P7 devices support the newer Rx completion version. This Rx completion though similar to the previous generation, provides some extra information for flow offload scenarios apart from the normal information.
Signed-off-by: Ajit Khaparde <ajit.khapa...@broadcom.com>
---
drivers/net/bnxt/bnxt_rxr.c | 87 ++++++++++++++++++++++++++++++++++-
drivers/net/bnxt/bnxt_rxr.h | 92 +++++++++++++++++++++++++++++++++++++
2 files changed, 177 insertions(+), 2 deletions(-)
diff --git a/drivers/net/bnxt/bnxt_rxr.c b/drivers/net/bnxt/bnxt_rxr.c
index 9d45065f28..59ea0121de 100644
--- a/drivers/net/bnxt/bnxt_rxr.c
+++ b/drivers/net/bnxt/bnxt_rxr.c
@@ -553,6 +553,41 @@ bnxt_parse_pkt_type(struct rx_pkt_cmpl *rxcmp, struct
rx_pkt_cmpl_hi *rxcmp1)
return bnxt_ptype_table[index];
}
+static void
+bnxt_parse_pkt_type_v3(struct rte_mbuf *mbuf,
+ struct rx_pkt_cmpl *rxcmp_v1,
+ struct rx_pkt_cmpl_hi *rxcmp1_v1)
+{
+ uint32_t flags_type, flags2, meta;
+ struct rx_pkt_v3_cmpl_hi *rxcmp1;
+ struct rx_pkt_v3_cmpl *rxcmp;
+ uint8_t index;
+
+ rxcmp = (void *)rxcmp_v1;
+ rxcmp1 = (void *)rxcmp1_v1;
+
+ flags_type = rte_le_to_cpu_16(rxcmp->flags_type);
+ flags2 = rte_le_to_cpu_32(rxcmp1->flags2);
+ meta = rte_le_to_cpu_32(rxcmp->metadata1_payload_offset);
+
+ /* TODO */
+ /* Validate ptype table indexing at build time. */
+ /* bnxt_check_ptype_constants_v3(); */
+
+ /*
+ * Index format:
+ * bit 0: Set if IP tunnel encapsulated packet.
+ * bit 1: Set if IPv6 packet, clear if IPv4.
+ * bit 2: Set if VLAN tag present.
+ * bits 3-6: Four-bit hardware packet type field.
+ */
+ index = BNXT_CMPL_V3_ITYPE_TO_IDX(flags_type) |
+ BNXT_CMPL_V3_VLAN_TO_IDX(meta) |
+ BNXT_CMPL_V3_IP_VER_TO_IDX(flags2);
+
+ mbuf->packet_type = bnxt_ptype_table[index];
+}
+
static void __rte_cold
bnxt_init_ol_flags_tables(struct bnxt_rx_queue *rxq)
{
@@ -716,6 +751,43 @@ bnxt_get_rx_ts_p5(struct bnxt *bp, uint32_t rx_ts_cmpl)
ptp->rx_timestamp = pkt_time;
}
+static uint32_t
+bnxt_ulp_set_mark_in_mbuf_v3(struct bnxt *bp, struct rx_pkt_cmpl_hi *rxcmp1,
+ struct rte_mbuf *mbuf, uint32_t *vfr_flag)
+{
+ struct rx_pkt_v3_cmpl_hi *rxcmp1_v3 = (void *)rxcmp1;
+ uint32_t flags2, meta, mark_id = 0;
+ /* revisit the usage of gfid/lfid if mark action is supported.
