mbox series

[v9,0/2] phy: mediatek: Add PCIe PHY driver

Message ID 20220520064920.27313-1-jianjun.wang@mediatek.com
Headers show
Series phy: mediatek: Add PCIe PHY driver | expand

Message

Jianjun Wang (王建军) May 20, 2022, 6:49 a.m. UTC 
These series patches add support for PCIe PHY driver on MediaTek chipsets.

Changes in v9:
1. Check if the return value is -ENOMEM when reading efuse data fails.

Changes in v8:
1. Use "device_property_present()" to increase human readability;
2. Use "GPL" as recommended in commit bf7fbeeae6db ("module: Cure
   the MODULE_LICENSE "GPL" vs. "GPL v2" bogosity").

Changes in v7:
1. Add bitfield.h header to fix the build error on non-arm64 platforms.

Changes in v6:
1. Remove unnecessary header files;
2. Use FILELD_PREP in bitfield.h to set value.

Changes in v5:
1. Fix typo in kerneldoc: "eFues" => "eFuse".

Changes in v4:
1. Fix no return when calling dev_err_probe.

Changes in v3:
1. Add introductions for structure members;
2. Add SoC dependent data;
3. Dynamically allocate efuse data;
4. Check return value if it's an -EPROBE_DEFER.

Changes in v2:
1. Add specific compatible name;
2. Read NVMEM data at probe time;
3. Fix typos.

Jianjun Wang (2):
  dt-bindings: phy: mediatek: Add YAML schema for PCIe PHY
  phy: mediatek: Add PCIe PHY driver

 .../bindings/phy/mediatek,pcie-phy.yaml       |  75 +++++
 drivers/phy/mediatek/Kconfig                  |  11 +
 drivers/phy/mediatek/Makefile                 |   1 +
 drivers/phy/mediatek/phy-mtk-pcie.c           | 267 ++++++++++++++++++
 4 files changed, 354 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/phy/mediatek,pcie-phy.yaml
 create mode 100644 drivers/phy/mediatek/phy-mtk-pcie.c

Comments

AngeloGioacchino Del Regno May 20, 2022, 8:57 a.m. UTC | #1 
Il 20/05/22 08:49, Jianjun Wang ha scritto:
> Add PCIe GEN3 PHY driver support on MediaTek chipsets.
> 
> Signed-off-by: Jianjun Wang <jianjun.wang@mediatek.com>

Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Chunfeng Yun (云春峰) May 23, 2022, 8:46 a.m. UTC | #2 
On Fri, 2022-05-20 at 14:49 +0800, Jianjun Wang wrote:
> Add PCIe GEN3 PHY driver support on MediaTek chipsets.
> 
> Signed-off-by: Jianjun Wang <jianjun.wang@mediatek.com>
> ---
>  drivers/phy/mediatek/Kconfig        |  11 ++
>  drivers/phy/mediatek/Makefile       |   1 +
>  drivers/phy/mediatek/phy-mtk-pcie.c | 267
> ++++++++++++++++++++++++++++
>  3 files changed, 279 insertions(+)
>  create mode 100644 drivers/phy/mediatek/phy-mtk-pcie.c
> 
Reviewed-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
Rex-BC Chen (陳柏辰) May 23, 2022, 10:34 a.m. UTC | #3 
On Fri, 2022-05-20 at 14:49 +0800, Jianjun Wang wrote:
> Add PCIe GEN3 PHY driver support on MediaTek chipsets.
> 
> Signed-off-by: Jianjun Wang <jianjun.wang@mediatek.com>
> ---
>  drivers/phy/mediatek/Kconfig        |  11 ++
>  drivers/phy/mediatek/Makefile       |   1 +
>  drivers/phy/mediatek/phy-mtk-pcie.c | 267
> ++++++++++++++++++++++++++++

Reviewed-by: Rex-BC Chen <rex-bc.chen@mediatek.com>
Jianjun Wang (王建军) June 1, 2022, 2:21 a.m. UTC | #4 
Hello Maintainers,

Gentle ping for this patch series, if there is anything I can do to get
these patches merged, please let me know.

Thanks.

On Fri, 2022-05-20 at 14:49 +0800, Jianjun Wang wrote:
> These series patches add support for PCIe PHY driver on MediaTek
> chipsets.
> 
> Changes in v9:
> 1. Check if the return value is -ENOMEM when reading efuse data
> fails.
> 
> Changes in v8:
> 1. Use "device_property_present()" to increase human readability;
> 2. Use "GPL" as recommended in commit bf7fbeeae6db ("module: Cure
>    the MODULE_LICENSE "GPL" vs. "GPL v2" bogosity").
> 
> Changes in v7:
> 1. Add bitfield.h header to fix the build error on non-arm64
> platforms.
> 
> Changes in v6:
> 1. Remove unnecessary header files;
> 2. Use FILELD_PREP in bitfield.h to set value.
> 
> Changes in v5:
> 1. Fix typo in kerneldoc: "eFues" => "eFuse".
> 
> Changes in v4:
> 1. Fix no return when calling dev_err_probe.
> 
> Changes in v3:
> 1. Add introductions for structure members;
> 2. Add SoC dependent data;
> 3. Dynamically allocate efuse data;
> 4. Check return value if it's an -EPROBE_DEFER.
> 
> Changes in v2:
> 1. Add specific compatible name;
> 2. Read NVMEM data at probe time;
> 3. Fix typos.
> 
> Jianjun Wang (2):
>   dt-bindings: phy: mediatek: Add YAML schema for PCIe PHY
>   phy: mediatek: Add PCIe PHY driver
> 
>  .../bindings/phy/mediatek,pcie-phy.yaml       |  75 +++++
>  drivers/phy/mediatek/Kconfig                  |  11 +
>  drivers/phy/mediatek/Makefile                 |   1 +
>  drivers/phy/mediatek/phy-mtk-pcie.c           | 267
> ++++++++++++++++++
>  4 files changed, 354 insertions(+)
>  create mode 100644
> Documentation/devicetree/bindings/phy/mediatek,pcie-phy.yaml
>  create mode 100644 drivers/phy/mediatek/phy-mtk-pcie.c
>
Vinod Koul June 1, 2022, 5:32 a.m. UTC | #5 
On 01-06-22, 10:21, Jianjun Wang wrote:
> Hello Maintainers,
> 
> Gentle ping for this patch series, if there is anything I can do to get
> these patches merged, please let me know.

