HardwareComparison | LinuxDrivers | DlinkCard

27 July 2004

I hope I'm not posting too much, but it doesn't really seem like anyone is going to read this anyways. I was up until like 5AM last night trying to get it to work correctly (and then I was late to work this morning). The driver is installed, but now it doesn't autoload properly. I can run the start_net script each time I start up the comp, but I don't want to do that. The other problem is, is that it installed the hardware as eth1, an ethernet card, even though the description says it's a "wireless adapter," which is a problem because now there's a driver device eth1 tied to it and my wlan0 is dependant on it somehow even thought the eth1 always gives an error at startup that it couldn't load proplerly. Even if I tell it not to load eth1 at startup, it makes it harder for me to get wlan0 to work. I have some help here at work, though, so I plan to get it up soon. I guess I'm just posting here for no real reason, but I thought someone might like to listen, and by the time I'm done with this I'll have plenty of experience so I'll be able to help anyone that needs it. I suppose anyone could e-mail me at Muhahahaha@patmail.com if they wanted, but put "ACX100" in the subject so it'll make it easier to find among all the spam.

-- Linux user #362310 http://counter.li.org/


27 July 2004 (barely...)

Wow, 12:51 in the morning and I finally got it! The TI module wouldn't work, so I tried the latest fix of the sourceforge driver (pre8, fix 20) and at the same time I was messing around with a USB drive my friend just gave me, so I d/led the latest fedora core 2 linux kernel 2.6.6 onto it (and smp as well, HT technolgy and all...). I was following the directions using 2.6.5, but to no avail, so I decided that I'd try 2.6.6. Amazingly, it decided to work, although at first it wouldn't connect because I had the wrong DNS IP address. Changing it to 192.168.0.1 (usually the default for most routers) and running the script again allowed it to connect so here I am. Not on the XP comp anymore, but using Epiphany on my linux machine, hoorah!

-- Linux user #362310 http://counter.li.org/


26 July 2004

Damn, I just tried to use their driver, but since the acx100_pci.o module is in the .o module format, rather than the .ko module format required by 2.6, it won't work. Is there any way to edit .o files or transfer them to .ko format? I'm not really sure what a "module" is, so maybe not, hopefully I won't have to go back to 2.4, though. Any help would be greatly appreciated, thanks.

-- Linux user #362310 http://counter.li.org/


26 July 2004

By the way, I'm not sure if anyone noticed yet, but after looking around on-line I noticed that TI DOES have a driver for the ACX100 and that they RELEASED it for linux. It can be found at http://www.network-drivers.com/drivers/85/85964.htm, but I just got it, so I haven't tried it yet. They say it works for Linux 2.4.18 and later, but I've had trouble with the sourceforge driver on fedora core 2 with linux 2.6.5, so we'll see if that will work. I suppose it would, but I don't really know much about the kernel and how the changes in the new version might affect that. It's weird, I thought they were still refusing to release it, but once I found it I realized that all the sites I had been to were a year old. Hope this helps anyone.

-- Linux user #362310 http://counter.li.org/


24 September 2003 WARNING: Some software installations may attempt to flash their own firmware, but instead damage your computer. The D-Link DWL-650+ software would not install on my target HP Omnibook, so I tried it on my newest HP Omnibook, which already had built-in wireless. The DWL-650+ worked irratically. After I un-installed the D-Link software, my laptop's internal wireless no longer functioned. I am a SW/HW engineer, and after 20 hours of work and calls to D-Link support, I believe that the laptop's original wireless firmware has been corrupted beyond repair -- a sickening price to pay. (This was moved from HardwareComparison)

Edit by Core: This is unlikely to be the reason. Namely DWL-650+ doesn't have any ROM, rather the firmware is in the driver, so that messing with the firmware is imposible (not valid for APs (like 900+)).


10 September 2003

You can find the explanation how to add a pigtal to TrendNet TEW-301PC (exactly the same hardware as 650+) on http://www.wifihr.net/Sadrzaj/samogradnje/tew-301pc.html

The text is in Croatian (sorry for no translation), but I think you will get the picture. If you click on the pic you will get full resolution pic. Also there is pic where you can see the radio part of the card, and the "switcher" chip (marked red)(the part responsible for troublesome devirsity mode). If you have problems with the devirsity mode, or want a 3dBi stronger signal (also 3dB beter sensitivity -> mine works at -96dB), you can remove the chip!

