Fetching contributors…
Cannot retrieve contributors at this time
5180 lines (4255 sloc) 219 KB
CHANGELOG for KAME kit, 2000
$KAME: CHANGELOG.2000,v 1.3 2002/12/10 02:12:08 jinmei Exp $
Thu Dec 28 08:48:12 2000 SUMIKAWA Munechika <>
* freebsd4/sys/miscfs/procfs/procfs_ctl.c:
FreeBSD Security Advisory: FreeBSD-SA-00:77.procfs fix
Wed Dec 27 20:52:26 JST 2000
* netkey/key.c:
Added lifetime handler of the policy. There are two timer
"lifetime" and "validtime". "lifetime" means the lifetime of
the policy. "validtime" means the duration after the policy
has not been used. when these timer will be expired, key_spdexpire()
is called and the expiration message is sent to the userland.
the policy can be used without limitiation if both lifetime and
validtime are zero.
2000-12-26 JINMEI, Tatuya <>
* bsdi4/lib/libinet6/resolv: supported EDNS0. See log on
Wed Apr 26 12:00:59 JST 2000 by itojun for more details.
2000-12-25 JINMEI, Tatuya <>
* bsdi4/lib/libinet6: supported IPv6 transport of DNS resolver.
you should updated bsdi4/contrib as well.
This is just a first step. We have many TODO items on this.
Fri Dec 22 JST 2000
* traceroute6, ping6: revoke root privilege earlier.
2000-12-22 JINMEI, Tatuya <>
* kame/kame/pim6sd/pim6_proto.c: fixed typo in J/P message
* kame/kame/pim6sd/routesock.c: fixed alignment bugs for no SA_LEN
cases (nothing to do with BSD, though).
Both fixes were based on KAME PR sys/308 from
2000-12-22 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_mroute.c (ip6_mdq): correctly initialized
the "must be zero" field of upcall data. This fix is essential to
run IPv6 multicast routing daemon.
Report from:
2000-12-22 JINMEI, Tatuya <>
* kame/kame/rtadvd/config.c (getconfig): allowed hosts to
advertise RA only when router lifetime is set 0 on every
advertising interface. This restriction would avoid invalid RAs
advertised from a host by misconfiguration.
2000-12-21 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (icmp6_input): all *BSD called
icmp6_rip6_input(). rip6_input(), which was used by bsdi4 and
openbsd, was not suitable, because it would return unexpected
"unknown next header" errors upon receiving (e.g.) icmp6 echo
* kame/sys/netinet6/raw_ip6.c (rip6_input): removed icmp6 cases
according to the above change.
2000-12-21 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (init_ip6pktopts): added to avoid
forgetting initialization of packet options. This function is
called from ip6_pcbopt() and ip6_setpktoptions().
This change would fix a problem that traceroute6 source routing
failed by EMSGSIZE.
2000-12-21 JINMEI, Tatuya <>
* kame/sys/netkey/key.c (key_cmpspidx_withmask): applied the same
fix as one introduced on 2000-12-07 (see below) for inbond
Wed Dec 20 14:41:11 JST 2000
* netbsd/lib/libinet6/getaddrinfo.c: experimental support for A6
records, only for cases where all 128 bits are supplied in a single
A6 record (like "A6 0 ::1"). netbsd only. disabled by default.
Mon Dec 18 01:41:47 2000 SUMIKAWA Munechika <>
* freebsd4/usr.sbin/{syslogd, lpr}, freebsd4/usr.bin/logger: sync
with FreeBSD-current.
* freebsd3/ports/qpopper: upgrade to 3.1.2
Fri Dec 15 23:24:58 JST 2000
* racoon:
racoon uses syslog() if both the logging file and -F option are not
specified when racoon is launched. Now there are three way to log.
One is to the standard output. Second is to syslog. Third is to the
file specified by the option.
nuke YIPSDEBUG() macro, instead racoon use the value "loglevel"
as logging value. "loglevel" can be specified either of the following
capital word, and they are printed with each output as the tag name in
the logging file.
INFO: begin negotiation, SA establishment/deletion/expiration.
NOTIFY: just notifiable.
WARNING: not error strictly.
ERROR: system call error. also invalid parameter/format.
DEBUG1: debugging informatioin.
DEBUG2: too more verbose. e.g. parsing config.
XXX The logging level and the tag name are mixed. They may be separated.
ERROR should be separated because the error of IKE protocol and system
call are mixed.
XXX to be more checked.
Wed Dec 13 17:05:36 JST 2000
* netbsd/sbin/racoon: update racoon reachover makefile. compiles in
GSSAPI support if MKKERBEROS is set.
Tue Dec 12 08:56:39 PST 2000
* racoon/
* racoon/algorithm.c
* racoon/algorithm.h
* racoon/cfparse.y
* racoon/cftoken.l
* racoon/
* racoon/handler.c
* racoon/handler.h
* racoon/ipsec_doi.c
* racoon/isakmp.c
* racoon/isakmp.h
* racoon/isakmp_agg.c
* racoon/isakmp_base.c
* racoon/isakmp_ident.c
* racoon/main.c
* racoon/oakley.c
* racoon/oakley.h
* racoon/racoon.conf.5
* racoon/remoteconf.h
* racoon/strnames.c
Added files:
* racoon/gssapi.c
* racoon/gssapi.h
* racoon/README.gssapi
Implement GSSAPI using Kerberos 5 authentication for Phase 1
per dratf-ietf-ipsec-isakmp-gss-auth-06.txt. Implemented by
Frank van der Linden <>.
2000-12-12 JINMEI, Tatuya <>
* kame/sys/netinet6/raw_ip6.c: used by bsdi4 for better code
sharing. Basically, no new feature change was introduced by this
Snu Dec 11 JST 2000
* freebsd4/sys/i386/isa/if_lnc.c: cope with VMware ethernet chip
emulation, which loops back multicast packet to itself
(violates IFF_SIMPLEX behavior). now DAD on freebsd-on-VMware
works fine. thanks to: Florent Parent
<> for help testing.
Sat Dec 10 JST 2000
* racoon/isakmp.c: bounds-check bogus length in isakmp header.
Sat Dec 9 09:08:02 JST 2000
* sys/netinet6/icmp6.c (netbsd/openbsd):
implement high/low watermark on pmtud host route entries.
create up to hiwat host route entries, if icmp6 too big messages is
validated. create up to lowat host route entries, if too big message
is not validated (= traffic is from non-connected pcb).
XXX hiwat/lowat default values
Fri Dec 8 16:05:04 JST 2000
* netbsd/sys/nfs: avoid IPv6 path MTU discovery if UDP is used as the
transport layer. TODO: make NFS aware of path MTU discovery.
2000-12-07 JINMEI, Tatuya <>
* kame/sys/netinet6/ipsec.c (ipsec6_output_trans): when an ipsec
SA cannot be found while ipsec is required, send an icmp6
dst_unreach_admin error (instead of silent discard).
NOTE: Please be sure to update icmp6.c as well.
2000-12-07 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (icmp6_reflect):
- processed scoped addresses in a generic manner.
- used in6_selectsrc to determine the source address of the
reflected packet.
2000-12-07 JINMEI, Tatuya <>
* kame/sys/netkey/key.c (key_cmpspidx_withmask): compared
sin6_scope_id values only when both two values were
non-zero. Without this fix, ::/0 would not match fe80::1%ne0,
which could be a security hole.
TODO: there seem to be additional misuse about scope in this
file. We'll have to fix them eventually.
Wed Dec 6 00:22:31 JST 2000
* openbsd: use shared raw ip6 logic.
Tue Dec 5 10:44:53 JST 2000
* sys/netinet6/nd6_nbr.c: (non-bsdi) make sure we don't have DAD
process, for an IPv6 interface address on detached PCMCIA interfaces.
Mon Dec 4 23:09:51 JST 2000
* netbsd/pkgsrc: obsolete KAME pkgsrc tree, at least for now.
we should consult NetBSD pkgsrc directory. KAME pkgsrc uses
1.4.2 pkgsrc build framework, which is too old for use with 1.5.
Mon Dec 4 20:27:27 JST 2000
* netbsd/sys/netinet/udp_usrreq.c: fix IPv4 multicast input path.
NetBSD PR 11629.
Mon Dec 4 19:55:10 JST 2000
* sys/netinet6/in6_pcb.c (netbsd/openbsd):
re-introduce workaround for updating cached routes in unconnected
inpcb, on ICMPv6 redirect (otherwise, cached routes will not be
updated). should be revisited.
2000-12-04 JINMEI, Tatuya <>
* kame/kame/pim6[ds]d/mld6.c (make_mld6_msg):
* kame/kame/mld6query/mld6.c (make_msg):
* kame/kame/pim6dd/pim6.c (init_pim6):
made sure to use CMSG_LEN (not CMSG_SPACE) to set to cmsg_len.
based on a PR from URA Hiroshi <>
2000-12-03 Koji Kawano <>
* sys/netinet6/in6_prefix.c: missing copy vltime & pltime to
This effects that, if you use prefix command, you can't
add address to the interface or vltime = 0 even added.
Sun Dec 3 02:48:51 JST 2000
* sys/netinet6/ip6_mroute.h: to provide better binary compatibility
with *BSD-integrated tree, change operation #. you'll need to
recompile all multicast routing tools as well as the kernel.
Sat Dec 2 17:27:30 JST 2000
* bsdi3, bsdi4, freebsd2, netbsd: do not pass incomplete tcp header
to syn_cache_unreach.
Sat Dec 2 17:19:58 JST 2000
* netbsd: catch up with netbsd 1.5. MIP6/NATPT do not compile.
pvcbridge does not work for now.
NOTE: if you use anoncvs/cvsup to update from existing tree,
make sure to invoke "make TARGET=netbsd clean" before update.
symlink can make troubles with update.
Fri Dec 1 21:02:25 2000 SUMIKAWA Munechika <>
* freebsd4/ports/{ct,v6eval}: upgrade to 1.2.
Fri Dec 1 19:35:18 JST 2000
* openbsd: catch up with openbsd 2.8. MIP6/NATPT may not work.
NOTE: if you use anoncvs/cvsup to update from existing tree,
make sure to invoke "make TARGET=openbsd clean" before update.
symlink can make troubles with update.
2000-12-01 JINMEI, Tatuya <>
* *BSD/sys/netinet6/in6_pcb.c (in6_pcbnotify):
* *BSD/sys/netinet[6]/{tcp, udp, raw}XXX.c (ctlinput functions):
added clarifications and improvements:
- made the ctlinput mechanism more scope aware.
- considered flow label to detect PCBs to which should be notified a
given error (as well as usual detection items such as port
- removed meaningless special considerations for PRC_HOSTDEAD
2000-12-01 JINMEI, Tatuya <>
* kame/sys/netinet6/esp_input.c (esp6_ctlinput): called
pfctlinput2 to tell applications minimum information about an
error, even when the notification was encrypted. PCB notification
functions then might be able to detect an appropriate PCB using
source and destination addresses and IPv6 flow label.
2000-12-01 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_output): called pfctlinput2
when fragmenting the packet, in order to tell applications the
proper MTU of the outgoing interface for a particular destinaion.
The code would be more conformant to rfc2292bis with this change.
2000-12-01 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_input.c (pfctlinput2): was newly added to
allow the ctlinput mechanism to have additional
information. Although it could (or even should) be generalize to
be domain independent, we currently keep the function for INET6
only, because we don't know the whole effects of this.
2000-11-30 JINMEI, Tatuya <>
* bsdi4/sys/net/route.c (rtredirect): disabled the AF check in the
beginning of rtredirect, due to which redirection did not work for
IPv6. Although the check is actually a bit redundant, but
disabling the check does not have a bad effect.
2000-11-30 JINMEI, Tatuya <>
* {kame,bsdi4,openbsd,freebsd[34]}/sys/netinet6/in6_pcb.c
(in6_pcbnotify): removed a special treatment for rtchange cases,
which was necessary only when we used the ND6_WAITDELETE neighbor
cache state for an unreachable neighbor. This removal would not
affect the behavior, but make the code simpler.
2000-11-29 JINMEI, Tatuya <>
* kame/kame/pim6sd/mld6.c (init_mld6): allow multicasted packet
sent to the MLD6 socket to be loop-backed by default. We can do
this since we don't have to handle DMVRP messages via the MLD6
(based on a question from <>.)
2000-11-28 JINMEI, Tatuya <>
* kame/kame/ping6/ping6.c ({get, set}_pathmtu): newly added to
support application-level path MTU discovery. If the -m option is
specified, ping6 will try to discover path MTU for the destination
using the IPV6_RECVPATHMTU socket option, and adjust the MTU using
the IPV6_USE_MTU socket option. This would be useful for those
OSes that do not perform path MTU discovery unless corresponding
connected socket exists, like recent versions of NetBSD and
2000-11-27 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c: added a new ancillary data item
"IPV6_USE_MTU," which allows an application to specify the MTU of
the outgoing packet. This item is currently experimental (i.e. not
described in a standard document.)
Mon Nov 27 04:20:40 2000 SUMIKAWA Munechika <>
* freebsd3/ports/lftp: upgrade to 2.3.5.
2000-11-27 Koji Kawano <>
* /kame/freebsd[3,4]/ifconfig
1. add vltime & pltime parameter setting
2. add default thier value setting (tentative bug fixing)
2 is coz that not to add addresses.
2000-11-25 JINMEI, Tatuya <>
* kame/kame/pim6sd/route.c (add_leaf): just ignored an MLD report
if the node is not the DR for the corresponding subnet.
The fix was from
2000-11-25 JINMEI, Tatuya <>
* kame/kame/pim6sd/pim6_proto.c (receive_pim6_bootstrap): even
when the BSR changes, just schedule an immediate advertisemnet of
advertisement, instead of actually sending message, in order to
avoid sending the advertisement to the old BSR.
In response to: a comment from
2000-11-25 Koji Kawano <>
* kame/freebsdX/net/if_loop.c: fixed duplicated free of pkt's aux data.
This happens in case of that "ping6 -w -I lo0 ::1" for example.
Fri Nov 24 17:29:30 JST 2000
* libinet6/if_nametoindex.c: conform to RFC2553 (ENXIO if the interface
is not found).
Fri Nov 24 15:14:37 JST 2000
* libinet6: compile strl{cpy,cat} for better code sharing.
Fri Nov 24 01:43:21 2000 SUMIKAWA Munechika <>
* kame/freebsd4/usr.sbin/syslogd: IPv6 enabled syslogd from '-a' option does not work now.
2000-11-21 Koji Kawano <>
* kame/freebsd4: we now start to support 4.2-RELEASE.
Tue Nov 21 02:45:24 2000 SUMIKAWA Munechika <>
* kame/freebsd3/ports/wwwoffle: upgrade to 2.6.
2000-11-18 JINMEI, Tatuya <>
* [fno]bsd: supported the IPV6_RECVPATHMTU socket option. Not
compiled as of writing this log.
2000-11-18 JINMEI, Tatuya <>
* kame/kame/racoon/grabmyaddr.c (suitable_ifaddr6): made sure to
close the temporary socket. Without this, racoon will eventually
die upon the "too many open files" error.
Fri Nov 17 05:09:07 2000 SUMIKAWA Munechika <>
* freebsd3/ports/apache13: upgrade to 1.3.14.
Thu Nov 16 01:16:07 JST 2000
* freebsd2/ports/lukemftp: lukemftp (portable NetBSD ftp client),
version 1.5.
* freebsd2/usr.bin/Makefile: we no longer build ftp client, to decrease
our maintenance labor. use lukemftp, or some other ftp client.
2000-11-15 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (icmp6_input):
* bsdi4/sys/netinet6/raw_ipv6.c (rip6_ctlinput):
* bsdi4/sys/netinet/{tcp,udp}_usrreq.c ({tcp6,udp6}_ctlinput):
* bsdi4/sys/netinet6/in6_pcb.c (in6_pcbnotify):
IPV6_RECVPATHMTU socket option support.
Note that the current implementation is not fully conformant to
the latest rfc2292bis spec. The implementation uses a special
control structure to pass both the MTU value and the destination
to the application, while the spec says the corresponding
ancillary data just contains the MTU value.
We are now clarifying it in the IETF ipng ML, and the
implementation may change after the discussion.
TODO: transport layer support in other *BSD than bsdi4.
Mon Nov 13 04:28:33 2000 SUMIKAWA Munechika <>
* freebsd3/ports/ncftp3: upgrade to 3.0.2
Sun Nov 12 15:40:29 JST 2000
* bsdi4/libexec/ftpd/ftpd.c: update LPSV/EPSV result code to conform
2000-11-11 JINMEI, Tatuya <>
* kame/kame/rtadvd: introduced some improvements and new features
mainly for auto router renumbering (RR):
- joined all-routers site-local multicast group when RR is enabled,
and made the receiving interface configurable via a new command
line option "-M".
- fixed many bugs in handling RR messages.
- supported lifetimes that decrement in real time, and will expire
at a specified time in the future. 3 new parameters for
rtadvd.conf were introduced to handle such lifetimes. See
Thu Nov 9 20:14:00 JST 2000
* racoon:
check the address is suitable or not before using them when grabing
all addresses on each interfaces. tentative, duplicated and
detached are unsuitable at the moment.
Thu Nov 9 15:30:42 JST 2000
* racoon:
Use both the source address and the destination address to get
Thu Nov 9 09:53:54 JST 2000
* netbsd/libexec/{rshd,rlogind}: do not refer free'ed memory region.
KAME PR 302 from
Thu Nov 9 02:21:00 JST 2000
* *bsd*/sys/sys/mbuf.h: make sure we nuke mbuf pointed to by m_aux,
on MFREE(). under sys/kern, there are code which calls MFREE()
instead of m_free().
* sys/netinet6?/tcp6?_input.c: reject SYN, if the destination address is
a deprecated address and we forbid the use of deprecated address.
* sys/sys/mbuf.h: get rid of M_ANYCAST6. flag bit is a precious resource
(done for freebsd2/bsdi[34]/openbsd/netbsd)
Thu Nov 9 02:17:10 JST 2000
Ignored a acquire message if its destination address matches
a multicast address.
Thu Nov 9 01:30:20 JST 2000
* setkey:
Print a timestamp on each dump line by using -x option.
2000-11-08 JINMEI, Tatuya <>
* kame/kame/libinet6/if_indextoname.c (if_indextoname): made sure
to set ENXIO to errno if an appropriate interface is not found.
(to be conformant to RFC2553 and its bis)
Wed Nov 8 02:14:22 JST 2000
* racoon:
Fixed to bind a address which is added right now to the kernel.
2000-11-08 JINMEI, Tatuya <>
* bsdi3/sbin/ifconfig/ifconfig.c (in6_status):
* bsdi4/sbin/ifconfig/ifconfig.c (inet6_status):
* {net,open}bsd/sbin/ifconfig/ifconfig.c:
explicitly state that an address is "deprecated" if preferred
lifetime goes into 0.
2000-11-08 JINMEI, Tatuya <>
* kame/kame/rtadvd/rtadvd.c (sock_open): join ff05::2 if router
renumbering is enabled.
Wed Nov 8 01:27:40 JST 2000
* route6d: do not advertise cloned/dynamic routes.
Tue Nov 7 12:44:07 JST 2000
* netbsd/pkgsrc/mail/postfix: use 20001005 + IPv6 patch.
Tue Nov 7 01:49:35 2000 SUMIKAWA Munechika <>
* kame/kame/libinet6/resolv/res_init.c: fix memory leak.
Mon Nov 6 15:46:39 JST 2000
* sys/net/if_stf.c, sys/netinet*/in*gif.c: IFF_LINK2 now controls
the use of ingress filter on tunnel outer/inner source address.
2000-11-06 JINMEI, Tatuya <>
* {bsdi4,freebsd[34]}/sys/netinet/tcp_subr.c (tcp6_ctlinput):
* {bsdi4,openbsd}/sys/netinet/udp_usrreq.c (udp6_ctlinput):
* freebsd[34]/sys/netinet6/udp6_usrreq.c (udp6_ctlinput):
* kame/sys/netinet6/udp6_usrreq.c (udp6_ctlinput):
loosened validation for inner packets of icmp6 errors as much as
2000-11-06 JINMEI, Tatuya <>
* freebsd3/sys/netinet/ip_input.c (ip_input): reverted original
FreeBSD code fragment; twisted ip_id before and after calling
Reported by: Greg Troxel <>
Mon Nov 6 00:36:02 JST 2000
* sys/netinet6/udp6_output.c (bsdi4/openbsd): do not mix up
inbound and outbound flowlabel value.
Thu Nov 2 23:17:07 JST 2000
* sys/netinet6/esp_core.c: fix [13]DES on big endian machines.
report from
Wed Nov 1 23:58:34 JST 2000
* sys/crypto/twofish: fix runtime error check logic. it has been
broken for some time.
* regress/crypto: regression test for kernel crypto logic.
Wed Nov 1 17:24:31 JST 2000
* sys/netinet6/ip6_{in,out}put.c: implement flow label manipulation
setsockopt, as documented in draft-itojun-ipv6-flowlabel-api-00.txt.
the default value can be switched by net.inet6.ip6.auto_flowlabel.
Tue Oct 31 13:33:38 JST 2000
* sys/netinet6/ip6_{in,out}put.c: implement traffic class manipulation
API, as documented in draft-itojun-ipv6-tclass-api-01.txt.
2000-10-29 Shin'ichi Fujisawa <>
* Fixed NATPT memory leak. When TCP session is finished, memory
which holds tcp information is not freed.
2000-10-25 JINMEI, Tatuya <>
* kame/kame/rtadvd/if.c (if_getmtu): made some clarifications:
- first try getifaddrs(), and then do SIOCGIFMTU for safety. The
revised logic will relieve environments with poor getifaddrs()
- avoid referring to freed data by delaying to call freeifaddrs().
