git-lng-odp

git-lng-odp@lists.linaro.org

1240 discussions

Delhi Dream Girls Escort In Mahipalpur Delhi 9999211002 Call Girls

by delhivip23＠gmail.com

Welcome To Call Girls In Mahipalpur Delhi NCR 9999211002 .I’m living in Aerocity Delhi.NCR Escorts Service Door Step Delivery Indian, Russian Best Quality Full Educated And Full Cooperative Independent Call Girls Escort Services In Delhi Noida Gurgaon & Ghaziabad No Advance Only Cash On Delivery . Genuine Person Contact Me Delhi NCR Escort Service in Mahipalpur With 24*7 Free Hotel Delivery Be smart to choose genuine escort services in Paharganj. There are a few simple steps that help you in choosing genuine escorts service providers. We can understand that whoever comes to go wants a lot of fun for amazing pictures and memories. Clients can enhance that fun without any hesitation with our Model Escorts in Paharganj. Our agency consists of a great collection of call girls, for providing an amazing feel all night long. Apart from that our services are 24*7 available for the clients. So, customers can choose their partner according to their needs and their time. Every girl who is working or connected with us is highly trained and well-known for their professional behavior. Our Paharganj Call Girls service never let down their partners in any manner. Bring them to your room and let them do their job to make your night blissful. The major focus is only to provide the best sensual feel that you ever had in your life Aunty Sexy Aunty. Location: Delhi Gurgaon Noida Mahipalpur Aerocity Connaught Place Dwarka Vasant Kunj And Covering All Delhi Ncr. Hotel Provide.. • Hotel Jw Marriott Aerocity… • Hotel Novotel Aerocity…. • Hotel Andaz Delhi Aerocity.. •hotel Pullman Aerocity… • Hotel Roseate House Aerocity.. • Hotel Shangri-la’s — Eros Cp.. • Hotel Taj Mahal, New Delhi.. •hotel The Leela Palace New Delhi.. • Hotel Itc Maurya, Hotel, New Delhi • Hotel Radisson Blue Dwarka… • Hotel The Westin Gurgaon.. • Hotel Le Meridien Cp.. • Hotel Le Meridien Gurgaon. • Hotel Sheraton Saket.. • Hotel Hilton Gadana Inn Saket • Hotel Ibis Aerocity.. • Hotel Crowne Plaza Okhla.. • Hotel The Lodhi New Delhi. • Hotel The Leela Ambience Gurgaon.. • Hotel Oberor New Delhi. • Hotel Metropolitan Cp.. • Hotel Hyatt Regency Delhi. • Hotel Pride Plaza Aerocirt. • Hotel Holida Inn Aerocity. • Hotel Lemon Tree Aerocity.. • Hotel Red Fox Aerocity.. • Hotel Dee Marks Mahipalpur. • Hotel Vivanta By Taj Dwarka. • Hotel The Ashok Chanakypuri.. • Hotel The Grand Vasant Kunj. • Hotel Hotel Radisson Blue Marina Cp.. • Hotel The Lalit Cp.. • Hotel The Park Cp • Hotel Wlcome Dwarka.. • Hotel Shanti Palace Mahipalpur And All Hotel Also.. Our Service Only Is Only For Vip And Premium Class Gentlemen. Escorts Available at all 3/5/7 Star Hotels In Gurgaon Delhi Noida Gurgaon We Provide Young College girls/,Foreigner Escorts and High Profile Models , In Call & Out Call Available. Our Girls Can Directly Enter Any Star Hotel (Out Call) Experience call girl and all types Services here THANKS FOR VISITING♡♡♡♡♡ — — — — Booking 247 HRS. NOW CALL FOR ENJOY WITH NIGHT QUEEN ♡♡♡♡THANKS FOR VISITING♡♡♡♡

1 week, 6 days

Delhi Dream Girls Escort In Mahipalpur Delhi 9999211002 Call Girls

by delhivip23＠gmail.com

1 week, 6 days

Ratchaburi Dating Vibes: Best Ways to Meet People

by arafatmohd857＠gmail.com

Ratchaburi dating vibes are a mix of relaxed everyday life and friendly social energy. People here often appreciate sincerity, good conversation, and taking things at a comfortable pace. If you’re trying to meet someone in Ratchaburi, the best approach is to focus on connection first—show interest in their world, ask thoughtful questions, and keep messages respectful and natural. Start with places and moments where conversation happens easily. Think about casual cafés, weekend markets, community events, local festivals, and popular hangout streets where people naturally spend time. If you’re new to the area, your best “icebreaker” is curiosity: ask what they recommend, what the best weekend plan looks like, or what local spot they’d take a friend to. When you share small details about your own day, it makes the chat feel more real and less like a generic template. Once you start talking online, your messages should be light but specific. Instead of “Hey, how are you?”, try something like: “What do you enjoy doing in Ratchaburi when you have free time?” or “Is there a favorite food you always come back to?” Follow up with one easy question and one short personal detail. That rhythm tends to keep the conversation flowing. If their profile mentions work or availability, it’s okay to acknowledge it in a respectful way, not in a demanding or transactional tone. For example, you can say: “I saw your update about Girl accepting a job (สาวรับงาน). That sounds busy—what kind of work are you doing, and what do you usually do after work?” This kind of question shows you’re genuinely interested in them as a person. Finally, when you feel comfortable, suggest a simple first meet—coffee, a short walk, or a quick bite at a public place. Ratchaburi dating works best when both people feel relaxed, safe, and excited to learn more about each other. https://fiwfan.app/location/%E0%B8%A3%E0%B8%B1%E0%B8%8A%E0%B8%94%E0%B8%B2

1 month, 2 weeks

Legitimate Cryp-to/Bit-coin Reco-very Firm in 2026 ; Visit ZEUS CRYP-TO RECOVE-RY SERVICES

by johnlucas635uw＠gmail.com

I am incredibly grateful to ZEUS CRYP-TO RECOVE-RY SERVICES for their exceptional help. After losing $209,000.56 in Bitcoin last month, I felt hopeless. Their team was professional, transparent, and utilized advanced bloc=kchain forensics to track and successfully reco-ver my assets. They kept me updated throughout the entire process, providing peace of mind during a stressful time. I highly recommend their expertise to anyone facing similar cry-pto losses. Below is their contact information.. Website: zeusrecoveryservices.com Mail-Box: support @ zeusrecoveryservices .c om WhatsApp: +44-7841 9000-82

6 months, 1 week

Reliable and Professional Crypto Recovery Company – Betafort Recovery

by godfreyleonard474＠gmail.com

I am immensely grateful to Betafort Recovery for their exceptional service and I would highly recommend reaching out to Betafort recovery via Google to anyone who wants to recover scammed bitcoin, stolen cryptocurrency, funds lost to binary options forex, funds lost to fake online investment and any other form of online scam for consultation and assistance. Their professionalism, expertise, and commitment to customer satisfaction truly sets them apart. NO UPFRONT FEES REQUIRED. Email: info(a)betafort.online WhatsApp OR SIGNAL +17126004729 Web portal: https://betafort.online/

7 months

Hire SAFEGUARD RECOVERY, a Bitcoin recovery specialist, to retrieve stolen or lost coins.

by sophia fourest

I invested $320,000 in Tether (USDT) on a fraudulent website after falling for a romantic scam. I felt completely helpless and in need of assistance after realizing I had been duped. I started looking for a hacker online and found SAFEGUARD RECOVERY. I had optimism because of his professionalism and knowledge. I'm happy to report that SAFEGUARD RECOVERY successfully recovered my stolen money after working relentlessly to do so! I am immensely appreciative of their help and heartily urge anyone in a comparable circumstance to use their services. I'm grateful Email: safeguardbitcoin(a)consultant.com WhatsApp: +44 7426 168300 Website: https://safeguardbitcoin.wixsite.com/safeguard-bitcoin--1

7 months, 1 week

Hire SAFEGUARD RECOVERY, a Bitcoin recovery specialist, to retrieve stolen or lost coins.

