[Intel-gfx] [PATCH 01/10] drm/i915/guc: Precompute GuC shared data offset

[Michal Wajdeczko]

On Thu, 05 Oct 2017 11:13:40 +0200, Michał Winiarski  
<michal.winiarski at intel.com> wrote:

> We're using first page of kernel context state to share data with GuC,
> let's precompute the ggtt offset at GuC initialization time rather than
> everytime we're using GuC actions.
>
> Cc: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Jeff Mcgee <jeff.mcgee at intel.com>
> Cc: Michal Wajdeczko <michal.wajdeczko at intel.com>
> Cc: Oscar Mateo <oscar.mateo at intel.com>
> Signed-off-by: Michał Winiarski <michal.winiarski at intel.com>
> ---
>  drivers/gpu/drm/i915/i915_guc_submission.c | 4 ++--
>  drivers/gpu/drm/i915/intel_uc.c            | 4 ++++
>  drivers/gpu/drm/i915/intel_uc.h            | 3 +++
>  3 files changed, 9 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_guc_submission.c  
> b/drivers/gpu/drm/i915/i915_guc_submission.c
> index 04f1281d81a5..2c0aeee3143d 100644
> --- a/drivers/gpu/drm/i915/i915_guc_submission.c
> +++ b/drivers/gpu/drm/i915/i915_guc_submission.c
> @@ -1234,7 +1234,7 @@ int intel_guc_suspend(struct drm_i915_private  
> *dev_priv)
>  	/* any value greater than GUC_POWER_D0 */
>  	data[1] = GUC_POWER_D1;
>  	/* first page is shared data with GuC */

Please drop this now redundant comment

> -	data[2] = guc_ggtt_offset(ctx->engine[RCS].state) + LRC_GUCSHR_PN *  
> PAGE_SIZE;
> +	data[2] = guc->shared_data_offset;
> 	return intel_guc_send(guc, data, ARRAY_SIZE(data));
>  }
> @@ -1260,7 +1260,7 @@ int intel_guc_resume(struct drm_i915_private  
> *dev_priv)
>  	data[0] = INTEL_GUC_ACTION_EXIT_S_STATE;
>  	data[1] = GUC_POWER_D0;
>  	/* first page is shared data with GuC */

Please drop this now redundant comment

> -	data[2] = guc_ggtt_offset(ctx->engine[RCS].state) + LRC_GUCSHR_PN *  
> PAGE_SIZE;
> +	data[2] = guc->shared_data_offset;
> 	return intel_guc_send(guc, data, ARRAY_SIZE(data));
>  }
> diff --git a/drivers/gpu/drm/i915/intel_uc.c  
> b/drivers/gpu/drm/i915/intel_uc.c
> index e7875277ba97..f4893c85e54a 100644
> --- a/drivers/gpu/drm/i915/intel_uc.c
> +++ b/drivers/gpu/drm/i915/intel_uc.c
> @@ -173,6 +173,10 @@ static void guc_free_load_err_log(struct intel_guc  
> *guc)
>  static int guc_enable_communication(struct intel_guc *guc)
>  {
>  	struct drm_i915_private *dev_priv = guc_to_i915(guc);
> +	struct i915_gem_context *ctx = dev_priv->kernel_context;
> +
> +	guc->shared_data_offset = guc_ggtt_offset(ctx->engine[RCS].state) +
> +		LRC_GUCSHR_PN * PAGE_SIZE;

Hmm, I'm not sure that this is the best place for such initialization.
What about doing it directly in intel_uc_init_hw() or in dedicated inline ?

> 	if (HAS_GUC_CT(dev_priv))
>  		return intel_guc_enable_ct(guc);
> diff --git a/drivers/gpu/drm/i915/intel_uc.h  
> b/drivers/gpu/drm/i915/intel_uc.h
> index 4fa091e90b5f..10e8f0ed02e4 100644
> --- a/drivers/gpu/drm/i915/intel_uc.h
> +++ b/drivers/gpu/drm/i915/intel_uc.h
> @@ -121,6 +121,9 @@ struct intel_guc {
>  	/* To serialize the intel_guc_send actions */
>  	struct mutex send_mutex;
> +	/* Kernel context state ggtt offset, first page is shared with GuC */
> +	u32 shared_data_offset;

We want to keep all 'send' related members together.
Maybe better place for this new member will be after 'execbuf_client' ?

> +
>  	/* GuC's FW specific send function */
>  	int (*send)(struct intel_guc *guc, const u32 *data, u32 len);