Re: [PATCH 0/7] Prepare GVT for display modularization

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

On Thu, 18 Dec 2025, Ankit Nautiyal <ankit.k.nautiyal@intel.com> wrote:
> GVT currently relies on display internals through register macros and
> helpers like for_each_pipe(). This tight coupling makes modularization
> difficult because GVT should not access struct intel_display directly.
> Add an API for GVT code to expose DISPLAY_RUNTIME_INFO()->pipe_mask.
> This series introduces changes to make GVT independent of display internals
> while keeping existing macros usable:
>
> - Abstract offset calculations in display using
>   INTEL_DISPLAY_DEVICE_*_OFFSET() macros.
> - Add APIs for GVT to compute offsets and pipe mask via functions.
> - Update GVT to use these APIs by overriding helper macros and
>   for_each_pipe().

Oh, this doesn't handle intel_gvt_mmio_table.c, which is part of i915.

That can be a follow-up, I guess, but still needs to be addressed.

BR,
Jani.

>
> Rev2:
> - Remove conflicting headers and get rid of #ifdefs/#undefs in last
>   patch.
> - Wrap macro arguments in paranthesis.
> - Rename for_each_pipe to gvt_for_each_pipe.
>
> PS: I have not yet addressed the question about whether we need to start
>     using _MMIO_TRANS2() instead of_MMIO_PIPE2() in TRANSCONF() macro.
>     That likely needs a separate patch and discussion.
>     For now, I have kept the patch#4 to deal with the
>     -Werror=enum-conversion:
>      - drm/i915/gvt/display_helpers: Cast argument to enum pipe for
>        pipe-offset macro
>
> Ankit Nautiyal (7):
>   drm/i915/display: Abstract pipe/trans/cursor offset calculation
>   drm/i915/display: Add APIs to be used by gvt to get the register
>     offsets
>   drm/i915/gvt: Add header to use display offset functions in macros
>   drm/i915/gvt/display_helpers: Cast argument to enum pipe for
>     pipe-offset macro
>   drm/i915/gvt: Change for_each_pipe to use pipe_valid API
>   drm/i915/gvt: Use the appropriate header for the DPLL macro
>   drm/i915/gvt/display_helper: Get rid of #ifdef/#undefs
>
>  drivers/gpu/drm/i915/Makefile                 |  1 +
>  .../drm/i915/display/intel_display_device.h   | 17 +++++++
>  .../drm/i915/display/intel_display_limits.c   |  0
>  .../drm/i915/display/intel_display_reg_defs.h | 15 ++-----
>  drivers/gpu/drm/i915/display/intel_gvt_api.c  | 45 +++++++++++++++++++
>  drivers/gpu/drm/i915/display/intel_gvt_api.h  | 21 +++++++++
>  drivers/gpu/drm/i915/gvt/cmd_parser.c         |  2 +-
>  drivers/gpu/drm/i915/gvt/display.c            |  9 ++--
>  drivers/gpu/drm/i915/gvt/display_helpers.h    | 27 +++++++++++
>  drivers/gpu/drm/i915/gvt/fb_decoder.c         |  2 +-
>  drivers/gpu/drm/i915/gvt/handlers.c           |  2 +-
>  11 files changed, 123 insertions(+), 18 deletions(-)
>  create mode 100644 drivers/gpu/drm/i915/display/intel_display_limits.c
>  create mode 100644 drivers/gpu/drm/i915/display/intel_gvt_api.c
>  create mode 100644 drivers/gpu/drm/i915/display/intel_gvt_api.h
>  create mode 100644 drivers/gpu/drm/i915/gvt/display_helpers.h

-- 
Jani Nikula, Intel