Add Windows ARM64 support (#656)

* Add Windows ARM64 support

* Add Windows ARM64/ARM64EC CI workflows

src/snmalloc/aal/aal.h

@@ -14,9 +14,11 @@
 #include <cstdint>
 #include <utility>
 
-#if defined(__i386__) || defined(_M_IX86) || defined(_X86_) || \
+#if ( \
+  defined(__i386__) || defined(_M_IX86) || defined(_X86_) || \
   defined(__amd64__) || defined(__x86_64__) || defined(_M_X64) || \
-  defined(_M_AMD64)
+  defined(_M_AMD64)) && \
+  !defined(_M_ARM64EC)
 #  if defined(SNMALLOC_SGX)
 #    define PLATFORM_IS_X86_SGX
 #    define SNMALLOC_NO_AAL_BUILTINS
@@ -25,7 +27,8 @@
 #  endif
 #endif
 
-#if defined(__arm__) || defined(__aarch64__)
+#if defined(__arm__) || defined(__aarch64__) || defined(_M_ARM64) || \
+  defined(_M_ARM64EC)
 #  define PLATFORM_IS_ARM
 #endif
 

src/snmalloc/aal/aal_arm.h

@@ -1,6 +1,6 @@
 #pragma once
 
-#if defined(__aarch64__)
+#if defined(__aarch64__) || defined(_M_ARM64) || defined(_M_ARM64EC)
 #  define SNMALLOC_VA_BITS_64
 #  ifdef _MSC_VER
 #    include <arm64_neon.h>

src/snmalloc/ds_core/bits.h

@@ -158,7 +158,11 @@ namespace snmalloc
       SNMALLOC_ASSERT(x != 0); // Calling with 0 is UB on some implementations
 
 #if defined(_MSC_VER) && !defined(__clang__)
-#  ifdef _WIN64
+#  if defined(_M_ARM64) || defined(_M_ARM64EC)
+      unsigned long n = 0;
+      _BitScanForward64(&n, static_cast<unsigned __int64>(x));
+      return static_cast<size_t>(n);
+#  elif defined(_WIN64)
       return _tzcnt_u64(static_cast<unsigned __int64>(x));
 #  else
       return _tzcnt_u32(static_cast<unsigned int>(x));
@@ -203,7 +207,12 @@ namespace snmalloc
       overflow = __builtin_mul_overflow(x, y, &prod);
       return prod;
 #elif defined(_MSC_VER)
-#  ifdef _WIN64
+#  if defined(_M_ARM64) || defined(_M_ARM64EC)
+      size_t high_prod = __umulh(x, y);
+      size_t prod = x * y;
+      overflow = high_prod != 0;
+      return prod;
+#  elif defined(_WIN64)
       size_t high_prod;
       size_t prod = _umul128(x, y, &high_prod);
       overflow = high_prod != 0;