Patience my friend patience. This was received very late in cycle and I
will review after merge window closes..


> 
> Thanks.
> 
> On Fri, 2022-05-20 at 14:49 +0800, Jianjun Wang wrote:
> > These series patches add support for PCIe PHY driver on MediaTek
> > chipsets.
> > 
> > Changes in v9:
> > 1. Check if the return value is -ENOMEM when reading efuse data
> > fails.
> > 
> > Changes in v8:
> > 1. Use "device_property_present()" to increase human readability;
> > 2. Use "GPL" as recommended in commit bf7fbeeae6db ("module: Cure
> >    the MODULE_LICENSE "GPL" vs. "GPL v2" bogosity").
> > 
> > Changes in v7:
> > 1. Add bitfield.h header to fix the build error on non-arm64
> > platforms.
> > 
> > Changes in v6:
> > 1. Remove unnecessary header files;
> > 2. Use FILELD_PREP in bitfield.h to set value.
> > 
> > Changes in v5:
> > 1. Fix typo in kerneldoc: "eFues" => "eFuse".
> > 
> > Changes in v4:
> > 1. Fix no return when calling dev_err_probe.
> > 
> > Changes in v3:
> > 1. Add introductions for structure members;
> > 2. Add SoC dependent data;
> > 3. Dynamically allocate efuse data;
> > 4. Check return value if it's an -EPROBE_DEFER.
> > 
> > Changes in v2:
> > 1. Add specific compatible name;
> > 2. Read NVMEM data at probe time;
> > 3. Fix typos.
> > 
> > Jianjun Wang (2):
> >   dt-bindings: phy: mediatek: Add YAML schema for PCIe PHY
> >   phy: mediatek: Add PCIe PHY driver
> > 
> >  .../bindings/phy/mediatek,pcie-phy.yaml       |  75 +++++
> >  drivers/phy/mediatek/Kconfig                  |  11 +
> >  drivers/phy/mediatek/Makefile                 |   1 +
> >  drivers/phy/mediatek/phy-mtk-pcie.c           | 267
> > ++++++++++++++++++
> >  4 files changed, 354 insertions(+)
> >  create mode 100644
> > Documentation/devicetree/bindings/phy/mediatek,pcie-phy.yaml
> >  create mode 100644 drivers/phy/mediatek/phy-mtk-pcie.c
> >
Jianjun Wang (王建军) June 14, 2022, 2:38 a.m. UTC | #6 
Hi Vinod,

Sorry to disturb you, would you mind taking a look at this patch?

Thanks.

