Re: [PATCH 2/2] phy: qcom: Introduce new eDP PHY driver

From: sbillaka@codeaurora.org
To: Bjorn Andersson <bjorn.andersson@linaro.org>
Cc: Andy Gross <agross@kernel.org>,
	Kishon Vijay Abraham I <kishon@ti.com>,
	Vinod Koul <vkoul@kernel.org>, Rob Herring <robh+dt@kernel.org>,
	Rob Clark <robdclark@gmail.com>,
	Dmitry Baryshkov <dmitry.baryshkov@linaro.org>,
	Stephen Boyd <swboyd@chromium.org>,
	linux-arm-msm@vger.kernel.org, linux-phy@lists.infradead.org,
	devicetree@vger.kernel.org, linux-kernel@vger.kernel.org,
	kalyan_t@codeaurora.org, abhinavk@codeaurora.org
Subject: Re: [PATCH 2/2] phy: qcom: Introduce new eDP PHY driver
Date: Thu, 12 Aug 2021 05:53:43 +0530	[thread overview]
Message-ID: <04267264ac7904733552d6ca02a6797b@codeaurora.org> (raw)
In-Reply-To: <YRH0WFvWmemHIHqc@builder.lan>

On 2021-08-10 09:06, Bjorn Andersson wrote:
> On Mon 09 Aug 22:15 CDT 2021, sbillaka@codeaurora.org wrote:
>> On 2021-05-11 09:49, Bjorn Andersson wrote:
>> > diff --git a/drivers/phy/qualcomm/phy-qcom-edp.c
> [..]
>> > +#define DP_PHY_AUX_CFG0				0x0020
>> > +#define DP_PHY_AUX_CFG1				0x0024
>> > +#define DP_PHY_AUX_CFG2				0x0028
>> > +#define DP_PHY_AUX_CFG3				0x002c
>> > +#define DP_PHY_AUX_CFG4				0x0030
>> > +#define DP_PHY_AUX_CFG5				0x0034
>> > +#define DP_PHY_AUX_CFG6				0x0038
>> > +#define DP_PHY_AUX_CFG7				0x003c
>> > +#define DP_PHY_AUX_CFG8				0x0040
>> > +#define DP_PHY_AUX_CFG9				0x0044
>> 
>> The DP_PHY_AUX_CFG0 offset for sc8180x eDP phy is 0x0024.
>> Some of the eDP PHY offset addresses are shifted by 4 address 
>> locations,
>> compared to the DP QMP PHY offset addresses for sc8180x.
>> The DP_PHY_AUX_CFG* offsets for this eDP phy driver are as below:
>> 
>> #define DP_PHY_AUX_CFG0                         0x0024
>> #define DP_PHY_AUX_CFG1                         0x0028
>> #define DP_PHY_AUX_CFG2                         0x002c
>> #define DP_PHY_AUX_CFG3                         0x0030
>> #define DP_PHY_AUX_CFG4                         0x0034
>> #define DP_PHY_AUX_CFG5                         0x0038
>> #define DP_PHY_AUX_CFG6                         0x003c
>> #define DP_PHY_AUX_CFG7                         0x0040
>> #define DP_PHY_AUX_CFG8                         0x0044
>> #define DP_PHY_AUX_CFG9                         0x0048
>> 
> 
> I noticed this as well. During development I just used the numbers
> directly in the code and I must have screwed up as I replaced them with
> defined - and somehow missed this in the testing before posting.
> 
> Sorry about that.
> 
> [..]
>> > +static int qcom_edp_phy_init(struct phy *phy)
>> > +{
>> > +	struct qcom_edp *edp = phy_get_drvdata(phy);
>> > +	int ret;
>> > +
>> > +	ret = regulator_bulk_enable(ARRAY_SIZE(edp->supplies), edp->supplies);
>> > +	if (ret)
>> > +		return ret;
>> > +
>> > +	ret = clk_bulk_prepare_enable(ARRAY_SIZE(edp->clks), edp->clks);
>> > +	if (ret)
>> > +		goto out_disable_supplies;
>> 
>> I think the number of clk and regulator resources can vary based on
>> platform.
>> 
> 
> If that's the case we should replace the ARRAY_SIZE() with an integer.
> But I prefer to wait with that until the number actually is variable.
> 
> [..]
>> > +static int qcom_edp_phy_probe(struct platform_device *pdev)
>> > +{
>> > +	struct phy_provider *phy_provider;
>> > +	struct device *dev = &pdev->dev;
>> > +	struct qcom_edp *edp;
>> > +	int ret;
>> > +
>> > +	edp = devm_kzalloc(dev, sizeof(*edp), GFP_KERNEL);
>> > +	if (!edp)
>> > +		return -ENOMEM;
>> > +
>> > +	edp->dev = dev;
>> > +
>> > +	edp->edp = devm_platform_ioremap_resource(pdev, 0);
>> > +	if (IS_ERR(edp->edp))
>> > +		return PTR_ERR(edp->edp);
>> > +
>> > +	edp->tx0 = devm_platform_ioremap_resource(pdev, 1);
>> > +	if (IS_ERR(edp->tx0))
>> > +		return PTR_ERR(edp->tx0);
>> > +
>> > +	edp->tx1 = devm_platform_ioremap_resource(pdev, 2);
>> > +	if (IS_ERR(edp->tx1))
>> > +		return PTR_ERR(edp->tx1);
>> > +
>> > +	edp->pll = devm_platform_ioremap_resource(pdev, 3);
>> > +	if (IS_ERR(edp->pll))
>> > +		return PTR_ERR(edp->pll);
>> > +
>> > +	edp->clks[0].id = "aux";
>> > +	edp->clks[1].id = "cfg_ahb";
>> > +	ret = devm_clk_bulk_get(dev, ARRAY_SIZE(edp->clks), edp->clks);
>> > +	if (ret)
>> > +		return ret;
>> > +
>> > +	edp->supplies[0].supply = "vdda-phy";
>> > +	edp->supplies[1].supply = "vdda-pll";
>> > +	ret = devm_regulator_bulk_get(dev, ARRAY_SIZE(edp->supplies),
>> > edp->supplies);
>> > +	if (ret)
>> > +		return ret;
>> 
>> I believe, the combination of the number of regulator and clk 
>> resources may
>> vary based on the platform.
>> I think we should not fail probe if all these resources are not 
>> present in
>> the device tree file.
>> I think, these resources can be optional. We can get these resources 
>> if they
>> are present in the device tree file and enable them as required.
>> 
> 
> It's quite helpful to the DTS writer to actually encode in the driver
> which resources the driver expects and provide useful error messages
> when these expectations aren't met - so I think in line with most other
> drivers this should be decided based on the compatible.
> 
> What clocks and regulators do you have on sc7280?

We have only one clock (edp refclk) and one regulator (phy-0p9) for 
sc7280.

> 
> 
> Thanks for the feedback, I see that I have a few more pieces of 
> feedback
> from others that I need to incorporate. I'll make sure to do that and
> repost this patch shortly.

Okay. Thank you.

> 
> Regards,
> Bjorn