Mesa (staging/21.0): tgsi_exec: Fix NaN behavior of min and max

[GitLab Mirror]

Module: Mesa
Branch: staging/21.0
Commit: bfc8e0d6093c67eb7e08832102b877d0be763f49
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=bfc8e0d6093c67eb7e08832102b877d0be763f49

Author: Ian Romanick <ian.d.romanick at intel.com>
Date:   Thu Apr 22 17:46:39 2021 -0700

tgsi_exec: Fix NaN behavior of min and max

Modern shader APIs, like DX10 and GLSL 1.30, want min() and max() to
"cleanse" NaN.  If one source is NaN, the other value should be chosen.
If both sources are NaN, the result may be either.

There are many cases where TGSI is generate from NIR, and many
optimizations in NIR expect this behavior.  Not meeting these
expectations can lead to unexpected results.

Reviewed-by: Eric Anholt <eric at anholt.net>
Fixes: ffe58739da9 ("Softpipe: import TGSI tree. Not hooked-up yet.")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10419>
(cherry picked from commit 939bf7a4198cecff57510b7fe5d38ef9b5dd22be)

Conflicts:
	.gitlab-ci/piglit/softpipe-quick.txt

---

 .gitlab-ci/piglit/softpipe-quick.txt   |  8 ++------
 .pick_status.json                      |  2 +-
 src/gallium/auxiliary/tgsi/tgsi_exec.c | 32 ++++++++++++++++----------------
 3 files changed, 19 insertions(+), 23 deletions(-)

diff --git a/.gitlab-ci/piglit/softpipe-quick.txt b/.gitlab-ci/piglit/softpipe-quick.txt
index d500fb01306..915951dd1ad 100644
--- a/.gitlab-ci/piglit/softpipe-quick.txt
+++ b/.gitlab-ci/piglit/softpipe-quick.txt
@@ -1844,10 +1844,6 @@ spec/arb_transform_feedback_overflow_query/arb_transform_feedback_overflow_query
 spec/arb_transform_feedback_overflow_query/arb_transform_feedback_overflow_query-basic/arb_transform_feedback_overflow_query-buffer_object_no_overflow: skip
 spec/arb_transform_feedback_overflow_query/arb_transform_feedback_overflow_query-basic/arb_transform_feedback_overflow_query-buffer_object_no_overflow_single: skip
 spec/arb_transform_feedback_overflow_query/arb_transform_feedback_overflow_query-basic/arb_transform_feedback_overflow_query-buffer_object_single: skip
-spec/arb_transform_feedback_overflow_query/arb_transform_feedback_overflow_query-basic/arb_transform_feedback_overflow_query-conditional_render_1: fail
-spec/arb_transform_feedback_overflow_query/arb_transform_feedback_overflow_query-basic/arb_transform_feedback_overflow_query-conditional_render_no_overflow: fail
-spec/arb_transform_feedback_overflow_query/arb_transform_feedback_overflow_query-basic/arb_transform_feedback_overflow_query-conditional_render_no_overflow_single: fail
-spec/arb_transform_feedback_overflow_query/arb_transform_feedback_overflow_query-basic/arb_transform_feedback_overflow_query-simple_query_any: fail
 spec/arb_uniform_buffer_object/bufferstorage: skip
 spec/arb_vertex_program/vp-address-02: skip
 spec/arb_vertex_program/vp-address-04: skip
@@ -3083,8 +3079,8 @@ wgl/wgl-sanity: skip
 summary:
        name:  results
        ----  --------
-       pass:    26801
-       fail:      399
+       pass:    26805
+       fail:      395
       crash:       42
        skip:     2612
     timeout:        0
diff --git a/.pick_status.json b/.pick_status.json
index 6c12b57ace7..0ec63cc09ec 100644
--- a/.pick_status.json
+++ b/.pick_status.json
@@ -949,7 +949,7 @@
         "description": "tgsi_exec: Fix NaN behavior of min and max",
         "nominated": true,
         "nomination_type": 1,
-        "resolution": 0,
+        "resolution": 1,
         "master_sha": null,
         "because_sha": "ffe58739da9eee2e99682747cc8f26e412c87430"
     },
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index 08be2e95688..c35437343fe 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c
@@ -263,20 +263,20 @@ static void
 micro_dmax(union tgsi_double_channel *dst,
            const union tgsi_double_channel *src)
 {
-   dst->d[0] = src[0].d[0] > src[1].d[0] ? src[0].d[0] : src[1].d[0];
-   dst->d[1] = src[0].d[1] > src[1].d[1] ? src[0].d[1] : src[1].d[1];
-   dst->d[2] = src[0].d[2] > src[1].d[2] ? src[0].d[2] : src[1].d[2];
-   dst->d[3] = src[0].d[3] > src[1].d[3] ? src[0].d[3] : src[1].d[3];
+   dst->d[0] = src[0].d[0] > src[1].d[0] || isnan(src[1].d[0]) ? src[0].d[0] : src[1].d[0];
+   dst->d[1] = src[0].d[1] > src[1].d[1] || isnan(src[1].d[1]) ? src[0].d[1] : src[1].d[1];
+   dst->d[2] = src[0].d[2] > src[1].d[2] || isnan(src[1].d[2]) ? src[0].d[2] : src[1].d[2];
+   dst->d[3] = src[0].d[3] > src[1].d[3] || isnan(src[1].d[3]) ? src[0].d[3] : src[1].d[3];
 }
 
