mischief/openbsd-src
1
0
Fork 0
mirror of https://github.com/openbsd/src.git synced 2025-01-10 06:47:55 -08:00
Code
c1aa9554a3
openbsd-src/usr.sbin/bgpd/rde.h

723 lines
21 KiB
C
Raw Normal View History

When generating UPDATE handle the message size limit better. First of all warn that a prefix was dropped. In the generate an update code handle possible overflows of attributes and NLRI and withdraw the affected prefix. This way the peer will not have stale data. OK tb@
2024-09-25 07:46:51 -07:00
/* $OpenBSD: rde.h,v 1.306 2024/09/25 14:46:51 claudio Exp $ */
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
/*
2004 OK henning@
2004-01-06 02:51:14 -08:00
* Copyright (c) 2003, 2004 Claudio Jeker <claudio@openbsd.org> and
Change email address in copyright to one that is valid.
2012-10-28 06:16:11 -07:00
* Andre Oppermann <oppermann@networx.ch>
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
#ifndef __RDE_H__
#define __RDE_H__
#include <sys/types.h>
#include <sys/queue.h>
RDE update generation. First we queue all updates and withdraws on a per peer basis. A queue runner will dequeue and package those messages to valid bgp UPDATE messages and send them to the SE. Not yet done is per peer type attribute handling (like aspath prepends and nexthop modifications) and the queue runner could be a tad smarter. All in all this gives us a good starting point for the missing parts. OK henning@
2004-01-10 08:20:29 -08:00
#include <sys/tree.h>
Save some space in struct rib_entry so it is back to 64bytes (on 64bit archs). Doing this by folding the lock flag into a pointer and providing an accessor function for the rib pointer. This is an acceptable middle path for this important structure. OK benno@ on an earlier version
2017-01-24 15:38:12 -08:00
#include <stdint.h>
Calculate ASPATH_HEADER_SIZE correctly by using offsetof() instead of the sizeof calculation that did not respect possible padding bytes. OK sthen@ denis@
2018-10-25 08:49:42 -07:00
#include <stddef.h>
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
#include "bgpd.h"
Change struct prefix a bit. Make the aspath pointer opaque and remove the direct pointer to the pt_prefix struct. To getter functions prefix_aspath() and prefix_peer() added to make access of the opaque pointers possible. Looks good henning@ and benno@
2018-02-04 19:55:54 -08:00
#include "log.h"
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
/* rde internal structures */
enum peer_state {
PEER_NONE,
PEER_DOWN,
Better update error handling. Ensure that the session goes down and ignore all queued updates after an error. Looks good henning@
2004-01-13 05:34:56 -08:00
PEER_UP,
fix some typos ok henning otto
2004-07-13 10:57:20 -07:00
PEER_ERR /* error occurred going to PEER_DOWN state */
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
};
Major refactoring of the RIB handling code. Mainly change how the RIB is walked. rib_dump_r() is now an internal function and instead the code gets an additional callback for throttling the rib_dump code. This removes a lot of similar code used to make sure the RDE is not walking to fast and replaces it with simpler callbacks. The other big change is the removal of struct rib pointers in other data structures. The rib pointers are not stable because of a realloc() call happening when extending the array so instead use the RIB ID as a reference. Tested and OK denis@ and benno@
2018-10-24 01:26:37 -07:00
LIST_HEAD(prefix_list, prefix);
Switch from a LIST to TAILQ for the structure to store prefixes on a rib_entry. Mostly mechanical, this simplifies prefix_insert() and prefix_remove() since the redo queue can now just use TAILQ_INSERT_TAIL(). rde_softreconfig_sync_reeval() needs to use TAILQ_CONCAT() to move the list of prefixes over to the local TAILQ_HEAD to reapply them later. OK tb@
2022-03-22 03:53:08 -07:00
TAILQ_HEAD(prefix_queue, prefix);
Major refactoring of the RIB handling code. Mainly change how the RIB is walked. rib_dump_r() is now an internal function and instead the code gets an additional callback for throttling the rib_dump code. This removes a lot of similar code used to make sure the RDE is not walking to fast and replaces it with simpler callbacks. The other big change is the removal of struct rib pointers in other data structures. The rib pointers are not stable because of a realloc() call happening when extending the array so instead use the RIB ID as a reference. Tested and OK denis@ and benno@
2018-10-24 01:26:37 -07:00
RB_HEAD(rib_tree, rib_entry);
struct rib_entry {
RB_ENTRY(rib_entry) rib_e;
Switch from a LIST to TAILQ for the structure to store prefixes on a rib_entry. Mostly mechanical, this simplifies prefix_insert() and prefix_remove() since the redo queue can now just use TAILQ_INSERT_TAIL(). rde_softreconfig_sync_reeval() needs to use TAILQ_CONCAT() to move the list of prefixes over to the local TAILQ_HEAD to reapply them later. OK tb@
2022-03-22 03:53:08 -07:00
struct prefix_queue prefix_h;
Major refactoring of the RIB handling code. Mainly change how the RIB is walked. rib_dump_r() is now an internal function and instead the code gets an additional callback for throttling the rib_dump code. This removes a lot of similar code used to make sure the RDE is not walking to fast and replaces it with simpler callbacks. The other big change is the removal of struct rib pointers in other data structures. The rib pointers are not stable because of a realloc() call happening when extending the array so instead use the RIB ID as a reference. Tested and OK denis@ and benno@
2018-10-24 01:26:37 -07:00
struct pt_entry *prefix;
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint16_t rib_id;
uint16_t lock;
Major refactoring of the RIB handling code. Mainly change how the RIB is walked. rib_dump_r() is now an internal function and instead the code gets an additional callback for throttling the rib_dump code. This removes a lot of similar code used to make sure the RDE is not walking to fast and replaces it with simpler callbacks. The other big change is the removal of struct rib pointers in other data structures. The rib pointers are not stable because of a realloc() call happening when extending the array so instead use the RIB ID as a reference. Tested and OK denis@ and benno@
2018-10-24 01:26:37 -07:00
};
struct rib {
struct rib_tree tree;
There is no longer a reason to use two structs for RIBs where one is part of the other. Just merge struct rib_desc into struct rib. Makes code simpler. OK benno@
2019-08-14 00:39:04 -07:00
char name[PEER_DESCR_LEN];
struct filter_head *in_rules;
struct filter_head *in_rules_tmp;
Major refactoring of the RIB handling code. Mainly change how the RIB is walked. rib_dump_r() is now an internal function and instead the code gets an additional callback for throttling the rib_dump code. This removes a lot of similar code used to make sure the RDE is not walking to fast and replaces it with simpler callbacks. The other big change is the removal of struct rib pointers in other data structures. The rib pointers are not stable because of a realloc() call happening when extending the array so instead use the RIB ID as a reference. Tested and OK denis@ and benno@
2018-10-24 01:26:37 -07:00
u_int rtableid;
Improve RIB reload behaviour. Especially when the rtable changes or the route evaluation is modified. In both cases the softreconfig code will now walk the RIB and ensure that everything is in proper sync. Additionally remove 'route-collector yes|no' from the bgpd config, instead use 'rde rib Loc-RIB no evaluate' with the benefit that you can alter the setting now during runtime. Tested and OK benno@
2019-08-07 03:26:41 -07:00
u_int rtableid_tmp;
Restructure struct prefix a bit and move the rib pointer to the union that splits the normal RIB linkage vs the adjrib-out linkage. This is done to make a bit of space to put an extra add-path related id into the struct without blowing its size over 128 bytes. Long run this struct should be split up but the necessary changes are too large right now so this is the 2nd best option. OK benno@
2021-07-27 00:50:01 -07:00
enum reconf_action state, fibstate;
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint16_t id;
uint16_t flags;
uint16_t flags_tmp;
Major refactoring of the RIB handling code. Mainly change how the RIB is walked. rib_dump_r() is now an internal function and instead the code gets an additional callback for throttling the rib_dump code. This removes a lot of similar code used to make sure the RDE is not walking to fast and replaces it with simpler callbacks. The other big change is the removal of struct rib pointers in other data structures. The rib pointers are not stable because of a realloc() call happening when extending the array so instead use the RIB ID as a reference. Tested and OK denis@ and benno@
2018-10-24 01:26:37 -07:00
};
#define RIB_ADJ_IN 0
Change the Adj-RIB-Out to a per peer set of RB trees. The way RIB data structures are linked does not scale for the Adj-RIB-Out and so inserts and updates into the Adj-RIB-Out did not scale because of some linear list traversals in hot paths. A synthetic test with 4000 peers announcing one prefix each showed that the initial convergence time dropped from around 1 hout to around 6min. Note: because the Adj-RIB-Out is now per peer the order in which prefixes are dumped in 'bgpctl show rib out' changed. Tested and OK job@, benno@, phessler@
2019-07-17 03:13:26 -07:00
#define RIB_LOC_START 1
Major refactoring of the RIB handling code. Mainly change how the RIB is walked. rib_dump_r() is now an internal function and instead the code gets an additional callback for throttling the rib_dump code. This removes a lot of similar code used to make sure the RDE is not walking to fast and replaces it with simpler callbacks. The other big change is the removal of struct rib pointers in other data structures. The rib pointers are not stable because of a realloc() call happening when extending the array so instead use the RIB ID as a reference. Tested and OK denis@ and benno@
2018-10-24 01:26:37 -07:00
#define RIB_NOTFOUND 0xffff
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
/*
* How do we identify peers between the session handler and the rde?
* Currently I assume that we can do that with the neighbor_ip...
*/
Switch the rde_peer hashtable and peer list to a single RB tree. Only the RDE used a hashtable for lookups while the session engine switched from a list to RB tree some time ago. Use peer_foreach() in the mrt code instead of passing the peer list as an argument. OK benno@ tb@
2022-09-01 06:23:24 -07:00
RB_HEAD(peer_tree, rde_peer);
Use Adj-RIB-Out to push UPDATE messages to peers instead of having another set of RB trees of prefixes and atribute. Refactor most of the update code which removes some strange buffer handling. By building the output queue directly in the Adj-RIB-Out the top memory usage during startup is greatly reduced which should help busy server. Tested by phessler@ and myself
2019-01-20 18:07:56 -08:00
RB_HEAD(prefix_tree, prefix);
Change the Adj-RIB-Out to a per peer set of RB trees. The way RIB data structures are linked does not scale for the Adj-RIB-Out and so inserts and updates into the Adj-RIB-Out did not scale because of some linear list traversals in hot paths. A synthetic test with 4000 peers announcing one prefix each showed that the initial convergence time dropped from around 1 hout to around 6min. Note: because the Adj-RIB-Out is now per peer the order in which prefixes are dumped in 'bgpctl show rib out' changed. Tested and OK job@, benno@, phessler@
2019-07-17 03:13:26 -07:00
RB_HEAD(prefix_index, prefix);
Instead of processing all imsg when reading them store peer specific messages on a per peer queue. This queue is later processed one at a time resulting in a fairer processing of work and avoiding big table dumps to delay processing of other updates. OK denis@ benno@
2019-12-31 23:25:04 -08:00
struct iq;
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
struct rde_peer {
Switch the rde_peer hashtable and peer list to a single RB tree. Only the RDE used a hashtable for lookups while the session engine switched from a list to RB tree some time ago. Use peer_foreach() in the mrt code instead of passing the peer list as an argument. OK benno@ tb@
2022-09-01 06:23:24 -07:00
RB_ENTRY(rde_peer) entry;
Instead of processing all imsg when reading them store peer specific messages on a per peer queue. This queue is later processed one at a time resulting in a fairer processing of work and avoiding big table dumps to delay processing of other updates. OK denis@ benno@
2019-12-31 23:25:04 -08:00
SIMPLEQ_HEAD(, iq) imsg_queue;
RDE update generation. First we queue all updates and withdraws on a per peer basis. A queue runner will dequeue and package those messages to valid bgp UPDATE messages and send them to the SE. Not yet done is per peer type attribute handling (like aspath prepends and nexthop modifications) and the queue runner could be a tad smarter. All in all this gives us a good starting point for the missing parts. OK henning@
2004-01-10 08:20:29 -08:00
struct peer_config conf;
Instead of relaying struct peer from the SE to the RDE to fill out 10 stat numbers, just send the peerid and have the RDE response with the stats. The control code will then merge these counters into the real peer struct and send that to bgpctl. This reduces the number of bytes sent around a fair bit. OK tb@
2023-02-09 05:43:23 -08:00
struct rde_peer_stats stats;
Add to the IMSG_SESSION_UP message the local and remote address of the peer session. This is needed to set the correct nexthop address in outgoing bgp UPDATES. OK henning@
2004-01-11 12:13:00 -08:00
struct bgpd_addr remote_addr;
Store the local v4 and v6 address of a session. For IPv4 sessions the first IPv6 address with global scope of the interface used by the session is used. In any case the local endpoint of the session is used for one of the two no matter what scope. OK henning@
2004-11-11 05:06:44 -08:00
struct bgpd_addr local_v4_addr;
struct bgpd_addr local_v6_addr;
Reshuffle order in struct rde_peer a bit. No functional change.
