On Sat, Jul 26, 2014 at 09:01:55AM +0100, Chris Wilson wrote:
> In i915, we have a big mutex around our device struct - every time before
> we attempt to communicate with the GPU, we acquire the mutex. This makes
> it a convenient juncture to place our GPU error handling - before we take
> the mute
In i915, we have a big mutex around our device struct - every time before
we attempt to communicate with the GPU, we acquire the mutex. This makes
it a convenient juncture to place our GPU error handling - before we take
the mutex we first check whether the GPU is hung or whether we are in
the proc