* kame/kame/pim6sd: massive improvements including - sanity checks for incoming PIM messages - fixed a bug of (sometimes) sending invalid RP-adv messages - fixed a bug in RP group list management - count various statistics inside the daemon and dump them to a file on receiving a signal. A new option '-s' was added to the pim6stat command to support this.
* kame/kame/rtadvd/rtadvd.c: more sanity checks were added for recevied RSes and RAs in order to avoid trying to process an invalid message. Thanks to the TAHI project for pointing out the problem. 1999-12-03 JINMEI, Tatuya <email@example.com> * kame/sys/netinet6/nd6_rtr.c (nd6_ra_input): fixed a problem that valid lifetime always updated regardless of the result of "two hours problem" detection. Reported by the TAHI project (http://www.tahi.org/).
(sync w/ stable)
IFF_RUNNING|IFF_UP, wait for it becomes IFF_RUNNING|IFF_UP before sending DAD probes. some of wireless network cards does not become IFF_RUNNING even if it is made IFF_UP (due to a long chip initialization time - netbsd "awi" needs 2 or 3 seconds). the code does not do the trick if the driver sets IFF_RUNNING without care (like netbsd cnw). current code waits for 15 seconds. maybe we need to provide sysctl access to the timing parameter.
IFF_RUNNING|IFF_UP before sending DAD probes. some of wireless network cards does not become IFF_RUNNING even if it is made IFF_UP (due to chip initialization). the code does not do the trick if the driver sets IFF_RUNNING without care (like netbsd cnw). XXX sysctl knob?
behavior on "-inet6". this does not make sense, and this changes old behavior too much. you could not do "route get -inet6 ::1" with it.
and changes behavior too much. (you can never do "route get -inet6 ::1" with previous code)
behavior too much and is harmful against "route get".