2018-07-31 08:30:04 -07:00
struct capabilities capa;
Implement send side of RFC7911 ADD-PATH This allows to send out more then one path per perfix to a neighbor that supports add-path receive. OpenBGPD supports a few different modes to select which paths to send: - all: send all valid paths (the ones with a * in bgpctl output) - best: send out only the single best path - ecmp: send out paths that evaluate the same up and including the nexthop metric - as-wide-best: send out paths that evaluete the same up but not including the nexthop metric Currently ecmp and as-wide-best are the same. On top of this best, ecmp and as-wide-best allow to include extra paths (e.g. best plus 2) and for the multipath modes there is also a maximum (e.g. ecmp plus 2 max 4) OK tb@
2022-07-11 10:08:21 -07:00
struct addpath_eval eval;
Change the Adj-RIB-Out to a per peer set of RB trees. The way RIB data structures are linked does not scale for the Adj-RIB-Out and so inserts and updates into the Adj-RIB-Out did not scale because of some linear list traversals in hot paths. A synthetic test with 4000 peers announcing one prefix each showed that the initial convergence time dropped from around 1 hout to around 6min. Note: because the Adj-RIB-Out is now per peer the order in which prefixes are dumped in 'bgpctl show rib out' changed. Tested and OK job@, benno@, phessler@
2019-07-17 03:13:26 -07:00
struct prefix_index adj_rib_out;
Use Adj-RIB-Out to push UPDATE messages to peers instead of having another set of RB trees of prefixes and atribute. Refactor most of the update code which removes some strange buffer handling. By building the output queue directly in the Adj-RIB-Out the top memory usage during startup is greatly reduced which should help busy server. Tested by phessler@ and myself
2019-01-20 18:07:56 -08:00
struct prefix_tree updates[AID_MAX];
struct prefix_tree withdraws[AID_MAX];
Compile the output filter rules into per peer filter rules. especially on route-servers the output filters are in the hot path so reducing the number of rules to check has a big impact. I have seen a 25% to 30% speedup in my big IXP testbench. The output ruleset is applied and copied for each peer during config reload and when a peer is initially added. OK tb@
2023-03-09 23:57:15 -08:00
struct filter_head *out_rules;
Use Adj-RIB-Out to push UPDATE messages to peers instead of having another set of RB trees of prefixes and atribute. Refactor most of the update code which removes some strange buffer handling. By building the output queue directly in the Adj-RIB-Out the top memory usage during startup is greatly reduced which should help busy server. Tested by phessler@ and myself
2019-01-20 18:07:56 -08:00
time_t staletime[AID_MAX];
Convert bgpid, remote_bgpid and clusterid to host byte order. Before the RDE used host byte order for remote_bgpid but all the other code used network byte order. The reason for that was that bgpid was initially an IPv4 address but since RFC 6286 in 2011 this is much more relaxed and so it makes more sense to just treat them as numbers and so host byte order. OK tb@
2024-05-22 01:41:14 -07:00
uint32_t remote_bgpid;
Adjust pathid_assign() to be much faster in the common case. Use a per peer path_id_tx to assign to paths received from none add-path enabled peers. This skips two extra walks of the RIB prefix list and is a big speed-up when there are many regular sessions. If the session uses add-path recv then the old way of assigning random path_ids needs to be used. With input and OK tb@
2022-09-21 03:39:17 -07:00
uint32_t path_id_tx;
Improve IPv6 link-local address handling When a session is established determine the possible interface scope of that session. The scope is only set when the remote address is directly connected. This interface scope is passed to the RDE that uses this information when link-local nexthops are received. Again checking that a link-local nexthop is actually acceptable. OK tb@
2023-10-16 03:25:45 -07:00
unsigned int local_if_scope;
sort structs for most optimal alignment, help mickey, claudio ok
2004-12-23 07:08:43 -08:00
enum peer_state state;
Improve reload behaviour of RDE peer flags and export_type. Add an extra reload barrier (IMSG_RECONF_DRAIN) to the sync of the peer config from the session engine to the rde. Necessary to ensure that the peer config is up to date in the RDE before hitting reconfiguration. Store the export_type and the peer flags outside of peer->conf. Adjust all users of these two fields so they only look at the copies in peer. During reload check the values with the peer->conf to check for changes. If the export_type or the rde evaluate or transparent-as flags changed flush the Adj-RIB-Out for that peer and in a 2nd step rebuild the RIB from scratch. This results in a lot of UPDATE churn but these configs are not altered often. Fix multiple issues in the rde_softreconfig_in_done handler that resulted in multiple runs of the out stage of the softreconfig pipeline. OK benno@
2021-05-06 02:18:54 -07:00
enum export_type export_type;
Implement ASPA validation and reload logic on ASPA set changes. For this use the validation state (vstate) in struct prefix and struct filterstate to store both the ASPA and ROA validity. Introduce helper functions to set and get the various states for struct prefix and make sure struct filterstate is also setup properly. Change the ASPA state in rde_aspath to be AFI/AID and role independent by storing all 4 possible outcomes. Also add a ASPA generation count which is used to update the rde_aspath ASPA state cache on reloads. Rework the rde_aspa.c code to be AFI/AID and role independent. Doing this for roles is trivial but AFI switch goes deep and is so unnecessary. The reload is combined with the ROA reload logic and renamed to RPKI softreload. OK tb@
2023-01-24 03:28:41 -08:00
enum role role;
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint16_t loc_rib_id;
uint16_t short_as;
uint16_t mrt_idx;
uint8_t recv_eor; /* bitfield per AID */
uint8_t sent_eor; /* bitfield per AID */
uint8_t reconf_out; /* out filter changed */
uint8_t reconf_rib; /* rib changed */
uint8_t throttled;
uint8_t flags;
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
};
Add the needed logic to load the ASPA table from the rtr process into the RDE. The actual reload logic is missing to keep the diff small. OK tb@
2023-01-17 08:09:01 -08:00
struct rde_aspa;
Implement ASPA validation and reload logic on ASPA set changes. For this use the validation state (vstate) in struct prefix and struct filterstate to store both the ASPA and ROA validity. Introduce helper functions to set and get the various states for struct prefix and make sure struct filterstate is also setup properly. Change the ASPA state in rde_aspath to be AFI/AID and role independent by storing all 4 possible outcomes. Also add a ASPA generation count which is used to update the rde_aspath ASPA state cache on reloads. Rework the rde_aspa.c code to be AFI/AID and role independent. Doing this for roles is trivial but AFI switch goes deep and is so unnecessary. The reload is combined with the ROA reload logic and renamed to RPKI softreload. OK tb@
2023-01-24 03:28:41 -08:00
struct rde_aspa_state {
Remove per-AFI ASPA handling in bgpd internals With draft-ietf-sidrops-aspa-profile-16 and draft-ietf-sidrops-aspa-verification-15 the AFI dependence of ASPA records was dropped. So remove this complication form the code. This only removes the AFI handling internally in bgpd but still allows the old syntax in aspa-set tables. The optional address family is just ignored and records are merged together. For RTR sessions draft-ietf-sidrops-8210bis has not yet been updated so right now we still handle RTR sessions as specified there. The IPv4 and IPv6 ASPA entries are handled in two trees and merged together into one AFI independent tree. This is the best we can do for now until IETF updates draft-ietf-sidrops-8210bis. OK tb@ job@
2023-08-16 01:26:35 -07:00
uint8_t onlyup;
uint8_t downup;
Implement ASPA validation and reload logic on ASPA set changes. For this use the validation state (vstate) in struct prefix and struct filterstate to store both the ASPA and ROA validity. Introduce helper functions to set and get the various states for struct prefix and make sure struct filterstate is also setup properly. Change the ASPA state in rde_aspath to be AFI/AID and role independent by storing all 4 possible outcomes. Also add a ASPA generation count which is used to update the rde_aspath ASPA state cache on reloads. Rework the rde_aspa.c code to be AFI/AID and role independent. Doing this for roles is trivial but AFI switch goes deep and is so unnecessary. The reload is combined with the ROA reload logic and renamed to RPKI softreload. OK tb@
2023-01-24 03:28:41 -08:00
};
Add the needed logic to load the ASPA table from the rtr process into the RDE. The actual reload logic is missing to keep the diff small. OK tb@
2023-01-17 08:09:01 -08:00
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
#define AS_SET 1
#define AS_SEQUENCE 2
Accept but ignore (treat as withdraw) updates with AS_CONFED_* path segments. Bgpd does not support confederations but it is too extreme to close a session because a path contained such elements. OK henning@, sthen@
2010-11-18 04:18:31 -08:00
#define AS_CONFED_SEQUENCE 3
#define AS_CONFED_SET 4
Calculate ASPATH_HEADER_SIZE correctly by using offsetof() instead of the sizeof calculation that did not respect possible padding bytes. OK sthen@ denis@
2018-10-25 08:49:42 -07:00
#define ASPATH_HEADER_SIZE (offsetof(struct aspath, data))
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
struct aspath {
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint32_t source_as; /* cached source_as */
uint16_t len; /* total length of aspath in octets */
uint16_t ascnt; /* number of AS hops in data */
gcc-ism. We don't want it, we don't need it. OK henning@ noticed by deraadt@
2004-09-28 08:48:52 -07:00
u_char data[1]; /* placeholder for actual data */
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
};
enum attrtypes {
ATTR_UNDEF,
ATTR_ORIGIN,
ATTR_ASPATH,
ATTR_NEXTHOP,
ATTR_MED,
ATTR_LOCALPREF,
ATTR_ATOMIC_AGGREGATE,
Add basic support for communities. Currently it is only possible to filter on communities, e.g match from any community 24640:* set localpref 666 OK henning@
2004-03-11 09:12:51 -08:00
ATTR_AGGREGATOR,
RFC 2796 bgp route reflector support. This is very useful in conjunction with templates. looks good, go for it henning@
2004-05-21 08:36:40 -07:00
ATTR_COMMUNITIES,
ATTR_ORIGINATOR_ID,
First step at multiprotocol support, only partially done. OK henning@
2004-06-24 16:15:58 -07:00
ATTR_CLUSTER_LIST,
ATTR_MP_REACH_NLRI=14,
Make bgpd 4-byte AS compatible. All internal representations of AS numbers are now 4-byte instead of the old 2-byte numbers. The only exception are communities because they can not be switched. The RDE will inflate and deflate the ASPATH and AGGREGATOR attributes on demand and create the NEW_ASPATH and NEW_AGGREGATOR field whenever needed. Both old and new stile sessions are supported and can be mixed. Currently new stile sessions with the 4-byte AS number capability turned on are only enabled if one of the AS numbers involved is a 4-byte one. This is based on an initial diff by Geoff Huston gih (at) apnic (dot) net Cleanup, testing and bug-fixes by myself (via AS 3.10). Currently mrt table dumps are producing incompatible output this will be fixed afterwards -- this diff is already big enough. "get it in if you think it is ready" henning@
2007-04-23 06:04:24 -07:00
ATTR_MP_UNREACH_NLRI=15,
Add defines for extended communities. OK henning@
2008-01-23 00:11:32 -08:00
ATTR_EXT_COMMUNITIES=16,
Replace NEW_ASPATH/NEW_AGGREGATOR with the naming from RFC4893, AS4_PATH/AS4_AGGREGATOR. No binary change. ok claudio
2009-01-13 13:35:16 -08:00
ATTR_AS4_PATH=17,
Add support for draft-ietf-idr-large-community Joint work with Job Snijders, many thanks! OK benno@ deraadt@
2016-10-14 09:05:35 -07:00
ATTR_AS4_AGGREGATOR=18,
Huawei squatted on BGP Path Attribute 30, and Cisco squatted on 31. So, IANA moved the Large Communities attribute to 32, which is a nice pun on the problem it is solving.
2016-10-27 01:21:58 -07:00
ATTR_LARGE_COMMUNITIES=32,
Add support for RFC 9234 - Route Leak Prevention and Detection Using Roles With this it is possible to send a role in the OPEN message and if that was successful the RDE will add the new OTC attribute if necessary. OK tb@
2022-06-27 06:26:51 -07:00
ATTR_OTC=35,
Exit the attribute loop early if there are no unknown attributes left and the loop passed all attributes known by bgpd. Saves about 80% of time in up_generate_attr(). OK phessler@
2019-05-31 02:46:31 -07:00
ATTR_FIRST_UNKNOWN, /* after this all attributes are unknown */
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
};
/* attribute flags. 4 low order bits reserved */
#define ATTR_EXTLEN 0x10
#define ATTR_PARTIAL 0x20
#define ATTR_TRANSITIVE 0x40
#define ATTR_OPTIONAL 0x80
By default mask the reserved bits and the ext len bit in the attribute flags field. Some systems seem to start sending bad flags around which cause session failures in bgpd. Make sure that bgpd ignores the must be zero flags correctly and ensure that they are always reset to zero when sending updates out. Reported and patch tested by Laurent CARON, OK henning@
2012-08-12 07:24:56 -07:00
#define ATTR_RESERVED 0x0f
/* by default mask the reserved bits and the ext len bit */
#define ATTR_DEFMASK (ATTR_RESERVED | ATTR_EXTLEN)
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
/* default attribute flags for well known attributes */
RDE update generation. First we queue all updates and withdraws on a per peer basis. A queue runner will dequeue and package those messages to valid bgp UPDATE messages and send them to the SE. Not yet done is per peer type attribute handling (like aspath prepends and nexthop modifications) and the queue runner could be a tad smarter. All in all this gives us a good starting point for the missing parts. OK henning@
2004-01-10 08:20:29 -08:00
#define ATTR_WELL_KNOWN ATTR_TRANSITIVE
struct attr {
Switch the generic attribute cache to an RB tree. OK benno@ tb@
2022-08-31 07:29:36 -07:00
RB_ENTRY(attr) entry;
sort structs for most optimal alignment, help mickey, claudio ok
2004-12-23 07:08:43 -08:00
u_char *data;
Cache optional BGP attributes (mostly communities) and use a simple pointer plus a ref counter to link the attributes to the path object. Saves +/- 10M on 11 full feeds. Looks good Henning
2006-01-05 08:00:07 -08:00
int refcnt;
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint16_t len;
uint8_t flags;
uint8_t type;
Monster diff to get one step closer to IPv6 support. Cleanup path attribute handling. First of all kill struct attr_flags, all those infos are now in struct rde_aspath. Second move attribute parser functions into rde.c, rde_attr.c is shared between bgpd and bgpctl. Third reimplementation of the nexthop handling. Make it IPv6 ready and fix some major bug relating to "set nexthop". henning@ OK if it breaks nothing
2004-08-06 05:04:08 -07:00
};
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
struct rde_community {
Switch the DB of communities collections to a RB tree instead of an undersized hash table. OK tb@
2022-08-29 09:44:47 -07:00
RB_ENTRY(rde_community) entry;
int size;
int nentries;
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
int flags;
int refcnt;
Cleanup, remove some unneded spaces add some other where needed. No binary change according to clang
2019-06-17 06:35:42 -07:00
struct community *communities;
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
};
#define PARTIAL_COMMUNITIES 0x01
#define PARTIAL_LARGE_COMMUNITIES 0x02
#define PARTIAL_EXT_COMMUNITIES 0x04
Make bgpd 4-byte AS compatible. All internal representations of AS numbers are now 4-byte instead of the old 2-byte numbers. The only exception are communities because they can not be switched. The RDE will inflate and deflate the ASPATH and AGGREGATOR attributes on demand and create the NEW_ASPATH and NEW_AGGREGATOR field whenever needed. Both old and new stile sessions are supported and can be mixed. Currently new stile sessions with the 4-byte AS number capability turned on are only enabled if one of the AS numbers involved is a 4-byte one. This is based on an initial diff by Geoff Huston gih (at) apnic (dot) net Cleanup, testing and bug-fixes by myself (via AS 3.10). Currently mrt table dumps are producing incompatible output this will be fixed afterwards -- this diff is already big enough. "get it in if you think it is ready" henning@
2007-04-23 06:04:24 -07:00
#define F_ATTR_ORIGIN 0x00001
#define F_ATTR_ASPATH 0x00002
#define F_ATTR_NEXTHOP 0x00004
#define F_ATTR_LOCALPREF 0x00008
#define F_ATTR_MED 0x00010
#define F_ATTR_MED_ANNOUNCE 0x00020
#define F_ATTR_MP_REACH 0x00040
#define F_ATTR_MP_UNREACH 0x00080
Replace NEW_ASPATH/NEW_AGGREGATOR with the naming from RFC4893, AS4_PATH/AS4_AGGREGATOR. No binary change. ok claudio
2009-01-13 13:35:16 -08:00
#define F_ATTR_AS4BYTE_NEW 0x00100 /* AS4_PATH or AS4_AGGREGATOR */
Prefixes or actually pathes that would cause a rooting loops should not be dropped when parsed but instead be added to the RIB marked as not eligible. So the decision process does not pick them up as a valid route. Tested and some ideas by Tony Sarendal (tony (at) polarcap (dot) org)
2007-11-26 17:13:54 -08:00
#define F_ATTR_LOOP 0x00200 /* path would cause a route loop */
F_LOCAL and F_ORIGINAL are gone. The Adj-Rib-In is now a distinct tree. Fix pf table code by checking if the aspath has a pftableid set or not instead of doing the F_LOCAL dance. This works because the in the Adj-Rib-In it is impossible to set pftableid.
