Wm3236AQ - AR6K kernel Driver

[Arnaud Derasse]

Hello,

I've just gave a look at the qemu ar6k file. This file could help me to 
understand the ar6k behavior. It's a good thing, because we do not have 
any documentation.

I am using the last driver from you SVN repository.
If I check in the AR6k_VERSION.h file, the version number are the good 
ones :
#define __VER_MAJOR_ 2
#define __VER_MINOR_ 0
#define __VER_PATCH_ 0

The question is : Why do the AR6K respond with 0xA instead of 0xC ...
-> Target Info Byte Count : 0xA, host is 12

I receive the following from the AR6K :
0xFFFFFFFF 0x0000000A 0x40000039 0x00000001

0x40000039 correspond to Major version 2, Build number 57. This is the 
good version number according to Accton people who send us the chips.

The big problem we have is after , when the BMI is not responding anymmore.

Regards,

Arnaud Derasse

andrzej zaborowski a écrit :
> Hi,
>
> 2008/7/10 Arnaud Derasse <arnaud.derasse at wifidget.com>:
>   
>> We have some difficulties to make the linux driver working. Since Accton
>> advised us to use your driver at
>> http://svn.openmoko.org/developers/sameo/patches/ar6k-atheros-2.0/2.6.24/
>> I prefer ask questions on this mailing list.
>>
>> Well, : We are using the last driver for WM3236AQ with Firwmare 2.0. The
>> driver initialisation fails first in the BMITargetInfo function in bmi.c
>>
>> ar6000_available
>> BMI Write : Ask for Target Info
>> BMI Read : Read Target Version
>> Target Info Version : 0xFFFFFFFF
>> BMI Read : Read Target info bytecount
>> Target Info Byte Count : 0xA, host is 12
>> /home/wifidget/Work/Mx27ADS/kernels/linux-2.6-mx/drivers/sdio/function/wlan/ar6000/bmi/bmi.c:105:
>> Assertion targ_info->target_info_byte_count == sizeof(*targ_info) failed!
>> Kernel panic - not syncing: targ_info->target_info_byte_count
>> ==sizeof(*targ_info)
>>     
>
> It looks like the driver you're trying to use is for the older
> firmware though. The info above is correct for the newer firmware (I
> don't remember the exact number) but the error message suggests that
> the driver is expecting the ReadTargetVersion response in the older
> format.
>
> If I'm judging correctly from code,
> the older firmware returns one word: 0x11000044
> while new firmware returns: 0xFFFFFFFF 0x0000000c 0x11000044
> (First word is the guard value, second word is the bytecount and the
> third word is the version ID)
>
> This can be seen in the qemu-neo973 emulator code in the file
> hw/ar6000.c at svn.openmoko.org. I cannot look up the URL because I
> don't have access to a computer at the moment, other than a handheld.
>
> Regards
>