[Optimize Instructions] Simplify zero/sign extentions (special case) (#4009)

For signed or unsigned extension to 64-bits after lowering from partially filled 64-bit arguments:
```rust
i64.extend_i32_u(i32.wrap_i64(x))  =>  x   // where maxBits(x) <= 32
i64.extend_i32_s(i32.wrap_i64(x))  =>  x   // where maxBits(x) <= 31
```

2 files changed, 76 insertions, 11 deletions

diff --git a/src/passes/OptimizeInstructions.cpp b/src/passes/OptimizeInstructions.cpp
index 31ed9348a..2e6a2d48d 100644
--- a/src/passes/OptimizeInstructions.cpp
+++ b/src/passes/OptimizeInstructions.cpp
@@ -841,19 +841,35 @@ struct OptimizeInstructions
         }
       }
       {
-        Unary* inner;
+        // i64.extend_i32_s(i32.wrap_i64(x))  =>  x
+        //   where maxBits(x) <= 31
+        //
+        // i64.extend_i32_u(i32.wrap_i64(x))  =>  x
+        //   where maxBits(x) <= 32
         Expression* x;
-        if (getModule()->features.hasSignExt()) {
-          // i64.extend_i32_s(i32.wrap_i64(x))  =>  i64.extend32_s(x)
-          if (matches(curr,
-                      unary(ExtendSInt32, unary(&inner, WrapInt64, any(&x))))) {
-            inner->op = ExtendS32Int64;
-            inner->type = Type::i64;
-            inner->value = x;
-            return replaceCurrent(inner);
+        UnaryOp unaryOp;
+        if (matches(curr, unary(&unaryOp, unary(WrapInt64, any(&x))))) {
+          if (unaryOp == ExtendSInt32 || unaryOp == ExtendUInt32) {
+            auto maxBits = Bits::getMaxBits(x, this);
+            if ((unaryOp == ExtendSInt32 && maxBits <= 31) ||
+                (unaryOp == ExtendUInt32 && maxBits <= 32)) {
+              return replaceCurrent(x);
+            }
           }
         }
       }
+      if (getModule()->features.hasSignExt()) {
+        // i64.extend_i32_s(i32.wrap_i64(x))  =>  i64.extend32_s(x)
+        Unary* inner;
+        Expression* x;
+        if (matches(curr,
+                    unary(ExtendSInt32, unary(&inner, WrapInt64, any(&x))))) {
+          inner->op = ExtendS32Int64;
+          inner->type = Type::i64;
+          inner->value = x;
+          return replaceCurrent(inner);
+        }
+      }
     }
 
     if (Abstract::hasAnyReinterpret(curr->op)) {
diff --git a/test/lit/passes/optimize-instructions.wast b/test/lit/passes/optimize-instructions.wast
index 07f246b49..352e70e26 100644
--- a/test/lit/passes/optimize-instructions.wast
+++ b/test/lit/passes/optimize-instructions.wast
@@ -11252,7 +11252,7 @@
   ;; i32.wrap_i64(i64.extend_i32_s(x))  ==>  x
   ;; i32.wrap_i64(i64.extend_i32_u(x))  ==>  x
 
-  ;; CHECK:      (func $sign-and-zero-extention-elimination (param $x i32)
+  ;; CHECK:      (func $sign-and-zero-extention-elimination-1 (param $x i32)
   ;; CHECK-NEXT:  (drop
   ;; CHECK-NEXT:   (local.get $x)
   ;; CHECK-NEXT:  )
@@ -11260,10 +11260,59 @@
   ;; CHECK-NEXT:   (local.get $x)
   ;; CHECK-NEXT:  )
   ;; CHECK-NEXT: )
-  (func $sign-and-zero-extention-elimination (param $x i32)
+  (func $sign-and-zero-extention-elimination-1 (param $x i32)
     (drop (i32.wrap_i64 (i64.extend_i32_s (local.get $x))))
     (drop (i32.wrap_i64 (i64.extend_i32_u (local.get $x))))
   )
+  ;; i64.extend_i32_u(i32.wrap_i64(x))  =>  x,  where maxBits(x) <= 32
+  ;; i64.extend_i32_s(i32.wrap_i64(x))  =>  x,  where maxBits(x) <= 31
+
+  ;; CHECK:      (func $sign-and-zero-extention-elimination-2 (param $x i64)
+  ;; CHECK-NEXT:  (drop
+  ;; CHECK-NEXT:   (i64.and
+  ;; CHECK-NEXT:    (local.get $x)
+  ;; CHECK-NEXT:    (i64.const 4294967295)
+  ;; CHECK-NEXT:   )
+  ;; CHECK-NEXT:  )
+  ;; CHECK-NEXT:  (drop
+  ;; CHECK-NEXT:   (i64.and
+  ;; CHECK-NEXT:    (local.get $x)
+  ;; CHECK-NEXT:    (i64.const 2147483647)
+  ;; CHECK-NEXT:   )
+  ;; CHECK-NEXT:  )
+  ;; CHECK-NEXT:  (drop
+  ;; CHECK-NEXT:   (i64.extend_i32_u
+  ;; CHECK-NEXT:    (i32.wrap_i64
+  ;; CHECK-NEXT:     (local.get $x)
+  ;; CHECK-NEXT:    )
+  ;; CHECK-NEXT:   )
+  ;; CHECK-NEXT:  )
+  ;; CHECK-NEXT:  (drop
+  ;; CHECK-NEXT:   (i64.extend_i32_s
+  ;; CHECK-NEXT:    (i32.wrap_i64
+  ;; CHECK-NEXT:     (local.get $x)
+  ;; CHECK-NEXT:    )
+  ;; CHECK-NEXT:   )
+  ;; CHECK-NEXT:  )
+  ;; CHECK-NEXT:  (drop
+  ;; CHECK-NEXT:   (i64.extend_i32_s
+  ;; CHECK-NEXT:    (i32.wrap_i64
+  ;; CHECK-NEXT:     (i64.and
+  ;; CHECK-NEXT:      (local.get $x)
+  ;; CHECK-NEXT:      (i64.const 4294967295)
+  ;; CHECK-NEXT:     )
+  ;; CHECK-NEXT:    )
+  ;; CHECK-NEXT:   )
+  ;; CHECK-NEXT:  )
+  ;; CHECK-NEXT: )
+  (func $sign-and-zero-extention-elimination-2 (param $x i64)
+    (drop (i64.extend_i32_u (i32.wrap_i64 (i64.and (local.get $x) (i64.const 0x00000000FFFFFFFF)))))
+    (drop (i64.extend_i32_s (i32.wrap_i64 (i64.and (local.get $x) (i64.const 0x000000007FFFFFFF)))))
+
+    (drop (i64.extend_i32_u (i32.wrap_i64 (local.get $x)))) ;; skip
+    (drop (i64.extend_i32_s (i32.wrap_i64 (local.get $x)))) ;; skip
+    (drop (i64.extend_i32_s (i32.wrap_i64 (i64.and (local.get $x) (i64.const 0x00000000FFFFFFFF))))) ;; skip
+  )
   ;; CHECK:      (func $optimize-shifts (param $x i32) (param $y i32) (param $z i64) (param $w i64)
   ;; CHECK-NEXT:  (drop
   ;; CHECK-NEXT:   (local.get $x)