2023-09-15 01:31:23

by Hariprasad Kelam

[permalink] [raw]
Subject: [net-next Patch] octeontx2-pf: Tc flower offload support for MPLS

This patch extends flower offload support for MPLS protocol.
Due to hardware limitation, currently driver supports lse
depth up to 4.

Signed-off-by: Hariprasad Kelam <[email protected]>
Signed-off-by: Sunil Goutham <[email protected]>
---
.../net/ethernet/marvell/octeontx2/af/mbox.h | 6 +++
.../net/ethernet/marvell/octeontx2/af/npc.h | 8 +++
.../marvell/octeontx2/af/rvu_debugfs.c | 52 ++++++++++++++++++
.../marvell/octeontx2/af/rvu_npc_fs.c | 46 ++++++++++++++++
.../ethernet/marvell/octeontx2/nic/otx2_tc.c | 54 +++++++++++++++++++
5 files changed, 166 insertions(+)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mbox.h b/drivers/net/ethernet/marvell/octeontx2/af/mbox.h
index 6b5b06c2b4e9..b82a9f8c329b 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/mbox.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/mbox.h
@@ -1473,6 +1473,12 @@ struct flow_msg {
u8 next_header;
};
__be16 vlan_itci;
+#define OTX2_FLOWER_MASK_MPLS_LB GENMASK(31, 12)
+#define OTX2_FLOWER_MASK_MPLS_TC GENMASK(11, 9)
+#define OTX2_FLOWER_MASK_MPLS_BOS BIT(8)
+#define OTX2_FLOWER_MASK_MPLS_TTL GENMASK(7, 0)
+#define OTX2_FLOWER_MASK_MPLS_NON_TTL GENMASK(31, 8)
+ u32 mpls_lse[4];
};

struct npc_install_flow_req {
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/npc.h b/drivers/net/ethernet/marvell/octeontx2/af/npc.h
index de9fbd98dfb7..ab3e39eef2eb 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/npc.h
+++ b/drivers/net/ethernet/marvell/octeontx2/af/npc.h
@@ -206,6 +206,14 @@ enum key_fields {
NPC_SPORT_SCTP,
NPC_DPORT_SCTP,
NPC_IPSEC_SPI,
+ NPC_MPLS1_LBTCBOS,
+ NPC_MPLS1_TTL,
+ NPC_MPLS2_LBTCBOS,
+ NPC_MPLS2_TTL,
+ NPC_MPLS3_LBTCBOS,
+ NPC_MPLS3_TTL,
+ NPC_MPLS4_LBTCBOS,
+ NPC_MPLS4_TTL,
NPC_HEADER_FIELDS_MAX,
NPC_CHAN = NPC_HEADER_FIELDS_MAX, /* Valid when Rx */
NPC_PF_FUNC, /* Valid when Tx */
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
index d30e84803481..e71c3da52cfd 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_debugfs.c
@@ -2756,6 +2756,26 @@ static int rvu_dbg_npc_rx_miss_stats_display(struct seq_file *filp,

RVU_DEBUG_SEQ_FOPS(npc_rx_miss_act, npc_rx_miss_stats_display, NULL);

