at76c503a and Kernel 2.6.14

Hm,

did I miss an announcement? at76c503a does not build against 2.6.14 due
to minor changes in a) the usb stack (URB_ASYNC_UNLINK is now obsolete
as far as I can see) and b) the wireless extensions (spy_offset is
gone). The patch attached "works-for-me" by a) defining that constant 
to 0 and b) stealing the changes of wl3501_cs.c.

However I'd be glad if anybody with more insight into the driver could
take a look into this.

    Christoph

--- drivers/net/wireless/at76c503/at76c503.c	2005-08-29 21:43:05.000000000 +0200
+++ drivers/net/wireless/at76c503/at76c503.c	2005-11-10 00:46:47.000000000 +0100
 <at>  <at>  -205,7 +205,7  <at>  <at> 
 #endif

 #ifndef USB_ASYNC_UNLINK
-#define USB_ASYNC_UNLINK	URB_ASYNC_UNLINK
+#define USB_ASYNC_UNLINK	0
 #endif

 #ifndef FILL_BULK_URB
 <at>  <at>  -6699,10 +6699,7  <at>  <at> 
 	.standard	= (iw_handler *) at76c503_handlers,
 	.private	= (iw_handler *) at76c503_priv_handlers,
 	.private_args	= (struct iw_priv_args *) at76c503_priv_args,
-#if WIRELESS_EXT > 15

Re: at76c503a and Kernel 2.6.14

	I was wondering about this as well but put off my upgrade to 2.6.14 because of the compiliaton troubles. 
Thanks for the patch, hopefully it will be integrated :)

You wrote:

> Hm,
> 
> did I miss an announcement? at76c503a does not build against 2.6.14 due
> to minor changes in a) the usb stack (URB_ASYNC_UNLINK is now obsolete
> as far as I can see) and b) the wireless extensions (spy_offset is
> gone). The patch attached "works-for-me" by a) defining that constant 
> to 0 and b) stealing the changes of wl3501_cs.c.
> 
> However I'd be glad if anybody with more insight into the driver could
> take a look into this.
> 
>     Christoph

Re: at76c503a and Kernel 2.6.14

Linux 2.6.14 + at76c503a CVS (no patch) work almost fine for me.

The trick was to enable external firmware loading.  (udev, and hotplug)  
Here's how I install the firmware on my distro:
http://members.shaw.ca/nick.steeves/ports/atmel-firmware/Pkgfile

I forget where on the website it is, but one needs to disable the loading of 
built-in firmware, in the at76c503a CVS source.  I think one just needs to 
remove a few files...

The only problems I have are these:

1. Hotplug doesn't properly upload the firmware whenever I reboot.
2. Coldplug causes my kernel to OOPS. (thus I don't use it)
3. So I have to plug/replug my WUSB11 v2.6 each time I reboot, or else just 
run a script of mine to unload the at76c503 modules, and then reload them, 
and then bring the interface up.

So, does the patch which started this thread work for everyone?  Does this 
driver need the "generic 802.11 network stack" thingy for Linux kernel 
2.6.14?

Cheers,
Nick

On November 9, 2005 5:42, Christoph Biedl wrote:
> Hm,
>
> did I miss an announcement? at76c503a does not build against 2.6.14 due
> to minor changes in a) the usb stack (URB_ASYNC_UNLINK is now obsolete

Re: at76c503a and Kernel 2.6.14

Nick Steeves wrote...

> So, does the patch which started this thread work for everyone?

Everybody out there: Please test 

For the records, I'm using an USB adapter:
| ID 069a:0321 Askey Computer Corp. Dynalink WLL013 / Compex WLU11A 802.11b Adapter
which is labeled as "Siemens Gigaset USB Adapter 11".

> Does this driver need the "generic 802.11 network stack" thingy for
> Linux kernel 2.6.14?

No, as at76c503a was written before that stuff.  I've just tested it to
be on the safe side: "generic 802.11 network stack" disabled and
re-built the kernel, works.  However it seems to be a good idea to merge
the driver into that.  How are the chances to get this driver into the
main stream kernel anyway?

And of course my patch is nothing but a quick hack, it should take care
of the kernel version to have no change when building against 2.6.13.x
and earlier.

    Christoph

Re: at76c503a and Kernel 2.6.14

On November 10, 2005 4:08, Christoph Biedl wrote:
> Everybody out there: Please test 
It doesn't work for my:
Bus 001 Device 009: ID 077b:2219 Linksys WUSB11 V2.6 802.11b Adapter

with Linux-2.6.14.2.

Syslog says:
Nov 12 16:54:48 DigitalMercury kernel: /root/at76c503a/at76c503.c: wlan0: 
mgmt, but not beacon, subtype = 0
Nov 12 16:54:50 DigitalMercury 
kernel: /root/at76c503a/at76c503.c: /root/at76c503a/at76c503.c:2950 assertion 
dev->istate == INIT failed
Nov 12 16:54:51 DigitalMercury kernel: /root/at76c503a/at76c503.c: wlan0: 
set_mib: wait_completion failed with 2
Nov 12 16:54:51 DigitalMercury kernel: /root/at76c503a/at76c503.c: wlan0: 
set_mib (frag threshold) failed: -156
Nov 12 16:54:51 DigitalMercury kernel: /root/at76c503a/at76c503.c: wlan0: 
handle_mgmt_timeout_scan: Bad scan status: CMD_STATUS_UNKNOWN

Out of curiosity, why did your patch patch against a kernel tree?  Did you use 
that kernel_patch.sh script?  (I modified your patch to patch against my 
at76c503a dir)  I should mention that I'm using ad-hoc mode, with "Security 
mode:restricted", and 104bit WEP key.  (apparently 128bit, is really 104...)

> How are the chances to get this driver into the
> main stream kernel anyway?
I think that we would need a permanent maintainer to always make sure that our 
USB driver compiles with the latest kernel.  Interestingly enough, there is 
already at76c50{3,5} stuff in the kernel for PCI, and PCMCIA.  Both need the 

Re: at76c503a and Kernel 2.6.14

	Been short on time... will try to test before too long.

You wrote:

> Everybody out there: Please test 

Re: at76c503a and Kernel 2.6.14

Nick Steeves wrote...

> On November 10, 2005 4:08, Christoph Biedl wrote:
> > Everybody out there: Please test 
> It doesn't work for my:
> Bus 001 Device 009: ID 077b:2219 Linksys WUSB11 V2.6 802.11b Adapter
> 
> with Linux-2.6.14.2.

Too bad 

You're using WEP? I don't - I do not trust WEP and use a tunnel in
software, actually openvpn.

> Out of curiosity, why did your patch patch against a kernel tree?  Did you use 
> that kernel_patch.sh script?

I did the "make kernel_patch" that's included in the at76c503a source
since I prefer to have additional modules in my local kernel tree. The
patch indeed was against that one.

> > How are the chances to get this driver into the
> > main stream kernel anyway?
> I think that we would need a permanent maintainer to always make sure that our 
> USB driver compiles with the latest kernel.

Hm, is this project abandoned? I noticed that
http://at76c503a.berlios.de/ is empty which is at least confusing.
However I was very glad about this alternative to the atmelwlandriver
which was just a pain.