crypto: add intel avx2 wrapper for double sha256 64 byte code

xanimo · Apr 30, 2024 · 42cf500 · 42cf500
1 parent da5973e
commit 42cf500
Showing 1 changed file with 52 additions and 4 deletions.
diff --git a/src/crypto/sha256.cpp b/src/crypto/sha256.cpp
@@ -657,6 +657,7 @@ void TransformD64(unsigned char* out, const unsigned char* in)
 }
 
 typedef void (*transform_type) (uint32_t*, const unsigned char*, size_t);
+typedef void (*transform_type_avx2) (const void*, const uint64_t, void*);
 typedef void (*transform_d64_type)(unsigned char*, const unsigned char*);
 
 /** Define a function pointer for Transform */
@@ -666,6 +667,49 @@ transform_d64_type transfrom_ptr_d64_2way = nullptr;
 transform_d64_type transfrom_ptr_d64_4way = nullptr;
 transform_d64_type transfrom_ptr_d64_8way = nullptr;
 
+#if USE_AVX2
+transform_type_avx2 transform_ptr_avx2 = sha256_avx2;
+
+template<transform_type_avx2 tr>
+void TransformD64WrapperAVX2(unsigned char* out, const unsigned char* in)
+{
+    uint32_t s[8];
+    static const unsigned char padding1[64] = {
+        0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0,    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0,    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0,    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 0
+    };
+    unsigned char buffer2[64] = {
+        0,    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0,    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+        0,    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0
+    };
+    sha256::Initialize(s);
+    tr(in, 1, s);
+    tr(padding1, 1, s);
+    WriteBE32(buffer2 + 0, s[0]);
+    WriteBE32(buffer2 + 4, s[1]);
+    WriteBE32(buffer2 + 8, s[2]);
+    WriteBE32(buffer2 + 12, s[3]);
+    WriteBE32(buffer2 + 16, s[4]);
+    WriteBE32(buffer2 + 20, s[5]);
+    WriteBE32(buffer2 + 24, s[6]);
+    WriteBE32(buffer2 + 28, s[7]);
+    sha256::Initialize(s);
+    tr(buffer2, 1, s);
+    WriteBE32(out + 0, s[0]);
+    WriteBE32(out + 4, s[1]);
+    WriteBE32(out + 8, s[2]);
+    WriteBE32(out + 12, s[3]);
+    WriteBE32(out + 16, s[4]);
+    WriteBE32(out + 20, s[5]);
+    WriteBE32(out + 24, s[6]);
+    WriteBE32(out + 28, s[7]);
+}
+#endif
+
 template<transform_type tr>
 void TransformD64Wrapper(unsigned char* out, const unsigned char* in)
 {
@@ -810,11 +854,15 @@ bool AVXEnabled()
 void inline Initialize_transform_ptr(void)
 {
 #if USE_AVX2 && defined(__linux__)
-    if (__builtin_cpu_supports("avx2"))
-       sha256::transform_ptr = sha256::Transform_AVX2;
+    if (__builtin_cpu_supports("avx2")) {
+        sha256::transform_ptr = sha256::Transform_AVX2;
+        sha256::transfrom_ptr_d64 = sha256::TransformD64WrapperAVX2<sha256_avx2>;       
+    }
 #elif USE_AVX2 && defined(__WIN64__)
-    if (AVXEnabled)
-       sha256::transform_ptr = sha256::Transform_AVX2;
+    if (AVXEnabled) {
+        sha256::transform_ptr = sha256::Transform_AVX2;
+        sha256::transfrom_ptr_d64 = sha256::TransformD64WrapperAVX2<sha256_avx2>;       
+    }
 #endif
 #if defined(USE_ASM) && (defined(__x86_64__) || defined(__amd64__) || defined(__i386__))
     bool have_sse4 = false;