[v2,00/11] drm: Introduce DRM client library

Message

Thomas Zimmermann Oct. 2, 2024, 1:04 p.m. UTC

With the next DRM client coming soon (drm_log) and most of DRM's
fbdev emulation consolidated in a few places, it's time to provide
a single place for the clients.

The new module drm_client_lib.ko stores most of the in-kernel client
code. It is designed such that drivers can opt into client support,
but the presence of the client module depends on the user's kernel
configuration. Without selected clients, no client module will be
build.

Version 2 of this patchset is a significant rework of the patches,
so there are no R-bs kept. The base client code and client event
handling remains in the DRM core. This simplifies module dependencies.
But suspend and resume are now handled in client helpers instead of
direct calls to fbdev emulation. This breaks a cyclic dependency
among the involved modules. It also allows any client to process
suspend and resume events.

v2:
- rework of the overall design
- keep base client code in DRM core.

Thomas Zimmermann (11):
  drm/i915: Select DRM_CLIENT_SELECTION
  drm/xe: Select DRM_CLIENT_SELECTION
  drm/fbdev: Select fbdev I/O helpers from modules that require them
  drm/fbdev: Store fbdev module parameters in separate file
  drm/client: Move client event handlers to drm_client_event.c
  drm/client: Move suspend/resume into DRM client callbacks
  drm/amdgpu: Suspend and resume internal clients with client helpers
  drm/nouveau: Suspend and resume clients with client helpers
  drm/radeon: Suspend and resume clients with client helpers
  drm/client: Make client support optional
  drm/client: Add client-lib module

 Documentation/gpu/drm-client.rst              |   3 +
 drivers/gpu/drm/Kconfig                       |  39 +++-
 drivers/gpu/drm/Makefile                      |  20 +-
 drivers/gpu/drm/amd/amdgpu/Kconfig            |   1 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c    |  22 +-
 drivers/gpu/drm/drm_client.c                  | 121 -----------
 drivers/gpu/drm/drm_client_event.c            | 195 ++++++++++++++++++
 drivers/gpu/drm/drm_client_setup.c            |   3 +
 drivers/gpu/drm/drm_drv.c                     |   2 +-
 drivers/gpu/drm/drm_fb_helper.c               |  31 ---
 drivers/gpu/drm/drm_fbdev_client.c            |  30 ++-
 drivers/gpu/drm/drm_file.c                    |   2 +-
 drivers/gpu/drm/drm_internal.h                |   7 +
 drivers/gpu/drm/drm_kms_helper_common.c       |  38 ++++
 drivers/gpu/drm/drm_modeset_helper.c          |  14 +-
 drivers/gpu/drm/drm_probe_helper.c            |   2 +-
 drivers/gpu/drm/i915/Kconfig                  |   1 +
 .../drm/i915/display/intel_display_driver.c   |   2 +-
 drivers/gpu/drm/nouveau/nouveau_display.c     |   8 +-
 drivers/gpu/drm/nouveau/nouveau_vga.c         |   2 +-
 drivers/gpu/drm/radeon/radeon_device.c        |  19 +-
 drivers/gpu/drm/radeon/radeon_fbdev.c         |   6 -
 drivers/gpu/drm/radeon/radeon_mode.h          |   3 -
 drivers/gpu/drm/xe/Kconfig                    |   1 +
 include/drm/drm_client.h                      |  39 +++-
 include/drm/drm_client_event.h                |  27 +++
 26 files changed, 423 insertions(+), 215 deletions(-)
 create mode 100644 drivers/gpu/drm/drm_client_event.c
 create mode 100644 include/drm/drm_client_event.h

Comments

Jocelyn Falempe Oct. 3, 2024, 5:10 p.m. UTC | #1

On 02/10/2024 15:04, Thomas Zimmermann wrote:
> With the next DRM client coming soon (drm_log) and most of DRM's
> fbdev emulation consolidated in a few places, it's time to provide
> a single place for the clients.
> 
> The new module drm_client_lib.ko stores most of the in-kernel client
> code. It is designed such that drivers can opt into client support,
> but the presence of the client module depends on the user's kernel
> configuration. Without selected clients, no client module will be
> build.

I think I would rather rename drm_client.c to drm_client_lib.c, and have 
a drm_client.ko. Users may need to interact with module name more than 
with C file.

> 
> Version 2 of this patchset is a significant rework of the patches,
> so there are no R-bs kept. The base client code and client event
> handling remains in the DRM core. This simplifies module dependencies.
> But suspend and resume are now handled in client helpers instead of
> direct calls to fbdev emulation. This breaks a cyclic dependency
> among the involved modules. It also allows any client to process
> suspend and resume events.

Thanks a lot for this work. I rebased my drm_log proof of concept on top 
on this v2, and it works great.
I think I can add a suspend/resume callback to drm_log, just marking the 
console as CON_SUSPENDED should work, I will test that shortly.


Best regards,