diff mbox series

[v3,3/3] phy: freescale: Add i.MX8qm Mixel LVDS PHY support

Message ID 20220620123854.1138028-4-victor.liu@nxp.com (mailing list archive)
State New, archived
Headers show
Series phy: freescale: Add i.MX8qm Mixel LVDS PHY support | expand

Commit Message

Ying Liu June 20, 2022, 12:38 p.m. UTC
Add Freescale i.MX8qm LVDS PHY support.
The PHY IP is from Mixel, Inc.

Signed-off-by: Liu Ying <victor.liu@nxp.com>
---
v2->v3:
* No change.

v1->v2:
* Drop 'This patch' from commit message. (Krzysztof)
* Make dev_err_probe() function calls as one-liners. (Krzysztof)
* Drop unnecessary debug messages. (Krzysztof)
* Get phy_ref_clk without name specified due to 'clock-names' dropped from
  dt-binding.

 drivers/phy/freescale/Kconfig                 |   9 +
 drivers/phy/freescale/Makefile                |   1 +
 .../phy/freescale/phy-fsl-imx8qm-lvds-phy.c   | 440 ++++++++++++++++++
 3 files changed, 450 insertions(+)
 create mode 100644 drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c

Comments

Vinod Koul July 5, 2022, 8:54 a.m. UTC | #1
On 20-06-22, 20:38, Liu Ying wrote:
> Add Freescale i.MX8qm LVDS PHY support.
> The PHY IP is from Mixel, Inc.
> 
> Signed-off-by: Liu Ying <victor.liu@nxp.com>
> ---
> v2->v3:
> * No change.
> 
> v1->v2:
> * Drop 'This patch' from commit message. (Krzysztof)
> * Make dev_err_probe() function calls as one-liners. (Krzysztof)
> * Drop unnecessary debug messages. (Krzysztof)
> * Get phy_ref_clk without name specified due to 'clock-names' dropped from
>   dt-binding.
> 
>  drivers/phy/freescale/Kconfig                 |   9 +
>  drivers/phy/freescale/Makefile                |   1 +
>  .../phy/freescale/phy-fsl-imx8qm-lvds-phy.c   | 440 ++++++++++++++++++
>  3 files changed, 450 insertions(+)
>  create mode 100644 drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c
> 
> diff --git a/drivers/phy/freescale/Kconfig b/drivers/phy/freescale/Kconfig
> index f9c54cd02036..853958fb2c06 100644
> --- a/drivers/phy/freescale/Kconfig
> +++ b/drivers/phy/freescale/Kconfig
> @@ -8,6 +8,15 @@ config PHY_FSL_IMX8MQ_USB
>  	select GENERIC_PHY
>  	default ARCH_MXC && ARM64
>  
> +config PHY_MIXEL_LVDS_PHY
> +	tristate "Mixel LVDS PHY support"
> +	depends on OF
> +	select GENERIC_PHY
> +	select REGMAP_MMIO
> +	help
> +	  Enable this to add support for the Mixel LVDS PHY as found
> +	  on NXP's i.MX8qm SoC.
> +
>  config PHY_MIXEL_MIPI_DPHY
>  	tristate "Mixel MIPI DSI PHY support"
>  	depends on OF && HAS_IOMEM
> diff --git a/drivers/phy/freescale/Makefile b/drivers/phy/freescale/Makefile
> index 3518d5dbe8a7..cedb328bc4d2 100644
> --- a/drivers/phy/freescale/Makefile
> +++ b/drivers/phy/freescale/Makefile
> @@ -1,5 +1,6 @@
>  # SPDX-License-Identifier: GPL-2.0-only
>  obj-$(CONFIG_PHY_FSL_IMX8MQ_USB)	+= phy-fsl-imx8mq-usb.o
> +obj-$(CONFIG_PHY_MIXEL_LVDS_PHY)	+= phy-fsl-imx8qm-lvds-phy.o
>  obj-$(CONFIG_PHY_MIXEL_MIPI_DPHY)	+= phy-fsl-imx8-mipi-dphy.o
>  obj-$(CONFIG_PHY_FSL_IMX8M_PCIE)	+= phy-fsl-imx8m-pcie.o
>  obj-$(CONFIG_PHY_FSL_LYNX_28G)		+= phy-fsl-lynx-28g.o
> diff --git a/drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c b/drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c
> new file mode 100644
> index 000000000000..37f77115ddab
> --- /dev/null
> +++ b/drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c
> @@ -0,0 +1,440 @@
> +// SPDX-License-Identifier: GPL-2.0+
> +/*
> + * Copyright 2017-2020,2022 NXP
> + */
> +
> +#include <linux/bitfield.h>
> +#include <linux/bits.h>
> +#include <linux/clk.h>
> +#include <linux/mfd/syscon.h>
> +#include <linux/module.h>
> +#include <linux/of.h>
> +#include <linux/phy/phy.h>
> +#include <linux/platform_device.h>
> +#include <linux/pm_runtime.h>
> +#include <linux/regmap.h>
> +
> +#define REG_SET		0x4
> +#define REG_CLR		0x8
> +
> +#define PHY_CTRL	0x0
> +#define  M_MASK		GENMASK(18, 17)
> +#define  M(n)		FIELD_PREP(M_MASK, (n))
> +#define  CCM_MASK	GENMASK(16, 14)
> +#define  CCM(n)		FIELD_PREP(CCM_MASK, (n))
> +#define  CA_MASK	GENMASK(13, 11)
> +#define  CA(n)		FIELD_PREP(CA_MASK, (n))
> +#define  TST_MASK	GENMASK(10, 5)
> +#define  TST(n)		FIELD_PREP(TST_MASK, (n))
> +#define  CH_EN(id)	BIT(3 + (id))
> +#define  NB		BIT(2)
> +#define  RFB		BIT(1)
> +#define  PD		BIT(0)
> +
> +/* Power On Reset(POR) value */
> +#define  CTRL_RESET_VAL	(M(0x0) | CCM(0x4) | CA(0x4) | TST(0x25))
> +
> +/* PHY initialization value and mask */
> +#define  CTRL_INIT_MASK	(M_MASK | CCM_MASK | CA_MASK | TST_MASK | NB | RFB)
> +#define  CTRL_INIT_VAL	(M(0x0) | CCM(0x5) | CA(0x4) | TST(0x25) | RFB)
> +
> +#define PHY_STATUS	0x10
> +#define  LOCK		BIT(0)
> +
> +#define PHY_NUM		2
> +
> +#define MIN_CLKIN_FREQ	25000000

this is 25MHz, so lets write as 25 * MEGA (see units.h)

