Re: [PATCH 3/3] phy: rockchip: Add Samsung HDMI/DP Combo PHY driver

[Cristian Ciocaltea]

On 1/22/24 14:14, Sascha Hauer wrote:
> On Fri, Jan 19, 2024 at 09:38:03PM +0200, Cristian Ciocaltea wrote:
>> Add driver for the Rockchip HDMI/eDP TX Combo PHY found on RK3588 SoC.
>>
>> The PHY is based on a Samsung IP block and supports HDMI 2.1 TMDS, FRL
>> and eDP links.  The maximum data rate is 12Gbps (HDMI 2.1 FRL), while
>> the minimum is 250Mbps (HDMI 2.1 TMDS).
>>
>> Co-developed-by: Algea Cao <algea.cao@xxxxxxxxxxxxxx>
>> Signed-off-by: Algea Cao <algea.cao@xxxxxxxxxxxxxx>
>> Signed-off-by: Cristian Ciocaltea <cristian.ciocaltea@xxxxxxxxxxxxx>
>> ---
>>  drivers/phy/rockchip/Kconfig                  |    8 +
>>  drivers/phy/rockchip/Makefile                 |    1 +
>>  .../phy/rockchip/phy-rockchip-samsung-hdptx.c | 2045 +++++++++++++++++
>>  3 files changed, 2054 insertions(+)
>>  create mode 100644 drivers/phy/rockchip/phy-rockchip-samsung-hdptx.c
>>
>> diff --git a/drivers/phy/rockchip/Kconfig b/drivers/phy/rockchip/Kconfig
>> index 94360fc96a6f..95666ac6aa3b 100644
>> --- a/drivers/phy/rockchip/Kconfig
>> +++ b/drivers/phy/rockchip/Kconfig
>> @@ -83,6 +83,14 @@ config PHY_ROCKCHIP_PCIE
>>  	help
>>  	  Enable this to support the Rockchip PCIe PHY.
>>  
>> +config PHY_ROCKCHIP_SAMSUNG_HDPTX
>> +	tristate "Rockchip Samsung HDMI/DP Combo PHY driver"
>> +	depends on (ARCH_ROCKCHIP || COMPILE_TEST) && OF
>> +	select GENERIC_PHY
>> +	help
>> +	  Enable this to support the Rockchip HDMI/DP Combo PHY
>> +	  with Samsung IP block.
>> +
>>  config PHY_ROCKCHIP_SNPS_PCIE3
>>  	tristate "Rockchip Snps PCIe3 PHY Driver"
>>  	depends on (ARCH_ROCKCHIP && OF) || COMPILE_TEST
>> diff --git a/drivers/phy/rockchip/Makefile b/drivers/phy/rockchip/Makefile
>> index 7eab129230d1..3d911304e654 100644
>> --- a/drivers/phy/rockchip/Makefile
>> +++ b/drivers/phy/rockchip/Makefile
>> @@ -8,6 +8,7 @@ obj-$(CONFIG_PHY_ROCKCHIP_INNO_HDMI)	+= phy-rockchip-inno-hdmi.o
>>  obj-$(CONFIG_PHY_ROCKCHIP_INNO_USB2)	+= phy-rockchip-inno-usb2.o
>>  obj-$(CONFIG_PHY_ROCKCHIP_NANENG_COMBO_PHY)	+= phy-rockchip-naneng-combphy.o
>>  obj-$(CONFIG_PHY_ROCKCHIP_PCIE)		+= phy-rockchip-pcie.o
>> +obj-$(CONFIG_PHY_ROCKCHIP_SAMSUNG_HDPTX)	+= phy-rockchip-samsung-hdptx.o
>>  obj-$(CONFIG_PHY_ROCKCHIP_SNPS_PCIE3)	+= phy-rockchip-snps-pcie3.o
>>  obj-$(CONFIG_PHY_ROCKCHIP_TYPEC)	+= phy-rockchip-typec.o
>>  obj-$(CONFIG_PHY_ROCKCHIP_USB)		+= phy-rockchip-usb.o
>> diff --git a/drivers/phy/rockchip/phy-rockchip-samsung-hdptx.c b/drivers/phy/rockchip/phy-rockchip-samsung-hdptx.c
>> new file mode 100644
>> index 000000000000..d8171ea5ce2b
>> --- /dev/null
>> +++ b/drivers/phy/rockchip/phy-rockchip-samsung-hdptx.c
>> @@ -0,0 +1,2045 @@
>> +// SPDX-License-Identifier: GPL-2.0+
>> +/*
>> + * Copyright (c) 2021-2022 Rockchip Electronics Co., Ltd.
>> + * Copyright (c) 2024 Collabora Ltd.
>> + *
>> + * Author: Algea Cao <algea.cao@xxxxxxxxxxxxxx>
>> + * Author: Cristian Ciocaltea <cristian.ciocaltea@xxxxxxxxxxxxx>
>> + */
>> +#include <linux/bitfield.h>
>> +#include <linux/clk.h>
>> +#include <linux/delay.h>
>> +#include <linux/mfd/syscon.h>
>> +#include <linux/module.h>
>> +#include <linux/of.h>
>> +#include <linux/of_platform.h>
>> +#include <linux/phy/phy.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/rational.h>
>> +#include <linux/regmap.h>
>> +#include <linux/reset.h>
>> +
>> +#define GRF_HDPTX_CON0			0x00
>> +#define HDPTX_I_PLL_EN			BIT(7)
>> +#define HDPTX_I_BIAS_EN			BIT(6)
>> +#define HDPTX_I_BGR_EN			BIT(5)
>> +#define GRF_HDPTX_STATUS		0x80
>> +#define HDPTX_O_PLL_LOCK_DONE		BIT(3)
>> +#define HDPTX_O_PHY_CLK_RDY		BIT(2)
>> +#define HDPTX_O_PHY_RDY			BIT(1)
>> +#define HDPTX_O_SB_RDY			BIT(0)
>> +
>> +#define CMN_REG0000			0x0000
> 
> These register names are not particularly helpful. Maybe use a
> 
> #define CMN_REG(x)			((x) * 4)
> 
> Instead?

