diff options
Diffstat (limited to 'VNFs/DPPD-PROX/handle_aggregator.c')
-rw-r--r-- | VNFs/DPPD-PROX/handle_aggregator.c | 229 |
1 files changed, 229 insertions, 0 deletions
diff --git a/VNFs/DPPD-PROX/handle_aggregator.c b/VNFs/DPPD-PROX/handle_aggregator.c new file mode 100644 index 00000000..6434d759 --- /dev/null +++ b/VNFs/DPPD-PROX/handle_aggregator.c @@ -0,0 +1,229 @@ +/* +// Copyright (c) 2010-2017 Intel Corporation +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. +*/ + +#include <rte_ip.h> +#include <stdio.h> +#include <string.h> +#include <rte_version.h> + +#include "prox_lua.h" +#include "prox_lua_types.h" + +#include "lconf.h" +#include "task_base.h" +#include "task_init.h" +#include "defines.h" +#include "prefetch.h" +#include "qinq.h" +#include "prox_cfg.h" +#include "log.h" +#include "quit.h" +#include "prox_shared.h" +#include "mbuf_utils.h" +#include "handle_aggregator.h" + +#define PRIORITY_DHCP (HIGH_PRIORITY) + +#define TASK_STATS_ADD_DROP_TX_FAIL_PRIO(stats, ntx, prio) do { \ + (stats)->drop_tx_fail_prio[prio] += ntx; \ + } while(0) +#define TASK_STATS_ADD_TX_PRIO(stats, ntx, prio) do { \ + (stats)->rx_prio[prio] += ntx; \ + } while(0) \ + +static inline uint8_t detect_l4_priority(uint8_t l3_priority, const struct ipv4_hdr *ipv4_hdr) +{ + if (ipv4_hdr->next_proto_id == IPPROTO_UDP) { + const struct udp_hdr *udp = (const struct udp_hdr *)((const uint8_t *)ipv4_hdr + sizeof(struct ipv4_hdr)); + if (((udp->src_port == 0x67) && (udp->dst_port == 0x68)) || ((udp->src_port == 0x68) && (udp->dst_port == 0x67))) { + return PRIORITY_DHCP; + } + } + return l3_priority; +} + +static inline uint8_t detect_l3_priority(uint8_t l2_priority, const struct ipv4_hdr *ipv4_hdr) +{ + uint8_t dscp; + if ((ipv4_hdr->version_ihl >> 4) == 4) { + } else if ((ipv4_hdr->version_ihl >> 4) == 6) { + plog_warn("IPv6 Not implemented\n"); + return OUT_DISCARD; + } else { + plog_warn("Unexpected IP version\n"); + return OUT_DISCARD; + } + dscp = ipv4_hdr->type_of_service >> 2; + if (dscp) + return MAX_PRIORITIES - dscp - 1; + else + return l2_priority; +} + +static inline uint8_t detect_l2_priority(const struct qinq_hdr *pqinq) +{ + if (pqinq->cvlan.eth_proto != ETYPE_VLAN) { + plog_warn("Unexpected proto in QinQ = %#04x\n", pqinq->cvlan.eth_proto); + return OUT_DISCARD; + } + uint16_t svlan_priority = ntohs(pqinq->svlan.vlan_tci >> 13); + uint16_t cvlan_priority = ntohs(pqinq->cvlan.vlan_tci >> 13); + if (svlan_priority) + return svlan_priority; + else + return cvlan_priority; +} + +static inline void buffer_packet(struct task_aggregator *task, struct rte_mbuf *mbuf, uint8_t priority) +{ + struct task_base *tbase = (struct task_base *)task; + + struct task_buffer *prio = &task->priority[priority]; + if (prio->pkt_nb < BUFFER_LENGTH) { + prio->buffer[prio->pkt_pos] = mbuf; + prio->pkt_pos++; + if (prio->pkt_pos == BUFFER_LENGTH) + prio->pkt_pos = 0; + prio->pkt_nb++; + } else { + task->drop.buffer[task->drop.pkt_nb] = mbuf; + task->drop.pkt_nb++; + TASK_STATS_ADD_DROP_TX_FAIL_PRIO(&task->stats, 1, priority); + } +} + +static inline void handle_aggregator(struct task_aggregator *task, struct rte_mbuf *mbuf) +{ + struct ether_hdr *peth = rte_pktmbuf_mtod(mbuf, struct ether_hdr *); + uint8_t priority = 0; + const struct qinq_hdr *pqinq; + const struct ipv4_hdr *ipv4_hdr; + + const uint16_t eth_type = peth->ether_type; + switch (eth_type) { + case ETYPE_MPLSU: + case