无线客户端模式

本文概述了以ClientStation模式以无线连接设备的几种方法。

这种能力的用途很广泛,比如:

  • 将独立设备或network segment连接至wireless access point
  • 建立点对点连接;
  • 反转AP和客户端以适应驱动造成的限制;
  • 可以在不影响网络其余部分的情况下关闭独立设备。

OpenWrt支持多种客户端模式,包括使用WDS (Wireless Distribution System)的桥接模式, 客户端路由模式和仅在(老的)brcm-2.4硬件上实现的客户端桥接模式

WDS模式是无线802.11标准的一个非标扩展。它使用了4-address-format以允许基站之间透明网络桥接,实现了无线网络客户端在不同接入点之间漫游的无缝交接

由于其非标的本性,WDS在不同的无线驱动和制造商固件采用了不同的实现,以至于它们经常互不兼容。为使用WDS,所有部署的设备应当使用同样的软硬件以保证兼容性。

OpenWrt支持两种类型的WDS,根据无线芯片和驱动分类如下:

  • Broadcom WDS - 支持使用私有wl.o驱动的Broadcom无线芯片
  • AP-to-Sta WDS (“4addr mode”) - 同时支持Madwifimac80211所支持的无线设备(比如Atheros无线芯片)

WDS的最大优势是数据链路层支持的透明桥接以及跨无线连接的广播 - 网络连接的所有设备组成一个统一的广播域.

The setup of Broadcom WDS is explained in the recipe article WDS (Broadcom).

The setup of Madwifi or mac80211 WDS is explained in the recipe article WDS (atheros).

This option is the preferred approach for wireless chipsets that support the Linux mac80211 wireless drivers (e.g. Atheros wireless chipsets). If the file /etc/config/wireless looks like the following, then mac80211 drivers are in use.

config wifi-device 'radio0'
        option type 'mac80211'
        ...

The routed client mode is the most generic wireless option. It is supported by all chipsets and drivers and requires no special modifications. The downside of routed client mode is the inability to bridge network segments or relay broadcast traffic. This affects for example the Windows Network Neighbourhood where hosts from two network segments cannot “see” each other without a domain controller. Connection by IP address or host name is still possible.

With routed client mode there are two possibilities to implement the network topology, depending on the specific requirements.

Using masquerading (NAT) on a client router connects a network segment behind the client to an existing wireless network without further modifications to the access point. The downside is that hosts on the AP side cannot access hosts behind the client router. Masqueraded

Hosts from the client network (red) are able to reach hosts in the AP network (blue), the client router masquerades outgoing traffic. Hosts on the AP side only see the client router, all traffic originating from client hosts uses 192.168.1.30 as source address. No direct connection from LAN Host 1 or 2 to Client Host 1 or 2 is possible.

See the Routed Client (Using MASQUERADE) article for configuration instructions.

This option requires a static route on the AP pointing to the subnet behind the client router using the client router's IP on the AP network as a gateway. This allows hosts on both segments to reach each other directly, but it requires administrative access to the AP in order to configure the static route.

Routed network topology

Hosts from the client network (red) are able to directly communicate with hosts in the AP network (blue) and vice versa. The rectangles represent static route entries. See the Routed Client (Using Routing) article for configuration instructions.

The bridged client mode is a proprietary Broadcom extension called WET (Wireless Ethernet Transceiver) mode. It is mostly Layer 2 transparent but has some disadvantages that may hinder network connectivity under certain circumstances (see technical background section).

WET Mode

All hosts are within the same subnet, W-LAN and LAN are bridged on both the AP and the client router. For a configuration example of bridged client mode, read the Bridged Client (Broadcom) article.

It is possible to achieve a bridge-like client mode setup with the help of relayd.

Relayd Topology

The setup is explained in the Routed Client with relayd (Pseudobridge) article.

Transparent client bridging or bridged client mode is not possible with vanilla OpenWrt on all platforms except brcm-2.4 with the proprietary Broadcom driver. It is possible to achieve this goal by using WDS, the background is explained below.

There are 3rd-party patches to implement ARP-NAT for platforms other than brcm-2.4 but those are not part of OpenWrt. The Kamikaze based Gargoyle firmware has support for it but is not officially supported by the OpenWrt developers.

