[PATCH v1] net/ice: fix incorrect reading of PHY timestamp

[PATCH v1] net/ice: fix incorrect reading of PHY timestamp

[Soumyadeep Hore]

In ICE PMD, previously the ready bitmap checking before reading
PHY timestamp was not present. This caused incorrect Tx
timestamping.

The ready bitmap checking is enabled and PHY timestamp is read once
the ready bitmap gives positive value.

Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
Cc: stable@dpdk.org

Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>
---
 drivers/net/ice/ice_ethdev.c | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
index 70298ac330..efc6db4346 100644
--- a/drivers/net/ice/ice_ethdev.c
+++ b/drivers/net/ice/ice_ethdev.c
@@ -6597,10 +6597,26 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
 	struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
 	struct ice_adapter *ad =
 			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
-	uint64_t ts_ns, tstamp;
+	uint64_t ts_ns, tstamp, tstamp_ready = 0;
+	uint64_t start_time, curr_time;
 	const uint64_t mask = 0xFFFFFFFF;
 	int ret;
 
+	start_time = rte_get_timer_cycles() / (rte_get_timer_hz() / 1000);
+
+	while (!(tstamp_ready & BIT_ULL(0))) {
+		ret = ice_get_phy_tx_tstamp_ready(hw, ad->ptp_tx_block, &tstamp_ready);
+		if (ret) {
+			PMD_DRV_LOG(ERR, "Failed to get phy ready for timestamp");
+			return -1;
+		}
+		curr_time = rte_get_timer_cycles() / (rte_get_timer_hz() / 1000);
+		if (curr_time - start_time > 1000) {
+			PMD_DRV_LOG(ERR, "Timeout to get phy ready for timestamp");
+			return -1;
+		}
+	}
+
 	ret = ice_read_phy_tstamp(hw, ad->ptp_tx_block, ad->ptp_tx_index, &tstamp);
 	if (ret || tstamp == 0) {
 		PMD_DRV_LOG(ERR, "Failed to read phy timestamp");
-- 
2.43.0

Re: [PATCH v1] net/ice: fix incorrect reading of PHY timestamp

[Singh, Aman Deep]

[-- Attachment #1: Type: text/plain, Size: 1907 bytes --]


On 25-10-2024 15:56, Soumyadeep Hore wrote:
> In ICE PMD, previously the ready bitmap checking before reading
> PHY timestamp was not present. This caused incorrect Tx
> timestamping.
>
> The ready bitmap checking is enabled and PHY timestamp is read once
> the ready bitmap gives positive value.
>
> Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
> Cc:stable@dpdk.org
>
> Signed-off-by: Soumyadeep Hore<soumyadeep.hore@intel.com>

Acked-by: Aman Singh<aman.deep.singh@intel.com>

> ---
>   drivers/net/ice/ice_ethdev.c | 18 +++++++++++++++++-
>   1 file changed, 17 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
> index 70298ac330..efc6db4346 100644
> --- a/drivers/net/ice/ice_ethdev.c
> +++ b/drivers/net/ice/ice_ethdev.c
> @@ -6597,10 +6597,26 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
>   	struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
>   	struct ice_adapter *ad =
>   			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
> -	uint64_t ts_ns, tstamp;
> +	uint64_t ts_ns, tstamp, tstamp_ready = 0;
> +	uint64_t start_time, curr_time;
>   	const uint64_t mask = 0xFFFFFFFF;
>   	int ret;
>   
> +	start_time = rte_get_timer_cycles() / (rte_get_timer_hz() / 1000);
> +
> +	while (!(tstamp_ready & BIT_ULL(0))) {
> +		ret = ice_get_phy_tx_tstamp_ready(hw, ad->ptp_tx_block, &tstamp_ready);
> +		if (ret) {
> +			PMD_DRV_LOG(ERR, "Failed to get phy ready for timestamp");
> +			return -1;
> +		}
> +		curr_time = rte_get_timer_cycles() / (rte_get_timer_hz() / 1000);
> +		if (curr_time - start_time > 1000) {
> +			PMD_DRV_LOG(ERR, "Timeout to get phy ready for timestamp");
> +			return -1;
> +		}
> +	}
> +
>   	ret = ice_read_phy_tstamp(hw, ad->ptp_tx_block, ad->ptp_tx_index, &tstamp);
>   	if (ret || tstamp == 0) {
>   		PMD_DRV_LOG(ERR, "Failed to read phy timestamp");

