X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=lib%2Fflow.c;h=4d472312b4511fe93c7228ff9c67f76f07affde7;hb=a00ca915ff30a5cb7372cdf145205e339873cd4a;hp=2d62a12d96f676a5f31815b519c5ec75ac97a156;hpb=572b70687b27ef6fdd1ed90d1ac932d6de35701f;p=sliver-openvswitch.git diff --git a/lib/flow.c b/lib/flow.c index 2d62a12d9..4d472312b 100644 --- a/lib/flow.c +++ b/lib/flow.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2008, 2009, 2010, 2011 Nicira Networks. + * Copyright (c) 2008, 2009, 2010, 2011, 2012 Nicira Networks. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. @@ -148,7 +148,9 @@ parse_ipv6(struct ofpbuf *packet, struct flow *flow) flow->ipv6_dst = nh->ip6_dst; tc_flow = get_unaligned_be32(&nh->ip6_flow); - flow->nw_tos = (ntohl(tc_flow) >> 4) & IP_DSCP_MASK; + flow->nw_tos = ntohl(tc_flow) >> 20; + flow->ipv6_label = tc_flow & htonl(IPV6_LABEL_MASK); + flow->nw_ttl = nh->ip6_hlim; flow->nw_proto = IPPROTO_NONE; while (1) { @@ -201,9 +203,14 @@ parse_ipv6(struct ofpbuf *packet, struct flow *flow) } /* We only process the first fragment. */ - if ((frag_hdr->ip6f_offlg & IP6F_OFF_MASK) != htons(0)) { - nexthdr = IPPROTO_FRAGMENT; - break; + if (frag_hdr->ip6f_offlg != htons(0)) { + if ((frag_hdr->ip6f_offlg & IP6F_OFF_MASK) == htons(0)) { + flow->nw_frag = FLOW_NW_FRAG_ANY; + } else { + flow->nw_frag |= FLOW_NW_FRAG_LATER; + nexthdr = IPPROTO_FRAGMENT; + break; + } } } } @@ -245,8 +252,8 @@ parse_icmpv6(struct ofpbuf *b, struct flow *flow) /* The ICMPv6 type and code fields use the 16-bit transport port * fields, so we need to store them in 16-bit network byte order. */ - flow->icmp_type = htons(icmp->icmp6_type); - flow->icmp_code = htons(icmp->icmp6_code); + flow->tp_src = htons(icmp->icmp6_type); + flow->tp_dst = htons(icmp->icmp6_code); if (icmp->icmp6_code == 0 && (icmp->icmp6_type == ND_NEIGHBOR_SOLICIT || @@ -305,7 +312,9 @@ invalid: } -/* Initializes 'flow' members from 'packet', 'tun_id', and 'ofp_in_port'. +/* Initializes 'flow' members from 'packet', 'skb_priority', 'tun_id', and + * 'ofp_in_port'. + * * Initializes 'packet' header pointers as follows: * * - packet->l2 to the start of the Ethernet header. @@ -320,19 +329,19 @@ invalid: * - packet->l7 to just past the TCP or UDP or ICMP header, if one is * present and has a correct length, and otherwise NULL. */ -int -flow_extract(struct ofpbuf *packet, ovs_be64 tun_id, uint16_t ofp_in_port, - struct flow *flow) +void +flow_extract(struct ofpbuf *packet, uint32_t skb_priority, ovs_be64 tun_id, + uint16_t ofp_in_port, struct flow *flow) { struct ofpbuf b = *packet; struct eth_header *eth; - int retval = 0; COVERAGE_INC(flow_extract); memset(flow, 0, sizeof *flow); flow->tun_id = tun_id; flow->in_port = ofp_in_port; + flow->skb_priority = skb_priority; packet->l2 = b.data; packet->l3 = NULL; @@ -340,7 +349,7 @@ flow_extract(struct ofpbuf *packet, ovs_be64 