Skip to contentSkip to conversation

support for hwtstamp_get in phy - part 2

PATCH Superseded Vadim Fedorenko 3 participants 13 messages Nov 29, 2025 – Dec 1, 2025

Subsystems (3)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

v2 cover support for hwtstamp_get in phy - part 2

There are 2 drivers which had some inconsistency in HW timestamping configuration logic. Improve both of them and add hwtstamp_get() callback.

v1 -> v2:
* add checks and return error in case of unsupported TX tiemstamps
* get function for lan8814 and lan8841 are the same - keep only one

Vadim Fedorenko (4):
net: phy: micrel: improve HW timestamping config logic
net: phy: micrel: add HW timestamp configuration reporting
net: phy: microchip_rds_ptp: improve HW ts config logic
net: phy: microchip_rds_ptp: add HW timestamp configuration reporting

 drivers/net/phy/micrel.c            | 35 ++++++++++++++++++++++++-----
 drivers/net/phy/microchip_rds_ptp.c | 34 +++++++++++++++++++++++-----
 2 files changed, 58 insertions(+), 11 deletions(-)
Signature · 1 line
2.47.3
#1 lore raw

PATCH v2 1/4 net: phy: micrel: improve HW timestamping config logic

The driver was adjusting stored values independently of what was actually supported and configured. Improve logic to store values once all checks are passing

---
 drivers/net/phy/micrel.c | 21 +++++++++++++++------
 1 file changed, 15 insertions(+), 6 deletions(-)
drivers/net/phy/micrel.c +15 -6
index 05de68b9f719..1ada05dd305c 100644
@@ -3157,9 +3157,6 @@ static int lan8814_hwtstamp_set(struct mii_timestamper *mii_ts,
 	int txcfg = 0, rxcfg = 0;
 	int pkt_ts_enable;
 
-	ptp_priv->hwts_tx_type = config->tx_type;
-	ptp_priv->rx_filter = config->rx_filter;
-
 	switch (config->rx_filter) {
 	case HWTSTAMP_FILTER_NONE:
 		ptp_priv->layer = 0;
@@ -3187,6 +3184,18 @@ static int lan8814_hwtstamp_set(struct mii_timestamper *mii_ts,
 		return -ERANGE;
 	}
 
+	switch (config->rx_filter) {
+	case HWTSTAMP_TX_OFF:
+	case HWTSTAMP_TX_ON:
+	case HWTSTAMP_TX_ONESTEP_SYNC:
+		break;
+	default:
+		return -ERANGE;
+	}
+
+	ptp_priv->hwts_tx_type = config->tx_type;
+	ptp_priv->rx_filter = config->rx_filter;
+
 	if (ptp_priv->layer & PTP_CLASS_L2) {
 		rxcfg = PTP_RX_PARSE_CONFIG_LAYER2_EN_;
 		txcfg = PTP_TX_PARSE_CONFIG_LAYER2_EN_;
@@ -5051,9 +5060,6 @@ static int lan8841_hwtstamp_set(struct mii_timestamper *mii_ts,
 	int txcfg = 0, rxcfg = 0;
 	int pkt_ts_enable;
 
-	ptp_priv->hwts_tx_type = config->tx_type;
-	ptp_priv->rx_filter = config->rx_filter;
-
 	switch (config->rx_filter) {
 	case HWTSTAMP_FILTER_NONE:
 		ptp_priv->layer = 0;
@@ -5081,6 +5087,9 @@ static int lan8841_hwtstamp_set(struct mii_timestamper *mii_ts,
 		return -ERANGE;
 	}
 
+	ptp_priv->hwts_tx_type = config->tx_type;
+	ptp_priv->rx_filter = config->rx_filter;
+
 	/* Setup parsing of the frames and enable the timestamping for ptp
 	 * frames
 	 */
Signature · 1 line
2.47.3
maintainer Reviewed-by on v2 1/4
#2 lore raw

Re: [PATCH net-next v2 1/4] net: phy: micrel: improve HW timestamping config logic

