+ * @param[in] attributes
+ * Pointer to flow attributes
+ * @param[in, out] flow
+ * Pointer to the rte_flow structure.
+ * @param[out] error
+ * Pointer to error structure.
+ *
+ * @return
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+static int
+mlx5_flow_attributes(struct rte_eth_dev *dev,
+ const struct rte_flow_attr *attributes,
+ struct rte_flow *flow,
+ struct rte_flow_error *error)
+{
+ uint32_t priority_max =
+ ((struct priv *)dev->data->dev_private)->config.flow_prio;
+
+ if (attributes->group)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ATTR_GROUP,
+ NULL,
+ "groups is not supported");
+ if (attributes->priority >= priority_max)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ATTR_PRIORITY,
+ NULL,
+ "priority out of range");
+ if (attributes->egress)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ATTR_EGRESS,
+ NULL,
+ "egress is not supported");
+ if (attributes->transfer)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ATTR_TRANSFER,
+ NULL,
+ "transfer is not supported");
+ if (!attributes->ingress)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ATTR_INGRESS,
+ NULL,
+ "ingress attribute is mandatory");
+ flow->attributes = *attributes;
+ return 0;
+}
+
+/**
+ * Verify the @p item specifications (spec, last, mask) are compatible with the
+ * NIC capabilities.
+ *
+ * @param[in] item
+ * Item specification.
+ * @param[in] mask
+ * @p item->mask or flow default bit-masks.
+ * @param[in] nic_mask
+ * Bit-masks covering supported fields by the NIC to compare with user mask.
+ * @param[in] size
+ * Bit-masks size in bytes.
+ * @param[out] error
+ * Pointer to error structure.
+ *
+ * @return
+ * 0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+static int
+mlx5_flow_item_acceptable(const struct rte_flow_item *item,
+ const uint8_t *mask,
+ const uint8_t *nic_mask,
+ unsigned int size,
+ struct rte_flow_error *error)
+{
+ unsigned int i;
+
+ assert(nic_mask);
+ for (i = 0; i < size; ++i)
+ if ((nic_mask[i] | mask[i]) != nic_mask[i])
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ITEM,
+ item,
+ "mask enables non supported"
+ " bits");
+ if (!item->spec && (item->mask || item->last))
+ return rte_flow_error_set(error, EINVAL,
+ RTE_FLOW_ERROR_TYPE_ITEM,
+ item,
+ "mask/last without a spec is not"
+ " supported");
+ if (item->spec && item->last) {
+ uint8_t spec[size];
+ uint8_t last[size];
+ unsigned int i;
+ int ret;
+
+ for (i = 0; i < size; ++i) {
+ spec[i] = ((const uint8_t *)item->spec)[i] & mask[i];
+ last[i] = ((const uint8_t *)item->last)[i] & mask[i];
+ }
+ ret = memcmp(spec, last, size);
+ if (ret != 0)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ITEM,
+ item,
+ "range is not supported");
+ }
+ return 0;
+}
+
+/**
+ * Add a verbs specification into @p flow.
+ *
+ * @param[in, out] flow
+ * Pointer to flow structure.
+ * @param[in] src
+ * Create specification.
+ * @param[in] size
+ * Size in bytes of the specification to copy.
+ */
+static void
+mlx5_flow_spec_verbs_add(struct rte_flow *flow, void *src, unsigned int size)
+{
+ if (flow->verbs.specs) {
+ void *dst;
+
+ dst = (void *)(flow->verbs.specs + flow->verbs.size);
+ memcpy(dst, src, size);
+ ++flow->verbs.attr->num_of_specs;
+ }
+ flow->verbs.size += size;
+}
+
+/**
+ * Convert the @p item into a Verbs specification after ensuring the NIC
+ * will understand and process it correctly.
+ * If the necessary size for the conversion is greater than the @p flow_size,
+ * nothing is written in @p flow, the validation is still performed.
+ *
+ * @param[in] item
+ * Item specification.
+ * @param[in, out] flow
+ * Pointer to flow structure.
+ * @param[in] flow_size
+ * Size in bytes of the available space in @p flow, if too small, nothing is
+ * written.
+ * @param[out] error
+ * Pointer to error structure.
+ *
+ * @return
+ * On success the number of bytes consumed/necessary, if the returned value
+ * is lesser or equal to @p flow_size, the @p item has fully been converted,
+ * otherwise another call with this returned memory size should be done.
+ * On error, a negative errno value is returned and rte_errno is set.
