tests: fix out-of-tree build

Fix pytest with $(top_srcdir) != "."

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

Merge pull request #194 from opensourcerouting/coverity-lib-2

more lib/ coverity fixes

lib: rework vty_use_backup_config()

Like config_write(), this should use rename(), even though atomicity is
not a real issue here.

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

lib: fix remaining coverity issues

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

Merge pull request #192 from donaldsharp/snmp1

debian: Fix to not auto-install snmp MIB

Merge remote-tracking branch 'origin/stable/2.0'

Merge pull request #191 from opensourcerouting/ospfclient-incl

[stable] ospfclient: #include sources

debian: Fix to not auto-install snmp MIB

Only install the snmp MIB for zebra if we
are building the debian package with SNMP.

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

ospfclient: #include sources

Pulling in source files from another directory breaks automake's
distclean target, and there seems to be no good fix for this...
(particularly since we need -fPIC build for a DSO here, while ospfd
compiles for an executable...)

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

Merge remote-tracking branch 'origin/stable/2.0'

Merge pull request #189 from opensourcerouting/config-write-sync

lib: use fsync() for config writes, plug fd leak

Merge pull request #164 from opensourcerouting/pytest2

Rework test runners to use pytest

Merge pull request #190 from opensourcerouting/config-write-sync-stable

[stable] lib: use fsync() for config writes, plug fd leak

lib: use fsync() for config writes, plug fd leak

sync() has a HUGE impact on systems that perform actual I/O, i.e. real
servers...

Also, we were leaking a fd on each config write ever since
c5e69a0 "lib/vty: add separate output fd support to VTYs"
(by myself :( ...)

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

Merge branch 'frr/pull/186' (pimd register length)

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

lib: use fsync() for config writes, plug fd leak

sync() has a HUGE impact on systems that perform actual I/O, i.e. real
servers...

Also, we were leaking a fd on each config write ever since
c5e69a0 "lib/vty: add separate output fd support to VTYs"
(by myself :( ...)

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

Merge pull request #185 from opensourcerouting/coverity-lib-1

Coverity foo

tests: add pytest testrunners

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

tests: reorganize tests hierarchically

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

tests: remove dejagnu

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

lib: csv: plug memleaks

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

lib: fix some coverity NULL check warnings

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

lib: remove unused str2in6_addr()

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

lib: remove unused cmd_concat_strvec()

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

Merge pull request #182 from opensourcerouting/ldpd-coverity-scan

ldpd: fix issues reported by coverity scan

Merge pull request #181 from opensourcerouting/zebra-static-uninstall-mpls

zebra: fix uninstallation of LDP labeled static routes

Merge branch 'frr/pull/183' ("Coverity")

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

pimd: Send proper length register packets

PIM Register packets are 20 bytes too long.

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

Merge branch 'master' of https://github.com/freerangerouting/frr into coverity

bgpd: Fix miss-indented code.

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

Merge branch 'frr/pull/180' ("Warnings")

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

zebra: Don't leak socket with failure condition

Don't leak a socket when we are unable to set it
as non-blocking and warn the user as appropriate.

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

ldpd: fix issues reported by coverity scan

Fix two small memleaks in the CLI code and check the return values of
getsockname() and getpeername().

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

zebra: fix uninstallation of LDP labeled static routes

Problem:
1 - Add a static route: "ip route 10.0.0.0/24 172.16.1.1";
2 - Receive an LDP mapping for 10.0.0.0/24 from 172.16.1.1;
3 - Remove the static route: "no ip route 10.0.0.0/24 172.16.1.1".
4 - Static route is removed but not uninstalled from the kernel.

What happens is that, on static_uninstall_route(), we can't find the
route we want to uninstall because it has an LDP label where the original
static route doesn't have any MPLS label.

To fix this, we can just stop calling static_nexthop_label_same() and
remove this function. It's impossible to have two routes for the same
prefix with the same distance and same nexthop address. This means that
we can lookup the correct route to uninstall without having to check
its labels.

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

Merge pull request #179 from donaldsharp/more_quagga_fixups

*: Found some instances of Quagga.conf

pimd: Cleanup dest buffer to small

Cleanup the in_ifname and out_ifname buffers
to be large enough to hold the actual interface
names.

Additionally move the common variables to be defined
once, instead of inside of multiple for loops

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

bgpd: Fix usage of uninitialized dn_flag[]

dn_flag was tested inside of a 'use_json' block
further down in the function.  Move initialization
of the value to outside of the !use_json if statement

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

bgpd: Fix passage of uninitialized value to skiplist_next

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

bgpd: Fix uninitialized data

Calling zapi_ipv[4|6]_route assumes we
set the instance appropriately.