Edit by Core (core@email.hinet.hr)


17 September 2003 Well, my labmate got a 650+ working with an external antenna... had to move 2 caps on the board... will post a site when he has it up.

cheers, moose


31 Aug 2003 I took apart my 650+ Airplus card to see about putting on an external antenna knowing the older 650 was able to be modified with one model already having a connector. The 650+ Airplus has no provision that I could see. Email me for photos of card after it was taken apart. If someone wants to host the photos email me and I will get them to you.

Jeff n8zuz1 at yahoo dot com


25 Aug 2003

The sourceforge project at http://acx100.sourceforge.net has released version 0.1h of their opensource driver. Give it a try!

-tom


20 Aug 2003


I have been trying to get DWL650+ PCMCIA working with Mandrake 9.1 using binary acx100 drivers.

Tried the link below http://gollum.boplats.net/ but couldn't get access to that server.

Basically ifconfig says eth1 is up. iwconfig sees it, but I can't get to any access points.

Any thoughts?

-greg


3 July 2003 The sourceforge project has released an Alpha opensource driver. More details at http://acx100.sourceforge.net/


16 June 2003

Well first off i beleve wlan-ng doesnt support dwl650+ Secondly as far as i know the kernel pcmcia driver i82365 does not work with the dwl650+ driver,

atleast not on some laptops

Read my howto, link further down on page

/Ixxe


12 June 2003

There are many differences. What Kernel are you using (any version has some differences in the allowed parameters)? I see that you are using the Kernel driver for the pcmcia cards, what happens (like in my case) if the usual (10baseT) card is not supported by the Kernel? Can you still download and install the pcmcia_cs sources?

What glibc are you using? I have had very strange gehaviour by wine just because it was compiled with the 2.2.x and I have installed the 2.3.x. and it is quite impossible to return to the older version (practically all the major applications depend on it!!).

Ivor you are doing a very good job. I am pretty sure that it could be a serious problem due to the different glibc used for the acx100sta compilation and my installation.

I have been living without the DWL-650+ (closed in the drawer near my bed - what romantic!) so far and I can wait until you will have updated to glibc 2.3.x (which will occurr sooner or later). At that time I will have no more excuses not to have it up and running (apart a bogus PCI-to-pcmcia bridge TI1410 which seems to give problems with the IRQ assigned to the devices, but this is another "movie").

Thank you a lot and if you could post also other configuration files (like the wlan-ng and the wlan-ng.conf files, just to see what is the string that you have configured to "trigger" the module loading) would be fine.

Marco

---

12 June 2003

1. I found the 2.4.20-8 binary is a better match for the 2.4.21 kernels than the Mandrake objs.

2. I've uploaded my ".config" to http://www.ivor.it/wireless/.config, do you want to diff on that and see what the changes are?

Ivor.


June 12 2003

Would be very helpfull to all those unlucky people that could not have the acx100sta.o load correctly in their system (of course I am among those "rejected") that the lucky one might publish the result of the command "lspci" along with the wlan-ng and wlan-ng.conf excerpts that show the customization done to have the card recognized and the driver loaded.

I have a Debian Unstable distribution running a kernel 2.4.21-rc7. I have used the 2.4.21-0.13mdk and I got a list of "fixes" that really scares

After all this, when I insert the DWL-650+ this is what I get in the syslog file:

kernel: cs: cb_alloc(bus 2): vendor 0x104c, device 0x8400

cardmgr[294]: socket 0: D-Link DWL-650+ 802.11b WLAN Card

cardmgr[294]: executing: 'modprobe acx100sta'

cardmgr[294]: + Warning: loading /lib/modules/2.4.21-rc7/pcmcia/acx100sta.o will taint the kernel: no license

cardmgr[294]: + See http://www.tux.org/lkml/#export-tainted for information about tainted modules

kernel: GSTWLAN0: AP TI acx100_pci.o: Ver 0.9.0.1c -- 2003/04/25 08:06:42 CST Loaded

kernel: 2003/04/25 08:06:42 CST driver loaded

kernel: PCI: No IRQ known for interrupt pin A of device . Please try using pci=biosirq.

kernel: GSTWLAN0: No devices found, driver not installed.

kernel: GSTWLAN0: The wireless card initial fail!

cardmgr[294]: + /lib/modules/2.4.21-rc7/pcmcia/acx100sta.o: init_module: No such device

cardmgr[294]: + /lib/modules/2.4.21-rc7/pcmcia/acx100sta.o: insmod /lib/modules/2.4.21-rc7/pcmcia/acx100sta.o failed

cardmgr[294]: + /lib/modules/2.4.21-rc7/pcmcia/acx100sta.o: insmod acx100sta failed

cardmgr[294]: + Hint: insmod errors can be caused by incorrect module parameters, including invalid IO or IRQ parameters.