Quoted 5 lines · Vadim Fedorenko

The driver was adjusting stored values independently of what was actually supported and configured. Improve logic to store values once all checks are passing

Andrew

#3 lore raw

Re: [PATCH net-next v2 1/4] net: phy: micrel: improve HW timestamping config logic

In reply to Vadim Fedorenko
Quoted 24 lines · On Sat, 29 Nov 2025 19:53:31 +0000 Vadim Fedorenko <[email protected]>…

The driver was adjusting stored values independently of what was actually supported and configured. Improve logic to store values once all checks are passing

---
 drivers/net/phy/micrel.c | 21 +++++++++++++++------
 1 file changed, 15 insertions(+), 6 deletions(-)
drivers/net/phy/micrel.c +0 -3
index 05de68b9f719..1ada05dd305c 100644
@@ -3157,9 +3157,6 @@ static int lan8814_hwtstamp_set(struct mii_timestamper
 *mii_ts, int txcfg = 0, rxcfg = 0;
 	int pkt_ts_enable;
 
-	ptp_priv->hwts_tx_type = config->tx_type;
-	ptp_priv->rx_filter = config->rx_filter;
-
 	switch (config->rx_filter) {
 	case HWTSTAMP_FILTER_NONE:
 		ptp_priv->layer = 0;

drivers/net/phy/micrel.c static int lan8814_hwtstamp_set(struct mii_timestamper · line 3187

Quoted 4 lines
drivers/net/phy/micrel.c +1 -0
@@ -3187,6 +3184,18 @@ static int lan8814_hwtstamp_set(struct mii_timestamper
 *mii_ts, return -ERANGE;
 	}
 
+	switch (config->rx_filter) {

You want to check tx_type here, not rx_filter.

Quoted 25 lines
drivers/net/phy/micrel.c +11 -3
+	case HWTSTAMP_TX_OFF:
+	case HWTSTAMP_TX_ON:
+	case HWTSTAMP_TX_ONESTEP_SYNC:
+		break;
+	default:
+		return -ERANGE;
+	}
+
+	ptp_priv->hwts_tx_type = config->tx_type;
+	ptp_priv->rx_filter = config->rx_filter;
+
 	if (ptp_priv->layer & PTP_CLASS_L2) {
 		rxcfg = PTP_RX_PARSE_CONFIG_LAYER2_EN_;
 		txcfg = PTP_TX_PARSE_CONFIG_LAYER2_EN_;
@@ -5051,9 +5060,6 @@ static int lan8841_hwtstamp_set(struct mii_timestamper
 *mii_ts, int txcfg = 0, rxcfg = 0;
 	int pkt_ts_enable;
 
-	ptp_priv->hwts_tx_type = config->tx_type;
-	ptp_priv->rx_filter = config->rx_filter;
-
 	switch (config->rx_filter) {
 	case HWTSTAMP_FILTER_NONE:
 		ptp_priv->layer = 0;

drivers/net/phy/micrel.c static int lan8841_hwtstamp_set(struct mii_timestamper · line 5091

Quoted 5 lines
drivers/net/phy/micrel.c +2 -0
@@ -5081,6 +5087,9 @@ static int lan8841_hwtstamp_set(struct mii_timestamper
 *mii_ts, return -ERANGE;
 	}
 
+	ptp_priv->hwts_tx_type = config->tx_type;
+	ptp_priv->rx_filter = config->rx_filter;

I there a reason to not add the check in the hwtstamp ops for lan8841 as well? because the issue is also present.

