DPDK flow_classify 源码阅读
Posted zcplayground
tags:
篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了DPDK flow_classify 源码阅读相关的知识，希望对你有一定的参考价值。
todo
/* SPDX-License-Identifier: BSD-3-Clause
* Copyright(c) 2017 Intel Corporation
*/
#include <stdint.h>
#include <inttypes.h>
#include <getopt.h>
#include <rte_eal.h>
#include <rte_ethdev.h>
#include <rte_cycles.h>
#include <rte_lcore.h>
#include <rte_mbuf.h>
#include <rte_flow.h>
#include <rte_flow_classify.h>
#include <rte_table_acl.h>
#define RX_RING_SIZE 1024
#define TX_RING_SIZE 1024
#define NUM_MBUFS 8191
#define MBUF_CACHE_SIZE 250
#define BURST_SIZE 32
#define MAX_NUM_CLASSIFY 30
#define FLOW_CLASSIFY_MAX_RULE_NUM 91
#define FLOW_CLASSIFY_MAX_PRIORITY 8
#define FLOW_CLASSIFIER_NAME_SIZE 64
#define COMMENT_LEAD_CHAR ('#')
#define OPTION_RULE_IPV4 "rule_ipv4"
#define RTE_LOGTYPE_FLOW_CLASSIFY RTE_LOGTYPE_USER3
#define flow_classify_log(format, ...) RTE_LOG(ERR, FLOW_CLASSIFY, format, ##__VA_ARGS__)
#define uint32_t_to_char(ip, a, b, c, d) do { *a = (unsigned char)(ip >> 24 & 0xff); *b = (unsigned char)(ip >> 16 & 0xff); *c = (unsigned char)(ip >> 8 & 0xff); *d = (unsigned char)(ip & 0xff); } while (0)
enum {
CB_FLD_SRC_ADDR,
CB_FLD_DST_ADDR,
CB_FLD_SRC_PORT,
CB_FLD_SRC_PORT_DLM,
CB_FLD_SRC_PORT_MASK,
CB_FLD_DST_PORT,
CB_FLD_DST_PORT_DLM,
CB_FLD_DST_PORT_MASK,
CB_FLD_PROTO,
CB_FLD_PRIORITY,
CB_FLD_NUM,
};
static struct{
const char *rule_ipv4_name;
} parm_config;
const char cb_port_delim[] = ":";
static const struct rte_eth_conf port_conf_default = {
.rxmode = {
.max_rx_pkt_len = ETHER_MAX_LEN,
.ignore_offload_bitfield = 1,
},
};
struct flow_classifier {
struct rte_flow_classifier *cls;
};
struct flow_classifier_acl {
struct flow_classifier cls;
} __rte_cache_aligned;
/* ACL field definitions for IPv4 5 tuple rule */
enum {
PROTO_FIELD_IPV4, // 0
SRC_FIELD_IPV4, // 1
DST_FIELD_IPV4, // 2
SRCP_FIELD_IPV4, // 3
DSTP_FIELD_IPV4, // 4
NUM_FIELDS_IPV4 // 5
};
enum {
PROTO_INPUT_IPV4,
SRC_INPUT_IPV4,
DST_INPUT_IPV4,
SRCP_DESTP_INPUT_IPV4
};
/* 数据结构 rte_acl_field_def：ACL访问控制表的字段的定义
ACL规则中的每个字段都有一个关联定义。有五个，分别是：类型，大小，字段的索引（指示哪一个字段），输入索引（0-N）和（距离字段开始处的）偏移量。
*/
static struct rte_acl_field_def ipv4_defs[NUM_FIELDS_IPV4] = {
/* first input field - always one byte long. */
{
.type = RTE_ACL_FIELD_TYPE_BITMASK,
.size = sizeof(uint8_t),
.field_index = PROTO_FIELD_IPV4,
.input_index = PROTO_INPUT_IPV4,
.offset = sizeof(struct ether_hdr) +
offsetof(struct ipv4_hdr, next_proto_id),
},
/* next input field (IPv4 source address) - 4 consecutive bytes. */
{
/* rte_flow uses a bit mask for IPv4 addresses */
.type = RTE_ACL_FIELD_TYPE_BITMASK,
.size = sizeof(uint32_t),
.field_index = SRC_FIELD_IPV4,
.input_index = SRC_INPUT_IPV4,
.offset = sizeof(struct ether_hdr) +
offsetof(struct ipv4_hdr, src_addr),
},
/* next input field (IPv4 destination address) - 4 consecutive bytes. */
{
/* rte_flow uses a bit mask for IPv4 addresses */
.type = RTE_ACL_FIELD_TYPE_BITMASK,
.size = sizeof(uint32_t),
.field_index = DST_FIELD_IPV4,
.input_index = DST_INPUT_IPV4,
.offset = sizeof(struct ether_hdr) +
offsetof(struct ipv4_hdr, dst_addr),
},
/*
* Next 2 fields (src & dst ports) form 4 consecutive bytes.
