If you want to have WiFi cheaply running on Linux try this solution

[The Linuxator]

Update : n0cmonkey found this recommended website for these type of adapters take a look.

After noticing in this thread how easily a Linux system can run WiFi utilizing an Intel mini-PCI g or b mini-PCI card ( Due to the fact that intel has supported a project for making drivers for Linux) which are avaiable all over the place for cheap like here'. for laptops, SFF and desktops.

Ppl might be saying well how is this thing going to run on a desktop well grab a mini-PCI PCI adapter and you should be set to go.

Note the only issue is that this mini-PCI PCI adapter that I have found googling around is't something that I was able to find anywhere that I usally shop around online, so if somone can point out a place for buying these badboys the problem should be solved right.

So what about SFF ? Well I have noticed that SFF such as this already have a mini-PCI slots available, also if yours doesn't have any then if the PCI adapter is possible then you can use the PCI ADAPTER card in there.

Your input is much appreciated.

 

[n0cmonkey]

RALink has some mini-pci adapters out there:

Amigo AWI-922W
Billionton MIWLGRL
Gigabyte GN-WIKG
MSI MP54G2
MSI MS-6833
Tonze PC-620C
Zinwell ZWX-G360

RALink is more F/OSS friendly than Intel. 😉

 

[The Linuxator]

Originally posted by: n0cmonkey
RALink has some mini-pci adapters out there:

Amigo AWI-922W
Billionton MIWLGRL
Gigabyte GN-WIKG
MSI MP54G2
MSI MS-6833
Tonze PC-620C
Zinwell ZWX-G360

RALink is more F/OSS friendly than Intel. 😉

Well I am talking from personal experience and lets face it the intel ones are thee most widespread among all of those due to the fact that most laptops are going to come with an intel mini-PCI of some sort, and the way the intel 2200 BG worked out for me yesterday I wouldn't recommend anything else I haven't tried yet. but it would be great if I can find any mini-PCI pci adapter so I can grab an extra 2200 BG for my HTPC.

 

[n0cmonkey]

Originally posted by: The Linuxator

Originally posted by: n0cmonkey
RALink has some mini-pci adapters out there:

Amigo AWI-922W
Billionton MIWLGRL
Gigabyte GN-WIKG
MSI MP54G2
MSI MS-6833
Tonze PC-620C
Zinwell ZWX-G360

RALink is more F/OSS friendly than Intel. 😉

Well I am talking from personal experience and lets face it the intel ones are thee most widespread among all of those due to the fact that most laptops are going to come with an intel mini-PCI of some sort, and the way the intel 2200 BG worked out for me yesterday I wouldn't recommend anything else I haven't tried yet. but it would be great if I can find any mini-PCI pci adapter so I can grab an extra 2200 BG for my HTPC.

Of course, recommend what you know. I was just offering an alternative. I think I even found an RALink based adapter (the name doesn't match up exactly to my list, but the chip in the pic has the RALink info). MSI MP54G4 -$20.50. The 2200BG is $27.99 at newegg. 😉

Expect RealTek and RALink to gain more ground in the US as companies see that they are cheaper. Of course, don't expect them in centrino laptops. 😉

Check the other thread, I left a link for a PCI to MiniPCI adapter there. 😉

 

[The Linuxator]

Does RAlink the chipset manfacturer release any drivers for Linux like Intel does or do they just release info concerning the chipset type , so that Linux developers can write drivers for them ?

 

[n0cmonkey]

Originally posted by: The Linuxator
Does RAlink the chipset manfacturer release any drivers for Linux like Intel does or do they just release info concerning the chipset type , so that Linux developers can write drivers for them ?

I'm not sure. There are independant projects for RALink drivers, and I know they got documentation (or atleast OpenBSD developers did).

RALink also provides drivers on their website for Windows, Linux, and Mac OS X (rare!). I'm not sure whether they wrote the drivers or if they got their driver from the independant group.

Who developped the code? Don't know, but it's all GPLed. 😉

The code in the BSDs is of course, not GPLed, but falls under a BSD license. But everyone seems to be happy. :beer:

 

[The Linuxator]

Is Intel's code GPLed ? I don't know but I , I don't think they would do that.
Either way if it works I don't have an issue weither it's GPLed or not just as long as it's stable I am fine.