2009-05-17 07:45:25 -07:00
#define F_PREFIX_ANNOUNCED 0x00400
#define F_ANN_DYNAMIC 0x00800
Add support for RFC 9234 - Route Leak Prevention and Detection Using Roles With this it is possible to send a role in the OPEN message and if that was successful the RDE will add the new OTC attribute if necessary. OK tb@
2022-06-27 06:26:51 -07:00
#define F_ATTR_OTC 0x01000 /* OTC present */
Add F_CTL_LEAKED and F_CTL_INELIGIBLE flags for bgpctl to show leaked and ineligible paths. While there rename F_PREF_OTC_LOOP to F_PREF_OTC_LEAK since this indicates that a route leak was detected. OK tb@
2023-03-13 09:52:41 -07:00
#define F_ATTR_OTC_LEAK 0x02000 /* otc leak, not eligible */
spelling fixes; from paul tagliamonte any parts of his diff not taken are noted on tech
2022-12-28 13:30:15 -08:00
#define F_ATTR_PARSE_ERR 0x10000 /* parse error, not eligible */
Add plumbing inside of the prefix handling to put prefixes on either the prefixes or updates list depending on flags passed to the functions. While there also introduce a similar flag for rde_aspath and adjust path_compare so that this flag is ignored (liked the linked one). OK benno@
2018-02-06 16:02:02 -08:00
#define F_ATTR_LINKED 0x20000 /* if set path is on various lists */
Small step in supporting the Adj-RIB-In additionaly to the Local-RIB. First step is to define two flags F_LOCAL and F_ORIGINAL. These flags are used to distinguish prefix in the Local-RIB and those in the Adj- RIB-In. Adapt prefix API and add additional checks so that no Adj-RIB- In prefixes get mistakenly selected. Currently no F_ORIGINAL prefixes are created but this may change soon. Looks good Henning.
2006-01-14 14:39:49 -08:00
RDE update generation. First we queue all updates and withdraws on a per peer basis. A queue runner will dequeue and package those messages to valid bgp UPDATE messages and send them to the SE. Not yet done is per peer type attribute handling (like aspath prepends and nexthop modifications) and the queue runner could be a tad smarter. All in all this gives us a good starting point for the missing parts. OK henning@
2004-01-10 08:20:29 -08:00
#define ORIGIN_IGP 0
#define ORIGIN_EGP 1
#define ORIGIN_INCOMPLETE 2
The default localpreference is 100 and not 0. Found and ok henning@
2004-07-28 09:02:14 -07:00
#define DEFAULT_LPREF 100
Monster diff to get one step closer to IPv6 support. Cleanup path attribute handling. First of all kill struct attr_flags, all those infos are now in struct rde_aspath. Second move attribute parser functions into rde.c, rde_attr.c is shared between bgpd and bgpctl. Third reimplementation of the nexthop handling. Make it IPv6 ready and fix some major bug relating to "set nexthop". henning@ OK if it breaks nothing
2004-08-06 05:04:08 -07:00
struct rde_aspath {
Switch rde_aspath to a RB tree instead of a hash table. OK tb@
2022-08-29 09:43:07 -07:00
RB_ENTRY(rde_aspath) entry;
Cache optional BGP attributes (mostly communities) and use a simple pointer plus a ref counter to link the attributes to the path object. Saves +/- 10M on 11 full feeds. Looks good Henning
2006-01-05 08:00:07 -08:00
struct attr **others;
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
struct aspath *aspath;
Implement ASPA validation and reload logic on ASPA set changes. For this use the validation state (vstate) in struct prefix and struct filterstate to store both the ASPA and ROA validity. Introduce helper functions to set and get the various states for struct prefix and make sure struct filterstate is also setup properly. Change the ASPA state in rde_aspath to be AFI/AID and role independent by storing all 4 possible outcomes. Also add a ASPA generation count which is used to update the rde_aspath ASPA state cache on reloads. Rework the rde_aspa.c code to be AFI/AID and role independent. Doing this for roles is trivial but AFI switch goes deep and is so unnecessary. The reload is combined with the ROA reload logic and renamed to RPKI softreload. OK tb@
2023-01-24 03:28:41 -08:00
struct rde_aspa_state aspa_state;
Instead of calling SipHash24_Update() in path_hash for each element of struct rde_aspath define aspath_hashstart and aspath_hashend and update all values in one call. Inspired by struct process and its ps_startcopy. OK deraadt@
2020-01-09 07:50:34 -08:00
int refcnt;
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint32_t flags; /* internally used */
uint32_t med; /* multi exit disc */
uint32_t lpref; /* local pref */
uint32_t weight; /* low prio lpref */
uint16_t rtlabelid; /* route label id */
uint16_t pftableid; /* pf table id */
uint8_t origin;
uint8_t others_len;
Implement ASPA validation and reload logic on ASPA set changes. For this use the validation state (vstate) in struct prefix and struct filterstate to store both the ASPA and ROA validity. Introduce helper functions to set and get the various states for struct prefix and make sure struct filterstate is also setup properly. Change the ASPA state in rde_aspath to be AFI/AID and role independent by storing all 4 possible outcomes. Also add a ASPA generation count which is used to update the rde_aspath ASPA state cache on reloads. Rework the rde_aspa.c code to be AFI/AID and role independent. Doing this for roles is trivial but AFI switch goes deep and is so unnecessary. The reload is combined with the ROA reload logic and renamed to RPKI softreload. OK tb@
2023-01-24 03:28:41 -08:00
uint8_t aspa_generation;
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
};
enum nexthop_state {
RDE part of the nexthop verification puzzle. The RDE just tracks the nexthop IPs and reacts on nexthop messages from the parent. ok henning@
2003-12-25 15:22:13 -08:00
NEXTHOP_LOOKUP,
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
NEXTHOP_UNREACH,
Change nexthop_update to run the list walk over all prefixes to run asynchronously and therefor other tasks can make progress at the same time. Additionally prefixes belonging to a RIB which does not run the the decision process are no longer linked into the nexthop list. This replaces the early return in prefix_updateall() and reduces the time spent in nexthop_update(). OK benno@
2019-06-20 06:18:19 -07:00
NEXTHOP_REACH,
Add comment that NEXTHOP_FLAPPED is only set on oldstate of a nexthop.
2022-08-03 01:56:23 -07:00
NEXTHOP_FLAPPED /* only used by oldstate */
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
};
move struct nexthop definition to rde.h
2003-12-26 13:30:20 -08:00
struct nexthop {
Switch nexthop hash to a RB tree. OK benno@
2022-08-30 11:50:21 -07:00
RB_ENTRY(nexthop) entry;
Change nexthop_update to run the list walk over all prefixes to run asynchronously and therefor other tasks can make progress at the same time. Additionally prefixes belonging to a RIB which does not run the the decision process are no longer linked into the nexthop list. This replaces the early return in prefix_updateall() and reduces the time spent in nexthop_update(). OK benno@
2019-06-20 06:18:19 -07:00
TAILQ_ENTRY(nexthop) runner_l;
Move nexthop and nexthop flags from the rde_aspath to struct prefix. struct prefix will be slowly becomming the hub of the rib. OK phessler@ job@
2018-08-03 09:31:22 -07:00
struct prefix_list prefix_h;
Change nexthop_update to run the list walk over all prefixes to run asynchronously and therefor other tasks can make progress at the same time. Additionally prefixes belonging to a RIB which does not run the the decision process are no longer linked into the nexthop list. This replaces the early return in prefix_updateall() and reduces the time spent in nexthop_update(). OK benno@
2019-06-20 06:18:19 -07:00
struct prefix *next_prefix;
sort structs for most optimal alignment, help mickey, claudio ok
2004-12-23 07:08:43 -08:00
struct bgpd_addr exit_nexthop;
struct bgpd_addr true_nexthop;
struct bgpd_addr nexthop_net;
move struct nexthop definition to rde.h
2003-12-26 13:30:20 -08:00
#if 0
Make it possible to announce own networks. In the RDE these prefixes are attached to a pseudo peer and inserted like all other prefixes into the RIB. OK henning@
2004-01-17 11:35:35 -08:00
/*
* currently we use the boolean nexthop state, this could be exchanged
First step at multiprotocol support, only partially done. OK henning@
2004-06-24 16:15:58 -07:00
* with a variable cost with a max for unreachable.
Make it possible to announce own networks. In the RDE these prefixes are attached to a pseudo peer and inserted like all other prefixes into the RIB. OK henning@
2004-01-17 11:35:35 -08:00
*/
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint32_t costs;
move struct nexthop definition to rde.h
2003-12-26 13:30:20 -08:00
#endif
Properly start reference counting struct nexthop. This removes the need for some ugly workaround to make sure nexthop objects don't disapear while still being referenced. During initial lookup of a nexthop a extra reference is pulled but even that is now a bit cleaner than before. Tested by job@, dennis@, benno@ OK job@ dennis@
2018-06-25 07:28:33 -07:00
int refcnt;
Monster diff to get one step closer to IPv6 support. Cleanup path attribute handling. First of all kill struct attr_flags, all those infos are now in struct rde_aspath. Second move attribute parser functions into rde.c, rde_attr.c is shared between bgpd and bgpctl. Third reimplementation of the nexthop handling. Make it IPv6 ready and fix some major bug relating to "set nexthop". henning@ OK if it breaks nothing
2004-08-06 05:04:08 -07:00
enum nexthop_state state;
Change nexthop_update to run the list walk over all prefixes to run asynchronously and therefor other tasks can make progress at the same time. Additionally prefixes belonging to a RIB which does not run the the decision process are no longer linked into the nexthop list. This replaces the early return in prefix_updateall() and reduces the time spent in nexthop_update(). OK benno@
2019-06-20 06:18:19 -07:00
enum nexthop_state oldstate;
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint8_t nexthop_netlen;
uint8_t flags;
Monster diff to get one step closer to IPv6 support. Cleanup path attribute handling. First of all kill struct attr_flags, all those infos are now in struct rde_aspath. Second move attribute parser functions into rde.c, rde_attr.c is shared between bgpd and bgpctl. Third reimplementation of the nexthop handling. Make it IPv6 ready and fix some major bug relating to "set nexthop". henning@ OK if it breaks nothing
2004-08-06 05:04:08 -07:00
#define NEXTHOP_CONNECTED 0x01
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
};
Make the RDE IPv6 ready missing is the message handling. The internal prefix tree changed form a hash table to a per AF RB tree. OK henning@ some ideas are from Brent Graveland.
2004-06-22 13:28:58 -07:00
/* generic entry without address specific part */
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
struct pt_entry {
Make the RDE IPv6 ready missing is the message handling. The internal prefix tree changed form a hash table to a per AF RB tree. OK henning@ some ideas are from Brent Graveland.
2004-06-22 13:28:58 -07:00
RB_ENTRY(pt_entry) pt_e;
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint8_t aid;
uint8_t prefixlen;
Put the size of the pt_entry object into the struct itself. Increase the refcnt to a 32bit int and while there reorder the vpn specific structs a bit so the IPv4 and IPv6 types are more equal. OK tb@
2023-03-30 05:11:18 -07:00
uint16_t len;
uint32_t refcnt;
More pt_entry cleanup, move structure definitions to rde_prefix.c and by that make them private. Remove no longer used AID_PTSIZE define. OK tb@
2023-03-28 06:30:31 -07:00
uint8_t data[4]; /* data depending on aid */
add support for IPv6 VPN routes The kernel bits are missing as of now. With input from claudio@ and kn@ OK claudio@
2018-12-30 05:53:07 -08:00
};
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
struct prefix {
Change the Adj-RIB-Out to a per peer set of RB trees. The way RIB data structures are linked does not scale for the Adj-RIB-Out and so inserts and updates into the Adj-RIB-Out did not scale because of some linear list traversals in hot paths. A synthetic test with 4000 peers announcing one prefix each showed that the initial convergence time dropped from around 1 hout to around 6min. Note: because the Adj-RIB-Out is now per peer the order in which prefixes are dumped in 'bgpctl show rib out' changed. Tested and OK job@, benno@, phessler@
2019-07-17 03:13:26 -07:00
union {
struct {
Switch from a LIST to TAILQ for the structure to store prefixes on a rib_entry. Mostly mechanical, this simplifies prefix_insert() and prefix_remove() since the redo queue can now just use TAILQ_INSERT_TAIL(). rde_softreconfig_sync_reeval() needs to use TAILQ_CONCAT() to move the list of prefixes over to the local TAILQ_HEAD to reapply them later. OK tb@
2022-03-22 03:53:08 -07:00
TAILQ_ENTRY(prefix) rib;
LIST_ENTRY(prefix) nexthop;
Restructure struct prefix a bit and move the rib pointer to the union that splits the normal RIB linkage vs the adjrib-out linkage. This is done to make a bit of space to put an extra add-path related id into the struct without blowing its size over 128 bytes. Long run this struct should be split up but the necessary changes are too large right now so this is the 2nd best option. OK benno@
2021-07-27 00:50:01 -07:00
struct rib_entry *re;
Change the Adj-RIB-Out to a per peer set of RB trees. The way RIB data structures are linked does not scale for the Adj-RIB-Out and so inserts and updates into the Adj-RIB-Out did not scale because of some linear list traversals in hot paths. A synthetic test with 4000 peers announcing one prefix each showed that the initial convergence time dropped from around 1 hout to around 6min. Note: because the Adj-RIB-Out is now per peer the order in which prefixes are dumped in 'bgpctl show rib out' changed. Tested and OK job@, benno@, phessler@
2019-07-17 03:13:26 -07:00
} list;
struct {
RB_ENTRY(prefix) index, update;
} tree;
} entry;
Add a direct pointer from struct prefix to struct pt_entry. This changes makes it possible to not use the struct rib_entry pointer which will be used to optimize the Adj-RIB-Out. Also adjust pt_ref() and pt_unref() so that the code can be written a bit more compact. Also prefix_cmp() no longer needs to go via rib_compare() and calls pt_prefix_cmp() directly. OK phessler@
2019-06-21 22:44:05 -07:00
struct pt_entry *pt;
Just put the aspath and the peer directly into the prefix struct. Needed for upcoming changes.
2018-06-27 06:14:44 -07:00
struct rde_aspath *aspath;
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
struct rde_community *communities;
Just put the aspath and the peer directly into the prefix struct. Needed for upcoming changes.
