Merge optimized implementations into base class

Various algorithms had an optimized implementation (for SSE2, AVX2, etc)
which was offered alongside the 'base' implementation. This is
admittedly very useful for testing, but it breaks user expectations in
bad ways. See GH #477 for background.

Now encrypting with `AES_128` (say) just runs whatever implementation
is best on the current processor/build.

2 files changed, 47 insertions, 13 deletions

diff --git a/src/lib/block/noekeon/noekeon.cpp b/src/lib/block/noekeon/noekeon.cpp
index 01f7491f3..5e7c0229e 100644
--- a/src/lib/block/noekeon/noekeon.cpp
+++ b/src/lib/block/noekeon/noekeon.cpp
@@ -7,6 +7,7 @@
 
 #include <botan/noekeon.h>
 #include <botan/loadstor.h>
+#include <botan/cpuid.h>
 
 namespace Botan {
 
@@ -85,6 +86,19 @@ const byte Noekeon::RC[] = {
 */
 void Noekeon::encrypt_n(const byte in[], byte out[], size_t blocks) const
    {
+#if defined(BOTAN_HAS_NOEKEON_SIMD)
+   if(CPUID::has_simd_32())
+      {
+      while(blocks >= 4)
+         {
+         simd_encrypt_4(in, out);
+         in += 4 * BLOCK_SIZE;
+         out += 4 * BLOCK_SIZE;
+         blocks -= 4;
+         }
+      }
+#endif
+
    for(size_t i = 0; i != blocks; ++i)
       {
       u32bit A0 = load_be<u32bit>(in, 0);
@@ -123,6 +137,32 @@ void Noekeon::encrypt_n(const byte in[], byte out[], size_t blocks) const
 */
 void Noekeon::decrypt_n(const byte in[], byte out[], size_t blocks) const
    {
+#if defined(BOTAN_HAS_NOEKEON_SIMD)
+   if(CPUID::has_simd_32())
+      {
+      /*
+      const size_t blocks4 = blocks / 4;
+      const size_t blocks_left = blocks % 4;
+
+      in += blocks4 * BLOCK_SIZE;
+      out += blocks4 * BLOCK_SIZE;
+      blocks = blocks % 4;
+
+      BOTAN_PARALLEL_FOR(size_t i = 0; i < blocks4; ++i)
+         {
+         simd_encrypt_4(in + i*4*BLOCK_SIZE, out + i*4*BLOCK_SIZE);
+         }
+      */
+      while(blocks >= 4)
+         {
+         simd_decrypt_4(in, out);
+         in += 4 * BLOCK_SIZE;
+         out += 4 * BLOCK_SIZE;
+         blocks -= 4;
+         }
+      }
+#endif
+
    for(size_t i = 0; i != blocks; ++i)
       {
       u32bit A0 = load_be<u32bit>(in, 0);
diff --git a/src/lib/block/noekeon/noekeon.h b/src/lib/block/noekeon/noekeon.h
index 4a3b9de0c..30c15a001 100644
--- a/src/lib/block/noekeon/noekeon.h
+++ b/src/lib/block/noekeon/noekeon.h
@@ -15,7 +15,7 @@ namespace Botan {
 /**
 * Noekeon
 */
-class BOTAN_DLL Noekeon : public Block_Cipher_Fixed_Params<16, 16>
+class BOTAN_DLL Noekeon final : public Block_Cipher_Fixed_Params<16, 16>
    {
    public:
       void encrypt_n(const byte in[], byte out[], size_t blocks) const override;
@@ -24,23 +24,17 @@ class BOTAN_DLL Noekeon : public Block_Cipher_Fixed_Params<16, 16>
       void clear() override;
       std::string name() const override { return "Noekeon"; }
       BlockCipher* clone() const override { return new Noekeon; }
-   protected:
+   private:
+#if defined(BOTAN_HAS_NOEKEON_SIMD)
+      void simd_encrypt_4(const byte in[], byte out[]) const;
+      void simd_decrypt_4(const byte in[], byte out[]) const;
+#endif
+
       /**
       * The Noekeon round constants
       */
       static const byte RC[17];
 
-      /**
-      * @return const reference to encryption subkeys
-      */
-      const secure_vector<u32bit>& get_EK() const { return m_EK; }
-
-      /**
-      * @return const reference to decryption subkeys
-      */
-      const secure_vector<u32bit>& get_DK() const { return m_DK; }
-
-   private:
       void key_schedule(const byte[], size_t) override;
       secure_vector<u32bit> m_EK, m_DK;
    };