* They share the same input index.
*/
{
/* rte_flow uses a bit mask for protocol ports */
.type = RTE_ACL_FIELD_TYPE_BITMASK,
.size = sizeof(uint16_t),
.field_index = SRCP_FIELD_IPV4,
.input_index = SRCP_DESTP_INPUT_IPV4,
.offset = sizeof(struct ether_hdr) +
sizeof(struct ipv4_hdr) +
offsetof(struct tcp_hdr, src_port),
},
{
/* rte_flow uses a bit mask for protocol ports */
.type = RTE_ACL_FIELD_TYPE_BITMASK,
.size = sizeof(uint16_t),
.field_index = DSTP_FIELD_IPV4,
.input_index = SRCP_DESTP_INPUT_IPV4,
.offset = sizeof(struct ether_hdr) +
sizeof(struct ipv4_hdr) +
offsetof(struct tcp_hdr, dst_port),
},
};
/* flow classify data */
static int num_classify_rules; // rules数组的下标
static struct rte_flow_classify_rule *rules[MAX_NUM_CLASSIFY]; // rules 数组
static struct rte_flow_classify_ipv4_5tuple_stats ntuple_stats;
static struct rte_flow_classify_stats classify_stats = {
.stats = (void **)&ntuple_stats
};
/* parameters for rte_flow_classify_validate and
* rte_flow_classify_table_entry_add functions
*/
static struct rte_flow_item eth_item = { RTE_FLOW_ITEM_TYPE_ETH,
0, 0, 0 };
static struct rte_flow_item end_item = { RTE_FLOW_ITEM_TYPE_END,
0, 0, 0 };
/* sample actions:
* "actions count / end"
*/
struct rte_flow_query_count count = {
.reset = 1,
.hits_set = 1,
.bytes_set = 1,
.hits = 0,
.bytes = 0,
};
static struct rte_flow_action count_action = { RTE_FLOW_ACTION_TYPE_COUNT,
&count};
static struct rte_flow_action end_action = { RTE_FLOW_ACTION_TYPE_END, 0};
static struct rte_flow_action actions[2];
/* sample attributes */
static struct rte_flow_attr attr;
/* flow_classify.c: * Based on DPDK skeleton forwarding example. */
/*
* Initializes a given port using global settings and with the RX buffers
* coming from the mbuf_pool passed as a parameter.
