Re: IGMP fragmentation bug - another behavior

[Paul.Murphy () GEMINI-RESEARCH CO UK (Paul Murphy)]

> I've compiled kod.c on Solaris SPARC 2.6:
> SunOS 5.6 Generic_105181-12 sun4u sparc SUNW,Ultra-1
> with gcc 2.8.1

I tried this as well, and gave up trying to get it to run on Solaris
after a few minutes.

> after executing kod to some win host Solaris hme driver has been
> choked and was in nonoperational mode about 1 min. After this
> hme interface was reinitialized:
> and machine continues work without problems.

Strange - rather than try this, I'll believe you - our system with an
HME interface is a production system, so no playing around...

> But on SPARCstation 4 Model 110 with le card there are no problems
with >executing kod. Interface not frozen... Also target win host
machine continues >normal work.

That's not surprising - the program compiles successfully, and although
it reports that it is sending packets, Snoop reported that nothing was
sent to the target machine.  The reason for this is that Solaris does
not appear to support Raw IGMP - the "sendto" call fails, but since the
program doesn't bother to check whether it was successful, it carries on
regardless....

Checking the return status of the "sendto", and reading the error
number, we get  EPROTONOSUPPORT - suggesting that raw IGMP can't be sent
over this interface.  To my mind, this is a _good_thing_ (TM).

Paul.

-----------------------------------------------------------------------------
Paul Murphy - Head of I.T., Gemini Research Ltd
162 Science Park, Cambridge CB4 4GH
Tel. 01223 435305 Fax. 01223 435301
http://www.gemini-research.co.uk/