[Mesa-dev] [PATCH 3/8] i965/msaa: Allocate MCS buffer when CMS MSAA is in use.

[Paul Berry]

On 9 July 2012 13:36, Chad Versace <chad.versace at linux.intel.com> wrote:

> On 07/06/2012 03:29 PM, Paul Berry wrote:
> > To implement Gen7's CMS MSAA layout, we need an extra buffer, the MCS
> > (Multisample Control Surface) buffer.  This patch introduces code for
> > allocating and deallocating the buffer, and storing a pointer to it in
> > the intel_mipmap_tree struct.
> >
> > No functional change, since the CMS layout is not enabled yet.
> > ---
> >  src/mesa/drivers/dri/intel/intel_fbo.c         |    9 +++++
> >  src/mesa/drivers/dri/intel/intel_mipmap_tree.c |   45
> ++++++++++++++++++++++++
> >  src/mesa/drivers/dri/intel/intel_mipmap_tree.h |   14 +++++++
> >  3 files changed, 68 insertions(+), 0 deletions(-)
> >
> > diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c
> b/src/mesa/drivers/dri/intel/intel_fbo.c
> > index a801bb7..6530b99 100644
> > --- a/src/mesa/drivers/dri/intel/intel_fbo.c
> > +++ b/src/mesa/drivers/dri/intel/intel_fbo.c
> > @@ -282,6 +282,15 @@ intel_alloc_renderbuffer_storage(struct gl_context
> * ctx, struct gl_renderbuffer
> >        }
> >     }
> >
> > +   if (irb->mt->msaa_layout == INTEL_MSAA_LAYOUT_CMS) {
> > +      bool ok = intel_miptree_alloc_mcs(intel, irb->mt, rb->NumSamples);
> > +      if (!ok) {
> > +         intel_miptree_release(&irb->mt->hiz_mt);
> > +         intel_miptree_release(&irb->mt);
> > +         return false;
> > +      }
> > +   }
> > +
> >     return true;
> >  }
>
> FYI, though it causes no harm, there's no need to release the hiz miptree
> separately from its parent. The miptree's destructor takes care of that.
>

Good point.  I'll remove it.


>
>
> > +   /* From the Ivy Bridge PRM, Vol 2 Part 1 p326:
> > +    *
> > +    *     When MCS buffer is enabled and bound to MSRT, it is required
> that it
> > +    *     is cleared prior to any rendering.
> > +    *
> > +    * Since we don't use the MCS buffer for any purpose other than
> rendering,
> > +    * it makes sense to just clear it immediately upon allocation.
> > +    */
> > +   void *data = intel_region_map(intel, mt->mcs_mt->region, 0);
> > +   memset(data, 0, mt->mcs_mt->region->bo->size);
> > +   intel_region_unmap(intel, mt->mcs_mt->region);
>
> But the clear value for a 4x MCS is 0xff. Am I missing something?
>

Yes, you're right.  I had a convoluted reason for believing that
initializing to 0 was better, but it's not standing up to scrutiny.  I'll
change it to 0xff.


>
> -Chad
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20120709/d15fb9e2/attachment.html>