Skip to content
Fetching contributors…
Cannot retrieve contributors at this time
903 lines (756 sloc) 39.1 KB
$KAME: CHANGELOG,v 1.2183 2002/06/10 12:00:16 jinmei Exp $
2002-06-10 JINMEI, Tatuya <>
* freebsd4/ports/ppp/: made the interface ID selection more
compliant to RFC 2472;
- first IEEE EUI-48/64 identifiers are checked
- use the full 64 bits when generating a random ID
Fri Jun 7 22:44:34 JST 2002
* sys/netinet/ip_input.c: look at rmx_mtu on IPsec tunnel MTU
computation. From: David Waitzman <>
2002-06-06 JINMEI, Tatuya <>
* kame/kame/rtadvd: when an advertised prefix configured from the
kernel has been added or invalidated, notice the change in a short
2002-06-06 Shin'ichi Fujisawa <>
* kame/sys/netinet6/natpt_*.[ch]:
* kame/kame/natptconfig/*.[chly]:
Picked up a modification minute of April and May, because I
forgot to change this CHANGELOG.
- Revise the following conformity with rfc2765 and rfc2766.
- ICMPv[46] header incorrectness.
- IPv[46] header incorrectness which is inside of the
ICMPv[46] payload.
- Support unknown protocol translation.
- Fragmentation-related incorrectness.
- Changed the expiration time of translation slot entry
(icmp=10, udp=10, tcp=1800).
- "natptconfig" command does not read kernel symbol table
any more. "show rules" and "show xlate" subcommand read
'/dev/kmem' as before.
2002-06-03 JINMEI, Tatuya <>
* freebsd4/ports/ppp: upgraded to the latest snap kit as of today
with KAME-original patches.
Mon Jun 3 09:52:03 JST 2002
* sys/netinet6/nd6.c: in nd6_setmtu0(), do not hardcode if_mtu
values (let sys/net layer handle them). special cases are IFT_ARC
and IFT_FDDI, which needs special handling in L3.
2002-05-31 JINMEI, Tatuya <>
* kame/kame/rtadvd: changed the behavior when an interface-direct
prefix being advertised was removed from the kernel;
rtadvd will continue to advertise the prefix with zero lifetimes
rather than to stop advertising the prefix.
This will help a receiving host renumber by deprecating the
address derived from the old prefix.
Fri May 31 12:56:52 JST 2002
* sys/netinet6/ip6_output.c: implement RFC2460 section 5, last
paragraph (ICMPv6 too big with < 1280).
Thu May 30 14:54:25 JST 2002
* kame/sys/netkey/key.c:
- the memory leak was fixed when an error happened during
the kernel made a sadb_delete message.
- the SA type in the sadb_delete message was fixed when the kernel
deleted the SA and made the message to tell it userland programs.
these are reported by <>
2002-05-30 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_setmtu0): separated from
nd6_setmtu() to handle two different cases; interface (and its MTU)
initialization and changing the MTU of an interface.
This change also fixed a bug that could destroy kernel stack
(when changing the interface MTU by hand) introduced yesterday.
2002-05-30 JINMEI, Tatuya <>
* kame/kame/dhcp6: [dhcp6c] added an ability to specify the SLA
length in the configuration file.
Wed May 29 21:02:59 JST 2002
* kame/kame/pim6sd/config.c,vif.[ch],routesock.c: fixed RPF
algorithm bugs for P2P connected routes and static interface
routes (reported from
Tue May 28 19:26:52 JST 2002
* netbsd/freebsd2: provide arc4random(9) and use it.
Tue May 28 07:38:29 JST 2002
* sys/netinet/tcp_subr.c: PMTUD blackhole detection. sync w/netbsd-
Tue May 28 06:46:23 JST 2002
* sys/netinet6/frag6.c: enforce max # of fragments (not fragment
queues) to defend against DoS. similar logic should be incorporated
into IPv4 reassembly code (openbsd has one already).
Tue May 28 00:08:08 JST 2002
* sys/netinet6/in6_ifattach.c: attach per-interface data structure
to if_afdata[AF_INET6] member via dom_ifattach framework.
Mon May 27 07:22:51 JST 2002
* sys/sys/if.h: add framework to have per-domain/af data onto
struct ifnet. as discussed on bsd-api-discuss.
2002-05-25 JINMEI, Tatuya <>
* kame/sys/netinet6/scope6.c (scope6_init): newly added to
inialize scope zone IDs for as many interfaces as possible.
This is important when an interface is somehow used for IPv6
before activating (i.e. making it up) the interface.