Regards,

Signature · 3 lines
Köry Maincent, Bootlin
Embedded Linux and kernel engineering
https://bootlin.com
#4 lore raw

Re: [PATCH net-next v2 1/4] net: phy: micrel: improve HW timestamping config logic

Quoted 62 lines · Kory Maincent
Quoted 24 lines · On Sat, 29 Nov 2025 19:53:31 +0000 Vadim Fedorenko <[email protected]>…

The driver was adjusting stored values independently of what was actually supported and configured. Improve logic to store values once all checks are passing

---
  drivers/net/phy/micrel.c | 21 +++++++++++++++------
  1 file changed, 15 insertions(+), 6 deletions(-)
drivers/net/phy/micrel.c +0 -3
index 05de68b9f719..1ada05dd305c 100644
@@ -3157,9 +3157,6 @@ static int lan8814_hwtstamp_set(struct mii_timestamper
 *mii_ts, int txcfg = 0, rxcfg = 0;
  	int pkt_ts_enable;
 
-	ptp_priv->hwts_tx_type = config->tx_type;
-	ptp_priv->rx_filter = config->rx_filter;
-
  	switch (config->rx_filter) {
  	case HWTSTAMP_FILTER_NONE:
  		ptp_priv->layer = 0;

drivers/net/phy/micrel.c static int lan8814_hwtstamp_set(struct mii_timestamper · line 3187

Quoted 4 lines
drivers/net/phy/micrel.c +1 -0
@@ -3187,6 +3184,18 @@ static int lan8814_hwtstamp_set(struct mii_timestamper
 *mii_ts, return -ERANGE;
  	}
 
+	switch (config->rx_filter) {

You want to check tx_type here, not rx_filter.

Quoted 28 lines
drivers/net/phy/micrel.c +11 -3
+	case HWTSTAMP_TX_OFF:
+	case HWTSTAMP_TX_ON:
+	case HWTSTAMP_TX_ONESTEP_SYNC:
+		break;
+	default:
+		return -ERANGE;
+	}
+
+	ptp_priv->hwts_tx_type = config->tx_type;
+	ptp_priv->rx_filter = config->rx_filter;
+
  	if (ptp_priv->layer & PTP_CLASS_L2) {
  		rxcfg = PTP_RX_PARSE_CONFIG_LAYER2_EN_;
  		txcfg = PTP_TX_PARSE_CONFIG_LAYER2_EN_;
@@ -5051,9 +5060,6 @@ static int lan8841_hwtstamp_set(struct mii_timestamper
 *mii_ts, int txcfg = 0, rxcfg = 0;
  	int pkt_ts_enable;
 
-	ptp_priv->hwts_tx_type = config->tx_type;
-	ptp_priv->rx_filter = config->rx_filter;
-
  	switch (config->rx_filter) {
  	case HWTSTAMP_FILTER_NONE:
  		ptp_priv->layer = 0;
@@ -5081,6 +5087,9 @@ static int lan8841_hwtstamp_set(struct mii_timestamper
 *mii_ts, return -ERANGE;
  	}
 
Quoted 5 lines

drivers/net/phy/micrel.c static int lan8841_hwtstamp_set(struct mii_timestamper · line 5091

Quoted 2 lines
drivers/net/phy/micrel.c +2 -0
+	ptp_priv->hwts_tx_type = config->tx_type;
+	ptp_priv->rx_filter = config->rx_filter;

I there a reason to not add the check in the hwtstamp ops for lan8841 as well? because the issue is also present.

Hm... I thought I added it... Well, looks like I shouldn't work during vacation :) Going to add it in v3, thanks!

Quoted 2 lines

Regards,

#5 lore raw

PATCH v2 2/4 net: phy: micrel: add HW timestamp configuration reporting

In reply to Vadim Fedorenko

The driver stores HW timestamping configuration and can technically report it. Add callback to do it.

---
 drivers/net/phy/micrel.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
drivers/net/phy/micrel.c +14 -0
index 1ada05dd305c..b48553edbfa4 100644
@@ -3147,6 +3147,18 @@ static void lan8814_flush_fifo(struct phy_device *phydev, bool egress)
 	lanphy_read_page_reg(phydev, LAN8814_PAGE_PORT_REGS, PTP_TSU_INT_STS);
 }
 
