Re: [Intel-gfx] [CI 1/2] drm/i915/step: Add macro magic for handling steps

[Jani Nikula <jani.nikula@linux.intel.com>]

On Mon, 19 Jul 2021, Anusha Srivatsa <anusha.srivatsa@intel.com> wrote:
> With the addition of stepping info for
> all platforms, lets use macros for handling them
> and autogenerating code for all steps at a time.
>
> Suggested-by: Matt Roper <matthew.d.roper@intel.com>
> Cc: Lucas De Marchi <lucas.demarchi@intel.com>
> Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
> Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
> ---
>  drivers/gpu/drm/i915/intel_step.c | 14 ++++++++++++
>  drivers/gpu/drm/i915/intel_step.h | 37 +++++++++++++++++++------------
>  2 files changed, 37 insertions(+), 14 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/intel_step.c b/drivers/gpu/drm/i915/intel_step.c
> index 9fcf17708cc8..d150d138e889 100644
> --- a/drivers/gpu/drm/i915/intel_step.c
> +++ b/drivers/gpu/drm/i915/intel_step.c
> @@ -182,3 +182,17 @@ void intel_step_init(struct drm_i915_private *i915)
>  
>  	RUNTIME_INFO(i915)->step = step;
>  }
> +
> +#define STEP_NAME_CASE(name)	\
> +	(case STEP_##name:	\
> +		return #name)
> +
> +const char *intel_step_name(enum intel_step step)
> +{
> +	switch (step) {
> +	STEP_NAME_LIST(STEP_NAME_CASE);
> +
> +	default:
> +		return "**";
> +	}
> +}
> diff --git a/drivers/gpu/drm/i915/intel_step.h b/drivers/gpu/drm/i915/intel_step.h
> index 88a77159703e..f6641e2a3c77 100644
> --- a/drivers/gpu/drm/i915/intel_step.h
> +++ b/drivers/gpu/drm/i915/intel_step.h
> @@ -15,30 +15,39 @@ struct intel_step_info {
>  	u8 display_step;
>  };
>  
> +#define STEP_ENUM_VAL(name)  STEP_##name,
> +
> +#define STEP_NAME_LIST(func)		\
> +	func(A0)			\
> +	func(A1)			\
> +	func(A2)			\
> +	func(B0)			\
> +	func(B1)			\
> +	func(B2)			\
> +	func(C0)			\
> +	func(C1)			\
> +	func(D0)			\
> +	func(D1)			\
> +	func(E0)			\
> +	func(F0)			\
> +	func(G0)			\
> +	func(H0)			\
> +	func(I0)			\
> +	func(I1)			\
> +	func(J0)
> +
>  /*
>   * Symbolic steppings that do not match the hardware. These are valid both as gt
>   * and display steppings as symbolic names.
>   */
>  enum intel_step {
>  	STEP_NONE = 0,
> -	STEP_A0,
> -	STEP_A2,
> -	STEP_B0,
> -	STEP_B1,
> -	STEP_C0,
> -	STEP_D0,
> -	STEP_D1,
> -	STEP_E0,
> -	STEP_F0,
> -	STEP_G0,
> -	STEP_H0,
> -	STEP_I0,
> -	STEP_I1,
> -	STEP_J0,
> +	STEP_NAME_LIST(STEP_ENUM_VAL)
>  	STEP_FUTURE,
>  	STEP_FOREVER,
>  };

It's been merged already, and so be it.

I'll just note that the main reason I don't like adding this kind of
macro abstractions is that it throws off source code tagging tools like
ctags and gnu global. I can no longer navigate the definition and all
the references of STEP_* using the tools, and it degenerates to git grep
which is much worse as it doesn't know about C syntax or semantics. It's
not always about how the code looks, it's also about how people can
navigate the code base.

BR,
Jani.


>  
>  void intel_step_init(struct drm_i915_private *i915);
> +const char *intel_step_name(enum intel_step step);
>  
>  #endif /* __INTEL_STEP_H__ */

-- 
Jani Nikula, Intel Open Source Graphics Center