TODO: we'll still need to handle the case where an interface is
dynamically attached.
2002-05-23 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6[cs].conf.5: provided man pages for
configuration files of DHCPv6 client and server.
2002-05-22 JINMEI, Tatuya <>
* kame/kame/dhcp6: supported renewal and rebinding of delegated
Wed May 22 19:23:40 JST 2002
* kame/revlookupd: a daemon to translate DNS reverse lookup (PTR query)
into ICMPv6 node information query. internet draft for this
will be issued shortly.
Tue May 21 21:19:30 PDT 2002
* openbsd: upgrade to 3.1. NOTE: sctp does not work yet.
Mon May 20 14:48:02 JST 2002
* sys/netinet/sctp_usrreq.c (and other sctp related code):
upgraded to latest shipment from randall and folks.
2002-05-17 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6c: when constructing a reply message,
included prefixes given in the corresponding advertisement message.
* kame/kame/dhcp6/dhcp6s.c (server6_react_request): chose
delegated prefixes based on the ones specified in the request
2002-05-17 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6.h: changed option type values that have
not been officially assigned to the values documented in NTTC's
service specification (as of today).
Note: the current version has lost interoperability with former
versions of KAME's dhcp6. Please upgrade all clients and servers
to the latest version.
Note: we'll change the temporary values if different type values
are officially assigned.
2002-05-17 JINMEI, Tatuya <>
* kame/kame/dhcp6: supported the preference option; servers can
specify the preference value for a particular interface. Clients
can parse the preference option and take it into account for
choosing the best server.
2002-05-16 JINMEI, Tatuya <>
* kame/kame/dhcp6: supported the "normal" sequence of
client-server exchange; solicit, advertise(s), request, and then
reply. There are still some restrictions; unicasted requests are
not supported. Address allocation is not supported (nor will be.)
When the client received multiple advertisements, it would just
pick the server that had the highest preference (though preference
option is not supported).
2002-05-09 JINMEI, Tatuya <>
* kame/kame/dhcp6: implemented a limited support of prefix
delegation based on draft-troan-dhcpv6-opt-prefix-delegation-01.txt.
A client sends a solicit with a rapid commit option and an option
request option specifying a prefix delegation option. A server
responds to the solicit with a reply containing a prefix
delegation option. The client accepts the reply and configure an
address on specified interfaces using the delegated prefix.
Thu May 9 18:41:28 JST 2002
* faithd: drop rsh/rlogin protocol support.
Tue May 7 18:37:13 JST 2002
* kame/racoon:
fixed the behavior when
- when the "unique" level policy was used, racoon in the responder side
mis-configured the SAs. <> reported and
donated the patch.
2002-05-01 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.h: declared sockaddr in all supported OSes
(KAME PR 418).
2002-05-01 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6s.c: added an ability to respond to a
solicit message with a rapid commit option.
2002-05-01 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6c.c (client6_send): ensured that the
transaction-ID unchanged in retransmissions of a message according
to Section 15.1 of dhcpv6-24.
2002-04-30 JINMEI, Tatuya <>
* kame/kame/dhcp6: added an ability to send solicit messages with
a rapid commit option. A configuration file for dhcp6c was
introduced to include the option.
2002-04-28 JINMEI, Tatuya <>
* kame/kame/rtadvd/config.c (getconfig): if "addrs" was explicitly
set to 0, enabled prefix autoconfiguration from the kernel.
* kame/kame/rtadvd/rtadvd.c (main): set LOG_PERROR for openlog()
when running in the foreground.
Suggested by: Andrew White <>
Thu Apr 25 19:03:16 JST 2002
* kame/kame/racoon:
fixed the key length of the null encryption for phase 2.
racoon configured incorrect key length of the null encryption
to the kernel.
2002-04-25 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6c.c (client6_recvreply):
tightened DUID validation.
* kame/kame/dhcp6/dhcp6s.c (server6_react_informreq):
processed server and client DUID correctly.
2002-04-25 JINMEI, Tatuya <>
* kame/kame/dhcp6/common.c (get_dhcp6_option): newly added as a
shared routine to parse DHCPv6 options.
2002-04-24 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6c.c: corrected handling DUID for the client
side according to dhcpv6-24.
Wed Apr 24 21:23:43 JST 2002
* faithd: correct connect(2) error handling.
From: Yukiyo Akisada <>
handle ECONNABORTED on accept(2).
2002-04-24 JINMEI, Tatuya <>
* kame/kame/dhcp6/dhcp6.h: removed the "dh6_servaddr" member from
the dhcp6{} structure, which had been removed from the
Note that interoperability with implementations in older KAME
snaps had been lost by this change.
