$OpenBSD: BUGS,v 1.14 2002/06/09 08:13:06 todd Exp $
$EOM: BUGS,v 1.38 2000/02/18 08:47:35 niklas Exp $

Until we have a bug-tracking system setup, we might just add bugs to this
file:
------------------------------------------------------------------------------
* message_drop frees the message, this is sometimes wrong and can cause
  duplicate frees, for example when a proposal does not get chosen. [fixed]

* Notifications should be their own exchanges, otherwise the IV gets
  disturbed. [fixed]

* We need a death timeout on half-ready SAs just like exchanges.  At the
  moment we leak SAs.

* When we establish a phase 2 exchange we seem to get the wrong IV set,
  according to SSH's logs. [fixed]

* If a phase 1 SA negotiation terminates with a cause that is to be sent in
  a NOTIFY to the peer, we get multiple free calls on the cleanup of the
  informational exchange. [fixed]

* IKE mandates that a HASH should be added to informational exchanges in
  phase 2. [fixed]

* Message_send requires an exchange to exist, and potentially it tries to
  encrypt a message multiple times when retransmitting. [fixed]

* Multiple protocol proposals seems to fail. [fixed]

* The initiator fails to match the responders choice of protocol suite with
  the correct one of its own when several are offered. [fixed]

* Duplicate specified sections is not detected. [fixed]

* Quick mode establishments via UI using -P bind-addr gets "Address already in
  use".

* Not chosen proposals should be deleted from the protos list in the sa
  structure. [fixed]

* Setting SPIs generates "Invalid argument" errors due to one tunnel endpoint
  being INADDR_ANY. [fixed]

* ipsec_proto structs are never allocated. [fixed]

* Remove SPIs of unused proposals. [fixed]

* If the first proposal is turned down, the initiator gets confused.

* Renegotiation after a failed phase 1 fails.

* Phase 1 rekey event removal seems to be done twice. [fixed]

* PF_ENCAP expirations does not find the proper phase 2 SA to remove. [fixed]

* ISAKMP SA expirations should have a soft/hard timeout just like IPsec ones.
  The soft one should put a watchdog on the SA, and start a renegotiation as
  soon as something used the SA.  Hard ones should just clean it up, no
  renegotiation at all. [fixed]

* ISAKMP SAs does not get removed after rekeying. [fixed]

* On-demand PF_ENCAP SAs does not get reestablished. [fixed]

* Rekeying is now done automatically on expirations, it should not.  The
  SAs should be brought up on-demand just like the first time.

* Notifications regarding exchange errors seems to not have the right SPI,
  at least not in phase 1, in NO_PROPOSAL_CHOSEN.

* Outgoing informational exchanges when we use INVALID_PAYLOAD_TYPE
  cause a DOI error.

* In Linux select(2) of named pipes seems broken as they will return as
  readable even when nothing is there after one read has succeeded.

* I have seen INITIAL-CONTACT sent on the second Main Mode.

* When ID mismatches occur, coredumps may follow, investigate!

* ESP+AH does not work properly

* Looping QM seen (due to lost sendpackets from other participant?)

* Teardown from UI does not remove exchanges.

* Wrong error message when policy check fails.

* Restransmit of QM (packet 1) after INFO/PAYLOAD_MALFORMED was received.

* SIGSEGV after sa_enter: sa added to sa list, trigged by DELETE notify (Linux)

* Passive connections, undefined local&remote IDs will cause IKE peer IDs
  to be used.

* host route support in KLIPS does not work properly

* When not having compiled in support for a certain crypto algorithm and
  the config file still tells us to propose it, we segfault.