GCM and CTR optimizations

In CTR, special case for counter widths of special interest.

In GHASH, uses a 4x reduction technique suggested by Intel.

Split out GHASH to its own source file and header.

With these changes GCM is over twice as fast on Skylake and about
50% faster on Westmere.

3 files changed, 17 insertions, 6 deletions

diff --git a/src/lib/mac/gmac/gmac.cpp b/src/lib/mac/gmac/gmac.cpp
index be27aba4a..a4e84f57b 100644
--- a/src/lib/mac/gmac/gmac.cpp
+++ b/src/lib/mac/gmac/gmac.cpp
@@ -7,6 +7,8 @@
  */
 
 #include <botan/gmac.h>
+#include <botan/ghash.h>
+#include <botan/block_cipher.h>
 
 namespace Botan {
 
@@ -28,6 +30,13 @@ void GMAC::clear()
    m_initialized = false;
    }
 
+GMAC::~GMAC() { /* for unique_ptr */ }
+
+Key_Length_Specification GMAC::key_spec() const
+   {
+   return m_cipher->key_spec();
+   }
+
 std::string GMAC::name() const
    {
    return "GMAC(" + m_cipher->name() + ")";
diff --git a/src/lib/mac/gmac/gmac.h b/src/lib/mac/gmac/gmac.h
index ef54a42bf..83094a5bc 100644
--- a/src/lib/mac/gmac/gmac.h
+++ b/src/lib/mac/gmac/gmac.h
@@ -10,11 +10,12 @@
 #define BOTAN_GMAC_H_
 
 #include <botan/mac.h>
-#include <botan/gcm.h>
-#include <botan/block_cipher.h>
 
 namespace Botan {
 
+class BlockCipher;
+class GHASH;
+
 /**
 * GMAC
 *
@@ -29,10 +30,7 @@ class BOTAN_PUBLIC_API(2,0) GMAC final : public MessageAuthenticationCode
       size_t output_length() const override;
       MessageAuthenticationCode* clone() const override;
 
-      Key_Length_Specification key_spec() const override
-         {
-         return m_cipher->key_spec();
-         }
+      Key_Length_Specification key_spec() const override;
 
       /**
       * Creates a new GMAC instance.
@@ -44,6 +42,8 @@ class BOTAN_PUBLIC_API(2,0) GMAC final : public MessageAuthenticationCode
       GMAC(const GMAC&) = delete;
       GMAC& operator=(const GMAC&) = delete;
 
+      virtual ~GMAC();
+
    private:
       void add_data(const uint8_t[], size_t) override;
       void final_result(uint8_t[]) override;
diff --git a/src/lib/mac/mac.cpp b/src/lib/mac/mac.cpp
index 053e36b6f..65107470b 100644
--- a/src/lib/mac/mac.cpp
+++ b/src/lib/mac/mac.cpp
@@ -19,10 +19,12 @@
 
 #if defined(BOTAN_HAS_GMAC)
   #include <botan/gmac.h>
+  #include <botan/block_cipher.h>
 #endif
 
 #if defined(BOTAN_HAS_HMAC)
   #include <botan/hmac.h>
+  #include <botan/hash.h>
 #endif
 
 #if defined(BOTAN_HAS_POLY1305)