 static void
 micro_dmin(union tgsi_double_channel *dst,
            const union tgsi_double_channel *src)
 {
-   dst->d[0] = src[0].d[0] < src[1].d[0] ? src[0].d[0] : src[1].d[0];
-   dst->d[1] = src[0].d[1] < src[1].d[1] ? src[0].d[1] : src[1].d[1];
-   dst->d[2] = src[0].d[2] < src[1].d[2] ? src[0].d[2] : src[1].d[2];
-   dst->d[3] = src[0].d[3] < src[1].d[3] ? src[0].d[3] : src[1].d[3];
+   dst->d[0] = src[0].d[0] < src[1].d[0] || isnan(src[1].d[0]) ? src[0].d[0] : src[1].d[0];
+   dst->d[1] = src[0].d[1] < src[1].d[1] || isnan(src[1].d[1]) ? src[0].d[1] : src[1].d[1];
+   dst->d[2] = src[0].d[2] < src[1].d[2] || isnan(src[1].d[2]) ? src[0].d[2] : src[1].d[2];
+   dst->d[3] = src[0].d[3] < src[1].d[3] || isnan(src[1].d[3]) ? src[0].d[3] : src[1].d[3];
 }
 
 static void
@@ -1357,10 +1357,10 @@ micro_max(union tgsi_exec_channel *dst,
           const union tgsi_exec_channel *src0,
           const union tgsi_exec_channel *src1)
 {
-   dst->f[0] = src0->f[0] > src1->f[0] ? src0->f[0] : src1->f[0];
-   dst->f[1] = src0->f[1] > src1->f[1] ? src0->f[1] : src1->f[1];
-   dst->f[2] = src0->f[2] > src1->f[2] ? src0->f[2] : src1->f[2];
-   dst->f[3] = src0->f[3] > src1->f[3] ? src0->f[3] : src1->f[3];
+   dst->f[0] = src0->f[0] > src1->f[0] || isnan(src1->f[0]) ? src0->f[0] : src1->f[0];
+   dst->f[1] = src0->f[1] > src1->f[1] || isnan(src1->f[1]) ? src0->f[1] : src1->f[1];
+   dst->f[2] = src0->f[2] > src1->f[2] || isnan(src1->f[2]) ? src0->f[2] : src1->f[2];
+   dst->f[3] = src0->f[3] > src1->f[3] || isnan(src1->f[3]) ? src0->f[3] : src1->f[3];
 }
 
 static void
@@ -1368,10 +1368,10 @@ micro_min(union tgsi_exec_channel *dst,
           const union tgsi_exec_channel *src0,
           const union tgsi_exec_channel *src1)
 {
-   dst->f[0] = src0->f[0] < src1->f[0] ? src0->f[0] : src1->f[0];
-   dst->f[1] = src0->f[1] < src1->f[1] ? src0->f[1] : src1->f[1];
-   dst->f[2] = src0->f[2] < src1->f[2] ? src0->f[2] : src1->f[2];
-   dst->f[3] = src0->f[3] < src1->f[3] ? src0->f[3] : src1->f[3];
+   dst->f[0] = src0->f[0] < src1->f[0] || isnan(src1->f[0]) ? src0->f[0] : src1->f[0];
+   dst->f[1] = src0->f[1] < src1->f[1] || isnan(src1->f[1]) ? src0->f[1] : src1->f[1];
+   dst->f[2] = src0->f[2] < src1->f[2] || isnan(src1->f[2]) ? src0->f[2] : src1->f[2];
+   dst->f[3] = src0->f[3] < src1->f[3] || isnan(src1->f[3]) ? src0->f[3] : src1->f[3];
 }
 
 static void