> > > generally, if outgoing interface decision is changed from lo0 after
> > > udp_output check it, it can be a problem.
> > 
> > Isn't this the case for all hardware-accelerated checkhsums?
> 
> yes.
> only difference is, for lo checksum omission,
> the decision is done a little earlier.

So, if I had a rule in ipf.conf:

pass out on ex0 (...) to fxp0

whch should cause the matching packets to leave through fxp0 instead of
ex0, and hw checksums enabled on ex0 and disabled on fxp0, they would go
out on fxp0 with wrong checksums?

Bye	Pavel