Fedora 17; rt73usb wireless stuck disabled

gregscott · #1 11th October 2012, 05:32 AM

I am at a dead-end here.

I have an ultra small form factor system here running Fedora 17. It has an rt73usb wireless B/G card inside. It's not a laptop - there is no wireless on/off switch on the front and no BIOS settings for wireless. It also has 5 wired NIC ports.

This box is a prototype for a small router/bridge/Access Point. That's why the 5 NIC slots and wireless. I love this hardware if I can get it working.

I loaded Fedora 17 a couple days ago and Fedora recognized everything. I tinkered a little with hostapd and then made the mistake of moving the box. The power cord came loose and I restarted. After the restart, that wireless is disabled and I haven't found any way to enable it yet.

Details - it's physically there - I think that Qcom device is my wireless card.

[root@my-fw ~]# lsusb
Bus 001 Device 003: ID 18e8:6229 Qcom RT2573
Bus 002 Device 002: ID 0624:0307 Avocent Corp.
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 002 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 004 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
Bus 005 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
[root@my-fw ~]#

I found and installed a utility named rfkill. Here is what rfkill tells me:

[root@my-fw ~]# rfkill list all
0: phy0: Wireless LAN
Soft blocked: no
Hard blocked: yes
[root@my-fw ~]#
[root@my-fw ~]# rfkill unblock all
[root@my-fw ~]#
[root@my-fw ~]# rfkill list all
0: phy0: Wireless LAN
Soft blocked: no
Hard blocked: yes
[root@my-fw ~]#

Hard blocked apparently means it's hardware disabled.

Trying to run hostapd by hand breaks, so forget about starting it as a service:

[root@my-fw ~]# hostapd /etc/hostapd/hostapd.conf
Configuration file: /etc/hostapd/hostapd.conf
Could not set interface mon.wlan0 flags: Operation not possible due to RF-kill
nl80211: Failed to set interface wlan0 into AP mode
nl80211 driver initialization failed.
rmdir[ctrl_interface]: No such file or directory
ELOOP: remaining socket: sock=4 eloop_data=0x822d878 user_data=0x822eb70 handler=0x8075010
ELOOP: remaining socket: sock=6 eloop_data=0x8230280 user_data=(nil) handler=0x807f1b0
[root@my-fw ~]#

Falling back - just trying to enable it at all breaks:

[root@my-fw ~]# ip link set dev wlan0 up
RTNETLINK answers: Operation not possible due to RF-kill
[root@my-fw ~]#

NetworkManager is running - looking at its GUI, it tells me the wireless device is hardware disabled, but gives me no interface to enable it. It also says Airplane mode is on, but when I click "Off", next time I bring up NetworkManager, Airplane mode shows on again.

I've also tried:

rmmod rt73usb
modprobe rt73usb

With no change in symptoms. When I rmmod rt73usb, the wlan0 device goes away. And then when I modprobe it back, wlan0 reappears.

There is an RPM package loaded with the rt73 firmware.

[root@my-fw ~]#
[root@my-fw ~]# rpm -qa | grep rt73
rt73usb-firmware-1.8-9.fc17.noarch
[root@my-fw ~]#

And, sure enough, there's a 7 year old .bin file sitting in /lib/firmware.

[root@my-fw ~]# ls -al /lib/firmware | grep rt73
-rw-r--r--. 1 root root 2048 Nov 30 2005 rt73.bin
[root@my-fw ~]#

Here is some more diagnostic info, all leading to the same conclusion:

[root@my-fw NetworkManager]# dmesg | grep rt73
[ 5.915823] Registered led device: rt73usb-phy0::radio
[ 5.915882] Registered led device: rt73usb-phy0::assoc
[ 5.915952] Registered led device: rt73usb-phy0::quality
[ 5.916537] usbcore: registered new interface driver rt73usb
[root@my-fw NetworkManager]#
[root@my-fw NetworkManager]#
[root@my-fw NetworkManager]# dmesg | grep wlan
[root@my-fw NetworkManager]#
[root@my-fw NetworkManager]#
[root@my-fw NetworkManager]# iwconfig
p4p1 no wireless extensions.

wlan0 IEEE 802.11bg Mode:Master Tx-Power=off
Retry long limit:7 RTS thr:off Fragment thr:off
Power Management:on

p3p1 no wireless extensions.

p2p1 no wireless extensions.

p2p2 no wireless extensions.

p1p1 no wireless extensions.

lo no wireless extensions.

virbr0 no wireless extensions.