2002-04-24 JINMEI, Tatuya <>
* kame/kame/ping6/ping6.c (main): changed the behaivor of the -m
option as follows:
- when the option is specified once, the default behavior (send
packets at the minimum MTU) will be disabled for unicast
- when the option is more than once, it will be disabled for both
unicast and multicast packets.
2002-04-24 JINMEI, Tatuya <>
* kame/kame/ndp/ndp.c (ifinfo): used the new member of the
nd_ifinfo structure (see the next log entry) to see if the
structure had been initialized.
2002-04-24 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_ifattach): obsoleted the usage of
nd_ifinfo.linkmtu as the flag if the structure had been
initialized. Introduced a new member "initialized" for this
specific purpose instead.
Note: this change had broken binary compatibility to ndp(8) with
the -i option. Be sure to upgrade the application as well.
2002-04-22 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_setmtu): warned if configured (or
detected) interface MTU is smaller than IPv6 minimum link MTU.
2002-04-22 JINMEI, Tatuya <>
* kame/sys/netinet6/in6_ifattach.c (in6_ifattach): do not update
in6_maxmtu in this function, in order to avoid using the MTU for a
loopback MTU mistakenly.
2002-04-22 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_ifattach): do not synchronize an ND
link MTU with the corresponding interface MTU, since this
synchronization had a bad side effect when increasing the
interface MTU. Instead, when detecting the link's "real" MTU
always consider those two values, and take the smaller one.
(many code points, such as ip6_output(), that relied on the
previous behavior were rewritten accordingly.)
2002-04-22 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_output): if there has been no NS
for a neighbor after entering the INCOMPLETE state, send the
first solicitation in nd6_output(), regardless of the timer value.
This change will effectively improve the response time to the
first NS.
2002-04-19 JINMEI, Tatuya <>
* freebsd4/sys/netinet/tcp_subr.c (tcp_respond): if
IPV6_USE_MIN_MTU is specified for the option tell ip6_output to
fragment the packet at the minimum MTU.
2002-04-19 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c: changed the policy about
performing path MTU discovery; from now on multicast packets will
be sent at the IPv6 minimum link MTU by default. The semantics of
IPV6_USE_MIN_MTU was changed accordingly, as will be described in
the forthcoming rfc2292bis-07.
2002-04-19 SUZUKI, Shinsuke <>
* kame/kame/pim6sd/mld6.c
does not log warning againt MLD message from ::.
(sync with the change of kame/sys/netinet6/nd6_nbr.c at Rev 1.76)
2002-04-18 JINMEI, Tatuya <>
* freebsd4/sys/netinet/ip_output.c (ip_output): always freed
the route entry stored in iproute (if it was used) regardless of
the IPSEC kernel compilation option.
This should be a better fix to FreeBSD-SA-02:21.tcpip, and should
apply the kernel without the IPSEC option.
2002-04-13 JINMEI, Tatuya <>
* kame/sys/netinet6/{icmp6.c, nd6.c}:
* netbsd/sys/netinet/if_arp.c (arp_rtdrain):
* netbsd/sys/net/route.c:
ported the cached route management routines from freebsd4.
2002-04-12 JINMEI, Tatuya <>
* kame/sys/netinet6/: use nd_ifinfo.linkmtu instead of
ifp->if_mtu to handle link MTU issues wrt IPv6, whenever possible.
Some OS dependent code for freebsd4 and bsdi4 was also modified in
this manner. Some for netbsd and openbsd was untouched.
2002-04-10 JINMEI, Tatuya <>
* freebsd4/sys/netinet/if_ether.c (arp_rtdrain):
* kame/sys/netinet6/nd6.c (nd6_rtdrain):
allowed the case where these functions received a NULL llinfo in
order to avoid kernel panic in some situations.
2002-04-09 JINMEI, Tatuya <>
* freebsd4/sys/netinet/if_ether.c (arp_rtdrain): newly added as
the hook for arp entries in the route cache queue. Since arp does
not have much state, the detection algorithm cannot be very smart;
we only keep entries that have already been resolved.
(see also: nd6_rtdrain below)
2002-04-09 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (icmp6_input, icmp6_mtudisc_update):
dropped icmp6 too big messages that reported an MTU less than the
minimum MTU (i.e., 1280 bytes). We believe such an error is only
meaningful when we're acting as an SIIT(-like) client, which we do
not support.