+static int lan8814_hwtstamp_get(struct mii_timestamper *mii_ts,
+				struct kernel_hwtstamp_config *config)
+{
+	struct kszphy_ptp_priv *ptp_priv =
+			  container_of(mii_ts, struct kszphy_ptp_priv, mii_ts);
+
+	config->tx_type = ptp_priv->hwts_tx_type;
+	config->rx_filter = ptp_priv->rx_filter;
+
+	return 0;
+}
+
 static int lan8814_hwtstamp_set(struct mii_timestamper *mii_ts,
 				struct kernel_hwtstamp_config *config,
 				struct netlink_ext_ack *extack)
@@ -4399,6 +4411,7 @@ static void lan8814_ptp_init(struct phy_device *phydev)
 	ptp_priv->mii_ts.rxtstamp = lan8814_rxtstamp;
 	ptp_priv->mii_ts.txtstamp = lan8814_txtstamp;
 	ptp_priv->mii_ts.hwtstamp_set = lan8814_hwtstamp_set;
+	ptp_priv->mii_ts.hwtstamp_get = lan8814_hwtstamp_get;
 	ptp_priv->mii_ts.ts_info  = lan8814_ts_info;
 
 	phydev->mii_ts = &ptp_priv->mii_ts;
@@ -5934,6 +5947,7 @@ static int lan8841_probe(struct phy_device *phydev)
 	ptp_priv->mii_ts.rxtstamp = lan8841_rxtstamp;
 	ptp_priv->mii_ts.txtstamp = lan8814_txtstamp;
 	ptp_priv->mii_ts.hwtstamp_set = lan8841_hwtstamp_set;
+	ptp_priv->mii_ts.hwtstamp_get = lan8814_hwtstamp_get;
 	ptp_priv->mii_ts.ts_info = lan8841_ts_info;
 
 	phydev->mii_ts = &ptp_priv->mii_ts;
Signature · 1 line
2.47.3
Reviewed-by on v2 2/4
#7 lore raw

Re: [PATCH net-next v2 2/4] net: phy: micrel: add HW timestamp configuration reporting

In reply to Vadim Fedorenko
Quoted 4 lines · On Sat, 29 Nov 2025 19:53:32 +0000 Vadim Fedorenko <[email protected]>…

The driver stores HW timestamping configuration and can technically report it. Add callback to do it.

Thank you!

Signature · 3 lines
Köry Maincent, Bootlin
Embedded Linux and kernel engineering
https://bootlin.com
#8 lore raw

PATCH v2 3/4 net: phy: microchip_rds_ptp: improve HW ts config logic

In reply to Vadim Fedorenko

The driver stores new HW timestamping configuration values unconditionally and may create inconsistency with what is actually configured in case of error. Improve the logic to store new values only once everything is configured.

---
 drivers/net/phy/microchip_rds_ptp.c | 21 ++++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)
drivers/net/phy/microchip_rds_ptp.c +16 -5
index 4c6326b0ceaf..6a7a0bb95301 100644
@@ -488,9 +488,6 @@ static int mchp_rds_ptp_hwtstamp_set(struct mii_timestamper *mii_ts,
 	unsigned long flags;
 	int rc;
 
-	clock->hwts_tx_type = config->tx_type;
-	clock->rx_filter = config->rx_filter;
-
 	switch (config->rx_filter) {
 	case HWTSTAMP_FILTER_NONE:
 		clock->layer = 0;
@@ -518,6 +515,15 @@ static int mchp_rds_ptp_hwtstamp_set(struct mii_timestamper *mii_ts,
 		return -ERANGE;
 	}
 
+	switch (config->rx_filter) {
+	case HWTSTAMP_TX_ONESTEP_SYNC:
+	case HWTSTAMP_TX_ON:
+	case HWTSTAMP_TX_OFF:
+		break;
+	default:
+		return -ERANGE;
+	}
+
 	/* Setup parsing of the frames and enable the timestamping for ptp
 	 * frames
 	 */
@@ -553,7 +559,7 @@ static int mchp_rds_ptp_hwtstamp_set(struct mii_timestamper *mii_ts,
 	if (rc < 0)
 		return rc;
 