On Wed, 2022-06-01 at 11:02 +0530, Vinod Koul wrote:
> On 01-06-22, 10:21, Jianjun Wang wrote:
> > Hello Maintainers,
> > 
> > Gentle ping for this patch series, if there is anything I can do to
> > get
> > these patches merged, please let me know.
> 
> Patience my friend patience. This was received very late in cycle and
> I
> will review after merge window closes..
> 
> 
> > 
> > Thanks.
> > 
> > On Fri, 2022-05-20 at 14:49 +0800, Jianjun Wang wrote:
> > > These series patches add support for PCIe PHY driver on MediaTek
> > > chipsets.
> > > 
> > > Changes in v9:
> > > 1. Check if the return value is -ENOMEM when reading efuse data
> > > fails.
> > > 
> > > Changes in v8:
> > > 1. Use "device_property_present()" to increase human readability;
> > > 2. Use "GPL" as recommended in commit bf7fbeeae6db ("module: Cure
> > >    the MODULE_LICENSE "GPL" vs. "GPL v2" bogosity").
> > > 
> > > Changes in v7:
> > > 1. Add bitfield.h header to fix the build error on non-arm64
> > > platforms.
> > > 
> > > Changes in v6:
> > > 1. Remove unnecessary header files;
> > > 2. Use FILELD_PREP in bitfield.h to set value.
> > > 
> > > Changes in v5:
> > > 1. Fix typo in kerneldoc: "eFues" => "eFuse".
> > > 
> > > Changes in v4:
> > > 1. Fix no return when calling dev_err_probe.
> > > 
> > > Changes in v3:
> > > 1. Add introductions for structure members;
> > > 2. Add SoC dependent data;
> > > 3. Dynamically allocate efuse data;
> > > 4. Check return value if it's an -EPROBE_DEFER.
> > > 
> > > Changes in v2:
> > > 1. Add specific compatible name;
> > > 2. Read NVMEM data at probe time;
> > > 3. Fix typos.
> > > 
> > > Jianjun Wang (2):
> > >   dt-bindings: phy: mediatek: Add YAML schema for PCIe PHY
> > >   phy: mediatek: Add PCIe PHY driver
> > > 
> > >  .../bindings/phy/mediatek,pcie-phy.yaml       |  75 +++++
> > >  drivers/phy/mediatek/Kconfig                  |  11 +
> > >  drivers/phy/mediatek/Makefile                 |   1 +
> > >  drivers/phy/mediatek/phy-mtk-pcie.c           | 267
> > > ++++++++++++++++++
> > >  4 files changed, 354 insertions(+)
> > >  create mode 100644
> > > Documentation/devicetree/bindings/phy/mediatek,pcie-phy.yaml
> > >  create mode 100644 drivers/phy/mediatek/phy-mtk-pcie.c
> > > 
> 
>
Vinod Koul June 17, 2022, 12:17 a.m. UTC | #7 
On 20-05-22, 14:49, Jianjun Wang wrote:
> Add PCIe GEN3 PHY driver support on MediaTek chipsets.
> 
> Signed-off-by: Jianjun Wang <jianjun.wang@mediatek.com>
> ---
>  drivers/phy/mediatek/Kconfig        |  11 ++
>  drivers/phy/mediatek/Makefile       |   1 +
>  drivers/phy/mediatek/phy-mtk-pcie.c | 267 ++++++++++++++++++++++++++++
>  3 files changed, 279 insertions(+)
>  create mode 100644 drivers/phy/mediatek/phy-mtk-pcie.c
> 
> diff --git a/drivers/phy/mediatek/Kconfig b/drivers/phy/mediatek/Kconfig
> index 55f8e6c048ab..387ed1b3f2cc 100644
> --- a/drivers/phy/mediatek/Kconfig
> +++ b/drivers/phy/mediatek/Kconfig
> @@ -55,3 +55,14 @@ config PHY_MTK_MIPI_DSI
>  	select GENERIC_PHY
>  	help
>  	  Support MIPI DSI for Mediatek SoCs.
> +
> +config PHY_MTK_PCIE
> +	tristate "MediaTek PCIe-PHY Driver"
> +	depends on ARCH_MEDIATEK || COMPILE_TEST
> +	depends on OF
> +	select GENERIC_PHY
> +	help
> +	  Say 'Y' here to add support for MediaTek PCIe PHY driver.
> +	  This driver create the basic PHY instance and provides initialize
> +	  callback for PCIe GEN3 port, it supports software efuse
> +	  initialization.
> diff --git a/drivers/phy/mediatek/Makefile b/drivers/phy/mediatek/Makefile
> index ace660fbed3a..788c13147f63 100644
> --- a/drivers/phy/mediatek/Makefile
> +++ b/drivers/phy/mediatek/Makefile
> @@ -6,6 +6,7 @@
>  obj-$(CONFIG_PHY_MTK_TPHY)		+= phy-mtk-tphy.o
>  obj-$(CONFIG_PHY_MTK_UFS)		+= phy-mtk-ufs.o
>  obj-$(CONFIG_PHY_MTK_XSPHY)		+= phy-mtk-xsphy.o
> +obj-$(CONFIG_PHY_MTK_PCIE)		+= phy-mtk-pcie.o