2018-06-27 06:14:44 -07:00
struct rde_peer *peer;
Move nexthop and nexthop flags from the rde_aspath to struct prefix. struct prefix will be slowly becomming the hub of the rib. OK phessler@ job@
2018-08-03 09:31:22 -07:00
struct nexthop *nexthop; /* may be NULL */
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
time_t lastchange;
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint32_t path_id;
uint32_t path_id_tx;
Introduce 'rde rib Loc-RIB include filtered' a feature that includes filtered prefixes in the Loc-RIB This includes filtered prefixes into the Loc-RIB but they are marked ineligible so nothing will select them but it is possible to show them in bgpctl. So 'bppctl show rib filtered' will return all prefixes filtered out by the input filters. OK tb@
2024-08-14 12:09:51 -07:00
uint16_t flags;
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint8_t validation_state;
uint8_t nhflags;
Introduce a decision metric (dmetric) that classifies the relation of this prefix with respect to its previous one. Currently the plan is to distinguish the best prefix (only one), ecmp prefixes (currently the same as as-wide-multipath), as-wide-multipath prefixes, valid prefixes and invalid prefixes. This information will be used to implement add-path send but also for ecmp support in bgpd. OK tb@
2022-07-07 05:16:04 -07:00
int8_t dmetric; /* decision metric */
Introduce 'rde rib Loc-RIB include filtered' a feature that includes filtered prefixes in the Loc-RIB This includes filtered prefixes into the Loc-RIB but they are marked ineligible so nothing will select them but it is possible to show them in bgpctl. So 'bppctl show rib filtered' will return all prefixes filtered out by the input filters. OK tb@
2024-08-14 12:09:51 -07:00
};
#define PREFIX_FLAG_WITHDRAW 0x0001 /* enqueued on withdraw queue */
#define PREFIX_FLAG_UPDATE 0x0002 /* enqueued on update queue */
#define PREFIX_FLAG_DEAD 0x0004 /* locked but removed */
#define PREFIX_FLAG_STALE 0x0008 /* stale entry (graceful reload) */
#define PREFIX_FLAG_MASK 0x000f /* mask for the prefix types */
#define PREFIX_FLAG_ADJOUT 0x0010 /* prefix is in the adj-out rib */
#define PREFIX_FLAG_EOR 0x0020 /* prefix is EoR */
#define PREFIX_NEXTHOP_LINKED 0x0040 /* prefix is linked onto nexthop list */
#define PREFIX_FLAG_LOCKED 0x0080 /* locked by rib walker */
#define PREFIX_FLAG_FILTERED 0x0100 /* prefix is filtered (ineligible) */
Introduce a decision metric (dmetric) that classifies the relation of this prefix with respect to its previous one. Currently the plan is to distinguish the best prefix (only one), ecmp prefixes (currently the same as as-wide-multipath), as-wide-multipath prefixes, valid prefixes and invalid prefixes. This information will be used to implement add-path send but also for ecmp support in bgpd. OK tb@
2022-07-07 05:16:04 -07:00
#define PREFIX_DMETRIC_NONE 0
#define PREFIX_DMETRIC_INVALID 1
#define PREFIX_DMETRIC_VALID 2
#define PREFIX_DMETRIC_AS_WIDE 3
#define PREFIX_DMETRIC_ECMP 4
#define PREFIX_DMETRIC_BEST 5
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
Change the Adj-RIB-Out to a per peer set of RB trees. The way RIB data structures are linked does not scale for the Adj-RIB-Out and so inserts and updates into the Adj-RIB-Out did not scale because of some linear list traversals in hot paths. A synthetic test with 4000 peers announcing one prefix each showed that the initial convergence time dropped from around 1 hout to around 6min. Note: because the Adj-RIB-Out is now per peer the order in which prefixes are dumped in 'bgpctl show rib out' changed. Tested and OK job@, benno@, phessler@
2019-07-17 03:13:26 -07:00
/* possible states for nhflags */
Move nexthop and nexthop flags from the rde_aspath to struct prefix. struct prefix will be slowly becomming the hub of the rib. OK phessler@ job@
2018-08-03 09:31:22 -07:00
#define NEXTHOP_SELF 0x01
#define NEXTHOP_REJECT 0x02
#define NEXTHOP_BLACKHOLE 0x04
#define NEXTHOP_NOMODIFY 0x08
Properly handle nexthop state changes in the decision process In rev 1.90 of rde_decide.c the re->active cache of the best prefix was replaced with a call to prefix_best(). This introduced a bug because the nexthop state at that time may have changed already. As a result when a nexthop became unreachable prefix_evaluate() had oldbest = NULL and newbest = NULL and did not withdraw the prefix from FIB and Adj-RIB-Out. To fix this store the nexthop state per prefix and introduce prefix_evaluate_nexthop() which removes the prefix from the decision list, updates the nexthop state of the prefix and reinserts the prefix. Doing this ensures that prefix_best() always reports the same result once the decison process is done. prefix_best() and prefix_eligible() only depend on data stored on the prefix itself. OK tb@
2022-07-25 09:37:55 -07:00
#define NEXTHOP_MASK 0x0f
#define NEXTHOP_VALID 0x80
Move nexthop and nexthop flags from the rde_aspath to struct prefix. struct prefix will be slowly becomming the hub of the rib. OK phessler@ job@
2018-08-03 09:31:22 -07:00
Introduce a struct filterstate and pass this to rde_filter instead of passing the asp. This is no longer using asp_get() and should be therefor be a bit lighter. Will also allow to put more stuff into the state. OK denis@ sthen@
2018-07-09 07:08:48 -07:00
struct filterstate {
Move nexthop and nexthop flags from the rde_aspath to struct prefix. struct prefix will be slowly becomming the hub of the rib. OK phessler@ job@
2018-08-03 09:31:22 -07:00
struct rde_aspath aspath;
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
struct rde_community communities;
Move nexthop into struct filterstate (including nexthop flags) and use them whereever possible. In some places (path_update, softreconfig_out handler) the nexthop state is temporarily folded back into the rde_aspath. Tested and OK benno@
2018-07-22 09:59:08 -07:00
struct nexthop *nexthop;
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint8_t nhflags;
Add the validation state to the filterstate struct. Removes vstate argument from rde_filter(). Rename prefix_vstate() to prefix_roa_vstate(). OK tb@
2023-01-11 09:10:25 -08:00
uint8_t vstate;
Introduce a struct filterstate and pass this to rde_filter instead of passing the asp. This is no longer using asp_get() and should be therefor be a bit lighter. Will also allow to put more stuff into the state. OK denis@ sthen@
2018-07-09 07:08:48 -07:00
};
Refactor the code that generates updates so that up_generate_updates is only called in one spot. rde_generate_updates() gets a enum eval_mode argument to discern the different cases. peer_generate_update() uses the eval_mode to skip the update if it is not needed. While there also add an extra AID check in IMSG_REFRESH case to make sure the requested AID is actually available for this peer. OK tb@
2022-07-07 03:46:54 -07:00
enum eval_mode {
EVAL_DEFAULT,
EVAL_ALL,
EVAL_RECONF,
};
Track some (memory) statistics in the RDE. Accessible via bgpctl.
2006-01-03 14:49:17 -08:00
extern struct rde_memstats rdemem;
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
/* prototypes */
Reorder the headerfiles a bit. Move the bgpd specific prototypes out of mrt.h. Now mrt.h includes only protocol specific defines.
2011-09-18 02:31:25 -07:00
/* mrt.c */
Switch the rde_peer hashtable and peer list to a single RB tree. Only the RDE used a hashtable for lookups while the session engine switched from a list to RB tree some time ago. Use peer_foreach() in the mrt code instead of passing the peer list as an argument. OK benno@ tb@
2022-09-01 06:23:24 -07:00
int mrt_dump_v2_hdr(struct mrt *, struct bgpd_config *);
Switch rde_generate_update and rde_send_kroute to accept a struct rib instead of the id. For this we move the rtableid into struct rib. Also move the update code in rib.c up to where the kroute code is. Makes more senses like that.
2017-01-24 16:15:38 -08:00
void mrt_dump_upcall(struct rib_entry *, void *);
Reorder the headerfiles a bit. Move the bgpd specific prototypes out of mrt.h. Now mrt.h includes only protocol specific defines.
2011-09-18 02:31:25 -07:00
Send updates to the main process so they can be added to the kernel routing table. OK henning@
2003-12-23 07:59:02 -08:00
/* rde.c */
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
void rde_update_err(struct rde_peer *, uint8_t , uint8_t,
Start converting the message parser to use the new ibuf api. Rewrite rde_update_dispatch() to use ibufs. Because of this rde_update_err(), rde_get_mp_nexthop(), nlri_get_prefix() and friends are switched to use ibufs. For rde_attr_parse() a minimal change was done for now. OK tb@
2024-01-23 08:13:35 -08:00
struct ibuf *);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
void rde_update_log(const char *, uint16_t,
whitespace found during a read-thru; ok claudio
2022-07-28 06:11:48 -07:00
const struct rde_peer *, const struct bgpd_addr *,
const struct bgpd_addr *, uint8_t);
Improve RIB reload behaviour. Especially when the rtable changes or the route evaluation is modified. In both cases the softreconfig code will now walk the RIB and ensure that everything is in proper sync. Additionally remove 'route-collector yes|no' from the bgpd config, instead use 'rde rib Loc-RIB no evaluate' with the benefit that you can alter the setting now during runtime. Tested and OK benno@
2019-08-07 03:26:41 -07:00
void rde_send_kroute_flush(struct rib *);
Switch rde_generate_update and rde_send_kroute to accept a struct rib instead of the id. For this we move the rtableid into struct rib. Also move the update code in rib.c up to where the kroute code is. Makes more senses like that.
2017-01-24 16:15:38 -08:00
void rde_send_kroute(struct rib *, struct prefix *, struct prefix *);
void rde_send_nexthop(struct bgpd_addr *, int);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
void rde_pftable_add(uint16_t, struct prefix *);
void rde_pftable_del(uint16_t, struct prefix *);
Switch rde_generate_update and rde_send_kroute to accept a struct rib instead of the id. For this we move the rtableid into struct rib. Also move the update code in rib.c up to where the kroute code is. Makes more senses like that.
2017-01-24 16:15:38 -08:00
Introduce 'rde evaluate all' a mode to work around path hiding in IXP route-server environments. By default only the best path is sent to peers and if that path is filtered then the path is hidden for that peer. On route-servers this is sometimes not desried. For this 'rde evaluate all' will cause the evaluation process to fall back to alternate routes and will redistribute the first non-filtered path to the peer. This is very similar to per-peer RIBs but accomplishes the same effect without the massive increase in memory usage. Compared to the default mode this requires more CPU resources but it is probably less than what per-peer RIBs would require. 'rde evaluate all' can be set and reset globally, on groups and on idividual neighbors. It is not limited to route-server configs but route loops are possible if not properly used. OK benno@
2021-03-02 01:45:07 -08:00
int rde_evaluate_all(void);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint32_t rde_local_as(void);
Switch rde_generate_update and rde_send_kroute to accept a struct rib instead of the id. For this we move the rtableid into struct rib. Also move the update code in rib.c up to where the kroute code is. Makes more senses like that.
2017-01-24 16:15:38 -08:00
int rde_decisionflags(void);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
void rde_peer_send_rrefresh(struct rde_peer *, uint8_t, uint8_t);
Move peer related code from rde.c to rde_peer.c. Change peer_foreach() to just walk the peer list instead of iterating over the peer hash table. Also change peer_down() arguments so that it can be used as a peer_foreach() callback (which is then used in rde_shutdown()). OK benno@
2020-01-09 05:31:52 -08:00
int rde_match_peer(struct rde_peer *, struct ctl_neighbor *);
Send updates to the main process so they can be added to the kernel routing table. OK henning@
2003-12-23 07:59:02 -08:00
Instead of processing all imsg when reading them store peer specific messages on a per peer queue. This queue is later processed one at a time resulting in a fairer processing of work and avoiding big table dumps to delay processing of other updates. OK denis@ benno@
2019-12-31 23:25:04 -08:00
/* rde_peer.c */
Rename and move functions used to get per-peer settings to the hopefully better names peer_has_as4byte() and peer_accept_no_as_set(). Move them to rde_peer.c where all other peer functions live. OK sthen@
2021-05-27 07:32:08 -07:00
int peer_has_as4byte(struct rde_peer *);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
int peer_has_add_path(struct rde_peer *, uint8_t, int);
Rename and move functions used to get per-peer settings to the hopefully better names peer_has_as4byte() and peer_accept_no_as_set(). Move them to rde_peer.c where all other peer functions live. OK sthen@
2021-05-27 07:32:08 -07:00
int peer_accept_no_as_set(struct rde_peer *);
Compile the output filter rules into per peer filter rules. especially on route-servers the output filters are in the hot path so reducing the number of rules to check has a big impact. I have seen a 25% to 30% speedup in my big IXP testbench. The output ruleset is applied and copied for each peer during config reload and when a peer is initially added. OK tb@
2023-03-09 23:57:15 -08:00
void peer_init(struct filter_head *);
Move peer related code from rde.c to rde_peer.c. Change peer_foreach() to just walk the peer list instead of iterating over the peer hash table. Also change peer_down() arguments so that it can be used as a peer_foreach() callback (which is then used in rde_shutdown()). OK benno@
2020-01-09 05:31:52 -08:00
void peer_shutdown(void);
void peer_foreach(void (*)(struct rde_peer *, void *), void *);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
struct rde_peer *peer_get(uint32_t);
struct rde_peer *peer_match(struct ctl_neighbor *, uint32_t);
Compile the output filter rules into per peer filter rules. especially on route-servers the output filters are in the hot path so reducing the number of rules to check has a big impact. I have seen a 25% to 30% speedup in my big IXP testbench. The output ruleset is applied and copied for each peer during config reload and when a peer is initially added. OK tb@
2023-03-09 23:57:15 -08:00
struct rde_peer *peer_add(uint32_t, struct peer_config *, struct filter_head *);
struct filter_head *peer_apply_out_filter(struct rde_peer *,
struct filter_head *);
Move peer related code from rde.c to rde_peer.c. Change peer_foreach() to just walk the peer list instead of iterating over the peer hash table. Also change peer_down() arguments so that it can be used as a peer_foreach() callback (which is then used in rde_shutdown()). OK benno@
2020-01-09 05:31:52 -08:00
Pass struct rib_entry to rde_generate_updates() instead of struct rib. With this the newbest and oldbest arguments can go since the infromation is part of the rib_entry. Especially the prefix in the rib_entry is always valid so simplify some code in various functions below to use this information. OK tb@
2023-02-13 10:07:53 -08:00
void rde_generate_updates(struct rib_entry *, struct prefix *,
struct prefix *, enum eval_mode);
Handle IMSG_SESSION_* messages immediatly when received and do not put them on the per peer imsg queue. This is mainly for IMSG_SESSION_DOWN. Delaying the session down can race against IMSG_SESSION_ADD which is handled immediatly and as a result an establised connection may be removed in the RDE because of it. The various graceful restart imsgs need similar treatment for similar reasons. In the end when a session is reset/closed the RDE needs to stop all work and flush the per peer imsg queue. With this only update and route refresh messages are handled via the imsg queue. OK tb@
2022-08-26 07:10:52 -07:00
void peer_up(struct rde_peer *, struct session_up *);
Move peer related code from rde.c to rde_peer.c. Change peer_foreach() to just walk the peer list instead of iterating over the peer hash table. Also change peer_down() arguments so that it can be used as a peer_foreach() callback (which is then used in rde_shutdown()). OK benno@
2020-01-09 05:31:52 -08:00
void peer_down(struct rde_peer *, void *);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
void peer_flush(struct rde_peer *, uint8_t, time_t);
