The prefix hash table in BGP used the same hash function as the rtable.
When a batch of routes are exported during feed/flush to the BGP, they
all have similar hash values, so they are all crowded in a few slots in
the BGP prefix table (which is much smaller - around the size of the
batch - and uses higher bits from hash values), making it much slower due
to excessive collisions. Use a different hash function to avoid this.
Also, increase the batch size to fill 4k BGP packets and increase minimum
BGP bucket and prefix hash sizes to avoid back and forth resizing during
flushes.
This leads to order of magnitude faster flushes (on my test data).
When shutting down a Babel instance we send a wildcard retraction to make
sure all peers can quickly switch to other route origins. Add another small
optimisation borrowed from babeld: sending a Hello message (along with the
retraction) with a very low interval.
This will cause neighbours to modify their expiry timers for the node's
state to quickly time it out, thus conserving resources in the network.
The interface pointer was improperly converted to u32 and back. Fixing
this by explicitly allocating an adata structure for it. It's not so
memory efficient, we'll optimize this later.
Add BFD protocol option 'strict bind' to use separate listening socket
for each BFD interface bound to its address instead of using shared
listening sockets.
It is too cryptic to flush tmp_linpool in these cases and we don't want
anybody in the future to break this code by adding an allocation
somewhere which should persist over that flush.
Saving and restoring linpool state is safer.
A recent change in Babel causes ifaces to disappear after
reconfiguration. The patch fixes that.
Thanks to Johannes Kimmel for an insightful bugreport.
Implement flowspec validation procedure as described in RFC 8955 sec. 6
and RFC 9117. The Validation procedure enforces that only routers in the
forwarding path for a network can originate flowspec rules for that
network.
The patch adds new mechanism for tracking inter-table dependencies, which
is necessary as the flowspec validation depends on IP routes, and flowspec
rules must be revalidated when best IP routes change.
The validation procedure is disabled by default and requires that
relevant IP table uses trie, as it uses interval queries for subnets.
Attach a prefix trie to IP/VPN/ROA tables. Use it for net_route() and
net_roa_check(). This leads to 3-5x speedups for IPv4 and 5-10x
speedup for IPv6 of these calls.
TODO:
- Rebuild the trie during rt_prune_table()
- Better way to avoid trie_add_prefix() in net_get() for existing tables
- Make it configurable (?)
One of previous commits added error logging of invalid routes. This
also inadvertently caused error logging of route loops, which should
be ignored silently. Fix that.
Most error messages in attribute processing are in rx/decode step and
these use L_REMOTE log class. But there are few that are in tx/export
step and these should use L_ERR log class.
Use tx-specific macro (REJECT()) in tx/export code and rename field
err_withdraw to err_reject in struct bgp_export_state to ensure that
appropriate error reporting macros are called in proper contexts.
RFC 6810 and RFC 8210 specify that the "Max Length" value MUST NOT be
less than the Prefix Length element (underflow). On the other side,
overflow of the Max Length element also is possible, it being an 8-bit
unsigned integer allows for values larger than 32 or 128. This also
implicitly ensures there is no overflow of "Length" value.
When a PDU is received where the Max Length field is corrputed, the RTR
client (BIRD) should immediately terminate the session, flush all data
learned from that cache, and log an error for the operator.
Minor changes done by commiter.
Some cleanups and bugfixes to the previous patch, including:
- Fix rate limiting in index mismatch check
- Fix missing BABEL_AUTH_INDEX_LEN in auth_tx_overhead computation
- Fix missing auth_tx_overhead recalculation during reconfiguration
- Fix pseudoheader construction in babel_auth_sign() (sport vs fport)
- Fix typecasts for ptrdiffs in log messages
- Make auth log messages similar to corresponding RIP/OSPF ones
- Change auth log messages for events that happen during regular
operation to debug messages
- Switch meaning of babel_auth_check*() functions for consistency
with corresponding RIP/OSPF ones
- Remove requirement for min/max key length, only those required by
given MAC code are enforced
This implements support for MAC authentication in the Babel protocol, as
specified by RFC 8967. The implementation seeks to follow the RFC as close
as possible, with the only deliberate deviation being the addition of
support for all the HMAC algorithms already supported by Bird, as well as
the Blake2b variant of the Blake algorithm.