> +#define MAX_CLKIN_FREQ	165000000

here too

> +
> +#define PLL_LOCK_SLEEP		10
> +#define PLL_LOCK_TIMEOUT	1000
> +
> +struct mixel_lvds_phy {
> +	struct phy *phy;
> +	struct phy_configure_opts_lvds cfg;
> +	unsigned int id;
> +};
> +
> +struct mixel_lvds_phy_priv {
> +	struct regmap *regmap;
> +	struct mutex lock;	/* protect remap access and cfg of our own */
> +	struct clk *phy_ref_clk;
> +	struct mixel_lvds_phy *phys[PHY_NUM];
> +};
> +
> +static int mixel_lvds_phy_init(struct phy *phy)
> +{
> +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy->dev.parent);
> +
> +	mutex_lock(&priv->lock);
> +	regmap_update_bits(priv->regmap,
> +			   PHY_CTRL, CTRL_INIT_MASK, CTRL_INIT_VAL);
> +	mutex_unlock(&priv->lock);
> +
> +	return 0;
> +}
> +
> +static int mixel_lvds_phy_power_on(struct phy *phy)
> +{
> +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy->dev.parent);
> +	struct mixel_lvds_phy *lvds_phy = phy_get_drvdata(phy);
> +	struct mixel_lvds_phy *companion = priv->phys[lvds_phy->id ^ 1];
> +	struct phy_configure_opts_lvds *cfg = &lvds_phy->cfg;
> +	u32 val = 0;
> +	u32 locked;
> +	int ret;
> +
> +	ret = clk_prepare_enable(priv->phy_ref_clk);
> +	if (ret < 0) {
> +		dev_err(&phy->dev,
> +			"failed to enable PHY reference clock: %d\n", ret);
> +		return ret;
> +	}
> +
> +	/* The master PHY would power on the slave PHY. */
> +	if (cfg->is_slave)
> +		return 0;
> +
> +	mutex_lock(&priv->lock);
> +	if (cfg->bits_per_lane_and_dclk_cycle == 7) {
> +		if (cfg->differential_clk_rate < 44000000)
> +			val |= M(0x2);
> +		else if (cfg->differential_clk_rate < 90000000)
> +			val |= M(0x1);
> +		else
> +			val |= M(0x0);
> +	} else {
> +		val = NB;
> +
> +		if (cfg->differential_clk_rate < 32000000)
> +			val |= M(0x2);
> +		else if (cfg->differential_clk_rate < 63000000)
> +			val |= M(0x1);
> +		else
> +			val |= M(0x0);
> +	}
> +	regmap_update_bits(priv->regmap, PHY_CTRL, M_MASK | NB, val);
> +
> +	/*
> +	 * Enable two channels synchronously,
> +	 * if the companion PHY is a slave PHY.
> +	 */
> +	if (companion->cfg.is_slave)
> +		val = CH_EN(0) | CH_EN(1);
> +	else
> +		val = CH_EN(lvds_phy->id);
> +	regmap_write(priv->regmap, PHY_CTRL + REG_SET, val);
> +
> +	ret = regmap_read_poll_timeout(priv->regmap, PHY_STATUS, locked,
> +				       locked, PLL_LOCK_SLEEP,
> +				       PLL_LOCK_TIMEOUT);
> +	if (ret < 0) {
> +		dev_err(&phy->dev, "failed to get PHY lock: %d\n", ret);
> +		clk_disable_unprepare(priv->phy_ref_clk);
> +	}
> +	mutex_unlock(&priv->lock);
> +
> +	return ret;
> +}
> +
> +static int mixel_lvds_phy_power_off(struct phy *phy)
> +{
> +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy->dev.parent);
> +	struct mixel_lvds_phy *lvds_phy = phy_get_drvdata(phy);
> +
> +	mutex_lock(&priv->lock);
> +	regmap_write(priv->regmap, PHY_CTRL + REG_CLR, CH_EN(lvds_phy->id));
> +	mutex_unlock(&priv->lock);

No check for slave here?

> +
> +	clk_disable_unprepare(priv->phy_ref_clk);
> +
> +	return 0;
> +}
> +
> +static int mixel_lvds_phy_configure(struct phy *phy,
> +				    union phy_configure_opts *opts)
> +{
> +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy->dev.parent);
> +	struct phy_configure_opts_lvds *cfg = &opts->lvds;
> +	int ret;
> +
> +	ret = clk_set_rate(priv->phy_ref_clk, cfg->differential_clk_rate);
> +	if (ret)
> +		dev_err(&phy->dev,
> +			"failed to set PHY reference clock rate(%lu): %d\n",

this can fit in a single line (100 chars is okay)

