On Mon, 24 Oct 2011 19:23:53 -0600, Brian Paul <brian.e.p...@gmail.com> wrote:
> On Mon, Oct 24, 2011 at 5:15 PM, Eric Anholt <e...@anholt.net> wrote:
> > @@ -2322,6 +2363,16 @@ teximage(struct gl_context *ctx, GLuint dims,
> > return; /* error was recorded */
> > }
> >
> > + /* Allow a hardware driver to just strip out the border, to provide
> > + * reliable but slightly incorrect hardware rendering instead of
> > + * rarely-tested software fallback rendering.
> > + */
> > + if (border && ctx->Texture.StripTextureBorder) {
> > + strip_texture_border(&border, &width, &height, &depth, unpack,
> > + &unpack_no_border);
> > + unpack = &unpack_no_border;
> > + }
> > +
>
> I think we also need this code in the glCopyTexImage() path.
>
> -Brian
Good point. I made the fix for that, then made a testcase, then fixed
the fix.
I also ended up deleting a bit more Intel code, which caught some
silliness where we do mappings to store data for 0-sized textures,
which made the driver now assert in an ES2 conformance case after
deleting more Border code. I think this is a good fix for the
problem, by just bailing out of texstore instead of making the caller
avoid texstore in these cases.
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