For description of applicability, assumptions and security properties,
see RFC 8967 sections 1.1 and 1.2.
In preparation for adding authentication checks, refactor the TLV
walking code so it can be reused for a separate pass of the packet
for authentication checks.
Routes from downed protocols stay in rtable (until next rtable prune
cycle ends) and may be even exported to another protocol. In BGP case,
source BGP protocol is examined, although dynamic parts (including
neighbor entries) are already freed. That may lead to crash under some
race conditions. Ensure that freed neighbor entry is not accessed to
avoid this issue.
When an interface disappears, all the neighbors are freed as well. Seqno
requests were anyway not decoupled from them, leading to strange
segfaults. This fix adds a proper seqno request list inside neighbors to
make sure that no pointer to neighbor is kept after free.
The babel protocol code checks whether iface supports multicast, and
whether it has a link-local address assigned. However, it doesn not give
any feedback if any of those checks fail, it just silently ignores the
interface. Fix this by explicitly logging when multicast check fails.
Based on patch from Toke Høiland-Jørgensen, thanks!
Ifaces with host address (/32) were forced to be stubby, but now they
can be used as PtP or PtMP. For these ifaces we need to:
- Do not force stub mode
- Accept packets from any IP as local
- Accept any configured neighbor as local
- Detect ifaces properly as unnumbered
- Use ONLINK flag for nexthops
As specified in RFC 2328 8.1: "On physical point-to-point networks,
the IP destination is always set to the address AllSPFRouters."
Note that this likely break setups with multiple neighbors on a network
configured as PtP, which worked before. These should be configured as
PtMP.
Thanks to Senthil Kumar Nagappan for the original patch and to Joakim
Tjernlund for suggestions.
The flag makes sense just in external representation. It is reset during
BGP export, but keeping it internally broke MRT dumps for short attributes
that used it anyways.
Thanks to Simon Marsh for the bugreport and the patch.
BGP statistics code was preliminary and i wanted to replace it by
separate 'show X stats' command. The patch hides the preliminary
output in 'show protocols all' so it is not part of the released
version.
In OSPFv3, only Hello and DBDes packets contain flags specifying whether
RFC 7166 authentication trailer is used. Other packets are processed
based on stored authentication state in neighbor structure. Update this
state with each received Hello to handle authentication change from
reconfigurations.
Thanks to Joakim Tjernlund and Kenth Eriksson for the bugreport.
This is an implementation of draft-walton-bgp-hostname-capability-02.
It is implemented since quite some time for FRR and in datacenter, this
gives a nice output to avoid using IP addresses.
It is disabled by default. The hostname is retrieved from uname(2) and
can be overriden with "hostname" option. The domain name is never set
nor displayed.
Minor changes by committer.
Flag signalling that MP-BGP mode should be used got reset after first
batch of routes, so remaining routes were processed without that, leading
to missing MP_REACH_NLRI attribute.
Thanks to Piotr Wydrych for the bugreport.
Add fake MP_REACH_NLRI attribute with BGP next hop when encoding MRT
table dumps for IPv6 routes. That is necessary to encode next hop as
NEXT_HOP attribute is not used for MP-BGP.
Thanks to Santiago Aggio for the bugreport.
Direct BFD sessions needs to be dispatched not only by IP addresses, but
also by interfaces, in order to avoid collisions between neighbors with
the same IPv6 link-local addresses.
Extend BFD session hash_ip key by interface index to handle that. Use 0
for multihop sessions.
Thanks to Sebastian Hahn for the original patch.
It was mixed-up if hostname is IPv6 address, and reporting separate
values (like port) on separate lines fits better into key-value style
of 'show protocols all' output. Also, the patch simplifies transport
identification formatting (although it is unused now).
Thanks to Alarig Le Lay for the suggestion.
The option is not implemented since transition to 2.0 and no plan to add it.
Also remove some deprecated RTS_* valus from documentation.
Thanks to Sébastien Parisot for notification.
