Register implied expression reads in OpLoad/OpAccessChain.

This is required to avoid relying on complex sub-expression elimination
in compilers, and generates cleaner code.

The problem case is if a complex expression is used in an access chain,
like:

Composite comp = buffer[texture(...)];
vec4 a = comp.a + comp.b + comp.c;

Before, we did not have common subexpression tracking for
OpLoad/OpAccessChain, so we easily ended up with code like:

vec4 a = buffer[texture(...)].a + buffer[texture(...)].b + buffer[texture(...)].c;

A good compiler will optimize this, but we should not rely on it, and
forcing texture(...) to a temporary also looks better.

The solution is to add a vector "implied_expression_reads", which works
similarly to expression_dependencies. We also need an extra mechanism in
to_expression which lets us skip expression read checking and do it
later. E.g. for expr -> access chain -> load, we should only trigger
a read of expr when using the loaded expression.

1 files changed, 29 insertions, 0 deletions

diff --git a/shaders-hlsl/frag/complex-expression-in-access-chain.frag b/shaders-hlsl/frag/complex-expression-in-access-chain.frag
new file mode 100644
index 00000000..47f93931
--- /dev/null
+++ b/shaders-hlsl/frag/complex-expression-in-access-chain.frag
@@ -0,0 +1,29 @@
+#version 310 es
+precision mediump float;
+
+struct Foo
+{
+	vec4 a;
+	vec4 b;
+};
+
+layout(binding = 0) buffer UBO
+{
+	vec4 results[1024];
+};
+
+layout(binding = 1) uniform highp isampler2D Buf;
+layout(location = 0) flat in int vIn;
+layout(location = 1) flat in int vIn2;
+
+layout(location = 0) out vec4 FragColor;
+
+void main()
+{
+	ivec4 coords = texelFetch(Buf, ivec2(gl_FragCoord.xy), 0);
+	vec4 foo = results[coords.x % 16];
+
+	int c = vIn * vIn;
+	int d = vIn2 * vIn2;
+	FragColor = foo + foo + results[c + d];
+}