> > http://cr.opensolaris.org/~meem/clearview-noipmp-cr/
 >
 > usr/src/uts/common/inet/ip.h
 > 
 > 774-784: No extra spaces are needed in the bulk comments section as the 
 > second section is removed.

Indeed; fixed.

 > usr/src/uts/common/inet/ip/igmp.c
 > 
 > Do we need ipsec_out_t in igmp_sendpkt and igmpv3_sendrpt, if we are not 
 > using ipsec_out_attach_if?
 > Is it also used in new implementation?

The new implementation also relies on still setting ipsec_out_ill_index
here (e.g., if it's the IPMP interface index, then when we send we'll send
out the illgrp's nominated multicast interface).

 > I have checked this after talking to you on friday, and it is used in 
 > ip_wput_ire for ipsec_out_dontroute and ipsec_out_multicast_loop.  So, 
 > this is just an FYI.

OK.

 > usr/src/uts/common/inet/ip/ip.c
 > 
 > Some of the IPMP related ioctls are obsoleted. What about the userland 
 > applications which are using them?
 > Ex: rcm_daemon (MI_SETOINDEX option which will use SIOCSLIFOINDEX).

Right, the removal of that logic is part of the Phase 2 review.  For
Phase 1, I just left the ioctl numbers allocated so that things will
still compile, even if they don't work correctly.  It doesn't matter
too much since Phase 1 is just for our internal project use and will
never be integrated.
 > 
 > usr/src/uts/common/inet/ip/ip_if.c
 > 
 > Line: 13494
 >         Extra "when" in the comments.

Fixed.

 > usr/src/uts/common/inet/ip_ire.h
 > 
 > Line: 97
 >         +/* UNUSED                       0x1000  */
 > 
 > I guess it should be UNUSED 0x0400 - 0x0800.

Indeed; Jim caught this too.  Fixed.

Thanks :-)
--
meem

Reply via email to