Mitesh Kanjariya [Mon, 28 Aug 2017 23:52:29 +0000 (16:52 -0700)]
lib: new APIs for get/set system hostname/domainname
1. Change hostname_get to cmd_hostname_get
2. Change domainname_get to cmd_domainname_get
3. New API to set domainname
3. Provide a CLI command to set domainname
Mitesh Kanjariya [Sun, 20 Aug 2017 00:28:58 +0000 (17:28 -0700)]
zebra: lookup link by ifindex
Frr has an assumption that when interface A links to B,
we already know about B. But that might be true always.
It is probably purely depends on the configuration
and how the interfaces are hashed in Kernel.
FRR seems to sometimes get "A is linked to B" before it knows about B,
in that case, the linkage between the data structure for A & B won't be proper.
Mitesh Kanjariya [Fri, 18 Aug 2017 00:39:20 +0000 (17:39 -0700)]
zebra: Revert the fix for zebra: Refresh remote neighbors when they go stale
NUD_STALE flag is causing a build breakage,
we might have to define it somewhere in frr.
Reverting the fix for now untill we decide how to handle it correctly.
vivek [Mon, 14 Aug 2017 04:52:04 +0000 (21:52 -0700)]
zebra: Fix MAC change handling for a neighbor
When the MAC changes for a local neighbor, ensure that the neighbor data
structure as well as the link between the neighbor and MAC data structures
is updated correctly.
vivek [Thu, 10 Aug 2017 15:43:06 +0000 (08:43 -0700)]
bgpd: Register for label only the default instance
Ensure that the registration for the "in" label for a unicast prefix
is done only in the default instance. The zebra label manager as well
as other code in BGP only has support for assigning labels in the
default instance.
vivek [Thu, 10 Aug 2017 00:32:19 +0000 (17:32 -0700)]
bgpd: Cleanup NHT state when underlying VRF goes down
When the underlying VRF is deleted, ensure that state for the
next hops that BGP registers with zebra for tracking purposes is
properly updated. Otherwise BGP will not re-register the next hop
when the VRF is re-created, resulting in the next hop staying
unresolved.
vivek [Thu, 10 Aug 2017 00:17:00 +0000 (17:17 -0700)]
zebra: Refresh remote neighbors when they go stale
If we get an ageout notification from the kernel for EVPN-installed
neighbors, ensure that they are readded. Otherwise, while entries in
STALE state are usable, based on other kernel parameters they can
get deleted and adding them back only at delete can have other
undesirable performance consequences.
Note: This is the current Linux kernel behavior (to ageout EVPN
installed neighbors).
zebra: Handle MACIP requests when in transient conditions
When multiple events are happening, it is possible that remote
MACIP or other requests may be received when an interface is down
or removed from a bridge. Handle this correctly.
Mitesh Kanjariya [Thu, 17 Aug 2017 06:19:58 +0000 (23:19 -0700)]
bgpd: Ignore EVPN routes from CLAG peer when VNI comes up
There are two parts to this commit:
1. create a database of self tunnel-ip for used in martian nexthop check
In a CLAG setup, the tunnel-ip (VNI UP) notification comes before the clag-anycast-ip comes up in the system.
This was causing our self next hop check to fail and we were instaling routes with martian nexthop in zebra.
We need to keep this info in a seperate database for all local tunnel-ip.
This database will be used in parallel with the self next hop database to martian nexthop checks.
2. When a local VNI comes up, update the tunnel-ip database and filter routes in the RD table if necessary
In case of EVPN we might receive routes from clag peer before the clag-anycast ip and VNI is up on the system.
We will store the routes in the RD table for later processing.
When VNI comes UP, we loop thorugh all the routes and install them in zebra if required.
However, we were missing the martian nexthop check in this code path.
From now onwards, when a VNI comes UP,
we will first update the tunnel-ip database
We then loop through all the routes in RD table and apply martian next hop filter if required.