*/
static inline int
port_init(uint8_t port, struct rte_mempool *mbuf_pool)
{
struct rte_eth_conf port_conf = port_conf_default;
struct ether_addr addr;
const uint16_t rx_rings = 1, tx_rings = 1;
int retval;
uint16_t q;
struct rte_eth_dev_info dev_info;
struct rte_eth_txconf txconf;
if (!rte_eth_dev_is_valid_port(port))
return -1;
rte_eth_dev_info_get(port, &dev_info);
if (dev_info.tx_offload_capa & DEV_TX_OFFLOAD_MBUF_FAST_FREE)
port_conf.txmode.offloads |=
DEV_TX_OFFLOAD_MBUF_FAST_FREE;
/* Configure the Ethernet device. */
retval = rte_eth_dev_configure(port, rx_rings, tx_rings, &port_conf);
if (retval != 0)
return retval;
/* Allocate and set up 1 RX queue per Ethernet port. */
for (q = 0; q < rx_rings; q++) {
retval = rte_eth_rx_queue_setup(port, q, RX_RING_SIZE,
rte_eth_dev_socket_id(port), NULL, mbuf_pool);
if (retval < 0)
return retval;
}
txconf = dev_info.default_txconf;
txconf.txq_flags = ETH_TXQ_FLAGS_IGNORE;
txconf.offloads = port_conf.txmode.offloads;
/* Allocate and set up 1 TX queue per Ethernet port. */
for (q = 0; q < tx_rings; q++) {
retval = rte_eth_tx_queue_setup(port, q, TX_RING_SIZE,
rte_eth_dev_socket_id(port), &txconf);
if (retval < 0)
return retval;
}
/* Start the Ethernet port. */
retval = rte_eth_dev_start(port);
if (retval < 0)
return retval;
/* Display the port MAC address. */
rte_eth_macaddr_get(port, &addr);
printf("Port %u MAC: %02" PRIx8 " %02" PRIx8 " %02" PRIx8
" %02" PRIx8 " %02" PRIx8 " %02" PRIx8 "
",
port,
addr.addr_bytes[0], addr.addr_bytes[1],
addr.addr_bytes[2], addr.addr_bytes[3],
addr.addr_bytes[4], addr.addr_bytes[5]);
/* Enable RX in promiscuous mode for the Ethernet device. */
rte_eth_promiscuous_enable(port);
return 0;
}
/*
* The lcore main. This is the main thread that does the work, reading from
* an input port classifying the packets and writing to an output port.
*/
static __attribute__((noreturn)) void
lcore_main(struct flow_classifier *cls_app)
{
uint16_t port;
int ret;
int i = 0;
ret = rte_flow_classify_table_entry_delete(cls_app->cls,
rules[7]);
if (ret)
printf("table_entry_delete failed [7] %d
", ret);
else
printf("table_entry_delete succeeded [7]
");
/*
* Check that the port is on the same NUMA node as the polling thread
* for best performance.
*/
RTE_ETH_FOREACH_DEV(port)
if (rte_eth_dev_socket_id(port) > 0 &&
rte_eth_dev_socket_id(port) != (int)rte_socket_id()) {
printf("
");
printf("WARNING: port %u is on remote NUMA node
",
port);
printf("to polling thread.
");
printf("Performance will not be optimal.
");
}
printf("
Core %u forwarding packets. ", rte_lcore_id());
printf("[Ctrl+C to quit]
");
/* Run until the application is quit or killed. */
for (;;) {
/*
* Receive packets on a port, **classify them** and forward them
* on the paired port.
* The mapping is 0 -> 1, 1 -> 0, 2 -> 3, 3 -> 2, etc.
*/
RTE_ETH_FOREACH_DEV(port) {
/* Get burst of RX packets, from first port of pair. */
struct rte_mbuf *bufs[BURST_SIZE];
const uint16_t nb_rx = rte_eth_rx_burst(port, 0,
bufs, BURST_SIZE); // 收包
if (unlikely(nb_rx == 0))
continue;
for (i = 0; i < MAX_NUM_CLASSIFY; i++) {
if (rules[i]) { // 在classifier中查询特定的规则
// 收包之后，将感兴趣的流放到
ret = rte_flow_classifier_query(
cls_app->cls, // 流分类器句柄
// 要处理的数据包的 mbuf
// 数据包数量
// 规则
bufs, nb_rx, rules[i],
&classify_stats);
if (ret) // 返回 0 代表分类成功
printf(
"rule [%d] query failed ret [%d]
",
i, ret);
else {
printf(
"rule[%d] count=%"PRIu64"
",
i, ntuple_stats.counter1);
printf("proto = %d
",
ntuple_stats.ipv4_5tuple.proto);
}
}
}
/* Send burst of TX packets, to second port of pair. */
const uint16_t nb_tx = rte_eth_tx_burst(port ^ 1, 0,
bufs, nb_rx);
/* Free any unsent packets. */
if (unlikely(nb_tx < nb_rx)) {
uint16_t buf;
for (buf = nb_tx; buf < nb_rx; buf++)
rte_pktmbuf_free(bufs[buf]);
}
}
}
}
/*
* Parse IPv4 5 tuple rules file, ipv4_rules_file.txt.
