git.droids-corp.org / dpdk.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
net/cxgbe: support flow API for matching QinQ VLAN
[dpdk.git] / drivers / net / cxgbe / cxgbe_flow.c
diff --git a/drivers/net/cxgbe/cxgbe_flow.c b/drivers/net/cxgbe/cxgbe_flow.c
index 9070f49..cd833d0 100644 (file)
--- a/drivers/net/cxgbe/cxgbe_flow.c
+++ b/drivers/net/cxgbe/cxgbe_flow.c
@@ -87,6 +87,15 @@ cxgbe_tweak_filter_spec(struct adapter *adap,
                                fs->mask.ethtype = 0;
                        }
                        break;
+               case RTE_ETHER_TYPE_QINQ:
+                       if (adap->params.tp.ethertype_shift < 0 &&
+                           adap->params.tp.vnic_shift >= 0) {
+                               fs->val.ovlan_vld = 1;
+                               fs->mask.ovlan_vld = 1;
+                               fs->val.ethtype = 0;
+                               fs->mask.ethtype = 0;
+                       }
+                       break;
                default:
                        break;
                }
@@ -145,6 +154,9 @@ cxgbe_fill_filter_region(struct adapter *adap,
        if (tp->vlan_shift >= 0 && fs->mask.ivlan_vld)
                ntuple_mask |= (u64)(F_FT_VLAN_VLD | fs->mask.ivlan) <<
                               tp->vlan_shift;
+       if (tp->vnic_shift >= 0 && fs->mask.ovlan_vld)
+               ntuple_mask |= (u64)(F_FT_VLAN_VLD | fs->mask.ovlan) <<
+                              tp->vnic_shift;
 
        if (ntuple_mask != hash_filter_mask)
                return;
@@ -167,22 +179,6 @@ ch_rte_parsetype_eth(const void *dmask, const struct rte_flow_item *item,
        if (!spec)
                return 0;
 
-       /* Chelsio hardware supports matching on only one ethertype
-        * (i.e. either the outer or inner ethertype, but not both). If
-        * we already encountered VLAN item, then ensure that the outer
-        * ethertype is VLAN (0x8100) and don't overwrite the inner
-        * ethertype stored during VLAN item parsing. Note that if
-        * 'ivlan_vld' bit is set in Chelsio filter spec, then the
-        * hardware automatically only matches packets with outer
-        * ethertype having VLAN (0x8100).
-        */
-       if (fs->mask.ivlan_vld &&
-           be16_to_cpu(spec->type) != RTE_ETHER_TYPE_VLAN)
-               return rte_flow_error_set(e, EINVAL, RTE_FLOW_ERROR_TYPE_ITEM,
-                                         item,
-                                         "Already encountered VLAN item,"
-                                         " but outer ethertype is not 0x8100");
-
        /* we don't support SRC_MAC filtering*/
        if (!rte_is_zero_ether_addr(&mask->src))
                return rte_flow_error_set(e, ENOTSUP, RTE_FLOW_ERROR_TYPE_ITEM,
@@ -206,13 +202,9 @@ ch_rte_parsetype_eth(const void *dmask, const struct rte_flow_item *item,
                CXGBE_FILL_FS(idx, 0x1ff, macidx);
        }
 
-       /* Only set outer ethertype, if we didn't encounter VLAN item yet.
-        * Otherwise, the inner ethertype set by VLAN item will get
-        * overwritten.
-        */
-       if (!fs->mask.ivlan_vld)
-               CXGBE_FILL_FS(be16_to_cpu(spec->type),
-                             be16_to_cpu(mask->type), ethtype);
+       CXGBE_FILL_FS(be16_to_cpu(spec->type),
+                     be16_to_cpu(mask->type), ethtype);
+
        return 0;
 }
 
@@ -249,32 +241,48 @@ ch_rte_parsetype_vlan(const void *dmask, const struct rte_flow_item *item,
        /* If user has not given any mask, then use chelsio supported mask. */
        mask = umask ? umask : (const struct rte_flow_item_vlan *)dmask;
 
