p4factory/targets/basic_routing/p4src代码解读
headers.p4
header_type ethernet_t {
fields {
dstAddr : 48;
srcAddr : 48;
etherType : 16;
}
}
header_type ipv4_t {
fields {
version : 4;
ihl : 4;
diffserv : 8;
totalLen : 16;
identification : 16;
flags : 3;
fragOffset : 13;
ttl : 8;
protocol : 8;
hdrChecksum : 16;
srcAddr : 32;
dstAddr: 32;
}
}
定义了一个以太网包头，以及一个IPV4包头
parser.p4
parser start {
return parse_ethernet;
}
#define ETHERTYPE_IPV4 0x0800
header ethernet_t ethernet;
parser parse_ethernet {
extract(ethernet);
return select(latest.etherType) {
ETHERTYPE_IPV4 : parse_ipv4;
default: ingress;
}
}
header ipv4_t ipv4;
field_list ipv4_checksum_list { //提取包的field_list
ipv4.version;
ipv4.ihl;
ipv4.diffserv;
ipv4.totalLen;
ipv4.identification;
ipv4.flags;
ipv4.fragOffset;
ipv4.ttl;
ipv4.protocol;
ipv4.srcAddr;
ipv4.dstAddr;
}
field_list_calculation ipv4_checksum {
input {
ipv4_checksum_list;
}
algorithm : csum16;
output_width : 16;
} //通过实例计算出校验和
calculated_field ipv4.hdrChecksum {
verify ipv4_checksum;
update ipv4_checksum;
} //与该包的校验和想比较，看是否正确
parser parse_ipv4 {
extract(ipv4);
return ingress;
}
解释可以看这张图
报文进来后，直接到了parse_ethernet解析器，然后提取ethernet实例的etherType判断是否位0x800如果是在进入到parse_ipv4解析器否则，到达状态_condition_()
中间有个field_list_calculation ipv4_checksum 可以理解成是通过实例化的其他的field字段计算出checksum
calculated_field ipv4.hdrChecksum 计算出来的ipv4_checksum在于刚刚包的hdrChecksum作对比，判断是否一样，然后在更新他
basic_routing.p4
#include "headers.p4"
#include "parser.p4"
#define PORT_VLAN_TABLE_SIZE 32768
#define BD_TABLE_SIZE 65536
#define IPV4_LPM_TABLE_SIZE 16384
#define IPV4_HOST_TABLE_SIZE 131072
#define NEXTHOP_TABLE_SIZE 32768
#define REWRITE_MAC_TABLE_SIZE 32768
#define VRF_BIT_WIDTH 12
#define BD_BIT_WIDTH 16
#define IFINDEX_BIT_WIDTH 10
/* METADATA */
header_type ingress_metadata_t { //用户自定义元数据
fields {
vrf : VRF_BIT_WIDTH; /* VRF */
bd : BD_BIT_WIDTH; /* ingress BD */
nexthop_index : 16; /* final next hop index */
}
}
metadata ingress_metadata_t ingress_metadata;
action on_miss() {
}
action set_bd(bd) {
modify_field(ingress_metadata.bd, bd); //设置元数据的bd值
}
table port_mapping {
reads {
standard_metadata.ingress_port : exact;
}
actions {
set_bd;
}
size : PORT_VLAN_TABLE_SIZE;
}
action set_vrf(vrf) {
modify_field(ingress_metadata.vrf, vrf);
}
table bd {
reads {
ingress_metadata.bd : exact;
}
actions {
set_vrf;
}
size : BD_TABLE_SIZE;
}
action fib_hit_nexthop(nexthop_index) {
modify_field(ingress_metadata.nexthop_index, nexthop_index);
subtract_from_field(ipv4.ttl, 1); //让ipv4包ttl-1
}
table ipv4_fib {
reads {
ingress_metadata.vrf : exact;
ipv4.dstAddr : exact;
}
actions {
on_miss;
fib_hit_nexthop;
}
size : IPV4_HOST_TABLE_SIZE;
}
table ipv4_fib_lpm {
reads {
ingress_metadata.vrf : exact;
ipv4.dstAddr : lpm;
}
actions {
on_miss;
fib_hit_nexthop;
}
size : IPV4_LPM_TABLE_SIZE;
}
action set_egress_details(egress_spec) {
modify_field(standard_metadata.egress_spec, egress_spec);
}
table nexthop {
reads {
ingress_metadata.nexthop_index : exact;
}
actions {
on_miss;
set_egress_details;
}
size : NEXTHOP_TABLE_SIZE;
}
control ingress { //ingress流控制程序，规定了匹配过程
if (valid(ipv4)) {
/* derive ingress_metadata.bd */
apply(port_mapping);
/* derive ingress_metadata.vrf */
apply(bd);
/* fib lookup, set ingress_metadata.nexthop_index */
apply(ipv4_fib) {
on_miss {
apply(ipv4_fib_lpm);
}
}
/* derive standard_metadata.egress_spec from ingress_metadata.nexthop_index */
apply(nexthop);
}
}
action rewrite_src_dst_mac(smac, dmac) {
modify_field(ethernet.srcAddr, smac);
modify_field(ethernet.dstAddr, dmac);
}
table rewrite_mac {
reads {
ingress_metadata.nexthop_index : exact;
}
actions {
on_miss;
rewrite_src_dst_mac;
}
size : REWRITE_MAC_TABLE_SIZE;
}
control egress { //egress的流控制程序