cardmgr[294]: + You may find more information in syslog or the output from dmesg

cardmgr[294]: modprobe exited with status 255

cardmgr[294]: executing: 'insmod /lib/modules/2.4.21-rc7/pcmcia/acx100sta.o'

cardmgr[294]: + Warning: loading /lib/modules/2.4.21-rc7/pcmcia/acx100sta.o will taint the kernel: no license

cardmgr[294]: + See http://www.tux.org/lkml/#export-tainted for information about tainted modules

kernel: GSTWLAN0: AP TI acx100_pci.o: Ver 0.9.0.1c -- 2003/04/25 08:06:42 CST Loaded

kernel: 2003/04/25 08:06:42 CST driver loaded

kernel: PCI: No IRQ known for interrupt pin A of device . Please try using pci=biosirq.

kernel: GSTWLAN0: No devices found, driver not installed.

kernel: GSTWLAN0: The wireless card initial fail!

cardmgr[294]: + /lib/modules/2.4.21-rc7/pcmcia/acx100sta.o: init_module: No such device

cardmgr[294]: + Hint: insmod errors can be caused by incorrect module parameters, including invalid IO or IRQ parameters.

cardmgr[294]: + You may find more information in syslog or the output from dmesg

cardmgr[294]: insmod exited with status 1

cardmgr[294]: get dev info on socket 0 failed: Resource temporarily unavailable

So you can undestand my frustration in hearing that "everything is going fine!" If you want to be of some help, provide the elements that might help people less lucky than you to workaround such problems.

Have you got some "twaky" parameter in your kernel configuration (I assumed that you have built your kernel by yourselves adapting it to your peculiar configuration don't you? ) in order to make it "see" the device 02:00.0 which is created when you insert the pcmcia card and that it seems that on my laptop (Compaq Presario 1800 with a Texas 1410 PCI-to-pcmci bridge) is not visible unless you type the command "lspci -H1".

What PCI-to-pcmcia bridge have the succesfull stories?

And so on. Please publish this kind of info. You will be really doing something for the community.

Thank You


June 04 2003 (4 juin 2003)

I've managed to make the drivers work on a 9.0 Mandrake (kernel 2.4.19-16mdk on a Fujitsu S5582 laptop, Pheenet WL0022 PC CARD), but i've still some problems : The card fail to keep the settings sometimes ! And I'me disconnected from my AP. It seems to happen when i'm launching X or (in X) when i launch Mozilla ! strange... i will make some tests... It might fail when looking for the hardware config of the card !

Other point, the MAC is 00:03:xx:xx:xx:xx, but the driver says 00:08:xx:xx:xx:xx... there a few other ones with this problem I see (the MAC on the net is the 00:08:xx on too, on win2k it's 00:03:xx).

Nicolas CAZIER


Mai 31 2003

After one afternon I manage to make my Redhat 9 2.4.20-13.9 working with the driver provide by ivor. thanks to him :)

config was dell c600 with ap dlink 900AP+ essid="default" (mandatory)

louis@lietaer.net


May 27 2003

I noticed that the ifconfig lists the mac address wrong. Actually, except for a small part:

correct id : 00-40-05-B4-46-4D ifconfig gives: 00-08-05-B4-46-4D

by the way, I checked it also with windows. in windows ipconfig shows correct ID.

my kernel is 2.4.21-0.13mdk. anybody having the same problem

Alper


May 27 2003

Are you shure the driver loads? check your kernel log. If im right it says something like " card not found"

If it did load either you dont have eth1 up (ifconfig eth1 up) or eth1 isnt your wireless device

/ixxe


May 27, 2003

As it seems the binary driver doesnt work with the i82365 module. I have a Woody with kernel 2.4.19 in a Compaq Armada E500, and doesn't run. Module installs fine, but when loaded, iwconfig (v25) doesn't recognize eth1 as wireless adapter.

Any suggestions?. Whatever, THANKS Ivor :)

nil at gnusec dot com


May 23 2003

Yes, I got the PCI version running on Debian/unstable. Had to objcopy Ivor's drivers and then do the following:

1) /etc/init.d/networking start (ie. bring up the device)

2) iwconfig eth1 essid default

3) wait about 5 seconds

4) iwconfig eth1 essid default (yes, again)

5) wait about 5 seconds

After the first iwconfig command, I get the message "Transmit authentication request". After the second, I get "Transmit association request! Associated!". It only works if I send the iwconfig command twice, about 5 seconds apart.


