eigrpd: Correct route send to zebra

We were not passing the instance and were also passing
a bogus distance?

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

eigrpd: Fix eigrp interface header

The switch over to using the new cli grammar
caused the 'show ip eigrp int' command to not
display a header in some cases.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

vtysh: Rehook lib/command.c:cmd_init

Vtysh was not running lib/command.c commands for eigrp

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

eigrpd: Update .gitignore

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

eigrpd: Fix to compile under gcc

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

eigrpd: Fix compilation for gcc on some platforms.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

eigrpd: Add missing vrf registration piece

Add vrf registration so that we receive
information about the default vrf.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

vtysh: Fix eigrp ordering in 'show run'

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

eigrp: Fix Compile issues

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

vtysh: Fix misplaced (

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

eigrp: Initial Commit

Please Note, I will be redoing this commit message with
more information.

Additionally I will rework the lib/* changes into their
own commits.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

Merge pull request #262 from donaldsharp/pim_jp_fixes

Pim jp fixes

pimd: Don't double include Source

When we are adding to a J/P Agg list, don't
double add the source.

Ticket: CM-15189
Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

pimd: Fix various ifdown/ifup scenarios w/ J/P Agg

There exists situations where we may have cleaned not
properly cleaned up the various J/P aggregation lists.
This commit fixes those issues.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

Merge pull request #261 from opensourcerouting/lib_cleanup

startup, option parsing & logging refactor

*: fix some option descriptions and make -e work

There were some typos in command line help, also the "e:" spec was
missing for getopt in zebra/bgpd (even before the refactor).

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

lib: time: remove leftover declarations

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

lib: hide away logging internals

... no need to have struct zlog generally-exposed.

A few files get to include log_int.h because they use zlog/vzlog.

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

lib: remove remaining struct zlog * args

These don't serve any purpose either.

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

lib: simplify protocol/log naming

The protocols enum serves no purpose other than adding potential for
bugs and making it complicated to add a new protocol... nuke.

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

lib: ditch struct zlog * argument on zlog/vzlog()

It's completely useless...

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

*: get rid of zlog(*, LOG_LEVEL, ...)

Result of running the following Coccinelle patch + fixups:
<<EOF
/* long-forms: zlog(NULL, <level>, ...)
 *    =>       zlog_level(...)
 */

@@
expression list args;
@@
- zlog(NULL, LOG_DEBUG, args)
+ zlog_debug(args)

@@
expression list args;
@@
- zlog(NULL, LOG_NOTICE, args)
+ zlog_notice(args)

@@
expression list args;
@@
- zlog(NULL, LOG_INFO, args)
+ zlog_info(args)

@@
expression list args;
@@
- zlog(NULL, LOG_WARNING, args)
+ zlog_warn(args)

@@
expression list args;
@@
- zlog(NULL, LOG_ERR, args)
+ zlog_err(args)

/* long-forms: zlog(base->log, <level>, ...)
 *    =>       zlog_level(...)
 */

@@
expression base;
expression list args;
@@
- zlog(base->log, LOG_DEBUG, args)
+ zlog_debug(args)

@@
expression base;
expression list args;
@@
- zlog(base->log, LOG_NOTICE, args)
+ zlog_notice(args)

@@
expression base;
expression list args;
@@
- zlog(base->log, LOG_INFO, args)
+ zlog_info(args)

@@
expression base;
expression list args;
@@
- zlog(base->log, LOG_WARNING, args)
+ zlog_warn(args)

@@
expression base;
expression list args;
@@
- zlog(base->log, LOG_ERR, args)
+ zlog_err(args)

EOF

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

*: add frr_run()

Contains the fetch-and-run-thread logic, and vty startup (which is the
last thing happening before entering the main loop).

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

*: add frr_config_fork()

Centralise read_config/daemonize/dryrun/pidfile/vty_serv into libfrr.

This also makes multi-instance pid/config handling available as part of
the library.  It's only wired up in ospfd, but the code is in lib/.

