Implement SDMA copy rect for gfx9.

Fix pitch overflow due to small element detection.
Add wide pitch 2D copy handling.
Cleanup code duplication.

Change-Id: I93b1584aba8e5964957eb7ab3544df806ca3e2f9

runtime/hsa-runtime/core/util/utils.h

@@ -176,6 +176,11 @@ static __forceinline T Min(const T& a, const T& b) {
   return (a > b) ? b : a;
 }
 
+template <class T, class... Arg>
+static __forceinline T Min(const T& a, const T& b, Arg... args) {
+  return Min(a, Min(b, args...));
+}
+
 /// @brief: Find out the max one of two inputs, input must support ">" operator.
 /// @param: a(Input), a reference to type T.
 /// @param: b(Input), a reference to type T.
@@ -185,6 +190,11 @@ static __forceinline T Max(const T& a, const T& b) {
   return (b > a) ? b : a;
 }
 
+template <class T, class... Arg>
+static __forceinline T Max(const T& a, const T& b, Arg... args) {
+  return Max(a, Max(b, args...));
+}
+
 /// @brief: Free the memory space which is newed previously.
 /// @param: ptr(Input), a pointer to memory space. Can't be NULL.
 /// @return: void.