[v2] net/vmxnet3: support per-queue stats for fewer queues

Message ID 20241025092734.1222248-1-mb@smartsharesystems.com (mailing list archive)
State Changes Requested, archived
Delegated to: Ferruh Yigit
Headers
Series [v2] net/vmxnet3: support per-queue stats for fewer queues |

Checks

Context Check Description
ci/checkpatch warning coding style issues
ci/loongarch-compilation success Compilation OK
ci/loongarch-unit-testing success Unit Testing PASS
ci/Intel-compilation success Compilation OK
ci/intel-Testing success Testing PASS
ci/github-robot: build success github build: passed
ci/intel-Functional success Functional PASS
ci/iol-mellanox-Performance success Performance Testing PASS
ci/iol-marvell-Functional success Functional Testing PASS
ci/iol-intel-Functional success Functional Testing PASS
ci/iol-broadcom-Performance success Performance Testing PASS
ci/iol-compile-arm64-testing success Testing PASS
ci/iol-unit-arm64-testing pending Testing pending
ci/iol-unit-amd64-testing fail Testing issues
ci/iol-compile-amd64-testing success Testing PASS
ci/iol-sample-apps-testing success Testing PASS

Commit Message

Morten Brørup Oct. 25, 2024, 9:27 a.m. UTC
Removed the requirement that the configured number of queues to provide
statistics for (RTE_ETHDEV_QUEUE_STAT_CNTRS) cannot be less than the
driver's max number of supported transmit queues (VMXNET3_MAX_TX_QUEUES).

Also improved support for virtual hardware version 6.

Signed-off-by: Morten Brørup <mb@smartsharesystems.com>
---
v2:
* Virtual hardware version 6 supports more queues; updated some arrays
  accordingly.
* Added support for larger MTU with virtual hardware version 6.
---
 drivers/net/vmxnet3/vmxnet3_ethdev.c | 34 +++++++++++++++++-----------
 drivers/net/vmxnet3/vmxnet3_ethdev.h |  4 ++--
 2 files changed, 23 insertions(+), 15 deletions(-)
  

Comments

Morten Brørup Oct. 27, 2024, 8:47 a.m. UTC | #1
Jochen, Ronak, (Roger,)

Please review the below patch, so it can go into the 24.11 LTS release.

