I have a sand device with a techstop Monoprice USB 3.0 USB-ethernet dongle plugged into the left Type-A port.  I am running a test that suspends and resumes the sand every ~40 seconds, to try to track down a WPA-EAP problem:

#!/bin/bash

set -euxo pipefail

cd /usr/local/lib/flimflam/test
./disconnect-service "Android Statue" || true

while :; do
	date
	./connect-service "Android Statue"
	sleep 30
	./get-service 7f1a7ab0-1b9e-4260-9940-256a180f1902 | grep "State.*online"
	./disconnect-service "Android Statue"

	now=`cat /sys/class/rtc/rtc0/since_epoch`
	echo $((now+10)) > /sys/class/rtc/rtc0/wakealarm
	echo mem > /sys/power/state
done


After about 24 hours, the ethernet dongle vanishes without a trace.  It is no longer in `lsusb` and eth0 is missing.  There are no errors in the log.

On a normal resume I see:

2017-09-01T01:03:59.569935+00:00 INFO kernel: [98174.611124] PM: suspend exit 2017-09-01 01:03:59.569056092 UTC
2017-09-01T01:03:59.573957+00:00 INFO avahi-daemon[2123]: Withdrawing address record for 100.127.1.54 on eth0.
2017-09-01T01:03:59.574036+00:00 INFO avahi-daemon[2123]: Leaving mDNS multicast group on interface eth0.IPv4 with address 100.127.1.54.
2017-09-01T01:03:59.579223+00:00 INFO avahi-daemon[2123]: Interface eth0.IPv4 no longer relevant for mDNS.
[...]
2017-09-01T01:04:02.942882+00:00 INFO kernel: [98177.984535] r8152 2-2:1.0 eth0: carrier on
2017-09-01T01:04:02.962891+00:00 INFO kernel: [98178.004161] wlan0: RX AssocResp from 00:25:9c:13:bd:e9 (capab=0x11 status=0 aid=1)
2017-09-01T01:04:02.964337+00:00 INFO kernel: [98178.005603] wlan0: associated
2017-09-01T01:04:02.968462+00:00 ERR shill[22058]: [ERROR:wifi.cc(1352)] WiFi wlan0 EAPEventTask with no current service.
2017-09-01T01:04:03.259496+00:00 INFO avahi-daemon[2123]: Joining mDNS multicast group on interface eth0.IPv4 with address 100.127.1.54.
2017-09-01T01:04:03.259563+00:00 INFO avahi-daemon[2123]: New relevant interface eth0.IPv4 for mDNS.
2017-09-01T01:04:03.259575+00:00 INFO avahi-daemon[2123]: Registering new address record for 100.127.1.54 on eth0.IPv4.

But beyond line 42768 in messages.1 in the attached tarball, I no longer see any sign of life from eth0 or the r8152 driver.  I see a usb 2-2 disconnect at line 42885 and a usb 2-2 detection at line 42909, but no mention of eth0 or r8152.

This has happened twice so far (it's actually interfering with my WPA test since I started the WPA test over the network).