+ */
+static int
+mlx5_flow_item_eth(const struct rte_flow_item *item, struct rte_flow *flow,
+ const size_t flow_size, struct rte_flow_error *error)
+{
+ const struct rte_flow_item_eth *spec = item->spec;
+ const struct rte_flow_item_eth *mask = item->mask;
+ const struct rte_flow_item_eth nic_mask = {
+ .dst.addr_bytes = "\xff\xff\xff\xff\xff\xff",
+ .src.addr_bytes = "\xff\xff\xff\xff\xff\xff",
+ .type = RTE_BE16(0xffff),
+ };
+ const unsigned int size = sizeof(struct ibv_flow_spec_eth);
+ struct ibv_flow_spec_eth eth = {
+ .type = IBV_FLOW_SPEC_ETH,
+ .size = size,
+ };
+ int ret;
+
+ if (flow->layers & MLX5_FLOW_LAYER_OUTER_L2)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ITEM,
+ item,
+ "L2 layers already configured");
+ if (!mask)
+ mask = &rte_flow_item_eth_mask;
+ ret = mlx5_flow_item_acceptable(item, (const uint8_t *)mask,
+ (const uint8_t *)&nic_mask,
+ sizeof(struct rte_flow_item_eth),
+ error);
+ if (ret)
+ return ret;
+ flow->layers |= MLX5_FLOW_LAYER_OUTER_L2;
+ if (size > flow_size)
+ return size;
+ if (spec) {
+ unsigned int i;
+
+ memcpy(ð.val.dst_mac, spec->dst.addr_bytes, ETHER_ADDR_LEN);
+ memcpy(ð.val.src_mac, spec->src.addr_bytes, ETHER_ADDR_LEN);
+ eth.val.ether_type = spec->type;
+ memcpy(ð.mask.dst_mac, mask->dst.addr_bytes, ETHER_ADDR_LEN);
+ memcpy(ð.mask.src_mac, mask->src.addr_bytes, ETHER_ADDR_LEN);
+ eth.mask.ether_type = mask->type;
+ /* Remove unwanted bits from values. */
+ for (i = 0; i < ETHER_ADDR_LEN; ++i) {
+ eth.val.dst_mac[i] &= eth.mask.dst_mac[i];
+ eth.val.src_mac[i] &= eth.mask.src_mac[i];
+ }
+ eth.val.ether_type &= eth.mask.ether_type;
+ }
+ mlx5_flow_spec_verbs_add(flow, ð, size);
+ return size;
+}
+
+/**
+ * Convert the @p pattern into a Verbs specifications after ensuring the NIC
+ * will understand and process it correctly.
+ * The conversion is performed item per item, each of them is written into
+ * the @p flow if its size is lesser or equal to @p flow_size.
+ * Validation and memory consumption computation are still performed until the
+ * end of @p pattern, unless an error is encountered.
+ *
+ * @param[in] pattern
+ * Flow pattern.
+ * @param[in, out] flow
+ * Pointer to the rte_flow structure.
+ * @param[in] flow_size
+ * Size in bytes of the available space in @p flow, if too small some
+ * garbage may be present.
+ * @param[out] error
+ * Pointer to error structure.
+ *
+ * @return
+ * On success the number of bytes consumed/necessary, if the returned value
+ * is lesser or equal to @p flow_size, the @pattern has fully been
+ * converted, otherwise another call with this returned memory size should
+ * be done.
+ * On error, a negative errno value is returned and rte_errno is set.
+ */
+static int
+mlx5_flow_items(const struct rte_flow_item pattern[],
+ struct rte_flow *flow, const size_t flow_size,
+ struct rte_flow_error *error)
+{
+ int remain = flow_size;
+ size_t size = 0;
+
+ for (; pattern->type != RTE_FLOW_ITEM_TYPE_END; pattern++) {
+ int ret = 0;
+
+ switch (pattern->type) {
+ case RTE_FLOW_ITEM_TYPE_VOID:
+ break;
+ case RTE_FLOW_ITEM_TYPE_ETH:
+ ret = mlx5_flow_item_eth(pattern, flow, remain, error);
+ break;
+ default:
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ITEM,
+ pattern,
+ "item not supported");
+ }
+ if (ret < 0)
+ return ret;
+ if (remain > ret)
+ remain -= ret;
+ else
+ remain = 0;
+ size += ret;
+ }
+ if (!flow->layers) {
+ const struct rte_flow_item item = {
+ .type = RTE_FLOW_ITEM_TYPE_ETH,
+ };
+
+ return mlx5_flow_item_eth(&item, flow, flow_size, error);
+ }
+ return size;
+}
+
+/**
+ * Convert the @p action into a Verbs specification after ensuring the NIC
+ * will understand and process it correctly.