This and Kconfig is sorted in alphabetical order, pls do so here as well

>  
>  phy-mtk-hdmi-drv-y			:= phy-mtk-hdmi.o
>  phy-mtk-hdmi-drv-y			+= phy-mtk-hdmi-mt2701.o
> diff --git a/drivers/phy/mediatek/phy-mtk-pcie.c b/drivers/phy/mediatek/phy-mtk-pcie.c
> new file mode 100644
> index 000000000000..7f29d43442bf
> --- /dev/null
> +++ b/drivers/phy/mediatek/phy-mtk-pcie.c
> @@ -0,0 +1,267 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright (c) 2022 MediaTek Inc.
> + * Author: Jianjun Wang <jianjun.wang@mediatek.com>
> + */
> +
> +#include <linux/bitfield.h>
> +#include <linux/module.h>
> +#include <linux/nvmem-consumer.h>
> +#include <linux/of_device.h>
> +#include <linux/phy/phy.h>
> +#include <linux/platform_device.h>
> +#include <linux/slab.h>
> +
> +#include "phy-mtk-io.h"
> +
> +#define PEXTP_ANA_GLB_00_REG		0x9000
> +/* Internal Resistor Selection of TX Bias Current */
> +#define EFUSE_GLB_INTR_SEL		GENMASK(28, 24)
> +
> +#define PEXTP_ANA_LN0_TRX_REG		0xa000
> +
> +#define PEXTP_ANA_TX_REG		0x04
> +/* TX PMOS impedance selection */
> +#define EFUSE_LN_TX_PMOS_SEL		GENMASK(5, 2)
> +/* TX NMOS impedance selection */
> +#define EFUSE_LN_TX_NMOS_SEL		GENMASK(11, 8)
> +
> +#define PEXTP_ANA_RX_REG		0x3c
> +/* RX impedance selection */
> +#define EFUSE_LN_RX_SEL			GENMASK(3, 0)
> +
> +#define PEXTP_ANA_LANE_OFFSET		0x100
> +
> +/**
> + * struct mtk_pcie_lane_efuse - eFuse data for each lane
> + * @tx_pmos: TX PMOS impedance selection data
> + * @tx_nmos: TX NMOS impedance selection data
> + * @rx_data: RX impedance selection data
> + * @lane_efuse_supported: software eFuse data is supported for this lane
> + */
> +struct mtk_pcie_lane_efuse {
> +	u32 tx_pmos;
> +	u32 tx_nmos;
> +	u32 rx_data;
> +	bool lane_efuse_supported;
> +};
> +
> +/**
> + * struct mtk_pcie_phy_data - phy data for each SoC
> + * @num_lanes: supported lane numbers
> + * @sw_efuse_supported: support software to load eFuse data
> + */
> +struct mtk_pcie_phy_data {
> +	int num_lanes;
> +	bool sw_efuse_supported;
> +};
> +
> +/**
> + * struct mtk_pcie_phy - PCIe phy driver main structure
> + * @dev: pointer to device
> + * @phy: pointer to generic phy
> + * @sif_base: IO mapped register base address of system interface
> + * @data: pointer to SoC dependent data
> + * @sw_efuse_en: software eFuse enable status
> + * @efuse_glb_intr: internal resistor selection of TX bias current data
> + * @efuse: pointer to eFuse data for each lane
> + */
> +struct mtk_pcie_phy {
> +	struct device *dev;
> +	struct phy *phy;
> +	void __iomem *sif_base;
> +	const struct mtk_pcie_phy_data *data;
> +
> +	bool sw_efuse_en;
> +	u32 efuse_glb_intr;
> +	struct mtk_pcie_lane_efuse *efuse;
> +};
> +
> +static void mtk_pcie_efuse_set_lane(struct mtk_pcie_phy *pcie_phy,
> +				    unsigned int lane)
> +{
> +	struct mtk_pcie_lane_efuse *data = &pcie_phy->efuse[lane];
> +	void __iomem *addr;
> +
> +	if (!data->lane_efuse_supported)
> +		return;
> +
> +	addr = pcie_phy->sif_base + PEXTP_ANA_LN0_TRX_REG +
> +	       lane * PEXTP_ANA_LANE_OFFSET;
> +
> +	mtk_phy_update_bits(addr + PEXTP_ANA_TX_REG, EFUSE_LN_TX_PMOS_SEL,
> +			    FIELD_PREP(EFUSE_LN_TX_PMOS_SEL, data->tx_pmos));
> +
> +	mtk_phy_update_bits(addr + PEXTP_ANA_TX_REG, EFUSE_LN_TX_NMOS_SEL,
> +			    FIELD_PREP(EFUSE_LN_TX_NMOS_SEL, data->tx_nmos));
> +
> +	mtk_phy_update_bits(addr + PEXTP_ANA_RX_REG, EFUSE_LN_RX_SEL,
> +			    FIELD_PREP(EFUSE_LN_RX_SEL, data->rx_data));
> +}
> +
> +/**
> + * mtk_pcie_phy_init() - Initialize the phy
> + * @phy: the phy to be initialized
> + *
> + * Initialize the phy by setting the efuse data.
> + * The hardware settings will be reset during suspend, it should be
> + * reinitialized when the consumer calls phy_init() again on resume.
> + */
> +static int mtk_pcie_phy_init(struct phy *phy)
> +{
> +	struct mtk_pcie_phy *pcie_phy = phy_get_drvdata(phy);
> +	int i;
> +
> +	if (!pcie_phy->sw_efuse_en)
> +		return 0;
> +
> +	/* Set global data */
> +	mtk_phy_update_bits(pcie_phy->sif_base + PEXTP_ANA_GLB_00_REG,
> +			    EFUSE_GLB_INTR_SEL,
> +			    FIELD_PREP(EFUSE_GLB_INTR_SEL, pcie_phy->efuse_glb_intr));
> +
> +	for (i = 0; i < pcie_phy->data->num_lanes; i++)
> +		mtk_pcie_efuse_set_lane(pcie_phy, i);
> +
> +	return 0;
> +}
> +
> +static const struct phy_ops mtk_pcie_phy_ops = {
> +	.init	= mtk_pcie_phy_init,
