/*
- * Copyright (c) 2008, 2009, 2010, 2011, 2012, 2013 Nicira, Inc.
+ * Copyright (c) 2008, 2009, 2010, 2011, 2012, 2013, 2014 Nicira, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
struct dpif_flow_stats;
struct ds;
struct flow_wildcards;
-struct miniflow;
struct minimask;
struct ofpbuf;
/* This sequence number should be incremented whenever anything involving flows
* or the wildcarding of flows changes. This will cause build assertion
* failures in places which likely need to be updated. */
-#define FLOW_WC_SEQ 21
+#define FLOW_WC_SEQ 24
#define FLOW_N_REGS 8
BUILD_ASSERT_DECL(FLOW_N_REGS <= NXM_NX_MAX_REGS);
odp_port_t odp_port;
};
+/* Maximum number of supported MPLS labels. */
+#define FLOW_MAX_MPLS_LABELS 3
+
/*
-* A flow in the network.
-*
-* The meaning of 'in_port' is context-dependent. In most cases, it is a
-* 16-bit OpenFlow 1.0 port number. In the software datapath interface (dpif)
-* layer and its implementations (e.g. dpif-linux, dpif-netdev), it is instead
-* a 32-bit datapath port number.
-*/
+ * A flow in the network.
+ *
+ * Must be initialized to all zeros to make any compiler-induced padding
+ * zeroed. Helps also in keeping unused fields (such as mutually exclusive
+ * IPv4 and IPv6 addresses) zeroed out.
+ *
+ * The meaning of 'in_port' is context-dependent. In most cases, it is a
+ * 16-bit OpenFlow 1.0 port number. In the software datapath interface (dpif)
+ * layer and its implementations (e.g. dpif-linux, dpif-netdev), it is instead
+ * a 32-bit datapath port number.
+ *
+ * The fields are organized in four segments to facilitate staged lookup, where
+ * lower layer fields are first used to determine if the later fields need to
+ * be looked at. This enables better wildcarding for datapath flows.
+ */
struct flow {
+ /* L1 */
struct flow_tnl tunnel; /* Encapsulating tunnel parameters. */
ovs_be64 metadata; /* OpenFlow Metadata. */
+ uint32_t regs[FLOW_N_REGS]; /* Registers. */
+ uint32_t skb_priority; /* Packet priority for QoS. */
+ uint32_t pkt_mark; /* Packet mark. */
+ union flow_in_port in_port; /* Input port.*/
+
+ /* L2 */
+ uint8_t dl_src[6]; /* Ethernet source address. */
+ uint8_t dl_dst[6]; /* Ethernet destination address. */
+ ovs_be16 dl_type; /* Ethernet frame type. */
+ ovs_be16 vlan_tci; /* If 802.1Q, TCI | VLAN_CFI; otherwise 0. */
+ ovs_be32 mpls_lse[FLOW_MAX_MPLS_LABELS]; /* MPLS label stack entry. */
+
+ /* L3 */
struct in6_addr ipv6_src; /* IPv6 source address. */
struct in6_addr ipv6_dst; /* IPv6 destination address. */
struct in6_addr nd_target; /* IPv6 neighbor discovery (ND) target. */
- uint32_t skb_priority; /* Packet priority for QoS. */
- uint32_t regs[FLOW_N_REGS]; /* Registers. */
+ ovs_be32 ipv6_label; /* IPv6 flow label. */
ovs_be32 nw_src; /* IPv4 source address. */
ovs_be32 nw_dst; /* IPv4 destination address. */
- ovs_be32 ipv6_label; /* IPv6 flow label. */
- union flow_in_port in_port; /* Input port.*/
- uint32_t pkt_mark; /* Packet mark. */
- ovs_be32 mpls_lse; /* MPLS label stack entry. */
- ovs_be16 vlan_tci; /* If 802.1Q, TCI | VLAN_CFI; otherwise 0. */
- ovs_be16 dl_type; /* Ethernet frame type. */
- ovs_be16 tp_src; /* TCP/UDP/SCTP source port. */
- ovs_be16 tp_dst; /* TCP/UDP/SCTP destination port. */
- uint8_t dl_src[6]; /* Ethernet source address. */
- uint8_t dl_dst[6]; /* Ethernet destination address. */
- uint8_t nw_proto; /* IP protocol or low 8 bits of ARP opcode. */
+ uint8_t nw_frag; /* FLOW_FRAG_* flags. */
uint8_t nw_tos; /* IP ToS (including DSCP and ECN). */
+ uint8_t nw_ttl; /* IP TTL/Hop Limit. */
+ uint8_t nw_proto; /* IP protocol or low 8 bits of ARP opcode. */
uint8_t arp_sha[6]; /* ARP/ND source hardware address. */
uint8_t arp_tha[6]; /* ARP/ND target hardware address. */
- uint8_t nw_ttl; /* IP TTL/Hop Limit. */
- uint8_t nw_frag; /* FLOW_FRAG_* flags. */
+ ovs_be16 tcp_flags; /* TCP flags. With L3 to avoid matching L4. */
+ ovs_be16 pad; /* Padding. */
+
+ /* L4 */
+ ovs_be16 tp_src; /* TCP/UDP/SCTP source port. */
+ ovs_be16 tp_dst; /* TCP/UDP/SCTP destination port.
