status of at76-usb driver

status of at76-usb driver

[Claudio]

Hi!

Since my hardware stopped working properly due to a distro (ubuntu btw) upgrade and I was not able to quickly fix it, I started investigating (again) in \"the case of at76-usb\" and read through this mailing-list, the wiki and had a look at the changes in the git tree. I am not sure if I understood everything correctly.

Can anyone sum up what the current state for users of at76-usb devices is please?

It seems like the working code has been removed from the kernel in order to rewrite it anywhere in the future and adapt it to this new mac80211 thing. Which means that currently users (like me) have to use an old version of linux (like 2.4.22) to make use of their devices. Hopefully I am wrong...

Kalle, Milan, Dan, <else>?

Regards,

Claudio

Re: status of at76-usb driver

[Kalle Valo]

Claudio <der.claudio@aon.at> writes:

> Hi!

Hello,

> Since my hardware stopped working properly due to a distro (ubuntu
> btw) upgrade and I was not able to quickly fix it

I'm sorry to hear this. Can you tell me what Ubuntu version and at76
hardware you have? I have Ubuntu running on one of my test laptops, I
could test it myself.

> I started investigating (again) in \"the case of at76-usb\" and read
> through this mailing-list, the wiki and had a look at the changes in
> the git tree. I am not sure if I understood everything correctly.
>
> Can anyone sum up what the current state for users of at76-usb devices
> is please?
>
> It seems like the working code has been removed from the kernel in
> order to rewrite it anywhere in the future and adapt it to this new
> mac80211 thing. Which means that currently users (like me) have to use
> an old version of linux (like 2.4.22) to make use of their devices.
> Hopefully I am wrong...

The current state is this:

A long time ago I started porting at76_usb to use mac80211, the IEEE
802.11 stack in Linux. It's still buggy, biggest problem is that for
some reason I have to replug the device constantly. I guess firmware
is crashing because of a driver bug, or something like that.
Unfortunately I haven't worked that much on the problem and haven't
been able to solve it yet. As soon as I get good enough stability I
will resubmit it to wireless-testing tree, but I'm not there quite
yet.

Fortunately Greg created staging tree some time ago and included the
original at76_usb driver to the tree:

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=tree;f=drivers/staging/at76_usb;hb=HEAD

So starting from 2.6.28 (please correct me if I'm wrong here) the
original (and working!) at76_usb driver should be included already in
the kernel, it's just up to the distributions to enable it.

Grep, I noticed that the TODO file in your tree doesn't mention
anything about my mac80211 port. I'll send you a patch fixing that.

Also I'm going to rename the mac80211 port to something else for just
to avoid the confusion with the original driver. Last thing I want to
see is having two in-tree drivers with identical names.

-- 
Kalle Valo

Re: status of at76-usb driver

[Pavel Roskin]

On Sun, 2008-12-28 at 15:26 +0200, Kalle Valo wrote:

> Also I'm going to rename the mac80211 port to something else for just
> to avoid the confusion with the original driver. Last thing I want to
> see is having two in-tree drivers with identical names.

I asked to rename the driver that was imported into the staging tree,
but my request was ignored.  I guess nobody really cares about the name
conflict.  I'll appreciate if we avoid further renames.  Once the
mac80211 port is working properly, it should be included in to the next
kernel release, and the driver from the staging tree should be removed
in the same kernel release.

-- 
Regards,
Pavel Roskin

Re: status of at76-usb driver

[Greg KH]

On Sun, Dec 28, 2008 at 04:28:03PM -0500, Pavel Roskin wrote:
> On Sun, 2008-12-28 at 15:26 +0200, Kalle Valo wrote:
> 
> > Also I'm going to rename the mac80211 port to something else for just
> > to avoid the confusion with the original driver. Last thing I want to
> > see is having two in-tree drivers with identical names.
> 
> I asked to rename the driver that was imported into the staging tree,
> but my request was ignored.

You did?  I don't see that in my email archives, sorry, did you cc: me?

> I guess nobody really cares about the name conflict.

What name conflict?  Is there in in-kernel with this same name for a
module?  If so, let me know and I'll be glad to rename the staging one.

> I'll appreciate if we avoid further renames.

renames are trivial :)

> Once the mac80211 port is working properly, it should be included in
> to the next kernel release, and the driver from the staging tree
> should be removed in the same kernel release.