* kame/kame/rtadvd/rtadvd.conf.5: insistently noted that the
configuration file can be omitted for those who do not read
2000-10-24 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6_rtr.c (defrouter_select): do not install a
route to the default interface as default route if the node acts
as a router, because such default route might conflict with a
default route installed from a user application (e.g. routing
2000-10-24 JINMEI, Tatuya <>
* kame/kame/v6test/getconfig.c (make_{tcp,udp}): TCP/UDP data
buffer can be specified.
based on a patch from: Tomomi Suzuki <>
2000-10-24 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.c (in6_ifscrub): checked the existence of
dstaddr before calling rtinit(). This fix will solve the KAME PR
Sun Oct 22 23:08:22 JST 2000
* kame/sys/netinet6/in6_pcb.c (freebsd2/bsdi3):
don't do anything special against IPv4 mapped address, as we don't
support them.
Sat Oct 21 01:40:20 JST 2000
* bindtest: add summary mode (-s). from yoshfuji.
Fri Oct 20 04:20:12 JST 2000
* openbsd: enable IPv6 PTMUD DoS prevention. validates TCP6/UDP6 PMTUD
by existence of connected pcb.
Fri Oct 20 03:14:55 JST 2000
* ping6: by default, fragment packet into 1280 bytes (IPv6 minimum MTU).
-m will suppress the behavior (send it as is, and obey path MTU
Thu Oct 19 14:12:04 JST 2000
* racoon:
- check whether the initial contact is acceptable.
when both ends are using "use" as the policy level, both ends initiate
the phase 1 and send initial-contacts under each phase 1 SA. In this
case, each phase 1 SA will be banished by the another initial-contact.
Simply calling getcontacted() is the solution to avoid the case.
2000-10-19 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_mroute.c (add_m6fc): made sure to reset
the outgoing interfaces list in all cases where a multicast
routing cache is created or updated.
In response to a report from Hoerdt Mickael
Thu Oct 19 10:37:05 JST 2000
* sys/netinet6/icmp6.c: (netbsd only for now) try to prevent
DoS attack using path MTU messages. validate ICMPv6 too big messages
using content of TCP/UDP connection table and/or IPsec SAs.
framework by sync with netbsd-current.
TODO: think about sendto(2) cases, raw ip6 cases and others.
aren't we too restrictive given the fact PMTUD is mandatory for IPv6?
TODO: other platforms
Wed Oct 18 18:54:50 JST 2000
* racoon:
- fixed to parse the timer directive in the racoon configuration.
- the identifier in phase 1 transmitted by peer compared with the user
expecting. If they are not same, racoon just warns it at the moment.
- Moved the place of calling ipsec_checkid1() after parsing the payload
transmitted by peer.
- bark if DH computation failed.
- added two function to handle the certificate.
eay_str2asn1dn(), convert the string into DER.
eay_cmp_asn1dn(), compare two subjectname.
2000/10/18 18:15:25 JST
the 2nd round of the ALTQ cleanup:
* add a new discipline, PRIQ (priority queueing).
its minor device number is 11.
* kernel config options:
prepend "ALTQ_" to all the ALTQ kernel config options.
(e.g., CBQ -> ALTQ_CBQ)
enable RED/RIO on HFSC/CBQ/PRIQ when RED/RIO is defined.
* change ioctl numbers to be more consistent among disciplines.
* altqd:
divide altqd.c into altqd.c and libaltq2.c.
libaltq2.c contains functions and variables required by libaltq.
* remove the accounting mode from headers and userland tools.
(it was already removed from the kernel)
* other misc cleanup
as usual, don't forget to install new include files and rebuild
userland tools.
2000-10-17 Shin'ichi Fujisawa <>
* NAT-PT can translate fragmented IPv4 packet to IPv4.
You need to write 'options NATPT_NAT' and 'options NATPT_FRAGMENT'
into kernel configuration file, and recompile kernel.
This function is experimental, and tested only FreeBSD 3.5.
2000/10/17 19:47:21 JST
* freebsd4/sys/alpha/alpha exception.s:
* freebsd4/sys/alpha/include asm.h:
import the fix from FreeBSD-current that allows to boot large
kernels on alpha.
without this fix, GENERIC.KAME kernels hang after printing
the copyright.
Mon Oct 16 05:13:56 2000 SUMIKAWA Munechika <>
* freebsd3/ports/mozilla: upgrade to M18
2000-10-15 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (icmp6_mtudisc_update): update path
MTU information only when a corresponding host route already
exists. Currently for bsdi only.
This is an experimental fix for the KAME PR sys/292. We may choose
a different way in the future.
2000-10-15 JINMEI, Tatuya <>
* bsdi4/sys/netinet/tcp_subr.c (tcp6_ctlinput):
* bsdi4/sys/netinet/udp_usrreq.c (udp6_ctlinput):
added validation for the TCP or UDP headers of inner
packets. Without this, a bad remote user could make the kernel
(itojun seemed to merge the fix to other *BSDs)
Sun Oct 15 19:18:45 2000 SUMIKAWA Munechika <>
* freebsd3/ports/squid11: use latest IPv6 patch.
* freebsd3/ports/lftp: upgrade to 2.3.4.
* freebsd3/ports/sendmail: sendmail 8.11 now is official version
so I combine sendmail6 and sendmail.beta into new port 'sendmail'.
Sun Oct 15 17:49:15 JST 2000
* netbsd/pkgsrc/net/{libpcap,tcpdump}, freebsd[23]/ports/{libpcap,tcpdump}:
use 10/9 weekly snapshot.
Thu Oct 12 04:56:24 JST 2000
* kame/racoon: Added two directives, my_identifier and peers_identifer.
It can be define the identifiers by each peers. Identifier directive
will obsoleted in near future.
XXX peers_identifier should be compared with the value included in ID
payload sent from the peer.
XXX asn1dn shoule be able to be specified by user.
Wed Oct 11 JST 2000
* sys/net/if_gif.c: fix uninitialized variable, which can lead to
kernel panic.
Tue Oct 10 23:54:17 JST 2000
* sys/netinet6/in6_proto.c: don't call tcp_init twice, if
we share tcp stack between IPv4/v6 and we are compiling a kernel
with dual stack configuration (netbsd/openbsd/freebsd[34]/bsdi4).
* sys/netinet6/icmp6.c: nuke net.inet6.icmp6.errratelimit.
* openbsd/sys/conf/GENERIC: move KAME mandatory flags into
conf/GENERIC, so that they will get defined for all kernel
configuration files.
Tue Oct 10 15:43:58 JST 2000
* kame/rtsold/rtsock.c: watch routing socket for pcmcia/cardbus
interface removal. cease RS if an interface gets removed.
(affects NetBSD 1.5 or higher)
Tue Oct 10 03:50:32 2000 SUMIKAWA Munechika <>
* freebsd3/ports/wu-ftpd: upgrade to latest IPv6 patch. EPSV now
works. TODO: -s speaks IPv4 only.
Sat Oct 7 16:01:12 JST 2000
* ping6, traceroute, traceroute6, mtrace6, icmp6dump: plug fd_set hole.
see openbsd select(2). from deraadt@openbsd.
* traceroute: repair savestr() hole.
Thu Oct 5 12:31:12 JST 2000
* sys/netkey/key.c: get rid of all uses of random() in IPsec code,
where possible (netbsd/freebsd4). for other operating systems, we
have no good random number source like rnd(4), and random() will
be used. you will see one-line warning when you start using IPsec.
XXX we may need to think about how IV should be stirred. the current
code uses too much entropy out of rnd(4).
Thu Oct 5 07:41:18 JST 2000
* *bsd*/sys/net/if.c: make sure we have root privilege on SIOCSPHY*
operation. remove SIOCSPHY* handling from in{,6}_control.
Wed Oct 4 20:39:16 JST 2000
* sys/netkey/key.c: supply two sysctl variables,
net.key.{esp,ah}_keymin, that control how ACQUIRE messages are
formed. algorithms/key length smaller than the configured value
will be filtered out.
Wed Oct 4 07:30:10 JST 2000
* sys/net/pfkeyv2.h, racoon: move AES (rijndael) protocol # to the
official one. note that you now lost interoperability between old
racoon and new racoon, when you negotiate rijndael.
also note that AES final document is yet to become FIPS standard,
so there's some uncertainity window, for like 3 months.
Wed Oct 4 00:29:42 2000 SUMIKAWA Munechika <>
* freebsd3/ports/lftp: upgrade to 2.3.3.
* freebsd3/ports/{libpcap,tcpdump}: upgrade to 2000/10/02
weekly snap.
* freebsd3/ports/wu-ftpd: upgrade to 2.6.1.
Tue Oct 3 JST 2000
* libipsec/pfkey.c: tighten header chasing. we should change some
of the function signatures for better protection.
2000-10-04 JINMEI, Tatuya <>
* bsdi4/usr.sbin/netstat/main.c:
* kame/kame/netstat/ipsec.c (ipsec6_stats0):
printed statistics on IPv6 IPsec by `netstat -s [-p ipsec6]'.
2000-10-03 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6_rtr.c (defrouter_addifreq): use rtrequest1
with a proper interface address for bsdi4, instead of rtrequest.
This is necessary for the function to work correctly for a p2p
2000-10-03 JINMEI, Tatuya <>
* bsdi3/sys/netinet/tcp_subr.c (ipsec4_hdrsiz_tcp):
* bsdi4/sys/netinet/tcp_subr.c (ipsec_hdrsiz_tcp):
properly set the ip_hl field before calling ipsec4_hdrsiz().
Without this, the policy matching routine would misunderstand the
IP header length, and would failed to detect the position of the
TCP header. As a consequence, policy matching using specific TCP
ports would fail.
2000-10-03 JINMEI, Tatuya <>
* kame/sys/crypto/rijndael/rijndael-api-fst.c: avoided assert,
memcpy, and memset, which are not available in the kernel of some
BSD variants. It would rather be better to use them from a
standardization point of view, but it could be undesirable for
portability among *BSD kernels.
2000-10-02 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6_nbr.c (nd6_ns_input): added several minor
- do not 'goto bad' unless taddr6 is propery set.
- adjust the pointer to ip6_hdr after IP6_EXTHDR_{GET, CHECK} for
- goto bad if check for hoplimit fails (to record the header
2000-10-02 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_input.c (ip6_savecontrol): check if a
routing header is contained, in order to store destination option
headers at proper postions.
XXX: the check routine is not very efficient.
2000-10-02 JINMEI, Tatuya <>
* bsdi[34]/sys/kern/uipc_syscalls.c (sendit): check if cmsg_len is
0, in order to avoid an infinite loop.
2000-10-02 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_setpktoptions): supported
the IPV6_RTHDRDSTOPTS case, which was just forgotten.
2000-10-02 JINMEI, Tatuya <>
* bsdi4/sbin/ifconfig/ifconfig.c (main): set ND6_INFINITE_LIFETIME
to ia6t_[pv]ltime as the default values, so that manually
configured IPv6 addresses have infinite lifetime.
2000-10-01 JINMEI, Tatuya <>
* kame/kame/pim6sd/timer.c: fixed a bug that missed necessary
(applied a patch from
Sun Sep 30 JST 2000
* racoon: disable rc5/idea support by default, due to patent twists.
Fri Sep 29 12:12:16 2000 SUMIKAWA Munechika <>
* freebsd3/ports/fetchmail: upgrade to 5.5.3.
* freebsd3/ports/lftp: upgrade to 2.3.2.
Wed Sep 27 00:56:03 2000 SUMIKAWA Munechika <>
* freebsd3/ports/wget: use latest IPv6 patch(9/15).
Tue Sep 26 17:40:30 JST 2000
* sys/netkey/key.c: attach sadb_comb for IPComp SADB_ACQUIRE message.
the use of sadb_comb is not on RFC2367; it was discussed in
pf_key mailing list and is one of the candidate behavior under
current header file declaration.
Tue Sep 26 07:56:17 JST 2000
* libinet6/getnameinfo.c: off-by-1 error in string length valildation.
From: Pavel Kankovsky <>
NOTE: all users are suggested to upgrade their userland.
Sun Sep 24 07:31:53 JST 2000
* netbsd/pkgsrc/net/{libpcap,tcpdump}: use 2000/9/18 weekly snap
Sun Sep 24 04:15:52 2000 SUMIKAWA Munechika <>
* freebsd3/ports/wwwoffle: upgrade to 2.5e.
* freebsd3/ports/netcat: netcat 1.10. you must select address
family to use -4/-6 option explicitly.
Sun Sep 24 00:38:51 2000 SUMIKAWA Munechika <>
* freebsd3/ports/tcp_wrapper: upgrade to latest IPv6 patch.
* freebsd3/ports/lynx: upgrade to 2.8.4dev10
Fri Sep 23 JST 2000
* racoon:
- Converted to binary from string if need. openssl returns a string
even if object should be a binary.
- fixed to attach the notify of responder life time. It always was
- enabled to get multiple subjectAltNames. verified the ID in the
ID payload and one of the subjectAltNames.
- don't send the attribute of lifetime as bytes when it is not defined
in configuration file.
- nuke certaltname. It always define explicitly name in the
racoon.conf, for example fqdn, user_fqdn. don't get identifier
from the subjectAltName.
- fixed base mode with RSA signature in initiator case.
- enable to switch sending CR payload or not.
- ignored multiple cert payload. at the moment, racoon take first cert.
But it should be took all cert payload.
Sat Sep 23 JST 2000
* racoon: (1) attach key length attribute to rijndael/twofish.
they are variable-length ciphers, key length attribute is mandatory.
(2) disable kmpstat and racoon admin port, as there's no
authentication at all (bad idea for security tools...)
* sys/netkey/key.c: attach sensible default lifetime to sadb_comb.
Fri Sep 22 23:47:24 JST 2000
* sys/netkey/key.c:
Fixed the meaning of "created". Then the behavior of
net.key.blockacq_count is fixed.
Fri Sep 22 JST 2000
* sys/netinet6/ipsec.c: clarify policy lookup. always pull port
number from the packet. previous code looked up port number from
pcb, and did not work right on sendto() with explicit address.
* sys/netinet6/ip6_forwarding.c: do not try to look at port numbers,
as we are forwarding and the packet could be a fragment.
* sys/netinet6/ipcomp_*.c: (1) stabilize deflate engine. (2) remove
too strong assumption on mbuf chain, make it friendly with
m_pulldown cases. (3) compute outbound byte lifetime based on
uncompressed packet. the behavior now meets inbound behavior.
RFC2401 page 23. (4) fix mbuf leak (compression error cases).
* racoon: (1) support short phase 2 SPI (16bit instead of 32bit) for
IPComp, to meet the latest IPComp draft. need more checks.
(2) stabilize behavior when many phase 1 proposals are presetned.
(3) make it possible to build with boehm GC.
(4) do not attach IDci/IDcr payload on phase 2 negotiation, if
we are negotiating transport mode SAs. this should help "encrypt
everyone's traffic" server configuration, like below:
spdadd ::/0 ::/0 any -P out ipsec esp/transport//use;
spdadd ::/0 ::/0 any -P ah ipsec esp/transport//use;
the source of this problem is the dual meaning of policy src/dst
address space pair ("::/0 ::/0" in the above example).
there are two meanings/interpretation of this at this moment:
- cloud behind the gateway (tunnel mode case only)
- src/dst pair for which i would like to enforce ipsec
we may need to split them into two distinct configuration,
if there's demand.
* sys/netkey/keysock.c: stabilize behavior in heavy SADB_ACQUIRE
Thu Sep 21 03:18:11 JST 2000
* sys/netinet6/esp_core.c: make it possible to use rijndael and twofish
at the same time. this obsoletes kernel compile option
* sys/netkey/key.c: relax SPI boundary check for IPComp. at this
moment no check is made against wellknown IPComp CPI.
need to revisit.
* sys/netinet6/route6.c: repair cut and paste bug in address validation.
from francis dupont.
Wed Sep 20 22:21:57 2000 SUMIKAWA Munechika <>
* kame/freebsd3/ports/bind9: upgrade to official release 9.0.0
Wed Sep 20 03:10:09 JST 2000
* sys/netkey/key.c: make proposal/combination payload on SADB_ACQUIRE
message (kernel -> userland) optional, to support ipcomp ACQUIRE
messages. this makes KAME PF_KEY to violate RFC2367. however,
as ipcomp is not covered by RFC2367 (we cannot attach
proposal/combination for ipcomp case), we have no other choice.
* netbsd/sys/netinet/tcp_subr.c: allocate mbuf cluster for tcp header
template, only when necessary. this avoids too aggressive memory
usage on heavy load web server, for example.
From: Kevin Lahey <>
Tue Sep 19 07:43:42 JST 2000
* freebsd[234], bsdi[34] tcp code: repair IPsec header size prediction
(which avoids fragmentation on IPsec'ed TCP packet).
was broken by Sep 19 00:59:30 change.
Tue Sep 19 06:01:33 JST 2000
* sys/netinet6/esp_core.c: repair IPsec blowfish-cbc (BF_encrypt takes
host endian value!). Aug28 commit log was wrong. blowfish-cbc has
been broken from Aug28 till this commit.
* sys/netinet6/esp_core.c: nuke old ESP core logic.
Tue Sep 19 02:50:41 JST 2000
* netbsd/pkgsrc/net/bind9: use 9.0.0 (first official release)
Tue Sep 19 00:59:30 JST 2000
* sys/netinet6/ipsec.c: on ipsec policy lookup, properly handle traffic
on AF_INET6 socket with IPv4 mapped address (eg IPv4 on wire).
NOTE: we still have issue with port number lookup (KAME PR 287).
Sat Sep 16 23:28:32 JST 2000
* freebsd2/ports/wget, netbsd/pkgsrc/net/wget: use latest IPv6 patch
(9/15). it repairs -I option.
2000-09-15 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (icmp6_redirect_input): examined the
gateway (from the routing table) only when the address family of
the gateway is AF_INET6.
In response to a report from Jacques A. Vidrine <> in
the freebsd-net ML.
Thu Sep 14 10:40:12 2000 SUMIKAWA Munechika <>
* kame/freebsd3/ports/lfp: upgrade to 2.2.6.
Thu Sep 14 09:24:47 JST 2000
* kame/faithd: repair EPRT translation. cope with PASV result
without paren.
Thu Sep 14 04:01:35 2000 SUMIKAWA Munechika <>
* freebsd[234]: hardcode NEW_STRUCT_ROUTE to 1
* freebsd4/usr.sbin/lpr: IPv6 support from kame/freebsd3
2000-09-13 JINMEI, Tatuya <>
* bsdi4/usr.sbin/racoon: was just added for build. It can be
compiled (at least in my environment) but has not been tested.
Wed Sep 13 15:01:17 JST 2000
* kame/kame/racoon:
Added some switches into racoon.conf;
- verifying a certificate or not.
- verifying the subjectAltName and the sucjectName.
- sending a certificate or not.
Wed Sep 13 14:56:23 JST 2000
* kame/racoon/doc/FAQ: add racoon FAQ. will be improved in the near
Tue Sep 12 23:48:58 JST 2000
* kame/kame/racoon:
- Fixed the parameter of ipsecdoi_checkph1proposal() of initiator side.
RFC 2409 says the last byte is not included the padding length.
- Fixed to parse the responder's proposal on initiator. Error happened
that the responder responded second proposal.
Tue Sep 12 17:52:01 JST 2000
* sys/netinet6/esp_input.c: workaround to recent IPv6 ESP problem
(started early Sep with ESP core crypto engine rewrite).
perform m_pullup2 equivalent on IPv6 ESP transport mode, to avoid
non-continuous mbuf. (this is not necessary for m_pulldown case)
KAME PR 283. Thanks to TAHI guys and
2000-09-12 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_forward.c (ip6_forward): if the incoming
interface equals to the outgoing one, and the link attached to the
interface is point-to-point, then immediately drop the packet and
send an ICMPv6 destination unreachable error message.
This is currently experimental and needs the PROHIBIT_P2PREDIRECT
kernel compilation option.
Tue Sep 12 14:21:27 JST 2000
* kame/faithd: change default behavior with no arguments.
old behavior (relay telnet session) does not seem like a sensible
default - the behavior is historical. now faithd will die with
usage on no argument.
WARNING: you may need to change startup scripts, if you rely upon
old "no argument" behavior.
Tue Sep 12 00:37:48 JST 2000
* kame/kame/racoon:
Enabled to negotiate unique SAs. For example, you can configured
each SAs of TCP and UDP between two nodes.
Mon Sep 11 14:54:04 JST 2000
* netbsd/pkgsrc/net/bind9: use bind 9.0.0rc5.
2000-09-10 JINMEI, Tatuya <>
* kame/kame/pim6sd/pim6_proto.c (receive_pim6_join_prune):
improved the algorithm of random delay calculation; the result of
the previous algorithm was always 0 in spite of using random().
This fix was based on a comment from
2000-09-10 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_mroute.c (pim6_input): added validation
for the IP version of inner packets.
2000-09-10 JINMEI, Tatuya <>
* kame/kame/pim6sd/pim6_proto.c (receive_pim6_register): improved
- added validation for the length of register messages.
- added validation for the IP version of inner packets, according
to the recent discussion in the IETF pim ML.
Fri Sep 8 15:58:38 JST 2000
* freebsd3/ports/apache13, netbsd/pkgsrc/www/apache13:
apache 1.3.12 + latest IPv6 patch.
Fixed to parse the hostinfo when proxy case.
Thu Sep 7 14:03:06 JST 2000
* rtadvd/config.c: use getifaddrs to grab list of configured interface
address prefixes. previous code used routing table dump, and can be
confused by misconfigured routing table (result in bogus prefix
information option to be attached to RA).
suppress bogus warning on source link-layer address option, attached
to RA packets.
XXX uses RTM_ADD/DELETE to check for new interface addresses. should
2000-09-07 JINMEI, Tatuya <>
* kame/kame/ping6/ping6.c (pr_nodeaddr): printed address lifetime
according to the latest specification.
This change was based on a patch from Hideaki YOSHIFUJI
2000-09-07 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (ni6_store_addrs): conformed to the
latest (05 or later) specification:
- put TTL (valid lifetime) of each address before the address
- put preferred addresses before deprecated ones.
