[PATCH 1/2] [ARM] Kirkwood: enhance TLCK detection

[Lennert Buytenhek]

On Thu, Oct 21, 2010 at 01:03:26PM +0200, Saeed Bishara wrote:

> >> >> -	if ((dev == MV88F6281_DEV_ID && (rev == MV88F6281_REV_A0 ||
> >> >> -					rev == MV88F6281_REV_A1)) ||
> >> >> -	    (dev == MV88F6282_DEV_ID))
> >> >> -		return 200000000;
> >> >> +	if (dev == MV88F6281_DEV_ID || dev == MV88F6282_DEV_ID)
> >> >> +		if ((readl(SAMPLE_AT_RESET) >> 21 & 1) == 0)
> >> >> +			return 200000000;
> >> >>  
> >> >>  	return 166666667;
> >> >>  }
> >> >
> >> >Can we not just make this:
> >> >
> >> >	if ((readl(SAMPLE_AT_RESET) >> 21 & 1) == 0)
> >> >		return 200000000;
> >> >	else
> >> >		return 166666667;
> >> >
> >> >(I don't have the necessary docs, Saeed might have to 
> >answer this one.)
> >>
> >> Lennert, the docs and our reference code shows that the 
> >original code of Simor is correct
> >
> >So what does it mean if bit 21 is 0 on 6180/6192?
> >
> >If [21] == 0 is not supported on 6180/6192 because those parts
> >officially only support 166 MHz TCLK operation, then that doesn't
> >need a test for the part number, I think.
>
> The spec says that bit 21 for 6180 is for internal testing, so lets
> don't assume any thing about it.

OK, I agree.