Hannu I think.. it was one of Hannu's changes that touched all this. It
may be a merge error.
> Alan, do you believe me now when I wrote the current
> sound driver API is way too complex? 8)
I know it is, but we are stuck with it 8). A lot of it actually is the
grown rather than designed implementation of said API.
> NB: the sound driver has a vmalloc/vfree in the ioctl
> code path - not exactly what you want for speed, especially
> as there are quite speed critical ioctls (GETxPTR and TRIGGER).
> Anyone any objections against removing those and using
> put_user/get_user and friends in the ioctl routines?
If there are no atomicity issues go ahead
Alan