+ * for now, only VFR is using mark and the metadata is the SVIF
+ * (a small number)
+ */
+ bool gfid = false;
+ int rc = 0;
+
+ flags2 = rte_le_to_cpu_32(rxcmp1_v3->flags2);
+
+ switch (flags2 & RX_PKT_V3_CMPL_HI_FLAGS2_META_FORMAT_MASK) {
+ case RX_PKT_V3_CMPL_HI_FLAGS2_META_FORMAT_CHDR_DATA:
+ /* Only supporting Metadata for ulp now */
+ meta = rxcmp1_v3->metadata2;
+ break;
+ default:
+ goto skip_mark;
+ }
+
+ rc = ulp_mark_db_mark_get(bp->ulp_ctx, gfid, meta, vfr_flag, &mark_id);
+ if (!rc) {
+ /* Only supporting VFR for now, no Mark actions */
+ if (vfr_flag && *vfr_flag)
+ return mark_id;
+ }
+
+skip_mark:
+ mbuf->hash.fdir.hi = 0;
+
+ return 0;
+}
+
static uint32_t
bnxt_ulp_set_mark_in_mbuf(struct bnxt *bp, struct rx_pkt_cmpl_hi *rxcmp1,
struct rte_mbuf *mbuf, uint32_t *vfr_flag)
@@ -892,7 +964,8 @@ static int bnxt_rx_pkt(struct rte_mbuf **rx_pkt,
*rx_pkt = mbuf;
goto next_rx;
} else if ((cmp_type != CMPL_BASE_TYPE_RX_L2) &&
- (cmp_type != CMPL_BASE_TYPE_RX_L2_V2)) {
+ (cmp_type != CMPL_BASE_TYPE_RX_L2_V2) &&
+ (cmp_type != CMPL_BASE_TYPE_RX_L2_V3)) {
rc = -EINVAL;
goto next_rx;
}
@@ -929,6 +1002,16 @@ static int bnxt_rx_pkt(struct rte_mbuf **rx_pkt,
bp->ptp_all_rx_tstamp)
bnxt_get_rx_ts_p5(rxq->bp, rxcmp1->reorder);
+ if (cmp_type == CMPL_BASE_TYPE_RX_L2_V3) {
+ bnxt_parse_csum_v3(mbuf, rxcmp1);
+ bnxt_parse_pkt_type_v3(mbuf, rxcmp, rxcmp1);
+ bnxt_rx_vlan_v3(mbuf, rxcmp, rxcmp1);
+ if (BNXT_TRUFLOW_EN(bp))
+ mark_id = bnxt_ulp_set_mark_in_mbuf_v3(rxq->bp, rxcmp1,
+ mbuf, &vfr_flag);
+ goto reuse_rx_mbuf;
+ }
+
if (cmp_type == CMPL_BASE_TYPE_RX_L2_V2) {
bnxt_parse_csum_v2(mbuf, rxcmp1);
bnxt_parse_pkt_type_v2(mbuf, rxcmp, rxcmp1);
@@ -1066,7 +1149,7 @@ uint16_t bnxt_recv_pkts(void *rx_queue, struct rte_mbuf
**rx_pkts,
if (CMP_TYPE(rxcmp) == CMPL_BASE_TYPE_HWRM_DONE) {
PMD_DRV_LOG(ERR, "Rx flush done\n");
} else if ((CMP_TYPE(rxcmp) >= CMPL_BASE_TYPE_RX_TPA_START_V2)
&&
- (CMP_TYPE(rxcmp) <= RX_TPA_V2_ABUF_CMPL_TYPE_RX_TPA_AGG)) {
+ (CMP_TYPE(rxcmp) <= CMPL_BASE_TYPE_RX_TPA_START_V3))
{
rc = bnxt_rx_pkt(&rx_pkts[nb_rx_pkts], rxq, &raw_cons);
if (!rc)
nb_rx_pkts++;
diff --git a/drivers/net/bnxt/bnxt_rxr.h b/drivers/net/bnxt/bnxt_rxr.h
index af53bc0c25..439d29a07f 100644
--- a/drivers/net/bnxt/bnxt_rxr.h
+++ b/drivers/net/bnxt/bnxt_rxr.h
@@ -386,4 +386,96 @@ bnxt_parse_pkt_type_v2(struct rte_mbuf *mbuf,
mbuf->packet_type = pkt_type;
}
+
+/* Thor2 specific code for RX completion parsing */
+#define RX_PKT_V3_CMPL_FLAGS2_IP_TYPE_SFT 8
+#define RX_PKT_V3_CMPL_METADATA1_VALID_SFT 15
+
+#define BNXT_CMPL_V3_ITYPE_TO_IDX(ft) \
+ (((ft) & RX_PKT_V3_CMPL_FLAGS_ITYPE_MASK) >> \
+ (RX_PKT_V3_CMPL_FLAGS_ITYPE_SFT - BNXT_PTYPE_TBL_TYPE_SFT))
+
+#define BNXT_CMPL_V3_VLAN_TO_IDX(meta) \
+ (((meta) & (1 << RX_PKT_V3_CMPL_METADATA1_VALID_SFT)) >> \
+ (RX_PKT_V3_CMPL_METADATA1_VALID_SFT - BNXT_PTYPE_TBL_VLAN_SFT))
