On Tue, Jan 31, 2006 at 12:14:53AM -0800, Darren Reed wrote:
> > The IPsec team has an unfunded project to get rid of remaining STREAMS
> > infrastucture related to IPsec (we were waiting for FireEngine, plus we do
> > have other fish to fry). I believe Surya will lay the groundwork for a LOT
> > of potential crap-removal by allowing someone to kill most of ip_newroute().
>
> And what about code like this in ip_wput_ire():
>
> if (!next_mp) {
> ipsec_out_process(q,
> first_mp, ire, ill_index);
> ...
> return;
> }
> ipsec_out_process(q, first_mp, ire,
> ill_index);
>
> Is there a reason why ip_wput_ire() shouldn't be heavily rototilled?
> (apart from funding, that is)
Apart from funding, and maybe some corner case that escapes me (global
vs. conn_t cached policy, perhaps), I see no reason not to take a rototiller
to ip_wput_ire().
Dan
_______________________________________________
networking-discuss mailing list
[email protected]