When a "real" driver is included, I will be glad to delete the staging
driver.  Just let me know.

thanks,

greg k-h

Re: status of at76-usb driver

[Kalle Valo]

Greg KH <gregkh@suse.de> writes:

> On Sun, Dec 28, 2008 at 04:28:03PM -0500, Pavel Roskin wrote:
>> On Sun, 2008-12-28 at 15:26 +0200, Kalle Valo wrote:
>> 
>> > Also I'm going to rename the mac80211 port to something else for just
>> > to avoid the confusion with the original driver. Last thing I want to
>> > see is having two in-tree drivers with identical names.
>> 
>> I asked to rename the driver that was imported into the staging tree,
>> but my request was ignored.
>
> You did?  I don't see that in my email archives, sorry, did you cc: me?
>
>> I guess nobody really cares about the name conflict.
>
> What name conflict?  Is there in in-kernel with this same name for a
> module?

Not yet, but there will be once I manage to get the mac80211 port
usable.

> If so, let me know and I'll be glad to rename the staging one.

I can rename the mac80211 port, it's less confusing for the users.
Please don't rename the staging driver.

>> Once the mac80211 port is working properly, it should be included in
>> to the next kernel release, and the driver from the staging tree
>> should be removed in the same kernel release.
>
> When a "real" driver is included, I will be glad to delete the staging
> driver.  Just let me know.

I have to disagree with Pavel here. We should have a proper transition
period, because I'm sure that there will be problems with the new
driver. The staging driver doesn't create any harm, quite the opposite
actually.

Greg, I'll let you know when I think the staging driver can be
removed.

-- 
Kalle Valo

Re: status of at76-usb driver

[Claudio]

>> Since my hardware stopped working properly due to a distro (ubuntu 
>> btw) upgrade and I was not able to quickly fix it 
> 
> I'm sorry to hear this. Can you tell me what Ubuntu version and at76 
> hardware you have? I have Ubuntu running on one of my test laptops, I 
> could test it myself. 

my at76 hardware: Netgear MA101 rev.B

experiences:
- works with Ubuntu 7.10 with stand-alone at76-usb-0.17 package
- AFAIR didn't work with Ubuntu 7.10 out-of-the-box
- didn't work with Ubuntu 8.10 (even didn't when I tried to install the
  old stand-alone version)

By "does not work" I mean it initializes correctly and sometimes(!)
displays our network, but won't establish a connection.

I am curious hearing about your experience.

Claudio


---
Die aonKombi zum Sensationspreis!
Breitband-Internet + Mobiltelefonie + Festnetz um insgesamt nur EUR 19,90* pro Monat.
Und jetzt sogar mit Gratis-Herstellung!* Infos und Aktionsbedigungen hier: http://www.telekom.at/aonKombi
 

Re: status of at76-usb driver

[Kalle Valo]

Claudio <der.claudio@aon.at> writes:

> my at76 hardware: Netgear MA101 rev.B
>
> experiences:
> - works with Ubuntu 7.10 with stand-alone at76-usb-0.17 package
> - AFAIR didn't work with Ubuntu 7.10 out-of-the-box
> - didn't work with Ubuntu 8.10 (even didn't when I tried to install the
>   old stand-alone version)
>
> By "does not work" I mean it initializes correctly and sometimes(!)
> displays our network, but won't establish a connection.
>
> I am curious hearing about your experience.

I tried twice with latest Ubuntu 8.10 (updated yesterday) and kernel
oopsed both time for me:

[  406.516067] usb 2-2: new full speed USB device using uhci_hcd and address 2
[  406.678113] usb 2-2: configuration #1 chosen from 1 choice
[  407.294144] at76_usb: Atmel at76c50x USB Wireless LAN Driver 0.14beta1 loading
[  407.297435] firmware: requesting atmel_at76c503-rfmd.bin
[  407.379144] usbcore: registered new interface driver at76_usb
[  409.544088] at76_usb: wlan0 disconnecting
[  409.546070] at76_usb: at76_usb disconnected
[  409.660069] usb 2-2: reset full speed USB device using uhci_hcd and address 2
[  409.804597] usb 2-2: device firmware changed
[  409.806621] usb 2-2: USB disconnect, address 2
[  409.809482] firmware: requesting atmel_at76c503-rfmd.bin
[  409.817171] at76_usb: wlan0 disconnecting
[  409.819601] BUG: unable to handle kernel NULL pointer dereference at 00000000
[  409.819617] IP: [<c037da0b>] __up+0xb/0x40
[  409.819635] *pde = 00000000 
[  409.819645] Oops: 0000 [#1] SMP 
[  409.819654] Modules linked in: at76_usb af_packet radeon drm bridge rfcomm stp bnep sco l2cap bluetooth thinkpad_acpi rfkill led_class nvram ppdev ipv6 speedstep_ich speedstep_lib cpufreq_powersave cpufreq_conservative cpufreq_stats cpufreq_ondemand freq_table cpufreq_userspace sbs container sbshc wmi pci_slot iptable_filter ip_tables x_tables lp pcmcia evdev battery ac psmouse serio_raw video output snd_intel8x0 snd_ac97_codec ac97_bus parport_pc snd_pcm_oss snd_mixer_oss parport snd_pcm irda crc_ccitt snd_seq_dummy pcspkr yenta_socket rsrc_nonstatic pcmcia_core snd_seq_oss snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq snd_timer snd_seq_device button snd iTCO_wdt iTCO_vendor_support soundcore snd_page_alloc intel_agp agpgart shpchp pci_hotplug ext3 jbd mbcache sr_mod cdrom sd_mod crc_t10dif sg pata_acpi ata_piix ata_generic libata uhci_hcd e100 mii scsi_mod usbcore dock thermal processor fan fbcon tileblit font bitblit softcursor fuse
[  409.819837] 
[  409.819843] Pid: 6, comm: events/0 Not tainted (2.6.27-9-generic #1)
[  409.819850] EIP: 0060:[<c037da0b>] EFLAGS: 00010046 CPU: 0
[  409.819859] EIP is at __up+0xb/0x40
[  409.819865] EAX: 00000000 EBX: ddc7b528 ECX: 00000000 EDX: 00000000
[  409.819871] ESI: 00000246 EDI: e0d367b0 EBP: df065ec8 ESP: df065ec8
[  409.819877]  DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
[  409.819885] Process events/0 (pid: 6, ti=df064000 task=df053ed0 task.ti=df064000)
[  409.819891] Stack: df065ed8 c014c102 ddc7b690 ddc7b480 df065f6c e0d36a24 df065f08 df053ed0 
[  409.819910]        c046e340 df065f14 c0102df6 df05420c c1489980 df065f1c ddc7b480 ddc7b53c 
[  409.819928]        ddc7b528 ddc7b538 c0513d00 df053ed0 c046e340 df065fa4 c037ca49 df065f90 
[  409.819947] Call Trace:
[  409.819953]  [<c014c102>] ? up+0x42/0x50
[  409.819968]  [<e0d36a24>] ? kevent+0x274/0x1710 [at76_usb]
[  409.819990]  [<c0102df6>] ? __switch_to+0xa6/0x160
[  409.820002]  [<c037ca49>] ? schedule+0x429/0x790
[  409.820016]  [<c0143814>] ? queue_delayed_work_on+0xb4/0xe0
[  409.820030]  [<c0192350>] ? vmstat_update+0x0/0x30
[  409.820038]  [<e0d367b0>] ? kevent+0x0/0x1710 [at76_usb]
[  409.820038]  [<c0143645>] ? run_workqueue+0x95/0x160
[  409.820038]  [<c0147576>] ? finish_wait+0x16/0x70
[  409.820038]  [<c01438e8>] ? worker_thread+0x88/0xf0
[  409.820038]  [<c01474b0>] ? autoremove_wake_function+0x0/0x50
[  409.820038]  [<c0143860>] ? worker_thread+0x0/0xf0
[  409.820038]  [<c0147141>] ? kthread+0x41/0x80
[  409.820038]  [<c0147100>] ? kthread+0x0/0x80
[  409.820038]  [<c0105297>] ? kernel_thread_helper+0x7/0x10
[  409.820038]  =======================
[  409.820038] Code: 55 89 e5 53 e8 b7 78 d8 ff 89 c3 e8 80 f4 ff ff 89 d8 90 ff 00 79 05 e8 b4 0a 00 00 5b 5d c3 00 55 89 e5 e8 98 78 d8 ff 8b 40 08 <8b> 08 8b 50 04 89 51 04 89 0a c7 40 04 00 02 20 00 c7 00 00 01 
[  409.820038] EIP: [<c037da0b>] __up+0xb/0x40 SS:ESP 0068:df065ec8
[  409.820038] ---[ end trace 6ce3c49c2e0a5efe ]---

-- 
Kalle Valo

Re: status of at76-usb driver

[Claudio]

Kalle Valo <kalle.valo@iki.fi> wrote:
> Claudio <der.claudio@aon.at> writes:
>
>> my at76 hardware: Netgear MA101 rev.B
>>
>> experiences:
>> - works with Ubuntu 7.10 with stand-alone at76-usb-0.17 package
>> - AFAIR didn\'t work with Ubuntu 7.10 out-of-the-box
>> - didn\'t work with Ubuntu 8.10 (even didn\'t when I tried to install the
>>   old stand-alone version)

Last one is wrong. See below.