-	if (clock->hwts_tx_type == HWTSTAMP_TX_ONESTEP_SYNC)
+	if (config->tx_type == HWTSTAMP_TX_ONESTEP_SYNC)
 		/* Enable / disable of the TX timestamp in the SYNC frames */
 		rc = mchp_rds_phy_modify_mmd(clock, MCHP_RDS_PTP_TX_MOD,
 					     MCHP_RDS_PTP_PORT,
@@ -587,8 +593,13 @@ static int mchp_rds_ptp_hwtstamp_set(struct mii_timestamper *mii_ts,
 	/* Now enable the timestamping interrupts */
 	rc = mchp_rds_ptp_config_intr(clock,
 				      config->rx_filter != HWTSTAMP_FILTER_NONE);
+	if (rc < 0)
+		return rc;
 
-	return rc < 0 ? rc : 0;
+	clock->hwts_tx_type = config->tx_type;
+	clock->rx_filter = config->rx_filter;
+
+	return 0;
 }
 
 static int mchp_rds_ptp_ts_info(struct mii_timestamper *mii_ts,
Signature · 1 line
2.47.3
maintainer Reviewed-by on v2 3/4
#9 lore raw

Re: [PATCH net-next v2 3/4] net: phy: microchip_rds_ptp: improve HW ts config logic

Quoted 6 lines · Vadim Fedorenko

The driver stores new HW timestamping configuration values unconditionally and may create inconsistency with what is actually configured in case of error. Improve the logic to store new values only once everything is configured.

Andrew

#10 lore raw

Re: [PATCH net-next v2 3/4] net: phy: microchip_rds_ptp: improve HW ts config logic

In reply to Vadim Fedorenko
Quoted 25 lines · On Sat, 29 Nov 2025 19:53:33 +0000 Vadim Fedorenko <[email protected]>…

The driver stores new HW timestamping configuration values unconditionally and may create inconsistency with what is actually configured in case of error. Improve the logic to store new values only once everything is configured.

---
 drivers/net/phy/microchip_rds_ptp.c | 21 ++++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)
(quoted diff) +0 -0

b/drivers/net/phy/microchip_rds_ptp.c index 4c6326b0ceaf..6a7a0bb95301 100644

drivers/net/phy/microchip_rds_ptp.c +0 -3
@@ -488,9 +488,6 @@ static int mchp_rds_ptp_hwtstamp_set(struct
 mii_timestamper *mii_ts, unsigned long flags;
 	int rc;
 
-	clock->hwts_tx_type = config->tx_type;
-	clock->rx_filter = config->rx_filter;
-
 	switch (config->rx_filter) {
 	case HWTSTAMP_FILTER_NONE:
 		clock->layer = 0;

drivers/net/phy/microchip_rds_ptp.c static int mchp_rds_ptp_hwtstamp_set(struct · line 518

Quoted 4 lines
drivers/net/phy/microchip_rds_ptp.c +1 -0
@@ -518,6 +515,15 @@ static int mchp_rds_ptp_hwtstamp_set(struct
 mii_timestamper *mii_ts, return -ERANGE;
 	}
 
+	switch (config->rx_filter) {

You want to check tx_type here not rx_filter.

drivers/net/phy/microchip_rds_ptp.c static int mchp_rds_ptp_hwtstamp_set(struct mii_timestamper *mii_ts, · line 526

Quoted 8 lines
drivers/net/phy/microchip_rds_ptp.c +8 -0
+	case HWTSTAMP_TX_ONESTEP_SYNC:
+	case HWTSTAMP_TX_ON:
+	case HWTSTAMP_TX_OFF:
+		break;
+	default:
+		return -ERANGE;
+	}
+

Regards,

Signature · 3 lines
Köry Maincent, Bootlin
Embedded Linux and kernel engineering
https://bootlin.com
#11 lore raw

Re: [PATCH net-next v2 3/4] net: phy: microchip_rds_ptp: improve HW ts config logic

Quoted 31 lines · Kory Maincent
Quoted 28 lines · On Sat, 29 Nov 2025 19:53:33 +0000 Vadim Fedorenko <[email protected]>…

The driver stores new HW timestamping configuration values unconditionally and may create inconsistency with what is actually configured in case of error. Improve the logic to store new values only once everything is configured.

