>>>>>I think when colin has tidied things up we should check in the 'final' 
>>>>>2.4
>>>>>version on balloonbaord and start a 2.6.recent which we are actaully 
>>>>>going
>>>>>to try and push upstream. Then tcl/engdept/anyone can just keep thier own
>>>>>diff if they need one.
>>>
>>This final version. Will it contain (suitably config managed) the samosa 
>>stuff? I would hope so as its very handy for balloon3 prototyping. If 
>>anyone wants to rip off the SM socket they have a samosa bus so it is 
>>not entirely a forked board.
> 
> 
> We haven't been taking code out, so it's still there. Not currently tested.
> 
Ok thanks.

> So on a 2.052 board one can have samosa simply by wiring up to the pads of
> the Smartmedia socket? So in fact it should be fine to leave samosa compiled
> in by default? If that is true we need to find out why the kernels we have
> been building don't appear to boot if samosa is configured. Are there
> corresponding CPLD changes? I had been under the mistaken impression that
> samosa implied hardware changes.
> 
There are some hardware changes with necessary CPLD changes due to new 
handshake lines and UDC control lines (connect/disconnect) but I think 
samosa per se is not affected. I have a niggle about a subtle 
interaction that may be a problem but this may well be realted to boards 
with a SM card inserted being mis-identified.

> 
>>>Indeed. And we appear to be making some progress on this front. Good stuff
>>>colin.
> 
> 
>>Thats all well and good so long at there is UDC support. If there isn't 
>>for sa1110 then TCL cannot use 2.6 in balloon2.
> 
> 
> 'Progress' does not imply 'finished'. Obviously broken USB is something that
> must be fixed, but it is not a prerequisite to getting our various houses in
> order and starting pushing up to mainline. Waiting till everything is
> finished is the classic reason why one gets left behind with a fork. We can
> work on both these things in parallel.
> 
Agreed entirely. My point was that although we have had 2.6.x running 
for some time - I have published that and is readily available - what 
was stopping TCL adopting it was the broken UDC support.

> [re-directed to list due to generally useful info]
> 
> Wookey
Nick