tun_id, uint16_t ofp_in_port, packet->l7 = NULL; if (b.size < sizeof *eth) { - return 0; + return; } /* Link layer. */ @@ -360,12 +369,22 @@ flow_extract(struct ofpbuf *packet, ovs_be64 tun_id, uint16_t ofp_in_port, if (flow->dl_type == htons(ETH_TYPE_IP)) { const struct ip_header *nh = pull_ip(&b); if (nh) { + packet->l4 = b.data; + flow->nw_src = get_unaligned_be32(&nh->ip_src); flow->nw_dst = get_unaligned_be32(&nh->ip_dst); - flow->nw_tos = nh->ip_tos & IP_DSCP_MASK; flow->nw_proto = nh->ip_proto; - packet->l4 = b.data; - if (!IP_IS_FRAGMENT(nh->ip_frag_off)) { + + flow->nw_tos = nh->ip_tos; + if (IP_IS_FRAGMENT(nh->ip_frag_off)) { + flow->nw_frag = FLOW_NW_FRAG_ANY; + if (nh->ip_frag_off & htons(IP_FRAG_OFF_MASK)) { + flow->nw_frag |= FLOW_NW_FRAG_LATER; + } + } + flow->nw_ttl = nh->ip_ttl; + + if (!(nh->ip_frag_off & htons(IP_FRAG_OFF_MASK))) { if (flow->nw_proto == IPPROTO_TCP) { parse_tcp(packet, &b, flow); } else if (flow->nw_proto == IPPROTO_UDP) { @@ -373,20 +392,16 @@ flow_extract(struct ofpbuf *packet, ovs_be64 tun_id, uint16_t ofp_in_port, } else if (flow->nw_proto == IPPROTO_ICMP) { const struct icmp_header *icmp = pull_icmp(&b); if (icmp) { - flow->icmp_type = htons(icmp->icmp_type); - flow->icmp_code = htons(icmp->icmp_code); + flow->tp_src = htons(icmp->icmp_type); + flow->tp_dst = htons(icmp->icmp_code); packet->l7 = b.data; } } - } else { - retval = 1; } } } else if (flow->dl_type == htons(ETH_TYPE_IPV6)) { - - retval = parse_ipv6(&b, flow); - if (retval) { - return 0; + if (parse_ipv6(&b, flow)) { + return; } packet->l4 = b.data; @@ -419,8 +434,6 @@ flow_extract(struct ofpbuf *packet, ovs_be64 tun_id, uint16_t ofp_in_port, } } } - - return retval; } /* For every bit of a field that is wildcarded in 'wildcards', sets the @@ -431,7 +444,7 @@ flow_zero_wildcards(struct flow *flow, const struct flow_wildcards *wildcards) const flow_wildcards_t wc = wildcards->wildcards; int i; - BUILD_ASSERT_DECL(FLOW_WC_SEQ == 1); + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 9); for (i = 0; i < FLOW_N_REGS; i++) { flow->regs[i] &= wildcards->reg_masks[i]; @@ -444,14 +457,10 @@ flow_zero_wildcards(struct flow *flow, const struct flow_wildcards *wildcards) } flow->vlan_tci &= wildcards->vlan_tci_mask; if (wc & FWW_DL_TYPE) { - flow->dl_type = 0; - } - if (wc & FWW_TP_SRC) { - flow->tp_src = 0; - } - if (wc & FWW_TP_DST) { - flow->tp_dst = 0; + flow->dl_type = htons(0); } + flow->tp_src &= wildcards->tp_src_mask; + flow->tp_dst &= wildcards->tp_dst_mask; if (wc & FWW_DL_SRC) { memset(flow->dl_src, 0, sizeof flow->dl_src); } @@ -465,9 +474,19 @@ flow_zero_wildcards(struct flow *flow, const struct flow_wildcards *wildcards) if (wc & FWW_NW_PROTO) { flow->nw_proto = 0; } - if (wc & FWW_NW_TOS) { - flow->nw_tos = 0; + if (wc & FWW_IPV6_LABEL) { + flow->ipv6_label = htonl(0); + } + if (wc & FWW_NW_DSCP) { + flow->nw_tos &= ~IP_DSCP_MASK; + } + if (wc & FWW_NW_ECN) { + flow->nw_tos &= ~IP_ECN_MASK; } + if (wc & FWW_NW_TTL) { + flow->nw_ttl = 0; + } + flow->nw_frag &= wildcards->nw_frag_mask; if (wc & FWW_ARP_SHA) { memset(flow->arp_sha, 0, sizeof flow->arp_sha); } @@ -481,6 +500,22 @@ flow_zero_wildcards(struct flow *flow, const struct flow_wildcards *wildcards) if (wc & FWW_ND_TARGET) { memset(&flow->nd_target, 0, sizeof flow->nd_target); } + flow->skb_priority = 0; +} + +/* Initializes 'fmd' with the metadata found in 'flow'. */ +void +flow_get_metadata(const struct flow *flow, struct flow_metadata *fmd) +{ + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 9); + + fmd->tun_id = flow->tun_id; + fmd->tun_id_mask = htonll(UINT64_MAX); + + memcpy(fmd->regs, flow->regs, sizeof fmd->regs); + memset(fmd->reg_masks, 0xff, sizeof fmd->reg_masks); + + fmd->in_port = flow->in_port; } char * @@ -494,43 +529,54 @@ flow_to_string(const struct flow *flow) void flow_format(struct ds *ds, const struct flow *flow) { - ds_put_format(ds, "tunnel%#"PRIx64":in_port%04"PRIx16":tci(", - ntohll(flow->tun_id), flow->in_port); + ds_put_format(ds, "priority:%"PRIu32 + ",tunnel:%#"PRIx64 + ",in_port:%04"PRIx16, + flow->skb_priority, + ntohll(flow->tun_id), + flow->in_port); + + ds_put_format(ds, ",tci("); if (flow->vlan_tci) { - ds_put_format(ds, "vlan%"PRIu16",pcp%d", + ds_put_format(ds, "vlan:%"PRIu16",pcp:%d", vlan_tci_to_vid(flow->vlan_tci), vlan_tci_to_pcp(flow->vlan_tci)); } else { ds_put_char(ds, '0'); } - ds_put_format(ds, ") mac"ETH_ADDR_FMT"->"ETH_ADDR_FMT - " type%04"PRIx16, + ds_put_format(ds, ") mac("ETH_ADDR_FMT"->"ETH_ADDR_FMT + ") type:%04"PRIx16, ETH_ADDR_ARGS(flow->dl_src), ETH_ADDR_ARGS(flow->dl_dst), ntohs(flow->dl_type)); if (flow->dl_type == htons(ETH_TYPE_IPV6)) { - ds_put_format(ds, " proto%"PRIu8" tos%"PRIu8" ipv6", - flow->nw_proto, flow->nw_tos); + ds_put_format(ds, " label:%#"PRIx32" proto:%"PRIu8" tos:%#"PRIx8 + " ttl:%"PRIu8" ipv6(", + ntohl(flow->ipv6_label), flow->nw_proto, + flow->nw_tos, flow->nw_ttl); print_ipv6_addr(ds, &flow->ipv6_src); ds_put_cstr(ds, "->"); print_ipv6_addr(ds, &flow->ipv6_dst); - + ds_put_char(ds, ')'); } else { - ds_put_format(ds, " proto%"PRIu8 - " tos%"PRIu8 - " ip"IP_FMT"->"IP_FMT, - flow->nw_proto, - flow->nw_tos, - IP_ARGS(&flow->nw_src), - IP_ARGS(&flow->nw_dst)); + ds_put_format(ds, " proto:%"PRIu8" tos:%#"PRIx8" ttl:%"PRIu8 + " ip("IP_FMT"->"IP_FMT")", + flow->nw_proto, flow->nw_tos, flow->nw_ttl, + IP_ARGS(&flow->nw_src), IP_ARGS(&flow->nw_dst)); + } + if (flow->nw_frag) { + ds_put_format(ds, " frag(%s)", + flow->nw_frag == FLOW_NW_FRAG_ANY ? "first" + : flow->nw_frag == (FLOW_NW_FRAG_ANY | FLOW_NW_FRAG_LATER) + ? "later" : ""); } if (flow->tp_src || flow->tp_dst) { - ds_put_format(ds, " port%"PRIu16"->%"PRIu16, + ds_put_format(ds, " port(%"PRIu16"->%"PRIu16")", ntohs(flow->tp_src), ntohs(flow->tp_dst)); } if (!eth_addr_is_zero(flow->arp_sha) || !eth_addr_is_zero(flow->arp_tha)) { - ds_put_format(ds, " arp_ha"ETH_ADDR_FMT"->"ETH_ADDR_FMT, + ds_put_format(ds, " arp_ha("ETH_ADDR_FMT"->"ETH_ADDR_FMT")", ETH_ADDR_ARGS(flow->arp_sha), ETH_ADDR_ARGS(flow->arp_tha)); } @@ -550,6 +596,8 @@ flow_print(FILE *stream, const struct flow *flow) void flow_wildcards_init_catchall(struct flow_wildcards *wc) { + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 9); + wc->wildcards = FWW_ALL; wc->tun_id_mask = htonll(0); wc->nw_src_mask = htonl(0); @@ -558,7 +606,10 @@ flow_wildcards_init_catchall(struct flow_wildcards *wc) wc->ipv6_dst_mask = in6addr_any; memset(wc->reg_masks, 0, sizeof wc->reg_masks); wc->vlan_tci_mask = htons(0); - wc->zero = 0; + wc->nw_frag_mask = 0; + wc->tp_src_mask = htons(0); + wc->tp_dst_mask = htons(0); + memset(wc->zeros, 0, sizeof wc->zeros); } /* Initializes 'wc' as an exact-match set of wildcards; that is, 'wc' does not @@ -566,6 +617,8 @@ flow_wildcards_init_catchall(struct flow_wildcards *wc) void flow_wildcards_init_exact(struct flow_wildcards *wc) { + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 9); + wc->wildcards = 0; wc->tun_id_mask = htonll(UINT64_MAX); wc->nw_src_mask = htonl(UINT32_MAX); @@ -574,7 +627,10 @@ flow_wildcards_init_exact(struct flow_wildcards *wc) wc->ipv6_dst_mask = in6addr_exact; memset(wc->reg_masks, 0xff, sizeof wc->reg_masks); wc->vlan_tci_mask = htons(UINT16_MAX); - wc->zero = 0; + wc->nw_frag_mask = UINT8_MAX; + wc->tp_src_mask = htons(UINT16_MAX); + wc->tp_dst_mask = htons(UINT16_MAX); + memset(wc->zeros, 0, sizeof wc->zeros); } /* Returns true if 'wc' is exact-match, false if 'wc' wildcards any bits or @@ -584,15 +640,18 @@ flow_wildcards_is_exact(const struct flow_wildcards *wc) { int i; - BUILD_ASSERT_DECL(FLOW_WC_SEQ == 1); + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 9); if (wc->wildcards || wc->tun_id_mask != htonll(UINT64_MAX) || wc->nw_src_mask != htonl(UINT32_MAX) || wc->nw_dst_mask != htonl(UINT32_MAX) + || wc->tp_src_mask != htons(UINT16_MAX) + || wc->tp_dst_mask != htons(UINT16_MAX) || wc->vlan_tci_mask != htons(UINT16_MAX) || !ipv6_mask_is_exact(&wc->ipv6_src_mask) - || !ipv6_mask_is_exact(&wc->ipv6_dst_mask)) { + || !ipv6_mask_is_exact(&wc->ipv6_dst_mask) + || wc->nw_frag_mask != UINT8_MAX) { return false; } @@ -612,15 +671,18 @@ flow_wildcards_is_catchall(const struct flow_wildcards *wc) { int i; - BUILD_ASSERT_DECL(FLOW_WC_SEQ == 1); + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 9); if (wc->wildcards != FWW_ALL || wc->tun_id_mask != htonll(0) || wc->nw_src_mask != htonl(0) || wc->nw_dst_mask != htonl(0) + || wc->tp_src_mask != htons(0) + || wc->tp_dst_mask != htons(0) || wc->vlan_tci_mask != htons(0) || !ipv6_mask_is_any(&wc->ipv6_src_mask) - || !ipv6_mask_is_any(&wc->ipv6_dst_mask)) { + || !ipv6_mask_is_any(&wc->ipv6_dst_mask) + || wc->nw_frag_mask != 0) { return false; } @@ -643,6 +705,8 @@ flow_wildcards_combine(struct flow_wildcards *dst, { int i; + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 9); + dst->wildcards = src1->wildcards | src2->wildcards; dst->tun_id_mask = src1->tun_id_mask & src2->tun_id_mask; dst->nw_src_mask = src1->nw_src_mask & src2->nw_src_mask; @@ -655,6 +719,8 @@ flow_wildcards_combine(struct flow_wildcards *dst, dst->reg_masks[i] = src1->reg_masks[i] & src2->reg_masks[i]; } dst->vlan_tci_mask = src1->vlan_tci_mask & src2->vlan_tci_mask; + dst->tp_src_mask = src1->tp_src_mask & src2->tp_src_mask; + dst->tp_dst_mask = src1->tp_dst_mask & src2->tp_dst_mask; } /* Returns a hash of the wildcards in 'wc'. */ @@ -664,7 +730,7 @@ flow_wildcards_hash(const struct flow_wildcards *wc, uint32_t basis) /* If you change struct flow_wildcards and thereby trigger this * assertion, please check that the new struct flow_wildcards has no holes * in it before you update the assertion. */ - BUILD_ASSERT_DECL(sizeof *wc == 56 + FLOW_N_REGS * 4); + BUILD_ASSERT_DECL(sizeof *wc == 64 + FLOW_N_REGS * 4); return hash_bytes(wc, sizeof *wc, basis); } @@ -676,13 +742,17 @@ flow_wildcards_equal(const struct flow_wildcards *a, { int i; + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 9); + if (a->wildcards != b->wildcards || a->tun_id_mask != b->tun_id_mask || a->nw_src_mask != b->nw_src_mask || a->nw_dst_mask != b->nw_dst_mask || a->vlan_tci_mask != b->vlan_tci_mask || !ipv6_addr_equals(&a->ipv6_src_mask, &b->ipv6_src_mask) - || !ipv6_addr_equals(&a->ipv6_dst_mask, &b->ipv6_dst_mask)) { + || !ipv6_addr_equals(&a->ipv6_dst_mask, &b->ipv6_dst_mask) + || a->tp_src_mask != b->tp_src_mask + || a->tp_dst_mask != b->tp_dst_mask) { return false; } @@ -704,6 +774,8 @@ flow_wildcards_has_extra(const struct flow_wildcards *a, int i; struct in6_addr ipv6_masked; + BUILD_ASSERT_DECL(FLOW_WC_SEQ == 9); + for (i = 0; i < FLOW_N_REGS; i++) { if ((a->reg_masks[i] & b->reg_masks[i]) != b->reg_masks[i]) { return true; @@ -724,67 +796,9 @@ flow_wildcards_has_extra(const struct flow_wildcards *a, || (a->tun_id_mask & b->tun_id_mask) != b->tun_id_mask || (a->nw_src_mask & b->nw_src_mask) != b->nw_src_mask || (a->nw_dst_mask & b->nw_dst_mask) != b->nw_dst_mask - || (a->vlan_tci_mask & b->vlan_tci_mask) != b->vlan_tci_mask); -} - -static bool -set_nw_mask(ovs_be32 *maskp, ovs_be32 mask) -{ - if (ip_is_cidr(mask)) { - *maskp = mask; - return true; - } else { - return false; - } -} - -/* Sets the IP (or ARP) source wildcard mask to CIDR 'mask' (consisting of N - * high-order 1-bit and 32-N low-order 0-bits). Returns true if successful, - * false if 'mask' is not a CIDR mask. */ -bool -flow_wildcards_set_nw_src_mask(struct flow_wildcards *wc, ovs_be32 mask) -{ - return set_nw_mask(&wc->nw_src_mask, mask); -} - -/* Sets the IP (or ARP) destination wildcard mask to CIDR 