Things not covered in this commit but are required:
This processing is needed in general when an address becomes a connected address.
We need to loop through all the routes in BGP and apply martian nexthop filter if necessary.
This will be taken care in a seperate bug
Chirag Shah [Tue, 15 Aug 2017 01:57:46 +0000 (18:57 -0700)]
ospfd: Fix ospfd crash in free_nexthop
Fix ANVL-OSPF-5.1 reported ospfd crash.
vertex_nexthop_free was added as valgrind reported potential
memory leak, but in some cases nexthop would not be available
freed.
The actual nexthop free is part of ospf_canonical_nexthops_free(),
upon trying to free, qfree checks mtype count becomes 0 and asserts.
Removing vertex_nexthop_free() from ospf_spf_flush_parents().
Signed-off-by: Chirag Shah <chirag@cumulusnetworks.com>
David Lamparter [Tue, 15 Aug 2017 09:17:36 +0000 (11:17 +0200)]
tests: fix clippy include paths / double-build
A mismatch between the make rule and the include path causes dependency
tracking to try to build the clippy.c file twice (at the same time),
which results in spurious build failures.
Fixes: #971 Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
Donald Sharp [Mon, 14 Aug 2017 23:06:39 +0000 (19:06 -0400)]
pimd: Ensure cleanup of S,G on ifchannel failure to create
There exists a path for ifchannel creation that if a S,G
fails to create and a corresponding *,G ifchannel is there,
the S,G will be deleted but we were leaving the S,G in the
*,G ifchannel sources list. Remove from the list in this case
Ticket: CM-17605 Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
Brian Rak [Mon, 14 Aug 2017 21:22:03 +0000 (17:22 -0400)]
bgpd: Check for per-peer outbound configuration, in addition to the peer-group config
When displaying the config, bgpd only checked for the existance of a peer-group prefix-list before
deciding to not display the outbound prefix-list. This commit updates the outbound prefix-list
logic to match the inbound.
Chirag Shah [Mon, 14 Aug 2017 00:41:01 +0000 (17:41 -0700)]
pimd: pim bfd vrf aware
BFD replay for all interfaces was only considering VRF_DEFAULT interface
list.
The change will walk all VRFs, interface list per VRF to register pim
neighbor with BFD.
Signed-off-by: Chirag shah <chirag@cumulusnetworks.com>
Chirag Shah [Mon, 14 Aug 2017 00:28:11 +0000 (17:28 -0700)]
pimd: Fix pim bfd config order dependecny
Upon frr restart vtysh replays pim bfd configuration
cli before pim sm command, this results in failure
of pim bfd config on the VIF as it is not pim enabled yet.
Inside pim bfd configuration, call to create PIM vif
handler if it is not created yet.
no ip pim bfd, if pim is not enabled display warning and return.
Signed-off-by: Chirag Shah <chirag@cumulusnetworks.com>
David Lamparter [Fri, 11 Aug 2017 16:54:26 +0000 (18:54 +0200)]
lib: fix prefix list trie corruption
The specific code here needs to establish an absolute order of more
specific to less specific possible matches in a prefix list. This is
indirectly checked by an assert on insertion, because the "next best"
entry is required to be consistent even when joining multiple chains
of candidates.
Unfortunately, trie_install_fn() would insert entries too far ahead in
the chain if another entry with higher sequence number was seen. This
breaks the trie and (rightfully) triggers the assertion failure on
insert.
Fixes: #937 Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
David Lamparter [Fri, 11 Aug 2017 16:54:26 +0000 (18:54 +0200)]
lib: fix prefix list trie corruption
The specific code here needs to establish an absolute order of more
specific to less specific possible matches in a prefix list. This is
indirectly checked by an assert on insertion, because the "next best"
entry is required to be consistent even when joining multiple chains
of candidates.
Unfortunately, trie_install_fn() would insert entries too far ahead in
the chain if another entry with higher sequence number was seen. This
breaks the trie and (rightfully) triggers the assertion failure on
insert.