[-- Attachment #2: Type: text/html, Size: 2659 bytes --]

Re: [PATCH v1] net/ice: fix incorrect reading of PHY timestamp

[Bruce Richardson]

On Tue, Oct 29, 2024 at 04:07:45PM +0530, Singh, Aman Deep wrote:
>    On 25-10-2024 15:56, Soumyadeep Hore wrote:
> 
> In ICE PMD, previously the ready bitmap checking before reading
> PHY timestamp was not present. This caused incorrect Tx
> timestamping.
> 
> The ready bitmap checking is enabled and PHY timestamp is read once
> the ready bitmap gives positive value.
> 
> Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
> Cc: [1]stable@dpdk.org
> 
> Signed-off-by: Soumyadeep Hore [2]<soumyadeep.hore@intel.com>
> 
> Acked-by: Aman Singh [3]<aman.deep.singh@intel.com>
> 
Is this a v5 of this patch, or a resubmission of v1? There are 3 different
v1's of this patch, plus a v2, v3 and v4 listed in patchwork, see [1].
I had feedback on v3 of this patch, which I believe is still relevant to
this version.

Please keep all revisions of a patch in the one thread and with increasing
version number, saves us reviewers getting confused! :-) 
Also, with each new version, please include a summary below the cutline of
what has changed from previous, so we know what to look out for when
reviewing.

/Bruce

[1] https://patches.dpdk.org/project/dpdk/list/?series=&submitter=&state=*&q=fix+incorrect+reading+of+PHY&archive=&delegate=

[PATCH v2] net/ice: fix incorrect reading of PHY timestamp

[Soumyadeep Hore]

In ICE PMD, previously the ready bitmap checking before reading
PHY timestamp was not present. This caused incorrect Tx
timestamping.

The ready bitmap checking is enabled and PHY timestamp is read once
the ready bitmap gives positive value.

Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
Cc: stable@dpdk.org

Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>
---
v2:
- Addressed Bruce's comments
---
 drivers/net/ice/ice_ethdev.c | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
index 70298ac330..6d0d37b3a0 100644
--- a/drivers/net/ice/ice_ethdev.c
+++ b/drivers/net/ice/ice_ethdev.c
@@ -6597,10 +6597,26 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
 	struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
 	struct ice_adapter *ad =
 			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
-	uint64_t ts_ns, tstamp;
+	uint64_t ts_ns, tstamp, tstamp_ready = 0;
+	uint64_t end_time;
 	const uint64_t mask = 0xFFFFFFFF;
 	int ret;
 
+	end_time = rte_get_timer_cycles() + rte_get_timer_hz();
+
+	while (!(tstamp_ready & BIT_ULL(0))) {
+		ret = ice_get_phy_tx_tstamp_ready(hw, ad->ptp_tx_block, &tstamp_ready);
+		if (ret) {
+			PMD_DRV_LOG(ERR, "Failed to get phy ready for timestamp");
+			return -1;
+		}
+
+		if (rte_get_timer_cycles() > end_time) {
+			PMD_DRV_LOG(ERR, "Timeout to get phy ready for timestamp");
+			return -1;
+		}
+	}
+
 	ret = ice_read_phy_tstamp(hw, ad->ptp_tx_block, ad->ptp_tx_index, &tstamp);
 	if (ret || tstamp == 0) {
 		PMD_DRV_LOG(ERR, "Failed to read phy timestamp");
-- 
2.43.0

Re: [PATCH v2] net/ice: fix incorrect reading of PHY timestamp

[Bruce Richardson]

