[ath5k] Please add support for AR5418 (details included)

Hi,

I suppose the device will need to be reverse engineered since OpenHAL don't support it. So it might take time.

But you can still use the old madwifi driver (there is a special branch with support for AR5008 which is the same as AR5418). But you might have a problem with the bug #1017. Then you could use (as I do) ndiswrapper.

Replying to Mildred:

But you can still use the old madwifi driver (there is a special branch with support for AR5008 which is the same as AR5418).

This branch is no more, it has been merged to trunk months ago.

Okay, I have a Lenovo X60 (Japanese version) with the AR5418, running Fedora 8, and it doesn't work.

After a moderate amount of pain, I got the wireless running (and sort of managed with NetworkManager, though it was a little erratic) with madwifi and Fedora 7, six months or so ago.

Unfortunately, I had a disk crash a couple of weeks ago, and when I reinstalled, I went with Fedora 8 instead of Fedora 7...and no WLAN.

I even went so far as to mess with the madwifi kernel module sources trying to convince the driver to recognize my chip and treat the same as an older one, but no dice. (I probably was missing some spot in the source where the device identification happens; I tried adding it to ar_device in if_ath.c, and to ath_pci_id_table[] in if_ath_pci.c, but no luck.)

This is with

kernel-2.6.23.15-137.fc8 madwifi-0.9.3.3-1.lvn8 kmod-madwifi-0.9.3.3-15.lvn8

From dmesg:

ath_hal: module license 'Proprietary' taints kernel. ath_hal: 0.9.18.0 (AR5210, AR5211, AR5212, RF5111, RF5112, RF2413, RF5413) wlan: 0.8.4.2 (0.9.3.3) ath_pci: 0.9.4.5 (0.9.3.3)

From lspci:

03:00.0 Network controller: Atheros Communications, Inc. AR5418 802.11a/b/g/n Wireless PCI Express Adapter (rev 01)

From modprobe.conf:

alias wlan0 ath_pci

[rdv@localhost madwifi-0.9.3.3]$ lsmod | grep ath ath_pci 84384 0 wlan 172740 1 ath_pci ath_hal 193488 1 ath_pci dm_multipath 18249 0 dm_mod 46209 9 dm_multipath,dm_snapshot,dm_zero,dm_mirror

A little more detail:

[root@localhost Desktop]# lspci -v 03:00.0 Network controller: Atheros Communications, Inc. AR5418 802.11a/b/g/n Wireless PCI Express Adapter (rev 01)

Subsystem: Atheros Communications, Inc. Unknown device 0033 Flags: fast devsel, IRQ 22 Memory at edf00000 (64-bit, non-prefetchable) [size=64K] Capabilities: [40] Power Management version 2 Capabilities: [50] Message Signalled Interrupts: Mask- 64bit- Queue=0/0 Enable- Capabilities: [60] Express Legacy Endpoint, MSI 00 Capabilities: [90] MSI-X: Enable- Mask- TabSize?=1 Kernel modules: ath5k

...but no lock, never comes up with the device.

Any suggestions? My track record for buying PCMCIA adapters off the shelf here in Japan having them work out of the box is also terrible, so I'm reluctant to go fishing in Akihabara without a clear indication of the chip I'm getting, which is hard to come by...

Replying to rdv@sfc.wide.ad.jp:

madwifi-0.9.3.3-1.lvn8

As stated clearly in the release notes of 0.9.3.3, there is no AR5008 support in this version. You need trunk.

In addition: this ticket is about ath5k, not MadWifi, and you are asking for support, not reporting a bug. Please contact our regular support channels for that.

Thanks for the quick response, after two weeks of messing with it I got it working in twenty minutes once I knew that about madwifi-trunk and that the 5418 is similar to the 5008. Apologies for being slightly off base in the forum. I had equated the term "trunk" with "latest official release", not realizing that I should be following, e.g. the instructions here: www.thinkwiki.org/wiki/How_to_checkout_and_install_madwifi_experimental_driver_for_ar5008 It now works on my Lenovo X60.

