ldpd(8): Fix address of misaligned packed members.

PR kern/56895

Make use of packed member addreses non-fatal for clang, too

ldpd: Discard warning about accessing packed member

I don't see a good way of fixing this right now.

Attempt to write a manpage for ldpd.conf

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.3.6; 1.3.12;
* add config file so one can control id, timers and label assignment and
use neighbour specific options - XXX: needs documentation
* add peer authentication using TCP_MD5SIG. Interoperability tested with
Cisco IOS
* use SLIST_FOREACH_SAFE when deleting labels instead of re-looping.

Don't assign CFLAGS. Use default WARNS from usr.sbin/Makefile.inc.

Add ldpd, a RFC 3036 compatible LDP speaker.

* add config file so one can control id, timers and label assignment and
use neighbour specific options - XXX: needs documentation
* add peer authentication using TCP_MD5SIG. Interoperability tested with
Cisco IOS
* use SLIST_FOREACH_SAFE when deleting labels instead of re-looping.

Add ldpd, a RFC 3036 compatible LDP speaker.

fix ctype cast.

Deduplicate the code and refactor in a way that gcc-14 is not confused.

Avoid common symbol definitions

need <sys/stat.h>

branches: 1.11.8;
CVE 1129617: negative arg

CID 1125875: Fix memory leak

CID 1125876: File descriptor leak

sync behaviour with man page: accept spaces, empty lines and comments

allow setting transport addresses for interfaces into config file
also move passive-interface functionality under interface block
report the correct line for config parsing errors

sprinkle more const

* add a new keyword for config file - passive-if and check if it's
allowed to use the interface before join/send mcast
* check if interface supports multicast before join/send mcast

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.3.2; 1.3.8;
* add no-default-route option into configuration file and check it before
rejecting default-route
* exit process in case of error into the main loop
* complete FSM for RTM_CHANGE
* Check if we overflow pollfd array

branches: 1.2.4;
- detect calloc failure
- int -> size_t
- cache strlen
- proper cast for ctype macro

* add config file so one can control id, timers and label assignment and
use neighbour specific options - XXX: needs documentation
* add peer authentication using TCP_MD5SIG. Interoperability tested with
Cisco IOS
* use SLIST_FOREACH_SAFE when deleting labels instead of re-looping.

Avoid common symbol definitions

allow setting transport addresses for interfaces into config file
also move passive-interface functionality under interface block
report the correct line for config parsing errors

sprinkle more const

* add a new keyword for config file - passive-if and check if it's
allowed to use the interface before join/send mcast
* check if interface supports multicast before join/send mcast

branches: 1.1.6; 1.1.12;
* add config file so one can control id, timers and label assignment and
use neighbour specific options - XXX: needs documentation
* add peer authentication using TCP_MD5SIG. Interoperability tested with
Cisco IOS
* use SLIST_FOREACH_SAFE when deleting labels instead of re-looping.

ldpd(8): Fix address of misaligned packed members.

PR kern/56895

branches: 1.15.18; 1.15.28;
Merge riastradh-drm2 to HEAD.

don't connect on first hello, there are chances that ours is not seen yet
setproctitle with ldp id - useful for rump kernels testing
fix a memory leak in ldp_peer_new
don't holddown if already holded down
peer sockets are now non-blocking
connected routes deletes are now processed
check if peer is connected before attempting to sending label mappings

branches: 1.13.2;
correct loopback address test
correct addr add/del handling

Local addresses vector is not needed anymore

Constify a little bit

Stop confusing peer hello source with peer transport address

Don't try to connect to peers before sending hellos

Fix the transport address TLV mess I created after INET6 convert
Use a single loop to decrement and check hello list keepalives
Display transport address in show hello output

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.5.2; 1.5.8;
obey RFC3036 3.5.2 regarding holdtimes
wire SIGINT

Use IN_LOOPBACKNET where is the case
Don't forget to freeifaddrs() in case of error
Update a couple of debug messages

branches: 1.3.4;
fix possibly uninitialized variable

- no cast for malloc
- malloc + memset = calloc
- sizeof(type) -> sizeof(*var)
- small indents

Add ldpd, a RFC 3036 compatible LDP speaker.

