Fix get invokeNative float ret value issue with clang compiler (#1248)

When using clang compiler, the f32/f64 return value might be
invalid when calling invokeNative asm code. Declare the return
type of invokeNative as void, and set volatile for the converted
function pointers to resolve the issue.

Author: wenyongh

core/iwasm/common/wasm_runtime_common.c

@@ -3049,7 +3049,7 @@ wasm_runtime_invoke_native_raw(WASMExecEnv *exec_env, void *func_ptr,
     || defined(BUILD_TARGET_RISCV32_ILP32D)                          \
     || defined(BUILD_TARGET_RISCV32_ILP32) || defined(BUILD_TARGET_ARC)
 typedef void (*GenericFunctionPointer)();
-int64
+void
 invokeNative(GenericFunctionPointer f, uint32 *args, uint32 n_stacks);
 
 typedef float64 (*Float64FuncPtr)(GenericFunctionPointer, uint32 *, uint32);
@@ -3058,13 +3058,16 @@ typedef int64 (*Int64FuncPtr)(GenericFunctionPointer, uint32 *, uint32);
 typedef int32 (*Int32FuncPtr)(GenericFunctionPointer, uint32 *, uint32);
 typedef void (*VoidFuncPtr)(GenericFunctionPointer, uint32 *, uint32);
 
-static Float64FuncPtr invokeNative_Float64 =
+static volatile Float64FuncPtr invokeNative_Float64 =
     (Float64FuncPtr)(uintptr_t)invokeNative;
-static Float32FuncPtr invokeNative_Float32 =
+static volatile Float32FuncPtr invokeNative_Float32 =
     (Float32FuncPtr)(uintptr_t)invokeNative;
-static Int64FuncPtr invokeNative_Int64 = (Int64FuncPtr)(uintptr_t)invokeNative;
-static Int32FuncPtr invokeNative_Int32 = (Int32FuncPtr)(uintptr_t)invokeNative;
-static VoidFuncPtr invokeNative_Void = (VoidFuncPtr)(uintptr_t)invokeNative;
+static volatile Int64FuncPtr invokeNative_Int64 =
+    (Int64FuncPtr)(uintptr_t)invokeNative;
+static volatile Int32FuncPtr invokeNative_Int32 =
+    (Int32FuncPtr)(uintptr_t)invokeNative;
+static volatile VoidFuncPtr invokeNative_Void =
+    (VoidFuncPtr)(uintptr_t)invokeNative;
 
 #if defined(BUILD_TARGET_ARM_VFP) || defined(BUILD_TARGET_THUMB_VFP)
 #define MAX_REG_INTS 4
@@ -3520,7 +3523,7 @@ wasm_runtime_invoke_native(WASMExecEnv *exec_env, void *func_ptr,
     || defined(BUILD_TARGET_THUMB) || defined(BUILD_TARGET_MIPS) \
     || defined(BUILD_TARGET_XTENSA)
 typedef void (*GenericFunctionPointer)();
-int64
+void
 invokeNative(GenericFunctionPointer f, uint32 *args, uint32 sz);
 
 typedef float64 (*Float64FuncPtr)(GenericFunctionPointer f, uint32 *, uint32);
@@ -3529,13 +3532,16 @@ typedef int64 (*Int64FuncPtr)(GenericFunctionPointer f, uint32 *, uint32);
 typedef int32 (*Int32FuncPtr)(GenericFunctionPointer f, uint32 *, uint32);
 typedef void (*VoidFuncPtr)(GenericFunctionPointer f, uint32 *, uint32);
 
-static Int64FuncPtr invokeNative_Int64 = (Int64FuncPtr)invokeNative;
-static Int32FuncPtr invokeNative_Int32 = (Int32FuncPtr)(uintptr_t)invokeNative;
-static Float64FuncPtr invokeNative_Float64 =
+static volatile Int64FuncPtr invokeNative_Int64 =
+    (Int64FuncPtr)(uintptr_t)invokeNative;
+static volatile Int32FuncPtr invokeNative_Int32 =
+    (Int32FuncPtr)(uintptr_t)invokeNative;
+static volatile Float64FuncPtr invokeNative_Float64 =
     (Float64FuncPtr)(uintptr_t)invokeNative;
-static Float32FuncPtr invokeNative_Float32 =
+static volatile Float32FuncPtr invokeNative_Float32 =
     (Float32FuncPtr)(uintptr_t)invokeNative;
-static VoidFuncPtr invokeNative_Void = (VoidFuncPtr)(uintptr_t)invokeNative;
+static volatile VoidFuncPtr invokeNative_Void =
+    (VoidFuncPtr)(uintptr_t)invokeNative;
 
 static inline void
 word_copy(uint32 *dest, uint32 *src, unsigned num)
@@ -3762,33 +3768,25 @@ typedef uint32x4_t __m128i;
 #endif /* end of WASM_ENABLE_SIMD != 0 */
 
 typedef void (*GenericFunctionPointer)();
-#if defined(__APPLE__) || defined(__MACH__)
-/**
- * Define the return type as 'void' in MacOS, since after converting
- * 'int64 invokeNative' into 'float64 invokeNative_Float64', the
- * return value passing might be invalid, the caller reads the return
- * value from register rax but not xmm0.
- */
 void
 invokeNative(GenericFunctionPointer f, uint64 *args, uint64 n_stacks);
-#else
-int64
-invokeNative(GenericFunctionPointer f, uint64 *args, uint64 n_stacks);
-#endif
 
 typedef float64 (*Float64FuncPtr)(GenericFunctionPointer, uint64 *, uint64);
 typedef float32 (*Float32FuncPtr)(GenericFunctionPointer, uint64 *, uint64);
 typedef int64 (*Int64FuncPtr)(GenericFunctionPointer, uint64 *, uint64);
 typedef int32 (*Int32FuncPtr)(GenericFunctionPointer, uint64 *, uint64);
 typedef void (*VoidFuncPtr)(GenericFunctionPointer, uint64 *, uint64);
 
-static Float64FuncPtr invokeNative_Float64 =
+static volatile Float64FuncPtr invokeNative_Float64 =
     (Float64FuncPtr)(uintptr_t)invokeNative;
-static Float32FuncPtr invokeNative_Float32 =
+static volatile Float32FuncPtr invokeNative_Float32 =
     (Float32FuncPtr)(uintptr_t)invokeNative;
-static Int64FuncPtr invokeNative_Int64 = (Int64FuncPtr)(uintptr_t)invokeNative;
-static Int32FuncPtr invokeNative_Int32 = (Int32FuncPtr)(uintptr_t)invokeNative;
-static VoidFuncPtr invokeNative_Void = (VoidFuncPtr)(uintptr_t)invokeNative;
+static volatile Int64FuncPtr invokeNative_Int64 =
+    (Int64FuncPtr)(uintptr_t)invokeNative;
+static volatile Int32FuncPtr invokeNative_Int32 =
+    (Int32FuncPtr)(uintptr_t)invokeNative;
+static volatile VoidFuncPtr invokeNative_Void =
+    (VoidFuncPtr)(uintptr_t)invokeNative;
 
 #if WASM_ENABLE_SIMD != 0
 typedef v128 (*V128FuncPtr)(GenericFunctionPointer, uint64 *, uint64);