[quagga-dev 3127] [PATCH] netlink fixes

Hasso Tepper hasso at estpak.ee
Fri Apr 8 12:52:50 BST 2005


Some things annoyed me some time already in netlink debug output.

1) IPv6 routes were not logged correctly.
2) Useless (wrong) info was logged with routes add/remove messages.
3) IFA_ADDRESS debug was logged in starnge format - eth0/24 10.10.10.1

Attached patch fixes this. It also fixes one unsigned/signed warning in 
rt_netlink.c. I'm not yet sure how to fix two more warnings in this file.

One question raised while testing patch though. Testing was done in laptop 
with 10.10.10.0/24 and dead:beaf::/64 networks in eth0. Two static routes 
were added - "ip route 10.30.30.0/24 10.10.10.1" and "ipv6 route 3ffe::/64 
dead:beaf::1". With this patch applied debug output:

2005/04/08 14:46:10 ZEBRA: netlink_route_multipath() (single hop): 
RTM_NEWROUTE 10.30.30.0/24, type IPv4 nexthop
2005/04/08 14:46:10 ZEBRA: netlink_route_multipath() (single hop): nexthop 
via 10.10.10.1 if 2

2005/04/08 14:46:45 ZEBRA: netlink_route_multipath() (single hop): 
RTM_NEWROUTE 3ffe::/64, type IPv6 nexthopIPv6 nexthop with ifindex
2005/04/08 14:46:45 ZEBRA: netlink_route_multipath() (single hop): nexthop 
via dead:beef::2 if 2

Why both "IPv6 nexthop" and "IPv6 nexthop with ifindex" in case of IPv6? 
Shouldn't IPv6 nexthop be enough in case of global nexthop?


-- 
Hasso Tepper
Elion Enterprises Ltd.
WAN administrator
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ht-20050408-rt_netlink.patch
Type: text/x-diff
Size: 14398 bytes
Desc: not available
URL: <http://lists.quagga.net/pipermail/quagga-dev/attachments/20050408/0921659a/attachment-0001.bin>


More information about the Quagga-dev mailing list