The patch add support for per-channel debug flags, currently just
'states', 'routes', and 'filters'. Flag 'states' is used for channel
state changes, remaining two for routes passed through the channel.
The per-protocol debug flags 'routes'/'filters' still enable reporting
of routes for all channels, to keep existing behavior.
The patch causes minor changes in some log messages.
When config structures are copied due to template application,
we need to reset list node structure before calling add_tail().
Thanks to Mikael Magnusson for patches.
The babel protocol code was initialising objects returned from the slab
allocator by assigning to each of the struct members individually, but
wasn't touching the NODE member while doing so. This leads to warnings on
debug builds since commit:
baac700906 ("List expensive check.")
To fix this, introduce an sl_allocz() variant of the slab allocator which
will zero out the memory before returning it, and switch all the babel call
sites to use this version. The overhead for doing this should be negligible
for small objects, and in the case of babel, the largest object being
allocated was being zeroed anyway, so we can drop the memset in
babel_read_tlv().
Add support for proper handling of multiple routes with the same network
to the static protocol. Routes are distinguished by internal index, which
is assigned automatically (sequentially for routes within each network).
Having different route preference or igp_metric attribute is optional.
When a new link-LSA is originated, we need to notify intra-area-prefix-LSA
handling, like when a new link-LSA is received. Otherwise a new network
prefix added to a DR is not propagated immediately.
Thanks to Bala Sajja for the bugreport.
Merge multiple BFD option blocks in BGP configs instead of using the last
one. That is necessary for proper handling of templates when BFD options
are used both in a BGP template and in a BGP protocol derived from that
template.
BFD session options are configured per interface in BFD protocol. This
patch allows to specify them also per-request in protocols requesting
sessions (currently limited to BGP).
Add 'ignore max length' option to RPKI protocol, which ignores received
max length in ROA records and instead uses max value (32 or 128). This
may be useful for implementing loose RPKI check for blackholes.
Sometimes multicast OSPF packet is received when neighbor adjacency is
not established. Such packet should be ignored earlier in packet
processing as otherwise it causes strange error messages when OSPFv3
authentication is enabled.
Most commands like 'show ospf neighbors' fail when protocol is not
specified and there are multiple instances of given protocol type.
This is annoying in BIRD 2, as many protocols have IPv4 and IPv6
instances. The patch changes that by showing output from all protocol
instances of appropriate type.
Note that the patch also removes terminating cli_msg() call from these
commands and moves it to the common iterating code.
Compare the new timing parameters with the old configuration, not with
the temporary state of the current connection.
The timing values in struct rpki_cache is updated by a version 1 End Of
Data PDU, unless this behavior is suppressed by the configuration
explicitly by the "keep" keyword. Consequently, every reconfiguration
of BIRD triggers a reconnection even if it is not necessary.
If the next hop of a route is not a reachable address, the route should be
installed as onlink. This enables a configuration common in mesh networks
where the mesh interface is assigned a /32 and babel handles the routing by
installing onlink routes.
Thanks to Toke Hoiland-Jorgensen for the patch.
This issue has a long history. In 2012, we changed data field for
unnumbered PtP links from iface id (specified by RFC) to IP address based
on reports of bugs in Quagga that required it, and we used out-of-band
information to distinquish unnumberred PtPs with the same local IP
address.
Then with OSPF graceful restart implementation, we found that we can no
longer use out-of-band information, and we need to use only LSAdb info
for routing table calculation, but i forgot to finish handling of this
case, so multiple unnumbered PtPs with the same local IP addresses were
broken.
Considering that even recent Mikrotik RouterOS has broken next hop
calculation that depends on IP address in PtP link data field, we
cannot just switch back to the iface id for unnumbered PtP links.
The patch makes two changes: First, it goes back to use out-of-band
(position) info for distinguishing local interfaces in SPF when graceful
restart is not enabled, while still uses LSAdb-only approach for SPF
calculation when graceful restart is enabled.
Second, it adds OSPF interface option 'ptp address', which controls
whether IP address or iface id is used in data field. It is enabled
by default except for unnumbered PtP links with enabled graceful
restart.
Thanks to Kenth Eriksson for the bugreport and Joakim Tjernlund for
suggestions.