Constify a little bit

Don't try to connect to peers before sending hellos

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.1.6; 1.1.12;
Add ldpd, a RFC 3036 compatible LDP speaker.

Store local bindings in a rbtree

* use a flag in label struct in order to determine if we should treat the
associated routes using RTF_HOST
* clean the mess in flush_mpls_routes that could result in accessing
uninitialized memory

* rework the reattachment code
* build an implnull label when a route is added and let mpls_add_label
to take care of relabelling
* take out last piece of the code where we do route lookups - we should have
every information we need in labels
* tested these changes using a number of rump kernels and some couple of
thousands of flapping routes

Make sure labels are always updated when a route is added and when a peer
is added
Rework mpls_add_label according to that so no route refresh is done anymore
Use poll when reading the PF_ROUTE socket
setsockopt SO_USELOOPBACK on the PF_ROUTE socket
Output some information on SIGINFO
Allow map changing for a ldp peer
Finally fix the connected routes admission into labels
Correct the route trigger when a label map is received

branches: 1.7.2;
retire union_ntoa, replace it with satos
check for valid sizes on PF_ROUTE socket
minor comment update

sprinkle more const

Constify a little bit

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.3.6; 1.3.12;
* add config file so one can control id, timers and label assignment and
use neighbour specific options - XXX: needs documentation
* add peer authentication using TCP_MD5SIG. Interoperability tested with
Cisco IOS
* use SLIST_FOREACH_SAFE when deleting labels instead of re-looping.

- no cast for malloc
- malloc + memset = calloc
- sizeof(type) -> sizeof(*var)
- small indents

Add ldpd, a RFC 3036 compatible LDP speaker.

Store local bindings in a rbtree

* use a flag in label struct in order to determine if we should treat the
associated routes using RTF_HOST
* clean the mess in flush_mpls_routes that could result in accessing
uninitialized memory

* rework the reattachment code
* build an implnull label when a route is added and let mpls_add_label
to take care of relabelling
* take out last piece of the code where we do route lookups - we should have
every information we need in labels
* tested these changes using a number of rump kernels and some couple of
thousands of flapping routes

Make sure labels are always updated when a route is added and when a peer
is added
Rework mpls_add_label according to that so no route refresh is done anymore
Use poll when reading the PF_ROUTE socket
setsockopt SO_USELOOPBACK on the PF_ROUTE socket
Output some information on SIGINFO
Allow map changing for a ldp peer
Finally fix the connected routes admission into labels
Correct the route trigger when a label map is received

branches: 1.4.2;
sprinkle more const

Constify a little bit

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.1.6; 1.1.12;
Add ldpd, a RFC 3036 compatible LDP speaker.

Remove accidentally added file

Add ldpd, a RFC 3036 compatible LDP speaker.

use predefined group consts/macros instead of ptoning strings

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.3.2; 1.3.8;
obey RFC3036 3.5.2 regarding holdtimes
wire SIGINT

branches: 1.2.4;
* add config file so one can control id, timers and label assignment and
use neighbour specific options - XXX: needs documentation
* add peer authentication using TCP_MD5SIG. Interoperability tested with
Cisco IOS
* use SLIST_FOREACH_SAFE when deleting labels instead of re-looping.

Add ldpd, a RFC 3036 compatible LDP speaker.

Use rbtree for storing peers FEC label bindings

Store local bindings in a rbtree

Make sure labels are always updated when a route is added and when a peer
is added
Rework mpls_add_label according to that so no route refresh is done anymore
Use poll when reading the PF_ROUTE socket
setsockopt SO_USELOOPBACK on the PF_ROUTE socket
Output some information on SIGINFO
Allow map changing for a ldp peer
Finally fix the connected routes admission into labels
Correct the route trigger when a label map is received

branches: 1.11.2;
retire union_ntoa, replace it with satos
check for valid sizes on PF_ROUTE socket
minor comment update

Use satos instead of inet_ntop
Fix an alloc problem

Fix the transport address TLV mess I created after INET6 convert
Use a single loop to decrement and check hello list keepalives
Display transport address in show hello output

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.7.6;
avoid reusing inet_addr twice.

