BIND 10 #1104: support TSIG in DNS (Request) ACL
BIND 10 Development
do-not-reply at isc.org
Tue Aug 2 10:07:10 UTC 2011
#1104: support TSIG in DNS (Request) ACL
-------------------------------------+-------------------------------------
Reporter: | Owner: jinmei
jinmei | Status: reviewing
Type: task | Milestone:
Priority: major | Sprint-20110802
Component: | Resolution:
xfrout | Sensitive: 0
Keywords: | Sub-Project: DNS
Defect Severity: N/A | Estimated Difficulty: 5
Feature Depending on Ticket: | Total Hours: 0
Add Hours to Ticket: 0 |
Internal?: 0 |
-------------------------------------+-------------------------------------
Changes (by stephen):
* owner: stephen => jinmei
Comment:
First off, I agree with the suggestion to close this ticket and open a new
one.
> what I suggested is to introduce a "verified" flag to the TSIGRecord
class, not TSIGContext
My mistake, apologies.
> I don't think it a good idea to deceive the compiler with mutable in
this case. When we set the "verified" flag of TSIGRecord, it actually
changes externally observable behavior
Only if we are allowing for the possibility that the contents of the TSIG
keyring may have changed between two calls to verify(). If we are using
the paradigm that all calls to verify() for a particular record are
against the same instance of the keyring, if the first call to verify()
succeeds, all subsequent calls will succeed. In this case the TSIGrecord
is still logically "const", the mutable flag being used to cache a result
to avoid expensive computation.
> IMO, if an application wants to handle TSIG, it should explicitly do it
at an appropriate position of the code.
But there is a separation between ACLs and the application; the ACL is set
by the user, the packet is checked against it, then the (rest of the)
application is invoked. If we restrict the TSIG check to just the name,
we leave open the possibility that an incorrect packet may get through the
check.
--
Ticket URL: <http://bind10.isc.org/ticket/1104#comment:16>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development
More information about the bind10-tickets
mailing list