1)rutilt: "Can't get STA config.."2)CTRL-EVENT-DIS

Live forum: http://rt2x00.serialmonkey.com/viewtopic.php?t=4530

impdonkey

03-01-2008 09:56:05

1) I have a USB ID 148f2573 stick. It has work unpredictably on my FC8 by means of iwconfig tools, in Ad-Hoc mode. Right now i manage to connect via wpa_supplicant ( ... -D ralink -i rausb0, etc.) (driver rt73usb)

In the config file

ap_scan=2

network={
ssid="****"
key_mgmt=NONE
wep_key0
wep_key1
wep_key2
wep_tx_keyidx=0
priority=5
auth_alg=SHARED

}

I have to "ifconfig $DEV up" first, then run wpa_supplicant, and it works; i connect online for a minute or two, and then

CTRL-EVENT-DISCONNECTED - Disconnect event - remove keys

and i'm no longer online.

??
----------------
2) rutilt

I've downloaded, compiled, and tried every version since 0.09.
Each version of rutilt i run gives me exact same pop-up

Can't get STA config through special ioctl. Operation not supported by the driver.

Please help!

(kernel 2.6.23.1-42.fc8 (4k stack -- hope it's not the stack again!)
thanks

impdonkey

03-01-2008 19:22:07

Sorry, I am not being whimsically impatient i am configuring a system for a friend, and time is running out for me to give it back to him (and demo the "glorious powers" of Linux to him for the first time (his first time ) )

I'd really like to set up the GUI wifi configurator -- rutils for him

Could you PLEEEZE help with this matter

(again, currently when I launch "rutilt", I get "Can't get STA config through special ioctl. Operation not supported by the driver. ")

(the other problem i mentioned in my previous post may not be as relevant - in Infrastructure mode this device has worked before for me (although not perfectly), and without the supplicant component. I just thought that the supplicant component would bring it to perfect stability)

Thanks a lot!

Starcrasher

04-01-2008 00:22:22

If you are using rt73usb module, it means that you are using the new rt2x00 driver. You certainly noticed that some other rt2x00 modules are loaded.
For this driver, you need to use -D wext option for wpa_supplicant.
You can't use this driver with RutilT.

For more stability you may prefer to use the latest legacy rt73 driver in CVS which are compatible with wpa_supplicant using option -D ralink. This driver has only a single module named rt73. It can be used with RutilT but, in this case, you can not use wpa_supplicant.

If you want to install the legacy driver, think about removing or blacklist the nextgen one to avoid conflict.

impdonkey

04-01-2008 16:58:20

If you are using rt73usb module, it means that you are using the new rt2x00 driver. You certainly noticed that some other rt2x00 modules are loaded.
For this driver, you need to use -D wext option for wpa_supplicant.
You can't use this driver with RutilT.
[/quote2z0cg6fm]

thank you thank you thank you (for the reply & clarification)

i don't have the machine at the moment to test and apply your instructions, but will likely have it again, for maintenance, because, speaking of multiple versions of drivers -- i believe I completely FUBARed the WIFI on that machine by frantically installing a variety of drivers and using blacklisting (it suddenly occurred to me that I don't know the proper way to remove/purge compiled kernel modules and, strangely, googling didn't help -- all i can think of is deleting *.ko files, but i have a hunch that may break something, and need to edit /etc/modules.conf as well. Plus, there are dependency trees with modules, and i don't know how to purge without breaking something...)

So, yesterday, all my damage manifested itself no sooner and no later than when i was doing my demo (naturally, after two weeks of rock-solid system stability characteristic of linux, under my heavy usage and configuration, it HAD to go into nasty lockups and freezes EXACTLY when I was introducing and touting the system to a highly valued friend with lots of community clout/influence... (!??! shock) -- sorry to digress off topic -- but what were the odds!? )

Anyhow... aside from having incorrect syntax in /etc/modules.d/blacklist, i also didn't know exactly which modules to blacklist. Bewilderingly, yesterday i broke even the previously perfectly working madwifi, which i had also installed on this machine to temporarily use my own, madwifi-compliant wifi card. The Madwifi card/driver had peaceful coexisted with the ralink card/driver -- i had BOTH plugged in simultaneously at one point -- no problems, no lockups or slowdowns, with iwconfig listing both, i believe.

