Jason Thorpe wrote:

> On Apr 27, 2007, at 8:56 PM, Michael Lorenz wrote:
> >
> > looks like the routines in dev/i2c/i2c_bitbang.c don't support  
> > synchronisation ( as in - when raising SCL wait until the SCL line  
> > is actually up )
> 
> Well, some devices supported by that code can't really tell, IIRC,  
> which is why it has delay loops instead.
> 
> > which gives endless troubles with slow i2c devices, like DDC2  
> > monitors.
> > Any objections against adding that ( with reasonable bail-out timer  
> > of course ) to i2c_bitbang_read_byte() and i2c_bitbang_write_byte() ?

The i2c spec doesn't mention timeouts at all, but a few SMBus
controllers I've seen have timeouts in the range of 25-35ms.

Cheers,
Simon.