> +	.owner	= THIS_MODULE,
> +};
> +
> +static int mtk_pcie_efuse_read_for_lane(struct mtk_pcie_phy *pcie_phy,
> +					unsigned int lane)
> +{
> +	struct mtk_pcie_lane_efuse *efuse = &pcie_phy->efuse[lane];
> +	struct device *dev = pcie_phy->dev;
> +	char efuse_id[16];
> +	int ret;
> +
> +	snprintf(efuse_id, sizeof(efuse_id), "tx_ln%d_pmos", lane);
> +	ret = nvmem_cell_read_variable_le_u32(dev, efuse_id, &efuse->tx_pmos);
> +	if (ret)
> +		return dev_err_probe(dev, ret, "Failed to read %s\n", efuse_id);
> +
> +	snprintf(efuse_id, sizeof(efuse_id), "tx_ln%d_nmos", lane);
> +	ret = nvmem_cell_read_variable_le_u32(dev, efuse_id, &efuse->tx_nmos);
> +	if (ret)
> +		return dev_err_probe(dev, ret, "Failed to read %s\n", efuse_id);
> +
> +	snprintf(efuse_id, sizeof(efuse_id), "rx_ln%d", lane);
> +	ret = nvmem_cell_read_variable_le_u32(dev, efuse_id, &efuse->rx_data);
> +	if (ret)
> +		return dev_err_probe(dev, ret, "Failed to read %s\n", efuse_id);
> +
> +	if (!(efuse->tx_pmos || efuse->tx_nmos || efuse->rx_data))
> +		return dev_err_probe(dev, -EINVAL,
> +				     "No eFuse data found for lane%d, but dts enable it\n",
> +				     lane);
> +
> +	efuse->lane_efuse_supported = true;
> +
> +	return 0;
> +}
> +
> +static int mtk_pcie_read_efuse(struct mtk_pcie_phy *pcie_phy)
> +{
> +	struct device *dev = pcie_phy->dev;
> +	bool nvmem_enabled;
> +	int ret, i;
> +
> +	/* nvmem data is optional */
> +	nvmem_enabled = device_property_present(dev, "nvmem-cells");
> +	if (!nvmem_enabled)
> +		return 0;
> +
> +	ret = nvmem_cell_read_variable_le_u32(dev, "glb_intr",
> +					      &pcie_phy->efuse_glb_intr);
> +	if (ret)
> +		return dev_err_probe(dev, ret, "Failed to read glb_intr\n");
> +
> +	pcie_phy->sw_efuse_en = true;
> +
> +	pcie_phy->efuse = devm_kzalloc(dev, pcie_phy->data->num_lanes *
> +				       sizeof(*pcie_phy->efuse), GFP_KERNEL);
> +	if (!pcie_phy->efuse)
> +		return -ENOMEM;
> +
> +	for (i = 0; i < pcie_phy->data->num_lanes; i++) {
> +		ret = mtk_pcie_efuse_read_for_lane(pcie_phy, i);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	return 0;
> +}
> +
> +static int mtk_pcie_phy_probe(struct platform_device *pdev)
> +{
> +	struct device *dev = &pdev->dev;
> +	struct phy_provider *provider;
> +	struct mtk_pcie_phy *pcie_phy;
> +	int ret;
> +
> +	pcie_phy = devm_kzalloc(dev, sizeof(*pcie_phy), GFP_KERNEL);
> +	if (!pcie_phy)
> +		return -ENOMEM;
> +
> +	pcie_phy->sif_base = devm_platform_ioremap_resource_byname(pdev, "sif");
> +	if (IS_ERR(pcie_phy->sif_base))
> +		return dev_err_probe(dev, PTR_ERR(pcie_phy->sif_base),
> +				     "Failed to map phy-sif base\n");
> +
> +	pcie_phy->phy = devm_phy_create(dev, dev->of_node, &mtk_pcie_phy_ops);
> +	if (IS_ERR(pcie_phy->phy))
> +		return dev_err_probe(dev, PTR_ERR(pcie_phy->phy),
> +				     "Failed to create PCIe phy\n");
> +
> +	pcie_phy->dev = dev;
> +	pcie_phy->data = of_device_get_match_data(dev);
> +	if (!pcie_phy->data)
> +		return dev_err_probe(dev, -EINVAL, "Failed to get phy data\n");
> +
> +	if (pcie_phy->data->sw_efuse_supported) {
> +		/*
> +		 * Failed to read the efuse data is not a fatal problem,
> +		 * ignore the failure and keep going.
> +		 */
> +		ret = mtk_pcie_read_efuse(pcie_phy);
> +		if (ret == -EPROBE_DEFER || ret == -ENOMEM)
> +			return ret;
> +	}
> +
> +	phy_set_drvdata(pcie_phy->phy, pcie_phy);
> +
> +	provider = devm_of_phy_provider_register(dev, of_phy_simple_xlate);
> +	if (IS_ERR(provider))
> +		return dev_err_probe(dev, PTR_ERR(provider),
> +				     "PCIe phy probe failed\n");
> +
> +	return 0;
> +}
> +
> +static const struct mtk_pcie_phy_data mt8195_data = {
> +	.num_lanes = 2,
> +	.sw_efuse_supported = true,
> +};
> +
> +static const struct of_device_id mtk_pcie_phy_of_match[] = {
> +	{ .compatible = "mediatek,mt8195-pcie-phy", .data = &mt8195_data },
> +	{ },
> +};
> +MODULE_DEVICE_TABLE(of, mtk_pcie_phy_of_match);
> +
> +static struct platform_driver mtk_pcie_phy_driver = {
> +	.probe	= mtk_pcie_phy_probe,
> +	.driver	= {
> +		.name = "mtk-pcie-phy",
> +		.of_match_table = mtk_pcie_phy_of_match,
> +	},
> +};
> +module_platform_driver(mtk_pcie_phy_driver);
> +
> +MODULE_DESCRIPTION("MediaTek PCIe PHY driver");
> +MODULE_AUTHOR("Jianjun Wang <jianjun.wang@mediatek.com>");
> +MODULE_LICENSE("GPL");
> -- 
> 2.18.0
Jianjun Wang (王建军) June 17, 2022, 1:33 a.m. UTC | #8 
Hi Vinod,

