Message-ID: <51B5FC31.5050208@st.com>
Date: Mon, 10 Jun 2013 17:17:53 +0100
From: Srinivas KANDAGATLA <srinivas.kandagatla@st.com>
Reply-To: srinivas.kandagatla@st.com
Organization: STMicroelectronics
User-Agent: Mozilla/5.0 (X11; Linux i686; rv:17.0) Gecko/20130308 Thunderbird/17.0.4
MIME-Version: 1.0
To: Arnd Bergmann <arnd@arndb.de>
Cc: Mauro Carvalho Chehab <mchehab@redhat.com>, linux-doc@vger.kernel.org,
        linux@arm.linux.org.uk, Samuel Ortiz <sameo@linux.intel.com>,
        Stephen Gallimore <stephen.gallimore@st.com>,
        linux-serial@vger.kernel.org, Grant Likely <grant.likely@linaro.org>,
        devicetree-discuss@lists.ozlabs.org,
        Rob Herring <rob.herring@calxeda.com>,
        Stuart Menefy <stuart.menefy@st.com>, Mark Brown <broonie@kernel.org>,
        John Stultz <john.stultz@linaro.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        linux-arm-kernel@lists.infradead.org,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        linux-kernel@vger.kernel.org,
        Andrew Morton <akpm@linux-foundation.org>,
        "David S. Miller" <davem@davemloft.net>
Subject: Re: [PATCH v2 07/11] ARM:stixxxx: Add STiH416 SOC support
References: <1370855828-5318-1-git-send-email-srinivas.kandagatla@st.com> <1370856425-7135-1-git-send-email-srinivas.kandagatla@st.com> <3733079.FxbR14kFrs@wuerfel>
In-Reply-To: <3733079.FxbR14kFrs@wuerfel>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 3677
Lines: 102

Thankyou for your comments.
On 10/06/13 14:52, Arnd Bergmann wrote:
> On Monday 10 June 2013 10:27:05 Srinivas KANDAGATLA wrote:
> 
>> +	soc {
>> +		pin-controller-sbc {
>> +			#address-cells	= <1>;
>> +			#size-cells	= <1>;
>> +			compatible	= "st,stih416-pinctrl", "simple-bus";
> 
> Why is this both its own device with a compatible string and a
> "simple-bus" at the same time? Wouldn't it be simpler to just
> scan the child device nodes from the "st,stih416-pinctrl"
> driver instead of having a separate platform_driver for them?
Am happy to get rid of gpio platform_driver, But looking at the existing
pinctrl drivers like at91, they do it exactly like this.

Also having a gpio platform driver ties the resources to driver in a
neat way.

> 
>> +			st,retime-in-delay	= <0 300 500 750 1000 1250 1500 1750 2000 2250 2500 2750 3000 3250>;
>> +			st,retime-out-delay	= <0 300 500 750 1000 1250 1500 1750 2000 2250 2500 2750 3000 3250>;
>> +			st,syscfg		= <&syscfg_sbc>;
>> +			st,syscfg-offsets	= <0 40 50 60 100>;
>> +			ranges;
>> +			PIO0: pinctrl@fe610000 {
>> +				#gpio-cells = <1>;
>> +				compatible = "st,stixxxx-gpio";
>> +				gpio-controller;
>> +				reg = <0xfe610000 0x100>;
>> +				st,bank-name  = "PIO0";
>> +				st,retime-pin-mask = <0xff>;
>> +			};
>> +			PIO1: pinctrl@fe611000 {
>> +				#gpio-cells	= <1>;
>> +				compatible	= "st,stixxxx-gpio";
>> +				gpio-controller;
>> +				reg = <0xfe611000 0x100>;
>> +				st,bank-name  = "PIO1";
>> +				st,retime-pin-mask = <0xff>;
>> +			};
> 
> What is in the ranges between these registers? It seems you have
> 256 bytes for each pinctrl node, with 4kb spacing. I wonder if
> it would make sense to declare the entire range to belong to a single
> pinctrl device. At least since all of the registers are in a single
> range, you could add a property like
> 
> 	ranges = <0 0xfe610000 0x10000>;
> 
> and use relative addresses in the sub-nodes.
> 
OK, I will change to use ranges.
> Please don't use identifiers with 'xxx' in them. Instead use numbers
> of actual chips, ideally using the first one that this is compatible
> with.

Ok, I will change st,stixxxx-gpio to st,stih415-gpio.

> 
>> +		syscfg_sbc:syscfg@fe600000{
>> +			compatible	= "st,stih416-syscfg";
>> +			reg		= <0xfe600000 0x1000>;
>> +			syscfg-range	= <0 999>;
>> +			syscfg-name	= "SYSCFG_SBC";
>> +		};
>> +		syscfg_front:syscfg@fee10000{
>> +			compatible	= "st,stih416-syscfg";
>> +			reg		= <0xfee10000 0x1000>;
>> +			syscfg-range	= <1000 999>;
>> +			syscfg-name	= "SYSCFG_FRONT";
>> +		};
> 
> Did you mean to declare ranges excluding 1000 and 2000 here?
> Normally I would expect inclusive ranges like syscfg-range=<0 1000>;
> 
These numbers are from data sheet so I used it as it is.

> What is the idea of the 'syscfg-name'? If the nodes are all different,
The idea of having syscfg-name is to lookup any sysconf bank(regmap)
from code which do not have reference to phandle from device trees.

> I would expect them to have distinct "compatible" values and not
> need them.
Yes, If we have distinct compatible we would not need them, but there
will be 5-10 compatibility list for each SOC.
It looks like its going to be much neater Am going to try this change
and see how it looks like.
> 
> 	Arnd
> _______________________________________________
> devicetree-discuss mailing list
> devicetree-discuss@lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/devicetree-discuss
> 

--
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/