From 483eb2f56657e8e7f419ab1a4fab8dce9ade8609 Mon Sep 17 00:00:00 2001 From: Daniel Baumann Date: Sat, 27 Apr 2024 20:24:20 +0200 Subject: Adding upstream version 14.2.21. Signed-off-by: Daniel Baumann --- src/seastar/dpdk/examples/multi_process/Makefile | 45 + .../multi_process/client_server_mp/Makefile | 44 + .../client_server_mp/mp_client/Makefile | 48 + .../client_server_mp/mp_client/client.c | 302 +++++ .../client_server_mp/mp_server/Makefile | 61 + .../client_server_mp/mp_server/args.c | 172 +++ .../client_server_mp/mp_server/args.h | 39 + .../client_server_mp/mp_server/init.c | 304 +++++ .../client_server_mp/mp_server/init.h | 72 ++ .../client_server_mp/mp_server/main.c | 318 +++++ .../multi_process/client_server_mp/shared/common.h | 87 ++ .../examples/multi_process/l2fwd_fork/Makefile | 50 + .../dpdk/examples/multi_process/l2fwd_fork/flib.c | 312 +++++ .../dpdk/examples/multi_process/l2fwd_fork/flib.h | 149 +++ .../dpdk/examples/multi_process/l2fwd_fork/main.c | 1283 ++++++++++++++++++++ .../dpdk/examples/multi_process/simple_mp/Makefile | 50 + .../dpdk/examples/multi_process/simple_mp/main.c | 155 +++ .../examples/multi_process/simple_mp/mp_commands.c | 166 +++ .../examples/multi_process/simple_mp/mp_commands.h | 44 + .../examples/multi_process/symmetric_mp/Makefile | 50 + .../examples/multi_process/symmetric_mp/main.c | 471 +++++++ 21 files changed, 4222 insertions(+) create mode 100644 src/seastar/dpdk/examples/multi_process/Makefile create mode 100644 src/seastar/dpdk/examples/multi_process/client_server_mp/Makefile create mode 100644 src/seastar/dpdk/examples/multi_process/client_server_mp/mp_client/Makefile create mode 100644 src/seastar/dpdk/examples/multi_process/client_server_mp/mp_client/client.c create mode 100644 src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/Makefile create mode 100644 src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/args.c create mode 100644 src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/args.h create mode 100644 src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/init.c create mode 100644 src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/init.h create mode 100644 src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/main.c create mode 100644 src/seastar/dpdk/examples/multi_process/client_server_mp/shared/common.h create mode 100644 src/seastar/dpdk/examples/multi_process/l2fwd_fork/Makefile create mode 100644 src/seastar/dpdk/examples/multi_process/l2fwd_fork/flib.c create mode 100644 src/seastar/dpdk/examples/multi_process/l2fwd_fork/flib.h create mode 100644 src/seastar/dpdk/examples/multi_process/l2fwd_fork/main.c create mode 100644 src/seastar/dpdk/examples/multi_process/simple_mp/Makefile create mode 100644 src/seastar/dpdk/examples/multi_process/simple_mp/main.c create mode 100644 src/seastar/dpdk/examples/multi_process/simple_mp/mp_commands.c create mode 100644 src/seastar/dpdk/examples/multi_process/simple_mp/mp_commands.h create mode 100644 src/seastar/dpdk/examples/multi_process/symmetric_mp/Makefile create mode 100644 src/seastar/dpdk/examples/multi_process/symmetric_mp/main.c (limited to 'src/seastar/dpdk/examples/multi_process') diff --git a/src/seastar/dpdk/examples/multi_process/Makefile b/src/seastar/dpdk/examples/multi_process/Makefile new file mode 100644 index 00000000..6b315cc0 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/Makefile @@ -0,0 +1,45 @@ +# BSD LICENSE +# +# Copyright(c) 2010-2014 Intel Corporation. All rights reserved. +# All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of Intel Corporation nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +ifeq ($(RTE_SDK),) +$(error "Please define RTE_SDK environment variable") +endif + +# Default target, can be overriden by command line or environment +RTE_TARGET ?= x86_64-native-linuxapp-gcc + +include $(RTE_SDK)/mk/rte.vars.mk + +DIRS-$(CONFIG_RTE_EXEC_ENV_LINUXAPP) += client_server_mp +DIRS-$(CONFIG_RTE_EXEC_ENV_LINUXAPP) += simple_mp +DIRS-$(CONFIG_RTE_EXEC_ENV_LINUXAPP) += symmetric_mp + +include $(RTE_SDK)/mk/rte.extsubdir.mk diff --git a/src/seastar/dpdk/examples/multi_process/client_server_mp/Makefile b/src/seastar/dpdk/examples/multi_process/client_server_mp/Makefile new file mode 100644 index 00000000..89cc6bf8 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/client_server_mp/Makefile @@ -0,0 +1,44 @@ +# BSD LICENSE +# +# Copyright(c) 2010-2014 Intel Corporation. All rights reserved. +# All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of Intel Corporation nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +ifeq ($(RTE_SDK),) +$(error "Please define RTE_SDK environment variable") +endif + +# Default target, can be overriden by command line or environment +RTE_TARGET ?= x86_64-native-linuxapp-gcc + +include $(RTE_SDK)/mk/rte.vars.mk + +DIRS-$(CONFIG_RTE_EXEC_ENV_LINUXAPP) += mp_client +DIRS-$(CONFIG_RTE_EXEC_ENV_LINUXAPP) += mp_server + +include $(RTE_SDK)/mk/rte.extsubdir.mk diff --git a/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_client/Makefile b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_client/Makefile new file mode 100644 index 00000000..2688fed0 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_client/Makefile @@ -0,0 +1,48 @@ +# BSD LICENSE +# +# Copyright(c) 2010-2014 Intel Corporation. All rights reserved. +# All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of Intel Corporation nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +ifeq ($(RTE_SDK),) +$(error "Please define RTE_SDK environment variable") +endif + +# Default target, can be overriden by command line or environment +include $(RTE_SDK)/mk/rte.vars.mk + +# binary name +APP = mp_client + +# all source are stored in SRCS-y +SRCS-y := client.c + +CFLAGS += $(WERROR_FLAGS) -O3 +CFLAGS += -I$(SRCDIR)/../shared + +include $(RTE_SDK)/mk/rte.extapp.mk diff --git a/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_client/client.c b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_client/client.c new file mode 100644 index 00000000..01b535c2 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_client/client.c @@ -0,0 +1,302 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2016 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "common.h" + +/* Number of packets to attempt to read from queue */ +#define PKT_READ_SIZE ((uint16_t)32) + +/* our client id number - tells us which rx queue to read, and NIC TX + * queue to write to. */ +static uint8_t client_id = 0; + +#define MBQ_CAPACITY 32 + +/* maps input ports to output ports for packets */ +static uint8_t output_ports[RTE_MAX_ETHPORTS]; + +/* buffers up a set of packet that are ready to send */ +struct rte_eth_dev_tx_buffer *tx_buffer[RTE_MAX_ETHPORTS]; + +/* shared data from server. We update statistics here */ +static volatile struct tx_stats *tx_stats; + + +/* + * print a usage message + */ +static void +usage(const char *progname) +{ + printf("Usage: %s [EAL args] -- -n \n\n", progname); +} + +/* + * Convert the client id number from a string to an int. + */ +static int +parse_client_num(const char *client) +{ + char *end = NULL; + unsigned long temp; + + if (client == NULL || *client == '\0') + return -1; + + temp = strtoul(client, &end, 10); + if (end == NULL || *end != '\0') + return -1; + + client_id = (uint8_t)temp; + return 0; +} + +/* + * Parse the application arguments to the client app. + */ +static int +parse_app_args(int argc, char *argv[]) +{ + int option_index, opt; + char **argvopt = argv; + const char *progname = NULL; + static struct option lgopts[] = { /* no long options */ + {NULL, 0, 0, 0 } + }; + progname = argv[0]; + + while ((opt = getopt_long(argc, argvopt, "n:", lgopts, + &option_index)) != EOF){ + switch (opt){ + case 'n': + if (parse_client_num(optarg) != 0){ + usage(progname); + return -1; + } + break; + default: + usage(progname); + return -1; + } + } + return 0; +} + +/* + * Tx buffer error callback + */ +static void +flush_tx_error_callback(struct rte_mbuf **unsent, uint16_t count, + void *userdata) { + int i; + uint8_t port_id = (uintptr_t)userdata; + + tx_stats->tx_drop[port_id] += count; + + /* free the mbufs which failed from transmit */ + for (i = 0; i < count; i++) + rte_pktmbuf_free(unsent[i]); + +} + +static void +configure_tx_buffer(uint8_t port_id, uint16_t size) +{ + int ret; + + /* Initialize TX buffers */ + tx_buffer[port_id] = rte_zmalloc_socket("tx_buffer", + RTE_ETH_TX_BUFFER_SIZE(size), 0, + rte_eth_dev_socket_id(port_id)); + if (tx_buffer[port_id] == NULL) + rte_exit(EXIT_FAILURE, "Cannot allocate buffer for tx on port %u\n", + (unsigned) port_id); + + rte_eth_tx_buffer_init(tx_buffer[port_id], size); + + ret = rte_eth_tx_buffer_set_err_callback(tx_buffer[port_id], + flush_tx_error_callback, (void *)(intptr_t)port_id); + if (ret < 0) + rte_exit(EXIT_FAILURE, "Cannot set error callback for " + "tx buffer on port %u\n", (unsigned) port_id); +} + +/* + * set up output ports so that all traffic on port gets sent out + * its paired port. Index using actual port numbers since that is + * what comes in the mbuf structure. + */ +static void +configure_output_ports(const struct port_info *ports) +{ + int i; + if (ports->num_ports > RTE_MAX_ETHPORTS) + rte_exit(EXIT_FAILURE, "Too many ethernet ports. RTE_MAX_ETHPORTS = %u\n", + (unsigned)RTE_MAX_ETHPORTS); + for (i = 0; i < ports->num_ports - 1; i+=2){ + uint8_t p1 = ports->id[i]; + uint8_t p2 = ports->id[i+1]; + output_ports[p1] = p2; + output_ports[p2] = p1; + + configure_tx_buffer(p1, MBQ_CAPACITY); + configure_tx_buffer(p2, MBQ_CAPACITY); + + } +} + +/* + * This function performs routing of packets + * Just sends each input packet out an output port based solely on the input + * port it arrived on. + */ +static void +handle_packet(struct rte_mbuf *buf) +{ + int sent; + const uint8_t in_port = buf->port; + const uint8_t out_port = output_ports[in_port]; + struct rte_eth_dev_tx_buffer *buffer = tx_buffer[out_port]; + + sent = rte_eth_tx_buffer(out_port, client_id, buffer, buf); + if (sent) + tx_stats->tx[out_port] += sent; + +} + +/* + * Application main function - loops through + * receiving and processing packets. Never returns + */ +int +main(int argc, char *argv[]) +{ + const struct rte_memzone *mz; + struct rte_ring *rx_ring; + struct rte_mempool *mp; + struct port_info *ports; + int need_flush = 0; /* indicates whether we have unsent packets */ + int retval; + void *pkts[PKT_READ_SIZE]; + uint16_t sent; + + if ((retval = rte_eal_init(argc, argv)) < 0) + return -1; + argc -= retval; + argv += retval; + + if (parse_app_args(argc, argv) < 0) + rte_exit(EXIT_FAILURE, "Invalid command-line arguments\n"); + + if (rte_eth_dev_count() == 0) + rte_exit(EXIT_FAILURE, "No Ethernet ports - bye\n"); + + rx_ring = rte_ring_lookup(get_rx_queue_name(client_id)); + if (rx_ring == NULL) + rte_exit(EXIT_FAILURE, "Cannot get RX ring - is server process running?\n"); + + mp = rte_mempool_lookup(PKTMBUF_POOL_NAME); + if (mp == NULL) + rte_exit(EXIT_FAILURE, "Cannot get mempool for mbufs\n"); + + mz = rte_memzone_lookup(MZ_PORT_INFO); + if (mz == NULL) + rte_exit(EXIT_FAILURE, "Cannot get port info structure\n"); + ports = mz->addr; + tx_stats = &(ports->tx_stats[client_id]); + + configure_output_ports(ports); + + RTE_LOG(INFO, APP, "Finished Process Init.\n"); + + printf("\nClient process %d handling packets\n", client_id); + printf("[Press Ctrl-C to quit ...]\n"); + + for (;;) { + uint16_t i, rx_pkts; + uint8_t port; + + rx_pkts = rte_ring_dequeue_burst(rx_ring, pkts, + PKT_READ_SIZE, NULL); + + if (unlikely(rx_pkts == 0)){ + if (need_flush) + for (port = 0; port < ports->num_ports; port++) { + sent = rte_eth_tx_buffer_flush(ports->id[port], client_id, + tx_buffer[port]); + if (unlikely(sent)) + tx_stats->tx[port] += sent; + } + need_flush = 0; + continue; + } + + for (i = 0; i < rx_pkts; i++) + handle_packet(pkts[i]); + + need_flush = 1; + } +} diff --git a/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/Makefile b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/Makefile new file mode 100644 index 00000000..c29e4783 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/Makefile @@ -0,0 +1,61 @@ +# BSD LICENSE +# +# Copyright(c) 2010-2014 Intel Corporation. All rights reserved. +# All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of Intel Corporation nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +ifeq ($(RTE_SDK),) +$(error "Please define RTE_SDK environment variable") +endif + +# Default target, can be overriden by command line or environment +RTE_TARGET ?