[PATCH v5 1/2] dt-bindings: ufs: Document static TX Equalization settings properties

[PATCH v5 1/2] dt-bindings: ufs: Document static TX Equalization settings properties

[Can Guo]

UFS v5.0/UFSHCI v5.0 add HS-G6 support (46.6 Gbps/lane) via UniPro v3.0
and M-PHY v6.0. In these specs, TX Equalization is defined for all High
Speed Gears (not only HS-G6) to compensate channel loss and improve signal
integrity at high speed operation.

For HS-G6, M-PHY uses PAM4 1b1b line coding, Pre-Coding may also be
required depending on channel characteristics.

Add vendor-neutral DT properties:

- patternProperties for txeq-preshoot-g[1-6] and txeq-deemphasis-g[1-6]
- fixed property tx-precode-enable-g6

Each property is a uint32 array of per-lane tuples:
<Host_Lane0 Device_Lane0>, [<Host_Lane1 Device_Lane1>]

Accept 2 or 4 values (x1/x2 lane configs). PreShoot and DeEmphasis values
are 0..7. Precode enable values are 0/1 and only applicable to HS-G6.

Acked-by: Manivannan Sadhasivam <mani@kernel.org>
Reviewed-by: Bean Huo <beanhuo@micron.com>
Signed-off-by: Can Guo <can.guo@oss.qualcomm.com>
---
 .../devicetree/bindings/ufs/ufs-common.yaml   | 45 +++++++++++++++++++
 1 file changed, 45 insertions(+)

diff --git a/Documentation/devicetree/bindings/ufs/ufs-common.yaml b/Documentation/devicetree/bindings/ufs/ufs-common.yaml
index ed97f5682509..d90cf25adfa5 100644
--- a/Documentation/devicetree/bindings/ufs/ufs-common.yaml
+++ b/Documentation/devicetree/bindings/ufs/ufs-common.yaml
@@ -105,6 +105,51 @@ properties:
       Restricts the UFS controller to rate-a or rate-b for both TX and
       RX directions.
 
+  tx-precode-enable-g6:
+    $ref: /schemas/types.yaml#/definitions/uint32-array
+    oneOf:
+      - minItems: 2
+        maxItems: 2
+      - minItems: 4
+        maxItems: 4
+    items:
+      enum: [0, 1]
+    description: |
+      Static TX Precode enable values for HS-G6 only.
+      Values are specified as per-lane tuples:
+      <Host_Lane0 Device_Lane0>, [<Host_Lane1 Device_Lane1>].
+
+patternProperties:
+  "^txeq-preshoot-g[1-6]$":
+    $ref: /schemas/types.yaml#/definitions/uint32-array
+    oneOf:
+      - minItems: 2
+        maxItems: 2
+      - minItems: 4
+        maxItems: 4
+    items:
+      minimum: 0
+      maximum: 7
+    description: |
+      Static TX Equalization PreShoot values for High Speed Gears.
+      Values are specified as per-lane tuples:
+      <Host_Lane0 Device_Lane0>, [<Host_Lane1 Device_Lane1>].
+
+  "^txeq-deemphasis-g[1-6]$":
+    $ref: /schemas/types.yaml#/definitions/uint32-array
+    oneOf:
+      - minItems: 2
+        maxItems: 2
+      - minItems: 4
+        maxItems: 4
+    items:
+      minimum: 0
+      maximum: 7
+    description: |
+      Static TX Equalization DeEmphasis values for High Speed Gears.
+      Values are specified as per-lane tuples:
+      <Host_Lane0 Device_Lane0>, [<Host_Lane1 Device_Lane1>].
+
 dependencies:
   freq-table-hz: [ clocks ]
   operating-points-v2: [ clocks, clock-names ]
-- 
2.34.1

[PATCH v5 2/2] scsi: ufs: core: Add support for static TX Equalization settings

[Can Guo]

Static TX Equalization settings and TX Precode enable indication from DT
properties txeq-preshoot-g[1-6], txeq-deemphasis-g[1-6], and
tx-precode-enable-g6 are board-specific baseline values. Values are
provided as per-lane tuples:

<Host_Lane0 Device_Lane0>, [<Host_Lane1 Device_Lane1>]

Parse DT u32 properties with explicit range checks by using
of_property_count_u32_elems()/of_property_read_u32_array().

