[Bug 904097] New: Use of more than one VT produces silent system crash
http://bugzilla.opensuse.org/show_bug.cgi?id=904097 Bug ID: 904097 Summary: Use of more than one VT produces silent system crash Classification: openSUSE Product: openSUSE Distribution Version: 13.2 Hardware: x86-64 OS: openSUSE 13.2 Status: NEW Severity: Critical Priority: P5 - None Component: Other Assignee: bnc-team-screening@forge.provo.novell.com Reporter: stakanov@freenet.de QA Contact: qa-bugs@suse.de Found By: --- Blocker: --- Lenovo X201. Create two or more users in kde. Work in one. Leave the other(s) open. Wait for a while (like watch a TG in stream) or read in the browser. You move the mouse all normal. You do: change VT with alt+ctrl+Fx (following my settings that is F7 and F8 with two users etc. So switch from the VT where you are to the other. Result: immediate, repeatable, stable - the system crashes silently. Just a white cursor on the left high corner of the VGA screen, HDMI is constant off. And even this only if you pull the hdmi cable and plug it in again or if you detach the device from the ultra base docking station. No key can wake up the machine. The ventilator continuous to run at speed as before. If you reset the machine no reaction. Closing lid and reopen: no reaction. Alt-ctrl-canc: no reaction. You have to hard reset with the power button. Then you will find that of course data has gone, programs have not been shut down correctly (i.e. firefox) so it did crash. I have no clue where the error could be found. So if you need logs..tell me what to look for. There are some ACPI messages in /var/log/messages. But now evident warning or something that I could recover as crash log. This is a very crippling bug. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #1 from Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #2 from Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Bernhard Wiedemann
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Bernhard Wiedemann
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Stefan Dirsch
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #4 from Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #5 from Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #6 from Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #9 from Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #10 from Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #12 from Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #13 from Takashi Iwai
Created attachment 612885 [details] output of var/log/messages (appears from an event tiggered with keyboard
This is from the /var/log/messages at the moment of screen crash and the machine staying "black screen dead but alive" afterwards up to reboot. I did cut when the time stamp changed after reboot.
The output is cut off, and doesn't contain the important bits, unfortunately. Maybe you can try alt-sysrq-w instead of alt-sysrq-t.
Second attachment is from the systemd command you gave me. Finally FYI, I tried now the kernel from stable, 3.17.2 and freezes and crashing of screen with VT seems to have stopped. I updated kernel firmware too. I will be able to confirm after trying a bit. However, if you see something that allows to fix a bug, I am more then available to drop back to original kernel and crash the poor beast until it asks for mercy. You just tell what you need, I do.
OK, could you clarify which kernel version caused the problem exactly? The rpm -qi shows the git commit ID, too. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #14 from Takashi Iwai
Created attachment 612886 [details] output of journalctl (example for sata freeze)
This seems to have stopped too after kernel update. Kernel was naturally 3.16.6-2
This looks irrelevant frm this bug itself. If it matters, please open another bug report. Thanks. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #15 from Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #17 from Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #18 from Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #19 from Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #20 from Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #23 from Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Bruno Pesavento
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #25 from Bruno Pesavento
BTW: with 3.17 this function is sound. So this problem came up between the two versions, or this was introduced with 3.16.x and then was corrected in 3.17.x
Sorry, apparently I missed comments 21 to 23. Maybe I witnessed something different, discard my notes. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #26 from Takashi Iwai
(In reply to Stakanov Schufter from comment #23)
BTW: with 3.17 this function is sound. So this problem came up between the two versions, or this was introduced with 3.16.x and then was corrected in 3.17.x
Sorry, apparently I missed comments 21 to 23. Maybe I witnessed something different, discard my notes.
In your case, the problem might be irrelevant with the stall in page flip. As you already noticed, there was a crash of X server. This might blocking the further usage of graphics and consoles. Did you try to remote-login while testing it? I guess the system is still alive and remotely available but no graphics and VT are controllable. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #27 from Bruno Pesavento
(In reply to Bruno Pesavento from comment #25)
In your case, the problem might be irrelevant with the stall in page flip. As you already noticed, there was a crash of X server. This might blocking the further usage of graphics and consoles.
Did you try to remote-login while testing it? I guess the system is still alive and remotely available but no graphics and VT are controllable.
Thanks for your clues, definitely an Xorg issue unrelated to page flip. Updating to the latest xorg-server and booting with safe settings fixed the problem. I'll join/open another bug report if I find something useful. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #29 from Stakanov Schufter
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Takashi Iwai
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #31 from Bruno Pesavento
Yes, could you try the latest KMP again? This contains a fix of hanging drm_read(). See the changelog of rpm whether it contains today's change.
There are different causes leading to X stall. One is the vblank page flip hang in i915 driver and another is some race of drm_read and stall of X. Let's see what you're seeing the second one I fixed today...
This latest KMP adds stability to a fix for the crash described in comment #27, likely thanks to the drm_read fix. That #27 problem with 965GM is fixed by explicitly loading the dri2 module _before_ glx gets loaded, or disabling dri entirely. But without your last KMP, Xorg still crashes occasionally on VT switch. I got similar stability with kernel 3.17.6, maybe because it includes a similar patch in i915.ko? -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
--- Comment #33 from Bruno Pesavento
http://bugzilla.opensuse.org/show_bug.cgi?id=904097
Takashi Iwai
participants (1)
-
bugzilla_noreply@novell.com