> From: Morten Brørup [mailto:mb@smartsharesystems.com]
> Sent: Friday, 25 October 2024 11.28
> 
> Removed the requirement that the configured number of queues to provide
> statistics for (RTE_ETHDEV_QUEUE_STAT_CNTRS) cannot be less than the
> driver's max number of supported transmit queues
> (VMXNET3_MAX_TX_QUEUES).
> 
> Also improved support for virtual hardware version 6.
> 
> Signed-off-by: Morten Brørup <mb@smartsharesystems.com>
> ---
> v2:
> * Virtual hardware version 6 supports more queues; updated some arrays
>   accordingly.
> * Added support for larger MTU with virtual hardware version 6.
> ---
>  drivers/net/vmxnet3/vmxnet3_ethdev.c | 34 +++++++++++++++++-----------
>  drivers/net/vmxnet3/vmxnet3_ethdev.h |  4 ++--
>  2 files changed, 23 insertions(+), 15 deletions(-)
> 
> diff --git a/drivers/net/vmxnet3/vmxnet3_ethdev.c
> b/drivers/net/vmxnet3/vmxnet3_ethdev.c
> index 78fac63ab6..1752c58069 100644
> --- a/drivers/net/vmxnet3/vmxnet3_ethdev.c
> +++ b/drivers/net/vmxnet3/vmxnet3_ethdev.c
> @@ -1470,42 +1470,52 @@ vmxnet3_dev_stats_get(struct rte_eth_dev *dev,
> struct rte_eth_stats *stats)
>  	struct vmxnet3_hw *hw = dev->data->dev_private;
>  	struct UPT1_TxStats txStats;
>  	struct UPT1_RxStats rxStats;
> +	uint64_t packets, bytes;
> 
>  	VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD,
> VMXNET3_CMD_GET_STATS);
> 
>  	for (i = 0; i < hw->num_tx_queues; i++) {
>  		vmxnet3_tx_stats_get(hw, i, &txStats);
> 
> -		stats->q_opackets[i] = txStats.ucastPktsTxOK +
> +		packets = txStats.ucastPktsTxOK +
>  			txStats.mcastPktsTxOK +
>  			txStats.bcastPktsTxOK;
> 
> -		stats->q_obytes[i] = txStats.ucastBytesTxOK +
> +		bytes = txStats.ucastBytesTxOK +
>  			txStats.mcastBytesTxOK +
>  			txStats.bcastBytesTxOK;
> 
> -		stats->opackets += stats->q_opackets[i];
> -		stats->obytes += stats->q_obytes[i];
> +		stats->opackets += packets;
> +		stats->obytes += bytes;
>  		stats->oerrors += txStats.pktsTxError +
> txStats.pktsTxDiscard;
> +
> +		if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
> +			stats->q_opackets[i] = packets;
> +			stats->q_obytes[i] = bytes;
> +		}
>  	}
> 
>  	for (i = 0; i < hw->num_rx_queues; i++) {
>  		vmxnet3_rx_stats_get(hw, i, &rxStats);
> 
> -		stats->q_ipackets[i] = rxStats.ucastPktsRxOK +
> +		packets = rxStats.ucastPktsRxOK +
>  			rxStats.mcastPktsRxOK +
>  			rxStats.bcastPktsRxOK;
> 
> -		stats->q_ibytes[i] = rxStats.ucastBytesRxOK +
> +		bytes = rxStats.ucastBytesRxOK +
>  			rxStats.mcastBytesRxOK +
>  			rxStats.bcastBytesRxOK;
> 
> -		stats->ipackets += stats->q_ipackets[i];
> -		stats->ibytes += stats->q_ibytes[i];
> -
> -		stats->q_errors[i] = rxStats.pktsRxError;
> +		stats->ipackets += packets;
> +		stats->ibytes += bytes;
>  		stats->ierrors += rxStats.pktsRxError;
>  		stats->imissed += rxStats.pktsRxOutOfBuf;
> +
> +		if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
> +			stats->q_ipackets[i] = packets;
> +			stats->q_ibytes[i] = bytes;
> +			stats->q_errors[i] = rxStats.pktsRxError;
> +		}
>  	}
> 
>  	return 0;
> @@ -1521,8 +1531,6 @@ vmxnet3_dev_stats_reset(struct rte_eth_dev *dev)
> 
>  	VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD,
> VMXNET3_CMD_GET_STATS);
> 
> -	RTE_BUILD_BUG_ON(RTE_ETHDEV_QUEUE_STAT_CNTRS <
> VMXNET3_MAX_TX_QUEUES);
> -
>  	for (i = 0; i < hw->num_tx_queues; i++) {
>  		vmxnet3_hw_tx_stats_get(hw, i, &txStats);
>  		memcpy(&hw->snapshot_tx_stats[i], &txStats,
> @@ -1566,7 +1574,7 @@ vmxnet3_dev_info_get(struct rte_eth_dev *dev,
>  	dev_info->min_rx_bufsize = 1518 + RTE_PKTMBUF_HEADROOM;
>  	dev_info->max_rx_pktlen = 16384; /* includes CRC, cf MAXFRS
> register */
>  	dev_info->min_mtu = VMXNET3_MIN_MTU;
> -	dev_info->max_mtu = VMXNET3_MAX_MTU;
> +	dev_info->max_mtu = VMXNET3_VERSION_GE_6(hw) ? VMXNET3_V6_MAX_MTU
> : VMXNET3_MAX_MTU;
>  	dev_info->speed_capa = RTE_ETH_LINK_SPEED_10G;
>  	dev_info->max_mac_addrs = VMXNET3_MAX_MAC_ADDRS;
> 
> diff --git a/drivers/net/vmxnet3/vmxnet3_ethdev.h
> b/drivers/net/vmxnet3/vmxnet3_ethdev.h
> index 2b3e2c4caa..e9ded6663d 100644
> --- a/drivers/net/vmxnet3/vmxnet3_ethdev.h
> +++ b/drivers/net/vmxnet3/vmxnet3_ethdev.h
> @@ -121,8 +121,8 @@ struct vmxnet3_hw {
>  #define VMXNET3_VFT_TABLE_SIZE     (VMXNET3_VFT_SIZE *
> sizeof(uint32_t))
>  	UPT1_TxStats
> saved_tx_stats[VMXNET3_EXT_MAX_TX_QUEUES];
>  	UPT1_RxStats
> saved_rx_stats[VMXNET3_EXT_MAX_RX_QUEUES];
> -	UPT1_TxStats          snapshot_tx_stats[VMXNET3_MAX_TX_QUEUES];
> -	UPT1_RxStats          snapshot_rx_stats[VMXNET3_MAX_RX_QUEUES];
> +	UPT1_TxStats
> snapshot_tx_stats[VMXNET3_EXT_MAX_TX_QUEUES];
> +	UPT1_RxStats
> snapshot_rx_stats[VMXNET3_EXT_MAX_RX_QUEUES];
>  	uint16_t              tx_prod_offset;
>  	uint16_t              rx_prod_offset[2];
>  	/* device capability bit map */
> --
> 2.43.0
  