+ * Keep last for the BUILD_ASSERT_DECL below */
};
BUILD_ASSERT_DECL(sizeof(struct flow) % 4 == 0);
#define FLOW_U32S (sizeof(struct flow) / 4)
/* Remember to update FLOW_WC_SEQ when changing 'struct flow'. */
-BUILD_ASSERT_DECL(sizeof(struct flow) == sizeof(struct flow_tnl) + 152 &&
- FLOW_WC_SEQ == 21);
+BUILD_ASSERT_DECL(offsetof(struct flow, tp_dst) + 2
+ == sizeof(struct flow_tnl) + 164
+ && FLOW_WC_SEQ == 24);
+
+/* Incremental points at which flow classification may be performed in
+ * segments.
+ * This is located here since this is dependent on the structure of the
+ * struct flow defined above:
+ * Each offset must be on a distint, successive U32 boundary srtictly
+ * within the struct flow. */
+enum {
+ FLOW_SEGMENT_1_ENDS_AT = offsetof(struct flow, dl_src),
+ FLOW_SEGMENT_2_ENDS_AT = offsetof(struct flow, ipv6_src),
+ FLOW_SEGMENT_3_ENDS_AT = offsetof(struct flow, tp_src),
+};
+BUILD_ASSERT_DECL(FLOW_SEGMENT_1_ENDS_AT % 4 == 0);
+BUILD_ASSERT_DECL(FLOW_SEGMENT_2_ENDS_AT % 4 == 0);
+BUILD_ASSERT_DECL(FLOW_SEGMENT_3_ENDS_AT % 4 == 0);
+BUILD_ASSERT_DECL( 0 < FLOW_SEGMENT_1_ENDS_AT);
+BUILD_ASSERT_DECL(FLOW_SEGMENT_1_ENDS_AT < FLOW_SEGMENT_2_ENDS_AT);
+BUILD_ASSERT_DECL(FLOW_SEGMENT_2_ENDS_AT < FLOW_SEGMENT_3_ENDS_AT);
+BUILD_ASSERT_DECL(FLOW_SEGMENT_3_ENDS_AT < sizeof(struct flow));
+
+extern const uint8_t flow_segment_u32s[];
/* Represents the metadata fields of struct flow. */
struct flow_metadata {
struct flow *);
void flow_zero_wildcards(struct flow *, const struct flow_wildcards *);
+void flow_unwildcard_tp_ports(const struct flow *, struct flow_wildcards *);
void flow_get_metadata(const struct flow *, struct flow_metadata *);
char *flow_to_string(const struct flow *);
void format_flags(struct ds *ds, const char *(*bit_to_string)(uint32_t),
uint32_t flags, char del);
+void format_flags_masked(struct ds *ds, const char *name,
+ const char *(*bit_to_string)(uint32_t),
+ uint32_t flags, uint32_t mask);
void flow_format(struct ds *, const struct flow *);
void flow_print(FILE *, const struct flow *);
void flow_set_vlan_vid(struct flow *, ovs_be16 vid);
void flow_set_vlan_pcp(struct flow *, uint8_t pcp);
-void flow_set_mpls_label(struct flow *flow, ovs_be32 label);
-void flow_set_mpls_ttl(struct flow *flow, uint8_t ttl);
-void flow_set_mpls_tc(struct flow *flow, uint8_t tc);
-void flow_set_mpls_bos(struct flow *flow, uint8_t stack);
+int flow_count_mpls_labels(const struct flow *, struct flow_wildcards *);
+int flow_count_common_mpls_labels(const struct flow *a, int an,
+ const struct flow *b, int bn,
+ struct flow_wildcards *wc);
+void flow_push_mpls(struct flow *, int n, ovs_be16 mpls_eth_type,
+ struct flow_wildcards *);
+bool flow_pop_mpls(struct flow *, int n, ovs_be16 eth_type,
+ struct flow_wildcards *);
+void flow_set_mpls_label(struct flow *, int idx, ovs_be32 label);
+void flow_set_mpls_ttl(struct flow *, int idx, uint8_t ttl);
