[opensuse] X + keyboard freeze, sysrq fun
Been a while, but I just got hit with the "X, keyboard and mouse all freeze at once" - no getting to another VT. But I could ssh in from another machine. As always when this happens, nothing useable in any logs. Trying to `sudo kill -9 pidof X`, etc.. did absolutely nothing. After a reboot, I went googl'ing for sysrq, which led me to discover that (a) my keyboard is not mapped for it (fixable), (b) /proc/sysrq-trigger would be just fine over ssh and (c) I can't figure out where output of sysrq's like 'l' (backtrace for all CPUS) goes - if only so that I can tell I've got things set up properly. The web says "goes to console", but on Leap, where the heck is that? If it is hidden by X on 1, is there a way to get it into the journal (journalctl)? So, I hereby solicit advice on: 1) clever ways to kill things that are locking up my system from an ssh session 2) clever ways to use /proc/sysrq-trigger, esp. if I can figure out if it is even working... 3) Debugging any of the above. After a dozen of these over the past 5 years or so, with no trace left in logs, I've kind of given up, but hey. TIA Michael -- Michael Fischer michael@visv.net -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Sunday, 2018-02-11 at 00:08 -0500, Michael Fischer wrote:
Been a while, but I just got hit with the "X, keyboard and mouse all freeze at once" - no getting to another VT. But I could ssh in from another machine. As always when this happens, nothing useable in any logs. Trying to `sudo kill -9 pidof X`, etc.. did absolutely nothing.
Maybe you can order the system to go to virtual console one over the ssh: # chvt 1
After a reboot, I went googl'ing for sysrq, which led me to discover that (a) my keyboard is not mapped for it (fixable), (b) /proc/sysrq-trigger would be just fine over ssh and (c) I can't figure out where output of sysrq's like 'l' (backtrace for all CPUS) goes - if only so that I can tell I've got things set up properly. The web says "goes to console", but on Leap, where the heck is that? If it is hidden by X on 1, is there a way to get it into the journal (journalctl)?
So, I hereby solicit advice on:
1) clever ways to kill things that are locking up my system from an ssh session
Did you try "init 3"? - -- Cheers, Carlos E. R. (from openSUSE 42.3 x86_64 "Malachite" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAlqAA1AACgkQtTMYHG2NR9U93gCggTIUSowFx93B99QerrHPtRal kWsAoJjtkibATPmZ9LE77aWfCYvFdO2z =EwCk -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
* Michael Fischer <michael@visv.net> [02-11-18 00:10]:
Been a while, but I just got hit with the "X, keyboard and mouse all freeze at once" - no getting to another VT. But I could ssh in from another machine. As always when this happens, nothing useable in any logs. Trying to `sudo kill -9 pidof X`, etc.. did absolutely nothing.
as root: systemctl isolate multi-user or init 3 -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Registered Linux User #207535 @ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Sun, Feb 11, Patrick Shanahan wrote:
* Michael Fischer <michael@visv.net> [02-11-18 00:10]:
Been a while, but I just got hit with the "X, keyboard and mouse all freeze at once" - no getting to another VT. But I could ssh in from another machine. As always when this happens, nothing useable in any logs. Trying to `sudo kill -9 pidof X`, etc.. did absolutely nothing.
as root: systemctl isolate multi-user or init 3
Replying to both Patrick and Carlos: I run my systems from 3, then manually run `startx`. What would be the effect of `init 3` or its systemd equivalent in this case? Of course, I did try this over ssh, doing `init 1`, with the predictable comic result... "oh...yeah, oops". What does the CTRL-ALT-BACKSPACE, BACKSPACE sequence actually do "under the hood"? If my keyboard were responsive, I'd have tried that. Thanks. Michael -- Michael Fischer michael@visv.net -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
* Michael Fischer <michael@visv.net> [02-11-18 11:40]:
On Sun, Feb 11, Patrick Shanahan wrote:
* Michael Fischer <michael@visv.net> [02-11-18 00:10]:
Been a while, but I just got hit with the "X, keyboard and mouse all freeze at once" - no getting to another VT. But I could ssh in from another machine. As always when this happens, nothing useable in any logs.