PR/45662: Henning Petersen: Overlapping buffer in lpd_command.c

branches: 1.5.2;
Introduce 'show labels' command that reports about labels received from
unsolicited label-maps. Statify.

branches: 1.4.4;
show in neighbour information if peer is MD5 authenticated

* add config file so one can control id, timers and label assignment and
use neighbour specific options - XXX: needs documentation
* add peer authentication using TCP_MD5SIG. Interoperability tested with
Cisco IOS
* use SLIST_FOREACH_SAFE when deleting labels instead of re-looping.

fix uninitialized variable by elimination.

Add ldpd, a RFC 3036 compatible LDP speaker.

Merge riastradh-drm2 to HEAD.

Make sure labels are always updated when a route is added and when a peer
is added
Rework mpls_add_label according to that so no route refresh is done anymore
Use poll when reading the PF_ROUTE socket
setsockopt SO_USELOOPBACK on the PF_ROUTE socket
Output some information on SIGINFO
Allow map changing for a ldp peer
Finally fix the connected routes admission into labels
Correct the route trigger when a label map is received

branches: 1.2.2; 1.2.8; 1.2.12;
Introduce 'show labels' command that reports about labels received from
unsolicited label-maps. Statify.

branches: 1.1.4;
Add ldpd, a RFC 3036 compatible LDP speaker.

retire union_ntoa, replace it with satos
check for valid sizes on PF_ROUTE socket
minor comment update

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.2.2;
Add __printflike attribution to use vprintf and friends with an argument
as format string.

branches: 1.1.6;
Add ldpd, a RFC 3036 compatible LDP speaker.

Be a little more strict when sending notifications and checking PDU

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.3.2;
Add __printflike attribution to use vprintf and friends with an argument
as format string.

branches: 1.2.2;
* add no-default-route option into configuration file and check it before
rejecting default-route
* exit process in case of error into the main loop
* complete FSM for RTM_CHANGE
* Check if we overflow pollfd array

branches: 1.1.4;
Add ldpd, a RFC 3036 compatible LDP speaker.

ldpd(8): Fix address of misaligned packed members.

PR kern/56895

s/successfuly/successfully/ in comment. No functional change.

Avoid common symbol definitions

branches: 1.16.18; 1.16.28;
Use rbtree for storing peers FEC label bindings

don't connect on first hello, there are chances that ours is not seen yet
setproctitle with ldp id - useful for rump kernels testing
fix a memory leak in ldp_peer_new
don't holddown if already holded down
peer sockets are now non-blocking
connected routes deletes are now processed
check if peer is connected before attempting to sending label mappings

Make sure labels are always updated when a route is added and when a peer
is added
Rework mpls_add_label according to that so no route refresh is done anymore
Use poll when reading the PF_ROUTE socket
setsockopt SO_USELOOPBACK on the PF_ROUTE socket
Output some information on SIGINFO
Allow map changing for a ldp peer
Finally fix the connected routes admission into labels
Correct the route trigger when a label map is received

branches: 1.13.2;
Local addresses vector is not needed anymore

Constify a little bit

move code around a little bit in order to call get_ldp_peer correctly
check also for transport_address in get_ldp_peer

* Don't assume INET in connection path
* Lookup in hello list in order to get the correct LDP ID, instead of
transport address
* Improve an error message

correct the sockaddr compare
modify a couple of debug messages

Be a little more strict when sending notifications and checking PDU

Use satos instead of inet_ntop
Fix an alloc problem

re-enable peer authentication

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.3.6; 1.3.12;
* add config file so one can control id, timers and label assignment and
use neighbour specific options - XXX: needs documentation
* add peer authentication using TCP_MD5SIG. Interoperability tested with
Cisco IOS
* use SLIST_FOREACH_SAFE when deleting labels instead of re-looping.

- no cast for malloc
- malloc + memset = calloc
- sizeof(type) -> sizeof(*var)
- small indents

Add ldpd, a RFC 3036 compatible LDP speaker.