> I tried twice with latest Ubuntu 8.10 (updated yesterday) and kernel
> oopsed both time for me:

So it did for me.

But the at76_usb that comes with Ubuntu 8.10 is actually a (very old)
0.14beta1 version. Try $ strings /lib/modules/`uname \\
-r`/kernel/ubuntu/misc/wireless/at76/at76_usb.ko | egrep \'0.1[47]\'

Ad 1: It seems like Ubuntu ships an outdated version of the old
stand-alone driver.

Ad 2: I\'ve found a work-around (I\'ve tried this earlier but it did not
work for any reason): adapting and compiling the 0.17 version of the old
stand-alone driver (http://developer.berlios.de/projects/at76c503a/) ->
It works again.

Since this is not the perfect solution I would like to find out how to
enable the at76 driver from the staging tree and then vote for doing the
same in the Ubuntu kernel package. Well, how do I enable this driver
anyway?

Or what do you think we should do?

Claudio

Re: status of at76-usb driver

[Kalle Valo]

On Fri, Jan 2, 2009 at 9:12 PM, Claudio <der.claudio@aon.at> wrote:
> Kalle Valo <kalle.valo@iki.fi> wrote:
>> I tried twice with latest Ubuntu 8.10 (updated yesterday) and kernel
>> oopsed both time for me:
>
> So it did for me.
>
> But the at76_usb that comes with Ubuntu 8.10 is actually a (very old)
> 0.14beta1 version. Try $ strings /lib/modules/`uname \\
> -r`/kernel/ubuntu/misc/wireless/at76/at76_usb.ko | egrep \'0.1[47]\'
>
> Ad 1: It seems like Ubuntu ships an outdated version of the old
> stand-alone driver.
>
> Ad 2: I\'ve found a work-around (I\'ve tried this earlier but it did not
> work for any reason): adapting and compiling the 0.17 version of the old
> stand-alone driver (http://developer.berlios.de/projects/at76c503a/) ->
> It works again.
>
> Since this is not the perfect solution I would like to find out how to
> enable the at76 driver from the staging tree and then vote for doing the
> same in the Ubuntu kernel package. Well, how do I enable this driver
> anyway?

I think at76_usb was included to 2.6.28 and I doubt that ubuntu uses
so new kernel yet.

> Or what do you think we should do?

Better to use the stand-alone driver from berlios for now and wait for
ubuntu to upgrade 2.6.28.

Kalle

Re: status of at76-usb driver

[John W. Linville]

On Sun, Dec 28, 2008 at 03:26:55PM +0200, Kalle Valo wrote:

> Fortunately Greg created staging tree some time ago and included the
> original at76_usb driver to the tree:
> 
> http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=tree;f=drivers/staging/at76_usb;hb=HEAD
> 
> So starting from 2.6.28 (please correct me if I'm wrong here) the
> original (and working!) at76_usb driver should be included already in
> the kernel, it's just up to the distributions to enable it.
> 
> Grep, I noticed that the TODO file in your tree doesn't mention
> anything about my mac80211 port. I'll send you a patch fixing that.
> 
> Also I'm going to rename the mac80211 port to something else for just
> to avoid the confusion with the original driver. Last thing I want to
> see is having two in-tree drivers with identical names.

OK, I still have patches to recreate the original port of at76_usb
to mac80211 and another to keep-up with a later mac80211 API change
on the at76 branch of wireless-testing.  I had intended to send them
to Greg now that the mac80211 API change has made it upstream.

Since you mentioned a) that you are working on the port again and b)
you are going to rename the driver for mac80211, should I just drop
my at76 branch of wireless testing for good?

John
-- 
John W. Linville		Linux should be at the core
linville@tuxdriver.com			of your literate lifestyle.