Wed Sep 6 16:32:42 2000 SUMIKAWA Munechika <>
* freebsd3/ports/fetchmail: update to 5.5.1
* freebsd3/ports/w3m: IPv6 ftp support
Wed Sep 6 16:20:45 JST 2000
* netbsd/pkgsrc/mail/fetchmail: update to 5.5.1
Wed Sep 6 16:08:52 2000 SUMIKAWA Munechika <>
* freebsd3/ports/xmms: xmms-1.2.2 with IPv6 patch
Mon Sep 4 10:47:06 2000 SUMIKAWA Munechika <>
* freebsd3/ports/lynx: upgrade to 2.8.4dev9
Fri Sep 1 22:14:27 2000 SUMIKAWA Munechika <>
* freebsd3/ports/rsync: use latest IPv6 patch.
- freeaddrinfo(NULL) was executed when reverse looking up of
the client was failed in server mode.
- -4/-6 options did not work in rsync client mode.
Reported by:
Submitted by:
Fri Sep 1 JST 2000
* kame/rtadvd/if.c: repair interface MTU lookup.
Fri Sep 1 02:30:02 JST 2000
* {net,open}bsd/lib/libinet6/getaddrinfo.c: on /etc/hosts lookup, set
"official host name" (the leftmost hostname) into ai_canonname.
this is to synchronize with practice in gethostby*.
comment from ume.
* sys/crypto: repair behavior on LP64 architecture. blowfish and
des did not work at all (or behave strange), due to issues in SSLeay.
* kame/racoon: allow comment line in pre-shared key setting.
Thu Aug 31 23:47:31 JST 2000
* kame/kame/racoon:
Enable to send either the DN or SubjectAltName as ID payload.
Compared between the value of ID payload and the name of the
certificate if it needed.
Thu Aug 31 20:04:27 JST 2000
* freebsd2/ports/sendmail.beta: upgrade to 8.11.0.
Thu Aug 31 15:18:39 JST 2000
* kame/kame/pfkey.c:
- Fixed to check if kernel supports the algorithm. the routine
overrided the map of algorithms supported whenever a user calls
- added ipsec_check_keylen2(). It's called with sa-type instead
- added pfkey_set_supported(). It's called with sadb_msg buffer,
and make the map.
Thu Aug 31 01:57:06 2000 SUMIKAWA Munechika <>
* freebsd3/ports/heimdal: upgrade to 0.3b but now it is broken due
to use old 2292API.
2000-08-30 JINMEI, Tatuya <>
* bsdi4/sys/netinet/ip_input.c (ip_forward): made the code
friendly with a BSDI's official patch.
From: Takashi Taniguchi <>
2000-08-30 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (ni6_input):
* kame/sys/netinet6/icmp6.c (ni6_addrs):
* kame/sys/netinet6/icmp6.c (ni6_store_addrs):
some corrections and clarfications about nodeinfo node addresses:
- fixed misuse of the A flag
- fixed a bug to produce truncated responses
- checked the subject address instead of ip6_dst in ni6_addrs
- updated comment based on the latest draft(06)
Wed Aug 30 15:02:27 JST 2000
* kame/ping6/ping6.c, sys/netinet6/in6_ifattach.c:
downcase hostname/DNS name, before computing NI group address.
From: yoshfuji
2000-08-29 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6_rtr.c (prelist_update): reverted the
original two-hour rule for updating address lifetime in RFC 2462.
We've adopted the "Jim Bound algorithm", but we found no strong
reason to do so through recent discussions.
Tue Aug 29 16:29:49 JST 2000
* netbsd/pkgsrc/net/{libpcap,tcpdump}: use 2000/8/28 weekly snap
* freebsd2/ports/{libpcap,tcpdump}: use 2000/8/28 weekly snap
2000-08-29 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6_rtr.c (prelist_update): clarification
about the "two-hour" rule in RFC 2462 (currently disabled,
- do not take into account if the pakcet is authenticated unless
storedlifetime <= TWOHOUR.
- changed the semantics of storedlifetime
from "Valid Lifetime in the previously received advertisement"
to "*remaining* lifetime of the stored entry".
Tue Aug 29 01:45:33 JST 2000
* sys/netinet6/esp_core.c: experimental twofish/rijndael (AES
candidate) support in kernel IPsec, based on
draft-ietf-ipsec-ciph-aes-cbc-00.txt. only either of them can be
enabled. by default, rijndael is used. if you define
IPSEC_ESP_TWOFISH in your kernel configuration file, twofish is used.
(this is due to library namespace issue - we are using the sample
code submitted to AES, and they use conflicting symbols)
TODO: racoon support
Mon Aug 28 23:22:59 JST 2000
* sys/netinet6/esp_core.c: revisit ESP crypto algorithm handling.
use block cipher code from esp_cbc_{en,de}crypt(), so that we can
easily switch the block cipher code, and reuse mbuf chasing part.
* sys/netinet6/esp_core.c: it looks that the old blowfish support code
has some bug. switch to the new codebase. old codebase and
new codebase does not interop with each other. the code will be
bringed into *BSD after we go through real interop tests.
[[the commit log was found to be wrong - blowfish was broken from
Aug28 to Sep19]]
Mon Aug 28 02:25:14 JST 2000
* sys/netkey/key.c: stir ESP IV better. on some of operating system
platforms, kernel malloc() gives region with almost constant content.
Sun Aug 27 21:25:44 JST 2000
* sys/netinet6/esp_core.c: ESP internal function API change;
algo->{en,de}crypt frees mbuf on error.
2000-08-25 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.c (in6_lifaddr_ioctl): made
SIOC[DG]LIFADDR scope-aware; fe80::/64 (not /10) will be accepted
to specify a link-local address.
In response to a report from Francis Dupont
Thu Aug 24 20:17:26 JST 2000
* kame/kame/racoon:
It is possible to verify the signer chain in the certificate.
You must have all of the certificate of each authorities before
the certificate verification.
Also racoon can output a readable certificate to the logfile during
XXX the caomparison between ID value and Subject{,Alt}Name is not yet.
Thu Aug 24 16:07:40 JST 2000
* netbsd/pkgsrc/net/bind9, freebsd3/ports/bind9: use bind
Thu Aug 24 16:01:26 JST 2000
* kame/kame/racoon:
Fixed IV processing. IV mismatching happened when a peer sent a
encrypted informational exchange on phase 1. Also added a comment
about IV processing in handler.h. And deleted ivd in IV hander
because it is useless.
Wed Aug 23 22:42:19 JST 2000
* kame/kame/racoon:
Added the behavior of PFS selection in the case of "strict"/"claim".
If PFS is not required by the responder, the responder obeys the
proposal. If PFS is required by both sides and if the responder's
group is not equal to the initiator's one, then the responder reject
the proposal.
Wed Aug 23 18:13:50 JST 2000
* add ALTQ support to the dc driver and ethernet bridging of openbsd.
Wed Aug 23 15:20:59 JST 2000
* kame/kame/racoon:
Added a level of "exact" to "proposal_check" directive. When the
responder define it, the responder will reject the proposal if each
lifetime/lifebytes/pfs are not same between the initiator and the
Wed Aug 23 14:18:44 JST 2000
* netbsd/sbin/ping6: provide stddev value at the end, to follow
practice in netbsd ping(8).
Wed Aug 23 13:19:59 2000 SUMIKAWA Munechika <>
* kame/freebsd3/ports/squid11: use latest IPv6 patch.
Wed Aug 23 12:30:02 JST 2000
* sys/netinet6/route6.c: dropped COMPAT_RFC2292 clause.
it still have binary backward compatible behavior with RFC2292,
modulo strict/loose bit.
Wed Aug 23 10:43:07 JST 2000
* kame/kame/racoon:
Enabled to check PFS group strictly. It can be done by
defining "strict" to "proposal_check" in "remote" directive.
Tue Aug 22 17:19:46 JST 2000
* kame/sys/netinet6/in6_ifattach.c (in6_ifattach): application
using routing-socket (e.g. zebra) can detect automatic addition
of linklocal I/F address by RTM_NEWADDR message.
Mon Aug 21 13:33:12 2000 SUMIKAWA Munechika <>
* kame/freebsd4/ports/{ct, v6eval}: TAHI test tools
Mon Aug 21 19:17:17 JST 2000
* freebsd3/ports/apache13: apache 1.3.12 + latest IPv6 patch
2000-08-21 JINMEI, Tatuya <>
* kame/sys/netinet6/route6.c (ip6_rthdr0): made the code friendly
with the old advanced API (RFC2292). Note that the kernel
compilation option "COMPAT_RFC2292" is necessary to enable this.
2000-08-21 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6_rtr.c (prelist_update): corrected the
interpretation of "StoredLifetime" in the "Jim Bound algorithm".
Pointed out by OKABE Nobuo <>.
Mon Aug 21 13:33:12 2000 SUMIKAWA Munechika <>
* kame/freebsd3/ports/{ct, v6eval}: upgrade to 1.1
2000-08-21 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_mroute.c (ip6_mforward): do not forward
packets with the unspecified source address. This is rather more
serious than unicast cases, because some MLD packets can be sent
with the unspecified source address (although such packets must
normally set 1 to the hop limit field).
Sun Aug 20 20:14:36 2000 SUMIKAWA Munechika <>
* freebsd3/ports/zebra: upgrade to 0.88.
Sun Aug 20 20:23:03 JST 2000
* freebsd3/ports/rsync, netbsd/pkgsrc/net/rsync: upgrade to 2.4.5.
Sun Aug 19 JST 2000
* sys/netinet6/ip6_output.c: repair mbuf manipulation in outgoing
jumbogram support
* sys/netinet6: some pedant fix for memory access alignment constraint.
* sys/netinet6/ip6_output.c: add diagnostic code for advanced API
memory allocation (there seem to be some memory leak, need
Fri Aug 18 23:53:15 JST 2000
* sys/crypto/rc5: drop IPsec RC5 support.
Fri Aug 18 20:29:23 2000 Shin'ichi Fujisawa <>
* remove kame/kame/sys/netpm/*.[ch] because this code is obsolete.
NOTICE: If you want to cleanup old symlinks,
perform "make TARGET=netbsd clean", in the top directory.
Of course, you should select TARGET name for proper one.
Fri Aug 18 14:22:15 JST 2000
* kame/sys/netinet6/in6.c: always assign prefixlen 128
for non-linklocal prefix of P2P and loopback I/F.
This feature is enabled only when you define
Fri Aug 18 11:34:01 JST 2000
* netbsd/pkgsrc/net/zebra: upgrade to 0.88.
Thu Aug 17 19:27:34 JST 2000
* netbsd/pkgsrc/net/zebra: upgrade to 0.88pre2.
Thu Aug 17 10:19:13 JST 2000
* openbsd: change kernel headers to use "NEW_STRUCT_ROUTE" definition.
kernel can panic if you don't define it in kernel configuration file.
Wed Aug 16 15:41:28 JST 2000
* netbsd/pkgsrc/net/wget: upgrade to latest IPv6 patch.
IPv6 FTP is now available.
2000-08-16 JINMEI, Tatuya <>
* bsdi4/sbin/ifconfig/ifconfig.c: set some IPv6 related parameters
to a correct variable. Without this fix, for example, an IPv6
anycast address couldn't be configured with the ifconfing command.
Tue Aug 15 22:28:09 2000 SUMIKAWA Munechika <>
* freebsd3/ports/apache13: apache 1.3.12 + latest IPv6 patch
* freebsd3/ports/mrt: mrt 2.2.2a-Aug11
* freebsd3/ports/lynx: upgrade to 2.7.4dev7 which IPv6 patch was
* freebsd3/ports/python: use latest IPv6 patch.
Tue Aug 15 18:16:14 JST 2000
* netbsd/pkgsrc/lang/python: use latest IPv6 patch.
* netbsd/pkgsrc/lang/python-current: python 1.6b1 + IPv6 patch
* netbsd/pkgsrc/www/apache13: apache 1.3.12 + latest IPv6 patch
* netbsd/pkgsrc/net/mrt: mrt 2.2.2a-Aug11
Tue Aug 15 16:41:23 JST 2000
* netbsd: change kernel headers to use "NEW_STRUCT_ROUTE" definition.
kernel can panic if you don't define it in kernel configuration file.
Tue Aug 15 16:39:43 2000 SUMIKAWA Munechika <>
* freebsd3/ports/postfix: postfix 20000531 + IPv6 patch.
Tue Aug 15 16:24:51 JST 2000
* kame/sys/net/if_stf.c, kame/sys/netinet/in_gif.c:
repair IN_MULTICAST() check for bsdi/freebsd.
IN_MULTICAST() and IN_CLASS[A-D] macro has different endianness
in *BSDs.
Mon Aug 14 21:22:20 JST 2000
* netbsd/pkgsrc/devel/boehm-gc: upgrade to 5.0a7
* netbsd/pkgsrc/www/w3m: upgrade to 0.1.10
Mon Aug 14 18:15:54 JST 2000
* freebsd{2,3}/sbin/ifconfig/ifconfig.c
+ don't assign dummy address to an interface when it's brought
up by "ifconfig up". (same fix has to be done on FreeBSD-4.0)
Mon Aug 14 16:07:45 JST 2000
* netbsd/pkgsrc/security/openssl: update to 0.9.5a.
* netbsd/pkgsrc/mail/postfix-current: postfix 20000531 + IPv6 patch.
Mon Aug 14 13:05:11 2000 SUMIKAWA Munechika <>
* kame/freebsd3/ports/mozilla: upgrade to M17.
Sun Aug 13 11:09:44 JST 2000
* kame/rtsold/rtsold.c: add -a flag, to autoprobe interface.
suggested by
Sun Aug 13 09:45:09 JST 2000
* kame/route6d: repair LP64 problem. from
2000-08-12 JINMEI, Tatuya <>
* kame/kame/pim6sd: implemented an experimental (and KAME
proprietary) hello option "additional addresses".
* kame/kame/pim6sd/route.c (set_incoming): when upstream router
determination, consider neighbor's additional addresses got
through the new option. This solves a case of mismatch between RP
and PIM neighbor.
2000-08-12 JINMEI, Tatuya <>
* kame/kame/pim6sd/main.c (cleanup): upon receipt of SIGTERM,
inform all neighbors the termination by sending a hello message
with 0 holdtime.
2000-08-12 JINMEI, Tatuya <>
* freebsd3/usr.bin/netstat/if.c (intpr): supported IPv6 scoped
addresses format.
2000-08-12 JINMEI, Tatuya <>
* kame/kame/pim6sd/pim6_proto.c (receive_pim6_bootstrap): added a
pair of braces to avoid unintentional failure.
In response to the PR kit/279 from Daniel Elphick
2000-08-11 JINMEI, Tatuya <>
* freebsd3/ports/bind9/: upgraded to 9.0.0rc2.
2000-08-11 JINMEI, Tatuya <>
* freebsd4/include/.prepare: was added to ignore ifaddrs.h
conflict between the FreeBSD's original version and the KAME's
shared version.
In response to a report from Tetsuya Isaki
2000-08-11 JINMEI, Tatuya <>
* kame/kame/ping6/ping6.c (pr_pack): printed multple sets of
bitmaps for NI supported QTYPEs.
(from Hideaki YOSHIFUJI <>)
Fri Aug 11 12:00:47 JST 2000
* bsdi3/sys/netiso: if NEW_STRUCT_ROUTE is defined, just use
`struct route' instead of struct route_iso.
(compilable, but not tested)
Thu Aug 10 04:42:35 JST 2000
* kame/kame/racoon:
Implemented responder lifetime selection. It does not conform to
the description of RFC2407 completely. There are three type of the
behavior by specifing the value of "proposal_check" in "remote"
In phase 2 case, a RESPONDER-LIFETIME notify payload includes into
the exchange if needed. To send the payload is not processed in
phase 1 case yet.
Thu Aug 10 01:19:58 JST 2000
* netbsd/pkgsrc/net/bind9: upgrade to 9.0.0rc2.
Wed Aug 9 22:37:14 JST 2000
* kame/ping6/ping6.c: support compresed DNS label string. be more
picky about DNS label validation.
Mon Aug 7 23:59:51 JST 2000
* libinet6/resolv/res_send.c: correct sendto() timeout retry loop.
old code made infinite loop in EINTR-busy environment.
NetBSD PR 6410.
2000-08-07 JINMEI, Tatuya <>
* bsdi4/sys/netiso: if NEW_STRUCT_ROUTE is defined, just use
`struct route' instead of struct route_iso.
(compilable, but not tested)
Sun Aug 6 05:41:47 2000 SUMIKAWA Munechika <>
* kame/freebsd4: we now start to support 4.1-RELEASE.
Fri Aug 4 00:16:00 JST 2000
* kame/ping6/ping6.c: support icmp6 node information
"supported query types" query, by ping6 -t.
Thu Aug 3 05:36:23 JST 2000
* kame/kame/racoon:
Improved to check a packet which is processed or not. The way is
to compare between the hash of a packet has been processed and the
list of hashs which are processed before.
Thu Aug 3 03:00:36 JST 2000
* kame/sys/{netkey/{key.c,keydb.h},netinet6/ipsec.h}:
Improved lifetime handling. It uses real time instead of tick.
XXX year 2038 problem remains.
Thu Aug 3 00:49:58 JST 2000
* kame/kame/racoon:
Tryed sending DELETE message against ISAKMP/IPsec SAs deleted
before racoon terminates.
Thu Aug 2 JST 2000
* sys/netinet6/icmp6.c: implement icmp6 nodeinfo "supported qtypes".
* sys/netinet6: remove bogus DIAGNOSTIC cases for sizeof(ro.ro_dst).
Tue Aug 1 23:53:36 2000 SUMIKAWA Munechika <>
* freebsd3/ports/ja-mnews: upgrade to 1.22PL4.
2000-08-01 JINMEI, Tatuya <>
* bsdi4/sys/netiso/tp_inet.c (tpip_output_dg):
* bsdi4/sys/netiso/tp_inet.c (tpip_ctlinput):
adjusted arguments to ip_output() and in_pcbnotify() for the
KAME code.
Tue Aug 1 04:03:29 JST 2000
* kame/kame/racoon:
No weak key check perform on racoon. She leaves this check to kernel.
Tue Aug 1 00:20:44 2000 SUMIKAWA Munechika <>
* kame/freebsd3/lib/libftpio: try IPv4 if IPv6 connecting was
From: FreeBSD-current
Mon Jul 31 23:26:19 2000 SUMIKAWA Munechika <>
* kame/freebsd4/sys/net/if_types.h: move private use IFT_xx to
0xf0 for syncing FreeBSD-current.
* kame/freebsd3/sys/net/if_types.h: sort IFT_xx for syncing
WARNING: you must recomplie all KAME applications.
2000-07-30 JINMEI, Tatuya <>
* kame/sys/netinet6/(various files): made FAKE_LOOPBACK_IF default
(as previously announced). At this moment, the older behavior can
be specified by the "OLD_LOOPBACK_IF" kernel compilation option.
2000-07-30 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6_rtr.c: made ND6_USE_RTSOCK option default.
2000-07-30 JINMEI, Tatuya <>
* kame/kame/pim6sd/vif.c (init_vifs):
* kame/kame/pim6sd/cfparse.y (phyint_config):
made sure to check if a valid global address exists after parsing
the configuration file, in order to avoid unexpected hang up.
From: Kengo NAGAHASHI <>
Sat Jul 29 JST 2000
* freebsd2/bsdi3: repair NEW_STRUCT_ROUTE support.
Sat Jul 29 07:57:41 2000 SUMIKAWA Munechika <>
* freebsd3/ports/{libpcap,tcpdump}: use 2000/7/24 weekly snap
Fri 28 Jul 18:56:23 JST 2000
* make altqd aware of token bucket regulators
- install a token bucket regulator when an interface is
attached unless there is already one installed.
- remove the installed tbr when the interface is detached.
Fri Jul 28 12:46:55 JST 2000
* netbsd/sys/netinet/tcp_output.c: add missing call to tcp6_quench().
* GENERIC.v6 config file for all *BSD:
enable NEW_STRUCT_ROUTE and FAKE_LOOPBACK_IF for torture-testing,
which will be enabled by default soon.
Thu Jul 27 22:40:51 JST 2000
* sys/netinet6/ip6_forward.c: do not forward packets with unspecified
IPv6 source. this is not on the documents, but it seems to have got
rough consensus on ipngwg (July 2000).
Thu Jul 27 12:16:52 JST 2000
* tcp6 layer (all *BSD):
be proactive about unspecified IPv6 source address. pcb layer uses
unspecified address (::) to mean "unbounded" or "unconnected",
and can be confused by packets from outside.
Wed Jul 26 JST 2000
* sys/netinet6: repair faith support. there are couple of mistakes
we had: (1) m->m_pkthdr.rcvif no longer points to faith* interface,
however, tcp/udp/icmp6 layer tried to check rcvif. this results in
mysterious icmp6 reply from faith relaying node, for translated
destinations. (2) in many places "faith.h" was not included
Wed 26 Jul 19:55:19 JST 2000
* add tbrconfig(8), a tool to configure a token bucket regulator
for an interface.
(need ALTQ configured in the kernel. A SMP box should have
ALTQ_NOPCC not to use processor cycle counter)
tbrconfig(8) is originally written for tuning ALTQ, but
can be used as a handy tool to rate-limit an interface.
for example, to rate-limit the fxp0 interface up to 25Mbps,
# tbrconfig fxp0 25M auto
see tbrconfig(8) for more details.
Wed Jul 26 12:52:50 JST 2000
* kame/kame/racoon:
Fixed to get a destination address when a packet is received from
a peer. It happened when a interface had similar multiple addresses.
Wed Jul 26 11:08:23 JST 2000
* sys/netinet6/ip6_input.c: reject IPv6 packet with IPv4 mapped
address in the header. this makes more sense as we cannot be
put into SIIT environment (no BSD supports non-INET kernel
compilation). This is an undo for a change on Mar 28 2000.
Wed Jul 26 11:01:41 JST 2000
* netbsd/pkgsrc/net/{libpcap,tcpdump}: use 2000/7/24 weekly snap
2000-07-25 JINMEI, Tatuya <>
* freebsd3/usr.bin/telnet/commands.c (tn): set the source address
specfied by the -s option correctly.