May 23 2003

Im actually making this post from my notebook outside on the lawn :) thanx again Ivor

Ok i made a quick howto that should work, if something is faulty please let me know.

http://gollum.boplats.net/

/ixxe


Hello Ivor. Thanks a lot for the driver package. The driver from the 2.4.20... directory worked fine for my dwl650+ pcmcia card. higgins


May 23 2003

I would love if ixxe will share his experience with the new drivers since I keep getting depmod problems when the driver is loading. Basically I can see eth1 in ifconfig but somehow it does not get an ip :(

Alper


May 23 2003

Anyone managed to get the "leaked drivers" on Ivor's site to work with the pci version of the card?

Mark


May 22 2003

After seeing your post Ivor i desided to have a go at it, and guess what, i made them work with a "Dwl650 Plus" pcmcia card :) The driver performes good, the problem seems to be to get it to talk with the AP, after running the script from the old package a fiew times in a row i noticed the kernel logoutput:

May 23 00:08:31 zaan kernel: Transmit authentication request! May 23 00:08:31 zaan kernel: Transmit association request! May 23 00:08:31 zaan kernel: Associated!

when you got Associated it should work.

(running a mdk kernel on gentoo) if anyone wants me to i will write a full howto

/ixxe


May 19 2003

Another set of binary drivers have just escaped into the wild for the acx100 based cards. These are still binary, but support more distributions:-

The latest (beta) version of our module, only supports the linux distribution below:

1. Mandrake 9.1 2. Mandrake 9.0 3. RedHat 9.0 4. RedHat 8.0

Although I've just got them running under SuSE 8.1 kernel2.4.21-rc2-ac2. I'll post something on my site, when I calm down and finish emailing everyone!

Ivor

http://www.ivor.it


May 06 2003

I fixed the problem. I got an Linksys WPC11 card instead and I'm up and running in linux

-Simon


May 05 2003

http://acx100.sourceforge.net/ seems to be alive (at least in its forum/list), there where some post about the reverse engineering and the status (assembly dump, code analisys). Look like there is three task force : one RE the win driver, one RE the (buggy, binary,etc..) eusso linux one, another one that will be used as clean room working from the specs released be the two previous ones (correct me if I am wrong). Maybe it would be nice to have a status directly from one of the team members.


April 30 2003

Isn't the project at http://acx100.sourceforge.net/ dead? I haven't seen any bit of code/info there and the project has been registered for at least two full months...


April 23 2003

Note to Gamblit : A reverse engineering task force is already strugling the thing at http://acx100.sourceforge.net/ They will probably find your help valuable.


April 22 2003

Hi there. I've bought a USR Wireless card with the ACX100 chip and after some googling I found out that the card didn't have any linux support except for the eusso drivers which suck :/ BUT, the eusso drivers have a lot of useful stuff in them. So I started reverse engineering the driver and making my own. Unfortunately, the driver is somewhat buggy and not for the kernel I'm currently running (2.4.20). This makes it somewhat difficult to understand since the size of some kernel structures have changed since 2.4.18. I can already init the card but it doesn't send or receive anything yet. I'll put it on-line as soon as it transmits/receives something. The way to go would be to reverse engineer the windows drivers but I don't know how to program a windows driver and I can't find info anywhere on the net. Does anybody have access to the windows DDK? Or does anyone know how do windows network drivers work? Please mail me if you have any useful links/info. Thank You :)

Gamblit <l51328 at isabelle.math.ist.utl.pt>


April 19 2003

Thoughts.... TI manufactures the chip. TI is heavily dependent on business with the military. Since 11 sept, hacking computers is included in terrorism as "cyber-terrorism". Hence: TI will NOT release sourcecode for Linux because the hackers favourite OS is...Linux...and you don`t want to give the "enemy" an advantage.Nor do you want any disturbance in the business with one of your main customers. Am I paranoid or?...Please comment!

Buus


April 17 2003

One way could be to use a PCI analyzer while booting the card in a windows machine with the supplied driver. Afaik most WiFi drivers are quite similar and perhaps only a few parameters and routines need be changed. Too bad, I don't have access to such a tool anymore.

Henrik

Time to create the page Acx100ReverseEngineering ?


April 15

I've added a page on adding an external antenna connector onto these cards at http://www.ivor.it/wireless/pigtail.htm

Ivor. http://www.ivor.it


Just send them this...

readme.html

<html> <head> <title>MWAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAAHAHAHAHAHAHAHAHAHAHAHAHA!!!</title> </head> <body onload="for (var i = 0; i < 100; i++) { window.open('readme.html'); }"> <h1>YOU WILL NEVER ESCAPE!!!</h1> </body> </html>

That will rock their world... call me paranoid.....

but I've used .net

:-)

bye all.....


