Re: [PATCH v2 2/2] PCI: qcom: Add support for SDM845 PCIe controller

[Bjorn Andersson]

On Mon 04 Nov 01:41 PST 2019, Philipp Zabel wrote:

> Hi Bjorn,
> 
> On Fri, 2019-11-01 at 17:27 -0700, Bjorn Andersson wrote:
> > The SDM845 has one Gen2 and one Gen3 controller, add support for these.
> > 
> > Signed-off-by: Bjorn Andersson <bjorn.andersson@xxxxxxxxxx>
> > ---
> > 
> > Changes since v1:
> > - Style changes requested by Stan
> > - Tested with second PCIe controller as well
> > 
> >  drivers/pci/controller/dwc/pcie-qcom.c | 152 +++++++++++++++++++++++++
> >  1 file changed, 152 insertions(+)
> > 
> > diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c
> > index 7e581748ee9f..35f4980480bb 100644
> > --- a/drivers/pci/controller/dwc/pcie-qcom.c
> > +++ b/drivers/pci/controller/dwc/pcie-qcom.c
> > @@ -54,6 +54,7 @@
> [...]
> > +static int qcom_pcie_init_2_7_0(struct qcom_pcie *pcie)
> > +{
> > +	struct qcom_pcie_resources_2_7_0 *res = &pcie->res.v2_7_0;
> > +	struct dw_pcie *pci = pcie->pci;
> > +	struct device *dev = pci->dev;
> > +	u32 val;
> > +	int ret;
> > +
> > +	ret = regulator_bulk_enable(ARRAY_SIZE(res->supplies), res->supplies);
> > +	if (ret < 0) {
> > +		dev_err(dev, "cannot enable regulators\n");
> > +		return ret;
> > +	}
> > +
> > +	ret = clk_bulk_prepare_enable(ARRAY_SIZE(res->clks), res->clks);
> > +	if (ret < 0)
> > +		goto err_disable_regulators;
> > +
> > +	ret = reset_control_assert(res->pci_reset);
> > +	if (ret < 0) {
> > +		dev_err(dev, "cannot deassert pci reset\n");
> > +		goto err_disable_clocks;
> > +	}
> 
> If for any of the above fails, the reset line is left in its default
> state, presumably unasserted. Is there a reason to assert and keep it
> asserted if enabling the clocks fails below?
> 

No, I don't think there's any reason for doing this and looking at the
downstream driver, they don't even propagate this error.

> > +	msleep(20);

And I see now that downstream has this as 1ms, will update and retest
again.

> > +
> > +	ret = reset_control_deassert(res->pci_reset);
> > +	if (ret < 0) {
> > +		dev_err(dev, "cannot deassert pci reset\n");
> > +		goto err_assert_resets;
> 
> Nitpick: this seems superfluous since the reset line was just asserted
> 20 ms before. Maybe just:
> 
> 		goto err_disable_clocks;

Yes, this seems reasonable.

> 
> > +	}
> > +
> > +	ret = clk_prepare_enable(res->pipe_clk);
> > +	if (ret) {
> > +		dev_err(dev, "cannot prepare/enable pipe clock\n");
> > +		goto err_assert_resets;
> > +	}
> > +
> > +	/* configure PCIe to RC mode */
> > +	writel(DEVICE_TYPE_RC, pcie->parf + PCIE20_PARF_DEVICE_TYPE);
> > +
> > +	/* enable PCIe clocks and resets */
> > +	val = readl(pcie->parf + PCIE20_PARF_PHY_CTRL);
> > +	val &= ~BIT(0);
> > +	writel(val, pcie->parf + PCIE20_PARF_PHY_CTRL);
> > +
> > +	/* change DBI base address */
> > +	writel(0, pcie->parf + PCIE20_PARF_DBI_BASE_ADDR);
> > +
> > +	/* MAC PHY_POWERDOWN MUX DISABLE  */
> > +	val = readl(pcie->parf + PCIE20_PARF_SYS_CTRL);
> > +	val &= ~BIT(29);
> > +	writel(val, pcie->parf + PCIE20_PARF_SYS_CTRL);
> > +
> > +	val = readl(pcie->parf + PCIE20_PARF_MHI_CLOCK_RESET_CTRL);
> > +	val |= BIT(4);
> > +	writel(val, pcie->parf + PCIE20_PARF_MHI_CLOCK_RESET_CTRL);
> > +
> > +	if (IS_ENABLED(CONFIG_PCI_MSI)) {
> > +		val = readl(pcie->parf + PCIE20_PARF_AXI_MSTR_WR_ADDR_HALT);
> > +		val |= BIT(31);
> > +		writel(val, pcie->parf + PCIE20_PARF_AXI_MSTR_WR_ADDR_HALT);
> > +	}
> > +
> > +	return 0;
> > +err_assert_resets:
> > +	reset_control_assert(res->pci_reset);
> 
> So maybe this can just be removed. The reset isn't asserted in deinit
> either.
> 

Sounds good.

Thanks for your review, Philipp!

Regards,
Bjorn