Make use of optimised bswap from ARIA

Signed-off-by: Dave Rodgman <dave.rodgman@arm.com>
diff --git a/library/aria.c b/library/aria.c
index 682d4ec..517e10a 100644
--- a/library/aria.c
+++ b/library/aria.c
@@ -98,47 +98,8 @@
  * modify byte order: ( A B C D ) -> ( D C B A ), i.e. change endianness
  *
  * This is submatrix P3 in [1] Appendix B.1
- *
- * Some compilers fail to translate this to a single instruction,
- * so let's provide asm versions for common platforms with C fallback.
  */
-#if defined(MBEDTLS_HAVE_ASM)
-#if defined(__arm__) /* rev available from v6 up */
-/* armcc5 --gnu defines __GNUC__ but doesn't support GNU's extended asm */
-#if defined(__GNUC__) && \
-    ( !defined(__ARMCC_VERSION) || __ARMCC_VERSION >= 6000000 ) && \
-    __ARM_ARCH >= 6
-static inline uint32_t aria_p3( uint32_t x )
-{
-    uint32_t r;
-    __asm( "rev %0, %1" : "=l" (r) : "l" (x) );
-    return( r );
-}
-#define ARIA_P3 aria_p3
-#elif defined(__ARMCC_VERSION) && __ARMCC_VERSION < 6000000 && \
-    ( __TARGET_ARCH_ARM >= 6 || __TARGET_ARCH_THUMB >= 3 )
-static inline uint32_t aria_p3( uint32_t x )
-{
-    uint32_t r;
-    __asm( "rev r, x" );
-    return( r );
-}
-#define ARIA_P3 aria_p3
-#endif
-#endif /* arm */
-#if defined(__GNUC__) && \
-    defined(__i386__) || defined(__amd64__) || defined( __x86_64__)
-static inline uint32_t aria_p3( uint32_t x )
-{
-    __asm( "bswap %0" : "=r" (x) : "0" (x) );
-    return( x );
-}
-#define ARIA_P3 aria_p3
-#endif /* x86 gnuc */
-#endif /* MBEDTLS_HAVE_ASM && GNUC */
-#if !defined(ARIA_P3)
-#define ARIA_P3(x) ARIA_P2( ARIA_P1 ( x ) )
-#endif
+#define ARIA_P3(x) MBEDTLS_BSWAP32(x)
 
 /*
  * ARIA Affine Transform