[Intel-gfx] [PATCH 5/6] drm/i915/uc: move uc_resume under gt_resume

[Chris Wilson]

Quoting Daniele Ceraolo Spurio (2019-07-30 00:47:26)
> intel_uc is part of intel_gt so it makes logical sense for it to be
> resumed as part of it. Note that, since gt_resume is also called during
> the init flow, a state variable has been added to intel_uc to avoid
> asking an already running GuC to resume.
> 
> Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio at intel.com>
> Cc: Michal Wajdeczko <michal.wajdeczko at intel.com>
> Cc: Chris Wilson <chris at chris-wilson.co.uk>
> ---
>  drivers/gpu/drm/i915/gem/i915_gem_pm.c |  2 --
>  drivers/gpu/drm/i915/gt/intel_gt_pm.c  |  3 +++
>  drivers/gpu/drm/i915/gt/uc/intel_uc.c  | 10 ++++++++--
>  drivers/gpu/drm/i915/gt/uc/intel_uc.h  |  2 ++
>  4 files changed, 13 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_pm.c b/drivers/gpu/drm/i915/gem/i915_gem_pm.c
> index b5561cbdc5ea..25610de3961b 100644
> --- a/drivers/gpu/drm/i915/gem/i915_gem_pm.c
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_pm.c
> @@ -265,8 +265,6 @@ void i915_gem_resume(struct drm_i915_private *i915)
>         if (intel_gt_resume(&i915->gt))
>                 goto err_wedged;
>  
> -       intel_uc_resume(&i915->gt.uc);
> -
>         /* Always reload a context for powersaving. */
>         if (!i915_gem_load_power_context(i915))
>                 goto err_wedged;

This sequence itself does not belong to i915_gem_resume() and needs to
be lifted to intel_gt_resume (that is establishing the baseline GT power
context). So I would rather postpone this to try and avoid having to
introduce bool suspended if at all possible.

Another rule of thumb to consider is that we should be able to throw gt
initialisation into an async task (and I'm considering an async task
per engine, although for a large part we can achieve asynchronicity via
HW queues).
-Chris