Avoid common symbol definitions

Use rbtree for storing peers FEC label bindings

don't connect on first hello, there are chances that ours is not seen yet
setproctitle with ldp id - useful for rump kernels testing
fix a memory leak in ldp_peer_new
don't holddown if already holded down
peer sockets are now non-blocking
connected routes deletes are now processed
check if peer is connected before attempting to sending label mappings

branches: 1.6.2;
Local addresses vector is not needed anymore

Constify a little bit

* Don't assume INET in connection path
* Lookup in hello list in order to get the correct LDP ID, instead of
transport address
* Improve an error message

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.1.6; 1.1.12;
Add ldpd, a RFC 3036 compatible LDP speaker.

Grammar. Sort sections.

Mention IPv6, operation mode and re-section a bit

New sentence, new line. Bump date for previous.

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.5.2; 1.5.8;
Mention that it doesn't support Path Vectors and Hop Counts
Mention RFC3037 and xref ldpd.conf

Sort SYNOPSIS, use Pa for path.

* add config file so one can control id, timers and label assignment and
use neighbour specific options - XXX: needs documentation
* add peer authentication using TCP_MD5SIG. Interoperability tested with
Cisco IOS
* use SLIST_FOREACH_SAFE when deleting labels instead of re-looping.

Add all options to SYNOPSIS. New sentence, new line. Some rewording.
Use more macros.

Add ldpd, a RFC 3036 compatible LDP speaker.

branches: 1.2.4; 1.2.8;
Sort SEE ALSO. Grammar consistency.

Attempt to write a manpage for ldpd.conf

Store local bindings in a rbtree

* don't request explicitly for uid 0 as various system policies may make that
unnecessary
* correct the mplsif message. We don't need an IP address assigned on it
anymore

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.6.2; 1.6.8;
return EXIT_SUCCESS and EXIT_FAILURE instead on 0 and -1

* add no-default-route option into configuration file and check it before
rejecting default-route
* exit process in case of error into the main loop
* complete FSM for RTM_CHANGE
* Check if we overflow pollfd array

branches: 1.4.4;
Make usage match SYNOPSIS.

* add config file so one can control id, timers and label assignment and
use neighbour specific options - XXX: needs documentation
* add peer authentication using TCP_MD5SIG. Interoperability tested with
Cisco IOS
* use SLIST_FOREACH_SAFE when deleting labels instead of re-looping.

Add all options to usage.

Add ldpd, a RFC 3036 compatible LDP speaker.

Use rbtree for storing peers FEC label bindings

* use a flag in label struct in order to determine if we should treat the
associated routes using RTF_HOST
* clean the mess in flush_mpls_routes that could result in accessing
uninitialized memory

* rework the reattachment code
* build an implnull label when a route is added and let mpls_add_label
to take care of relabelling
* take out last piece of the code where we do route lookups - we should have
every information we need in labels
* tested these changes using a number of rump kernels and some couple of
thousands of flapping routes

Make sure labels are always updated when a route is added and when a peer
is added
Rework mpls_add_label according to that so no route refresh is done anymore
Use poll when reading the PF_ROUTE socket
setsockopt SO_USELOOPBACK on the PF_ROUTE socket
Output some information on SIGINFO
Allow map changing for a ldp peer
Finally fix the connected routes admission into labels
Correct the route trigger when a label map is received

branches: 1.9.2;
Constify a little bit

allow INET6 gateways, split some long lines

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.6.2; 1.6.8;
* add no-default-route option into configuration file and check it before
rejecting default-route
* exit process in case of error into the main loop
* complete FSM for RTM_CHANGE
* Check if we overflow pollfd array

branches: 1.5.2;
use RT_ROUNDUP provided by net/route.h

branches: 1.4.2;
Fix more memory leaks in case malloc fails

Fix memory leak in error case. Found by cppcheck.

XXX: so_tag has same problem?

- no cast for malloc
- malloc + memset = calloc
- sizeof(type) -> sizeof(*var)
- small indents

Add ldpd, a RFC 3036 compatible LDP speaker.

Merge riastradh-drm2 to HEAD.

Make sure labels are always updated when a route is added and when a peer
is added
Rework mpls_add_label according to that so no route refresh is done anymore
Use poll when reading the PF_ROUTE socket
setsockopt SO_USELOOPBACK on the PF_ROUTE socket
Output some information on SIGINFO
Allow map changing for a ldp peer
Finally fix the connected routes admission into labels
Correct the route trigger when a label map is received

branches: 1.3.2;
Constify a little bit

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.1.6; 1.1.12;
Add ldpd, a RFC 3036 compatible LDP speaker.

fix various typos in comments.