by sophia fourest

7 months, 1 week

[COMMIT] branch master updated. v1.21.0.0-428-g519555018

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, master has been updated via 519555018eb8beeaf8bbd704d0d24b60eddf4d20 (commit) via 1d09a45ec333bf237edff731e78fa1c68dc49092 (commit) via 831de57416098df6946d2eb97132fe236d0c7e42 (commit) via 181773542664bddff206fedb62ba94c38118457d (commit) from 324a2a2c7c1a9d35323483b679eedfebcb310827 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 519555018eb8beeaf8bbd704d0d24b60eddf4d20 Author: Matias Elo <matias.elo(a)nokia.com> Date: Tue Feb 25 16:16:29 2020 +0200 example: don't enable input hashing with only one input queue Signed-off-by: Matias Elo <matias.elo(a)nokia.com> Reviewed-by: Petri Savolainen <petri.savolainen(a)nokia.com> diff --git a/example/l3fwd/odp_l3fwd.c b/example/l3fwd/odp_l3fwd.c index 79613d672..0f8b8525c 100644 --- a/example/l3fwd/odp_l3fwd.c +++ b/example/l3fwd/odp_l3fwd.c @@ -803,17 +803,18 @@ static void setup_worker_qconf(app_args_t *args) in_queue_param.op_mode = ODP_PKTIO_OP_MT_UNSAFE; out_queue_param.op_mode = ODP_PKTIO_OP_MT_UNSAFE; - in_queue_param.hash_enable = 1; - in_queue_param.hash_proto.proto.ipv4 = 1; - in_queue_param.hash_proto.proto.ipv4_tcp = 1; - in_queue_param.hash_proto.proto.ipv4_udp = 1; - in_queue_param.num_queues = port->rxq_idx; if (port->rxq_idx > port->nb_rxq) { in_queue_param.num_queues = port->nb_rxq; in_queue_param.op_mode = ODP_PKTIO_OP_MT; } + in_queue_param.hash_enable = (in_queue_param.num_queues > 1) ? + 1 : 0; + in_queue_param.hash_proto.proto.ipv4 = 1; + in_queue_param.hash_proto.proto.ipv4_tcp = 1; + in_queue_param.hash_proto.proto.ipv4_udp = 1; + if (odp_pktin_queue_config(port->pktio, &in_queue_param)) ODPH_ABORT("Fail to config input queue for port %s\n", name); diff --git a/example/switch/odp_switch.c b/example/switch/odp_switch.c index ccb975f6c..bcc87680e 100644 --- a/example/switch/odp_switch.c +++ b/example/switch/odp_switch.c @@ -304,7 +304,7 @@ static int create_pktio(const char *dev, int idx, int num_rx, int num_tx, mode_tx = ODP_PKTIO_OP_MT; } - pktin_param.hash_enable = 1; + pktin_param.hash_enable = (num_rx > 1) ? 1 : 0; pktin_param.hash_proto.proto.ipv4_tcp = 1; pktin_param.hash_proto.proto.ipv4_udp = 1; pktin_param.num_queues = num_rx; diff --git a/test/performance/odp_l2fwd.c b/test/performance/odp_l2fwd.c index fc1f04bd1..4bacc5309 100644 --- a/test/performance/odp_l2fwd.c +++ b/test/performance/odp_l2fwd.c @@ -759,7 +759,7 @@ static int create_pktio(const char *dev, int idx, int num_rx, int num_tx, mode_tx = ODP_PKTIO_OP_MT; } - pktin_param.hash_enable = 1; + pktin_param.hash_enable = (num_rx > 1) ? 1 : 0; pktin_param.hash_proto.proto.ipv4_udp = 1; pktin_param.num_queues = num_rx; pktin_param.op_mode = mode_rx; diff --git a/test/performance/odp_pktio_ordered.c b/test/performance/odp_pktio_ordered.c index 4ac6060b7..46e6c1861 100644 --- a/test/performance/odp_pktio_ordered.c +++ b/test/performance/odp_pktio_ordered.c @@ -642,7 +642,7 @@ static int create_pktio(const char *dev, int idx, int num_rx, int num_tx, mode_tx = ODP_PKTIO_OP_MT; } - pktin_param.hash_enable = 1; + pktin_param.hash_enable = (num_rx > 1) ? 1 : 0; pktin_param.hash_proto.proto.ipv4_udp = 1; pktin_param.num_queues = num_rx; pktin_param.op_mode = mode_rx; commit 1d09a45ec333bf237edff731e78fa1c68dc49092 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Feb 20 16:49:04 2020 +0200 test: packet_gen: test application during make check Packet_gen sends 500 packets through both loopback interfaces. Test lasts 5 sec. In total 1000 packets are sent and the same amount (or minus the last sent burst) should be received. Test passes when at least 800 packets are received. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Matias Elo <matias.elo(a)nokia.com> diff --git a/test/performance/Makefile.am b/test/performance/Makefile.am index 36bc7cbd6..4b32573cd 100644 --- a/test/performance/Makefile.am +++ b/test/performance/Makefile.am @@ -20,6 +20,7 @@ COMPILE_ONLY = odp_l2fwd \ odp_scheduling TESTSCRIPTS = odp_l2fwd_run.sh \ + odp_packet_gen_run.sh \ odp_sched_latency_run.sh \ odp_sched_pktio_run.sh \ odp_scheduling_run.sh diff --git a/test/performance/odp_packet_gen_run.sh b/test/performance/odp_packet_gen_run.sh new file mode 100755 index 000000000..579509a07 --- /dev/null +++ b/test/performance/odp_packet_gen_run.sh @@ -0,0 +1,70 @@ +#!/bin/sh +# +# Copyright (c) 2020, Nokia +# All rights reserved. +# +# SPDX-License-Identifier: BSD-3-Clause +# + +# directory where test binaries have been built +TEST_DIR="${TEST_DIR:-$PWD}" + +# directory where test sources are, including scripts +TEST_SRC_DIR=$(dirname $0) + +PATH=$TEST_DIR:$PATH + +# exit codes expected by automake for skipped tests +TEST_SKIPPED=77 + +VALIDATION_TESTDIR=platform/$ODP_PLATFORM/test/validation +PLATFORM_VALIDATION=${TEST_SRC_DIR}/../../$VALIDATION_TESTDIR + +# Use installed pktio env or for make check take it from platform directory +if [ -f "./pktio_env" ]; then + . ./pktio_env +elif [ "$ODP_PLATFORM" = "" ]; then + echo "$0: error: ODP_PLATFORM must be defined" + # not skipped as this should never happen via "make check" + exit 1 +elif [ -f ${PLATFORM_VALIDATION}/api/pktio/pktio_env ]; then + . ${PLATFORM_VALIDATION}/api/pktio/pktio_env +else + echo "BUG: unable to find pktio_env!" + echo "pktio_env has to be in current directory or " + echo "in platform/\$ODP_PLATFORM/test." + echo "ODP_PLATFORM=\"$ODP_PLATFORM\"" + exit 1 +fi + +run_packet_gen() +{ + setup_pktio_env clean # install trap to call cleanup_pktio_env + + if [ $? -ne 0 ]; then + echo "setup_pktio_env error $?" + exit $TEST_SKIPPED + fi + + # Runs 500 * 10ms = 5 sec + # Sends 500 packets through both interfaces => total 1000 packets + odp_packet_gen${EXEEXT} -i $IF0,$IF1 -b 1 -g 10000000 -q 500 + + ret=$? + + if [ $ret -eq -1 ]; then + echo "FAIL: test failed" + elif [ $ret -eq -2 ]; then + echo "FAIL: too few packets received" + fi + + cleanup_pktio_env + + exit $ret +} + +case "$1" in + setup) setup_pktio_env ;; + cleanup) cleanup_pktio_env ;; + *) run_packet_gen ;; +esac commit 831de57416098df6946d2eb97132fe236d0c7e42 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Wed Feb 12 15:28:50 2020 +0200 test: packet_gen: new simple packet generator Added new simple and performance optimized packet generator. It may be used e.g. in pair with l2fwd. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Matias Elo <matias.elo(a)nokia.com> diff --git a/test/performance/.gitignore b/test/performance/.gitignore index 41e18facc..a4e78b9af 100644 --- a/test/performance/.gitignore +++ b/test/performance/.gitignore @@ -6,6 +6,7 @@ odp_cpu_bench odp_crypto odp_ipsec odp_l2fwd +odp_packet_gen odp_pktio_ordered odp_pktio_perf odp_pool_perf diff --git a/test/performance/Makefile.am b/test/performance/Makefile.am index 29aee55bd..36bc7cbd6 100644 --- a/test/performance/Makefile.am +++ b/test/performance/Makefile.am @@ -13,6 +13,7 @@ EXECUTABLES = odp_bench_packet \ odp_timer_perf COMPILE_ONLY = odp_l2fwd \ + odp_packet_gen \ odp_pktio_ordered \ odp_sched_latency \ odp_sched_pktio \ @@ -39,6 +40,7 @@ odp_bench_packet_SOURCES = odp_bench_packet.c odp_cpu_bench_SOURCES = odp_cpu_bench.c odp_crypto_SOURCES = odp_crypto.c odp_ipsec_SOURCES = odp_ipsec.c +odp_packet_gen_SOURCES = odp_packet_gen.c odp_pktio_ordered_SOURCES = odp_pktio_ordered.c dummy_crc.h odp_sched_latency_SOURCES = odp_sched_latency.c odp_sched_pktio_SOURCES = odp_sched_pktio.c diff --git a/test/performance/odp_packet_gen.c b/test/performance/odp_packet_gen.c new file mode 100644 index 000000000..4c1a1b49a --- /dev/null +++ b/test/performance/odp_packet_gen.c @@ -0,0 +1,1313 @@ +/* Copyright (c) 2020, Nokia + * All rights reserved. + * + * SPDX-License-Identifier: BSD-3-Clause + */ + +#include <stdio.h> +#include <string.h> +#include <stdint.h> +#include <inttypes.h> +#include <signal.h> +#include <stdlib.h> +#include <getopt.h> + +#include <odp_api.h> +#include <odp/helper/odph_api.h> + +#define MAX_PKTIOS 32 +#define MAX_PKTIO_NAME 255 +#define RX_THREAD 1 +#define TX_THREAD 2 + +/* Minimum number of packets to receive in CI test */ +#define MIN_RX_PACKETS_CI 800 + +typedef struct test_options_t { + uint64_t gap_nsec; + uint64_t quit; + uint32_t num_rx; + uint32_t num_tx; + uint32_t num_cpu; + uint32_t num_pktio; + uint32_t num_pkt; + uint32_t pkt_len; + uint32_t burst_size; + uint32_t bursts; + uint32_t ipv4_src; + uint32_t ipv4_dst; + uint16_t udp_src; + uint16_t udp_dst; + + struct { + uint32_t udp_src; + uint32_t udp_dst; + } c_mode; + + char pktio_name[MAX_PKTIOS][MAX_PKTIO_NAME + 1]; + char ipv4_src_s[24]; + char ipv4_dst_s[24]; + +} test_options_t; + +typedef struct thread_arg_t { + void *global; + int tx_thr; + + /* pktout queue per pktio interface (per thread) */ + odp_pktout_queue_t pktout[MAX_PKTIOS]; + +} thread_arg_t; + +typedef struct ODP_ALIGNED_CACHE thread_stat_t { + uint64_t time_nsec; + uint64_t rx_timeouts; + uint64_t rx_packets; + uint64_t rx_bytes; + + uint64_t tx_timeouts; + uint64_t tx_packets; + uint64_t tx_drops; + + int thread_type; + +} thread_stat_t; + +typedef struct test_global_t { + test_options_t test_options; + odp_atomic_u32_t exit_test; + odp_barrier_t barrier; + odp_cpumask_t cpumask; + odp_pool_t pool; + uint64_t drained; + odph_thread_t thread_tbl[ODP_THREAD_COUNT_MAX]; + thread_stat_t stat[ODP_THREAD_COUNT_MAX]; + thread_arg_t thread_arg[ODP_THREAD_COUNT_MAX]; + + struct { + odph_ethaddr_t eth_src; + odph_ethaddr_t eth_dst; + odp_pktio_t pktio; + odp_pktout_queue_t pktout[ODP_THREAD_COUNT_MAX]; + int started; + + } pktio[MAX_PKTIOS]; + +} test_global_t; + +static test_global_t *test_global; + +static void print_usage(void) +{ + printf("\n" + "ODP packet generator\n" + "\n" + "Usage: odp_packet_gen [options]\n" + "\n" + " Mandatory:\n" + " -i, --interface <name> Packet IO interfaces. Comma-separated list of\n" + " interface names (no spaces) e.g. eth0,eth1.\n" + " At least one interface is required.\n" + "\n" + " Optional:\n" + " -e, --eth_dst <mac> Destination MAC address. Comma-separated list of\n" + " addresses (no spaces), one address per packet IO\n" + " interface e.g. AA:BB:CC:DD:EE:FF,11:22:33:44:55:66\n" + " Default per interface: 02:00:00:A0:B0:CX, where X = 0,1,...