CTX714 to a good home

CTX714 to a good home

[Howard Chu]

Hi,
   I recently bought a Creatix CTX714 in the hopes of using it in a Thinkpad 
T23. After much experimentation I've been unable to get it working, and am 
giving up. I bought it for $8 on eBay, but the shipping (from Hong Kong to Los 
Angeles) was over $20 so it's not worthwhile for me to return it for a refund. 
I'd like to send it to any interested developer to wrestle with and see if 
they can get it working. lspci identifies it as 1260:3390, and the p54pci 
driver talks to it, but I was never able to see any APs with it.

More background: the T23 had a 56K modem in the mini-PCI slot. I wanted to 
find a combo wifi/modem card with 802.11g and V.90, and the CTX714 appeared to 
be the only candidate, so when I saw one on eBay, I jumped on it.

The card came with a mini-CDROM with a Windows driver. The driver included a 
fullmac firmware, 1.0.1.0. The Creatix docs say that this card uses the same 
chipset/driver as their CTX405 card. Their docs also said that the CTX405 
would only work using ndiswrapper on Linux. (There are barely any CTX714 docs.)

I've already gone thru all the motions of trying the 1.0.1.0 and 1.0.4.3 
firmware with the original prism54 driver, as well as just trying the default 
softmac firmware for the p54pci driver, and ndiswrapper with the Windows 
driver. (On Ubuntu Jaunty, 2.6.28.x kernel.) I've had absolutely no luck. The 
drivers see the device, register an interface, and they show the correct MAC 
address for the card, but I never see any other stations. In fact the card 
didn't work under Windows either, so possibly there's a problem with the card 
itself, but I don't know how to diagnose this any further. And possibly it 
just won't work in any Thinkpad. The eBay ad said it would only work in two 
models of Medion laptop. I could give it a try in my Asus laptop, but I don't 
want to spend any more time on it; I've bought a Thinkpad T42 (which comes 
with a wifi card) to replace the T23 and will be selling the T23. If you want 
this CTX714 and are willing to try to make it work, email me. I'll send it to 
the first responder.

-- 
   -- Howard Chu
   CTO, Symas Corp.           http://www.symas.com
   Director, Highland Sun     http://highlandsun.com/hyc/
   Chief Architect, OpenLDAP  http://www.openldap.org/project/

orinoco_cs and WPA-EAP?

[Howard Chu]

Sigh, nothing's ever easy. After abandoning the CTX714 I mentioned in an 
earlier email, I find that the Thinkpad T42 I got from eBay came with some 
kind of Orinoco 802.11B mini-PCI card installed. Ubuntu Jaunty's 2.6.28 kernel 
loads up the orinoco_cs driver for this card. It appears that this driver only 
supports WPA-PSK, and I use WPA-EAP on my home wlan. Has anyone done any work 
on supporting WPA-EAP for this card?

As an exercise in futility, I also built a wlags49-h1-cs driver and 
wpa_supplicant-0.6.9 with Hermes driver support. After patching it all to 
compile and run on my 2.6.28 kernel, I found that it also only supports 
WPA-PSK. I'm going to guess that this was obviously a waste of time and these 
patches aren't interesting/useful going forward, since the original Lucent 
code is so ancient...

-- 
   -- Howard Chu
   CTO, Symas Corp.           http://www.symas.com
   Director, Highland Sun     http://highlandsun.com/hyc/
   Chief Architect, OpenLDAP  http://www.openldap.org/project/

Re: orinoco_cs and WPA-EAP?

[Dave]

Howard Chu wrote:
> Sigh, nothing's ever easy. After abandoning the CTX714 I mentioned in an 
> earlier email, I find that the Thinkpad T42 I got from eBay came with some 
> kind of Orinoco 802.11B mini-PCI card installed. Ubuntu Jaunty's 2.6.28 kernel 
> loads up the orinoco_cs driver for this card. It appears that this driver only 
> supports WPA-PSK, and I use WPA-EAP on my home wlan. Has anyone done any work 
> on supporting WPA-EAP for this card?

