nir: add a pass to clamp gl_PointSize to a range

The OpenGL and OpenGL ES specs require that implementations clamp the
value of gl_PointSize to an implementation-depedent range. This pass
is useful for any GPU hardware that doesn't do this automatically
for either one or both sides of the range, such as V3D.

v2:
 - Turn into a generic NIR pass (Eric).
 - Make the pass work before lower I/O so we can use the deref variable
   to inspect if we are writing to gl_PointSize (Eric).
 - Make the pass take the range to clamp as parameter and allow it
   to clamp to both sides of the range or just one side.
 - Make the pass report progress.

v3:
 - Fix copyright header (Eric)
 - use fmin/fmax instead of bcsel to clamp (Eric)

Reviewed-by: Eric Anholt <eric@anholt.net>

diff --git a/src/compiler/nir/meson.build b/src/compiler/nir/meson.build
index 9b527b22f7c196bc4452d8c742f3f599788b2f88..336bf2ef5439bce684007aeaee60a012b32d1e9a 100644 (file)
--- a/src/compiler/nir/meson.build
+++ b/src/compiler/nir/meson.build
@@ -148,6 +148,7 @@ files_libnir = files(
   'nir_lower_passthrough_edgeflags.c',
   'nir_lower_patch_vertices.c',
   'nir_lower_phis_to_scalar.c',
+  'nir_lower_point_size.c',
   'nir_lower_regs_to_ssa.c',
   'nir_lower_returns.c',
   'nir_lower_scratch.c',

diff --git a/src/compiler/nir/nir.h b/src/compiler/nir/nir.h
index 314b696f7386e07484a992552d9bf9fd077cbe28..09a3be5527e44997241a49865182ae4151ad101b 100644 (file)
--- a/src/compiler/nir/nir.h
+++ b/src/compiler/nir/nir.h
@@ -3886,6 +3886,8 @@ bool nir_lower_doubles(nir_shader *shader, const nir_shader *softfp64,
                        nir_lower_doubles_options options);
 bool nir_lower_pack(nir_shader *shader);
 
+bool nir_lower_point_size(nir_shader *shader, float min, float max);
+
 typedef enum {
    nir_lower_interpolation_at_sample = (1 << 1),
    nir_lower_interpolation_at_offset = (1 << 2),

diff --git a/src/compiler/nir/nir_lower_point_size.c b/src/compiler/nir/nir_lower_point_size.c
new file mode 100644 (file)
index 0000000..d614bb2
--- /dev/null
+++ b/src/compiler/nir/nir_lower_point_size.c
@@ -0,0 +1,124 @@
+/*
+ * Copyright © 2019 Raspberry Pi
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "nir_builder.h"
+
+/** @file nir_lower_point_size.c
+ *
+ * The OpenGL spec requires that implementations clamp gl_PointSize to an
+ * implementation-dependant point size range. The OpenGL ES 3.0 spec further
+ * requires that this range must match GL_ALIASED_POINT_SIZE_RANGE.
+ * Some hardware such as V3D don't clamp to a valid range automatically so
+ * the driver must clamp the point size written by the shader manually to a
+ * valid range.
+ */
+
+static void
+lower_point_size_instr(nir_builder *b, nir_instr *psiz_instr,
+                       float min, float max)
+{
+   b->cursor = nir_before_instr(psiz_instr);
+
+   nir_intrinsic_instr *instr = nir_instr_as_intrinsic(psiz_instr);
+
+   /* Some fixed function vertex programs generate PSIZ as a vec4
+    * instead of a scalar, where the actual point size is stored in the
+    * first component.
+    */
+   assert(instr->src[1].is_ssa);
+   nir_ssa_def *psiz = nir_channel(b, instr->src[1].ssa, 0);
+
+   if (min > 0.0f)
+      psiz = nir_fmax(b, psiz, nir_imm_float(b, min));
+
+   if (max > 0.0f)
+      psiz = nir_fmin(b, psiz, nir_imm_float(b, max));
+
+   nir_ssa_def *src_chans[4];
+   src_chans[0] = psiz;
+   for (int i = 1; i < instr->src[1].ssa->num_components; i++)
+      src_chans[i] = nir_channel(b, instr->src[1].ssa, i);
+   nir_ssa_def *lowered_src =
+      nir_vec(b, src_chans, instr->src[1].ssa->num_components);
+
+   nir_instr_rewrite_src(&instr->instr, &instr->src[1],
+                         nir_src_for_ssa(lowered_src));
+}
+
+static bool
+instr_is_point_size(const nir_instr *instr)
+{
+   if (instr->type != nir_instr_type_intrinsic)
+      return false;
+
+   nir_intrinsic_instr *intr = nir_instr_as_intrinsic(instr);
+   if (intr->intrinsic != nir_intrinsic_store_deref)
+      return false;
+
+   nir_deref_instr *deref = nir_src_as_deref(intr->src[0]);
+   nir_variable *var = nir_deref_instr_get_variable(deref);
+   if (var->data.location != VARYING_SLOT_PSIZ)
+      return false;
+
+   return true;
+}
+
+/**
+ * Clamps gl_PointSize to the range [min, max]. If either min or max are not
+ * greater than 0 then no clamping is done for that side of the range.
+ */
+bool
+nir_lower_point_size(nir_shader *s, float min, float max)
+{
+   assert(s->info.stage != MESA_SHADER_FRAGMENT &&
+          s->info.stage != MESA_SHADER_COMPUTE);
+
+   assert(min > 0.0f || max > 0.0f);
+   assert(min <= 0.0f || max <= 0.0f || min <= max);
+
+   bool progress = false;
+   nir_foreach_function(function, s) {
+      if (!function->impl)
+         continue;
+
+      nir_builder b;
+      nir_builder_init(&b, function->impl);
+
+      nir_foreach_block(block, function->impl) {
+         nir_foreach_instr_safe(instr, block) {
+            if (instr_is_point_size(instr)) {
+               lower_point_size_instr(&b, instr, min, max);
+               progress = true;
+            }
+         }
+      }
+
+      if (progress) {
+         nir_metadata_preserve(function->impl,
+                               nir_metadata_block_index |
+                               nir_metadata_dominance);
+      }
+   }
+
+   return progress;
+}