[PATCH] net/mlx5: relax headroom assertion

[PATCH] net/mlx5: relax headroom assertion

[Dmitry Kozlyuk]

A debug assertion in Single-Packet Receive Queue (SPRQ) mode
required all Rx mbufs to have a 128 byte headroom,
based on the assumption that rte_pktmbuf_init() sets it.
However, rte_pktmbuf_init() may set a smaller headroom
if the dataroom is insufficient, e.g. this is a natural case
for split buffer segments. The headroom can also be larger.
Only check the headroom size when vectored Rx routines
are used because they rely on it. Relax the assertion
to require sufficient headroom size, not an exact one.

Fixes: a0a45e8af723 ("net/mlx5: configure Rx queue for buffer split")
Cc: viacheslavo@nvidia.com
Cc: stable@dpdk.org

Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
---
 drivers/net/mlx5/mlx5_rxq.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index f77d42dedf..baf676d3b4 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -139,6 +139,7 @@ rxq_alloc_elts_sprq(struct mlx5_rxq_ctrl *rxq_ctrl)
 	unsigned int elts_n = mlx5_rxq_mprq_enabled(&rxq_ctrl->rxq) ?
 		(1 << rxq_ctrl->rxq.elts_n) * (1 << rxq_ctrl->rxq.strd_num_n) :
 		(1 << rxq_ctrl->rxq.elts_n);
+	bool has_vec_support = mlx5_rxq_check_vec_support(&rxq_ctrl->rxq) > 0;
 	unsigned int i;
 	int err;
 
@@ -160,8 +161,9 @@ rxq_alloc_elts_sprq(struct mlx5_rxq_ctrl *rxq_ctrl)
 			rte_errno = ENOMEM;
 			goto error;
 		}
-		/* Headroom is reserved by rte_pktmbuf_alloc(). */
-		MLX5_ASSERT(DATA_OFF(buf) == RTE_PKTMBUF_HEADROOM);
+		/* Only vectored Rx routines rely on headroom size. */
+		MLX5_ASSERT(!has_vec_support ||
+			    DATA_OFF(buf) >= RTE_PKTMBUF_HEADROOM);
 		/* Buffer is supposed to be empty. */
 		MLX5_ASSERT(rte_pktmbuf_data_len(buf) == 0);
 		MLX5_ASSERT(rte_pktmbuf_pkt_len(buf) == 0);
@@ -174,7 +176,7 @@ rxq_alloc_elts_sprq(struct mlx5_rxq_ctrl *rxq_ctrl)
 		(*rxq_ctrl->rxq.elts)[i] = buf;
 	}
 	/* If Rx vector is activated. */
-	if (mlx5_rxq_check_vec_support(&rxq_ctrl->rxq) > 0) {
+	if (has_vec_support) {
 		struct mlx5_rxq_data *rxq = &rxq_ctrl->rxq;
 		struct rte_mbuf *mbuf_init = &rxq->fake_mbuf;
 		struct rte_pktmbuf_pool_private *priv =
-- 
2.25.1

RE: [PATCH] net/mlx5: relax headroom assertion

[Raslan Darawsheh]

Hi,

> -----Original Message-----
> From: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
> Sent: Tuesday, December 28, 2021 11:21 AM
> To: dev@dpdk.org
> Cc: Slava Ovsiienko <viacheslavo@nvidia.com>; stable@dpdk.org; Matan
> Azrad <matan@nvidia.com>
> Subject: [PATCH] net/mlx5: relax headroom assertion
> 
> A debug assertion in Single-Packet Receive Queue (SPRQ) mode required all
> Rx mbufs to have a 128 byte headroom, based on the assumption that
> rte_pktmbuf_init() sets it.
> However, rte_pktmbuf_init() may set a smaller headroom if the dataroom is
> insufficient, e.g. this is a natural case for split buffer segments. The
> headroom can also be larger.
> Only check the headroom size when vectored Rx routines are used because
> they rely on it. Relax the assertion to require sufficient headroom size, not an
> exact one.
> 
> Fixes: a0a45e8af723 ("net/mlx5: configure Rx queue for buffer split")
> Cc: viacheslavo@nvidia.com
> Cc: stable@dpdk.org
> 
> Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
> Acked-by: Matan Azrad <matan@nvidia.com>

Patch rebased and applied to next-net-mlx,

Kindest regards,
Raslan Darawsheh