Handle IMSG_SESSION_* messages immediatly when received and do not put them on the per peer imsg queue. This is mainly for IMSG_SESSION_DOWN. Delaying the session down can race against IMSG_SESSION_ADD which is handled immediatly and as a result an establised connection may be removed in the RDE because of it. The various graceful restart imsgs need similar treatment for similar reasons. In the end when a session is reset/closed the RDE needs to stop all work and flush the per peer imsg queue. With this only update and route refresh messages are handled via the imsg queue. OK tb@
2022-08-26 07:10:52 -07:00
void peer_stale(struct rde_peer *, uint8_t, int);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
void peer_dump(struct rde_peer *, uint8_t);
void peer_begin_rrefresh(struct rde_peer *, uint8_t);
Clean up header a bit, remove peer_recv_eor and peer_send_eor prototypes and order prototypes like the functions in rde_peer.c
2020-01-09 06:44:55 -08:00
void peer_imsg_push(struct rde_peer *, struct imsg *);
int peer_imsg_pop(struct rde_peer *, struct imsg *);
int peer_imsg_pending(void);
void peer_imsg_flush(struct rde_peer *);
Instead of processing all imsg when reading them store peer specific messages on a per peer queue. This queue is later processed one at a time resulting in a fairer processing of work and avoiding big table dumps to delay processing of other updates. OK denis@ benno@
2019-12-31 23:25:04 -08:00
Introduce peer_is_up() and use it instead of peer->state == PEER_UP checks also enqueue update and rrfresh imsgs only if the peer is up and flush them once this is no longer the case. OK tb@
2024-08-28 06:21:39 -07:00
static inline int
peer_is_up(struct rde_peer *peer)
{
return (peer->state == PEER_UP);
}
Switch the rde_peer hashtable and peer list to a single RB tree. Only the RDE used a hashtable for lookups while the session engine switched from a list to RB tree some time ago. Use peer_foreach() in the mrt code instead of passing the peer list as an argument. OK benno@ tb@
2022-09-01 06:23:24 -07:00
RB_PROTOTYPE(peer_tree, rde_peer, entry, peer_cmp);
Add basic functions to print aspath in human readable format. OK henning@
2004-02-16 04:58:45 -08:00
/* rde_attr.c */
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
int attr_writebuf(struct ibuf *, uint8_t, uint8_t, void *,
whitespace found during a read-thru; ok claudio
2022-07-28 06:11:48 -07:00
uint16_t);
Cache optional BGP attributes (mostly communities) and use a simple pointer plus a ref counter to link the attributes to the path object. Saves +/- 10M on 11 full feeds. Looks good Henning
2006-01-05 08:00:07 -08:00
void attr_shutdown(void);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
int attr_optadd(struct rde_aspath *, uint8_t, uint8_t,
whitespace found during a read-thru; ok claudio
2022-07-28 06:11:48 -07:00
void *, uint16_t);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
struct attr *attr_optget(const struct rde_aspath *, uint8_t);
Prepare the ground for allowing temporary aspath object living on the stack. To do this path_copy() gets a second argument (dst, src) and a new function path_prep() is introduced to initialize an aspath object. The current path_copy() usage is replaced with path_copy(path_get(), asp) which does the same. Additionally some const where added to the *_copy functions to make it more obvious which is the source and target. Also the pftable_ref() and rtlabel_ref() functions return now the id instead of void. OK sthen@
2018-06-29 04:45:50 -07:00
void attr_copy(struct rde_aspath *, const struct rde_aspath *);
Cache optional BGP attributes (mostly communities) and use a simple pointer plus a ref counter to link the attributes to the path object. Saves +/- 10M on 11 full feeds. Looks good Henning
2006-01-05 08:00:07 -08:00
int attr_compare(struct rde_aspath *, struct rde_aspath *);
void attr_freeall(struct rde_aspath *);
It is no longer allowed to change attributes in place. This corrupts the attribute cache. Instead remove attribute and readd it after beeing changed. This should fix the "att_diff: equal attributes encountered" error seen by Tony Sarendal. OK henning@
2006-04-12 07:05:46 -07:00
void attr_free(struct rde_aspath *, struct attr *);
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
struct aspath *aspath_get(void *, uint16_t);
Instead of a global aspath cache copy the aspath attribute per rde_aspath struct. It uses a bit more memory but improves performance a lot on really big systems because aspath_get() becomes a very hot function. OK tb@
2022-08-29 11:18:55 -07:00
struct aspath *aspath_copy(struct aspath *);
Cleanup aspath specific functions and api. Mainly switch to a refcnt based allocation. This helps to save a bit of RAM. looks good henning@
2004-08-05 11:44:19 -07:00
void aspath_put(struct aspath *);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
u_char *aspath_deflate(u_char *, uint16_t *, int *);
Make bgpd 4-byte AS compatible. All internal representations of AS numbers are now 4-byte instead of the old 2-byte numbers. The only exception are communities because they can not be switched. The RDE will inflate and deflate the ASPATH and AGGREGATOR attributes on demand and create the NEW_ASPATH and NEW_AGGREGATOR field whenever needed. Both old and new stile sessions are supported and can be mixed. Currently new stile sessions with the 4-byte AS number capability turned on are only enabled if one of the AS numbers involved is a 4-byte one. This is based on an initial diff by Geoff Huston gih (at) apnic (dot) net Cleanup, testing and bug-fixes by myself (via AS 3.10). Currently mrt table dumps are producing incompatible output this will be fixed afterwards -- this diff is already big enough. "get it in if you think it is ready" henning@
2007-04-23 06:04:24 -07:00
void aspath_merge(struct rde_aspath *, struct attr *);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint32_t aspath_neighbor(struct aspath *);
int aspath_loopfree(struct aspath *, uint32_t);
RDE update generation. First we queue all updates and withdraws on a per peer basis. A queue runner will dequeue and package those messages to valid bgp UPDATE messages and send them to the SE. Not yet done is per peer type attribute handling (like aspath prepends and nexthop modifications) and the queue runner could be a tad smarter. All in all this gives us a good starting point for the missing parts. OK henning@
2004-01-10 08:20:29 -08:00
int aspath_compare(struct aspath *, struct aspath *);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
int aspath_match(struct aspath *, struct filter_as *, uint32_t);
u_char *aspath_prepend(struct aspath *, uint32_t, int, uint16_t *);
u_char *aspath_override(struct aspath *, uint32_t, uint32_t,
uint16_t *);
Implement two new filters, max-as-len and max-as-seq. The first is limiting the length of an AS path (matches if the path is longer then the specified lenght) the second matches when a sequence of the same AS number is longer then the specified length). max-as-len is good to protect crappy comercial bgp boxes from other crappy comercial bgp boxes. max-as-seq was a feature request from SwissIX and maybe EuroIX to find and filter prepends. Additinal testing and OK sthen@
2010-05-17 09:08:20 -07:00
int aspath_lenmatch(struct aspath *, enum aslen_spec, u_int);
Start reworking community handling. Merge standard communities and large communities into one filter_community struct and allow it that more then one community can be used in filter rules (currently up to 3). Also rework the code handling bgpctl show rib commands. The special IMSG types for the various filters are gone and the code is in general simpler. OK job@, phessler@
2018-11-28 00:32:26 -08:00
Move some basic accessors of aspath to rde.h and make them static inline. OK tb@
2022-12-14 04:37:15 -08:00
static inline u_char *
aspath_dump(struct aspath *aspath)
{
return (aspath->data);
}
static inline uint16_t
aspath_length(struct aspath *aspath)
{
return (aspath->len);
}
static inline uint32_t
aspath_origin(struct aspath *aspath)
{
return (aspath->source_as);
}
Change the Adj-RIB-Out to a per peer set of RB trees. The way RIB data structures are linked does not scale for the Adj-RIB-Out and so inserts and updates into the Adj-RIB-Out did not scale because of some linear list traversals in hot paths. A synthetic test with 4000 peers announcing one prefix each showed that the initial convergence time dropped from around 1 hout to around 6min. Note: because the Adj-RIB-Out is now per peer the order in which prefixes are dumped in 'bgpctl show rib out' changed. Tested and OK job@, benno@, phessler@
2019-07-17 03:13:26 -07:00
/* rde_community.c */
mrt dumps lost communities after the community rewrite. Readd them by dumping them explicitly. Tested by and OK benno@
2019-06-23 23:39:49 -07:00
int community_match(struct rde_community *, struct community *,
Start reworking community handling. Merge standard communities and large communities into one filter_community struct and allow it that more then one community can be used in filter rules (currently up to 3). Also rework the code handling bgpctl show rib commands. The special IMSG types for the various filters are gone and the code is in general simpler. OK job@, phessler@
2018-11-28 00:32:26 -08:00
struct rde_peer *);
Implement a max communities filter match When max-communities X is set on a filterrule the filter will match when more than X communities are present in the path. In other words max-communities 0 means no communities are allowed and max-communities 3 limits it up to 3 communities. There is max-communities, max-ext-communities and max-large-communities for each of the 3 community attributes. These three max checks can be used together. OK tb@ job@
2022-05-31 02:45:33 -07:00
int community_count(struct rde_community *, uint8_t type);
mrt dumps lost communities after the community rewrite. Readd them by dumping them explicitly. Tested by and OK benno@
2019-06-23 23:39:49 -07:00
int community_set(struct rde_community *, struct community *,
Start reworking community handling. Merge standard communities and large communities into one filter_community struct and allow it that more then one community can be used in filter rules (currently up to 3). Also rework the code handling bgpctl show rib commands. The special IMSG types for the various filters are gone and the code is in general simpler. OK job@, phessler@
2018-11-28 00:32:26 -08:00
struct rde_peer *);
mrt dumps lost communities after the community rewrite. Readd them by dumping them explicitly. Tested by and OK benno@
2019-06-23 23:39:49 -07:00
void community_delete(struct rde_community *, struct community *,
Start reworking community handling. Merge standard communities and large communities into one filter_community struct and allow it that more then one community can be used in filter rules (currently up to 3). Also rework the code handling bgpctl show rib commands. The special IMSG types for the various filters are gone and the code is in general simpler. OK job@, phessler@
2018-11-28 00:32:26 -08:00
struct rde_peer *);
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
Convert the community parsers to the new ibuf api. This converts community_add(), community_large_add() and community_ext_add() and as a result removes some hacks from rde_attr_add() and rde_attr_parse(). OK tb@
2024-01-24 06:51:11 -08:00
int community_add(struct rde_community *, int, struct ibuf *);
int community_large_add(struct rde_community *, int, struct ibuf *);
int community_ext_add(struct rde_community *, int, int, struct ibuf *);
Update OpenBGPD to use new ibuf API. This replaces the old way of using a static buffer and a len to build UPDATEs with a pure ibuf solution. The result is much cleaner and a lot of almost duplicate code can be removed because often a version for ibufs and one for this static buffer was implemented (e.g. for mrt or bgpctl). With and OK tb@
2023-07-12 07:45:42 -07:00
int community_writebuf(struct rde_community *, uint8_t, int, struct ibuf *);
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
void communities_shutdown(void);
struct rde_community *communities_lookup(struct rde_community *);
struct rde_community *communities_link(struct rde_community *);
Cleanup, remove some unneded spaces add some other where needed. No binary change according to clang
2019-06-17 06:35:42 -07:00
void communities_unlink(struct rde_community *);
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
int communities_equal(struct rde_community *, struct rde_community *);
void communities_copy(struct rde_community *, struct rde_community *);
void communities_clean(struct rde_community *);
static inline struct rde_community *
Unify the way object in the RDE are reference counted. The affected structures are pt_entry, rde_aspath, rde_communities, and nexthop. The functions are always called *_ref and *_unref also the behaviour when the last reference is removed is unified and now the object is removed inside of the unref function. The actual bean-counting is not modified by this diff. OK benno@
2019-07-01 00:07:08 -07:00
communities_ref(struct rde_community *comm)
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
{
if (comm->refcnt == 0)
fatalx("%s: not-referenced community", __func__);
comm->refcnt++;
rdemem.comm_refs++;
return comm;
}
static inline void
Unify the way object in the RDE are reference counted. The affected structures are pt_entry, rde_aspath, rde_communities, and nexthop. The functions are always called *_ref and *_unref also the behaviour when the last reference is removed is unified and now the object is removed inside of the unref function. The actual bean-counting is not modified by this diff. OK benno@
2019-07-01 00:07:08 -07:00
communities_unref(struct rde_community *comm)
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
{
if (comm == NULL)
return;
rdemem.comm_refs--;
Unify the way object in the RDE are reference counted. The affected structures are pt_entry, rde_aspath, rde_communities, and nexthop. The functions are always called *_ref and *_unref also the behaviour when the last reference is removed is unified and now the object is removed inside of the unref function. The actual bean-counting is not modified by this diff. OK benno@
2019-07-01 00:07:08 -07:00
if (--comm->refcnt == 1) /* last ref is hold internally */
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
communities_unlink(comm);
}
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
int community_to_rd(struct community *, uint64_t *);
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
Reorder the headerfiles a bit. Move the bgpd specific prototypes out of mrt.h. Now mrt.h includes only protocol specific defines.
2011-09-18 02:31:25 -07:00
/* rde_decide.c */
Remove the active prefix cache in struct rib_entry. I need the space and it also makes less sense to track this with ECMP or add-path. Replace the re->active access with prefix_best(re) which does the check on the spot. Feedback and OK tb@
2022-03-21 10:35:56 -07:00
int prefix_eligible(struct prefix *);
struct prefix *prefix_best(struct rib_entry *);
void prefix_evaluate(struct rib_entry *, struct prefix *,
whitespace found during a read-thru; ok claudio
2022-07-28 06:11:48 -07:00
struct prefix *);
Properly handle nexthop state changes in the decision process In rev 1.90 of rde_decide.c the re->active cache of the best prefix was replaced with a call to prefix_best(). This introduced a bug because the nexthop state at that time may have changed already. As a result when a nexthop became unreachable prefix_evaluate() had oldbest = NULL and newbest = NULL and did not withdraw the prefix from FIB and Adj-RIB-Out. To fix this store the nexthop state per prefix and introduce prefix_evaluate_nexthop() which removes the prefix from the decision list, updates the nexthop state of the prefix and reinserts the prefix. Doing this ensures that prefix_best() always reports the same result once the decison process is done. prefix_best() and prefix_eligible() only depend on data stored on the prefix itself. OK tb@
2022-07-25 09:37:55 -07:00
void prefix_evaluate_nexthop(struct prefix *, enum nexthop_state,
whitespace found during a read-thru; ok claudio
2022-07-28 06:11:48 -07:00
enum nexthop_state);
Reorder the headerfiles a bit. Move the bgpd specific prototypes out of mrt.h. Now mrt.h includes only protocol specific defines.