= x86_64-native-linuxapp-gcc + +include $(RTE_SDK)/mk/rte.vars.mk + +ifneq ($(CONFIG_RTE_EXEC_ENV),"linuxapp") +$(error This application can only operate in a linuxapp environment, \ +please change the definition of the RTE_TARGET environment variable) +endif + +# binary name +APP = mp_server + +# all source are stored in SRCS-y +SRCS-y := main.c init.c args.c + +INC := $(wildcard *.h) + +CFLAGS += $(WERROR_FLAGS) -O3 +CFLAGS += -I$(SRCDIR)/../shared + +# for newer gcc, e.g. 4.4, no-strict-aliasing may not be necessary +# and so the next line can be removed in those cases. +EXTRA_CFLAGS += -fno-strict-aliasing + +include $(RTE_SDK)/mk/rte.extapp.mk diff --git a/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/args.c b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/args.c new file mode 100644 index 00000000..bf8c666c --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/args.c @@ -0,0 +1,172 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include +#include +#include +#include +#include +#include + +#include +#include + +#include "common.h" +#include "args.h" +#include "init.h" + +/* global var for number of clients - extern in header */ +uint8_t num_clients; + +static const char *progname; + +/** + * Prints out usage information to stdout + */ +static void +usage(void) +{ + printf( + "%s [EAL options] -- -p PORTMASK -n NUM_CLIENTS [-s NUM_SOCKETS]\n" + " -p PORTMASK: hexadecimal bitmask of ports to use\n" + " -n NUM_CLIENTS: number of client processes to use\n" + , progname); +} + +/** + * The ports to be used by the application are passed in + * the form of a bitmask. This function parses the bitmask + * and places the port numbers to be used into the port[] + * array variable + */ +static int +parse_portmask(uint8_t max_ports, const char *portmask) +{ + char *end = NULL; + unsigned long pm; + uint8_t count = 0; + + if (portmask == NULL || *portmask == '\0') + return -1; + + /* convert parameter to a number and verify */ + pm = strtoul(portmask, &end, 16); + if (end == NULL || *end != '\0' || pm == 0) + return -1; + + /* loop through bits of the mask and mark ports */ + while (pm != 0){ + if (pm & 0x01){ /* bit is set in mask, use port */ + if (count >= max_ports) + printf("WARNING: requested port %u not present" + " - ignoring\n", (unsigned)count); + else + ports->id[ports->num_ports++] = count; + } + pm = (pm >> 1); + count++; + } + + return 0; +} + +/** + * Take the number of clients parameter passed to the app + * and convert to a number to store in the num_clients variable + */ +static int +parse_num_clients(const char *clients) +{ + char *end = NULL; + unsigned long temp; + + if (clients == NULL || *clients == '\0') + return -1; + + temp = strtoul(clients, &end, 10); + if (end == NULL || *end != '\0' || temp == 0) + return -1; + + num_clients = (uint8_t)temp; + return 0; +} + +/** + * The application specific arguments follow the DPDK-specific + * arguments which are stripped by the DPDK init. This function + * processes these application arguments, printing usage info + * on error. + */ +int +parse_app_args(uint8_t max_ports, int argc, char *argv[]) +{ + int option_index, opt; + char **argvopt = argv; + static struct option lgopts[] = { /* no long options */ + {NULL, 0, 0, 0 } + }; + progname = argv[0]; + + while ((opt = getopt_long(argc, argvopt, "n:p:", lgopts, + &option_index)) != EOF){ + switch (opt){ + case 'p': + if (parse_portmask(max_ports, optarg) != 0){ + usage(); + return -1; + } + break; + case 'n': + if (parse_num_clients(optarg) != 0){ + usage(); + return -1; + } + break; + default: + printf("ERROR: Unknown option '%c'\n", opt); + usage(); + return -1; + } + } + + if (ports->num_ports == 0 || num_clients == 0){ + usage(); + return -1; + } + + if (ports->num_ports % 2 != 0){ + printf("ERROR: application requires an even number of ports to use\n"); + return -1; + } + return 0; +} diff --git a/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/args.h b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/args.h new file mode 100644 index 00000000..23af1bd3 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/args.h @@ -0,0 +1,39 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef _ARGS_H_ +#define _ARGS_H_ + +int parse_app_args(uint8_t max_ports, int argc, char *argv[]); + +#endif /* ifndef _ARGS_H_ */ diff --git a/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/init.c b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/init.c new file mode 100644 index 00000000..ad941a7a --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/init.c @@ -0,0 +1,304 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "common.h" +#include "args.h" +#include "init.h" + +#define MBUFS_PER_CLIENT 1536 +#define MBUFS_PER_PORT 1536 +#define MBUF_CACHE_SIZE 512 + +#define RTE_MP_RX_DESC_DEFAULT 512 +#define RTE_MP_TX_DESC_DEFAULT 512 +#define CLIENT_QUEUE_RINGSIZE 128 + +#define NO_FLAGS 0 + +/* The mbuf pool for packet rx */ +struct rte_mempool *pktmbuf_pool; + +/* array of info/queues for clients */ +struct client *clients = NULL; + +/* the port details */ +struct port_info *ports; + +/** + * Initialise the mbuf pool for packet reception for the NIC, and any other + * buffer pools needed by the app - currently none. + */ +static int +init_mbuf_pools(void) +{ + const unsigned num_mbufs = (num_clients * MBUFS_PER_CLIENT) \ + + (ports->num_ports * MBUFS_PER_PORT); + + /* don't pass single-producer/single-consumer flags to mbuf create as it + * seems faster to use a cache instead */ + printf("Creating mbuf pool '%s' [%u mbufs] ...\n", + PKTMBUF_POOL_NAME, num_mbufs); + pktmbuf_pool = rte_pktmbuf_pool_create(PKTMBUF_POOL_NAME, num_mbufs, + MBUF_CACHE_SIZE, 0, RTE_MBUF_DEFAULT_BUF_SIZE, rte_socket_id()); + + return pktmbuf_pool == NULL; /* 0 on success */ +} + +/** + * Initialise an individual port: + * - configure number of rx and tx rings + * - set up each rx ring, to pull from the main mbuf pool + * - set up each tx ring + * - start the port and report its status to stdout + */ +static int +init_port(uint8_t port_num) +{ + /* for port configuration all features are off by default */ + const struct rte_eth_conf port_conf = { + .rxmode = { + .mq_mode = ETH_MQ_RX_RSS + } + }; + const uint16_t rx_rings = 1, tx_rings = num_clients; + const uint16_t rx_ring_size = RTE_MP_RX_DESC_DEFAULT; + const uint16_t tx_ring_size = RTE_MP_TX_DESC_DEFAULT; + + uint16_t q; + int retval; + + printf("Port %u init ... ", (unsigned)port_num); + fflush(stdout); + + /* Standard DPDK port initialisation - config port, then set up + * rx and tx rings */ + if ((retval = rte_eth_dev_configure(port_num, rx_rings, tx_rings, + &port_conf)) != 0) + return retval; + + for (q = 0; q < rx_rings; q++) { + retval = rte_eth_rx_queue_setup(port_num, q, rx_ring_size, + rte_eth_dev_socket_id(port_num), + NULL, pktmbuf_pool); + if (retval < 0) return retval; + } + + for ( q = 0; q < tx_rings; q ++ ) { + retval = rte_eth_tx_queue_setup(port_num, q, tx_ring_size, + rte_eth_dev_socket_id(port_num), + NULL); + if (retval < 0) return retval; + } + + rte_eth_promiscuous_enable(port_num); + + retval = rte_eth_dev_start(port_num); + if (retval < 0) return retval; + + printf( "done: \n"); + + return 0; +} + +/** + * Set up the DPDK rings which will be used to pass packets, via + * pointers, between the multi-process server and client processes. + * Each client needs one RX queue. + */ +static int +init_shm_rings(void) +{ + unsigned i; + unsigned socket_id; + const char * q_name; + const unsigned ringsize = CLIENT_QUEUE_RINGSIZE; + + clients = rte_malloc("client details", + sizeof(*clients) * num_clients, 0); + if (clients == NULL) + rte_exit(EXIT_FAILURE, "Cannot allocate memory for client program details\n"); + + for (i = 0; i < num_clients; i++) { + /* Create an RX queue for each client */ + socket_id = rte_socket_id(); + q_name = get_rx_queue_name(i); + clients[i].rx_q = rte_ring_create(q_name, + ringsize, socket_id, + RING_F_SP_ENQ | RING_F_SC_DEQ ); /* single prod, single cons */ + if (clients[i].rx_q == NULL) + rte_exit(EXIT_FAILURE, "Cannot create rx ring queue for client %u\n", i); + } + return 0; +} + +/* Check the link status of all ports in up to 9s, and print them finally */ +static void +check_all_ports_link_status(uint8_t port_num, uint32_t port_mask) +{ +#define CHECK_INTERVAL 100 /* 100ms */ +#define MAX_CHECK_TIME 90 /* 9s (90 * 100ms) in total */ + uint8_t portid, count, all_ports_up, print_flag = 0; + struct rte_eth_link link; + + printf("\nChecking link status"); + fflush(stdout); + for (count = 0; count <= MAX_CHECK_TIME; count++) { + all_ports_up = 1; + for (portid = 0; portid < port_num; portid++) { + if ((port_mask & (1 << ports->id[portid])) == 0) + continue; + memset(&link, 0, sizeof(link)); + rte_eth_link_get_nowait(ports->id[portid], &link); + /* print link status if flag set */ + if (print_flag == 1) { + if (link.link_status) + printf("Port %d Link Up - speed %u " + "Mbps - %s\n", ports->id[portid], + (unsigned)link.link_speed, + (link.link_duplex == ETH_LINK_FULL_DUPLEX) ? + ("full-duplex") : ("half-duplex\n")); + else + printf("Port %d Link Down\n", + (uint8_t)ports->id[portid]); + continue; + } + /* clear all_ports_up flag if any link down */ + if (link.link_status == ETH_LINK_DOWN) { + all_ports_up = 0; + break; + } + } + /* after finally printing all link status, get out */ + if (print_flag == 1) + break; + + if (all_ports_up == 0) { + printf("."); + fflush(stdout); + rte_delay_ms(CHECK_INTERVAL); + } + + /* set the print_flag if all ports up or timeout */ + if (all_ports_up == 1 || count == (MAX_CHECK_TIME - 1)) { + print_flag = 1; + printf("done\n"); + } + } +} + +/** + * Main init function for the multi-process server app, + * calls subfunctions to do each stage of the initialisation. + */ +int +init(int argc, char *argv[]) +{ + int retval; + const struct rte_memzone *mz; + uint8_t i, total_ports; + + /* init EAL, parsing EAL args */ + retval = rte_eal_init(argc, argv); + if (retval < 0) + return -1; + argc -= retval; + argv += retval; + + /* get total number of ports */ + total_ports = rte_eth_dev_count(); + + /* set up array for port data */ + mz = rte_memzone_reserve(MZ_PORT_INFO, sizeof(*ports), + rte_socket_id(), NO_FLAGS); + if (mz == NULL) + rte_exit(EXIT_FAILURE, "Cannot reserve memory zone for port information\n"); + memset(mz->addr, 0, sizeof(*ports)); + ports = mz->addr; + + /* parse additional, application arguments */ + retval = parse_app_args(total_ports, argc, argv); + if (retval != 0) + return -1; + + /* initialise mbuf pools */ + retval = init_mbuf_pools(); + if (retval != 0) + rte_exit(EXIT_FAILURE, "Cannot create needed mbuf pools\n"); + + /* now initialise the ports we will use */ + for (i = 0; i < ports->num_ports; i++) { + retval = init_port(ports->id[i]); + if (retval != 0) + rte_exit(EXIT_FAILURE, "Cannot initialise port %u\n", + (unsigned)i); + } + + check_all_ports_link_status(ports->num_ports, (~0x0)); + + /* initialise the client queues/rings for inter-eu comms */ + init_shm_rings(); + + return 0; +} diff --git a/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/init.h b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/init.h new file mode 100644 index 00000000..7333614d --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/init.h @@ -0,0 +1,72 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef _INIT_H_ +#define _INIT_H_ + +/* + * #include + * #include "args.h" + */ + +/* + * Define a client structure with all needed info, including + * stats from the clients. + */ +struct client { + struct rte_ring *rx_q; + unsigned