2002-04-08 JINMEI, Tatuya <>
* kame/sys/netinet6/in6_proto.c: added a new sysctl knob
"net.inet6.ip6.pmtu_expire" to configure the timer value (in
(only available in freebsd4)
2002-04-08 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (icmp6_mtudisc_timeout): enabled the
function for FreeBSD to purge stale path MTU information.
2002-04-06 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (icmp6_mtudisc_update): revised the
case that the reported path MTU is less than the minimum MTU so
that the minimum MTU would be used instead. The former behavior,
which disabled path MTU discovery, was probably just derived from
the IPv4 code, and was meaningless for IPv6, because IPv6 does not
have the "don't fragment" bit. The previous behavior may even be
exploited by some DoS attacks that intentionally send a too big
message with an MTU less than the minimum.
2002-04-06 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_lookup): specified RTF_CACHE for a
newly created route so that the route entry would be subject to
cached route management.
2002-04-06 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_rtdrain): added as the callback
function for each ND entry as a cached route, and decides
whether the entry should really be removed based on the ND state.
Entries which are reachable, delay or probe will not be removed.
An ND cache to a default route will not be removed either.
[FreeBSD4 only]
2002-04-06 JINMEI, Tatuya <>
* freebsd4/sys/net/route.c (rtredirect): set RTF_CACHE before
calling rtrequest() so that routes created by redirect will be
subject to management of cached routes.
2002-04-06 JINMEI, Tatuya <>
* freebsd4/sys/net/route.c: made cached route management more
- we now have 3 parameters to limit the cache; max, hiwat, and
lowat, all of which are configured automatically based on the
kernel memory limit.
- rt_draincache() is called when the number of entries are being
beyond "hiwat", and the function tries to keep the entries
between "lowat" and "hiwat".
- do not regard static routes as cache.
- allow ifa_rtrequest() functions to add a timer entry so that
they can set a refined cleanup function.
2002-04-01 JINMEI, Tatuya <>
* bsdi4/sys/net/route.c: added cached routes management (see
2002-04-01 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_lookup): checked if rt_llinfo is
non-NULL to avoid returning an entry that has the LLINFO flag but
is not a neighbor cache entry.
* kame/sys/netinet6/nd6.c (nd6_is_addr_neighbor): removed the
check for the pointer, because the check is now done in nd6_lookup().
2002-03-29 JINMEI, Tatuya <>
* {freebsd4, netbsd, openbsd}/sys/net/route.c: added an ability to
manage the number of route entries that can be regarded as cache,
in order to prevent local/remote DoS. This is a preliminary and
experimental support, which will need more work. The same
function for bsdi will be committed next week.
2002-03-29 JINMEI, Tatuya <>
* freebsd4/sys/net/route.c: ported routing entry timer from
NetBSD (which was originally derived from BSD/OS).
2002-03-29 JINMEI, Tatuya <>
* kame/kame/pim6sd/mld6v2.c (make_mld6v2_msg):
* kame/kame/pim6sd/mld6.c (make_mld6_msg):
* kame/kame/pim6dd/mld6.c (make_mld6_msg):
always initialized destination address structures.
Otherwise, the strucuture may contain garbages in sin6_scope_id,
which might be rejected in recent versions of kernel.
2002-03-29 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6_rtr.c (defrouter_addifreq): set
RTF_CLONING when installing a default route to an interface
in order to perform ND correctly. Previously this flag was
implicitly set from the corresponding ifa_flags, which was dropped
by a fix on 2002-02-26. We could revert the former change, but I
belive we should set the flag explicitly.
2002-03-27 Shin'ichi Fujisawa <>
* kame/sys/netinet6/natpt_*.[ch]:
- Fix some fragmentation related bugs.
- Revise the following fragmentation related nonconformance
to RFC2765.
- Set DF flag when IPv6->IPv4 translation and there is no
IPv6 fragment header. According to RFC2765 4.1.
- Add fragment header when IPv4 packet does not set DF flag
(Tested ICMPv4->ICMPv6 translation only). According to
RFC2765 3.1.
- Translate IPv4 packet (ICMP Echo/Echoreply) which DF flag
is not set and larger than 1232 bytes (excluding IPv4
header) into two IPv6 packet.
2002-03-25 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_purge): zero-cleard the
corresponding ND info. Otherwise, parameters for removed
interfaces could be reused for a newly-created interface.
2002-03-18 JINMEI, Tatuya <>
* kame/sys/netinet6/ah_input.c (ah6_ctlinput):
* kame/sys/netinet6/esp_input.c (esp6_ctlinput):
corrected arguments to key_allocsa(). This fix is very important
if you use IPsec, because key_allocsa() in recent snaps has a
strong validation check which will cause kernel panic against
bogus values.