 

[n0cmonkey]

Originally posted by: The Linuxator
Is Intel's code GPLed ? I don't know but I , I don't think they would do that.
Either way if it works I don't have an issue weither it's GPLed or not just as long as it's stable I am fine.

Intel's code is GPLed too, but the devs are still missing the documentation to go with it.

I care about freedom. If the software isn't free, it takes a lot for me to use it. 😉

 

[The Linuxator]

Originally posted by: n0cmonkey

Originally posted by: The Linuxator
Is Intel's code GPLed ? I don't know but I , I don't think they would do that.
Either way if it works I don't have an issue weither it's GPLed or not just as long as it's stable I am fine.

Intel's code is GPLed too, but the devs are still missing the documentation to go with it.

I care about freedom. If the software isn't free, it takes a lot for me to use it. 😉

I agree 100 % but what I meant was that to have driver that works is better than having nothing at all :Q so GPLing the code is one step to freedom.

 

[drag]

Originally posted by: n0cmonkey

Originally posted by: The Linuxator
Does RAlink the chipset manfacturer release any drivers for Linux like Intel does or do they just release info concerning the chipset type , so that Linux developers can write drivers for them ?

I'm not sure. There are independant projects for RALink drivers, and I know they got documentation (or atleast OpenBSD developers did).

RALink also provides drivers on their website for Windows, Linux, and Mac OS X (rare!). I'm not sure whether they wrote the drivers or if they got their driver from the independant group.

Who developped the code? Don't know, but it's all GPLed. 😉

The code in the BSDs is of course, not GPLed, but falls under a BSD license. But everyone seems to be happy. :beer:

Ralink released some documentation and they wrote their own GPL'd drivers after having them under closed source licenses for a bit. The BSD's wrote their own drivers.. There were numerious maintainance and endin (spl?) issues with the Ralink drivers... One of the major problems with hardware developers like that is that they only realy have the resources to quickly release drivers 'that work' and move onto the next project. Since the drivers have been released the linux devs developed a generic/universal 802.11x protocol stack for the Linux kernel. There is a ongoing project to rewrite the Ralink drivers to use this stack and make a universal driver for all of Ralink 802.11 cards (rt2400 (802.11b), rt2500 (802.11g mini-pci, pci, pcmcia (I think) ), rt2570 (802.11g USB). This driver is refered to as "rt2x00 beta', it's had it first release just recently, it's only realy able to load the module and do some other basic stuff.. it's not able to realy form a network yet.

However the rt2400, rt2500, and rt2570 drivers work just fine for x86 arch and have had numerious bug fixes since their original release by Ralink. They aren't able to do 'Master' mode so won't be able to run as a wireless access point under Linux, but can do ad-hoc and managed modes just fine for per to per and client wireless access just fine.

For other-then-x86 a temporary solution for them is the ural-linux driver and is a port from FreeBSD for USB devices. I use it for my ibook right now. Works so-so. Borrows a lot of code from mad-wifi stuff, I beleive.

*bsd driver:
http://damien.bergamini.free.fr/ral/

Linux ural port:
http://damien.bergamini.free.fr/ral/

rt2x00 project:
http://rt2x00.serialmonkey.com/

rt2400 and rt2500 tested configurations:
http://rt2x00.serialmonkey.com/wiki/index.php/Hardware

attempt at a list of rt2500 devices:
http://ralink.rapla.net/

Be aware that with many hardware manufacturers that they use different chipsets with same name cards with different revisions.. So 'Namebrand Card A' (ver 1) may work, but 'Namebrand Card A' (ver 1.1) may be completely different card.

Also with Ralink cards there is no need for firmware or binary blob stuff, and they are VERY inexpensive.

Intel cards may be more suitable for many situations. There is the issue those nasty card revisions you have to watch out for. (I've returned cards before that had the wrong chipset) Also they won't be completely stable until the rt2x00 beta driver is finished. Also, although I don't know this, the intel card may support Master mode, which is very handy for setting up wireless routers. Then again the USB ralink stuff is pretty handy, too.

