RTL8189FS: Fix P2P-GO advertising

[Azq2]

Description

Most of Realtek drivers have a bug: P2P IE not passed to AP beacon.
This causes problems, when P2P-GO looks like a regular AP. All P2P capable devices can't see my P2P-GO in the list.

I found a driver where P2P-GO works fine (according to user reviews): https://github.com/aircrack-ng/rtl8812au
And found these differences. After I backported these changes my P2P-GO works fine.

Also this changes is totally safe, because affects only P2P-related code

This patch contains my patches to the original driver repo:

1. rtl8189fs: Fix p2p-go advertising jwrdegoede/rtl8189ES_linux#87
2. rtl8189fs: Fix uninitialized cfg80211_chan_def jwrdegoede/rtl8189ES_linux#88

But the owner is too busy to review that :(

How Has This Been Tested?

1. Create /etc/wpa_supplicant/wpa_supplicant-wlan0.conf

ctrl_interface=/var/run/wpa_supplicant
update_config=1
device_name=orangepilite
device_type=1-0050F204-1
config_methods=virtual_push_button
p2p_no_group_iface=1

2. Setup DHCP for the wlan0: /etc/systemd/network/wlan0.network

[Match]
Name=wlan0

[Network]
Address=10.7.7.7/24
DHCPServer=true
IPMasquerade=ipv4

[DHCPServer]
PoolOffset=100
PoolSize=20
EmitDNS=yes
DNS=1.1.1.1

3. Setup services

# We don't need NM
sudo systemctl disable NetworkManager
sudo systemctl stop NetworkManager

# For test we need systemd-networkd
sudo systemctl enable systemd-networkd
sudo systemctl start systemd-networkd

# Useless without NM
sudo systemctl stop wpa_supplicant
sudo systemctl disable wpa_supplicant

# Enable wpa_supplicant for wlan0
sudo systemctl enable wpa_supplicant@wlan0
sudo systemctl start wpa_supplicant@wlan0

4. Now create P2P-GO:

sudo wpa_cli p2p_group_add

5. Open on android "Wifi Direct" section in settings.
   Now you can see "orangepilite" device in the list. Just click it!

6. Accept join req from your phone to P2P-GO

sudo wpa_cli wps_pbc

In wpa_supplicant logs you can see something like that:

<3>P2P-DEVICE-FOUND xx:xx:xx:xx:xx:xx p2p_dev_addr=xx:xx:xx:xx:xx:xx pri_dev_type=10-0050F204-5 name='Android_6d7f' config_methods=0x188 dev_capab=0x25 group_capab=0x0 new=1
<3>P2P-PROV-DISC-PBC-REQ xx:xx:xx:xx:xx:xx p2p_dev_addr=xx:xx:xx:xx:xx:xx pri_dev_type=10-0050F204-5 name='Android_6d7f' config_methods=0x188 dev_capab=0x25 group_capab=0x0

<3>WPS-PBC-ACTIVE 
<3>P2P-PROV-DISC-PBC-REQ xx:xx:xx:xx:xx:xx p2p_dev_addr=xx:xx:xx:xx:xx:xx pri_dev_type=10-0050F204-5 name='Android_6d7f' config_methods=0x188 dev_capab=0x25 group_capab=0x0
<3>CTRL-EVENT-SUBNET-STATUS-UPDATE status=0
<3>CTRL-EVENT-EAP-STARTED xx:xx:xx:xx:xx:xx
<3>CTRL-EVENT-EAP-PROPOSED-METHOD vendor=0 method=1
<3>CTRL-EVENT-EAP-PROPOSED-METHOD vendor=14122 method=254
<3>WPS-REG-SUCCESS xx:xx:xx:xx:xx:xx xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
<3>WPS-PBC-DISABLE 
<3>WPS-SUCCESS 
<3>CTRL-EVENT-EAP-FAILURE xx:xx:xx:xx:xx:xx
<3>CTRL-EVENT-SUBNET-STATUS-UPDATE status=0
<3>AP-STA-CONNECTED xx:xx:xx:xx:xx:xx p2p_dev_addr=xx:xx:xx:xx:xx:xx
<3>EAPOL-4WAY-HS-COMPLETED xx:xx:xx:xx:xx:xx

That's successful connection.

Checklist:

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published in downstream modules

[Azq2]

I think, this patch must be ported to all realtek drivers. That easily.
Having working P2P in the armbian is a great feature.

But I have only RTL8189FS :(

[igorpecovnik]

I think, this patch must be ported to all realtek drivers. That easily.

Agree, but we are also very low on time :( I hope patch just applies elsewhere too but probably not ...

But I have only RTL8189FS :(

We have around 25 different USB wireless chips in the lab for testing purposes. Plan is to share them via usbip and run automated tests with. But since I am as busy as Hans :) adaptors just sit there doing nothing :(

[Azq2]

Sounds cool :)

If this PR will be successful, I can try to adapt this patch to other RTL wifi's.
Theoretically, this  is safe, because it affects only P2P-GO code, which is currently broken anyway.
And all testing will be delegated to end-users, which want to use P2P.

Also I have tinkerboard with RTL8723BS. With same bug. I'll send a patch for it too.

[hzyitc]

We have around 25 different USB wireless chips in the lab for testing purposes. Plan is to share them via usbip and run automated tests with. But since I am as busy as Hans :) adaptors just sit there doing nothing :(

If necessary, I can try to help.

[The-going]

Plan is to share them via usbip and run automated tests with.

Do I understand correctly, we are talking about this one?
linux-mainline/6.1/tools/usb/usbip/*

[igorpecovnik]

Idea was summarised in https://armbian.atlassian.net/browse/INFRA-3

Do I understand correctly, we are talking about this one?

Yes, but until test scripting is not production ready, it can run on x86 machinery - where this just works.

[The-going]

Yes, but until test scripting is not production ready, it can run on x86 machinery - where this just works.

I will formulate the question differently.
Do I need to implement a package build for usbip? When we assemble the target kernel (current, edge).

[igorpecovnik]

Do I need to implement a package build for usbip? When we assemble the target kernel (current, edge).

yes. usbip client is needed. assembled on the device or packaged. But its not urgent as we only need this, when this test system is functioning. I think on x86 just works as I was already testing this by hand few months ago.

[The-going]

yes. usbip client is needed. assembled on the device or packaged.

Good. I'll do that, too.