* Expected format:
* <src_ipv4_addr>'/'<masklen> <space> * <dst_ipv4_addr>'/'<masklen> <space> * <src_port> <space> ":" <src_port_mask> <space> * <dst_port> <space> ":" <dst_port_mask> <space> * <proto>'/'<proto_mask> <space> * <priority>
*/
static int
get_cb_field(char **in, uint32_t *fd, int base, unsigned long lim,
char dlm)
{
unsigned long val;
char *end;
errno = 0;
val = strtoul(*in, &end, base);
if (errno != 0 || end[0] != dlm || val > lim)
return -EINVAL;
*fd = (uint32_t)val;
*in = end + 1;
return 0;
}
static int
parse_ipv4_net(char *in, uint32_t *addr, uint32_t *mask_len)
{
uint32_t a, b, c, d, m;
if (get_cb_field(&in, &a, 0, UINT8_MAX, '.'))
return -EINVAL;
if (get_cb_field(&in, &b, 0, UINT8_MAX, '.'))
return -EINVAL;
if (get_cb_field(&in, &c, 0, UINT8_MAX, '.'))
return -EINVAL;
if (get_cb_field(&in, &d, 0, UINT8_MAX, '/'))
return -EINVAL;
if (get_cb_field(&in, &m, 0, sizeof(uint32_t) * CHAR_BIT, 0))
return -EINVAL;
addr[0] = IPv4(a, b, c, d);
mask_len[0] = m;
return 0;
}
static int
parse_ipv4_5tuple_rule(char *str, struct rte_eth_ntuple_filter *ntuple_filter)
{
int i, ret;
char *s, *sp, *in[CB_FLD_NUM];
static const char *dlm = "
";
int dim = CB_FLD_NUM;
uint32_t temp;
s = str;
for (i = 0; i != dim; i++, s = NULL) {
in[i] = strtok_r(s, dlm, &sp);
if (in[i] == NULL)
return -EINVAL;
}
ret = parse_ipv4_net(in[CB_FLD_SRC_ADDR],
&ntuple_filter->src_ip,
&ntuple_filter->src_ip_mask);
if (ret != 0) {
flow_classify_log("failed to read source address/mask: %s
",
in[CB_FLD_SRC_ADDR]);
return ret;
}
ret = parse_ipv4_net(in[CB_FLD_DST_ADDR],
&ntuple_filter->dst_ip,
&ntuple_filter->dst_ip_mask);
if (ret != 0) {
flow_classify_log("failed to read source address/mask: %s
",
in[CB_FLD_DST_ADDR]);
return ret;
}
if (get_cb_field(&in[CB_FLD_SRC_PORT], &temp, 0, UINT16_MAX, 0))
return -EINVAL;
ntuple_filter->src_port = (uint16_t)temp;
if (strncmp(in[CB_FLD_SRC_PORT_DLM], cb_port_delim,
sizeof(cb_port_delim)) != 0)
return -EINVAL;
if (get_cb_field(&in[CB_FLD_SRC_PORT_MASK], &temp, 0, UINT16_MAX, 0))
return -EINVAL;
ntuple_filter->src_port_mask = (uint16_t)temp;
if (get_cb_field(&in[CB_FLD_DST_PORT], &temp, 0, UINT16_MAX, 0))
return -EINVAL;
ntuple_filter->dst_port = (uint16_t)temp;
if (strncmp(in[CB_FLD_DST_PORT_DLM], cb_port_delim,
sizeof(cb_port_delim)) != 0)
return -EINVAL;
if (get_cb_field(&in[CB_FLD_DST_PORT_MASK], &temp, 0, UINT16_MAX, 0))
return -EINVAL;
ntuple_filter->dst_port_mask = (uint16_t)temp;
if (get_cb_field(&in[CB_FLD_PROTO], &temp, 0, UINT8_MAX, '/'))
return -EINVAL;
ntuple_filter->proto = (uint8_t)temp;
if (get_cb_field(&in[CB_FLD_PROTO], &temp, 0, UINT8_MAX, 0))
return -EINVAL;
ntuple_filter->proto_mask = (uint8_t)temp;
if (get_cb_field(&in[CB_FLD_PRIORITY], &temp, 0, UINT16_MAX, 0))
return -EINVAL;
ntuple_filter->priority = (uint16_t)temp;
if (ntuple_filter->priority > FLOW_CLASSIFY_MAX_PRIORITY)
ret = -EINVAL;
return ret;
}