KAME PR sys/272.
Tue 25 Jul 19:12:42 JST 2000
* the first round of ALTQ interface mega cleanup:
- cleanup the system interface to the rest of the kernel.
(the second round cleans up altq internals.)
* bring in the new output queue model.
the new model removes "#ifdef ALTQ" from most of the
drivers, and thus, touches all the drivers used by ALTQ.
- the type of if_snd in struct ifnet is changed from
struct ifqueue to struct ifaltq.
- the altq related fields are moved from struct ifnet
to struct ifaltq.
- use the newly introduced IFQ_XXX macros instead of
IF_XXX macros
* the new altq mechanism consists of classifier, queueing
discipline, and token bucket regulator.
- add token bucket regulator to control network devices.
tbr decouples driver tuning from disciplines.
* other changes:
- removed 2 kernel config options
ALTQ_ACCOUNT: should be part of classifier
AFMAP: doesn't really belong to altq
- added kernel option
ALTQ_NOPPC: do not use processor cycle counter
- support alpha architecture on netbsd/openbsd.
make use of alpha_rpcc() for a high resolution clock.
- add new driver support: ex, wi, tun, ppp
- some other style(9) cleanup
Mon Jul 24 23:34:26 JST 2000
* *bsd*/usr.bin/telnet/commands.c: do not use IPV6_PKTOPTIONS in
2292bis API environment. IPv6 source route is temporarily disabled.
* libinet6/if_nameindex.c: malloc length computation bug. from kjc.
Mon Jul 24 09:31:25 JST 2000
* sys/netinet6/in6.h: try to provide binary backward compatibility for
RFC2292 API, deployed into freebsd40/netbsd15/openbsd27.
TODO: check 2292/bis-on-TCP behavior.
Mon Jul 24 08:22:29 2000 SUMIKAWA Munechika <>
* freebsd3/ports/netperf: use the latest IPv6 patch (7/21).
Sat Jul 22 01:05:03 JST 2000
* kame/kame/racoon:
Trying to conform with draft-ietf-mobileip-ipv6-12. If "support_mip6"
in remote directive is on, both values of ID payloads in phase 2
exchange are always used as the addresses of end-point of IPsec-SAs.
Sat Jul 22 JST 2000
* sys/netinet6/esp_core.c: cache intermediate key for ESP encryption
algorithms, into SAs. this will drastically improve performance
for algorithms with long key setup time (blowfish).
KAME PR 229. suggested by sommerfeld.
Fri Jul 21 22:37:44 JST 2000
* netbsd/pkgsrc/mail/sendmail.beta: use 8.11.0.
* netbsd/pkgsrc/net/netperf: use the latest IPv6 patch (7/21).
2000-07-21 JINMEI, Tatuya <>
* kame/kame/bgpd/bgp.c: clarification about the advanced API;
do not use IPV6_PKTOPTIONS in pure RFC2292 (i.e. not 2292bis)
Thu Jul 20 01:52:59 2000 SUMIKAWA Munechika <>
* freebsd3/ports/{libpcap,tcpdump}: use 7/17 weekly snap.
* freebsd3/ports/mtr: use latest IPv6 patch.
Wed Jul 19 JST 2000
* kame/kame/racoon:
- when racoon is running local test mode, AND initial contact is
received, don't delete phase 2 SA.
- more improve to delete phase 1 SA on receiving initial contact.
Wed Jul 19 22:22:47 JST 2000
* kame/rtsold/rtsol.c, kame/rtadvd/rtadvd.c: ip6_var.h requires
* netbsd/pkgsrc/net/{libpcap,tcpdump}: use 7/17 weekly snap.
Tue Jul 18 JST 2000
* kame/kame/racoon:
- don't send notify message in processing of information exchange.
- fixed a possible memory leak when error occure.
- added some message about comparing each values in proposal.
- improved to delete phase1 SA when delete message is received OR
initial contact is received.
Mon Jul 17 JST 2000
* kame/kame/racoon:
- commented when id payload of subnet type with full bit masked.
Mon Jul 17 18:45:11 JST 2000
* sys/netkey/key.c: allow ESP with no authentication. it was
mistakenly forbidden with the recent key.c change.
Mon Jul 17 03:31:47 2000 SUMIKAWA Munechika <>
* freebsd3/ports/ruby: upgrade to 1.4.5.
Sun Jul 16 JST 2000
* kame/kame/racoon:
- commented when id payload of subnet type with full bit masked.
- racoon.conf are updated about padding option.
Sun Jul 16 16:44:24 JST 2000
* sys/netinet6/ip6_forward.c: enable IPSEC_IPV6FWD case by default.
(actually, previous default behavior was wrong as it transmits
forwarded packets in clear, even though the policy asks for
Sun Jul 16 14:32:58 JST 2000
* kame/ping6:
- check duplicated replies for node information query (*)
- fill nonce field for node information query (*)
- use generic icmp6 printer for echo/echo reply/ni reply/ni reply,
when ident/nonce does not match
- clarlify outgoing packet construction a bit
(*) based on patch from yoshfuji, cleaned up by itojun
Sun Jul 16 13:26:12 JST 2000
* sys/netinet6/icmp6.c: ICMPv6 node information query now based on
06 draft (ping6 -w).
Sun Jul 16 00:57:23 JST 2000
* sys/netinet6/in6.h; do not pull sys/queue.h in (it is not
necessary anyways)
* sys/net/pfkeyv2.h: correct conformance to RFC2367 (SADB_[EA]ALG_xx
symbol name). beware: the change breaks backward compatibility.
setkey and racoon MUST be recompiled after updating include files.
Sat Jul 15 JST 2000
* kame/kame/racoon:
- fixed the place of checking whether delete payload is protected
when delete notification is received.
- When a IKE node receive a delete payload, the node always delete
outbount SAs only. So destination IP address have to check before
SA will be deleted.
- Delete phase 1 handler when a delete notification has been received.
Renamed purge_spi() to purge_ipsec_spi() accompanied with above
- commented about the reason why we don't send delete payload
for outbound SAs.
- added a option to randomize values in a padding. clarified to
randomize length.
- fixed some of memory leak.
Sat Jul 15 13:51:59 JST 2000
* kame/route6d: if a routing entry exists for aggregate prefix (-A),
do not overwrite it (exit with error). it should be a safer behavir.
Fri Jul 14 JST 2000
* kame/kame/racoon:
- enable to switch sending initial contact.
- try to send delete payload on phase 1 when phase 1 sa is deleted
- fixed to compare address family in two sockaddrs.
- fixed making a length of IPv6 ID payload. it was used a length of
struct in_addr.
- fixed a part of sending a notification on phase 2.
- fixed making a ID payload of type of IPv4 address. There was
unnecessary space in a address part.
Thu Jul 13 22:27:18 JST 2000
* openbsd/sys/netinet6/raw_ipv6.c: enable IPv6 multicast routing
related setsockopt.
* kame/sys/netinet6/ip6_mroute.c: to enable openbsd users to perform
netstat -g, make mif6table a non-static variable. on openbsd
file static variables will not appear in kernel symbol table.
Thu Jul 13 16:09:57 JST 2000
* netbsd/pkgsrc/net/bind9: upgrade to 9.0.0rc1.
Thu Jul 13 01:39:26 JST 2000
* sys/kern/uipc_mbuf2.c: cleanup m_pulldown statistics.
(1) PULLDOWN_STAT is now a global compilation option (should be
put into kernel configuration file). (2) m_pulldown statistics
now belong to mbstat, and available via netstat -m (instead of
netstat -sn -f inet6). suggested by jinmei.
2000-07-12 JINMEI, Tatuya <>
* *bsd*/sys/net/route.h: redefined the route structure so that it
can support protcols that have large socket address (e.g. IPv6).
Currently, this is enabled only with the NEW_STRUCT_ROUTE kernel
compilation option, but will be default once stabilized.
* many files mainly under the netinet and netinet6 directories
were also modified with this change.
2000-07-12 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.c (in6_ifloop_request):
do not force rtrequest() to return an rtentry when executing the
DELETE operation, in order to avoid overdecreasing the refcnt.
Older versions might cause leak of rtentry when you delete an IPv6
address (via ifconfig, ndp -P, or something).
Fortunately, address deletion is not issued so often, the bug is
effectively not very serious. However, if you have chance to
update your kernel, it is of course recommended to apply this fix.
In particular, KAME's dtcp or ppp (for IPv6) users are highly
recommended to upgrade the kernel.
2000-07-11 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.h: commented out the definition of
ND6_LLINFO_WAITDELETE, which is not used any more.
* kame/sys/netinet6/nd6.c:
* kame/sys/netinet6/mip6_md.c:
* kame/kame/ndp/ndp.c:
removed ND6_LLINFO_WAITDELETE cases according to the above change.
2000-07-11 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_mroute.h: corrected the type of a member
of if_set{} from fd_mask to if_mask.
In response to PR sys/266 from
2000-07-10 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6_rtr.c (rt6_deleteroute): do not
(automatically) delete the static route in rt6_deleteroute(), even
if it uses a dead router.
2000-07-10 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c: some clarifications about neighbor
cache manipulation (this change does not affect behavior from the
user side, though):
- removed the ND6_LLINFO_WAITDELETE status. Actually, we can just
call rtrequest(RTM_DELETE) for an unreachable
neighbor. Reference to the neighbor cache entry from a cahced
route will be freed at the next time the route is used.
- also, we do not have to call pfctlinput(PRC_HOSTDEAD) in
nd6_free() for the same reason.
- do not set/refer the RTF_REJECT flag in neighbor cache
manipulation. It was just for (IPv4) arp-flooding prevention,
which is not necessary ND for IPv6.
2000-07-10 JINMEI, Tatuya <>
* kame/kame/route6d/route6d.c: removed "ifndef ADVAPI"
parts. Since the advanced API has already been standardized,
implemented, and deployed, we don't need to take care of the older
kernel behavior (which is even confusing).
* *BSD/usr.sbin/route6d/Makefile: removed the -DADVAPI flag
according the change.
Mon Jul 10 14:43:40 2000 SUMIKAWA Munechika <>
* freebsd3/ports/{libpcap,tcpdump}: use 7/3 snapshot.
* freebsd3/ports/w3m: upgrade to 0.1.11.p.
Sun Jul 9 21:50:54 JST 2000
* *bsd*/sys/netinet/tcp_input.c, kame/sys/netinet6/tcp6_input.c:
be more cautious about tcp option length field. drop bogus ones
not sure if there is a real threat or not, but it seems that there's
possibility for overrun/underrun (like non-NOP option with
optlen > cnt). the bug is from 4.4BSD.
Sun Jul 9 13:39:22 JST 2000
* libinet6/getaddrinfo.c: do not mistakenly accept empty scopeid.
Sun Jul 9 12:29:24 JST 2000
* freebsd4/sys/net/if_ethersubr.c: repair IPV6_JOIN_GROUP(::).
Sat Jul 8 12:11:34 JST 2000
* netbsd/pkgsrc/net/bind9: use bind 9.0.0b5
* netbsd/pkgsrc/net/{libpcap,tcpdump}: use 7/3 snapshot.
Sat Jul 8 10:57:36 JST 2000
* {netbsd,openbsd}/usr.sbin/inetd: allow square-bracket for the first
element on inetd.conf, to disambiguate IPv6 address and colon
* openbsd/usr.sbin/inetd: handle IPv6 address in first element on
inetd.conf line.
Sat Jul 8 09:43:26 JST 2000
* {bsdi3,openbsd,netbsd}/libexec/ftpd: plug setproctitle issue in
CERT Advisory CA-2000-13. NOTE: bsdi3 uses wu-ftpd. it may have
other vulnerabilities left in the code.
* netbsd/usr.sbin/inetd: improve error handling on getaddrinfo
(determine listening socket address). hints from enami.
Fri Jul 7 21:39:33 JST 2000
* various places: audit use of printf-like functions, including
errx?, warnx?, setproctitle, and syslog. if we pass user-supplied
variable alone to these functions, they can be hosed by malicious
%-format string. from openbsd.
Thu Jul 6 20:43:57 JST 2000
* openbsd/sys/netinet/tcp_*.c: remove IPv4 mapped support completely
from inbound packet processing. there were some corner cases not
covered by the code, and it caused SEGV due to inconsistency in
address family. sync with openbsd-current.
2000-07-06 JINMEI, Tatuya <>
* kame/kame/rtadvd/if.c (if_getflags): made sure to close a
temporary socket to avoid making garbage sockets.
Wed Jul 5 12:08:16 JST 2000
* bsdi3/sys/conf/files.i386, bsdi3/sys/conf/GENERIC.KAME,
bsdi3/usr.sbin/wiconfig, bsdi3/usr.sbin/Makefile
ported WaveLAN driver and its configuration program from bsdi4
(geertj permitted it. Thanks!)
Wed Jul 5 11:30:39 JST 2000
* {netbsd,openbsd,freebsd4}/lib/libinet6/getaddrinfo.c,
return EAI_NODATA, instead of EAI_NONAME, on name resolution errors.
EAI_NONAME does not make sense in these situations
From: enami
Wed Jul 5 11:02:03 JST 2000
* freebsd4: add netstat -sn -f pfkey.
Wed Jul 5 10:40:53 JST 2000
* freebsd[234]: split IPv6 path MTU discovery-related sysctl from
net.inet.ip tree. FreeBSD SYSCTL_xxx does not have a way to report
duplicated definition into the same variable, it bites us many
Wed Jul 5 02:25:11 JST 2000
* kame/kame/racoon:
This message is sent by single notify message after phase 1 established
immediately. It means the message is not included last exchange on
phase 1. So it can be sent by responder on aggressive/base mode.
If there is no remote address in contacted list, racoon sends the
message to peer. If the message is received, racoon deletes all
IPsec-SAs relatived to peer's address. It takes place both initiator
and responder side.
Tue Jul 4 21:36:16 JST 2000
* kame/kame/racoon:
Racoon usually runs in background. If you specify -F option, you make
her running in foreground.
2000-07-04 JINMEI, Tatuya <>
* kame/sys/netinet6/mip6.c (mip6_add_ifaddr): use in6_update_ifa()
to assign an address instead of coping code from in6.c
2000-07-04 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.c (in6_update_ifa): newly added to update
parameters of an IPv6 interface address.
Basically, this function does nothing new, but made in6_control()
2000-07-04 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.c (in6_control): completely obsoleted
We are quite confident there is no application that used these
commands, but if one exists, please let us know.
Tue Jul 4 18:33:20 JST 2000
* kame/kame/racoon:
don't delete phase 1/2 handler if some internal error occurs.
2000-07-04 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.c (in6_is_ifloop_auto): removed an `ifdef'
part for openbsd, which made the function always return 0.
2000-07-04 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.c (in6_unlink_ifa): newly added to release
various links for in6_ifaddr when deleting an address.
This function is also called from in6_control(), in order to
prevent the kernel from keeping a garbage structure on failure of
address addition.
Tue Jul 4 13:26:56 JST 2000
* kame/sys/key.c:
A patch from <> applied.
- fixed a interval to call key_timehandler.
- fixed a typo.
- added a value to be returned when some error happen.
2000-07-04 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.c (in6_control): added several
improvements for sharing a single prefix with multiple addresses:
- install an interface direct route only when there's no shared
prefix. We'll never see unexpected EEXIST errors with this fix.
- call in6_ifaddloop()/in6_ifremloop() whenever necessary.
- do not call in6_ifaddloop()/in6_ifremloop() unless necessary.
- added several clarifications according to the ipv6 address
2000-07-04 JINMEI, Tatuya <>
* kame/kame/bgpd/main.c (main_listen_accept): set the receiving
interface when accepting an on-link bgp connection.
* kame/kame/bgpd/bgp.c (bgp_process_open): detected a proper peer
for an incoming IBGP open message with link-local address.
These changes enabled an IBGP peer using link-local addresses.
Suggested by: Tomomi Suzuki <>
Tue Jul 4 10:25:13 JST 2000
* kame/kame/racoon:
- Process to send a delete notify message only when phase 2 has
been established.
- added "dead" flag to a schedule. It is used to mark a schedule
already dead. don't delete a schedule at multiple place.
Tue Jul 4 08:44:11 JST 2000
* netbsd/usr.sbin/inetd: remove duplicated ipsec initialization code
(used on SIGHUP).
2000-07-03 JINMEI, Tatuya <>
* kame/kame/bgpd/bgp.c (connect_try): made sure to zero-clear a
newly allocated buffer.
Report from: Tomomi Suzuki <>
Mon Jul 3 11:50:12 JST 2000
* kame/sys/netinet/icmp6.h: avoid bitfields in router renumbering packet
declaration. XXX standards?
2000-07-02 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_input.c (ip6_input): immediately discarded
a packet to an unready (i.e. tentative or duplicated) address with
This change reflected recent discussion in the ipngwg ML.
Sun Jul 2 11:24:52 JST 2000
* netbsd/sys/netinet/tcp_input.c, kame/sys/netinet6/in6_pcb.c:
repair netbsd faith support. (1) tcp6_input dropped faith'ed
connections (2) in6_pcblookup_connect() was too strict.
Sat Jul 1 20:57:57 JST 2000
* kame/faithd: make it possible to invoke faithd(8) from inetd(8).
benefits: allows us to access-control inbound traffic by using
possible drawbacks: inetd mode has no chance for multi-connection-
per-single-process enhancement. current faithd(8) needs 1
process per 1 connection anyways.
Fri Jun 30 17:45:23 JST 2000
* freebsd[34]/usr.bin/whois.c:
ported whois for IPv6/4.
Thu Jun 29 16:24:35 JST 2000
*/sys/netinet/in.c, kame/sys/netinet6/in6.c:
inhibit EEXIST from in{,6}_ifinit(). history: (1) 4.4BSD ignores
return value from in_ifinit() completely. (2) previous kame code
tried to handle error case better, the change raised bogus EEXIST
to the userland on two-address-from-same-prefix assignment.
Thu Jun 29 10:14:47 JST 2000
* faithd/faithd.c, natptd/main.c, natptlog/natptlog.c: be more careful
about arg to syslog(3), to prevent possible buffer overrun.
2000-06-28 JINMEI, Tatuya <>
* kame/kame/dhcp6: several minor improvements:
- daemonized dhcp6c.
- reactivated dhcp6c agains a SIGHUP signal or change of the
default route.
- changed logging based on syslog(8).
2000-06-28 SUZUKI Shinsuke <>
* kame/sys/netinet6/ip6_fw.c
ip6fw works on FreeBSD-4.0 + KAME, too.
Wed Jun 28 15:01:09 JST 2000
* kame/kame/racoon:
improved sending a notify message including delete payload.
It's sent when one of below situations happens:
o receiving SADB_DELETE message from kernel.
o receiving SADB_FLUSH message from kernel.
o flushing phase2 handler.
Wed Jun 28 01:16:41 JST 2000 SUZUKI Shinsuke <>
* freebsd4/INSTALL
write up configuration-related matters for FreeBSD-4.0
Wed Jun 28 01:12:59 JST 2000
* libinet6/name6.c: correct error handling in DNS name lookups.
Tue Jun 27 23:12:54 JST 2000 SUZUKI Shinsuke <>
* bsdi3/sbin/ifconfig/ifconfig.c: fixed error trap when given name
corresponds to multiple v6 addresses.
Tue Jun 27 14:01:39 2000 SUMIKAWA Munechika <>
* freebsd3/ports/emacs: upgrade to 20.7
Tue Jun 27 00:32:20 JST 2000
* netbsd/sys/arch/*/conf/GENERIC.v6: enable PULLDOWN_TEST for all
architectures. this is done because mbuf pullup code in
sys/net/if_loop.c has been found to be a source of performance hit,
and PULLDOWN_TEST code is found to be stable enough.
* netbsd/sys/sys/mbuf.h: recover 4.4BSD MINCLSIZE.
Sun Jun 25 21:11:19 2000 SUMIKAWA Munechika <>
* start to support FreeBSD 3.5-RELEASE. 3.4-RELEASE is obsolete.
Sat Jun 24 23:41:31 JST 2000
* freebsd4/lib/libinet6/getnameinfo.c: correct NIS lookup. from ume.
Sat Jun 24 16:43:58 JST 2000
* netbsd/pkgsrc/net/{libpcap,tcpdump}: upgrade to 6/19.
Sat Jun 24 02:20:33 2000 SUMIKAWA Munechika <>
* freebsd3/ports/wyvern: new port, a simple web server
* freebsd3/ports/tcpd: new tcpd from Artur Frysiak <>
* freebsd3/ports/wget: fix security hole and use latest IPv6 patch
* freebsd3/ports/vnc: use latest IPv6 patch
Fri Jun 23 19:49:28 2000 SUMIKAWA Munechika <>
* freebsd3/ports/gbatnav: new port, a battleship game
* freebsd3/ports/mmosaic: upgrade to 3.6.2.
Thu Jun 22 17:40:37 JST 2000
* kame/sys/netkey:
delete sadb_x_ident_id_addr. don't send a pair of addresses
by including acquire message to a user.
* kame/kame/racoon/pfkey.c:
SADB_EXT_IDENTITY_{SRC,DST} is not required to parse SADB_ACQUIRE
message any more.
Thu Jun 22 17:45:40 JST 2000
* openbsd/sys/dev/ic/xl.c: disable multicast hash filer setup on 905B,
since the code does not do the right thing. (sync with
Thu Jun 22 03:45:41 JST 2000
* kame/sys/netinet6/raw_ip6.c, {bsdi4,openbsd}/sys/netinet6/raw_ipv6.c:
correct RFC2292bis interface selection support for
multicast packets. KAME PR261.
Wed Jun 21 17:07:55 JST 2000
* sys/netinet6/in6_src.c: make in6_recoverscope() friendly with
!FAKE_LOOPBACK_IF compilation. with previous code in6_recoverscope()
may fail to convert kernel internal representation into sockaddr_in6.
Wed Jun 21 03:18:47 2000 SUMIKAWA Munechika <>
* freebsd3/ports/leafnode+: IPv6 enabled leafnode+-2.10
Wed Jun 21 03:00:16 JST 2000
* openbsd/sys/dev/pcmcia/if_wi.c: make IPv6 work on wavelan cards.