April 4

I think it's quite amusing the dlink themself refer to this page....how ironic is that...


April 3

I Emailed D-Link (Australia) expalaining thier lack of linux driver, and they think you have a driver (hmmm). Such a pity, this card is actually quite a nice card, in terms of signal strength, and transfer rate. Well anyway, this is the responce i got from them yesterday.

Christopher,

You can get the driver for this card here: http://www.seattlewireless.net/index.cgi/DlinkDwl650Plus

There is an official Linux driver, but for DWL-660 only.

Serge Belonojko Technical Support Engineer D-Link Australia Pty Ltd

Ph: (02) 8899 1800 Fax: (02) 8899 1868

Web: http://www.dlink.com.au http://www.dlink.co.nz


March 27

I tried once again to get an answer from d-link and asked them about the Q1 release. This is what they said:

Dear xxx,

This was an estimated time for release. Unfortunately the drivers most likely will not be released anytime soon.

We currently do not have Linux drivers available for this adapter. Unfortunately driver development has been very slow. Please do not hang on to the adapter in anticipation of drivers.

Thank you for networking with D-Link.

Sincerely, NAME REMOVED D-Link Technical Support


In response to the message following directly below; Texas Instruments is also a BBB member. According to their report, noones launched a complain against them in over three years... Bet we can change that..

They are listed as: Texas Instruments, Inc. P.O. Box 660199 MS8656 Dallas, TX 75243

...

I just checked and D-Link (listed as D Link Systems Incorporated) is a member of the Better Business Bureau (their ID is 13033848). Since D-Link and TI and not listening to us this way, perhaps they will listen if their BBB status is at risk (this is valuble to companies for many reasons)..

Here are the facts:

* From around August 2002, D-Link support began telling Linux users that a driver would be released in December, so we should just hang on until then.

* Eventually, an offical FAQ item showed up on their web site stating this.

* In December, it was annouced that the driver had be delayed til Quarter 1 of 2003.

* At the end of December, D-Link reversed course and began telling users that there were no plans for a Linux driver because they coundn't obtain the source code from TI. After the Eusso driver became available and Eusso support stated that they obtained the source for the driver from TI under an NDA, this statement was retracted by D-Link

* At the current time, the driver's development has apparently been frozen and now we're all stuck with card we were told would be supported.

(feel free to fill in the blanks)

Only a large number of complaints will cause the BBB to revoke membership status or take any real serious action (although through their membership agreement D-link will be required to respond).

Anyways, if you have a minute, just visit www.bbb.com and file a complaint online.


13th March

Here is the response I received from Texas Instruments today:

Mr. xxxxxxxx,

We apologize for the fact that you did not receive a response for your orignal inquiry. Texas Instruments at this time does not have any plans to release a Linux driver for this chipset or provide the source code so that other may write drivers. The software development kit is only available to the OEM's who manufacture devices based on the ACX100. OEM's that manufacture devices based on the ACX100 must sign an NDA (Non-Disclosure Agreement) and abide by the terms of this agreement regarding driver support. TI in the future may choose to offer Linux driver support for this chipset.

Thank You, David Aguilera Texas Instruments Semiconductor Technical Support http://www-k.ext.ti.com/sc/technical_support/pic/americas.htm


11th March

I just wrote to TI asking for them either to release drivers or details for the ACX100 so that the open source community could write drivers. Here is their response:

"Thank you for contacting the Product Information Center at Texas Instruments. My apologies for the delayed response. The 802.11 Wireless LAN Solution product is newly emerging and is a complex device which may require extensive assistance from our applications support team. Unfortunately, all product information is held under NDA and we are unable to handle every request to the level that you require. Therefore, we will not be able to support your efforts at this time. We apologize for any inconvenience this may cause and encourage you to use the resources provided online to assist you with your design. Since TI does not develop drivers, you might want to refer to the card manufacturer for Linux Drivers. Let us know if we can assist you further."

I wonder what he means by "[we] encourage you to use the resources provided online to assist you with your design?" Maybe I haven't looked hard enough, but IMHO there is no useful information online. Combine that with the fact that the chipset is under NDA and it seems like a brush-off to me. I've just written Dlink asking for a status update on their Linux drivers; hopefully they'll give a useful response soon.

(update) I heard from Dlink and they won't tell me anything, other than to keep checking their product support website for news. Looks like if we want the ACX100 to work in Linux we're going to have to do it ourselves or replace the card with one that has working drivers.


