add support for simulating wifi

[Quackdoc]

@warlice it will need some testing first

[warlice]

I think it's need an auto-adjusting mechanizm like detecting the cfg80211 and virt_wifi ko, if exists, set a property into waydroid_base.prop.

[aleasto]

This is not working for me:
logcat.txt

$ sudo waydroid shell -- ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host 
       valid_lft forever preferred_lft forever
2: wifi_eth@if29: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether 00:16:3e:f9:d3:03 brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet6 fe80::216:3eff:fef9:d303/64 scope link 
       valid_lft forever preferred_lft forever
3: wlan0@wifi_eth: <BROADCAST,MULTICAST> mtu 1500 qdisc pfifo_fast state DOWN group default qlen 1000
    link/ether 00:16:3e:f9:d3:03 brd ff:ff:ff:ff:ff:ff

[warlice]

It lack of an automative action of connecting virtwifi.

[aleasto]

I'm not sure I understand.
What do we need to do? Why is it not in this PR? Is it working for you?

[TriMoon]

This needs rework.

1. Do NOT rename eth0 like that, people will not be happy to have their main ethernet connection be renamed like that.
   • Not to mention that systems can have different names instead of eth0.
     Most distros have nowerdays changed to use eno1 as default for example...
2. Not all systems have their wifi named wlan0.
   It's better to find an automated way of finding what name the wifi card(s) use.
   Maybe something like:

   wificards=( $(iwconfig 2>/dev/null | grep 'IEEE 802.11' | cut -d ' ' -f 1) )
   for wifi in "${wificards[@]}"; do
    ...
   done

Last but not least, i would prefer it much more to use netdev files to setup this virt_wifi link.

[TriMoon]

Besides i don't really understand the need for all this renaming of links anyhow 🤷‍♀️

[JamiKettunen]

If that part is to be moved to the host couldn't the virtual wlan setup be handled by the already existing https://github.com/waydroid/waydroid/blob/main/data/scripts/waydroid-net.sh? I'd prefer not adding dependencies on random systemd components for core fixes like "apps not detecting any networking" if that can be avoided

[TriMoon]

Agree about the place to move that part.
But strongly disagree with the "random systemd components" with same arguments for usage of non-systemd solutions on a systemd based system which most distro's use these days...

• Pssttt they already make use of systemd stuff, see waydroid-container.service 😉

If they want to be totally politically correct they should implement all this stuff they do now in scripts, inside a compiled C++ binary that directly calls kernel functions...

[JamiKettunen]

The init script is different because that's just the entrypoint and easily replicatable anywhere, if other waydroid parts depend on systemd handling something it'll just be broken elsewhere silently

[supechicken]

@TriMoon This eth0 is inside of Waydroid's own network namespace (which is a veth pair connected to the waydroid0 bridge in the host), and it is not related to the host's network devices

[aleasto]

@supechicken do you have this working? Can you help me understand what I'm missing?

[supechicken]

@supechicken do you have this working? Can you help me understand what I'm missing?

I haven't started trying this yet 😅 (will leave a comment here when got progress)

[sujoshua]

Any progress ?

[echthesia]

@aleasto I'm not sure this approach is even possible without changes to virt_wifi.ko. As best I understand from my attempts to make it work earlier, the wiphy created by the module is always created in the default (host) namespace and doesn't support being moved to the container namespace. Since the wiphy (with VirtWifi SSID, etc.) ends up on the host, at best you achieve nothing for the container and at worst you break the host's Wi-Fi connection.