> +			cfg->differential_clk_rate, ret);
> +
> +	return ret;
> +}
> +
> +/* Assume the master PHY's configuration set is cached first. */
> +static int mixel_lvds_phy_check_slave(struct phy *slave_phy)
> +{
> +	struct device *dev = &slave_phy->dev;
> +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(dev->parent);
> +	struct mixel_lvds_phy *slv = phy_get_drvdata(slave_phy);
> +	struct mixel_lvds_phy *mst = priv->phys[slv->id ^ 1];
> +	struct phy_configure_opts_lvds *mst_cfg = &mst->cfg;
> +	struct phy_configure_opts_lvds *slv_cfg = &slv->cfg;
> +
> +	if (mst_cfg->bits_per_lane_and_dclk_cycle !=
> +	    slv_cfg->bits_per_lane_and_dclk_cycle) {
> +		dev_err(dev, "number bits mismatch(mst: %u vs slv: %u)\n",
> +			mst_cfg->bits_per_lane_and_dclk_cycle,
> +			slv_cfg->bits_per_lane_and_dclk_cycle);
> +		return -EINVAL;
> +	}
> +
> +	if (mst_cfg->differential_clk_rate !=
> +	    slv_cfg->differential_clk_rate) {
> +		dev_err(dev, "dclk rate mismatch(mst: %lu vs slv: %lu)\n",
> +			mst_cfg->differential_clk_rate,
> +			slv_cfg->differential_clk_rate);
> +		return -EINVAL;
> +	}
> +
> +	if (mst_cfg->lanes != slv_cfg->lanes) {
> +		dev_err(dev, "lanes mismatch(mst: %u vs slv: %u)\n",
> +			mst_cfg->lanes, slv_cfg->lanes);
> +		return -EINVAL;
> +	}
> +
> +	if (mst_cfg->is_slave == slv_cfg->is_slave) {
> +		dev_err(dev, "master PHY is not found\n");
> +		return -EINVAL;
> +	}
> +
> +	return 0;
> +}
> +
> +static int mixel_lvds_phy_validate(struct phy *phy, enum phy_mode mode,
> +				   int submode, union phy_configure_opts *opts)
> +{
> +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy->dev.parent);
> +	struct mixel_lvds_phy *lvds_phy = phy_get_drvdata(phy);
> +	struct phy_configure_opts_lvds *cfg = &opts->lvds;
> +	int ret = 0;
> +
> +	if (mode != PHY_MODE_LVDS) {
> +		dev_err(&phy->dev, "invalid PHY mode(%d)\n", mode);
> +		return -EINVAL;
> +	}
> +
> +	if (cfg->bits_per_lane_and_dclk_cycle != 7 &&
> +	    cfg->bits_per_lane_and_dclk_cycle != 10) {
> +		dev_err(&phy->dev, "invalid bits per data lane(%u)\n",
> +			cfg->bits_per_lane_and_dclk_cycle);
> +		return -EINVAL;
> +	}
> +
> +	if (cfg->lanes != 4 && cfg->lanes != 3) {
> +		dev_err(&phy->dev, "invalid data lanes(%u)\n", cfg->lanes);
> +		return -EINVAL;
> +	}
> +
> +	if (cfg->differential_clk_rate < MIN_CLKIN_FREQ ||
> +	    cfg->differential_clk_rate > MAX_CLKIN_FREQ) {
> +		dev_err(&phy->dev, "invalid differential clock rate(%lu)\n",
> +			cfg->differential_clk_rate);
> +		return -EINVAL;
> +	}
> +
> +	mutex_lock(&priv->lock);
> +	/* cache configuration set of our own for check */
> +	memcpy(&lvds_phy->cfg, cfg, sizeof(*cfg));
> +
> +	if (cfg->is_slave) {
> +		ret = mixel_lvds_phy_check_slave(phy);
> +		if (ret)
> +			dev_err(&phy->dev,
> +				"failed to check slave PHY: %d\n", ret);

very ugly, single line pls

> +	}
> +	mutex_unlock(&priv->lock);
> +
> +	return ret;
> +}
> +
> +static const struct phy_ops mixel_lvds_phy_ops = {
> +	.init = mixel_lvds_phy_init,
> +	.power_on = mixel_lvds_phy_power_on,
> +	.power_off = mixel_lvds_phy_power_off,
> +	.configure = mixel_lvds_phy_configure,
> +	.validate = mixel_lvds_phy_validate,
> +	.owner = THIS_MODULE,
> +};
> +
> +static int mixel_lvds_phy_reset(struct device *dev)
> +{
> +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(dev);
> +	int ret;
> +
> +	ret = pm_runtime_get_sync(dev);

pm_runtime_resume_and_get() pls

> +	if (ret < 0) {
> +		dev_err(dev, "failed to get PM runtime: %d\n", ret);
> +		return ret;
> +	}
> +
> +	regmap_write(priv->regmap, PHY_CTRL, CTRL_RESET_VAL);
> +
> +	ret = pm_runtime_put(dev);

this seems to be done only around reset, why not in on/off method?