+#define RVU_DBG_PRINT_MPLS_TTL(pkt, mask) \
+do { \
+ seq_printf(s, "%ld ", FIELD_GET(OTX2_FLOWER_MASK_MPLS_TTL, pkt)); \
+ seq_printf(s, "mask 0x%lx\n", FIELD_GET(OTX2_FLOWER_MASK_MPLS_TTL, mask)); \
+} while (0) \
+
+#define RVU_DBG_PRINT_MPLS_LBTCBOS(_pkt, _mask) \
+do { \
+ typeof(_pkt) (pkt) = (_pkt); \
+ typeof(_mask) (mask) = (_mask); \
+ seq_printf(s, "%ld %ld %ld\n", \
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_LB, pkt), \
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_TC, pkt), \
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_BOS, pkt)); \
+ seq_printf(s, "\tmask 0x%lx 0x%lx 0x%lx\n", \
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_LB, mask), \
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_TC, mask), \
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_BOS, mask)); \
+} while (0) \
+
static void rvu_dbg_npc_mcam_show_flows(struct seq_file *s,
struct rvu_npc_mcam_rule *rule)
{
@@ -2836,6 +2856,38 @@ static void rvu_dbg_npc_mcam_show_flows(struct seq_file *s,
seq_printf(s, "0x%x ", ntohl(rule->packet.spi));
seq_printf(s, "mask 0x%x\n", ntohl(rule->mask.spi));
break;
+ case NPC_MPLS1_LBTCBOS:
+ RVU_DBG_PRINT_MPLS_LBTCBOS(rule->packet.mpls_lse[0],
+ rule->mask.mpls_lse[0]);
+ break;
+ case NPC_MPLS1_TTL:
+ RVU_DBG_PRINT_MPLS_TTL(rule->packet.mpls_lse[0],
+ rule->mask.mpls_lse[0]);
+ break;
+ case NPC_MPLS2_LBTCBOS:
+ RVU_DBG_PRINT_MPLS_LBTCBOS(rule->packet.mpls_lse[1],
+ rule->mask.mpls_lse[1]);
+ break;
+ case NPC_MPLS2_TTL:
+ RVU_DBG_PRINT_MPLS_TTL(rule->packet.mpls_lse[1],
+ rule->mask.mpls_lse[1]);
+ break;
+ case NPC_MPLS3_LBTCBOS:
+ RVU_DBG_PRINT_MPLS_LBTCBOS(rule->packet.mpls_lse[2],
+ rule->mask.mpls_lse[2]);
+ break;
+ case NPC_MPLS3_TTL:
+ RVU_DBG_PRINT_MPLS_TTL(rule->packet.mpls_lse[2],
+ rule->mask.mpls_lse[2]);
+ break;
+ case NPC_MPLS4_LBTCBOS:
+ RVU_DBG_PRINT_MPLS_LBTCBOS(rule->packet.mpls_lse[3],
+ rule->mask.mpls_lse[3]);
+ break;
+ case NPC_MPLS4_TTL:
+ RVU_DBG_PRINT_MPLS_TTL(rule->packet.mpls_lse[3],
+ rule->mask.mpls_lse[3]);
+ break;
default:
seq_puts(s, "\n");
break;
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
index 237f82082ebe..8c3c1a1e10a6 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_npc_fs.c
@@ -43,6 +43,14 @@ static const char * const npc_flow_names[] = {
[NPC_DPORT_SCTP] = "sctp destination port",
[NPC_LXMB] = "Mcast/Bcast header ",
[NPC_IPSEC_SPI] = "SPI ",
+ [NPC_MPLS1_LBTCBOS] = "lse depth 1 label tc bos",
+ [NPC_MPLS1_TTL] = "lse depth 1 ttl",
+ [NPC_MPLS2_LBTCBOS] = "lse depth 2 label tc bos",
+ [NPC_MPLS2_TTL] = "lse depth 2 ttl",
+ [NPC_MPLS3_LBTCBOS] = "lse depth 3 label tc bos",
+ [NPC_MPLS3_TTL] = "lse depth 3 ttl",
+ [NPC_MPLS4_LBTCBOS] = "lse depth 4 label tc bos",
+ [NPC_MPLS4_TTL] = "lse depth 4",
[NPC_UNKNOWN] = "unknown",
};

@@ -528,6 +536,14 @@ do { \

NPC_SCAN_HDR(NPC_IPSEC_SPI, NPC_LID_LD, NPC_LT_LD_AH, 4, 4);
NPC_SCAN_HDR(NPC_IPSEC_SPI, NPC_LID_LE, NPC_LT_LE_ESP, 0, 4);
+ NPC_SCAN_HDR(NPC_MPLS1_LBTCBOS, NPC_LID_LC, NPC_LT_LC_MPLS, 0, 3);
+ NPC_SCAN_HDR(NPC_MPLS1_TTL, NPC_LID_LC, NPC_LT_LC_MPLS, 3, 1);
+ NPC_SCAN_HDR(NPC_MPLS2_LBTCBOS, NPC_LID_LC, NPC_LT_LC_MPLS, 4, 3);
+ NPC_SCAN_HDR(NPC_MPLS2_TTL, NPC_LID_LC, NPC_LT_LC_MPLS, 7, 1);
+ NPC_SCAN_HDR(NPC_MPLS3_LBTCBOS, NPC_LID_LC, NPC_LT_LC_MPLS, 8, 3);
+ NPC_SCAN_HDR(NPC_MPLS3_TTL, NPC_LID_LC, NPC_LT_LC_MPLS, 11, 1);
+ NPC_SCAN_HDR(NPC_MPLS4_LBTCBOS, NPC_LID_LC, NPC_LT_LC_MPLS, 12, 3);
+ NPC_SCAN_HDR(NPC_MPLS4_TTL, NPC_LID_LC, NPC_LT_LC_MPLS, 15, 1);