Problem with bridging in a plain AP-to-STA setup

The 802.11 standard only uses three MAC addresses for frames transmitted between the Access Point and the Station. Frames transmitted from the Station to the AP don't include the ethernet source MAC of the requesting host and response frames are missing the destination ethernet MAC to address the target host behind the client bridge.

  1. Bridged Host sends a packet to the Target host
  2. Frame is relayed via the W-LAN Client and the MAC address of the transmitting wireless adapter is used as source MAC, the sending ethernet MAC is discarded
  3. W-LAN AP receives the frame and redirects it to the Target
  4. Target receives the frame and generates a response
  5. Target responds to the received frame using the (wrong) source MAC as destination
  6. W-LAN AP relays the frame to the W-LAN Client with the given destination MAC
  7. W-LAN Client receives the frame and assumes it is the final destination since it's wireless MAC is used in the frame, the packet is not forwarded
  8. Bridged Host never sees a response frame since the W-LAN Client became the destination, no connection is possible

Bridging possible by using WDS

If WDS is used, both the AP and the Station switch to the 4-address-mode which enables transparent bridging on the client side.

  1. Bridged Host sends a packet to the Target host
  2. Frame is relayed via the W-LAN Client, the sending ethernet MAC is preserved
  3. W-LAN AP receives the frame and redirects it to the Target using the second source MAC as sender address
  4. Target receives the frame and generates a response
  5. Target responds to the received frame using the given source MAC as destination
  6. W-LAN AP relays the frame to the W-LAN Client with the right MAC as second destination address
  7. W-LAN Client receives the frame and redirects it to the final destination using the second destination MAC as target
  8. Bridged Host receives the response frame, connection is established

The proprietary wl.o Broadcom wireless driver implements an ARP-NAT (Layer 2 address translation) mechanism called WET mode. ARP-NAT is comparable to Masquerading used on Layer 3 to connect multiple hosts using only one globally routed public IP address.

However, the address translation used by the Broadcom driver is not fully transparent and can cause various hard to debug network issues:

  • A host on the AP side may not be able to reach a host on the Station side until the other side made an initial connection attempt
  • Layer 3 routing on the Station side using a host behind the AP as gateway is unreliable or does not work at all
  • The mapping entries in the ARP-NAT table may time out causing sudden connection loss
  • The translation table may run out of space resulting in connection problems for some hosts on the Station side

:!: FIXME Bridged client mode only works on brcm-2.4 all other platforms need non-standard patches or other workarounds

There are no bridged and routed modes on Kamikaze, per se. Instead, multiple interfaces are bridged with an entry in /etc/config/network like this:

config 'interface' option 'type' 'bridge' option 'ifname' 'eth0.0' *..

Then in /etc/config/wireless, set the network to the same network specified in the bridge:

config 'wifi-device ' *.. config 'wifi-iface' *.. option 'network'

Alternatively, but a little less flexibly, you can use this line in /etc/config/network:

# athx for Atheros, or wl0 for Broadcom option 'ifname' 'eth0.0 ath0'

For routed mode, the wireless device needs to be used in a normal network configuration in /etc/config/network. Then, iptables rules are used to forward packets between the networks. The default gateway on each network (this is routing; you're connecting two networks together) needs to forward packets destined for the other network to the wifi router, or each host on each network needs to know that the wifi router is the router for packets to the respective network.

Both Broadcom and Atheros chipsets support scanning with the iwlist command. This command will scan all interfaces for networks:

iwlist scanning
  • ifconfig
  • iwconfig
  • wpa_cli

If you're having trouble connecting to either a hidden AP or a virtual AP (usually because wpa_supplicant doesn't list it in a scan), make sure these options are set correctly in the wpa_supplicant config file:

  • ap_scan (See the example wpa_supplicant.conf for more details)
    • 1: wpa_supplicant initiates scanning and AP selection
    • 0: driver takes care of scanning, AP selection, and IEEE 802.11 association parameters
    • 2: like 0, but associate with APs using security policy and SSID (but not BSSID).
  • scan_ssid
    • 0: do not scan this SSID with specific Probe Request frames (default)
    • 1: scan with SSID-specific Probe Request frames (this can be used to find APs that do not accept broadcast SSID or use multiple SSIDs; this will add latency to scanning, so enable this only when needed)
This website uses cookies. By using the website, you agree with storing cookies on your computer. Also you acknowledge that you have read and understand our Privacy Policy. If you do not agree leave the website.More information about cookies
  • Last modified: 2018/03/03 20:33
  • by