At L21736 ip_if.c, shouldn't the check be
if (!breinfo[i].bi_willdie || bireinfo[i].bi_haverep)
continue;
That would match the assert in the subsequent execution of
ipif_consider_bcast(). Otherwise looks good.
Thirumalai
Peter Memishian wrote:
>Folks,
>
>As part of the Clearview IPMP work, I've had to rework a bunch of the IP
>broadcast and I_{P}{UN}LINK code. While the changes are motivated by the
>requirement to keep the IPMP-related codepaths simple, I believe they
>stand on their own (as evidenced by the fact that they eliminate about 130
>lines of headache-inducing logic from IP). As such, I'm hoping to put
>these changes back to Nevada ASAP.
>
>Thirumalai has already agreed to look at these changes (thanks, Thiru!),
>but I'd like at least one additional reviewer. The changes are at:
>
> http://cr.opensolaris.org/~meem/ipmp-testaddr/
>
>The only file that needs review is ip_if.c. (The in.mpathd-related
>changes have already been reviewed -- and yes, they're independent of the
>IP changes but ended up here for testing convenience.)
>
>Note: these changes do not attempt to change the any of the (questionable)
>existing aspects of the broadcast IRE design, such as having loopback and
>non-loopback broadcast IREs, or having classful broadcast IREs. The goal
>is just to simplify the code for the existing design.
>
>I'd like feedback by the end of the week.
>
>Thanks!
>
>
_______________________________________________
networking-discuss mailing list
[email protected]