2011-09-18 02:31:25 -07:00
/* rde_filter.c */
Instead of passing a struct prefix pointer to rde_filter() pass the 4 values prefix_peer, prefix_vstate and prefix/prefixlen to the function. This removes some ugly hacks in cases where the prefix was not available. Also adjust the order of arguments of rde_attr_set() to match rde_filter(). OK benno@
2019-08-13 05:16:20 -07:00
void rde_apply_set(struct filter_set_head *, struct rde_peer *,
Compile the output filter rules into per peer filter rules. especially on route-servers the output filters are in the hot path so reducing the number of rules to check has a big impact. I have seen a 25% to 30% speedup in my big IXP testbench. The output ruleset is applied and copied for each peer during config reload and when a peer is initially added. OK tb@
2023-03-09 23:57:15 -08:00
struct rde_peer *, struct filterstate *, u_int8_t);
Split rde_filterstate_prep() into three functions. - rde_filterstate_init(): initialize a filterstate to default values - rde_filterstate_copy(): copy from a filterstate into a new state object - rde_filterstate_prep(): set filtersate based on prefix passed as argument. This makes the code a bit easier to read. OK tb@
2023-01-12 09:35:51 -08:00
void rde_filterstate_init(struct filterstate *);
void rde_filterstate_prep(struct filterstate *, struct prefix *);
void rde_filterstate_copy(struct filterstate *, struct filterstate *);
Implement ASPA validation and reload logic on ASPA set changes. For this use the validation state (vstate) in struct prefix and struct filterstate to store both the ASPA and ROA validity. Introduce helper functions to set and get the various states for struct prefix and make sure struct filterstate is also setup properly. Change the ASPA state in rde_aspath to be AFI/AID and role independent by storing all 4 possible outcomes. Also add a ASPA generation count which is used to update the rde_aspath ASPA state cache on reloads. Rework the rde_aspa.c code to be AFI/AID and role independent. Doing this for roles is trivial but AFI switch goes deep and is so unnecessary. The reload is combined with the ROA reload logic and renamed to RPKI softreload. OK tb@
2023-01-24 03:28:41 -08:00
void rde_filterstate_set_vstate(struct filterstate *, uint8_t, uint8_t);
Instead of passing a struct prefix pointer to rde_filter() pass the 4 values prefix_peer, prefix_vstate and prefix/prefixlen to the function. This removes some ugly hacks in cases where the prefix was not available. Also adjust the order of arguments of rde_attr_set() to match rde_filter(). OK benno@
2019-08-13 05:16:20 -07:00
void rde_filterstate_clean(struct filterstate *);
Compile the output filter rules into per peer filter rules. especially on route-servers the output filters are in the hot path so reducing the number of rules to check has a big impact. I have seen a 25% to 30% speedup in my big IXP testbench. The output ruleset is applied and copied for each peer during config reload and when a peer is initially added. OK tb@
2023-03-09 23:57:15 -08:00
int rde_filter_skip_rule(struct rde_peer *, struct filter_rule *);
int rde_filter_equal(struct filter_head *, struct filter_head *);
Instead of passing a struct prefix pointer to rde_filter() pass the 4 values prefix_peer, prefix_vstate and prefix/prefixlen to the function. This removes some ugly hacks in cases where the prefix was not available. Also adjust the order of arguments of rde_attr_set() to match rde_filter(). OK benno@
2019-08-13 05:16:20 -07:00
void rde_filter_calc_skip_steps(struct filter_head *);
Pass a struct prefix to rde_filter and remove some of the arguments which are now fetched from struct prefix. Currently some trickery is needed for export default-route but this will get better once Adj-RIB-Out comes. OK benno@
2018-06-28 01:07:21 -07:00
enum filter_actions rde_filter(struct filter_head *, struct rde_peer *,
Add the validation state to the filterstate struct. Removes vstate argument from rde_filter(). Rename prefix_vstate() to prefix_roa_vstate(). OK tb@
2023-01-11 09:10:25 -08:00
struct rde_peer *, struct bgpd_addr *, uint8_t,
Instead of passing a struct prefix pointer to rde_filter() pass the 4 values prefix_peer, prefix_vstate and prefix/prefixlen to the function. This removes some ugly hacks in cases where the prefix was not available. Also adjust the order of arguments of rde_attr_set() to match rde_filter(). OK benno@
2019-08-13 05:16:20 -07:00
struct filterstate *);
Reorder the headerfiles a bit. Move the bgpd specific prototypes out of mrt.h. Now mrt.h includes only protocol specific defines.
2011-09-18 02:31:25 -07:00
/* rde_prefix.c */
Unify the way object in the RDE are reference counted. The affected structures are pt_entry, rde_aspath, rde_communities, and nexthop. The functions are always called *_ref and *_unref also the behaviour when the last reference is removed is unified and now the object is removed inside of the unref function. The actual bean-counting is not modified by this diff. OK benno@
2019-07-01 00:07:08 -07:00
void pt_init(void);
void pt_shutdown(void);
void pt_getaddr(struct pt_entry *, struct bgpd_addr *);
Extend the pt_entry api to handle flowspec. Introduce pt_get_flow() and pt_add_flow() to lookup and insert flowspec objects. Add pt_getflowspec() which works somewhat similar to pt_getaddr() to extract the flowspec NLRI from a pt_entry. Make pt_getaddr() to return the destination prefix of the flowspec rule and handle flowspec in pt_write(). OK tb@
2023-04-19 00:09:47 -07:00
int pt_getflowspec(struct pt_entry *, uint8_t **);
Unify the way object in the RDE are reference counted. The affected structures are pt_entry, rde_aspath, rde_communities, and nexthop. The functions are always called *_ref and *_unref also the behaviour when the last reference is removed is unified and now the object is removed inside of the unref function. The actual bean-counting is not modified by this diff. OK benno@
2019-07-01 00:07:08 -07:00
struct pt_entry *pt_fill(struct bgpd_addr *, int);
struct pt_entry *pt_get(struct bgpd_addr *, int);
struct pt_entry *pt_add(struct bgpd_addr *, int);
Extend the pt_entry api to handle flowspec. Introduce pt_get_flow() and pt_add_flow() to lookup and insert flowspec objects. Add pt_getflowspec() which works somewhat similar to pt_getaddr() to extract the flowspec NLRI from a pt_entry. Make pt_getaddr() to return the destination prefix of the flowspec rule and handle flowspec in pt_write(). OK tb@
2023-04-19 00:09:47 -07:00
struct pt_entry *pt_get_flow(struct flowspec *);
struct pt_entry *pt_add_flow(struct flowspec *);
Unify the way object in the RDE are reference counted. The affected structures are pt_entry, rde_aspath, rde_communities, and nexthop. The functions are always called *_ref and *_unref also the behaviour when the last reference is removed is unified and now the object is removed inside of the unref function. The actual bean-counting is not modified by this diff. OK benno@
2019-07-01 00:07:08 -07:00
void pt_remove(struct pt_entry *);
struct pt_entry *pt_lookup(struct bgpd_addr *);
int pt_prefix_cmp(const struct pt_entry *, const struct pt_entry *);
Update OpenBGPD to use new ibuf API. This replaces the old way of using a static buffer and a len to build UPDATEs with a pure ibuf solution. The result is much cleaner and a lot of almost duplicate code can be removed because often a version for ibufs and one for this static buffer was implemented (e.g. for mrt or bgpctl). With and OK tb@
2023-07-12 07:45:42 -07:00
int pt_writebuf(struct ibuf *, struct pt_entry *, int, int, uint32_t);
Change struct prefix a bit. Make the aspath pointer opaque and remove the direct pointer to the pt_prefix struct. To getter functions prefix_aspath() and prefix_peer() added to make access of the opaque pointers possible. Looks good henning@ and benno@
2018-02-04 19:55:54 -08:00
Add a direct pointer from struct prefix to struct pt_entry. This changes makes it possible to not use the struct rib_entry pointer which will be used to optimize the Adj-RIB-Out. Also adjust pt_ref() and pt_unref() so that the code can be written a bit more compact. Also prefix_cmp() no longer needs to go via rib_compare() and calls pt_prefix_cmp() directly. OK phessler@
2019-06-21 22:44:05 -07:00
static inline struct pt_entry *
Change struct prefix a bit. Make the aspath pointer opaque and remove the direct pointer to the pt_prefix struct. To getter functions prefix_aspath() and prefix_peer() added to make access of the opaque pointers possible. Looks good henning@ and benno@
2018-02-04 19:55:54 -08:00
pt_ref(struct pt_entry *pt)
{
++pt->refcnt;
if (pt->refcnt == 0)
fatalx("pt_ref: overflow");
Add a direct pointer from struct prefix to struct pt_entry. This changes makes it possible to not use the struct rib_entry pointer which will be used to optimize the Adj-RIB-Out. Also adjust pt_ref() and pt_unref() so that the code can be written a bit more compact. Also prefix_cmp() no longer needs to go via rib_compare() and calls pt_prefix_cmp() directly. OK phessler@
2019-06-21 22:44:05 -07:00
return pt;
Change struct prefix a bit. Make the aspath pointer opaque and remove the direct pointer to the pt_prefix struct. To getter functions prefix_aspath() and prefix_peer() added to make access of the opaque pointers possible. Looks good henning@ and benno@
2018-02-04 19:55:54 -08:00
}
Unify the way object in the RDE are reference counted. The affected structures are pt_entry, rde_aspath, rde_communities, and nexthop. The functions are always called *_ref and *_unref also the behaviour when the last reference is removed is unified and now the object is removed inside of the unref function. The actual bean-counting is not modified by this diff. OK benno@
2019-07-01 00:07:08 -07:00
static inline void
Change struct prefix a bit. Make the aspath pointer opaque and remove the direct pointer to the pt_prefix struct. To getter functions prefix_aspath() and prefix_peer() added to make access of the opaque pointers possible. Looks good henning@ and benno@
2018-02-04 19:55:54 -08:00
pt_unref(struct pt_entry *pt)
{
if (pt->refcnt == 0)
fatalx("pt_unref: underflow");
Unify the way object in the RDE are reference counted. The affected structures are pt_entry, rde_aspath, rde_communities, and nexthop. The functions are always called *_ref and *_unref also the behaviour when the last reference is removed is unified and now the object is removed inside of the unref function. The actual bean-counting is not modified by this diff. OK benno@
2019-07-01 00:07:08 -07:00
if (--pt->refcnt == 0)
pt_remove(pt);
Change struct prefix a bit. Make the aspath pointer opaque and remove the direct pointer to the pt_prefix struct. To getter functions prefix_aspath() and prefix_peer() added to make access of the opaque pointers possible. Looks good henning@ and benno@
2018-02-04 19:55:54 -08:00
}
Reorder the headerfiles a bit. Move the bgpd specific prototypes out of mrt.h. Now mrt.h includes only protocol specific defines.
2011-09-18 02:31:25 -07:00
Add basic functions to print aspath in human readable format. OK henning@
2004-02-16 04:58:45 -08:00
/* rde_rib.c */
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
extern uint16_t rib_size;
Rework most of the RDE to allow multiple RIBs. This is mostly preparation work by changing the way the RDE DB is built. struct prefix and struct pt_entry are simplified and extended with a rib_entry where the decision tree is run on. From now on a prefix can only reside on one particular RIB which simplifies the code a bit. Currently there are two fixed ribs (adj-rib-in and the local-rib) which needs to be made more dynamic in upcomming commits. This is work in progress, the RDE seems to work for me and sthen@ (no flames comming out of our testrouters but there is still a lot missing) Move into the tree to simplify developement -- henning@
2009-05-17 05:25:15 -07:00
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
struct rib *rib_new(char *, u_int, uint16_t);
Adjust how RIB are reloaded when their flags (esp. no evaluate) changes. First flush all affected Adj-RIB-Out and then in a second step re-evaluate the RIB itself. The no evaluate case becomes simpler. Fix the way prefixes are re-evaluated, the list remove needs to be explict and not part of prefix_evaluate() as in most other cases since this list is not part of the rib_entry. OK tb@
2022-03-21 06:33:20 -07:00
int rib_update(struct rib *);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
struct rib *rib_byid(uint16_t);
uint16_t rib_find(char *);
Hopefully the last of the struct rib rototilling. Peer just points to a struct rib and not rib_desc since the full descriptor is almost never needed. This should now allow the update code to be changed.