Yes, sounds good.

>> +
>> +static int hdptx_lcpll_frl_mode_config(struct rockchip_hdptx_phy *hdptx,
>> +				       u32 rate)
>> +{
>> +	u32 bit_rate = rate & DATA_RATE_MASK;
>> +	u8 color_depth = (rate & COLOR_DEPTH_MASK) ? 1 : 0;
>> +	const struct lcpll_config *cfg = lcpll_cfg;
>> +
>> +	for (; cfg->bit_rate != ~0; cfg++)
>> +		if (bit_rate == cfg->bit_rate)
>> +			break;
> 
> You could use ARRAY_SIZE() to iterate over the array and save the extra
> entry at the end. Likewise for the other arrays used in the driver.

Sure, will do.

>> +
>> +	if (cfg->bit_rate == ~0)
>> +		return -EINVAL;
>> +
> 
>> +static int rockchip_hdptx_phy_power_on(struct phy *phy)
>> +{
>> +	struct rockchip_hdptx_phy *hdptx = phy_get_drvdata(phy);
>> +	int bus_width = phy_get_bus_width(hdptx->phy);
>> +	int bit_rate = bus_width & DATA_RATE_MASK;
> 
> What is going on here? bus_width is set to 8 in probe() using
> phy_set_bus_width(), but the value you pull out of phy_get_bus_width()
> is expected to contain the bit_rate and several other flags.
> 
> It looks like you are tunneling flags from some other driver using this
> field. Isn't there a better way to accomplish this? If not, I think this
> needs some explanation.

Indeed, sorry for missing a comment here.  The flags are set by the
bridge driver to enable 10-bit color depth, FRL and EARC.  So far I
couldn't find an alternative approach to pass custom data using the PHY API.

> At least the variable should be renamed. it's called "bus_width" and it's
> passed to functions like hdptx_lcpll_frl_mode_config() which has this
> parameter named "rate" which is quite confusing.

I think for the initial support it's not really necessary to implement
all those features.  Andy, should we drop them until a better solution
is found?

> Sascha

Thanks for the review,
Cristian