always happens, you have made a bug report? what logs, what video card, what openSUSE version, odd repos
Trying to `sudo kill -9 pidof X`, etc.. did absolutely nothing.
as root: systemctl isolate multi-user or init 3
Replying to both Patrick and Carlos: I run my systems from 3, then manually run `startx`. What would be the effect of `init 3` or its systemd equivalent in this case?
none, but, somehow you failed to mention that pertinent fact. and now "init 3" is an equivalent for "systemctl isolate multi-user", not the reverse did your "kill" operation actually kill a process? what ",etc" was tried? and i "assume" 'did absolutely nothing' actually means it didn't do what you expected or ...
Of course, I did try this over ssh, doing `init 1`, with the predictable comic result... "oh...yeah, oops".
yeah, no network :)
What does the CTRL-ALT-BACKSPACE, BACKSPACE sequence actually do "under the hood"? If my keyboard were responsive, I'd have tried that.
are you asking and answering or is the quoting obscured? is your keyboard freezing causing the problem? is your mouse still active? you ask a lot but provide very little information :( -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Registered Linux User #207535 @ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Sun, Feb 11, Patrick Shanahan wrote:
* Michael Fischer <michael@visv.net> [02-11-18 11:40]:
On Sun, Feb 11, Patrick Shanahan wrote:
* Michael Fischer <michael@visv.net> [02-11-18 00:10]:
Been a while, but I just got hit with the "X, keyboard and mouse all freeze at once" - no getting to another VT. But I could ssh in from another machine. As always when this happens, nothing useable in any logs.
always happens, you have made a bug report?
what logs, what video card, what openSUSE version, odd repos
It happens say with a frequency of say, once every 5 or 6 months over 5 years. Has not been enough to provoke me to a bug report. Would also cover 3 different machines, with different graphics cards, drivers, kernels, SuSE versions, etc.. This is also why I didn't bother to provide logs, etc. as there is too much history and variety to pin it down to the current setup. I was kind of hoping someone knew a good trick for coping with this sort of problem, short of a reboot. In case it is of use, the current setup is: Model: "ATI Ellesmere [Radeon RX 470/480]" Driver: "amdgpu" SuSE Leap 42.3 kernel 4.14.6-1.g45f120a-default Logs I examined were ~/.xsession-errors /var/log/Xorg.0.log `journalctl` which, besides xsession-errors being full of very repetetive browser junk, told me nothing of any error conditions.
Trying to `sudo kill -9 pidof X`, etc.. did absolutely nothing.
as root: systemctl isolate multi-user or init 3
Replying to both Patrick and Carlos: I run my systems from 3, then manually run `startx`. What would be the effect of `init 3` or its systemd equivalent in this case?
none, but, somehow you failed to mention that pertinent fact. and now "init 3" is an equivalent for "systemctl isolate multi-user", not the reverse
Sorry, and understood about systemd back-compat commands. And thank you for clarifying that `init 3` to a system brought up to that runlevel in the first place would have no effect.
did your "kill" operation actually kill a process? what ",etc" was tried? and i "assume" 'did absolutely nothing' actually means it didn't do what you expected or ...
`kill -9 pidof X` didn't kill X. Screen didn't change, and over the ssh session, `ps` showed X still there, with the same pid. This is what I meant by "did abosolutely nothing". As ~/.xsession-errors was full of noise about browser problems, I killed firefox and chrome processes, which did work (pids no longer in `ps`, sound cut out from the movie I was watching in chrome), but the screen did not update (i.e. the browser windows were still there, unchanged).
What does the CTRL-ALT-BACKSPACE, BACKSPACE sequence actually do "under the hood"? If my keyboard were responsive, I'd have tried that.
are you asking and answering or is the quoting obscured?
Hmm. Asking. Sorry if that somehow came out with confusing quoting.
is your keyboard freezing causing the problem? is your mouse still active?