Last night, I inserted the madwifi card (alone, after reboot!) and iwconfig would hang and system would slow down to a grind; extraction of card led to freezing.

And, i swear, all that had changed, since normal madwifi functionality, are the many installations of ralink drivers (and some blacklisting). Granted, i have installed a good variety of them, including even the RT61, while I tried briefly to make a miniPCI card work (apparently that one is hopelessly unsupported)

If someone could be so kind as to tell me how i can undo this mess, I would be incredibly grateful. (Short of reinstalling the OS!)

(Speaking of iwconfig / iwlist prior to the lockups, when I did "iwlist scan", it said rausb0 does not support scanning. Any way to scan with your drivers/utilities? I was hoping RutilT might)


For more stability you may prefer to use the latest legacy rt73 driver in CVS which are compatible with wpa_supplicant using option -D ralink.
This driver has only a single module named rt73. It can be used with RutilT but, in this case, you can not use wpa_supplicant.

If you want to install the legacy driver, think about removing or blacklist the nextgen one to avoid conflict.[/quote2z0cg6fm]

So, the trade-off for me is Either have RutilT without the wpa_supplicant, or the other way around? Is the wpa_supplicant what gives this driver greater stability, or is its stability independent of the wpa_supplicant?

My hardware is a USB stick -- are you saying it /will/ work with both the single-module rt73 and the multi-module rt73usb ?

Are there plans to provide BOTH RutilT support and wpa_supplicant in one driver?

BTW, I don't know what the benefits of wpa_supplicant are, exactly.

thanks again

Starcrasher

04-01-2008 18:52:11

In fact, you can use three different drivers.
The one you can grab from Ralink site, the rt73 CVS driver you can grab from the download page, or the new rt2x00 driver (rt73usb).
But of course, you need to choose one only.

From my point of view, stability is better without wpa_supplicant at the moment. wpa_supplicant is mainly a tool for authentication. It's not needed with the legacy driver. Most of time, you can use iwpriv or RutilT to do this.

I don't think it will be possible to use RutilT and wpa_supplicant at the same time because wpa_supplicant act as a daemon. But, there are already some graphical tools to operate with it.

The main reason to use the legacy driver is the fact the nextgen driver is not completely ready. Even if a huge work as already be done on it, leading to it's inclusion in next kernel, a lot of functionalities need more work. But when it will reach maturity, it will be more than a complete replacement and support for the legacy driver will be dropped. Don't expect that for this year but maybe next one.

I'm only a user just like you, all I know about these drivers and wpa_supplicant is the result of experience and the reading the various documentations included in tarballs and this forum.

impdonkey

04-01-2008 22:03:48

In fact, you can use three different drivers.
The one you can grab from Ralink site, the rt73 CVS driver you can grab from the download page, or the new rt2x00 driver (rt73usb).
But of course, you need to choose one only.
[/quote8mnkeqs5]

i think i installed all three, and they were all listed by lsmod. Oops! oops

Still, I wonder why then, suddenly, the madwifi drivers and card caused a lockup. Oh well. I won't have another go at it until Monday.


From my point of view, stability is better without wpa_supplicant at the moment. wpa_supplicant is mainly a tool for authentication. It's not needed with the legacy driver. Most of time, you can use iwpriv or RutilT to do this.
[/quote8mnkeqs5]

thanks. That probably saves me a lot of research time.