+
+#define BNXT_CMPL_V3_IP_VER_TO_IDX(f2) \
+ (((f2) & RX_PKT_V3_CMPL_HI_FLAGS2_IP_TYPE) >> \
+ (RX_PKT_V3_CMPL_FLAGS2_IP_TYPE_SFT - BNXT_PTYPE_TBL_IP_VER_SFT))
+
+#define RX_CMP_V3_VLAN_VALID(rxcmp) \
+ (((struct rx_pkt_v3_cmpl *)rxcmp)->metadata1_payload_offset & \
+ RX_PKT_V3_CMPL_METADATA1_VALID)
+
+#define RX_CMP_V3_METADATA0_VID(rxcmp1) \
+ ((((struct rx_pkt_v3_cmpl_hi *)rxcmp1)->metadata0) & \
+ (RX_PKT_V3_CMPL_HI_METADATA0_VID_MASK | \
+ RX_PKT_V3_CMPL_HI_METADATA0_DE | \
+ RX_PKT_V3_CMPL_HI_METADATA0_PRI_MASK))
+
+static inline void bnxt_rx_vlan_v3(struct rte_mbuf *mbuf,
+ struct rx_pkt_cmpl *rxcmp,
+ struct rx_pkt_cmpl_hi *rxcmp1)
+{
+ if (RX_CMP_V3_VLAN_VALID(rxcmp)) {
+ mbuf->vlan_tci = RX_CMP_V3_METADATA0_VID(rxcmp1);
+ mbuf->ol_flags |= RTE_MBUF_F_RX_VLAN |
RTE_MBUF_F_RX_VLAN_STRIPPED;
+ }
+}
+
+#define RX_CMP_V3_L4_CS_ERR(err) \
+ (((err) & RX_PKT_CMPL_ERRORS_MASK) \
+ & (RX_PKT_CMPL_ERRORS_L4_CS_ERROR))
+#define RX_CMP_V3_L3_CS_ERR(err) \
+ (((err) & RX_PKT_CMPL_ERRORS_MASK) \
+ & (RX_PKT_CMPL_ERRORS_IP_CS_ERROR))
+#define RX_CMP_V3_T_IP_CS_ERR(err) \
+ (((err) & RX_PKT_CMPL_ERRORS_MASK) \
+ & (RX_PKT_CMPL_ERRORS_T_IP_CS_ERROR))
+#define RX_CMP_V3_T_L4_CS_ERR(err) \
+ (((err) & RX_PKT_CMPL_ERRORS_MASK) \
+ & (RX_PKT_CMPL_ERRORS_T_L4_CS_ERROR))
+#define RX_PKT_CMPL_CALC \
+ (RX_PKT_CMPL_FLAGS2_IP_CS_CALC | \
+ RX_PKT_CMPL_FLAGS2_L4_CS_CALC | \
+ RX_PKT_CMPL_FLAGS2_T_IP_CS_CALC | \
+ RX_PKT_CMPL_FLAGS2_T_L4_CS_CALC)
+
+static inline uint64_t
+bnxt_parse_csum_fields_v3(uint32_t flags2, uint32_t error_v2)
+{
+ uint64_t ol_flags = 0;
+
+ if (flags2 & RX_PKT_CMPL_CALC) {
+ if (unlikely(RX_CMP_V3_L4_CS_ERR(error_v2)))
+ ol_flags |= RTE_MBUF_F_RX_L4_CKSUM_BAD;
+ else
+ ol_flags |= RTE_MBUF_F_RX_L4_CKSUM_GOOD;
+ if (unlikely(RX_CMP_V3_L3_CS_ERR(error_v2)))
+ ol_flags |= RTE_MBUF_F_RX_IP_CKSUM_BAD;
+ if (unlikely(RX_CMP_V3_T_L4_CS_ERR(error_v2)))
+ ol_flags |= RTE_MBUF_F_RX_OUTER_L4_CKSUM_BAD;
+ else
+ ol_flags |= RTE_MBUF_F_RX_OUTER_L4_CKSUM_GOOD;
+ if (unlikely(RX_CMP_V3_T_IP_CS_ERR(error_v2)))
+ ol_flags |= RTE_MBUF_F_RX_OUTER_IP_CKSUM_BAD;
+ if (!(ol_flags & (RTE_MBUF_F_RX_IP_CKSUM_BAD |
RTE_MBUF_F_RX_OUTER_IP_CKSUM_BAD)))
+ ol_flags |= RTE_MBUF_F_RX_IP_CKSUM_GOOD;
+ } else {
+ /* Unknown is defined as 0 for all packets types hence using
below for all */
+ ol_flags |= RTE_MBUF_F_RX_IP_CKSUM_UNKNOWN;
+ }
+ return ol_flags;
+}
+
+static inline void
+bnxt_parse_csum_v3(struct rte_mbuf *mbuf, struct rx_pkt_cmpl_hi *rxcmp1)
+{
+ struct rx_pkt_v3_cmpl_hi *v3_cmp =
+ (struct rx_pkt_v3_cmpl_hi *)(rxcmp1);
+ uint16_t error_v2 = rte_le_to_cpu_16(v3_cmp->errors_v2);
+ uint32_t flags2 = rte_le_to_cpu_32(v3_cmp->flags2);
+
+ mbuf->ol_flags |= bnxt_parse_csum_fields_v3(flags2, error_v2);
+}
#endif /* _BNXT_RXR_H_ */
--
2.39.2 (Apple Git-143)
smime.p7s
Description: S/MIME Cryptographic Signature
