From mboxrd@z Thu Jan  1 00:00:00 1970
From: dinh.linux@gmail.com (Dinh Nguyen)
Date: Thu, 26 Sep 2013 22:27:55 -0500
Subject: [PATCH 1/2] dts: socfpga: Add ethernet support on Altera's SOCFPGA
 platform
In-Reply-To: <5244C171.2040705@wwwdotorg.org>
References: <1380057120-27108-1-git-send-email-dinguyen@altera.com>
 <5244C171.2040705@wwwdotorg.org>
Message-ID: <5244FB3B.2050906@gmail.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

Hi Stephen,

On 9/26/13 6:21 PM, Stephen Warren wrote:
> On 09/24/2013 03:11 PM, dinguyen at altera.com wrote:
>> From: Dinh Nguyen <dinguyen@altera.com>
>>
>> The STMMAC Ethernet controller in SOCFPGA requires setting a register for
>> the phy-mode that is outside of the ethernet IP. This register resides in
>> the System Manager block. So we define a new DTS binding
>> "altr,sysmgr-phy-mask". This binding's property is a bitmask that can be
>> used to set the correct register bit.
>> diff --git a/Documentation/devicetree/bindings/net/stmmac-altr-socfpga.txt b/Documentation/devicetree/bindings/net/stmmac-altr-socfpga.txt
>> +* altr,sysmgr-phy-mask: This property contains the bitmask that is needed to
>> +	set the appropriate register bits for the phy-mode in the System Manager.
>> +	The value should be:
>> +		-Ethernet Controller 1 (gmac0) = 0x3
>> +		-Ethernet Controller 2 (gmac1) = 0xC
> Wouldn't you need a phandle to the sysmgr node so that the driver could
> be located, and a register number within its register block too? Or,
> does sysmgr know which register to poke? If so, couldn't the API take
> just a device index rather than a bitmask instead, and calculate the
> mask itself?
The function that pokes this register is a platform specific init 
function call from the driver, so the function knows which register in 
the the sysmgr to poke. The first byte of the register has controls for 
both ethernet controllers. I thought it would be better to use a mask to 
be used to differentiate between each controller.

Thanks,
Dinh