\n" + " -r, --num_rx Number of receive threads. Default: 1\n" + " -t, --num_tx Number of transmit threads. Default: 1\n" + " -n, --num_pkt Number of packets in the pool. Default: 1000\n" + " -l, --len Packet length. Default: 512\n" + " -b, --burst_size Transmit burst size. Default: 8\n" + " -x, --bursts Number of bursts per one transmit round. Default: 1\n" + " -g, --gap Gap between transmit rounds in nsec. Default: 1000000\n" + " Transmit packet rate per interface:\n" + " num_tx * burst_size * bursts * (10^9 / gap)\n" + " -s, --ipv4_src IPv4 source address. Default: 192.168.0.1\n" + " -d, --ipv4_dst IPv4 destination address. Default: 192.168.0.2\n" + " -o, --udp_src UDP source port. Default: 10000\n" + " -p, --udp_dst UDP destination port. Default: 20000\n" + " -c, --c_mode <counts> Counter mode for incrementing UDP port numbers.\n" + " Specify the number of port numbers used starting from\n" + " udp_src/udp_dst. Comma-serarated (no spaces) list of\n" + " count values: <udp_src count>,<udp_dst count>\n" + " Default value: 0,0\n" + " -q, --quit Quit after this many transmit rounds.\n" + " Default: 0 (don't quit)\n" + " -h, --help This help\n" + "\n"); +} + +static int parse_options(int argc, char *argv[], test_global_t *global) +{ + int opt, i, len, str_len, long_index; + unsigned long int count; + uint32_t min_packets, num_tx_pkt; + char *name, *str, *end; + test_options_t *test_options = &global->test_options; + int ret = 0; + uint8_t default_eth_dst[6] = {0x02, 0x00, 0x00, 0xa0, 0xb0, 0xc0}; + + static const struct option longopts[] = { + {"interface", required_argument, NULL, 'i'}, + {"eth_dst", required_argument, NULL, 'e'}, + {"num_rx", required_argument, NULL, 'r'}, + {"num_tx", required_argument, NULL, 't'}, + {"num_pkt", required_argument, NULL, 'n'}, + {"len", required_argument, NULL, 'l'}, + {"burst_size", required_argument, NULL, 'b'}, + {"bursts", required_argument, NULL, 'x'}, + {"gap", required_argument, NULL, 'g'}, + {"ipv4_src", required_argument, NULL, 's'}, + {"ipv4_dst", required_argument, NULL, 'd'}, + {"udp_src", required_argument, NULL, 'o'}, + {"udp_dst", required_argument, NULL, 'p'}, + {"c_mode", required_argument, NULL, 'c'}, + {"quit", required_argument, NULL, 'q'}, + {"help", no_argument, NULL, 'h'}, + {NULL, 0, NULL, 0} + }; + + static const char *shortopts = "+i:e:r:t:n:l:b:x:g:s:d:o:p:c:q:h"; + + test_options->num_pktio = 0; + test_options->num_rx = 1; + test_options->num_tx = 1; + test_options->num_pkt = 1000; + test_options->pkt_len = 512; + test_options->burst_size = 8; + test_options->bursts = 1; + test_options->gap_nsec = 1000000; + strncpy(test_options->ipv4_src_s, "192.168.0.1", + sizeof(test_options->ipv4_src_s)); + strncpy(test_options->ipv4_dst_s, "192.168.0.2", + sizeof(test_options->ipv4_dst_s)); + odph_ipv4_addr_parse(&test_options->ipv4_src, test_options->ipv4_src_s); + odph_ipv4_addr_parse(&test_options->ipv4_dst, test_options->ipv4_dst_s); + test_options->udp_src = 10000; + test_options->udp_dst = 20000; + test_options->c_mode.udp_src = 0; + test_options->c_mode.udp_dst = 0; + test_options->quit = 0; + + for (i = 0; i < MAX_PKTIOS; i++) { + memcpy(global->pktio[i].eth_dst.addr, default_eth_dst, 6); + global->pktio[i].eth_dst.addr[5] += i; + } + + while (1) { + opt = getopt_long(argc, argv, shortopts, longopts, &long_index); + + if (opt == -1) + break; + + switch (opt) { + case 'i': + i = 0; + str = optarg; + str_len = strlen(str); + + while (str_len > 0) { + len = strcspn(str, ","); + str_len -= len + 1; + + if (i == MAX_PKTIOS) { + printf("Error: Too many interfaces\n"); + ret = -1; + break; + } + + if (len > MAX_PKTIO_NAME) { + printf("Error: Too long interface name %s\n", + str); + ret = -1; + break; + } + + name = test_options->pktio_name[i]; + memcpy(name, str, len); + str += len + 1; + i++; + } + + test_options->num_pktio = i; + + break; + case 'e': + i = 0; + str = optarg; + str_len = strlen(str); + + while (str_len > 0) { + odph_ethaddr_t *dst = &global->pktio[i].eth_dst; + + len = strcspn(str, ","); + str_len -= len + 1; + + if (i == MAX_PKTIOS) { + printf("Error: Too many MAC addresses\n"); + ret = -1; + break; + } + + if (odph_eth_addr_parse(dst, str)) { + printf("Error: Bad MAC address: %s\n", + str); + ret = -1; + break; + } + + str += len + 1; + i++; + } + break; + case 'r': + test_options->num_rx = atoi(optarg); + break; + case 't': + test_options->num_tx = atoi(optarg); + break; + case 'n': + test_options->num_pkt = atoi(optarg); + break; + case 'l': + test_options->pkt_len = atoi(optarg); + break; + case 'b': + test_options->burst_size = atoi(optarg); + break; + case 'x': + test_options->bursts = atoi(optarg); + break; + case 'g': + test_options->gap_nsec = atoll(optarg); + break; + case 's': + if (odph_ipv4_addr_parse(&test_options->ipv4_src, + optarg)) { + printf("Error: Bad IPv4 source address: %s\n", + optarg); + ret = -1; + } + strncpy(test_options->ipv4_src_s, optarg, + sizeof(test_options->ipv4_src_s)); + break; + case 'd': + if (odph_ipv4_addr_parse(&test_options->ipv4_dst, + optarg)) { + printf("Error: Bad IPv4 destination address: %s\n", + optarg); + ret = -1; + } + strncpy(test_options->ipv4_dst_s, optarg, + sizeof(test_options->ipv4_dst_s)); + break; + case 'c': + count = strtoul(optarg, &end, 0); + test_options->c_mode.udp_src = count; + + end++; + count = strtoul(end, NULL, 0); + test_options->c_mode.udp_dst = count; + break; + case 'q': + test_options->quit = atoll(optarg); + break; + case 'h': + /* fall through */ + default: + print_usage(); + ret = -1; + break; + } + } + + if (test_options->num_pktio == 0) { + printf("Error: At least one packet IO interface is needed.\n"); + printf(" Use -i <name> to specify interfaces.\n"); + ret = -1; + } + + if (test_options->num_rx < 1 || test_options->num_tx < 1) { + printf("Error: At least one rx and tx thread needed.\n"); + ret = -1; + } + + test_options->num_cpu = test_options->num_rx + test_options->num_tx; + + /* Pool needs to have enough packets for all tx side bursts and + * one rx side burst */ + num_tx_pkt = test_options->burst_size * test_options->bursts; + min_packets = (test_options->num_pktio * test_options->num_tx * + num_tx_pkt) + + (test_options->num_pktio * test_options->num_rx * + test_options->burst_size); + + if (test_options->num_pkt < min_packets) { + printf("Error: Pool needs to have at least %u packets\n", + min_packets); + ret = -1; + } + + if (test_options->gap_nsec) { + double gap_hz = 1000000000.0 / test_options->gap_nsec; + + if (gap_hz > (double)odp_time_local_res()) { + printf("\nWARNING: Burst gap exceeds time counter resolution " + "%" PRIu64 "\n\n", odp_time_local_res()); + } + } + + if (test_options->c_mode.udp_dst && + num_tx_pkt % test_options->c_mode.udp_dst) + printf("\nWARNING: Transmit packet count is not evenly divisible by UDP destination port count.\n\n"); + + if (test_options->c_mode.udp_src && + num_tx_pkt % test_options->c_mode.udp_src) + printf("\nWARNING: Transmit packet count is not evenly divisible by UDP source port count.\n\n"); + + return ret; +} + +static int set_num_cpu(test_global_t *global) +{ + int ret; + test_options_t *test_options = &global->test_options; + int num_cpu = test_options->num_cpu; + + /* One thread used for the main thread */ + if (num_cpu > ODP_THREAD_COUNT_MAX - 1) { + printf("Error: Too many threads. API supports max %i.\n", + ODP_THREAD_COUNT_MAX - 1); + return -1; + } + + ret = odp_cpumask_default_worker(&global->cpumask, num_cpu); + + if (ret != num_cpu) { + int cpu; + + /* Normally we want to use only worker threads */ + if (ret > 1) { + printf("Error: Too many workers. Maximum supported %i.\n", + ret); + return -1; + } + + /* When number of workers is very limited (e.g. ODP project CI), + * we try to use any CPUs available. */ + ret = odp_cpumask_all_available(&global->cpumask); + if (ret < num_cpu) { + printf("Error: Not enough CPUs. Maximum supported %i.\n", + ret); + return -1; + } + + /* Remove extra CPUs from the mask */ + cpu = odp_cpumask_first(&global->cpumask); + while (ret > num_cpu) { + odp_cpumask_clr(&global->cpumask, cpu); + cpu = odp_cpumask_first(&global->cpumask); + ret--; + } + } + + odp_barrier_init(&global->barrier, num_cpu + 1); + + return 0; +} + +static int open_pktios(test_global_t *global) +{ + odp_pool_capability_t pool_capa; + odp_pktio_capability_t pktio_capa; + odp_pool_param_t pool_param; + odp_pool_t pool; + odp_pktio_param_t pktio_param; + odp_pktio_t pktio; + odp_pktio_config_t pktio_config; + odp_pktin_queue_param_t pktin_param; + odp_pktout_queue_param_t pktout_param; + char *name; + uint32_t i, seg_len; + int j; + test_options_t *test_options = &global->test_options; + uint32_t num_rx = test_options->num_rx; + int num_tx = test_options->num_tx; + uint32_t num_pktio = test_options->num_pktio; + uint32_t num_pkt = test_options->num_pkt; + uint32_t pkt_len = test_options->pkt_len; + odp_pktout_queue_t pktout[num_tx]; + + printf("\nODP packet generator\n"); + printf(" quit test after %" PRIu64 " rounds\n", + test_options->quit); + printf(" num rx threads %u\n", num_rx); + printf(" num tx threads %i\n", num_tx); + printf(" num packets %u\n", num_pkt); + printf(" packet length %u bytes\n", pkt_len); + printf(" tx burst size %u\n", test_options->burst_size); + printf(" tx bursts %u\n", test_options->bursts); + printf(" tx burst gap %" PRIu64 " nsec\n", + test_options->gap_nsec); + printf(" clock resolution %" PRIu64 " Hz\n", odp_time_local_res()); + printf(" IPv4 source %s\n", test_options->ipv4_src_s); + printf(" IPv4 destination %s\n", test_options->ipv4_dst_s); + printf(" UDP source %u\n", test_options->udp_src); + printf(" UDP destination %u\n", test_options->udp_dst); + printf(" UDP src count %u\n", test_options->c_mode.udp_src); + printf(" UDP dst count %u\n", test_options->c_mode.udp_dst); + printf(" num pktio %u\n", num_pktio); + + printf(" interfaces names: "); + for (i = 0; i < num_pktio; i++) { + if (i > 0) + printf(" "); + printf("%s\n", test_options->pktio_name[i]); + } + + printf(" destination MACs: "); + for (i = 0; i < num_pktio; i++) { + uint8_t *eth_dst = global->pktio[i].eth_dst.addr; + + if (i > 0) + printf(" "); + printf("%02x:%02x:%02x:%02x:%02x:%02x\n", + eth_dst[0], eth_dst[1], eth_dst[2], + eth_dst[3], eth_dst[4], eth_dst[5]); + } + printf("\n"); + + global->pool = ODP_POOL_INVALID; + + if (odp_pool_capability(&pool_capa)) { + printf("Error: Pool capability failed.\n"); + return -1; + } + + if (pool_capa.pkt.max_num && + num_pkt > pool_capa.pkt.max_num) { + printf("Error: Too many packets. Max %u supported.\n", + pool_capa.pkt.max_num); + return -1; + } + + if (pool_capa.pkt.max_len && + pkt_len > pool_capa.pkt.max_len) { + printf("Error: Too large packets. Max %u supported length.\n", + pool_capa.pkt.max_len); + return -1; + } + + seg_len = ODPH_ETHHDR_LEN + ODPH_IPV4HDR_LEN + ODPH_UDPHDR_LEN; + if (pool_capa.pkt.max_seg_len && + seg_len > pool_capa.pkt.max_seg_len) { + printf("Error: Max segment length is too small %u\n", + pool_capa.pkt.max_seg_len); + return -1; + } + + /* Create pool */ + odp_pool_param_init(&pool_param); + pool_param.type = ODP_POOL_PACKET; + pool_param.pkt.num = num_pkt; + pool_param.pkt.len = pkt_len; + pool_param.pkt.seg_len = seg_len; + + pool = odp_pool_create("packet gen pool", &pool_param); + + if (pool == ODP_POOL_INVALID) { + printf("Error: Pool create failed.