2002-03-18 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_output): do not update the
"dst" address, which is the next-hop address, for multicast
destinations. Otherwise, the layer 2 destination address would be
2002-03-17 JINMEI, Tatuya <>
* freebsd4/sys/netinet6/udp6_usrreq.c (udp6_input): corrected
address matching rules for incoming multicast packets, as
suggested by Konstantin KABASSANOV <>
Without the change, multicast applications that binds the
receiving socket with the multicast address would fail to accept
incoming packets.
Fri Mar 15 19:45:40 JST 2002
* sys/netinet6/frag6.c: implement real lock around IPv6 reassembly
(netbsd/openbsd only)
Fri Mar 15 18:28:20 JST 2002
* netbsd/sys/netinet/tcp_subr.c: have tcp6_drain().
2002-03-10 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_mroute.c:
- made sure to attach packet addresses to every mbuf even after
- added a supplement function m_copy_withpktaddrs() for this
Thanks to: "Konstantin KABASSANOV" <>
for finding the problem and testing patches.
2002-03-03 JINMEI, Tatuya <>
* kame/kame/wru: added the -z option to specify the zone of the
destination, particularly for the default destination, ff02::1.
2002-03-03 JINMEI, Tatuya <>
* bsdi4/sys/netinet/tcp_subr.c (tcp6_ctlinput): loosened the
assertion check for the "inner" source. The check was so strong
that it caused kernel panic when the function called from
ip6_output() via pfctlinput2().
bsdi4 KAME snap users are recommended to apply this fix.
Sun Mar 3 01:51:54 JST 2002 Keiichi SHIMA <>
* kame/sys/netinet6/{in6.c|mip6.c|mip_var.h|nd6_rtr.c}
check the CoA when p2p address (like gif) is set. this enables
you to use MIP6 with molec.
2002-03-02 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_ctloutput, and related
- applied recent clarifications in rfc2292bis:
+ ignored IPV6_DONTFRAG and IPV6_RECVPATHMTU for TCP sockets
+ disallowed to set non-unspecified address by IPV6_PKTINFO for
TCP sockets
- corrected the return value from getsocketopt for IPV6_DONTFRAG and
2002-03-02 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (icmp6_redirect_output): included
a target link layer address option in the redirect message for an
on-link destination as well as for a better router. This behavior
should be correct according to the specification.
Tue Feb 26 16:05:53 2002 SUMIKAWA Munechika <>
* freebsd4/ports/ppp: awkhulhak ppp
2002-02-26 JINMEI, Tatuya <>
* freebsd4/sys/netinet6/in6_pcb.c (in6_mapped_sockaddr,
in6_mapped_peeraddr): do not convert the unspecified address (::)
to the mapped address form. Otherwise, get{peer, sock}name(2) for
an AF_INET6 wildcard address would return ::ffff:
2002-02-26 JINMEI, Tatuya <>
* freebsd4/sys/netinet/tcp_syncache.c (syncache_socket): when
making a PCB entry from a syn cache, do not forget to copy
* freebsd4/sys/netinet/tcp_usrreq.c (tcp6_usr_connect): set
inc_isipv6 when connecting to a non-mapped AF_INET6 address.
All FreeBSD users are recommended to apply this fix. Without
these changes, we would fail to update a stale cached route.
2002-02-25 SUZUKI, Shinsuke <>
* sys/netinet6/in6_src.c (in6_selectsrc):
fixed a bug that sendmsg() on raw socket sometimes fails by ENXIO,
due to the inconsistency between embedded zone-id and the calculated
outgoing interface-id.
Mon Feb 25 10:58:09 JST 2002
* sys/netinet/ip_input.c: enforce ipsec policy checking on forwarding
case (the portion was lost during transition to PR_LASTHDR).
From: Greg Troxel <>
2002-02-24 SUZUKI, Shinsuke <>
* kame/route6d/route6d.c:
not remove global addresses on loopback interface from routing
table by route aging.
2002-02-23 JINMEI, Tatuya <>
* kame/sys/netinet6/nd6.c (nd6_rtrequest): rejected the process of
RESOLVE when the interface does not need neighbor caches.
Otherwise, this function would mistakenly try to make a neighbor
cache for an stf interface.
Based on a report from Ross Finlayson <> at the
freebsd-net ML.
2002-02-23 SUZUKI, Shinsuke <>
* kame/pim6sd/{pim6_proto.c, route.c}:
pim6sd doesn't crash now when receiving (*,*,RP) entry.
Thu Feb 21 23:48:38 JST 2002
* kame/kame/racoon:
to specify the identifier in "sainfo" directive is deprecated.
the identifier should be always made from SPD.
