Re: [PATCH 2/5] arm: shmobile: r7s72100: add i2c clocks

[Simon Horman <horms@verge.net.au>]

On Wed, Dec 18, 2013 at 01:15:42PM +0100, Wolfram Sang wrote:
> 
> On Wed, Dec 18, 2013 at 03:53:45PM +0400, Sergei Shtylyov wrote:
> > On 18-12-2013 15:43, Wolfram Sang wrote:
> > 
> > >>>@@ -173,6 +179,10 @@ static struct clk_lookup lookups[] = {
> > >>>  	CLKDEV_CON_ID("mtu2_fck", &mstp_clks[MSTP33]),
> > 
> > >>>  	/* ICK */
> > >>>+	CLKDEV_DEV_ID("fcfee000.i2c", &mstp_clks[MSTP97]),
> > >>>+	CLKDEV_DEV_ID("fcfee400.i2c", &mstp_clks[MSTP96]),
> > >>>+	CLKDEV_DEV_ID("fcfee800.i2c", &mstp_clks[MSTP95]),
> > >>>+	CLKDEV_DEV_ID("fcfeec00.i2c", &mstp_clks[MSTP94]),
> > 
> > >>    These belong to some other place, the group marked by /* ICK */
> > >>is only for CLKDEV_ICK_ID().
> > 
> > >So, I'll create a /* DEV */ prefix?
> > 
> >    I really don't know. Other places have /* MSTP */ comment in this
> > case despite all clocks, CLKDEV_DEV_ID() and CLKDEV_ICK_ID() are
> > really MSTP clocks. I considered the idea of separating
> > CLKDEV_ICK_ID() under /* ICK */ comment silly from the very start
> > but Simon didn't listen to me.
> 
> I am puzzled, too. ICK is a type of registration and not a clock domain.
> Also, there is 'mtu2_fck' which is under ICK as well as MSTP? Looks
> wrong. From what I understand now, removing the /* ICK */ comment would
> be easiest and proper?

I'm not sure that I really understand what all the fuss is about.

As I understand things the convention that prevails for
MSTP clocks under mach-shmobile is as follows:

1. Clocks not registered by CLKDEV_ICK_ID() are grouped together
   under /* MSTP */ followed by:
2. Clocks registered using CLKDEV_ICK_ID() are grouped together
   under /* ICK */

I am unsure of the historical reason for this but it does
seem to be consistent.