When adaptive TX Equalization is used, these static settings are not final:

- If valid settings are retrieved from qTxEQGnSettings/wTxEQGnSettingsExt,
  those retrieved settings override static DT settings.
- If retrieval is not available/valid, TX EQTR runs and trained settings
  override static DT settings.

So static DT settings are a fallback and are intended for cases where
adaptive TX Equalization is not enabled/used. Adaptive TX Equalization
remains the primary path when enabled.

No behavior changes for platforms that do not provide these properties.

Signed-off-by: Can Guo <can.guo@oss.qualcomm.com>
---
 drivers/ufs/core/ufs-txeq.c      |  10 ++-
 drivers/ufs/host/ufshcd-pltfrm.c | 137 +++++++++++++++++++++++++++++++
 include/ufs/ufshcd.h             |   2 +
 3 files changed, 148 insertions(+), 1 deletion(-)

diff --git a/drivers/ufs/core/ufs-txeq.c b/drivers/ufs/core/ufs-txeq.c
index 4b264adfdf49..b645fe5f6d95 100644
--- a/drivers/ufs/core/ufs-txeq.c
+++ b/drivers/ufs/core/ufs-txeq.c
@@ -1297,7 +1297,13 @@ int ufshcd_config_tx_eq_settings(struct ufs_hba *hba,
 	}
 
 	params = &hba->tx_eq_params[gear - 1];
-	if (!params->is_valid || force_tx_eqtr) {
+	/*
+	 * TX EQTR must run for the following cases:
+	 * 1. TX EQ settings are invalid.
+	 * 2. TX EQ settings are valid but static, i.e., populated from DT.
+	 * 3. TX EQTR procedure is forced.
+	 */
+	if (!params->is_valid || params->is_static || force_tx_eqtr) {
 		int ret;
 
 		ret = ufshcd_tx_eqtr(hba, params, pwr_mode);
@@ -1310,6 +1316,7 @@ int ufshcd_config_tx_eq_settings(struct ufs_hba *hba,
 		/* Mark TX Equalization settings as valid */
 		params->is_valid = true;
 		params->is_trained = true;
+		params->is_static = false;
 		params->is_applied = false;
 	}
 
@@ -1495,6 +1502,7 @@ static void ufshcd_extract_tx_eq_settings_attrs(struct ufs_hba *hba, u8 gear)
 	}
 
 	params->is_valid = true;
+	params->is_static = false;
 }
 
 void ufshcd_retrieve_tx_eq_settings(struct ufs_hba *hba)
diff --git a/drivers/ufs/host/ufshcd-pltfrm.c b/drivers/ufs/host/ufshcd-pltfrm.c
index c2dafb583cf5..c498ea6a86a3 100644
--- a/drivers/ufs/host/ufshcd-pltfrm.c
+++ b/drivers/ufs/host/ufshcd-pltfrm.c
@@ -210,6 +210,141 @@ static void ufshcd_init_lanes_per_dir(struct ufs_hba *hba)
 	}
 }
 
