[Mesa-dev] [PATCH] i965: Replace checks for rb->Name with FlipY

[Fritz Koenig]

On Thu, Aug 23, 2018 at 11:01 AM Emil Velikov <emil.l.velikov at gmail.com> wrote:
>
> On Thu, 23 Aug 2018 at 18:41, Emil Velikov <emil.l.velikov at gmail.com> wrote:
> >
> > Hi Fritz,
> >
> > On Wed, 22 Aug 2018 at 23:01, Fritz Koenig <frkoenig at google.com> wrote:
> > >
> > > In the GL_MESA_framebuffer_flip_y implementation
> > > _mesa_is_winsys_fbo checks were replaced with
> > > FlipY checks.  rb->Name is also used to determine
> > > if a buffer is winsys.
> >
> > This seems like it fixes your earlier patch, so you normally want a
> > fixes tag. Something like the following should be fine.
> > It adds nice reference tracking, allowing the release manager to track
> > and land in the correct stable branches.
> >
> > Fixes: ab05dd183cc ("i965: implement GL_MESA_framebuffer_flip_y [v3]")
> >

Thanks for the heads up on this, I'll work on getting a changed commit
message out.

> Quick grep shows a number of rb->Name instances in intel_fbo.c - DBG
> and assert statements.
> If they should stay as-is, perhaps it'll be better to add some comments?
>
The one assert I see is in intel_alloc_window_storage and that one
should stay.  That function should only be called with a windows
system framebuffer.

The debug statements are a little trickier.  In intel_map_renderbuffer
it is trivial to change to printing the flip_y out, but in
intel_unmap_renderbuffer it's harder to do because flip_y is not
present.  It's associated with a framebuffer, not a renderbuffer.  I
don't think adding flip_y and removing rb->Name is desired here.
rb->Name is an integer, those two statements usefulness appears to
come from correlating mapping and unmapping of buffers.  So I don't
think removing rb->Name would be worthwhile.

> Just a fly-by idea.
> -Emil