No idea if keyboard freezing is *causing* the problem. To me it seems like more of a symptom. As stated in the original email, each of the screen, mouse and keyboard freeze together. There is simply no interactivity whatsoever.
you ask a lot but provide very little information :(
Sorry, but the system gave me very little information (logs, etc.) to pass along. I appreciate you trying in spite of the low level of such information. Michael -- Michael Fischer michael@visv.net -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Sunday, 2018-02-11 at 12:26 -0500, Michael Fischer wrote:
On Sun, Feb 11, Patrick Shanahan wrote:
did your "kill" operation actually kill a process? what ",etc" was tried? and i "assume" 'did absolutely nothing' actually means it didn't do what you expected or ...
`kill -9 pidof X` didn't kill X. Screen didn't change, and over the ssh session, `ps` showed X still there, with the same pid. This is what I meant by "did abosolutely nothing".
Maybe the process went "zombie".
As ~/.xsession-errors was full of noise about browser problems, I killed firefox and chrome processes, which did work (pids no longer in `ps`, sound cut out from the movie I was watching in chrome), but the screen did not update (i.e. the browser windows were still there, unchanged).
Yes, that happens till the X process really dies. And -9 means it doesn't get a chance to restore the screen. I suggest using "ps afxu | less -S" to find the hierarchy of processes and kill the parents. But if the video driver is hung badly... maybe nothing restores it.
What does the CTRL-ALT-BACKSPACE, BACKSPACE sequence actually do "under the hood"? If my keyboard were responsive, I'd have tried that.
are you asking and answering or is the quoting obscured?
Hmm. Asking. Sorry if that somehow came out with confusing quoting.
is your keyboard freezing causing the problem? is your mouse still active?
No idea if keyboard freezing is *causing* the problem. To me it seems like more of a symptom. As stated in the original email, each of the screen, mouse and keyboard freeze together. There is simply no interactivity whatsoever.
Yes, it is a symptom. IMO :-) - -- Cheers, Carlos E. R. (from openSUSE 42.3 x86_64 "Malachite" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAlqAzFoACgkQtTMYHG2NR9VdjACfSE6d3F7a4jtkuuqLXA6Gxrmn hjcAn2ePsVDp323sRCEns+XeUih1O0rb =gxv6 -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Mon, Feb 12, Carlos E. R. wrote:
As ~/.xsession-errors was full of noise about browser problems, I killed firefox and chrome processes, which did work (pids no longer in `ps`, sound cut out from the movie I was watching in chrome), but the screen did not update (i.e. the browser windows were still there, unchanged).
Yes, that happens till the X process really dies. And -9 means it doesn't get a chance to restore the screen.
I suggest using "ps afxu | less -S" to find the hierarchy of processes and kill the parents.
But if the video driver is hung badly... maybe nothing restores it.
Yes, this is what I'm afraid of. Maybe next time I should try an `rmmod` - `modprobe` cycle.
No idea if keyboard freezing is *causing* the problem. To me it seems like more of a symptom. As stated in the original email, each of the screen, mouse and keyboard freeze together. There is simply no interactivity whatsoever.
Yes, it is a symptom. IMO :-)
How would this play with the theory that the video driver is hung? i.e., I hit CTRL-ALT-F2 and the machine "thinks" it gave me VT2, but I can't tell because the screen doesn't update? But if I typed my login, etc.. blind, I could get a terminal and send commands without having to ssh....sorry, brain hurting now. Thanks for the suggestions. Michael -- Michael Fischer michael@visv.net -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
* Michael Fischer <michael@visv.net> [02-14-18 12:21]:
On Mon, Feb 12, Carlos E. R. wrote:
As ~/.xsession-errors was full of noise about browser problems, I killed firefox and chrome processes, which did work (pids no longer in `ps`, sound cut out from the movie I was watching in chrome), but the screen did not update (i.e. the browser windows were still there, unchanged).
Yes, that happens till the X process really dies. And -9 means it doesn't get a chance to restore the screen.
I suggest using "ps afxu | less -S" to find the hierarchy of processes and kill the parents.
But if the video driver is hung badly... maybe nothing restores it.
Yes, this is what I'm afraid of. Maybe next time I should try an `rmmod` - `modprobe` cycle.
