On 2026/08/19 14:28, [email protected] wrote: > Le Sat, Aug 15, 2026 at 12:37:50PM -0400, Ronald Dahlgren a écrit : > > Hello ports, > > > > librewolf has been segfaulting after profile selection for about a > > week now (maybe 10 days?). The behavior started after a `sysupgrade > > -s` and has not been resolved with subsequent updates, so I thought > > I'd report here. The first time I try to start it, the profile > > selection screen appears. I choose a profile and a moment later it > > segfaults. On subsequent runs, I can select the profile and the main > > browser window will appear but all tabs have the "Gah. Your tab just > > crashed." content. Even `about:blank` shows the same "gah" message. > > `about:config` loads. > > > > If I select another profile, it will core dump the first time, and > > then subsequently open the browser window with "Gah. Your tab just > > crashed." messages. > > > > If I create a new profile, it will open the browser window with "Gah. > > Your tab just crashed." messages. There is _some_ sort of interaction > > with profiles here, but I can't tell if it's meaningful or not. > > > > It drops a core, but there isn't much helpful here: > > this hints at pledge violations, maybe in a child process, refer to the > pkg-readme for the steps to provide more useful information (eg use > egdb, install debug packages, learn to decipher ktrace...) > > > ATTENTION: default value of option mesa_glthread overridden by environment. > > librewolf[69235]: pledge "proc", syscall 195 > > librewolf[13877]: pledge "proc", syscall 195 > > ATTENTION: default value of option mesa_glthread overridden by environment. > > librewolf[57614]: pledge "proc", syscall 195 > > librewolf[6793]: pledge "proc", syscall 195 >
195 is setrlimit(). try running under ktrace -di and showing a few screens full of kdump output from a crashing pid before the crash (e.g. kdump -p 69235 | tail -300 or something if the above run had been done under ktrace). also if you run "while true; do ps axwww|grep [l]ibrewolf >> /tmp/ps; done" while triggering it, you might catch the process arguments that might let you figure out what type of process the crashing pid is.