On Thu, 2022-06-16 at 17:17 -0700, Vinod Koul wrote:
> On 20-05-22, 14:49, Jianjun Wang wrote:
> > Add PCIe GEN3 PHY driver support on MediaTek chipsets.
> > 
> > Signed-off-by: Jianjun Wang <jianjun.wang@mediatek.com>
> > ---
> >  drivers/phy/mediatek/Kconfig        |  11 ++
> >  drivers/phy/mediatek/Makefile       |   1 +
> >  drivers/phy/mediatek/phy-mtk-pcie.c | 267
> > ++++++++++++++++++++++++++++
> >  3 files changed, 279 insertions(+)
> >  create mode 100644 drivers/phy/mediatek/phy-mtk-pcie.c
> > 
> > diff --git a/drivers/phy/mediatek/Kconfig
> > b/drivers/phy/mediatek/Kconfig
> > index 55f8e6c048ab..387ed1b3f2cc 100644
> > --- a/drivers/phy/mediatek/Kconfig
> > +++ b/drivers/phy/mediatek/Kconfig
> > @@ -55,3 +55,14 @@ config PHY_MTK_MIPI_DSI
> >  	select GENERIC_PHY
> >  	help
> >  	  Support MIPI DSI for Mediatek SoCs.
> > +
> > +config PHY_MTK_PCIE
> > +	tristate "MediaTek PCIe-PHY Driver"
> > +	depends on ARCH_MEDIATEK || COMPILE_TEST
> > +	depends on OF
> > +	select GENERIC_PHY
> > +	help
> > +	  Say 'Y' here to add support for MediaTek PCIe PHY driver.
> > +	  This driver create the basic PHY instance and provides
> > initialize
> > +	  callback for PCIe GEN3 port, it supports software efuse
> > +	  initialization.
> > diff --git a/drivers/phy/mediatek/Makefile
> > b/drivers/phy/mediatek/Makefile
> > index ace660fbed3a..788c13147f63 100644
> > --- a/drivers/phy/mediatek/Makefile
> > +++ b/drivers/phy/mediatek/Makefile
> > @@ -6,6 +6,7 @@
> >  obj-$(CONFIG_PHY_MTK_TPHY)		+= phy-mtk-tphy.o
> >  obj-$(CONFIG_PHY_MTK_UFS)		+= phy-mtk-ufs.o
> >  obj-$(CONFIG_PHY_MTK_XSPHY)		+= phy-mtk-xsphy.o
> > +obj-$(CONFIG_PHY_MTK_PCIE)		+= phy-mtk-pcie.o
> 
> This and Kconfig is sorted in alphabetical order, pls do so here as
> well

