Date: Wed, 1 Jun 2011 00:08:23 +0800
From: Shawn Guo <shawn.guo@freescale.com>
To: Arnd Bergmann <arnd@arndb.de>
CC: Olof Johansson <olof@lixom.net>, Shawn Guo <shawn.guo@linaro.org>,
        <linux-kernel@vger.kernel.org>, <grant.likely@secretlab.ca>,
        <kernel@pengutronix.de>, <linux-arm-kernel@lists.infradead.org>,
        <patches@linaro.org>
Subject: Re: [PATCH 2/4] ARM: mxc: migrate mach-mx5 gpio driver to gpio-mxc
Message-ID: <20110531160822.GA32411@S2100-06.ap.freescale.net>
References: <1306767139-24763-1-git-send-email-shawn.guo@linaro.org>
 <201105311328.18012.arnd@arndb.de>
 <20110531151019.GA32275@S2100-06.ap.freescale.net>
 <201105311727.55985.arnd@arndb.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Disposition: inline
In-Reply-To: <201105311727.55985.arnd@arndb.de>
User-Agent: Mutt/1.5.21 (2010-09-15)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2641
Lines: 59

On Tue, May 31, 2011 at 05:27:55PM +0200, Arnd Bergmann wrote:
> On Tuesday 31 May 2011, Shawn Guo wrote:
> > 
> > On Tue, May 31, 2011 at 01:28:17PM +0200, Arnd Bergmann wrote:
> > > On Tuesday 31 May 2011, Shawn Guo wrote:
> > > > > Just open-code the mxc_add_mxc_gpio() by moving the individual calls to
> > > > > mxc_add_gpio() into the respective callers. Having a global
> > > > > mxc_add_mxc_gpio() function that does something different for each
> > > > > caller seems entirely pointless to me.
> > > > > 
> > > > Right now, mxc_add_mxc_gpio() is a postcore_initcall.  Moving
> > > > individual mxc_add_gpio() call into irq_init function does not work.
> > > > And I need to find a proper caller for each SoC to call mxc_add_gpio
> > > > to register gpio devices.
> > >  
> > > Why not init_machine? That is an arch_initcall(), so it's probably close
> > > enough.
> > > 
> > The init_machine is mostly a board specific function than SoC specific
> > one.  That is to say we will call mxc_add_gpio() in every single board
> > init function even for the same SoC.
> 
> But the machine is the only place that knows what SOC is being used.
> Your patch right now detects it by looking at the CPU type that is
> also being set by the board file using the init_early call, which
> is a bit silly.
> 
I would say that the CPU type is being set by SoC file (e.g.
mach-mx5/mm.c) than board file.  All the things in mach-mx5/mm.c are
common to any mx51 based boards.

> I would leave e.g. the imx51_register_gpios() in place, but only
> change the definition and the caller to
> 
> void __init imx51_register_gpios(void)
> {
>         mxc_add_gpio(0, MX51_GPIO1_BASE_ADDR, SZ_16K, MX51_MXC_INT_GPIO1_LOW, MX51_MXC_INT_GPIO1_HIGH);
>         mxc_add_gpio(1, MX51_GPIO2_BASE_ADDR, SZ_16K, MX51_MXC_INT_GPIO2_LOW, MX51_MXC_INT_GPIO2_HIGH);
>         mxc_add_gpio(2, MX51_GPIO3_BASE_ADDR, SZ_16K, MX51_MXC_INT_GPIO3_LOW, MX51_MXC_INT_GPIO3_HIGH);
>         mxc_add_gpio(3, MX51_GPIO4_BASE_ADDR, SZ_16K, MX51_MXC_INT_GPIO4_LOW, MX51_MXC_INT_GPIO4_HIGH);
> }
> 
> Then you can call that function from each i.mx51 based board, or
> from a new imx51_soc_init() function that groups multiple such
> functions.
> 
It will anyway touch every single board file (a lot) to have this SoC
specific device registration plugged in.  If you still think it's
worthy to do so, I can turn around.

-- 
Regards,
Shawn

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/