Michael Stokes, who is primarily responsible for debugging my initial
WPA efforts, reported getting WPA-EAP working with a Toshiba Wireless
WLAN v01.01 shortly after we got WPA-PSK working. However I haven't had
 independent confirmation of this.

The driver has evolved since then, so it's possible that we've broken
something.

Note that the card only supports TKIP, so it won't work if the AP
doesn't offer it. At that time there seemed to be an issue with
wpa_supplicant attempting to use CCMP when offered by the AP, despite it
also offering TKIP - the workaround being to explicitly specify TKIP in
wpa_supplicant.conf.



Regards,

Dave.

PS. My contact with Michael has only been in private mail, so I've bcc'd
him (rather than cc'ing in case he want to keep his email off public
lists) in the hope that he can supply you with the necessary hints to
get it working.

Re: orinoco_cs and WPA-EAP?

[Howard Chu]

Dave wrote:
> Howard Chu wrote:
>> Sigh, nothing's ever easy. After abandoning the CTX714 I mentioned in an
>> earlier email, I find that the Thinkpad T42 I got from eBay came with some
>> kind of Orinoco 802.11B mini-PCI card installed. Ubuntu Jaunty's 2.6.28 kernel
>> loads up the orinoco_cs driver for this card. It appears that this driver only
>> supports WPA-PSK, and I use WPA-EAP on my home wlan. Has anyone done any work
>> on supporting WPA-EAP for this card?
>
> Michael Stokes, who is primarily responsible for debugging my initial
> WPA efforts, reported getting WPA-EAP working with a Toshiba Wireless
> WLAN v01.01 shortly after we got WPA-PSK working. However I haven't had
>   independent confirmation of this.
>
> The driver has evolved since then, so it's possible that we've broken
> something.
>
> Note that the card only supports TKIP, so it won't work if the AP
> doesn't offer it. At that time there seemed to be an issue with
> wpa_supplicant attempting to use CCMP when offered by the AP, despite it
> also offering TKIP - the workaround being to explicitly specify TKIP in
> wpa_supplicant.conf.

Thanks, that may have been the issue I was facing at first. My AP was 
configured for CCMP. I've now set it to TKIP and seem to be getting further. 
But now when I start wpa_supplicant I see a stream of "Information frame lost" 
messages in dmesg that pretty much never end (interspersed once in a while 
with ratelimit messages). I wonder if this laptop's mini-PCI slot is hosed or 
something. Also, one of the 2 antenna cable's connector came off, and I've 
tried to crimp the cable back into the connector but it may not be making full 
contact. The joys of buying used hardware...

> Regards,
>
> Dave.
>
> PS. My contact with Michael has only been in private mail, so I've bcc'd
> him (rather than cc'ing in case he want to keep his email off public
> lists) in the hope that he can supply you with the necessary hints to
> get it working.
>


-- 
   -- Howard Chu
   CTO, Symas Corp.           http://www.symas.com
   Director, Highland Sun     http://highlandsun.com/hyc/
   Chief Architect, OpenLDAP  http://www.openldap.org/project/

Re: orinoco_cs and WPA-EAP?

[Dave]

Howard Chu wrote:
> Dave wrote:
>> Howard Chu wrote:
>>> Sigh, nothing's ever easy. After abandoning the CTX714 I mentioned in an
>>> earlier email, I find that the Thinkpad T42 I got from eBay came with some
>>> kind of Orinoco 802.11B mini-PCI card installed. Ubuntu Jaunty's 2.6.28 kernel
>>> loads up the orinoco_cs driver for this card. It appears that this driver only
>>> supports WPA-PSK, and I use WPA-EAP on my home wlan. Has anyone done any work
>>> on supporting WPA-EAP for this card?
>>
>> Note that the card only supports TKIP, so it won't work if the AP
>> doesn't offer it.
> 
> Thanks, that may have been the issue I was facing at first. My AP was 
> configured for CCMP. I've now set it to TKIP and seem to be getting further. 
> But now when I start wpa_supplicant I see a stream of "Information frame lost" 
> messages in dmesg that pretty much never end (interspersed once in a while 
> with ratelimit messages). I wonder if this laptop's mini-PCI slot is hosed or 
> something.

