* [PATCH] net/mlx5: fix MPRQ stride size to accommodate the headroom
@ 2023-05-31 18:43 Alexander Kozyrev
2023-05-31 19:00 ` [PATCH v2] " Alexander Kozyrev
0 siblings, 1 reply; 4+ messages in thread
From: Alexander Kozyrev @ 2023-05-31 18:43 UTC (permalink / raw)
To: dev; +Cc: stable, rasland, viacheslavo, matan, michaelba
The space for the headroom is reserved at the end of every MPRQ stride
for the next packet. The Rx burst logic is to copy any overlapping
packet data if there is an overlap with this reserved headroom space.
But it is not possible if the headroom size is bigger than the whole
stride. Adjust the stride size to make sure the stride size is greater
than the headroom size.
Fixes: 34776af600df ("net/mlx5: fix MPRQ stride devargs adjustment")
Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
---
drivers/net/mlx5/mlx5.c | 1 +
drivers/net/mlx5/mlx5_rxq.c | 41 +++++++++++++++++++++++++------------
2 files changed, 29 insertions(+), 13 deletions(-)
diff --git a/drivers/net/mlx5/mlx5.c b/drivers/net/mlx5/mlx5.c
index a75fa1b7f0..592e4053c5 100644
--- a/drivers/net/mlx5/mlx5.c
+++ b/drivers/net/mlx5/mlx5.c
@@ -2597,6 +2597,7 @@ mlx5_port_args_config(struct mlx5_priv *priv, struct mlx5_kvargs_ctrl *mkvlist,
config->mprq.max_memcpy_len = MLX5_MPRQ_MEMCPY_DEFAULT_LEN;
config->mprq.min_rxqs_num = MLX5_MPRQ_MIN_RXQS;
config->mprq.log_stride_num = MLX5_MPRQ_DEFAULT_LOG_STRIDE_NUM;
+ config->mprq.log_stride_size = MLX5_MPRQ_DEFAULT_LOG_STRIDE_SIZE;
config->log_hp_size = MLX5_ARG_UNSET;
config->std_delay_drop = 0;
config->hp_delay_drop = 0;
diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index ad8fd13cbe..da31b9a106 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -1604,23 +1604,38 @@ mlx5_mprq_prepare(struct rte_eth_dev *dev, uint16_t idx, uint16_t desc,
} else {
*actual_log_stride_num = config->mprq.log_stride_num;
}
- if (config->mprq.log_stride_size) {
- /* Checks if chosen size of stride is in supported range. */
- if (config->mprq.log_stride_size > log_max_stride_size ||
- config->mprq.log_stride_size < log_min_stride_size) {
- *actual_log_stride_size = log_def_stride_size;
+ /* Checks if chosen size of stride is in supported range. */
+ if (config->mprq.log_stride_size > log_max_stride_size ||
+ config->mprq.log_stride_size < log_min_stride_size) {
+ *actual_log_stride_size = log_def_stride_size;
+ DRV_LOG(WARNING,
+ "Port %u Rx queue %u size of a stride for Multi-Packet RQ is out of range, setting default value (%u)",
+ dev->data->port_id, idx,
+ RTE_BIT32(log_def_stride_size));
+ } else {
+ *actual_log_stride_size = config->mprq.log_stride_size;
+ }
+ /* Make the stride fit the mbuf size by default. */
+ if (*actual_log_stride_size == MLX5_MPRQ_DEFAULT_LOG_STRIDE_SIZE) {
+ if (min_mbuf_size <= RTE_BIT32(log_max_stride_size)) {
DRV_LOG(WARNING,
- "Port %u Rx queue %u size of a stride for Multi-Packet RQ is out of range, setting default value (%u)",
- dev->data->port_id, idx,
- RTE_BIT32(log_def_stride_size));
+ "Port %u Rx queue %u size of a stride for Multi-Packet RQ is adjusted to match the mbuf size (%u)",
+ dev->data->port_id, idx, min_mbuf_size);
+ *actual_log_stride_size = log2above(min_mbuf_size);
} else {
- *actual_log_stride_size = config->mprq.log_stride_size;
+ goto unsupport;
}
- } else {
- if (min_mbuf_size <= RTE_BIT32(log_max_stride_size))
- *actual_log_stride_size = log2above(min_mbuf_size);
- else
+ }
+ /* Make sure the stride size is greater than the headroom. */
+ if (RTE_BIT32(*actual_log_stride_size) < RTE_PKTMBUF_HEADROOM) {
+ if (RTE_BIT32(log_max_stride_size) > RTE_PKTMBUF_HEADROOM) {
+ DRV_LOG(WARNING,
+ "Port %u Rx queue %u size of a stride for Multi-Packet RQ is adjusted to accomodate the headroom (%u)",
+ dev->data->port_id, idx, RTE_PKTMBUF_HEADROOM);
+ *actual_log_stride_size = log2above(RTE_PKTMBUF_HEADROOM);
+ } else {
goto unsupport;
+ }
}
log_stride_wqe_size = *actual_log_stride_num + *actual_log_stride_size;
/* Check if WQE buffer size is supported by hardware. */
--
2.18.2
^ permalink raw reply [flat|nested] 4+ messages in thread* [PATCH v2] net/mlx5: fix MPRQ stride size to accommodate the headroom
2023-05-31 18:43 [PATCH] net/mlx5: fix MPRQ stride size to accommodate the headroom Alexander Kozyrev
@ 2023-05-31 19:00 ` Alexander Kozyrev
2023-07-18 9:12 ` Slava Ovsiienko
2023-07-18 15:02 ` Raslan Darawsheh
0 siblings, 2 replies; 4+ messages in thread
From: Alexander Kozyrev @ 2023-05-31 19:00 UTC (permalink / raw)
To: dev; +Cc: stable, rasland, viacheslavo, matan, michaelba
The space for the headroom is reserved at the end of every MPRQ stride
for the next packet. The Rx burst logic is to copy any overlapping
packet data if there is an overlap with this reserved headroom space.