On Wed, Oct 30, 2024 at 02:16:11AM +0000, Soumyadeep Hore wrote:
> In ICE PMD, previously the ready bitmap checking before reading
> PHY timestamp was not present. This caused incorrect Tx
> timestamping.
> 
> The ready bitmap checking is enabled and PHY timestamp is read once
> the ready bitmap gives positive value.
> 
> Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>
> ---
> v2:
> - Addressed Bruce's comments
> ---
>  drivers/net/ice/ice_ethdev.c | 18 +++++++++++++++++-
>  1 file changed, 17 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
> index 70298ac330..6d0d37b3a0 100644
> --- a/drivers/net/ice/ice_ethdev.c
> +++ b/drivers/net/ice/ice_ethdev.c
> @@ -6597,10 +6597,26 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
>  	struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
>  	struct ice_adapter *ad =
>  			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
> -	uint64_t ts_ns, tstamp;
> +	uint64_t ts_ns, tstamp, tstamp_ready = 0;
> +	uint64_t end_time;
>  	const uint64_t mask = 0xFFFFFFFF;
>  	int ret;
>  
> +	end_time = rte_get_timer_cycles() + rte_get_timer_hz();
> +

Patch code itself looks fine, but don't you think that 1sec is an
excessively long time to wait? Maybe do a V3 with a lower timeout
threshold.

/Bruce

[PATCH v3] net/ice: fix incorrect reading of PHY timestamp

[Soumyadeep Hore]

In ICE PMD, previously the ready bitmap checking before reading
PHY timestamp was not present. This caused incorrect Tx
timestamping.

The ready bitmap checking is enabled and PHY timestamp is read once
the ready bitmap gives positive value.

Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
Cc: stable@dpdk.org

Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>
---
v3:
- Decreased the end time delay from 1 second to 10 microseconds
---
v2:
- Addressed Bruce's comments
---
 drivers/net/ice/ice_ethdev.c | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
index 70298ac330..3c768b6d0b 100644
--- a/drivers/net/ice/ice_ethdev.c
+++ b/drivers/net/ice/ice_ethdev.c
@@ -6597,10 +6597,27 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
 	struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
 	struct ice_adapter *ad =
 			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
-	uint64_t ts_ns, tstamp;
+	uint64_t ts_ns, tstamp, tstamp_ready = 0;
+	uint64_t end_time;
 	const uint64_t mask = 0xFFFFFFFF;
 	int ret;
 
+	/* Set the end time with a delay of 10 microseconds */
+	end_time = rte_get_timer_cycles() + (rte_get_timer_hz() / 100000);
+
+	while (!(tstamp_ready & BIT_ULL(0))) {
+		ret = ice_get_phy_tx_tstamp_ready(hw, ad->ptp_tx_block, &tstamp_ready);
+		if (ret) {
+			PMD_DRV_LOG(ERR, "Failed to get phy ready for timestamp");
+			return -1;
+		}
+
+		if (rte_get_timer_cycles() > end_time) {
+			PMD_DRV_LOG(ERR, "Timeout to get phy ready for timestamp");
+			return -1;
+		}
+	}
+
 	ret = ice_read_phy_tstamp(hw, ad->ptp_tx_block, ad->ptp_tx_index, &tstamp);
 	if (ret || tstamp == 0) {
 		PMD_DRV_LOG(ERR, "Failed to read phy timestamp");
-- 
2.43.0

Re: [PATCH v3] net/ice: fix incorrect reading of PHY timestamp

[Bruce Richardson]

On Mon, Nov 04, 2024 at 10:31:12AM +0000, Soumyadeep Hore wrote:
> In ICE PMD, previously the ready bitmap checking before reading
> PHY timestamp was not present. This caused incorrect Tx
> timestamping.
> 
> The ready bitmap checking is enabled and PHY timestamp is read once
> the ready bitmap gives positive value.
> 
> Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>
> ---
> v3:
> - Decreased the end time delay from 1 second to 10 microseconds
> ---
> v2:
> - Addressed Bruce's comments
> ---
>  drivers/net/ice/ice_ethdev.c | 19 ++++++++++++++++++-
>  1 file changed, 18 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
> index 70298ac330..3c768b6d0b 100644
> --- a/drivers/net/ice/ice_ethdev.c
> +++ b/drivers/net/ice/ice_ethdev.c
> @@ -6597,10 +6597,27 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
>  	struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
>  	struct ice_adapter *ad =
>  			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
> -	uint64_t ts_ns, tstamp;
> +	uint64_t ts_ns, tstamp, tstamp_ready = 0;
> +	uint64_t end_time;
>  	const uint64_t mask = 0xFFFFFFFF;
>  	int ret;
>  
> +	/* Set the end time with a delay of 10 microseconds */
> +	end_time = rte_get_timer_cycles() + (rte_get_timer_hz() / 100000);
> +
> +	while (!(tstamp_ready & BIT_ULL(0))) {

Nit: in DPDK, we recommend using "== 0" explicitly for numeric comparisons.
The "!" should only be used for boolean values, not ints.

