Created attachment 1673842 [details] Dmesg log Description of problem: When i click on the menu grid icon, gnome-shell freezes and crash, then user session immediately is closed. When i restart the user session, When I restart the session, the ABRT notification appears stating that /usr/bin/gnome-shell crashes. I thought this error was associated with Mutter, so I disabled the animations, but these freezes persisted. Version-Release number of selected component (if applicable): gnome-shell.3.36.0 kernel 5.6.0-0.rc7 Steps to Reproduce: 1. Click on activities hot corner. 2. Next, click on the menu-grid icon. Actual results: Gnome-shell freezes and crashes. Additional info: In the dmesg log, the following error appears: gnome-shell[1500]: segfault at 8 ip 00007f089da6b66b sp 00007ffe67725850 error 4 in libmutter-cogl-6.so.0.0.0 (deleted)[7f089da4a000+54000]
Do you have a backtrace from the crash? I.e. one you see via coredumpctl? The journal log seems to contain a lot of errors from the radeon kernel driver.
(In reply to Jonas Ådahl from comment #1) > Do you have a backtrace from the crash? I.e. one you see via coredumpctl? > > The journal log seems to contain a lot of errors from the radeon kernel Yeap, but this not affect the system in basics tasks. I've already reported and barred a ticket for these errors. https://bugzilla.redhat.com/show_bug.cgi?id=1816461 With regard to bracktrace, it fails to debug (I'll try to fix it). At the moment I have little information here.
Created attachment 1673845 [details] coredumpctl-debug.txt
Created attachment 1673846 [details] coredumpctl-list.txt
Created attachment 1673861 [details] Bracktrace
Created attachment 1673862 [details] core_backtrace
Created attachment 1673863 [details] cpuinfo
Created attachment 1673864 [details] dso
Created attachment 1673865 [details] environ
Created attachment 1673866 [details] exploitable
Created attachment 1673867 [details] limits
Created attachment 1673868 [details] maps
Created attachment 1673869 [details] mountinfo
Created attachment 1673870 [details] open_fds
Created attachment 1673871 [details] proc_pid
Created attachment 1673872 [details] var_log_msg
When you mean you click on the menu grid you mean the one on the left after having opened the overview that makes the application icons show up? Does it happen every time you open it?
(In reply to Jonas Ådahl from comment #17) > When you mean you click on the menu grid you mean the one on the left after > having opened the overview that makes the application icons show up? yes. >Does it happen every time you open it? Yes, with the window composition active. With animations turned off it's kind of random.
Do you have any extensions enabled?
(In reply to Jonas Ådahl from comment #19) > Do you have any extensions enabled? No. 100% vanilla
Would you mind if I prepare a debug rpm of mutter for you to run, reproduce, then attach the result? I have yet to be able to reproduce this issue myself, and sadly the backtrace isn't much to go on.
No problem at all
Made two builds, one for mutter and one for gnome-shell. Installing them will result in *a lot* of spam from gnome-shell in the journal. After having installed, please log out, then log back in, reproduce the issue. Then log back in, look up the pid of the crashed gnome-shell process (use coredumpctl). Then run the command journalctl _PID=<pid> and replace <pid> with the PID of the gnome-shell that had crashed. The gnome-shell build is available at https://koji.fedoraproject.org/koji/taskinfo?taskID=42784199 and the mutter build at https://koji.fedoraproject.org/koji/taskinfo?taskID=42784238 . When you had enough of the journal spamming, you can revert back to the original state by running sudo dnf downgrade mutter gnome-shell
(In reply to Jonas Ådahl from comment #23) > .. Then run the command > > journalctl _PID=<pid> > To be more precis, run journalctl _PID=<pid> > gnome-shell.log and attach the gnome-shell.log file here.
(In reply to Jonas Ådahl from comment #24) > (In reply to Jonas Ådahl from comment #23) > > .. Then run the command > > > > journalctl _PID=<pid> > > > > To be more precis, run > > journalctl _PID=<pid> > gnome-shell.log > > and attach the gnome-shell.log file here. The file is quite big, so I'll give you a link to access the file >>https://drive.google.com/open?id=1Ow1t9JxIbKEl9AtYKZ9UIClMJDdGrvXz
(In reply to Gabriel Elyas from comment #25) > (In reply to Jonas Ådahl from comment #24) > > (In reply to Jonas Ådahl from comment #23) > > > .. Then run the command > > > > > > journalctl _PID=<pid> > > > > > > > To be more precis, run > > > > journalctl _PID=<pid> > gnome-shell.log > > > > and attach the gnome-shell.log file here. > > The file is quite big, so I'll give you a link to access the file > > >>https://drive.google.com/open?id=1Ow1t9JxIbKEl9AtYKZ9UIClMJDdGrvXz Thanks! Could you attach a new backtrace for the same pid as well? Best would be if you first open it using `coredumpctl gdb <pid>` then ran `bt full` in the opened gdb console.