/* Bypass comment and empty lines */
static inline int
is_bypass_line(char *buff)
{
int i = 0;
/* comment line */
if (buff[0] == COMMENT_LEAD_CHAR)
return 1;
/* empty line */
while (buff[i] != ' ') {
if (!isspace(buff[i]))
return 0;
i++;
}
return 1;
}
static uint32_t
convert_depth_to_bitmask(uint32_t depth_val)
{
uint32_t bitmask = 0;
int i, j;
for (i = depth_val, j = 0; i > 0; i--, j++)
bitmask |= (1 << (31 - j));
return bitmask;
}
static int
add_classify_rule(struct rte_eth_ntuple_filter *ntuple_filter,
struct flow_classifier *cls_app) // 一层封装
{
int ret = -1;
int key_found;
struct rte_flow_error error;
struct rte_flow_item_ipv4 ipv4_spec;
struct rte_flow_item_ipv4 ipv4_mask;
struct rte_flow_item ipv4_udp_item;
struct rte_flow_item ipv4_tcp_item;
struct rte_flow_item ipv4_sctp_item;
struct rte_flow_item_udp udp_spec;
struct rte_flow_item_udp udp_mask;
struct rte_flow_item udp_item;
struct rte_flow_item_tcp tcp_spec;
struct rte_flow_item_tcp tcp_mask;
struct rte_flow_item tcp_item;
struct rte_flow_item_sctp sctp_spec;
struct rte_flow_item_sctp sctp_mask;
struct rte_flow_item sctp_item;
struct rte_flow_item pattern_ipv4_5tuple[4];
struct rte_flow_classify_rule *rule;
uint8_t ipv4_proto;
if (num_classify_rules >= MAX_NUM_CLASSIFY) {
printf(
"
INFO: classify rule capacity %d reached
",
num_classify_rules);
return ret;
}
/* set up parameters for validate and add */
memset(&ipv4_spec, 0, sizeof(ipv4_spec));
ipv4_spec.hdr.next_proto_id = ntuple_filter->proto;
ipv4_spec.hdr.src_addr = ntuple_filter->src_ip;
ipv4_spec.hdr.dst_addr = ntuple_filter->dst_ip;
ipv4_proto = ipv4_spec.hdr.next_proto_id;
memset(&ipv4_mask, 0, sizeof(ipv4_mask));
ipv4_mask.hdr.next_proto_id = ntuple_filter->proto_mask;
ipv4_mask.hdr.src_addr = ntuple_filter->src_ip_mask;
ipv4_mask.hdr.src_addr =
convert_depth_to_bitmask(ipv4_mask.hdr.src_addr);
ipv4_mask.hdr.dst_addr = ntuple_filter->dst_ip_mask;
ipv4_mask.hdr.dst_addr =
convert_depth_to_bitmask(ipv4_mask.hdr.dst_addr);
switch (ipv4_proto) {
case IPPROTO_UDP:
ipv4_udp_item.type = RTE_FLOW_ITEM_TYPE_IPV4;
ipv4_udp_item.spec = &ipv4_spec;
ipv4_udp_item.mask = &ipv4_mask;
ipv4_udp_item.last = NULL;
udp_spec.hdr.src_port = ntuple_filter->src_port;
udp_spec.hdr.dst_port = ntuple_filter->dst_port;
udp_spec.hdr.dgram_len = 0;
udp_spec.hdr.dgram_cksum = 0;
udp_mask.hdr.src_port = ntuple_filter->src_port_mask;
udp_mask.hdr.dst_port = ntuple_filter->dst_port_mask;
udp_mask.hdr.dgram_len = 0;
udp_mask.hdr.dgram_cksum = 0;
udp_item.type = RTE_FLOW_ITEM_TYPE_UDP;
udp_item.spec = &udp_spec;
udp_item.mask = &udp_mask;
udp_item.last = NULL;
attr.priority = ntuple_filter->priority;
pattern_ipv4_5tuple[1] = ipv4_udp_item;
pattern_ipv4_5tuple[2] = udp_item;
break;
case IPPROTO_TCP:
ipv4_tcp_item.type = RTE_FLOW_ITEM_TYPE_IPV4;
ipv4_tcp_item.spec = &ipv4_spec;
ipv4_tcp_item.mask = &ipv4_mask;
ipv4_tcp_item.last = NULL;
memset(&tcp_spec, 0, sizeof(tcp_spec));
tcp_spec.hdr.src_port = ntuple_filter->src_port;
tcp_spec.hdr.dst_port = ntuple_filter->dst_port;
memset(&tcp_mask, 0, sizeof(tcp_mask));
tcp_mask.hdr.src_port = ntuple_filter->src_port_mask;
tcp_mask.hdr.dst_port = ntuple_filter->dst_port_mask;
tcp_item.type = RTE_FLOW_ITEM_TYPE_TCP;
tcp_item.spec = &tcp_spec;
tcp_item.mask = &tcp_mask;
tcp_item.last = NULL;
attr.priority = ntuple_filter->priority;
pattern_ipv4_5tuple[1] = ipv4_tcp_item;
pattern_ipv4_5tuple[2] = tcp_item;
break;
case IPPROTO_SCTP:
ipv4_sctp_item.type = RTE_FLOW_ITEM_TYPE_IPV4;
ipv4_sctp_item.spec = &ipv4_spec;
ipv4_sctp_item.mask = &ipv4_mask;
ipv4_sctp_item.last = NULL;
sctp_spec.hdr.src_port = ntuple_filter->src_port;
sctp_spec.hdr.dst_port = ntuple_filter->dst_port;
sctp_spec.hdr.cksum = 0;
sctp_spec.hdr.tag = 0;
sctp_mask.hdr.src_port = ntuple_filter->src_port_mask;
sctp_mask.hdr.dst_port = ntuple_filter->dst_port_mask;
sctp_mask.hdr.cksum = 0;
sctp_mask.hdr.tag = 0;
sctp_item.type = RTE_FLOW_ITEM_TYPE_SCTP;
sctp_item.spec = &sctp_spec;
sctp_item.mask = &sctp_mask;
sctp_item.last = NULL;
attr.priority = ntuple_filter->priority;
pattern_ipv4_5tuple[1] = ipv4_sctp_item;
pattern_ipv4_5tuple[2] = sctp_item;
break;
default:
return ret;
}
attr.ingress = 1;
pattern_ipv4_5tuple[0] = eth_item;
pattern_ipv4_5tuple[3] = end_item;
actions[0] = count_action;
actions[1] = end_action;
/* Validate and add rule */
// 验证这条规则
ret = rte_flow_classify_validate(cls_app->cls, &attr,
pattern_ipv4_5tuple, actions, &error);
if (ret) { // 成功时返回 0
printf("table entry validate failed ipv4_proto = %u
",
ipv4_proto);
return ret;
}
// 调用 rte_flow_classify_table_entry_add() 将规则添加到 rte_flow_classifier 对象中的table。
/* 五个参数
1. classifier的句柄
2. 流规则属性
3. 模式规范
4. 关联的操作
5. 如果规则已经存在则返回1，否则返回0。
6. 仅在出错的情况下初始化这个结构。
*/
rule = rte_flow_classify_table_entry_add(
cls_app->cls, &attr, pattern_ipv4_5tuple,
actions, &key_found, &error);
if (rule == NULL) { // 成功时的有效句柄，否则为NULL
printf("table entry add failed ipv4_proto = %u
",
ipv4_proto);
ret = -1;
return ret;
}
rules[num_classify_rules] = rule;
num_classify_rules++;
return 0;
}
static int
add_rules(const char *rule_path, struct flow_classifier *cls_app) // 封装一层
{
FILE *fh;
char buff[LINE_MAX];
unsigned int i = 0;
unsigned int total_num = 0;
struct rte_eth_ntuple_filter ntuple_filter; // 用于定义n-tuple过滤器条目
int ret;
fh = fopen(rule_path, "rb"); // 打开 ipv4_rules_file.txt
if (fh == NULL)
rte_exit(EXIT_FAILURE, "%s: fopen %s failed
", __func__,
rule_path);
ret = fseek(fh, 0, SEEK_SET);
if (ret)
rte_exit(EXIT_FAILURE, "%s: fseek %d failed
", __func__,
ret);
i = 0;
while (fgets(buff, LINE_MAX, fh) != NULL) { // 读取一行内容
i++;
if (is_bypass_line(buff)) // 如果是注释行 or 空行就跳过
continue;
if (total_num >= FLOW_CLASSIFY_MAX_RULE_NUM - 1) { // 最大规则数量限制
printf("
INFO: classify rule capacity %d reached
",
total_num);
break;
}
if (parse_ipv4_5tuple_rule(buff, &ntuple_filter) != 0) // 过规则的 parser 解析出 rule