Re: status of at76-usb driver

[Kalle Valo]

On Tue, Jan 6, 2009 at 10:15 PM, John W. Linville
<linville@tuxdriver.com> wrote:
> On Sun, Dec 28, 2008 at 03:26:55PM +0200, Kalle Valo wrote:
>
>> Also I'm going to rename the mac80211 port to something else for just
>> to avoid the confusion with the original driver. Last thing I want to
>> see is having two in-tree drivers with identical names.
>
> OK, I still have patches to recreate the original port of at76_usb
> to mac80211 and another to keep-up with a later mac80211 API change
> on the at76 branch of wireless-testing.  I had intended to send them
> to Greg now that the mac80211 API change has made it upstream.

I have to disagree here. It's better not to change at76_usb in staging
to avoid any breakage. Like I said, I will submit the mac80211 port
with a new name (most probably at76c50x-usb) as soon as I get it
stable enough. It's good idea to have a working driver in staging tree
until the mac80211 has been verified to be working.

> Since you mentioned a) that you are working on the port again and b)
> you are going to rename the driver for mac80211, should I just drop
> my at76 branch of wireless testing for good?

Please do, because I will submit at7650x-usb patches from scratch. But
don't remove at76 branch from the legacy tree, it's useful for me:

http://git.kernel.org/?p=linux/kernel/git/linville/wireless-legacy.git;a=shortlog;h=at76

Kalle

Re: status of at76-usb driver

[Claudio]

Kalle Valo <kalle.valo@iki.fi> wrote:
> On Tue, Jan 6, 2009 at 10:15 PM, John W. Linville
> <linville@tuxdriver.com> wrote:
>> On Sun, Dec 28, 2008 at 03:26:55PM +0200, Kalle Valo wrote:
>>
>>> Also I\'m going to rename the mac80211 port to something else for just
>>> to avoid the confusion with the original driver. Last thing I want to
>>> see is having two in-tree drivers with identical names.
>>
>> OK, I still have patches to recreate the original port of at76_usb
>> to mac80211 and another to keep-up with a later mac80211 API change
>> on the at76 branch of wireless-testing.  I had intended to send them
>> to Greg now that the mac80211 API change has made it upstream.
>
> I have to disagree here. It\'s better not to change at76_usb in staging
> to avoid any breakage. Like I said, I will submit the mac80211 port
> with a new name (most probably at76c50x-usb) as soon as I get it
> stable enough. It\'s good idea to have a working driver in staging tree
> until the mac80211 has been verified to be working.

I think so too.

>> Since you mentioned a) that you are working on the port again and b)
>> you are going to rename the driver for mac80211, should I just drop
>> my at76 branch of wireless testing for good?
>
> Please do, because I will submit at7650x-usb patches from scratch. But
> don\'t remove at76 branch from the legacy tree, it\'s useful for me:

I am looking forward for testing your new mac80211 version. Please tell
me where the development happens, so that I can contribute to it.

Regarding Ubuntu: They are still shipping an ancient version (0.14beta1!).
Sorry for considering blaming you for the problems. The current version
of the staging-tree/directory(?) works well.

Claudio

Re: status of at76-usb driver

[Kalle Valo]

On Thu, Jan 8, 2009 at 4:56 PM, Claudio <der.claudio@aon.at> wrote:
> Kalle Valo <kalle.valo@iki.fi> wrote:
>> I have to disagree here. It\'s better not to change at76_usb in staging
>> to avoid any breakage. Like I said, I will submit the mac80211 port
>> with a new name (most probably at76c50x-usb) as soon as I get it
>> stable enough. It\'s good idea to have a working driver in staging tree
>> until the mac80211 has been verified to be working.

s/mac80211/mac80211 port/

Sigh, me and my infamous typos.

> I think so too.
>
>>> Since you mentioned a) that you are working on the port again and b)
>>> you are going to rename the driver for mac80211, should I just drop
>>> my at76 branch of wireless testing for good?
>>
>> Please do, because I will submit at7650x-usb patches from scratch. But
>> don\'t remove at76 branch from the legacy tree, it\'s useful for me:
>
> I am looking forward for testing your new mac80211 version. Please tell
> me where the development happens, so that I can contribute to it.

The development is slow :) I will post any patches to here as I have
them. Currently I'm working on usb protocol decoder so that I can
verify the differences between the original and the mac80211 port.

> Regarding Ubuntu: They are still shipping an ancient version (0.14beta1!).
> Sorry for considering blaming you for the problems.

No problem :)

> The current version of the staging-tree/directory(?) works well.

I didn't find any problems, but I only briefly tested it.

Kalle