Comment 79 for bug 1861936

@Juerg:

My config for my network setup, which is an AP via wlan0, not bridged and station via USB Wifi dongle:

-------------------------------------------------
#!/bin/bash

# Error management
set -o pipefail
set -o nounset
set -x

apt install -y hostapd dnsmasq
systemctl unmask hostapd
systemctl enable hostapd

systemctl stop hostapd
systemctl stop dnsmasq

echo -e "interface wlan0\n\tstatic ip_address=192.168.10.1/24\n\tnohook wpa_supplicant\n" >> /etc/dhcpcd.conf

systemctl restart dhcpcd
cp /home/pi/hostapd.conf /etc/hostapd/
sed -i 's/^#DAEMON_CONF=.*$/DAEMON_CONF="\/etc\/hostapd\/hostapd.conf"/' /etc/default/hostapd

mv /etc/dnsmasq.conf /etc/dnsmasq.conf.orig
cp /home/pi/dnsmasq.conf /etc/

systemctl unmask hostapd
systemctl enable hostapd
systemctl start hostapd
service dnsmasq start

-------------------------------------------------

Unlike the top post, I'm not using IPV6 so that is not a factor. However, the kernel
log --- CUT HERE --- ... NETDEV WATCHDOG occurs for me.

As I state above in my original post:

-------------------------------------------------
When I do connect, I am sending a lot of data. Hard to pin what actions are triggering the NETDEV WATCHDOG backtrace, but it leaves wlan1 in a state where it cannot come up again:

SIOCSIFFLAGS: No such device

ip address shows wlan1: <BROADCAST,MULTICAST> mtu 1500 qdisc mq state DOWN group default qlen 1000

vcgencmd shows:
Aug 6 2020 16:24:09
version af3...d5b337 (clean) (release) (start)
-------------------------------------------

Seems to me a driver issue when there is a flood of traffic. The older version
of firmware I posted above solves the problem.