Add sha256_arm_shani to build system

Also rename AArch64 intrinsics to ARMv8 intrinsics
as these are not necessarily limited to 64-bit

Author: prusnak

configure.ac

@@ -569,10 +569,11 @@ CXXFLAGS="$TEMP_CXXFLAGS"
 
 # ARM
 AX_CHECK_COMPILE_FLAG([-march=armv8-a+crc+crypto], [ARM_CRC_CXXFLAGS="-march=armv8-a+crc+crypto"], [], [$CXXFLAG_WERROR])
+AX_CHECK_COMPILE_FLAG([-march=armv8-a+crc+crypto], [ARM_SHANI_CXXFLAGS="-march=armv8-a+crc+crypto"], [], [$CXXFLAG_WERROR])
 
 TEMP_CXXFLAGS="$CXXFLAGS"
 CXXFLAGS="$CXXFLAGS $ARM_CRC_CXXFLAGS"
-AC_MSG_CHECKING([for AArch64 CRC32 intrinsics])
+AC_MSG_CHECKING([for ARMv8 CRC32 intrinsics])
 AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[
     #include <arm_acle.h>
     #include <arm_neon.h>
@@ -589,6 +590,24 @@ AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[
 )
 CXXFLAGS="$TEMP_CXXFLAGS"
 
+TEMP_CXXFLAGS="$CXXFLAGS"
+CXXFLAGS="$CXXFLAGS $ARM_SHANI_CXXFLAGS"
+AC_MSG_CHECKING([for ARMv8 SHA-NI intrinsics])
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[
+    #include <arm_acle.h>
+    #include <arm_neon.h>
+  ]],[[
+    uint32x4_t a, b, c;
+    vsha256h2q_u32(a, b, c);
+    vsha256hq_u32(a, b, c);
+    vsha256su0q_u32(a, b);
+    vsha256su1q_u32(a, b, c);
+  ]])],
+ [ AC_MSG_RESULT([yes]); enable_arm_shani=yes; AC_DEFINE([ENABLE_ARM_SHANI], [1], [Define this symbol to build code that uses ARMv8 SHA-NI intrinsics]) ],
+ [ AC_MSG_RESULT([no])]
+)
+CXXFLAGS="$TEMP_CXXFLAGS"
+
 fi
 
 CPPFLAGS="$CPPFLAGS -DHAVE_BUILD_INFO"
@@ -1808,6 +1827,7 @@ AM_CONDITIONAL([ENABLE_SSE41], [test "$enable_sse41" = "yes"])
 AM_CONDITIONAL([ENABLE_AVX2], [test "$enable_avx2" = "yes"])
 AM_CONDITIONAL([ENABLE_X86_SHANI], [test "$enable_x86_shani" = "yes"])
 AM_CONDITIONAL([ENABLE_ARM_CRC], [test "$enable_arm_crc" = "yes"])
+AM_CONDITIONAL([ENABLE_ARM_SHANI], [test "$enable_arm_shani" = "yes"])
 AM_CONDITIONAL([USE_ASM], [test "$use_asm" = "yes"])
 AM_CONDITIONAL([WORDS_BIGENDIAN], [test "$ac_cv_c_bigendian" = "yes"])
 AM_CONDITIONAL([USE_NATPMP], [test "$use_natpmp" = "yes"])
@@ -1866,6 +1886,7 @@ AC_SUBST(CLMUL_CXXFLAGS)
 AC_SUBST(AVX2_CXXFLAGS)
 AC_SUBST(X86_SHANI_CXXFLAGS)
 AC_SUBST(ARM_CRC_CXXFLAGS)
+AC_SUBST(ARM_SHANI_CXXFLAGS)
 AC_SUBST(LIBTOOL_APP_LDFLAGS)
 AC_SUBST(USE_SQLITE)
 AC_SUBST(USE_BDB)

src/Makefile.am

@@ -50,6 +50,10 @@ if ENABLE_X86_SHANI
 LIBBITCOIN_CRYPTO_X86_SHANI = crypto/libbitcoin_crypto_x86_shani.a
 LIBBITCOIN_CRYPTO += $(LIBBITCOIN_CRYPTO_X86_SHANI)
 endif
+if ENABLE_ARM_SHANI
+LIBBITCOIN_CRYPTO_ARM_SHANI = crypto/libbitcoin_crypto_arm_shani.a
+LIBBITCOIN_CRYPTO += $(LIBBITCOIN_CRYPTO_ARM_SHANI)
+endif
 
