Received-SPF: pass (google.com: domain of linux-wireless-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Tue, 20 Apr 2021 06:32:44 +0530 (IST)
Message-Id: <20210420.063244.1157699536418751229.enometh@meer.net>
To:     linux-wireless@vger.kernel.org
Subject: rtw88 strange killall behaviour with nm
From:   Madhu <enometh@meer.net>
Mime-Version: 1.0
Content-Type: Text/Plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Precedence: bulk


I was using the rtw88 on (lspci: Realtek. RTL8822CE 802.11ac PCIe on a
Lenovo Thinkbook) fairly normally on gentoo (kernel 5.10.30) with
netifrc scripts.  Recently I tried to move the scripts to
networkmanager connection files.

1. With NM the WPA-PSK connection to my AP (dlink router - india)
keeps dropping (citing a CLASS2_FRAME_FROM_NONAUTH_STA error)

2. With a specific connection file - if I activate the connection with
nmcli conn activate the kernel apparently terminates all processes on
the system except the NetworkManager process. (For added fun this
crashes the drm video + keyboard if Xorg is running requiring a
powercycle)

How could this behaviour (all tasks being terminated) possibly happen?
Is it b related specifically to the rtw88 module?  Have you heard of
anything like this before?

[Other Issues I had were

3. occasional RIPs "purge skb(s) not reported by firmware"

4. RIPs when I try to run kismet which may be related to:

5. Usually I see only 2-3 APs from this card on linux. On a b43 driver
older laptop I can usually see 10-15 APs at the same time. I suspect
the kismet RIPS happen when processing these other APs
]

In any case I'm intrigued by the (2) kernel repeatably killing off all
my processes when nmcli fires up.  This is a bit hard to debug as
there is nothing helpful in dmesg afaict. (and syslog exits early)

How is this even possible. (Offhand I can't spot where nm is calling
killall)