Problem with rt73 and hostapd error MGMT (TX callback) fail

Live forum: http://rt2x00.serialmonkey.com/viewtopic.php?t=5005

ZX81

21-09-2008 12:01:49

Hello,

i compiled the rt73usb and the hostapd deamon with the latest git sources.
I can run hostapd and the wlan device switch into the master mode.
On an other PC I see the AP with the SSID name but I can't connect
to the hostapd AP.
In the hostapd Log are a lot of MGMT (TX callback) fail errors
and I asume that they are the reason why the connection fails.

I found a post on the gmane.linux.kernel.wireless.general forum
from Ivo van Doorn with the date 2008-08-15
The posting contains a conversation about the topic but without a solution.
So my question is, are there news regarding the MGMT (TX callback) fail problem?

IvD

21-09-2008 12:19:49

A solution was thought of, and it will be implemented, although the details for the implementation are at this time not completely clear yet.

ZX81

21-09-2008 12:33:32

Thanks for the quick reply.

But if I understand that correct, until the fix will be implemented,
within the rt73usb driver or the mac80211, the hostapd support
of the rt73usb driver is broken and will not work at the moment?

Rgeards
Markus

IvD

21-09-2008 13:00:05

that is right.

ZX81

21-09-2008 13:11:27

Thanks a lot for the answer.
This will save me a lot of time, I am sitting here the last two days
and try how to get these things to work )

Regards
Markus

stringZ

06-12-2008 18:22:31

I have the same problem. When (days? weeks? months?) will this issue be fixed at USB chips?

IvD

06-12-2008 19:00:50

no idea.

stringZ

07-12-2008 15:21:14

As far as I know the windows driver does support master mode and RalinkTech also encourages open source developers to do their best in improving their drivers for Linux/UNIX. This may sound stupid to You, but I need to ask Have you contacted with Ralink with these kinds of issues?

By the way, may the windows driver of Ralink supports master mode but the one for Linux does not, I already tried. Is there any other solution to bring this device into master mode?

Thanks
stringZ

IvD

07-12-2008 15:53:29

As far as I know the windows driver does support master mode and RalinkTech[/quotex8pkvs9a]

They do? Perhaps some experimental versions...


also encourages open source developers to do their best in improving their drivers for Linux/UNIX. This may sound stupid to You, but I need to ask Have you contacted with Ralink with these kinds of issues?
[/quotex8pkvs9a]

Well if they are interested in advancing the Linux drivers then they are free to contribute their code, but they are releasing Linux drivers without Master mode support. So that means they leave it up to us to conjure up some sort of support based on guesses on how it should work for Ralink hardware.
And yes, we have contacted Ralink about more direct input from them for the in-kernel drivers, but that happen only for the next-generation wireless card they produce, and even that is not sure. Overall they prefer the release of their own out-of-kernel Linux driver which this project maintains as the "legacy driver" (which still doesn't include master mode, because Ralink didn't put it in there)

Oh and before people start complaining that I am not complete in my answer They did develop an Master mode capable driver together with ASUS but that was for rt2560 only and was completely closed source.


By the way, may the windows driver of Ralink supports master mode but the one for Linux does not, I already tried. Is there any other solution to bring this device into master mode?[/quotex8pkvs9a]

A Use Ad-hoc mode
B Buy an AP and use managed mode

Also could you point me to a windows Ralink driver with Master mode support, I have several windows drivers from them and none of them support master mode.

stringZ

07-12-2008 17:06:06

Thanks for your answer.