client_id; + /* these stats hold how many packets the client will actually receive, + * and how many packets were dropped because the client's queue was full. + * The port-info stats, in contrast, record how many packets were received + * or transmitted on an actual NIC port. + */ + struct { + volatile uint64_t rx; + volatile uint64_t rx_drop; + } stats; +}; + +extern struct client *clients; + +/* the shared port information: port numbers, rx and tx stats etc. */ +extern struct port_info *ports; + +extern struct rte_mempool *pktmbuf_pool; +extern uint8_t num_clients; +extern unsigned num_sockets; +extern struct port_info *ports; + +int init(int argc, char *argv[]); + +#endif /* ifndef _INIT_H_ */ diff --git a/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/main.c b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/main.c new file mode 100644 index 00000000..c2b0261d --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/client_server_mp/mp_server/main.c @@ -0,0 +1,318 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "common.h" +#include "args.h" +#include "init.h" + +/* + * When doing reads from the NIC or the client queues, + * use this batch size + */ +#define PACKET_READ_SIZE 32 + +/* + * Local buffers to put packets in, used to send packets in bursts to the + * clients + */ +struct client_rx_buf { + struct rte_mbuf *buffer[PACKET_READ_SIZE]; + uint16_t count; +}; + +/* One buffer per client rx queue - dynamically allocate array */ +static struct client_rx_buf *cl_rx_buf; + +static const char * +get_printable_mac_addr(uint8_t port) +{ + static const char err_address[] = "00:00:00:00:00:00"; + static char addresses[RTE_MAX_ETHPORTS][sizeof(err_address)]; + + if (unlikely(port >= RTE_MAX_ETHPORTS)) + return err_address; + if (unlikely(addresses[port][0]=='\0')){ + struct ether_addr mac; + rte_eth_macaddr_get(port, &mac); + snprintf(addresses[port], sizeof(addresses[port]), + "%02x:%02x:%02x:%02x:%02x:%02x\n", + mac.addr_bytes[0], mac.addr_bytes[1], mac.addr_bytes[2], + mac.addr_bytes[3], mac.addr_bytes[4], mac.addr_bytes[5]); + } + return addresses[port]; +} + +/* + * This function displays the recorded statistics for each port + * and for each client. It uses ANSI terminal codes to clear + * screen when called. It is called from a single non-master + * thread in the server process, when the process is run with more + * than one lcore enabled. + */ +static void +do_stats_display(void) +{ + unsigned i, j; + const char clr[] = { 27, '[', '2', 'J', '\0' }; + const char topLeft[] = { 27, '[', '1', ';', '1', 'H','\0' }; + uint64_t port_tx[RTE_MAX_ETHPORTS], port_tx_drop[RTE_MAX_ETHPORTS]; + uint64_t client_tx[MAX_CLIENTS], client_tx_drop[MAX_CLIENTS]; + + /* to get TX stats, we need to do some summing calculations */ + memset(port_tx, 0, sizeof(port_tx)); + memset(port_tx_drop, 0, sizeof(port_tx_drop)); + memset(client_tx, 0, sizeof(client_tx)); + memset(client_tx_drop, 0, sizeof(client_tx_drop)); + + for (i = 0; i < num_clients; i++){ + const volatile struct tx_stats *tx = &ports->tx_stats[i]; + for (j = 0; j < ports->num_ports; j++){ + /* assign to local variables here, save re-reading volatile vars */ + const uint64_t tx_val = tx->tx[ports->id[j]]; + const uint64_t drop_val = tx->tx_drop[ports->id[j]]; + port_tx[j] += tx_val; + port_tx_drop[j] += drop_val; + client_tx[i] += tx_val; + client_tx_drop[i] += drop_val; + } + } + + /* Clear screen and move to top left */ + printf("%s%s", clr, topLeft); + + printf("PORTS\n"); + printf("-----\n"); + for (i = 0; i < ports->num_ports; i++) + printf("Port %u: '%s'\t", (unsigned)ports->id[i], + get_printable_mac_addr(ports->id[i])); + printf("\n\n"); + for (i = 0; i < ports->num_ports; i++){ + printf("Port %u - rx: %9"PRIu64"\t" + "tx: %9"PRIu64"\n", + (unsigned)ports->id[i], ports->rx_stats.rx[i], + port_tx[i]); + } + + printf("\nCLIENTS\n"); + printf("-------\n"); + for (i = 0; i < num_clients; i++){ + const unsigned long long rx = clients[i].stats.rx; + const unsigned long long rx_drop = clients[i].stats.rx_drop; + printf("Client %2u - rx: %9llu, rx_drop: %9llu\n" + " tx: %9"PRIu64", tx_drop: %9"PRIu64"\n", + i, rx, rx_drop, client_tx[i], client_tx_drop[i]); + } + + printf("\n"); +} + +/* + * The function called from each non-master lcore used by the process. + * The test_and_set function is used to randomly pick a single lcore on which + * the code to display the statistics will run. Otherwise, the code just + * repeatedly sleeps. + */ +static int +sleep_lcore(__attribute__((unused)) void *dummy) +{ + /* Used to pick a display thread - static, so zero-initialised */ + static rte_atomic32_t display_stats; + + /* Only one core should display stats */ + if (rte_atomic32_test_and_set(&display_stats)) { + const unsigned sleeptime = 1; + printf("Core %u displaying statistics\n", rte_lcore_id()); + + /* Longer initial pause so above printf is seen */ + sleep(sleeptime * 3); + + /* Loop forever: sleep always returns 0 or <= param */ + while (sleep(sleeptime) <= sleeptime) + do_stats_display(); + } + return 0; +} + +/* + * Function to set all the client statistic values to zero. + * Called at program startup. + */ +static void +clear_stats(void) +{ + unsigned i; + + for (i = 0; i < num_clients; i++) + clients[i].stats.rx = clients[i].stats.rx_drop = 0; +} + +/* + * send a burst of traffic to a client, assuming there are packets + * available to be sent to this client + */ +static void +flush_rx_queue(uint16_t client) +{ + uint16_t j; + struct client *cl; + + if (cl_rx_buf[client].count == 0) + return; + + cl = &clients[client]; + if (rte_ring_enqueue_bulk(cl->rx_q, (void **)cl_rx_buf[client].buffer, + cl_rx_buf[client].count, NULL) == 0){ + for (j = 0; j < cl_rx_buf[client].count; j++) + rte_pktmbuf_free(cl_rx_buf[client].buffer[j]); + cl->stats.rx_drop += cl_rx_buf[client].count; + } + else + cl->stats.rx += cl_rx_buf[client].count; + + cl_rx_buf[client].count = 0; +} + +/* + * marks a packet down to be sent to a particular client process + */ +static inline void +enqueue_rx_packet(uint8_t client, struct rte_mbuf *buf) +{ + cl_rx_buf[client].buffer[cl_rx_buf[client].count++] = buf; +} + +/* + * This function takes a group of packets and routes them + * individually to the client process. Very simply round-robins the packets + * without checking any of the packet contents. + */ +static void +process_packets(uint32_t port_num __rte_unused, + struct rte_mbuf *pkts[], uint16_t rx_count) +{ + uint16_t i; + uint8_t client = 0; + + for (i = 0; i < rx_count; i++) { + enqueue_rx_packet(client, pkts[i]); + + if (++client == num_clients) + client = 0; + } + + for (i = 0; i < num_clients; i++) + flush_rx_queue(i); +} + +/* + * Function called by the master lcore of the DPDK process. + */ +static void +do_packet_forwarding(void) +{ + unsigned port_num = 0; /* indexes the port[] array */ + + for (;;) { + struct rte_mbuf *buf[PACKET_READ_SIZE]; + uint16_t rx_count; + + /* read a port */ + rx_count = rte_eth_rx_burst(ports->id[port_num], 0, \ + buf, PACKET_READ_SIZE); + ports->rx_stats.rx[port_num] += rx_count; + + /* Now process the NIC packets read */ + if (likely(rx_count > 0)) + process_packets(port_num, buf, rx_count); + + /* move to next port */ + if (++port_num == ports->num_ports) + port_num = 0; + } +} + +int +main(int argc, char *argv[]) +{ + /* initialise the system */ + if (init(argc, argv) < 0 ) + return -1; + RTE_LOG(INFO, APP, "Finished Process Init.\n"); + + cl_rx_buf = calloc(num_clients, sizeof(cl_rx_buf[0])); + + /* clear statistics */ + clear_stats(); + + /* put all other cores to sleep bar master */ + rte_eal_mp_remote_launch(sleep_lcore, NULL, SKIP_MASTER); + + do_packet_forwarding(); + return 0; +} diff --git a/src/seastar/dpdk/examples/multi_process/client_server_mp/shared/common.h b/src/seastar/dpdk/examples/multi_process/client_server_mp/shared/common.h new file mode 100644 index 00000000..631c4632 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/client_server_mp/shared/common.h @@ -0,0 +1,87 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef _COMMON_H_ +#define _COMMON_H_ + +#define MAX_CLIENTS 16 + +/* + * Shared port info, including statistics information for display by server. + * Structure will be put in a memzone. + * - All port id values share one cache line as this data will be read-only + * during operation. + * - All rx statistic values share cache lines, as this data is written only + * by the server process. (rare reads by stats display) + * - The tx statistics have values for all ports per cache line, but the stats + * themselves are written by the clients, so we have a distinct set, on different + * cache lines for each client to use. + */ +struct rx_stats{ + uint64_t rx[RTE_MAX_ETHPORTS]; +} __rte_cache_aligned; + +struct tx_stats{ + uint64_t tx[RTE_MAX_ETHPORTS]; + uint64_t tx_drop[RTE_MAX_ETHPORTS]; +} __rte_cache_aligned; + +struct port_info { + uint8_t num_ports; + uint8_t id[RTE_MAX_ETHPORTS]; + volatile struct rx_stats rx_stats; + volatile struct tx_stats tx_stats[MAX_CLIENTS]; +}; + +/* define common names for structures shared between server and client */ +#define MP_CLIENT_RXQ_NAME "MProc_Client_%u_RX" +#define PKTMBUF_POOL_NAME "MProc_pktmbuf_pool" +#define MZ_PORT_INFO "MProc_port_info" + +/* + * Given the rx queue name template above, get the queue name + */ +static inline const char * +get_rx_queue_name(unsigned id) +{ + /* buffer for return value. Size calculated by %u being replaced + * by maximum 3 digits (plus an extra byte for safety) */ + static char buffer[sizeof(MP_CLIENT_RXQ_NAME) + 2]; + + snprintf(buffer, sizeof(buffer) - 1, MP_CLIENT_RXQ_NAME, id); + return buffer; +} + +#define RTE_LOGTYPE_APP RTE_LOGTYPE_USER1 + +#endif diff --git a/src/seastar/dpdk/examples/multi_process/l2fwd_fork/Makefile b/src/seastar/dpdk/examples/multi_process/l2fwd_fork/Makefile new file mode 100644 index 00000000..ff257a35 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/l2fwd_fork/Makefile @@ -0,0 +1,50 @@ +# BSD LICENSE +# +# Copyright(c) 2010-2014 Intel Corporation. All rights reserved. +# All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of Intel Corporation nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +ifeq ($(RTE_SDK),) +$(error "Please define RTE_SDK environment variable") +endif + +# Default target, can be overriden by command line or environment +RTE_TARGET ?