2000-06-20 JINMEI, Tatuya <>
* kame/sys/net/if_gif.c (gif_ioctl): made sure to cast the
argument to in6_aliasreq{} for the SIOCSIFPHYADDR_IN6 command.
Without this, the validation check would reject correct requests;
i.e. you couldn't configure IPv6 physical addresses.
In response to a report from FUJIURA Toyonori <>
2000-06-20 SUZUKI Shinsuke <>
* kame/kame/ndp/ndp.c (delete, get)
supported <link-local_addr>%<link ID> on "ndp -d" and "ndp (addr)".
Tue Jun 20 14:55:10 2000 SUMIKAWA Munechika <>
* freebsd3/ports/openssh: upgrade to 2.1.0
* freebsd3/ports/bind9: upgrade to 9.0.0b4
* freebsd3/ports/lftp: upgrade to 2.2.3
* freebsd3/ports/mozilla: upgrade to M16
Tue Jun 20 12:49:28 JST 2000
* sys/netinet6/in6_proto.c: disable rate limitation for ICMPv6 error,
since (1) it makes no sense to put less-than-10ms value to here
due to UNIX timer resolution, and (2) it seems wrong to rate-limit
without considering content of the payload (like ICMPV6 type/code).
we still have pps limitation. based on comments from kjc.
Tue Jun 20 11:44:19 JST 2000
* netbsd/pkgsrc/net/bind9: use bind 9.0.0b4.
2000-06-19 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_ctloutput): corrected logic
of error detection for sooptcopyin(). This will fix the problem
that traditional RFC2292 compatible mode did not work for some
socket options (e.g. IPV6_PKTINFO) on freebsd[34].
Mon Jun 19 18:23:15 JST 2000
* kame/kame/racoon:
A path name in configuration file is always complemented if it is
not begin from slash(/). If it's begin from slash, a path name
never be complemented.
Mon Jun 19 16:51:24 JST 2000
* kame/kame/racoon:
If "non_auth" is defined in racoon.conf, any transform of AH proposal
including "non_auth" is not sent to the peer.
2000-06-19 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_output): jumped to the
"freehdrs" label before making the header chain in order to avoid
possible memory leak.
In response to the KAME problem report sys/259.
Mon Jun 19 07:41:31 JST 2000
* libinet6/resolv/res_{init,send}.c: be more backward-compatible with
past behavior. some userland code may not initialize
Mon Jun 19 04:42:55 JST 2000
* netbsd/pkgsrc/net/{libpcap,tcpdump}: use 6/12 snapshot from
Mon Jun 19 04:15:23 2000 SUMIKAWA Munechika <>
* freebsd4/sys: Security fix "FreeBSD-SA-00:25 FreeBSD/Alpha
platform lacks kernel pseudo-random number generator, some
applications fail to detect this."
2000-06-18 JINMEI, Tatuya <>
* freebsd[34]/sys/netinet6/udp6_usrreq.c (udp6_attach):
initialized inp_ip_ttl in udp6_attach for mapped addresses.
in response to a report from Hideaki YOSHIFUJI
<> (KAME-snap 2738)
2000-06-16 JINMEI, Tatuya <>
* {netbsd, openbsd}/usr.bin/netstat/inet6.c (icmp6_stats):
printed number of icmp6 error messages not sent due to rate
Fri Jun 16 05:01:24 JST 2000
* openbsd/sys: sync with OpenBSD 2.7
TODO: userland cleanup. tests (i386/conf/GENERIC.KAME compiles but
not tested).
kame/openbsd/ports does not work. we may want to remove those.
If you wish to upgrade to KAME-on-OpenBSD 2.7, make sure to
perform "make clean" at the top level to nuke symlinks, like:
% make TARGET=openbsd clean update prepare
make VERY sure that you use kame-supplied tools (like
/usr/local/v6/sbin/ping6) instead of normal ones (/sbin/ping6)
if you use KAME-enabled kernel. there are API changes between them.
Thu Jun 15 22:41:16 JST 2000
* kame/sys/netkey/key.c: correct compilation without IPSEC_ESP.
From: Matthias Drochner <>
Thu Jun 15 21:22:35 JST 2000
* kame/sys/netkey/key.c:
Fixed extension length of two key extension when dumping SA.
setkey was failed to print keys if SA has both encryption and
authentication keys.
Thu Jun 15 14:44:30 JST 2000
* kame/kame/racoon:
CR payload is only made if signature authentication method is applied.
Thu Jun 15 13:29:29 JST 2000
* kame/kame/racoon/cfparse.y:
In racoon.conf, the path of configuration file is complemented by
include directive only if there is no '/' in the path.
Thu Jun 15 13:08:25 JST 2000
* sys/netinet6/ipsec.[ch]: net.inet.ipsec.inbound_call_ike sysctl
MIB is now gone for good.
Thu Jun 15 10:01:47 JST 2000
* libinet6/resolv/res_init.c: make _res.nsaddr_list initialization
more conservative when resolv.conf is missing (or there is no
"nameserver" line). previous code chokes on IPv4-only kernel.
merge from netbsd-current.
2000-06-15 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6_rtr.c (defrouter_msg): was added to tell
user processes changes about the default router (including
deletion). This function would be called from defrouter_addreq,
defrouter_addifreq, and defrouter_delreq.
Note: this is currently experimental and is only enabled with the
ND6_USE_RTSOCK kernel compilation option.
Thu Jun 15 02:18:24 2000 SUMIKAWA Munechika <>
* freebsd3/ports/zebra: upgrade to 0.87.
* freebsd3/ports/ruby: use 1.4.4.
Thu Jun 15 02:07:53 JST 2000
* kame/man/man4/inet6.4, *bsd*/usr.sbin/inetd/inetd.8:
update wording on IPv4 mapped address and tcp4/tcp6 interaction.
Wed Jun 14 23:35:03 JST 2000
* {netbsd,openbsd}/libexec/ftpd: correct STAT command output for LPSV.
* libinet6/resolv/res_query.c: change member name for struct res_target.
"class" conflicts with C++ reserved identifier.
From: Graham Wheeler <>
2000-06-14 SUZUKI Shinsuke <>
* sys/net/if_dummy.c, freebsd4/sys/conf/files:
dummy I/F is available on FreeBSD4.0, too.
* freebsd4/sys/conf/options, freebsd4/sys/conf/files
FreeBSD-4.0 KAME with MIP6 option is available.
add mbuf flag M_PROTO6 and use it for M_MIP6TUNNEL.
Wed Jun 14 20:14:47 JST 2000
* sys/netinet6/esp_core.c: pass encryption failure code up to ESP
engine, just in case the encryption routine fails.
Tue Jun 13 21:11:28 JST 2000
* netbsd/pkgsrc/net/zebra: upgrade to 0.87.
Tue Jun 13 20:08:38 JST 2000
* openbsd/sys/netinet/udp_usrreq.c, openbsd/sys/netinet6/raw_ipv6.c:
correct scoped address handling.
2000-06-13 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.c (in6_purgeaddr): tried to restore an
interface direct route if we have an address that shares the same
prefix with the deleted address.
This would improve behavior in multi-address environments;
if you assigned multiple addresses that shared a same prefix and
then remove one of them, the interface direct route corresponding
to the address would still remain.
2000-06-13 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.c (in6_ifremloop): always called
in6_ifloop_request regardless of the result of
in6_is_ifloop_auto(), in order to make sure to invalidate a stale
route entry for a deleted address.
Note: bsdi doesn't need this fix.
2000-06-13 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_input.c (ip6_input): avoided to use the
cached route for forwarding, if it is down.
This fix would prevent the input routine from accepting a packet
to already removed address.
Tue Jun 13 14:29:12 JST 2000
* netbsd/sbin/setkey: move setkey from usr.sbin/setkey to sbin/setkey.
we need it for encrypted NFS (sync better with netbsd-current).
Tue Jun 13 14:07:24 JST 2000
* kame/racoon/safefile.c: be more picky about secret file permission.
now pre-shared key file (psk.txt) is required to be owned by the
uid running racoon (= root), and must not be accessible by others
(like 0400).
2000-06-13 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6_rtr.c (in6_ifadd): made sure to gain a
reference counter to in6_ifaddr for the autoconfigured address
in bsdi and freebsd2 cases.
This fix would be important to those OSes, since the older code
woulde cause duplicated free when the lifetime for the address
2000-06-13 JINMEI, Tatuya <>
* kame/kame/dhcp6/common.c (getifaddr): corrected arguments to
In response to a report from Hajimu UMEMOTO<>
Tue Jun 13 03:39:42 JST 2000
* kame/kame/racoon:
send error message against sadb_acquire message to the kernel
when IPsec-SA negotiation fail.
Mon Jun 12 16:52:29 JST 2000
* netbsd/pkgsrc/lang/ruby: use 1.4.4.
Mon Jun 12 14:53:22 JST 2000
* kame/kame/racoon:
Add to handle CR payload on main/aggressive mode.
No Certificate Authority field is included to CR payload at the moment.
Becuase any certificate authority are accepted without any check.
Mon Jun 12 JST 2000
* sys/netkey/key.c: transmit SADB_X_SA2 from kernel to userland
on SADB_ADD and SADB_UPDATE. without it latest racoon does not work.
Mon Jun 12 12:34:02 JST 2000
* libinet6/getnameinfo.c: use EAI_xx for error code. rfc2553bis
suggests it. the commit corrects old behavior on invalid socket,
where getnameinfo returned 0 (success).
Mon Jun 12 08:51:25 JST 2000
* sys/netkey/key.c: correct prefix length match on destination address.
the code used source prefix len on comparision by mistake.
From: Ronald van der Pol <>
2000-06-12 JINMEI, Tatuya <>
* *bsd*/usr.[s]bin/netstat/inet6.c (icmp6_stats): printed new
statistics counters about error messages (see below).
2000-06-12 JINMEI, Tatuya <>
* kame/sys/netinet/icmp6.h: added a new structure icmp6errstat{}
to count more precise statistics of error messages to be generated.
* kame/sys/netinet6/icmp6.c (icmp6_errcount): added as a new
function in order to count the precise statistics.
2000-06-12 JINMEI, Tatuya <>
* *bsd*/usr.[s]bin/netstat/inet6.c (ip6_stats): printed new
statistics counters of forward cache for incoming packets (see
2000-06-12 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_var.h (ip6stat): added members to count
statistics of forward cache for incoming packets.
* kame/sys/netinet6/ip6_input.c (ip6_input): counted the
Sun Jun 11 17:03:07 JST 2000
* {netbsd/pkgsrc/net,freebsd[23]/ports}/{libpcap,tcpdump}: use
2000/6/5 weekly SNAP from
* netbsd/pkgsrc/net/ethereal: upgrade to 0.8.9.
Sat Jun 10 23:11:41 JST 2000
* kame/sys/netkey/key.c:
* kame/kame/libipsec/pfkey.c:
* kame/kame/setkey/parse.y:
Obsoleted mode specification. You don't need to specify a mode of
SA when you get/delete SA by setkey. Also, it's not need to do
Sat Jun 10 15:42:25 JST 2000
* kame/sys/{netkey/key.c,net/pfkeyv2.h}:
* kame/kame/{stekey,racoon,libipsec}:
Moved mode/reqid to new extension from sadb_msg header.
New extension is defined sadb_x_sa2 structure.
Fri Jun 9 10:10:08 JST 2000
* *bsd*/sys/netinet6/in6_pcb.c, kame/sys/netinet6/in6_src.c:
add in6_embedscope() and in6_recoverscope(), to avoid code duplicate
for KAME scopeid hack.
* {bsdi4,openbsd}/sys/netinet6/in6_pcb.c: add scope consideration
to in6_pcbbind.
Thu Jun 8 22:58:24 JST 2000
* kame/ping6/ping6.c: make sure to clear ni_flags on ping6 -w
or ping6 -W. From: yoshfuji
Thu Jun 8 21:40:55 JST 2000
* *bsd*/sys/netinet6/{udp6_output,raw_ip6,in6_pcb}.c:
do not overwrite sockaddr on PRU_SEND nor PRU_CONNECT.
(the change is very important)
Wed Jun 7 22:21:43 JST 2000
* sys/netinet6/in6_prefix.c: check interface matches when adding
new interface addresses, based on new link-local address and
prefix information. old code added new interface address without
checking interface match and ended up many bogus new interface
Tue Jun 6 22:41:24 JST 2000
* bsdi4/sys/netinet/tcp_input.c: fix INP_IPV6_MAPPED configuration
onto inpcb. the mistake caused ND to behave badly (ND tries to
update reachability info for IPv4 mapped address - one more reason
to hate IPv4 mapped address).
Tue Jun 6 22:16:18 JST 2000
* netbsd/pkgsrc/www/wwwoffle: upgrade to 2.5e.
Tue Jun 6 21:53:54 JST 2000
* bsdi4/sbin/sysctl: correctly handle IPv6 ipsec/pim entries.
2000-06-06 JINMEI, Tatuya <>
* freebsd3/libexec/tftpd/tftpd.c: do not connect tftp session
sockets. Instead, check consistency of the client's local port.
In response to PR: sys/254
2000-06-06 JINMEI, Tatuya <>
* freebsd3/usr.bin/tftp/tftp.c:
- do not blindly copy server's port, but make sure to check its
consistency during a session.
- copy the whole received sockaddr (i.e. not only port) for the
next sendto(), which would be safer if the server has bound its
local address.
2000-06-06 JINMEI, Tatuya <>
* freebsd3/ports/bind9, netbsd/pkgsrc/net/bind9:
upgraded to bind-9.0.0b3. KAME users are recommended to use this port,
because the original version has an API compatibility issue in a
multi-address environment.
2000-06-06 JINMEI, Tatuya <>
* bsdi3/sbin/sysctl/sysctl.c:
* {netbsd, openbsd}/usr.sbin/sysctl/sysctl.c:
synch with the change below. (Fortunately, there are no binary
compatibility issues).
2000-06-06 JINMEI, Tatuya <>
* kame/sys/netinet6/pim6_var.h: renamed macro definitions about
pim6 sysctls to avoid possible conflict.
Suggested by: Hitoshi Asaeda <>
Mon Jun 5 21:56:35 JST 2000
* traceroute6/traceroute6.c: add -f (skip first hops). the option
is common to freebsd4/bsdi4/openbsd/netbsd traceroute(8).
2000-06-05 JINMEI, Tatuya <>
* kame/kame/traceroute6/traceroute6.c (main): set a minimum set of
socket options as soon as opening a raw socket to avoid
unintentionally receiving packets without ancillary data.
Note: this could actually happen when we used an unreachable IPv6
DNS server.
2000-06-05 JINMEI, Tatuya <>
* kame/sys/netinet6/in6_src.c (in6_pcbsetport): supported FreeBSD
3 and later as a separate function, the code which was a part of
2000-06-05 JINMEI, Tatuya <>
* kame/sys/netinet6/udp6_output.c (udp6_output): supported FreeBSD
3 and later.
* freebsd[34]/sys/netinet6/udp6_usrreq.c: removed original
* freebsd[34]/sys/conf/files: added netinet6/udp6_output.c
2000-06-05 JINMEI, Tatuya <>
* freebsd[34]/sys/netinet6/udp6_usrreq.c (udp6_send): added some
validation checks for the passed sockaddr. The previous code seemed
a bit naive.
Sun Jun 4 22:28:30 JST 2000
* libinet6/resolv/res_send.c: remove MULTICASTQUERY6 case.
it is unnecessary since we have mdnsd, and it was mistake in BIND9
that BIND9 replied to multicast queries.
Sun Jun 4 21:57:47 JST 2000
* sys/netinet6/nd6.c: revise upper-layer reachability confirmation
hint processing. count hints from upper-layer, and if the count
reaches maximum, try to probe reachability by real ND.
sysctl net.inet6.icmp6.nd6_maxnudhint defines maximum # of subsequent
upper-layer hints to be accepted.
we have removed check against process privilege for IPV6_REACHCONF.
we currently have two sources for hints: (1) IPV6_REACHCONF
(2292bis setsockopt), and (2) hints from tcp_input.
it is still questionable if they are really trustworthy.
for example, rogue userland program can use IPV6_REACHCONF to confuse
ND process (which is system-wide cache). also, tcp_input can be
hosed by hijack attempts.
Sun Jun 4 12:29:41 JST 2000
* sys/netinet6/mip6*: sync with latest MIPv6 patch from ericsson guys.
- New addition to eager movement detection: now a second level
provided to enhance handoff time a lot. As soon as a
new prefix is heard the Mobile Node changes default
router. Less stable, router reachability is not assured,
but this can make handoff go in less than 0.5 sec. Also:
nasty trick used is to run nd6_timer() five times per
second. Use "mip6config -e 2" to test.
- More consistent function of mip6config and mip6stat.
- Fixed some bugs in mip6stat.
- Code clean up.
Sun Jun 4 01:12:52 2000 SUMIKAWA Munechika <>
* {freebsd3/ports, netbsd/pkgsrc/net}/p5-Socket6: upgrade to 0.07.
Sat Jun 3 08:08:43 2000 SUMIKAWA Munechika <>
* freebsd3/ports/{libpcap,tcpdump}: use 2000/5/29 weekly snap.
* freebsd3/ports/mod_perl: mod_perl-1.24.tar.gz
2000-06-02 JINMEI, Tatuya <>
* bsdi4/sys/netinet/tcp_subr.c (tcp6_ctlinput): separated from
tcp_ctlinput. We believe a separated function would make sense
in this case.
Additionaly, the new function now calls syn_cache_unreach().
IMO, there's been no special reason not to call it.
* openbsd/sys/netinet/tcp_subr.c (tcp6_ctlinput): applied the same
fix as bsdi4.
Also, the ctlinput function now calls in6_pcbnotify(), which was
unintentionally (IMO) prohibited.
2000-06-02 JINMEI, Tatuya <>
* {bsdi4, freebsd[34], openbsd}/sys/netinet6/in6_pcb.c
* kame/sys/netinet6/in6_pcb.c (in6_pcbnotify):
corrected the algorithm to detect whether in6_rtchage() should
be called.
Fri Jun 2 21:08:22 JST 2000
* freebsd2/ports/{libpcap,tcpdump}: use 2000/5/29 weekly snap.
Thu Jun 1 04:47:42 JST 2000
* kame/kame/racoon:
- You must explicitly specify the file name of certificate and
private key if you use a certificate. See README.certificate.
- fixed to make a index of SPD in responder side when ID type in ID
paload is subnet, or when there is no ID payload.
- supported to negotiate the IPsec-SA of ESP without authentication.
- Added the direction as an item to compare policy.
- fixed some crash problems.
Thu Jun 1 02:43:31 JST 2000
* kame/mdnsd: multicast DNS resolver, as specified in
draft-aboba-dnsext-mdns-00.txt. Note that spec conformance is
still very low.
* netbsd/pkgsrc/net/{libpcap,tcpdump}: use 5/29 weekly snap.
2000-05-31 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_mloopback): make a deep copy
of the IPv6 header only in an M_EXT case. Practically, we'll
rarely need such an extra copy.
Tue May 30 23:56:04 JST 2000
* sys/netinet6/ip6_output.c: make scopeid handling in multicast
loopback case (ip6_mloopback) consistent with other cases.
Tue May 30 20:54:23 JST 2000
* libinet6/resolv/res_send.c: with #define MULTICASTQUERY6, allow
multicast address in /etc/resolv.conf "nameserver" line. By
configuring resolv.conf(5) like below:
nameserver ff02::1%3
DNS queries will reach all nodes on the link with link ID "3".
With the change, the resolver code changes some of its behaviors:
- the resolver code does not check src/dst pair match, if we have
multicast address in resolv.conf. It may or may not raise security
- the resolver code does not use connected udp socket for multicast
highly experimental - note that this change DOES NOT attempt to
support multicast DNS queries, as drafted in
2000-05-30 JINMEI, Tatuya <>
* kame/kame/faithd/faithd.c (main): allocated enough memory for
serverpath and serverarg to prevent possible buffer overrun.
Tue May 30 11:18:03 JST 2000
* freebsd[23]/ports/popper, netbsd/pkgsrc/mail/qpopper:
mark the port broken for security issue. users are suggested to
remove the installed binary.[]
Tue May 30 01:29:04 JST 2000
* freebsd4/sys/net/if_loop.c: check BPF data link layer type (DLT_xx)
on call to bpf_mtap() in if_simloop().
XXX if_simloop() call from ip6_output() causes junk to be injected
to bpf buffer. we may need some fix separately from this commit.
Tue May 30 00:11:17 JST 2000
* {bsdi4,netbsd}/usr.bin/ftp: internally convert IPv4 mapped address
into real IPv4 address. IPv4 mapped address adds too much confusion
to FTP protocol handling.
Suggested by: ume
Mon May 29 08:34:58 JST 2000
* kame/sys/netinet/in6_pcb.c (affects netbsd only):
bind(2) on AF_INET6 socket with IPv4 mapped address will raise error,
instead of making port number duplicate on pcb struct.
Mon May 29 02:26:25 JST 2000
* bsdi4/sys/netinet/tcp_input.c: correct critical typo. prior to the
change, if we have inbound IPv4 TCP traffic to AF_INET6 socket.
inp_laddr6 will have ::ff0f:xx.yy.zz.uu, not ::ffff:xx.yy.zz.uu.
* bsdi4/usr.sbin/netstat/inet.c: do not print IPv4 mapped address
as normal IPv4 address, on netstat -an. print it as is (as IPv4
Mon May 29 00:43:15 JST 2000
* sys/netinet6/ip6_output.c (affects netbsd only):
correct setsockopt(IPV6_BINDV6ONLY) behavior.
Mon May 29 00:04:15 JST 2000
* bsdi4/sys/netinet/in_pcb.c: remove NRL bind(2) ordering constraint.
NRL code forbids wildcard IPv4 bind after wildcard IPv6 bind on
the same port number. it was mistake. now bind(2) is free from
ordering constraint.
