Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <31d459e4-57c0-85d1-2d88-e5bf0bed6604@quicinc.com>
Date:   Wed, 31 Aug 2022 18:41:46 +0530
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:91.0) Gecko/20100101
 Thunderbird/91.12.0
Subject: Re: [PATCH v11 2/3] phy: qcom-snps: Add support for overriding phy
 tuning parameters
Content-Language: en-US
To:     Bjorn Andersson <andersson@kernel.org>
CC:     Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        Rob Herring <robh+dt@kernel.org>,
        Andy Gross <agross@kernel.org>,
        Bjorn Andersson <bjorn.andersson@linaro.org>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Stephen Boyd <swboyd@chromium.org>,
        "Doug Anderson" <dianders@chromium.org>,
        Matthias Kaehlcke <mka@chromium.org>,
        "Vinod Koul" <vkoul@kernel.org>,
        Wesley Cheng <quic_wcheng@quicinc.com>,
        <devicetree@vger.kernel.org>, <linux-arm-msm@vger.kernel.org>,
        <linux-usb@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
        <linux-phy@lists.infradead.org>, <quic_pkondeti@quicinc.com>,
        <quic_ppratap@quicinc.com>, <quic_vpulyala@quicinc.com>
References: <1658384954-9506-1-git-send-email-quic_kriskura@quicinc.com>
 <1658384954-9506-3-git-send-email-quic_kriskura@quicinc.com>
 <20220830203518.pty67fyefho4ewgw@builder.lan>
From:   Krishna Kurapati PSSNV <quic_kriskura@quicinc.com>
In-Reply-To: <20220830203518.pty67fyefho4ewgw@builder.lan>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 8/31/2022 2:05 AM, Bjorn Andersson wrote:
> On Thu, Jul 21, 2022 at 11:59:13AM +0530, Krishna Kurapati wrote:
>> Add support for overriding electrical signal tuning parameters for
>> SNPS HS Phy.
>>
>> Signed-off-by: Krishna Kurapati <quic_kriskura@quicinc.com>
>> Reviewed-by: Pavankumar Kondeti <quic_pkondeti@quicinc.com>
>> ---
>>   drivers/phy/qualcomm/phy-qcom-snps-femto-v2.c | 262 +++++++++++++++++++++++++-
>>   1 file changed, 260 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/phy/qualcomm/phy-qcom-snps-femto-v2.c b/drivers/phy/qualcomm/phy-qcom-snps-femto-v2.c
>> index 5d20378..24264ad 100644
>> --- a/drivers/phy/qualcomm/phy-qcom-snps-femto-v2.c
>> +++ b/drivers/phy/qualcomm/phy-qcom-snps-femto-v2.c
>> @@ -52,6 +52,12 @@
>>   #define USB2_SUSPEND_N				BIT(2)
>>   #define USB2_SUSPEND_N_SEL			BIT(3)
>>   
>> +#define USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X0		(0x6c)
>> +#define USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X1		(0x70)
>> +#define USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X2		(0x74)
>> +#define USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X3		(0x78)
>> +#define PARAM_OVRD_MASK				0xFF
>> +
>>   #define USB2_PHY_USB_PHY_CFG0			(0x94)
>>   #define UTMI_PHY_DATAPATH_CTRL_OVERRIDE_EN	BIT(0)
>>   #define UTMI_PHY_CMN_CTRL_OVERRIDE_EN		BIT(1)
>> @@ -60,12 +66,69 @@
>>   #define REFCLK_SEL_MASK				GENMASK(1, 0)
>>   #define REFCLK_SEL_DEFAULT			(0x2 << 0)
>>   
>> +#define HS_DISCONNECT_MASK			GENMASK(2, 0)
>> +#define SQUELCH_DETECTOR_MASK			GENMASK(7, 5)
>> +
>> +#define HS_AMPLITUDE_MASK			GENMASK(3, 0)
>> +#define PREEMPHASIS_DURATION_MASK		BIT(5)
>> +#define PREEMPHASIS_AMPLITUDE_MASK		GENMASK(7, 6)
>> +
>> +#define HS_RISE_FALL_MASK			GENMASK(1, 0)
>> +#define HS_CROSSOVER_VOLTAGE_MASK		GENMASK(3, 2)
>> +#define HS_OUTPUT_IMPEDANCE_MASK		GENMASK(5, 4)
>> +
>> +#define LS_FS_OUTPUT_IMPEDANCE_MASK		GENMASK(3, 0)
>> +
>>   static const char * const qcom_snps_hsphy_vreg_names[] = {
>>   	"vdda-pll", "vdda33", "vdda18",
>>   };
>>   
>>   #define SNPS_HS_NUM_VREGS		ARRAY_SIZE(qcom_snps_hsphy_vreg_names)
>>   
>> +struct override_param {
>> +	s32	value;
>> +	u8	reg;
>
Hi Bjorn,

> It wasn't immediately obvious to me that reg meant "register value to be
> written for the given value specified in DT".