= x86_64-native-linuxapp-gcc + +include $(RTE_SDK)/mk/rte.vars.mk + +# binary name +APP = l2fwd_fork + +# all source are stored in SRCS-y +SRCS-y := main.c flib.c + +CFLAGS += -O3 +CFLAGS += $(WERROR_FLAGS) + +include $(RTE_SDK)/mk/rte.extapp.mk diff --git a/src/seastar/dpdk/examples/multi_process/l2fwd_fork/flib.c b/src/seastar/dpdk/examples/multi_process/l2fwd_fork/flib.c new file mode 100644 index 00000000..85bbc2d3 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/l2fwd_fork/flib.c @@ -0,0 +1,312 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "flib.h" + +#define SIG_PARENT_EXIT SIGUSR1 + +struct lcore_stat { + pid_t pid; /**< pthread identifier */ + lcore_function_t *f; /**< function to call */ + void *arg; /**< argument of function */ + slave_exit_notify *cb_fn; +} __rte_cache_aligned; + + +static struct lcore_stat *core_cfg; +static uint16_t *lcore_cfg = NULL; + +/* signal handler to be notified after parent leaves */ +static void +sighand_parent_exit(int sig) +{ + printf("lcore = %u : Find parent leaves, sig=%d\n", rte_lcore_id(), + sig); + printf("Child leaving\n"); + exit(0); + + return; +} + +/** + * Real function entrance ran in slave process + **/ +static int +slave_proc_func(void) +{ + struct rte_config *config; + unsigned slave_id = rte_lcore_id(); + struct lcore_stat *cfg = &core_cfg[slave_id]; + + if (prctl(PR_SET_PDEATHSIG, SIG_PARENT_EXIT, 0, 0, 0, 0) != 0) + printf("Warning: Slave can't register for being notified in" + "case master process exited\n"); + else { + struct sigaction act; + memset(&act, 0 , sizeof(act)); + act.sa_handler = sighand_parent_exit; + if (sigaction(SIG_PARENT_EXIT, &act, NULL) != 0) + printf("Fail to register signal handler:%d\n", SIG_PARENT_EXIT); + } + + /* Set slave process to SECONDARY to avoid operation like dev_start/stop etc */ + config = rte_eal_get_configuration(); + if (NULL == config) + printf("Warning:Can't get rte_config\n"); + else + config->process_type = RTE_PROC_SECONDARY; + + printf("Core %u is ready (pid=%d)\n", slave_id, (int)cfg->pid); + + exit(cfg->f(cfg->arg)); +} + +/** + * function entrance ran in master thread, which will spawn slave process and wait until + * specific slave exited. + **/ +static int +lcore_func(void *arg __attribute__((unused))) +{ + unsigned slave_id = rte_lcore_id(); + struct lcore_stat *cfg = &core_cfg[slave_id]; + int pid, stat; + + if (rte_get_master_lcore() == slave_id) + return cfg->f(cfg->arg); + + /* fork a slave process */ + pid = fork(); + + if (pid == -1) { + printf("Failed to fork\n"); + return -1; + } else if (pid == 0) /* child */ + return slave_proc_func(); + else { /* parent */ + cfg->pid = pid; + + waitpid(pid, &stat, 0); + + cfg->pid = 0; + cfg->f = NULL; + cfg->arg = NULL; + /* Notify slave's exit if applicable */ + if(cfg->cb_fn) + cfg->cb_fn(slave_id, stat); + return stat; + } +} + +static int +lcore_id_init(void) +{ + int i; + /* Setup lcore ID allocation map */ + lcore_cfg = rte_zmalloc("LCORE_ID_MAP", + sizeof(uint16_t) * RTE_MAX_LCORE, + RTE_CACHE_LINE_SIZE); + + if(lcore_cfg == NULL) + rte_panic("Failed to malloc\n"); + + for (i = 0; i < RTE_MAX_LCORE; i++) { + if (rte_lcore_is_enabled(i)) + lcore_cfg[i] = 1; + } + return 0; +} + +int +flib_assign_lcore_id(void) +{ + unsigned i; + int ret; + + /** + * thread assigned a lcore id previously, or a slave thread. But still have + * a bug here: If the core mask includes core 0, and that core call this + * function, it still can get a new lcore id. + **/ + if (rte_lcore_id() != 0) + return -1; + + do { + /* Find a lcore id not used yet, avoid to use lcore ID 0 */ + for (i = 1; i < RTE_MAX_LCORE; i++) { + if (lcore_cfg[i] == 0) + break; + } + if (i == RTE_MAX_LCORE) + return -1; + + /* Assign new lcore id to this thread */ + + ret = rte_atomic16_cmpset(&lcore_cfg[i], 0, 1); + } while (unlikely(ret == 0)); + + RTE_PER_LCORE(_lcore_id) = i; + return i; +} + +void +flib_free_lcore_id(unsigned lcore_id) +{ + /* id is not valid or belongs to pinned core id */ + if (lcore_id >= RTE_MAX_LCORE || lcore_id == 0 || + rte_lcore_is_enabled(lcore_id)) + return; + + lcore_cfg[lcore_id] = 0; +} + +int +flib_register_slave_exit_notify(unsigned slave_id, + slave_exit_notify *cb) +{ + if (cb == NULL) + return -EFAULT; + + if (!rte_lcore_is_enabled(slave_id)) + return -ENOENT; + + core_cfg[slave_id].cb_fn = cb; + + return 0; +} + +enum slave_stat +flib_query_slave_status(unsigned slave_id) +{ + if (!rte_lcore_is_enabled(slave_id)) + return ST_FREEZE; + /* pid only be set when slave process spawned */ + if (core_cfg[slave_id].pid != 0) + return ST_RUN; + else + return ST_IDLE; +} + +int +flib_remote_launch(lcore_function_t *f, + void *arg, unsigned slave_id) +{ + if (f == NULL) + return -1; + + if (!rte_lcore_is_enabled(slave_id)) + return -1; + + /* Wait until specific lcore state change to WAIT */ + rte_eal_wait_lcore(slave_id); + + core_cfg[slave_id].f = f; + core_cfg[slave_id].arg = arg; + + return rte_eal_remote_launch(lcore_func, NULL, slave_id); +} + +int +flib_mp_remote_launch(lcore_function_t *f, void *arg, + enum rte_rmt_call_master_t call_master) +{ + int i; + + RTE_LCORE_FOREACH_SLAVE(i) { + core_cfg[i].arg = arg; + core_cfg[i].f = f; + } + + return rte_eal_mp_remote_launch(lcore_func, NULL, call_master); +} + +int +flib_init(void) +{ + if ((core_cfg = rte_zmalloc("core_cfg", + sizeof(struct lcore_stat) * RTE_MAX_LCORE, + RTE_CACHE_LINE_SIZE)) == NULL ) { + printf("rte_zmalloc failed\n"); + return -1; + } + + if (lcore_id_init() != 0) { + printf("lcore_id_init failed\n"); + return -1; + } + + return 0; +} diff --git a/src/seastar/dpdk/examples/multi_process/l2fwd_fork/flib.h b/src/seastar/dpdk/examples/multi_process/l2fwd_fork/flib.h new file mode 100644 index 00000000..711e3b6d --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/l2fwd_fork/flib.h @@ -0,0 +1,149 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef __FLIB_H +#define __FLIB_H + +/* callback function pointer when specific slave leaves */ +typedef void (slave_exit_notify)(unsigned slaveid, int stat); + +enum slave_stat{ + ST_FREEZE = 1, + ST_IDLE, + ST_RUN, + ST_ZOMBIE, /* Not implemented yet */ +}; + +/** + * Initialize the fork lib. + * + * @return + * - 0 : fork lib initialized successfully + * - -1 : fork lib initialized failed + */ +int flib_init(void); + +/** + * Check that every SLAVE lcores are in WAIT state, then call + * flib_remote_launch() for all of them. If call_master is true + * (set to CALL_MASTER), also call the function on the master lcore. + * + * @param f: + * function pointer need to run + * @param arg: + * argument for f to carry + * @param call_master + * - SKIP_MASTER : only launch function on slave lcores + * - CALL_MASTER : launch function on master and slave lcores + * @return + * - 0 : function execute successfully + * - -1 : function execute failed + */ +int flib_mp_remote_launch(lcore_function_t *f, + void *arg, enum rte_rmt_call_master_t call_master); + +/** + * Send a message to a slave lcore identified by slave_id to call a + * function f with argument arg. + * + * @param f: + * function pointer need to run + * @param arg: + * argument for f to carry + * @param slave_id + * slave lcore id to run on + * @return + * - 0 : function execute successfully + * - -1 : function execute failed + */ +int flib_remote_launch(lcore_function_t *f, + void *arg, unsigned slave_id); + +/** + * Query the running stat for specific slave, wont' work in with master id + * + * @param slave_id: + * lcore id which should not be master id + * @return + * - ST_FREEZE : lcore is not in enabled core mask + * - ST_IDLE : lcore is idle + * - ST_RUN : lcore is running something + */ +enum slave_stat +flib_query_slave_status(unsigned slave_id); + +/** + * Register a callback function to be notified in case specific slave exit. + * + * @param slave_id: + * lcore id which should not be master id + * @param cb: + * callback pointer to register + * @return + * - 0 : function execute successfully + * - -EFAULT : argument error + * - -ENOENT : slave_id not correct + */ +int flib_register_slave_exit_notify(unsigned slave_id, + slave_exit_notify *cb); + +/** + * Assign a lcore ID to non-slave thread. Non-slave thread refers to thread that + * not created by function rte_eal_remote_launch or rte_eal_mp_remote_launch. + * These threads can either bind lcore or float among differnt lcores. + * This lcore ID will be unique in multi-thread or multi-process DPDK running + * environment, then it can benefit from using the cache mechanism provided in + * mempool library. + * After calling successfully, use rte_lcore_id() to get the assigned lcore ID, but + * other lcore funtions can't guarantee to work correctly. + * + * @return + * - -1 : can't assign a lcore id with 3 possibilities. + * - it's not non-slave thread. + * - it had assign a lcore id previously + * - the lcore id is running out. + * - > 0 : the assigned lcore id. + */ +int flib_assign_lcore_id(void); + +/** + * Free the lcore_id that assigned in flib_assign_lcore_id(). + * call it in case non-slave thread is leaving or left. + * + * @param lcore_id + * The identifier of the lcore, which MUST be between 1 and + * RTE_MAX_LCORE-1. + */ +void flib_free_lcore_id(unsigned lcore_id); + +#endif /* __FLIB_H */ diff --git a/src/seastar/dpdk/examples/multi_process/l2fwd_fork/main.c b/src/seastar/dpdk/examples/multi_process/l2fwd_fork/main.c new file mode 100644 index 00000000..d922522f --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/l2fwd_fork/main.c @@ -0,0 +1,1283 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2016 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#define _GNU_SOURCE +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "flib.h" + +#define RTE_LOGTYPE_L2FWD RTE_LOGTYPE_USER1 +#define MBUF_NAME "mbuf_pool_%d" +#define MBUF_DATA_SIZE RTE_MBUF_DEFAULT_BUF_SIZE +#define NB_MBUF 8192 +#define RING_MASTER_NAME "l2fwd_ring_m2s_" +#define RING_SLAVE_NAME "l2fwd_ring_s2m_" +#define MAX_NAME_LEN 32 +/* RECREATE flag indicate needs initialize resource and launch slave_core again */ +#define SLAVE_RECREATE_FLAG 0x1 +/* RESTART flag indicate needs restart port and send START command again */ +#define SLAVE_RESTART_FLAG 0x2 +#define INVALID_MAPPING_ID ((unsigned)LCORE_ID_ANY) +/* Maximum message buffer per slave */ +#define NB_CORE_MSGBUF 32 +enum l2fwd_cmd{ + CMD_START, + CMD_STOP, +}; + +#define MAX_PKT_BURST 32 +#define BURST_TX_DRAIN_US 100 /* TX drain every ~100us */ + +/* + * Configurable number of RX/TX ring descriptors + */ +#define RTE_TEST_RX_DESC_DEFAULT 128 +#define RTE_TEST_TX_DESC_DEFAULT 512 +static uint16_t nb_rxd = RTE_TEST_RX_DESC_DEFAULT; +static uint16_t nb_txd = RTE_TEST_TX_DESC_DEFAULT; + +/* ethernet addresses of ports */ +static struct ether_addr l2fwd_ports_eth_addr[RTE_MAX_ETHPORTS]; + +/* mask of enabled ports */ +static uint32_t l2fwd_enabled_port_mask = 0; + +/* list of enabled ports */ +static uint32_t l2fwd_dst_ports[RTE_MAX_ETHPORTS]; + +static unsigned int l2fwd_rx_queue_per_lcore = 1; + +struct mbuf_table { + unsigned len; + struct rte_mbuf *m_table[MAX_PKT_BURST]; +}; + +#define MAX_RX_QUEUE_PER_LCORE 16 +#define MAX_TX_QUEUE_PER_PORT 16 +struct lcore_queue_conf { + unsigned n_rx_port; + unsigned rx_port_list[MAX_RX_QUEUE_PER_LCORE]; +} __rte_cache_aligned; +struct lcore_queue_conf lcore_queue_conf[RTE_MAX_LCORE]; + +struct rte_eth_dev_tx_buffer *tx_buffer[RTE_MAX_ETHPORTS]; + +struct lcore_resource_struct { + int enabled; /* Only set in case this lcore involved into packet forwarding */ + int flags; /* Set only slave need to restart or recreate */ + unsigned lcore_id; /* lcore ID */ + unsigned pair_id; /* dependency lcore ID on port */ + char ring_name[2][MAX_NAME_LEN]; + /* ring[0] for master send cmd, slave read */ + /* ring[1] for slave send ack, master read */ + struct rte_ring *ring[2]; + int port_num; /* Total port numbers */ + uint8_t port[RTE_MAX_ETHPORTS]; /* Port id for that lcore to receive packets */ +}__attribute__((packed)) __rte_cache_aligned; + +static struct lcore_resource_struct lcore_resource[RTE_MAX_LCORE]; +static struct rte_mempool *message_pool; +static rte_spinlock_t res_lock = RTE_SPINLOCK_INITIALIZER; +/* use floating processes */ +static int float_proc = 0; +/* Save original cpu affinity */ +struct cpu_aff_arg{ + cpu_set_t set; + size_t size; +}cpu_aff; + +static const struct rte_eth_conf port_conf = { + .rxmode = { + .split_hdr_size = 0, + .header_split = 0, /**< Header Split disabled */ + .hw_ip_checksum = 0, /**< IP checksum offload disabled */ + .hw_vlan_filter = 0, /**< VLAN filtering disabled */ + .jumbo_frame = 0, /**< Jumbo Frame Support disabled */ + .hw_strip_crc = 1, /**< CRC stripped by hardware */ + }, + .txmode = { + .mq_mode = ETH_MQ_TX_NONE, + }, +}; + +static struct rte_mempool * l2fwd_pktmbuf_pool[RTE_MAX_ETHPORTS]; + +/* Per-port statistics struct */ +struct l2fwd_port_statistics { + uint64_t tx; + uint64_t rx; + uint64_t dropped; +} __rte_cache_aligned; +struct l2fwd_port_statistics *port_statistics; +/** + * pointer to lcore ID mapping array, used to return lcore id in case slave + * process exited unexpectedly, use only floating process option applied + **/ +unsigned *mapping_id; + +/* A tsc-based timer responsible for triggering statistics printout */ +#define TIMER_MILLISECOND 2000000ULL /* around 1ms at 2 Ghz */ +#define MAX_TIMER_PERIOD 86400 /* 1 day max */ +static int64_t timer_period = 10 * TIMER_MILLISECOND * 1000; /* default period is 10 seconds */ + +static int l2fwd_launch_one_lcore(void *dummy); + +/* Print out statistics on packets dropped */ +static void +print_stats(void) +{ + uint64_t total_packets_dropped, total_packets_tx, total_packets_rx; + unsigned portid; + + total_packets_dropped = 0; + total_packets_tx = 0; + total_packets_rx = 