I'm sick of people being misinformed that Broadcom WLAN does not have Linux compatibility

[beer]

PSA: http://www.broadcom.com/support/802.11/linux_sta.php

Don't know if you have a supported Broadcom chip?
bcm4311 = PCI Express single-lane (newest laptops only)
bcm4318 - mini-PCI
bcm4306C0 - look at the silicon chip. It should say 4306; the stepping on the silicon should read 'P30' to indicate C0 rev; if the stepping reads '20' you're just running the 4306B0 and it might not work.the B0 had separate MACs for both A and G side, and this isn't a supported platform, so YMMV

Cheers!

 

[spyordie007]

Even before that werent there beta/community drivers? I did wireless with my old laptop and I'm pretty sure that it's MiniPCI wireless card was a Broadcom.

 

[n0cmonkey]

$ file *
bcmsrom.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_channel.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_phy.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_rate_sel.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped

It won't make it into the kernel, it isn't open. Sorry, unsupported.

 

[beer]

Originally posted by: n0cmonkey

$ file *
bcmsrom.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_channel.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_phy.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_rate_sel.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped

It won't make it into the kernel, it isn't open. Sorry, unsupported.

There are reasons why it can't be open. Blame the FCC. The WLAN architecture is different than Intel's or Atheros' and without the precompiled binaries the FCC compliance would not be satisfactory. This is explained in the readme.

It provides functionaliy which is what people generally complain about. And it's certainly better than the reverse-engineered 4301 driver that is out there now.

 

[drag]

Originally posted by: beer

Originally posted by: n0cmonkey

$ file *
bcmsrom.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_channel.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_phy.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_rate_sel.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped

It won't make it into the kernel, it isn't open. Sorry, unsupported.

There are reasons why it can't be open. Blame the FCC. The WLAN architecture is different than Intel's or Atheros' and without the precompiled binaries the FCC compliance would not be satisfactory. This is explained in the readme.

Bullshit. Utter Bullshit.

Ralink and many other manufacturers make cards that are completely compatable with 100% open source drivers. They release documentation and code.

If Ralink and Realtek can do it.

It provides functionaliy which is what people generally complain about. And it's certainly better than the reverse-engineered 4301 driver that is out there now.

No it's not.

The reverse engineered drivers works out of the box on 2.6.17 kernels. I am using it right now. You just need to rip the firmware out of a OSX or Windows XP driver.

It's superior because:
1. It's a part of the kernel by default.
2. It's open source.
3. Works on more then just a handfull of linux distros and kernel configurations
4. Works on more then just x86
5. Is supported by it's developers.

If you need a reason to use it I suggest you look at the huge number of security holes being discovered in these 'superior' wireless drivers companies like broadcom supply to Windows, OS X, and Linux.
http://www.infoworld.com/article/06/06/21/79536_HNwifibreach_1.html

What I am tired of is people settling for closed source and driver wrappers because they semi-work.


AND YOU KNOW WHAT HAPPENNED ABOUT FOUR MONTHS AGO??

People using the reverse engineered documentation on Broadcom chipsets finally had working drivers.

Broadcom only released it to try to prevent people from needing to help improve the open source stuff. This is from what I gather.

Broadcom has had drivers that worked well in Linux for YEARS. And they are crossplatform. This is because Wifi in Linux is big business for linux-based networking equipment. Hell even Microsoft uses Linux wifi stuff at their campus. Linux works well on wireless and it's because of manufacturers like Broadcom being dicks about their drivers and their hardware manufacture is what causes huge problems for people starting out with Linux on the desktop.

There is a hundred times more regulation in automobiles. A lot more IP. A lot more patented interfaces and industry secrets. YET automobile manufacturers sell detailed shop manuals, specifications, and even scematics on their stuff. Selling automobiles without giving third parties the ability to work on those cars is almost unthinkable...

Same thing with televisions, radios, microwaves or any complex consumer devices.

YET people have learned that computers and computer hardware isn't like that. They've been taught to 'put up and shut up' and take it up the *** all the f-ing time. It's just plain crap. I am not going to put up with it, and nobody else should either. Screw binary blobs, screw closed source drivers. Manufacturers have a obligation to allow users to know how to work the devices they buy. Buy publishing the programming interfaces there is no reason why they need to describe in detail how the firmware or the hardware acts. They can still keep their 'IP' and they can still meet regulations on radio devices and they can still be competative.