+void flow_set_mpls_tc(struct flow *, int idx, uint8_t tc);
+void flow_set_mpls_bos(struct flow *, int idx, uint8_t stack);
+void flow_set_mpls_lse(struct flow *, int idx, ovs_be32 lse);
void flow_compose(struct ofpbuf *, const struct flow *);
uint32_t flow_hash_in_minimask(const struct flow *, const struct minimask *,
uint32_t basis);
+uint32_t flow_hash_in_minimask_range(const struct flow *,
+ const struct minimask *,
+ uint8_t start, uint8_t end,
+ uint32_t *basis);
\f
/* Wildcards for a flow.
*
};
void flow_wildcards_init_catchall(struct flow_wildcards *);
-void flow_wildcards_init_exact(struct flow_wildcards *);
+
+void flow_wildcards_clear_non_packet_fields(struct flow_wildcards *);
bool flow_wildcards_is_catchall(const struct flow_wildcards *);
void flow_wildcards_fold_minimask(struct flow_wildcards *,
const struct minimask *);
+void flow_wildcards_fold_minimask_range(struct flow_wildcards *,
+ const struct minimask *,
+ uint8_t start, uint8_t end);
uint32_t flow_wildcards_hash(const struct flow_wildcards *, uint32_t basis);
bool flow_wildcards_equal(const struct flow_wildcards *,
const struct flow_wildcards *);
uint32_t flow_hash_symmetric_l4(const struct flow *flow, uint32_t basis);
+/* Initialize a flow with random fields that matter for nx_hash_fields. */
+void flow_random_hash_fields(struct flow *);
void flow_mask_hash_fields(const struct flow *, struct flow_wildcards *,
enum nx_hash_fields);
uint32_t flow_hash_fields(const struct flow *, enum nx_hash_fields,
/* Compressed flow. */
#define MINI_N_INLINE (sizeof(void *) == 4 ? 7 : 8)
-#define MINI_N_MAPS DIV_ROUND_UP(FLOW_U32S, 32)
+BUILD_ASSERT_DECL(FLOW_U32S <= 64);
/* A sparse representation of a "struct flow".
*
*
* Elements in 'values' are allowed to be zero. This is useful for "struct
* minimatch", for which ensuring that the miniflow and minimask members have
- * same 'map' allows optimization .
+ * same 'map' allows optimization. This allowance applies only to a miniflow
+ * that is not a mask. That is, a minimask may NOT have zero elements in
+ * its 'values'.
*/
struct miniflow {
+ uint64_t map;
uint32_t *values;
uint32_t inline_values[MINI_N_INLINE];
- uint32_t map[MINI_N_MAPS];
};
void miniflow_init(struct miniflow *, const struct flow *);
uint32_t miniflow_hash(const struct miniflow *, uint32_t basis);
uint32_t miniflow_hash_in_minimask(const struct miniflow *,
const struct minimask *, uint32_t basis);
+uint64_t miniflow_get_map_in_range(const struct miniflow *miniflow,
+ uint8_t start, uint8_t end,
+ unsigned int *offset);
+
\f
/* Compressed flow wildcards. */
/* A sparse representation of a "struct flow_wildcards".
*
- * See the large comment on struct miniflow for details. */
+ * See the large comment on struct miniflow for details.
+ *
+ * Note: While miniflow can have zero data for a 1-bit in the map,
+ * a minimask may not! We rely on this in the implementation. */
struct minimask {
struct miniflow masks;
};