2002-02-20 SUZUKI, Shinsuke <>
* freebsd4/sys/net/if.c ():
fixed a kernel crash that occurs when you enable IPv6 Multicast
on VLAN interface.
2002-02-19 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_setmoptions): fixed kernel
panic in the case of IPV6_LEAVE_GROUP without an interface.
The kernel would choose the group based on the address only with
Note, however, that ambiguity on the scope zone of the address
would not be allowed by default. This particularly means that you
should always specify the interface for interface or link local
Based on: a report from Tomomi Suzuki <>
2002-02-19 JINMEI, Tatuya <>
* kame/sys/netinet6/tcp6_subr.c (tcp6_respond): called
ip6_getpktaddrs() correctly. NULL pointers would be referred to
without this fix.
Based on: a report from Tomomi Suzuki <>
2002-02-19 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_input.c (ip6_init2): called nd6_ifattach()
for the loopback interface, in order to make sure to initialize
the nd_ifinfo structure for the interface.
2002/02/19 16:05:42 JST
* reduce differences from netbsd-current and openbsd-3.0:
- use ALTQ_DECL() for altq only variables.
- make IFQ_ENQUEUE() take 4 args even for the non-altq case
- the return type of altq_etherclassify() is changed from
int to void.
the function prototype is moved to if_altq.h.
* add more altq supported drivers and link types.
most of the remaining drivers are supported by now.
2002-02-18 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.h: stopped defining the route_in6{}
structure for bsdi4 regardless of local configuration options.
We'll never need this because BSD/OS has merged the "new" route{}
structure, which can store all socket addresses.
* kame/sys/netinet6/{mld6.c, route6.h}: always included route.h
based on the change above.
2002-02-18 Shin'ichi Fujisawa <>
* kame/sys/netinet6/natpt_*.[ch]:
* kame/kame/natptconfig/*.[chly58]:
- About natptcofig command
- Add "-q" option to suppress error message.
- Change it to return non ZERO value when detects syntax error.
- NAT-PT rule has rule number.
You can omit rule number when setting NAT-PT. In this case,
behaviour of natptconfig command is same as previous.
Rules can be deleted individually; Rule number can be
- see natptconfig(8) and natpt.conf(5) for more detail.
2002-02-14 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_input.c (ip6_input): cleanup; removed a
duplicated check for mapped source or destination addresses.
2002-02-14 JINMEI, Tatuya <>
* kame/sys/netinet6/mld6.c (mld6_start_listening): added an
assertion in the case of this function was called before
2002-02-14 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_input.c (ip6_init2): stopped calling
in6_ifattach(lo0). We in fact do not need to call in6_ifattach()
at this stage. Additionally, in6_ifattach() has a bad effect of
sending packets even though some parts of the kernel are not ready
for sending.
2002-02-09 JINMEI, Tatuya <>
* kame/sys/{netinet6, netkey}/: changed the definition of
in6_multi{} to contain a full sockaddr_in6 of the multicast
address. Related functions were also modified accordingly.
* kame/kame/ifmcstat/ifmcstat.c (in6_multientry):
* {netbsd, openbsd}/usr.bin/netstat/if.c (intpr):
Modified according to the change above. These tools should
be rebuilt as well as the kernel.
Mon Feb 8 JST 2002
* freebsd4: sync with 4.5-RELEASE
2002-02-04 JINMEI, Tatuya <>
* bsdi4/sys/netinet6/in6_pcb.c (in6_pcbbind): correctly separated
IPv4-mapped address case in in6_pcbbind(). Reported from bsdi.
2002-02-04 JINMEI, Tatuya <>
* kame/kame/ping6: supported a new option '-g gateway' to allow a
specific next hop.
2002-02-03 JINMEI, Tatuya <>
* bsdi4/sbin/sysctl/sysctl.c (sysctl_key): supported net.key.*
2002-02-02 JINMEI, Tatuya <>
* freebsd4/sys/netinet6/udp6_usrreq.c (udp6_ctlinput):
* openbsd/sys/netinet/udp_usrreq.c (udp6_ctlinput):
* bsdi4/sys/netinet/udp_usrreq.c (udp6_ctlinput):
corrected arguments to ip6_pcbnotify().
2002-02-02 JINMEI, Tatuya <>
* freebsd4/sys/netinet/tcp_subr.c (tcp_respond): correctly reset
returned IPv6 header. This is essential when the original packet
contains an IPv6 extension header.
