[Mesa-dev] [PATCH] i965/gen5+: Fix incorrect miptree layout for non-power-of-two cubemaps.
eric at anholt.net
Sun Jul 31 19:25:11 PDT 2011
On Sat, 30 Jul 2011 21:35:52 -0700, Kenneth Graunke <kenneth at whitecape.org> wrote:
> For power-of-two sizes, h0 == mt->height0 since it's already a multiple
> of two. However, for NPOT, they're different; h1 should be computed
> based on the original size.
> Fixes piglit test "cubemap npot" and oglconform_31 test "textureNPOT".
> NOTE: This is a candidate for stable release branches.
> Signed-off-by: Kenneth Graunke <kenneth at whitecape.org>
> src/mesa/drivers/dri/i965/brw_tex_layout.c | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
> Note that the piglit test referenced isn't committed yet; I sent it to the
> piglit mailing list.
> diff --git a/src/mesa/drivers/dri/i965/brw_tex_layout.c b/src/mesa/drivers/dri/i965/brw_tex_layout.c
> index f462f32..46a417a 100644
> --- a/src/mesa/drivers/dri/i965/brw_tex_layout.c
> +++ b/src/mesa/drivers/dri/i965/brw_tex_layout.c
> @@ -60,7 +60,7 @@ GLboolean brw_miptree_layout(struct intel_context *intel,
> * given in Volume 1 of the BSpec.
> h0 = ALIGN(mt->height0, align_h);
> - h1 = ALIGN(minify(h0), align_h);
> + h1 = ALIGN(minify(mt->height0), align_h);
> qpitch = (h0 + h1 + (intel->gen >= 7 ? 12 : 11) * align_h);
> if (mt->compressed)
> qpitch /= 4;
This looks wrong to me. The height of a level L is ALIGN(height0 >> L,
j) according to SNB PRM vol1, 126.96.36.199 "Computing MIP level sizes".
Note that our calculation of j is wrong for a bunch of hardware/format
combos -- I wonder if that was the issue you were looking at?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 197 bytes
Desc: not available
More information about the mesa-dev