+/**
+ * ufshcd_parse_tx_eq_settings_for_gear - Parse static TX EQ DT settings for one gear
+ * @hba: per adapter instance
+ * @gear: target HS gear
+ * @num_elems: expected number of elements per property
+ *
+ * Reads the txeq-preshoot-gN, txeq-deemphasis-gN, and (for G6)
+ * tx-precode-enable-gN device-tree properties and, if all are valid, stores
+ * them as static TX Equalization settings for the given gear.
+ */
+static void ufshcd_parse_tx_eq_settings_for_gear(struct ufs_hba *hba,
+						 int gear, const u32 num_elems)
+{
+	u32 precode_en[UFS_MAX_LANES * 2] = { 0 };
+	const u32 lpd = hba->lanes_per_direction;
+	struct ufshcd_tx_eq_params *params;
+	u32 deemphasis[UFS_MAX_LANES * 2];
+	u32 preshoot[UFS_MAX_LANES * 2];
+	struct device *dev = hba->dev;
+	char prop_name[MAX_PROP_SIZE];
+	int i, err, lane, count;
+
+	snprintf(prop_name, MAX_PROP_SIZE, "txeq-preshoot-g%d", gear);
+	count = of_property_count_u32_elems(dev->of_node, prop_name);
+	if (count <= 0)
+		return;
+
+	if (count != num_elems) {
+		dev_err(dev, "Property %s has invalid count (%d), expecting %u\n",
+			prop_name, count, num_elems);
+		return;
+	}
+
+	err = of_property_read_u32_array(dev->of_node, prop_name, preshoot, num_elems);
+	if (err) {
+		dev_err(dev, "Failed to read %s property, %d\n", prop_name, err);
+		return;
+	}
+
+	for (i = 0; i < count; i++) {
+		if (preshoot[i] >= TX_HS_NUM_PRESHOOT) {
+			dev_err(dev, "An invalid TX EQ PreShoot (%d) provided in %s property\n",
+				preshoot[i], prop_name);
+			return;
+		}
+	}
+
+	snprintf(prop_name, MAX_PROP_SIZE, "txeq-deemphasis-g%d", gear);
+	count = of_property_count_u32_elems(dev->of_node, prop_name);
+	if (count <= 0) {
+		dev_err(dev, "Missing required %s property\n", prop_name);
+		return;
+	}
+
+	if (count != num_elems) {
+		dev_err(dev, "Property %s has invalid count (%d), expecting %u\n",
+			prop_name, count, num_elems);
+		return;
+	}
+
+	err = of_property_read_u32_array(dev->of_node, prop_name, deemphasis, num_elems);
+	if (err) {
+		dev_err(dev, "Failed to read %s property, %d\n", prop_name, err);
+		return;
+	}
+
+	for (i = 0; i < count; i++) {
+		if (deemphasis[i] >= TX_HS_NUM_DEEMPHASIS) {
+			dev_err(dev, "An invalid TX EQ DeEmphasis (%d) provided in %s property\n",
+				deemphasis[i], prop_name);
+			return;
+		}
+	}
+
+	if (gear == UFS_HS_G6) {
+		snprintf(prop_name, MAX_PROP_SIZE, "tx-precode-enable-g%d", gear);
+		count = of_property_count_u32_elems(dev->of_node, prop_name);
+		if (count > 0) {
+			if (count != num_elems) {
+				dev_err(dev, "Property %s has invalid count (%d), expecting %u\n",
+					prop_name, count, num_elems);
+				return;
+			}
+
+			err = of_property_read_u32_array(dev->of_node, prop_name,
+							 precode_en, num_elems);
+			if (err) {
+				dev_err(dev, "Failed to read %s property, %d\n",
+					prop_name, err);
+				return;
+			}
+
+			for (i = 0; i < count; i++) {
+				if (precode_en[i] > 1) {
+					dev_err(dev, "An invalid PrecodeEn (%d) provided in %s property\n",
+						precode_en[i], prop_name);
+					return;
+				}
+			}
+		}
+	}
+
+	params = &hba->tx_eq_params[gear - 1];
+	for (lane = 0; lane < lpd; lane++) {
+		params->host[lane].preshoot = preshoot[lane * 2];
+		params->host[lane].deemphasis = deemphasis[lane * 2];
+		params->host[lane].precode_en = precode_en[lane * 2];
+
+		params->device[lane].preshoot = preshoot[lane * 2 + 1];
+		params->device[lane].deemphasis = deemphasis[lane * 2 + 1];
+		params->device[lane].precode_en = precode_en[lane * 2 + 1];
+	}
+
+	params->is_valid = true;
+	params->is_static = true;
+}
+
+static void ufshcd_parse_static_tx_eq_settings(struct ufs_hba *hba)
+{
+	const u32 lpd = hba->lanes_per_direction;
+	const u32 num_elems = lpd * 2;
+	int gear;
+
+	if (!lpd) {
+		return;
+	} else if (lpd > UFS_MAX_LANES) {
+		dev_warn(hba->dev, "lanes_per_direction (%u) exceeds UFS_MAX_LANES (%u)\n",
+			 lpd, UFS_MAX_LANES);
+		return;
+	}
+
+	for (gear = UFS_HS_G1; gear <= UFS_HS_GEAR_MAX; gear++)
+		ufshcd_parse_tx_eq_settings_for_gear(hba, gear, num_elems);
+}
+
 /**
  * ufshcd_parse_clock_min_max_freq  - Parse MIN and MAX clocks freq
  * @hba: per adapter instance
@@ -528,6 +663,8 @@ int ufshcd_pltfrm_init(struct platform_device *pdev,
 
