DHCP client nonstandard behavior

Christian Treczoks ct at braehler.com
Wed Jan 21 12:45:57 UTC 2009


Dear Listmembers,

I am new to this mailing list, and the question might have been asked 
before, although I could not find anything in the archives regarding my 
problem.

I'm using an RTOS (Keil RL-ARM) with a simple TCP/IP stack (Library, no 
source) for my project. Now I need to implement DHCP client functionality 
into my project, but not as the designer of the stack intended it to be 
(they only support DHCP when you start the system, you can disable it 
during runtime, but not re-enable it).

So I read the docs and DHCP sufficiently easy to implement, if it wasn't 
for one point: The RFCs state that the clients sender address is to be 
0.0.0.0 and the recipient address should be 255.255.255.255. This is ok 
for the initial DISCOVER and REQUEST packets, but when I need to renew the 
lease, I can't create a UDP packet with 0.0.0.0 as the sender address - 
the library uses the systems' IP address as the sender address.

If I would send a REQUEST package (for a renewal) with a valid, previously 
negotiated sender address that is not 0.0.0.0 (but everything else ok), 
would this cause problems with existing DHCP servers?

Yours,
Christian Treczoks

Research and Development
Auf der Alten Burg 6
53639 Königswinter

Tel.: +49 (0)2244 930-282
Fax.: +49 (0)2244 930-410
E-Mail: ct at braehler.com
http://www.braehler.com
--------------------------------------------------------------------
Hauptsitz / Headquarters
Brähler ICS Konferenztechnik International Congress Service AG
Wahlfelder Mühle 3, 53639 Königswinter, Germany
Handelsregister: Amtsgericht Siegburg HRB 6980
Vorstand: Michael Brähler, Rita Degen, Hans Henrik Friis
Vorsitzende des Aufsichtsrates: Erika Brähler
*************************************************************
Visit us at Integrated Systems Europe in Amsterdam
Booth #3A118 from 3rd to 5th of February 2009
*************************************************************



More information about the dhcp-hackers mailing list