Skip to contentSkip to conversation

Atheros and rt2x00 driver

DISCUSSION Daniel J Blueman 6 participants 10 messages Aug 17, 2005 – Aug 31, 2005

lore
Share & tools
lore
#0 lore raw
Quoted 5 lines · Jon Jahren

Hello, I'm new to the mailling list, and couldn't find any traces of discussing this anywhere. I was wondering why neither the atheros driver
http://madwifi.sourceforge.net, or the rt2x00 driver
http://rt2x00.serialmonkey.com /wiki/index.php/Main_Page is included in the kernel?

There is a good chance the rt2x00 driver will get into the kernel tree in time, since there is no firmware to upload - Ralink Tech (www.ralink.com.tw) took a design decision to incorporate the firmware into an EEPROM on-board, allowing their driver to be GPL'd, and the rt2x00 is a Linux-specific rewrite which is stabilising well.
___
Daniel J Blueman

#1 lore raw

Daniel J Blueman <[email protected]> wrote:
->
-> There is a good chance the rt2x00 driver will get into the kernel tree -> in time, since there is no firmware to upload - Ralink Tech -> (www.ralink.com.tw) took a design decision to incorporate the firmware -> into an EEPROM on-board, allowing their driver to be GPL'd, and the -> rt2x00 is a Linux-specific rewrite which is stabilising well.

the same applies to atheros. they got stuff either in eeprom or in driver. no firmware necessary.

there is an ongoing project for atheros cards.
work in progress located at http://mateusz.agrest.org/atheros/

/M

Signature · 4 lines
  @..@   Mateusz Berezecki 
 (----)  [email protected] http://mateusz.agrest.org
( >__< ) PGP: 5F1C 86DF 89DB BFE9 899E 8CBE EB60 B7A7 43F9 5808 
^^ ~~ ^^
#2 lore raw
Quoted 8 lines · Mateusz Berezecki

Daniel J Blueman <[email protected]> wrote:
->
-> There is a good chance the rt2x00 driver will get into the kernel tree -> in time, since there is no firmware to upload - Ralink Tech -> (www.ralink.com.tw) took a design decision to incorporate the firmware -> into an EEPROM on-board, allowing their driver to be GPL'd, and the -> rt2x00 is a Linux-specific rewrite which is stabilising well.

Quoted 5 lines

the same applies to atheros. they got stuff either in eeprom or in driver. no firmware necessary.

there is an ongoing project for atheros cards.
work in progress located at http://mateusz.agrest.org/atheros/

There is still the open question of whether this is legal enough to include in the kernel :(

I really would have preferred a cleanroom approach, like that taken by the forcedeth driver authors.

Jeff

#3 lore raw

* Jeff Garzik:

Quoted 2 lines

There is still the open question of whether this is legal enough to include in the kernel :(

Are you referring to FTC issues, or potential copyright/trade secret issues?

The FTC issues are shared by many (most?) wireless drivers. The copyright/trade secret issues might be worked around by basing the work on the OpenBSD version of that driver (and someone is actually working on that).

#4 lore raw

Florian Weimer <[email protected]> wrote:
-> The FTC issues are shared by many (most?) wireless drivers. The -> copyright/trade secret issues might be worked around by basing the -> work on the OpenBSD version of that driver (and someone is actually -> working on that).

the problem with openbsd version of the hal is that it is - sorry to say that - fundamentally broken, at least it was last time I was checking. It misses just too much functionality. Apart from that, the work I'm doing is partially based on that openbsd stuff :) (no, this doesn't contradict what I wrote above)

Mateusz

Signature · 4 lines
  @..@   Mateusz Berezecki 
 (----)  [email protected] http://mateusz.agrest.org
( >__< ) PGP: 5F1C 86DF 89DB BFE9 899E 8CBE EB60 B7A7 43F9 5808 
^^ ~~ ^^
#5 lore raw

* Mateusz Berezecki:

Quoted 9 lines

Florian Weimer <[email protected]> wrote:
-> The FTC issues are shared by many (most?) wireless drivers. The -> copyright/trade secret issues might be worked around by basing the -> work on the OpenBSD version of that driver (and someone is actually -> working on that).

the problem with openbsd version of the hal is that it is - sorry to say that - fundamentally broken, at least it was last time I was checking.