 

[The Linuxator]

Why would a manfacturer be reluctant to make their drivers open sourced ? I mean wouldn't making them open sourced under the GPL allow ppl to make enhancements to it and therefore making there hardware more wanted ? Or are they reluctant because exposing their drivers would expose certain flaws or technologies that they don't want the competetion to know about ?

 

[n0cmonkey]

Originally posted by: The Linuxator
Or are they reluctant because exposing their drivers would expose certain flaws or technologies that they don't want the competetion to know about ?

I think that's the reason. They don't abstract things well enough to allow for open drivers that don't give the good stuff away. Or they're stuck in the old paradigm, thinking that these programming interfaces are a great secret to have.

 

[n0cmonkey]

Originally posted by: drag
Ralink released some documentation and they wrote their own GPL'd drivers after having them under closed source licenses for a bit. The BSD's wrote their own drivers.. There were numerious maintainance and endin (spl?) issues with the Ralink drivers... One of the major problems with hardware developers like that is that they only realy have the resources to quickly release drivers 'that work' and move onto the next project. Since the drivers have been released the linux devs developed a generic/universal 802.11x protocol stack for the Linux kernel. There is a ongoing project to rewrite the Ralink drivers to use this stack and make a universal driver for all of Ralink 802.11 cards (rt2400 (802.11b), rt2500 (802.11g mini-pci, pci, pcmcia (I think) ), rt2570 (802.11g USB). This driver is refered to as "rt2x00 beta', it's had it first release just recently, it's only realy able to load the module and do some other basic stuff.. it's not able to realy form a network yet.

However the rt2400, rt2500, and rt2570 drivers work just fine for x86 arch and have had numerious bug fixes since their original release by Ralink. They aren't able to do 'Master' mode so won't be able to run as a wireless access point under Linux, but can do ad-hoc and managed modes just fine for per to per and client wireless access just fine.

For other-then-x86 a temporary solution for them is the ural-linux driver and is a port from FreeBSD for USB devices. I use it for my ibook right now. Works so-so. Borrows a lot of code from mad-wifi stuff, I beleive.

*bsd driver:
http://damien.bergamini.free.fr/ral/

Linux ural port:
http://damien.bergamini.free.fr/ral/

rt2x00 project:
http://rt2x00.serialmonkey.com/

rt2400 and rt2500 tested configurations:
http://rt2x00.serialmonkey.com/wiki/index.php/Hardware

attempt at a list of rt2500 devices:
http://ralink.rapla.net/

Be aware that with many hardware manufacturers that they use different chipsets with same name cards with different revisions.. So 'Namebrand Card A' (ver 1) may work, but 'Namebrand Card A' (ver 1.1) may be completely different card.

Also with Ralink cards there is no need for firmware or binary blob stuff, and they are VERY inexpensive.

Intel cards may be more suitable for many situations. There is the issue those nasty card revisions you have to watch out for. (I've returned cards before that had the wrong chipset) Also they won't be completely stable until the rt2x00 beta driver is finished. Also, although I don't know this, the intel card may support Master mode, which is very handy for setting up wireless routers. Then again the USB ralink stuff is pretty handy, too.

Heh, I figured the Linux drivers would be a bit more advanced than that. 😛

 

[drag]

Well there are a few reasons.

One reason is that they may license code, patents, or other technology from third parties. If they pay to use closed source software it doesn't usually give them the right to release it themselves. For instance Nvidia is probably at least partially that way, they have a very high-quality OpenGL implimentation and their cards are designed to use it. They've gotten a lot of engineers that originally worked for SGI at least, and I wouldn't be suprised if they liscenced code also. It goes to show that closed source software can screw people both ways in some ways.

Another reason is fear of exposing hardware. Although usually not true, people may be paraniod that telling people how to use hardware will allow them to figure out how the hardware works and replicate it easier, or at least make reverse-engineering easier. Also it may show off flaws in the hardware or show BS claims.. like something a company claims is implimented in hardware is realy done, or partially done, in the cpu by the drivers.

Although rare, but with wifi cards specificly there are legal restrictions. Since they are radio devices they are covered under the FCC and if it's easy for the end-user to use the cards outside restrictive frequencies and power outputs the FCC will require that the enduser must obtain a license to use the devices before they can legally use it. Obviously this would have a chilling effect on sales. 😉 Right now people either use restrictions built into the hardware, firmware, or software (drivers).