But it is not possible if the headroom size is bigger than the whole
stride. Adjust the stride size to make sure the stride size is greater
than the headroom size.
Fixes: 34776af600df ("net/mlx5: fix MPRQ stride devargs adjustment")
Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
---
drivers/net/mlx5/mlx5.c | 1 +
drivers/net/mlx5/mlx5_rxq.c | 41 +++++++++++++++++++++++++------------
2 files changed, 29 insertions(+), 13 deletions(-)
diff --git a/drivers/net/mlx5/mlx5.c b/drivers/net/mlx5/mlx5.c
index a75fa1b7f0..592e4053c5 100644
--- a/drivers/net/mlx5/mlx5.c
+++ b/drivers/net/mlx5/mlx5.c
@@ -2597,6 +2597,7 @@ mlx5_port_args_config(struct mlx5_priv *priv, struct mlx5_kvargs_ctrl *mkvlist,
config->mprq.max_memcpy_len = MLX5_MPRQ_MEMCPY_DEFAULT_LEN;
config->mprq.min_rxqs_num = MLX5_MPRQ_MIN_RXQS;
config->mprq.log_stride_num = MLX5_MPRQ_DEFAULT_LOG_STRIDE_NUM;
+ config->mprq.log_stride_size = MLX5_MPRQ_DEFAULT_LOG_STRIDE_SIZE;
config->log_hp_size = MLX5_ARG_UNSET;
config->std_delay_drop = 0;
config->hp_delay_drop = 0;
diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index ad8fd13cbe..20e03146a5 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -1604,23 +1604,38 @@ mlx5_mprq_prepare(struct rte_eth_dev *dev, uint16_t idx, uint16_t desc,
} else {
*actual_log_stride_num = config->mprq.log_stride_num;
}
- if (config->mprq.log_stride_size) {
- /* Checks if chosen size of stride is in supported range. */
- if (config->mprq.log_stride_size > log_max_stride_size ||
- config->mprq.log_stride_size < log_min_stride_size) {
- *actual_log_stride_size = log_def_stride_size;
+ /* Checks if chosen size of stride is in supported range. */
+ if (config->mprq.log_stride_size > log_max_stride_size ||
+ config->mprq.log_stride_size < log_min_stride_size) {
+ *actual_log_stride_size = log_def_stride_size;
+ DRV_LOG(WARNING,
+ "Port %u Rx queue %u size of a stride for Multi-Packet RQ is out of range, setting default value (%u)",
+ dev->data->port_id, idx,
+ RTE_BIT32(log_def_stride_size));
+ } else {
+ *actual_log_stride_size = config->mprq.log_stride_size;
+ }
+ /* Make the stride fit the mbuf size by default. */
+ if (*actual_log_stride_size == MLX5_MPRQ_DEFAULT_LOG_STRIDE_SIZE) {
+ if (min_mbuf_size <= RTE_BIT32(log_max_stride_size)) {
DRV_LOG(WARNING,
- "Port %u Rx queue %u size of a stride for Multi-Packet RQ is out of range, setting default value (%u)",
- dev->data->port_id, idx,
- RTE_BIT32(log_def_stride_size));
+ "Port %u Rx queue %u size of a stride for Multi-Packet RQ is adjusted to match the mbuf size (%u)",
+ dev->data->port_id, idx, min_mbuf_size);
+ *actual_log_stride_size = log2above(min_mbuf_size);
} else {
- *actual_log_stride_size = config->mprq.log_stride_size;
+ goto unsupport;
}
- } else {
- if (min_mbuf_size <= RTE_BIT32(log_max_stride_size))
- *actual_log_stride_size = log2above(min_mbuf_size);
- else
+ }
+ /* Make sure the stride size is greater than the headroom. */
+ if (RTE_BIT32(*actual_log_stride_size) < RTE_PKTMBUF_HEADROOM) {
+ if (RTE_BIT32(log_max_stride_size) > RTE_PKTMBUF_HEADROOM) {
+ DRV_LOG(WARNING,
+ "Port %u Rx queue %u size of a stride for Multi-Packet RQ is adjusted to accommodate the headroom (%u)",
+ dev->data->port_id, idx, RTE_PKTMBUF_HEADROOM);
+ *actual_log_stride_size = log2above(RTE_PKTMBUF_HEADROOM);
+ } else {
goto unsupport;
