Slashdot Mirror
Explaining WLAN Chips' Poor Linux Support
morcheeba writes "Kernel Traffic is reporting (mirror mirror list) that 'Some WLAN Chip Specs Secret To Protect Military Communications.' While this is stretching it a bit -- these radios are generally limited to a narrow frequency range and few modulation types -- software can cause illegal radio operation, especially when the laws vary by country. Is Linux support for 802.11g and Centrino chipsets going to be delayed by manufacturers afraid of FCC harassment? An interesting discussion on the future of Openness in radio chipsets." Interesting comments from Alan Cox in here about just how flexible some of these chips are.
And this won't be an issue
What people need to realize is that nothing in software can be secure. It is far to easy to crack. You may have 100 developers working on a secure WiFi driver, but there are thousands of bored hackers out there waiting to tear it apart. If something needs to be secure, do it in hardware.
On a side note, I've not had any trouble getting my WiFi hardware to work on my slackware laptop, but I understand that some chipsets can be more difficult to setup than others.
I would expect such blatant racism on Fark, but on Slashdot? Mods please ban this asshole.
Yeah - that reminds me of the wifi combo card Im using - its a prototype card that I've had for about 3 months that does A, B, and G. The software limits radio strength normally based on the country that you're in due to FCC regulations. However, the developer software allows you to set the power transmit - something thats actually illegal according to the FCC (the consumer is not allowed to sett the power level of the card).
why *should* card companies spend money to write drivers for linux when there are going to be enthusiasts that do it for them?
It makes no sense from a company viewpoint, other than to get the support from the 5% of people that are using thier cards with linux. But even so, if an independent driver is written, those people using linux are still going to be buying the company's cards with no out of pocket expense to the company.
It's kinda messed up.
Though this whole thing seems silly to me anyways. Why don't the broadband chipset makers just provide the specs on how to load a binary firmware. I'm using it right now in order to use my cheap USB<-Serial adapter, I'm certain thatproviding a binary firmware isn't going to upset anyone other than the most hardcore GPL only people.
Marxism is the opiate of dumbasses
Atmel released an open source driver that can be found here.
I have been buying 802.11 cards like crazy trying to find one that will work well with Linux for mobile self-propelled linux boxes.
WHat I have found is that almost all the new cards out there, brand name, no name, etc, are based on the broadcom chip, for which there is no driver.
Now it's true that there are wierd FCC rules, such as the one that the antenna connectors must be proprietary, as if that makes a difference, but that doesn't explain certain things.
For instance, the SMC, Siemens, and Linksys cards all USED to work. The new cards from these vendors, such as the Linksys WPC11, don't work, but have the SAME MODEL NUMBER even though they are entirely different cards. They all give the same codes or similar when inserted, they all have similar antenna shapes, they all have two dimples in the bottom of the antenna where one dimple has a bump from the injection molder.
Now, the only difference I can see on the BOX between the old and the new ones is that the new ones mention Windows XP. So, can it be that MS would only support chipsets with proprietary specs? It sure looks that way. I really can't understand why multiple vendors would completely change the card and keep the same model number. This makes no sense. I think it's as simple as not supporting linux compatible hardware in each release of windows. Not so far fetched, how many product boxes do you see that even MENTION linux? You're not gonna get that 'designed for windows XP' logo if you don't do as you're told.
Now, the older cards work just fine, I have a prism card and it's great. Problem is I only have one, which serves no purpose at all. ALso, aparantly the Netgear card DOES work, but not well, under linux, and Cisco's Aironets are supposed to work fine, though they cost twice as much and I'll gladly pay, but I have yet to find a retail channel for these (help)
So I've bought six different types of cards ranging in price from $49 to $79 and they are ALL broadcom products. You can see the similarities in the physical construction of the card as soon as you take it out of the box. Slight differences in antenna shape, but always with the broadcommy squareness.