Separate nexthop caches from the routing table

By this change, nexthop caches (IP-MAC address pair) are not stored
in the routing table anymore. Instead nexthop caches are stored in
each network interface; we already have lltable/llentry data structure
for this purpose. This change also obsoletes the concept of cloning/cloned
routes. Cloned routes no longer exist while cloning routes still exist
with renamed to connected routes.

Noticeable changes are:
- Nexthop caches aren't listed in route show/netstat -r
- sysctl(NET_RT_DUMP) doesn't return them
- If RTF_LLDATA is specified, it returns nexthop caches
- Several definitions of routing flags and messages are removed
- RTF_CLONING, RTF_XRESOLVE, RTF_LLINFO, RTF_CLONED and RTM_RESOLVE
- RTF_CONNECTED is added
- It has the same value of RTF_CLONING for backward compatibility
- route's -xresolve, -[no]cloned and -llinfo options are removed
- -[no]cloning remains because it seems there are users
- -[no]connected is introduced and recommended
to be used instead of -[no]cloning
- route show/netstat -r drops some flags
- 'L' and 'c' are not seen anymore
- 'C' now indicates a connected route
- Gateway value of a route of an interface address is now not
a L2 address but "link#N" like a connected (cloning) route
- Proxy ARP: "arp -s ... pub" doesn't create a route

You can know details of behavior changes by seeing diffs under tests/.

Proposed on tech-net and tech-kern:
http://mail-index.netbsd.org/tech-net/2016/03/11/msg005701.html

zeroize the rest of the comparable struct before adding it to the rb tree

Store local bindings in a rbtree

check route messages sizes more carefully
cache pid
treat every message at a time, even if there are more messages to read
interpret also cloning routes

The latter two should fix the ldp_regen test

* use a flag in label struct in order to determine if we should treat the
associated routes using RTF_HOST
* clean the mess in flush_mpls_routes that could result in accessing
uninitialized memory

don't connect on first hello, there are chances that ours is not seen yet
setproctitle with ldp id - useful for rump kernels testing
fix a memory leak in ldp_peer_new
don't holddown if already holded down
peer sockets are now non-blocking
connected routes deletes are now processed
check if peer is connected before attempting to sending label mappings

* rework the reattachment code
* build an implnull label when a route is added and let mpls_add_label
to take care of relabelling
* take out last piece of the code where we do route lookups - we should have
every information we need in labels
* tested these changes using a number of rump kernels and some couple of
thousands of flapping routes

Make sure labels are always updated when a route is added and when a peer
is added
Rework mpls_add_label according to that so no route refresh is done anymore
Use poll when reading the PF_ROUTE socket
setsockopt SO_USELOOPBACK on the PF_ROUTE socket
Output some information on SIGINFO
Allow map changing for a ldp peer
Finally fix the connected routes admission into labels
Correct the route trigger when a label map is received

branches: 1.16.2;
create binding for connected routes on startup
decay an warning message to debug only

retire union_ntoa, replace it with satos
check for valid sizes on PF_ROUTE socket
minor comment update

correct loopback address test
correct addr add/del handling

add code for address and address withdraw messages
get rid of unused fields in tlv struct

sprinkle more const

speed up a little bit the string<->prefixlen transforms

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.9.2;
Add __printflike attribution to use vprintf and friends with an argument
as format string.

branches: 1.8.2;
Correct prefixlen->sockaddr calculation. Now ldpd may finally allow MPLS
forwarding for default route.

shorten some execution paths related to check_route() and get_route()
statify a bit
don't assume AF_INET in a debug message

* add no-default-route option into configuration file and check it before
rejecting default-route
* exit process in case of error into the main loop
* complete FSM for RTM_CHANGE
* Check if we overflow pollfd array

branches: 1.5.2;
get rid of unused variable/assignments

use RT_ROUNDUP provided by net/route.h

branches: 1.3.2;
check from_cidr_to_union() result

- no cast for malloc
- malloc + memset = calloc
- sizeof(type) -> sizeof(*var)
- small indents

Add ldpd, a RFC 3036 compatible LDP speaker.