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

tests: Fix warning in compile

Fix bgp_mp_attr_test.c warning in compile
on fedora 25 and clang.

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

lib: Fix warning -> error build issue

When compiling ./buildtest.sh the gcc compiler
on fedora 25 complains about indentation

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

*: Found some instances of Quagga.conf

Cleanup Quagga.conf -> Frr.conf

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

Merge remote-tracking branch 'origin/stable/2.0'

Merge pull request #176 from opensourcerouting/mtype-underflow-backtrace

lib: print backtrace on MTYPE count underflow

Merge pull request #178 from opensourcerouting/ospf_lib_private

make libfrrospf private

Merge pull request #174 from LabNConsulting/working/master/patch-set/issue-172

master: Fix encap attribute on wire, protect against broken object.

Merge pull request #173 from LabNConsulting/working/2.0/patch-set/issue-172

2.0: Fix encap attribute on wire, protect against broken object.

ospfd: make libfrrospf a private non-dynamic lib

There's no point in installing libospf anymore, now that
libfrrospfclient doesn't link it.

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

ospfd: isolate ospfapiclient lib from ospf lib

The OSPF API-client library doesn't really use anything from libospf
except some small dumping tools.  Isolate these into a separate file
and detangle the ospfapiclient library.

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

Merge pull request #161 from donaldsharp/stableMcLibrary

Stable mc library

bgpd: Allow views to 'pretend' resolve nexthops

Views are supposed to be independent tables that have no connection to
routing tables.  Since View's are 'independent' there is no way to do
'real' nexthop resolution since connected routes and interfaces are
associated with a particular table (or really vrf).  So when we have a
bgp instance assume that nexthops specified are actually valid, since we
are propagating what our neighbors are telling us.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
[cherry-picked from master 6fa1ab2eecb39b35256739f7083196d6f556f639]
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>

Merge branch 'frr/pull/157' ("Fixinator")

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

lib: print backtrace on MTYPE count underflow

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

Merge branch 'frr/pull/165' ("Pim vrf")

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

Merge branch 'frr/pull/166' ("bgpd: fix array oob ...")

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

Merge branch 'frr/pull/152' ("Lib fixes")

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

bgpd rfapi: advertise encap attribute when TT is valid and not MPLS.
     Ensure tunnel type set based on rx'ed
     Ignore bad/mpls encap types.

Signed-off-by: Lou Berger <lberger@labn.net>

Merge pull request #171 from donaldsharp/double_free

lib: Don't use malloc!

Merge pull request #169 from opensourcerouting/ldpd-label-alloc-rework

ldpd: update local labels when necessary

bgpd rfapi: advertise encap attribute when TT is valid and not MPLS.
     Ensure tunnel type set based on rx'ed
     Ignore bad/mpls encap types.

Signed-off-by: Lou Berger <lberger@labn.net>

zebra: fix installation of LDP labels on static routes

If the ifindex of a nexthop is not zero, we can't assume that its type is
NEXTHOP_TYPE_IPV4_IFINDEX or NEXTHOP_TYPE_IPV6_IFINDEX. Nexthops of type
NEXTHOP_TYPE_IPV[46] can have their ifindex set by the nexthop_active()
function.

With that said, we need to me more flexible when comparing nexthops
on mpls_ftn_update() to make sure that we'll always find the correct
route/nexthop to update.

Regression introduced by commit 88d88a ("zebra/ldpd: allow MPLS ECMP on
unnumbered interfaces") and found by ANVL.

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

lib: Fixup CR comment

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

lib: Don't use malloc!

thread.c was using malloc.  This was found by the
assert added to memory.c.

Fixes issues #170

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

ldpd: update local labels when necessary

ldpd allocates null labels for directly connected routes. If a connected
route is removed (interface goes down) and an IGP learned route takes its
place in the RIB, ldpd must update the local label of the associated FEC
entry with a non-null label. The same applies for the other way around
(an interface goes up and a connected route is selected in favour of an
IGP route). Labels should be dynamic and change when necessary.

Additionally, this patch fixes the processing of route delete messages
from zebra. Route delete messages don't contain any nexthop, meaning that
whenever we receive such messages we must delete all nexthop previously
received.

Based on a patch from Bingen Eguzkitza <bingen@voltanet.io>.
Signed-off-by: Renato Westphal <renato@opensourcerouting.org>

ldpd: fix silly bug introduced by a recent commit

This call to fatalx() was making ldpd crash when a pseudowire was
configured.

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

zebra: fix indentation problems introduced by wrong conflict resolution

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

Merge pull request #167 from qlyoung/vtysh-subs-stable

vtysh: Quagga.conf -> Frr.conf

vtysh: Quagga.conf -> Frr.conf

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

bgpd: fix array oob for `show ip bgp`

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

pimd: Refactor pim_sock_open to just need the ifp pointer

In all cases pim_sock_open was called, we just passed
in the pim_ifp->primary_address, which is accessible
from the interface pointer.  So just pass that in.