No idea if keyboard freezing is *causing* the problem. To me it seems like more of a symptom. As stated in the original email, each of the screen, mouse and keyboard freeze together. There is simply no interactivity whatsoever.
Yes, it is a symptom. IMO :-)
How would this play with the theory that the video driver is hung? i.e., I hit CTRL-ALT-F2 and the machine "thinks" it gave me VT2, but I can't tell because the screen doesn't update? But if I typed my login, etc.. blind, I could get a terminal and send commands without having to ssh....sorry, brain hurting now.
what video card and what driver? -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Registered Linux User #207535 @ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Wed, Feb 14, Patrick Shanahan wrote:
what video card and what driver?
Given upthread: Model: "ATI Ellesmere [Radeon RX 470/480]" Driver: "amdgpu" SuSE Leap 42.3 kernel 4.14.6-1.g45f120a-default Michael -- Michael Fischer michael@visv.net -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 11/02/18 17:26, Michael Fischer wrote:
is your mouse still active? No idea if keyboard freezing is *causing* the problem. To me it seems
is your keyboard freezing causing the problem? like more of a symptom. As stated in the original email, each of the screen, mouse and keyboard freeze together. There is simply no interactivity whatsoever.
you ask a lot but provide very little information :(
Sorry, but the system gave me very little information (logs, etc.) to pass along. I appreciate you trying in spite of the low level of such information.
Are you running plasma? Well, obviously your old systems weren't ... My system (Toshiba laptop) I have exactly that problem - it locks up and I end up pressing the power button to reset it. BUT. That video bug mentioned here seems - the fix for that seems noticeably to have helped. My problems started when I upgraded from 42.2 to 42.3 iirc. If I leave the system long enough, it seems to come back. I run xosview and that has the load go through the roof for no obvious reason every now and then, which knackers response. I blamed it on the network, but now I actually suspect that whatever the real problem is starves the network of cpu, which causes the link to crash, which causes ... My GUESS as to the problem is that something is overloading X, which is why you don't get any logs or anything like that - part of the system has just got itself into a tight do-nothing loop that is chewing up all the resources ... Cheers, Wol -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Wed, Feb 14, Wols Lists wrote:
Are you running plasma? Well, obviously your old systems weren't ...
No. FVWM.
My GUESS as to the problem is that something is overloading X, which is why you don't get any logs or anything like that - part of the system has just got itself into a tight do-nothing loop that is chewing up all the resources ...
`top` showed a load of 1.2. Admittedly that doesn't necessarily reflect the GPU... Michael -- Michael Fischer michael@visv.net -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Sunday, 2018-02-11 at 11:38 -0500, Michael Fischer wrote:
On Sun, Feb 11, Patrick Shanahan wrote:
* Michael Fischer <michael@visv.net> [02-11-18 00:10]:
Been a while, but I just got hit with the "X, keyboard and mouse all freeze at once" - no getting to another VT. But I could ssh in from another machine. As always when this happens, nothing useable in any logs. Trying to `sudo kill -9 pidof X`, etc.. did absolutely nothing.
as root: systemctl isolate multi-user or init 3
Replying to both Patrick and Carlos: I run my systems from 3, then manually run `startx`. What would be the effect of `init 3` or its systemd equivalent in this case?
Sorry, we did not know that, you didn't say, and it is atypical (and not reccomended, in fact).
Of course, I did try this over ssh, doing `init 1`, with the predictable comic result... "oh...yeah, oops".
What does the CTRL-ALT-BACKSPACE, BACKSPACE sequence actually do "under the hood"? If my keyboard were responsive, I'd have tried that.
Kills suddenly the graphical session entirely and forcibly. I don't know more detail than that. - -- Cheers, Carlos E. R. (from openSUSE 42.3 x86_64 "Malachite" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iEYEARECAAYFAlqAyjUACgkQtTMYHG2NR9UixACeL5IpTICwzWSF2nIbaLkyCH/j S5QAnRj1aU2X55wCKye9jJ3wOpSsMPzA =J23H -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 12/02/18 00:56, Carlos E. R. wrote:
What does the CTRL-ALT-BACKSPACE, BACKSPACE sequence actually do "under the hood"? If my keyboard were responsive, I'd have tried that.
