Al 07/04/11 14:16, En/na Luca Olivetti ha escrit:
> Al 04/04/2011 20:21, En/na Luca Olivetti ha escrit:
>> Al 04/04/11 20:17, En/na Luca Olivetti ha escrit:
>>> Al 04/04/11 19:57, En/na Felix Fietkau ha escrit:
>>>> On 2011-04-04 7:51 PM, Luca Olivetti wrote:
>>>>> Al 04/04/11 19:48, En/na Luca Olivetti ha escrit:
>>>>>>   The following patch adds ath9k support to the arv7518 board.
>>>>>
>>>>> I forgot to say that it needs this patch in mac80211
>>>>>
>>>>> http://patchwork.midlink.org/patch/727/
>>>>>
>>>>> without it, initialization of the card would fail (no other harm should be
>>>>> done).
>>>> I think we should make the endian check unconditional in ath9k. No need to 
>>>> add another field to ath9k_platform.h
>>>
>>> Well, I cannot say, I just wanted to avoid to breaking working devices.
>>
>> Which I probably did (since I didn't patch, e.g., ar71xx, to add the field) 
>> :(
> 
> Well, what's the final decision? The new field or the unconditional check? 
> (And why atheros made it conditional in the first place?).
> Also, I checked and the endianness of the various ar71xx devices using ath9k 
> should be the same as mine, so I don't understand why it works on those 
> devices (does it?) and it doesn't on mine.
> Note that, in the fixup code, the same magic as the ar71xx (0xa55a) works, 
> while I had to change
> 
> __raw_writel(val, mem + reg);
> 
> to
> 
> __raw_writel(cpu_to_le32(val), mem + reg);
> 
> I'm quite confused.
> 
> Next we can discuss the performance problem (maybe the root cause is the 
> same, e.g. I'm doing something stupid without realizing).

No decision yet?

Bye
-- 
Luca
_______________________________________________
openwrt-devel mailing list
[email protected]
https://lists.openwrt.org/mailman/listinfo/openwrt-devel

Reply via email to