newbus bus access routines and bus_space_barrier()

[Adrian Chadd]

On 16 October 2011 11:45, Warner Losh <imp at bsdimp.com> wrote:
> Usually they are needed, but we get away without them often because they are needed in a limited set of circumstances and we have memory barriers in our locking primitives.

Right. I'll just change the ath driver register ops to enforce a
read/write barrier on each read or write.
It'd be nice to teach the driver and the HAL about "correct" barriers
but I just don't have the time. :(

(FYI: apparently linux's ioread32/iowrite32 operations enforce
barriers on each op.)



Adrian