[quoteh15ec5et]And yes, we have contacted Ralink about more direct input from them for the in-kernel drivers, but that happen only for the next-generation wireless card they produce, and even that is not sure.[/quoteh15ec5et]
So that's their dirty business, unfortunately, I completely see that (


[quoteh15ec5et]Also could you point me to a windows Ralink driver with Master mode support, I have several windows drivers from them and none of them support master mode.
[/quoteh15ec5et]
Official drivers from Ralink, like the one I use on Windows [bh15ec5et]IS_AP_STA_7x[/bh15ec5et] (can be reached from [urlh15ec5et]http://www.ralinktech.com/ralink/Home/Support/Windows.html[/urlh15ec5et] - the second one from the top) do support AP mode. There's a tool (Ralink configuration Tool - replaces Wireless Zero Configuration) that gets installed along with the driver and loads into the system tray, so when I right click on it there's a shortcut menu item "Switch to AP mode". When I click it I get an options dialog where I can configure my USB Dongle to be an access point.

Here are a little detailed information which comfirm the presence of AP mode...

[attachment=0h15ec5et]ralink-softap-directory.png[/attachmenth15ec5et]

[attachment=1h15ec5et]ralink-drivers-directory.png[/attachmenth15ec5et]

[attachment=2h15ec5et]ralink-apui.png[/attachmenth15ec5et]

Greets,
stringZ

IvD

07-12-2008 18:28:57

Hmm apparently that was added recently when they merged the drivers for rt73 and rt2570.
Well their legacy driver still lacks master mode support, and that is where the in-kernel drivers is based on.

But the legacy drivers don't fit into the kernel/mac80211 design so their trick to determining if a frame was send succesfully or not would probably not work. Additionally I doubt that their solution would work with hostapd which could cause a lot of people complaining about that as well.

Other drivers (from other manufacturers) have the same issue with determining TX status as well with USB hardware, so I hope a generic solution will be found and merged into the kernel soon.

Vern

08-12-2008 02:22:49

stringZ

What is your interest in master mode support?

Thanks,

stringZ

08-12-2008 10:10:57

[quote1bq8g8y7]Other drivers (from other manufacturers) have the same issue with determining TX status as well with USB hardware, so I hope a generic solution will be found and merged into the kernel soon.[/quote1bq8g8y7]

I'm eagerly waiting for that moment )

[quote1bq8g8y7]What is your interest in master mode support?[/quote1bq8g8y7]

I have a homeserver which is also a gateway and a router with two ethernet cards (one connected to the internet, the other connected to the LAN) runing Debian Linux. This manages my LAN and NAT. I would like to extend this to wireless and make an access point out of my homeserver, that's why. Since my motherboard (Pentium I / Gigabyte) is not PCI 2.2 compliant, only PCI 2.1 devices run properly and manufacturers don't give a shit about writing it down in their specifications what kinds of PCI slots would their product be working in (e.g. I have another card, Linksys Wireless.G, with Broadcom chipset (bcm43xx) and it's not working in this machine, it's listed in `lspci` as an Unknown Device and I can't do anything about it), so the only way I can connect a wireless device to my homeserver is USB.

Greets
stringZ

mgessner

10-12-2008 21:51:38

Howdy,

I'm curious about when this will work because I'm trying to create an access point for a product and we're trying to use the Ralink chipset.

One of their PDF advertisements says that linux host AP support exists.

We don't have the ability to use PCI, so we're kind of pushed into using the USB stuff by schedule and hardware.

Thanks,

-mgessner

IvD

10-12-2008 22:43:08

As I said, Ralink might have some solution for the USB TX status reporting, but I doubt it is a correct behavior. Secondly I have serious doubts if their AP mode implementation would work with hostapd.

Vern

11-12-2008 16:56:31

I'm curious about when this will work because I'm trying to create an access point for a product and we're trying to use the Ralink chipset.[/quote3ctbbxxx]Is this a commercial development effort you're working on?