> +		ret = ice_get_phy_tx_tstamp_ready(hw, ad->ptp_tx_block, &tstamp_ready);
> +		if (ret) {
> +			PMD_DRV_LOG(ERR, "Failed to get phy ready for timestamp");
> +			return -1;
> +		}
> +
> +		if (rte_get_timer_cycles() > end_time) {
> +			PMD_DRV_LOG(ERR, "Timeout to get phy ready for timestamp");
> +			return -1;
> +		}

Sorry for the last minute feedback here, but shouldn't these two conditions
be the other way around? Right now, if you call ice_get_phy_tx_tstamp_ready
just as the timer expires, and if the timestamp is actually ready this
time, you will exit with error instead of handling the timestamp.

Also, a very minor issue, but since you always want to go through the loop
at least once, a do { } while(0) might be a better construct. It would
avoid you having to initialize tstamp_ready to zero (not that it matters).

	do {
		if (rte_get_timer_cycles() > end_time) {
			/* log error and return */
		}

		if (ice_get_phy_tx_tstamp_ready(...) != 0) {
			/* error log and return */
		}
	} while((tstamp_ready & BIT_ULL(0)) == 0);

> +	}
> +
>  	ret = ice_read_phy_tstamp(hw, ad->ptp_tx_block, ad->ptp_tx_index, &tstamp);
>  	if (ret || tstamp == 0) {
>  		PMD_DRV_LOG(ERR, "Failed to read phy timestamp");
> -- 

Regards,
/Bruce

[PATCH v4] net/ice: fix incorrect reading of PHY timestamp

[Soumyadeep Hore]

In ICE PMD, previously the ready bitmap checking before reading
PHY timestamp was not present. This caused incorrect Tx
timestamping.

The ready bitmap checking is enabled and PHY timestamp is read once
the ready bitmap gives positive value.

Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
Cc: stable@dpdk.org

Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>
---
v4:
- Addressed Bruce comments for do while loop introduction
---
v3:
- Decreased the end time delay from 1 second to 10 microseconds
---
v2:
- Addressed Bruce's comments
---
 drivers/net/ice/ice_ethdev.c | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
index 70298ac330..7e2c6107ff 100644
--- a/drivers/net/ice/ice_ethdev.c
+++ b/drivers/net/ice/ice_ethdev.c
@@ -6597,10 +6597,27 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
 	struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
 	struct ice_adapter *ad =
 			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
-	uint64_t ts_ns, tstamp;
+	uint64_t ts_ns, tstamp, tstamp_ready = 0;
+	uint64_t end_time;
 	const uint64_t mask = 0xFFFFFFFF;
 	int ret;
 
+	/* Set the end time with a delay of 10 microseconds */
+	end_time = rte_get_timer_cycles() + (rte_get_timer_hz() / 100000);
+
+	do {
+		ret = ice_get_phy_tx_tstamp_ready(hw, ad->ptp_tx_block, &tstamp_ready);
+		if (ret) {
+			PMD_DRV_LOG(ERR, "Failed to get phy ready for timestamp");
+			return -1;
+		}
+
+		if (rte_get_timer_cycles() > end_time) {
+			PMD_DRV_LOG(ERR, "Timeout to get phy ready for timestamp");
+			return -1;
+		}
+	} while ((tstamp_ready & BIT_ULL(0)) == 0);
+
 	ret = ice_read_phy_tstamp(hw, ad->ptp_tx_block, ad->ptp_tx_index, &tstamp);
 	if (ret || tstamp == 0) {
 		PMD_DRV_LOG(ERR, "Failed to read phy timestamp");
-- 
2.43.0

Re: [PATCH v4] net/ice: fix incorrect reading of PHY timestamp

[Bruce Richardson]

