On Fri, Oct 23, 2026 at 12:41:15PM +0530, Jayant Shekhar wrote:
> 
> On 15-10-2026 18:48, Dmitry Baryshkov wrote:
> > On Thu, Oct 15, 2026 at 01:08:26PM +0530, Jayant Shekhar wrote:
> > > The firmware-owned DCP pipeline still needs normal DRM atomic objects and
> > > event handling so it can participate in the MSM DRM device.
> > > 
> > > Add DCP plane and CRTC objects, validate framebuffer and plane state,
> > > allocate firmware display pipes, calculate bandwidth and clock votes, and
> > > translate visible planes and CRTC state into HFI commands. Route VSYNC and
> > > frame-scan completion events back to DRM and retain CRTC-local display
> > > session state.
> > > 
> > > Signed-off-by: Jayant Shekhar <[email protected]>
> > > Assisted-by: LLM
> > > ---
> > >   drivers/gpu/drm/msm/Makefile           |    3 +
> > >   drivers/gpu/drm/msm/dcp/dcp_crtc.c     |  733 ++++++++++++++++++++++
> > >   drivers/gpu/drm/msm/dcp/dcp_crtc.h     |  104 ++++
> > >   drivers/gpu/drm/msm/dcp/dcp_dpu_kms.c  |   27 +
> > >   drivers/gpu/drm/msm/dcp/dcp_dpu_kms.h  |   49 ++
> > >   drivers/gpu/drm/msm/dcp/dcp_hfi.h      |    2 +
> > >   drivers/gpu/drm/msm/dcp/dcp_hfi_core.c |   21 +
> > >   drivers/gpu/drm/msm/dcp/dcp_plane.c    | 1073 
> > > ++++++++++++++++++++++++++++++++
> > >   drivers/gpu/drm/msm/dcp/dcp_plane.h    |   60 ++
> > >   9 files changed, 2072 insertions(+)
> > > 

> > > *p++ = lid;
> > > + for (i = 0; i < HFI_MAX_PLANES; i++)
> > > +         *p++ = lower_32_bits(iova[i]);
> > > + for (i = 0; i < HFI_MAX_PLANES; i++)
> > > +         *p++ = upper_32_bits(iova[i]);
> > Who has invented this? Splitting addresses into two distinct arrays.
> > 
> 
> The field order is defined by the firmware HFI ABI, not chosen by the
> driver. struct hfi_plane_buff defines the wire layout as:
> 
> u32 addr_l[HFI_MAX_PLANES];
> u32 addr_h[HFI_MAX_PLANES];
> ..

can it be fixed before it is landed? If somebody made a strange decision
it doesn't mean that it can't be changed.


-- 
With best wishes
Dmitry

Reply via email to