Ferruh Yigit Nov. 1, 2024, 12:28 a.m. UTC | #2
On 10/25/2024 10:27 AM, Morten Brørup wrote:
> Removed the requirement that the configured number of queues to provide
> statistics for (RTE_ETHDEV_QUEUE_STAT_CNTRS) cannot be less than the
> driver's max number of supported transmit queues (VMXNET3_MAX_TX_QUEUES).
> 
> Also improved support for virtual hardware version 6.
> 
> Signed-off-by: Morten Brørup <mb@smartsharesystems.com>
> ---
> v2:
> * Virtual hardware version 6 supports more queues; updated some arrays
>   accordingly.
> * Added support for larger MTU with virtual hardware version 6.
> ---
>  drivers/net/vmxnet3/vmxnet3_ethdev.c | 34 +++++++++++++++++-----------
>  drivers/net/vmxnet3/vmxnet3_ethdev.h |  4 ++--
>  2 files changed, 23 insertions(+), 15 deletions(-)
> 
> diff --git a/drivers/net/vmxnet3/vmxnet3_ethdev.c b/drivers/net/vmxnet3/vmxnet3_ethdev.c
> index 78fac63ab6..1752c58069 100644
> --- a/drivers/net/vmxnet3/vmxnet3_ethdev.c
> +++ b/drivers/net/vmxnet3/vmxnet3_ethdev.c
> @@ -1470,42 +1470,52 @@ vmxnet3_dev_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
>  	struct vmxnet3_hw *hw = dev->data->dev_private;
>  	struct UPT1_TxStats txStats;
>  	struct UPT1_RxStats rxStats;
> +	uint64_t packets, bytes;
>  
>  	VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD, VMXNET3_CMD_GET_STATS);
>  
>  	for (i = 0; i < hw->num_tx_queues; i++) {
>  		vmxnet3_tx_stats_get(hw, i, &txStats);
>  
> -		stats->q_opackets[i] = txStats.ucastPktsTxOK +
> +		packets = txStats.ucastPktsTxOK +
>  			txStats.mcastPktsTxOK +
>  			txStats.bcastPktsTxOK;
>  
> -		stats->q_obytes[i] = txStats.ucastBytesTxOK +
> +		bytes = txStats.ucastBytesTxOK +
>  			txStats.mcastBytesTxOK +
>  			txStats.bcastBytesTxOK;
>  
> -		stats->opackets += stats->q_opackets[i];
> -		stats->obytes += stats->q_obytes[i];
> +		stats->opackets += packets;
> +		stats->obytes += bytes;
>  		stats->oerrors += txStats.pktsTxError + txStats.pktsTxDiscard;
> +
> +		if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
> +			stats->q_opackets[i] = packets;
> +			stats->q_obytes[i] = bytes;
> +		}
>  	}
>  
>  	for (i = 0; i < hw->num_rx_queues; i++) {
>  		vmxnet3_rx_stats_get(hw, i, &rxStats);
>  
> -		stats->q_ipackets[i] = rxStats.ucastPktsRxOK +
> +		packets = rxStats.ucastPktsRxOK +
>  			rxStats.mcastPktsRxOK +
>  			rxStats.bcastPktsRxOK;
>  
> -		stats->q_ibytes[i] = rxStats.ucastBytesRxOK +
> +		bytes = rxStats.ucastBytesRxOK +
>  			rxStats.mcastBytesRxOK +
>  			rxStats.bcastBytesRxOK;
>  
> -		stats->ipackets += stats->q_ipackets[i];
> -		stats->ibytes += stats->q_ibytes[i];
> -
> -		stats->q_errors[i] = rxStats.pktsRxError;
> +		stats->ipackets += packets;
> +		stats->ibytes += bytes;
>  		stats->ierrors += rxStats.pktsRxError;
>  		stats->imissed += rxStats.pktsRxOutOfBuf;
> +
> +		if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
> +			stats->q_ipackets[i] = packets;
> +			stats->q_ibytes[i] = bytes;
> +			stats->q_errors[i] = rxStats.pktsRxError;
> +		}
>  	}
>

Ack, this is also a fix, since queue size can be up to
'VMXNET3_EXT_MAX_RX_QUEUES' (32) for version 6, even if
'RTE_ETHDEV_QUEUE_STAT_CNTRS > VMXNET3_MAX_TX_QUEUES',
'RTE_ETHDEV_QUEUE_STAT_CNTRS ' still can be smaller than
'VMXNET3_EXT_MAX_RX_QUEUES', which will cause accessing out of boundary
to queue stats array.