On Tue, Nov 05, 2024 at 10:14:19AM +0000, Soumyadeep Hore wrote:
> In ICE PMD, previously the ready bitmap checking before reading
> PHY timestamp was not present. This caused incorrect Tx
> timestamping.
> 
> The ready bitmap checking is enabled and PHY timestamp is read once
> the ready bitmap gives positive value.
> 
> Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>
> ---
> v4:
> - Addressed Bruce comments for do while loop introduction
> ---
> v3:
> - Decreased the end time delay from 1 second to 10 microseconds
> ---
> v2:
> - Addressed Bruce's comments
> ---
>  drivers/net/ice/ice_ethdev.c | 19 ++++++++++++++++++-
>  1 file changed, 18 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
> index 70298ac330..7e2c6107ff 100644
> --- a/drivers/net/ice/ice_ethdev.c
> +++ b/drivers/net/ice/ice_ethdev.c
> @@ -6597,10 +6597,27 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
>  	struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
>  	struct ice_adapter *ad =
>  			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
> -	uint64_t ts_ns, tstamp;
> +	uint64_t ts_ns, tstamp, tstamp_ready = 0;
> +	uint64_t end_time;
>  	const uint64_t mask = 0xFFFFFFFF;
>  	int ret;
>  
> +	/* Set the end time with a delay of 10 microseconds */
> +	end_time = rte_get_timer_cycles() + (rte_get_timer_hz() / 100000);
> +
> +	do {
> +		ret = ice_get_phy_tx_tstamp_ready(hw, ad->ptp_tx_block, &tstamp_ready);
> +		if (ret) {
> +			PMD_DRV_LOG(ERR, "Failed to get phy ready for timestamp");
> +			return -1;
> +		}
> +
> +		if (rte_get_timer_cycles() > end_time) {
> +			PMD_DRV_LOG(ERR, "Timeout to get phy ready for timestamp");
> +			return -1;
> +		}
> +	} while ((tstamp_ready & BIT_ULL(0)) == 0);
> +

Unfortunately, this is still not quite right. The important change in my
last feedback was not the introduction of the do-while, but the switching
of the two blocks inside the while loop, or the adding of an additional
check to the timer expiry. You need to check that the bit is
still not set before checking for timer expiry. When "get_timer_cycles() >
end_time", it may also be the case that (tstamp_ready & BIT_ULL(0)) != 0, in
which case you want to continue rather than returning error.

/Bruce

[PATCH v5] net/ice: fix incorrect reading of PHY timestamp

[Soumyadeep Hore]

In ICE PMD, previously the ready bitmap checking before reading
PHY timestamp was not present. This caused incorrect Tx
timestamping.

The ready bitmap checking is enabled and PHY timestamp is read once
the ready bitmap gives positive value.

Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
Cc: stable@dpdk.org

Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>
---
v5:
- Addressed edge case as specified in Bruce's comment
---
v4:
- Addressed Bruce comments for do while loop introduction
---
v3:
- Decreased the end time delay from 1 second to 10 microseconds
---
v2:
- Addressed Bruce's comments
---
 drivers/net/ice/ice_ethdev.c | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
index 70298ac330..4a7fb2b656 100644
--- a/drivers/net/ice/ice_ethdev.c
+++ b/drivers/net/ice/ice_ethdev.c
@@ -6597,10 +6597,27 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
 	struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
 	struct ice_adapter *ad =
 			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
-	uint64_t ts_ns, tstamp;
+	uint64_t ts_ns, tstamp, tstamp_ready = 0;
+	uint64_t end_time;
 	const uint64_t mask = 0xFFFFFFFF;
 	int ret;
 
+	/* Set the end time with a delay of 10 microseconds */
+	end_time = rte_get_timer_cycles() + (rte_get_timer_hz() / 100000);
+
+	do {
+		ret = ice_get_phy_tx_tstamp_ready(hw, ad->ptp_tx_block, &tstamp_ready);
+		if (ret) {
+			PMD_DRV_LOG(ERR, "Failed to get phy ready for timestamp");
+			return -1;
+		}
+
+		if ((tstamp_ready & BIT_ULL(0)) == 0 && rte_get_timer_cycles() > end_time) {
+			PMD_DRV_LOG(ERR, "Timeout to get phy ready for timestamp");
+			return -1;
+		}
+	} while ((tstamp_ready & BIT_ULL(0)) == 0);
+
 	ret = ice_read_phy_tstamp(hw, ad->ptp_tx_block, ad->ptp_tx_index, &tstamp);
 	if (ret || tstamp == 0) {
 		PMD_DRV_LOG(ERR, "Failed to read phy timestamp");
-- 
2.43.0

