Thanks for spotting this! This is a much simpler test case:
$ (ping www.ubuntu.com &); sleep 1; sudo pkill -SEGV ping; sleep 1; ls -l /var/crash/_bin_ping.*.crash -rw-r----- 1 martin whoopsie 57649 Okt 23 12:53 /var/crash/_bin_ping.1000.crash
Indeed most of /proc/<pid>/* are owned by root:root in this case, and "status" shows euid=0. I agree that we should make the .crash file owned by the effective instead of real UID.
Thanks for spotting this! This is a much simpler test case:
$ (ping www.ubuntu.com &); sleep 1; sudo pkill -SEGV ping; sleep 1; ls -l /var/crash/ _bin_ping. *.crash _bin_ping. 1000.crash
-rw-r----- 1 martin whoopsie 57649 Okt 23 12:53 /var/crash/
Indeed most of /proc/<pid>/* are owned by root:root in this case, and "status" shows euid=0. I agree that we should make the .crash file owned by the effective instead of real UID.