[possible regression] pc386/ne2000 networking is not running on rtems-trunk

[Keith Robertson]

Karel Gardas wrote:
> On Fri, 14 Jan 2005, Joel Sherrill <joel at OARcorp.com> wrote:
>>>Running is:
>>>4-6-branch + toolchain recommended for 4.6.x release. Counfigured with
>>>--target=i386-rtems --prefix=<something> --enable-cxx
>>>
>>>Broken is:
>>>trunk + Ralf's toolchain: gcc3.4.3 + binutils2.15 + newlib1.13 + Ralf's
>>>patch, configured as 4-6-branch with different installation prefix.
>>>
>>>dnsdemo is the same source code compiled for both these versions.
>>
>>Does netdemo run?  Do any of the other networking tests run for you?
> 
> 
> Netdemo runs partially only:
> 
> on the 46-branch configuration statistic and udp are working, tcp not. --
> it just prints `Create socket.\nBind socket.\nConnect socket.\n' and
> that's all, after this demo looks (at least console) blocked

Hi Karel,

It's possibly worth re-checking that you've set the DATA_SINK_HOST to a 
valid machine that accepts connections on the discard port (9).  The 
console appears to lock for sometime if you don't (while waiting for a 
timeout presumably).

Additionally, when you do get it working the 't' option sends a fair 
chunk of data.  It takes a good 5-10s before you'll get focus back in 
the console.  You should see plenty of traffic to DATA_SINK_HOST:9 
throughout this period though.

Cheers,

Keith