-       CXGBE_FILL_FS(1, 1, ivlan_vld);
-       if (!spec)
-               return 0; /* Wildcard, match all VLAN */
-
-       /* Chelsio hardware supports matching on only one ethertype
-        * (i.e. either the outer or inner ethertype, but not both).
-        * If outer ethertype is already set and is not VLAN (0x8100),
-        * then don't proceed further. Otherwise, reset the outer
-        * ethertype, so that it can be replaced by inner ethertype.
-        * Note that the hardware will automatically match on outer
-        * ethertype 0x8100, if 'ivlan_vld' bit is set in Chelsio
-        * filter spec.
+       if (!fs->mask.ethtype)
+               return rte_flow_error_set(e, EINVAL, RTE_FLOW_ERROR_TYPE_ITEM,
+                                         item,
+                                         "Can't parse VLAN item without knowing ethertype");
+
+       /* If ethertype is already set and is not VLAN (0x8100) or
+        * QINQ(0x88A8), then don't proceed further. Otherwise,
+        * reset the outer ethertype, so that it can be replaced by
+        * innermost ethertype. Note that hardware will automatically
+        * match against VLAN or QINQ packets, based on 'ivlan_vld' or
+        * 'ovlan_vld' bit set in Chelsio filter spec, respectively.
         */
        if (fs->mask.ethtype) {
-               if (fs->val.ethtype != RTE_ETHER_TYPE_VLAN)
+               if (fs->val.ethtype != RTE_ETHER_TYPE_VLAN &&
+                   fs->val.ethtype != RTE_ETHER_TYPE_QINQ)
                        return rte_flow_error_set(e, EINVAL,
                                                  RTE_FLOW_ERROR_TYPE_ITEM,
                                                  item,
-                                                 "Outer ethertype not 0x8100");
+                                                 "Ethertype must be 0x8100 or 0x88a8");
+       }
 
-               fs->val.ethtype = 0;
-               fs->mask.ethtype = 0;
+       if (fs->val.ethtype == RTE_ETHER_TYPE_QINQ) {
+               CXGBE_FILL_FS(1, 1, ovlan_vld);
+               if (spec) {
+                       CXGBE_FILL_FS(be16_to_cpu(spec->tci),
+                                     be16_to_cpu(mask->tci), ovlan);
+
+                       fs->mask.ethtype = 0;
+                       fs->val.ethtype = 0;
+               }
+       } else if (fs->val.ethtype == RTE_ETHER_TYPE_VLAN) {
+               CXGBE_FILL_FS(1, 1, ivlan_vld);
+               if (spec) {
+                       CXGBE_FILL_FS(be16_to_cpu(spec->tci),
+                                     be16_to_cpu(mask->tci), ivlan);
+
+                       fs->mask.ethtype = 0;
+                       fs->val.ethtype = 0;
+               }
        }
 
-       CXGBE_FILL_FS(be16_to_cpu(spec->tci), be16_to_cpu(mask->tci), ivlan);
-       if (spec->inner_type)
+       if (spec)
                CXGBE_FILL_FS(be16_to_cpu(spec->inner_type),
                              be16_to_cpu(mask->inner_type), ethtype);
 
@@ -351,8 +359,7 @@ ch_rte_parsetype_ipv4(const void *dmask, const struct rte_flow_item *item,
                                          item, "ttl/tos are not supported");
 
        if (fs->mask.ethtype &&
-           (fs->val.ethtype != RTE_ETHER_TYPE_VLAN &&
-            fs->val.ethtype != RTE_ETHER_TYPE_IPV4))
+           (fs->val.ethtype != RTE_ETHER_TYPE_IPV4))
                return rte_flow_error_set(e, EINVAL, RTE_FLOW_ERROR_TYPE_ITEM,
                                          item,
                                          "Couldn't find IPv4 ethertype");
@@ -385,8 +392,7 @@ ch_rte_parsetype_ipv6(const void *dmask, const struct rte_flow_item *item,
                                          "tc/flow/hop are not supported");
 
        if (fs->mask.ethtype &&
-           (fs->val.ethtype != RTE_ETHER_TYPE_VLAN &&
-            fs->val.ethtype != RTE_ETHER_TYPE_IPV6))
+           (fs->val.ethtype != RTE_ETHER_TYPE_IPV6))
                return rte_flow_error_set(e, EINVAL, RTE_FLOW_ERROR_TYPE_ITEM,
                                          item,
                                          "Couldn't find IPv6 ethertype");
@@ -907,10 +913,12 @@ cxgbe_rtef_parse_items(struct rte_flow *flow,
                        continue;
                default:
                        /* check if item is repeated */
-                       if (repeat[i->type])
+                       if (repeat[i->type] &&
+                           i->type != RTE_FLOW_ITEM_TYPE_VLAN)
                                return rte_flow_error_set(e, ENOTSUP,
                                                RTE_FLOW_ERROR_TYPE_ITEM, i,
-                                               "parse items cannot be repeated (except void)");
+                                               "parse items cannot be repeated(except void/vlan)");
+
                        repeat[i->type] = 1;
 
                        /* No spec found for this pattern item. Skip it */
DPDK repo used for reviews