0; + + const char clr[] = { 27, '[', '2', 'J', '\0' }; + const char topLeft[] = { 27, '[', '1', ';', '1', 'H','\0' }; + + /* Clear screen and move to top left */ + printf("%s%s", clr, topLeft); + + printf("\nPort statistics ===================================="); + + for (portid = 0; portid < RTE_MAX_ETHPORTS; portid++) { + /* skip disabled ports */ + if ((l2fwd_enabled_port_mask & (1 << portid)) == 0) + continue; + printf("\nStatistics for port %u ------------------------------" + "\nPackets sent: %24"PRIu64 + "\nPackets received: %20"PRIu64 + "\nPackets dropped: %21"PRIu64, + portid, + port_statistics[portid].tx, + port_statistics[portid].rx, + port_statistics[portid].dropped); + + total_packets_dropped += port_statistics[portid].dropped; + total_packets_tx += port_statistics[portid].tx; + total_packets_rx += port_statistics[portid].rx; + } + printf("\nAggregate statistics ===============================" + "\nTotal packets sent: %18"PRIu64 + "\nTotal packets received: %14"PRIu64 + "\nTotal packets dropped: %15"PRIu64, + total_packets_tx, + total_packets_rx, + total_packets_dropped); + printf("\n====================================================\n"); +} + +static int +clear_cpu_affinity(void) +{ + int s; + + s = sched_setaffinity(0, cpu_aff.size, &cpu_aff.set); + if (s != 0) { + printf("sched_setaffinity failed:%s\n", strerror(errno)); + return -1; + } + + return 0; +} + +static int +get_cpu_affinity(void) +{ + int s; + + cpu_aff.size = sizeof(cpu_set_t); + CPU_ZERO(&cpu_aff.set); + + s = sched_getaffinity(0, cpu_aff.size, &cpu_aff.set); + if (s != 0) { + printf("sched_getaffinity failed:%s\n", strerror(errno)); + return -1; + } + + return 0; +} + +/** + * This fnciton demonstrates the approach to create ring in first instance + * or re-attach an existed ring in later instance. + **/ +static struct rte_ring * +create_ring(const char *name, unsigned count, + int socket_id,unsigned flags) +{ + struct rte_ring *ring; + + if (name == NULL) + return NULL; + + /* If already create, just attached it */ + if (likely((ring = rte_ring_lookup(name)) != NULL)) + return ring; + + /* First call it, create one */ + return rte_ring_create(name, count, socket_id, flags); +} + +/* Malloc with rte_malloc on structures that shared by master and slave */ +static int +l2fwd_malloc_shared_struct(void) +{ + port_statistics = rte_zmalloc("port_stat", + sizeof(struct l2fwd_port_statistics) * RTE_MAX_ETHPORTS, + 0); + if (port_statistics == NULL) + return -1; + + /* allocate mapping_id array */ + if (float_proc) { + int i; + mapping_id = rte_malloc("mapping_id", sizeof(unsigned) * RTE_MAX_LCORE, + 0); + + if (mapping_id == NULL) + return -1; + + for (i = 0 ;i < RTE_MAX_LCORE; i++) + mapping_id[i] = INVALID_MAPPING_ID; + } + return 0; +} + +/* Create ring which used for communicate among master and slave */ +static int +create_ms_ring(unsigned slaveid) +{ + unsigned flag = RING_F_SP_ENQ | RING_F_SC_DEQ; + struct lcore_resource_struct *res = &lcore_resource[slaveid]; + unsigned socketid = rte_socket_id(); + + /* Always assume create ring on master socket_id */ + /* Default only create a ring size 32 */ + snprintf(res->ring_name[0], MAX_NAME_LEN, "%s%u", + RING_MASTER_NAME, slaveid); + if ((res->ring[0] = create_ring(res->ring_name[0], NB_CORE_MSGBUF, + socketid, flag)) == NULL) { + printf("Create m2s ring %s failed\n", res->ring_name[0]); + return -1; + } + + snprintf(res->ring_name[1], MAX_NAME_LEN, "%s%u", + RING_SLAVE_NAME, slaveid); + if ((res->ring[1] = create_ring(res->ring_name[1], NB_CORE_MSGBUF, + socketid, flag)) == NULL) { + printf("Create s2m ring %s failed\n", res->ring_name[1]); + return -1; + } + + return 0; +} + +/* send command to pair in paired master and slave ring */ +static inline int +sendcmd(unsigned slaveid, enum l2fwd_cmd cmd, int is_master) +{ + struct lcore_resource_struct *res = &lcore_resource[slaveid]; + void *msg; + int fd = !is_master; + + /* Only check master, it must be enabled and running if it is slave */ + if (is_master && !res->enabled) + return -1; + + if (res->ring[fd] == NULL) + return -1; + + if (rte_mempool_get(message_pool, &msg) < 0) { + printf("Error to get message buffer\n"); + return -1; + } + + *(enum l2fwd_cmd *)msg = cmd; + + if (rte_ring_enqueue(res->ring[fd], msg) != 0) { + printf("Enqueue error\n"); + rte_mempool_put(message_pool, msg); + return -1; + } + + return 0; +} + +/* Get command from pair in paired master and slave ring */ +static inline int +getcmd(unsigned slaveid, enum l2fwd_cmd *cmd, int is_master) +{ + struct lcore_resource_struct *res = &lcore_resource[slaveid]; + void *msg; + int fd = !!is_master; + int ret; + /* Only check master, it must be enabled and running if it is slave */ + if (is_master && (!res->enabled)) + return -1; + + if (res->ring[fd] == NULL) + return -1; + + ret = rte_ring_dequeue(res->ring[fd], &msg); + + if (ret == 0) { + *cmd = *(enum l2fwd_cmd *)msg; + rte_mempool_put(message_pool, msg); + } + return ret; +} + +/* Master send command to slave and wait until ack received or error met */ +static int +master_sendcmd_with_ack(unsigned slaveid, enum l2fwd_cmd cmd) +{ + enum l2fwd_cmd ack_cmd; + int ret = -1; + + if (sendcmd(slaveid, cmd, 1) != 0) + rte_exit(EXIT_FAILURE, "Failed to send message\n"); + + /* Get ack */ + while (1) { + ret = getcmd(slaveid, &ack_cmd, 1); + if (ret == 0 && cmd == ack_cmd) + break; + + /* If slave not running yet, return an error */ + if (flib_query_slave_status(slaveid) != ST_RUN) { + ret = -ENOENT; + break; + } + } + + return ret; +} + +/* restart all port that assigned to that slave lcore */ +static int +reset_slave_all_ports(unsigned slaveid) +{ + struct lcore_resource_struct *slave = &lcore_resource[slaveid]; + int i, ret = 0; + + /* stop/start port */ + for (i = 0; i < slave->port_num; i++) { + char buf_name[RTE_MEMPOOL_NAMESIZE]; + struct rte_mempool *pool; + printf("Stop port :%d\n", slave->port[i]); + rte_eth_dev_stop(slave->port[i]); + snprintf(buf_name, RTE_MEMPOOL_NAMESIZE, MBUF_NAME, slave->port[i]); + pool = rte_mempool_lookup(buf_name); + if (pool) + printf("Port %d mempool free object is %u(%u)\n", slave->port[i], + rte_mempool_avail_count(pool), + (unsigned int)NB_MBUF); + else + printf("Can't find mempool %s\n", buf_name); + + printf("Start port :%d\n", slave->port[i]); + ret = rte_eth_dev_start(slave->port[i]); + if (ret != 0) + break; + } + return ret; +} + +static int +reset_shared_structures(unsigned slaveid) +{ + int ret; + /* Only port are shared resource here */ + ret = reset_slave_all_ports(slaveid); + + return ret; +} + +/** + * Call this function to re-create resource that needed for slave process that + * exited in last instance + **/ +static int +init_slave_res(unsigned slaveid) +{ + struct lcore_resource_struct *slave = &lcore_resource[slaveid]; + enum l2fwd_cmd cmd; + + if (!slave->enabled) { + printf("Something wrong with lcore=%u enabled=%d\n",slaveid, + slave->enabled); + return -1; + } + + /* Initialize ring */ + if (create_ms_ring(slaveid) != 0) + rte_exit(EXIT_FAILURE, "failed to create ring for slave %u\n", + slaveid); + + /* drain un-read buffer if have */ + while (getcmd(slaveid, &cmd, 1) == 0); + while (getcmd(slaveid, &cmd, 0) == 0); + + return 0; +} + +static int +recreate_one_slave(unsigned slaveid) +{ + int ret = 0; + /* Re-initialize resource for stalled slave */ + if ((ret = init_slave_res(slaveid)) != 0) { + printf("Init slave=%u failed\n", slaveid); + return ret; + } + + if ((ret = flib_remote_launch(l2fwd_launch_one_lcore, NULL, slaveid)) + != 0) + printf("Launch slave %u failed\n", slaveid); + + return ret; +} + +/** + * remapping resource belong to slave_id to new lcore that gets from flib_assign_lcore_id(), + * used only floating process option applied. + * + * @param slaveid + * original lcore_id that apply for remapping + */ +static void +remapping_slave_resource(unsigned slaveid, unsigned map_id) +{ + + /* remapping lcore_resource */ + memcpy(&lcore_resource[map_id], &lcore_resource[slaveid], + sizeof(struct lcore_resource_struct)); + + /* remapping lcore_queue_conf */ + memcpy(&lcore_queue_conf[map_id], &lcore_queue_conf[slaveid], + sizeof(struct lcore_queue_conf)); +} + +static int +reset_pair(unsigned slaveid, unsigned pairid) +{ + int ret; + if ((ret = reset_shared_structures(slaveid)) != 0) + goto back; + + if((ret = reset_shared_structures(pairid)) != 0) + goto back; + + if (float_proc) { + unsigned map_id = mapping_id[slaveid]; + + if (map_id != INVALID_MAPPING_ID) { + printf("%u return mapping id %u\n", slaveid, map_id); + flib_free_lcore_id(map_id); + mapping_id[slaveid] = INVALID_MAPPING_ID; + } + + map_id = mapping_id[pairid]; + if (map_id != INVALID_MAPPING_ID) { + printf("%u return mapping id %u\n", pairid, map_id); + flib_free_lcore_id(map_id); + mapping_id[pairid] = INVALID_MAPPING_ID; + } + } + + if((ret = recreate_one_slave(slaveid)) != 0) + goto back; + + ret = recreate_one_slave(pairid); + +back: + return ret; +} + +static void +slave_exit_cb(unsigned slaveid, __attribute__((unused))int stat) +{ + struct lcore_resource_struct *slave = &lcore_resource[slaveid]; + + printf("Get slave %u leave info\n", slaveid); + if (!slave->enabled) { + printf("Lcore=%u not registered for it's exit\n", slaveid); + return; + } + rte_spinlock_lock(&res_lock); + + /* Change the state and wait master to start them */ + slave->flags = SLAVE_RECREATE_FLAG; + + rte_spinlock_unlock(&res_lock); +} + +static void +l2fwd_simple_forward(struct rte_mbuf *m, unsigned portid) +{ + struct ether_hdr *eth; + void *tmp; + unsigned dst_port; + int sent; + struct rte_eth_dev_tx_buffer *buffer; + + dst_port = l2fwd_dst_ports[portid]; + eth = rte_pktmbuf_mtod(m, struct ether_hdr *); + + /* 02:00:00:00:00:xx */ + tmp = ð->d_addr.addr_bytes[0]; + *((uint64_t *)tmp) = 0x000000000002 + ((uint64_t)dst_port << 40); + + /* src addr */ + ether_addr_copy(&l2fwd_ports_eth_addr[dst_port], ð->s_addr); + + buffer = tx_buffer[dst_port]; + sent = rte_eth_tx_buffer(dst_port, 0, buffer, m); + if (sent) + port_statistics[dst_port].tx += sent; +} + +/* main processing loop */ +static void +l2fwd_main_loop(void) +{ + struct rte_mbuf *pkts_burst[MAX_PKT_BURST]; + struct rte_mbuf *m; + int sent; + unsigned lcore_id; + uint64_t prev_tsc, diff_tsc, cur_tsc; + unsigned i, j, portid, nb_rx; + struct lcore_queue_conf *qconf; + const uint64_t drain_tsc = (rte_get_tsc_hz() + US_PER_S - 1) / US_PER_S * + BURST_TX_DRAIN_US; + struct rte_eth_dev_tx_buffer *buffer; + + prev_tsc = 0; + + lcore_id = rte_lcore_id(); + + qconf = &lcore_queue_conf[lcore_id]; + + if (qconf->n_rx_port == 0) { + RTE_LOG(INFO, L2FWD, "lcore %u has nothing to do\n", lcore_id); + return; + } + + RTE_LOG(INFO, L2FWD, "entering main loop on lcore %u\n", lcore_id); + + for (i = 0; i < qconf->n_rx_port; i++) { + portid = qconf->rx_port_list[i]; + RTE_LOG(INFO, L2FWD, " -- lcoreid=%u portid=%u\n", lcore_id, + portid); + } + + while (1) { + enum l2fwd_cmd cmd; + cur_tsc = rte_rdtsc(); + + if (unlikely(getcmd(lcore_id, &cmd, 0) == 0)) { + sendcmd(lcore_id, cmd, 0); + + /* If get stop command, stop forwarding and exit */ + if (cmd == CMD_STOP) { + return; + } + } + + /* + * TX burst queue drain + */ + diff_tsc = cur_tsc - prev_tsc; + if (unlikely(diff_tsc > drain_tsc)) { + + for (i = 0; i < qconf->n_rx_port; i++) { + + portid = l2fwd_dst_ports[qconf->rx_port_list[i]]; + buffer = tx_buffer[portid]; + + sent = rte_eth_tx_buffer_flush(portid, 0, buffer); + if (sent) + port_statistics[portid].tx += sent; + + } + + prev_tsc = cur_tsc; + } + + /* + * Read packet from RX queues + */ + for (i = 0; i < qconf->n_rx_port; i++) { + + portid = qconf->rx_port_list[i]; + nb_rx = rte_eth_rx_burst((uint8_t) portid, 0, + pkts_burst, MAX_PKT_BURST); + + port_statistics[portid].rx += nb_rx; + + for (j = 0; j < nb_rx; j++) { + m = pkts_burst[j]; + rte_prefetch0(rte_pktmbuf_mtod(m, void *)); + l2fwd_simple_forward(m, portid); + } + } + } +} + +static int +l2fwd_launch_one_lcore(__attribute__((unused)) void *dummy) +{ + unsigned lcore_id = rte_lcore_id(); + + if (float_proc) { + unsigned flcore_id; + + /* Change