summaryrefslogtreecommitdiff
path: root/usr.sbin
AgeCommit message (Collapse)Author
2005-02-15let ospfd "tag" its routes with RTF_PROTO2, bgpd uses RTF_PROTO1 alreadyHenning Brauer
2005-02-15Recognize RFC 3947 NAT-T VID. hshoexer@ okHakan Olsson
2005-02-15New command: pkg_mklocatedb, that can be used to create locate(1)-styleMarc Espie
databases for packages (duh, isn't that obvious from the name ?) Tweaks based on comments from naddy/jmc.
2005-02-13Don't close the dbm until after we have copied the datum retrieved by theAaron Campbell
fetch into a local buffer. Adapted from r1.10 of the same file in Apache CVS. henning@ ok
2005-02-10Work around a timing issue in the db exchange phase. The DB descriptionClaudio Jeker
packets comming from the net are looped through the RDE and may get delayed because of this. The result is that the neighbor FSM ends in state FULL instead of LOADING and so the LSDB is not correctly synchronized. Issue found by norby@ OK norby@
2005-02-10Cleanup the neighbor FSM. Use log_warnx() where appropriate and don'tClaudio Jeker
use the correct enum instead of 0 in the nbr_fsm_tbl.
2005-02-10Minor punctuation nit.Joel Knight
ok henning@
2005-02-09Fix premature ageing of self originating LSA. Still not perfect but betterClaudio Jeker
than before where it just failed to work.
2005-02-09mini KNFClaudio Jeker
2005-02-09NBR_STA_ACTIVE is just ~NBR_STA_DOWN instead of some hardcoded magic.Claudio Jeker
2005-02-09Use log_warnx() in lsa_check() to report bad LSA settings.Claudio Jeker
2005-02-09Rework of self originated LSA. First of all the code to generate the LSAClaudio Jeker
is rewritten to use the dynamic buffer api introduced lately. Add missing parts in the RDE to merge these LSA originated from the OE into the LSDB with correct seq_num, etc. Add hooks to the neighbor and interface fsm so that the LSA get updated if needed. This is not perfect, still many things are missing. OK norby@
2005-02-09Fix interface fsm by inserting dummy events and actions as placeholderClaudio Jeker
for 0. Action 0 was IF_ACT_STRT and that's why if_act_start() got called multiple times. The inappropriate hack from the previous commit is no longer needed.
2005-02-09Don't elect neighbors that are down and fix a stupid copy-pasto.Claudio Jeker
2005-02-09Initialize iface->self only once. if_act_start() may be called multipleClaudio Jeker
times for a interface.
2005-02-09Print the interface state from the view of the neighbor in ospfctl sh nei.Claudio Jeker
The idea is to print DR, BDR or DROther dependent on the type of the neighbor. This is only done for multicast capable networks.
2005-02-09Prevent the ADJ_TMOUT from complaining about an unexpected event.Claudio Jeker
From norby@
2005-02-09Fix the restart DD case. In case of a timeout recheck the adjacency and goClaudio Jeker
to state NBR_STA_XSTRT and issue a new initial DB description packet or go to NBR_STA_2_WAY respectivly. OK norby@
2005-02-09When sending out hellos, copy the bdr and dr addresses into iface->self.Claudio Jeker
Without this the election process causes strange results as a router that announces itself as dr/bdr is prefered over one that doesn't.
2005-02-09If the router self is the originator of the LS update flood it outClaudio Jeker
no matter what. This solves an issue where self originated updates where not flooded because the router was the DR.
2005-02-09Once again a round of bugfixes in the db exchange process. This isClaudio Jeker
one of the worst documented parts of the RFC. The initial packet sent by the slave may have no flags set. Don't enforce that the M bit is set but instead check that the I bit and MS bit are unset. The master should only issue NBR_EVT_XCHNG_DONE if it has sent at least one packet with M bit cleared else the slave may get stuck in state Exchange. In NBR_STA_LOAD the db_tx_timer() still needs to send db descriptions out. This will be the last packet sent with the cleared M bit.
2005-02-09Add buf_seek() as buf_reserve() fails if a buf_add()/buf_reserve() isClaudio Jeker
called afterwards as it may realloc() the buffer and so the returned pointer is bogus. Needed by the upcomming originate LSA code. OK henning@
2005-02-09Print the neighbor address and not the local interface address inClaudio Jeker
ospfctl show nei.
2005-02-09need to send IMSG_NETWORK_DONE after sending networks announcement.Claudio Jeker
Similar to the fix commited by henning@ a few hours ago. OK henning@
2005-02-09cleanup and unifdef'ing, no change in object filesHenning Brauer
work by Daniel Ouellet <daniel@presscom.net>
2005-02-09in case some md5 don't match, keep the offending files around as aMarc Espie
borked package (with fixed md5). Makes it easier to locate those files and erase them later if needed... this also makes collisions explicit if user forgets and tries to install a new package later.
2005-02-09need to send IMSG_NETWORK_DONE after sending networks and associated filterHenning Brauer
sets, otherwise local netyworks get withdrawn after config reload; misbehaviour noticed by peter.galbavy@knowtion.net, claudio ok
2005-02-08Change default metric to 10, reflects what is stated in ospfd.conf(5).Esben Norby
ok claudio@
2005-02-08* plug memory leaks in some paths. From Andrey MatveevPatrick Latifi
ok otto
2005-02-08The LSA seq_num should be unsigned as it is expected by ntohl/htonl() evenClaudio Jeker
though the seq_num needs to be compared as signed number. No idea how IETF came up with such a stupid idea especially because the seq_num is not allowed to wrap. requested by henning@ OK henning@
2005-02-07make updatedepends slightly safer: note which packages may have bogusMarc Espie
dependencies, so that if you do one big pkg_add -r *, then you can check which packages may have issues thanks to forced updates.
2005-02-07Fix error message: set-state, not setstate.Ryan Thomas McBride
2005-02-07Fix protos, and a KNF nit. From Andrey Matveev <andrushock@korovino.net>Ryan Thomas McBride
2005-02-07spelling fixDavid Krause
2005-02-07spelling fixesDavid Krause
2005-02-07take size of metainfo into account when checking if the file system isMarc Espie
large enough before installing a package.
2005-02-05connect ospfd and ospfctlHenning Brauer
2005-02-04update `dst port', `ip proto', `ether proto', and SEE ALSO;Jason McIntyre
ok mcbride@
2005-02-04unfuck amd64Peter Valchev
2005-02-04Use log_warn() insteade of log_warnx() after sendto() and if_set_mcast()Claudio Jeker
errors as errno was set by the call.
2005-02-04Add sanity checks for AS-external LSA. Refine some other checks to be asClaudio Jeker
strict as possible. OK norby@
2005-02-03Simplify the ifstated syntax:Marco Pfatschbacher
"carp0 link up" => carp0.link.up "and" => && "or" => || * Allow one line actions after if statements without braces. * Remove unecessary parentheses in the example config. ok mcbride@
2005-02-03Implement simple duplicate suppression of peer errors; ok henning@Darren Tucker
2005-02-03add ospfd.conf.5 to SEE ALSO;Jason McIntyre
2005-02-03no need for -compact/.Pp in these lists;Jason McIntyre
2005-02-02various fixes;Jason McIntyre
2005-02-02Add manpage ospfd.conf(5).Esben Norby
ok claudio@ henning@ jmc@ jaredy@
2005-02-02Change keyword fibupdate to fib-update, for consistency with bgpd.conf(5).Esben Norby
ok claudio@ henning@
2005-02-02Typo in constant.Esben Norby
ok claudio@ henning@
2005-02-02excessive ()Henning Brauer