Description of problem: After the upgrade F36 -> F37-> F38, openbox start to crash randomly, closing the session and getting back to the display manager. Unfortunately I do not have a clear reproducer as the trigger is changing all the time. Following actions triggers the crash randomly: * switch between windows * open new window (e.g. start firefox) * highlight text * move a window between displays * switch between workspaces I've tried to limit the number of running application to minimum to reduce the noise for the investigation (e.g. tint2 has been oftenly visible in crash logs, so I stopped using it, still seeing the crashes). Also I put selinux into permissive mode, without any change around the original issue. Version-Release number of selected component (if applicable): * openbox-3.6.1-21.fc38.x86_64 Additional info: * Up-to-date F38 * context menu does not appear in terminal (xterm, terminator) after clicking right mouse button * Display manager: lightdm * 3 active displays * Oftenly can be visible the following error log, which is sometimes separated by time from crashes, so I guess it's not relevant: pam_systemd(login:session): Failed to release session: Access denied * Despite the changes in the system, abrt always produce the same crashdump, removing other crashes as duplicates. Today, hit 19 chrashes in ~5h. Openbox crashes usually in ~2min, but I had lucky ~2h session without issues. * Most of the memory is free (64 GB RAM, ~4 GB used) * systemd-oomd is disabled & masked Additional data put in attachments
Created attachment 1950731 [details] journalctl-snippet.log
Created attachment 1950743 [details] xsession-errors after the crash
Created attachment 1950744 [details] core_backtrace
Created attachment 1950745 [details] coredump.tgz
Created attachment 1952708 [details] list of downgraded rpms I've tried to downgrade number of packages without the success to fix the bug - the list of downgraded packages from the history of dnf transactions in attachment. I tried to downgrade more packages but I've started to hit various additional problems and I had not enough time to continue with experiments in more details, so sharing at least this. (e.g. I haven't tried to downgrade lightdm - just realized now..) Currently I do not have the system to reproduce the issue as it seems I successfully downgraded back to F37 - I am actually surprised the system survived that and seems to be working as before, without seeing unusual errors. When I have a time, I will try a clean installation to see whether the same is happening on a clean F38 system.
*** Bug 2190035 has been marked as a duplicate of this bug. ***
The core backtrace { "address": 94693188868051 , "build_id": "07cc0b5ced9a6f556aace94d1c24ce80901e98c7" , "build_id_offset": 124883 , "function_name": "client_calc_layer" , "file_name": "/usr/bin/openbox" } is exactly the one from bug #2190055 and https://retrace.fedoraproject.org/faf/reports/648708/ I've left a patch which works for me in the other bug.
*** Bug 2190055 has been marked as a duplicate of this bug. ***
There is an upstream bug report: https://bugzilla.icculus.org/show_bug.cgi?id=6669 It also has a proposed patch, slightly smaller than the one in our bug #2190055. It seems it wasn't accepted yet (or I don't know which is the right upstream repository). If you can verify that it fixes the issue for you, I can add it to our package.
Had a look at the one from https://bugs.archlinux.org/task/77853 and https://bugzilla-attachments.icculus.org/attachment.cgi?id=3646 and it copies the full list before iterating. The one with the loop in https://bbs.archlinux.org/viewtopic.php?id=284299 is too intransparent for me (I can't prove adhoc that it will terminate). My version only copies part of the list, not the full list, so it might be ever so slightly faster and needs less memory. However I'm not sure it is NULL-safe around freeing the list. The documenation for g_list_free does not seem to specify. However, I've been using my code for a day, seems ok. The one in https://bugzilla-attachments.icculus.org/attachment.cgi?id=3647 "repairs" the list after corrupting it. I can't prove adhoc that the code would be fine in all cases. Taking the patch from the arch tracker should be good I guess.
FEDORA-2023-6bf51a3399 has been submitted as an update to Fedora 38. https://bodhi.fedoraproject.org/updates/FEDORA-2023-6bf51a3399
FEDORA-2023-6bf51a3399 has been pushed to the Fedora 38 testing repository. Soon you'll be able to install the update with the following command: `sudo dnf upgrade --enablerepo=updates-testing --refresh --advisory=FEDORA-2023-6bf51a3399` You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2023-6bf51a3399 See also https://fedoraproject.org/wiki/QA:Updates_Testing for more information on how to test updates.
*** Bug 2192188 has been marked as a duplicate of this bug. ***
FEDORA-2023-6bf51a3399 has been pushed to the Fedora 38 stable repository. If problem still persists, please make note of it in this bug report.