[root@my-fw NetworkManager]#

I am at a dead-end - everything is telling me the card is sitting there disabled. How do I enable it and ensure it stays enabled? Surely there's some means to tell this card to turn itself back on. Or should I be looking for a different wireless card to fit in this teeny tiny enclosure?

thanks

- Greg Scott

Skull One · #2 11th October 2012, 08:23 AM

You will not be able to to anything until you have the hardware unblocked.
It remembers me something...
Look at your driver parameters ('modinfo rt73usb'). There is this:

Code:

parm:           nohwcrypt:Disable hardware encryption. (bool)

Try to load the driver with it, I guess using:

Code:

rmmod rt73usb
modprobe rt73usb nohwcrypt

And then, check the output of 'rkfill'.

gregscott · #3 11th October 2012, 03:12 PM

I think we're on the right track but this didn't make any difference.

[root@my-fw ~]# rmmod rt73usb
[root@my-fw ~]# modprobe rt73usb nohwcrypt
[root@my-fw ~]# rfkill list all
1: phy1: Wireless LAN
Soft blocked: no
Hard blocked: yes
[root@my-fw ~]#

Somehow or another, I need to tell that driver to tell that device to enable itself. I wonder if there is anything documenting what we can pass to that driver? Or maybe there's a better driver.

- Greg Scott

Skull One · #4 11th October 2012, 04:13 PM

Damn! I cannot remember... how was it already?
Maybe 'modprobe rt73usb nohwcrypt=1'?
Could you post the relevant part of 'dmesg' after the remove/add driver?

---------- Post added at 03:13 PM ---------- Previous post was at 03:07 PM ----------

You could also try the parameters of 'rfkill':

Code:

rmmode rfkill
modprobe rfkill default_state=1

gregscott · #5 11th October 2012, 04:22 PM

I've been tinkering with various rmmod and modprobe scenarios, so dmesg will have lots of messages. I found a website that claims to have driver parameters, but this is turning into a dead-end. You'll see below. I pasted in all the dmesg messages from the past few minutes below.

[root@my-fw ~]#
[root@my-fw ~]# dmesg | more
[ 0.000000] Initializing cgroup subsys cpuset
[ 0.000000] Initializing cgroup subsys cpu
[ 0.000000] Linux version 3.5.4-2.fc17.i686.PAE (mockbuild@) (gcc version 4.7
.2 20120921 (Red Hat 4.7.2-2) (GCC) ) #1 SMP Wed Sep 26 22:10:23 UTC 2012
.
.
.
[ 1317.307917] modem-manager[784]: segfault at 0 ip b771677c sp bfc5a600 error 4 in libmm-plugin-linktop.so[b7715000+3000]
[36663.994945] usbcore: deregistering interface driver rt73usb
[36689.106048] usb 1-3: >reset high-speed USB device number 3 using ehci_hcd
[36689.575151] ieee80211 phy1: >Selected rate control algorithm 'minstrel_ht'
[36689.575762] Registered led device: rt73usb-phy1::radio
[36689.575824] Registered led device: rt73usb-phy1::assoc
[36689.575883] Registered led device: rt73usb-phy1::quality
[36689.576502] usbcore: registered new interface driver rt73usb
[39510.602294] usbcore: deregistering interface driver rt73usb
[39590.440499] rt73usb: Unknown parameter `firmName'
[39617.780047] usb 1-3: >reset high-speed USB device number 3 using ehci_hcd
[39618.248983] ieee80211 phy2: >Selected rate control algorithm 'minstrel_ht'
[39618.249658] Registered led device: rt73usb-phy2::radio
[39618.249708] Registered led device: rt73usb-phy2::assoc
[39618.249756] Registered led device: rt73usb-phy2::quality
[39618.251109] usbcore: registered new interface driver rt73usb
[39650.883247] usbcore: deregistering interface driver rt73usb
[39654.508964] rt73usb: Unknown parameter `firmName'
[39705.306639] rt73usb: Unknown parameter `firmName'
[39710.288045] usb 1-3: >reset high-speed USB device number 3 using ehci_hcd
[39710.756954] ieee80211 phy3: >Selected rate control algorithm 'minstrel_ht'
[39710.757686] Registered led device: rt73usb-phy3::radio
[39710.757729] Registered led device: rt73usb-phy3::assoc
[39710.757785] Registered led device: rt73usb-phy3::quality
[39710.758631] usbcore: registered new interface driver rt73usb
[root@my-fw ~]#