2017-01-24 19:21:55 -08:00
void rib_free(struct rib *);
Introduce a real Adj-RIB-Out. At the same time remove the update_rib introduced before 6.4 because it now can be replaced with the real RIB. Main changes are: - simplified 'show rib' handling since everything is now a real RIB - path_update() is now returning if a prefix was not modified, added or moved - softreconfig out case is simpler since path_update does all the magic now - Adjust shutdown code to work with the Adj-RIB-Out Tested and OK denis@, benno@
2018-11-04 04:34:54 -08:00
void rib_shutdown(void);
Pass a pt_entry pointer to rib_get() and rib_add(). Add rib_get_addr() to behave like rib_get() did before. OK tb@
2023-04-07 06:49:03 -07:00
struct rib_entry *rib_get(struct rib *, struct pt_entry *);
struct rib_entry *rib_get_addr(struct rib *, struct bgpd_addr *, int);
Change the Adj-RIB-Out to a per peer set of RB trees. The way RIB data structures are linked does not scale for the Adj-RIB-Out and so inserts and updates into the Adj-RIB-Out did not scale because of some linear list traversals in hot paths. A synthetic test with 4000 peers announcing one prefix each showed that the initial convergence time dropped from around 1 hout to around 6min. Note: because the Adj-RIB-Out is now per peer the order in which prefixes are dumped in 'bgpctl show rib out' changed. Tested and OK job@, benno@, phessler@
2019-07-17 03:13:26 -07:00
struct rib_entry *rib_match(struct rib *, struct bgpd_addr *);
Major refactoring of the RIB handling code. Mainly change how the RIB is walked. rib_dump_r() is now an internal function and instead the code gets an additional callback for throttling the rib_dump code. This removes a lot of similar code used to make sure the RDE is not walking to fast and replaces it with simpler callbacks. The other big change is the removal of struct rib pointers in other data structures. The rib pointers are not stable because of a realloc() call happening when extending the array so instead use the RIB ID as a reference. Tested and OK denis@ and benno@
2018-10-24 01:26:37 -07:00
int rib_dump_pending(void);
void rib_dump_runner(void);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
int rib_dump_new(uint16_t, uint8_t, unsigned int, void *,
Cleanup, remove some unneded spaces add some other where needed. No binary change according to clang
2019-06-17 06:35:42 -07:00
void (*)(struct rib_entry *, void *),
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
void (*)(void *, uint8_t),
Major refactoring of the RIB handling code. Mainly change how the RIB is walked. rib_dump_r() is now an internal function and instead the code gets an additional callback for throttling the rib_dump code. This removes a lot of similar code used to make sure the RDE is not walking to fast and replaces it with simpler callbacks. The other big change is the removal of struct rib pointers in other data structures. The rib pointers are not stable because of a realloc() call happening when extending the array so instead use the RIB ID as a reference. Tested and OK denis@ and benno@
2018-10-24 01:26:37 -07:00
int (*)(void *));
Introduce tree walkers that only walk a subtree of the RIB. In some cases only a "small" part of the RIB needs to be looked at. Like bgpctl show rib 10/8 or-longer that only needs to travers nodes under 10/8 all other RIB entries do not matter. By setting the start node to the RB_NFIND(10/8) the all nodes below this point can be skipped. Using prefix_compare() while walking the tree with RB_NEXT() the walker know when it steps outside of the 10/8 subtree and stops. With this the or-longer commands become a lot faster. Looks good to tb@
2022-09-12 03:03:17 -07:00
int rib_dump_subtree(uint16_t, struct bgpd_addr *, uint8_t,
unsigned int count, void *arg,
void (*)(struct rib_entry *, void *),
void (*)(void *, uint8_t),
int (*)(void *));
Change the Adj-RIB-Out to a per peer set of RB trees. The way RIB data structures are linked does not scale for the Adj-RIB-Out and so inserts and updates into the Adj-RIB-Out did not scale because of some linear list traversals in hot paths. A synthetic test with 4000 peers announcing one prefix each showed that the initial convergence time dropped from around 1 hout to around 6min. Note: because the Adj-RIB-Out is now per peer the order in which prefixes are dumped in 'bgpctl show rib out' changed. Tested and OK job@, benno@, phessler@
2019-07-17 03:13:26 -07:00
void rib_dump_terminate(void *);
Rework most of the RDE to allow multiple RIBs. This is mostly preparation work by changing the way the RDE DB is built. struct prefix and struct pt_entry are simplified and extended with a rib_entry where the decision tree is run on. From now on a prefix can only reside on one particular RIB which simplifies the code a bit. Currently there are two fixed ribs (adj-rib-in and the local-rib) which needs to be made more dynamic in upcomming commits. This is work in progress, the RDE seems to work for me and sthen@ (no flames comming out of our testrouters but there is still a lot missing) Move into the tree to simplify developement -- henning@
2009-05-17 05:25:15 -07:00
Implement a way to announce flowspec rules without hitting Adj-RIB-In and Loc-RIB. Flowspec objects are collected in a single flowrib RIB and then directly distributed into the various Adj-RIB-Outs. For this to work add a bypass in the filter logic (flowspec AFI/SAFI are currently accepted without any rule). The filter language lacks a way to allow prefixes based on AFI/SAFI which is the minimum needed. OK tb@
2023-04-19 06:23:33 -07:00
extern struct rib flowrib;
Save some space in struct rib_entry so it is back to 64bytes (on 64bit archs). Doing this by folding the lock flag into a pointer and providing an accessor function for the rib pointer. This is an acceptable middle path for this important structure. OK benno@ on an earlier version
2017-01-24 15:38:12 -08:00
static inline struct rib *
re_rib(struct rib_entry *re)
{
Implement a way to announce flowspec rules without hitting Adj-RIB-In and Loc-RIB. Flowspec objects are collected in a single flowrib RIB and then directly distributed into the various Adj-RIB-Outs. For this to work add a bypass in the filter logic (flowspec AFI/SAFI are currently accepted without any rule). The filter language lacks a way to allow prefixes based on AFI/SAFI which is the minimum needed. OK tb@
2023-04-19 06:23:33 -07:00
if (re->prefix->aid == AID_FLOWSPECv4 ||
re->prefix->aid == AID_FLOWSPECv6)
return &flowrib;
Major refactoring of the RIB handling code. Mainly change how the RIB is walked. rib_dump_r() is now an internal function and instead the code gets an additional callback for throttling the rib_dump code. This removes a lot of similar code used to make sure the RDE is not walking to fast and replaces it with simpler callbacks. The other big change is the removal of struct rib pointers in other data structures. The rib pointers are not stable because of a realloc() call happening when extending the array so instead use the RIB ID as a reference. Tested and OK denis@ and benno@
2018-10-24 01:26:37 -07:00
return rib_byid(re->rib_id);
Save some space in struct rib_entry so it is back to 64bytes (on 64bit archs). Doing this by folding the lock flag into a pointer and providing an accessor function for the rib pointer. This is an acceptable middle path for this important structure. OK benno@ on an earlier version
2017-01-24 15:38:12 -08:00
}
Shutdown the RDE cleanly on exit. Plug some memleaks. OK henning@
2004-03-11 06:22:22 -08:00
void path_shutdown(void);
Prepare the ground for allowing temporary aspath object living on the stack. To do this path_copy() gets a second argument (dst, src) and a new function path_prep() is introduced to initialize an aspath object. The current path_copy() usage is replaced with path_copy(path_get(), asp) which does the same. Additionally some const where added to the *_copy functions to make it more obvious which is the source and target. Also the pftable_ref() and rtlabel_ref() functions return now the id instead of void. OK sthen@
2018-06-29 04:45:50 -07:00
struct rde_aspath *path_copy(struct rde_aspath *, const struct rde_aspath *);
struct rde_aspath *path_prep(struct rde_aspath *);
Monster diff to get one step closer to IPv6 support. Cleanup path attribute handling. First of all kill struct attr_flags, all those infos are now in struct rde_aspath. Second move attribute parser functions into rde.c, rde_attr.c is shared between bgpd and bgpctl. Third reimplementation of the nexthop handling. Make it IPv6 ready and fix some major bug relating to "set nexthop". henning@ OK if it breaks nothing
2004-08-06 05:04:08 -07:00
struct rde_aspath *path_get(void);
Introduce a struct filterstate and pass this to rde_filter instead of passing the asp. This is no longer using asp_get() and should be therefor be a bit lighter. Will also allow to put more stuff into the state. OK denis@ sthen@
2018-07-09 07:08:48 -07:00
void path_clean(struct rde_aspath *);
Monster diff to get one step closer to IPv6 support. Cleanup path attribute handling. First of all kill struct attr_flags, all those infos are now in struct rde_aspath. Second move attribute parser functions into rde.c, rde_attr.c is shared between bgpd and bgpctl. Third reimplementation of the nexthop handling. Make it IPv6 ready and fix some major bug relating to "set nexthop". henning@ OK if it breaks nothing
2004-08-06 05:04:08 -07:00
void path_put(struct rde_aspath *);
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
Fix minor issues with IPv6 dumps and add a function for dumping the RIB table protocol independent. This new dump format is not (yet) supported by the mrtd route_btoa tool. OK henning@
2004-08-13 07:03:20 -07:00
#define PREFIX_SIZE(x) (((x) + 7) / 8 + 1)
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
struct prefix *prefix_get(struct rib *, struct rde_peer *, uint32_t,
Clean up prefix flag handling. First of all the dynamic networks no longer need this and are now treated equally to the network statement in the config. This makes bgpctl network delete <net> also remove a network which was defined in the config. While there remove the other use of flag which was done to support Adj-RIB-Out but the direction we're taking is no longer needing that. Makes code simpler again. OK benno@
2018-09-09 05:33:51 -07:00
struct bgpd_addr *, int);
For add-path send the Adj-RIB-Out needs to handle multiple paths per prefix. For this extend the RB trees of the Adj-RIB-Out to also consider the path_id. Add functions to lookup a prefix without path_id so that bgpctl works. Rename functions so that all Adj-RIB-Out specific functions start with prefix_adjout_ For now the path_id_tx in the Adj-RIB-Out is forced to 0 since up_generate_updates() is not ready to handle more than one path per prefix. OK tb@
2022-02-25 03:36:54 -08:00
struct prefix *prefix_adjout_get(struct rde_peer *, uint32_t,
Switch prefix_adjout_get and new prefix_adjout_first to use a pt_entry as argument instead of the bgpd_addr + prefixlen. Do the same with prefix_adjout_update but leave prefix_adjout_lookup and prefix_adjout_match since those are used by bgpctl code that does not use pt_entry structs. With this most of the update code no longer needs struct bgpd_addr and pt_getaddr(). OK tb@
2023-03-29 03:46:11 -07:00
struct pt_entry *);
struct prefix *prefix_adjout_first(struct rde_peer *, struct pt_entry *);
struct prefix *prefix_adjout_next(struct rde_peer *, struct prefix *);
For add-path send the Adj-RIB-Out needs to handle multiple paths per prefix. For this extend the RB trees of the Adj-RIB-Out to also consider the path_id. Add functions to lookup a prefix without path_id so that bgpctl works. Rename functions so that all Adj-RIB-Out specific functions start with prefix_adjout_ For now the path_id_tx in the Adj-RIB-Out is forced to 0 since up_generate_updates() is not ready to handle more than one path per prefix. OK tb@
2022-02-25 03:36:54 -08:00
struct prefix *prefix_adjout_lookup(struct rde_peer *, struct bgpd_addr *,
whitespace found during a read-thru; ok claudio
2022-07-28 06:11:48 -07:00
int);
Switch prefix_adjout_get and new prefix_adjout_first to use a pt_entry as argument instead of the bgpd_addr + prefixlen. Do the same with prefix_adjout_update but leave prefix_adjout_lookup and prefix_adjout_match since those are used by bgpctl code that does not use pt_entry structs. With this most of the update code no longer needs struct bgpd_addr and pt_getaddr(). OK tb@
2023-03-29 03:46:11 -07:00
struct prefix *prefix_adjout_match(struct rde_peer *, struct bgpd_addr *);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
int prefix_update(struct rib *, struct rde_peer *, uint32_t,
Introduce 'rde rib Loc-RIB include filtered' a feature that includes filtered prefixes in the Loc-RIB This includes filtered prefixes into the Loc-RIB but they are marked ineligible so nothing will select them but it is possible to show them in bgpctl. So 'bppctl show rib filtered' will return all prefixes filtered out by the input filters. OK tb@
2024-08-14 12:09:51 -07:00
uint32_t, struct filterstate *, int, struct bgpd_addr *,
int);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
int prefix_withdraw(struct rib *, struct rde_peer *, uint32_t,
Clean up prefix flag handling. First of all the dynamic networks no longer need this and are now treated equally to the network statement in the config. This makes bgpctl network delete <net> also remove a network which was defined in the config. While there remove the other use of flag which was done to support Adj-RIB-Out but the direction we're taking is no longer needing that. Makes code simpler again. OK benno@
2018-09-09 05:33:51 -07:00
struct bgpd_addr *, int);
Implement a way to announce flowspec rules without hitting Adj-RIB-In and Loc-RIB. Flowspec objects are collected in a single flowrib RIB and then directly distributed into the various Adj-RIB-Outs. For this to work add a bypass in the filter logic (flowspec AFI/SAFI are currently accepted without any rule). The filter language lacks a way to allow prefixes based on AFI/SAFI which is the minimum needed. OK tb@
2023-04-19 06:23:33 -07:00
int prefix_flowspec_update(struct rde_peer *, struct filterstate *,
struct pt_entry *, uint32_t);
int prefix_flowspec_withdraw(struct rde_peer *, struct pt_entry *);
void prefix_flowspec_dump(uint8_t, void *,
void (*)(struct rib_entry *, void *),
void (*)(void *, uint8_t));
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
void prefix_add_eor(struct rde_peer *, uint8_t);
Pass path_id_tx to the Adj-RIB-Out Adjust prefix_adjout_update() to properly handle path_id_tx. Move the lookup of the prefix out of prefix_adjout_update() and to up_generate_updates(). While that code uses prefix_adjout_lookup() to find the current prefix in the Adj-RIB-Out and add-path aware function will use prefix_adjout_get(). In up_generate_default() just use 0 for path_id_tx since for this peer that is the only prefix installed into the Adj-RIB-Out. OK tb@
2022-07-08 03:01:52 -07:00
void prefix_adjout_update(struct prefix *, struct rde_peer *,
Switch prefix_adjout_get and new prefix_adjout_first to use a pt_entry as argument instead of the bgpd_addr + prefixlen. Do the same with prefix_adjout_update but leave prefix_adjout_lookup and prefix_adjout_match since those are used by bgpctl code that does not use pt_entry structs. With this most of the update code no longer needs struct bgpd_addr and pt_getaddr(). OK tb@
2023-03-29 03:46:11 -07:00
struct filterstate *, struct pt_entry *, uint32_t);
Refactor prefix_adjout_withdraw() Just pass the prefix to be withdrawn to the function and move the lookup up. Adjust how the various accounting vars are updated so that the values are decremented in the right cases. Do the same accounting dance for prefix_adjout_destroy(). Adjust rde_up_flush_upcall() to directly call prefix_adjout_withdraw() without calling it via up_generate_updates(). OK tb@
2022-03-02 06:44:46 -08:00
void prefix_adjout_withdraw(struct prefix *);
Correct prefix_adjout_destroy() prototype
2022-03-02 06:49:25 -08:00
void prefix_adjout_destroy(struct prefix *);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
int prefix_dump_new(struct rde_peer *, uint8_t, unsigned int,
Change the Adj-RIB-Out to a per peer set of RB trees. The way RIB data structures are linked does not scale for the Adj-RIB-Out and so inserts and updates into the Adj-RIB-Out did not scale because of some linear list traversals in hot paths. A synthetic test with 4000 peers announcing one prefix each showed that the initial convergence time dropped from around 1 hout to around 6min. Note: because the Adj-RIB-Out is now per peer the order in which prefixes are dumped in 'bgpctl show rib out' changed. Tested and OK job@, benno@, phessler@
2019-07-17 03:13:26 -07:00
void *, void (*)(struct prefix *, void *),
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
void (*)(void *, uint8_t), int (*)(void *));
Introduce tree walkers that only walk a subtree of the RIB. In some cases only a "small" part of the RIB needs to be looked at. Like bgpctl show rib 10/8 or-longer that only needs to travers nodes under 10/8 all other RIB entries do not matter. By setting the start node to the RB_NFIND(10/8) the all nodes below this point can be skipped. Using prefix_compare() while walking the tree with RB_NEXT() the walker know when it steps outside of the 10/8 subtree and stops. With this the or-longer commands become a lot faster. Looks good to tb@
2022-09-12 03:03:17 -07:00
int prefix_dump_subtree(struct rde_peer *, struct bgpd_addr *,
uint8_t, unsigned int, void *,
void (*)(struct prefix *, void *),
void (*)(void *, uint8_t), int (*)(void *));
Implement reception of multiple paths per BGP session. This is one side of RFC7911 and the send portion will follow. The path-id is extracted from the NLRI encoding an put into struct prefix. To do this the prefix_by_peer() function gets a path-id argument. If a session is not path-id enabled this argument will be always 0. If a session is path-id enabled the value is taken from the NLRI and can be anything, including 0. The value has no meaning in itself. Still to make sure the decision process is able to break a tie the path-id is checked as the last step (this is not part of the RFC but required). OK benno@
2021-08-09 01:15:34 -07:00
struct prefix *prefix_bypeer(struct rib_entry *, struct rde_peer *,
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
uint32_t);
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
void prefix_destroy(struct prefix *);
Change struct prefix a bit. Make the aspath pointer opaque and remove the direct pointer to the pt_prefix struct. To getter functions prefix_aspath() and prefix_peer() added to make access of the opaque pointers possible. Looks good henning@ and benno@
2018-02-04 19:55:54 -08:00
Use Adj-RIB-Out to push UPDATE messages to peers instead of having another set of RB trees of prefixes and atribute. Refactor most of the update code which removes some strange buffer handling. By building the output queue directly in the Adj-RIB-Out the top memory usage during startup is greatly reduced which should help busy server. Tested by phessler@ and myself
2019-01-20 18:07:56 -08:00
RB_PROTOTYPE(prefix_tree, prefix, entry, prefix_cmp)
Move nexthop and nexthop flags from the rde_aspath to struct prefix. struct prefix will be slowly becomming the hub of the rib. OK phessler@ job@
2018-08-03 09:31:22 -07:00
static inline struct rde_peer *
prefix_peer(struct prefix *p)
{
return (p->peer);
}
Change struct prefix a bit. Make the aspath pointer opaque and remove the direct pointer to the pt_prefix struct. To getter functions prefix_aspath() and prefix_peer() added to make access of the opaque pointers possible. Looks good henning@ and benno@
2018-02-04 19:55:54 -08:00
static inline struct rde_aspath *
prefix_aspath(struct prefix *p)
{
Just put the aspath and the peer directly into the prefix struct. Needed for upcoming changes.
