Al 05/03/11 22:04, En/na Luca Olivetti ha escrit:
> Al 05/03/11 21:25, En/na John Crispin ha escrit:
>> On 05/03/11 21:20, Luca Olivetti wrote:
>>> Al 05/03/11 21:16, En/na John Crispin ha escrit:
>>>
>>>>> Ok, I have to tell it where the firmware files are on the command line:
>>>>>
>>>>> root@OpenWrt:/lib/firmware# tapidemo  --path /lib/firmware/
>>>>> TAPI Demo Version 5.0.1.27
>>>>> VAD is set to: IFX_TAPI_ENC_VAD_NOVAD - no voice activity detection.
>>>>> Err with ioctl cmd SIOCGIFADDR Cannot assign requested address. (File: 
>>>>> ./tapidemo.c, line: 3657)
>>>>> CompileIFX_MPS: CPU1 base address is invalid!
>>>>> IFX_MPS: firmware download error (-1)!
>>>>> d for EASY50712.
>>>>> Used board(s) EASY50712.
>>>>> Filesystem firmware binary used (/lib/firmware/danube_firmware.bin).
>>>>> Filesystem BBD binary used (/lib/firmware/danube_bbd_fxs.bin).
>>>>> Err, Failed download FW. (File: ./device_vmmc.c, line: 718)
>>>>> Err, Channel 0 is not initialized. (File: ./board_easy50712.c, line: 563)
>>>>> Err, Board Initialization failed (Idx 0). (File: ./tapidemo.c, line: 4331)
>>>>>
>>>>>
>>>>> Bye
>>>>
>>>>
>>>> hi,
>>>>
>>>> what openwrt revision are you using ?
>>>
>>> 25626
>>>
>>> Bye
>>
>> Hi,
>>
>> can you try if it works if ath9k is not loaded please
> 
> Just unloading ath modules: no difference.
> Rebuilding the image with the "arv7518_register_ath9k" call removed: no 
> difference.
> 
> Isn't it due to cp1_base assignment being commented out in 
> arch/mips/lantiq/setup.c?

As already discussed privately, the problem was due to a missing initialization
of cp1_base (a casualty of the switch from ifxmips to lantiq), which apparently 
is the memory address reserved for the firmware.
I moved the function ifx_get_cp1_base from setup.c to devices.c, returning the 
address
assigned in lq_register_tapi.
I also added the call to lq_register_tapi (which also went missing in the 
switch from
ifxmips to lantiq) and now tapidemo works.
Nothing to do with ath9k (which in spite of the lack of help it works, even if 
it
has hideous performance).
Will post patches once the pending ones are applied.

Bye
-- 
Luca
_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel

Reply via email to