---
  drivers/net/phy/microchip_rds_ptp.c | 21 ++++++++++++++++-----
  1 file changed, 16 insertions(+), 5 deletions(-)
(quoted diff) +0 -0

b/drivers/net/phy/microchip_rds_ptp.c index 4c6326b0ceaf..6a7a0bb95301 100644

drivers/net/phy/microchip_rds_ptp.c +0 -3
@@ -488,9 +488,6 @@ static int mchp_rds_ptp_hwtstamp_set(struct
 mii_timestamper *mii_ts, unsigned long flags;
  	int rc;
 
-	clock->hwts_tx_type = config->tx_type;
-	clock->rx_filter = config->rx_filter;
-
  	switch (config->rx_filter) {
  	case HWTSTAMP_FILTER_NONE:
  		clock->layer = 0;
@@ -518,6 +515,15 @@ static int mchp_rds_ptp_hwtstamp_set(struct
 mii_timestamper *mii_ts, return -ERANGE;
  	}
 
Quoted 3 lines

drivers/net/phy/microchip_rds_ptp.c static int mchp_rds_ptp_hwtstamp_set(struct mii_timestamper *mii_ts, · line 518

Quoted 1 line
drivers/net/phy/microchip_rds_ptp.c +1 -0
+	switch (config->rx_filter) {

You want to check tx_type here not rx_filter.

Damn it, copy-paste didn't work well :(

Quoted 11 lines

drivers/net/phy/microchip_rds_ptp.c static int mchp_rds_ptp_hwtstamp_set(struct mii_timestamper *mii_ts, · line 526

Quoted 9 lines
drivers/net/phy/microchip_rds_ptp.c +8 -0
+	case HWTSTAMP_TX_ONESTEP_SYNC:
+	case HWTSTAMP_TX_ON:
+	case HWTSTAMP_TX_OFF:
+		break;
+	default:
+		return -ERANGE;
+	}
+

Regards,

#12 lore raw

PATCH v2 4/4 net: phy: microchip_rds_ptp: add HW timestamp configuration reporting

In reply to Vadim Fedorenko

The driver stores HW timestamping configuration and can technically report it. Add callback to do it.

---
 drivers/net/phy/microchip_rds_ptp.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)
drivers/net/phy/microchip_rds_ptp.c +13 -0
index 6a7a0bb95301..ad066d66a467 100644
@@ -476,6 +476,18 @@ static bool mchp_rds_ptp_rxtstamp(struct mii_timestamper *mii_ts,
 	return true;
 }
 
+static int mchp_rds_ptp_hwtstamp_get(struct mii_timestamper *mii_ts,
+				     struct kernel_hwtstamp_config *config)
+{
+	struct mchp_rds_ptp_clock *clock =
+				container_of(mii_ts, struct mchp_rds_ptp_clock,
+					     mii_ts);
+	config->tx_type = clock->hwts_tx_type;
+	config->rx_filter = clock->rx_filter;
+
+	return 0;
+}
+
 static int mchp_rds_ptp_hwtstamp_set(struct mii_timestamper *mii_ts,
 				     struct kernel_hwtstamp_config *config,
 				     struct netlink_ext_ack *extack)
@@ -1293,6 +1305,7 @@ struct mchp_rds_ptp_clock *mchp_rds_ptp_probe(struct phy_device *phydev, u8 mmd,
 	clock->mii_ts.rxtstamp = mchp_rds_ptp_rxtstamp;
 	clock->mii_ts.txtstamp = mchp_rds_ptp_txtstamp;
 	clock->mii_ts.hwtstamp_set = mchp_rds_ptp_hwtstamp_set;
+	clock->mii_ts.hwtstamp_get = mchp_rds_ptp_hwtstamp_get;
 	clock->mii_ts.ts_info = mchp_rds_ptp_ts_info;
 
 	phydev->mii_ts = &clock->mii_ts;
Signature · 1 line
2.47.3