If you don't understand what 'firmware' is.. it's also called 'microcode' sometimes. Older and more expensive cards have on-board proccessors. For instance the earlier Prism54 cards have built-in embedded proccessors. These are custom made asic chips that are designed to provide the fequencies or whatnot. The firmware is the actual code that these proccessors are desgined to run. So firmware wouldn't be run in the drivers.. the drivers would be what the kernel uses to control the hardware.. but the firmware would run on the actual card/device itself. Originally this firmware would be stored onboard the device in Flash or CMOS ROM.

In order to save costs hardware makers began eliminating these ROM chips and using features in the OS itself to 'load' the firmware into memory. (note that this originally caused big problems with Linux and I've noticed that devices that used loadable firmware via the driver used to be the most instable devices in Windows)

Nowadays makers are even eliminating the ASIC proccessor. Now they are emulating the proccessor on the central cpu.. So it's a bit of a mess that you load the firmware on a virtual proccessor.


Other reasons that people don't release drivers is thru simple ignorance. Many times you have small businesses with only a handfull of programmers and they don't realy understand how OSS stuff works. Or they don't think there is much of a demand with it. Or they don't understand that once they put the effort and work into getting the driver into the kernel then the kernel developers take responsability for it and it reduces the cost and effort of maintaining a driver by a lot. Many makers have small budgets and small amount of programmers and they idea of making 'proper' code that they are willing to expose the public is a bit intimidating. Often the code they produce is shoddy and isn't very good and fixing it may cost to much or something like that.

Other reason is simple hostility. They don't like the idea of free software and they beleive in the Microsoft-style idea of 'intellectual property'. A lot of them make lots of money from closed source software and Microsoft Windows, they are big fans of it.

9 times out of 10 it's a big win though to GPL code. Especially if you can get it into the kernel then it works by default for all future revisions of the kernel. Then all you need to do is test 'rc' kernels with your hardware and whatever 'certified' configurations you want to support directly or whatnot. Also kernel devs offer generic features like the 802.11x protocol stack or filesystem journalling features for driver makers to use that would help to reduce code complexity and cost.

 

[drag]

Originally posted by: n0cmonkey

Originally posted by: drag
Ralink released some documentation and they wrote their own GPL'd drivers after having them under closed source licenses for a bit. The BSD's wrote their own drivers.. There were numerious maintainance and endin (spl?) issues with the Ralink drivers... One of the major problems with hardware developers like that is that they only realy have the resources to quickly release drivers 'that work' and move onto the next project. Since the drivers have been released the linux devs developed a generic/universal 802.11x protocol stack for the Linux kernel. There is a ongoing project to rewrite the Ralink drivers to use this stack and make a universal driver for all of Ralink 802.11 cards (rt2400 (802.11b), rt2500 (802.11g mini-pci, pci, pcmcia (I think) ), rt2570 (802.11g USB). This driver is refered to as "rt2x00 beta', it's had it first release just recently, it's only realy able to load the module and do some other basic stuff.. it's not able to realy form a network yet.

However the rt2400, rt2500, and rt2570 drivers work just fine for x86 arch and have had numerious bug fixes since their original release by Ralink. They aren't able to do 'Master' mode so won't be able to run as a wireless access point under Linux, but can do ad-hoc and managed modes just fine for per to per and client wireless access just fine.

For other-then-x86 a temporary solution for them is the ural-linux driver and is a port from FreeBSD for USB devices. I use it for my ibook right now. Works so-so. Borrows a lot of code from mad-wifi stuff, I beleive.

*bsd driver:
http://damien.bergamini.free.fr/ral/

Linux ural port:
http://damien.bergamini.free.fr/ral/

rt2x00 project:
http://rt2x00.serialmonkey.com/

rt2400 and rt2500 tested configurations:
http://rt2x00.serialmonkey.com/wiki/index.php/Hardware

attempt at a list of rt2500 devices:
http://ralink.rapla.net/

Be aware that with many hardware manufacturers that they use different chipsets with same name cards with different revisions.. So 'Namebrand Card A' (ver 1) may work, but 'Namebrand Card A' (ver 1.1) may be completely different card.

