[Intel-gfx] [RFC PATCH 0/7] Convert to atomic, part 4.

[Maarten Lankhorst]

The only thing missing after part 3 is restoring atomic readout and making
suspend/restore. With fixes for all identified causes of regressions from
atomic suspend being done in convert to atomic, part 3 it's time to worry
about getting atomic suspend/restore working again.

I do this in a few steps, first I re-introduce the old commit for hw state
readout, with fixes. After that I convert the atomic readout to a full initial
modeset, and finally I try to re-enable fastboot optimizations again by default,
in a less hacky way.

This is a RFC because of the following possible issues:
 - mode->clock may not be read out correctly. It can differ slightly from the
   actually set value, which forces a modeset anyway when it could be avoided.
 - Not tested with DRRS support on DP, might be broken with fastboot?

Maarten Lankhorst (7):
  drm/i915: Do not update pfit state when toggling crtc enabled.
  drm/i915: Read hw state into an atomic state struct, try 2.
  All changes from try2.
  drm/i915: enable fastboot for everyone
  drm/i915: Update power domains only on affected crtc's.
  drm/i915: Always reset in intel_crtc_restore_mode
  drm/i915: Make intel_display_suspend atomic, try 2.

 drivers/gpu/drm/i915/i915_dma.c      |   12 +-
 drivers/gpu/drm/i915/i915_drv.c      |    2 +-
 drivers/gpu/drm/i915/i915_drv.h      |    7 +-
 drivers/gpu/drm/i915/i915_params.c   |    5 -
 drivers/gpu/drm/i915/intel_atomic.c  |   23 +-
 drivers/gpu/drm/i915/intel_display.c | 1302 ++++++++++++++++++++--------------
 drivers/gpu/drm/i915/intel_dp.c      |    2 +-
 drivers/gpu/drm/i915/intel_drv.h     |   21 +-
 drivers/gpu/drm/i915/intel_fbdev.c   |   22 +-
 drivers/gpu/drm/i915/intel_lvds.c    |    2 +-
 10 files changed, 786 insertions(+), 612 deletions(-)

-- 
2.1.0