\n"); + return -1; + } + + global->pool = pool; + + odp_pktio_param_init(&pktio_param); + pktio_param.in_mode = ODP_PKTIN_MODE_SCHED; + pktio_param.out_mode = ODP_PKTOUT_MODE_DIRECT; + + for (i = 0; i < num_pktio; i++) + global->pktio[i].pktio = ODP_PKTIO_INVALID; + + /* Open and configure interfaces */ + for (i = 0; i < num_pktio; i++) { + name = test_options->pktio_name[i]; + pktio = odp_pktio_open(name, pool, &pktio_param); + + if (pktio == ODP_PKTIO_INVALID) { + printf("Error (%s): Pktio open failed.\n", name); + return -1; + } + + global->pktio[i].pktio = pktio; + + odp_pktio_print(pktio); + + if (odp_pktio_capability(pktio, &pktio_capa)) { + printf("Error (%s): Pktio capability failed.\n", name); + return -1; + } + + if (num_rx > pktio_capa.max_input_queues) { + printf("Error (%s): Too many RX threads. Interface supports max %u input queues.\n", + name, pktio_capa.max_input_queues); + return -1; + } + + if (num_tx > (int)pktio_capa.max_output_queues) { + printf("Error (%s): Too many TX threads. Interface supports max %u output queues.\n", + name, pktio_capa.max_output_queues); + return -1; + } + + if (odp_pktio_mac_addr(pktio, + &global->pktio[i].eth_src.addr, + ODPH_ETHADDR_LEN) != ODPH_ETHADDR_LEN) { + printf("Error (%s): MAC address read failed.\n", name); + return -1; + } + + odp_pktio_config_init(&pktio_config); + pktio_config.parser.layer = ODP_PROTO_LAYER_ALL; + + odp_pktio_config(pktio, &pktio_config); + + odp_pktin_queue_param_init(&pktin_param); + + pktin_param.queue_param.sched.prio = ODP_SCHED_PRIO_DEFAULT; + pktin_param.queue_param.sched.sync = ODP_SCHED_SYNC_PARALLEL; + pktin_param.queue_param.sched.group = ODP_SCHED_GROUP_ALL; + pktin_param.hash_enable = 1; + pktin_param.hash_proto.proto.ipv4_udp = 1; + pktin_param.num_queues = num_rx; + + if (odp_pktin_queue_config(pktio, &pktin_param)) { + printf("Error (%s): Pktin config failed.\n", name); + return -1; + } + + odp_pktout_queue_param_init(&pktout_param); + pktout_param.op_mode = ODP_PKTIO_OP_MT_UNSAFE; + pktout_param.num_queues = num_tx; + + if (odp_pktout_queue_config(pktio, &pktout_param)) { + printf("Error (%s): Pktout config failed.\n", name); + return -1; + } + + if (odp_pktout_queue(pktio, pktout, num_tx) != num_tx) { + printf("Error (%s): Pktout queue request failed.\n", + name); + return -1; + } + + for (j = 0; j < num_tx; j++) + global->pktio[i].pktout[j] = pktout[j]; + } + + return 0; +} + +static int start_pktios(test_global_t *global) +{ + uint32_t i; + test_options_t *test_options = &global->test_options; + uint32_t num_pktio = test_options->num_pktio; + + for (i = 0; i < num_pktio; i++) { + if (odp_pktio_start(global->pktio[i].pktio)) { + printf("Error (%s): Pktio start failed.\n", + test_options->pktio_name[i]); + + return -1; + } + + global->pktio[i].started = 1; + } + + return 0; +} + +static int stop_pktios(test_global_t *global) +{ + uint32_t i; + odp_pktio_t pktio; + int ret = 0; + test_options_t *test_options = &global->test_options; + uint32_t num_pktio = test_options->num_pktio; + + for (i = 0; i < num_pktio; i++) { + pktio = global->pktio[i].pktio; + + if (pktio == ODP_PKTIO_INVALID || global->pktio[i].started == 0) + continue; + + if (odp_pktio_stop(pktio)) { + printf("Error (%s): Pktio stop failed.\n", + test_options->pktio_name[i]); + ret = -1; + } + } + + return ret; +} + +static int close_pktios(test_global_t *global) +{ + uint32_t i; + odp_pktio_t pktio; + test_options_t *test_options = &global->test_options; + uint32_t num_pktio = test_options->num_pktio; + int ret = 0; + + for (i = 0; i < num_pktio; i++) { + pktio = global->pktio[i].pktio; + + if (pktio == ODP_PKTIO_INVALID) + continue; + + if (odp_pktio_close(pktio)) { + printf("Error (%s): Pktio close failed.\n", + test_options->pktio_name[i]); + ret = -1; + } + } + + if (global->pool != ODP_POOL_INVALID && + odp_pool_destroy(global->pool)) { + printf("Error: Pool destroy failed.\n"); + ret = -1; + } + + return ret; +} + +static int rx_thread(void *arg) +{ + int i, thr, num; + uint32_t exit_test; + odp_time_t t1, t2; + thread_arg_t *thread_arg = arg; + test_global_t *global = thread_arg->global; + uint64_t rx_timeouts = 0; + uint64_t rx_packets = 0; + uint64_t rx_bytes = 0; + uint64_t nsec = 0; + int ret = 0; + int clock_started = 0; + int paused = 0; + int max_num = 32; + odp_event_t ev[max_num]; + + thr = odp_thread_id(); + global->stat[thr].thread_type = RX_THREAD; + + /* Start all workers at the same time */ + odp_barrier_wait(&global->barrier); + + while (1) { + num = odp_schedule_multi_no_wait(NULL, ev, max_num); + + exit_test = odp_atomic_load_u32(&global->exit_test); + if (num == 0 && exit_test) { + if (paused == 0) { + odp_schedule_pause(); + paused = 1; + } else { + /* Exit schedule loop after schedule paused and + * no more packets received */ + break; + } + } + + if (num == 0) { + rx_timeouts++; + continue; + } + + if (!clock_started) { + t1 = odp_time_local(); + clock_started = 1; + } + + for (i = 0; i < num; i++) { + odp_packet_t pkt; + + pkt = odp_packet_from_event(ev[i]); + rx_bytes += odp_packet_len(pkt); + } + + odp_event_free_multi(ev, num); + rx_packets += num; + } + + t2 = odp_time_local(); + + if (clock_started) + nsec = odp_time_diff_ns(t2, t1); + + /* Update stats*/ + global->stat[thr].time_nsec = nsec; + global->stat[thr].rx_timeouts = rx_timeouts; + global->stat[thr].rx_packets = rx_packets; + global->stat[thr].rx_bytes = rx_bytes; + + return ret; +} + +static void drain_queues(test_global_t *global) +{ + odp_event_t ev; + uint64_t wait_time = odp_schedule_wait_time(100 * ODP_TIME_MSEC_IN_NS); + + while ((ev = odp_schedule(NULL, wait_time)) != ODP_EVENT_INVALID) { + global->drained++; + odp_event_free(ev); + } +} + +static int init_packets(test_global_t *global, int pktio, + odp_packet_t packet[], uint32_t num, uint16_t seq) +{ + odp_packet_t pkt; + uint32_t i, j, pkt_len, seg_len, payload_len; + void *data; + uint8_t *u8; + odph_ethhdr_t *eth; + odph_ipv4hdr_t *ip; + odph_udphdr_t *udp; + test_options_t *test_options = &global->test_options; + uint32_t hdr_len = ODPH_ETHHDR_LEN + ODPH_IPV4HDR_LEN + ODPH_UDPHDR_LEN; + uint16_t udp_src = test_options->udp_src; + uint16_t udp_dst = test_options->udp_dst; + uint32_t udp_src_cnt = 0; + uint32_t udp_dst_cnt = 0; + + for (i = 0; i < num; i++) { + pkt = packet[i]; + pkt_len = odp_packet_len(pkt); + seg_len = odp_packet_seg_len(pkt); + data = odp_packet_data(pkt); + payload_len = pkt_len - hdr_len; + + if (seg_len < hdr_len) { + printf("Error: First segment too short %u\n", seg_len); + return -1; + } + + /* Ethernet */ + eth = data; + memcpy(eth->dst.addr, global->pktio[pktio].eth_dst.addr, 6); + memcpy(eth->src.addr, global->pktio[pktio].eth_src.addr, 6); + eth->type = odp_cpu_to_be_16(ODPH_ETHTYPE_IPV4); + + /* IPv4 */ + ip = (odph_ipv4hdr_t *)((uint8_t *)data + ODPH_ETHHDR_LEN); + memset(ip, 0, ODPH_IPV4HDR_LEN); + ip->ver_ihl = ODPH_IPV4 << 4 | ODPH_IPV4HDR_IHL_MIN; + ip->tot_len = odp_cpu_to_be_16(pkt_len - ODPH_ETHHDR_LEN); + ip->id = odp_cpu_to_be_16(seq + i); + ip->ttl = 64; + ip->proto = ODPH_IPPROTO_UDP; + ip->src_addr = odp_cpu_to_be_32(test_options->ipv4_src); + ip->dst_addr = odp_cpu_to_be_32(test_options->ipv4_dst); + ip->chksum = ~odp_chksum_ones_comp16(ip, ODPH_IPV4HDR_LEN); + + /* UDP */ + udp = (odph_udphdr_t *)((uint8_t *)data + ODPH_ETHHDR_LEN + + ODPH_IPV4HDR_LEN); + memset(udp, 0, ODPH_UDPHDR_LEN); + udp->src_port = odp_cpu_to_be_16(udp_src); + udp->dst_port = odp_cpu_to_be_16(udp_dst); + udp->length = odp_cpu_to_be_16(payload_len + ODPH_UDPHDR_LEN); + udp->chksum = 0; + + /* Init UDP payload until the end of the first segment */ + u8 = data; + u8 += hdr_len; + for (j = 0; j < seg_len - hdr_len; j++) + u8[j] = j; + + /* Insert UDP checksum */ + odp_packet_l3_offset_set(pkt, ODPH_ETHHDR_LEN); + odp_packet_l4_offset_set(pkt, + ODPH_ETHHDR_LEN + ODPH_IPV4HDR_LEN); + odp_packet_has_eth_set(pkt, 1); + odp_packet_has_ipv4_set(pkt, 1); + odp_packet_has_udp_set(pkt, 1); + udp->chksum = odph_ipv4_udp_chksum(pkt); + + /* Increment port numbers */ + if (test_options->c_mode.udp_src) { + udp_src_cnt++; + if (udp_src_cnt < test_options->c_mode.udp_src) { + udp_src++; + } else { + udp_src = test_options->udp_src; + udp_src_cnt = 0; + } + } + if (test_options->c_mode.udp_dst) { + udp_dst_cnt++; + if (udp_dst_cnt < test_options->c_mode.udp_dst) { + udp_dst++; + } else { + udp_dst = test_options->udp_dst; + udp_dst_cnt = 0; + } + } + } + + return 0; +} + +static inline int send_burst(odp_pktout_queue_t pktout, odp_packet_t pkt[], + int burst_size) { + int i, sent; + int num = burst_size; + odp_packet_t pkt_ref[burst_size]; + + for (i = 0; i < burst_size; i++) { + pkt_ref[i] = odp_packet_ref_static(pkt[i]); + + if (pkt_ref[i] == ODP_PACKET_INVALID) { + num = i; + break; + } + } + + if (odp_unlikely(num == 0)) + return 0; + + sent = odp_pktout_send(pktout, pkt_ref, num); + + if (odp_unlikely(sent < 0)) + sent = 0; + + if (odp_unlikely(sent != num)) { + uint32_t num_drop = num - sent; + + odp_packet_free_multi(&pkt_ref[sent], num_drop); + } + + return sent; +} + +static int tx_thread(void *arg) +{ + int i, thr, tx_thr, num_alloc; + uint32_t exit_test; + odp_time_t t1, t2, next_tmo; + uint64_t diff_ns, t1_nsec; + thread_arg_t *thread_arg = arg; + test_global_t *global = thread_arg->global; + test_options_t *test_options = &global->test_options; + odp_pool_t pool = global->pool; + uint64_t gap_nsec = test_options->gap_nsec; + uint64_t quit = test_options->quit; + uint64_t tx_timeouts = 0; + uint64_t tx_packets = 0; + uint64_t tx_drops = 0; + int ret = 0; + int burst_size = test_options->burst_size; + int bursts = test_options->bursts; + uint32_t num_tx = test_options->num_tx; + uint32_t pkt_len = test_options->pkt_len; + int num_pktio = test_options->num_pktio; + int num_pkt = num_pktio * bursts * burst_size; + odp_pktout_queue_t pktout[num_pktio]; + odp_packet_t pkt[num_pkt]; + + thr = odp_thread_id(); + tx_thr = thread_arg->tx_thr; + global->stat[thr].thread_type = TX_THREAD; + + num_alloc = odp_packet_alloc_multi(pool, pkt_len, pkt, num_pkt); + + if (num_alloc != num_pkt) { + printf("Error: alloc of %u packets failed (%i)\n", + num_pkt, thr); + ret = -1; + } + + /* Initialize packets per pktio interface */ + for (i = 0; ret == 0 && i < num_pktio; i++) { + int f = i * bursts * burst_size; + + pktout[i] = thread_arg->pktout[i]; + if (init_packets(global, i, &pkt[f], bursts * burst_size, f)) { + ret = -1; + break; + } + } + + /* Start all workers at the same time */ + odp_barrier_wait(&global->barrier); + + t1 = odp_time_local(); + + /* Start TX burst at different per thread offset */ + t1_nsec = odp_time_to_ns(t1) + gap_nsec + (tx_thr * gap_nsec / num_tx); + + while (ret == 0) { + exit_test = odp_atomic_load_u32(&global->exit_test); + if (exit_test) + break; + + if (quit && tx_timeouts >= quit) { + odp_atomic_inc_u32(&global->exit_test); + break; + } + + if (gap_nsec) { + uint64_t nsec = t1_nsec + tx_timeouts * gap_nsec; + + next_tmo = odp_time_local_from_ns(nsec); + odp_time_wait_until(next_tmo); + } + tx_timeouts++; + + /* Send bursts to each pktio */ + for (i = 0; i < num_pktio; i++) { + int sent, j; + int first = i * bursts * burst_size; + + for (j = 0; j < bursts; j++) { + sent = send_burst(pktout[i], + &pkt[first + j * burst_size], + burst_size); + + if (odp_unlikely(sent < 0)) { + ret = -1; + tx_drops += burst_size; + break; + } + + tx_packets += sent; + if (odp_unlikely(sent < burst_size)) + tx_drops += burst_size - sent; + } + } + } + + t2 = odp_time_local(); + diff_ns = odp_time_diff_ns(t2, t1); + + if (num_alloc > 0) + odp_packet_free_multi(pkt, num_alloc); + + /* Update stats */ + global->stat[thr].time_nsec = diff_ns; + global->stat[thr].tx_timeouts = tx_timeouts; + global->stat[thr].tx_packets = tx_packets; + global->stat[thr].tx_drops = tx_drops; + + return ret; +} + +static int start_workers(test_global_t *global, odp_instance_t instance) +{ + odph_thread_common_param_t thr_common; + int i, j, ret, tx_thr; + test_options_t *test_options = &global->test_options; + int num_pktio = test_options->num_pktio; + int num_rx = test_options->num_rx; + int num_cpu = test_options->num_cpu; + odph_thread_param_t thr_param[num_cpu]; + + memset(global->thread_tbl, 0, sizeof(global->thread_tbl)); + memset(thr_param, 0, sizeof(thr_param)); + memset(&thr_common, 0, sizeof(thr_common)); + + thr_common.instance = instance; + thr_common.cpumask = &global->cpumask; + + /* Receive threads */ + for (i = 0; i < num_rx; i++) { + thr_param[i].start = rx_thread; + thr_param[i].arg = &global->thread_arg[i]; + thr_param[i].thr_type = ODP_THREAD_WORKER; + } + + /* Transmit threads */ + tx_thr = 0; + for (i = num_rx; i < num_cpu; i++) { + for (j = 0; j < num_pktio; j++) { + odp_pktout_queue_t pktout; + + global->thread_arg[i].tx_thr = tx_thr; + + /* Dedicate a pktout queue per pktio interface + * (per TX thread) */ + pktout = global->pktio[j].pktout[tx_thr]; + global->thread_arg[i].pktout[j] = pktout; + } + + thr_param[i].start = tx_thread; + thr_param[i].arg = &global->thread_arg[i]; + thr_param[i].thr_type = ODP_THREAD_WORKER; + tx_thr++; + } + + ret = odph_thread_create(global->thread_tbl, &thr_common, thr_param, + num_cpu); + + if (ret != num_cpu) { + printf("Error: thread create failed %i\n", ret); + return -1; + } + + return 0; +} + +static int print_stat(test_global_t *global) +{ + int i, num_thr; + double rx_pkt_ave, rx_mbit_per_sec, tx_mbit_per_sec; + test_options_t *test_options = &global->test_options; + int num_rx = test_options->num_rx; + int num_tx = test_options->num_tx; + uint32_t pkt_len = test_options->pkt_len; + uint64_t rx_nsec_sum = 0; + uint64_t rx_pkt_sum = 0; + uint64_t rx_byte_sum = 0; + uint64_t rx_tmo_sum = 0; + uint64_t tx_nsec_sum = 0; + uint64_t tx_pkt_sum = 0; + uint64_t tx_drop_sum = 0; + uint64_t tx_tmo_sum = 0; + double rx_pkt_per_sec = 0.0; + double rx_byte_per_sec = 0.0; + double rx_pkt_len = 0.0; + double rx_sec = 0.0; + double tx_pkt_per_sec = 0.0; + double tx_sec = 0.0; + + printf("\nRESULTS PER THREAD\n"); + printf(" rx thread:\n"); + printf(" 1 2 3 4 5 6 7 8\n"); + printf(" ---------------------------------------------------------------------------------------\n"); + printf(" "); + + num_thr = 0; + for (i = 0; i < ODP_THREAD_COUNT_MAX; i++) { + if (global->stat[i].thread_type != RX_THREAD) + continue; + + if (num_thr && (num_thr % 8) == 0) + printf("\n "); + + printf("%10" PRIu64 " ", global->stat[i].rx_packets); + num_thr++; + } + + printf("\n\n"); + + printf(" tx thread:\n"); + printf(" 1 2 3 4 5 6 7 8\n"); + printf(" ---------------------------------------------------------------------------------------\n"); + printf(" "); + + num_thr = 0; + for (i = 0; i < ODP_THREAD_COUNT_MAX; i++) { + if (global->stat[i].thread_type != TX_THREAD) + continue; + + if (num_thr && (num_thr % 8) == 0) + printf("\n "); + + printf("%10" PRIu64 " ", global->stat[i].tx_packets); + num_thr++; + } + + printf("\n\n"); + + for (i = 0; i < ODP_THREAD_COUNT_MAX; i++) { + if (global->stat[i].thread_type == RX_THREAD) { + rx_tmo_sum += global->stat[i].rx_timeouts; + rx_pkt_sum += global->stat[i].rx_packets; + rx_byte_sum += global->stat[i].rx_bytes; + rx_nsec_sum += global->stat[i].time_nsec; + + } else if (global->stat[i].thread_type == TX_THREAD) { + tx_tmo_sum += global->stat[i].tx_timeouts; + tx_pkt_sum += global->stat[i].tx_packets; + tx_drop_sum += global->stat[i].tx_drops; + tx_nsec_sum += global->stat[i].time_nsec; + } + } + + rx_pkt_ave = (double)rx_pkt_sum / num_rx; + rx_sec = rx_nsec_sum / 1000000000.0; + tx_sec = tx_nsec_sum / 1000000000.0; + + /* Packets and bytes per thread per sec */ + if (rx_nsec_sum) { + rx_pkt_per_sec = (1000000000.0 * (double)rx_pkt_sum) / + (double)rx_nsec_sum; + + rx_byte_per_sec = 1000000000.0; + rx_byte_per_sec *= (rx_byte_sum + 24 * rx_pkt_sum); + rx_byte_per_sec /= (double)rx_nsec_sum; + } + + if (tx_nsec_sum) { + tx_pkt_per_sec = (1000000000.0 * (double)tx_pkt_sum) / + (double)tx_nsec_sum; + } + + /* Total Mbit/s */ + rx_mbit_per_sec = (num_rx * 8 * rx_byte_per_sec) / 1000000.0; + tx_mbit_per_sec = (num_tx * 8 * tx_pkt_per_sec * (pkt_len + 24)) + / 1000000.0; + + if (rx_pkt_sum) + rx_pkt_len = (double)rx_byte_sum / rx_pkt_sum; + + printf("TOTAL (%i rx and %i tx threads)\n", num_rx, num_tx); + printf(" rx timeouts: %" PRIu64 "\n", rx_tmo_sum); + printf(" rx time spent (sec): %.3f\n", rx_sec); + printf(" rx packets: %" PRIu64 "\n", rx_pkt_sum); + printf(" rx packets drained: %" PRIu64 "\n", global->drained); + printf(" rx packets per thr: %.1f\n", rx_pkt_ave); + printf(" rx packets per thr per sec: %.1f\n", rx_pkt_per_sec); + printf(" rx packets per sec: %.1f\n", num_rx * rx_pkt_per_sec); + printf(" rx ave packet len: %.1f\n", rx_pkt_len); + printf(" rx Mbit/s: %.1f\n", rx_mbit_per_sec); + printf("\n"); + printf(" tx timeouts: %" PRIu64 "\n", tx_tmo_sum); + printf(" tx time spent (sec): %.3f\n", tx_sec); + printf(" tx packets: %" PRIu64 "\n", tx_pkt_sum); + printf(" tx dropped packets: %" PRIu64 "\n", tx_drop_sum); + printf(" tx packets per thr per sec: %.1f\n", tx_pkt_per_sec); + printf(" tx packets per sec: %.1f\n", num_tx * tx_pkt_per_sec); + printf(" tx Mbit/s: %.1f\n", tx_mbit_per_sec); + printf("\n"); + + if (rx_pkt_sum < MIN_RX_PACKETS_CI) + return -1; + + return 0; +} + +static void sig_handler(int signo) +{ + (void)signo; + + if (test_global == NULL) + return; + + odp_atomic_add_u32(&test_global->exit_test, 1); +} + +int main(int argc, char **argv) +{ + odp_instance_t instance; + odp_init_t init; + test_global_t *global; + odp_shm_t shm; + int i; + int ret = 0; + + signal(SIGINT, sig_handler); + + /* List features not to be used */ + odp_init_param_init(&init); + init.not_used.feat.cls = 1; + init.not_used.feat.compress = 1; + init.not_used.feat.crypto = 1; + init.not_used.feat.ipsec = 1; + init.not_used.feat.timer = 1; + init.not_used.feat.tm = 1; + + /* Init ODP before calling anything else */ + if (odp_init_global(&instance, &init, NULL)) { + printf("Error: Global init failed.\n"); + return -1; + } + + /* Init this thread */ + if (odp_init_local(instance, ODP_THREAD_CONTROL)) { + printf("Error: Local init failed.\n"); + return -1; + } + + shm = odp_shm_reserve("packet_gen_global", sizeof(test_global_t), + ODP_CACHE_LINE_SIZE, 0); + + if (shm == ODP_SHM_INVALID) { + printf("Error: SHM reserve failed.\n"); + return -1; + } + + global = odp_shm_addr(shm); + test_global = global; + + memset(global, 0, sizeof(test_global_t)); + odp_atomic_init_u32(&global->exit_test, 0); + + for (i = 0; i < ODP_THREAD_COUNT_MAX; i++) + global->thread_arg[i].global = global; + + if (parse_options(argc, argv, global)) { + ret = -1; + goto term; + } + + odp_schedule_config(NULL); + + if (set_num_cpu(global)) { + ret = -1; + goto term; + } + + if (open_pktios(global)) { + ret = -1; + goto term; + } + + if (start_pktios(global)) { + ret = -1; + goto term; + } + + /* Start worker threads */ + start_workers(global, instance); + + /* Wait until workers have started. */ + odp_barrier_wait(&global->barrier); + + /* Wait workers to exit */ + odph_thread_join(global->thread_tbl, + global->test_options.num_cpu); + + if (stop_pktios(global)) + ret = -1; + + drain_queues(global); + + if (close_pktios(global)) + ret = -1; + + if (print_stat(global)) + ret = -2; + +term: + if (odp_shm_free(shm)) { + printf("Error: SHM free failed.\n"); + return -1; + } + + if (odp_term_local()) { + printf("Error: term local failed.\n"); + return -1; + } + + if (odp_term_global(instance)) { + printf("Error: term global failed.\n"); + return -1; + } + + return ret; +} commit 181773542664bddff206fedb62ba94c38118457d Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Wed Feb 12 15:31:28 2020 +0200 test: timer_perf: fix help text Fixed copy-paste error from help text. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Matias Elo <matias.elo(a)nokia.com> diff --git a/test/performance/odp_timer_perf.c b/test/performance/odp_timer_perf.c index 5a2317aab..7834d4e56 100644 --- a/test/performance/odp_timer_perf.c +++ b/test/performance/odp_timer_perf.c @@ -79,9 +79,9 @@ test_global_t test_global; static void print_usage(void) { printf("\n" - "Scheduler performance test\n" + "Timer performance test\n" "\n" - "Usage: odp_sched_perf [options]\n" + "Usage: odp_timer_perf [options]\n" "\n" " -c, --num_cpu Number of CPUs (worker threads). 0: all available CPUs. Default: 1\n" " -n, --num_tp Number of timer pools. Default: 1\n" ----------------------------------------------------------------------- Summary of changes: example/l3fwd/odp_l3fwd.c | 11 +- example/switch/odp_switch.c | 2 +- test/performance/.gitignore | 1 + test/performance/Makefile.am | 3 + test/performance/odp_l2fwd.c | 2 +- test/performance/odp_packet_gen.c | 1313 ++++++++++++++++++++ ...dp_sched_pktio_run.sh => odp_packet_gen_run.sh} | 55 +- test/performance/odp_pktio_ordered.c | 2 +- test/performance/odp_timer_perf.c | 4 +- 9 files changed, 1340 insertions(+), 53 deletions(-) create mode 100644 test/performance/odp_packet_gen.c copy test/performance/{odp_sched_pktio_run.sh => odp_packet_gen_run.sh} (58%) hooks/post-receive --