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

*: centralize more into frr_init()

Move CLI/VTY/Memory accounting init into frr_*

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

*: add frr_init() infrastructure

Start centralising startup & option parsing into the library.

FRR_DAEMON_INFO is a bit weird, but it will become useful later (e.g.
for killing the ZLOG_* enum, and having the daemon name available)

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

Merge pull request #257 from opensourcerouting/nhrpd

nhrpd

Merge pull request #254 from donaldsharp/shutdown3

Shutdown take 3

Merge branch 'frr/pull/250' ("bgpd, zebra: Add ifindex to NEXTHOP_TYPE_IPV4")

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

build: only default-enable nhrpd on Linux

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

cumulus, tools: Allow nhrpd to be brought up

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

Merge branch 'frr/pull/243' ("Pim jp agg")

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

pimd: Add copyright Information to pim_jp_agg.[ch]

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

distro/redhat: package nhrpd

Signed-off-by: Timo Teräs <timo.teras@iki.fi>
[DL: adapted for FRR, changed port number 2612->2610]
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

nhrpd: workaround old kernel vs. glibc definition conflics

fixes https://bugzilla.quagga.net/show_bug.cgi?id=908

With kernel-headers-3.10.0 we have:

In file included from netlink_gre.c:15:0:
/usr/include/linux/ipv6.h:19:8: error: redefinition of 'struct in6_pktinfo'
 struct in6_pktinfo {
        ^
In file included from netlink_gre.c:10:0:
/usr/include/netinet/in.h:536:8: note: originally defined here
 struct in6_pktinfo
        ^
In file included from netlink_gre.c:15:0:
/usr/include/linux/ipv6.h:24:8: error: redefinition of 'struct ip6_mtuinfo'
 struct ip6_mtuinfo {
        ^
In file included from netlink_gre.c:10:0:
/usr/include/netinet/in.h:543:8: note: originally defined here
 struct ip6_mtuinfo

So instead of libc's netinet/in.h include kernel's linux/in.h
and the add sys/socket.h for struct sockaddr since it does not
seem to be defined in kernel headers.

Signed-off-by: Timo Teräs <timo.teras@iki.fi>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

doc: add initial nhrpd documentation

[DL: resolved renaming & conflicts]
Signed-off-by: Timo Teräs <timo.teras@iki.fi>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

nhrpd: include headers in dist tarball

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

nhrpd: change VTY port 2612->2610

2612 is already in use for ldpd.
(Change to SERVICES file resolved in rebasing)

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

nhrpd: adapt to API changes

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

nhrpd: implement next hop resolution protocol

This provides DMVPN support and integrates to strongSwan. Please read
README.nhrpd and README.kernel for more details.

[DL: cherry-picked from dafa05e65fe4b3b3ed5525443f554215ba14f42c]
[DL: merge partially resolved, this commit will not build.]
Signed-off-by: Timo Teräs <timo.teras@iki.fi>
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

Merge branch 'frr/pull/255' ("vtysh: Fix cli help string to have only 1 mention of vty_socket")

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

Merge branch 'frr/pull/246' ("Pim command cleanup")

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

Merge branch 'frr/pull/240' ("lib, bgpd: clean up headers")

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

Merge branch 'frr/pull/236' ("tools: frr-reload.py needs to treat "mpls" as a single line context")

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

Merge branch 'frr/pull/228' ("doc: add CLI implementation documentation")

Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

Merge pull request #251 from donaldsharp/bgp_view

BGP Display

vtysh: Fix cli help string to have only 1 mention of vty_socket

When you run 'vtysh -h' the option '--vty_socket' is listed twice.
Fixes issue #253

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

bgpd: Only send data if we have something to send

When shutting down, only send with data if we have
something to send.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

bgpd: Add json output for last Notification message

When we send a shutdown message, add json
support to let us see it there too.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

bgpd: TX shutdown message

Signed-off-by: Christian Franke <chris@opensourcerouting.org>

bgpd: RX shutdown message in "show bgp neighbor"