Look at it this way..
If you have 5 manufaturers release devices and 2 manufacturers release open specs and documentation to programmers and 3 don't and the 3 make up all sorts of excuses about why they can't.. Doesn't it seem that either those 3 manufacturers are either full of crap or made huge errors in hardware design that those other 2 makers didn't?

 

[Nothinman]

There are reasons why it can't be open. Blame the FCC. The WLAN architecture is different than Intel's or Atheros' and without the precompiled binaries the FCC compliance would not be satisfactory. This is explained in the readme.

Then they should have designed the chip better. Hell Intel's latest drivers require a userspace daemon to enforce the FCC regulations and have documented the interface so that GPL'd daemons can be written. FCC regulations are no execuse for binary-only kernel code.

It provides functionaliy which is what people generally complain about. And it's certainly better than the reverse-engineered 4301 driver that is out there now.

And what about people in areas where the regulations are different? If they use the drivers you linked to they end up breaking the rules.

 

[Kyanzes]

I'm sick of people being misinformed that Broadcom WLAN does not have Linux compatibility

It doesn't? I can't believe this... I should have seen this coming, one cannot trust Broadcom anymore.

 

[drag]

Originally posted by: Kyanzes

I'm sick of people being misinformed that Broadcom WLAN does not have Linux compatibility

It doesn't? I can't believe this... I should have seen this coming, one cannot trust Broadcom anymore.

It didn't until AFTER:
A. People reversed engineered chipset documentation from broadcom binary-only drivers for the Linksys wifi routers
B. Linux devs using that documentation wrote drivers for the hardware.

Once people had working open source drivers then broadcom felt that they could 'support' linux.

Anyways the broadcom drivers he linked to still have those crappy binary blobs, is unsupported, is only usefull on a small number of linux distros, and is only usefull on one hardware platform. There is no way to file bug reports or fix the drivers either. It's completely unsupported by broadcom and says so on their website.

 

[chango]

www.linuxant.com Use your windows drivers. Worked for me.

 

[n0cmonkey]

Originally posted by: beer
There are reasons why it can't be open. Blame the FCC. The WLAN architecture is different than Intel's or Atheros' and without the precompiled binaries the FCC compliance would not be satisfactory. This is explained in the readme.

It's a poor excuse, and not believable. Too many other companies have played nicely to believe that drivel.

 

[n0cmonkey]

Originally posted by: chango
www.linuxant.com Use your windows drivers. Worked for me.

That's dirtier than using broadcom's blobs and lies.

 

[Nothinman]

www.linuxant.com Use your windows drivers. Worked for me.

That's crap, there's real OSS drivers available so there's no reason to taint yourself by using ndiswrapper or linuxtant.

 

[bersl2]

Originally posted by: drag

Originally posted by: beer

Originally posted by: n0cmonkey

$ file *
bcmsrom.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_channel.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_phy.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
wlc_rate_sel.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped

It won't make it into the kernel, it isn't open. Sorry, unsupported.

There are reasons why it can't be open. Blame the FCC. The WLAN architecture is different than Intel's or Atheros' and without the precompiled binaries the FCC compliance would not be satisfactory. This is explained in the readme.

Bullshit. Utter Bullshit.

Ralink and many other manufacturers make cards that are completely compatable with 100% open source drivers. They release documentation and code.

If Ralink and Realtek can do it.

It provides functionaliy which is what people generally complain about. And it's certainly better than the reverse-engineered 4301 driver that is out there now.

No it's not.

The reverse engineered drivers works out of the box on 2.6.17 kernels. I am using it right now. You just need to rip the firmware out of a OSX or Windows XP driver.

It's superior because:
1. It's a part of the kernel by default.
2. It's open source.
3. Works on more then just a handfull of linux distros and kernel configurations
4. Works on more then just x86
5. Is supported by it's developers.

If you need a reason to use it I suggest you look at the huge number of security holes being discovered in these 'superior' wireless drivers companies like broadcom supply to Windows, OS X, and Linux.
http://www.infoworld.com/article/06/06/21/79536_HNwifibreach_1.html

What I am tired of is people settling for closed source and driver wrappers because they semi-work.


AND YOU KNOW WHAT HAPPENNED ABOUT FOUR MONTHS AGO??

People using the reverse engineered documentation on Broadcom chipsets finally had working drivers.

Broadcom only released it to try to prevent people from needing to help improve the open source stuff. This is from what I gather.

Broadcom has had drivers that worked well in Linux for YEARS. And they are crossplatform. This is because Wifi in Linux is big business for linux-based networking equipment. Hell even Microsoft uses Linux wifi stuff at their campus. Linux works well on wireless and it's because of manufacturers like Broadcom being dicks about their drivers and their hardware manufacture is what causes huge problems for people starting out with Linux on the desktop.