 	ufshcd_init_lanes_per_dir(hba);
 
+	ufshcd_parse_static_tx_eq_settings(hba);
+
 	err = ufshcd_parse_operating_points(hba);
 	if (err) {
 		dev_err(dev, "%s: OPP parse failed %d\n", __func__, err);
diff --git a/include/ufs/ufshcd.h b/include/ufs/ufshcd.h
index f48d6416e299..c01824576472 100644
--- a/include/ufs/ufshcd.h
+++ b/include/ufs/ufshcd.h
@@ -359,6 +359,7 @@ struct ufshcd_tx_eqtr_record {
  * @is_valid: True if parameter contains valid TX Equalization settings
  * @is_applied: True if settings have been applied to UniPro of both sides
  * @is_trained: True if parameters obtained from TX EQTR procedure
+ * @is_static: True if settings are static
  */
 struct ufshcd_tx_eq_params {
 	struct ufshcd_tx_eq_settings host[UFS_MAX_LANES];
@@ -367,6 +368,7 @@ struct ufshcd_tx_eq_params {
 	bool is_valid;
 	bool is_applied;
 	bool is_trained;
+	bool is_static;
 };
 
 /**
-- 
2.34.1

Re: [PATCH v5 1/2] dt-bindings: ufs: Document static TX Equalization settings properties

[Peter Wang (王信友)]

On Thu, 2026-05-28 at 18:14 -0700, Can Guo wrote:
> UFS v5.0/UFSHCI v5.0 add HS-G6 support (46.6 Gbps/lane) via UniPro
> v3.0
> and M-PHY v6.0. In these specs, TX Equalization is defined for all
> High
> Speed Gears (not only HS-G6) to compensate channel loss and improve
> signal
> integrity at high speed operation.
> 
> For HS-G6, M-PHY uses PAM4 1b1b line coding, Pre-Coding may also be
> required depending on channel characteristics.
> 
> Add vendor-neutral DT properties:
> 
> - patternProperties for txeq-preshoot-g[1-6] and txeq-deemphasis-g[1-
> 6]
> - fixed property tx-precode-enable-g6
> 
> Each property is a uint32 array of per-lane tuples:
> <Host_Lane0 Device_Lane0>, [<Host_Lane1 Device_Lane1>]
> 
> Accept 2 or 4 values (x1/x2 lane configs). PreShoot and DeEmphasis
> values
> are 0..7. Precode enable values are 0/1 and only applicable to HS-G6.
> 
> Acked-by: Manivannan Sadhasivam <mani@kernel.org>
> Reviewed-by: Bean Huo <beanhuo@micron.com>
> Signed-off-by: Can Guo <can.guo@oss.qualcomm.com>
> ---

