I'm testing a fix right now that inits (and frees ) AFI_ETHER tables...
On 9/5/2016 11:46 AM, Paul Jakma wrote: > On Mon, 5 Sep 2016, Paul Jakma wrote: > >> Also, if I go back to the commit before, i.e. 'bgpd, zebra: Use next >> hop tracking for connected routes too' plus 'bgpd: Remove unused and >> leaking code' effectively, then it passes Martin's test but 'show ip >> nexthop' can still crash it: >> ==16194== Process terminating with default action of signal 6 (SIGABRT) >> ==16194== at 0x5A1B6F5: raise (in /usr/lib64/libc-2.23.so) >> ==16194== by 0x5A1D2F9: abort (in /usr/lib64/libc-2.23.so) >> ==16194== by 0x4E70EC0: core_handler (sigevent.c:242) >> ==16194== by 0x5A1B76F: ??? (in /usr/lib64/libc-2.23.so) >> ==16194== by 0x43A7C5: bgp_table_top (bgp_table.h:160) >> ==16194== by 0x43A7C5: show_ip_bgp_nexthop_table (bgp_nexthop.c:424) > This one is trivial it seems, just a missing null check in the loop in > the command. > > regards, _______________________________________________ Quagga-dev mailing list Quagga-dev@lists.quagga.net https://lists.quagga.net/mailman/listinfo/quagga-dev