[U-Boot-Users] CS8900 Problem on an ARM system

[Guennadi Liakhovetski]

I made an interesting observation... Out of frustration I dumped all 
chip's register (luckily, there are only about 20 of them), and what I see 
there: on first attempts the PolarityOK bit is set, and reception doesn't 
work. Then after some time the bit gets cleared, and - voila - it works! 
The chip does have automatic polarity detection, if I disable it, it 
_never_ receives a frame, that's what made me suspicious in the first 
place. And now the confirmation. I'll have to check if the board is 
connected with a crossover cable to the switch... But why does the 
automatic detection need so long??? I just tried again just waiting for a 
couple of minutes without doing anything, and already on the first access 
the chip had polarity correctly and ping worked. So, it either needs about 
a minute idle time (ok, maybe there are some broadcasts taking place in 
this time), or some activity. That also explains why pinging the board 
from outside helps too.

But why so long???

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.

DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: +49-8142-66989-0 Fax: +49-8142-66989-80  Email: office at denx.de