>
Yes, reg meant the register value to be written corresponding to the 
dt_val provided.

> I think if you changed "reg" to "reg_val" it would have been.
>Sure. Will make this change in the next version.

>> +};
>> +
>> +#define OVERRIDE_PARAM(bps, val) {	\
> 
> TBH, I don't find this macro to add any value. In order to understand
> the tables below I need to look at this macro to know that the two
> integers passed to the macro are just stuffed into the array - there's
> nothing more to it.
> 
> As such I would prefer if you simply wrote:
> 
Sure. I just didn't want to make it look like some random values, but 
since these are coming straight out of datasheet and the macro is just 
to assign values, I can remove this in the next version.

> static const struct override_param hs_disconnect_sc7280[] = {
> 	{ -272, 0 },
> 	{ 0, 1 },
> 	{ 317, 2 },
> 	...
> }
> 
>> +	.value = bps,			\
>> +	.reg = val,			\
>> +}
>> +
>> +struct override_param_map {
>> +	const struct override_param *param_table;
>> +	u8 table_size;
>> +	u8 reg_offset;
>> +	u8 param_mask;
>> +};
>> +
>> +#define OVERRIDE_PARAM_MAP(table, num_elements, offset, mask)		\
>> +{									\
>> +	.param_table = table,						\
>> +	.table_size = num_elements,					\
>> +	.reg_offset = offset,						\
>> +	.param_mask = mask,						\
>> +}
> 
> Here you could have used the preprocessor to fill in table_size (by just
> making it ARRAY_SIZE(table)). But as with OVERRIDE_PARAM I would prefer
> if you simply do:
> 
I used ARRAY_SIZE(table) in the code below to indicate the number of 
input data elements present for each property.
> 
> static const struct override_param_map sc7280[] = {
> 	{
> 		hs_disconnect_sc7280,
> 		ARRAY_SIZE(hs_disconnect_sc7280),
> 		USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X0,
> 		HS_DISCONNECT_MASK),
> 	},
> 	{
> 		squelch_det_threshold_sc7280,
> 		ARRAY_SIZE(squelch_det_threshold_sc7280),
> 		USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X0,
> 		SQUELCH_DETECTOR_MASK),
> 	},
> 	...
> }
> 
>> +
>> +struct phy_override_seq {
>> +	bool	need_update;
>> +	u8	offset;
>> +	u8	value;
>> +	u8	mask;
>> +};
>> +
>> +static const char * const phy_seq_props[] = {
>> +	"qcom,hs-disconnect-bp",
> 
> The ordering of this list needs to match the order of
> override_param_map[] and there's nothing indicating this in the code.
> 
> I was considering suggesting that you add a enum/define and do
> 	[SQUELCH_DETECTOR_BP] = "qcom,squelch-detector-bp",
> 	...
> and then do the same in the override_param_map array.
> >
> But I think it will be cleaner if you add a const char const pointer to
> override_param_map and just specify these strings in the
> override_param_map array.
> 
> Each entry will grow by a pointer, but multiple copies of the same
> strings (when added in the future) should be combined by the compiler.
>
IIUC, you want me to remove this array of const char*'s and embed them 
in the override_param_map and iterate through it without using this 
const phy_seq_props as a reference.