Sun May 28 15:33:11 JST 2000
* netbsd/pkgsrc/net/{libpcap,tcpdump}: use 2000/5/22 weekly snap.
Sat May 27 23:05:00 JST 2000
* netbsd/pkgsrc/net/ethereal: upgrade to 0.8.8.
2000-05-27 JINMEI, Tatuya <>
* kame/kame/dtcp/dtcpc.rb: added a new option "-r (static|solict)"
to specifiy how the default route should be configured.
2000-05-27 JINMEI, Tatuya <>
* kame/kame/rtadvd: became more conscious of interface status
(up or down);
- when an interface becomes down, stop the corresponding timer.
- when an interface becomes up again, reinitialize the status and
restart the timer.
- simply discard data received on a daed interface.
- dump if interface is up or dump
Sat May 27 19:07:58 JST 2000
* sys/netinet6/icmp6.c: add net.inet6.icmp6.errppslimit sysctl MIB.
it will let you limit ICMPv6 error messages by pps, not interval.
still not sure what is the best way to perform rate limitation...
change default parameter for icmp6 error rate/pps limitation
to 100usec interval and 200pps (an icmp6 error packet needs to
pass both tests to leave the node).
Sat May 27 15:18:58 JST 2000
* kame/ping6/ping6.c: support -N option to probe NI group address at
ease. for example:
% ping6 -N -I ne2 lychee
will probe NI group address for nodename "lychee", instead of the
address resolved for "lychee".
Sat May 27 14:01:29 JST 2000
* sys/netinet6/in6_ifattach.c: on in6_ifattach(), join NI group address
specified in icmp-name-lookup-04 and beyond.
TODO: cope with hostname change events, like sethostname(3).
Fri May 26 00:02:15 JST 2000
* netbsd/usr.bin/ftp/fetch.c: do not transmit scope identifier
on HTTP Host: directive. disallow scoped address if the access is
via proxy.
* openbsd/usr.bin/ftp/fetch.c: do not transmit scope identifier on
HTTP Host: directive. correct Host: directive passing if the
access is via proxy.
TODO: disallow scoped address if the access is via proxy.
Thu May 25 23:16:01 JST 2000
* bsdi4/usr.bin/telnet/commands.c: use getaddrinfo to be scoped-address
Thu May 25 21:56:47 JST 2000
* bsdi4/sys/i386/pci/if_exp.c: do not use interrupt for initializing
multicast filter. this will make exp driver to be friendly with
IPv6 initialization sequence. hints from netbsd fxp driver.
2000-05-25 JINMEI, Tatuya <>
* bsdi4/sys/net/route.c (ifa_ifwithroute): reverted a KAME
specific change introduced on 2000-02-25 (see this log file).
The change was an ad-hoc hack and broke BSDI's original
intention, so we've wanted to clarify the code.
2000-05-25 JINMEI, Tatuya <>
* kame/sys/netinet6/in6_ifattach.c (in6_ifattach,
in6_ifattach_addaddr): use rtrequest1() when assigning IPv6
addresses since this function is more conscious of interfaces than
rtrequest() and is suitable for link-local addresses.
Thu May 25 01:25:39 JST 2000
* kame/sys/{netkey/key.c,netinet6/ipsec.c}:
* kame/kame/setkey:
supported IPsec with link local address.
XXX to be supported by racoon.
Wed May 24 22:42:40 JST 2000
* bsdi4/sys/net/if_ppp.c: IPv6 support. highly experimental.
* bsdi4/usr.bin/ppp: IPv6CP support. highly experimental. solicit
test/experience reports.
2000-05-24 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6s.c: the "-n dnsserv" option can now be
specified multiple times, when a user wants to set more than one
Wed May 24 18:59:55 JST 2000
* kame/kame/racoon:
fix to check vendor id. HASH function for vendor id is always MD5.
Wed May 24 18:09:12 JST 2000
* sys/netinet6/in6.h: check parameter type to IN6_ARE_ADDR_EQUAL().
this should avoid typecast bugs like below.
2000-05-24 JINMEI, Tatuya <>
* {bsdi4,openbsd}/sys/netinet6/in6_pcb.c (in6_pcbnotify):
corrected the 2nd argument IN6_ARE_ADDR_EQUAL(), which was
unintentionally "struct in6_addr **"".
This fix is quite important for stability and users of those OSes
are recommended to apply it.
Wed May 24 14:32:44 JST 2000
* kame/kame/racoon:
supported pre-shared key written by hex string.
Wed May 24 01:29:46 JST 2000
* netbsd/usr.sbin/pppd: pppd with IPv6 support. from netbsd-current.
no on-demand dialing support at this moment.
* netbsd/sys/net/if_ppp.c: async ppp kernel code with IPv6 support.
from netbsd-current. it will change ioctl API slightly,
and stock netbsd142 pppd may choke.
Tue May 23 21:51:00 JST 2000
* kame/kame/racoon:
fix to get local address to be used by sending address. It was
enbugged when some macro was removed.
Tue May 23 07:41:08 JST 2000
* kame/kame/racoon:
don't bark when the life duration of attribute in SA payload is
encoded TV format. It's allowed by RFC.
Tue May 23 07:14:59 JST 2000
* kame/kame/racoon:
some fixes.
- initialize prefixlen at ipsecdoi_id2sockaddr.
- check sainfo duplicated.
- improve sainfo2str() to return "anonymous" if sainfo is anonymous
- fix to parase port number specification.
- change output level during parsing payload.
Tue May 23 01:46:39 2000 Shin'ichi Fujisawa <>
* Remove "options PM" from GENERIC.v6.
Now, the facility of v4NAT was integrated by NATPT (some NAT
function is not implemented yet), please use NATPT when you want
to use v4NAT.
Tue May 23 00:27:52 JST 2000
* sys/netinet6/udp6_var.h: make sysctl symbol name meet with
IPv4 counterpart. freebsd2: net.inet6.udp6.maxdgram,
netbsd: udp6.sendspace, bsdi3: udp6.sendmax.
Mon May 22 21:08:55 JST 2000
* sys/netinet6/icmp6.c: disallow negative numbers for ICMPv6 rate limit
interval. use ratecheck(9) if exists (this will make
netbsd15/openbsd27 merger easier).
Mon May 22 17:54:29 JST 2000
* sys/netinet6/{ah,esp}_output.c: correct rare error case handling
(do not leak mbuf/do not emit incomplete packet).
* sys/netinet6/ah_*.c: cope with memory allocation failure more
2000-05-22 JINMEI, Tatuya <>
* kame/kame/scopedaddrtest/scopedaddrtest.c: a new option "-N" was
added to specify the NI_NUMERICSCOPE flag for getnameinfo().
2000-05-22 JINMEI, Tatuya <>
* kame/kame/libinet6/getnameinfo.c (ip6_sa2str): supported
the NI_NUMERICSCOPE flag if available.
2000-05-22 JINMEI, Tatuya <>
* {freebsd[234],bsdi3,{net,open}bsd}/include/netdb.h
* bsdi4/contrib/bind/src/include/netdb.h (NI_NUMERICSCOPE):
the new flag was added as specified in rfc2553bis-00.
Mon May 22 13:19:29 JST 2000
* sys/netinet6/icmp6.c: to conform better with icmp-name-lookup-05,
make the validation rule more strict for node information query.
now, IPv6 destination address must be anycast/unicast for the node,
or link-local multicast.
there still are spec non-conformance - see comments in icmp6.c.
Mon May 22 12:07:28 JST 2000
* kame/gifconfig/gifconfig.c: correct gif outer (physical) address
printing, on non-IPv6 kernel.
Sat May 20 04:59:26 JST 2000
* sys/netinet6/ip6_{forward,output,input}.c:
correct link-local scoped address handling on loopback interface.
broken by FAKE_LOOPBACK_IF code committed on 17 May 2000.
2000-05-19 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_mroute.c (ip6_mdq): loosened scope check
in ip6_mdq() for packets forwarded to a register I/F. Without
this, such packets could be mistakenly discarded.
In response to: a report from Noriaki Takamiya
Fri May 19 11:41:05 JST 2000
* sys/netinet6/ip6_mroute.[ch]: add backward compatibility code
for 2000/4/12 mrt6msg change (is not used in kame tree).
freebsd40 and openbsd27 are shipped with old declaration, so we
need to provide backward compat to some degree.
Fri May 19 09:26:28 2000 SUMIKAWA Munechika <>
* freebsd4/{usr.bin/fetch, lib/libftpio}: support IPv6.
Fri May 19 00:30:04 JST 2000
* kame/pim6dd/vif.c: allocate uv_querier struct as necessary on
initialization. it will avoid SEGV right after invocation.
2000-05-19 JINMEI, Tatuya <>
* kame/sys/netinet6/scope6.c (scope6_addr2default): was added to
get the default scope zone for a specified address.
2000-05-19 JINMEI, Tatuya <>
* kame/sys/netinet6/udp6_output.c (udp6_output):
* kame/sys/netinet6/raw_ip6.c (rip6_usrreq):
* {bsdi4, openbsd}/sys/netinet6/raw_ipv6.c:
* {freebsd[34]}/sys/netinet6/raw_ip6.c:
* {bsdi4, freebsd[34], openbsd}/sys/netinet6/in6_pcb.c:
filled default scope zone in sin6_scope_id if unspecified.
Currently, these are enabled only with the ENABLE_DEFAULT_SCOPE
kernel compilation option.
2000-05-18 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_forward.c (ip6_forward): zero-cleared
embedded link zone IDs in forwarded packets. We forgot this, since
old kernel did not allow link-scope packets to be forwarded.
2000-05-18 JINMEI, Tatuya <>
* kame/kame/ndp/ndp.c (set): supported <link-local_addr>%<link ID>
for "ndp -s".
Thu May 18 21:15:32 2000 SUMIKAWA Munechika <>
* freebsd3/ports/fetchmail: upgrade to 5.4.0
* freebsd3/ports/heimdal: upgrade to 0.2t.
* freebsd3/ports/w3m: upgrade to 0.1.9.
* freebsd3/ports/netperf: use latest IPv6 patch.
Thu May 18 16:34:18 JST 2000
* netbsd/pkgsrc/mail/fetchmail: upgrade to 5.4.0.
Wed May 17 23:58:14 JST 2000
* {netbsd/pkgsrc, freebsd[23]/ports}/net/{libpcap,tcpdump}: use 5/15
weekly snap.
Wed May 17 21:50:27 JST 2000
* {freebsd[234],netbsd}/sys/netinet/ip_output.c:
call key_spdacquire() if there is no policy entry for IPv4.
key_spacquire() sends a setup policy request to the policy
negotiating daemon or something.
2000-05-17 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_output):
* kame/sys/netinet6/ip6_output.c (ip6_output):
made scoped addresses more friendly with loopback interfaces;
if a packet containing scoped addresses in the source or
destination address fields is going to be sent on a loopback
interface, pass looutput() the interface to which the scoped
address(es) belongs.
For example, you won't see "lo0" when you ping to your own
link-local address.
This is currently experimental and enabled only when the
FAKE_LOOPBACK_IF kernel compile option is specified.
Wed May 17 JST 2000
* sys/netinet6/icmp6.c: correct node information lookup, when
the query is formatted in 05 draft format and IPv6 multicast
address is attached as subject. this will make node to reply to
"ping6 -w ff02::1" like before (actually, validation rule is not
really conformant to 05 draft. see comments in source for details).
Wed May 17 19:09:51 2000 SUMIKAWA Munechika <>
* kame/freebsd3/ports/{squid11, qmail}: upgrade to latest IPv6
* kame/freebsd3/ports/{quake6, quakedata}: use IPv6-migrated
2000-05-17 JINMEI, Tatuya <>
* kame/kame/scope6config/scope6config.c: added a new syntax
"scope6config default" to see the default zones.
2000-05-17 JINMEI, Tatuya <>
* kame/sys/netinet6/{in6.c, nd6_rtr.c, scope6.c}: introduced the
notion of "default scope zones":
- store the default zones in scope6_ids[0]
- reflect change of default interface to default link
- added a new ioctl SIOCGSCOPE6DEF to get the default values
2000-05-17 SUZUKI Shinsuke <>
* bsdi3/sys/net/if_tun.c:
- fixed mbuf allocation bug on tun?
- enabled IPv6 addressing on tun?
* {freebsd[23]/ports,netbsd/pkgsrc/net}/gated-ipv6
updated to May 11's snapshot
Wed May 17 10:09:42 JST 2000
* kame/sys/net/if_gif.c: with "options XBONEHACK", allow multiple
gif tunnels to be configured with the same pair of outer addresses.
From: Lars Eggert <>
* kame/sys/net/if_gif.c: always disallow multiple multi-dest tunnels.
2000-05-16 JINMEI, Tatuya <>
* kame/kame/dhcp6/common.c (getifaddr): corrected the 2nd argument
to in6_matchflags().
* kame/kame/dhcp6/dhcp6s.c (server6_react_solicit,
server6_react_request): corrected scope validation for incoming
Tue May 16 23:05:56 JST 2000
* kame/route6d/route6d.c: memory leak and bzero() correction.
* kame/route6d/route6d.c: correct interface address handling
for global address on p2p interface. there seem to be two major
models in this arena: cisco (use addr/plen) and gated (advert
dest/128 and addr/128). the code supports both behavior,
switchable at compilation time. default behavior is "gated".
we may need to revisit the code later. Comments from suz on p2p
routing model.
Tue May 16 09:44:50 JST 2000
* netbsd/usr.bin/ftp/fetch.c: parse RFC2372 ftp URL correctly.
(the code is different from netbsd-current code - netbsd-current
code works just right, and our version will go away when we switch
to kame/netbsd15)
Mon May 15 22:39:20 JST 2000
* openbsd/usr.bin/ftp/fetch.c: parse RFC2372 ftp URL correctly.
2000-05-15 JINMEI, Tatuya <>
* kame/kame/v6test/conf/ext.conf (jumbolack): added to see if the
target is coformant to the 1st validation in Section 3. of rfc
2000-05-15 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_input.c (ip6_input): if ip6_plen == 0 &&
HbH is included && no jumbo opt, return icmp6 error and discard
packet. This is a new validation check added in RFC 2675.
Mon May 15 18:25:52 JST 2000
* sys/netinet6/icmp6.c: support DNS subject name in NI6 query.
receiving node will supress reply, if DNS subject name does not
match gethostname(3). truncated nodename case is little bit
questionable, see below.
Mon May 15 15:48:44 JST 2000
* sys/netinet6/icmp6.c, kame/ping6/ping6.c:
use DNS name encoding for NI6 replies (DNS name).
the kernel code tries to be compatible with 03 draft (will reply
with pascal string to old "ping6 -w" or current "ping6 -W").
truncated nodename case is little bit questionable - many nodes are
configured with non-FQDN nodename, and we're not sure if we can
treat such name as FQDN for the node.
Mon May 15 02:35:41 2000 SUMIKAWA Munechika <>
* freebsd3/ports/{libpcap,tcpdump}: use 2000/5/8
weekly snap
Sun May 14 JST 2000
* sys/netkey: completely rewrite message management, to avoid possible
buffer overrun with corrupted pfkey data stream. since only root
processes are able to connect to pfkey socket, we don't consider
this as vulnerability (if a bad guy can inject corrupted data stream
to pfkey, he has root privilege and he can do many other things
as well). rewrite still in progress.
if you see any breakage with IPsec related operations, please report.
* sys/netkey: attach better proposal/combination payload to acquire
message from the kernel.
2000-05-14 JINMEI, Tatuya <>
* bsdi4/usr.sbin/netstat: printed per-interface IPv6 statistics.
Please try
% netstat [-p [ip6|icmp6] | -f inet6] -s -I if_name
for printing statistics on a specified interface.
2000-05-14 JINMEI, Tatuya <>
* {bsdi4, openbsd}/sys/netinet6/raw_ipv6.c (rip6_output):
- fixed memory leak for outgoing ancillary data
- use KAME's checksum offset option for outgoing packets
- count statistics for outgoing ICMPv6 packets
- be more friendly with nervous compilers
- code cleanup
Sat May 13 JST 2000
* */usr.bin/telnet/commands.c: improve error message on invalid
port name. gai_strerror() does not give enough detail.
* netbsd/usr.sbin/inetd/inetd.c: improve error message on address
family mismatch.
* sys/netinet6/icmp6.c: improve icmp6 node information support.
we have not been conformant to the latest specification (05).
the current code is a mixture of 05 draft and 03 draft.
need more fixes (see comment in code for detail).
2000-05-13 JINMEI, Tatuya <>
* kame/kame/gifconfig/gifconfig.c (in6_status): use getnameinfo
(instead of inet_ntop) to print IPv6 addresses.
2000-05-13 JINMEI, Tatuya <>
* {bsdi4,openbsd}/sys/netinet6/in6_pcb.c (in6_setsockaddr): move
the embedded link ID of a link-local address to the sin6_scope_id
field in order to conceal the embedding hack from users.
2000-05-13 JINMEI, Tatuya <>
* kame/kame/dhcp6: supported the latest draft (dhcpv6-15).
Note that this version is not compatible with the previous one,
and thus you should update whole your DHCPv6 system including
clients, relays and servers.
Thu May 11 19:28:22 JST 2000
* kame/kame/racoon:
- fix to get lifetime/lifebyte from doi attribute.
- fix to lifetime/lifebyte handling.
lifetime of 0 should not be included a packet. 0 of lifebyte should
not be included in a packet although 0 means not to care of the
lifetime of bytes. If there is such case, it is rejected.
- modify racoon.conf.5.
- update version to 20000501 means proposal parser is fixed.
They are reported from <>.
Wed May 10 14:07:22 JST 2000
* sys/netinet/* (all platforms): correct more boundary checking issues
in ip options processing. This was 4.4BSD issue, and all *BSD
platforms are not very careful in parsing ip options.
Wed May 10 02:07:33 JST 2000
* sbin/ping6: correct outbound NI query packet with -aw.
we should have attached Subject data, which was defined in
04 draft. (we are now in 05)
* sys/netinet6/icmp6.c: improve inbound NI query packet validation.
Tue May 9 22:11:56 JST 2000
* all platforms: correct NetBSD Security Advisory 2000-002, problem (2).
this applies to all BSD-based operating systems. Kernel upgrade
is suggested if you run serious services with KAME kernel.
Mon May 8 23:24:34 JST 2000
* netbsd/pkgsrc/net/{libpcap,tcpdump}: use 2000/5/8
weekly snap.
Mon May 8 23:08:24 JST 2000
* libinet6/name6.c: correct resolver error code handling (h_errno,
like last arg for getipnodebyname).
Mon May 8 17:20:17 JST 2000
* freebsd4: ipsec now compiles (not tested at all).
brave testers wanted.
Sat May 6 01:48:14 2000 SUMIKAWA Munechika <>
* freebsd[23]/ports/{libpcap, tcpdump}: upgrade to 2000/05/01.
* freebsd[23]/ports/tcp_wrapper: upgrade to latest IPv6 patch.
* {freebsd[23]/ports,netbsd/pkgsrc/net}/mtr: upgrade to 0.42.
* freebsd3/ports/vat6: upgrade to latest IPv6 patch.
Sat May 6 00:30:39 JST 2000
* */include/netdb.h, kame/libinet6/getnameinfo.c:
2nd arg of getnameinfo() is socklen_t, not size_t.
this was changed somewhere before rfc2553. we forgot to catch
up with the change.
Sat May 6 00:55:30 2000 SUMIKAWA Munechika <>
* freebsd3/ports/mozilla: upgrade to M15. From: ume
Fri May 5 23:19:10 2000 SUMIKAWA Munechika <>
* freebsd3: INET6 and IPSEC* defines are moved from opt_inet.h to
opt_inet6.h and opt_ipsec.h
Fri May 5 15:59:21 JST 2000
* kame/sys/netinet6/in6_ifattach.c: stabilize in6_ifdetach() again.
previous code touched uninitialized pointers.
* openbsd/sys/net/if.c: cope with pcmcia card removal.
2000-05-04 JINMEI, Tatuya <>
* kame/kame/pim6[ds]d/mld6_proto.c (accept_listener_query):
corrected incoming MLD query message handling.
Thu May 4 13:10:16 JST 2000
* openbsd/lib/libinet6/getaddrinfo.c: translate resolver error codes
in h_errno into getaddrinfo error code. without this, EAI_FAIL
is always returned and it made no sense.
Mon May 1 19:38:32 JST 2000
* netbsd/pkgsrc/net/{libpcap,tcpdump}: use 2000/5/1 weekly snap.
Mon May 1 18:24:04 JST 2000
* sys/netinet6/ip6_output.c: avoid panic on ip6fw kernel configuration
on IPv6 packet outputs (usually, DAD on reboot).
Mon May 1 14:05:02 JST 2000
* kame/libinet6/name6.c: use BIND 4.9.7 DNS answer parser to remove
any of vulnerability old name6.c might have.
Mon May 1 12:22:53 JST 2000
* kame/libinet6: have more sanity check against validity of CNAME.
(there were some advisory about this in the past, IIRC)
2000-04-30 JINMEI, Tatuya <>
* kame/kame/pim6[ds]d: did timeout the MLD querier correctly.
2000-04-30 JINMEI, Tatuya <>
* bsdi4/sys/netinet6/raw_ipv6.c: call ip6_mrouter_done() when the
ip6_mrouter socket is closed in order to avoid remaining a garbage
socket in the kernel.
2000-04-29 JINMEI, Tatuya <>
* freebsd3/ports/bind9/Makefile: upgraded to bind-9.0.0b2.
2000-04-29 JINMEI, Tatuya <>
* kame/kame/pim6sd/cfparse.y:
* kame/kame/pim6sd/pim6_proto.c (receive_pim6_bootstrap):
ignored node-local and link-local group prefixes in the
configuration file and received bootstrap messages.
2000-04-29 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_output): do not try NUD unless the
gateway is a real neighbor. We can now enable NUD on a p2p link
even when
- assigning a pair of source and destinaion addresses on the p2p
link (with the /128 mask), or
- installing a direct route to the p2p link using an AF_LINK
gateway or a node's own address on another interface.