'mask' (consisting of - * N high-order 1-bit and 32-N low-order 0-bits). Returns true if successful, - * false if 'mask' is not a CIDR mask. */ -bool -flow_wildcards_set_nw_dst_mask(struct flow_wildcards *wc, ovs_be32 mask) -{ - return set_nw_mask(&wc->nw_dst_mask, mask); -} - -static bool -set_ipv6_mask(struct in6_addr *maskp, const struct in6_addr *mask) -{ - if (ipv6_is_cidr(mask)) { - *maskp = *mask; - return true; - } else { - return false; - } -} - -/* Sets the IPv6 source wildcard mask to CIDR 'mask' (consisting of N - * high-order 1-bit and 128-N low-order 0-bits). Returns true if successful, - * false if 'mask' is not a CIDR mask. */ -bool -flow_wildcards_set_ipv6_src_mask(struct flow_wildcards *wc, - const struct in6_addr *mask) -{ - return set_ipv6_mask(&wc->ipv6_src_mask, mask); -} - -/* Sets the IPv6 destination wildcard mask to CIDR 'mask' (consisting of - * N high-order 1-bit and 128-N low-order 0-bits). Returns true if - * successful, false if 'mask' is not a CIDR mask. */ -bool -flow_wildcards_set_ipv6_dst_mask(struct flow_wildcards *wc, - const struct in6_addr *mask) -{ - return set_ipv6_mask(&wc->ipv6_dst_mask, mask); + || (a->vlan_tci_mask & b->vlan_tci_mask) != b->vlan_tci_mask + || (a->tp_src_mask & b->tp_src_mask) != b->tp_src_mask + || (a->tp_dst_mask & b->tp_dst_mask) != b->tp_dst_mask); } /* Sets the wildcard mask for register 'idx' in 'wc' to 'mask'. @@ -883,7 +897,7 @@ flow_hash_symmetric_l4(const struct flow *flow, uint32_t basis) }; ovs_be16 eth_type; ovs_be16 vlan_tci; - ovs_be16 tp_addr; + ovs_be16 tp_port; uint8_t eth_addr[ETH_ADDR_LEN]; uint8_t ip_proto; } fields; @@ -903,7 +917,7 @@ flow_hash_symmetric_l4(const struct flow *flow, uint32_t basis) fields.ipv4_addr = flow->nw_src ^ flow->nw_dst; fields.ip_proto = flow->nw_proto; if (fields.ip_proto == IPPROTO_TCP) { - fields.tp_addr = flow->tp_src ^ flow->tp_dst; + fields.tp_port = flow->tp_src ^ flow->tp_dst; } } else if (fields.eth_type == htons(ETH_TYPE_IPV6)) { const uint8_t *a = &flow->ipv6_src.s6_addr[0]; @@ -915,7 +929,7 @@ flow_hash_symmetric_l4(const struct flow *flow, uint32_t basis) } fields.ip_proto = flow->nw_proto; if (fields.ip_proto == IPPROTO_TCP) { - fields.tp_addr = flow->tp_src ^ flow->tp_dst; + fields.tp_port = flow->tp_src ^ flow->tp_dst; } } return hash_bytes(&fields, sizeof fields, basis); @@ -957,6 +971,44 @@ flow_hash_fields_valid(enum nx_hash_fields fields) || fields == NX_HASH_FIELDS_SYMMETRIC_L4; } +/* Sets the VLAN VID that 'flow' matches to 'vid', which is interpreted as an + * OpenFlow 1.0 "dl_vlan" value: + * + * - If it is in the range 0...4095, 'flow->vlan_tci' is set to match + * that VLAN. Any existing PCP match is unchanged (it becomes 0 if + * 'flow' previously matched packets without a VLAN header). + * + * - If it is OFP_VLAN_NONE, 'flow->vlan_tci' is set to match a packet + * without a VLAN tag. + * + * - Other