ETYPE_MPLSM: + break; + case ETYPE_8021ad: + pqinq = rte_pktmbuf_mtod(mbuf, const struct qinq_hdr *); + if ((priority = detect_l2_priority(pqinq)) == OUT_DISCARD) + break; + ipv4_hdr = (const struct ipv4_hdr *)(pqinq + 1); + if ((priority = detect_l3_priority(priority, ipv4_hdr)) == OUT_DISCARD) + break; + if ((priority = detect_l4_priority(priority, ipv4_hdr)) == OUT_DISCARD) + break; + break; + case ETYPE_VLAN: + break; + case ETYPE_IPv4: + ipv4_hdr = (const struct ipv4_hdr *)(peth+1); + if ((priority = detect_l3_priority(LOW_PRIORITY, ipv4_hdr)) == OUT_DISCARD) + break; + if ((priority = detect_l4_priority(priority, ipv4_hdr)) == OUT_DISCARD) + break; + break; + case ETYPE_IPv6: + break; + case ETYPE_ARP: + break; + default: + break; + } + if (priority == OUT_DISCARD) { + task->drop.buffer[task->drop.pkt_nb] = mbuf; + task->drop.pkt_nb++; + return; + } + buffer_packet(task, mbuf, priority); +} + +static int handle_aggregator_bulk(struct task_base *tbase, struct rte_mbuf **mbufs, uint16_t n_pkts) +{ + struct task_aggregator *task = (struct task_aggregator *)tbase; + + uint16_t j; + uint32_t drop_bytes = 0; +#ifdef PROX_PREFETCH_OFFSET + for (j = 0; j < PROX_PREFETCH_OFFSET && j < n_pkts; ++j) { + prefetch_nta(mbufs[j]); + } + for (j = 1; j < PROX_PREFETCH_OFFSET && j < n_pkts; ++j) { + prefetch_nta(rte_pktmbuf_mtod(mbufs[j - 1], void *)); + } +#endif + for (j = 0; j + PREFETCH_OFFSET < n_pkts; ++j) { +#ifdef PROX_PREFETCH_OFFSET + prefetch_nta(mbufs[j + PREFETCH_OFFSET]); + prefetch_nta(rte_pktmbuf_mtod(mbufs[j + PREFETCH_OFFSET - 1], void *)); +#endif + handle_aggregator(task, mbufs[j]); + } +#ifdef PROX_PREFETCH_OFFSET + prefetch_nta(rte_pktmbuf_mtod(mbufs[n_pkts - 1], void *)); + for (; j < n_pkts; ++j) { + handle_aggregator(task, mbufs[j]); + } +#endif + + for (int i = 0 ; i < task->drop.pkt_nb; i++) { + drop_bytes += mbuf_wire_size(task->drop.buffer[i]); + rte_pktmbuf_free(task->drop.buffer[i]); + } + TASK_STATS_ADD_DROP_TX_FAIL(&tbase->aux->stats, task->drop.pkt_nb); + TASK_STATS_ADD_DROP_BYTES(&tbase->aux->stats, drop_bytes); + task->drop.pkt_nb = 0; + + for (int priority = 0; priority < MAX_PRIORITIES; priority++) { + struct task_buffer *prio = &task->priority[priority]; + if (prio->pkt_nb) { + uint8_t n = 0; + if (prio->pkt_pos > prio->pkt_nb) { + struct rte_mbuf **buf = prio->buffer + prio->pkt_pos - prio->pkt_nb; + n = tbase->aux->tx_pkt_try(&task->base, buf, prio->pkt_nb); + } else { + struct rte_mbuf **buf = prio->buffer + BUFFER_LENGTH + prio->pkt_pos - prio->pkt_nb; + n = tbase->aux->tx_pkt_try(&task->base, buf, prio->pkt_nb - prio->pkt_pos); + if (n == (prio->pkt_nb - prio->pkt_pos)) + n += tbase->aux->tx_pkt_try(&task->base, prio->buffer, prio->pkt_pos); + } + prio->pkt_nb -=n; + TASK_STATS_ADD_TX_PRIO(&task->stats, n, priority); + if (prio->pkt_nb) + break; + } + } + return 0; +} + +static void init_task_aggregator(struct task_base *tbase, struct task_args *targ) +{ + struct task_aggregator *task = (struct task_aggregator *)tbase; + const int socket_id = rte_lcore_to_socket_id(targ->lconf->id); +} + +static struct task_init task_init_aggregator = { + .mode_str = "aggreg", + .init = init_task_aggregator, + .handle = handle_aggregator_bulk, + .flag_features = TASK_FEATURE_NEVER_DISCARDS, + .size = sizeof(struct task_aggregator) +}; + +__attribute__((constructor)) static void reg_task_aggregator(void) +{ + reg_task(&task_init_aggregator); +} |