On Sep 3 2016 01:05, Takashi Iwai wrote:
But, the addition of the flag is helpful alone. Could you concentrate on that and resubmit?
Before starting it, I'd like to wait for comments from Clemens Ladisch, Charles Keepax and Vinod Koul, in a view of application interfaces stable, safe and tolerant for long term usage. I'm unwilling to use my time more for light-minded work done to bring confusions to hardware- abstraction layers for user land applications.
But, if possible, I'd request you to re-think about my proposal to return processed byte of TLV packet payload to user land, again. It looks to be intrusive, but in my taste, it doesn't bring much impacts to applications, than adding new flags to kernel/userspace interfaces.
The length of processed bytes from drivers in ALSA SoC part might help developers to get current value of 'struct soc_bytes_ext.max' of each control element set. As of kernel 4.8, soc-da7218/nau8825/wm5102 modules implements the coefficients, and it's already in ALSA topology design. I believe it better to assist developers for the feature, than judging it's coarse and abuse for good APIs.
Regards
Takashi Sakamoto