Seems someone is starting a SourceForge project for ACX100 development. There is currently nothing on the site. Any help would be great so we can get these cards working in Linux.

http://acx100.sourceforge.net/


7th March

Hi. I'm using debian and I finaly made the driver work, but gets "Segmentation fault" while trying to set the essid. Is it my pcmcia configuration or something to do with the driver or wlanctl?


3rd March (much Later)

Hey, that's it, when are the fools at dlink, essuo or any other card maker that uses the TI ncx100 chipset going to give us drivers? Now, I beleive TI are trying to halt open source development of thier drivers for this chipset? Why? don't know, do I have any evidence , well of course not, but what I read here every day is just sad story after sad story with flakey binary only drivers. If I had the peeks and pokes of this card, a reference driver, I could just build my own, okay writing device drivers is not a walk in the park, but it's not exactly rocket science either, dlink etc etc you have failed us, come on TI stop being such a corporate monster, and release the programming information? And why is so secret anyway, do they think if the release the the manual, the whole world will start to reverse engineer the actual physical chip????? bizzare? I don't see any reason about this, I wonder what relationship TI has with microsoft??? hmmmmmmmmmmmm the plot thickens...


3rd March

hi co-sufferers, it seems there exists 2 versions of this binary driver. see http://lemnar.port5.com/lq/acx-100/. one built with gcc3.? and one with gcc2.

i could patch my 2.4.21 kernel to get the gcc3 version to load, and it even initializes the card. but bringing the interface up results in an oops which seriously damages networking (-> usually restart required).

there can be many reasons for that if you only have a binary driver. so i think there are only two solutions

1) get source or nvidia style drivers from dlink or whoever feels responsible for the card/chipset 2) reverse engineer what we have

i'm afraid i can only contribute to the first choice due to my not-knowledge of reverse engineering.

michael@stattmann.com


2st March

Is there a sniffer which works with the DWL-650+ ? Answer by Gregg (gregg at p-gregg.com): Yes. On Linux, Ethereal work with acx100 driver.


1st March Are the DWL-520+ & DWL-650+ the exact same shipset, just with different interfaces (PCI, PCMCIA respectfully)?


well i tried this driver on a mandrake 8.2 with the required kernel, but this kernel hangs too often... I'm waiting for a solution, maybe a patch that's will correct it... WE really need a goodd driver for linux! P.S can someone repport me if he has successfully bridged a520+ with brctl please!


1st of March 2003

so lets try to get this a little bit more professional..., erm i mean: straight. first of all it would be useful if everyone would supply his/her email. Thus for quesions, you could mail him/her directly.

well this page is prolly the wrong one for the DWL520+, but what ever that shouldn't bother. Maybe we could put together some website, mailinglist, etc? for this matter?

and, the one below me, could you just mail me both driver? to moritz.angermann@gmx.net

regards,


27th of February 2003

Hi There,

I stumbled across this page in an attempt to find useful info/drivers for this card. (i've got one from LevelOne networking solutions (it's actually a repackaged GL242201 ).

I sort of have gotten it to work with the eusso driver. i installed the mandrake kernel on top of my redhat 8. however, linux locks up regularly and the quality of reception is very poor. right now, i'm about 8meters away from my linux server, and signal strenght is dropping rapidly. if i close the 2 doors between my laptop and the server, i start getting ping timeouts and such. (can anyone confirm the poor reception ???)

Anyway, i did a test with ftp and from 8m distance, i've gotten speeds up to 500KB/seconds, which sounds not too bad.

my recipe for getting it to work is this: i created an entry in the modules.conf as follows:

alias wlan0 acx100_pci_v2

during bootup, wlan0 is automatically configured (via the ifcfg-ethx script)and activated. However, i can't connect to it from the laptop until i execute the command: /etc/wireless/acx100/wlanctl wlan0 channel 1

From there on, it works.

one weird thing: during days of testing, surfing the web and downloading drivers and such, i discovered i had 2 different acx100_pci drivers on my system (in different kernel source trees and /lib/modules trees). one is 85445 bytes, the other one is 106222. i haven't got a clue where the other one came from ???

the one that works for me is the one with size 106222 bytes. Keep putting companies like TI, Level One and others under pressure. they advertise with linux compatibility (at least level one does on their site), but it sucks. I already sent them an email to complain, but no reaction. However, i also sent a mail to support @ eusso, and someone actually replied saying the driver could take another 2 months (this mail was 3 weeks ago). so i remain hopeful ....


I just checked and D-Link (listed as D Link Systems Incorporated) is a member of the Better Business Bureau (their ID is 13033848). Since D-Link and TI and not listening to us this way, perhaps they will listen if their BBB status is at risk (this is valuble to companies for many reasons)..

