chromeos4-row13-rack8-host5 fails network_WlanDriver.
Usually this is symptomatic of (a) NIC not being found on the bus
(b) driver not being present/loaded. Neither of the two is true here.
Not sure there is much debugging we can do here, since this system+wifi chip is not actively supported. But we should consider replacing this DUT, at least.
localhost /var/log # grep -i mwifiex *
messages:2018-06-11T21:11:37.541850+00:00 NOTICE kernel: [ 15.419451] mwifiex: info: mwifiex_add_card rx work enabled, cpus 4 :
messages:2018-06-11T21:11:37.541883+00:00 INFO kernel: [ 15.529411] mwifiex_sdio mmc1:0001:1: FW already running! Skip FW dnld
messages:2018-06-11T21:13:27.603931+00:00 INFO kernel: [ 125.486512] mwifiex_sdio mmc1:0001:1: FW failed to be active in time
messages:2018-06-11T21:13:27.603983+00:00 INFO kernel: [ 125.486544] mwifiex_sdio mmc1:0001:1: info: mwifiex_fw_dpc: unregister device
messages:2018-06-11T21:13:27.603999+00:00 INFO kernel: [ 125.488221] mwifiex_sdio mmc1:0001:1: info: FREE_CMD_BUF: cmd_pool is null
Comment 1 by briannorris@chromium.org
, Jun 12 2018Labels: OS-Chrome
There's a reason I added an exception to make this test non-fatal on kitty when running in BVT: EXCEPTION_BOARDS = [ # Exhibits very similar symptoms to http://crbug.com/693724 , # b/65858242, b/36264732. 'nyan_kitty', ] BTW, this test passes occasionally on this host. I'm also not sure replacing the DUT will actually fix this device. It's known that there are some behaviors in some Marvell firmwares where external noise at the "wrong" time can cause the firmware to fail to initialize at all, which would yield symptoms like this. They believe they fixed this in 8997 (kevin, bob), but it wouldn't be surprising if the problem exists on older devices. IOW, the problem may be a combination of bad firmware and bad surroundings, which won't be fixed by DUT replacement. I think Kevin previously started to debug some Snow problems too, and found that we had a bunch of devices in Wifi reset loops in one portion of the lab, and the best he could figure is that the surrounding RF environment was triggering some FW bugs.