Signed-off-by: Christian Franke <chris@opensourcerouting.org>

lib: add zlog_sanitize function

Signed-off-by: Christian Franke <chris@opensourcerouting.org>

bgpd: Remove dead code

Remove the accidently left code.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

bgpd, zebra: Add ifindex to NEXTHOP_TYPE_IPV6

In the future we plan to update Nexthop tracking to better
handle ipv6 lla.  This commit will set this up for that.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

bgpd: Fix more assignments inside of if statement

Code inspection showed a bunch more spots where
we were assigning the safi inside of a if statement

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

bgpd: Fix broken if statement

When outputting routes associated with a 'struct bgp'
do not force the safi to bee SAFI_EVPN and cause output
to be hosed.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

bgpd, zebra: Add ifindex to NEXTHOP_TYPE_IPV4

When passing up NEXTHOP_TYPE_IPV4 pass up the ifindex as well
Zebra already stores this data by passing it up PIM will be
able to use NEXTHOP_TYPE_IPV4 without having to do a recursive
lookup.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
Reviewed-by: Don Slice <dslice@cumulusnetworks.com>
Reviewed-by: Chirag Shah <chirag@cumulusnetworks.com>

Merge pull request #249 from opensourcerouting/ldpd-rfcs

ldpd: implement new RFCs and other assorted enhancements

ldpd: add missing htonl and ntohl in PF_KEY code

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: always check if the received labels are valid

We were doing some sanity checks only for labels of Label Mapping
messages. Now do the same checks for labels of Label Release and Label
Withdraw messages as well.

While here, plug a small memleak in the error path of recv_labelmessage().

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: remove unnecessary memcpy's

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: send VPLS MAC withdrawals

RFC 4762 says that MAC address withdrawal messages can be used to
improve convergence time in VPLS networks. This patch makes ldpd send
MAC withdrawals whenever a non-pseudowire interface pertaining to a
VPLS goes down. The processing of received MAC withdrawals will be
implemented later.

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: fix parsing of optional tlvs in address messages

We were aborting the session upon receipt of MAC Address Withdrawal
messages. Now make the parser aware that optional TLVs are possible in
address messages.

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: implement RFC 5919 (LDP End-of-LIB)

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: implement RFC 6667 (Typed Wildcard FEC for PWid)

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: implement RFC 5918 (Typed Wildcard FEC)

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: implement RFC 5561 (LDP Capabilities)

This patch per-se doesn't introduce any useful functionality, but prepares
the ground for new enhancements to ldpd (i.e. implementation of new RFCs
that make use of LDP capabilities).

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: implement support for PWid group wildcards

This was missing from our original RFC 4447 VPLS implementation. Now
ldpd understands group wildcards as mandated by the RFC, but we still
don't send them ourselves. I can't see any case in which sending a group
wildcard would be useful, but nonetheless this patch provides a function
called lde_send_labelwithdraw_pwid_wcard() which is ready to be used in
the future anytime we feel like it might be useful.

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: fix processing of Label Withdraw messages

Whenever we receive a Label Withdraw message with an optional Label
TLV, we should check if this label matches the label previously
received from this neighbor for this FEC. If they don't match then we
shouldn't uninstall the previous label from the kernel. This fixes a
misinterpretation from the "Receive Label Withdraw" algorithm described
in the A.1.5 section of RFC 5036.

Also, simplify the check of pending withdraws in lde_check_release()
and lde_check_release_wcard().

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: the PW Status is an unknown TLV

Without this fix, any LDP speaker that doesn't implement RFC 4447 will
shut down the session upon receipt of a PWid Label Mapping (unless the
use of the PW-Status TLV is disabled via configuration).

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: send correct status code on error condition

The Unknown TLV status code is reserved for cases where we don't know
how to process a given TLV and its Unknown TLV bit is not set.

