Re: [PATCH 4/5] drm/i915/gvt: Change for_each_pipe to use pipe_mask API

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

On Mon, 15 Dec 2025, Ankit Nautiyal <ankit.k.nautiyal@intel.com> wrote:
> Add a new API to get pipe_mask from DISPLAY_RUNTIME_INFO() for GVT.
> Update the for_each_pipe() macro in GVT to call this API, instead of
> accessing DISPLAY_RUNTIME_INFO()->pipe_mask directly.
>
> This keeps the macro usable in GVT without exposing display internals
> and prepares for display modularization.
>
> Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
> ---
>  drivers/gpu/drm/i915/display/intel_gvt_api.c | 6 ++++++
>  drivers/gpu/drm/i915/display/intel_gvt_api.h | 1 +
>  drivers/gpu/drm/i915/gvt/display_helpers.h   | 7 +++++++
>  3 files changed, 14 insertions(+)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_gvt_api.c b/drivers/gpu/drm/i915/display/intel_gvt_api.c
> index 8abea318fbc2..0b09bbf2c29a 100644
> --- a/drivers/gpu/drm/i915/display/intel_gvt_api.c
> +++ b/drivers/gpu/drm/i915/display/intel_gvt_api.c
> @@ -32,3 +32,9 @@ u32 intel_display_device_mmio_base(struct intel_display *display)
>  	return DISPLAY_MMIO_BASE(display);
>  }
>  EXPORT_SYMBOL_GPL(intel_display_device_mmio_base);
> +
> +u8 intel_display_runtime_info_pipe_mask(struct intel_display *display)
> +{
> +	return DISPLAY_RUNTIME_INFO(display)->pipe_mask;
> +}

I don't think gvt needs to know it's about "runtime info". Maybe make it
just intel_display_device_pipe_mask()?

Though I'm also wondering about making it even more abstracted with
something like intel_display_device_pipe_valid(), and using that for the
various other cases that check pipes in GVT. But maybe the patch at hand
is a good start.

> +EXPORT_SYMBOL_GPL(intel_display_runtime_info_pipe_mask);
> diff --git a/drivers/gpu/drm/i915/display/intel_gvt_api.h b/drivers/gpu/drm/i915/display/intel_gvt_api.h
> index e9a1122a988d..8ceda30a969b 100644
> --- a/drivers/gpu/drm/i915/display/intel_gvt_api.h
> +++ b/drivers/gpu/drm/i915/display/intel_gvt_api.h
> @@ -16,5 +16,6 @@ u32 intel_display_device_pipe_offset(struct intel_display *display, enum pipe pi
>  u32 intel_display_device_trans_offset(struct intel_display *display, enum transcoder trans);
>  u32 intel_display_device_cursor_offset(struct intel_display *display, enum pipe pipe);
>  u32 intel_display_device_mmio_base(struct intel_display *display);
> +u8 intel_display_runtime_info_pipe_mask(struct intel_display *display);
>  
>  #endif /* __INTEL_GVT_API_H__ */
> diff --git a/drivers/gpu/drm/i915/gvt/display_helpers.h b/drivers/gpu/drm/i915/gvt/display_helpers.h
> index 6f68a1e8751a..d11ebb03b946 100644
> --- a/drivers/gpu/drm/i915/gvt/display_helpers.h
> +++ b/drivers/gpu/drm/i915/gvt/display_helpers.h
> @@ -36,4 +36,11 @@ struct display;
>  #define INTEL_DISPLAY_DEVICE_CURSOR_OFFSET(display, pipe) \
>  	intel_display_device_cursor_offset((display), (pipe))
>  
> +#ifdef for_each_pipe

Ditto about ifdefs here as with previous patch.

> +#undef for_each_pipe
> +#endif
> +#define for_each_pipe(display, __p) \
> +	for ((__p) = 0; (__p) < I915_MAX_PIPES; (__p)++) \
> +		for_each_if(intel_display_runtime_info_pipe_mask((display)) & BIT(__p))
> +
>  #endif /* __DISPLAY_HELPERS_H__ */

-- 
Jani Nikula, Intel