Hi Neale,

Thanks for response.  On a quick tangent, your comment about unnecessary work 
got me thinking about these lines at the top of ip4_local_check_src():

vnet_buffer (b)->ip.fib_index =
    vnet_buffer (b)->sw_if_index[VLIB_TX] != ~0 ?
    vnet_buffer (b)->sw_if_index[VLIB_TX] : vnet_buffer (b)->ip.fib_index;

Wouldn't these be unnecessary given that ip_lookup_set_buffer_fib_index() does 
this already?  I guess it depends on what assumptions ip4-local, and 
ip4_local_check_src() specifically, are allowed to make.

Regarding the patch - My initial thought is to call 
ip_lookup_set_buffer_fib_index() directly (from ip4-options) rather than 
copy/paste the two fib_index lines.  If you prefer not using that function let 
me know.

Regarding test_igmp.py - My interface is not in a non-default VRF.  Triggering 
this issue requires setting the fib_index metadata to something invalid 
(recommend ~0) and then reusing that buffer for an IGMP with router alert.

Thanks,
Jeff
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#10860): https://lists.fd.io/g/vpp-dev/message/10860
Mute This Topic: https://lists.fd.io/mt/27373338/21656
Group Owner: [email protected]
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to