> +	if (ret < 0)
> +		dev_err(dev, "failed to put PM runtime: %d\n", ret);
> +
> +	return ret;
> +}
> +
> +static struct phy *mixel_lvds_phy_xlate(struct device *dev,
> +					struct of_phandle_args *args)
> +{
> +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(dev);
> +	unsigned int phy_id;
> +
> +	if (args->args_count != 1) {
> +		dev_err(dev,
> +			"invalid argument number(%d) for 'phys' property\n",
> +			args->args_count);
> +		return ERR_PTR(-EINVAL);
> +	}
> +
> +	phy_id = args->args[0];
> +
> +	if (phy_id >= PHY_NUM) {
> +		dev_err(dev, "invalid PHY index(%d)\n", phy_id);
> +		return ERR_PTR(-ENODEV);
> +	}
> +
> +	return priv->phys[phy_id]->phy;
> +}
> +
> +static int mixel_lvds_phy_probe(struct platform_device *pdev)
> +{
> +	struct device *dev = &pdev->dev;
> +	struct phy_provider *phy_provider;
> +	struct mixel_lvds_phy_priv *priv;
> +	struct mixel_lvds_phy *lvds_phy;
> +	struct phy *phy;
> +	int i;
> +	int ret;
> +
> +	if (!dev->of_node)
> +		return -ENODEV;
> +
> +	priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
> +	if (!priv)
> +		return -ENOMEM;
> +
> +	priv->regmap = syscon_node_to_regmap(dev->of_node->parent);
> +	if (IS_ERR(priv->regmap))
> +		return dev_err_probe(dev, PTR_ERR(priv->regmap),
> +				     "failed to get regmap\n");
> +
> +	priv->phy_ref_clk = devm_clk_get(dev, NULL);
> +	if (IS_ERR(priv->phy_ref_clk))
> +		return dev_err_probe(dev, PTR_ERR(priv->phy_ref_clk),
> +				     "failed to get PHY reference clock\n");
> +
> +	mutex_init(&priv->lock);
> +
> +	dev_set_drvdata(dev, priv);
> +
> +	pm_runtime_enable(dev);
> +
> +	ret = mixel_lvds_phy_reset(dev);
> +	if (ret) {
> +		dev_err(dev, "failed to do POR reset: %d\n", ret);
> +		return ret;
> +	}
> +
> +	for (i = 0; i < PHY_NUM; i++) {
> +		lvds_phy = devm_kzalloc(dev, sizeof(*lvds_phy), GFP_KERNEL);
> +		if (!lvds_phy) {
> +			ret = -ENOMEM;
> +			goto err;
> +		}
> +
> +		phy = devm_phy_create(dev, NULL, &mixel_lvds_phy_ops);
> +		if (IS_ERR(phy)) {
> +			ret = PTR_ERR(phy);
> +			dev_err(dev, "failed to create PHY for channel%d: %d\n",
> +				i, ret);
> +			goto err;
> +		}
> +
> +		lvds_phy->phy = phy;
> +		lvds_phy->id = i;
> +		priv->phys[i] = lvds_phy;
> +
> +		phy_set_drvdata(phy, lvds_phy);
> +	}
> +
> +	phy_provider = devm_of_phy_provider_register(dev, mixel_lvds_phy_xlate);
> +	if (IS_ERR(phy_provider)) {
> +		ret = PTR_ERR(phy_provider);
> +		dev_err(dev, "failed to register PHY provider: %d\n", ret);
> +		goto err;
> +	}
> +
> +	return 0;
> +err:
> +	pm_runtime_disable(dev);
> +
> +	return ret;
> +}
> +
> +static int mixel_lvds_phy_remove(struct platform_device *pdev)
> +{
> +	pm_runtime_disable(&pdev->dev);
> +
> +	return 0;
> +}
> +
> +static int __maybe_unused mixel_lvds_phy_runtime_suspend(struct device *dev)
> +{
> +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(dev);
> +
> +	/* power down */
> +	mutex_lock(&priv->lock);
> +	regmap_write(priv->regmap, PHY_CTRL + REG_SET, PD);
> +	mutex_unlock(&priv->lock);
> +
> +	return 0;
> +}
> +
> +static int __maybe_unused mixel_lvds_phy_runtime_resume(struct device *dev)
> +{
> +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(dev);
> +
> +	/* power up + control initialization */
> +	mutex_lock(&priv->lock);
> +	regmap_update_bits(priv->regmap, PHY_CTRL,
> +			   CTRL_INIT_MASK | PD, CTRL_INIT_VAL);
> +	mutex_unlock(&priv->lock);
> +
> +	return 0;
> +}
> +
> +static const struct dev_pm_ops mixel_lvds_phy_pm_ops = {
> +	SET_RUNTIME_PM_OPS(mixel_lvds_phy_runtime_suspend,
> +			   mixel_lvds_phy_runtime_resume, NULL)
> +};
> +
> +static const struct of_device_id mixel_lvds_phy_of_match[] = {
> +	{ .compatible = "fsl,imx8qm-lvds-phy" },
> +	{ /* sentinel */ }
> +};
> +MODULE_DEVICE_TABLE(of, mixel_lvds_phy_of_match);
> +
> +static struct platform_driver mixel_lvds_phy_driver = {
> +	.probe	= mixel_lvds_phy_probe,
> +	.remove	= mixel_lvds_phy_remove,
> +	.driver = {
> +		.pm = &mixel_lvds_phy_pm_ops,
> +		.name = "mixel-lvds-phy",
> +		.of_match_table	= mixel_lvds_phy_of_match,
> +	}
> +};
> +module_platform_driver(mixel_lvds_phy_driver);
> +
> +MODULE_DESCRIPTION("Mixel LVDS PHY driver");
> +MODULE_AUTHOR("Liu Ying <victor.liu@nxp.com>");
> +MODULE_LICENSE("GPL");
> -- 
> 2.25.1
Ying Liu July 5, 2022, 3:10 p.m. UTC | #2
On Tue, 2022-07-05 at 14:24 +0530, Vinod Koul wrote:
> On 20-06-22, 20:38, Liu Ying wrote:
> > Add Freescale i.MX8qm LVDS PHY support.
> > The PHY IP is from Mixel, Inc.
> > 
> > Signed-off-by: Liu Ying <victor.liu@nxp.com>
> > ---
> > v2->v3:
> > * No change.
> > 
> > v1->v2:
> > * Drop 'This patch' from commit message. (Krzysztof)
> > * Make dev_err_probe() function calls as one-liners. (Krzysztof)
> > * Drop unnecessary debug messages. (Krzysztof)
> > * Get phy_ref_clk without name specified due to 'clock-names'
> > dropped from
> >   dt-binding.
> > 
> >  drivers/phy/freescale/Kconfig                 |   9 +
> >  drivers/phy/freescale/Makefile                |   1 +
> >  .../phy/freescale/phy-fsl-imx8qm-lvds-phy.c   | 440
> > ++++++++++++++++++
> >  3 files changed, 450 insertions(+)
> >  create mode 100644 drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c
> > 
> > diff --git a/drivers/phy/freescale/Kconfig
> > b/drivers/phy/freescale/Kconfig
> > index f9c54cd02036..853958fb2c06 100644
> > --- a/drivers/phy/freescale/Kconfig
> > +++ b/drivers/phy/freescale/Kconfig
> > @@ -8,6 +8,15 @@ config PHY_FSL_IMX8MQ_USB
> >  	select GENERIC_PHY
> >  	default ARCH_MXC && ARM64
> >  
> > +config PHY_MIXEL_LVDS_PHY
> > +	tristate "Mixel LVDS PHY support"
> > +	depends on OF
> > +	select GENERIC_PHY
> > +	select REGMAP_MMIO
> > +	help
> > +	  Enable this to add support for the Mixel LVDS PHY as found
> > +	  on NXP's i.MX8qm SoC.
> > +
> >  config PHY_MIXEL_MIPI_DPHY
> >  	tristate "Mixel MIPI DSI PHY support"
> >  	depends on OF && HAS_IOMEM
> > diff --git a/drivers/phy/freescale/Makefile
> > b/drivers/phy/freescale/Makefile
> > index 3518d5dbe8a7..cedb328bc4d2 100644
> > --- a/drivers/phy/freescale/Makefile
> > +++ b/drivers/phy/freescale/Makefile
> > @@ -1,5 +1,6 @@
> >  # SPDX-License-Identifier: GPL-2.0-only
> >  obj-$(CONFIG_PHY_FSL_IMX8MQ_USB)	+= phy-fsl-imx8mq-usb.o
> > +obj-$(CONFIG_PHY_MIXEL_LVDS_PHY)	+= phy-fsl-imx8qm-lvds-phy.o
> >  obj-$(CONFIG_PHY_MIXEL_MIPI_DPHY)	+= phy-fsl-imx8-mipi-dphy.o
> >  obj-$(CONFIG_PHY_FSL_IMX8M_PCIE)	+= phy-fsl-imx8m-pcie.o
> >  obj-$(CONFIG_PHY_FSL_LYNX_28G)		+= phy-fsl-lynx-28g.o
> > diff --git a/drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c
> > b/drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c
> > new file mode 100644
> > index 000000000000..37f77115ddab
> > --- /dev/null
> > +++ b/drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c
> > @@ -0,0 +1,440 @@
> > +// SPDX-License-Identifier: GPL-2.0+
> > +/*
> > + * Copyright 2017-2020,2022 NXP
> > + */
> > +
> > +#include <linux/bitfield.h>
> > +#include <linux/bits.h>
> > +#include <linux/clk.h>
> > +#include <linux/mfd/syscon.h>
> > +#include <linux/module.h>
> > +#include <linux/of.h>
> > +#include <linux/phy/phy.h>
> > +#include <linux/platform_device.h>
> > +#include <linux/pm_runtime.h>
> > +#include <linux/regmap.h>
> > +
> > +#define REG_SET		0x4
> > +#define REG_CLR		0x8
> > +
> > +#define PHY_CTRL	0x0
> > +#define  M_MASK		GENMASK(18, 17)
> > +#define  M(n)		FIELD_PREP(M_MASK, (n))
> > +#define  CCM_MASK	GENMASK(16, 14)
> > +#define  CCM(n)		FIELD_PREP(CCM_MASK, (n))
> > +#define  CA_MASK	GENMASK(13, 11)
> > +#define  CA(n)		FIELD_PREP(CA_MASK, (n))
> > +#define  TST_MASK	GENMASK(10, 5)
> > +#define  TST(n)		FIELD_PREP(TST_MASK, (n))
> > +#define  CH_EN(id)	BIT(3 + (id))
> > +#define  NB		BIT(2)
> > +#define  RFB		BIT(1)
> > +#define  PD		BIT(0)
> > +
> > +/* Power On Reset(POR) value */
> > +#define  CTRL_RESET_VAL	(M(0x0) | CCM(0x4) | CA(0x4) |
> > TST(0x25))
> > +
> > +/* PHY initialization value and mask */
> > +#define  CTRL_INIT_MASK	(M_MASK | CCM_MASK | CA_MASK | TST_MASK
> > | NB | RFB)
> > +#define  CTRL_INIT_VAL	(M(0x0) | CCM(0x5) | CA(0x4) |
> > TST(0x25) | RFB)
> > +
> > +#define PHY_STATUS	0x10
> > +#define  LOCK		BIT(0)
> > +
> > +#define PHY_NUM		2
> > +
> > +#define MIN_CLKIN_FREQ	25000000
> 
> this is 25MHz, so lets write as 25 * MEGA (see units.h)

