Fix a bug that causes a struct field to be initialized twice. (#8619)

We insert field initialization logic at the beginning of every ctor in
`synthesizeCtorBody`, but then immediately inserts another round of
initialization again for explicit ctors in `maybeInsertDefaultInitExpr`,
both called from `SemanticsDeclBodyVisitor::visitAggTypeDecl` right next
to each other.

The fix is to remove `maybeInsertDefaultInitExpr`.

This change also enhances the address aliasing analysis, so that for the
following case:
```
this->member1 = 0;
this->member2 = 0;
this->member1 = param;
```
We can still remove the first assignment to `this->member1` despite
seeing `this->member2=0`, since it is easy to know that `this->member2`
cannot alias with `this->member1`.

Closes #8600.

1 files changed, 40 insertions, 0 deletions

diff --git a/tests/optimization/func-resource-result/init-copy-eliminate.slang b/tests/optimization/func-resource-result/init-copy-eliminate.slang
new file mode 100644
index 000000000..c7c8eadcf
--- /dev/null
+++ b/tests/optimization/func-resource-result/init-copy-eliminate.slang
@@ -0,0 +1,40 @@
+//TEST:SIMPLE(filecheck=CHECK):-target cuda -entry computeMain -stage compute
+
+// Test on a regression that caused a field to be initialized twice in the ctor.
+
+// CHECK-COUNT-1: callOnce{{.*}}(int(1000))
+// CHECK-COUNT-1: {{.*}}->origin_0 = 
+int callOnce(int x)
+{ 
+    *result += 1.0f;  // cause some side effect so the call is not eliminated
+     return 999;
+}
+
+public struct Ray
+{
+    public int calledOnce = callOnce(1000);
+    public float3 origin = {};
+    public float t_min = 0;
+    public float3 dir = {};
+    public float t_max = 0;
+
+    public __init(float3 origin, float3 dir, float t_min = 0.f, float t_max = 1000.0)
+    {
+        this.origin = origin;
+        this.dir = dir;
+        this.t_min = t_min;
+        this.t_max = t_max;
+    }
+
+    public RayDesc to_ray_desc() { return { origin, t_min, dir, t_max }; }
+};
+
+uniform float* result;
+
+[numthreads(1,1,1)]
+void computeMain()
+{
+    Ray r = Ray(float3(1,2,3), float3(4,5,6));
+    RayDesc rd = r.to_ray_desc();
+    *result = rd.Direction.x;
+}