So, it kind of got a problem, so I remade things. >> gnome-shell-NEW.log >> https://drive.google.com/open?id=1fUjPneBIFkAYu_96UEIyO8B2s6mx-dEB >> coredumpctl gdb <pid> (bt full) >> https://drive.google.com/open?id=1GVRhu-axBJvS0QqF8WLtHbXwV9SPpL3YF5wNJAY0fzM
(In reply to Gabriel Elyas from comment #27) > So, it kind of got a problem, so I remade things. > > > >> gnome-shell-NEW.log >> https://drive.google.com/open?id=1fUjPneBIFkAYu_96UEIyO8B2s6mx-dEB > > >> coredumpctl gdb <pid> (bt full) >> https://drive.google.com/open?id=1GVRhu-axBJvS0QqF8WLtHbXwV9SPpL3YF5wNJAY0fzM I hope this has not hindered you.
(In reply to Gabriel Elyas from comment #28) > (In reply to Gabriel Elyas from comment #27) > > So, it kind of got a problem, so I remade things. > > > > > > >> gnome-shell-NEW.log >> https://drive.google.com/open?id=1fUjPneBIFkAYu_96UEIyO8B2s6mx-dEB > > > > >> coredumpctl gdb <pid> (bt full) >> https://drive.google.com/open?id=1GVRhu-axBJvS0QqF8WLtHbXwV9SPpL3YF5wNJAY0fzM > > I hope this has not hindered you. Not at all, just got too late in the evening. For the "bt full", what you'd need to do is first coredumpctl gdb 49520 Then press enter until you see only (gdb) on a single line. After that, type "bt full" and press enter. Then press enter again until you see "(gdb)" on a single line again. Also, can you make sure you have debug info packages installed too? I.e. run "sudo dnf debuginfo-install mutter gnome-shell" before running the coredumpctl command.
BTW, I've created two new scratch builds that have more useful debug output. Could you replace the ones you installed with these? mutter: https://koji.fedoraproject.org/koji/taskinfo?taskID=42794720 gnome-shell: https://koji.fedoraproject.org/koji/taskinfo?taskID=42794722
(In reply to Jonas Ådahl from comment #30) > BTW, I've created two new scratch builds that have more useful debug output. > Could you replace the ones you installed with these? > > mutter: https://koji.fedoraproject.org/koji/taskinfo?taskID=42794720 > gnome-shell: https://koji.fedoraproject.org/koji/taskinfo?taskID=42794722 No problem. Here it is : >> gnome-shell-new^2.log >> https://drive.google.com/open?id=1klPwbW9V_2pnrqH4xqJjQnTfTDfdHr6F >> coredumpctl(bt full) >> https://drive.google.com/open?id=1djUEV74ilYJNdqid4dtJpFSQ9M_Va_s5-WX_IYin56s
Could you install the debuginfo RPMs from the two scratch builds too, then rerun the "coredumpctl" part?
(In reply to Jonas Ådahl from comment #32) > Could you install the debuginfo RPMs from the two scratch builds too, then > rerun the "coredumpctl" part? yeap. >>>>> gnome-shell-new^3.log >>>>> https://drive.google.com/open?id=1KqTW9acDF3pQ7bk5Un3vfaimoh9mB8tJ >>>>> coredumpctl_bt-full >>>>>>> https://drive.google.com/open?id=1YlNVB0r5d0eNApejvMHVDaIt97568SBNRvOSQTRwkF8
Here is a test build to see if the issue is resolved: https://koji.fedoraproject.org/koji/taskinfo?taskID=42872652 None of the other scratch builds should be installed.
Now everything is working properly. I didn't have any freezes or breaks.
(In reply to Gabriel Elyas from comment #35) > Now everything is working properly. I didn't have any freezes or breaks. Thanks !!
(In reply to Gabriel Elyas from comment #36) > (In reply to Gabriel Elyas from comment #35) > > Now everything is working properly. I didn't have any freezes or breaks. Thanks !! Thanks a lot for both helping out with debugging and testing the fix. I'm going to close this bug as a duplicate of bug 1813515, as it likely the same, with the same backtrace. I'm marking this as duplicate, as the other one get lots of duplicates from the automatic crash reporting system. *** This bug has been marked as a duplicate of bug 1813515 ***