BIND 10 #3279: DHCPv4 server should discard packets with non-matching server id
BIND 10 Development
do-not-reply at isc.org
Tue Jan 14 15:28:39 UTC 2014
#3279: DHCPv4 server should discard packets with non-matching server id
-------------------------------------+-------------------------------------
Reporter: marcin | Owner: tmark
Type: defect | Status:
Priority: medium | reviewing
Component: dhcp4 | Milestone: DHCP-
Keywords: | Kea1.0-alpha
Sensitive: 0 | Resolution:
Sub-Project: DHCP | CVSS Scoring:
Estimated Difficulty: 16 | Defect Severity:
Total Hours: 10 | Medium
| Feature Depending on Ticket:
| Add Hours to Ticket: 2
| Internal?: 0
-------------------------------------+-------------------------------------
Changes (by marcin):
* owner: marcin => tmark
Comment:
Replying to [comment:7 tmark]:
> The changes are quite straight forward, I only have a few
questions/comments:
>
> General question:
>
> * In looking at IfaceMgr::hasOpenSocket(), have you done any impact
assessment on performance? Granted we are not likely to have more than a
few interfaces, with maybe one address per interface.
I added a comment to say that I am aware of performance penalty. But I
really don't think it is going to be significant if we just open sockets
on a couple of interfaces. If the performance penalty is significant, we
can think about caching the server identifiers, but I don't want to
complicate this code more than it is complicated now.
>
> * Should we consider making this check configurable?
Unless I misunderstand this question, my answer is no. Accepting or
rejecting server identifier is a core functionality of DHCPv4 described in
RFC2131 (section 4.1). How could I make this configurable?
>
>
----------------------------------------------------------------------------------------
> Dhcpv4Srv::acceptServerId(const Pkt4Ptr& pkt)
>
> You test the address list for size = 1 and return false if it is not.
I'm
> assuming there is no valid case for it have more than 1 entry. You may
want
> to add a quick comment on this above the size test.
Added a comment. There should be no case that the size of the list is
unequal 1, unless the client has sent a malformed option which we would
(and want to) catch here.
>
>
-----------------------------------------------------------------------------------------
>
> Unit tests run with valgrind on Fedora 19. cppcheck passes on OS-X.
>
>
>
Thank you QA team!
--
Ticket URL: <https://bind10.isc.org/ticket/3279#comment:8>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development
More information about the bind10-tickets
mailing list