Also with Ralink cards there is no need for firmware or binary blob stuff, and they are VERY inexpensive.

Intel cards may be more suitable for many situations. There is the issue those nasty card revisions you have to watch out for. (I've returned cards before that had the wrong chipset) Also they won't be completely stable until the rt2x00 beta driver is finished. Also, although I don't know this, the intel card may support Master mode, which is very handy for setting up wireless routers. Then again the USB ralink stuff is pretty handy, too.

Heh, I figured the Linux drivers would be a bit more advanced than that. 😛

*shrug*.

I don't like how the rt2x00 devs feel the need to do rewrites like that. Especially from scratch... the 'rt2x00 beta' is the second effort they've made to rewrite it. But at least they maintain the rt2400, rt2500, and rt2570 drivers and do bug fixes.

 

[bersl2]

Originally posted by: drag

Originally posted by: n0cmonkey

Originally posted by: drag
Ralink released some documentation and they wrote their own GPL'd drivers after having them under closed source licenses for a bit. The BSD's wrote their own drivers.. There were numerious maintainance and endin (spl?) issues with the Ralink drivers... One of the major problems with hardware developers like that is that they only realy have the resources to quickly release drivers 'that work' and move onto the next project. Since the drivers have been released the linux devs developed a generic/universal 802.11x protocol stack for the Linux kernel. There is a ongoing project to rewrite the Ralink drivers to use this stack and make a universal driver for all of Ralink 802.11 cards (rt2400 (802.11b), rt2500 (802.11g mini-pci, pci, pcmcia (I think) ), rt2570 (802.11g USB). This driver is refered to as "rt2x00 beta', it's had it first release just recently, it's only realy able to load the module and do some other basic stuff.. it's not able to realy form a network yet.

However the rt2400, rt2500, and rt2570 drivers work just fine for x86 arch and have had numerious bug fixes since their original release by Ralink. They aren't able to do 'Master' mode so won't be able to run as a wireless access point under Linux, but can do ad-hoc and managed modes just fine for per to per and client wireless access just fine.

For other-then-x86 a temporary solution for them is the ural-linux driver and is a port from FreeBSD for USB devices. I use it for my ibook right now. Works so-so. Borrows a lot of code from mad-wifi stuff, I beleive.

*bsd driver:
http://damien.bergamini.free.fr/ral/

Linux ural port:
http://damien.bergamini.free.fr/ral/

rt2x00 project:
http://rt2x00.serialmonkey.com/

rt2400 and rt2500 tested configurations:
http://rt2x00.serialmonkey.com/wiki/index.php/Hardware

attempt at a list of rt2500 devices:
http://ralink.rapla.net/

Be aware that with many hardware manufacturers that they use different chipsets with same name cards with different revisions.. So 'Namebrand Card A' (ver 1) may work, but 'Namebrand Card A' (ver 1.1) may be completely different card.

Also with Ralink cards there is no need for firmware or binary blob stuff, and they are VERY inexpensive.

Intel cards may be more suitable for many situations. There is the issue those nasty card revisions you have to watch out for. (I've returned cards before that had the wrong chipset) Also they won't be completely stable until the rt2x00 beta driver is finished. Also, although I don't know this, the intel card may support Master mode, which is very handy for setting up wireless routers. Then again the USB ralink stuff is pretty handy, too.

Heh, I figured the Linux drivers would be a bit more advanced than that. 😛

*shrug*.

I don't like how the rt2x00 devs feel the need to do rewrites like that. Especially from scratch... the 'rt2x00 beta' is the second effort they've made to rewrite it. But at least they maintain the rt2400, rt2500, and rt2570 drivers and do bug fixes.

The reason they want to rewrite the drivers in the first place is that the current rt2[45]00 drivers are more-or-less ports of the NDIS ones IIRC---even if this is not true, the code and design doesn't feel Unix-esque to me. The reason they started a second rewrite is to take advantage of a common 802.11 stack that other wireless driver projects seem to have standardized on. I believe the goal is to merge this common 802.11 stack into the Kernel at some point, along with the individual drivers that use it.