There are three common ways how to encode IPv6 link-local-only next hops:
(:: ll), (ll), and (ll ll). We use the first one but we should accept all
three. The patch fixes handling of the last one.
Thanks to Sebastian Hahn for the bugreport.
The RFC 5575 does not explicitly reject flowspec rules without dst part,
it just requires dst part in validation procedure for feasibility, which
we do not implement anyway. Thus flow without dst prefix is syntactically
valid, but unfeasible (if feasibilty testing is done).
Thanks to Alex D. for the bugreport.
When dynamic BGP with remote range is configured, MD5SIG needs to use
newer socket option (TCP_MD5SIG_EXT) to specify remote addres range for
listening socket.
Thanks to Adam Kułagowski for the suggestion.
Recent changes in neighbor code caused RIP to access neighbor field which
is NULL during interface/neighbor removal and caused crash when debug
messages are enabled. Use correct field to get iface from neighbor.
During NLRI parsing of IPv6 Flowspec, dst prefix was not properly
extracted from NLRI, therefore a received flow was stored in a different
position in flowspec routing table, and was not reachable by command
'show route <flow>'.
Add proper prefix part accessors to flowspec code and use them from BGP
NLRI parsing code.
Thanks to Alex D. for the bugreport.
This is optional check described in RFC 4271. Although this can be also
done by filters, it is widely implemented option in BGP implementations.
Thanks to Eugene Bogomazov for the original patch.
Transitive extended communities should be removed on external sessions,
the old code them in all cases.
Thanks to Jean-Daniel Pauget for the original patch.
Change of some options requires route refresh, but when import table is
active, channel reload is done from it instead of doing full route
refresh. So in this case we request it internally.
The bfd_reconfigure_neighbors() returned after first reconfigured
neighbor instead of continuing with the next one.
Thanks to Winston Chen for the bugreport and a patch.
There is an improper check for valid message size, which may lead to
stack overflow and buffer leaks to log when a large message is received.
Thanks to Daniel McCarney for bugreport and analysis.
Instead of having large stack buffer for max amount of AFI/SAFI pairs.
The old code is not correct w.r.t. extendeded option length, as more
AFI/SAFI pairs may fit into the capability option.
The patch implements optional internal export table to a channel and
hooks it to BGP so it can be used as Adj-RIB-Out. When enabled, all
exported (post-filtered) routes are stored there. An export table can be
examined using e.g. 'show route export table bgp1.ipv4'.
Several BGP channel options (including 'next hop self') could be
reconfigured without session reset, with just route refeed/refresh.
The patch improves reconfiguration code to do it that way.
The 'deterministic med' option is implemented by suppressing other than
best-in-group routes (grouped by ASN) from best route selection. This
interferes with 'merge paths' as supressed routes are no longer mergable
with best route. This is fixed by suppressing only those routes that are
not mergable with best-in-group route.
Protocol can have specified VRF, in such case it is restricted to a set
of ifaces associated with the VRF, otherwise it can use all interfaces.
The patch allows to specify VRF as 'default', in which case it is
restricted to a set of iface not associated with any VRF.
Per RFC 3101, N-bit signalling NSSA support should be used only in Hello
packets, not in DBDES packets. BIRD since 2.0.4 verifies N-bit in
neighbor structure, which is learned from DBDES packets, therefore
NSSA-LSAs are not propagated to proper implementations of RFC 3101.
This patch fixes that. Both removing the check and removing N-bit from
DBDES packet. This will fix compatibility issues with proper
implementations, but causes compatibility issues with BIRD 2.0.4.
We need to flush learned external LSAs a bit later than other LSAs (after
first feed after end of the graceful restart) to avoid flap of external
routes.
If BGP has too many data to send and BIRD is slower than the link, TX is
always possible until all data is sent. This patch limits maximum number
of generated BGP messages in one iteration of TX hook.
Implement OSPFv2 (RFC 3623) and OSPFv3 (RFC 5187) graceful restart,
for both restarting and helper sides. Graceful restart is initiated
by 'graceful down' command.
When 'graceful down' command is entered, protocols are shut down
with regard to graceful restart. Namely Kernel protocol does
not remove routes and BGP protocol does not send notification,
just closes the connection.
