brw/algebraic: Allow mixed types in saturate constant folding

Prevents assertion failures in func.shader-ballot.basic.q0 and other
tests starting with "nir/algebraic: Optimize some b2f of integer
comparison".

Vector immediates, bfloat, and 8-bit floats are still not supported.

v2: Almost complete re-write based on suggestions from Ken.

v3: Don't retype() on a brw_imm_f value.

Fixes: f8e54d02f7 ("intel/compiler: Relax mixed type restriction for saturating immediates")
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/38978>

src/intel/compiler/brw/brw_opt_algebraic.cpp

@@ -411,22 +411,35 @@ brw_opt_algebraic(brw_shader &s)
          if (inst->src[0].file != IMM)
             break;
 
-         if (inst->saturate) {
-            /* Full mixed-type saturates don't happen.  However, we can end up
-             * with things like:
-             *
-             *    mov.sat(8) g21<1>DF       -1F
-             *
-             * Other mixed-size-but-same-base-type cases may also be possible.
-             */
-            if (inst->dst.type != inst->src[0].type &&
-                inst->dst.type != BRW_TYPE_DF &&
-                inst->src[0].type != BRW_TYPE_F)
-               UNREACHABLE("unimplemented: saturate mixed types");
-
-            if (brw_reg_saturate_immediate(&inst->src[0])) {
+         if (inst->saturate && brw_type_is_float(inst->dst.type)) {
+            switch (inst->src[0].type) {
+            case BRW_TYPE_UW:
+            case BRW_TYPE_UD:
+            case BRW_TYPE_UQ:
+               inst->src[0] = brw_imm_f(src_as_uint(inst->src[0]) != 0);
                inst->saturate = false;
                progress = true;
+               break;
+
+            case BRW_TYPE_W:
+            case BRW_TYPE_D:
+            case BRW_TYPE_Q:
+               inst->src[0] = brw_imm_f((int64_t)src_as_uint(inst->src[0]) > 0);
+               inst->saturate = false;
+               progress = true;
+               break;
+
+            case BRW_TYPE_HF:
+            case BRW_TYPE_F:
+            case BRW_TYPE_DF:
+               inst->src[0] = brw_imm_f(SATURATE(src_as_float(inst->src[0])));
+               inst->saturate = false;
+               progress = true;
+               break;
+
+            default:
+               /* VF, BF, BF8, HF8 and others are not handled. */
+               break;
             }
          }
          break;

src/intel/compiler/brw/brw_reg.cpp

@@ -8,69 +8,6 @@
 #include "brw_reg.h"
 #include "util/macros.h"
 
-bool
-brw_reg_saturate_immediate(brw_reg *reg)
-{
-   union {
-      unsigned ud;
-      int d;
-      float f;
-      double df;
-   } imm, sat_imm = { 0 };
-
-   const unsigned size = brw_type_size_bytes(reg->type);
-
-   /* We want to either do a 32-bit or 64-bit data copy, the type is otherwise
-    * irrelevant, so just check the size of the type and copy from/to an
-    * appropriately sized field.
-    */
-   if (size < 8)
-      imm.ud = reg->ud;
-   else
-      imm.df = reg->df;
-
-   switch (reg->type) {
-   case BRW_TYPE_UD:
-   case BRW_TYPE_D:
-   case BRW_TYPE_UW:
-   case BRW_TYPE_W:
-   case BRW_TYPE_UQ:
-   case BRW_TYPE_Q:
-      /* Nothing to do. */
-      return false;
-   case BRW_TYPE_F:
-      sat_imm.f = SATURATE(imm.f);
-      break;
-   case BRW_TYPE_DF:
-      sat_imm.df = SATURATE(imm.df);
-      break;
-   case BRW_TYPE_UB:
-   case BRW_TYPE_B:
-      UNREACHABLE("no UB/B immediates");
-   case BRW_TYPE_V:
-   case BRW_TYPE_UV:
-   case BRW_TYPE_VF:
-      UNREACHABLE("unimplemented: saturate vector immediate");
-   case BRW_TYPE_HF:
-      UNREACHABLE("unimplemented: saturate HF immediate");
-   default:
-      UNREACHABLE("invalid type");
-   }
-
-   if (size < 8) {
-      if (imm.ud != sat_imm.ud) {
-         reg->ud = sat_imm.ud;
-         return true;
-      }
-   } else {
-      if (imm.df != sat_imm.df) {
-         reg->df = sat_imm.df;
-         return true;
-      }
-   }
-   return false;
-}
-
 bool
 brw_reg_negate_immediate(brw_reg *reg)
 {

src/intel/compiler/brw/brw_reg.h

@@ -1322,7 +1322,6 @@ element_sz(struct brw_reg reg)
 int brw_float_to_vf(float f);
 float brw_vf_to_float(unsigned char vf);
 
-bool brw_reg_saturate_immediate(brw_reg *reg);
 bool brw_reg_negate_immediate(brw_reg *reg);
 bool brw_reg_abs_immediate(brw_reg *reg);
 

src/intel/compiler/brw/test_opt_algebraic.cpp

@@ -96,4 +96,53 @@ TEST_F(algebraic_test, mad_with_all_immediates)
    EXPECT_SHADERS_MATCH(bld, exp);
 }
 
+TEST_F(algebraic_test, mov_sat_neg_constant_type_d_conversion)
+{
+   brw_builder bld = make_shader(MESA_SHADER_FRAGMENT, 16);
+   brw_builder exp = make_shader(MESA_SHADER_FRAGMENT, 16);
 
+   brw_reg dst0 = vgrf(bld, exp, BRW_TYPE_F);
+
+   bld.MOV(dst0, brw_imm_d(-5))
+      ->saturate = true;
+
+   EXPECT_PROGRESS(brw_opt_algebraic, bld);
+
+   exp.MOV(dst0, brw_imm_f(0.0));
+
+   EXPECT_SHADERS_MATCH(bld, exp);
+}
+
+TEST_F(algebraic_test, mov_sat_neg_constant_type_ud_conversion)
+{
+   brw_builder bld = make_shader(MESA_SHADER_FRAGMENT, 16);
+   brw_builder exp = make_shader(MESA_SHADER_FRAGMENT, 16);
+
+   brw_reg dst0 = vgrf(bld, exp, BRW_TYPE_F);
+
+   bld.MOV(dst0, brw_imm_ud(-5))
+      ->saturate = true;
+
+   EXPECT_PROGRESS(brw_opt_algebraic, bld);
+
+   exp.MOV(dst0, brw_imm_f(1.0));
+
+   EXPECT_SHADERS_MATCH(bld, exp);
+}
+
+TEST_F(algebraic_test, mov_sat_pos_constant_type_conversion)
+{
+   brw_builder bld = make_shader(MESA_SHADER_FRAGMENT, 16);
+   brw_builder exp = make_shader(MESA_SHADER_FRAGMENT, 16);
+
+   brw_reg dst0 = vgrf(bld, exp, BRW_TYPE_F);
+
+   bld.MOV(dst0, brw_imm_ud(37))
+      ->saturate = true;
+
+   EXPECT_PROGRESS(brw_opt_algebraic, bld);
+
+   exp.MOV(dst0, brw_imm_f(1.0));
+
+   EXPECT_SHADERS_MATCH(bld, exp);
+}