>  
>  	return 0;
> @@ -1521,8 +1531,6 @@ vmxnet3_dev_stats_reset(struct rte_eth_dev *dev)
>  
>  	VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD, VMXNET3_CMD_GET_STATS);
>  
> -	RTE_BUILD_BUG_ON(RTE_ETHDEV_QUEUE_STAT_CNTRS < VMXNET3_MAX_TX_QUEUES);
> -
>  	for (i = 0; i < hw->num_tx_queues; i++) {
>  		vmxnet3_hw_tx_stats_get(hw, i, &txStats);
>  		memcpy(&hw->snapshot_tx_stats[i], &txStats,
> @@ -1566,7 +1574,7 @@ vmxnet3_dev_info_get(struct rte_eth_dev *dev,
>  	dev_info->min_rx_bufsize = 1518 + RTE_PKTMBUF_HEADROOM;
>  	dev_info->max_rx_pktlen = 16384; /* includes CRC, cf MAXFRS register */
>  	dev_info->min_mtu = VMXNET3_MIN_MTU;
> -	dev_info->max_mtu = VMXNET3_MAX_MTU;
> +	dev_info->max_mtu = VMXNET3_VERSION_GE_6(hw) ? VMXNET3_V6_MAX_MTU : VMXNET3_MAX_MTU;
>

Hi Morten,

The MTU update is unrelated with the main purpose of this patch, what do
you think to separate above to its own function.

>  	dev_info->speed_capa = RTE_ETH_LINK_SPEED_10G;
>  	dev_info->max_mac_addrs = VMXNET3_MAX_MAC_ADDRS;
>  
> diff --git a/drivers/net/vmxnet3/vmxnet3_ethdev.h b/drivers/net/vmxnet3/vmxnet3_ethdev.h
> index 2b3e2c4caa..e9ded6663d 100644
> --- a/drivers/net/vmxnet3/vmxnet3_ethdev.h
> +++ b/drivers/net/vmxnet3/vmxnet3_ethdev.h
> @@ -121,8 +121,8 @@ struct vmxnet3_hw {
>  #define VMXNET3_VFT_TABLE_SIZE     (VMXNET3_VFT_SIZE * sizeof(uint32_t))
>  	UPT1_TxStats	      saved_tx_stats[VMXNET3_EXT_MAX_TX_QUEUES];
>  	UPT1_RxStats	      saved_rx_stats[VMXNET3_EXT_MAX_RX_QUEUES];
> -	UPT1_TxStats          snapshot_tx_stats[VMXNET3_MAX_TX_QUEUES];
> -	UPT1_RxStats          snapshot_rx_stats[VMXNET3_MAX_RX_QUEUES];
> +	UPT1_TxStats          snapshot_tx_stats[VMXNET3_EXT_MAX_TX_QUEUES];
> +	UPT1_RxStats          snapshot_rx_stats[VMXNET3_EXT_MAX_RX_QUEUES];
>

Ack, probably above needs to be fixed anyway, even with
"VMXNET3_MAX_TX_QUEUES < RTE_ETHDEV_QUEUE_STAT_CNTRS" restriction (as
RTE_ETHDEV_QUEUE_STAT_CNTRS can be configured to be a bigger value
during build).
  

Patch

diff --git a/drivers/net/vmxnet3/vmxnet3_ethdev.c b/drivers/net/vmxnet3/vmxnet3_ethdev.c
index 78fac63ab6..1752c58069 100644
--- a/drivers/net/vmxnet3/vmxnet3_ethdev.c
+++ b/drivers/net/vmxnet3/vmxnet3_ethdev.c
@@ -1470,42 +1470,52 @@  vmxnet3_dev_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
 	struct vmxnet3_hw *hw = dev->data->dev_private;
 	struct UPT1_TxStats txStats;
 	struct UPT1_RxStats rxStats;
+	uint64_t packets, bytes;
 
 	VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD, VMXNET3_CMD_GET_STATS);
 
 	for (i = 0; i < hw->num_tx_queues; i++) {
 		vmxnet3_tx_stats_get(hw, i, &txStats);
 
-		stats->q_opackets[i] = txStats.ucastPktsTxOK +
+		packets = txStats.ucastPktsTxOK +
 			txStats.mcastPktsTxOK +
 			txStats.bcastPktsTxOK;
 
-		stats->q_obytes[i] = txStats.ucastBytesTxOK +
+		bytes = txStats.ucastBytesTxOK +
 			txStats.mcastBytesTxOK +
 			txStats.bcastBytesTxOK;
 