/* SMAC follows the DMAC(which is 6 bytes) */
NPC_SCAN_HDR(NPC_SMAC, NPC_LID_LA, la_ltype, la_start + 6, 6);
@@ -593,6 +609,11 @@ static void npc_set_features(struct rvu *rvu, int blkaddr, u8 intf)
/* for L2M/L2B/L3M/L3B, check if the type is present in the key */
if (npc_check_field(rvu, blkaddr, NPC_LXMB, intf))
*features |= BIT_ULL(NPC_LXMB);
+
+ for (hdr = NPC_MPLS1_LBTCBOS; hdr <= NPC_MPLS4_TTL; hdr++) {
+ if (npc_check_field(rvu, blkaddr, hdr, intf))
+ *features |= BIT_ULL(hdr);
+ }
}

/* Scan key extraction profile and record how fields of our interest
@@ -959,6 +980,31 @@ do { \
NPC_WRITE_FLOW(NPC_INNER_VID, vlan_itci, ntohs(pkt->vlan_itci), 0,
ntohs(mask->vlan_itci), 0);

+ NPC_WRITE_FLOW(NPC_MPLS1_LBTCBOS, mpls_lse,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_NON_TTL, pkt->mpls_lse[0]), 0,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_NON_TTL, mask->mpls_lse[0]), 0);
+ NPC_WRITE_FLOW(NPC_MPLS1_TTL, mpls_lse,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_TTL, pkt->mpls_lse[0]), 0,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_TTL, mask->mpls_lse[0]), 0);
+ NPC_WRITE_FLOW(NPC_MPLS2_LBTCBOS, mpls_lse,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_NON_TTL, pkt->mpls_lse[1]), 0,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_NON_TTL, mask->mpls_lse[1]), 0);
+ NPC_WRITE_FLOW(NPC_MPLS2_TTL, mpls_lse,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_TTL, pkt->mpls_lse[1]), 0,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_TTL, mask->mpls_lse[1]), 0);
+ NPC_WRITE_FLOW(NPC_MPLS3_LBTCBOS, mpls_lse,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_NON_TTL, pkt->mpls_lse[2]), 0,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_NON_TTL, mask->mpls_lse[2]), 0);
+ NPC_WRITE_FLOW(NPC_MPLS3_TTL, mpls_lse,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_TTL, pkt->mpls_lse[2]), 0,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_TTL, mask->mpls_lse[2]), 0);
+ NPC_WRITE_FLOW(NPC_MPLS4_LBTCBOS, mpls_lse,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_NON_TTL, pkt->mpls_lse[3]), 0,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_NON_TTL, mask->mpls_lse[3]), 0);
+ NPC_WRITE_FLOW(NPC_MPLS4_TTL, mpls_lse,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_TTL, pkt->mpls_lse[3]), 0,
+ FIELD_GET(OTX2_FLOWER_MASK_MPLS_TTL, mask->mpls_lse[3]), 0);
+
NPC_WRITE_FLOW(NPC_IPFRAG_IPV6, next_header, pkt->next_header, 0,
mask->next_header, 0);
npc_update_ipv6_flow(rvu, entry, features, pkt, mask, output, intf);
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
index fab9d85bfb37..760237f413fb 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_tc.c
@@ -27,6 +27,8 @@
#define CN10K_TLX_BURST_MANTISSA GENMASK_ULL(43, 29)
#define CN10K_TLX_BURST_EXPONENT GENMASK_ULL(47, 44)

+#define OTX2_UNSUPP_LSE_DEPTH GENMASK(6, 4)
+
struct otx2_tc_flow_stats {
u64 bytes;
u64 pkts;
@@ -519,6 +521,7 @@ static int otx2_tc_prepare_flow(struct otx2_nic *nic, struct otx2_tc_flow *node,
BIT_ULL(FLOW_DISSECTOR_KEY_IPV6_ADDRS) |
BIT_ULL(FLOW_DISSECTOR_KEY_PORTS) |
BIT(FLOW_DISSECTOR_KEY_IPSEC) |
+ BIT_ULL(FLOW_DISSECTOR_KEY_MPLS) |
BIT_ULL(FLOW_DISSECTOR_KEY_IP)))) {
netdev_info(nic->netdev, "unsupported flow used key 0x%llx",
dissector->used_keys);
@@ -738,6 +741,57 @@ static int otx2_tc_prepare_flow(struct otx2_nic *nic, struct otx2_tc_flow *node,
}
}