Re: [PATCH v5] net/ice: fix incorrect reading of PHY timestamp

[Bruce Richardson]

On Wed, Nov 06, 2024 at 08:22:29AM +0000, Soumyadeep Hore wrote:
> In ICE PMD, previously the ready bitmap checking before reading
> PHY timestamp was not present. This caused incorrect Tx
> timestamping.
> 
> The ready bitmap checking is enabled and PHY timestamp is read once
> the ready bitmap gives positive value.
> 
> Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>

Reviewed-by: Bruce Richardson <bruce.richardson@intel.com>

> ---
> v5:
> - Addressed edge case as specified in Bruce's comment
> ---
> v4:
> - Addressed Bruce comments for do while loop introduction
> ---
> v3:
> - Decreased the end time delay from 1 second to 10 microseconds
> ---
> v2:
> - Addressed Bruce's comments
> ---
>  drivers/net/ice/ice_ethdev.c | 19 ++++++++++++++++++-
>  1 file changed, 18 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
> index 70298ac330..4a7fb2b656 100644
> --- a/drivers/net/ice/ice_ethdev.c
> +++ b/drivers/net/ice/ice_ethdev.c
> @@ -6597,10 +6597,27 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
>  	struct ice_hw *hw = ICE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
>  	struct ice_adapter *ad =
>  			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
> -	uint64_t ts_ns, tstamp;
> +	uint64_t ts_ns, tstamp, tstamp_ready = 0;
> +	uint64_t end_time;
>  	const uint64_t mask = 0xFFFFFFFF;
>  	int ret;
>  
> +	/* Set the end time with a delay of 10 microseconds */
> +	end_time = rte_get_timer_cycles() + (rte_get_timer_hz() / 100000);
> +
> +	do {
> +		ret = ice_get_phy_tx_tstamp_ready(hw, ad->ptp_tx_block, &tstamp_ready);
> +		if (ret) {
> +			PMD_DRV_LOG(ERR, "Failed to get phy ready for timestamp");
> +			return -1;
> +		}
> +
> +		if ((tstamp_ready & BIT_ULL(0)) == 0 && rte_get_timer_cycles() > end_time) {
> +			PMD_DRV_LOG(ERR, "Timeout to get phy ready for timestamp");
> +			return -1;
> +		}
> +	} while ((tstamp_ready & BIT_ULL(0)) == 0);
> +
>  	ret = ice_read_phy_tstamp(hw, ad->ptp_tx_block, ad->ptp_tx_index, &tstamp);
>  	if (ret || tstamp == 0) {
>  		PMD_DRV_LOG(ERR, "Failed to read phy timestamp");
> -- 
> 2.43.0
> 

Re: [PATCH v5] net/ice: fix incorrect reading of PHY timestamp

[Bruce Richardson]

On Wed, Nov 06, 2024 at 10:59:47AM +0000, Bruce Richardson wrote:
> On Wed, Nov 06, 2024 at 08:22:29AM +0000, Soumyadeep Hore wrote:
> > In ICE PMD, previously the ready bitmap checking before reading
> > PHY timestamp was not present. This caused incorrect Tx
> > timestamping.
> > 
> > The ready bitmap checking is enabled and PHY timestamp is read once
> > the ready bitmap gives positive value.
> > 
> > Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
> > Cc: stable@dpdk.org
> > 
> > Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>
> 
> Reviewed-by: Bruce Richardson <bruce.richardson@intel.com>
> 
Applied to dpdk-next-net-intel.

Thanks,
/Bruce

Re: [PATCH v3] net/ice: fix incorrect reading of PHY timestamp

[Bruce Richardson]