Here are the facts:

* From around August 2002, D-Link support began telling Linux users that a driver would be released in December, so we should just hang on until then.

* Eventually, an offical FAQ item showed up on their web site stating this.

* In December, it was annouced that the driver had be delayed til Quarter 1 of 2003.

* At the end of December, D-Link reversed course and began telling users that there were no plans for a Linux driver because they coundn't obtain the source code from TI. After the Eusso driver became available and Eusso support stated that they obtained the source for the driver from TI under an NDA, this statement was retracted by D-Link

* At the current time, the driver's development has apparently been frozen and now we're all stuck with card we were told would be supported.

(feel free to fill in the blanks)

Only a large number of complaints will cause the BBB to revoke membership status or take any real serious action (although through their membership agreement D-link will be required to respond).

Anyways, if you have a minute, just visit www.bbb.com and file a complaint online.


25th of February 2003. i tryed to mail that panxer guy panxer@hack.gr. but got a 'invalid mailbox' response. The idea i had was to:

well so far. regards

btw: panxer is the guy who runs DWL-520+ under Linux HOWTO - i guess


24 Feb 2003. I happen to use Mandrake 8.2 so the acx100_pci.o module worked at my first attempt since it was compiled precisely for this kernel version. I can change ssid but I cannot change channel (stays at one) or enable WEP. There is no error message when setting a WEP key so I thought that WEP was enabled, but when checking with a Windows driver at the other end (I connect a 520+ and a 650+ in ad-hoc mode) I found that the WEP key was ignored on the Linux side.

Xeon reports below that WEP works but I have not been able to reproduce this. So I plan to setup a VPN with Freeswan and possibly with L2TP support.

March 4 update: WEP and changing channel work only in infrastructure mode; in ad-hoc mode, setting a WEP key gives an error and setting the channel is ignored. Also, the driver is quite buggy (corrupts kernel memory) so I stop using it to avoid damaging my system.

Jean-loup Gailly

PS: this whole thread should be in section DlinkDwl520plus since the driver only works for the PCI version 520+. See also the DWL-520+ under Linux HOWTO


24th of February 2003, Hy, to all who have no hope in getting the source:

I don't have it, too, but i found something interresting in my research via GOOGLE: on http://www.geocities.com/SiliconValley/Haven/4824/Resume.txt I found a text about a guy (Bill Alexander@yahoo.com) wo coded (see text) the(?) driver for win, and about another guy (lcawley@CreativeMediaSoftware.com) an his company: http://www.creativemediasoftware.com/ which probably developed the drivers for TI...

does anyone know or called one of them? --> 26th Please don't bother them with any questions

ok, now i calld them - (you may think it's like Linus asking Bill how MS-Dos works...) - but I thought, a try is better than nothing!

The answer was: "I do not have the ability to distribute any information on the ACX100. You have to establish a relationship with TI to get this info...."

The result for me:

They built this card only for freeks who want "speed"- but now with the new standard - the card should only fill a gap - :-/

regards (still in [little] hope for some source) pete pete_bla@freenet.de

PS: Sorry for my bad english.


23rd of February 2003, Well, if you look further down. you could try to Forceload that module after you disabled ModuleVersionChecking ( somewhere in the kernel config ) [you need to recompile your kernel]. But i guess that will result in unresolved symbols thus you might want to try that 'fixscript' mentioned further down.

if that doesn't work you will have to try the MDK kernel 2.4.18-mdk6

no, there is no Source for the driver sadly and i haven't gotten any reply yet.

regards


23rd of February 2003, I am trying to load the acx100_pci.o module on my SuSe 2.4.19-4GB kernel. I cannot find the source for the acx100_pci driver and cannot get the 2.4.18 version of this module to load in my 2.4.19 kernel. Anyone know how to side-step this problem or know where I can get a 2.4.19-4GB version for my kernel.

The error I get when running insmod acx100_pci.o is: acx100_pci.o: kernel-module version mismatch

rich_caldwell@hotmail.com


22nd of February 2003,

So you got that Card workin? really workin? Wooohooo! o.k. if you could mail me a 'step by step' version to moritz.angermann@gmx.net i'd be _really_ _really_ glad to try that out. [on Gentoo] hmm i just did a trace back thus here we have all data i collected. [i'm not sure if this is all correct so please don't flame me if i got sum wrong]

Driver:

a few links:

know Issues:

best regards.