Reviewed-by: Peter Wang <peter.wang@mediatek.com>

Re: [PATCH v5 2/2] scsi: ufs: core: Add support for static TX Equalization settings

[Peter Wang (王信友)]

On Thu, 2026-05-28 at 18:14 -0700, Can Guo wrote:
> +       for (i = 0; i < count; i++) {
> +               if (preshoot[i] >= TX_HS_NUM_PRESHOOT) {
> 
> 
> +       for (i = 0; i < count; i++) {
> +               if (deemphasis[i] >= TX_HS_NUM_DEEMPHASIS) {
> 
> +                       for (i = 0; i < count; i++) {
> +                               if (precode_en[i] > 1) {
> 

Hi Can,

I suggest using
for (i = 0; i < num_elems; i++) { ... }
instead of
for (i = 0; i < count; i++) { ... }
as it is more clear.

> 
> +static void ufshcd_parse_static_tx_eq_settings(struct ufs_hba *hba)
> +{
> +       const u32 lpd = hba->lanes_per_direction;
> +       const u32 num_elems = lpd * 2;
> +       int gear;
> +
> +       if (!lpd) {
> +               return;
> +       } else if (lpd > UFS_MAX_LANES) {

Unnecessary else if after return.

The others look good to me.

Thanks.
Peter

Re: [PATCH v5 2/2] scsi: ufs: core: Add support for static TX Equalization settings

[Can Guo]

On 5/29/2026 4:04 PM, Peter Wang (王信友) wrote:
>
> On Thu, 2026-05-28 at 18:14 -0700, Can Guo wrote:
> > +       for (i = 0; i < count; i++) {
> > +               if (preshoot[i] >= TX_HS_NUM_PRESHOOT) {
> > 
> > 
> > +       for (i = 0; i < count; i++) {
> > +               if (deemphasis[i] >= TX_HS_NUM_DEEMPHASIS) {
> > 
> > +                       for (i = 0; i < count; i++) {
> > +                               if (precode_en[i] > 1) {
> > 
>
> Hi Can,
>
> I suggest using
> for (i = 0; i < num_elems; i++) { ... }
> instead of
> for (i = 0; i < count; i++) { ... }
> as it is more clear.
Thanks for the review. Let me change to num_elems.
> > 
> > +static void ufshcd_parse_static_tx_eq_settings(struct ufs_hba *hba)
> > +{
> > +       const u32 lpd = hba->lanes_per_direction;
> > +       const u32 num_elems = lpd * 2;
> > +       int gear;
> > +
> > +       if (!lpd) {
> > +               return;
> > +       } else if (lpd > UFS_MAX_LANES) {
>
> Unnecessary else if after return.
You mean removing 'else if (lpd > UFS_MAX_LANES)' and the dev_warn()?

Thanks,
Can Guo.
>
> The others look good to me.
>
> Thanks.
> Peter
>
> ************* MEDIATEK Confidentiality Notice
>   ********************
> The information contained in this e-mail message (including any
> attachments) may be confidential, proprietary, privileged, or otherwise
> exempt from disclosure under applicable laws. It is intended to be
> conveyed only to the designated recipient(s). Any use, dissemination,
> distribution, printing, retaining or copying of this e-mail (including its
> attachments) by unintended recipient(s) is strictly prohibited and may
> be unlawful. If you are not an intended recipient of this e-mail, or believe
>   
> that you have received this e-mail in error, please notify the sender
> immediately (by replying to this e-mail), delete any and all copies of
> this e-mail (including any attachments) from your system, and do not
> disclose the content of this e-mail to any other person. Thank you!

