[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Problem with DGRAM multicasting


Yes, there are versions of the Ethernet driver that has this problem. I will
send you the latest driver.

Best Regards

-----Original Message-----
From: owner-dev-etrax@xxxxxxx.com">mailto:owner-dev-etrax@xxxxxxx.com] On Behalf Of Johan Neven
Sent: Friday, June 20, 2003 11:50 AM
To: Dalton Hamilton
Cc: dev-etrax
Subject: Re: Problem with DGRAM multicasting


Thanks for your fast reply.

The original source where I detected the problem contained the code to disable the loop-back, but with the same result. The code I inserted in the mail was the bare minimum to simulate the problem. It worked correctly on a i386 linux PC but failed on the Axis 100LX. I tried the reduced code with the line you gave me included again (to be sure) but with the same result.

It's a strange fenomenon. Running the process on the Axis through telnet with console output enabled it gives each 116 sento's an EWOULDBLOCK error and than continues. Through telnet with output to /dev/null, it stops totaly after 116 sendto's until touching a key in the telnet session.

I suppose it's a problem in the Etrax ethernet driver.

Thanks again for your suggestion.

Kind regards.

Johan Neven.

On Thu, 2003-06-19 at 17:45, Dalton Hamilton wrote:
> Hi -
> Just an idea, it looks as though you left yourself in the multicast
> loop (which is enabled by default) to receive the packets you're 
> sending.
> Therefore, the problem may be that your receive buffer is full since 
> all sent packets are copied to your receive buffer.
> You may want to do the following to see if that helps:
> to disable loopback:
> u_char loop=0;
> setsockopt(socket, IPPROTO_IP, IP_MULTICAST_LOOP, &loop, 
> sizeof(loop));
> You can find more infor at: 
> http://www.icewalkers.com/Linux/Howto/Multicast-HOWTO-6.html
> /Dalton
> On Thursday, June 19, 2003, at 08:35 AM, Johan Neven wrote:
> > Hello,
> >
> > During software development on a Axis 100LX I met some strange 
> > things with ethernet multicasting. When sending packets to a network 
> > through MC it was only possible to do a very limited number (116) of 
> > sendto's before there was an WOULDBLOCK error from sendto. The 
> > situation becomes unblocked again when there is some other network 
> > transaction on the Axis
> > (telnet). I send here a file with the concentrated code giving 
> > problems.
> > I hope the problem is reproductible with this information.
> >
> > When sending normal UniCast packets, everything seems fine because
> > there
> > is some kind of ack from the receiver and the situation is each time
> > unblocked (I suppose).
> >
> > It would be nice to have a solution or work-around for this problem.
> >
> > Thanks
> > --
> > Johan Neven <johan.neven@xxxxxxx.be>
> > <jn.c>
Johan Neven <johan.neven@xxxxxxx.be>