[Intel-gfx] [PATCH 0/4] SKL watermark algorithm updates

[Matt Roper]

These patches from Mahesh address places where our SKL-style watermark
programming does not match the current bspec rules, either because the bspec
was updated after the original code was written, or because new display
features (e.g., plane scaling) were added without corresponding updates to the
watermark code.

Mahesh wrote these patches a long time ago, but due to other churn in the
driver they never made it in.  The work here is all orthogonal to the atomic
watermark stuff I've been doing lately, but landing atomic watermarks did
require major changes to the flow of the code in this series, so these patches
have been nearly completely rewritten at this point.  You can thank Mahesh for
any bugs these fix, but blame me for any new bugs they contain.

At the moment I've dropped the final patches here for the arbitrated display
bandwidth workaround that was previously part of this series; I'm working on a
new method of obtaining the data we need for that workaround that's less
fragile than the DMI method previously attempted and will post that workaround
separately after this series lands.

Cc: Kumar, Mahesh <mahesh1.kumar at intel.com>


Kumar, Mahesh (3):
  drm/i915/skl+: calculate ddb minimum allocation (v4)
  drm/i915/skl+: calculate plane pixel rate (v4)
  drm/i915/skl+: Use scaling amount for plane data rate calculation (v3)

Matt Roper (1):
  drm/i915: Don't try to calculate relative data rates during hw readout

 drivers/gpu/drm/i915/intel_pm.c | 166 ++++++++++++++++++++++++++++++++--------
 1 file changed, 136 insertions(+), 30 deletions(-)

-- 
2.1.4