diff options
| author |   Xavier Simonart <xavier.simonart@intel.com> | 2020-05-10 21:04:08 +0200 |
|---|---|---|
| committer | Xavier Simonart <xavier.simonart@intel.com> | 2020-05-29 23:34:46 +0200 |
| commit | 1614130d60abfaa89a41ba8eed5f9bbf41d9a4f4 (patch) | |
| tree | a25af2c624a87aa850faa38a663f4554b0d498c7 /VNFs/DPPD-PROX/tx_pkt.c | |
| parent | 08fee9c5d2e1d1f3fe14d00683c2a4b7a17e7876 (diff) | |
Improve performance in l3 submode
Two cases where performance has been improved
- When using a gateway from a routing table (l3 submode),
store the mac within the next hop table, to avoid a hash_lookup.
This gives ~10% improvement.
- Read tsc only once per bulk (of up to 64 packets).
This gives ~10% improvement to swap, 4% to gen.
In addition a small fix has been added, preventig "No route" Error
to be written too aften.
Change-Id: I8a7ab74a32f09c8ff47f751ee91e84afee1b2147
Signed-off-by: Xavier Simonart <xavier.simonart@intel.com>
Diffstat (limited to 'VNFs/DPPD-PROX/tx_pkt.c')
| -rw-r--r-- | VNFs/DPPD-PROX/tx_pkt.c | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/VNFs/DPPD-PROX/tx_pkt.c b/VNFs/DPPD-PROX/tx_pkt.c index 2a4f53b2..51c1afa7 100644 --- a/VNFs/DPPD-PROX/tx_pkt.c +++ b/VNFs/DPPD-PROX/tx_pkt.c @@ -57,11 +57,12 @@ int tx_pkt_l3(struct task_base *tbase, struct rte_mbuf **mbufs, uint16_t n_pkts, const struct port_queue *port_queue = &tbase->tx_params_hw.tx_port_queue[0]; struct rte_mbuf *arp_mbuf = NULL; // used when one need to send both an ARP and a mbuf uint64_t *time; + uint64_t tsc = rte_rdtsc(); for (int j = 0; j < n_pkts; j++) { if ((out) && (out[j] >= OUT_HANDLED)) continue; - if (unlikely((rc = write_dst_mac(tbase, mbufs[j], &ip_dst, &time)) != SEND_MBUF)) { + if (unlikely((rc = write_dst_mac(tbase, mbufs[j], &ip_dst, &time, tsc)) != SEND_MBUF)) { if (j - first) { ret = tbase->aux->tx_pkt_l2(tbase, mbufs + first, j - first, out); ok += ret; |