If it's a usb keyboard you can try unplugging it and plugging it back in again, sometimes it works. Dave P -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Sun, Feb 11, Carlos E. R. wrote:
Replying to both Patrick and Carlos: I run my systems from 3, then manually run `startx`. What would be the effect of `init 3` or its systemd equivalent in this case?
Sorry, we did not know that, you didn't say, and it is atypical (and not reccomended, in fact).
Why is it not recommended? Ok, rephrase - is it actively *discouraged*? And if so, why? Thanks. Michael -- Michael Fischer michael@visv.net -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 El 2018-02-14 a las 12:21 -0500, Michael Fischer escribió:
On Sun, Feb 11, Carlos E. R. wrote:
Replying to both Patrick and Carlos: I run my systems from 3, then manually run `startx`. What would be the effect of `init 3` or its systemd equivalent in this case?
Sorry, we did not know that, you didn't say, and it is atypical (and not reccomended, in fact).
Why is it not recommended? Ok, rephrase - is it actively *discouraged*? And if so, why?
Well, startx is not maintained. Or not enough. Some toolchains are not aware that you have the seat, and thus you may not get permission to use sound or the cdrom. I use it only for diagnostics when X simply doesn't start, like a video problem. - -- Cheers Carlos E. R. (from openSUSE 42.3 x86_64 "Malachite" (Minas Tirith)) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iF4EAREIAAYFAlqF4VAACgkQja8UbcUWM1zq+gD+I6/yZOqvs1HXbHFTHlgXaDNy oEKxHvvi9fpHDwsGJboBAJB8bgYDi8zAFwwULFv1q8t8WwE1e1UUFCGGK/v2aLSF =I11Y -----END PGP SIGNATURE-----
Hello, On Sun, 11 Feb 2018, Michael Fischer wrote:
What does the CTRL-ALT-BACKSPACE, BACKSPACE sequence actually do "under the hood"? If my keyboard were responsive, I'd have tried that.
==== man 5 xorg.conf ==== Option "DontZap" "boolean" This disallows the use of the Terminate_Server XKB action (usu- ally on Ctrl+Alt+Backspace, depending on XKB options). This action is normally used to terminate the Xorg server. When this option is enabled, the action has no effect. Default: off. ==== man 1 xorg ==== Ctrl+Alt+Backspace Immediately kills the server -- no questions asked. It can be disabled by setting the DontZap/ZapWarning xorg.conf(5) file option to a TRUE value. It should be noted that zapping is triggered by the Termi- nate_Server action in the keyboard map. This action is not part of the default keymaps but can be enabled with the XKB option "terminate:ctrl_alt_bksp". ==== HTH, -dnh --
[internet destroyer] I want an Internet Battleship. Or an Internet Dreadnought. Oh, and an Internet SSBN. -- Niklas Karlsson Don't we all, but instead we end up with junks. -- Peter Corlett
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Thu, Feb 15, David Haller wrote:
Hello,
On Sun, 11 Feb 2018, Michael Fischer wrote:
What does the CTRL-ALT-BACKSPACE, BACKSPACE sequence actually do "under the hood"? If my keyboard were responsive, I'd have tried that.
==== man 5 xorg.conf ==== Option "DontZap" "boolean" This disallows the use of the Terminate_Server XKB action (usu- ally on Ctrl+Alt+Backspace, depending on XKB options). This action is normally used to terminate the Xorg server. When this option is enabled, the action has no effect. Default: off. ==== man 1 xorg ==== Ctrl+Alt+Backspace Immediately kills the server -- no questions asked. It can be disabled by setting the DontZap/ZapWarning xorg.conf(5) file option to a TRUE value.
It should be noted that zapping is triggered by the Termi- nate_Server action in the keyboard map. This action is not part of the default keymaps but can be enabled with the XKB option "terminate:ctrl_alt_bksp". ====
Hmm, thanks for that. Michael -- Michael Fischer michael@visv.net -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
participants (6)
-
Carlos E. R. -
Dave Plater -
David Haller -
Michael Fischer -
Patrick Shanahan -
Wols Lists