Sat Apr 29 02:11:18 JST 2000
* freebsd4: fix various kame/freebsd4 sync issue, including KAME
PR 244. there can be more issues left - pls report if you find one.
* */libinet6/getaddrinfo.c, kame/libinet6/name6.c: correct
FILTER_V4MAPPED code. avoid unaligned memory accesses.
2000-04-28 JINMEI, Tatuya <>
* kame/kame/pim6sd: filled the suggested hash mask length in
bootstrap messages. Also, that field is now configurable.
In response to a report from Olivier TOGNI
Fri Apr 28 19:30:18 JST 2000
* ports/pkgsrc for zebra (openbsd, freebsd[23], netbsd):
upgrade to 0.86.
2000-04-28 JINMEI, Tatuya <>
* kame/kame/traceroute6/traceroute6.c (main): always printed the
source address of proving packets.
Fri Apr 28 13:34:27 2000 SUMIKAWA Munechika <>
* freebsd[23]/ports/rsync, netbsd/pkgsrc/net/rsync: upgrade to 2.4.3
Fri Apr 28 00:41:36 JST 2000
* sys/netinet6/in6_ifattach.c: correct in6_ifdetach().
free oia, not ia. there are more problems, Lennart says.
From: Lennart Augustsson <>
2000-04-27 JINMEI, Tatuya <>
* bsdi4/sys/netinet/udp_usrreq.c: removed udp6_output() to use a
shared version defined in netinet6/udp6_output().
2000-04-27 JINMEI, Tatuya <>
* {bsdi4, openbsd}/sys/netinet/in_pcb.c (in_pcbnotify):
* openbsd/sys/netinet/in_pcb.c (in_pcbnotifyall):
prevented an infinite loop, which could happen if an IPv6 PCB
entry exists and notification to the PCB list occurs.
OpenBSD users who use the 20000424 snap release are strongly
recommended to apply this fix.
Thu Apr 27 12:32:34 JST 2000
* freebsd4/lib/libinet6/getaddrinfo.c: getaddrinfo.c obeys
configured resolve order. From: ume
Thu Apr 27 10:35:47 JST 2000
* libinet6/resolv/res_send.c: disallow $HOSTALIASES for setuid binary.
Thu Apr 27 05:06:35 JST 2000
* kame/kame/racoon:
vendorid directive is obsoleted. vendorid should not be configurable.
Wed Apr 26 21:23:49 JST 2000
* {net,open}bsd/lib/libinet6/getaddrinfo.c:
disable AI_ADDRCONFIG support since it does not handle PF_UNSPEC
case. From: ume
Wed Apr 26 19:20:54 JST 2000
* libinet6/resolv/res_{comp,send}.c: incorporate strict boundary
checking from BIND 4.9.7. simplify #ifdefs.
* libinet6/resolv/res_send.c: force fragment to minimum IPv6 MTU when
we send a UDP query to IPv6 DNS server. this avoids path MTU
discovery from affecting name resolution.
Wed Apr 26 17:44:18 JST 2000
* libinet6/resolv/res_query.c: clarify reasons for setbuf().
* freebsd4: sync with latest gif and stf code (compiled, need more
Wed Apr 26 12:00:59 JST 2000
* libinet6/resolv: allow users to switch use of EDNS0 query packet,
by "options edns0" in /etc/resolv.conf. BIND9 DNS server must be
listed on "nameserver" line if you use EDNS0 query packet, since
BIND4/8 will choke with EDNS0 packets. highly experimental.
(not available in freebsd4 and bsdi4 due to resolver differences)
Tue Apr 25 16:46:56 2000 Shin'ichi Fujisawa <>
* v4 traceroute (in NAT-PT w/ v4NAT) works.
Tue Apr 25 16:21:24 JST 2000
* netbsd/sys/netinet/ip_ipip.c: re-enable support for ipip interface.
sync with netbsd-current.
Tue Apr 25 11:39:22 JST 2000
* netbsd/usr.sbin/lpr: allow lp=port@host. NetBSD PR 9975.
* netbsd/usr.sbin/lpr: don't SEGV on rp=numerichost.
Tue Apr 25 04:09:09 JST 2000
* kame/kame/racoon:
* kame/kame/setkey:
In the case of ICMP{,V6}, check address family and upper layer protocol.
Tue Apr 25 03:54:32 JST 2000
* kame/kame/racoon:
enable VPN negotiation again.
XXX AS before, racoon can NOT negotiate the socket based SA.
2000-04-24 JINMEI, Tatuya <>
* bsdi4/sys/net/if.c (ifa_ifwithnet): reverted a change on
2000-02-25, since the logic is not related with our latest code.
Mon Apr 24 22:13:09 JST 2000
* libpcap/tcpdump pkgsrc (netbsd/freebsd2): upgrade to 4/24 snapshot.
Mon Apr 24 21:39:23 JST 2000
* kame/kame/racoon:
- enable to define some algorithm each algorithm class.
- use internal SPD, don't SPDGET in order to get SP entry.
- fix to negotiate phase 2 with multiple transform in a single protocol.
Mon Apr 24 17:22:51 JST 2000
* kame/kame/racoon:
improve phase 2 configuration and parse phase 2 SA payload.
Racoon takes security protocols to be negotiated at phase 2 from
kernel's SPD.
User can define algorithms, lifetime and pfs by "sainfo" directive.
XXX At the moment, racoon can NOT negotiate the SA for VPN. Because
I don't clarify to define the relation between the values of ID
payloads and a pair of src/dst which cause the SA negotiation to
begin. Also racoon can NOT negotiate the socket based SA. Because
there is not any SP entry for socket in the kernel's SPD.
Mon Apr 24 13:08:42 JST 2000
* netbsd/pkgsrc/mail/fetchmail: upgrade to 5.3.8.
Sun Apr 23 14:21:32 JST 2000
* freebsd2/usr.bin/ftp: use RFC2732 URL format for Host: HTTP directive.
* openbsd/usr.bin/ftp: support RFC2732 URL format (bracket around
IPv6 numeric address).
2000-04-23 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_mroute.c (ip6_mdq): do not check scope
boundary for packets through a PIM register tunnel. Without this
change, an RP will not be able to forward packets received on a
register tunnel. However, this blindly relies on the routing
daemon's behavior. We'll eventually need more reliable fix.
2000-04-21 JINMEI, Tatuya <>
* freebsd3/usr.bin/netstat/route.c (routename6): fill in the
sin6_socpe_id field of a local variable to support
<scope_addr>%<scope_id> extension.
Fri Apr 21 20:13:09 JST 2000
* openbsd/sys/netinet{,6}/in{,6}_pcb.c: on in{,6}_pcbnotify, make sure
to touch pcbs that meets address family of incoming packet.
2000-04-21 JINMEI, Tatuya <>
* bsdi4/sys/netinet/ip_output.c (in_mrejoin): skipped PF_INET6
entries. This is a fix of a serious bug and should be applied if
you use IPv6 multicasting.
Thu Apr 20 21:32:02 JST 2000
* sbin/route/route.c (freebsd[23], netbsd, bsdi[34]):
use scoped address notation when printing IPv6 link-local multicast
2000-04-20 JINMEI, Tatuya <>
* kame/kame/bgpd/ripng.c (rip_sockinit): do not bind ff02::9 on
the RIPng socket. Without this fix, bgpd can't accept unicasted
RIPng messages.
In response to a report from
Thu Apr 20 18:22:26 JST 2000
* libpcap/tcpdump pkgsrc (netbsd): upgrade to 4/17 snapshot.
* libpcap/tcpdump pkgsrc (openbsd): added port for 4/17 snapshot.
2000-04-19 JINMEI, Tatuya <>
* bsdi4/sys/netinet6/raw_ipv6.c (rip6_ctloutput):
- just (implicitly) call ip6_ctloutput for IPV6_CHECKSUM.
- added multicast routing socket options.
Wed Apr 19 17:18:25 2000 Shin'ichi Fujisawa <>
* v4NAT in NATPT works.
- v4 traceroute does not work yet. I will fix it soon.
- FTP active transfer process does not work.
When you want to use ftp, use passve mode instead of active.
Wed Apr 19 10:52:56 JST 2000
* sys/netinet{,6}/in{,6}_gif.c: do not attempt to transmit packet
if we are apparently making infinite loop.
* sys/netinet6/in6_gif.c: always fragment encapsulated IPv6 packet
to minimum IPv6 MTU. otherwise, inner layer (like IPv4 in
IPv4-in-IPv6) needs to retransmit to kick path MTU discovery in
the outer layer - too painful.
2000-04-18 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_mroute.c (ip6_mdq): added scope boundary
2000-04-18 JINMEI, Tatuya <>
* kame/kame/scope6config/scope6config.c: added to configure and
view the scope zone identifiers of a particular interface.
XXX: should eventually be merged into ifconfig or something.
2000-04-18 JINMEI, Tatuya <>
* kame/sys/netinet6/scope6.c: added to configure/mangement IPv6
scopes. Two new ioctls were also added to set/get scope zone
Tue Apr 18 13:12:40 JST 2000
* kame/sys/netinet6/udp6_output.c: for better code sharing, split
udp6_output() from udp6_usrreq.c.
Tue Apr 18 12:44:24 JST 2000
* kame/kame/racoon:
improve phase1 DH group check.
dh_group must be defined by each proposal in a configuration file.
2000-04-18 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6relay.c: added to support DHCPv6 relay
agent. The implementation currently handles DHCP Solicit and
Advertise messages only.
Tue Apr 18 03:10:01 2000 SUMIKAWA Munechika <>
* freebsd3/ports/mmosaic: multicast Mosaic.
Tue Apr 18 02:35:54 JST 2000
* kame/kame/racoon:
- use getifaddrs to get addresses.
- fix to define logging level by hexa-decimal string.
- tiny fix to print error/warn message during parsing configuration
2000-04-17 JINMEI, Tatuya <>
* kame/sys/netinet6/udp6_usrreq.c (udp6_output):
* bsdi4/sys/netinet/udp_usrreq.c (udp6_output):
considered byte order of an embedded link-identifier.
Mon Apr 17 20:32:27 JST 2000
* add ALTQ support to freebsd4.
Mon Apr 17 21:05:48 JST 2000
* sys/netinet/ip_encap.c: support priority between attached
interfaces. see comments for more detail.
* sys/net/if_gif.c: gif prioritized match support (between LINK0 and
!LINK0 - LINK0 is less preferred) is now recovered.
Mon Apr 17 15:52:19 JST 2000
* netbsd/pkgsrc/net/ethereal: upgrade to 0.8.7.
Mon Apr 17 13:32:48 JST 2000
* kame/ndp/ndp.c: remove leftover from IPv4 arp logic in set() and
delete(). this caused infinite loop in certain oruting setup.
Mon Apr 17 11:37:44 JST 2000
* sys/netinet/in_gif.c: reject packets with IPv4 broadcast address
on outer IPv4 source. this is suggested in ngtrans-mech-05.
2000-04-15 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6[cs].c: tiny clarifications;
- shutdown(0) for write only sockets
- do not set IPV6_MULTICAST_IF for a read-only socket.
Fri Apr 14 18:18:18 JST 2000
* ports/pkgsrc for libpcap and tcpdump (freebsd[23], netbsd):
upgrade to use 2000/4/3 snapshot.
Fri Apr 14 17:39:27 JST 2000
* sys/net/if_gif.c: simplify gif interface attachment to ip_encap
dispatcher. add martian/ingress filters against outer source,
as suggested in ngtrans-mech-05.
Fri Apr 14 13:26:44 JST 2000
* openbsd/sys/netinet/tcp_input.c: use path MTU value on routing
table when computing MSS, only if layer 3 supports path MTU
discovery. (openbsd has no IPv4 path mtu discovery. a node may
propose too large mss if we use rmx_mtu value)
From: Hugh Graham <>
Fri Apr 14 01:00:41 JST 2000
* sys/netinet6/ip6_output.c: require root privilege on IPV6_REACHCONF
ancillary message (experimental, from jinmei).
Thu Apr 13 23:13:52 JST 2000
* sys/netinet6/icmp6.c: correct icmp6 extension header traversal,
to avoid sending icmp6 error against icmp6 error. found by
TAHI test.
2000-04-13 JINMEI, Tatuya <>
* freebsd[34]/sys/netinet/tcp_input.c (tcp_input): do not call
ND6_HINT() if the connection status is LISTEN.
2000-04-13 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_nud_hint): do not change a
neighbor's status unless the old status is probably reachable
(i.e. the link-layer address has already been resolved).
In response to: KAME PR sys/235.
Wed Apr 12 22:11:02 2000 SUMIKAWA Munechika <>
* freebsd[23]/ports/ncftp3, netbsd/pkgsrc/net/ncftp3: upgrade to 3.0.1.
2000-04-12 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6[cs].c: tiny fixes to decrease compilation
- corrected arguments for inet_pton().
- supplemented a header file.
2000-04-12 JINMEI, Tatuya <>
* kame/kame/pim6[ds]d/mld6.c (accept_mld6): reflected the change
of mrt6msg{}; check the overlaid MLD6 type field instead of
msg_controllen to detect if received packets are control messages
or real MLD6 messages.
2000-04-12 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_mroute.h: changed the order of members of
mrt6msg{} to overlay the structure onto the MLD6 header (not IPv6
Suggested by:
Wed Apr 12 13:28:35 JST 2000
* netbsd/pkgsrc/mbone/vat6: pkgsrc for vat6 (IPv6-ready vat).
Wed Apr 12 01:24:40 JST 2000
* sys/netinet6/in6_ifattach.c: cleanup IPv6 interface initialization
in in6_ifattach(). be more persistent about initialization,
to cope with manually configured link-locals.
2000-04-11 JINMEI, Tatuya <>
* kame/kame/gifconfig/gifconfig.c: supported a new description
"gifconfig ifname delete", which removes already specified physical
addresses on the interface.
2000-04-11 JINMEI, Tatuya <>
* kame/sys/net/if_gif.c (gif_ioctl): supported a new ioctl
SIOCDIFPHYADDR in order to allow a user to remove
already specified physical addresses on a gif interface.
Tue Apr 11 14:58:10 JST 2000
* sys/netinet6/in6_ifattach.c: revise interface identifier lookup.
we no longer cache EUI64, due to KAME PR 236.
NOTE: you may see interface identifier changes due to lookup
policy change. be careful when you upgrade the kernel (you may
experience address mismatch with DNS database, for example)
Mon Apr 10 14:38:00 JST 2000
* netbsd/pkgsrc/mail/sendmail.beta, freebsd2/ports/sendmail.beta:
use 8.10.1. (not really beta already...)
Sat Apr 8 09:49:57 JST 2000
* gated-ipv6 ports/pkgsrc (netbsd/freebsd[23]): upgrade to 3/9 snapshot.
Fri Apr 7 23:14:53 JST 2000
* netbsd/pkgsrc/net/ethereal: upgrade to 0.8.6.
Thu Apr 6 14:41:26 JST 2000
* sys/netkey/key.c: reject spdadd if address family does not match
between SA address pair and SPD filter. for example, the following
attempt will return EINVAL:
# setkey -c
spdadd ::1 ::1 any -P out ipsec esp/tunnel/;
Thu Apr 6 14:51:03 JST 2000
* kame/kame/racoon/cfparse.y:
disable algorithm direcitive.
if there is in the config file, parser will bark as warning.
if strength is used the file, parser will stop to parse config file.
Tue Apr 4 17:48:20 JST 2000
* sys/netinet6/ip6_output.c: support IPV6_USE_MIN_MTU in 2292bis
advanced API. We support it in setsockopt (IPv6 UDP only) and
in ancillary data.
Mon Apr 3 12:36:01 JST 2000
* */lib/libpcap, */usr.sbin/tcpdump: remove to avoid duplicated effort
with if you want them, use pkgsrc/ports.
Mon Apr 3 12:15:32 JST 2000
* netbsd/pkgsrc/security/heimdal: use 0.2r.
Mon Apr 3 09:52:23 JST 2000
* libpcap/tcpdump pkgsrc/ports (freebsd[23], netbsd):
upgrade to 2000/3/27 weekly snapshot.
Sun Apr 2 22:05:03 JST 2000
* totd pkgsrc/ports (openbsd, netbsd, freebsd[23]): upgrade to 1.1p4.
Sat Apr 1 22:47:12 JST 2000
* sys/sys/socket.h (all platforms): avoid namespace pollution
Thu Mar 30 2000
* libinet6: by defining USE_EDNS0, resolver will make queries with
EDNS0 UDP buffer size report. (experimental since BIND4/8 chokes
with EDNS0 packet)
Thu Mar 30 14:10:35 2000 SUMIKAWA Munechika <>
* freebsd4: turn off gif and faith temporally.
basic catching up was done. I checked ping[46]/telnet6/ssh[46]
work on KAME-current/FreeBSD4.
(Now I have alpha box only. I do not check it on x86 environment)
gif, faith, altq, IPsec(encap4_input), mip6, natpt
* freebsd2/ports/emacs20: upgrade to 20.6.
Wed Mar 29 16:34:20 2000 SUMIKAWA Munechika <>
* freebsd3/ports/ncftp3: upgrade to 3.0.0.
* freebsd3: remove MAPPED_ADDR_ENABLED option to easily code
sharing with freebsd4. you can turn on/off mapped-addr
supporting by sysctl instead of it.
Wed Mar 29 16:00:24 JST 2000
* netbsd/pkgsrc/net/bind9: use bind 9.0.0beta2.
Tue Mar 28 23:09:59 JST 2000
* sys/netinet6/ip6_input.c: SIIT assumes that routers forward native
IPv6 packet with IPv4 mapped address just like for normal packets.
move IPv4 mapped address src/dst check downwards.
* sys/net/if_stf.c: make IPv4 ingress filtering rule more strict.
drop packet if there's no route found for IPv4 source, or
IPv4 compatible IPv6 source.
Tue Mar 28 21:45:54 2000 SUMIKAWA Munechika <>
* freebsd2/ports/ncftp3: upgrade to 3.0.0.
Tue Mar 28 10:35:50 JST 2000
* netbsd/pkgsrc/net/socks64: upgrade to latest patch.
Tue Mar 28 09:03:03 JST 2000
* netbsd/pkgsrc/devel/unproven-pthreads: upgrade to 0.12.
Tue Mar 28 00:40:38 JST 2000
* netbsd/pkgsrc/security/isakmpd: OpenBSD isakmpd snapshot as of
Sun Mar 26 20:05:55 JST 2000
* netbsd/pkgsrc/mail/fetchmail: upgrade to 5.3.4.
Sat Mar 25 16:55:37 2000 SUMIKAWA Munechika <>
* freebsd4/sys: import FreeBSD 4.0-RELEASE. This is just
virgin import. Migration does not ended yet.
Sat Mar 25 14:52:06 2000 SUMIKAWA Munechika <>
* freebsd[23]/ports/socks64: upgrade to latest patch.
Thu Mar 23 02:36:12 JST 2000
* kame/route6d: don't advertise reject/blackhole static routes.
2000-03-23 JINMEI, Tatuya <>
* bsdi4/sys/netinet/udp_usrreq.c (udp6_output): was copied from
netinet6/udp6_usrreq.c to separate UDP output routine from that of
IPv4, since KAME's UDP6 output had completely different logic from
original udp_output.
2000-03-23 JINMEI, Tatuya <>
* kame/sys/netinet6/in6_src.c (in6_pcbsetport): was moved from
in6_src.c for sharing with bsdi4.
2000-03-23 JINMEI, Tatuya <>
* kame/kame/bgpd: retained rejected or blackhole routes to prevent
accepting or advertising such routes unexpectedly.
2000-03-22 JINMEI, Tatuya <>
* freebsd[23]/sbin/route/route.c (getaddr): use getifaddrs() instead
of SIOCGIFCONF to examine all interface addresses.
2000-03-22 JINMEI, Tatuya <>
* {freebsd[23], bsdi[34]}/sbin/route/route.c: supported extended
format for IPv6 scoped addresses. For example, you can install an
IPv6 link-local gateway like
# route add add -inet6 3ffe:: fe80::1234%ed0 -prefixlen 16
instead of
# route add add -inet6 3ffe:: fe80:1::1234 -prefixlen 16
(where 1 embedded in the link-local address is the inteface index
of ed0).
Wed Mar 22 17:57:21 JST 2000
* netbsd: synchronize userland with 1.4.1 -> 1.4.2 changes.
TODO: boot floppies (netbsd/distrib).
Wed Mar 22 16:29:19 JST 2000
* {ports,pkgsrc}/net/bind8: upgrade to latest patch. this avoids
invalid use of netinet6/in6.h.
2000-03-22 JINMEI, Tatuya <>
* kame/kame/rtadvd/dump.c: supported the SIGUSR1 signal to let
rtadvd dump its internal status into a file upon receipt of the
Wed Mar 22 16:12:02 JST 2000
* kame/sys/netkey/key.c:
- improve key_spdget() to get a policy by policy id.
- add key_spddelete2() to delete policy by policy id.
* kame/sys/net/pfkeyv2.h:
- add a new member, sadb_x_policy_id, into sadb_x_policy structure.
- add a new message type, SADB_X_SPDDELETE2 in order to distinguish
* kame/kame/libipsec:
- simplify pfkey_send_spdget(). no require src/dst addresses.
- add pfkey_send_spddelete2() to delete policy by policy id.
Wed Mar 22 15:52:48 JST 2000
* netbsd/sys: synchronize with NetBSD 1.4.2. userland still needs
some work.
2000-03-21 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_ctloutput): set the
INP_IPV6_MCAST flag after ip6_setmoptions() for an NRL pcb.
Without this, bsdi4/openbsd would panic when an application using
IPv6 multicasting is terminated.
This fix would solve KAME PR sys/224.
Tue Mar 21 21:56:04 JST 2000
* freebsd[23]/sys/kern/uipc_socket.c: correct freebsd bug in sanity
check for socket timeout value. netbsd uipc_socket.c 1.18 -> 1.19.