In the case of Address Messages, the Address List TLV is mandatory and
should appear before any optional TLV in the message. If that doesn't
happen the correct status notification to send is "Missing Message
Parameters" (non-fatal error).

Also, add a missing htons when creating an Address List TLV. Since the
value of TLV_TYPE_ADDRLIST is 0x0101 this missing htons wasn't noticed
earlier.

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: create helper functions to log sent/received messages

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: provide more detailed output when logging notification messages

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: kill send_notification_nbr()

Be more clever and trigger the PDU SENT event inside send_notification()
when tcp->nbr is set. This way we can eliminate send_notification_nbr()
and always use send_notification() instead.

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: minor tweaks

Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: replace hand-rolled 'for' loop with specialized macro

No intentional functional change.

Original author: Kenneth R Westerback <krw@openbsd.org>
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

tools: frr-reload.py needs to treat "no " as a single line context

Signed-off-by: Daniel Walton <dwalton@cumulusnetworks.com>
Reviewed-by: Donald Sharp <sharpd@cumulusnetworks.com>

Merge pull request #245 from donaldsharp/ldpd_load

Ldpd load

lib: Cleanup missing help strings for 'show commandtree..'

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

cumulus: Add startup parameters for ldpd

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

Merge pull request #233 from donaldsharp/ecmp1

bgpd, zebra: Allow setting ecmp from daemon cli

bgp: Fix view/vrf lookup

This commit fixes the view/vrf lookup that got broken.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

pimd: Fix 'show ip pim state' command

The 'show ip pim state' command was not
working quite correctly from when
we moved over to the new vtysh-grammar.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

pimd: Hide 'ip multicast-routing' command

Automatically turn on pim if the daemon is turned on.
Hide the 'ip multicast-routing' command
Disable the 'no ip multicast-routing' command
Do not display the 'ip multicast-routing' command.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

pimd: Turn on igmp when igmp commands are entered

When you enter these commands, under an interface:

ip igmp version ...
ip igmp query-interval ...
ip igmp query-max-response-time ...
ip igmp query-max-response-time-dsec ...

These commands will now just turn on igmp for the
interface.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

cumulus: Add new daemons to daemons file

Add the existence of new daemons to the daemons
file.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

tools: Allow ldpd to be started

Allow ldpd to be started at runtime with the frr script.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

pimd: Join/Prune Aggregation

Add the ability for PIM to send Join/Prunes as an
aggregated message instead of individual messages
for each S,G.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

pimd: Track oil list totals a bit better.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

bgpd: remove unnecessary #include "vty.h"

Per previous commit, these are no longer necessary.

Signed-off-by: Quentin Young <qlyoung@cumulusnetworks.com>

bgpd: add vty.h to bgpd.h

`struct vty` referenced in body of bgpd.h, so it should be #include'd.

Signed-off-by: Quentin Young <qlyoung@cumulusnetworks.com>

lib: add stream.h, vty.h, zebra.h to plist.h

Since `afi_t` and `struct vty` are used in plist.h, the appropriate
headers for them should be included.

Signed-off-by: Quentin Young <qlyoung@cumulusnetworks.com>

tools: frr-reload.py needs to treat "mpls" as a single line context

Signed-off-by: Daniel Walton <dwalton@cumulusnetworks.com>
Reviewed-by: Donald Sharp <sharpd@cumulusnetworks.com>

Merge pull request #230 from qlyoung/remove-quagga_clkid

lib: remove unused quagga_clkid

Merge pull request #224 from donaldsharp/pim_patches_feb

Pim patches feb

Merge pull request #231 from qlyoung/monotime-retab

lib: monotime.h tabs -> spaces

Merge pull request #234 from donaldsharp/v6_multipath_rework1

V6 multipath rework1

Merge pull request #235 from donaldsharp/coverity1

Coverity

bgpd: Cleanup grabbag of coverity scan issues found

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

isisd: Clean up leaked memory

When in error path clean up the route table
created.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>