6 years, 6 months

[COMMIT] branch master updated. v1.21.0.0-424-g324a2a2c7

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, master has been updated via 324a2a2c7c1a9d35323483b679eedfebcb310827 (commit) via 9dc2dcff30a5bf8963b9947343cf79fe63013932 (commit) via f974cb930ffb6bc1f0084b23f8d2e877b75e44db (commit) via f1798c854568c4e521cf71842bb71e81a0904251 (commit) from 64b4c275b39373678eaad8294816e9937805de47 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 324a2a2c7c1a9d35323483b679eedfebcb310827 Author: Matias Elo <matias.elo(a)nokia.com> Date: Thu Feb 13 10:39:56 2020 +0200 build: check if libatomic is needed for __atomic_compare_exchange_n Double wide __atomic_compare_exchange_n() is required by ipfragreass example. Signed-off-by: Matias Elo <matias.elo(a)nokia.com> Reviewed-by: Petri Savolainen <petri.savolainen(a)nokia.com> diff --git a/example/Makefile.am b/example/Makefile.am index 28362f2c1..26723a9e4 100644 --- a/example/Makefile.am +++ b/example/Makefile.am @@ -3,7 +3,6 @@ SUBDIRS = classifier \ hello \ ipsec \ ipsec_api \ - ipfragreass \ ipsec_offload \ l2fwd_simple \ l3fwd \ @@ -15,3 +14,7 @@ SUBDIRS = classifier \ time \ timer \ traffic_mgmt + +if HAVE_DW_ATOMIC_CMP_EXC +SUBDIRS += ipfragreass +endif diff --git a/example/ipfragreass/Makefile.am b/example/ipfragreass/Makefile.am index b98e066f8..2cd61a39e 100644 --- a/example/ipfragreass/Makefile.am +++ b/example/ipfragreass/Makefile.am @@ -1,6 +1,6 @@ include $(top_srcdir)/example/Makefile.inc -LDADD += $(ATOMIC_LIBS) +LDADD += $(ATOMIC_LIBS) $(ATOMIC_LIBS_OPT) bin_PROGRAMS = odp_ipfragreass diff --git a/m4/odp_atomic.m4 b/m4/odp_atomic.m4 index 413dcbda9..c0068b5e4 100644 --- a/m4/odp_atomic.m4 +++ b/m4/odp_atomic.m4 @@ -14,6 +14,30 @@ if test "x$use_libatomic" = "xyes"; then ATOMIC_LIBS="-latomic" fi AC_SUBST([ATOMIC_LIBS]) + +# Double wide __atomic_compare_exchange_n is required by ipfragreass example +use_libatomic_opt=no; +have_atomic_cmp_exc=yes; + +AC_CHECK_SIZEOF([void *]) +AC_PREPROC_IFELSE( + [AC_LANG_SOURCE([ + #if SIZEOF_VOID_P == 8 + #error + #endif + ])], [plat64=no], [plat64=yes]) + +if test "x$plat64" = "xyes"; then + ODP_ATOMIC_NEEDED_128BIT_CMP_EXC([use_libatomic_opt=yes], [have_atomic_cmp_exc=no]) +else + ODP_ATOMIC_NEEDED_64BIT_CMP_EXC([use_libatomic_opt=yes], [have_atomic_cmp_exc=no]) +fi + +if test "x$use_libatomic_opt" = "xyes"; then + ATOMIC_LIBS_OPT="-latomic" +fi +AC_SUBST([ATOMIC_LIBS_OPT]) +AM_CONDITIONAL([HAVE_DW_ATOMIC_CMP_EXC], [test x$have_atomic_cmp_exc = xyes]) ]) # ODP_ATOMIC # ODP_ATOMIC_BUILTINS @@ -93,3 +117,64 @@ if test "x$odp_cv_atomic_needed_128bit" = "xyes" ; then [AC_MSG_FAILURE([__atomic_exchange_16 is not available])]) fi ]) # ODP_ATOMIC_NEEDED_128BIT + +# ODP_ATOMIC_NEEDED_64BIT_CMP_EXC([ACTION_IF_NEEDED], [ACTION_IF_NOT_AVAILABLE]) +# ------------------------------------------------------------------------------ +# +AC_DEFUN([ODP_ATOMIC_NEEDED_64BIT_CMP_EXC], [dnl +AC_CACHE_CHECK([whether -latomic is needed for 64-bit atomic compare exchange], + [odp_cv_atomic_needed_64bit_cmp_exc], [dnl +AC_LINK_IFELSE( + [AC_LANG_SOURCE([[ + #include <stdint.h> + static uint64_t loc; + int main(void) + { + uint64_t exp = 0; + uint64_t = __atomic_compare_exchange_n(&loc, &exp, 1, 1, + __ATOMIC_ACQUIRE, + __ATOMIC_RELAXED); + return 0; + } + ]])], + [odp_cv_atomic_needed_64bit_cmp_exc=no], + [odp_cv_atomic_needed_64bit_cmp_exc=yes])]) + +if test "x$odp_cv_atomic_needed_64bit_cmp_exc" = "xyes" ; then + AC_CHECK_LIB( + [atomic], [__atomic_compare_exchange_8], + [m4_default([$1], [:])], + [m4_default([$2], [:])]) +fi + +]) # ODP_ATOMIC_NEEDED_64BIT_CMP_EXC + +# ODP_ATOMIC_NEEDED_128BIT_CMP_EXC([ACTION_IF_NEEDED], [ACTION_IF_NOT_AVAILABLE]) +# ------------------------------------------------------------------------------- +# +AC_DEFUN([ODP_ATOMIC_NEEDED_128BIT_CMP_EXC], [dnl +AC_CACHE_CHECK([whether -latomic is needed for 128-bit atomic compare exchange], + [odp_cv_atomic_needed_128bit_cmp_exc], [dnl +AC_LINK_IFELSE( + [AC_LANG_SOURCE([[ + #include <stdint.h> + static __int128 loc; + int main(void) + { + __int128 exp = 0; + __int128 = __atomic_compare_exchange_n(&loc, &exp, 1, 1, + __ATOMIC_ACQUIRE, + __ATOMIC_RELAXED); + return 0; + } + ]])], + [odp_cv_atomic_needed_128bit_cmp_exc=no], + [odp_cv_atomic_needed_128bit_cmp_exc=yes])]) + +if test "x$odp_cv_atomic_needed_128bit_cmp_exc" = "xyes" ; then + AC_CHECK_LIB( + [atomic], [__atomic_compare_exchange_16], + [m4_default([$1], [:])], + [m4_default([$2], [:])]) +fi +]) # ODP_ATOMIC_NEEDED_128BIT_CMP_EXC commit 9dc2dcff30a5bf8963b9947343cf79fe63013932 Author: Matias Elo <matias.elo(a)nokia.com> Date: Wed Feb 12 16:24:45 2020 +0200 ci: use correct paths for pkg-config when cross compiling Signed-off-by: Matias Elo <matias.elo(a)nokia.com> Reviewed-by: Petri Savolainen <petri.savolainen(a)nokia.com> diff --git a/scripts/ci/build_arm64.sh b/scripts/ci/build_arm64.sh index 647dd29cf..abdc5acb8 100755 --- a/scripts/ci/build_arm64.sh +++ b/scripts/ci/build_arm64.sh @@ -11,4 +11,8 @@ else fi export CPPFLAGS="-I/usr/include/${TARGET_ARCH}/dpdk" +# Use target libraries +export PKG_CONFIG_PATH= +export PKG_CONFIG_LIBDIR=/usr/lib/${TARGET_ARCH}/pkgconfig + exec "$(dirname "$0")"/build.sh diff --git a/scripts/ci/build_armhf.sh b/scripts/ci/build_armhf.sh index c13acaa08..4ae0f19b1 100755 --- a/scripts/ci/build_armhf.sh +++ b/scripts/ci/build_armhf.sh @@ -15,4 +15,8 @@ export CXXFLAGS="-march=armv7-a" # No DPDK on ARMv7 export CONF="${CONF} --disable-dpdk" +# Use target libraries +export PKG_CONFIG_PATH= +export PKG_CONFIG_LIBDIR=/usr/lib/${TARGET_ARCH}/pkgconfig + exec "$(dirname "$0")"/build.sh diff --git a/scripts/ci/build_i386.sh b/scripts/ci/build_i386.sh index 17b6bf668..797dd454d 100755 --- a/scripts/ci/build_i386.sh +++ b/scripts/ci/build_i386.sh @@ -12,4 +12,8 @@ else fi export CPPFLAGS="-I/usr/include/i386-linux-gnu/dpdk" +# Use target libraries +export PKG_CONFIG_PATH= +export PKG_CONFIG_LIBDIR=/usr/lib/i386-linux-gnu/pkgconfig + exec "$(dirname "$0")"/build.sh diff --git a/scripts/ci/build_ppc64el.sh b/scripts/ci/build_ppc64el.sh index ca094c901..984481bb5 100755 --- a/scripts/ci/build_ppc64el.sh +++ b/scripts/ci/build_ppc64el.sh @@ -13,4 +13,8 @@ else fi export CPPFLAGS="-I/usr/include/${TARGET_ARCH}/dpdk" +# Use target libraries +export PKG_CONFIG_PATH= +export PKG_CONFIG_LIBDIR=/usr/lib/${TARGET_ARCH}/pkgconfig + exec "$(dirname "$0")"/build.sh commit f974cb930ffb6bc1f0084b23f8d2e877b75e44db Author: Matias Elo <matias.elo(a)nokia.com> Date: Tue Feb 11 15:37:45 2020 +0200 travis: change default ubuntu container version to 18.04 Separate DPDK v18.11 test has been removed. DPDK v18.11 is used now by default in Ubuntu 18.04 x86 builds. DPDK v17.11 is still used in Ubuntu 16.04 builds. Signed-off-by: Matias Elo <matias.elo(a)nokia.com> Reviewed-by: Petri Savolainen <petri.savolainen(a)nokia.com> diff --git a/.travis.yml b/.travis.yml index e75ec5dc6..d3fafcb0c 100644 --- a/.travis.yml +++ b/.travis.yml @@ -49,26 +49,22 @@ env: # you need to generate a new one at https://codecov.io specific for your # repo. - CODECOV_TOKEN=a733c34c-5f5c-4ff1-af4b-e9f5edb1ab5e - - OS=ubuntu_16.04 + - OS=ubuntu_18.04 - ARCH=x86_64 - CHECK=1 - NETMAP=0 matrix: - CHECK=0 CONF="CFLAGS=-O3" - CHECK=0 CONF="CFLAGS=-O0 --enable-debug --enable-debug-print" - - CHECK=0 OS=ubuntu_18.04 CONF="CFLAGS=-O3" - - CHECK=0 OS=ubuntu_18.04 CONF="CFLAGS=-O0 --enable-debug --enable-debug-print" - CHECK=0 CONF="--enable-lto" - CHECK=0 CONF="--enable-lto --disable-abi-compat" - - CHECK=0 OS=ubuntu_18.04 CONF="--enable-lto" - - CHECK=0 OS=ubuntu_18.04 CONF="--enable-lto --disable-abi-compat" - CHECK=0 ARCH=arm64 - CHECK=0 ARCH=armhf - - CHECK=0 ARCH=powerpc + - CHECK=0 ARCH=ppc64el - CHECK=0 ARCH=i386 - CHECK=0 ARCH=arm64 CONF="--disable-abi-compat" - CHECK=0 ARCH=armhf CONF="--disable-abi-compat" - - CHECK=0 ARCH=powerpc CONF="--disable-abi-compat" + - CHECK=0 ARCH=ppc64el CONF="--disable-abi-compat" - CHECK=0 ARCH=i386 CONF="--disable-abi-compat" - CONF="" - CONF="--disable-abi-compat" @@ -81,7 +77,7 @@ env: - CHECK=0 CONF="--enable-pcapng-support" - CHECK=0 OS=centos_7 - CONF="--without-openssl --without-pcap" - - OS=ubuntu_18.04 + - OS=ubuntu_16.04 matrix: exclude: @@ -93,10 +89,6 @@ matrix: env: CHECK=0 CONF="--enable-lto" - compiler: clang env: CHECK=0 CONF="--enable-lto --disable-abi-compat" - - compiler: clang - env: CHECK=0 OS=ubuntu_18.04 CONF="--enable-lto" - - compiler: clang - env: CHECK=0 OS=ubuntu_18.04 CONF="--enable-lto --disable-abi-compat" install: - if [ ${NETMAP} -eq 1 ] ; then @@ -117,14 +109,14 @@ script: docker run -i -t -v `pwd`:/odp --shm-size 8g -e CC="${CC}" -e CONF="${CONF}" - ${DOCKER_NAMESPACE}/travis-odp-lng-${OS} /odp/scripts/ci/build_${ARCH}.sh ; + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/build_${ARCH}.sh ; else echo "Running test" ; docker run --privileged -i -t -v `pwd`:/odp --shm-size 8g -e CC="${CC}" -e CONF="${CONF}" - ${DOCKER_NAMESPACE}/travis-odp-lng-${OS} /odp/scripts/ci/check.sh ; + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/check.sh ; fi jobs: include: @@ -137,7 +129,7 @@ jobs: -v `pwd`:/odp --shm-size 8g -e CODECOV_TOKEN="${CODECOV_TOKEN}" -e CC="${CC}" - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/coverage.