Will do.

> 
> > +#define MAX_CLKIN_FREQ	165000000
> 
> here too

Will do.

> > +
> > +#define PLL_LOCK_SLEEP		10
> > +#define PLL_LOCK_TIMEOUT	1000
> > +
> > 

[...]

> > +
> > +static int mixel_lvds_phy_power_off(struct phy *phy)
> > +{
> > +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy-
> > >dev.parent);
> > +	struct mixel_lvds_phy *lvds_phy = phy_get_drvdata(phy);
> > +
> > +	mutex_lock(&priv->lock);
> > +	regmap_write(priv->regmap, PHY_CTRL + REG_CLR, CH_EN(lvds_phy-
> > >id));
> > +	mutex_unlock(&priv->lock);
> 
> No check for slave here?

I don't worry too much about separate channel power off. But, it should
be fine to power off the two channels together if the companion is a
slave phy. So, will check for slave here.

> 
> > +
> > +	clk_disable_unprepare(priv->phy_ref_clk);
> > +
> > +	return 0;
> > +}
> > +
> > +static int mixel_lvds_phy_configure(struct phy *phy,
> > +				    union phy_configure_opts *opts)
> > +{
> > +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy-
> > >dev.parent);
> > +	struct phy_configure_opts_lvds *cfg = &opts->lvds;
> > +	int ret;
> > +
> > +	ret = clk_set_rate(priv->phy_ref_clk, cfg-
> > >differential_clk_rate);
> > +	if (ret)
> > +		dev_err(&phy->dev,
> > +			"failed to set PHY reference clock rate(%lu):
> > %d\n",
> 
> this can fit in a single line (100 chars is okay)

Will do.

> 
> > +			cfg->differential_clk_rate, ret);
> > +
> > +	return ret;
> > +}
> > +
> > 

[...]

> > +static int mixel_lvds_phy_validate(struct phy *phy, enum phy_mode
> > mode,
> > +				   int submode, union
> > phy_configure_opts *opts)
> > +{
> > +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy-
> > >dev.parent);
> > +	struct mixel_lvds_phy *lvds_phy = phy_get_drvdata(phy);
> > +	struct phy_configure_opts_lvds *cfg = &opts->lvds;
> > +	int ret = 0;
> > +
> > +	if (mode != PHY_MODE_LVDS) {
> > +		dev_err(&phy->dev, "invalid PHY mode(%d)\n", mode);
> > +		return -EINVAL;
> > +	}
> > +
> > +	if (cfg->bits_per_lane_and_dclk_cycle != 7 &&
> > +	    cfg->bits_per_lane_and_dclk_cycle != 10) {
> > +		dev_err(&phy->dev, "invalid bits per data lane(%u)\n",
> > +			cfg->bits_per_lane_and_dclk_cycle);
> > +		return -EINVAL;
> > +	}
> > +
> > +	if (cfg->lanes != 4 && cfg->lanes != 3) {
> > +		dev_err(&phy->dev, "invalid data lanes(%u)\n", cfg-
> > >lanes);
> > +		return -EINVAL;
> > +	}
> > +
> > +	if (cfg->differential_clk_rate < MIN_CLKIN_FREQ ||
> > +	    cfg->differential_clk_rate > MAX_CLKIN_FREQ) {
> > +		dev_err(&phy->dev, "invalid differential clock
> > rate(%lu)\n",
> > +			cfg->differential_clk_rate);
> > +		return -EINVAL;
> > +	}
> > +
> > +	mutex_lock(&priv->lock);
> > +	/* cache configuration set of our own for check */
> > +	memcpy(&lvds_phy->cfg, cfg, sizeof(*cfg));
> > +
> > +	if (cfg->is_slave) {
> > +		ret = mixel_lvds_phy_check_slave(phy);
> > +		if (ret)
> > +			dev_err(&phy->dev,
> > +				"failed to check slave PHY: %d\n",
> > ret);
> 
> very ugly, single line pls

Will do.

