[Intel-gfx] [PATCH 15/40] drm/i915: Priority boost for new clients

[Chris Wilson]

Quoting Tvrtko Ursulin (2018-09-25 09:57:11)
> 
> On 25/09/2018 09:26, Chris Wilson wrote:
> > Quoting Chris Wilson (2018-09-25 09:01:06)
> >> Quoting Tvrtko Ursulin (2018-09-24 11:29:52)
> >>>
> >>> On 19/09/2018 20:55, Chris Wilson wrote:
> >>>> diff --git a/drivers/gpu/drm/i915/i915_scheduler.h b/drivers/gpu/drm/i915/i915_scheduler.h
> >>>> index 7edfad0abfd7..93e43e263d8c 100644
> >>>> --- a/drivers/gpu/drm/i915/i915_scheduler.h
> >>>> +++ b/drivers/gpu/drm/i915/i915_scheduler.h
> >>>> @@ -19,12 +19,14 @@ enum {
> >>>>        I915_PRIORITY_INVALID = INT_MIN
> >>>>    };
> >>>>    
> >>>> -#define I915_USER_PRIORITY_SHIFT 0
> >>>> +#define I915_USER_PRIORITY_SHIFT 1
> >>>>    #define I915_USER_PRIORITY(x) ((x) << I915_USER_PRIORITY_SHIFT)
> >>>>    
> >>>>    #define I915_PRIORITY_COUNT BIT(I915_USER_PRIORITY_SHIFT)
> >>>>    #define I915_PRIORITY_MASK (-I915_PRIORITY_COUNT)
> >>>>    
> >>>> +#define I915_PRIORITY_NEWCLIENT      ((u8)BIT(0))
> >>>
> >>> Is the cast important and why?
> >>
> >> Unreliable memory says there was something iffy about the code generation
> >> at one point.
> > 
> > drivers/gpu/drm/i915/intel_lrc.c: In function ‘__unwind_incomplete_requests’:
> > drivers/gpu/drm/i915/intel_lrc.c:272:13: error: overflow in conversion from ‘long unsigned int’ to ‘int’ changes value from ‘18446744071562067969’ to ‘-2147483647’ [-Werror=overflow]
> >    int prio = I915_PRIORITY_INVALID | I915_PRIORITY_NEWCLIENT;
> 
> So correct cast would be (int)BIT(..), or maybe not use BIT for less 
> confusion?

It's a bit, I like them unsigned to avoid sign extension confusion most
of the time. (What am I saying, sign extension is already confusing and
no matter what you do, you always want the opposite.)
-Chris