On 02-Jan-99 Florian Lohoff wrote:
>> Florian, declance.c doesn't work on my '260, so I'd like to ask you a
>> R4000 related questions:
> It doesnt work on my /240 either ... It sends out packages and
> it gets the FIRST but no others - It receives the BOOTP request answer
> correct but the tcpdump output shows that it trys to ask my
> sunrpc for the port numbers for the mountd/nfsd which my machine
> answers but it doesnt seem to creep up the Kernel.
Exactly the same for my '260. The LANCE refuses to receive further packets
after having received 3 (three) packets. Can be checked with defining
>> o How long does it take for init to start? It takes several _minutes_
>> my '133 and tcpdump doesn't show any activity on the wire.
> Yup multiple minutes ...
Ok, then this is not processor dependant.
>> o Can you switch the runlevel? Doesn't work on my '133, machine hangs
> It looks like it does - It boots into and then shows me a change to
> 3 and trys then to start a mingetty on /dev/ttyS2 which doesnt seem to
> work :)
>> o Does mount work with a vanilla /bin/mount? It didn't on my '133. This
>> no big deal. Replacing /bin/mount with an rpm from ftp.cobaltmicro.com
>> _did_ work.
> It does ... I tried to replace the init by the one on CobaltMicro
> but that doesnt help so - What on earth is the init doing ?
I am beginning to wonder. I built a statically linked init for an R3000
target and this is *way* faster, but I doubt that this is a ld.so problem.
init executes some shell scripts and bash is dynamically linked...
Anyway, I am going to clean up my source tree a little bit.