There is a hundred times more regulation in automobiles. A lot more IP. A lot more patented interfaces and industry secrets. YET automobile manufacturers sell detailed shop manuals, specifications, and even scematics on their stuff. Selling automobiles without giving third parties the ability to work on those cars is almost unthinkable...

Same thing with televisions, radios, microwaves or any complex consumer devices.

YET people have learned that computers and computer hardware isn't like that. They've been taught to 'put up and shut up' and take it up the *** all the f-ing time. It's just plain crap. I am not going to put up with it, and nobody else should either. Screw binary blobs, screw closed source drivers. Manufacturers have a obligation to allow users to know how to work the devices they buy. Buy publishing the programming interfaces there is no reason why they need to describe in detail how the firmware or the hardware acts. They can still keep their 'IP' and they can still meet regulations on radio devices and they can still be competative.

Look at it this way..
If you have 5 manufaturers release devices and 2 manufacturers release open specs and documentation to programmers and 3 don't and the 3 make up all sorts of excuses about why they can't.. Doesn't it seem that either those 3 manufacturers are either full of crap or made huge errors in hardware design that those other 2 makers didn't?

Hmm... I wonder, could the unwillingness of certain chip companies to release their documentation have anything to do with, say, money? They have something we want, and they know it.

But this we all know. So I ask the question: How much do you think it would cost for someone to pay any given company to release the needed specifications? Yes, I know, money is very, very tight in the FOSS world, but what order of magnitude am I talking about? Thousands? Hundreds of thousands? Millions?

Money is the only language these guys understand. Is it even possible for us to speak it to the needed degree?

 

[drag]

It probably would require enough money to help pay off cross-licensing patent agreements.. which can be very very expensive. Millions to get out of a software patent licensing agreements. It's not going to happen. There is no point either.

It's much much cheaper just to buy stuff from people that do a decent job of supporting Linux.

But realy all that is needed is to do customer demand. Until Linux gets to the point were costs justify the risks to they simply aren't going to care.

The 'FOSS world' spends enough money on stuff that it makes it worth for smaller companies. Getting choosen over Intel or broadcom for things is a big deal for companies like Ralink. So they are happy to share if it means that they make advances in the market place. Advances small enough that intel or broadcom doesn't care about at this moment.

I think that for the desktop the turning point will be probably 10% of the market for Linux People just need to speak with their wallets. That's the money that companies care about. They want to sell their products. If selling their products mean that it has to be documented and open-enough then they will do that. They will just figure out how to do it.

Especially if people like HP or Dell begin selling Linux desktops/laptops.

Just think: Linux and Windows work equaly well on open hardware. Windows runs fine on closed stuff, but for Linux it causes problems for end users. There is plenty of inexpensive open(ish) hardware that HP/Dell/Whoever can get for any purpose, except for video cards.

(all hardware is basicly propriatory, buy it's nature, but I am talking about more open interfaces)

So if your Dell would you have seperate lines of computers just for Linux and Windows just because some chipset maker has a big IP stick up you-know-were? Or will they just purchase the same open(ish) hardware for both types of computers?

No of course not! It would be much cheaper to go with the hardware that works fine on both systems. So even if Linux got to only 10-20% market share then that mean that you'd have to have highly linux compatable hardware on 20-50% to fit in with the market realities of OEM computers.

Intel has learned that it pays to work with Linux. All their stuff has pretty open stuff. They have that oddball wifi thing which sucks, but otherwise their stuff is pretty linux friendly. I wouldn't say that Intel itself is Linux-friendly.. They realy aren't. They don't give a crap about FOSS. But they found out that it pays to support Linux properly (and they want people to know this). Lowers cost, increases demand. What more could you want? Just plain old capitalist forces. Even if it's small compared to Windows there is still money to be made. It's amazing what a little competition and market pressure will do (hint: AMD)

 

[n0cmonkey]

Originally posted by: bersl2
Hmm... I wonder, could the unwillingness of certain chip companies to release their documentation have anything to do with, say, money? They have something we want, and they know it.

But this we all know. So I ask the question: How much do you think it would cost for someone to pay any given company to release the needed specifications? Yes, I know, money is very, very tight in the FOSS world, but what order of magnitude am I talking about? Thousands? Hundreds of thousands? Millions?

Money is the only language these guys understand. Is it even possible for us to speak it to the needed degree?

A lot of people have already given them money. Everyone that has a broadcom card handed over some money to that company. The least they should get is support.