Also, you can order parts real cheap and configure them to put out a carrier on any frequency you want, so this really sounds like baloney. Not that it isn't the reason they're giving, but it isn't the reason. We're talking about less than 0.1 watt here. If the military doesn't like the signal they can just move to the livingroom. Seems to block it just fine from my experience.
So I have an awesome little laptop robot and I can only control it from windows or mac and I have another one that I haven't even bothered with because I can't operate them at the same time.
I don't feel very free to innovate.
Most of these cards have internal restrictions (such as firmware) preventing out-of-band operation. At least Prisms do.
Even with wide-open drivers, I don't think you can force most Prism cards out-of-band because the firmware restricts it. (Which sucks, because all it takes is a 55-question multiple choice test and you can legally run 802.11 cards out of the ISM band and at much higher powers - The 2.4 GHz amateur (ham) band is adjacent to the ISM band, and many cards can be reclassified under Part 97 rules.)
What I would love to see is open specs for a cable modem chipset - it would make a perfect exciter for an amateur data network if combined with a transverter.
retrorocket.o not found, launch anyway?
I needed a new router/firewall/AP, I bought the linksys g product, a couple of G cards, spent an afternoon looking for Broadcom support in linux and pitched Redhat for WXp as a result.
That laptop is mostly used for surfing the web and rdesktopping into other computers, I'd most likely put RH8 (9) BACK on it if Broadcom's g set was supported.
I can't be the only person that WANTS this. (and I DID lobby Broadcom and Linksys for driver support)
"Draco dormiens nunquam titillandus."
Proxim let a developer take a library written by them custom for him that allows him certain functions that he needs, it works with pcmcia, isa and pci cards. he just wrote a wrapper around this library and has no access to the internal workings of the actual chipset to protect their details for security I suppose. Maybe he can help someone find the way with these new chip makers since he worked out a deal with another company that was not going to release the internal details. From what I gathered from back when I used one, they built him the library and gave him the headers.
His source code distribution with prebuilt library is at:
http://www.komacke.com/distribution.html
I was able to use it for my laptop with a pcmcia card, my smp machine with pci card and pentium with an isa card on 2.2 and 2.4 kernels so the method obviously will work.
Why can't the frequency band and power be enforced in hardware?
There are potentially serious issues with 5 GHz. WLAN technologies and military radar avoidance. This is the case in both the U.S. and Europe. The IEEE is working on mechanisms to dynamically avoid these conflicts. I can understand why it could be bad to allow anybody out there with one of these devices to twiddle the knobs as they could cause serious problems. This doesn't, in my mind, rule out binary drivers.
I recently picked up a Sony Z1A with Centrino wireless built in. Its a great laptop though I'm still figuring out all the Linux stuff for it. I gave up on linux support for wireless after figuring out that the windows drivers don't even work well. They're great UNLESS you have other networks in the area.
For example, I have my home wireless network and my neighbor also has one that my laptop picks up. Because Centrino is "smart" enough to be "seamless" across networks, 5 minutes after connecting to my own network, windows lets me know that there are other networks available, prompts me to look at them, AND cuts off the current network connection!
Essentially the network connection must be reset every 4-5 minutes because it is being "smart" and cutting service at the sign of another provider.
I made it to a Sony Engineer (for fun) and they had no possible solution. I really just want to tell the network driver and/or built in windows wireless software to hold onto a network connection once its established... but NOOOOO, it wants to ask me to look at more and drops me from my current network.
The Sony guy tried to tell me that it would be okay. Riiiight, since I can't do anythign (IM, ftp, etc) that takes longer than 5 minutes to complete.
And this is in Windows...
unf.
I know OSS is about coders scratching an itch and writing drivers for stuff that they are interested in at the moment, but what about the serial drivers in the 2.5.6x kernel? Many of them are still using cli(), sti() and friends. This is deep juju, too deep for a code cleaner like me to do. Even generic-serial is still using the old method. I can fix other things like 'flags' should be unsigned long, but redoing irq handling needs people who know what is going on.