rte_exit(EXIT_FAILURE,
"%s Line %u: parse rules error
",
rule_path, i);
if (add_classify_rule(&ntuple_filter, cls_app) != 0) // 添加这条规则到 ACL 中
rte_exit(EXIT_FAILURE, "add rule error
");
total_num++;
}
fclose(fh);
return 0;
}
/* display usage */
static void
print_usage(const char *prgname)
{
printf("%s usage:
", prgname);
printf("[EAL options] -- --"OPTION_RULE_IPV4"=FILE: ");
printf("specify the ipv4 rules file.
");
printf("Each rule occupies one line in the file.
");
}
/* Parse the argument given in the command line of the application */
// 解析执行 flow_classify 的命令行参数
static int
parse_args(int argc, char **argv)
{
int opt, ret;
char **argvopt;
int option_index;
char *prgname = argv[0];
static struct option lgopts[] = {
{OPTION_RULE_IPV4, 1, 0, 0},
{NULL, 0, 0, 0}
};
argvopt = argv;
while ((opt = getopt_long(argc, argvopt, "",
lgopts, &option_index)) != EOF) {
switch (opt) {
/* long options */
case 0:
if (!strncmp(lgopts[option_index].name,
OPTION_RULE_IPV4,
sizeof(OPTION_RULE_IPV4)))
parm_config.rule_ipv4_name = optarg;
break;
default:
print_usage(prgname);
return -1;
}
}
if (optind >= 0)
argv[optind-1] = prgname;
ret = optind-1;
optind = 1; /* reset getopt lib */
return ret;
}
/*
* The main function, which does initialization and calls the lcore_main
* function.
*/
int
main(int argc, char *argv[])
{
struct rte_mempool *mbuf_pool;
uint8_t nb_ports;
uint16_t portid;
int ret;
int socket_id;
// 以下可以在 dpdk api data struct 中查看
struct rte_table_acl_params table_acl_params; // ACL 的参数
struct rte_flow_classify_table_params cls_table_params; // ACL中 table 的参数
struct flow_classifier *cls_app; // 分流器
struct rte_flow_classifier_params cls_params; // classifier 的参数
uint32_t size;
/* Initialize the Environment Abstraction Layer (EAL). */
ret = rte_eal_init(argc, argv); // 初始化 EAL
if (ret < 0)
rte_exit(EXIT_FAILURE, "Error with EAL initialization
");
argc -= ret;
argv += ret;
/* parse application arguments (after the EAL ones) */
ret = parse_args(argc, argv); // 解析除了 EAL 之外（也就是flow_classify）的命令行参数
if (ret < 0)
rte_exit(EXIT_FAILURE, "Invalid flow_classify parameters
");
/* Check that there is an even number of ports to send/receive on. */
nb_ports = rte_eth_dev_count(); // 网口数目必须是偶数
if (nb_ports < 2 || (nb_ports & 1))
rte_exit(EXIT_FAILURE, "Error: number of ports must be even
");
/* Creates a new mempool in memory to hold the mbufs. */
// 创建mempool
mbuf_pool = rte_pktmbuf_pool_create("MBUF_POOL", NUM_MBUFS * nb_ports,
MBUF_CACHE_SIZE, 0, RTE_MBUF_DEFAULT_BUF_SIZE, rte_socket_id());
if (mbuf_pool == NULL)
rte_exit(EXIT_FAILURE, "Cannot create mbuf pool
");
/* Initialize all ports. */
RTE_ETH_FOREACH_DEV(portid) // 端口初始化
if (port_init(portid, mbuf_pool) != 0)
rte_exit(EXIT_FAILURE, "Cannot init port %"PRIu8 "
",
portid);
if (rte_lcore_count() > 1) // 只需要一个逻辑核心
printf("
WARNING: Too many lcores enabled. Only 1 used.