> 
> > +	}
> > +	mutex_unlock(&priv->lock);
> > +
> > +	return ret;
> > +}
> > +
> > 

[...]

> > +
> > +static int mixel_lvds_phy_reset(struct device *dev)
> > +{
> > +	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(dev);
> > +	int ret;
> > +
> > +	ret = pm_runtime_get_sync(dev);
> 
> pm_runtime_resume_and_get() pls

Will do.

> 
> > +	if (ret < 0) {
> > +		dev_err(dev, "failed to get PM runtime: %d\n", ret);
> > +		return ret;
> > +	}
> > +
> > +	regmap_write(priv->regmap, PHY_CTRL, CTRL_RESET_VAL);
> > +
> > +	ret = pm_runtime_put(dev);
> 
> this seems to be done only around reset, why not in on/off method?

The reason is that phy-core.c does that (See phy_pm_runtime_get_sync
and phy_pm_runtime_put).

Thanks,
Liu Ying
diff mbox series

Patch

diff --git a/drivers/phy/freescale/Kconfig b/drivers/phy/freescale/Kconfig
index f9c54cd02036..853958fb2c06 100644
--- a/drivers/phy/freescale/Kconfig
+++ b/drivers/phy/freescale/Kconfig
@@ -8,6 +8,15 @@  config PHY_FSL_IMX8MQ_USB
 	select GENERIC_PHY
 	default ARCH_MXC && ARM64
 