Re: [PATCH v5 2/2] scsi: ufs: core: Add support for static TX Equalization settings

[Peter Wang (王信友)]

On Fri, 2026-05-29 at 16:12 +0800, Can Guo wrote:
> > > +static void ufshcd_parse_static_tx_eq_settings(struct ufs_hba
> > > *hba)
> > > +{
> > > +       const u32 lpd = hba->lanes_per_direction;
> > > +       const u32 num_elems = lpd * 2;
> > > +       int gear;
> > > +
> > > +       if (!lpd) {
> > > +               return;
> > > +       } else if (lpd > UFS_MAX_LANES) {
> > 
> > Unnecessary else if after return.
> You mean removing 'else if (lpd > UFS_MAX_LANES)' and the dev_warn()?
> 
> Thanks,
> Can Guo.
> 

Hi Can,

No, what I mean is that the check below aligns better with 
kernel code style:

if (!lpd)
    return;

if (lpd > UFS_MAX_LANES) {
    ...
}

Thanks
Peter

Re: [PATCH v5 2/2] scsi: ufs: core: Add support for static TX Equalization settings

[Can Guo]

On 5/29/2026 4:40 PM, Peter Wang (王信友) wrote:
>
> On Fri, 2026-05-29 at 16:12 +0800, Can Guo wrote:
> > > > +static void ufshcd_parse_static_tx_eq_settings(struct ufs_hba
> > > > *hba)
> > > > +{
> > > > +       const u32 lpd = hba->lanes_per_direction;
> > > > +       const u32 num_elems = lpd * 2;
> > > > +       int gear;
> > > > +
> > > > +       if (!lpd) {
> > > > +               return;
> > > > +       } else if (lpd > UFS_MAX_LANES) {
> > > 
> > > Unnecessary else if after return.
> > You mean removing 'else if (lpd > UFS_MAX_LANES)' and the dev_warn()?
> > 
> > Thanks,
> > Can Guo.
> > 
>
> Hi Can,
>
> No, what I mean is that the check below aligns better with
> kernel code style:
>
> if (!lpd)
>      return;
>
> if (lpd > UFS_MAX_LANES) {
>      ...
> }
OK, got it.

Thanks,
Can Guo.
>
> Thanks
> Peter
>
> ************* MEDIATEK Confidentiality Notice
>   ********************
> The information contained in this e-mail message (including any
> attachments) may be confidential, proprietary, privileged, or otherwise
> exempt from disclosure under applicable laws. It is intended to be
> conveyed only to the designated recipient(s). Any use, dissemination,
> distribution, printing, retaining or copying of this e-mail (including its
> attachments) by unintended recipient(s) is strictly prohibited and may
> be unlawful. If you are not an intended recipient of this e-mail, or believe
>   
> that you have received this e-mail in error, please notify the sender
> immediately (by replying to this e-mail), delete any and all copies of
> this e-mail (including any attachments) from your system, and do not
> disclose the content of this e-mail to any other person. Thank you!

Re: [PATCH v5 2/2] scsi: ufs: core: Add support for static TX Equalization settings

[Bean Huo]

On Fri, 2026-05-29 at 16:42 +0800, Can Guo wrote:
> > if (!lpd)
> >       return;
> > 
> > if (lpd > UFS_MAX_LANES) {
> >       ...
> > }
> OK, got it.
> 
> Thanks,
> Can Guo.

With this and Peter's num_elems rename addressed in v6, please add:

Reviewed-by: Bean Huo <beanhuo@micron.com>

Thanks,
Bean

Re: [PATCH v5 2/2] scsi: ufs: core: Add support for static TX Equalization settings

[Can Guo]

On 5/29/2026 6:19 PM, Bean Huo wrote:
> On Fri, 2026-05-29 at 16:42 +0800, Can Guo wrote:
>>> if (!lpd)
>>>        return;
>>>
>>> if (lpd > UFS_MAX_LANES) {
>>>        ...
>>> }
>> OK, got it.
>>
>> Thanks,
>> Can Guo.
> With this and Peter's num_elems rename addressed in v6, please add:
>
> Reviewed-by: Bean Huo <beanhuo@micron.com>
Thanks sir.

Can Guo.
>
> Thanks,
> Bean
>