/* set smac and dmac from ingress_metadata.nexthop_index */
apply(rewrite_mac);
}
这个也可以从图可得他想要实现的结果
上图两个团员可以看作是查表，线上文字代表执行相应的动作。下面解释一下这张图
首先一个有效ipv4实例进来后先进行端口的匹配，设置bd，然后接着匹配bd表，设置vrf，然后在匹配ipv4_fib这张表匹配到就到下一跳去处理，接下应该都比较好理解。
处理完ingress后在进入buffer区，然后在进行egress的处理，这里egress只有匹配一张表rewite_mac。
输出结果展示
Packet in on port 2 length 100; first bytes:
00010203 04050006 0708090a 08004500
new packet, len : 100, ingress : 2
rmt proc returns 0
ingress_pipeline: packet dequeued
parsing start
parsing parse_ethernet
parsing parse_ipv4
payload length: 66
all checksums are correct //检验和正确，开始进入相应匹配
Applying table port_mapping
Lookup key for port_mapping: //端口匹配
standard_metadata_ingress_port: 0x00000002,
table hit
**********
entry at index 1:
key:
standard_metadata_ingress_port: 0x00000002 (0 0 0 2),//key值
action:
set_bd //动作
action data:
bd: 0x0000000a (0 0 0 10), //等下要赋值的数据
**********
action set_bd
action data:
bd: 0x0000000a (0 0 0 10), //执行完成
executing next table for action
Applying table bd
Lookup key for bd:
ingress_metadata_bd: 0x0000000a, //bd匹配
table hit
**********
entry at index 0:
key:
ingress_metadata_bd: 0x0000000a (0 0 0 10), //找到了表项
action:
set_vrf
action data:
vrf: 0x0000001e (0 0 0 30),
**********
action set_vrf
action data:
vrf: 0x0000001e (0 0 0 30),
executing next table for action
Applying table ipv4_fib //匹配路由表
Lookup key for ipv4_fib:
ingress_metadata_vrf: 0x0000001e, ipv4_dstAddr: 0x0a000002,
table miss, applying default action
action on_miss //没有匹配到，执行on_miss操作
executing next table for action
Applying table ipv4_fib_lpm //执行ipv4路由表的另一种lpm匹配、
//与上面ipv4_fib区别的的是 ipv4_fib是
//精确匹配，而ipv4_fib_lpm执行的是三 //重匹配
//“动作-匹配表的每个表项都有一个掩码，将 //掩码和字段值进行逻辑与运算，再执行匹 //配。”
Lookup key for ipv4_fib_lpm:
ingress_metadata_vrf: 0x0000001e, ipv4_dstAddr: 0x0a000002,
table hit
**********
entry at index 1:
key:
ingress_metadata_vrf: 0x0000001e (0 0 0 30), ipv4_dstAddr: 0x0a000000 (10 0 0 0),
prefix_length:
40
action:
fib_hit_nexthop
action data:
nexthop_index: 0x00000003 (0 0 0 3),
**********
action fib_hit_nexthop
action data:
nexthop_index: 0x00000003 (0 0 0 3), //设置下一跳索引
executing next table for action
Applying table nexthop
Lookup key for nexthop:
ingress_metadata_nexthop_index: 0x00000003,
table hit
**********
entry at index 2:
key:
ingress_metadata_nexthop_index: 0x00000003 (0 0 0 3),
action:
set_egress_details
action data:
egress_spec: 0x00000003 (0 0 0 3),
**********
action set_egress_details
action data:
egress_spec: 0x00000003 (0 0 0 3),
executing next table for action //设置下一跳egress出端口
total length for outgoing pkt: 100
deparsing ethernet //逆解析ethernet解析器刚刚解析的过程
deparsing ipv4 //逆解析刚刚ipv4解析器逆解析的港城
total length for outgoing meta: 26 //总的要进入egress元数据长度为26
copying metadata //赋值元数据
deparsing standard_metadata //逆解析固有元数据
deparsing ingress_metadata //你解析用户自定义元数据
queuing system: packet dequeued
egress port set to 3
instance type set to 0
egress_pipeline: packet dequeued
parsing start
parsing parse_ethernet
parsing parse_ipv4 //然后在向刚刚那样解析一次
payload length: 66
extracting metadata //提取元数据
extracting all metadata for 0x7f9ef0000930
Applying table rewrite_mac
Lookup key for rewrite_mac:
ingress_metadata_nexthop_index: 0x00000003,
table hit
**********
entry at index 2:
key:
ingress_metadata_nexthop_index: 0x00000003 (0 0 0 3),
action:
rewrite_src_dst_mac
action data:
smac: 0x010303030303, dmac: 0x000303030303,
**********
action rewrite_src_dst_mac
action data:
smac: 0x010303030303, dmac: 0x000303030303, //重写源目mac
executing next table for action
total lenth for outgoing pkt: 100
deparsing ethernet
deparsing ipv4 //在逆解析一次
outgoing thread: packet dequeued
outgoing thread: sending pkt: Size[100]: Port[3] //包从端口3送出去