On 09/10/08 17:08, Eric Cheng wrote: > On Wed, Sep 10, 2008 at 04:23:06PM -0700, Darren Reed wrote: > >> Location: mac_sched.c >> Type: T >> Priority: 1 >> Comment: To what extent is header stacking in IPv6 being supported >> by Crossbow? The only one I see supported is IPPROTO_HOPOPTS. >> Of immediate concern is that this leaves out IPPROTO_FRAGMENT, >> IPPROTO_ROUTING and IPPROTO_DSTOPTS. Is further handling >> of IPv6 part of the plan for further work on Crossbow? (The exclusion >> of IPPROTO_FRAGMENT would appear to be the most pressing one.) >> >> > > this is being worked on. > > >> Location: mac_sched.c, mac_flow.c >> Type: T >> Priority: 3 >> Comment: In mac_sched.c we examine the header and say 'type=OTH' >> for fragments. How is this knowledge transferred through to the flow >> classification in mac_flow.c? >> >> > > actually, the mac_sched code is entered after flow classification. > so any pre-parsed state should be transferred the other way around > (mac_flow -> mac_sched). unfortunately this is not being > done now due to the current code structure, we'll try to see if > this is feasible in the upcoming weeks. >
In that case: Location: mac_flow.c Type: T Priority: 1 Comment: Throughout this code the assumption is made that the packet is not a fragment and that what follows the IP header is in fact a layer 4 protocol header. Darren -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/crossbow-discuss/attachments/20080910/5b4bc6ff/attachment.html>
