Wandboard Quad WiFi woes

[Alarmed]

Would like to report my configuration and share output in case this has been seen before.

$this->bbcode_second_pass_code('', '
root@alarm ~]# uname -a
Linux alarm 3.10.17-3-ARCH #1 SMP Tue Jul 15 20:38:48 MDT 2014 armv7l GNU/Linux

root@alarm ~]# ip link
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: eth0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast state DOWN mode DEFAULT group default qlen 1000
link/ether 00:1f:7b:b4:0b:16 brd ff:ff:ff:ff:ff:ff
3: wlan0: <NO-CARRIER,BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state DORMANT mode DORMANT group default qlen 1000
link/ether 40:2c:e3:be:3d:19 brd ff:ff:ff:ff:ff:ff
')

Dmesg output is weird:
$this->bbcode_second_pass_code('', '
[ 7.454976] brcmfmac: F1 signature read @0x18000000=0x9934329
[ 7.455784] brcmfmac: brcmf_sdio_chip_drivestrengthinit: No SDIO Drive strength init done for chip 4329 rev 3 pmurev 6
[ 7.491720] ERROR: v4l2 capture: slave not found!
[ 7.547273] unable to find transceiver of type USB3 PHY
[ 7.551276] dwc3 dwc3.0.auto: no usb3 phy configured
[ 7.554986] platform dwc3.0.auto: Driver dwc3 requests probe deferral
[ 7.720268] usb 1-1: reset high-speed USB device number 2 using ci_hdrc
[ 7.751979] brcmfmac: brcmf_c_preinit_dcmds: Firmware version = wl0: Sep 2 2011 14:48:19 version 4.220.48
[ 7.765693] brcmfmac: brcmf_fil_cmd_data: Failed err=-23
[ 7.769709] brcmfmac: brcmf_fws_init: failed to set bdcv2 tlv signaling
[ 7.778102] unable to find transceiver of type USB3 PHY
[ 7.778118] usbcore: registered new interface driver brcmfmac
[ 7.787416] dwc3 dwc3.0.auto: no usb3 phy configured
...
...
...
[10484.465112] brcmfmac: brcmf_fil_cmd_data: Failed err=-25
[10484.469148] brcmfmac: brcmf_run_escan: error (-25)
[10484.473043] brcmfmac: brcmf_fweh_event_worker: event handler failed (33)
[11149.458210] brcmfmac: brcmf_fil_cmd_data: Failed err=-25
[11149.462249] brcmfmac: brcmf_run_escan: error (-25)
[11149.466329] brcmfmac: brcmf_fweh_event_worker: event handler failed (33)
[11572.647459] brcmfmac: brcmf_fil_cmd_data: Failed err=-25
[11572.651477] brcmfmac: brcmf_run_escan: error (-25)
[11572.655290] brcmfmac: brcmf_fweh_event_worker: event handler failed (33)
[11874.929545] brcmfmac: brcmf_fil_cmd_data: Failed err=-25
[11874.933561] brcmfmac: brcmf_run_escan: error (-25)
[11874.937289] brcmfmac: brcmf_fweh_event_worker: event handler failed (33)
[12177.207986] brcmfmac: brcmf_fil_cmd_data: Failed err=-25
[12177.212004] brcmfmac: brcmf_run_escan: error (-25)
[12177.215891] brcmfmac: brcmf_fweh_event_worker: event handler failed (33)
[12630.626157] brcmfmac: brcmf_fil_cmd_data: Failed err=-25
[12630.630175] brcmfmac: brcmf_run_escan: error (-25)
[12630.634072] brcmfmac: brcmf_fweh_event_worker: event handler failed (33)
[12963.111774] brcmfmac: brcmf_fil_cmd_data: Failed err=-25
[12963.115790] brcmfmac: brcmf_run_escan: error (-25)
[12963.121760] brcmfmac: brcmf_fweh_event_worker: event handler failed (33)
[13114.250891] brcmfmac: brcmf_fil_cmd_data: Failed err=-25
[13114.254947] brcmfmac: brcmf_run_escan: error (-25)
[13114.258887] brcmfmac: brcmf_fweh_event_worker: event handler failed (33)
[13567.669820] brcmfmac: brcmf_fil_cmd_data: Failed err=-25
[13567.673879] brcmfmac: brcmf_run_escan: error (-25)
')

Had installed the following packages:

$this->bbcode_second_pass_code('', 'firmware-brcm43xx-1.1-2-armv7h.pkg.tar.xz
linux-firmware-20140603.a4f3bc0-1-any.pkg.tar.xz
')

I've noticed that if I open up the wandboard and press the reset button, wifi works alright for a bit. Then goes back to "no-carrier" mode.

[pepedog]

Does power cycling work?
https://groups.google.com/forum/m/#!top ... 3MhCOpv2g0

[Alarmed]

I have tried power cycling, and similar result as pressing reset. WiFi is detected and will go into "no carrier" mode randomly.

[pepedog]

The most likely cause of this happening is the brcm chipset is a piece of ####

[PLyttle]

Are you using a BCM4330? In that case the hardware interface is not completely covered by rfkill. The wlan wakeup interfacing has changed from a dedicated pin to pulsing the WL_REG_ON pin low.
There is a thread on it over at the wandboard forum, here... http://forums.wandboard.org/viewtopic.p ... d78c0aea60. but not a fix yet.
I keep wlan alive by pinging it once a minute as a workaround. Not pretty, but workable.

Pepedog, you've probably seen more of this chip, but all the problems on wandboard I see currently have more to do with implementation than anything else.
Speaking of which, why are the firmware binary files (brcmfmac4329/30-sdio.bin) contained in the linux-firmware package, while the text files (brcmfmac4329/30-sdio.txt) are in firmware-brcm43xx?
Now you need to install Bluetooth when you want to run Wlan, which is a bit odd.

LP

[pepedog]

I don't see why linux-firmware would be touched, it's an "any" package not modified from archlinux
I see your point, have seen arguments elsewhere that all 4 pieces should be builtin to kernel

[Alarmed]

Will it be a good idea to have a common meta package for each model of wandboard which takes care of some of these things? I.e. one package called wandboard-quad-c will include everything needed for wandboard-quad (including u-boot, kernel, firmware, modules).

[PLyttle]

I guess that would be a nightmare for package maintainers. You don't want the same item to be present in several packages. I feel for Pepedogs point of view. Archlinux wants so use generalized packages as much as possible. This creates transparency. Compiled-in firmware is not the right answer either. We want to end up mainline eventually. that would imply losing system dependencies, not adding them. I was of the impression that the aforementioned .txt files and the .bin files are a set, and thus should appear together. I'm probably wrong (again)

Part of the "woes" of wifi and bluetooth is hunting down the correct firmware, and using the correct file names. Users without a serial debug interface end up failing to notice the missing firmware and get frustrated. But that is what these forums are for

LP

[Alarmed]

> You don't want the same item to be present in several packages.

I thought the idea of a meta package is that it would simply be a pointer. Whichever maintainer is an expert on Quad for example, can make a list of packages needed for hardware to function on that device and submit that. Others can do the same. And these packages can be mutually exclusive.