Hi Dave, I am going to respond to all three of your report in a one message so I do not overwhelm the mailing list unnecessarily)) Firstly, thank you for the reports and patches. The Bridge log one and also the RIP authentication one are new to us, so we will fix them once our devs have some time on their hands. The stack-buffer-overflow in ipX_getbit functions is a know issue, that is mostly harmless, but you are right that it should be fixed. Thanks for the catches and happy routing! David David Petera (he/him) | BIRD Tech Support | CZ.NIC, z.s.p.o. On 11/2/26 03:31, dave seddon wrote:
G'day,
Another small one from the static analysis run.
In kbr_check_iface(), the warning for a bridge without VLAN filtering passes the interface name, but its format string has no place for it:
log(L_WARN "%s: Mismatch in VLAN filtering", p->p.name, i->name);
So the name is silently dropped. The patch makes it match the warning just above it:
"%s: Interface %s has no VLAN filtering"
Based on master (4bee51c8): https://github.com/randomizedcoder/bird/tree/fix-bridge-vlan-log
Tip: 50c4d0e7021b73ba6bd816b44ee74bc47bb8c23f
90155552 Build: Let a test exclude an object from its link 058bdeb3 Bridge: Add tests of the bridge interface check 50c4d0e7 Bridge: Name the interface in the VLAN filtering warning
The first commit is the same as in the RIP patch I've just sent; whichever goes in second can drop it.
The new proto/bridge/bridge_test.c has 12 cases over kbr_check_iface() (up/down, bridge or not, VLAN filtering on either side, missing attributes, a 15 character name), and checks both the result and the line logged. The tests are committed first, passing against the current code; the fix flips the one case with the warning.
All test executables pass with ASan/UBSan, except filter_test and trie_test, which fail the same way on master and are fixed by the trie series. The bridge tests also pass on big-endian s390x under qemu-user.
LLM tools assisted with the analysis and patches.
-- Regards, Dave Seddon