Do not pack struct rt_msg - this voids the alignement restrictions and
causes crashes on alignemen critical archs.
From Matt Thomas.

* rework the reattachment code
* build an implnull label when a route is added and let mpls_add_label
to take care of relabelling
* take out last piece of the code where we do route lookups - we should have
every information we need in labels
* tested these changes using a number of rump kernels and some couple of
thousands of flapping routes

branches: 1.5.2;
retire union_ntoa, replace it with satos
check for valid sizes on PF_ROUTE socket
minor comment update

sprinkle more const

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

Make sure there is enough space for sockaddr subtypes
Use u_char for IP_MULTICAST_LOOP as specified in in.h

branches: 1.1.6; 1.1.12;
Add ldpd, a RFC 3036 compatible LDP speaker.

Constify a little bit

branches: 1.2.6; 1.2.12;
- no cast for malloc
- malloc + memset = calloc
- sizeof(type) -> sizeof(*var)
- small indents

Add ldpd, a RFC 3036 compatible LDP speaker.

Constify a little bit

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.1.6; 1.1.12;
Add ldpd, a RFC 3036 compatible LDP speaker.

Constify a little bit

* Don't assume INET in connection path
* Lookup in hello list in order to get the correct LDP ID, instead of
transport address
* Improve an error message

Be a little more strict when sending notifications and checking PDU

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.1.6; 1.1.12;
Add ldpd, a RFC 3036 compatible LDP speaker.

Constify a little bit

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.1.6; 1.1.12;
Add ldpd, a RFC 3036 compatible LDP speaker.

ldpd(8): Fix address of misaligned packed members.

PR kern/56895

Avoid common symbol definitions

branches: 1.34.4; 1.34.14;
Use RO_MSGFILTER.

branches: 1.33.8; 1.33.12;
ensure variables are initialised

allow setting transport addresses for interfaces into config file
also move passive-interface functionality under interface block
report the correct line for config parsing errors

check route messages sizes more carefully
cache pid
treat every message at a time, even if there are more messages to read
interpret also cloning routes

The latter two should fix the ldp_regen test

don't connect on first hello, there are chances that ours is not seen yet
setproctitle with ldp id - useful for rump kernels testing
fix a memory leak in ldp_peer_new
don't holddown if already holded down
peer sockets are now non-blocking
connected routes deletes are now processed
check if peer is connected before attempting to sending label mappings

Make sure labels are always updated when a route is added and when a peer
is added
Rework mpls_add_label according to that so no route refresh is done anymore
Use poll when reading the PF_ROUTE socket
setsockopt SO_USELOOPBACK on the PF_ROUTE socket
Output some information on SIGINFO
Allow map changing for a ldp peer
Finally fix the connected routes admission into labels
Correct the route trigger when a label map is received

branches: 1.28.2;
Constify a little bit

First send initialize and keep alive after that. Fixes inter-operability
issues regarding session initialization.

move code around a little bit in order to call get_ldp_peer correctly
check also for transport_address in get_ldp_peer

* Don't assume INET in connection path
* Lookup in hello list in order to get the correct LDP ID, instead of
transport address
* Improve an error message

Don't try to connect to peers before sending hellos

Be a little more strict when sending notifications and checking PDU

permit notifications even if session not fully established

Fix the transport address TLV mess I created after INET6 convert
Use a single loop to decrement and check hello list keepalives
Display transport address in show hello output

* add a new keyword for config file - passive-if and check if it's
allowed to use the interface before join/send mcast
* check if interface supports multicast before join/send mcast

use predefined group consts/macros instead of ptoning strings

allow compile without -DINET6

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

Make sure there is enough space for sockaddr subtypes
Use u_char for IP_MULTICAST_LOOP as specified in in.h