That message is printed in response to a firmware interrupt. It
indicates "an unsolicited Frame Structure cannot be generated... because
there is not enough Frame Structure buffer space available".

I suspect that the beacons/probe responses/scan results are either too
large/too frequent for the card to handle, or the firmware can't
decipher them.

Before enabling wpa_supplicant, can you try scan for your AP (as root):

iwlist ethX scan essid YYYYY

I'm interested in what it returns, and whether dmesg reports 'Ext scan
results too large' or something similar.


Dave.

Re: orinoco_cs and WPA-EAP?

[Dan Williams]

On Fri, 2009-05-15 at 21:42 +0100, Dave wrote:
> Howard Chu wrote:
> > Dave wrote:
> >> Howard Chu wrote:
> >>> Sigh, nothing's ever easy. After abandoning the CTX714 I mentioned in an
> >>> earlier email, I find that the Thinkpad T42 I got from eBay came with some
> >>> kind of Orinoco 802.11B mini-PCI card installed. Ubuntu Jaunty's 2.6.28 kernel
> >>> loads up the orinoco_cs driver for this card. It appears that this driver only
> >>> supports WPA-PSK, and I use WPA-EAP on my home wlan. Has anyone done any work
> >>> on supporting WPA-EAP for this card?
> >>
> >> Note that the card only supports TKIP, so it won't work if the AP
> >> doesn't offer it.
> > 
> > Thanks, that may have been the issue I was facing at first. My AP was 
> > configured for CCMP. I've now set it to TKIP and seem to be getting further. 
> > But now when I start wpa_supplicant I see a stream of "Information frame lost" 
> > messages in dmesg that pretty much never end (interspersed once in a while 
> > with ratelimit messages). I wonder if this laptop's mini-PCI slot is hosed or 
> > something.
> 
> That message is printed in response to a firmware interrupt. It
> indicates "an unsolicited Frame Structure cannot be generated... because
> there is not enough Frame Structure buffer space available".
> 
> I suspect that the beacons/probe responses/scan results are either too
> large/too frequent for the card to handle, or the firmware can't
> decipher them.
> 
> Before enabling wpa_supplicant, can you try scan for your AP (as root):
> 
> iwlist ethX scan essid YYYYY
> 
> I'm interested in what it returns, and whether dmesg reports 'Ext scan
> results too large' or something similar.

Yeah a lot of APs will stuff a ton of info into the IEs these days,
including WPS, WPA, 11d, blah blah.  It's not out of the question that
the firmware writers didn't anticipate or handle large numbers of IEs in
the beacons.  That said, what firmware version is he using?  Maybe
Apple's firmware fixes that or something?  Would be interesting if the
reporter could boot up an old iBook with an Airport card and see if that
works.

Dan

Re: orinoco_cs and WPA-EAP?

[Howard Chu]

Dan Williams wrote:
> On Fri, 2009-05-15 at 21:42 +0100, Dave wrote:
>> That message is printed in response to a firmware interrupt. It
>> indicates "an unsolicited Frame Structure cannot be generated... because
>> there is not enough Frame Structure buffer space available".
>>
>> I suspect that the beacons/probe responses/scan results are either too
>> large/too frequent for the card to handle, or the firmware can't
>> decipher them.
>>
>> Before enabling wpa_supplicant, can you try scan for your AP (as root):
>>
>> iwlist ethX scan essid YYYYY

That works fine. In fact, just "iwlist scan" returns the same result - only 
the info for my AP is returned. (There are at least 6 APs visible from an 
another laptop sitting nearby.)

>> I'm interested in what it returns, and whether dmesg reports 'Ext scan
>> results too large' or something similar.

No, it just continues with the stream of "frame lost" messages:

eth2: Information frame lost. (10 times, followed by 1
__ratelimit: 2 callbacks suppressed  (then 10 more frame lost messages, and so on)

The number in "callbacks suppressed" varies pretty widely, from 2 to 32.