Thanks for your review, I'll send another version to fix this.

> 
> >  
> >  phy-mtk-hdmi-drv-y			:= phy-mtk-hdmi.o
> >  phy-mtk-hdmi-drv-y			+= phy-mtk-hdmi-mt2701.o
> > diff --git a/drivers/phy/mediatek/phy-mtk-pcie.c
> > b/drivers/phy/mediatek/phy-mtk-pcie.c
> > new file mode 100644
> > index 000000000000..7f29d43442bf
> > --- /dev/null
> > +++ b/drivers/phy/mediatek/phy-mtk-pcie.c
> > @@ -0,0 +1,267 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Copyright (c) 2022 MediaTek Inc.
> > + * Author: Jianjun Wang <jianjun.wang@mediatek.com>
> > + */
> > +
> > +#include <linux/bitfield.h>
> > +#include <linux/module.h>
> > +#include <linux/nvmem-consumer.h>
> > +#include <linux/of_device.h>
> > +#include <linux/phy/phy.h>
> > +#include <linux/platform_device.h>
> > +#include <linux/slab.h>
> > +
> > +#include "phy-mtk-io.h"
> > +
> > +#define PEXTP_ANA_GLB_00_REG		0x9000
> > +/* Internal Resistor Selection of TX Bias Current */
> > +#define EFUSE_GLB_INTR_SEL		GENMASK(28, 24)
> > +
> > +#define PEXTP_ANA_LN0_TRX_REG		0xa000
> > +
> > +#define PEXTP_ANA_TX_REG		0x04
> > +/* TX PMOS impedance selection */
> > +#define EFUSE_LN_TX_PMOS_SEL		GENMASK(5, 2)
> > +/* TX NMOS impedance selection */
> > +#define EFUSE_LN_TX_NMOS_SEL		GENMASK(11, 8)
> > +
> > +#define PEXTP_ANA_RX_REG		0x3c
> > +/* RX impedance selection */
> > +#define EFUSE_LN_RX_SEL			GENMASK(3, 0)
> > +
> > +#define PEXTP_ANA_LANE_OFFSET		0x100
> > +
> > +/**
> > + * struct mtk_pcie_lane_efuse - eFuse data for each lane
> > + * @tx_pmos: TX PMOS impedance selection data
> > + * @tx_nmos: TX NMOS impedance selection data
> > + * @rx_data: RX impedance selection data
> > + * @lane_efuse_supported: software eFuse data is supported for
> > this lane
> > + */
> > +struct mtk_pcie_lane_efuse {
> > +	u32 tx_pmos;
> > +	u32 tx_nmos;
> > +	u32 rx_data;
> > +	bool lane_efuse_supported;
> > +};
> > +
> > +/**
> > + * struct mtk_pcie_phy_data - phy data for each SoC
> > + * @num_lanes: supported lane numbers
> > + * @sw_efuse_supported: support software to load eFuse data
> > + */
> > +struct mtk_pcie_phy_data {
> > +	int num_lanes;
> > +	bool sw_efuse_supported;
> > +};
> > +
> > +/**
> > + * struct mtk_pcie_phy - PCIe phy driver main structure
> > + * @dev: pointer to device
> > + * @phy: pointer to generic phy
> > + * @sif_base: IO mapped register base address of system interface
> > + * @data: pointer to SoC dependent data
> > + * @sw_efuse_en: software eFuse enable status
> > + * @efuse_glb_intr: internal resistor selection of TX bias current
> > data
> > + * @efuse: pointer to eFuse data for each lane
> > + */
> > +struct mtk_pcie_phy {
> > +	struct device *dev;
> > +	struct phy *phy;
> > +	void __iomem *sif_base;
> > +	const struct mtk_pcie_phy_data *data;
> > +
> > +	bool sw_efuse_en;
> > +	u32 efuse_glb_intr;
> > +	struct mtk_pcie_lane_efuse *efuse;
> > +};
> > +
> > +static void mtk_pcie_efuse_set_lane(struct mtk_pcie_phy *pcie_phy,
> > +				    unsigned int lane)
> > +{
> > +	struct mtk_pcie_lane_efuse *data = &pcie_phy->efuse[lane];
> > +	void __iomem *addr;
> > +
> > +	if (!data->lane_efuse_supported)
> > +		return;
> > +
> > +	addr = pcie_phy->sif_base + PEXTP_ANA_LN0_TRX_REG +
> > +	       lane * PEXTP_ANA_LANE_OFFSET;
> > +
> > +	mtk_phy_update_bits(addr + PEXTP_ANA_TX_REG,
> > EFUSE_LN_TX_PMOS_SEL,
> > +			    FIELD_PREP(EFUSE_LN_TX_PMOS_SEL, data-
> > >tx_pmos));
> > +
> > +	mtk_phy_update_bits(addr + PEXTP_ANA_TX_REG,
> > EFUSE_LN_TX_NMOS_SEL,
> > +			    FIELD_PREP(EFUSE_LN_TX_NMOS_SEL, data-
> > >tx_nmos));
> > +
> > +	mtk_phy_update_bits(addr + PEXTP_ANA_RX_REG, EFUSE_LN_RX_SEL,
> > +			    FIELD_PREP(EFUSE_LN_RX_SEL, data-
> > >rx_data));
> > +}
> > +
> > +/**
> > + * mtk_pcie_phy_init() - Initialize the phy
> > + * @phy: the phy to be initialized
> > + *
> > + * Initialize the phy by setting the efuse data.
> > + * The hardware settings will be reset during suspend, it should
> > be
> > + * reinitialized when the consumer calls phy_init() again on
> > resume.
> > + */
> > +static int mtk_pcie_phy_init(struct phy *phy)
> > +{
> > +	struct mtk_pcie_phy *pcie_phy = phy_get_drvdata(phy);
> > +	int i;
> > +
> > +	if (!pcie_phy->sw_efuse_en)
> > +		return 0;
> > +
> > +	/* Set global data */
> > +	mtk_phy_update_bits(pcie_phy->sif_base + PEXTP_ANA_GLB_00_REG,
> > +			    EFUSE_GLB_INTR_SEL,
> > +			    FIELD_PREP(EFUSE_GLB_INTR_SEL, pcie_phy-
> > >efuse_glb_intr));
> > +
> > +	for (i = 0; i < pcie_phy->data->num_lanes; i++)
> > +		mtk_pcie_efuse_set_lane(pcie_phy, i);
> > +
> > +	return 0;
> > +}
> > +
> > +static const struct phy_ops mtk_pcie_phy_ops = {
> > +	.init	= mtk_pcie_phy_init,
