Re: [PATCH 05/11] spi/pxa2xx: make clock rate configurable from platform data

[Mika Westerberg <mika.westerberg@linux.intel.com>]

On Thu, Jan 10, 2013 at 02:23:25PM +0100, Rafael J. Wysocki wrote:
> On Thursday, January 10, 2013 03:07:40 PM Mika Westerberg wrote:
> > On Thu, Jan 10, 2013 at 12:51:59PM +0000, Mark Brown wrote:
> > > On Thu, Jan 10, 2013 at 01:54:41PM +0100, Rafael J. Wysocki wrote:
> > > > On Thursday, January 10, 2013 02:38:37 PM Mika Westerberg wrote:
> > > 
> > > > > 	3. We make the acpi_create_platform_device() match on, lets say
> > > > > 	   "INT33C" (a partial match), and in such case it assumes that we are
> > > > >            running on Lynxpoint. It will then create platform device for 'clk-lpt'.
> > > 
> > > > > 	4. Now the clk-lpt driver creates the clocks.
> > > 
> > > > > 	5. The SPI driver gets the clock it wants.
> > > 
> > > > That sounds reasonable to me.  Mark, what do you think?
> > > 
> > > Sounds sensible, yes - about what I'd expect.  Is it possible to match
> > > on CPUID or similar information (given that this is all in the SoC)
> > > instead of ACPI, that might be more robust I guess?
> > 
> > I can look into that but I'm not sure whether there are any other way to
> > detect are we running on Lynxpoint or not, except the device IDs (and even
> > that is not 100% guaranteed because of ACPI _CIDs).
> 
> Well, we only need the clock when the SPI controller is going to be used,
> so even if we have a reliable way to detect Lynxpoint, that may be not enough
> (the BIOS may not expose the SPI to us, for example, in which case it will be
> pointless to create the clock for it).

Good point. I'll do the checking in acpi_create_platform_device() based on
ACPI IDs so that we can be sure that the SPI controller is really there.