It's better than nothing, that is, it worked for us when we gave it a try. And it seems to be relatively unencumbered.

#6 lore raw

Florian Weimer <[email protected]> wrote:

-> >  the problem with openbsd version of the hal is that it is - sorry to
-> >  say that - fundamentally broken, at least it was last time I was
-> >  checking.

->
-> It's better than nothing, that is, it worked for us when we gave it a -> try. And it seems to be relatively unencumbered.

ok, maybe it sounded to harsh. What I really meant was that when I was doing my research on the atheros chipset I revealed some differences between what is actually in HAL and what is in openbsd driver. It could have been due to the update schedule of madwifi. It is always easier to be up to date with the vendor supported driver than with reverse engineered one. Anyways what I saw was that some parts were missing in openbsd. I don't know how much in overall.

Mateusz

Signature · 4 lines
  @..@   Mateusz Berezecki 
 (----)  [email protected] http://mateusz.agrest.org
( >__< ) PGP: 5F1C 86DF 89DB BFE9 899E 8CBE EB60 B7A7 43F9 5808 
^^ ~~ ^^
#7 lore raw
In reply to Mateusz Berezecki
Quoted 9 lines · Mateusz Berezecki

Florian Weimer <[email protected]> wrote:
-> The FTC issues are shared by many (most?) wireless drivers. The -> copyright/trade secret issues might be worked around by basing the -> work on the OpenBSD version of that driver (and someone is actually -> working on that).

the problem with openbsd version of the hal is that it is - sorry to say that - fundamentally broken, at least it was last time I was checking. It misses just too much functionality. Apart from that, the

What it can do? In particular, can it:
* send packets with arbitrary contents? In particular, packets shorter than 3-address 802.11 header? packets with WEP bit set? Does it allow to do WEP encoding by host instead of hal? Any weird limitations?
* receive packets?
* tune to the given channel (or freq)?

If it can do that, everything else IIRC can be done in software. Really, what prevents us from, say, beacons every 1/10s?

Quoted 2 lines

work I'm doing is partially based on that openbsd stuff :) (no, this doesn't contradict what I wrote above)

Nice to see you are in "release early" crowd.

http://mateusz.agrest.org/atheros/:
may I suggest using atheros-20050805 instead of atheros-08052005 (will maintain correct sorting order in 2006,2007...) --
vda

#8 lore raw

Denis Vlasenko <[email protected]> wrote:
-> What it can do? In particular, can it:
-> * send packets with arbitrary contents? In particular, packets

->   shorter than 3-address 802.11 header? packets with WEP bit set?
->   Does it allow to do WEP encoding by host instead of hal?
->   Any weird limitations?

-> * receive packets?
-> * tune to the given channel (or freq)?
->

I don't really know to which version you are referring to. If you wrote about OpenBSD then I can't give an authoritative answer.

-> If it can do that, everything else IIRC can be done in software. -> Really, what prevents us from, say, beacons every 1/10s?

Yes, that is a very nature of atheros radio chips. It is simplier to tell what you can't do in software with these chipsets. This complicates writing a driver a bit too ;-)

->
-> Nice to see you are in "release early" crowd.
->
-> http://mateusz.agrest.org/atheros/:
-> may I suggest using atheros-20050805 instead of atheros-08052005 -> (will maintain correct sorting order in 2006,2007...)

Hummm... these are old patches :) the new site is located at

http://www.ath-driver.org

As I'm going to put separate patch snapshots out there too, I will change the naming order.

Mateusz

Signature · 4 lines
  @..@   Mateusz Berezecki 
 (----)  [email protected] http://mateusz.agrest.org
( >__< ) PGP: 5F1C 86DF 89DB BFE9 899E 8CBE EB60 B7A7 43F9 5808 
^^ ~~ ^^
#9 lore raw
In reply to Daniel J Blueman
Quoted 3 lines · Daniel J Blueman

Ralink Tech (www.ralink.com.tw) took a design decision to incorporate the firmware into an EEPROM on-board, allowing their driver to be GPL'd

Binary only firmware and firmware loading is perfectly compatible with the GPL, as long as the vendor includes a license to redistribute the firmware. The problem was that vendors were distributing the firmware embedded in the driver code as a big hex string, without a separate license, which made the firmware fall under the GPL, which make the whole kernel undistributable as there's no source code for the firmware.

Lee