---------- Post added at 10:22 AM ---------- Previous post was at 10:14 AM ----------

Trying to rmmod rfkill looks complicated. Looks like everyone and their cousin latches on to it:

[root@my-fw ~]# rmmod rfkill
Error: Module rfkill is in use by: cfg80211 bluetooth
[root@my-fw ~]# rmmod cfg80211
Error: Module cfg80211 is in use by: mac80211 rt2x00lib
[root@my-fw ~]# rmmod mac80211
Error: Module mac80211 is in use by: rt2x00lib rt2x00usb

But it looks like these modules parse the switches when you load them - they give a misleading error message when the switch is bogus, but at least we know what the message looks like.

[root@my-fw ~]# rmmod rt73usb
[root@my-fw ~]# modprobe rt73usb greg
ERROR: could not insert 'rt73usb': No such file or directory
[root@my-fw ~]# modprobe rt73usb nohwcrypt=1
[root@my-fw ~]# rfkill list all
4: phy4: Wireless LAN
Soft blocked: no
Hard blocked: yes
[root@my-fw ~]#

So now if we can figure out a way to tell that driver to tell the hardware to enable itself I think we're in business.

- Greg

Skull One · #6 11th October 2012, 04:51 PM

There is a special function to source to handle dependancies:

Code:

. /usr/lib64/pm-utils/functions
modunload rfkill

By the way, I do not think that rfkill will do it, since it is hardware related.

You are not alone: http://unix.stackexchange.com/questi...-even-when-not
It looks like it is related to the firmware. And I wonder about the message 'rt73usb: Unknown parameter `firmName''
My only idea is to try with ndiswrapper, just in case.

gregscott · #7 11th October 2012, 04:56 PM

> By the way, I do not think that rfkill will do it, since it is hardware related.

I agree. We need to find a way to tell the hardware to turn itself back on. Or get a different brand device if this one is fatally flawed.

> And I wonder about the message 'rt73usb: Unknown parameter `firmName''

Oh - that was me. I found a link to some info for the rt73 driver. Not rt73usb but the older rt73 driver. One of the parameters was "firmName={filename}". So I tried it, thinking maybe if I reloaded the firmware that might turn it back on. But it was a dead end. "firmName" apparently is not a valid parameter for the newer rt73usb driver.

- Greg

Skull One · #8 11th October 2012, 05:55 PM

I have read the rfkill doc (http://www.mjmwired.net/kernel/Documentation/rfkill.txt).
I wonder about the communication between the driver and rfkill.
Maybe you can try the following:

Code:

. /usr/lib64/pm-utils/functions
modunload rfkill
modunload rt73usb
modprobe rt73usb nohwcrypt=1

---------- Post added at 04:55 PM ---------- Previous post was at 04:28 PM ----------

hum... looks like the problem appeared in kernel 3.4. (see http://comments.gmane.org/gmane.linu....general/93621)
So, what is your kernel? Maybe you could try with a kernel 3.3?
Edit: looks like the patch was incorporated in kernel 3.5.6

gregscott · #9 11th October 2012, 06:05 PM

No effect. I forgot to mention, this is a 32 bit system. The CPU is a dual core atom and it wouldn't boot my 64 bit Fedora DVD, so I built it with 32 bits.

Anyway, I don't think I unloaded the rfkill module below. For what it's worth, I have done several warm and cold reboots. I have a hunch, what's going on is, that driver loads up the .bin firmware file and it just keeps the hardware in whatever state it's in. I was unlucky enough to pull the power cord, the hardware probably "thought" somebody triggered a non-existant on/off switch, and now there is nothing to send it whatever it needs to turn itself back on. Just a hunch.

