X-Git-Url: http://git.droids-corp.org/?a=blobdiff_plain;f=drivers%2Fnet%2Fmlx5%2Fmlx5_rxtx_vec_altivec.h;h=9778b0bbc1c01a37a39c7813141c91df315b6513;hb=a3147ae9aff9d7fd8644083d7d9f87ba9cabc524;hp=9e5c6ee7ab40497f5de7dd2c854e8646c7f262fc;hpb=bdb8e5b1ea7be5ec55b87c4090e712316701f175;p=dpdk.git diff --git a/drivers/net/mlx5/mlx5_rxtx_vec_altivec.h b/drivers/net/mlx5/mlx5_rxtx_vec_altivec.h index 9e5c6ee7ab..9778b0bbc1 100644 --- a/drivers/net/mlx5/mlx5_rxtx_vec_altivec.h +++ b/drivers/net/mlx5/mlx5_rxtx_vec_altivec.h @@ -6,7 +6,6 @@ #ifndef RTE_PMD_MLX5_RXTX_VEC_ALTIVEC_H_ #define RTE_PMD_MLX5_RXTX_VEC_ALTIVEC_H_ -#include #include #include #include @@ -17,13 +16,14 @@ #include #include +#include + +#include "mlx5_defs.h" #include "mlx5.h" #include "mlx5_utils.h" #include "mlx5_rxtx.h" #include "mlx5_rxtx_vec.h" #include "mlx5_autoconf.h" -#include "mlx5_defs.h" -#include "mlx5_prm.h" #ifndef __INTEL_COMPILER #pragma GCC diagnostic ignored "-Wcast-qual" @@ -155,8 +155,9 @@ rxq_cq_decompress_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, const vector unsigned long shmax = {64, 64}; #endif - if (!(pos & 0x7) && pos + 8 < mcqe_n) - rte_prefetch0((void *)(cq + pos + 8)); + for (i = 0; i < MLX5_VPMD_DESCS_PER_LOOP; ++i) + if (likely(pos + i < mcqe_n)) + rte_prefetch0((void *)(cq + pos + i)); /* A.1 load mCQEs into a 128bit register. */ mcqe1 = (vector unsigned char)vec_vsx_ld(0, @@ -263,6 +264,19 @@ rxq_cq_decompress_v(struct mlx5_rxq_data *rxq, volatile struct mlx5_cqe *cq, elts[pos + 2]->hash.fdir.hi = flow_tag; elts[pos + 3]->hash.fdir.hi = flow_tag; } + if (rte_flow_dynf_metadata_avail()) { + const uint32_t meta = *RTE_FLOW_DYNF_METADATA(t_pkt); + + /* Check if title packet has valid metadata. */ + if (meta) { + MLX5_ASSERT(t_pkt->ol_flags & + PKT_RX_DYNF_METADATA); + *RTE_FLOW_DYNF_METADATA(elts[pos]) = meta; + *RTE_FLOW_DYNF_METADATA(elts[pos + 1]) = meta; + *RTE_FLOW_DYNF_METADATA(elts[pos + 2]) = meta; + *RTE_FLOW_DYNF_METADATA(elts[pos + 3]) = meta; + } + } pos += MLX5_VPMD_DESCS_PER_LOOP; /* Move to next CQE and invalidate consumed CQEs. */ @@ -615,8 +629,8 @@ rxq_burst_v(struct mlx5_rxq_data *rxq, struct rte_mbuf **pkts, uint16_t pkts_n, const vector unsigned short cqe_sel_mask2 = (vector unsigned short){0, 0, 0xffff, 0, 0, 0, 0, 0}; - assert(rxq->sges_n == 0); - assert(rxq->cqe_n == rxq->elts_n); + MLX5_ASSERT(rxq->sges_n == 0); + MLX5_ASSERT(rxq->cqe_n == rxq->elts_n); cq = &(*rxq->cqes)[cq_idx]; rte_prefetch0(cq); rte_prefetch0(cq + 1); @@ -646,7 +660,7 @@ rxq_burst_v(struct mlx5_rxq_data *rxq, struct rte_mbuf **pkts, uint16_t pkts_n, if (!pkts_n) return rcvd_pkt; /* At this point, there shouldn't be any remaining packets. */ - assert(rxq->decompressed == 0); + MLX5_ASSERT(rxq->decompressed == 0); /* * A. load first Qword (8bytes) in one loop. @@ -732,13 +746,13 @@ rxq_burst_v(struct mlx5_rxq_data *rxq, struct rte_mbuf **pkts, uint16_t pkts_n, /* A.1 load cqes. */ p3 = (unsigned int)((vector unsigned short)p)[3]; cqes[3] = (vector unsigned char)(vector unsigned long){ - *(__attribute__((__aligned__(8))) unsigned long *) + *(__rte_aligned(8) unsigned long *) &cq[pos + p3].sop_drop_qpn, 0LL}; rte_compiler_barrier(); p2 = (unsigned int)((vector unsigned short)p)[2]; cqes[2] = (vector unsigned char)(vector unsigned long){ - *(__attribute__((__aligned__(8))) unsigned long *) + *(__rte_aligned(8) unsigned long *) &cq[pos + p2].sop_drop_qpn, 0LL}; rte_compiler_barrier(); @@ -751,12 +765,12 @@ rxq_burst_v(struct mlx5_rxq_data *rxq, struct rte_mbuf **pkts, uint16_t pkts_n, /* A.1 load a block having op_own. */ p1 = (unsigned int)((vector unsigned short)p)[1]; cqes[1] = (vector unsigned char)(vector unsigned long){ - *(__attribute__((__aligned__(8))) unsigned long *) + *(__rte_aligned(8) unsigned long *) &cq[pos + p1].sop_drop_qpn, 0LL}; rte_compiler_barrier(); cqes[0] = (vector unsigned char)(vector unsigned long){ - *(__attribute__((__aligned__(8))) unsigned long *) + *(__rte_aligned(8) unsigned long *) &cq[pos].sop_drop_qpn, 0LL}; rte_compiler_barrier(); @@ -783,10 +797,10 @@ rxq_burst_v(struct mlx5_rxq_data *rxq, struct rte_mbuf **pkts, uint16_t pkts_n, vec_sel((vector unsigned short)cqes[2], (vector unsigned short)cqe_tmp1, cqe_sel_mask1); cqe_tmp2 = (vector unsigned char)(vector unsigned long){ - *(__attribute__((__aligned__(8))) unsigned long *) + *(__rte_aligned(8) unsigned long *) &cq[pos + p3].rsvd3[9], 0LL}; cqe_tmp1 = (vector unsigned char)(vector unsigned long){ - *(__attribute__((__aligned__(8))) unsigned long *) + *(__rte_aligned(8) unsigned long *) &cq[pos + p2].rsvd3[9], 0LL}; cqes[3] = (vector unsigned char) vec_sel((vector unsigned short)cqes[3], @@ -846,10 +860,10 @@ rxq_burst_v(struct mlx5_rxq_data *rxq, struct rte_mbuf **pkts, uint16_t pkts_n, vec_sel((vector unsigned short)cqes[0], (vector unsigned short)cqe_tmp1, cqe_sel_mask1); cqe_tmp2 = (vector unsigned char)(vector unsigned long){ - *(__attribute__((__aligned__(8))) unsigned long *) + *(__rte_aligned(8) unsigned long *) &cq[pos + p1].rsvd3[9], 0LL}; cqe_tmp1 = (vector unsigned char)(vector unsigned long){ - *(__attribute__((__aligned__(8))) unsigned long *) + *(__rte_aligned(8) unsigned long *) &cq[pos].rsvd3[9], 0LL}; cqes[1] = (vector unsigned char) vec_sel((vector unsigned short)cqes[1], @@ -1062,7 +1076,7 @@ rxq_burst_v(struct mlx5_rxq_data *rxq, struct rte_mbuf **pkts, uint16_t pkts_n, if (unlikely(!nocmp_n && comp_idx == MLX5_VPMD_DESCS_PER_LOOP)) return rcvd_pkt; /* Update the consumer indexes for non-compressed CQEs. */ - assert(nocmp_n <= pkts_n); + MLX5_ASSERT(nocmp_n <= pkts_n); rxq->cq_ci += nocmp_n; rxq->rq_pi += nocmp_n; rcvd_pkt += nocmp_n; @@ -1072,7 +1086,7 @@ rxq_burst_v(struct mlx5_rxq_data *rxq, struct rte_mbuf **pkts, uint16_t pkts_n, #endif /* Decompress the last CQE if compressed. */ if (comp_idx < MLX5_VPMD_DESCS_PER_LOOP && comp_idx == n) { - assert(comp_idx == (nocmp_n % MLX5_VPMD_DESCS_PER_LOOP)); + MLX5_ASSERT(comp_idx == (nocmp_n % MLX5_VPMD_DESCS_PER_LOOP)); rxq->decompressed = rxq_cq_decompress_v(rxq, &cq[nocmp_n], &elts[nocmp_n]); /* Return more packets if needed. */