Useful for implementation of agents implementing the SNMP-BGP MIB, which
requires the local AS of a session to be specified.
Thanks to Jan-Philipp Litza for the patch.
Support for dynamically spawning BGP protocols for incoming connections.
Use 'neighbor range' to specify range of valid neighbor addresses, then
incoming connections from these addresses spawn new BGP instances.
When BGP connection is opened, it may happen that rx hook (with remote
OPEN) is called before tx hook (for local OPEN). Therefore, we need to do
internal changes (like setting local_caps) synchronously with OPENSENT
transition and we need to ensure that OPEN is sent before KEEPALIVE.
Allow to specify just 'internal' or 'external' for remote neighbor
instead of specific ASN. In the second case that means BGP peers with
any non-local ASNs are accepted.
The temporary atttributes are no longer removed by ea_do_prune(), but
they are undefined by store_tmp_attrs() protocol hooks. This fixes
several bugs where temporary attributes were removed when they should
not or not removed when they should be. The flag EAF_TEMP is no longer
needed and was removed.
Update all protocol make_tmp_attrs() / store_tmp_attrs() hooks to use
helper functions and to handle unset attributes properly.
Also fix some related bugs like improper handling of empty eattr list.
Keep track of whether OSPF tmpattrs are actually defined for given route
(using flags in rte->pflags). That makes them behave more like real
eattrs so a protocol can define just a subset of them or they can be
undefined by filters.
Do not set ospf_metric2 for other than type 2 external OSPF routes and do
not set ospf_tag for non-external OSPF routes. That also fixes a bug
where internal/inter-area route propagated from one OSPF instance to
another is initiated with infinity ospf_metric2.
Thanks to Yaroslav Dronskii for the bugreport.
This is a major change of how the filters are interpreted. If everything
works how it should, it should not affect you unless you are hacking the
filters themselves.
Anyway, this change should make a huge improvement in the filter performance
as previous benchmarks showed that our major problem lies in the
recursion itself.
There are also some changes in nest and protocols, related mostly to
spreading const declarations throughout the whole BIRD and also to
refactored dynamic attribute definitions. The need of these came up
during the whole work and it is too difficult to split out these
not-so-related changes.
When area is reconfigured to a different type, we need to flush LSAs as
they may not be valid (e.g. NSSA-LSA for non-NSSA area). Also, when we
have have just one OSPF area and that changes type, we could restart OSPF
as there is no state to keep anyway. That solves issue with different
handling of external routes exported to OSPF based of main area type.
External LSAs originated by OSPF routers with VPN-PE behavior enabled are
marked by DN flag and they are ignored by other OSPF routers with VPN-PE
enabled.
Direct acknowledgements should be send as unicast to a corresponding
neighbor. Only delayed acks should be send as multicast to all/designated
routers.
Master may free last DBDES packet immediately. Slave must wait dead
interval before freeing last DBDES packet and then reject duplicate
DBDES packets with SeqNumberMismatch.
Since v2 we have multiple listening BGP sockets, and each BGP protocol
has associated one of them. Use listening socket that accepted the
incoming connection as a key in the dispatch process so only BGP
protocols assocaited with that listening socket can be selected.
This is necesary for proper dispatch when VRFs are used.
This protocol is highly experimental and nobody should use it in
production. Anyway it may help you getting some insight into what eats
so much time in filter processing.
In some circumstances (old LSA flushed but not acknowledged and not
removed) origination of a new LSA may wrongly triggers LSA collision
code. The patch fixes that.
Thanks to Asbjorn Mikkelsen for the bugreport and @mdelagueronniere
for the original patch.
Extend 'next hop keep' and 'next hop self' options to have boolean values
(enabled / disabled) and also values 'ibgp'/ 'ebgp' to restrict it to
routes received from IBGP / EBGP. This allows to have it enabled by
default in some cases, matches features of other implementations, and
allows to handle some strange cases like EBGP border router with 'next
hop self' also doing IBGP route reflecting.
Change default of 'next hop keep' to enabled for route servers, and
'ibgp' for route reflectors.
Update documentation for these options.
