Mesa (master): gallivm: fix sampling with texture offsets in SoA path

[Roland Scheidegger]

Module: Mesa
Branch: master
Commit: 71ff5af5dd308e4a53b66b7530cc01ec4bf5715e
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=71ff5af5dd308e4a53b66b7530cc01ec4bf5715e

Author: Roland Scheidegger <sroland at vmware.com>
Date:   Thu Oct 22 23:49:41 2015 +0200

gallivm: fix sampling with texture offsets in SoA path

When using nearest filtering and clamp / clamp to edge wrapping results could
be wrong for negative offsets. Fix this by adding the offset before doing
the conversion to int coords (could also use floor instead of trunc int
conversion but probably more complex on "typical" cpu).

This fixes the piglit texwrap offset failures with this filter/wrap combo
(which only leaves the linear/mirror repeat combination broken).

Reviewed-by: Dave Airlie <airlied at redhat.com>
Reviewed-by: Jose Fonseca <jfonseca at vmware.com>

---

 src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c |   11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c b/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c
index b5c06b6..125505e 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c
@@ -567,12 +567,13 @@ lp_build_sample_wrap_nearest(struct lp_build_sample_context *bld,
          coord = lp_build_mul(coord_bld, coord, length_f);
       }
 
+      if (offset) {
+         offset = lp_build_int_to_float(coord_bld, offset);
+         coord = lp_build_add(coord_bld, coord, offset);
+      }
       /* floor */
       /* use itrunc instead since we clamp to 0 anyway */
       icoord = lp_build_itrunc(coord_bld, coord);
-      if (offset) {
-         icoord = lp_build_add(int_coord_bld, icoord, offset);
-      }
 
       /* clamp to [0, length - 1]. */
       icoord = lp_build_clamp(int_coord_bld, icoord, int_coord_bld->zero,
@@ -2586,6 +2587,10 @@ lp_build_sample_soa_code(struct gallivm_state *gallivm,
       derived_sampler_state.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
       derived_sampler_state.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
    }
+   /*
+    * We could force CLAMP to CLAMP_TO_EDGE here if min/mag filter is nearest,
+    * so AoS path could be used. Not sure it's worth the trouble...
+    */
 
    min_img_filter = derived_sampler_state.min_img_filter;
    mag_img_filter = derived_sampler_state.mag_img_filter;