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/coverage.sh - stage: test env: TEST=scheduler_sp compiler: gcc @@ -148,7 +140,7 @@ jobs: -e CC="${CC}" -e CONF="" -e ODP_SCHEDULER=sp - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/check.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/check.sh - stage: test env: TEST=scheduler_scalable compiler: gcc @@ -159,7 +151,7 @@ jobs: -e CC="${CC}" -e CONF="" -e ODP_SCHEDULER=scalable - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/check.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/check.sh - stage: test env: TEST=process_mode install: @@ -173,19 +165,7 @@ jobs: -e CONF="" -e ODP_CONFIG_FILE=/odp/platform/linux-generic/test/process-mode.conf -e ODPH_PROC_MODE=1 - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/check.sh - - stage: test - env: TEST=dpdk_18.11 - install: - - true - compiler: gcc - script: - - if [ -z "${DOCKER_NAMESPACE}" ] ; then export DOCKER_NAMESPACE="opendataplane"; fi - - docker run --privileged -i -t - -v `pwd`:/odp --shm-size 8g - -e CC="${CC}" - -e CONF="" - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04-dpdk_18.11 /odp/scripts/ci/check.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/check.sh - stage: test env: TEST=inline_timer install: @@ -198,7 +178,7 @@ jobs: -e CC="${CC}" -e CONF="" -e ODP_CONFIG_FILE=/odp/platform/linux-generic/test/inline-timer.conf - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/check_inline_timer.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/check_inline_timer.sh - stage: test env: TEST=packet_align install: @@ -211,7 +191,7 @@ jobs: -e CC="${CC}" -e CONF="" -e ODP_CONFIG_FILE=/odp/platform/linux-generic/test/packet_align.conf - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/check_pktio.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/check_pktio.sh - stage: test env: TEST=distcheck compiler: gcc @@ -221,7 +201,7 @@ jobs: -v `pwd`:/odp --shm-size 8g -e CC="${CC}" -e CONF="--enable-user-guides" - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/distcheck.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/distcheck.sh - stage: test env: TEST=distcheck_nonabi compiler: gcc @@ -231,7 +211,7 @@ jobs: -v `pwd`:/odp --shm-size 8g -e CC="${CC}" -e CONF="--enable-user-guides --disable-abi-compat" - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/distcheck.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/distcheck.sh - stage: test env: TEST=out_of_tree compiler: gcc @@ -241,7 +221,7 @@ jobs: -v `pwd`:/odp --shm-size 8g -e CC="${CC}" -e CONF="" - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/out_of_tree.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/out_of_tree.sh - stage: "build only" env: TEST=documentation compiler: gcc @@ -314,7 +294,7 @@ jobs: - if [ -z "${DOCKER_NAMESPACE}" ] ; then export DOCKER_NAMESPACE="opendataplane"; fi - docker run -i -t -v `pwd`:/odp --shm-size 8g -e CC="${CC}" - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/build_${ARCH}.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/build_${ARCH}.sh - stage: "build only" env: ARCH=x86_64 compiler: clang @@ -324,7 +304,7 @@ jobs: - if [ -z "${DOCKER_NAMESPACE}" ] ; then export DOCKER_NAMESPACE="opendataplane"; fi - docker run -i -t -v `pwd`:/odp --shm-size 8g -e CC="${CC}" - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/build_${ARCH}.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/build_${ARCH}.sh - stage: "build only" env: ARCH=arm64 install: @@ -333,7 +313,7 @@ jobs: - if [ -z "${DOCKER_NAMESPACE}" ] ; then export DOCKER_NAMESPACE="opendataplane"; fi - docker run -i -t -v `pwd`:/odp -e CC="${CC}" - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/build_${ARCH}.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/build_${ARCH}.sh - stage: "build only" env: ARCH=i386 install: @@ -342,7 +322,7 @@ jobs: - if [ -z "${DOCKER_NAMESPACE}" ] ; then export DOCKER_NAMESPACE="opendataplane"; fi - docker run -i -t -v `pwd`:/odp --shm-size 8g -e CC="${CC}" - ${DOCKER_NAMESPACE}/travis-odp-lng-ubuntu_16.04 /odp/scripts/ci/build_${ARCH}.sh + ${DOCKER_NAMESPACE}/travis-odp-${OS}-${ARCH} /odp/scripts/ci/build_${ARCH}.sh - stage: test canfail: yes env: TEST=checkpatch diff --git a/scripts/ci/build_armhf.sh b/scripts/ci/build_armhf.sh index 837561f83..c13acaa08 100755 --- a/scripts/ci/build_armhf.sh +++ b/scripts/ci/build_armhf.sh @@ -9,8 +9,10 @@ else export CC="${TARGET_ARCH}-gcc" export CXX="${TARGET_ARCH}-g++" fi -export CPPFLAGS="-I/usr/include/${TARGET_ARCH}/dpdk" export CFLAGS="-march=armv7-a" export CXXFLAGS="-march=armv7-a" +# No DPDK on ARMv7 +export CONF="${CONF} --disable-dpdk" + exec "$(dirname "$0")"/build.sh diff --git a/scripts/ci/build_powerpc.sh b/scripts/ci/build_ppc64el.sh similarity index 61% rename from scripts/ci/build_powerpc.sh rename to scripts/ci/build_ppc64el.sh index a213ee1d3..ca094c901 100755 --- a/scripts/ci/build_powerpc.sh +++ b/scripts/ci/build_ppc64el.sh @@ -1,15 +1,16 @@ #!/bin/bash set -e -export TARGET_ARCH=powerpc-linux-gnu +export TARGET_ARCH=powerpc64le-linux-gnu if [ "${CC#clang}" != "${CC}" ] ; then export CC="clang --target=${TARGET_ARCH}" export CXX="clang++ --target=${TARGET_ARCH}" + # DPDK clang build broken + export CONF="${CONF} --disable-dpdk" else export CC="${TARGET_ARCH}-gcc" export CXX="${TARGET_ARCH}-g++" fi -# No DPDK on PowerPC -export CONF="${CONF} --disable-dpdk" +export CPPFLAGS="-I/usr/include/${TARGET_ARCH}/dpdk" exec "$(dirname "$0")"/build.sh commit f1798c854568c4e521cf71842bb71e81a0904251 Author: Matias Elo <matias.elo(a)nokia.com> Date: Thu Feb 13 21:15:24 2020 +0200 linux-gen: pcap: disable loop support in process mode Reopening pcap interface, which is needed to implemement loop support, causes pcap internal failure in process mode. Signed-off-by: Matias Elo <matias.elo(a)nokia.com> Reviewed-by: Petri Savolainen <petri.savolainen(a)nokia.com> diff --git a/platform/linux-generic/pktio/pcap.c b/platform/linux-generic/pktio/pcap.c index 4332348b6..6ef998717 100644 --- a/platform/linux-generic/pktio/pcap.c +++ b/platform/linux-generic/pktio/pcap.c @@ -29,7 +29,8 @@ * doesn't exist it will be created, if it does exist it will * be overwritten. * loops the number of times to iterate through the input file, set - * to 0 to loop indefinitely. The default value is 1. + * to 0 to loop indefinitely. The default value is 1. Looping is + * only supported in thread mode (ODP_MEM_MODEL_THREAD). * * The total length of the string is limited by PKTIO_NAME_LEN. */ @@ -38,6 +39,7 @@ #include <odp_api.h> #include <odp/api/plat/packet_inlines.h> +#include <odp_global_data.h> #include <odp_packet_internal.h> #include <odp_packet_io_internal.h> @@ -199,6 +201,10 @@ static int _pcapif_reopen(pkt_pcap_t *pcap) { char errbuf[PCAP_ERRBUF_SIZE]; + /* Reopen causes pcap internal failure in process mode */ + if (odp_global_ro.init_param.mem_model == ODP_MEM_MODEL_PROCESS) + return 1; + if (pcap->loops != 0 && ++pcap->loop_cnt >= pcap->loops) return 1; diff --git a/test/performance/odp_pktio_ordered_run.sh b/test/performance/odp_pktio_ordered_run.sh index 09dd2ab98..d695be289 100755 --- a/test/performance/odp_pktio_ordered_run.sh +++ b/test/performance/odp_pktio_ordered_run.sh @@ -11,7 +11,7 @@ TEST_DIR="${TEST_DIR:-$(dirname $0)}" DURATION=5 LOG=odp_pktio_ordered.log LOOPS=100000000 -PASS_PPS=5000 +PASS_PPS=100 PCAP_IN=`find . ${TEST_SRC_DIR} $(dirname $0) -name udp64.pcap -print -quit` PCAP_OUT=/dev/null ----------------------------------------------------------------------- Summary of changes: .travis.yml | 58 ++++++----------- example/Makefile.am | 5 +- example/ipfragreass/Makefile.am | 2 +- m4/odp_atomic.m4 | 85 +++++++++++++++++++++++++ platform/linux-generic/pktio/pcap.c | 8 ++- scripts/ci/build_arm64.sh | 4 ++ scripts/ci/build_armhf.sh | 8 ++- scripts/ci/build_i386.sh | 4 ++ scripts/ci/build_powerpc.sh | 15 ----- scripts/ci/{build_armhf.sh => build_ppc64el.sh} | 10 ++- test/performance/odp_pktio_ordered_run.sh | 2 +- 11 files changed, 139 insertions(+), 62 deletions(-) delete mode 100755 scripts/ci/build_powerpc.sh copy scripts/ci/{build_armhf.sh => build_ppc64el.sh} (59%) hooks/post-receive --