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

pimd: Allow a interface in a vrf to be configured.

This fixes the issue a crash when we have configured an interface
inside of a vrf, and apply pim commands to it.

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

Merge pull request #158 from LabNConsulting/working/2.0/patch-set/rfapi+vty

2.0: Fix a couple of merge related and rfapi/show vpn issues

Revert "bgpd: Put back original behavior for some show bgp commands"

This reverts commit 713de0685fc0ef882c97a6cb4bd149620f4dbdf4.

During Discussions with Lou, it was decided that his code
changes would handle this part, and my code changes would
fix up vtysh and bgp to be in sync again.

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

bgpd, vtysh: Fix vtysh able to handle some safi modes for bgp

This adds the ability for vtysh to handle v4 and v6
safi modes besides unicast and multicast.

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

bgp: remove HAVE_IPV6 from file (again!)

Signed-off-by: Lou Berger <lberger@labn.net>

bgpd: add missing json strings

Signed-off-by: Lou Berger <lberger@labn.net>

Merge pull request #159 from LabNConsulting/working/master/patch-set/rfapi+vty

Master: Fix a couple of issues -  rfapi/show vpn issues

*: Convert libospf -> libfrrospf

Convert the libospfapiclient -> libfrrospfapiclient
and libospf -> libfrrospf to prevent namespace
collision with Quagga.

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

*: Convert libfpm -> libfrrfpm

Convert the libfpm to libfrrfpm to prevent namespace
collision with Quagga.

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

ospf6d: Fix compile warnings

These compile issues were found by running ./buildtest.sh
on a fedora 25 box.

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

ospfd: Fix compile warnings

These compile issues were found w/ gcc and running
./buildtest.sh on a fedora 25 box.

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

ripd: Fix compile warning

This issue was found compiling with ./buildtest.sh
on fedora 25

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

*: Convert libzebra -> libfrr

The library libzebra that is installed with FRR will
conflict with Quagga.  So let's rename it to libfrr.

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

bgpd: get rid of 'unknown' in show safi
      Merge 2.0 change, adapted to master

Signed-off-by: Lou Berger <lberger@labn.net>

bgpd rfapi: fix TT handling broken on Sat Jan 28 18:57:28 2017 -0500

Signed-off-by: Lou Berger <lberger@labn.net>

bgpd: get rid of 'unknown' in show safi (broken in earlier merge)

Signed-off-by: Lou Berger <lberger@labn.net>

bgpd: cleanup/restore show ip|ipv6 bgp and show bgp ipv4 summary related commands

Signed-off-by: Lou Berger <lberger@labn.net>

bgpd rfapi: fix TT handling broken on Sat Jan 28 18:57:28 2017 -0500

Signed-off-by: Lou Berger <lberger@labn.net>

bgpd: Put back original behavior for some show bgp commands

The behavior of some show commands was switched from
showing the unicast data to all safi data.  This
is a change in behavior and needs to be put
back for 2.0

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

bgpd: Allow vtysh to work for address-family

extract.pl.in requires the lines that are ignored
to be consisten with what is in the *.c file
that we are ignoring from, else we get ambiguous
commands.

This allows vtysh to enter address-family sub mode
in bgp again.

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

Merge remote-tracking branch 'origin/stable/2.0'

Merge pull request #135 from LabNConsulting/working/master/patch-set/misc+vrf

Master: misc vnc/rfapi and VRF

Merge pull request #154 from LabNConsulting/working/master/patch-set/rr-part2

Mater: complete Issue#91 fix (regression now clean)

Merge pull request #153 from LabNConsulting/working/2.0/patch-set/rr-part2

2.0: complete Issue#91 fix (regression now clean)

Merge pull request #143 from LabNConsulting/working/2.0/patch/mpls-tt

2.0: bgp rfapi: fix resolution / lookup failure when using MPLS TT

Merge pull request #151 from opensourcerouting/ldpd-filtering

ldpd: introduce advanced filtering capabilities

bgpd rfapi: small debug log enhancements

Signed-off-by: G. Paul Ziemba <paulz@labn.net>

bgpd rfapi: RR replated (Issue #91) treat filtered update
     as implicit withdraw for rfapi

Signed-off-by: G. Paul Ziemba <paulz@labn.net>

bgpd rfapi: RR replated (Issue #91) treat filtered update
     as implicit withdraw for rfapi

Signed-off-by: G. Paul Ziemba <paulz@labn.net>

pimd: Fix fd leak

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