net/mlx5: fix typo in vectorized Rx comments

[ upstream commit 6e695b0 ]

Change "returing" to "returning".

Fixes: 2e542da ("net/mlx5: add Altivec Rx")
Fixes: 570acdb ("net/mlx5: add vectorized Rx/Tx burst for ARM")
Fixes: 3c2ddbd ("net/mlx5: separate shareable vector functions")

Signed-off-by: Sarosh Arif <sarosh.arif@emumba.com>

drivers/net/mlx5/mlx5_rxtx_vec_altivec.h

@@ -840,7 +840,7 @@ rxq_cq_process_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq,
 
 	/*
 	 * A. load first Qword (8bytes) in one loop.
-	 * B. copy 4 mbuf pointers from elts ring to returing pkts.
+	 * B. copy 4 mbuf pointers from elts ring to returning pkts.
 	 * C. load remaining CQE data and extract necessary fields.
 	 *    Final 16bytes cqes[] extracted from original 64bytes CQE has the
 	 *    following structure:

drivers/net/mlx5/mlx5_rxtx_vec_neon.h

@@ -593,7 +593,7 @@ rxq_cq_process_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq,
 	 * there's no instruction to count trailing zeros. __builtin_clzl() is
 	 * used instead.
 	 *
-	 * A. copy 4 mbuf pointers from elts ring to returing pkts.
+	 * A. copy 4 mbuf pointers from elts ring to returning pkts.
 	 * B. load 64B CQE and extract necessary fields
 	 *    Final 16bytes cqes[] extracted from original 64bytes CQE has the
 	 *    following structure:

drivers/net/mlx5/mlx5_rxtx_vec_sse.h

@@ -565,7 +565,7 @@ rxq_cq_process_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq,
 	const __m128i flow_mark_adj = _mm_set_epi32(rxq->mark * (-1), 0, 0, 0);
 	/*
 	 * A. load first Qword (8bytes) in one loop.
-	 * B. copy 4 mbuf pointers from elts ring to returing pkts.
+	 * B. copy 4 mbuf pointers from elts ring to returning pkts.
 	 * C. load remained CQE data and extract necessary fields.
 	 *    Final 16bytes cqes[] extracted from original 64bytes CQE has the
 	 *    following structure: