Re: [PATCH v2 4/5] PCI: qcom: Add Qualcomm PCIe controller driver

2015-11-09 Thread Stanimir Varbanov
Hi Bjorn,

On 11/06/2015 10:50 PM, Bjorn Andersson wrote:
> On Mon 04 May 05:42 PDT 2015, Stanimir Varbanov wrote:
> 
>> The PCIe driver reuse the Designware common code for host
>> and MSI initialization, and also program the Qualcomm
>> application specific registers.
>>
> 
> I want to get the ethernet on the ifc6410 running and this seems like
> the last patchset for the PCIe.
> 
>> Signed-off-by: Stanimir Varbanov 
> [..]
>> diff --git a/drivers/pci/host/pcie-qcom.c b/drivers/pci/host/pcie-qcom.c
>> new file mode 100644
>> index 000..4f083c6
>> --- /dev/null
>> +++ b/drivers/pci/host/pcie-qcom.c
>> @@ -0,0 +1,677 @@
>> +/*
>> + * Copyright (c) 2014, The Linux Foundation. All rights reserved.
> 
> Bump the year, it's the future now :)

yep, thanks for the valuable comments I will address all of them on next
version.

I will probably restart the work on the driver this week.

-- 
regards,
Stan
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html


Re: [PATCH v2 4/5] PCI: qcom: Add Qualcomm PCIe controller driver

2015-11-06 Thread Bjorn Andersson
On Mon 04 May 05:42 PDT 2015, Stanimir Varbanov wrote:

> The PCIe driver reuse the Designware common code for host
> and MSI initialization, and also program the Qualcomm
> application specific registers.
> 

I want to get the ethernet on the ifc6410 running and this seems like
the last patchset for the PCIe.

> Signed-off-by: Stanimir Varbanov 
[..]
> diff --git a/drivers/pci/host/pcie-qcom.c b/drivers/pci/host/pcie-qcom.c
> new file mode 100644
> index 000..4f083c6
> --- /dev/null
> +++ b/drivers/pci/host/pcie-qcom.c
> @@ -0,0 +1,677 @@
> +/*
> + * Copyright (c) 2014, The Linux Foundation. All rights reserved.

Bump the year, it's the future now :)

> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 and
> + * only version 2 as published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + */
> +
[..]
> +
> +struct qcom_pcie {
> + struct pcie_port pp;
> + struct device *dev;

You already have this device pointer in pp->dev.

> + union pcie_resources res;
> + void __iomem *parf;
> + void __iomem *dbi;
> + void __iomem *elbi;
> + struct phy *phy;
> + struct gpio_desc *reset;
> + unsigned int version;
> +};
> +
> +#define to_qcom_pcie(x)  container_of(x, struct qcom_pcie, pp)
> +
> +static inline void
> +writel_masked(void __iomem *addr, u32 clear_mask, u32 set_mask)
> +{
> + u32 val = readl(addr);
> +
> + val &= ~clear_mask;
> + val |= set_mask;
> + writel(val, addr);
> +}

There are no case where you do clear and set, so I would like that you
just inline this function where you need it. It adds 2 extra lines at a
few places, but a lot of clarity.

> +
> +static void qcom_ep_reset_assert_deassert(struct qcom_pcie *pcie, int assert)
> +{
> + int val, active_low;
> +
> + if (IS_ERR_OR_NULL(pcie->reset))
> + return;

This will be NULL or valid here, never ERR. So it's fine to call
gpiod_set_value() with this pointer without the check.

> +
> + active_low = gpiod_is_active_low(pcie->reset);
> +

gpiod_set_value() checks for FLAG_ACTIVE_LOW and will invert the value
first thing. So you do not need to do this manually.

> + if (assert)
> + val = !!active_low;
> + else
> + val = !active_low;
> +
> + gpiod_set_value(pcie->reset, val);
> +
> + usleep_range(PERST_DELAY_MIN_US, PERST_DELAY_MAX_US);

This doesn't seem to be called in a hot path, so you should be able to
simply msleep(1) here.

> +}
> +
> +static void qcom_ep_reset_assert(struct qcom_pcie *pcie)
> +{
> + qcom_ep_reset_assert_deassert(pcie, 1);

If we're down to this function just being a gpiod_set_value() and a
sleep we can inline it here instead of having this proxy function.

> +}
> +
> +static void qcom_ep_reset_deassert(struct qcom_pcie *pcie)
> +{
> + qcom_ep_reset_assert_deassert(pcie, 0);

Same here.

> +}
> +
> +static irqreturn_t qcom_pcie_msi_irq_handler(int irq, void *arg)
> +{
> + struct pcie_port *pp = arg;
> +
> + return dw_handle_msi_irq(pp);
> +}
> +
> +static int qcom_pcie_link_up(struct pcie_port *pp)
> +{
> + struct qcom_pcie *pcie = to_qcom_pcie(pp);
> + u32 val = readl(pcie->dbi + PCIE20_CAP_LINKCTRLSTATUS);
> +
> + return val & BIT(29) ? 1 : 0;

return !!(val & BIT(29));

Do we know what this bit 29 is?

> +}
> +
[..]
> +
> +static int qcom_pcie_get_resources_v1(struct qcom_pcie *pcie)
> +{
> + struct qcom_pcie_resources_v1 *res = &pcie->res.v1;
> + struct device *dev = pcie->dev;
> +
> + res->vdda = devm_regulator_get(dev, "vdda");
> + if (IS_ERR(res->vdda))
> + return PTR_ERR(res->vdda);
> +
> + res->iface = devm_clk_get(dev, "iface");
> + if (IS_ERR(res->iface))
> + return PTR_ERR(res->iface);
> +
> + res->aux = devm_clk_get(dev, "aux");
> + if (IS_ERR(res->aux) && PTR_ERR(res->aux) == -EPROBE_DEFER)

PTR_ERR() == x implies IS_ERR(), so you don't need both.

> + return -EPROBE_DEFER;
> + else if (IS_ERR(res->aux))
> + res->aux = NULL;
> +
> + res->master_bus = devm_clk_get(dev, "master_bus");
> + if (IS_ERR(res->master_bus))
> + return PTR_ERR(res->master_bus);
> +
> + res->slave_bus = devm_clk_get(dev, "slave_bus");
> + if (IS_ERR(res->slave_bus))
> + return PTR_ERR(res->slave_bus);
> +
> + res->core = devm_reset_control_get(dev, "core");
> + if (IS_ERR(res->core))
> + return PTR_ERR(res->core);
> +
> + return 0;
> +}
> +
> +static int qcom_pcie_enable_link_training(struct pcie_port *pp)
> +{
> + struct qcom_pcie *pcie = to_qcom_pcie(pp);
> + struct device *dev = pp->dev;
> +