>> +	"qcom,squelch-detector-bp",
>> +	"qcom,hs-amplitude-bp",
>> +	"qcom,pre-emphasis-duration-bp",
>> +	"qcom,pre-emphasis-amplitude-bp",
>> +	"qcom,hs-rise-fall-time-bp",
>> +	"qcom,hs-crossover-voltage-microvolt",
>> +	"qcom,hs-output-impedance-micro-ohms",
>> +	"qcom,ls-fs-output-impedance-bp",
>> +};
>> +
>>   /**
>>    * struct qcom_snps_hsphy - snps hs phy attributes
>>    *
>> @@ -91,6 +154,7 @@ struct qcom_snps_hsphy {
>>   
>>   	bool phy_initialized;
>>   	enum phy_mode mode;
>> +	struct phy_override_seq update_seq_cfg[ARRAY_SIZE(phy_seq_props)];
>>   };
>>   
>>   static inline void qcom_snps_hsphy_write_mask(void __iomem *base, u32 offset,
>> @@ -173,10 +237,147 @@ static int qcom_snps_hsphy_set_mode(struct phy *phy, enum phy_mode mode,
>>   	return 0;
>>   }
>>   
>> +static const struct override_param hs_disconnect_sc7280[] = {
>> +	OVERRIDE_PARAM(-272, 0),
>> +	OVERRIDE_PARAM(0, 1),
>> +	OVERRIDE_PARAM(317, 2),
>> +	OVERRIDE_PARAM(630, 3),
>> +	OVERRIDE_PARAM(973, 4),
>> +	OVERRIDE_PARAM(1332, 5),
>> +	OVERRIDE_PARAM(1743, 6),
>> +	OVERRIDE_PARAM(2156, 7),
>> +};
>> +
>> +static const struct override_param squelch_det_threshold_sc7280[] = {
>> +	OVERRIDE_PARAM(-2090, 7),
>> +	OVERRIDE_PARAM(-1560, 6),
>> +	OVERRIDE_PARAM(-1030, 5),
>> +	OVERRIDE_PARAM(-530, 4),
>> +	OVERRIDE_PARAM(0, 3),
>> +	OVERRIDE_PARAM(530, 2),
>> +	OVERRIDE_PARAM(1060, 1),
>> +	OVERRIDE_PARAM(1590, 0),
>> +};
>> +
>> +static const struct override_param hs_amplitude_sc7280[] = {
>> +	OVERRIDE_PARAM(-660, 0),
>> +	OVERRIDE_PARAM(-440, 1),
>> +	OVERRIDE_PARAM(-220, 2),
>> +	OVERRIDE_PARAM(0, 3),
>> +	OVERRIDE_PARAM(230, 4),
>> +	OVERRIDE_PARAM(440, 5),
>> +	OVERRIDE_PARAM(650, 6),
>> +	OVERRIDE_PARAM(890, 7),
>> +	OVERRIDE_PARAM(1110, 8),
>> +	OVERRIDE_PARAM(1330, 9),
>> +	OVERRIDE_PARAM(1560, 10),
>> +	OVERRIDE_PARAM(1780, 11),
>> +	OVERRIDE_PARAM(2000, 12),
>> +	OVERRIDE_PARAM(2220, 13),
>> +	OVERRIDE_PARAM(2430, 14),
>> +	OVERRIDE_PARAM(2670, 15),
>> +};
>> +
>> +static const struct override_param preemphasis_duration_sc7280[] = {
>> +	OVERRIDE_PARAM(10000, 1),
>> +	OVERRIDE_PARAM(20000, 0),
>> +};
>> +
>> +static const struct override_param preemphasis_amplitude_sc7280[] = {
>> +	OVERRIDE_PARAM(10000, 1),
>> +	OVERRIDE_PARAM(20000, 2),
>> +	OVERRIDE_PARAM(30000, 3),
>> +	OVERRIDE_PARAM(40000, 0),
>> +};
>> +
>> +static const struct override_param hs_rise_fall_time_sc7280[] = {
>> +	OVERRIDE_PARAM(-4100, 3),
>> +	OVERRIDE_PARAM(0, 2),
>> +	OVERRIDE_PARAM(2810, 1),
>> +	OVERRIDE_PARAM(5430, 0),
>> +};
>> +
>> +static const struct override_param hs_crossover_voltage_sc7280[] = {
>> +	OVERRIDE_PARAM(-31000, 1),
>> +	OVERRIDE_PARAM(0, 3),
>> +	OVERRIDE_PARAM(28000, 2),
>> +};
>> +
>> +static const struct override_param hs_output_impedance_sc7280[] = {
>> +	OVERRIDE_PARAM(-2300000, 3),
>> +	OVERRIDE_PARAM(0, 2),
>> +	OVERRIDE_PARAM(2600000, 1),
>> +	OVERRIDE_PARAM(6100000, 0),
>> +};
>> +
>> +static const struct override_param ls_fs_output_impedance_sc7280[] = {
>> +	OVERRIDE_PARAM(-1053, 15),
>> +	OVERRIDE_PARAM(-557, 7),
>> +	OVERRIDE_PARAM(0, 3),
>> +	OVERRIDE_PARAM(612, 1),
>> +	OVERRIDE_PARAM(1310, 0),
>> +};
>> +
>> +static const struct override_param_map sc7280[] = {
> 
> There's nothing ensuring that the loop below doesn't run off the end of
> this array. So when the next platform is added, there's no way to
> handle the fact that they might have a different set of properties.
