Search

DHCPv6 server does not format a DNS Domain Search List correctly by DerekMorr

Closed
as Tracking Help for as Tracking

1
Sign in to vote
0
Sign in to vote
Sign in
to vote
Type: Suggestion
ID: 345620
Opened: 5/22/2008 2:02:28 PM
Access Restriction: Public
0
Workaround(s)
The Windows Server 2008 SP1 DHCPv6 server does not format the DNS Domain Search List (DHCPv6 option 24) correctly. The format of this option is defined in RFC 3646, section 4. The exact encoding is defined in RFC 1035, section 3.1.

The WIDE and Fedora DHCPv6 clients misparse a DHCPv6 offer from a Windows 2008 DHCPv6 server. This prevents Unix and Linux clients from behaving correctly.
Details (expand)
Server Role:
DHCP Server
Please describe how you would like this behavior to change?
Windows 2008 encodes the search list with the length of each search domain, termating the list with a NULL. The correct behavior is to encode the length of each search domain label, and terminate each search domain with a NUL.

For example, using the search domains "its.psu.edu" and "psu.edu" -

Windows 2008 encoding:

0b (length = 11)
69 i
74 t
73 s
2e .
70 p
73 s
75 u
2e .
65 e
64 d
75 u
07 (length = 7)
70 p
73 s
75 u
2e .
65 e
64 d
75 u
00 NUL

The correct encoding is:

03 (length = 3)
69 i
74 t
73 s
03 (length = 3)
70 p
73 s
75 u
03 (length = 3)
65 e
64 d
75 u
00 NUL
03 (length = 3)
70 p
73 s
75 u
03 (length = 3)
65 e
64 d
75 u
00 NUL
Please provide a justification for why you think this change should be made?
Quoting from RFC 1035, section 3.1:

"Domain names in messages are expressed in terms of a sequence of labels. Each label is represented as a one octet length field followed by that number of octets. Since every domain name ends with the null label of the root, a domain name is terminated by a length byte of zero."

Windows does not conform to this standard. It encodes the length of the domain, not of the labels in each domain.
File Attachments
0 attachments
Sign in to post a comment.
Posted by DerekMorr on 5/27/2008 at 8:49 AM
I'm curious why this bug was closed. Is there a HotFix for the bug? Is there a patch in-progress for it?
Posted by Microsoft on 5/29/2008 at 9:08 AM
I'm checking to see if this is a documentation bug for you. If it is, I'll submit a bug report internally against it.

In the future, please report bugs to Microsoft Product Support Services or your support partner for Server 2008.

The site you submitted this to is for collecting suggestions and feature requests for Microsoft Windows Server 2008 products only.

Thanks,
Microsoft Windows Server 2008 Feedback Team
Posted by hamajun2.0 on 10/24/2008 at 9:50 AM
I also need to know Microsoft's official statement regarding this obvious RFC violation.

I've posted the same question to the Microsoft Windows DHCP team and am waiting for their response.

http://blogs.technet.com/teamdhcp/archive/2007/01/27/dhcpv6-stateless-and-stateful-server-in-windows-server-longhorn.aspx?CommentPosted=true#commentmessage

Thanks.