Give each version of SIMD_32 a public bswap()

3 files changed, 29 insertions, 11 deletions

diff --git a/src/utils/simd_32/simd_altivec.h b/src/utils/simd_32/simd_altivec.h
index c3d1a76dc..e1aa62002 100644
--- a/src/utils/simd_32/simd_altivec.h
+++ b/src/utils/simd_32/simd_altivec.h
@@ -168,6 +168,15 @@ class SIMD_Altivec
          return vec_nor(reg, reg);
          }
 
+      SIMD_Altivec bswap() const
+         {
+         __vector unsigned char perm = vec_lvsl(0, (u32bit*)0);
+
+         perm = vec_xor(perm, vec_splat_u8(3));
+
+         return SIMD_Altivec(vec_perm(reg, reg, perm));
+         }
+
       static void transpose(SIMD_Altivec& B0, SIMD_Altivec& B1,
                             SIMD_Altivec& B2, SIMD_Altivec& B3)
          {
diff --git a/src/utils/simd_32/simd_scalar.h b/src/utils/simd_32/simd_scalar.h
index 606923289..5fc20b462 100644
--- a/src/utils/simd_32/simd_scalar.h
+++ b/src/utils/simd_32/simd_scalar.h
@@ -9,6 +9,7 @@
 #define BOTAN_SIMD_SCALAR_H__
 
 #include <botan/loadstor.h>
+#include <botan/bswap.h>
 
 namespace Botan {
 
@@ -170,6 +171,14 @@ class SIMD_Scalar
          return SIMD_Scalar(~R0, ~R1, ~R2, ~R3);
          }
 
+      SIMD_Scalar bswap() const
+         {
+         return SIMD_Scalar(reverse_bytes(R0),
+                            reverse_bytes(R1),
+                            reverse_bytes(R2),
+                            reverse_bytes(R3));
+         }
+
       static void transpose(SIMD_Scalar& B0, SIMD_Scalar& B1,
                             SIMD_Scalar& B2, SIMD_Scalar& B3)
          {
diff --git a/src/utils/simd_32/simd_sse.h b/src/utils/simd_32/simd_sse.h
index fcfe6f203..c45d8032f 100644
--- a/src/utils/simd_32/simd_sse.h
+++ b/src/utils/simd_32/simd_sse.h
@@ -121,6 +121,17 @@ class SIMD_SSE2
          return _mm_xor_si128(reg, all_ones);
          }
 
+      SIMD_SSE2 bswap() const
+         {
+         __m128i T = reg;
+
+         T = _mm_shufflehi_epi16(T, _MM_SHUFFLE(2, 3, 0, 1));
+         T = _mm_shufflelo_epi16(T, _MM_SHUFFLE(2, 3, 0, 1));
+
+         return _mm_or_si128(_mm_srli_epi16(T, 8),
+                             _mm_slli_epi16(T, 8));
+         }
+
       static void transpose(SIMD_SSE2& B0, SIMD_SSE2& B1,
                             SIMD_SSE2& B2, SIMD_SSE2& B3)
          {
@@ -137,17 +148,6 @@ class SIMD_SSE2
    private:
       SIMD_SSE2(__m128i in) { reg = in; }
 
-      SIMD_SSE2 bswap() const
-         {
-         __m128i T = reg;
-
-         T = _mm_shufflehi_epi16(T, _MM_SHUFFLE(2, 3, 0, 1));
-         T = _mm_shufflelo_epi16(T, _MM_SHUFFLE(2, 3, 0, 1));
-
-         return _mm_or_si128(_mm_srli_epi16(T, 8),
-                             _mm_slli_epi16(T, 8));
-         }
-
       __m128i reg;
    };