Replying to rdv@sfc.wide.ad.jp:

Apologies for being slightly off base in the forum.

For the sake of correctness: this is no forum, but an issue tracker. It's meant for reporting bugs found in the drivers and is not suited for support requests. Hence my (probably harsh) response above.

Hi,

is there any progress on this issue? I'd love to use ath5k on my macbook :)

regards,

Moritz

Working just fine on MacBook Pro. I've checked out the latest trunk (r3561), ran make; make install; reboot; type my WEP key into Network Manager and was on. Thanks, devs!

Replying to eduffy:

Working just fine on MacBook Pro. I've checked out the latest trunk (r3561), ran make; make install; reboot; type my WEP key into Network Manager and was on. Thanks, devs!

but... are you using madwifi or ath5k driver?

With madwifi driver works a lot, after stop with a error and only with a reboot that the NIC works more some time. For me, in last days, the media is 4-5 hours of work, after, shutdown the machine and go to sleep. =/

Hi

I saw the ath5k code in kernel.org[1] and, if i understood correctly, haven't nothing for chips >=ar5416 ? If this is true, how is the roadmap, or where can i see about, of ath5k devel?

Thanks in advanced.

[1]git.kernel.org/?p=linux/kernel/git/linville/wireless-2.6.26.git;a=blob;f=drivers/net/wireless/ath5k/hw.c;h=87e782291a0180536cef3329de66cfc5ad32dec6;hb=master#l264

I can't see a way to subscribe to bugs without replying, so here's a reply. I have a Macbook Pro (v3, Santa Rosa) and am currently suffering madwifi. Here's to an imminent kernel release that features AR5418 support!

The just released Linux kernel 2.6.26 still says

myname linux # dmesg |grep -i ath
ath5k_pci 0000:02:00.0: registered as 'phy0'
ath5k phy0: Device not yet supported.

about my card

02:00.0 Network controller: Atheros Communications, Inc. AR5418 802.11a/b/g/n Wireless PCI Express Adapter (rev 01)

when is this card going to be supported? It's not like nobody owns a macbook. If I can help somehow, let me know.

regards,

Moritz

Replying to moritz at eysholdt dot de:

The just released Linux kernel 2.6.26 still says {{{ myname linux # dmesg |grep -i ath ath5k_pci 0000:02:00.0: registered as 'phy0' ath5k phy0: Device not yet supported. }}} about my card {{{ 02:00.0 Network controller: Atheros Communications, Inc. AR5418 802.11a/b/g/n Wireless PCI Express Adapter (rev 01) }}} when is this card going to be supported? It's not like nobody owns a macbook. If I can help somehow, let me know.

Well, the upcoming madwifi release will support it... so there's no need for any ath5k support at all!

Replying to anonymous:

Well, the upcoming madwifi release will support it... so there's no need for any ath5k support at all!

*cough* Of course ath5k support is needed for this chipset family, not only because ath5k will replace MadWifi in the future.

Replying to moritz at eysholdt dot de:

when is this card going to be supported? It's not like nobody owns a macbook.

Support will become available as soon as all necessary work has been done and is accepted for inclusion upstream. Please keep in mind that most of the work is done by volunteers in their spare time. It's not like you're talking to a company and/or you pay for this driver...

Support will become available as soon as all necessary work has been done and is accepted for inclusion upstream. Please keep in mind that most of the work is done by volunteers in their spare time. It's not like you're talking to a company and/or you pay for this driver...

I know, I'm an open source developer myself. That is why I asked if I could help out - the open source way of paying for good work of others.

Replying to mrenzmann:

Replying to anonymous:

Well, the upcoming madwifi release will support it... so there's no need for any ath5k support at all!

*cough* Of course ath5k support is needed for this chipset family, not only because ath5k will replace MadWifi in the future.

Well, with the current "new-patch-speed" ath5k isn't going to replace madwifi in the next two years! And Luis R. seems to be bussy with regdomain-changes...

