Mark Brown wrote:
That'd be Liam in the multi-component conversion, though obviously that didn't cause any immediate issues. The shared I/O code has been present since mid 2009 though, including the cache management. This kind of comes back to what I'm saying about making your code as idiomatic as possible, the more a given piece of code diverges from standard idioms the more likely it is that it will be unintentially broken by some other change.
My code *was* idiomatic the last time I touched it, which was early 2009.
FWIW using the standard stuff should just be a case providing defaults, calling snd_soc_set_cache_io() and removing your custom I/O functions.
Ok, I'll take a look at it tomorrow.
You'd think the guys at Crystal Semi would at least give me a free coffee mug or something.