> Yeah a lot of APs will stuff a ton of info into the IEs these days,
> including WPS, WPA, 11d, blah blah.  It's not out of the question that
> the firmware writers didn't anticipate or handle large numbers of IEs in
> the beacons.  That said, what firmware version is he using?  Maybe

Using 9.48, I so far haven't found 9.52 for download anywhere.

> Apple's firmware fixes that or something?  Would be interesting if the
> reporter could boot up an old iBook with an Airport card and see if that
> works.

-- 
   -- Howard Chu
   CTO, Symas Corp.           http://www.symas.com
   Director, Highland Sun     http://highlandsun.com/hyc/
   Chief Architect, OpenLDAP  http://www.openldap.org/project/

Re: orinoco_cs and WPA-EAP?

[Dave]

Howard Chu wrote:
> Dan Williams wrote:
>> On Fri, 2009-05-15 at 21:42 +0100, Dave wrote:
>>> Before enabling wpa_supplicant, can you try scan for your AP (as root):
>>>
>>> iwlist ethX scan essid YYYYY
> 
> That works fine. In fact, just "iwlist scan" returns the same result - only 
> the info for my AP is returned. (There are at least 6 APs visible from an 
> another laptop sitting nearby.)

The fact we can parse the scan result from your AP is good. It indicates
the AP doesn't transmit a large list of IEs which trip the card up. The
fact we can't see the other APs... I'll assume that's your dodgy antenna :)

>>> I'm interested in what it returns, and whether dmesg reports 'Ext scan
>>> results too large' or something similar.
> 
> No, it just continues with the stream of "frame lost" messages:
> 
> eth2: Information frame lost. (10 times, followed by 1
> __ratelimit: 2 callbacks suppressed  (then 10 more frame lost messages, and so on)

I had thought you didn't get the frame lost messages until you tried to
authenticate. If you always receive them, then it might just be due to
an AP you're not interested in (and you might be able to ignore these
warnings).

A few questions:

1. How far is wpa_supplicant getting in authentication? Use -dd
2. If it authenticates, did you check whether you could dhcp on eth2?

It would also be nice to identify the source of your frame lost
messages. Any chance you could have a look at the wireless traffic with
another card in monitor mode?


Regards,

Dave.

Re: orinoco_cs and WPA-EAP?

[Dan Williams]

On Fri, 2009-05-15 at 17:14 -0700, Howard Chu wrote:
> Dan Williams wrote:
> > On Fri, 2009-05-15 at 21:42 +0100, Dave wrote:
> >> That message is printed in response to a firmware interrupt. It
> >> indicates "an unsolicited Frame Structure cannot be generated... because
> >> there is not enough Frame Structure buffer space available".
> >>
> >> I suspect that the beacons/probe responses/scan results are either too
> >> large/too frequent for the card to handle, or the firmware can't
> >> decipher them.
> >>
> >> Before enabling wpa_supplicant, can you try scan for your AP (as root):
> >>
> >> iwlist ethX scan essid YYYYY
> 
> That works fine. In fact, just "iwlist scan" returns the same result - only 
> the info for my AP is returned. (There are at least 6 APs visible from an 
> another laptop sitting nearby.)
> 
> >> I'm interested in what it returns, and whether dmesg reports 'Ext scan
> >> results too large' or something similar.
> 
> No, it just continues with the stream of "frame lost" messages:
> 
> eth2: Information frame lost. (10 times, followed by 1
> __ratelimit: 2 callbacks suppressed  (then 10 more frame lost messages, and so on)
> 
> The number in "callbacks suppressed" varies pretty widely, from 2 to 32.
> 
> > Yeah a lot of APs will stuff a ton of info into the IEs these days,
> > including WPS, WPA, 11d, blah blah.  It's not out of the question that
> > the firmware writers didn't anticipate or handle large numbers of IEs in
> > the beacons.  That said, what firmware version is he using?  Maybe
> 
> Using 9.48, I so far haven't found 9.52 for download anywhere.

9.52 AFAIK is only available from Apple Airport drivers if you cut it
out of the .kext on a PPC Mac.

Dan