(FreeBSD PR should be issued separately)
From: Eric Lemiere <>
Tue Mar 21 18:23:14 JST 2000
* sys/netinet6/ip6_input.c: disable too strong check against
packets with IPv4 compatible address. this (wrongly) forbids
auto tunnel relaying case in RFC1933. we may want to re-enable it
whenever mech-xx discussion goes to more secure settings.
Mon Mar 20 21:16:30 JST 2000
* netbsd/pkgsrc/net/{libpcap,tcpdump},
freebsd[23]/ports/{libpcap,tcpdump}: upgrade to weekly snapshot
as of 2000/03/20.
* netbsd/pkgsrc/mail/sendmail.beta, freebsd2/ports/sendmail.beta:
upgrade to 8.10.0 (not really beta).
2000-03-18 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_output): just sent packets on a p2p
interface where NUD is disabled. Without this fix, the following
would not work:
# route add -inet6 default ::1 -ifp gif0
Also note that you'll have to do like
# ndp -i gif0 -nud
to disable NUD.
Sat Mar 18 12:09:15 JST 2000
* sync with latest ericsson mobile-ip6.
major changes include:
- options MIP6_{MN,HA} gone. SIOCSATTACH_MIP6 now selects MN or HA
functionality. Same usage as before.
- Memory leaks, bug and other irritating things found and eliminated.
- Code clean-up.
- Pretty stable thing now (in FreeBSD3), we tortured it a bit at
- Better source address selection.
- Delete neighbor cache working better?
- README and TODO updated.
- Proxy delete implemented.
Sat Mar 18 10:27:03 JST 2000
* netbsd/pkgsrc/net/mtr: install binary as non-suid root.
unbroken pkgsrc.
Sat Mar 18 01:16:26 JST 2000
* ports/packages for mtr and lynx: mark them BROKEN for vulnerability
reported on bugtraq. we recommend you to remove them from your KAME
box by using pkg_delete(8), if you have installed them.
Sat Mar 18 JST 2000
* {netbsd,openbsd}/sys/netinet/in.c, sys/netinet6/in6.c: backout
memory leak fix on interface address addition failure, added
Mar 11 2000. the fix was too strict and made it impossible to
add two interface addresses in same prefix.
KAME PR 218. From: Kazuto Ushioda <>
Thu Mar 16 22:33:52 JST 2000
* {netbsd,openbsd}/lib/libinet6/getaddrinfo.c: in res_queryN,
try to visit all of the requests given. don't stop even if
AAAA query returns SERVFAIL. some of old name servers return
SERVFAIL or NXDOMAIN on AAAA queries, if there's no AAAA records.
NetBSD PR: 9621
2000-03-16 JINMEI, Tatuya <>
* kame/kame/pim6sd/cfparse.y: correclty parsed data_timeout
In response to a report from <>
2000-03-16 JINMEI, Tatuya <>
* kame/kame/ndp/ndp.c: added additional arguments for the -i
option to set some ND-related parameters per i/f basis.
Note: you should update kernel and include files before
recompiling ndp.
2000-03-16 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c: performed NUD on p2p links. A new
member of nd_ifinio{} and a new ioctl (SIOCSIFINFO_FLAGS) were
added to turn on/off the feature.
2000-03-15 JINMEI, Tatuya <>
* kame/kame/ping6/ping6.c: added a new option 'H', which specifies
to try reverse-lookup of IPv6 addresses. ping6 now does not try
reverse-lookup by default. We hope that this change will decrease
a report like "ping6 does not work with a global address".
Tue Mar 14 10:04:42 JST 2000
* libinet6/getnameinfo.c: configure return value correctly, when
name lookup is asked and there's no DNS records.
KAME PR: 222, From: Noriaki Takamiya
Mon Mar 13 21:29:32 JST 2000
* {netbsd,openbsd}/lib/libinet6/getaddrinfo.c: fix error handling in
YP name lookup.
* {netbsd,openbsd}/lib/libinet6/getaddrinfo.c, kame/libinet6/name6.c:
filter out IPv4 mapped address in DNS/YP/hosts database, as they
are not supposed to appear on those places. There are real-world
example with incorrect use exist, and it makes trouble with
IPv6-only node.
Mon Mar 13 19:36:55 JST 2000
* kame/kame/racoon:
Add "strict_address" in listen directive. It requires all addresses
for ISAKMP to be bound.
Mon Mar 13 14:09:54 JST 2000
* netbsd/pkgsrc/net/{tcpdump,libpcap}, freebsd2/ports/{tcpdump,libpcap}:
upgrade to 20000228.
Sun Mar 12 13:58:34 JST 2000
* {openbsd,netbsd}/sys/netinet/in.c: undo interface address addition
attempt, when in_ifinit() fails.
Sun Mar 12 11:06:55 JST 2000
* {netbsd/pkgsrc,openbsd/ports}/net/zebra, freebsd2/ports/zebra:
upgrade to 0.85.
Sun Mar 12 00:25:06 JST 2000
* sys/net/if_stf.c: now compiles on all the platforms we have.
(freebsd3 compilation is not tested by myself). TODO: more tests
* {bsdi3,bsdi4,freebsd2,freebsd3}/sys/net/if_types.h: move private
interface types to 0xf0 and upwards.
Fri Mar 10 23:00:45 JST 2000
* {bsdi3,bsdi4,freebsd2,netbsd,openbsd}/sys/net/rtsock.c: fix problem
in route_output, when rnh_lookup returned RNF_ROOT radix node.
This seems to be BSD problem, not KAME problem.
(freebsd3 have already fixed this in a different manner)
openbsd/netbsd-current are fixed as well.
KAME PR: 217
Fri Mar 10 06:11:03 JST 2000
* sys/netinet/icmp6.h: fix member name of icmp6_filter to be
compatible with RFC2292. (member name is icmp6_filt, not icmp6_filter)
From: Francis Dupont
Fri Mar 10 01:05:05 JST 2000
* sys/netinet6/ah_core.c: add strict mbuf boundary checks for IPv4/v6
AH computation. now IPv4/v6 AH processing are m_pulldown safe.
Thu Mar 9 22:15:33 JST 2000
* kame/sys/netkey,netinet6:
implement pfkey function for policy management.
key_spdupdate() replace a policy entry.
key_spdsetidx() set a policy index.
key_spdget() gets a policy entry.
key_spdacquire() send request for starting IKE nagotiation.
XXX key_spdacquire() is not tested.
Thu Mar 9 22:08:17 JST 2000
* kame/kame/libipsec/pfkey.c:
implement pfkey libraries.
pfkey_send_spdupdate() replace a policy entry.
pfkey_send_spdsetidx() set a policy index.
pfkey_send_spdget() gets a policy entry.
Thu Mar 9 18:06:30 JST 2000
* sys/netinet6/ah_output.c: avoid mbuf memory leak, or unauthenticated
packet transmission, on IPv6 AH checksum failure.
Thu Mar 9 17:48:58 JST 2000
* kame/freebsd3/sys/netinet/ip_output.c
* kame/kame/sys/netinet6/ip6_output.c:
- Fix ipsec_get_policy().
- Avoid double free of mbuf. it has been free when error occured
at soopt_mcopyout().
They are only in FreeBSD3. Not tested yet in Others paltforms.
Thu Mar 9 17:40:46 JST 2000
* kame/kame/libipsec/test-policy.c:
Improved test program for ipsec_{set|get}_policy.
Thu Mar 9 15:06:53 2000 Shin'ichi Fujisawa <>
* Add v4 NAT into NATPT.
* Now, Options NATPT_NAT is necessary to use this function,
but this option becomes needless soon.
2000-03-09 JINMEI, Tatuya <>
* freebsd3/sys/netinet/tcp_input.c (tcp_input): did not try to
access to ip6_inputopts unless it's really an IPv6 packet.
In response to: a report from Noriaki Takamiya
Thu Mar 9 09:41:53 JST 2000
* sys/netinet6/ah_core.c: except the first AH, treat AH checksum as
immutable if there are multiple AH present on a packet.
note that AH spec does not talk about this case very much, and use
of multiple AH is very rare.
(no interoperability between KAME code prior to this change, and
after this change)
* bgpd/cfparse.y: fix endian issue. found by tahi team.
2000-03-08 JINMEI, Tatuya <>
* kame/kame/scopedaddrtest: was added to check the precise
behavior of KAME's scoped address support in the getXXXinfo
2000-03-08 JINMEI, Tatuya <>
* kame/kame/libinet6/getnameinfo.c (getnameinfo): tried to parse
IPv6 scoped addresses according to the new format
* kame/kame/libinet6/getnameinfo.c (ip6_parsenumeric): was added
for the above purpose.
Tue Mar 7 19:10:21 JST 2000
* kame/kame/racoon:
- Fixed to check OpenSSL version number. reported by
- Since we don't support pre-0.9.4 any more, drop INCLUDE_PATH_OPENSSL,
Tue Mar 7 11:36:03 JST 2000
* kame/sys/netkey/key.c:
enable SADB_X_SPDSETIDX. It can set a policy index into SPD.
2000-03-07 JINMEI, Tatuya <>
* kame/kame/pim6dd/pim6.c (init_pim6): corrected the initilization
process so that pim6dd does not access to NULL pointers.
From: Olivier TOGNI <>
Tue Mar 7 07:16:29 JST 2000
* kame/kame/setkey:
Fix replay window size as default to 0 when manual setting.
It means responder doesn't check sequence number.
Mon Mar 6 16:54:29 JST 2000
* kame/libinet6/getnameinfo.c: comment out NI_NOFQDN support,
since the behavior is specwise incorrect.
Mon Mar 6 13:38:59 JST 2000
* sys/netinet/ip_encap.c: add new api, encap_attach_func(), which
allows you to attach decapsulator with packet matcher function.
* sys/net/if_stf.c: 6to4 tunnel support. (compiles on netbsd only at
this moment)
Mon Mar 6 11:19:24 JST 2000
* netbsd/libexec/{rlogind,rshd}/{rlogind,rshd}.c:
reject connection from IPv4 mapped address (IPv4 connection via
AF_INET6 socket).
Sun Mar 5 11:41:24 JST 2000
* sys/netkey/key.c: correct SA lookup, so that we can always get/use
proper SA even if we use sendto(2).
From: Greg Troxel <>
Sat Mar 4 02:01:16 JST 2000
* netbsd/pkgsrc/net/ethereal: upgrade to 0.8.4.
Fri Mar 3 22:11:59 JST 2000
* {netbsd,openbsd}/sys/netinet/ip_input.c: remove merge mistake,
which affected TTL value for ressembled packet. NetBSD PR 9412.
Fri Mar 3 00:36:58 JST 2000
* netbsd/pkgsrc/www/w3m: upgrade to 0.1.6.
Thu Mar 2 21:30:50 2000 SUMIKAWA Munechika <>
* freebsd3/usr.bin/{fstat, sockstat}: support AF_INET6.
* freebsd3/ports/{v6eval, ct}: upgrade to 0.6 but this port is now
2000-03-02 JINMEI, Tatuya <>
* kame/kame/bgpd/dump.c: used %qu instead of %llu for printing
a u_quad_t value.
XXX: although %qu is not standardized, FreeBSD 3.1 and olders do
not handle %llu correctly.
Thu Mar 2 16:44:18 JST 2000
* sys/netinet6/in6.c: for IPv6 address, don't configure broadcast
address (ifa_dstaddr) in ifaddr structure. this is to avoid
returning broadcast addresses via routing socket.
From: Hideaki YOSHIFUJI <>
2000-03-02 JINMEI, Tatuya <>
* kame/kame/ifmcstat/ifmcstat.c (ifname): corrected invalid access
to kernel internal pointers in !(nbsd||obsd) cases.
Thu Mar 2 02:01:48 JST 2000
* sys/netinet6/mip6*.c: bring in update from ericsson for mobile-ip6.
Wed Mar 1 18:46:24 JST 2000
* kame/kame/setkey:
Fix to delete a SPD entry. "spddelete" can work.
Wed Mar 1 18:36:27 2000 SUMIKAWA Munechika <>
* freebsd3/ports/emacs: upgrade to 20.6.
Wed Mar 1 12:38:07 JST 2000
* libpcap/gencode.c: fix "ip host foo" or "ip6 host foo" where foo has
both A and AAAA. From: Bill Fenner, FreeBSD PR: 17083
Wed Mar 1 03:27:06 JST 2000
* libinet6/rthdr.c: fix alignment constraint in rthdr_*.
From: for freebsd-current
* */sys/kern/uipc_socket2.c: fix ancillary data padding at the end,
in sbcreatecontrol().
Wed Mar 1 01:30:43 JST 2000
* netbsd/sys/netinet/udp_usrreq.c: always use in4_cksum() for
inbound udp4 checksum.
* bsdi4/sys/netinet/udp_usrreq.c: always use in4_cksum() for inbound
udp4 checksum. fix ipsec checks on multicast. add jumbogram
consideration a bit. undo temporary pcbconnect correctly on
udp6 scope check failure.
Tue Feb 29 13:30:10 JST 2000
* */sys/netinet{,6}/in{,6}_proto.c: call encap_init(). From: jinmei
* */sys/netinet{,6}/tcp{,6}_subr.c: make sure to send valid TCP window
size option, even when we have socket buffer size >= 65536.
FreeBSD PR 16914.
Mon Feb 28 23:26:03 JST 2000
* kame/sys/netinet6/icmp6.c: fix ICMPv6 redirect input. the bug
can result in invalid ND entry.
Mon Feb 28 20:14:15 JST 2000
* kame/sys/netinet*/*.h: wrap xxCTL_VARS by #ifdef __bsdi__,
since it is only for BSDI.
Mon Feb 28 19:01:44 JST 2000
* netbsd/usr.sbin/lpr/lpd: fix breakage due to unsynchronized change
between rcmd.c and lpd.c (lpd should be reorganized to not use
2000-02-28 JINMEI, Tatuya <>
* kame/kame/ping6/ping6.c (main): added a new modifier "c" for the
-a option, which means target's IPv4 compatible and IPv4 mapped
addresses are required.
2000-02-28 JINMEI, Tatuya <>
* kame/sys/netinet/icmp6.h (NI_NODEADDR_FLAG_xxx): rearranged
according to ipngwg-icmp-name-lookups-05.
Note: this version is not compatible with previous ones.
2000-02-28 JINMEI, Tatuya <>
* kame/kame/{advapitest, bgpd, pim6dd, pim6sd, rtsold, rrenumd,
traceroute6, rtadvd}: treated CMSG_SPACE() as a function, not a
constant macro, and avoided to put it in a definition of a static
array as a result.
Sun Feb 27 18:27:51 JST 2000
* libinet6/resolv/res_init.c: allow extended scoped IPv6 address format
into "nameserver" line in /etc/resolv.conf.
Sun Feb 27 17:49:32 JST 2000
* {netbsd,openbsd}/lib/libinet6/getaddrinfo.c: add OS-specific
getaddrinfo, which keeps search order declared in /etc/nsswitch.conf
or /etc/resolv.conf.
Sun Feb 27 16:43:04 JST 2000
* */sys/sys/socket.h: use ALIGNBYTES, instead of sizeof(long) - 1,
for CMSG alignment. without it CMSG alignment is not spec
conformant if
ALIGNBYTES > sizeof(long) - 1.
this add dependency from sys/socket.h to machine/param.h - which may
not be right.
NOTE: recompile all IPv6 userland, if you are on architecture
with ALIGNBYTES != sizeof(long) - 1.
TODO: make it future adaptable by using sysctl hw.alignbytes.
(maybe for *bsd-current?)
Sat Feb 26 20:46:03 JST 2000
* sys/netinet6/esp_input.c: with IPv6 ESP, correctly handle
off != sizeof(ip6) case.
* sys/netinet6/ah_input.c: with IPv4 AH (m_pulldown case),
strip off AH from the packet. this is to make some of IPv4
transport layer code work correctly (specifically, ICMPv4 will
transmit wrong packet if we don't strip AH here)
2000-02-26 JINMEI, Tatuya <>
* bsdi4/sys/netinet/tcp_input.c (tcp_input): made the variable
"ip" consistent with "ip6" in order to avoid possible pointer
2000-02-26 JINMEI, Tatuya <>
* kame/sys/netinet/in_gif.c (in_gif_ioctl):
* kame/sys/netinet6/in6_gif.c (in6_gif_ioctl):
- prevented in[6]_gif_ioctl from making dangling pointers after
freeing old ones. Without this fix, kernel would panic when we
tried to execute gifconfig multiple times on a single gif.
- call encap_detach before encap_attach().
If detach was called after attach, we'd not be able to delete
an existing pair of phyaddrs.
2000-02-26 JINMEI, Tatuya <>
* bsdi[34]/sys/netinet/in.c (in_control): called gif_ioctl() for
gif related ioctls. We need this for initializing encapsulation
2000-02-26 JINMEI, Tatuya <>
* bsdi4/sbin/ifconfig/ifconfig.c: filled a valid pointer in
af_stab[MASK] of the afswitch structure for IPv6.
Disabled inet6_auto(), since KAME assigns a link-local address in
the kernel.
2000-02-26 JINMEI, Tatuya <>
* kame/kame/v6test/getconfig.c (make_ip6): made ipv6 version and
traffic class fields configurable.
* kame/kame/v6test/conf/ext.conf: added new tests "badver" and
"tclass" to aset non-default values to version and traffic class
fields, respectively.
Sat Feb 26 03:03:21 JST 2000
* netbsd/pkgsrc/security/heimdal: upgrade to 0.2q.
* openbsd: sys tree was broken for 4 days since "m->m_pkthdr.aux"
change. sorry for this. now we properly initialize aux pointer
and the kernel boots fine.
Fri Feb 25 18:03:58 JST 2000
* kame/traceroute/ifaddrlist.c: ignore, it is meaningless
as source. if you run ISC DHCP client and lease expires, the address
appears as interface address.
2000-02-25 JINMEI, Tatuya <>
* bsdi4/sys/net/if.c (ifa_ifwithnet):
* bsdi4/sys/net/route.c (ifa_ifwithroute):
loosened restriction on p2p interfaces. It was too strict for
KAME's IPv6 support.
2000-02-25 JINMEI, Tatuya <>
* {(net|open)bsd, freebsd3}/sys/netinet/tcp_input.c:
* kame/sys/netinet6/tcp6_input.c:
prevented from storing 2292bis IPv6 options in a listening
socket. Ancillary data on a listening socket was meaningless and
even harmful for applications.
2000-02-25 JINMEI, Tatuya <>
RFC2292bis support:
* bsdi4/sys/netinet/tcp_output.c (tcp_output):
passed IPv6 output options to ip6_ouput().
* bsdi4/sys/netinet/tcp_input.c:
passed specified IPv6 optional data to a user as ancillary data.
* bsdi4/sys/netinet/in_pcb.c (in_pcbdetach):
fixed memory leak of received IPv6 options.
2000-02-25 JINMEI, Tatuya <>
* kame/kame/libinet6/getnameinfo.c (ip6_sa2str, getnameinfo):
changes according to the recent discussion:
- always print scope_id of an IPv6 address with non-zero
sin6_scope_id (by default), even for a global unicast address.
- use numeric numbers as the last resort of a scope_id string.
Fri Feb 25 01:49:00 JST 2000
* */sys/net/rtsock.c: wrap BBN hack (honor ifp/ifa on RTM_ADD),
into #ifdef BBNHACK. it conflicts with proxy NDP and mobile-ip6.
(see CHANGELOG entry on Tue Nov 17 14:57:51)
Fri Feb 25 01:34:11 JST 2000
* sys/netinet6/in6.c: net.inet6.icmp6.nd6_proxyall was obsoleted.
use "ndp -s <v6> <mac> proxy" if you need the same functionality.
Fri Feb 25 01:34:37 2000 SUMIKAWA Munechika <>
* freebsd2/ports/openssh: make buildable.
* freebsd3/ports/heimdal: upgrade to 0.2q.
2000-02-24 JINMEI, Tatuya <>
* bsdi4/usr.sbin/{traceroute6, ping6}/Makefile: enabled IPsec
related definitions.
Thu Feb 24 18:59:14 JST 2000
* sys/netinet/ip_icmp.c (except openbsd):
don't transmit ICMPv4 packet back if the original packet was
encrypted. if we send those, we may mistakingly leak secret
information. From: jinmei
2000-02-24 JINMEI, Tatuya <>
* bsdi4/usr.sbin/vmstat: was just imported. We don't have to
modify the code, but it should be recompiled to support additional
memory types used in the kernel.
2000-02-24 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.c (in6_control): clarified address
management ioctls:
not very suitable in an IPv6 era, when we can assign multiple
addresses on a single interface.
- added strict validations for address families of arguments.
2000-02-24 JINMEI, Tatuya <>
* bsdi4/usr.sbin/netstat/{main.c, inet6.c(added)}: printed KAME's
statistics for IPv6 and ICMPv6 instead of original ones.
Thu Feb 24 JST 2000
* libinet6/getifaddrs.c: bring in freeifaddrs(). this is in sync
with bsdi4 change. From:
* bsdi4: enable kernel IPsec code by default. it looks to be working
fine now, thanks to jinmei's efforts.
2000-02-24 JINMEI, Tatuya <>
* kame/sys/netinet6/ah_input.c (ah4_input): prevented
re-translating the byte order of the ip_id field on bsdi4.
2000-02-24 JINMEI, Tatuya <>
* kame/kame/netstat/ipsec.c:
* bsdi4/usr.sbin/netstat:
modified to print KAME's IPsec statistics.
2000-02-24 JINMEI, Tatuya <>
* kame/kame/traceroute/traceroute.c (send_probe):
* bsdi4/usr.sbin/traceroute/Makefile:
changed a bit to calculate a UDP checksum correctly on bsdi4.
Wed Feb 24 08:49:50 JST 2000
* kame/kame/racoon:
enbale to define DH group by a number of group.
old way to define like "modp768" is kept.
Wed Feb 23 21:26:21 JST 2000
* kame/kame/racoon:
fixed to compute RSA signature. SIGN was computed including OID.
Now, racoon gets a interoperability with
TODO: must be check the signer of certificate.
2000-02-23 JINMEI, Tatuya