 $(LIBSECP256K1): $(wildcard secp256k1/src/*.h) $(wildcard secp256k1/src/*.c) $(wildcard secp256k1/include/*)
 	$(AM_V_at)$(MAKE) $(AM_MAKEFLAGS) -C $(@D) $(@F)
@@ -504,6 +508,12 @@ crypto_libbitcoin_crypto_x86_shani_a_CXXFLAGS += $(X86_SHANI_CXXFLAGS)
 crypto_libbitcoin_crypto_x86_shani_a_CPPFLAGS += -DENABLE_X86_SHANI
 crypto_libbitcoin_crypto_x86_shani_a_SOURCES = crypto/sha256_x86_shani.cpp
 
+crypto_libbitcoin_crypto_arm_shani_a_CXXFLAGS = $(AM_CXXFLAGS) $(PIE_FLAGS)
+crypto_libbitcoin_crypto_arm_shani_a_CPPFLAGS = $(AM_CPPFLAGS)
+crypto_libbitcoin_crypto_arm_shani_a_CXXFLAGS += $(ARM_SHANI_CXXFLAGS)
+crypto_libbitcoin_crypto_arm_shani_a_CPPFLAGS += -DENABLE_ARM_SHANI
+crypto_libbitcoin_crypto_arm_shani_a_SOURCES = crypto/sha256_arm_shani.cpp
+
 # consensus: shared between all executables that validate any consensus rules.
 libbitcoin_consensus_a_CPPFLAGS = $(AM_CPPFLAGS) $(BITCOIN_INCLUDES)
 libbitcoin_consensus_a_CXXFLAGS = $(AM_CXXFLAGS) $(PIE_FLAGS)

src/crypto/sha256.cpp

@@ -10,6 +10,16 @@
 
 #include <compat/cpuid.h>
 
+#if defined(__linux__) && defined(ENABLE_ARM_SHANI) && !defined(BUILD_BITCOIN_INTERNAL)
+#include <sys/auxv.h>
+#include <asm/hwcap.h>
+#endif
+
+#if defined(MAC_OSX) && defined(ENABLE_ARM_SHANI) && !defined(BUILD_BITCOIN_INTERNAL)
+#include <sys/types.h>
+#include <sys/sysctl.h>
+#endif
+
 #if defined(__x86_64__) || defined(__amd64__) || defined(__i386__)
 #if defined(USE_ASM)
 namespace sha256_sse4
@@ -39,6 +49,11 @@ namespace sha256_x86_shani
 void Transform(uint32_t* s, const unsigned char* chunk, size_t blocks);
 }
 
+namespace sha256_arm_shani
+{
+void Transform(uint32_t* s, const unsigned char* chunk, size_t blocks);
+}
+
 // Internal implementation code.
 namespace
 {
@@ -623,6 +638,37 @@ std::string SHA256AutoDetect()
 #endif
 #endif
 
+#if defined(ENABLE_ARM_SHANI) && !defined(BUILD_BITCOIN_INTERNAL)
+    bool have_arm_shani = false;
+
+#if defined(__linux__)
+#if defined(__arm__) // 32-bit
+    if (getauxval(AT_HWCAP2) & HWCAP2_SHA2) {
+        have_arm_shani = true;
+    }
+#endif
+#if defined(__aarch64__) // 64-bit
+    if (getauxval(AT_HWCAP) & HWCAP_SHA2) {
+        have_arm_shani = true;
+    }
+#endif
+#endif
+
+#if defined(MAC_OSX)
+    int val = 0;
+    size_t len = sizeof(val);
+    if (sysctlbyname("hw.optional.arm.FEAT_SHA256", &val, &len, nullptr, 0) == 0) {
+        have_arm_shani = val != 0;
+    }
+#endif
+
+    if (have_arm_shani) {
+        Transform = sha256_arm_shani::Transform;
+        TransformD64 = TransformD64Wrapper<sha256_arm_shani::Transform>;
+        ret = "arm_shani(1way)";
+    }
+#endif
+
     assert(SelfTest());
     return ret;
 }

src/crypto/sha256_arm_shani.cpp

@@ -0,0 +1,22 @@
+// Copyright (c) 2022 The Bitcoin Core developers
+// Distributed under the MIT software license, see the accompanying
+// file COPYING or http://www.opensource.org/licenses/mit-license.php.
+//
+// Based on https://github.com/noloader/SHA-Intrinsics/blob/master/sha256-arm.c,
+// Written and placed in public domain by Jeffrey Walton.
+// Based on code from ARM, and by Johannes Schneiders, Skip Hovsmith and
+// Barry O'Rourke for the mbedTLS project.
+
+#ifdef ENABLE_ARM_SHANI
+
+#include <cstdint>
+#include <cstddef>
+
+namespace sha256_arm_shani {
+void Transform(uint32_t* s, const unsigned char* chunk, size_t blocks)
+{
+
+}
+}
+
+#endif