liblzma: Rename crc32_aarch64.h to crc32_arm64.h.

Even though the proper name for the architecture is aarch64, this
project uses ARM64 throughout. So the rename is for consistency.

Additionally, crc32_arm64.h was slightly refactored for the following
changes:

   * Added MSVC, FreeBSD, and macOS support in
     is_arch_extension_supported().

   * crc32_arch_optimized() now checks the size when aligning the
     buffer.

   * crc32_arch_optimized() loop conditions were slightly modified to
     avoid both decrementing the size and incrementing the buffer
     pointer.

   * Use the intrinsic wrappers defined in <arm_acle.h> because GCC and
     Clang name them differently.

   * Minor spacing and comment changes.

1 files changed, 119 insertions, 0 deletions

diff --git a/src/liblzma/check/crc32_arm64.h b/src/liblzma/check/crc32_arm64.h
new file mode 100644
index 00000000..a1888ea7
--- /dev/null
+++ b/src/liblzma/check/crc32_arm64.h
@@ -0,0 +1,119 @@
+///////////////////////////////////////////////////////////////////////////////
+//
+/// \file       crc32_arm64.h
+/// \brief      CRC32 calculation with ARM64 optimization
+//
+//  Authors:    Chenxi Mao
+//              Jia Tan
+//
+//  This file has been put into the public domain.
+//  You can do whatever you want with this file.
+//
+///////////////////////////////////////////////////////////////////////////////
+
+
+#ifndef LZMA_CRC32_ARM64_H
+#define LZMA_CRC32_ARM64_H
+
+// MSVC always has the CRC intrinsics available when building for ARM64
+// there is no need to include any header files.
+#ifndef _MSC_VER
+#	include <arm_acle.h>
+#endif
+
+#if defined(CRC32_GENERIC) && defined(CRC32_ARCH_OPTIMIZED)
+#	if defined(HAVE_GETAUXVAL) || defined(HAVE_ELF_AUX_INFO)
+#		include <sys/auxv.h>
+#	elif defined(_WIN32)
+#		include <processthreadsapi.h>
+#	elif defined(__APPLE__) && defined(HAVE_SYSCTLBYNAME)
+#		include <sys/sysctl.h>
+#	endif
+#endif
+
+// Some EDG-based compilers support ARM64 and define __GNUC__
+// (such as Nvidia's nvcc), but do not support function attributes.
+//
+// NOTE: Build systems check for this too, keep them in sync with this.
+#if (defined(__GNUC__) || defined(__clang__)) && !defined(__EDG__)
+#	define crc_attr_target \
+        __attribute__((__target__("+crc")))
+#else
+#	define crc_attr_target
+#endif
+
+
+crc_attr_target
+static uint32_t
+crc32_arch_optimized(const uint8_t *buf, size_t size, uint32_t crc)
+{
+	crc = ~crc;
+
+	// Align the input buffer because this was shown to be
+	// significantly faster than unaligned accesses.
+	const size_t align_amount = my_min(size, (8 - (uintptr_t)buf) & 7);
+
+	for (const uint8_t *limit = buf + align_amount; buf < limit; ++buf)
+		crc = __crc32b(crc, *buf);
+
+	size -= align_amount;
+
+	// Process 8 bytes at a time. The end point is determined by
+	// ignoring the least significant three bits of size to ensure
+	// we do not process past the bounds of the buffer. This guarentees
+	// that limit is a multiple of 8 and is strictly less than size.
+	for (const uint8_t *limit = buf + (size & ~((size_t)7));
+			buf < limit; buf += 8)
+		crc = __crc32d(crc, aligned_read64le(buf));
+
+	// Process the remaining bytes that are not 8 byte aligned.
+	for (const uint8_t *limit = buf + (size & 7); buf < limit; ++buf)
+		crc = __crc32b(crc, *buf);
+
+	return ~crc;
+}
+
+
+#if defined(CRC32_GENERIC) && defined(CRC32_ARCH_OPTIMIZED)
+static inline bool
+is_arch_extension_supported(void)
+{
+#if defined(HAVE_GETAUXVAL)
+	return (getauxval(AT_HWCAP) & HWCAP_CRC32) != 0;
+
+#elif defined(HAVE_ELF_AUX_INFO)
+	unsigned long feature_flags;
+
+	elf_aux_info(AT_HWCAP, &feature_flags, sizeof(feature_flags));
+	return feature_flags & HWCAP_CRC32 != 0;
+
+#elif defined(_WIN32)
+	return IsProcessorFeaturePresent(
+			PF_ARM_V8_CRC32_INSTRUCTIONS_AVAILABLE);
+
+#elif defined(__APPLE__) && defined(HAVE_SYSCTLBYNAME)
+	int has_crc32 = 0;
+	size_t size = sizeof(has_crc32);
+
+	// The sysctlbyname() function requires a string identifier for the
+	// CPU feature it tests. The Apple documentation lists the string
+	// "hw.optional.armv8_crc32", which can be found here:
+	// (https://developer.apple.com/documentation/kernel/1387446-sysctlbyname/determining_instruction_set_characteristics#3915619)
+	int err = sysctlbyname("hw.optional.armv8_crc32", &has_crc32,
+			&size, NULL, 0);
+
+	return !err && has_crc32;
+
+#else
+	// If a runtime detection method cannot be found, then this must
+	// be a compile time error. The checks in crc_common.h should ensure
+	// a runtime detection method is always found if this function is
+	// built. It would be possible to just return false here, but this
+	// is inefficient for binary size and runtime since only the generic
+	// method could ever be used.
+#	error Runtime detection method unavailable.
+#endif
+}
+#endif
+
+#endif // LZMA_CRC32_ARM64_H