Re: [Intel-gfx] [PATCH v14 0/7] mtl: add support for pmdemand

["Govindapillai, Vinod" <vinod.govindapillai@intel.com>]

On Tue, 2023-06-06 at 13:32 +0300, Jani Nikula wrote:
> On Tue, 06 Jun 2023, Vinod Govindapillai <vinod.govindapillai@intel.com> wrote:
> > SAGV configuration support for MTL
> > 
> > v2: added one missing patch in the previous version
> > 
> > v3: chekcpatch warning fixes
> >     update index handling for the icl/tgl QGV point handling
> >     program pmdemand code simplified
> > 
> > v4: update to debufs and pipe values pmdemand regiters
> >     removed the macro usage in update_pmdemand_values
> > 
> > V5: Addressing comments from Gustavo and Jani
> >     And some other fixes for issues from CI
> > 
> > v6: Addressing some comments from Gustavo
> >     Updates to pmdemand state struct, active phys calculations
> >     Got rid of suppress warning patch from v5
> > 
> > v7: Rebased and updates to max ddiclk and active phys calculations
> > 
> > v8: updates to active phys calcuations
> > 
> > v9: Address styling issues
> > 
> > v10: Updates to phys calculation, pmdemand state initialization during
> >      HW readout / sanitization
> > 
> > v11: Fix CI checkpatch warnings
> > 
> > v12: Addressing comments
> > 
> > v13: Updates based on imre's comments to handle non serialize cases,
> >      updates tp phys mask during sanitize calls after HW readout
> > 
> > v14: check display version before accessig pmdemand functions
> 
> Okay, please stop sending new versions now.
> 
> We don't want to reach v14 on any patch series. It's a burden on the
> developer, the reviewers, the CI, everyone. Please adjust towards
> waiting for more review comments before sending another version, and
> making the most of each version sent. There are several versions of the
> series that didn't receive any feedback; why send another version?
> 
> In this case, AFAICT, some of the patches in the series have Reviewed-by
> but haven't been changed in the last ten versions of the series. Why
> haven't they been merged already? What's the point in resending them
> over and over again?

Yeah! This Pmdemand stuffs has been a nightmare! Apparently it looks like the original version
lacked many functionalities and had been many review comments! I overlooked some fixes which i only
noticed after i sent the patch. That might be the reason why some patch revisions are sent without
any review comments.

Now Imre is actively reviewing the pmdemand patch. Hopefully could be closed soon

Yes. The other patches in the series could be merged. I dont have the commit rights.
It would be nice if someone can merge those

BR
Vinod

> 
> 
> BR,
> Jani.
> 
> 
> > 
> > Mika Kahola (1):
> >   drm/i915/mtl: Add support for PM DEMAND
> > 
> > Vinod Govindapillai (6):
> >   drm/i915: fix the derating percentage for MTL
> >   drm/i915: update the QGV point frequency calculations
> >   drm/i915: store the peak bw per QGV point
> >   drm/i915: extract intel_bw_check_qgv_points()
> >   drm/i915: modify max_bw to return index to intel_bw_info
> >   drm/i915/mtl: find the best QGV point for the SAGV configuration
> > 
> >  drivers/gpu/drm/i915/Makefile                 |   1 +
> >  drivers/gpu/drm/i915/display/intel_bw.c       | 345 ++++++---
> >  drivers/gpu/drm/i915/display/intel_bw.h       |   6 +
> >  drivers/gpu/drm/i915/display/intel_display.c  |  14 +
> >  .../gpu/drm/i915/display/intel_display_core.h |  11 +
> >  .../drm/i915/display/intel_display_driver.c   |   7 +
> >  .../gpu/drm/i915/display/intel_display_irq.c  |  23 +-
> >  .../drm/i915/display/intel_display_power.c    |  14 +-
> >  .../drm/i915/display/intel_modeset_setup.c    |  36 +
> >  drivers/gpu/drm/i915/display/intel_pmdemand.c | 714 ++++++++++++++++++
> >  drivers/gpu/drm/i915/display/intel_pmdemand.h |  69 ++
> >  drivers/gpu/drm/i915/i915_reg.h               |  36 +-
> >  12 files changed, 1153 insertions(+), 123 deletions(-)
> >  create mode 100644 drivers/gpu/drm/i915/display/intel_pmdemand.c
> >  create mode 100644 drivers/gpu/drm/i915/display/intel_pmdemand.h
>