From mboxrd@z Thu Jan  1 00:00:00 1970
From: heiko@sntech.de (Heiko =?ISO-8859-1?Q?St=FCbner?=)
Date: Mon, 17 Feb 2014 23:48:45 +0100
Subject: [PATCH 08/12] clk: samsung: add clock controller driver for
 s3c2410, s3c2440 and s3c2442
In-Reply-To: <53028F1C.7090806@gmail.com>
References: <201312131356.40755.heiko@sntech.de> <5643318.r2EB20boll@phil>
 <53028F1C.7090806@gmail.com>
Message-ID: <1854479.bubKgKa3MN@phil>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

Am Montag, 17. Februar 2014, 23:37:16 schrieb Tomasz Figa:
> On 17.02.2014 22:05, Heiko St?bner wrote:
> > Hi Tomasz,
> > 
> > Am Sonntag, 9. Februar 2014, 20:34:26 schrieb Tomasz Figa:
> >>> +/*
> >>> + * fixed rate clocks generated outside the soc
> >>> + * Only necessary until the devicetree-move is complete
> >>> + */
> >>> +struct samsung_fixed_rate_clock s3c2410_common_frate_clks[] __initdata
> >>> =
> >>> {
> >>> +	FRATE(XTI, "xti", NULL, CLK_IS_ROOT, 0),
> >> 
> >> Do you need to assign this clock an ID? When DT is used, it's defined as
> >> a separate clock provider using fixed clock bindings, so it is not
> >> provided by this driver.
> > 
> > I think that I'll need this (and similar on the s3c2412). This is due to
> > the cpufreq driver for the earlier SoCs needing the xtal-frequency to
> > calculate the pll lock time (s3c24xx-cpufreq.c
> > s3c_cpufreq_update_loctkime).
> > 
> > Or do you have an idea on how to fix up the cpufreq driver in an easy way?
> 
> OK, I get the problem here. So you need the ID to register an alias.
> 
> Still, the ID would be invalid when booted with DT and xti clock
> registered using fixed rate clock bindings. So I'd suggest moving the
> XTI clock ID definition from your DT binding header back to your driver,
> leaving a comment in the header that this ID is reserved.
> 
> What do you think?

nice idea, thanks :-). I'll use it for the s3c2410 (+ variants) and s3c2412.


Heiko