questions about current RM9200 state

[Bernd Walter]

On Fri, Dec 08, 2006 at 10:32:00PM -0700, M. Warner Losh wrote:
> In message: <20061208114906.GJ54209 at cicely12.cicely.de>
>             Bernd Walter <ticso at cicely12.cicely.de> writes:
> : On Fri, Dec 08, 2006 at 11:42:28AM +0100, Bernd Walter wrote:
> : > On Thu, Dec 07, 2006 at 08:50:08PM -0700, M. Warner Losh wrote:
> : > > In message: <20061208015032.GF54209 at cicely12.cicely.de>
> : > >             Bernd Walter <ticso at cicely12.cicely.de> writes:
> : > > : But I can't probe the bus from userland:
> : > > : [56]arm9# ./iic_probe /dev/iic0 
> : > > : - and nothing listed.
> : > > 
> : > > twi doesn't support START/STOP ioctls that you use.  Hmmm, that's in
> : > > p4 that I've removed them because I never got that working.  I also
> : > > think the version in -head isn't the latest, working one.  I think you
> : > > do the same thing with a 0 byte transfer.  Except we don't detect
> : > > NOACK quite right yet, I'm afraid.  I'd expect an error each time you
> : > > tried.
> : > 
> : > Ah - Ok, will switch to 0 Byte transfers and retest.
> : 
> : iic.c:
> :         case I2CWRITE:
> :                 if (s->count <= 0) {
> :                         error = EINVAL;
> :                         break;
> :                 }
> : 
> : Well - it wouldn't work without NOACK detection anyway.
> 
> I think a zero-byte transfer would do it.

Unfortunately the TWI controller can't do zero transfers at all :(
I thought the RM9200 has different one, but it is the same ^$%@ as in
the AT91SAM7 series, which I already had some hard time with.
The TWI controller in the AVR series is much better IMHO.
I'll send you a diff later today to fix some small bugs that I've found
in the driver - those points aren't documented very well in the
handbook.

-- 
B.Walter                http://www.bwct.de      http://www.fizon.de
bernd at bwct.de           info at bwct.de            support at fizon.de