[root@my-fw ~]# . /usr/lib64/pm-utils/functions
-bash: /usr/lib64/pm-utils/functions: No such file or directory
[root@my-fw ~]#
[root@my-fw ~]# . /usr/lib/pm-utils/functions
[root@my-fw ~]# modunload rfkill
FATAL: Module bnep is in use.
bash: log: command not found...
FATAL: Module bnep is in use.
bash: log: command not found...
[root@my-fw ~]# modunload rt73usb
[root@my-fw ~]# modprobe rt73usb nohwcrypt=1
[root@my-fw ~]# rfkill list all
5: phy0: Wireless LAN
Soft blocked: no
Hard blocked: yes
[root@my-fw ~]#

- Greg

---------- Post added at 11:59 AM ---------- Previous post was at 11:55 AM ----------

Just saw this:

> hum... looks like the problem appeared in kernel 3.4. (see
> http://comments.gmane.org/gmane.linu....general/93621)
> So, what is your kernel? Maybe you could try with a kernel 3.3?

Going to an earlier kernel might not be so easy with Fedora 17. I suppose I could brew up a custom one, but what a load of trouble for a $20 wireless card. If that turns out to be a workaround, I think I'd rather look forward than backwards.

[root@my-fw ~]#
[root@my-fw ~]# more /proc/version
Linux version 3.5.4-2.fc17.i686.PAE (mockbuild@) (gcc version 4.7.2 20120921 (Re
d Hat 4.7.2-2) (GCC) ) #1 SMP Wed Sep 26 22:10:23 UTC 2012
[root@my-fw ~]#

---------- Post added at 12:05 PM ---------- Previous post was at 11:59 AM ----------

Hmmm - how do I follow this link?
> http://comments.gmane.org/gmane.linu....general/93621)

I get page not found even when going only to http://comments.gmane.org. https://comments.gmane.org also times out. There's a www.gmane.org with a bazillion threads around Linux.

- Greg

Skull One · #10 11th October 2012, 06:06 PM

The patch was incorporated in kernel 3.5.6.
It is the last kernel, so update and try it.

Edit: whoops! sorry for the buggy link...
http://permalink.gmane.org/gmane.lin....general/96561

gregscott · #11 11th October 2012, 06:20 PM

Running a yum update right now...there's a kernel 3.5.6-1.fc17 so we should know in a few minutes after I grab some food and a reboot.

- Greg

---------- Post added at 12:20 PM ---------- Previous post was at 12:09 PM ----------

Holy moley - I think the udpate did the trick! I really like how fast f17 boots, btw.
.
.
.
Installed:
kernel-PAE.i686 0:3.5.6-1.fc17 kernel-PAE-devel.i686 0:3.5.6-1.fc17

Updated:
GConf2.i686 0:3.2.5-3.fc17 cmake.i686 0:2.8.9-1.fc17
firefox.i686 0:16.0-1.fc17 git.i686 0:1.7.11.7-1.fc17
kernel-headers.i686 0:3.5.6-1.fc17 libcap-ng.i686 0:0.7-1.fc17
libnl3.i686 0:3.2.13-1.fc17 openldap.i686 0:2.4.32-3.fc17
openldap-clients.i686 0:2.4.32-3.fc17 openldap-devel.i686 0:2.4.32-3.fc17
perl-Git.noarch 0:1.7.11.7-1.fc17 taglib.i686 0:1.8-2.fc17
usbredir.i686 0:0.4.4-1.fc17 xulrunner.i686 0:16.0-2.fc17

Complete!
[root@my-fw ~]# reboot

Broadcast message from root@my-fw on pts/1 (Thu, 11 Oct 2012 12:14:52 -0500):

The system is going down for reboot NOW!

[root@my-fw ~]#
login as: root
root@10.10.10.103's password:
Last login: Thu Oct 11 09:05:58 2012 from tinahp100b.infrasupportetc.com
[root@my-fw ~]# rfkill list all
0: phy0: Wireless LAN
Soft blocked: no
Hard blocked: no
[root@my-fw ~]# ifup wlan0

Determining IP information for wlan0...^C
[root@my-fw ~]#

(I don't have any DHCP server that device will see.)

[root@my-fw ~]#
[root@my-fw ~]# hostapd /etc/hostapd/hostapd.conf
Configuration file: /etc/hostapd/hostapd.conf
Using interface wlan0 with hwaddr 00:0d:f0:94:74:eb and ssid 'GregLinuxAP'

Skull One · #12 11th October 2012, 06:26 PM

Good!
It was just a driver bug in the end

gregscott · #13 11th October 2012, 06:30 PM

Thanks for the help tracking it down.

- Greg