[Intel-gfx] [PATCH 03/18] drm/i915: Lock the engine while dumping the active request

[Chris Wilson]

Quoting Tvrtko Ursulin (2019-07-11 08:20:59)
> 
> On 11/07/2019 07:52, Chris Wilson wrote:
> > diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c
> > index 5489cd879315..f297a43df1e9 100644
> > --- a/drivers/gpu/drm/i915/i915_gpu_error.c
> > +++ b/drivers/gpu/drm/i915/i915_gpu_error.c
> > @@ -1463,6 +1464,7 @@ static void gem_record_rings(struct i915_gpu_state *error)
> >   
> 
> Aren't there allocations in this block?

They are already atomic due to this being called under stop_machine().

Next patch gives up on stop_machine(), excludes that which is not under
spinlock, and relies on the engine->active.lock for capturing the
request and friends.
-Chris