+config PHY_MIXEL_LVDS_PHY
+	tristate "Mixel LVDS PHY support"
+	depends on OF
+	select GENERIC_PHY
+	select REGMAP_MMIO
+	help
+	  Enable this to add support for the Mixel LVDS PHY as found
+	  on NXP's i.MX8qm SoC.
+
 config PHY_MIXEL_MIPI_DPHY
 	tristate "Mixel MIPI DSI PHY support"
 	depends on OF && HAS_IOMEM
diff --git a/drivers/phy/freescale/Makefile b/drivers/phy/freescale/Makefile
index 3518d5dbe8a7..cedb328bc4d2 100644
--- a/drivers/phy/freescale/Makefile
+++ b/drivers/phy/freescale/Makefile
@@ -1,5 +1,6 @@ 
 # SPDX-License-Identifier: GPL-2.0-only
 obj-$(CONFIG_PHY_FSL_IMX8MQ_USB)	+= phy-fsl-imx8mq-usb.o
+obj-$(CONFIG_PHY_MIXEL_LVDS_PHY)	+= phy-fsl-imx8qm-lvds-phy.o
 obj-$(CONFIG_PHY_MIXEL_MIPI_DPHY)	+= phy-fsl-imx8-mipi-dphy.o
 obj-$(CONFIG_PHY_FSL_IMX8M_PCIE)	+= phy-fsl-imx8m-pcie.o
 obj-$(CONFIG_PHY_FSL_LYNX_28G)		+= phy-fsl-lynx-28g.o
diff --git a/drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c b/drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c
new file mode 100644
index 000000000000..37f77115ddab
--- /dev/null
+++ b/drivers/phy/freescale/phy-fsl-imx8qm-lvds-phy.c
@@ -0,0 +1,440 @@ 
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2017-2020,2022 NXP
+ */
+
+#include <linux/bitfield.h>
+#include <linux/bits.h>
+#include <linux/clk.h>
+#include <linux/mfd/syscon.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/phy/phy.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+#include <linux/regmap.h>
+
+#define REG_SET		0x4
+#define REG_CLR		0x8
+
+#define PHY_CTRL	0x0
+#define  M_MASK		GENMASK(18, 17)
+#define  M(n)		FIELD_PREP(M_MASK, (n))
+#define  CCM_MASK	GENMASK(16, 14)
+#define  CCM(n)		FIELD_PREP(CCM_MASK, (n))
+#define  CA_MASK	GENMASK(13, 11)
+#define  CA(n)		FIELD_PREP(CA_MASK, (n))
+#define  TST_MASK	GENMASK(10, 5)
+#define  TST(n)		FIELD_PREP(TST_MASK, (n))
+#define  CH_EN(id)	BIT(3 + (id))
+#define  NB		BIT(2)
+#define  RFB		BIT(1)
+#define  PD		BIT(0)
+
+/* Power On Reset(POR) value */
+#define  CTRL_RESET_VAL	(M(0x0) | CCM(0x4) | CA(0x4) | TST(0x25))
+
+/* PHY initialization value and mask */
+#define  CTRL_INIT_MASK	(M_MASK | CCM_MASK | CA_MASK | TST_MASK | NB | RFB)
+#define  CTRL_INIT_VAL	(M(0x0) | CCM(0x5) | CA(0x4) | TST(0x25) | RFB)
+
+#define PHY_STATUS	0x10
+#define  LOCK		BIT(0)
+
+#define PHY_NUM		2
+
+#define MIN_CLKIN_FREQ	25000000
+#define MAX_CLKIN_FREQ	165000000
+
+#define PLL_LOCK_SLEEP		10
+#define PLL_LOCK_TIMEOUT	1000
+
+struct mixel_lvds_phy {
+	struct phy *phy;
+	struct phy_configure_opts_lvds cfg;
+	unsigned int id;
+};
+
+struct mixel_lvds_phy_priv {
+	struct regmap *regmap;
+	struct mutex lock;	/* protect remap access and cfg of our own */
+	struct clk *phy_ref_clk;
+	struct mixel_lvds_phy *phys[PHY_NUM];
+};
+
+static int mixel_lvds_phy_init(struct phy *phy)
+{
+	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy->dev.parent);
+
+	mutex_lock(&priv->lock);
+	regmap_update_bits(priv->regmap,
+			   PHY_CTRL, CTRL_INIT_MASK, CTRL_INIT_VAL);
+	mutex_unlock(&priv->lock);
+
+	return 0;
+}
+
+static int mixel_lvds_phy_power_on(struct phy *phy)
+{
+	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy->dev.parent);
+	struct mixel_lvds_phy *lvds_phy = phy_get_drvdata(phy);
+	struct mixel_lvds_phy *companion = priv->phys[lvds_phy->id ^ 1];
+	struct phy_configure_opts_lvds *cfg = &lvds_phy->cfg;
+	u32 val = 0;
+	u32 locked;
+	int ret;
+
+	ret = clk_prepare_enable(priv->phy_ref_clk);
+	if (ret < 0) {
+		dev_err(&phy->dev,
+			"failed to enable PHY reference clock: %d\n", ret);
+		return ret;
+	}
+
+	/* The master PHY would power on the slave PHY. */
+	if (cfg->is_slave)
+		return 0;
+
+	mutex_lock(&priv->lock);
+	if (cfg->bits_per_lane_and_dclk_cycle == 7) {
+		if (cfg->differential_clk_rate < 44000000)
+			val |= M(0x2);
+		else if (cfg->differential_clk_rate < 90000000)
+			val |= M(0x1);
+		else
+			val |= M(0x0);
+	} else {
+		val = NB;
+
+		if (cfg->differential_clk_rate < 32000000)
+			val |= M(0x2);
+		else if (cfg->differential_clk_rate < 63000000)
+			val |= M(0x1);
+		else
+			val |= M(0x0);
+	}
+	regmap_update_bits(priv->regmap, PHY_CTRL, M_MASK | NB, val);
+
+	/*
+	 * Enable two channels synchronously,
+	 * if the companion PHY is a slave PHY.
+	 */
+	if (companion->cfg.is_slave)
+		val = CH_EN(0) | CH_EN(1);
+	else
+		val = CH_EN(lvds_phy->id);
+	regmap_write(priv->regmap, PHY_CTRL + REG_SET, val);
+
+	ret = regmap_read_poll_timeout(priv->regmap, PHY_STATUS, locked,
+				       locked, PLL_LOCK_SLEEP,
+				       PLL_LOCK_TIMEOUT);
+	if (ret < 0) {
+		dev_err(&phy->dev, "failed to get PHY lock: %d\n", ret);
+		clk_disable_unprepare(priv->phy_ref_clk);
+	}
+	mutex_unlock(&priv->lock);
+
+	return ret;
+}
+
+static int mixel_lvds_phy_power_off(struct phy *phy)
+{
+	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy->dev.parent);
+	struct mixel_lvds_phy *lvds_phy = phy_get_drvdata(phy);
+
+	mutex_lock(&priv->lock);
+	regmap_write(priv->regmap, PHY_CTRL + REG_CLR, CH_EN(lvds_phy->id));
+	mutex_unlock(&priv->lock);
+
+	clk_disable_unprepare(priv->phy_ref_clk);
+
+	return 0;
+}
+
+static int mixel_lvds_phy_configure(struct phy *phy,
+				    union phy_configure_opts *opts)
+{
+	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy->dev.parent);
+	struct phy_configure_opts_lvds *cfg = &opts->lvds;
+	int ret;
+
+	ret = clk_set_rate(priv->phy_ref_clk, cfg->differential_clk_rate);
+	if (ret)
+		dev_err(&phy->dev,
+			"failed to set PHY reference clock rate(%lu): %d\n",
+			cfg->differential_clk_rate, ret);
+
+	return ret;
+}
+
+/* Assume the master PHY's configuration set is cached first. */
+static int mixel_lvds_phy_check_slave(struct phy *slave_phy)
+{
+	struct device *dev = &slave_phy->dev;
+	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(dev->parent);
+	struct mixel_lvds_phy *slv = phy_get_drvdata(slave_phy);
+	struct mixel_lvds_phy *mst = priv->phys[slv->id ^ 1];
+	struct phy_configure_opts_lvds *mst_cfg = &mst->cfg;
+	struct phy_configure_opts_lvds *slv_cfg = &slv->cfg;
+
+	if (mst_cfg->bits_per_lane_and_dclk_cycle !=
+	    slv_cfg->bits_per_lane_and_dclk_cycle) {
+		dev_err(dev, "number bits mismatch(mst: %u vs slv: %u)\n",
+			mst_cfg->bits_per_lane_and_dclk_cycle,
+			slv_cfg->bits_per_lane_and_dclk_cycle);
+		return -EINVAL;
+	}
+
+	if (mst_cfg->differential_clk_rate !=