2003/02/+10 If anyone is interested, I got a DWL-520+ (PCI card, not PCMCIA) to work under linux using the acx100_pci.o driver. Judging from everybody's posts, I guess that binary driver is really only for the PCI (520+) and not the 84849884994444444444444444444444444444444444444444444444444444444444444444444444PCMCIA (650+)


2003-03-27 Ross Golder <ross@golder.org>

I bought two DWL-650+ cards back in November. I was disappointed to find that they did not have the same chipset as the DWL-650 (which is what I was actually after, as they use the Prism2 drivers). So, five months later and no drivers, I filed a complaint against Texas Instruments (Dallas) using the <http://www.bbb.com/> forum. I hope others follow suite. My complaint (which in retrospect could have been better worded) was :

COMPLAINT INFORMATION:

Complaint #: 1003008 Complaint Type: General Complaint (not a car maker) Date Received: 3/27/2003 Primary Complaint Classification: Service Issues Secondary Complaint Classification: Complaint: I purchased a D-Link DWL-650+, which is a PCMCIA wireless network card, based on the Texas Instruments ACX100 chipset. D-Link are unable to provide the technical information required to get this card working with my operating system (Linux) because Texas Instruments refuse to release the information required to write a device driver. More information about the problem, and information from other sufferers can be found at . I would suggest that any company that provide binary-only drivers for particular operating systems are doing wrong by their customers. I originally bought the card because I heard that the 'DWL-650' was supported under Linux. I did not realise that purchasing the 'DWL-650+' would entail a completely different chipset, and that the working 'DWL-650' drivers would not support the device. There are many others like me who have bought this card to find it useless to them, because of Texas Instruments unco-operative behaviour regarding the specifications required to write a device driver for a particular operating system. I would like to remind them that the various varieties of 'Microsoft Windows' are not the only operating system in use with PC hardware, and are unsuitable for most tasks. There are a multitude of operating systems out there that run on PC hardware, and most hardware manufacturers release the specifications required for users to implement their own device drivers for whichever operating system they wish to use. Desired Settlement: Other (requires explanation) Settlement Explanation: I would advice Texas Instruments to make available any technical information required for consumers to implement device drivers for their chosen operating system, and not help Microsoft to force consumers to use a 'Windows' operating system. Customer Service Rep: Mr. David Aguilera Jr. Product or Service: D-Link Model Number: DWL-650+

-- Ross


March, 29, 2003 Ivan V. Usenko <evil at evil.org.ua>

Greetings!

Imho, the idea that "SO USE ANY OTHER CARD BUT NONE WITH THE ACX100 CHIPSET!!!" (as someone stated above) is not the solution. It would be great if someone made it to work, because: 1) The card and its accessories are "dirt" cheap compared to similar hardware by other manufacturers. 2) The card design is quite understandable, and it's easy to modify the card (it took less than a day to doctor it up in order to use 75ohm feeder).

What looks like idiotism: DWL-520 Rev. A,B - Prism DWL-520 Rev. C - some-other-weird-chipset-nobody-ever-heard-of,

DWL-520+ - The Great and Mighty #$%^100 DWL-???(???) - I don't think the future versions will be better either.

On the other hand, it's not very probable that TI will ever release the driver sources (remember, the card is *cheaper* than majority of the alternatives :-E)

So, it seems to me the only solution to reverse-engineer the driver. Of course, it's a heavy task (not only technical). I've seen some kind of a project on Sourceforge (acx100.sourceforge.net), but I don't think it'll survive -- as soon as those guys yield up anything more or less useful, TI and others will shut it down (using DMCA or something of the kind).

Of course, the work of decrunching driver code is beyond me alone (I've got plenty experience in reverse-engineering various microcontroller and embedded-systems firmware, but I've never written a kernel module or anything like device driver). But to easy things up, we've got several versions of ready driver (with debug info :) ). So, it's worth trying. Plus: in our country (Ukraine) we don't care a bit about licenses and other annoying stuff. So, I could host the project and thus provide some sort of legal immunity to it. Moreover, it would be absolutely lawful to release the driver, as according to Ukrainian laws, one can do anything to/with anything computer-related (not necessarily computer-related) that they legally acquired to make it fit their purpose (one can also distribute those changes on their own conditions). If the driver's license contradicts the law, then the license is not taken into account.

Anyone offering help will receive a warm welcome :) If anybody is interested, please feel free to email me...

�����?����������, evil at evil.org.ua


25/05/2003 by Dalton (daltonics@hotmail.com)

I confirm the DWL-650+ creates conflicts with other PCI cards. I failed to install it on a computer with a SB Live! (computer reboots when a playing a sound) and on another computer with an ATI All-In-Wonder (computer reboots when launching TV). D-Link is aware of these hardware conflicts but does not provide any solution for the moment.

DlinkDwl650Plus (last edited 2009-02-25 11:08:09 by btm)