> 
> If you add the property name to these elements, that will no longer be a
> problem (and you can add a {} entry at the end of the list and check for
> this when looping over the elements.
> 
>> +	OVERRIDE_PARAM_MAP(
>> +			hs_disconnect_sc7280,
>> +			ARRAY_SIZE(hs_disconnect_sc7280),
>> +			USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X0,
>> +			HS_DISCONNECT_MASK),
>> +
>> +	OVERRIDE_PARAM_MAP(
>> +			squelch_det_threshold_sc7280,
>> +			ARRAY_SIZE(squelch_det_threshold_sc7280),
>> +			USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X0,
>> +			SQUELCH_DETECTOR_MASK),
>> +
>> +	OVERRIDE_PARAM_MAP(
>> +			hs_amplitude_sc7280,
>> +			ARRAY_SIZE(hs_amplitude_sc7280),
>> +			USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X1,
>> +			HS_AMPLITUDE_MASK),
>> +
>> +	OVERRIDE_PARAM_MAP(
>> +			preemphasis_duration_sc7280,
>> +			ARRAY_SIZE(preemphasis_duration_sc7280),
>> +			USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X1,
>> +			PREEMPHASIS_DURATION_MASK),
>> +
>> +	OVERRIDE_PARAM_MAP(
>> +			preemphasis_amplitude_sc7280,
>> +			ARRAY_SIZE(preemphasis_amplitude_sc7280),
>> +			USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X1,
>> +			PREEMPHASIS_AMPLITUDE_MASK),
>> +
>> +	OVERRIDE_PARAM_MAP(
>> +			hs_rise_fall_time_sc7280,
>> +			ARRAY_SIZE(hs_rise_fall_time_sc7280),
>> +			USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X2,
>> +			HS_RISE_FALL_MASK),
>> +
>> +	OVERRIDE_PARAM_MAP(
>> +			hs_crossover_voltage_sc7280,
>> +			ARRAY_SIZE(hs_crossover_voltage_sc7280),
>> +			USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X2,
>> +			HS_CROSSOVER_VOLTAGE_MASK),
>> +
>> +	OVERRIDE_PARAM_MAP(
>> +			hs_output_impedance_sc7280,
>> +			ARRAY_SIZE(hs_output_impedance_sc7280),
>> +			USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X2,
>> +			HS_OUTPUT_IMPEDANCE_MASK),
>> +
>> +	OVERRIDE_PARAM_MAP(
>> +			ls_fs_output_impedance_sc7280,
>> +			ARRAY_SIZE(ls_fs_output_impedance_sc7280),
>> +			USB2_PHY_USB_PHY_HS_PHY_OVERRIDE_X3,
>> +			LS_FS_OUTPUT_IMPEDANCE_MASK),
>> +};
>> +
>>   static int qcom_snps_hsphy_init(struct phy *phy)
>>   {
>>   	struct qcom_snps_hsphy *hsphy = phy_get_drvdata(phy);
>> -	int ret;
>> +	int ret, i;
>>   
>>   	dev_vdbg(&phy->dev, "%s(): Initializing SNPS HS phy\n", __func__);
>>   
>> @@ -223,6 +424,14 @@ static int qcom_snps_hsphy_init(struct phy *phy)
>>   	qcom_snps_hsphy_write_mask(hsphy->base, USB2_PHY_USB_PHY_HS_PHY_CTRL1,
>>   					VBUSVLDEXT0, VBUSVLDEXT0);
>>   
>> +	for (i = 0; i < ARRAY_SIZE(hsphy->update_seq_cfg); i++) {
>> +		if (hsphy->update_seq_cfg[i].need_update)
>> +			qcom_snps_hsphy_write_mask(hsphy->base,
>> +					hsphy->update_seq_cfg[i].offset,
>> +					hsphy->update_seq_cfg[i].mask,
>> +					hsphy->update_seq_cfg[i].value);
>> +	}
>> +
>>   	qcom_snps_hsphy_write_mask(hsphy->base,
>>   					USB2_PHY_USB_PHY_HS_PHY_CTRL_COMMON2,
>>   					VREGBYPASS, VREGBYPASS);
>> @@ -280,7 +489,10 @@ static const struct phy_ops qcom_snps_hsphy_gen_ops = {
>>   static const struct of_device_id qcom_snps_hsphy_of_match_table[] = {
>>   	{ .compatible	= "qcom,sm8150-usb-hs-phy", },
>>   	{ .compatible	= "qcom,usb-snps-hs-5nm-phy", },
>> -	{ .compatible	= "qcom,usb-snps-hs-7nm-phy", },
>> +	{
>> +		.compatible	= "qcom,usb-snps-hs-7nm-phy",
>> +		.data		= &sc7280,
> 
> Perhaps name "qcom_snps_hsphy_7nm_override_map" or something
> along those lines instead of "sc7280"?