Thanks,

stringZ

12-12-2008 10:42:12

[quote368sa5g1]As I said, Ralink might have some solution for the USB TX status reporting, but I doubt it is a correct behavior. Secondly I have serious doubts if their AP mode implementation would work with hostapd.[/quote368sa5g1]

Well, to be honest I don't care about hostapd, I'd use something else if worked so if anyone knows another solution under Linux, don't hesitate to tell me.

zhilla

27-12-2008 10:33:05

OK, exactly the same problem, tested with 1x rt2500pci and 1x rt61pci card.
With rt61pci I have the "unknown vendor specific information element ignored (vendor OUI 001018 len=9)" line, not with rt2500pci one, but other is same.

offtopic
Also tested some Atheros and Broadcom cards with hostap, so if original poster is interested, I had most success (indoor, same room) with a Broadcom b43 card (minipci) - cca 1.3 MB/s (down to 600-700) while atheros 5212 was at 600-700 (but unreliable, down to 50), and atheros 5413 card did not work - but that was expected, see
https//kerneltrap.org/mailarchive/linu ... 064/thread
All this with 2.6.28 + compat-wireless 26-12-2008 + hostapd git 26-12-2008 + libnl 1.1 + slackware 12.2

edit b43 minipci on a minipci to pci adapter. those adapers usualy compensate for the pci 2.1 problem (no 3.3v) so that could be an even better solution than usb.

markryde

04-01-2009 15:42:38

Hello,
I am having the same messages (error MGMT (TX callback) fail)
when working with the latest git hostapd and wireless-testing.

Is there anything new about it ? is it going to be fixed (I wish I could know from where to start in order to fix it , really).
Regards,
Mark

statfs

22-01-2009 15:41:48

any news about this problem? Still the same?

Fy_hertz

24-04-2009 20:13:47

Hello,
So any news ? I having the same troubles !

IvD

24-04-2009 20:26:35

As far as I know there is somebody working on this issue, although I haven't heard from him since he started.

denilsonsa

22-08-2009 04:07:59

Looks like I still have this [incqyue2l]MGMT (TX callback) fail[/incqyue2l] issue, and I'm using rt61pci.

On [incqyue2l]hostapd[/incqyue2l] output, I receive several of these messages (and basically nothing else)

[codencqyue2l]STA 00:1c:bf:bb:e1:8c sent probe request for broadcast SSID
MGMT (TX callback) fail
mgmt::proberesp cb[/codencqyue2l]
I've described everything I had done, step-by-step at my blog
http//my.opera.com/CrazyTerabyte/blog/ ... nk-rt61pci

I still want to setup my desktop RaLink RT2561/RT61 PCI card as an access point. After I get it working working, I would also like to enable encryption.

ZX81

27-08-2009 03:09:53

Same problem here under wireless-testing (master-2009-08-25)

MGMT (TX callback) fail
mgmtproberesp cb

I use the wireless-testing.git and the mastermode seems to be
broken since version v2.6.31-rc2
At the moment I use the wireless-testing.git version v2.6.31-rc1
and the mastermode works very well here.
So you could try to use wireless-testing and check out
the tag v2.6.31-rc1

denilsonsa

05-09-2009 13:17:43

At the moment I use the wireless-testing.git version v2.6.31-rc1
and the mastermode works very well here.
So you could try to use wireless-testing and check out
the tag v2.6.31-rc1[/quote15emriux]
I've just tried the tag "v2.6.31-rc1" from this repo
git//git.kernel.org/pub/scm/linux/kern ... rt2x00.git

The hostapd messages changed to these
[code15emriux]MGMT (TX callback) ACK
mgmt::proberesp cb[/code15emriux]
But still no luck in my laptop connecting to the wi-fi.

denilsonsa

23-10-2009 16:51:34

2 months after my previous attempt, I've tried again... And now master mode appears to be working! (needs more testing, though)

As before, I've described everything at my blog
http//my.opera.com/CrazyTerabyte/blog/ ... de-finally

An special note to the last part of my blog post. I guess my previous attempt had failed because I had the CLIENT misconfigured. I can't really confirm this, but today I've noticed that I had some incorrect lines at wpa_supplicant.conf on my notebook, and these lines might have been the cause of failure.