+ * If the necessary size for the conversion is greater than the @p flow_size,
+ * nothing is written in @p flow, the validation is still performed.
+ *
+ * @param[in] action
+ * Action configuration.
+ * @param[in, out] flow
+ * Pointer to flow structure.
+ * @param[in] flow_size
+ * Size in bytes of the available space in @p flow, if too small, nothing is
+ * written.
+ * @param[out] error
+ * Pointer to error structure.
+ *
+ * @return
+ * On success the number of bytes consumed/necessary, if the returned value
+ * is lesser or equal to @p flow_size, the @p action has fully been
+ * converted, otherwise another call with this returned memory size should
+ * be done.
+ * On error, a negative errno value is returned and rte_errno is set.
+ */
+static int
+mlx5_flow_action_drop(const struct rte_flow_action *action,
+ struct rte_flow *flow, const size_t flow_size,
+ struct rte_flow_error *error)
+{
+ unsigned int size = sizeof(struct ibv_flow_spec_action_drop);
+ struct ibv_flow_spec_action_drop drop = {
+ .type = IBV_FLOW_SPEC_ACTION_DROP,
+ .size = size,
+ };
+
+ if (flow->fate)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ACTION,
+ action,
+ "multiple fate actions are not"
+ " supported");
+ if (size < flow_size)
+ mlx5_flow_spec_verbs_add(flow, &drop, size);
+ flow->fate |= MLX5_FLOW_FATE_DROP;
+ return size;
+}
+
+/**
+ * Convert the @p action into @p flow after ensuring the NIC will understand
+ * and process it correctly.
+ * The conversion is performed action per action, each of them is written into
+ * the @p flow if its size is lesser or equal to @p flow_size.
+ * Validation and memory consumption computation are still performed until the
+ * end of @p action, unless an error is encountered.
+ *
+ * @param[in] dev
+ * Pointer to Ethernet device structure.
+ * @param[in] actions
+ * Pointer to flow actions array.
+ * @param[in, out] flow
+ * Pointer to the rte_flow structure.
+ * @param[in] flow_size
+ * Size in bytes of the available space in @p flow, if too small some
+ * garbage may be present.
+ * @param[out] error
+ * Pointer to error structure.
+ *
+ * @return
+ * On success the number of bytes consumed/necessary, if the returned value
+ * is lesser or equal to @p flow_size, the @p actions has fully been
+ * converted, otherwise another call with this returned memory size should
+ * be done.
+ * On error, a negative errno value is returned and rte_errno is set.
+ */
+static int
+mlx5_flow_actions(struct rte_eth_dev *dev __rte_unused,
+ const struct rte_flow_action actions[],
+ struct rte_flow *flow, const size_t flow_size,
+ struct rte_flow_error *error)
+{
+ size_t size = 0;
+ int remain = flow_size;
+ int ret = 0;
+
+ for (; actions->type != RTE_FLOW_ACTION_TYPE_END; actions++) {
+ switch (actions->type) {
+ case RTE_FLOW_ACTION_TYPE_VOID:
+ break;
+ case RTE_FLOW_ACTION_TYPE_DROP:
+ ret = mlx5_flow_action_drop(actions, flow, remain,
+ error);
+ break;
+ default:
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_ACTION,
+ actions,
+ "action not supported");
+ }
+ if (ret < 0)
+ return ret;
+ if (remain > ret)
+ remain -= ret;
+ else
+ remain = 0;
+ size += ret;
+ }
+ if (!flow->fate)
+ return rte_flow_error_set(error, ENOTSUP,
+ RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
+ NULL,
+ "no fate action found");
+ return size;
+}
+
+/**
+ * Convert the @p attributes, @p pattern, @p action, into an flow for the NIC
+ * after ensuring the NIC will understand and process it correctly.
+ * The conversion is only performed item/action per item/action, each of
+ * them is written into the @p flow if its size is lesser or equal to @p
+ * flow_size.
+ * Validation and memory consumption computation are still performed until the
+ * end, unless an error is encountered.
+ *
+ * @param[in] dev
+ * Pointer to Ethernet device.
+ * @param[in, out] flow
+ * Pointer to flow structure.
+ * @param[in] flow_size
+ * Size in bytes of the available space in @p flow, if too small some
+ * garbage may be present.
+ * @param[in] attributes