it to floating process, also change it's lcore_id */ + clear_cpu_affinity(); + RTE_PER_LCORE(_lcore_id) = 0; + /* Get a lcore_id */ + if (flib_assign_lcore_id() < 0 ) { + printf("flib_assign_lcore_id failed\n"); + return -1; + } + flcore_id = rte_lcore_id(); + /* Set mapping id, so master can return it after slave exited */ + mapping_id[lcore_id] = flcore_id; + printf("Org lcore_id = %u, cur lcore_id = %u\n", + lcore_id, flcore_id); + remapping_slave_resource(lcore_id, flcore_id); + } + + l2fwd_main_loop(); + + /* return lcore_id before return */ + if (float_proc) { + flib_free_lcore_id(rte_lcore_id()); + mapping_id[lcore_id] = INVALID_MAPPING_ID; + } + return 0; +} + +/* display usage */ +static void +l2fwd_usage(const char *prgname) +{ + printf("%s [EAL options] -- -p PORTMASK -s COREMASK [-q NQ] -f\n" + " -p PORTMASK: hexadecimal bitmask of ports to configure\n" + " -q NQ: number of queue (=ports) per lcore (default is 1)\n" + " -f use floating process which won't bind to any core to run\n" + " -T PERIOD: statistics will be refreshed each PERIOD seconds (0 to disable, 10 default, 86400 maximum)\n", + prgname); +} + +static int +l2fwd_parse_portmask(const char *portmask) +{ + char *end = NULL; + unsigned long pm; + + /* parse hexadecimal string */ + pm = strtoul(portmask, &end, 16); + if ((portmask[0] == '\0') || (end == NULL) || (*end != '\0')) + return -1; + + if (pm == 0) + return -1; + + return pm; +} + +static unsigned int +l2fwd_parse_nqueue(const char *q_arg) +{ + char *end = NULL; + unsigned long n; + + /* parse hexadecimal string */ + n = strtoul(q_arg, &end, 10); + if ((q_arg[0] == '\0') || (end == NULL) || (*end != '\0')) + return 0; + if (n == 0) + return 0; + if (n >= MAX_RX_QUEUE_PER_LCORE) + return 0; + + return n; +} + +static int +l2fwd_parse_timer_period(const char *q_arg) +{ + char *end = NULL; + int n; + + /* parse number string */ + n = strtol(q_arg, &end, 10); + if ((q_arg[0] == '\0') || (end == NULL) || (*end != '\0')) + return -1; + if (n >= MAX_TIMER_PERIOD) + return -1; + + return n; +} + +/* Parse the argument given in the command line of the application */ +static int +l2fwd_parse_args(int argc, char **argv) +{ + int opt, ret; + char **argvopt; + int option_index; + char *prgname = argv[0]; + static struct option lgopts[] = { + {NULL, 0, 0, 0} + }; + int has_pmask = 0; + + argvopt = argv; + + while ((opt = getopt_long(argc, argvopt, "p:q:T:f", + lgopts, &option_index)) != EOF) { + + switch (opt) { + /* portmask */ + case 'p': + l2fwd_enabled_port_mask = l2fwd_parse_portmask(optarg); + if (l2fwd_enabled_port_mask == 0) { + printf("invalid portmask\n"); + l2fwd_usage(prgname); + return -1; + } + has_pmask = 1; + break; + + /* nqueue */ + case 'q': + l2fwd_rx_queue_per_lcore = l2fwd_parse_nqueue(optarg); + if (l2fwd_rx_queue_per_lcore == 0) { + printf("invalid queue number\n"); + l2fwd_usage(prgname); + return -1; + } + break; + + /* timer period */ + case 'T': + timer_period = l2fwd_parse_timer_period(optarg) * 1000 * TIMER_MILLISECOND; + if (timer_period < 0) { + printf("invalid timer period\n"); + l2fwd_usage(prgname); + return -1; + } + break; + + /* use floating process */ + case 'f': + float_proc = 1; + break; + + /* long options */ + case 0: + l2fwd_usage(prgname); + return -1; + + default: + l2fwd_usage(prgname); + return -1; + } + } + + if (optind >= 0) + argv[optind-1] = prgname; + + if (!has_pmask) { + l2fwd_usage(prgname); + return -1; + } + ret = optind-1; + optind = 1; /* reset getopt lib */ + return ret; +} + +/* Check the link status of all ports in up to 9s, and print them finally */ +static void +check_all_ports_link_status(uint8_t port_num, uint32_t port_mask) +{ +#define CHECK_INTERVAL 100 /* 100ms */ +#define MAX_CHECK_TIME 90 /* 9s (90 * 100ms) in total */ + uint8_t portid, count, all_ports_up, print_flag = 0; + struct rte_eth_link link; + + printf("\nChecking link status"); + fflush(stdout); + for (count = 0; count <= MAX_CHECK_TIME; count++) { + all_ports_up = 1; + for (portid = 0; portid < port_num; portid++) { + if ((port_mask & (1 << portid)) == 0) + continue; + memset(&link, 0, sizeof(link)); + rte_eth_link_get_nowait(portid, &link); + /* print link status if flag set */ + if (print_flag == 1) { + if (link.link_status) + printf("Port %d Link Up - speed %u " + "Mbps - %s\n", (uint8_t)portid, + (unsigned)link.link_speed, + (link.link_duplex == ETH_LINK_FULL_DUPLEX) ? + ("full-duplex") : ("half-duplex\n")); + else + printf("Port %d Link Down\n", + (uint8_t)portid); + continue; + } + /* clear all_ports_up flag if any link down */ + if (link.link_status == ETH_LINK_DOWN) { + all_ports_up = 0; + break; + } + } + /* after finally printing all link status, get out */ + if (print_flag == 1) + break; + + if (all_ports_up == 0) { + printf("."); + fflush(stdout); + rte_delay_ms(CHECK_INTERVAL); + } + + /* set the print_flag if all ports up or timeout */ + if (all_ports_up == 1 || count == (MAX_CHECK_TIME - 1)) { + print_flag = 1; + printf("done\n"); + } + } +} + +int +main(int argc, char **argv) +{ + struct lcore_queue_conf *qconf; + struct rte_eth_dev_info dev_info; + int ret; + uint8_t nb_ports; + uint8_t nb_ports_available; + uint8_t portid, last_port; + unsigned rx_lcore_id; + unsigned nb_ports_in_mask = 0; + unsigned i; + int flags = 0; + uint64_t prev_tsc, diff_tsc, cur_tsc, timer_tsc; + + /* Save cpu_affinity first, restore it in case it's floating process option */ + if (get_cpu_affinity() != 0) + rte_exit(EXIT_FAILURE, "get_cpu_affinity error\n"); + + /* Also tries to set cpu affinity to detect whether it will fail in child process */ + if(clear_cpu_affinity() != 0) + rte_exit(EXIT_FAILURE, "clear_cpu_affinity error\n"); + + /* init EAL */ + ret = rte_eal_init(argc, argv); + if (ret < 0) + rte_exit(EXIT_FAILURE, "Invalid EAL arguments\n"); + argc -= ret; + argv += ret; + + /* parse application arguments (after the EAL ones) */ + ret = l2fwd_parse_args(argc, argv); + if (ret < 0) + rte_exit(EXIT_FAILURE, "Invalid L2FWD arguments\n"); + + /*flib init */ + if (flib_init() != 0) + rte_exit(EXIT_FAILURE, "flib init error"); + + /** + * Allocated structures that slave lcore would change. For those that slaves are + * read only, needn't use malloc to share and global or static variables is ok since + * slave inherit all the knowledge that master initialized. + **/ + if (l2fwd_malloc_shared_struct() != 0) + rte_exit(EXIT_FAILURE, "malloc mem failed\n"); + + /* Initialize lcore_resource structures */ + memset(lcore_resource, 0, sizeof(lcore_resource)); + for (i = 0; i < RTE_MAX_LCORE; i++) + lcore_resource[i].lcore_id = i; + + nb_ports = rte_eth_dev_count(); + if (nb_ports == 0) + rte_exit(EXIT_FAILURE, "No Ethernet ports - bye\n"); + + /* create the mbuf pool */ + for (portid = 0; portid < nb_ports; portid++) { + /* skip ports that are not enabled */ + if ((l2fwd_enabled_port_mask & (1 << portid)) == 0) + continue; + char buf_name[RTE_MEMPOOL_NAMESIZE]; + flags = MEMPOOL_F_SP_PUT | MEMPOOL_F_SC_GET; + snprintf(buf_name, RTE_MEMPOOL_NAMESIZE, MBUF_NAME, portid); + l2fwd_pktmbuf_pool[portid] = + rte_pktmbuf_pool_create(buf_name, NB_MBUF, 32, + 0, MBUF_DATA_SIZE, rte_socket_id()); + if (l2fwd_pktmbuf_pool[portid] == NULL) + rte_exit(EXIT_FAILURE, "Cannot create mbuf pool\n"); + + printf("Create mbuf %s\n", buf_name); + } + + /* reset l2fwd_dst_ports */ + for (portid = 0; portid < RTE_MAX_ETHPORTS; portid++) + l2fwd_dst_ports[portid] = 0; + last_port = 0; + + /* + * Each logical core is assigned a dedicated TX queue on each port. + */ + for (portid = 0; portid < nb_ports; portid++) { + /* skip ports that are not enabled */ + if ((l2fwd_enabled_port_mask & (1 << portid)) == 0) + continue; + + if (nb_ports_in_mask % 2) { + l2fwd_dst_ports[portid] = last_port; + l2fwd_dst_ports[last_port] = portid; + } + else + last_port = portid; + + nb_ports_in_mask++; + + rte_eth_dev_info_get(portid, &dev_info); + } + if (nb_ports_in_mask % 2) { + printf("Notice: odd number of ports in portmask.\n"); + l2fwd_dst_ports[last_port] = last_port; + } + + rx_lcore_id = 0; + qconf = NULL; + + /* Initialize the port/queue configuration of each logical core */ + for (portid = 0; portid < nb_ports; portid++) { + struct lcore_resource_struct *res; + /* skip ports that are not enabled */ + if ((l2fwd_enabled_port_mask & (1 << portid)) == 0) + continue; + + /* get the lcore_id for this port */ + /* skip master lcore */ + while (rte_lcore_is_enabled(rx_lcore_id) == 0 || + rte_get_master_lcore() == rx_lcore_id || + lcore_queue_conf[rx_lcore_id].n_rx_port == + l2fwd_rx_queue_per_lcore) { + + rx_lcore_id++; + if (rx_lcore_id >= RTE_MAX_LCORE) + rte_exit(EXIT_FAILURE, "Not enough cores\n"); + } + + if (qconf != &lcore_queue_conf[rx_lcore_id]) + /* Assigned a new logical core in the loop above. */ + qconf = &lcore_queue_conf[rx_lcore_id]; + + qconf->rx_port_list[qconf->n_rx_port] = portid; + qconf->n_rx_port++; + + /* Save the port resource info into lcore_resource strucutres */ + res = &lcore_resource[rx_lcore_id]; + res->enabled = 1; + res->port[res->port_num++] = portid; + + printf("Lcore %u: RX port %u\n", rx_lcore_id, (unsigned) portid); + } + + nb_ports_available = nb_ports; + + /* Initialise each port */ + for (portid = 0; portid < nb_ports; portid++) { + /* skip ports that are not enabled */ + if ((l2fwd_enabled_port_mask & (1 << portid)) == 0) { + printf("Skipping disabled port %u\n", (unsigned) portid); + nb_ports_available--; + continue; + } + /* init port */ + printf("Initializing port %u... ", (unsigned) portid); + fflush(stdout); + ret = rte_eth_dev_configure(portid, 1, 1, &port_conf); + if (ret < 0) + rte_exit(EXIT_FAILURE, "Cannot configure device: err=%d, port=%u\n", + ret, (unsigned) portid); + + rte_eth_macaddr_get(portid,&l2fwd_ports_eth_addr[portid]); + + /* init one RX queue */ + fflush(stdout); + ret = rte_eth_rx_queue_setup(portid, 0, nb_rxd, + rte_eth_dev_socket_id(portid), + NULL, + l2fwd_pktmbuf_pool[portid]); + if (ret < 0) + rte_exit(EXIT_FAILURE, "rte_eth_rx_queue_setup:err=%d, port=%u\n", + ret, (unsigned) portid); + + /* init one TX queue on each port */ + fflush(stdout); + ret = rte_eth_tx_queue_setup(portid, 0, nb_txd, + rte_eth_dev_socket_id(portid), + NULL); + if (ret < 0) + rte_exit(EXIT_FAILURE, "rte_eth_tx_queue_setup:err=%d, port=%u\n", + ret, (unsigned) portid); + + /* Initialize TX buffers */ + tx_buffer[portid] = rte_zmalloc_socket("tx_buffer", + RTE_ETH_TX_BUFFER_SIZE(MAX_PKT_BURST), 0, + rte_eth_dev_socket_id(portid)); + if (tx_buffer[portid] == NULL) + rte_exit(EXIT_FAILURE, "Cannot allocate buffer for tx on port %u\n", + (unsigned) portid); + + rte_eth_tx_buffer_init(tx_buffer[portid], MAX_PKT_BURST); + + ret = rte_eth_tx_buffer_set_err_callback(tx_buffer[portid], + rte_eth_tx_buffer_count_callback, + &port_statistics[portid].dropped); + if (ret < 0) + rte_exit(EXIT_FAILURE, "Cannot set error callback for " + "tx buffer on port %u\n", (unsigned) portid); + + /* Start device */ + ret = rte_eth_dev_start(portid); + if (ret < 0) + rte_exit(EXIT_FAILURE, "rte_eth_dev_start:err=%d, port=%u\n", + ret, (unsigned) portid); + + printf("done: \n"); + + rte_eth_promiscuous_enable(portid); + + printf("Port %u, MAC address: %02X:%02X:%02X:%02X:%02X:%02X\n\n", + (unsigned) portid, + l2fwd_ports_eth_addr[portid].addr_bytes[0], + l2fwd_ports_eth_addr[portid].addr_bytes[1], + l2fwd_ports_eth_addr[portid].addr_bytes[2], + l2fwd_ports_eth_addr[portid].addr_bytes[3], + l2fwd_ports_eth_addr[portid].addr_bytes[4], + l2fwd_ports_eth_addr[portid].addr_bytes[5]); + + /* initialize port stats */ + //memset(&port_statistics, 0, sizeof(port_statistics)); + } + + if (!nb_ports_available) { + rte_exit(EXIT_FAILURE, + "All available ports are disabled. Please set portmask.