+	    slv_cfg->differential_clk_rate) {
+		dev_err(dev, "dclk rate mismatch(mst: %lu vs slv: %lu)\n",
+			mst_cfg->differential_clk_rate,
+			slv_cfg->differential_clk_rate);
+		return -EINVAL;
+	}
+
+	if (mst_cfg->lanes != slv_cfg->lanes) {
+		dev_err(dev, "lanes mismatch(mst: %u vs slv: %u)\n",
+			mst_cfg->lanes, slv_cfg->lanes);
+		return -EINVAL;
+	}
+
+	if (mst_cfg->is_slave == slv_cfg->is_slave) {
+		dev_err(dev, "master PHY is not found\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int mixel_lvds_phy_validate(struct phy *phy, enum phy_mode mode,
+				   int submode, union phy_configure_opts *opts)
+{
+	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(phy->dev.parent);
+	struct mixel_lvds_phy *lvds_phy = phy_get_drvdata(phy);
+	struct phy_configure_opts_lvds *cfg = &opts->lvds;
+	int ret = 0;
+
+	if (mode != PHY_MODE_LVDS) {
+		dev_err(&phy->dev, "invalid PHY mode(%d)\n", mode);
+		return -EINVAL;
+	}
+
+	if (cfg->bits_per_lane_and_dclk_cycle != 7 &&
+	    cfg->bits_per_lane_and_dclk_cycle != 10) {
+		dev_err(&phy->dev, "invalid bits per data lane(%u)\n",
+			cfg->bits_per_lane_and_dclk_cycle);
+		return -EINVAL;
+	}
+
+	if (cfg->lanes != 4 && cfg->lanes != 3) {
+		dev_err(&phy->dev, "invalid data lanes(%u)\n", cfg->lanes);
+		return -EINVAL;
+	}
+
+	if (cfg->differential_clk_rate < MIN_CLKIN_FREQ ||
+	    cfg->differential_clk_rate > MAX_CLKIN_FREQ) {
+		dev_err(&phy->dev, "invalid differential clock rate(%lu)\n",
+			cfg->differential_clk_rate);
+		return -EINVAL;
+	}
+
+	mutex_lock(&priv->lock);
+	/* cache configuration set of our own for check */
+	memcpy(&lvds_phy->cfg, cfg, sizeof(*cfg));
+
+	if (cfg->is_slave) {
+		ret = mixel_lvds_phy_check_slave(phy);
+		if (ret)
+			dev_err(&phy->dev,
+				"failed to check slave PHY: %d\n", ret);
+	}
+	mutex_unlock(&priv->lock);
+
+	return ret;
+}
+
+static const struct phy_ops mixel_lvds_phy_ops = {
+	.init = mixel_lvds_phy_init,
+	.power_on = mixel_lvds_phy_power_on,
+	.power_off = mixel_lvds_phy_power_off,
+	.configure = mixel_lvds_phy_configure,
+	.validate = mixel_lvds_phy_validate,
+	.owner = THIS_MODULE,
+};
+
+static int mixel_lvds_phy_reset(struct device *dev)
+{
+	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(dev);
+	int ret;
+
+	ret = pm_runtime_get_sync(dev);
+	if (ret < 0) {
+		dev_err(dev, "failed to get PM runtime: %d\n", ret);
+		return ret;
+	}
+
+	regmap_write(priv->regmap, PHY_CTRL, CTRL_RESET_VAL);
+
+	ret = pm_runtime_put(dev);
+	if (ret < 0)
+		dev_err(dev, "failed to put PM runtime: %d\n", ret);
+
+	return ret;
+}
+
+static struct phy *mixel_lvds_phy_xlate(struct device *dev,
+					struct of_phandle_args *args)
+{
+	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(dev);
+	unsigned int phy_id;
+
+	if (args->args_count != 1) {
+		dev_err(dev,
+			"invalid argument number(%d) for 'phys' property\n",
+			args->args_count);
+		return ERR_PTR(-EINVAL);
+	}
+
+	phy_id = args->args[0];
+
+	if (phy_id >= PHY_NUM) {
+		dev_err(dev, "invalid PHY index(%d)\n", phy_id);
+		return ERR_PTR(-ENODEV);
+	}
+
+	return priv->phys[phy_id]->phy;
+}
+
+static int mixel_lvds_phy_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct phy_provider *phy_provider;
+	struct mixel_lvds_phy_priv *priv;
+	struct mixel_lvds_phy *lvds_phy;
+	struct phy *phy;
+	int i;
+	int ret;
+
+	if (!dev->of_node)
+		return -ENODEV;
+
+	priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	priv->regmap = syscon_node_to_regmap(dev->of_node->parent);
+	if (IS_ERR(priv->regmap))
+		return dev_err_probe(dev, PTR_ERR(priv->regmap),
+				     "failed to get regmap\n");
+
+	priv->phy_ref_clk = devm_clk_get(dev, NULL);
+	if (IS_ERR(priv->phy_ref_clk))
+		return dev_err_probe(dev, PTR_ERR(priv->phy_ref_clk),
+				     "failed to get PHY reference clock\n");
+
+	mutex_init(&priv->lock);
+
+	dev_set_drvdata(dev, priv);
+
+	pm_runtime_enable(dev);
+
+	ret = mixel_lvds_phy_reset(dev);
+	if (ret) {
+		dev_err(dev, "failed to do POR reset: %d\n", ret);
+		return ret;
+	}
+
+	for (i = 0; i < PHY_NUM; i++) {
+		lvds_phy = devm_kzalloc(dev, sizeof(*lvds_phy), GFP_KERNEL);
+		if (!lvds_phy) {
+			ret = -ENOMEM;
+			goto err;
+		}
+
+		phy = devm_phy_create(dev, NULL, &mixel_lvds_phy_ops);
+		if (IS_ERR(phy)) {
+			ret = PTR_ERR(phy);
+			dev_err(dev, "failed to create PHY for channel%d: %d\n",
+				i, ret);
+			goto err;
+		}
+
+		lvds_phy->phy = phy;
+		lvds_phy->id = i;
+		priv->phys[i] = lvds_phy;
+
+		phy_set_drvdata(phy, lvds_phy);
+	}
+
+	phy_provider = devm_of_phy_provider_register(dev, mixel_lvds_phy_xlate);
+	if (IS_ERR(phy_provider)) {
+		ret = PTR_ERR(phy_provider);
+		dev_err(dev, "failed to register PHY provider: %d\n", ret);
+		goto err;
+	}
+
+	return 0;
+err:
+	pm_runtime_disable(dev);
+
+	return ret;
+}
+
+static int mixel_lvds_phy_remove(struct platform_device *pdev)
+{
+	pm_runtime_disable(&pdev->dev);
+
+	return 0;
+}
+
+static int __maybe_unused mixel_lvds_phy_runtime_suspend(struct device *dev)
+{
+	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(dev);
+
+	/* power down */
+	mutex_lock(&priv->lock);
+	regmap_write(priv->regmap, PHY_CTRL + REG_SET, PD);
+	mutex_unlock(&priv->lock);
+
+	return 0;
+}
+
+static int __maybe_unused mixel_lvds_phy_runtime_resume(struct device *dev)
+{
+	struct mixel_lvds_phy_priv *priv = dev_get_drvdata(dev);
+
+	/* power up + control initialization */
+	mutex_lock(&priv->lock);
+	regmap_update_bits(priv->regmap, PHY_CTRL,
+			   CTRL_INIT_MASK | PD, CTRL_INIT_VAL);
+	mutex_unlock(&priv->lock);
+
+	return 0;
+}
+
+static const struct dev_pm_ops mixel_lvds_phy_pm_ops = {
+	SET_RUNTIME_PM_OPS(mixel_lvds_phy_runtime_suspend,
+			   mixel_lvds_phy_runtime_resume, NULL)
+};
+
+static const struct of_device_id mixel_lvds_phy_of_match[] = {
+	{ .compatible = "fsl,imx8qm-lvds-phy" },
+	{ /* sentinel */ }
+};
+MODULE_DEVICE_TABLE(of, mixel_lvds_phy_of_match);
+
+static struct platform_driver mixel_lvds_phy_driver = {
+	.probe	= mixel_lvds_phy_probe,
+	.remove	= mixel_lvds_phy_remove,
+	.driver = {
+		.pm = &mixel_lvds_phy_pm_ops,
+		.name = "mixel-lvds-phy",
+		.of_match_table	= mixel_lvds_phy_of_match,
+	}
+};
+module_platform_driver(mixel_lvds_phy_driver);
+
+MODULE_DESCRIPTION("Mixel LVDS PHY driver");
+MODULE_AUTHOR("Liu Ying <victor.liu@nxp.com>");
+MODULE_LICENSE("GPL");