");
socket_id = rte_eth_dev_socket_id(0); // 返回 0 号网口所在的NUMA socket id号
/* Memory allocation */
// 为分流器 cls_app 分配内存
size = RTE_CACHE_LINE_ROUNDUP(sizeof(struct flow_classifier_acl));// 返回大于或等于宏定义参数的第一个缓存对齐值
cls_app = rte_zmalloc(NULL, size, RTE_CACHE_LINE_SIZE); // DPDK的malloc：从调用该函数的核上的同一个NUMA socket的大页面区域分配堆内存。
// zmalloc 就是清零 与 calloc 相似
/* rte_zmalloc 参数三个：
1. 指示这块区域分配给怎样的object类型。用于debug用途。可以写NULL
2. size (in bytes) to be allocated
3. align
if 0, 会返回一个适合任何类型变量的指针，就像 malloc
否则，返回一个内存区域是 align 的对齐倍数，显然最小对齐是高速缓存行大小，宏：RTE_CACHE_LINE_SIZE
*/
if (cls_app == NULL) // 分配内存失败
rte_exit(EXIT_FAILURE, "Cannot allocate classifier memory
");
// classifier 的参数 有两个： name 和 socket id
cls_params.name = "flow_classifier";
cls_params.socket_id = socket_id;
// 调用 rte_flow_classifier_create() 函数来创建rte_flow_classifier对象。
// 参数是 rte_flow_classifier_params 结构体指针
cls_app->cls = rte_flow_classifier_create(&cls_params);
if (cls_app->cls == NULL) { // 创建失败
rte_free(cls_app);
rte_exit(EXIT_FAILURE, "Cannot create classifier
");
}
/* initialise ACL table params */
// 填写 ACL 的初始化参数
// 四个字段：
table_acl_params.name = "table_acl_ipv4_5tuple"; // 名字
table_acl_params.n_rules = FLOW_CLASSIFY_MAX_RULE_NUM; // 表中最大ACL规则数量：91
table_acl_params.n_rule_fields = RTE_DIM(ipv4_defs); // （todo） 一条ACL规则中的字段数量
memcpy(table_acl_params.field_format, ipv4_defs, sizeof(ipv4_defs)); // （todo） ACL rule的详细specification
/* initialise table create params */
// 填写 表 的创建参数
// 三个字段：
cls_table_params.ops = &rte_table_acl_ops; //（todo）Table operations (specific to each table type)
cls_table_params.arg_create = &table_acl_params; // 传递给表的用于创建的参数 这里是ACL的初始化参数结构体的指针
cls_table_params.type = RTE_FLOW_CLASSIFY_TABLE_ACL_IP4_5TUPLE; // （todo）table type
// rte_flow_classify_table_create() 在classifier对象中创建表。
// 参数两个：1. 流分类器的指针 2. 表创建的参数
ret = rte_flow_classify_table_create(cls_app->cls, &cls_table_params);
if (ret) { // 返回值：成功时返回 0
rte_flow_classifier_free(cls_app->cls);
rte_free(cls_app);
rte_exit(EXIT_FAILURE, "Failed to create classifier table
");
}
/* read file of IPv4 5 tuple rules and initialize parameters
* for rte_flow_classify_validate and rte_flow_classify_table_entry_add
* API's.
*/
// 然后它读取ipv4_rules_file.txt文件，初始化rte_flow_classify_table_entry_add() API 的参数。此API将规则添加到ACL表。
if (add_rules(parm_config.rule_ipv4_name, cls_app)) {
rte_flow_classifier_free(cls_app->cls);
rte_free(cls_app);
rte_exit(EXIT_FAILURE, "Failed to add rules
");
}
/* Call lcore_main on the master core only. */ // todo
lcore_main(cls_app);
return 0;
}
以上是关于DPDK flow_classify 源码阅读的主要内容，如果未能解决你的问题，请参考以下文章