\n"); + } + + check_all_ports_link_status(nb_ports, l2fwd_enabled_port_mask); + + /* Record pair lcore */ + /** + * Since l2fwd example would create pair between different neighbour port, that's + * port 0 receive and forward to port 1, the same to port 1, these 2 ports will have + * dependency. If one port stopped working (killed, for example), the port need to + * be stopped/started again. During the time, another port need to wait until stop/start + * procedure completed. So, record the pair relationship for those lcores working + * on ports. + **/ + for (portid = 0; portid < nb_ports; portid++) { + uint32_t pair_port; + unsigned lcore = 0, pair_lcore = 0; + unsigned j, find_lcore, find_pair_lcore; + /* skip ports that are not enabled */ + if ((l2fwd_enabled_port_mask & (1 << portid)) == 0) + continue; + + /* Find pair ports' lcores */ + find_lcore = find_pair_lcore = 0; + pair_port = l2fwd_dst_ports[portid]; + for (i = 0; i < RTE_MAX_LCORE; i++) { + if (!rte_lcore_is_enabled(i)) + continue; + for (j = 0; j < lcore_queue_conf[i].n_rx_port;j++) { + if (lcore_queue_conf[i].rx_port_list[j] == portid) { + lcore = i; + find_lcore = 1; + break; + } + if (lcore_queue_conf[i].rx_port_list[j] == pair_port) { + pair_lcore = i; + find_pair_lcore = 1; + break; + } + } + if (find_lcore && find_pair_lcore) + break; + } + if (!find_lcore || !find_pair_lcore) + rte_exit(EXIT_FAILURE, "Not find port=%d pair\n", portid); + + printf("lcore %u and %u paired\n", lcore, pair_lcore); + lcore_resource[lcore].pair_id = pair_lcore; + lcore_resource[pair_lcore].pair_id = lcore; + } + + /* Create message buffer for all master and slave */ + message_pool = rte_mempool_create("ms_msg_pool", + NB_CORE_MSGBUF * RTE_MAX_LCORE, + sizeof(enum l2fwd_cmd), NB_CORE_MSGBUF / 2, + 0, + rte_pktmbuf_pool_init, NULL, + rte_pktmbuf_init, NULL, + rte_socket_id(), 0); + + if (message_pool == NULL) + rte_exit(EXIT_FAILURE, "Create msg mempool failed\n"); + + /* Create ring for each master and slave pair, also register cb when slave leaves */ + for (i = 0; i < RTE_MAX_LCORE; i++) { + /** + * Only create ring and register slave_exit cb in case that core involved into + * packet forwarding + **/ + if (lcore_resource[i].enabled) { + /* Create ring for master and slave communication */ + ret = create_ms_ring(i); + if (ret != 0) + rte_exit(EXIT_FAILURE, "Create ring for lcore=%u failed", + i); + + if (flib_register_slave_exit_notify(i, + slave_exit_cb) != 0) + rte_exit(EXIT_FAILURE, + "Register master_trace_slave_exit failed"); + } + } + + /* launch per-lcore init on every lcore except master */ + flib_mp_remote_launch(l2fwd_launch_one_lcore, NULL, SKIP_MASTER); + + /* print statistics 10 second */ + prev_tsc = cur_tsc = rte_rdtsc(); + timer_tsc = 0; + while (1) { + sleep(1); + cur_tsc = rte_rdtsc(); + diff_tsc = cur_tsc - prev_tsc; + /* if timer is enabled */ + if (timer_period > 0) { + + /* advance the timer */ + timer_tsc += diff_tsc; + + /* if timer has reached its timeout */ + if (unlikely(timer_tsc >= (uint64_t) timer_period)) { + + print_stats(); + /* reset the timer */ + timer_tsc = 0; + } + } + + prev_tsc = cur_tsc; + + /* Check any slave need restart or recreate */ + rte_spinlock_lock(&res_lock); + for (i = 0; i < RTE_MAX_LCORE; i++) { + struct lcore_resource_struct *res = &lcore_resource[i]; + struct lcore_resource_struct *pair = &lcore_resource[res->pair_id]; + + /* If find slave exited, try to reset pair */ + if (res->enabled && res->flags && pair->enabled) { + if (!pair->flags) { + master_sendcmd_with_ack(pair->lcore_id, CMD_STOP); + rte_spinlock_unlock(&res_lock); + sleep(1); + rte_spinlock_lock(&res_lock); + if (pair->flags) + continue; + } + if (reset_pair(res->lcore_id, pair->lcore_id) != 0) + rte_exit(EXIT_FAILURE, "failed to reset slave"); + res->flags = 0; + pair->flags = 0; + } + } + rte_spinlock_unlock(&res_lock); + } + +} diff --git a/src/seastar/dpdk/examples/multi_process/simple_mp/Makefile b/src/seastar/dpdk/examples/multi_process/simple_mp/Makefile new file mode 100644 index 00000000..31ec0c80 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/simple_mp/Makefile @@ -0,0 +1,50 @@ +# BSD LICENSE +# +# Copyright(c) 2010-2014 Intel Corporation. All rights reserved. +# All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of Intel Corporation nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +ifeq ($(RTE_SDK),) +$(error "Please define RTE_SDK environment variable") +endif + +# Default target, can be overriden by command line or environment +RTE_TARGET ?= x86_64-native-linuxapp-gcc + +include $(RTE_SDK)/mk/rte.vars.mk + +# binary name +APP = simple_mp + +# all source are stored in SRCS-y +SRCS-y := main.c mp_commands.c + +CFLAGS += -O3 +CFLAGS += $(WERROR_FLAGS) + +include $(RTE_SDK)/mk/rte.extapp.mk diff --git a/src/seastar/dpdk/examples/multi_process/simple_mp/main.c b/src/seastar/dpdk/examples/multi_process/simple_mp/main.c new file mode 100644 index 00000000..2843d94e --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/simple_mp/main.c @@ -0,0 +1,155 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +/* + * This sample application is a simple multi-process application which + * demostrates sharing of queues and memory pools between processes, and + * using those queues/pools for communication between the processes. + * + * Application is designed to run with two processes, a primary and a + * secondary, and each accepts commands on the commandline, the most + * important of which is "send", which just sends a string to the other + * process. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "mp_commands.h" + +#define RTE_LOGTYPE_APP RTE_LOGTYPE_USER1 + +static const char *_MSG_POOL = "MSG_POOL"; +static const char *_SEC_2_PRI = "SEC_2_PRI"; +static const char *_PRI_2_SEC = "PRI_2_SEC"; +const unsigned string_size = 64; + +struct rte_ring *send_ring, *recv_ring; +struct rte_mempool *message_pool; +volatile int quit = 0; + +static int +lcore_recv(__attribute__((unused)) void *arg) +{ + unsigned lcore_id = rte_lcore_id(); + + printf("Starting core %u\n", lcore_id); + while (!quit){ + void *msg; + if (rte_ring_dequeue(recv_ring, &msg) < 0){ + usleep(5); + continue; + } + printf("core %u: Received '%s'\n", lcore_id, (char *)msg); + rte_mempool_put(message_pool, msg); + } + + return 0; +} + +int +main(int argc, char **argv) +{ + const unsigned flags = 0; + const unsigned ring_size = 64; + const unsigned pool_size = 1024; + const unsigned pool_cache = 32; + const unsigned priv_data_sz = 0; + + int ret; + unsigned lcore_id; + + ret = rte_eal_init(argc, argv); + if (ret < 0) + rte_exit(EXIT_FAILURE, "Cannot init EAL\n"); + + if (rte_eal_process_type() == RTE_PROC_PRIMARY){ + send_ring = rte_ring_create(_PRI_2_SEC, ring_size, rte_socket_id(), flags); + recv_ring = rte_ring_create(_SEC_2_PRI, ring_size, rte_socket_id(), flags); + message_pool = rte_mempool_create(_MSG_POOL, pool_size, + string_size, pool_cache, priv_data_sz, + NULL, NULL, NULL, NULL, + rte_socket_id(), flags); + } else { + recv_ring = rte_ring_lookup(_PRI_2_SEC); + send_ring = rte_ring_lookup(_SEC_2_PRI); + message_pool = rte_mempool_lookup(_MSG_POOL); + } + if (send_ring == NULL) + rte_exit(EXIT_FAILURE, "Problem getting sending ring\n"); + if (recv_ring == NULL) + rte_exit(EXIT_FAILURE, "Problem getting receiving ring\n"); + if (message_pool == NULL) + rte_exit(EXIT_FAILURE, "Problem getting message pool\n"); + + RTE_LOG(INFO, APP, "Finished Process Init.\n"); + + /* call lcore_recv() on every slave lcore */ + RTE_LCORE_FOREACH_SLAVE(lcore_id) { + rte_eal_remote_launch(lcore_recv, NULL, lcore_id); + } + + /* call cmd prompt on master lcore */ + struct cmdline *cl = cmdline_stdin_new(simple_mp_ctx, "\nsimple_mp > "); + if (cl == NULL) + rte_exit(EXIT_FAILURE, "Cannot create cmdline instance\n"); + cmdline_interact(cl); + cmdline_stdin_exit(cl); + + rte_eal_mp_wait_lcore(); + return 0; +} diff --git a/src/seastar/dpdk/examples/multi_process/simple_mp/mp_commands.c b/src/seastar/dpdk/examples/multi_process/simple_mp/mp_commands.c new file mode 100644 index 00000000..8da244bb --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/simple_mp/mp_commands.c @@ -0,0 +1,166 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include "mp_commands.h" + +/**********************************************************/ + +struct cmd_send_result { + cmdline_fixed_string_t action; + cmdline_fixed_string_t message; +}; + +static void cmd_send_parsed(void *parsed_result, + __attribute__((unused)) struct cmdline *cl, + __attribute__((unused)) void *data) +{ + void *msg = NULL; + struct cmd_send_result *res = parsed_result; + + if (rte_mempool_get(message_pool, &msg) < 0) + rte_panic("Failed to get message buffer\n"); + snprintf((char *)msg, string_size, "%s", res->message); + if (rte_ring_enqueue(send_ring, msg) < 0) { + printf("Failed to send message - message discarded\n"); + rte_mempool_put(message_pool, msg); + } +} + +cmdline_parse_token_string_t cmd_send_action = + TOKEN_STRING_INITIALIZER(struct cmd_send_result, action, "send"); +cmdline_parse_token_string_t cmd_send_message = + TOKEN_STRING_INITIALIZER(struct cmd_send_result, message, NULL); + +cmdline_parse_inst_t cmd_send = { + .f = cmd_send_parsed, /* function to call */ + .data = NULL, /* 2nd arg of func */ + .help_str = "send a string to another process", + .tokens = { /* token list, NULL terminated */ + (void *)&cmd_send_action, + (void *)&cmd_send_message, + NULL, + }, +}; + +/**********************************************************/ + +struct cmd_quit_result { + cmdline_fixed_string_t quit; +}; + +static void cmd_quit_parsed(__attribute__((unused)) void *parsed_result, + struct cmdline *cl, + __attribute__((unused)) void *data) +{ + quit = 1; + cmdline_quit(cl); +} + +cmdline_parse_token_string_t cmd_quit_quit = + TOKEN_STRING_INITIALIZER(struct cmd_quit_result, quit, "quit"); + +cmdline_parse_inst_t cmd_quit = { + .f = cmd_quit_parsed, /* function to call */ + .data = NULL, /* 2nd arg of func */ + .help_str = "close the application", + .tokens = { /* token list, NULL terminated */ + (void *)&cmd_quit_quit, + NULL, + }, +}; + +/**********************************************************/ + +struct cmd_help_result { + cmdline_fixed_string_t help; +}; + +static void cmd_help_parsed(__attribute__((unused)) void *parsed_result, + struct cmdline *cl, + __attribute__((unused)) void *data) +{ + cmdline_printf(cl, "Simple demo example of multi-process in RTE\n\n" + "This is a readline-like interface that can be used to\n" + "send commands to the simple app. Commands supported are:\n\n" + "- send [string]\n" "- help\n" "- quit\n\n"); +} + +cmdline_parse_token_string_t cmd_help_help = + TOKEN_STRING_INITIALIZER(struct cmd_help_result, help, "help"); + +cmdline_parse_inst_t cmd_help = { + .f = cmd_help_parsed, /* function to call */ + .data = NULL, /* 2nd arg of func */ + .help_str = "show help", + .tokens = { /* token list, NULL terminated */ + (void *)&cmd_help_help, + NULL, + }, +}; + +/****** CONTEXT (list of instruction) */ +cmdline_parse_ctx_t simple_mp_ctx[] = { + (cmdline_parse_inst_t *)&cmd_send, + (cmdline_parse_inst_t *)&cmd_quit, + (cmdline_parse_inst_t *)&cmd_help, + NULL, +}; diff --git a/src/seastar/dpdk/examples/multi_process/simple_mp/mp_commands.h b/src/seastar/dpdk/examples/multi_process/simple_mp/mp_commands.h new file mode 100644 index 00000000..7e9a4ab2 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/simple_mp/mp_commands.h @@ -0,0 +1,44 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#ifndef _SIMPLE_MP_COMMANDS_H_ +#define _SIMPLE_MP_COMMANDS_H_ + +extern const unsigned string_size; +extern struct rte_ring *send_ring; +extern struct rte_mempool *message_pool; +extern volatile int quit; + +extern cmdline_parse_ctx_t simple_mp_ctx[]; + +#endif /* _SIMPLE_MP_COMMANDS_H_ */ diff --git a/src/seastar/dpdk/examples/multi_process/symmetric_mp/Makefile b/src/seastar/dpdk/examples/multi_process/symmetric_mp/Makefile new file mode 100644 index 00000000..c789f3c9 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/symmetric_mp/Makefile @@ -0,0 +1,50 @@ +# BSD LICENSE +# +# Copyright(c) 2010-2014 Intel Corporation. All rights reserved. +# All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of Intel Corporation nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +ifeq ($(RTE_SDK),) +$(error "Please define RTE_SDK environment variable") +endif + +# Default target, can be overriden by command line or environment +RTE_TARGET ?