In the previous version of dhcp(c)d, I had created a NMU with some
changes to make the checksum calculation correct. (Basically,
sizeof(ether_header) turned out 16 instead of 14 due to alignment
stuff).
I filed a bug report to the maintainer. Perhaps that got lost. I
will redo the change and submit another NMU tonight. (If that's
still the problem...)
-tor
>>>>> "Peter" == Peter Naulls <[EMAIL PROTECTED]> writes:
Peter> Perhaps someone knows about this:
Peter> I just discovered today (after ntl cable being down for 2
Peter> days) that dhcp-client was no longer working - that is, it
Peter> was apparently misunderstanding the response from the
Peter> server. The only clue I got was complaints about IP
Peter> checksums and from tcpdump complaining ip_hl < 5 (IP header
Peter> length presumably) when it was running. I downgraded some
Peter> likely packges that had changed in unstable in the last few
Peter> days, but no luck. The same setup on i386 Linux works just
Peter> fine.
Peter> In the end, I gave up and used pump instead, and that
Peter> worked fine.
Peter> So either, a package changed which I missed, or something's
Peter> different on ntl's network, which is causing problems in
Peter> ARM Linux.
Peter> Peter
Peter> --
Peter>
------------------------------------------------------------------------
Peter> Peter Naulls - [EMAIL PROTECTED] Homepage -
Peter> http://free.prohosting.com/~chocky/ Java for RISC OS and
Peter> ARM - http://free.prohosting.com/~chocky/java/ Debian Linux
Peter> on RiscPCs - http://www.erble.freeserve.co.uk/
Peter>
------------------------------------------------------------------------
Peter> -- To UNSUBSCRIBE, email to
Peter> [EMAIL PROTECTED] with a subject of
Peter> "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
--
Får i ulveklær