Use local-link address as source for hellos

Repair another %lu --> %zd - this one was caught by building for
evbmips port

fix the build; ssize_t is %zd not %lu

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.11.2; 1.11.8;
Use __dead

obey RFC3036 3.5.2 regarding holdtimes
wire SIGINT

clear the error condition
KNF a little bit

simplify the rtsock read sequence
apply a little bit of KNF

* add no-default-route option into configuration file and check it before
rejecting default-route
* exit process in case of error into the main loop
* complete FSM for RTM_CHANGE
* Check if we overflow pollfd array

branches: 1.6.2;
Packed pointers are non-sense

avoid advertising 0/8 transport address (conforming RFC1122
Section 3.2.1.3)

use SLIST_FOREACH_SAFE when deleting a peer in loop

branches: 1.3.2;
* add config file so one can control id, timers and label assignment and
use neighbour specific options - XXX: needs documentation
* add peer authentication using TCP_MD5SIG. Interoperability tested with
Cisco IOS
* use SLIST_FOREACH_SAFE when deleting labels instead of re-looping.

- no cast for malloc
- malloc + memset = calloc
- sizeof(type) -> sizeof(*var)
- small indents

Add ldpd, a RFC 3036 compatible LDP speaker.

Avoid common symbol definitions

Constify a little bit

Fix the transport address TLV mess I created after INET6 convert
Use a single loop to decrement and check hello list keepalives
Display transport address in show hello output

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.2.2; 1.2.8;
* add no-default-route option into configuration file and check it before
rejecting default-route
* exit process in case of error into the main loop
* complete FSM for RTM_CHANGE
* Check if we overflow pollfd array

branches: 1.1.4;
Add ldpd, a RFC 3036 compatible LDP speaker.

Add ldpd, a RFC 3036 compatible LDP speaker.

ldpd: remove unused bit-field member from anonymous union

The struct common_hello_tlv contained an anonymous union, which is a C11
feature. That union contained a bit-field, which is very unusual since
the C standards answer basically all interesting questions about
bit-fields in unions with "implementation-defined" or "unspecified".
GCC passes these questions further down by saying "Determined by ABI".

This was the only occurrence of a bit-field union member in the whole
NetBSD tree, and it caused the lint diagnostic 41 to be downgraded from
error to warning on 2021-09-17 since this exotic construct is allowed by
the C standards but practically unused.

No binary change.

add code for address and address withdraw messages
get rid of unused fields in tlv struct

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.1.6; 1.1.12;
Add ldpd, a RFC 3036 compatible LDP speaker.

Store local bindings in a rbtree

don't connect on first hello, there are chances that ours is not seen yet
setproctitle with ldp id - useful for rump kernels testing
fix a memory leak in ldp_peer_new
don't holddown if already holded down
peer sockets are now non-blocking
connected routes deletes are now processed
check if peer is connected before attempting to sending label mappings

* rework the reattachment code
* build an implnull label when a route is added and let mpls_add_label
to take care of relabelling
* take out last piece of the code where we do route lookups - we should have
every information we need in labels
* tested these changes using a number of rump kernels and some couple of
thousands of flapping routes

Make sure labels are always updated when a route is added and when a peer
is added
Rework mpls_add_label according to that so no route refresh is done anymore
Use poll when reading the PF_ROUTE socket
setsockopt SO_USELOOPBACK on the PF_ROUTE socket
Output some information on SIGINFO
Allow map changing for a ldp peer
Finally fix the connected routes admission into labels
Correct the route trigger when a label map is received

branches: 1.9.2;
send the actual prefix on the wire instead of struct headers
statify

sprinkle more const

Constify a little bit

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

* add initial IPv6 support - still incomplete at this moment, but it sends
out there IPv6 hellos. Interoperability not yet tested.
* sync man page with reality

branches: 1.4.2; 1.4.8;
we need to add mapping even if there is no route for the prefix

branches: 1.3.2;
stop processing a label mapping message if we don't have a route for it.
This way we avoid failing an assert later in mpls_add_label()

branches: 1.2.2;
- no cast for malloc
- malloc + memset = calloc
- sizeof(type) -> sizeof(*var)
- small indents

Add ldpd, a RFC 3036 compatible LDP speaker.

sprinkle more const

Constify a little bit

modify structures in order to accomodate IPv6 according to
draft-ietf-mpls-ldp-ipv6. Correct a little bit IPv6 hello path

branches: 1.1.6; 1.1.12;
Add ldpd, a RFC 3036 compatible LDP speaker.