Re: [PATCH] soundwire: stream: fix programming slave ports for non-continous port maps

[Krzysztof Kozlowski]

On 03/09/2024 17:17, Liao, Bard wrote:

>>>
>>> then dpn_prop[0].num = 1 and dpn_prop[1].num = 3. And we need to go
>>>
>>> throuth dpn_prop[0] and dpn_prop[1] instead of dpn_prop[1] and
>> dpn_prop[3].
>>>
>>
>> What are the source or sink ports in your case? Maybe you just generate
>> wrong mask?
> 
> I checked my mask is 0xa when I do aplay and it matches the sink_ports of
> the rt722 codec.
> 
>>
>> It's not only my patch which uses for_each_set_bit(). sysfs_slave_dpn
>> does the same.
> 
> What sysfs_slave_dpn does is 
>         i = 0;                          
>         for_each_set_bit(bit, &mask, 32) {
>                 if (bit == N) {
>                         return sprintf(buf, format_string,
>                                        dpn[i].field);
>                 }
>                 i++;
>         }                         
> It uses a variable "i" to represent the array index of dpn[i].
> But, it is for_each_set_bit(i, &mask, 32) in your patch and the variable "i"
> which represents each bit of the mask is used as the index of dpn_prop[i].
> 
> Again, the point is that the bits of mask is not the index of the dpn_prop[]
> array.

Yes, you are right. I think I cannot achieve my initial goal of using
same dpn array with different indices. My patch should be reverted, I
believe.

I'll send a revert, sorry for the mess.

Best regards,
Krzysztof