2018-06-27 06:14:44 -07:00
return (p->aspath);
Change struct prefix a bit. Make the aspath pointer opaque and remove the direct pointer to the pt_prefix struct. To getter functions prefix_aspath() and prefix_peer() added to make access of the opaque pointers possible. Looks good henning@ and benno@
2018-02-04 19:55:54 -08:00
}
Completley rewrite the community matching and handling code. All community attributes are put into a new data structure when parsing the UPDATE. The filter code can quickly lookup and modify this data structure. When creating an UPDATE the data is put back into wire format. Setups using a lot of communities benefit a lot from this. Input and OK benno@
2019-06-17 04:02:19 -07:00
static inline struct rde_community *
prefix_communities(struct prefix *p)
{
return (p->communities);
}
Move nexthop into struct filterstate (including nexthop flags) and use them whereever possible. In some places (path_update, softreconfig_out handler) the nexthop state is temporarily folded back into the rde_aspath. Tested and OK benno@
2018-07-22 09:59:08 -07:00
static inline struct nexthop *
prefix_nexthop(struct prefix *p)
{
Move nexthop and nexthop flags from the rde_aspath to struct prefix. struct prefix will be slowly becomming the hub of the rib. OK phessler@ job@
2018-08-03 09:31:22 -07:00
return (p->nexthop);
Move nexthop into struct filterstate (including nexthop flags) and use them whereever possible. In some places (path_update, softreconfig_out handler) the nexthop state is temporarily folded back into the rde_aspath. Tested and OK benno@
2018-07-22 09:59:08 -07:00
}
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
static inline uint8_t
Move nexthop and nexthop flags from the rde_aspath to struct prefix. struct prefix will be slowly becomming the hub of the rib. OK phessler@ job@
2018-08-03 09:31:22 -07:00
prefix_nhflags(struct prefix *p)
Change struct prefix a bit. Make the aspath pointer opaque and remove the direct pointer to the pt_prefix struct. To getter functions prefix_aspath() and prefix_peer() added to make access of the opaque pointers possible. Looks good henning@ and benno@
2018-02-04 19:55:54 -08:00
{
Properly handle nexthop state changes in the decision process In rev 1.90 of rde_decide.c the re->active cache of the best prefix was replaced with a call to prefix_best(). This introduced a bug because the nexthop state at that time may have changed already. As a result when a nexthop became unreachable prefix_evaluate() had oldbest = NULL and newbest = NULL and did not withdraw the prefix from FIB and Adj-RIB-Out. To fix this store the nexthop state per prefix and introduce prefix_evaluate_nexthop() which removes the prefix from the decision list, updates the nexthop state of the prefix and reinserts the prefix. Doing this ensures that prefix_best() always reports the same result once the decison process is done. prefix_best() and prefix_eligible() only depend on data stored on the prefix itself. OK tb@
2022-07-25 09:37:55 -07:00
return (p->nhflags & NEXTHOP_MASK);
}
static inline int
prefix_nhvalid(struct prefix *p)
{
return ((p->nhflags & NEXTHOP_VALID) != 0);
Change struct prefix a bit. Make the aspath pointer opaque and remove the direct pointer to the pt_prefix struct. To getter functions prefix_aspath() and prefix_peer() added to make access of the opaque pointers possible. Looks good henning@ and benno@
2018-02-04 19:55:54 -08:00
}
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
static inline uint8_t
Add the validation state to the filterstate struct. Removes vstate argument from rde_filter(). Rename prefix_vstate() to prefix_roa_vstate(). OK tb@
2023-01-11 09:10:25 -08:00
prefix_roa_vstate(struct prefix *p)
Implement origin validation in bgpd. This introduces two new tables, the roa-set for RPKI based origin validation and a origin-set which allows to lookup a source-as / prefix pair. For RPKI a config can be built like this: roa-set { 165.254.255.0/24 source-as 15562 193.0.0.0/21 maxlen 24 source-as 3333 } deny from any ovs invalid match from any ovs valid set community local-as:42 match from any ovs not-found set community local-as:43 Origin sets are similar but only match when the source-as / prefix pair is valid. match from any origin-set ARINDB set community local-as:44 Committing this now so that further work can be done in tree. OK benno@, job@
2018-09-29 01:11:11 -07:00
{
return (p->validation_state & ROA_MASK);
}
Implement ASPA validation and reload logic on ASPA set changes. For this use the validation state (vstate) in struct prefix and struct filterstate to store both the ASPA and ROA validity. Introduce helper functions to set and get the various states for struct prefix and make sure struct filterstate is also setup properly. Change the ASPA state in rde_aspath to be AFI/AID and role independent by storing all 4 possible outcomes. Also add a ASPA generation count which is used to update the rde_aspath ASPA state cache on reloads. Rework the rde_aspa.c code to be AFI/AID and role independent. Doing this for roles is trivial but AFI switch goes deep and is so unnecessary. The reload is combined with the ROA reload logic and renamed to RPKI softreload. OK tb@
2023-01-24 03:28:41 -08:00
static inline uint8_t
prefix_aspa_vstate(struct prefix *p)
{
return (p->validation_state >> 4);
}
static inline void
prefix_set_vstate(struct prefix *p, uint8_t roa_vstate, uint8_t aspa_vstate)
{
p->validation_state = roa_vstate & ROA_MASK;
p->validation_state |= aspa_vstate << 4;
}
Restructure struct prefix a bit and move the rib pointer to the union that splits the normal RIB linkage vs the adjrib-out linkage. This is done to make a bit of space to put an extra add-path related id into the struct without blowing its size over 128 bytes. Long run this struct should be split up but the necessary changes are too large right now so this is the 2nd best option. OK benno@
2021-07-27 00:50:01 -07:00
static inline struct rib_entry *
prefix_re(struct prefix *p)
{
if (p->flags & PREFIX_FLAG_ADJOUT)
return NULL;
return (p->entry.list.re);
}
Introduce 'rde rib Loc-RIB include filtered' a feature that includes filtered prefixes in the Loc-RIB This includes filtered prefixes into the Loc-RIB but they are marked ineligible so nothing will select them but it is possible to show them in bgpctl. So 'bppctl show rib filtered' will return all prefixes filtered out by the input filters. OK tb@
2024-08-14 12:09:51 -07:00
static inline int
prefix_filtered(struct prefix *p)
{
return ((p->flags & PREFIX_FLAG_FILTERED) != 0);
}
Shutdown the RDE cleanly on exit. Plug some memleaks. OK henning@
2004-03-11 06:22:22 -08:00
void nexthop_shutdown(void);
Change nexthop_update to run the list walk over all prefixes to run asynchronously and therefor other tasks can make progress at the same time. Additionally prefixes belonging to a RIB which does not run the the decision process are no longer linked into the nexthop list. This replaces the early return in prefix_updateall() and reduces the time spent in nexthop_update(). OK benno@
2019-06-20 06:18:19 -07:00
int nexthop_pending(void);
void nexthop_runner(void);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
void nexthop_modify(struct nexthop *, enum action_types, uint8_t,
struct nexthop **, uint8_t *);
Move nexthop and nexthop flags from the rde_aspath to struct prefix. struct prefix will be slowly becomming the hub of the rib. OK phessler@ job@
2018-08-03 09:31:22 -07:00
void nexthop_link(struct prefix *);
void nexthop_unlink(struct prefix *);
RDE part of the nexthop verification puzzle. The RDE just tracks the nexthop IPs and reacts on nexthop messages from the parent. ok henning@
2003-12-25 15:22:13 -08:00
void nexthop_update(struct kroute_nexthop *);
Just ignore RFC2545 and the silly idea of using link local addresses as nexthop. This makes the code a lot simpler. OK henning@
2004-08-12 03:24:16 -07:00
struct nexthop *nexthop_get(struct bgpd_addr *);
Properly start reference counting struct nexthop. This removes the need for some ugly workaround to make sure nexthop objects don't disapear while still being referenced. During initial lookup of a nexthop a extra reference is pulled but even that is now a bit cleaner than before. Tested by job@, dennis@, benno@ OK job@ dennis@
2018-06-25 07:28:33 -07:00
struct nexthop *nexthop_ref(struct nexthop *);
Unify the way object in the RDE are reference counted. The affected structures are pt_entry, rde_aspath, rde_communities, and nexthop. The functions are always called *_ref and *_unref also the behaviour when the last reference is removed is unified and now the object is removed inside of the unref function. The actual bean-counting is not modified by this diff. OK benno@
2019-07-01 00:07:08 -07:00
int nexthop_unref(struct nexthop *);
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
Major cleanup in rde_update.c. Merge equal code used in different places into own functions. Move up_dump_upcall() into rde.c and rename it rde_up_dump_upcall(). This is needed for the next step. up_test_update() tests if an update or withdraw is needed and up_generate() creates the updates. "get it in" henning@ (he is eager on softreconfig out)
2005-11-01 06:37:16 -08:00
/* rde_update.c */
Compile the output filter rules into per peer filter rules. especially on route-servers the output filters are in the hot path so reducing the number of rules to check has a big impact. I have seen a 25% to 30% speedup in my big IXP testbench. The output ruleset is applied and copied for each peer during config reload and when a peer is initially added. OK tb@
2023-03-09 23:57:15 -08:00
void up_generate_updates(struct rde_peer *, struct rib_entry *);
void up_generate_addpath(struct rde_peer *, struct rib_entry *);
void up_generate_addpath_all(struct rde_peer *, struct rib_entry *,
struct prefix *, struct prefix *);
void up_generate_default(struct rde_peer *, uint8_t);
Switch from u_intX_t types to stdint.h uintX_t. Mostly mechanical with a few reindents. OK florian@ tb@
2022-02-06 01:51:19 -08:00
int up_is_eor(struct rde_peer *, uint8_t);
When generating UPDATE handle the message size limit better. First of all warn that a prefix was dropped. In the generate an update code handle possible overflows of attributes and NLRI and withdraw the affected prefix. This way the peer will not have stale data. OK tb@
2024-09-25 07:46:51 -07:00
struct ibuf *up_dump_withdraws(struct rde_peer *, uint8_t);
struct ibuf *up_dump_update(struct rde_peer *, uint8_t);
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
Add ASPA validation functions to the RDE. This implements ASPA validation based on the current draft. Implementing this showed various weaknesses in the current ASPA draft which I hope to fix in the near future. Unlike the algorithm specified in the draft our version validates the AS_PATH attribute in a single path doing one or two lookups depending on the sessions BGP role. The code is not yet hooked up into the RDE (see the NOTYET blocks). Missing are reload logic, bgpctl integration and the loading of the merged ASPA set from the rtr process. OK tb@
2023-01-11 05:53:17 -08:00
/* rde_aspa.c */
Implement ASPA validation and reload logic on ASPA set changes. For this use the validation state (vstate) in struct prefix and struct filterstate to store both the ASPA and ROA validity. Introduce helper functions to set and get the various states for struct prefix and make sure struct filterstate is also setup properly. Change the ASPA state in rde_aspath to be AFI/AID and role independent by storing all 4 possible outcomes. Also add a ASPA generation count which is used to update the rde_aspath ASPA state cache on reloads. Rework the rde_aspa.c code to be AFI/AID and role independent. Doing this for roles is trivial but AFI switch goes deep and is so unnecessary. The reload is combined with the ROA reload logic and renamed to RPKI softreload. OK tb@
2023-01-24 03:28:41 -08:00
void aspa_validation(struct rde_aspa *, struct aspath *,
struct rde_aspa_state *);
Add ASPA validation functions to the RDE. This implements ASPA validation based on the current draft. Implementing this showed various weaknesses in the current ASPA draft which I hope to fix in the near future. Unlike the algorithm specified in the draft our version validates the AS_PATH attribute in a single path doing one or two lookups depending on the sessions BGP role. The code is not yet hooked up into the RDE (see the NOTYET blocks). Missing are reload logic, bgpctl integration and the loading of the merged ASPA set from the rtr process. OK tb@
2023-01-11 05:53:17 -08:00
struct rde_aspa *aspa_table_prep(uint32_t, size_t);
void aspa_add_set(struct rde_aspa *, uint32_t, const uint32_t *,
Remove per-AFI ASPA handling in bgpd internals With draft-ietf-sidrops-aspa-profile-16 and draft-ietf-sidrops-aspa-verification-15 the AFI dependence of ASPA records was dropped. So remove this complication form the code. This only removes the AFI handling internally in bgpd but still allows the old syntax in aspa-set tables. The optional address family is just ignored and records are merged together. For RTR sessions draft-ietf-sidrops-8210bis has not yet been updated so right now we still handle RTR sessions as specified there. The IPv4 and IPv6 ASPA entries are handled in two trees and merged together into one AFI independent tree. This is the best we can do for now until IETF updates draft-ietf-sidrops-8210bis. OK tb@ job@
2023-08-16 01:26:35 -07:00
uint32_t);
Add ASPA validation functions to the RDE. This implements ASPA validation based on the current draft. Implementing this showed various weaknesses in the current ASPA draft which I hope to fix in the near future. Unlike the algorithm specified in the draft our version validates the AS_PATH attribute in a single path doing one or two lookups depending on the sessions BGP role. The code is not yet hooked up into the RDE (see the NOTYET blocks). Missing are reload logic, bgpctl integration and the loading of the merged ASPA set from the rtr process. OK tb@
2023-01-11 05:53:17 -08:00
void aspa_table_free(struct rde_aspa *);
Add the needed logic to load the ASPA table from the rtr process into the RDE. The actual reload logic is missing to keep the diff small. OK tb@
2023-01-17 08:09:01 -08:00
void aspa_table_stats(const struct rde_aspa *,
struct ctl_show_set *);
int aspa_table_equal(const struct rde_aspa *,
const struct rde_aspa *);
void aspa_table_unchanged(struct rde_aspa *,
const struct rde_aspa *);
Add ASPA validation functions to the RDE. This implements ASPA validation based on the current draft. Implementing this showed various weaknesses in the current ASPA draft which I hope to fix in the near future. Unlike the algorithm specified in the draft our version validates the AS_PATH attribute in a single path doing one or two lookups depending on the sessions BGP role. The code is not yet hooked up into the RDE (see the NOTYET blocks). Missing are reload logic, bgpctl integration and the loading of the merged ASPA set from the rtr process. OK tb@
2023-01-11 05:53:17 -08:00
welcome, bgpd started by me some time ago with moral support from theo, the proceeded up to the point where the session engine worked correctly. claudio jeker joined then and did a lot of work in the RDE. it is not particulary usefull as application right now as parts are still missing but is imported to enable more people to work on it. status: BGP sessions get established fine, OPEN messages and then KEEPALIVEs exchanged etc. session FSM works fine; NOTIFICATIONs are handled fine, and all connection drops etc I provoked get handled fine. Incoming UPDATE messgages are parsed well and the data entered to the RIB, the decision process is not yet there, neither is outgoing UPDATEs or sync to the kernel routing table. not connected to the builds yet.
2003-12-17 03:46:54 -08:00
#endif /* __RDE_H__ */
Copy Permalink