源 IP 和目标 IP 的环回地址

Question 1

所有组合可能用于环回测试，至少在 Linux 中如此，因为在什么情况下会发生环回几乎完全取决于local路由表中的路由类型local(ID 255)。这些路由决定了您可以绑定/使用哪些地址作为源地址，以及哪些流量将被“捕获”在主机本身中（而不是实际发送/转发到另一台主机）。

$ sudo ip r add local 10.100.100.100 dev lo src 10.100.100.100

（您dev eth0也可以将例如用于上述情况。在单播情况下，本地路由的指定接口似乎并不重要。但它似乎会导致某些广播目的地产生差异，例如255.255.255.255，可能是因为它用于确定这种情况下广播链路/域。）

$ ip r show table local type local
local 10.100.100.100 dev lo scope host src 10.100.100.100
local 127.0.0.0/8 dev lo proto kernel scope host src 127.0.0.1
local 127.0.0.1 dev lo proto kernel scope host src 127.0.0.1
local 192.168.1.143 dev eth0 proto kernel scope host src 192.168.1.143

（src是个首选源地址，将用作不回复当您未指定目的地时，使用相应目的地进行通信，例如不带的 ping -I some_address。)

$ ping -c 1 10.100.100.100
PING 10.100.100.100 (10.100.100.100) 56(84) bytes of data.
64 bytes from 10.100.100.100: icmp_seq=1 ttl=64 time=0.193 ms

--- 10.100.100.100 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.193/0.193/0.193/0.000 ms

$ ping -c 1 -I 10.100.100.100 127.1.100.1
PING 127.1.100.1 (127.1.100.1) from 10.100.100.100 : 56(84) bytes of data.
64 bytes from 127.1.100.1: icmp_seq=1 ttl=64 time=0.209 ms

--- 127.1.100.1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.209/0.209/0.209/0.000 ms

$ ping -c 1 -I 127.100.100.100 10.100.100.100
PING 10.100.100.100 (10.100.100.100) from 127.100.100.100 : 56(84) bytes of data.
64 bytes from 10.100.100.100: icmp_seq=1 ttl=64 time=0.223 ms

--- 10.100.100.100 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.223/0.223/0.223/0.000 ms

$ sudo sysctl -w net.ipv4.icmp_echo_ignore_broadcasts=0
net.ipv4.icmp_echo_ignore_broadcasts = 0

$ ping -c 1 -I 10.100.100.100 -b 255.255.255.255
PING 255.255.255.255 (255.255.255.255) from 10.100.100.100 : 56(84) bytes of data.
64 bytes from 10.100.100.100: icmp_seq=1 ttl=64 time=0.199 ms

--- 255.255.255.255 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.199/0.199/0.199/0.000 ms

以下是tcpdump在同一时间在同一主机上运行的相应输出：

$ sudo tcpdump -eni any icmp
tcpdump: data link type LINUX_SLL2
tcpdump: verbose output suppressed, use -v[v]... for full protocol decode
listening on any, link-type LINUX_SLL2 (Linux cooked v2), snapshot length 262144 bytes
16:50:01.183360 lo    In  ifindex 1 00:00:00:00:00:00 ethertype IPv4 (0x0800), length 104: 10.100.100.100 > 10.100.100.100: ICMP echo request, id 29, seq 1, length 64
16:50:01.183422 lo    In  ifindex 1 00:00:00:00:00:00 ethertype IPv4 (0x0800), length 104: 10.100.100.100 > 10.100.100.100: ICMP echo reply, id 29, seq 1, length 64
16:50:19.194256 lo    In  ifindex 1 00:00:00:00:00:00 ethertype IPv4 (0x0800), length 104: 10.100.100.100 > 127.1.100.1: ICMP echo request, id 30, seq 1, length 64
16:50:19.194321 lo    In  ifindex 1 00:00:00:00:00:00 ethertype IPv4 (0x0800), length 104: 127.1.100.1 > 10.100.100.100: ICMP echo reply, id 30, seq 1, length 64
16:50:43.600107 lo    In  ifindex 1 00:00:00:00:00:00 ethertype IPv4 (0x0800), length 104: 127.100.100.100 > 10.100.100.100: ICMP echo request, id 31, seq 1, length 64
16:50:43.600173 lo    In  ifindex 1 00:00:00:00:00:00 ethertype IPv4 (0x0800), length 104: 10.100.100.100 > 127.100.100.100: ICMP echo reply, id 31, seq 1, length 64
16:51:24.764481 lo    In  ifindex 1 00:00:00:00:00:00 ethertype IPv4 (0x0800), length 104: 10.100.100.100 > 255.255.255.255: ICMP echo request, id 32, seq 1, length 64
16:51:24.764548 lo    In  ifindex 1 00:00:00:00:00:00 ethertype IPv4 (0x0800), length 104: 10.100.100.100 > 10.100.100.100: ICMP echo reply, id 32, seq 1, length 64

默认情况下，您不能绑定未被任何本地路由覆盖为目的地的地址：

$ ping -I 192.168.1.123 127.0.0.1
ping: bind: Cannot assign requested address

（尽管据我所知，您可以使用 sysctl 来实现这一点。）

由于本地路由是127.0.0.0/8自动添加的（作为前缀路由，因为127.0.0.1/8在lo虚拟的127.0.0.1如果接口处于 up 状态，则可以使用和（含）范围内的任意地址127.255.255.254作为源地址。（127.0.0.0是的子网 ID 127.0.0.0/8，127.255.255.255是 IP 子网的广播地址，因此应避免使用它们。）

虽然127.0.0.0/8预订的作为“环回地址”，这并不意味着只有涉及该块的地址时才会发生环回。

Answer