Fixes: #937 Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
David Lamparter [Fri, 11 Aug 2017 13:43:28 +0000 (15:43 +0200)]
lib: don't close stdio when running foreground
c9c8d0d ("lib: close stdin/out/err in non-terminal case") overshot its
goal and closes stdin/stdout/stderr even when a daemon is running in
foreground. That means stdout logging & exit memory reporting are both
broken.
Reported-by: Lou Berger <lberger@labn.net> Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
David Lamparter [Fri, 11 Aug 2017 12:25:56 +0000 (14:25 +0200)]
bgpd: don't sidestep vty code
afi_header_vty_out is sidestepping the vty code, writing straight to the
output (either stdout or the obuf), which results in newline translation
not being performed.
Easiest fix is replacing it with a macro. Longer-term, I have some old
code to add "prefaces" to the vty output, planning to dig that up.
Fixes: #949 ("bgpd show running doesn't show new lines") Reported-by: Lou Berger <lberger@labn.net> Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
Donald Sharp [Thu, 10 Aug 2017 20:13:45 +0000 (16:13 -0400)]
pimd: Remove mroute when RPF is gone
This does three things:
1) When we get a RPF_FAILURE, remove the mroute associated
with it.
-> This way when the RPF comes back we can just add the
mroute in as part of the normal scanning process.
2) When we do a ecmp_nexthop_search return 1 when we found
something we can use.
3) Ignore output from pim_update_rp_nh
-> When we do a ecmp_nexthop_search ignore the return
code and do not attempt to gather it up to return
to the calling function. It is just ignored
and we were not taking into account the what of
multiple RP's we were looking at.
Ticket: CM-17218 Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>
Chirag Shah [Fri, 23 Jun 2017 00:19:00 +0000 (17:19 -0700)]
ospfd: Fix conflict router-id, memory leaks
Remove assert in path of router_id_update:
Upon configuring same router-id as neighbor's
assert would cause a crash. Log a warning message
and neighborship would not come up.
Address memory leaks
Signed-off-by: Chirag Shah <chirag@cumulusnetworks.com>
Quentin Young [Thu, 10 Aug 2017 21:01:59 +0000 (17:01 -0400)]
bgpd: don't use DECLVAR without qobj pointer
Since we don't actually have a BGP context until we enter into the
`router bgp` node, debugging commands which use DECLVAR before that
happens will fail. This is just `no debug bgp` and `no debug bgp
updates`. Instead just iterate over all bgp instances and do the deed.
Signed-off-by: Quentin Young <qlyoung@cumulusnetworks.com>
Andrew Lunn [Wed, 9 Aug 2017 18:46:52 +0000 (13:46 -0500)]
eigrp: Don't dereference NULL timer in a Neighbour
Current, a eigrp_neighbor only has a t_holddown timer when in state
EIGRP_NEIGHBOR_PENDING and EIGRP_NEIGHBOR_UP. In state
EIGRP_NEIGHBOR_DOWN it could be a NULL pointer. Don't dereference the
timer when dumping the neighbour table without first checking it
exists. If it does not exist, display - instead of the remaining time.
Quentin Young [Thu, 10 Aug 2017 16:31:47 +0000 (12:31 -0400)]
lib, bgpd: fix `set ip next-hop peer-address`
This bgp-specific command had its positive form defined only in bgpd and
its negative form defined only in lib, which broke the whole rule for
other daemons.
Signed-off-by: Quentin Young <qlyoung@cumulusnetworks.com>
David Lamparter [Wed, 9 Aug 2017 18:30:34 +0000 (20:30 +0200)]
lib: some final MAC access-list fixes
- couldn't load back written configs because it was trying to parse
"any" as MAC address
- don't need special-casing in filter_match_zebra(), exact is going to
be 0 for AF_ETHERNET anyway
- some vty formatting was slightly different
- is_zero_mac now static to prefix.c
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
projects / mirror / frr.git / log