unmount and reboot (which were the only 2 things that I tried, with SysRq). I have gone back with journalctl and found the log, included in attachment. Main part looks like this: Jan 12 00:01:01 manji anacron[10747]: Anacron started on 2015-01-12 Jan 12 00:01:01 manji CROND[10739]: pam_unix(crond:session): session closed for user root Jan 12 00:01:01 manji anacron[10747]: Normal exit (0 jobs run) Jan 12 00:01:36 manji systemd-journal[1653]: Forwarding to syslog missed 3 messages. Jan 12 00:25:40 manji systemd-journal[1653]: Missed 16 kernel messages Jan 12 00:25:40 manji kernel: INFO: task Xorg.bin:3666 blocked for more than 120 seconds. Jan 12 00:25:40 manji kernel: Tainted: G IO 3.19.0-rc4-geaa27f3 #7 Jan 12 00:25:40 manji kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. Jan 12 00:25:40 manji kernel: Xorg.bin D ffff8800370b3808 12888 3666 3665 0x00000004 Jan 12 00:25:40 manji kernel: ffff8800370b3808 0001000000000002 0000000000009000 ffff88040b0a5aa0 Jan 12 00:25:40 manji kernel: ffff8800370b3fd8 ffff88040b9f8000 ffff88040b0a5aa0 ffff88040b0a5aa0 Jan 12 00:25:40 manji kernel: 0000000000000296 ffff88040b0a5aa0 7fffffffffffffff ffff8800370b3958 Jan 12 00:25:40 manji kernel: Call Trace: Jan 12 00:25:40 manji kernel: [<ffffffff81685794>] schedule+0x24/0x70 Jan 12 00:25:40 manji kernel: [<ffffffff8168974c>] schedule_timeout+0x1dc/0x250 Jan 12 00:25:40 manji kernel: [<ffffffff813d7c61>] ? radeon_fence_process+0x11/0x40 Jan 12 00:25:40 manji kernel: [<ffffffff813d7ccc>] ? radeon_fence_any_seq_signaled+0x3c/0x90 Jan 12 00:25:40 manji kernel: [<ffffffff813d7fb7>] radeon_fence_wait_seq_timeout.constprop.8+0x297/0x2f0 Jan 12 00:25:40 manji kernel: [<ffffffff8107a4e0>] ? wait_woken+0x90/0x90 Jan 12 00:25:40 manji kernel: [<ffffffff813d837f>] radeon_fence_wait_any+0x4f/0x60 Jan 12 00:25:40 manji kernel: [<ffffffff8144a585>] radeon_sa_bo_new+0x295/0x470 Jan 12 00:25:40 manji kernel: [<ffffffff813dba7c>] ? radeon_bo_move_notify+0x7c/0x90 Jan 12 00:25:40 manji kernel: [<ffffffff814a07a2>] radeon_ib_get+0x32/0xe0 Jan 12 00:25:40 manji kernel: [<ffffffff813f3ab0>] radeon_cs_ioctl+0x260/0x7a0 Jan 12 00:25:40 manji kernel: [<ffffffff8139504f>] drm_ioctl+0x18f/0x660 Jan 12 00:25:40 manji kernel: [<ffffffff8103158d>] ? __do_page_fault+0x14d/0x530 Jan 12 00:25:40 manji kernel: [<ffffffff813bb137>] radeon_drm_ioctl+0x47/0x80 Jan 12 00:25:40 manji kernel: [<ffffffff81163110>] do_vfs_ioctl+0x2e0/0x4e0 Jan 12 00:25:40 manji kernel: [<ffffffff81163391>] SyS_ioctl+0x81/0xa0 Jan 12 00:25:40 manji kernel: [<ffffffff8168a712>] system_call_fastpath+0x12/0x17 If you need any more information I'd be happy to provide.
does this still fail without all those cmdline parames shown in dmesg, particularly the ones related to radeon and forcing on ASPM?
I was unable to get it to fail again, even without changing those cmdline params. I am guessing it was a fluke maybe... or just really hard to reproduce. I will keep those params removed and will report back if it ever happens again. Thanks.
Disregard that (above comment), I don't think I was using the latest master because I had previously tried a reverting a commit and even though I checked out the 2 files it changed(thinking this alone undid the revert), I haven't noticed these git status messages: Your branch is behind 'origin/master' by 23 commits, and can be fast-forwarded. (use "git pull" to update your local branch) You are currently reverting commit 1b1f3e1. (all conflicts fixed: run "git revert --continue") (use "git revert --abort" to cancel the revert operation) Now that I aborted the revert and git pull-ed, will try to reproduce this and report back if I succeed!
I just realized that the behind 'origin/master' by 23 commits was because more commits were added after the rc4 git tag and my build automatically pulls latest master. So the revert that I did was likely correct the first time. Anyway since I cannot reproduce this anymore, I don't know what else to do.