MDEV-21362 do something with -fno-builtin-memcmp for rem0cmp.cc

Such CMake check is not relevant as the oldest supported gcc is 4.8 and
compiler issue was solved in gcc-4.6 already.
https://godbolt.org/z/7G64qo

cmp_data(): simplify code

storage/innobase/innodb.cmake

@@ -93,17 +93,6 @@ IF(CMAKE_CXX_COMPILER_ID MATCHES "GNU")
 ## As of Mar 15 2011 this flag causes 3573+ warnings. If you are reading this
 ## please fix them and enable the following code:
 #SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wconversion")
-
-  IF (CMAKE_SYSTEM_PROCESSOR MATCHES "x86_64" OR
-      CMAKE_SYSTEM_PROCESSOR MATCHES "i386")
-    INCLUDE(CheckCXXCompilerFlag)
-    CHECK_CXX_COMPILER_FLAG("-fno-builtin-memcmp" HAVE_NO_BUILTIN_MEMCMP)
-    IF (HAVE_NO_BUILTIN_MEMCMP)
-      # Work around http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43052
-      SET_SOURCE_FILES_PROPERTIES(${CMAKE_CURRENT_SOURCE_DIR}/rem/rem0cmp.cc
-	PROPERTIES COMPILE_FLAGS -fno-builtin-memcmp)
-    ENDIF()
-  ENDIF()
 ENDIF()
 
 # Enable InnoDB's UNIV_DEBUG in debug builds

storage/innobase/rem/rem0cmp.cc

@@ -449,64 +449,19 @@ cmp_data(
 				       data2, (unsigned) len2));
 	}
 
-	ulint	len;
-	int	cmp;
+	ulint len = std::min(len1, len2);
 
-	if (len1 < len2) {
-		len = len1;
-		len2 -= len;
-		len1 = 0;
-	} else {
-		len = len2;
-		len1 -= len;
-		len2 = 0;
-	}
-
-	if (len) {
-#if defined __i386__ || defined __x86_64__ || defined _M_IX86 || defined _M_X64
-		/* Compare the first bytes with a loop to avoid the call
-		overhead of memcmp(). On x86 and x86-64, the GCC built-in
-		(repz cmpsb) seems to be very slow, so we will be calling the
-		libc version. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43052
-		tracks the slowness of the GCC built-in memcmp().
-
-		We compare up to the first 4..7 bytes with the loop.
-		The (len & 3) is used for "normalizing" or
-		"quantizing" the len parameter for the memcmp() call,
-		in case the whole prefix is equal. On x86 and x86-64,
-		the GNU libc memcmp() of equal strings is faster with
-		len=4 than with len=3.
-
-		On other architectures than the IA32 or AMD64, there could
-		be a built-in memcmp() that is faster than the loop.
-		We only use the loop where we know that it can improve
-		the performance. */
-		for (ulint i = 4 + (len & 3); i > 0; i--) {
-			cmp = int(*data1++) - int(*data2++);
-			if (cmp) {
-				return(cmp);
-			}
+	int cmp = memcmp(data1, data2, len);
 
-			if (!--len) {
-				break;
-			}
-		}
-
-		if (len) {
-#endif /* IA32 or AMD64 */
-			cmp = memcmp(data1, data2, len);
-
-			if (cmp) {
-				return(cmp);
-			}
-
-			data1 += len;
-			data2 += len;
-#if defined __i386__ || defined __x86_64__ || defined _M_IX86 || defined _M_X64
-		}
-#endif /* IA32 or AMD64 */
+	if (cmp) {
+		return (cmp);
 	}
 
+	data1 += len;
+	data2 += len;
+	len1 -= len;
+	len2 -= len;
+
 	cmp = (int) (len1 - len2);
 
 	if (!cmp || pad == ULINT_UNDEFINED) {