How Linux sets up the network in VMware

If your XXX card is not supported by the kernel, and you fear that your own compiled kernel will cause system damage, you can try this method ^_^

VMware Three network connection Internet settings:

1.bridge:

Default Use Vmnet0

The IP settings of the virtual machine are not used with the host network segment and the rest are the same as the host:

For example, host IP is 10.70.54.31, set virtual machine IP to 10.70.54.22. Netmask,broadcast,gateway,dns is the same as the host to achieve the virtual machine <---> Host virtual machine <----> Internet communications.

2.nat:

Default Use Vmnet8

Set up the virtual machine to use DHCP to access the Internet, Windows Select "Automatically get IP", Linux to open the DHCP service can

You can also manually set:

IP settings and vmnet8 with the network segment, gateway set into VMNET8 Gateway (/etc/vmware/vmnet8/nat/nat.conf) can be found in the gateway Vmnet8, usually xxx.xxx.xxx.2.

The Netmask,broadcast setting is the same as VMNET8, and the DNS settings are the same as the host.

such as Vmnet8 ip:172.16.249.1 gw:172.16.249.2

Virtual machine settings: ip:172.16.249.100 gw:172.16.249.2

3.host-only:

Default Use Vmnet1

The virtual machine IP settings and Vmnet1 with the network segment, gateway set to Vmnet1 IP, the rest of the same settings and Vmnet1, DNS settings and host the same

such as Vmnet1 ip:172.16.245.1

Virtual machine settings: ip:172.16.245.100 gateway:172.16.245.1

This enables the virtual machine <---> host communication, but the virtual machine <---> The Internet is still unable to communicate

Virtual machine and Internet communication:

1. Open Host Routing function

2. Set the iptables, make the host become a NAT server

1. Echo 1 >/proc/sys/net/ipv4/ip_forward This allows the host to have the routing function

2. Iptables-t nat-a postrouting-o eth0-s 172.16.245.0/24-j Masquerade

This rule means that the source IP from the 172.16.245.0/24 packet is disguised as eth0 IP, and the virtual machine communicates with the Internet.

If the network interface is ppp+ or PPPoE, it needs to be modified to-o PPPoE

Of course, S 172.16.245.0/24 also need to modify according to the actual situation

QEMU two ways to access the Internet:

User Mode network:

This way to achieve a virtual machine online is very simple, similar to the VMware Nat,qemu startup when the-user-net parameters, virtual machines using DHCP, you can communicate with the Internet, but this way the virtual machine and host communication is not convenient.

Tap/tun Network:

This is a bit more complicated than user mode, but it's easy to set up a virtual machine <--> an Internet virtual machine <--> host communication.

This way, a host-only,qemu like VMware is set up to use the TUN/TAP device to add a virtual network device (TUN0) to the host, and then configure it like a real network card.

First, tap/tuns devices are supported in the kernel:

Device Drivers--->

Networking Support--->

[M] Universal Tun/tap device driver Support

If the current kernel does not support it, you can recompile the module by adding it simply:

The current kernel profile CP to the kernel source directory:

[Root@lfs ~] #cp/boot/config-[kernel-version]/usr/src/linux

[Root@lfs ~] #cd/usr/src/linux

Configure the kernel to select the Tun/tap module (M) as shown above:

[Root@lfs ~] #make menuconfig

Re-compile only module (M), do not compile core (*) supported Dongdong:

[Root@lfs ~] #make Modules

After compiling, the Tun.ko can be found under/usr/src/linux/drivers/net:

[Root@lfs NET] #ls-L/usr/src/linux/drivers/net/tun.ko

-rw-r--r--1 root root 11116 Mar 20:29/usr/src/linux/drivers/net/tun.ko

[Root@lfs net]#

It is CP to the current kernel's module directory at the appropriate location:

[Root@lfs NET] #cp/usr/src/linux/drivers/net/tun.ko/lib/modules/' uname-r '/kernel/drivers/net

To re-establish a module dependency:

[Root@lfs NET] #depmod

Now it's ready to load:

[Root@lfs NET] #modprobe Tun

Check:

[Root@lfs NET] #lsmod |grep Tun

Tun 8704 0

[Root@lfs net]#

Ok. Success does not recompile the entire kernel join special module support

If your XX card is not supported by the kernel, you can compile into a module, you are afraid of recompiling the kernel after the problem can be used to compile only the module you need, and then manually install to the appropriate location, and then load it.

This compiles the speed to also be quicker than compiles the entire kernel, does not have any damage to the system, may use on the XX card. ^_^

Pay attention to three points:

1. Kernel source code must be exactly the same as the current kernel version, otherwise the compiled module is not available.

2. Note Only make modules (compiled module), no make Modules_install (automatic installation module to/lib/modules)

3. You must run Depmod before loading the newly compiled module, otherwise modprobe cannot find it

In fact, using the current kernel configuration file (/boot/config-[kernel-version]), only to add the modules you need, do not do any other changes, make modules_install should not have problems.

But the safest way is to manually install it, control in their own hands more down-to-earth:-)

OK, back to qemu Internet problem

If you use a Udev management device (typically the distribution of the 2.6.x kernel already uses udev), the/dev/net directory is automatically created after Modprobe Tun, and the Tun device is built to do the relevant linking:

[Root@lfs NET] #ls-L/dev/net/tun

lrwxrwxrwx 1 root 6 15:35/dev/net/tun->. /tun

[Root@lfs net]#

If it's unfortunate, you don't see it, you need to do the work yourself manually.

[Root@lfs ~] #mkdir/dev/net

[Root@lfs ~] #mknod/dev/net/tun c 10 200

OK, the related device is ready and you need a TUN/TAP initialization script:

/etc/qemu-ifup:

#!/bin/sh

/sbin/ifconfig $172.20.0.1

Then give Qemu-ifup x Execute permission to put it under/etc.

This script can only be executed by the root user, and if you need to use QEMU for the average user, you need to change to sudo/sbin/ifconfig ... Then set sudo related permissions.

When QEMU is started, it adds a virtual network device (TUN0) to the host:

[Root@lfs ~] #ifconfig tun0

Tun0 Link encap:ethernet hwaddr 0A3:8A:5D:97:CD

inet addr:172.20.0.1 bcast:172.20.255.255 mask:255.255.0.0

Up broadcast RUNNING multicast mtu:1500 metric:1

RX packets:0 errors:0 dropped:0 overruns:0 frame:0

TX packets:0 errors:0 dropped:0 overruns:0 carrier:0

collisions:0 txqueuelen:500

RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)

[Root@lfs ~]#

You can now start Qemu to configure the virtual machine network parameters, just like VMware host-only:

IP and tun0 the same network segment, Gateway for tun0 IP remaining parameters and tun0 the same, DNS and host the same:

tun0:ip:172.20.0.1 broadcast:172.20.255.255 netmask:255.255.0.0

qemu:ip:172.20.0.100 broadcast:172.20.255.255 netmask:255.255.0.0 gateway:172.20.0.1

Like Host-only, this only implements the virtual machine <----> communication between the hosts, but also need to set up Router,nat to connect to the Internet

[Root@lfs ~] #echo 1 >/proc/sys/net/ipv4/ip_forward

[Root@lfs ~] #iptables-t nat-a postrouting-o eth0-s 172.20.0.0/24-j Masquerade

[Root@lfs ~]#

OK, Virtual machine <---> Host virtual machine <----> Internet communication is complete.