My bad, I left it blant. I will rename this appropriately in the next 
version.
> 
>> +	},
>>   	{ .compatible	= "qcom,usb-snps-femto-v2-phy",	},
>>   	{ }
>>   };
>> @@ -291,6 +503,51 @@ static const struct dev_pm_ops qcom_snps_hsphy_pm_ops = {
>>   			   qcom_snps_hsphy_runtime_resume, NULL)
>>   };
>>   
>> +static void qcom_snps_hsphy_override_param_update_val(
>> +			const struct override_param_map map,
>> +			s32 dt_val, struct phy_override_seq *seq_entry)
>> +{
>> +	int i;
>> +
>> +	/*
>> +	 * Param table for each param is in increasing order
>> +	 * of dt values. We need to iterate over the list to
>> +	 * select the entry that has equal or the next highest value.
>> +	 */
>> +	for (i = 0; i < map.table_size - 1; i++) {
>> +		if (map.param_table[i].value == dt_val)
>> +			break;
>> +	}
>> +
>> +	seq_entry->need_update = true;
>> +	seq_entry->offset = map.reg_offset;
>> +	seq_entry->mask = map.param_mask;
>> +	seq_entry->value =  map.param_table[i].reg << __ffs(map.param_mask);
> 
> I would prefer that you include linux/bitfield.h and use:
> 
> 	u32_encode_bits(map.param_table[i].reg, map.param_mask);
> 
> instead of writing this yourself (even if it looks correct).
> 
> There's a double space after '='.
> 
> Regards,
> Bjorn
> 
Thanks,
Krishna,
>> +}
>> +
>> +static void qcom_snps_hsphy_read_override_param_seq(struct device *dev)
>> +{
>> +	struct device_node *node = dev->of_node;
>> +	s32 val;
>> +	int ret, i;
>> +	struct qcom_snps_hsphy *hsphy;
>> +	const struct override_param_map *cfg = of_device_get_match_data(dev);
>> +
>> +	hsphy = dev_get_drvdata(dev);
>> +
>> +	for (i = 0; i < ARRAY_SIZE(phy_seq_props); i++) {
>> +		ret = of_property_read_s32(node, phy_seq_props[i], &val);
>> +		if (ret)
>> +			continue;
>> +
>> +		qcom_snps_hsphy_override_param_update_val(cfg[i], val,
>> +					&hsphy->update_seq_cfg[i]);
>> +		dev_dbg(&hsphy->phy->dev, "Read param: %s dt_val: %d reg_val: 0x%x\n",
>> +			phy_seq_props[i], val, hsphy->update_seq_cfg[i].value);
>> +
>> +	}
>> +}
>> +
>>   static int qcom_snps_hsphy_probe(struct platform_device *pdev)
>>   {
>>   	struct device *dev = &pdev->dev;
>> @@ -352,6 +609,7 @@ static int qcom_snps_hsphy_probe(struct platform_device *pdev)
>>   
>>   	dev_set_drvdata(dev, hsphy);
>>   	phy_set_drvdata(generic_phy, hsphy);
>> +	qcom_snps_hsphy_read_override_param_seq(dev);
>>   
>>   	phy_provider = devm_of_phy_provider_register(dev, of_phy_simple_xlate);
>>   	if (!IS_ERR(phy_provider))
>> -- 
>> 2.7.4
>>