2002-02-02 JINMEI, Tatuya <>
* bsdi4/usr.sbin/netstat/inet.c (ipsec6_stats): supported printing
statistics for IPsec over IPv6.
2002-02-02 JINMEI, Tatuya <>
* many kernel files: totally revised IPv6 scoped address
architecture in the kernel.
- sticked to use sockaddr_in6, not in6_addr, as much as possible
- did not refer to ip6_src/ip6_dst (which may have ambiguity on
scope zones), but used full sockaddr_in6 structures attached to
the packet
- replaced special cases for link-local addresses with code that
used generic functions such as in6_addr2zoneid()
- additional cleanups mainly for scoped address handling
The change is so big and we'll need some more time to stabilize
the code. It is not recommended to use the latest code for
purposes that need stable behavior.
2002-01-29 Shin'ichi Fujisawa <>
* kame/sys/netinet6/natpt_{defs,rule,soctl,usrreq}.[ch]:
* kame/kame/natptconfig/misc.c:
- Change NATPT related ioctl macro name which begin with
"SIOC" to name which begin with "NATPT". "SIOC" as ioctl
macro name prefix is too general.
Mon Jan 28 17:19:19 JST 2002
* kame/sys/netnet6/{mip6.c,mip6_binding.c,mip6.h}
- fix a bug in the processing routine of the authentication data
- change the default security policy for protecting bu/ba.
if compiled for ID-15, IPsec = 0 and authdata = 1.
if compiled for ID-13, IPsec = 1.
Mon Jan 28 14:09:21 JST 2002
* bsdi4: upgrade base version to BSD/OS 4.3.
2002-01-26 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_ctloutput): handled
getsockopt(IPV6_RTHDR) correctly.
In response to: KAME PR 403 from
Sat Jan 26 11:50:54 JST 2002
* kame/kame/racoon:
the port number in the phase1 identifier is set 500
when the identifier type is the ip address
although it is described ambiguity in RFC2407 4.6.2.
2002-01-23 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_output): checked the
IPV6_MINMTU flag in ip6_output as well as the IP6PO_MINMTU flag in
the outgoing packet options. The former can still be set by
icmp6_reflect or gif_output.
Mon Jan 21 21:07:25 JST 2002
* kame/sys/netinet6/mip6*,kame/kame/mip6control/
add swithces to enable/disable
- the check code of bu/ba if they are protected by the ipsec.
- the check code of bu/ba if they are protected by the authdata.
to enable/disable those switches, use mip6cotrol.
2002-01-21 JINMEI, Tatuya <>
* kame/sys/netinet6/in6_pcb.h:
* {bsdi4, freebsd, openbsd}/sys/netinet/in_pcb.h:
- added foreign/local socket address structures to in[6]pcb{} in order
to have scope zone information of IPv6 addresses
- changed [fl]addr and [fl]port as shortcut macro
* kame/sys/netinet6/in6_pcb.c (in6_pcballoc):
* {bsdi4, freebsd, openbsd}/sys/netinet/in_pcb.c (in_pcballoc):
- set sin6_family and sin6_len when allocating an AF_INET6 pcb
Note: this change implicitly affected applications that referred
to the in[6]pcb structure (e.g. netstat). Be sure to update
header files and recompile all KAME applications.
2002-01-21 JINMEI, Tatuya <>
* kame/kame/wru: allowed the user to omit the hostname, in which
case the command used ff02::1 disambiguating the link zone using
the default interface.
2002-01-21 JINMEI, Tatuya <>
* kame/sys/netinet6/route6.c (ip6_rthdr0): made the source routing
code more scope-aware:
- it now considers all type of scopes (i.e. not only link-local
- it conforms to the forwarding rule described in the scoping arch
2002-01-21 SUZUKI, Shinsuke <>
* kame/kame/pim6sd/mldv2_proto.c:
- fixed a bug that MLDv2 Report crashes pim6sd.
2002-01-20 JINMEI, Tatuya <>
improved the support of IPv6 scoped addresses:
* kame/sys/netinet6/ip6_var.h: added sockaddr_in6 structures to
record source and destinaion addresses with scope information.
* kame/sys/netinet6/ip6_input.c (ip6_input): recorded the
* kame/sys/netinet6/ip6_forward.c(ip6_forward): used the recorded
addresses to check scope breakage and to get a route.
2002-01-20 JINMEI, Tatuya <>
* kame/sys/netinet6/udp6_output.c (udp6_output): corrected the
length argument to in_cksum (bsdi4 only).
