guado_moblab: network_WlanDriver: Found no recognized wireless device |
||||||||
Issue descriptionhttps://stainless.corp.google.com/search?exclude_cts=true&exclude_non_release=true&board=%5Eguado%5C_moblab%24&test=%5Enetwork%5C_WlanDriver%24&view=matrix&col=build&row=model&first_date=2018-02-08&last_date=2018-02-14 I guess these don't have Wifi, but they have USE=wifi enabled.
,
Feb 14 2018
,
Feb 14 2018
Any tips on how we should rule out devices like this? Specific board exceptions? Labels in the lab?
,
Feb 14 2018
Since we want this test to run in bvt-cq / bvt-inline, labels are probably not best way forward. dchan@ / kmshelton@ - aren't moblab guado's same as regular guado's with modified image? If so, then these should have the wifi module (actual HW) onboard as well. Maybe its disabled?
,
Feb 14 2018
,
Feb 14 2018
yes, correct re: moblab == stock guado hw with different system image FYI, fw testing no longer uses moblabs for test parallelization (in the process of switching to labstations, which is guado with another image), as there are no dev resources for FAFT on moblab. cc'ing haddowk Moblab TL for authoritative moblab info re: whether the moblab system image disables wifi on moblabs
,
Feb 14 2018
We do not use wifi on moblab. Moblab code does some complex networking to get two Ethernet ports working one being a connection to the WAN and one being a private LAN, so although I am not aware of us code/setting that expressly disable wifi I am not surprised it does not work.
,
Feb 14 2018
I was mixed up when reading through the logs. There *is* a Wifi device: 02:00.0 0280: 8086:08b1 (rev bb) Subsystem: 8086:4070 ... Kernel driver in use: iwlwifi Kernel modules: iwlwifi so it's not like people modified these Guados. But it looks like moblab specifically blacklists it for shill: 2018-02-14T06:41:52.759131-08:00 NOTICE moblab-network-bridge-init[3043]: starting shill with lxcbr0,wlan0 and eth0 blacklisted I didn't even remember that was possible. I guess we could probably add checks for blacklisted interfaces.
,
Feb 14 2018
That's the best I could come up with for now: https://chromium-review.googlesource.com/c/chromiumos/third_party/autotest/+/919701 I don't think shill supports querying blacklisted devices very nicely.
,
Feb 21 2018
The following revision refers to this bug: https://chromium.googlesource.com/chromiumos/third_party/autotest/+/13036ce3355081c62855783fb20cd79192be01cb commit 13036ce3355081c62855783fb20cd79192be01cb Author: Brian Norris <briannorris@chromium.org> Date: Wed Feb 21 06:49:58 2018 network_WlanDriver: ask the system directly for interfaces, not shill It's possible for shill to ignore certain network interfaces (e.g., with moblab, we find that moblab-network-bridge-init is blacklisting wlan0, among others), causing this test to fail, when in fact, the Wifi driver is working just fine. I couldn't find a way to query shill to determine whether a device is "blacklisted" -- the best I could come up with was to look for a Device type of "unknown", except this required figuring out the device name without shill's help (shill doesn't enumerate blacklisted devices for us). Instead, I figured that because we're really looking for just kernel support in this test (and not really testing for shill behaviors), I think it makes sense to just query the kernel directly. So, let's directly list out /sys/class/net/ and look for Wifi devices. BUG= chromium:812064 TEST=network_WlanDriver, on guado_moblab, kevin, etc. Change-Id: I726d74f67e41c19f2bcb4d87737fce7cdabf5fc2 Signed-off-by: Brian Norris <briannorris@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/919701 Reviewed-by: Kevin Cernekee <cernekee@chromium.org> Reviewed-by: Eric Caruso <ejcaruso@chromium.org> [modify] https://crrev.com/13036ce3355081c62855783fb20cd79192be01cb/client/common_lib/cros/network/interface.py [modify] https://crrev.com/13036ce3355081c62855783fb20cd79192be01cb/client/site_tests/network_WlanDriver/network_WlanDriver.py
,
Feb 21 2018
,
Mar 5 2018
|
||||||||
►
Sign in to add a comment |
||||||||
Comment 1 by briannorris@chromium.org
, Feb 14 2018