6 years, 6 months

[COMMIT] branch master updated. v1.21.0.0-420-g64b4c275

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, master has been updated via 64b4c275b39373678eaad8294816e9937805de47 (commit) from 2356affd7f5b0c96b66e9b6c3174175bbfa00bd1 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 64b4c275b39373678eaad8294816e9937805de47 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Wed Feb 19 11:13:50 2020 +0200 linux-gen: timer: fix timer pool create sync with inline timer scan Set highest_tp_idx only after timer pool has been fully initialized. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-by: Matias Elo <matias.elo(a)nokia.com> diff --git a/platform/linux-generic/odp_timer.c b/platform/linux-generic/odp_timer.c index 3a82ee84..5fa02011 100644 --- a/platform/linux-generic/odp_timer.c +++ b/platform/linux-generic/odp_timer.c @@ -307,9 +307,6 @@ static odp_timer_pool_t timer_pool_new(const char *name, tp_idx = i; timer_global->num_timer_pools++; - if (tp_idx > timer_global->highest_tp_idx) - timer_global->highest_tp_idx = tp_idx; - odp_ticketlock_unlock(&timer_global->lock); sz0 = ROUNDUP_CACHE_LINE(sizeof(timer_pool_t)); @@ -377,7 +374,10 @@ static odp_timer_pool_t timer_pool_new(const char *name, } tp->tp_idx = tp_idx; odp_spinlock_init(&tp->lock); + tp->start_time = odp_time_global(); + odp_ticketlock_lock(&timer_global->lock); + /* Inline timer scan may find the timer pool after this */ timer_global->timer_pool[tp_idx] = tp; if (timer_global->num_timer_pools == 1) @@ -391,13 +391,18 @@ static odp_timer_pool_t timer_pool_new(const char *name, } odp_ticketlock_unlock(&timer_global->lock); + if (!odp_global_rw->inline_timers) { if (tp->param.clk_src == ODP_CLOCK_CPU) itimer_init(tp); + } else { + /* Update the highest index for inline timer scan */ + odp_ticketlock_lock(&timer_global->lock); + if (tp_idx > timer_global->highest_tp_idx) + timer_global->highest_tp_idx = tp_idx; + odp_ticketlock_unlock(&timer_global->lock); } - tp->start_time = odp_time_global(); - return timer_pool_to_hdl(tp); } ----------------------------------------------------------------------- Summary of changes: platform/linux-generic/odp_timer.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-) hooks/post-receive --

6 years, 6 months

← Newer
1
2
3
4
5
6
7
...
124
Older →

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

git-lng-odp