values of 'vid' should not be used. */ +void +flow_set_vlan_vid(struct flow *flow, ovs_be16 vid) +{ + if (vid == htons(OFP_VLAN_NONE)) { + flow->vlan_tci = htons(0); + } else { + vid &= htons(VLAN_VID_MASK); + flow->vlan_tci &= ~htons(VLAN_VID_MASK); + flow->vlan_tci |= htons(VLAN_CFI) | vid; + } +} + +/* Sets the VLAN PCP that 'flow' matches to 'pcp', which should be in the + * range 0...7. + * + * This function has no effect on the VLAN ID that 'flow' matches. + * + * After calling this function, 'flow' will not match packets without a VLAN + * header. */ +void +flow_set_vlan_pcp(struct flow *flow, uint8_t pcp) +{ + pcp &= 0x07; + flow->vlan_tci &= ~htons(VLAN_PCP_MASK); + flow->vlan_tci |= htons((pcp << VLAN_PCP_SHIFT) | VLAN_CFI); +} + /* Puts into 'b' a packet that flow_extract() would parse as having the given * 'flow'. * @@ -974,7 +1026,7 @@ flow_compose(struct ofpbuf *b, const struct flow *flow) } if (flow->vlan_tci & htons(VLAN_CFI)) { - eth_push_vlan(b, flow->vlan_tci & ~htons(VLAN_CFI)); + eth_push_vlan(b, flow->vlan_tci); } if (flow->dl_type == htons(ETH_TYPE_IP)) { @@ -987,25 +1039,38 @@ flow_compose(struct ofpbuf *b, const struct flow *flow) ip->ip_src = flow->nw_src; ip->ip_dst = flow->nw_dst; - if (flow->nw_proto == IPPROTO_TCP) { - struct tcp_header *tcp; - - b->l4 = tcp = ofpbuf_put_zeros(b, sizeof *tcp); - tcp->tcp_src = flow->tp_src; - tcp->tcp_dst = flow->tp_dst; - } else if (flow->nw_proto == IPPROTO_UDP) { - struct udp_header *udp; - - b->l4 = udp = ofpbuf_put_zeros(b, sizeof *udp); - udp->udp_src = flow->tp_src; - udp->udp_dst = flow->tp_dst; - } else if (flow->nw_proto == IPPROTO_ICMP) { - struct icmp_header *icmp; - - b->l4 = icmp = ofpbuf_put_zeros(b, sizeof *icmp); - icmp->icmp_type = ntohs(flow->tp_src); - icmp->icmp_code = ntohs(flow->tp_dst); + if (flow->nw_frag & FLOW_NW_FRAG_ANY) { + ip->ip_frag_off |= htons(IP_MORE_FRAGMENTS); + if (flow->nw_frag & FLOW_NW_FRAG_LATER) { + ip->ip_frag_off |= htons(100); + } } + if (!(flow->nw_frag & FLOW_NW_FRAG_ANY) + || !(flow->nw_frag & FLOW_NW_FRAG_LATER)) { + if (flow->nw_proto == IPPROTO_TCP) { + struct tcp_header *tcp; + + b->l4 = tcp = ofpbuf_put_zeros(b, sizeof *tcp); + tcp->tcp_src = flow->tp_src; + tcp->tcp_dst = flow->tp_dst; + tcp->tcp_ctl = TCP_CTL(0, 5); + } else if (flow->nw_proto == IPPROTO_UDP) { + struct udp_header *udp; + + b->l4 = udp = ofpbuf_put_zeros(b, sizeof *udp); + udp->udp_src = flow->tp_src; + udp->udp_dst = flow->tp_dst; + } else if (flow->nw_proto == IPPROTO_ICMP) { + struct icmp_header *icmp; + + b->l4 = icmp = ofpbuf_put_zeros(b, sizeof *icmp); + icmp->icmp_type = ntohs(flow->tp_src); + icmp->icmp_code = ntohs(flow->tp_dst); + } + } + + ip->ip_tot_len = htons((uint8_t *) b->data + b->size + - (uint8_t *) b->l3); } else if (flow->dl_type == htons(ETH_TYPE_IPV6)) { /* XXX */ } else if (flow->dl_type == htons(ETH_TYPE_ARP)) {