= x86_64-native-linuxapp-gcc + +include $(RTE_SDK)/mk/rte.vars.mk + +# binary name +APP = symmetric_mp + +# all source are stored in SRCS-y +SRCS-y := main.c + +CFLAGS += -O3 +CFLAGS += $(WERROR_FLAGS) + +include $(RTE_SDK)/mk/rte.extapp.mk diff --git a/src/seastar/dpdk/examples/multi_process/symmetric_mp/main.c b/src/seastar/dpdk/examples/multi_process/symmetric_mp/main.c new file mode 100644 index 00000000..0990d965 --- /dev/null +++ b/src/seastar/dpdk/examples/multi_process/symmetric_mp/main.c @@ -0,0 +1,471 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2014 Intel Corporation. All rights reserved. + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS + * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, + * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT + * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE + * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +/* + * Sample application demostrating how to do packet I/O in a multi-process + * environment. The same code can be run as a primary process and as a + * secondary process, just with a different proc-id parameter in each case + * (apart from the EAL flag to indicate a secondary process). + * + * Each process will read from the same ports, given by the port-mask + * parameter, which should be the same in each case, just using a different + * queue per port as determined by the proc-id parameter. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#define RTE_LOGTYPE_APP RTE_LOGTYPE_USER1 + +#define NB_MBUFS 64*1024 /* use 64k mbufs */ +#define MBUF_CACHE_SIZE 256 +#define PKT_BURST 32 +#define RX_RING_SIZE 128 +#define TX_RING_SIZE 512 + +#define PARAM_PROC_ID "proc-id" +#define PARAM_NUM_PROCS "num-procs" + +/* for each lcore, record the elements of the ports array to use */ +struct lcore_ports{ + unsigned start_port; + unsigned num_ports; +}; + +/* structure to record the rx and tx packets. Put two per cache line as ports + * used in pairs */ +struct port_stats{ + unsigned rx; + unsigned tx; + unsigned drop; +} __attribute__((aligned(RTE_CACHE_LINE_SIZE / 2))); + +static int proc_id = -1; +static unsigned num_procs = 0; + +static uint8_t ports[RTE_MAX_ETHPORTS]; +static unsigned num_ports = 0; + +static struct lcore_ports lcore_ports[RTE_MAX_LCORE]; +static struct port_stats pstats[RTE_MAX_ETHPORTS]; + +/* prints the usage statement and quits with an error message */ +static void +smp_usage(const char *prgname, const char *errmsg) +{ + printf("\nError: %s\n",errmsg); + printf("\n%s [EAL options] -- -p " + "--"PARAM_NUM_PROCS" " + " --"PARAM_PROC_ID" \n" + "-p : a hex bitmask indicating what ports are to be used\n" + "--num-procs: the number of processes which will be used\n" + "--proc-id : the id of the current process (id < num-procs)\n" + "\n", + prgname); + exit(1); +} + + +/* signal handler configured for SIGTERM and SIGINT to print stats on exit */ +static void +print_stats(int signum) +{ + unsigned i; + printf("\nExiting on signal %d\n\n", signum); + for (i = 0; i < num_ports; i++){ + const uint8_t p_num = ports[i]; + printf("Port %u: RX - %u, TX - %u, Drop - %u\n", (unsigned)p_num, + pstats[p_num].rx, pstats[p_num].tx, pstats[p_num].drop); + } + exit(0); +} + +/* Parse the argument given in the command line of the application */ +static int +smp_parse_args(int argc, char **argv) +{ + int opt, ret; + char **argvopt; + int option_index; + unsigned i, port_mask = 0; + char *prgname = argv[0]; + static struct option lgopts[] = { + {PARAM_NUM_PROCS, 1, 0, 0}, + {PARAM_PROC_ID, 1, 0, 0}, + {NULL, 0, 0, 0} + }; + + argvopt = argv; + + while ((opt = getopt_long(argc, argvopt, "p:", \ + lgopts, &option_index)) != EOF) { + + switch (opt) { + case 'p': + port_mask = strtoull(optarg, NULL, 16); + break; + /* long options */ + case 0: + if (strncmp(lgopts[option_index].name, PARAM_NUM_PROCS, 8) == 0) + num_procs = atoi(optarg); + else if (strncmp(lgopts[option_index].name, PARAM_PROC_ID, 7) == 0) + proc_id = atoi(optarg); + break; + + default: + smp_usage(prgname, "Cannot parse all command-line arguments\n"); + } + } + + if (optind >= 0) + argv[optind-1] = prgname; + + if (proc_id < 0) + smp_usage(prgname, "Invalid or missing proc-id parameter\n"); + if (rte_eal_process_type() == RTE_PROC_PRIMARY && num_procs == 0) + smp_usage(prgname, "Invalid or missing num-procs parameter\n"); + if (port_mask == 0) + smp_usage(prgname, "Invalid or missing port mask\n"); + + /* get the port numbers from the port mask */ + for(i = 0; i < rte_eth_dev_count(); i++) + if(port_mask & (1 << i)) + ports[num_ports++] = (uint8_t)i; + + ret = optind-1; + optind = 1; /* reset getopt lib */ + + return ret; +} + +/* + * Initialises a given port using global settings and with the rx buffers + * coming from the mbuf_pool passed as parameter + */ +static inline int +smp_port_init(uint8_t port, struct rte_mempool *mbuf_pool, uint16_t num_queues) +{ + struct rte_eth_conf port_conf = { + .rxmode = { + .mq_mode = ETH_MQ_RX_RSS, + .split_hdr_size = 0, + .header_split = 0, /**< Header Split disabled */ + .hw_ip_checksum = 1, /**< IP checksum offload enabled */ + .hw_vlan_filter = 0, /**< VLAN filtering disabled */ + .jumbo_frame = 0, /**< Jumbo Frame Support disabled */ + .hw_strip_crc = 1, /**< CRC stripped by hardware */ + }, + .rx_adv_conf = { + .rss_conf = { + .rss_key = NULL, + .rss_hf = ETH_RSS_IP, + }, + }, + .txmode = { + .mq_mode = ETH_MQ_TX_NONE, + } + }; + const uint16_t rx_rings = num_queues, tx_rings = num_queues; + struct rte_eth_dev_info info; + int retval; + uint16_t q; + + if (rte_eal_process_type() == RTE_PROC_SECONDARY) + return 0; + + if (port >= rte_eth_dev_count()) + return -1; + + printf("# Initialising port %u... ", (unsigned)port); + fflush(stdout); + + rte_eth_dev_info_get(port, &info); + info.default_rxconf.rx_drop_en = 1; + + retval = rte_eth_dev_configure(port, rx_rings, tx_rings, &port_conf); + if (retval < 0) + return retval; + + for (q = 0; q < rx_rings; q ++) { + retval = rte_eth_rx_queue_setup(port, q, RX_RING_SIZE, + rte_eth_dev_socket_id(port), + &info.default_rxconf, + mbuf_pool); + if (retval < 0) + return retval; + } + + for (q = 0; q < tx_rings; q ++) { + retval = rte_eth_tx_queue_setup(port, q, TX_RING_SIZE, + rte_eth_dev_socket_id(port), + NULL); + if (retval < 0) + return retval; + } + + rte_eth_promiscuous_enable(port); + + retval = rte_eth_dev_start(port); + if (retval < 0) + return retval; + + return 0; +} + +/* Goes through each of the lcores and calculates what ports should + * be used by that core. Fills in the global lcore_ports[] array. + */ +static void +assign_ports_to_cores(void) +{ + + const unsigned lcores = rte_eal_get_configuration()->lcore_count; + const unsigned port_pairs = num_ports / 2; + const unsigned pairs_per_lcore = port_pairs / lcores; + unsigned extra_pairs = port_pairs % lcores; + unsigned ports_assigned = 0; + unsigned i; + + RTE_LCORE_FOREACH(i) { + lcore_ports[i].start_port = ports_assigned; + lcore_ports[i].num_ports = pairs_per_lcore * 2; + if (extra_pairs > 0) { + lcore_ports[i].num_ports += 2; + extra_pairs--; + } + ports_assigned += lcore_ports[i].num_ports; + } +} + +/* Main function used by the processing threads. + * Prints out some configuration details for the thread and then begins + * performing packet RX and TX. + */ +static int +lcore_main(void *arg __rte_unused) +{ + const unsigned id = rte_lcore_id(); + const unsigned start_port = lcore_ports[id].start_port; + const unsigned end_port = start_port + lcore_ports[id].num_ports; + const uint16_t q_id = (uint16_t)proc_id; + unsigned p, i; + char msgbuf[256]; + int msgbufpos = 0; + + if (start_port == end_port){ + printf("Lcore %u has nothing to do\n", id); + return 0; + } + + /* build up message in msgbuf before printing to decrease likelihood + * of multi-core message interleaving. + */ + msgbufpos += snprintf(msgbuf, sizeof(msgbuf) - msgbufpos, + "Lcore %u using ports ", id); + for (p = start_port; p < end_port; p++){ + msgbufpos += snprintf(msgbuf + msgbufpos, sizeof(msgbuf) - msgbufpos, + "%u ", (unsigned)ports[p]); + } + printf("%s\n", msgbuf); + printf("lcore %u using queue %u of each port\n", id, (unsigned)q_id); + + /* handle packet I/O from the ports, reading and writing to the + * queue number corresponding to our process number (not lcore id) + */ + + for (;;) { + struct rte_mbuf *buf[PKT_BURST]; + + for (p = start_port; p < end_port; p++) { + const uint8_t src = ports[p]; + const uint8_t dst = ports[p ^ 1]; /* 0 <-> 1, 2 <-> 3 etc */ + const uint16_t rx_c = rte_eth_rx_burst(src, q_id, buf, PKT_BURST); + if (rx_c == 0) + continue; + pstats[src].rx += rx_c; + + const uint16_t tx_c = rte_eth_tx_burst(dst, q_id, buf, rx_c); + pstats[dst].tx += tx_c; + if (tx_c != rx_c) { + pstats[dst].drop += (rx_c - tx_c); + for (i = tx_c; i < rx_c; i++) + rte_pktmbuf_free(buf[i]); + } + } + } +} + +/* Check the link status of all ports in up to 9s, and print them finally */ +static void +check_all_ports_link_status(uint8_t port_num, uint32_t port_mask) +{ +#define CHECK_INTERVAL 100 /* 100ms */ +#define MAX_CHECK_TIME 90 /* 9s (90 * 100ms) in total */ + uint8_t portid, count, all_ports_up, print_flag = 0; + struct rte_eth_link link; + + printf("\nChecking link status"); + fflush(stdout); + for (count = 0; count <= MAX_CHECK_TIME; count++) { + all_ports_up = 1; + for (portid = 0; portid < port_num; portid++) { + if ((port_mask & (1 << portid)) == 0) + continue; + memset(&link, 0, sizeof(link)); + rte_eth_link_get_nowait(portid, &link); + /* print link status if flag set */ + if (print_flag == 1) { + if (link.link_status) + printf("Port %d Link Up - speed %u " + "Mbps - %s\n", (uint8_t)portid, + (unsigned)link.link_speed, + (link.link_duplex == ETH_LINK_FULL_DUPLEX) ? + ("full-duplex") : ("half-duplex\n")); + else + printf("Port %d Link Down\n", + (uint8_t)portid); + continue; + } + /* clear all_ports_up flag if any link down */ + if (link.link_status == ETH_LINK_DOWN) { + all_ports_up = 0; + break; + } + } + /* after finally printing all link status, get out */ + if (print_flag == 1) + break; + + if (all_ports_up == 0) { + printf("."); + fflush(stdout); + rte_delay_ms(CHECK_INTERVAL); + } + + /* set the print_flag if all ports up or timeout */ + if (all_ports_up == 1 || count == (MAX_CHECK_TIME - 1)) { + print_flag = 1; + printf("done\n"); + } + } +} + +/* Main function. + * Performs initialisation and then calls the lcore_main on each core + * to do the packet-processing work. + */ +int +main(int argc, char **argv) +{ + static const char *_SMP_MBUF_POOL = "SMP_MBUF_POOL"; + int ret; + unsigned i; + enum rte_proc_type_t proc_type; + struct rte_mempool *mp; + + /* set up signal handlers to print stats on exit */ + signal(SIGINT, print_stats); + signal(SIGTERM, print_stats); + + /* initialise the EAL for all */ + ret = rte_eal_init(argc, argv); + if (ret < 0) + rte_exit(EXIT_FAILURE, "Cannot init EAL\n"); + argc -= ret; + argv += ret; + + /* determine the NIC devices available */ + if (rte_eth_dev_count() == 0) + rte_exit(EXIT_FAILURE, "No Ethernet ports - bye\n"); + + /* parse application arguments (those after the EAL ones) */ + smp_parse_args(argc, argv); + + proc_type = rte_eal_process_type(); + mp = (proc_type == RTE_PROC_SECONDARY) ? + rte_mempool_lookup(_SMP_MBUF_POOL) : + rte_pktmbuf_pool_create(_SMP_MBUF_POOL, NB_MBUFS, + MBUF_CACHE_SIZE, 0, RTE_MBUF_DEFAULT_BUF_SIZE, + rte_socket_id()); + if (mp == NULL) + rte_exit(EXIT_FAILURE, "Cannot get memory pool for buffers\n"); + + if (num_ports & 1) + rte_exit(EXIT_FAILURE, "Application must use an even number of ports\n"); + for(i = 0; i < num_ports; i++){ + if(proc_type == RTE_PROC_PRIMARY) + if (smp_port_init(ports[i], mp, (uint16_t)num_procs) < 0) + rte_exit(EXIT_FAILURE, "Error initialising ports\n"); + } + + if (proc_type == RTE_PROC_PRIMARY) + check_all_ports_link_status((uint8_t)num_ports, (~0x0)); + + assign_ports_to_cores(); + + RTE_LOG(INFO, APP, "Finished Process Init.\n"); + + rte_eal_mp_remote_launch(lcore_main, NULL, CALL_MASTER); + + return 0; +} -- cgit v1.2.3