I don't think it will be possible to use RutilT and wpa_supplicant at the same time because wpa_supplicant act as a daemon. But, there are already some graphical tools to operate with it.
[/quote8mnkeqs5]

(for the record, i am quite OK with CLI, and usually try to recommend its virtues to others; it's just that for a complete newcomer to linux for whom years of pulldown menus and icon clicking has made any alternative unthinkable, a GUI tool such as RutilT would mitigate the "culture shock". That, compounded with the fact that networking and WiFi in particular has a learning curve in and of itself. A GUI layout is good for the newcomer's quick orientation.)


The main reason to use the legacy driver is the fact the nextgen driver is not completely ready. Even if a huge work has already be done on it, leading to its inclusion in the next kernel, a lot of functionalities need more work. But when it reaches maturity, it will be more than a complete replacement and support for the legacy driver will be dropped. Don't expect that for this year but maybe next one.

I'm only a user just like you, all I know about these drivers and wpa_supplicant is the result of experience and the reading the various documentations included in tarballs and this forum.[/quote8mnkeqs5]

The above explanation is very helpful. Your experience is brought to bear. Thanks.

Spy84464

06-01-2008 18:45:06

RutilT is able to configure rt73 stick (managed by our legacy driver) to connect to networks which use the following encryptions None, WEP, WPAPSK or WPA2PSK. If the network uses an authentication server (EAP, LEAP etc..., these are sometimes called WPA enterprise), you'll need to use wpa_supplicant.
Building a front-end to wpa_supplicant within RutilT is something I would really like to do, but I don't have the time at the moment unfortunately.
If you want to remove some kernel modules, just delete them (or move them away for backup purpose), then run "depmod -a".

impdonkey

08-01-2008 09:46:15

I cleaned out the various module versions mentioned previously, and downloaded and installed rt73-cvs-2008010800.

Initially, RutilT did come up, and seemed to detect networks fine.

However, when I try to add a detected network to the profiles and then click on "Apply", it crashes. Here is a crash report http//www.flight.us/bugs/rutilt-bugreport.txt

Other times it has crashed on startup, with this report

http//www.flight.us/bugs/rutilt-bugreport.txt

The above crash happened when I tried to start rutilt after successfully bringing up and configuring rausb0 (to fully functional state) by means of iwconfig (including encryption, essid, etc.).

Also, should I migrate this thread to the RutilT forum ?

thanks again.

impdonkey

08-01-2008 10:16:44

forgot to add

when the said CVS RT73 module is up and working the load average jumps way up.

In X/Gnome/KDE/Firefox/etc it jumps from 0.40 or so, to over 3.5 (!)

in CLI it jumped from 0.3 to about 2.00 (while downloading at 166 kB/s)

Could one do something to avoid this load average spike?

This is a P4 2.2GHz, 512MB Ram.

thanks

Spy84464

11-01-2008 22:04:17

Hello,
Which version of you RutilT are you running? There's something familiar in your logs, but the issue has been fixed with the latest release.

Regards,
Romain

impdonkey

16-01-2008 15:50:31

I compiled from cvs-2008010800

Also, a serious and frustrating issue

I am now certain that my madwifi functionality gets hosed by the installation of the above cvs driver.

It has been the exact same cause-and-effect on *three* distinct systems (albeit similar systems -- FC6, FC7.9 and FC8)

On all three machines I had madwifi drivers working without a glitch, with netgear and belkin cards. After installing the above cvs rt driver, the ralink usb stick works very well, on all three. HOWEVER! now, when i insert my madwifi card and do my tried and true "wlanconfig ath0 create wlandev wifi0 ... [etc]", and bring up the device with "ifconfig $DEV up" my system freezes stone cold (sorry about the oxymoron)

Again, the rt drivers work wonderfully, but, it appears, something somehow gets demolished in the madwifi realm. I have no clue where the overlap/interference might be; i would think the madwifi and rt* modules to be distinct and not encroach on each other -- apparently that's not the case.

I will try the very latest CVS, and report results