2002-01-20 JINMEI, Tatuya <>
* freebsd4/sys/netinet/in_pcb.c (in_pcbladdr):
* {freebsd4, netbsd, openbsd}/sys/netinet/ip_output.c (ip_output):
* {netbsd, openbsd}/sys/netinet/in_pcb.c (in_selectsrc):
- check the address family of the destination cached in a PCB.
- clear the cached destination before getting another cached
route. Otherwise, garbage in the padding space (which might be
filled in if it was used for IPv6) could annoy rtalloc.
Note for OpenBSD: though Openbsd actually does not need these
fixes since it does not support IPv4-mapped IPv6 addresses, the
fixes are at least not harmful and will avoid possible bugs in the
future if it will ever support mapped addresses.
2002-01-20 JINMEI, Tatuya <>
* kame/sys/netinet6/in6_src.c (in6_selectroute): checked the
address family of a cached destination, in case of sharing the
cache with IPv4.
In response to: a bug report from Vineet Goel in the snap-users
2002-01-20 SUZUKI, Shinsuke <>
* kame/kame/pim6{dd,sd}/mld6.c:
fixed a bug that (*,G) entry diminishes due to a failure of
MLD Query advertisement.
2002-01-18 Shin'ichi Fujisawa <>
* kame/sys/netinet6/natpt_*.[chly]:
- Change a method of modification or reference of NAT-PT local
variable. Natptconfig show subcommand does not read
/dev/kmem any more when refer to variable.
2002-01-12 JINMEI, Tatuya <>
* kame/sys/netinet6/icmp6.c (icmp6_rip6_input): if the received
data is small enough but in an mbuf cluster, copy the data to a
separate mbuf that does not use a cluster.
This change will reduce the possiblity of packet loss in the
socket layer.
2002-01-12 JINMEI, Tatuya <>
* kame/kame/wru: was added as a separate command for ICMPv6 node
information queries. The first motivation of this command is to
provide the ability to handle the queries and responses with
platforms that do not support 'ping6 -w'. Though this command is
just a subset of 'ping6 -w' at this moment, it has some useful
properties comparing to the existing command. For example, wru
only sends a single query even for a multicast destination. In
this case, it collects replies during a certain amount of period,
prints the result, and then exits.
2002-01-11 JINMEI, Tatuya <>
* kame/sys/netinet6/in6_src.c (in6_selectif): do not use a
rejected or black hole route to pick the outgoing interface.
Otherwise, we could disambiguate a scoped destination with the
invalid route and would see some confusing results.
2002-01-11 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_input.c (ip6_input): do not accept packets
to a destination that has a rejected or blackhole host route.
2002-01-10 JINMEI, Tatuya <>
* (many kernel files): cleaned up the code to receive extension
headers (and other optional information) on TCP sockets. As a
result, the kernel TCP stack does not pass the information to
applications. All related functions were removed with this change,
while the "imputopts" member in the PCB structure was remained
just in case.
2002-01-10 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_output.c (ip6_clearpktopts): avoided NULL
pointer reference. The reference should have occurred in very
rare cases, but it is recommended to upgrade the kernel the
20011231 snap and later.
2002-01-10 JINMEI, Tatuya <>
* kame/sys/netinet6/ip6_input.c (ip6_savecontrol): do not compare
the previous value of each option to decide if the option should
be passed to application. This part has been removed from the
advanced API spec.
2002/01/07 21:26:21 JST
add ECN (Explicit Congestion Notification) support (RFC3168).
the ECN implementation consists of 3 independent components:
- marking mechanism in ALTQ
- tunnel-egress and fragment reassenbly rules in layer-3
- TCP mechanisms
to enable ECN support in TCP, build a kernel with TCP_ECN
and turn it on by sysctl -w net.inet.tcp.ecn=1
netstat(1) shows the ECN related statistics.
a simple example setup for ALTQ to mark packets on interface xl0:
interface xl0 bandwidth 3M red ecn
2002-01-03 JINMEI, Tatuya <>
* bsdi4/sys/netinet/ip_icmp.c (icmp_input): passed applications
icmp6 error packets that contained ipv6 over ipv4 packets,
even if the error packet did not have enough length to store the
whole IPv6 packets. traceroute -6 in fact needed this behavior.
2002-01-02 JINMEI, Tatuya <>
* kame/kame/traceroute: added a new option "-6" to send IPv6 over
IPv4 probe packets. This option will specifically be useful to
see if an intermediate router rejects the tunneled packets.
2002-01-02 JINMEI, Tatuya <>
* kame/sys/netinet6/in6.h: restricted the use of
IPV6_RECVRTHDRDSTOPTS in the kernel. This option was obsoleted in
Jump to Line
Something went wrong with that request. Please try again.