> > +	.owner	= THIS_MODULE,
> > +};
> > +
> > +static int mtk_pcie_efuse_read_for_lane(struct mtk_pcie_phy
> > *pcie_phy,
> > +					unsigned int lane)
> > +{
> > +	struct mtk_pcie_lane_efuse *efuse = &pcie_phy->efuse[lane];
> > +	struct device *dev = pcie_phy->dev;
> > +	char efuse_id[16];
> > +	int ret;
> > +
> > +	snprintf(efuse_id, sizeof(efuse_id), "tx_ln%d_pmos", lane);
> > +	ret = nvmem_cell_read_variable_le_u32(dev, efuse_id, &efuse-
> > >tx_pmos);
> > +	if (ret)
> > +		return dev_err_probe(dev, ret, "Failed to read %s\n",
> > efuse_id);
> > +
> > +	snprintf(efuse_id, sizeof(efuse_id), "tx_ln%d_nmos", lane);
> > +	ret = nvmem_cell_read_variable_le_u32(dev, efuse_id, &efuse-
> > >tx_nmos);
> > +	if (ret)
> > +		return dev_err_probe(dev, ret, "Failed to read %s\n",
> > efuse_id);
> > +
> > +	snprintf(efuse_id, sizeof(efuse_id), "rx_ln%d", lane);
> > +	ret = nvmem_cell_read_variable_le_u32(dev, efuse_id, &efuse-
> > >rx_data);
> > +	if (ret)
> > +		return dev_err_probe(dev, ret, "Failed to read %s\n",
> > efuse_id);
> > +
> > +	if (!(efuse->tx_pmos || efuse->tx_nmos || efuse->rx_data))
> > +		return dev_err_probe(dev, -EINVAL,
> > +				     "No eFuse data found for lane%d,
> > but dts enable it\n",
> > +				     lane);
> > +
> > +	efuse->lane_efuse_supported = true;
> > +
> > +	return 0;
> > +}
> > +
> > +static int mtk_pcie_read_efuse(struct mtk_pcie_phy *pcie_phy)
> > +{
> > +	struct device *dev = pcie_phy->dev;
> > +	bool nvmem_enabled;
> > +	int ret, i;
> > +
> > +	/* nvmem data is optional */
> > +	nvmem_enabled = device_property_present(dev, "nvmem-cells");
> > +	if (!nvmem_enabled)
> > +		return 0;
> > +
> > +	ret = nvmem_cell_read_variable_le_u32(dev, "glb_intr",
> > +					      &pcie_phy-
> > >efuse_glb_intr);
> > +	if (ret)
> > +		return dev_err_probe(dev, ret, "Failed to read
> > glb_intr\n");
> > +
> > +	pcie_phy->sw_efuse_en = true;
> > +
> > +	pcie_phy->efuse = devm_kzalloc(dev, pcie_phy->data->num_lanes *
> > +				       sizeof(*pcie_phy->efuse),
> > GFP_KERNEL);
> > +	if (!pcie_phy->efuse)
> > +		return -ENOMEM;
> > +
> > +	for (i = 0; i < pcie_phy->data->num_lanes; i++) {
> > +		ret = mtk_pcie_efuse_read_for_lane(pcie_phy, i);
> > +		if (ret)
> > +			return ret;
> > +	}
> > +
> > +	return 0;
> > +}
> > +
> > +static int mtk_pcie_phy_probe(struct platform_device *pdev)
> > +{
> > +	struct device *dev = &pdev->dev;
> > +	struct phy_provider *provider;
> > +	struct mtk_pcie_phy *pcie_phy;
> > +	int ret;
> > +
> > +	pcie_phy = devm_kzalloc(dev, sizeof(*pcie_phy), GFP_KERNEL);
> > +	if (!pcie_phy)
> > +		return -ENOMEM;
> > +
> > +	pcie_phy->sif_base =
> > devm_platform_ioremap_resource_byname(pdev, "sif");
> > +	if (IS_ERR(pcie_phy->sif_base))
> > +		return dev_err_probe(dev, PTR_ERR(pcie_phy->sif_base),
> > +				     "Failed to map phy-sif base\n");
> > +
> > +	pcie_phy->phy = devm_phy_create(dev, dev->of_node,
> > &mtk_pcie_phy_ops);
> > +	if (IS_ERR(pcie_phy->phy))
> > +		return dev_err_probe(dev, PTR_ERR(pcie_phy->phy),
> > +				     "Failed to create PCIe phy\n");
> > +
> > +	pcie_phy->dev = dev;
> > +	pcie_phy->data = of_device_get_match_data(dev);
> > +	if (!pcie_phy->data)
> > +		return dev_err_probe(dev, -EINVAL, "Failed to get phy
> > data\n");
> > +
> > +	if (pcie_phy->data->sw_efuse_supported) {
> > +		/*
> > +		 * Failed to read the efuse data is not a fatal
> > problem,
> > +		 * ignore the failure and keep going.
> > +		 */
> > +		ret = mtk_pcie_read_efuse(pcie_phy);
> > +		if (ret == -EPROBE_DEFER || ret == -ENOMEM)
> > +			return ret;
> > +	}
> > +
> > +	phy_set_drvdata(pcie_phy->phy, pcie_phy);
> > +
> > +	provider = devm_of_phy_provider_register(dev,
> > of_phy_simple_xlate);
> > +	if (IS_ERR(provider))
> > +		return dev_err_probe(dev, PTR_ERR(provider),
> > +				     "PCIe phy probe failed\n");
> > +
> > +	return 0;
> > +}
> > +
> > +static const struct mtk_pcie_phy_data mt8195_data = {
> > +	.num_lanes = 2,
> > +	.sw_efuse_supported = true,
> > +};
> > +
> > +static const struct of_device_id mtk_pcie_phy_of_match[] = {
> > +	{ .compatible = "mediatek,mt8195-pcie-phy", .data =
> > &mt8195_data },
> > +	{ },
> > +};
> > +MODULE_DEVICE_TABLE(of, mtk_pcie_phy_of_match);
> > +
> > +static struct platform_driver mtk_pcie_phy_driver = {
> > +	.probe	= mtk_pcie_phy_probe,
> > +	.driver	= {
> > +		.name = "mtk-pcie-phy",
> > +		.of_match_table = mtk_pcie_phy_of_match,
> > +	},
> > +};
> > +module_platform_driver(mtk_pcie_phy_driver);
> > +
> > +MODULE_DESCRIPTION("MediaTek PCIe PHY driver");
> > +MODULE_AUTHOR("Jianjun Wang <jianjun.wang@mediatek.com>");
> > +MODULE_LICENSE("GPL");
> > -- 
> > 2.18.0
> 
>