Fedora Account System
Red Hat Associate
Red Hat Customer
Version-Release number of selected component: ebtables-legacy-2.0.10-33.fc31 Additional info: reporter: libreport-2.10.0 backtrace_rating: 4 cmdline: /usr/sbin/ebtables --concurrent -L crash_function: dirname executable: /usr/sbin/ebtables-legacy journald_cursor: s=2e73494b77804b0ca188d8f6dbd0fd0d;i=17c15f4;b=d0e36c8337e441fbbac05f48f7adfaad;m=458d9716d2;t=58606054ffcc1;x=4133f70baf421e36 kernel: 5.1.0-0.rc3.git1.1.fc31.x86_64 rootdir: / runlevel: N 5 type: CCpp uid: 0 Truncated backtrace: Thread no. 1 (4 frames) #0 dirname at dirname.c:71 #1 lock_file at libebtc.c:148 #2 ebt_get_kernel_table at libebtc.c:168 #3 do_command at ebtables.c:719
Created attachment 1553697 [details] File: backtrace
Created attachment 1553698 [details] File: cgroup
Created attachment 1553699 [details] File: core_backtrace
Created attachment 1553700 [details] File: cpuinfo
Created attachment 1553701 [details] File: dso_list
Created attachment 1553702 [details] File: environ
Created attachment 1553703 [details] File: exploitable
Created attachment 1553704 [details] File: limits
Created attachment 1553705 [details] File: maps
Created attachment 1553706 [details] File: mountinfo
Created attachment 1553707 [details] File: open_fds
Created attachment 1553708 [details] File: proc_pid_status
Observed this too.
Problem caused by a combination of two things: * Bug in lockfile path customization support (I broke upstream). * Upstream default lockfile path in build which doesn't exist in Fedora. Fix sent upstream, Fedora received a new build which restores the old lockfile location (/run/ebtables.lock) - this effectively avoids the crash, so no need to wait for upstream fix.