[Intel-gfx] [PATCH] drm/i915/gt: Prune inlines

[Jani Nikula]

On Wed, 13 Jan 2021, Chris Wilson <chris at chris-wilson.co.uk> wrote:
> Remove all the manual inlines from non-critical sections in gt/
>
> add/remove: 2/0 grow/shrink: 0/3 up/down: 762/-1473 (-711)
> Function                                     old     new   delta
> mi_set_context.isra                            -     602    +602
> write_dma_entry                                -     160    +160
> __set_pd_entry                               214      69    -145
> clear_pd_entry                               190      42    -148
> ring_request_alloc                          2021     841   -1180
> Total: Before=1605086, After=1604375, chg -0.04%
>
> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Jani Nikula <jani.nikula at intel.com>

> @@ -98,13 +98,13 @@ static inline void __timeline_mark_unlock(struct intel_context *ce,
>  
>  #else
>  
> -static inline unsigned long __timeline_mark_lock(struct intel_context *ce)
> +static unsigned long __timeline_mark_lock(struct intel_context *ce)
>  {
>  	return 0;
>  }
>  
> -static inline void __timeline_mark_unlock(struct intel_context *ce,
> -					  unsigned long flags)
> +static void __timeline_mark_unlock(struct intel_context *ce,
> +				   unsigned long flags)
>  {
>  }

I don't mind inlines for stubs, but probably doesn't make a difference.

> @@ -113,9 +114,10 @@ clear_pd_entry(struct i915_page_directory * const pd,
>  {
>  	GEM_BUG_ON(atomic_read(px_used(pd)) == 0);
>  
> -	write_dma_entry(px_base(pd), idx, scratch->encode);
> -	pd->entry[idx] = NULL;
>  	atomic_dec(px_used(pd));
> +	pd->entry[idx] = NULL;
> +
> +	write_dma_entry(px_base(pd), idx, scratch->encode);

With the changes in this hunk removed,

Reviewed-by: Jani Nikula <jani.nikula at intel.com>


-- 
Jani Nikula, Intel Open Source Graphics Center