Re: [PATCH] PCI: exynos: Adapt to clk_bulk_* APIs

['Manivannan Sadhasivam']

On Wed, Nov 15, 2023 at 10:06:19AM +0100, Marek Szyprowski wrote:
> Hi!
> 
> On 15.11.2023 07:40, Shradha Todi wrote:
> >> -----Original Message-----
> >> From: Manivannan Sadhasivam [mailto:manivannan.sadhasivam@xxxxxxxxxx]
> >> Sent: 27 October 2023 19:19
> >> To: Shradha Todi<shradha.t@xxxxxxxxxxx>
> >> Cc:jingoohan1@xxxxxxxxx;lpieralisi@xxxxxxxxxx;kw@xxxxxxxxx;
> >> robh@xxxxxxxxxx;bhelgaas@xxxxxxxxxx;krzysztof.kozlowski@xxxxxxxxxx;
> >> alim.akhtar@xxxxxxxxxxx;linux-pci@xxxxxxxxxxxxxxx; linux-arm-
> >> kernel@xxxxxxxxxxxxxxxxxxx;linux-samsung-soc@xxxxxxxxxxxxxxx; linux-
> >> kernel@xxxxxxxxxxxxxxx;pankaj.dubey@xxxxxxxxxxx
> >> Subject: Re: [PATCH] PCI: exynos: Adapt to clk_bulk_* APIs
> >>
> >> On Mon, Oct 09, 2023 at 11:52:16AM +0530, Shradha Todi wrote:
> >>> There is no need to hardcode the clock info in the driver as driver
> >>> can rely on the devicetree to supply the clocks required for the
> >>> functioning of the peripheral. Get rid of the static clock info and
> >>> obtain the platform supplied clocks. The total number of clocks
> >>> supplied is obtained using the devm_clk_bulk_get_all() API and used
> >>> for the rest of the clk_bulk_* APIs.
> >>>
> >>> Signed-off-by: Shradha Todi<shradha.t@xxxxxxxxxxx>
> >>> ---
> >>>   drivers/pci/controller/dwc/pci-exynos.c | 46
> >>> ++++++-------------------
> >>>   1 file changed, 11 insertions(+), 35 deletions(-)
> >>>
> >>> diff --git a/drivers/pci/controller/dwc/pci-exynos.c
> >>> b/drivers/pci/controller/dwc/pci-exynos.c
> >>> index 9e42cfcd99cc..023cf41fccd7 100644
> >>> --- a/drivers/pci/controller/dwc/pci-exynos.c
> >>> +++ b/drivers/pci/controller/dwc/pci-exynos.c
> >>> @@ -54,8 +54,8 @@
> >>>   struct exynos_pcie {
> >>>   	struct dw_pcie			pci;
> >>>   	void __iomem			*elbi_base;
> >>> -	struct clk			*clk;
> >>> -	struct clk			*bus_clk;
> >>> +	struct clk_bulk_data		*clks;
> >>> +	int				clk_cnt;
> >>>   	struct phy			*phy;
> >>>   	struct regulator_bulk_data	supplies[2];
> >>>   };
> >>> @@ -65,30 +65,18 @@ static int exynos_pcie_init_clk_resources(struct
> >> exynos_pcie *ep)
> >>>   	struct device *dev = ep->pci.dev;
> >>>   	int ret;
> >>>
> >>> -	ret = clk_prepare_enable(ep->clk);
> >>> -	if (ret) {
> >>> -		dev_err(dev, "cannot enable pcie rc clock");
> >>> +	ret = devm_clk_bulk_get_all(dev, &ep->clks);
> >>> +	if (ret < 0)
> >> Please use !(ret) here and below to be consistent with the driver.
> >>
> > In this case, only negative values indicate failure. Hence we cannot use (!ret) here.
> >
> >>>   		return ret;
> >>> -	}
> >>>
> >>> -	ret = clk_prepare_enable(ep->bus_clk);
> >>> -	if (ret) {
> >>> -		dev_err(dev, "cannot enable pcie bus clock");
> >>> -		goto err_bus_clk;
> >>> -	}
> >>> +	ep->clk_cnt = ret;
> >> Since clk_cnt is "int", you can just use it directly instead of "ret".
> >>
> > Thanks for this suggestion! Will take care in v2.
> >
> >>> -	return 0;
> >>> -
> >>> -err_bus_clk:
> >>> -	clk_disable_unprepare(ep->clk);
> >>> -
> >>> -	return ret;
> >>> +	return clk_bulk_prepare_enable(ep->clk_cnt, ep->clks);
> >>>   }
> >>>
> >>>   static void exynos_pcie_deinit_clk_resources(struct exynos_pcie *ep)
> >>> {
> >>> -	clk_disable_unprepare(ep->bus_clk);
> >>> -	clk_disable_unprepare(ep->clk);
> >>> +	clk_bulk_disable_unprepare(ep->clk_cnt, ep->clks);
> >>>   }
> >>>
> >>>   static void exynos_pcie_writel(void __iomem *base, u32 val, u32 reg)
> >>> @@ -332,17 +320,9 @@ static int exynos_pcie_probe(struct
> >> platform_device *pdev)
> >>>   	if (IS_ERR(ep->elbi_base))
> >>>   		return PTR_ERR(ep->elbi_base);
> >>>
> >>> -	ep->clk = devm_clk_get(dev, "pcie");
> >>> -	if (IS_ERR(ep->clk)) {
> >>> -		dev_err(dev, "Failed to get pcie rc clock\n");
> >>> -		return PTR_ERR(ep->clk);
> >>> -	}
> >>> -
> >>> -	ep->bus_clk = devm_clk_get(dev, "pcie_bus");
> >>> -	if (IS_ERR(ep->bus_clk)) {
> >>> -		dev_err(dev, "Failed to get pcie bus clock\n");
> >>> -		return PTR_ERR(ep->bus_clk);
> >>> -	}
> >>> +	ret = exynos_pcie_init_clk_resources(ep);
> >>> +	if (ret < 0)
> >>> +		return ret;
> >>>
> >>>   	ep->supplies[0].supply = "vdd18";
> >>>   	ep->supplies[1].supply = "vdd10";
> >>> @@ -351,10 +331,6 @@ static int exynos_pcie_probe(struct
> >> platform_device *pdev)
> >>>   	if (ret)
> >>>   		return ret;
> >>>
> >>> -	ret = exynos_pcie_init_clk_resources(ep);
> >>> -	if (ret)
> >>> -		return ret;
> >>> -
> >>>   	ret = regulator_bulk_enable(ARRAY_SIZE(ep->supplies), ep-
> >>> supplies);
> >>>   	if (ret)
> >> You need to disable_unprepare() clocks in error path here and above.
> >>
> > Thanks for pointing out! Will take care in v2.
> 
> 
> Maybe it would make sense to add devm_clk_bulk_get_all_enabled() to 
> clock framework, similar to the existing devm_clk_get_enabled()?
> 
> It is really a common pattern to get all clocks and enable them for the 
> time of driver operation.
> 

Right. Someone may argue that the users would need to check the number of clocks
returned by the devm_clk_bulk_get_all() API, before enabling them. But I don't
think the drivers should check those values as they need to rely on the
firmware (unless accessing the clocks manually later). Even for those cases, the
individual APIs can be used.

So IMO it is worth to give it a shot.

- Mani

> Best regards
> -- 
> Marek Szyprowski, PhD
> Samsung R&D Institute Poland

-- 
மணிவண்ணன் சதாசிவம்