Drop GCM to POWER7 on PowerPC

GCM can do some bulk XOR's using the SIMD unit. However, we still need loads and stores to be fast. Fast loads and stores of unaligned data requires the VSX unit

1 files changed, 11 insertions, 11 deletions

diff --git a/gcm.cpp b/gcm.cpp
index 59236ac2..473477c4 100644
--- a/gcm.cpp
+++ b/gcm.cpp
@@ -75,8 +75,8 @@ extern void GCM_Xor16_SSE2(byte *a, const byte *b, const byte *c);
 extern void GCM_Xor16_NEON(byte *a, const byte *b, const byte *c);

 #endif

 

-#if CRYPTOPP_ALTIVEC_AVAILABLE

-extern void GCM_Xor16_ALTIVEC(byte *a, const byte *b, const byte *c);

+#if CRYPTOPP_POWER7_AVAILABLE

+extern void GCM_Xor16_POWER7(byte *a, const byte *b, const byte *c);

 #endif

 

 #if CRYPTOPP_CLMUL_AVAILABLE

@@ -213,11 +213,11 @@ void GCM_Base::SetKeyWithoutResync(const byte *userKey, size_t keylength, const
                     for (k=1; k<j; k++)

                         GCM_Xor16_NEON(mulTable+i*256*16+(j+k)*16, mulTable+i*256*16+j*16, mulTable+i*256*16+k*16);

             else

-#elif CRYPTOPP_ALTIVEC_AVAILABLE

-            if (HasAltivec())

+#elif CRYPTOPP_POWER7_AVAILABLE

+            if (HasPower7())

                 for (j=2; j<=0x80; j*=2)

                     for (k=1; k<j; k++)

-                        GCM_Xor16_ALTIVEC(mulTable+i*256*16+(j+k)*16, mulTable+i*256*16+j*16, mulTable+i*256*16+k*16);

+                        GCM_Xor16_POWER7(mulTable+i*256*16+(j+k)*16, mulTable+i*256*16+j*16, mulTable+i*256*16+k*16);

             else

 #endif

                 for (j=2; j<=0x80; j*=2)

@@ -277,13 +277,13 @@ void GCM_Base::SetKeyWithoutResync(const byte *userKey, size_t keylength, const
                         GCM_Xor16_NEON(mulTable+1024+i*256+(j+k)*16, mulTable+1024+i*256+j*16, mulTable+1024+i*256+k*16);

                     }

             else

-#elif CRYPTOPP_ALTIVEC_AVAILABLE

-            if (HasAltivec())

+#elif CRYPTOPP_POWER7_AVAILABLE

+            if (HasPower7())

                 for (j=2; j<=8; j*=2)

                     for (k=1; k<j; k++)

                     {

-                        GCM_Xor16_ALTIVEC(mulTable+i*256+(j+k)*16, mulTable+i*256+j*16, mulTable+i*256+k*16);

-                        GCM_Xor16_ALTIVEC(mulTable+1024+i*256+(j+k)*16, mulTable+1024+i*256+j*16, mulTable+1024+i*256+k*16);

+                        GCM_Xor16_POWER7(mulTable+i*256+(j+k)*16, mulTable+i*256+j*16, mulTable+i*256+k*16);

+                        GCM_Xor16_POWER7(mulTable+1024+i*256+(j+k)*16, mulTable+1024+i*256+j*16, mulTable+1024+i*256+k*16);

                     }

             else

 #endif

@@ -369,8 +369,8 @@ unsigned int GCM_Base::OptimalDataAlignment() const
         HasSSE2() ? 16 :

 #elif CRYPTOPP_ARM_NEON_AVAILABLE

         HasNEON() ? 4 :

-#elif CRYPTOPP_ALTIVEC_AVAILABLE

-        HasAltivec() ? 16 :

+#elif CRYPTOPP_POWER7_AVAILABLE

+        HasPower7() ? 16 :

 #endif

         GetBlockCipher().OptimalDataAlignment();

 }