-		stats->opackets += stats->q_opackets[i];
-		stats->obytes += stats->q_obytes[i];
+		stats->opackets += packets;
+		stats->obytes += bytes;
 		stats->oerrors += txStats.pktsTxError + txStats.pktsTxDiscard;
+
+		if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
+			stats->q_opackets[i] = packets;
+			stats->q_obytes[i] = bytes;
+		}
 	}
 
 	for (i = 0; i < hw->num_rx_queues; i++) {
 		vmxnet3_rx_stats_get(hw, i, &rxStats);
 
-		stats->q_ipackets[i] = rxStats.ucastPktsRxOK +
+		packets = rxStats.ucastPktsRxOK +
 			rxStats.mcastPktsRxOK +
 			rxStats.bcastPktsRxOK;
 
-		stats->q_ibytes[i] = rxStats.ucastBytesRxOK +
+		bytes = rxStats.ucastBytesRxOK +
 			rxStats.mcastBytesRxOK +
 			rxStats.bcastBytesRxOK;
 
-		stats->ipackets += stats->q_ipackets[i];
-		stats->ibytes += stats->q_ibytes[i];
-
-		stats->q_errors[i] = rxStats.pktsRxError;
+		stats->ipackets += packets;
+		stats->ibytes += bytes;
 		stats->ierrors += rxStats.pktsRxError;
 		stats->imissed += rxStats.pktsRxOutOfBuf;
+
+		if (i < RTE_ETHDEV_QUEUE_STAT_CNTRS) {
+			stats->q_ipackets[i] = packets;
+			stats->q_ibytes[i] = bytes;
+			stats->q_errors[i] = rxStats.pktsRxError;
+		}
 	}
 
 	return 0;
@@ -1521,8 +1531,6 @@  vmxnet3_dev_stats_reset(struct rte_eth_dev *dev)
 
 	VMXNET3_WRITE_BAR1_REG(hw, VMXNET3_REG_CMD, VMXNET3_CMD_GET_STATS);
 
-	RTE_BUILD_BUG_ON(RTE_ETHDEV_QUEUE_STAT_CNTRS < VMXNET3_MAX_TX_QUEUES);
-
 	for (i = 0; i < hw->num_tx_queues; i++) {
 		vmxnet3_hw_tx_stats_get(hw, i, &txStats);
 		memcpy(&hw->snapshot_tx_stats[i], &txStats,
@@ -1566,7 +1574,7 @@  vmxnet3_dev_info_get(struct rte_eth_dev *dev,
 	dev_info->min_rx_bufsize = 1518 + RTE_PKTMBUF_HEADROOM;
 	dev_info->max_rx_pktlen = 16384; /* includes CRC, cf MAXFRS register */
 	dev_info->min_mtu = VMXNET3_MIN_MTU;
-	dev_info->max_mtu = VMXNET3_MAX_MTU;
+	dev_info->max_mtu = VMXNET3_VERSION_GE_6(hw) ? VMXNET3_V6_MAX_MTU : VMXNET3_MAX_MTU;
 	dev_info->speed_capa = RTE_ETH_LINK_SPEED_10G;
 	dev_info->max_mac_addrs = VMXNET3_MAX_MAC_ADDRS;
 
diff --git a/drivers/net/vmxnet3/vmxnet3_ethdev.h b/drivers/net/vmxnet3/vmxnet3_ethdev.h
index 2b3e2c4caa..e9ded6663d 100644
--- a/drivers/net/vmxnet3/vmxnet3_ethdev.h
+++ b/drivers/net/vmxnet3/vmxnet3_ethdev.h
@@ -121,8 +121,8 @@  struct vmxnet3_hw {
 #define VMXNET3_VFT_TABLE_SIZE     (VMXNET3_VFT_SIZE * sizeof(uint32_t))
 	UPT1_TxStats	      saved_tx_stats[VMXNET3_EXT_MAX_TX_QUEUES];
 	UPT1_RxStats	      saved_rx_stats[VMXNET3_EXT_MAX_RX_QUEUES];
-	UPT1_TxStats          snapshot_tx_stats[VMXNET3_MAX_TX_QUEUES];
-	UPT1_RxStats          snapshot_rx_stats[VMXNET3_MAX_RX_QUEUES];
+	UPT1_TxStats          snapshot_tx_stats[VMXNET3_EXT_MAX_TX_QUEUES];
+	UPT1_RxStats          snapshot_rx_stats[VMXNET3_EXT_MAX_RX_QUEUES];
 	uint16_t              tx_prod_offset;
 	uint16_t              rx_prod_offset[2];
 	/* device capability bit map */