Subject: Re: [PATCH] phy: rockchip-emmc: fix compile issue on arm64 platform
To: Kishon Vijay Abraham I <kishon@ti.com>
References: <1457425453-1996-1-git-send-email-shawn.lin@rock-chips.com>
 <570659EA.9000706@ti.com> <57066137.4040807@ti.com>
Cc: shawn.lin@rock-chips.com, linux-kernel@vger.kernel.org,
        Heiko Stuebner <heiko@sntech.de>, linux-rockchip@lists.infradead.org
From: Shawn Lin <shawn.lin@rock-chips.com>
Message-ID: <570705FE.2000202@rock-chips.com>
Date: Fri, 8 Apr 2016 09:14:38 +0800
User-Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:38.0) Gecko/20100101
 Thunderbird/38.7.0
MIME-Version: 1.0
In-Reply-To: <57066137.4040807@ti.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 3492
Lines: 97

在 2016/4/7 21:31, Kishon Vijay Abraham I 写道:
> Hi,
>
> On Thursday 07 April 2016 06:30 PM, Kishon Vijay Abraham I wrote:
>> Hi,
>>
>> On Tuesday 08 March 2016 01:54 PM, Shawn Lin wrote:
>>> This patch rename "reg" property to "reg_offset".
>>> We rename it to fix the compile issue on ARM64 platform:
>>> (reg_format): "reg" property in /phy has invalid length (4 bytes)
>>> (#address-cells == 2, #size-cells == 2)
>>
>> Is the same node used for both ARM32 and ARM64 platforms?
>>
>> Thanks
>> Kishon
>>>
>>> This's because "reg" is very special one which should keep the
>>> *-cells with its parent node and can't be overwrited even if we
>>> do that explicitly. On 32-bit plafform, the default *-cells
>>> fit for what we assign to "reg". But that's not correct for 64-bit
>>> platform. So we can see two possible solutions to fix this problem:
>>> A) make phy-rockchip-emmc as a child phy node and overwrite its
>>> parent's #address-cells and #size-cells.
>>> B) avoid using this special property.
>>>
>>> we use it just for passing on a offset for different Socs, and there's
>>> no requirement to change the code to make phy-rockchip-emmc as a child
>>> node. so choose option B) is sane.
>
> I just looked at the Heiko's patch and it makes more sense to have the binding
> that he described in his patch [1]. Can you fix it accordingly?

yes, Heiko's patch is more reasonable to me. With his patch applied,
this issue is gone if assigning address-cells in dt for grf.

So we can drop this patch.

>
> I seem to have only this patch and Heiko's patch for this -rc cycle. Once you
> send your patch, I can send a pull request to Greg.
>
> Thanks
> Kishon
>
> [1]- > https://patchwork.ozlabs.org/patch/601580/
>>>
>>> Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
>>> ---
>>>
>>>   Documentation/devicetree/bindings/phy/rockchip-emmc-phy.txt | 4 ++--
>>>   drivers/phy/phy-rockchip-emmc.c                             | 2 +-
>>>   2 files changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/Documentation/devicetree/bindings/phy/rockchip-emmc-phy.txt b/Documentation/devicetree/bindings/phy/rockchip-emmc-phy.txt
>>> index 61916f1..ed964ef 100644
>>> --- a/Documentation/devicetree/bindings/phy/rockchip-emmc-phy.txt
>>> +++ b/Documentation/devicetree/bindings/phy/rockchip-emmc-phy.txt
>>> @@ -6,7 +6,7 @@ Required properties:
>>>    - rockchip,grf : phandle to the syscon managing the "general
>>>      register files"
>>>    - #phy-cells: must be 0
>>> - - reg: PHY configure reg address offset in "general
>>> + - reg_offset: PHY configure reg address offset in "general
>>>      register files"
>>>
>>>   Example:
>>> @@ -14,6 +14,6 @@ Example:
>>>   emmcphy: phy {
>>>   	compatible = "rockchip,rk3399-emmc-phy";
>>>   	rockchip,grf = <&grf>;
>>> -	reg = <0xf780>;
>>> +	reg_offset = <0xf780>;
>>>   	#phy-cells = <0>;
>>>   };
>>> diff --git a/drivers/phy/phy-rockchip-emmc.c b/drivers/phy/phy-rockchip-emmc.c
>>> index 887b4c2..3f55c0d 100644
>>> --- a/drivers/phy/phy-rockchip-emmc.c
>>> +++ b/drivers/phy/phy-rockchip-emmc.c
>>> @@ -186,7 +186,7 @@ static int rockchip_emmc_phy_probe(struct platform_device *pdev)
>>>   	if (!rk_phy)
>>>   		return -ENOMEM;
>>>
>>> -	if (of_property_read_u32(dev->of_node, "reg", &reg_offset)) {
>>> +	if (of_property_read_u32(dev->of_node, "reg_offset", &reg_offset)) {
>>>   		dev_err(dev, "missing reg property in node %s\n",
>>>   			dev->of_node->name);
>>>   		return -EINVAL;
>>>
>
>
>


-- 
Best Regards
Shawn Lin