BIND 10 #931: Implement signing part in b10-auth

[BIND 10 Development]

#931: Implement signing part in b10-auth
-------------------------------------+-------------------------------------
                   Reporter:         |                 Owner:  jinmei
  vorner                             |                Status:  reviewing
                       Type:  task   |             Milestone:
                   Priority:  major  |  Sprint-20110531
                  Component:         |            Resolution:
  Unclassified                       |             Sensitive:  0
                   Keywords:         |           Sub-Project:  DNS
            Defect Severity:  N/A    |  Estimated Difficulty:  0.0
Feature Depending on Ticket:  tsig   |           Total Hours:  0
        Add Hours to Ticket:  0      |
                  Internal?:  0      |
-------------------------------------+-------------------------------------
Changes (by vorner):

 * owner:  vorner => jinmei


Comment:

 Hello

 As we are no longer in hurry, I tried to address everything (if I missed
 something, it's probably I overlooked it). The new code is in trac931_2,
 as I did some history rewrites. The current changes are:
  * It is the same up to a8307030f7af9fc88e3e66b6eefcc89f6b6e15c5.
  * The fix for sending correct module name in addRemoteConfig got a test
 (actually, it's not a new test, but extended the original). It's now
 directly in the commit with the fix, but except adding the check, the
 commit is the same.
  * I've thrown out my fixes and included yours (I didn't throw out the
 size_t → uint32_t change, but I have no idea how to test it, when it
 didn't produce any bugs, only looked completely wrong).
  * I added test for the double-read bug.
  * Added a test for the signature and invalid request (using unknown
 opcode).
  * Then there's a merge from master, it brings another parameter for the
 addRemoteConfig function (the original #931 was based on older version of
 #925).
  * The addRemoteConfig got some more attention, I split part of it into
 separate function and added some spaces and comments, but I don't know if
 there's a way to clarify it more.
  * Implemented the part with auth server not referring directly to the
 global keyring variable.

 I hope it looks better now.

 Regarding the changelog, the sha1-hmac is there for the exact reason that
 it's a different format than dig, so they can see it. I think confusion
 from the changelog would be smaller than confusion when the software
 rejects their keys.

 So, what about this?
 {{{
 Authoritative server can now sign the answers using TSIG
 (configured in tsig_keys/keys, list of strings like
 "name:<base64-secret>:sha1-hmac"). It doesn't use them for
 ACL yet, only verifies them and signs if the request is signed.
 }}}

 Thanks

-- 
Ticket URL: <http://bind10.isc.org/ticket/931#comment:12>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development