On Fri, Oct 25, 2024 at 07:32:11AM +0000, Soumyadeep Hore wrote:
> In ICE PMD, previously the ready bitmap checking before reading
> PHY timestamp was not present. This caused incorrect Tx
> timestamping.
> 
> The ready bitmap checking is enabled and PHY timestamp is read once
> the ready bitmap gives positive value.
> 
> Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>
> ---
>  drivers/net/ice/ice_ethdev.c | 18 +++++++++++++++++-
>  1 file changed, 17 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
> index 7b1bd163a2..e0db47cf28 100644
> --- a/drivers/net/ice/ice_ethdev.c
> +++ b/drivers/net/ice/ice_ethdev.c
> @@ -6517,12 +6517,28 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
>  	struct ice_adapter *ad =
>  			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
>  	uint8_t lport;
> -	uint64_t ts_ns, ns, tstamp;
> +	uint64_t ts_ns, ns, tstamp, tstamp_ready = 0;
> +	uint64_t start_time, curr_time;
>  	const uint64_t mask = 0xFFFFFFFF;
>  	int ret;
>  
>  	lport = hw->port_info->lport;
>  
> +	start_time = rte_get_timer_cycles() / (rte_get_timer_hz() / 1000);

Why all the division by 1000?

> +
> +	while (!(tstamp_ready & BIT_ULL(0))) {
> +		ret = ice_get_phy_tx_tstamp_ready(hw, lport, &tstamp_ready);
> +		if (ret) {
> +			PMD_DRV_LOG(ERR, "Failed to get phy ready for timestamp");
> +			return -1;
> +		}
> +		curr_time = rte_get_timer_cycles() / (rte_get_timer_hz() / 1000);
> +		if (curr_time - start_time > 1000) {

Is 1 second not a very long time to wait for this? Surely even milliseconds
is a very long delay in this case.

In terms of the logic, rather than constantly comparing vs the start time
and doing lots of division, I think it would be simpler to just set a max
end-time. For example, to keep the current 1s limit:

	uint64_t end_time = rte_get_timer_cycles?() + rte_get_timer_hz()
	...
	if (rte_get_timer_cycles() > end_time) {
	     ...
	}

> +			PMD_DRV_LOG(ERR, "Timeout to get phy ready for timestamp");
> +			return -1;
> +		}
> +	}
> +
>  	ret = ice_read_phy_tstamp(hw, lport, 0, &tstamp);
>  	if (ret) {
>  		PMD_DRV_LOG(ERR, "Failed to read phy timestamp");
> -- 
> 2.43.0
> 

[PATCH v3] net/ice: fix incorrect reading of PHY timestamp

[Soumyadeep Hore]

In ICE PMD, previously the ready bitmap checking before reading
PHY timestamp was not present. This caused incorrect Tx
timestamping.

The ready bitmap checking is enabled and PHY timestamp is read once
the ready bitmap gives positive value.

Fixes: 881169950d80 ("net/ice/base: implement initial PTP support for E830")
Cc: stable@dpdk.org

Signed-off-by: Soumyadeep Hore <soumyadeep.hore@intel.com>
---
 drivers/net/ice/ice_ethdev.c | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ice/ice_ethdev.c b/drivers/net/ice/ice_ethdev.c
index 7b1bd163a2..e0db47cf28 100644
--- a/drivers/net/ice/ice_ethdev.c
+++ b/drivers/net/ice/ice_ethdev.c
@@ -6517,12 +6517,28 @@ ice_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
 	struct ice_adapter *ad =
 			ICE_DEV_PRIVATE_TO_ADAPTER(dev->data->dev_private);
 	uint8_t lport;
-	uint64_t ts_ns, ns, tstamp;
+	uint64_t ts_ns, ns, tstamp, tstamp_ready = 0;
+	uint64_t start_time, curr_time;
 	const uint64_t mask = 0xFFFFFFFF;
 	int ret;
 
 	lport = hw->port_info->lport;
 
+	start_time = rte_get_timer_cycles() / (rte_get_timer_hz() / 1000);
+
+	while (!(tstamp_ready & BIT_ULL(0))) {
+		ret = ice_get_phy_tx_tstamp_ready(hw, lport, &tstamp_ready);
+		if (ret) {
+			PMD_DRV_LOG(ERR, "Failed to get phy ready for timestamp");
+			return -1;
+		}
+		curr_time = rte_get_timer_cycles() / (rte_get_timer_hz() / 1000);
+		if (curr_time - start_time > 1000) {
+			PMD_DRV_LOG(ERR, "Timeout to get phy ready for timestamp");
+			return -1;
+		}
+	}
+
 	ret = ice_read_phy_tstamp(hw, lport, 0, &tstamp);
 	if (ret) {
 		PMD_DRV_LOG(ERR, "Failed to read phy timestamp");
-- 
2.43.0