+ }
}
log_stride_wqe_size = *actual_log_stride_num + *actual_log_stride_size;
/* Check if WQE buffer size is supported by hardware. */
--
2.18.2
^ permalink raw reply [flat|nested] 4+ messages in thread* RE: [PATCH v2] net/mlx5: fix MPRQ stride size to accommodate the headroom
2023-05-31 19:00 ` [PATCH v2] " Alexander Kozyrev
@ 2023-07-18 9:12 ` Slava Ovsiienko
2023-07-18 15:02 ` Raslan Darawsheh
1 sibling, 0 replies; 4+ messages in thread
From: Slava Ovsiienko @ 2023-07-18 9:12 UTC (permalink / raw)
To: Alexander Kozyrev, dev
Cc: stable, Raslan Darawsheh, Matan Azrad, Michael Baum
> -----Original Message-----
> From: Alexander Kozyrev <akozyrev@nvidia.com>
> Sent: Wednesday, May 31, 2023 10:00 PM
> To: dev@dpdk.org
> Cc: stable@dpdk.org; Raslan Darawsheh <rasland@nvidia.com>; Slava
> Ovsiienko <viacheslavo@nvidia.com>; Matan Azrad <matan@nvidia.com>;
> Michael Baum <michaelba@nvidia.com>
> Subject: [PATCH v2] net/mlx5: fix MPRQ stride size to accommodate the
> headroom
>
> The space for the headroom is reserved at the end of every MPRQ stride for
> the next packet. The Rx burst logic is to copy any overlapping packet data if
> there is an overlap with this reserved headroom space.
> But it is not possible if the headroom size is bigger than the whole stride.
> Adjust the stride size to make sure the stride size is greater than the
> headroom size.
>
> Fixes: 34776af600df ("net/mlx5: fix MPRQ stride devargs adjustment")
>
> Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
^ permalink raw reply [flat|nested] 4+ messages in thread* RE: [PATCH v2] net/mlx5: fix MPRQ stride size to accommodate the headroom
2023-05-31 19:00 ` [PATCH v2] " Alexander Kozyrev
2023-07-18 9:12 ` Slava Ovsiienko
@ 2023-07-18 15:02 ` Raslan Darawsheh
1 sibling, 0 replies; 4+ messages in thread
From: Raslan Darawsheh @ 2023-07-18 15:02 UTC (permalink / raw)
To: Alexander Kozyrev, dev; +Cc: stable, Slava Ovsiienko, Matan Azrad, Michael Baum
Hi,
> -----Original Message-----
> From: Alexander Kozyrev <akozyrev@nvidia.com>
> Sent: Wednesday, May 31, 2023 10:00 PM
> To: dev@dpdk.org
> Cc: stable@dpdk.org; Raslan Darawsheh <rasland@nvidia.com>; Slava
> Ovsiienko <viacheslavo@nvidia.com>; Matan Azrad <matan@nvidia.com>;
> Michael Baum <michaelba@nvidia.com>
> Subject: [PATCH v2] net/mlx5: fix MPRQ stride size to accommodate the
> headroom
>
> The space for the headroom is reserved at the end of every MPRQ stride for
> the next packet. The Rx burst logic is to copy any overlapping packet data if
> there is an overlap with this reserved headroom space.
> But it is not possible if the headroom size is bigger than the whole stride.
> Adjust the stride size to make sure the stride size is greater than the headroom
> size.
>
> Fixes: 34776af600df ("net/mlx5: fix MPRQ stride devargs adjustment")
>
> Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
> ---
Patch applied to next-net-mlx,
Kindest regards,
Raslan Darawsheh
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2023-07-18 15:02 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-05-31 18:43 [PATCH] net/mlx5: fix MPRQ stride size to accommodate the headroom Alexander Kozyrev
2023-05-31 19:00 ` [PATCH v2] " Alexander Kozyrev
2023-07-18 9:12 ` Slava Ovsiienko
2023-07-18 15:02 ` Raslan Darawsheh
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).