Replying to anonymous:

That is why I asked if I could help out - the open source way of paying for good work of others.

Thanks for your offer. Anyone who is interested to help with ath5k development should subscribe to the ath5k-devel mailing list. A good start is to follow (and, where possible, contribute to) ongoing discussions, test patches that are posted there, and so on. Offering your help on the list is another option.

Replying to anonymous:

Well, with the current "new-patch-speed" ath5k isn't going to replace madwifi in the next two years!

What is your contribution again to speed up the development of ath5k?

ath5k development is a community effort, and since it requires a lot of reverse engineering due to the lack of publically available chipset documentation it's a lot of work. The more people help, the faster MadWifi can be disposed. And no, you don't need to have any programming knowledge to help, there's a lot of other stuff you could do.

Replying to mrenzmann:

Replying to anonymous:

Well, with the current "new-patch-speed" ath5k isn't going to replace madwifi in the next two years!

What is your contribution again to speed up the development of ath5k?

Nothing. In fact I'm for the opposite direction, as there are lots of problems with the ath5k & mac80211 combo. And some of critical ones are in the mac80211-stack.

e.g: - Regdomain/TPC/DFS (dot11d/dot11h) support is next to nil.

Where is channel 13? What about the 5Ghz DFS/TPC frequencies? And this is a minefield! Especially since mac80211 doesn't enforce your countries limits, you could end with a big fine!

- rate control algorithms?

PID is a joke compared to minstrel or sample. Intel has even written their own rate code for their iwlwifi driver instead of "fixing the old one".

- powersaving features don't exist either.

Ask any laptop battery about PSPOLL/UAPSD.

- MAC-ACL (iptables doesn't filter MLMEs) / WDS /

VAPs (and not just cooked monitor devices!)

- Accesspoint mode doesn't work at all if you last name isn't Berg,

Busch, Wu, Slaby, Kossifidis or Rodriguez.

- all the plugs for the nice bells and whistles like

TurboG/A,FastFrame? or eXtended Rates are missing.

Most of this stuff has been in the net80211-stack for ages!

So, good luck with your ongoing contributions, the problem isn't reverse engineering the driver (the HAL isn't really obfuscated and is much smaller than the blob from a green gfx maker.), but writing a good foundation! And don't tell me that it's hard to read the 802.11 specifications and write some lines of code for it!

looks like ath9k is going to support the card:

http://www.heise-online.co.uk/open/Kernel-log-ath9k-driver-for-new-Atheros-WLAN-chips-discussions-of-Nvidia-drivers--/news/111148
http://wireless.kernel.org/en/users/Drivers/ath9k

I made ath9k work for my AR5418 and unencrypted wifi using basically theses commands:

cd /usr/src/
git clone git://git.kernel.org/pub/scm/linux/kernel/git/mcgrof/ath9k.git 
git clone git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-testing.git
cd ath9k
./gen-patch.sh
cd ../wireless-testing
patch -p1 < ../ath9k/add-ath9k.patch

cp /usr/src/my-old-linux-kernel/.config .
make oldconfig
make menuconfig
make bzImage modules modules_install

I ddn't get WPA to work, haven't tried WEP

Both, ath5k and ath9k, are part of the kernel some time now. It's better for everyone to have bug reports for those drivers in the kernel bugzilla, as they will likely get better attention there. Hence from now on we don't accept new tickets for ath5k/ath9k here.

We talked to Linux Wireless developers about how to best handle tickets that already have been filed here for ath5k/ath9k. The original idea was to copy those bugs over to the kernel bugzilla. But then many of these tickets have not seen any activity for a while, and it's not clear if the reported issue still exists or if the reporter is even still around to answer bugs.

In the end we agreed that it would be best to ask people to open a new ticket at bugzilla.kernel.org for those issues that still exist in the latest version of ath5k/ath9k, and close related tickets here. This is what I do now (closing as invalid, since that seems to be the best matching resolution available).

Sorry for the inconvenience, and thank you for your understanding.