+ if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_MPLS)) {
+ struct flow_match_mpls match;
+ u8 bit;
+
+ flow_rule_match_mpls(rule, &match);
+
+ if (match.mask->used_lses & OTX2_UNSUPP_LSE_DEPTH) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "unsupported LSE depth for MPLS match offload");
+ return -EOPNOTSUPP;
+ }
+
+ for_each_set_bit(bit, (unsigned long *)&match.mask->used_lses,
+ FLOW_DIS_MPLS_MAX) {
+ /* check if any of the fields LABEL,TC,BOS are set */
+ if (*((u32 *)&match.mask->ls[bit]) & 0xffffff00) {
+ /* Hardware will capture 4 byte MPLS header into two
+ * fields NPC_MPLSX_LBTCBOS and NPC_MPLSX_TTL. Derive
+ * the associated NPC key based on header index and offset.
+ */
+
+ req->features |= BIT_ULL(NPC_MPLS1_LBTCBOS + 2 * bit);
+ flow_spec->mpls_lse[bit] =
+ FIELD_PREP(OTX2_FLOWER_MASK_MPLS_LB,
+ match.key->ls[bit].mpls_label) |
+ FIELD_PREP(OTX2_FLOWER_MASK_MPLS_TC,
+ match.key->ls[bit].mpls_tc) |
+ FIELD_PREP(OTX2_FLOWER_MASK_MPLS_BOS,
+ match.key->ls[bit].mpls_bos);
+
+ flow_mask->mpls_lse[bit] =
+ FIELD_PREP(OTX2_FLOWER_MASK_MPLS_LB,
+ match.mask->ls[bit].mpls_label) |
+ FIELD_PREP(OTX2_FLOWER_MASK_MPLS_TC,
+ match.mask->ls[bit].mpls_tc) |
+ FIELD_PREP(OTX2_FLOWER_MASK_MPLS_BOS,
+ match.mask->ls[bit].mpls_bos);
+ }
+
+ if (match.mask->ls[bit].mpls_ttl) {
+ req->features |= BIT_ULL(NPC_MPLS1_TTL + 2 * bit);
+ flow_spec->mpls_lse[bit] |=
+ FIELD_PREP(OTX2_FLOWER_MASK_MPLS_TTL,
+ match.key->ls[bit].mpls_ttl);
+ flow_mask->mpls_lse[bit] |=
+ FIELD_PREP(OTX2_FLOWER_MASK_MPLS_TTL,
+ match.mask->ls[bit].mpls_ttl);
+ }
+ }
+ }
+
return otx2_tc_parse_actions(nic, &rule->action, req, f, node);
}

--
2.17.1


2023-09-16 15:06:57

by Simon Horman

[permalink] [raw]
Subject: Re: [net-next Patch] octeontx2-pf: Tc flower offload support for MPLS

On Thu, Sep 14, 2023 at 04:36:55PM +0530, Hariprasad Kelam wrote:

...

> @@ -738,6 +741,57 @@ static int otx2_tc_prepare_flow(struct otx2_nic *nic, struct otx2_tc_flow *node,
> }
> }
>
> + if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_MPLS)) {
> + struct flow_match_mpls match;
> + u8 bit;
> +
> + flow_rule_match_mpls(rule, &match);
> +
> + if (match.mask->used_lses & OTX2_UNSUPP_LSE_DEPTH) {
> + NL_SET_ERR_MSG_MOD(extack,
> + "unsupported LSE depth for MPLS match offload");
> + return -EOPNOTSUPP;
> + }
> +
> + for_each_set_bit(bit, (unsigned long *)&match.mask->used_lses,
> + FLOW_DIS_MPLS_MAX) {
> + /* check if any of the fields LABEL,TC,BOS are set */
> + if (*((u32 *)&match.mask->ls[bit]) & 0xffffff00) {

Hi Hariprasad,

I wonder if we could avoid using the magic number 0xffffff00 above.
Perhaps ~MPLS_LS_TTL_MASK is appropriate?

Otherwise this patch looks good to me.

Reviewed-by: Simon Horman <[email protected]>

...