Also, there is the desire to unify the driver for all devices. The rt24xx and rt25xx can't be that different under the hood...

 

[drag]

Makes sense. I know that the original ralink drivers had next to no chance to getting into the kernel. Also I like having a standard 802.11 stack, that's goodness.

It's a more instinctive adversion to anything 'rewrite from scratch' that I have. 🙂

 

[The Linuxator]

Hey the only problem tha I have faced so far with my Fedora Core 4 installtion on my thinkpad with the Intel 2200 BG is that I am able to connect to networks like mine that is running WEP 128( with MAC adress filtering) but my colleges WiFI network runs a specefice cisco protocol called LEAP I can only connect to them through windows ( by intel wireless utilities) I am not able to configure my card under fedora for that purpose any help would be appreciated. 😉

 

[drag]

Originally posted by: The Linuxator
Hey the only problem tha I have faced so far with my Fedora Core 4 installtion on my thinkpad with the Intel 2200 BG is that I am able to connect to networks like mine that is running WEP 128( with MAC adress filtering) but my colleges WiFI network runs a specefice cisco protocol called LEAP I can only connect to them through windows ( by intel wireless utilities) I am not able to configure my card under fedora for that purpose any help would be appreciated. 😉

Alright.. Don't know much about it but some quick google searches sent me on the right track.

LEAP is a cisco-specific implimentation of EAP from the 802.1x protocol.

802.1x is a 2nd layer (data-link) way to authenticate users. It's one of those things were your required to supply a password and username in order to access the wireless link. It would be made up of three parts.. a Authenticator that takes care of the authentication on the network, a Supplicant that is the program that allows you to log into the network and then a radius server to keep track of all the authentication information and such.

There is a client (or supplicant) that claims to be compatable with LEAP and other 802.1x stuff. It's called xsupplicant and it's from the Open1x project. They have a sample script for LEAP networks.

PEAP is the EAP version that Microsoft uses, I beleive.

I don't know if it needs any special support in the hardware or drivers or anything like that. But I think it needs it. You might want to check out Intel's driver website about that.

Here is the docs for xsupplicant.
http://sourceforge.net/docman/display_doc.php?docid=24379&group_id=60236

Another supplicant I've run across.
http://hostap.epitest.fi/wpa_supplicant/

 

[n0cmonkey]

Originally posted by: drag
Well there are a few reasons.

One reason is that they may license code, patents, or other technology from third parties. If they pay to use closed source software it doesn't usually give them the right to release it themselves. For instance Nvidia is probably at least partially that way, they have a very high-quality OpenGL implimentation and their cards are designed to use it. They've gotten a lot of engineers that originally worked for SGI at least, and I wouldn't be suprised if they liscenced code also. It goes to show that closed source software can screw people both ways in some ways.

Another reason is fear of exposing hardware. Although usually not true, people may be paraniod that telling people how to use hardware will allow them to figure out how the hardware works and replicate it easier, or at least make reverse-engineering easier. Also it may show off flaws in the hardware or show BS claims.. like something a company claims is implimented in hardware is realy done, or partially done, in the cpu by the drivers.

All of this can be handled with an abstracted programming interface. 😉

Although rare, but with wifi cards specificly there are legal restrictions. Since they are radio devices they are covered under the FCC and if it's easy for the end-user to use the cards outside restrictive frequencies and power outputs the FCC will require that the enduser must obtain a license to use the devices before they can legally use it. Obviously this would have a chilling effect on sales. 😉 Right now people either use restrictions built into the hardware, firmware, or software (drivers).

The firmware or hardware should handle this, it doesn't belong in a driver. And hell, the FCC is a tyrant, the airwaves are no longer free. Unless you have a metric buttload of money. 🙁

 

[n0cmonkey]

Originally posted by: bersl2
Also, there is the desire to unify the driver for all devices. The rt24xx and rt25xx can't be that different under the hood...

OpenBSD has one driver for all the ral based cards. Except the USB based ones. Apparently they wanted to squeek ural support in _right_ before they sent the master cd to get pressed, and didn't want to possibly compromise the integrity of the ral driver. IIRC though, they said it was all basically the same. 😛