[00/13] drm/amd/display: more drm_edid to AMD display driver

Message

Melissa Wen April 11, 2025, 8:08 p.m. UTC

Hi,

Siqueira and I worked on a solution to reduce the usage of drm_edid_raw
in the AMD display driver, since the current guideline in the DRM
subsystem is to stop handling raw edid data in driver-specific
implementation and use opaque `drm_edid` object with common-code
helpers.

In short, this series approaches the issue of maintaining DC as an
OS-agnostic component by creating a mid layer to isolate `drm_edid`
helpers that need to be called in the DC code, while allowing other OSes
to implement their specific implementation. This is an extension of [1].

- Patch 1 addresses a possible leak added by previous migration to
  drm_edid.
- Patches 2-6 use common-code, drm_edid helpers to parse edid
  capabilities instead of driver-specific solutions.
- Patches 7-8 are groundwork to reduce the noise of Linux/DRM specific
  code in the DC shared code
- Patch 9 creates a mid layer to make DC embraces different ways of
  handling EDID by platforms.
- Patch 10-11 move open-coded management of raw EDID data to the mid
  layer created before.
- Patch 12 adds drm_edid to dc_sink struct and a mid-layer helper to
  free `drm_edid`.
- Patch 13 switch dc_edid to drm_edid across the driver in a way that
  the DC shared code is little affected by Linux specific stuff.

There are three specific points where we still use drm_edid_raw() in the
driver:
1. raw edid data for write EDID checksum in DP_TEST_EDID_CHECKSUM via
   drm_dp_dpcd_write(), that AFAIK there is no common code solution yet;
2. open-coded connectivity log for dc link detection, that maybe can be
   moved to drm (?);
3. open-coded parser that I suspect is a lot of duplicated code, but
   needs careful examining.

I suggest to address those points in a next phase for regression control.

[1] https://lore.kernel.org/amd-gfx/[e-mail]/

Let me know yours thoughts!

Melissa

Melissa Wen (11):
  drm/amd/display: make sure drm_edid stored in aconnector doesn't leak
  drm/amd/display: use drm_edid_product_id for parsing EDID product info
  drm/amd/display: parse display name from drm_eld
  drm/amd/display: get panel id with drm_edid helper
  drm/amd/display: get SAD from drm_eld when parsing EDID caps
  drm/amd/display: get SADB from drm_eld when parsing EDID caps
  drm/amd/display: simplify dm_helpers_parse_edid_caps signature
  drm/amd/display: change DC functions to accept private types for edid
  drm/edid: introduce a helper that compares edid data from two drm_edid
  drm/amd/display: add drm_edid to dc_sink
  drm/amd/display: move dc_sink from dc_edid to drm_edid

Rodrigo Siqueira (2):
  drm/amd/display: add a mid-layer file to handle EDID in DC
  drm/amd/display: create a function to fill dc_sink with edid data

 .../gpu/drm/amd/display/amdgpu_dm/Makefile    |   1 +
 .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c |  33 +++---
 .../amd/display/amdgpu_dm/amdgpu_dm_helpers.c | 104 ++++++++----------
 .../display/amdgpu_dm/amdgpu_dm_mst_types.c   |  21 ++--
 .../gpu/drm/amd/display/amdgpu_dm/dc_edid.c   |  34 ++++++
 .../gpu/drm/amd/display/amdgpu_dm/dc_edid.h   |  15 +++
 .../drm/amd/display/dc/core/dc_link_exports.c |   9 +-
 drivers/gpu/drm/amd/display/dc/core/dc_sink.c |   3 +
 drivers/gpu/drm/amd/display/dc/dc.h           |  12 +-
 drivers/gpu/drm/amd/display/dc/dm_helpers.h   |   7 +-
 drivers/gpu/drm/amd/display/dc/inc/link.h     |   9 +-
 .../drm/amd/display/dc/link/link_detection.c  |  30 ++---
 .../drm/amd/display/dc/link/link_detection.h  |   9 +-
 drivers/gpu/drm/drm_edid.c                    |  18 +++
 include/drm/drm_edid.h                        |   2 +
 15 files changed, 165 insertions(+), 142 deletions(-)
 create mode 100644 drivers/gpu/drm/amd/display/amdgpu_dm/dc_edid.c
 create mode 100644 drivers/gpu/drm/amd/display/amdgpu_dm/dc_edid.h