[cmake] enable clang-cl and WoA builds (#348)

Compilation and CMake fixes for both Windows on ARM and clang-cl, meaning Windows can now be built on both MSVC and clang on both amd64 and aarch64.

Compiling on clang is *dramatically* faster so this should be useful for CI.

Co-authored-by: crueter <crueter@eden-emu.dev>
Co-authored-by: crueter <crueter@crueter.xyz>
Reviewed-on: #348
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>
Reviewed-by: crueter <crueter@eden-emu.dev>
Co-authored-by: lizzie <lizzie@eden-emu.dev>
Co-committed-by: lizzie <lizzie@eden-emu.dev>

externals/CMakeLists.txt

@@ -63,7 +63,14 @@ if (mbedtls_ADDED)
     if (NOT MSVC)
         target_compile_options(mbedcrypto PRIVATE
             -Wno-unused-but-set-variable
-            -Wno-string-concatenation)
+            -Wno-string-concatenation
+        )
+    elseif(CXX_CLANG)
+        foreach(TARGET mbedtls mbedcrypto mbedx509)
+            target_compile_options(${TARGET} PRIVATE
+                -w
+            )
+        endforeach()
     endif()
 endif()
 
@@ -84,6 +91,8 @@ if(MSVC AND USE_CCACHE AND sirit_ADDED)
     list(FILTER _opts EXCLUDE REGEX "/Zi")
     list(APPEND _opts "/Z7")
     set_target_properties(sirit PROPERTIES COMPILE_OPTIONS "${_opts}")
+elseif(MSVC AND CXX_CLANG)
+    target_compile_options(sirit PRIVATE -Wno-error=unused-command-line-argument)
 endif()
 
 # httplib
@@ -136,8 +145,16 @@ add_subdirectory(nx_tzdb)
 # VMA
 AddJsonPackage(vulkan-memory-allocator)
 
-if (VulkanMemoryAllocator_ADDED AND MSVC)
-    target_compile_options(VulkanMemoryAllocator INTERFACE /wd4189)
+if (VulkanMemoryAllocator_ADDED)
+    if (CXX_CLANG)
+        target_compile_options(VulkanMemoryAllocator INTERFACE
+            -Wno-unused-variable
+        )
+    elseif(MSVC)
+        target_compile_options(VulkanMemoryAllocator INTERFACE
+            /wd4189
+        )
+    endif()
 endif()
 
 if (NOT TARGET LLVM::Demangle)

externals/cpmfile.json

@@ -1,12 +1,9 @@
 {
     "mbedtls": {
-        "repo": "Mbed-TLS/mbedtls",
-        "sha": "8c88150ca1",
-        "hash": "769ad1e94c570671071e1f2a5c0f1027e0bf6bcdd1a80ea8ac970f2c86bc45ce4e31aa88d6d8110fc1bed1de81c48bc624df1b38a26f8b340a44e109d784a966",
-        "find_args": "MODULE",
-        "patches": [
-            "0001-cmake-version.patch"
-        ]
+        "repo": "eden-emulator/mbedtls",
+        "sha": "ce4f81f4a9",
+        "hash": "f2e7f887651b28745e508149214d409fd7cfdb92cb94b4146b47ff1e0fc09e47143f203ac18e34c2c1814b5bd031d04c74828676c0d4342920a2ddb7fd35e9a5",
+        "find_args": "MODULE"
     },
     "spirv-headers": {
         "package": "SPIRV-Headers",
@@ -29,18 +26,12 @@
     },
     "cpp-jwt": {
         "version": "1.4",
-        "repo": "arun11299/cpp-jwt",
-        "sha": "a54fa08a3b",
-        "hash": "a90f7e594ada0c7e49d5ff9211c71097534e7742a8e44bf0851b0362642a7271d53f5d83d04eeaae2bad17ef3f35e09e6818434d8eaefa038f3d1f7359d0969a",
+        "repo": "crueter/cpp-jwt",
+        "sha": "9eaea6328f",
+        "hash": "e237d92c59ebbf0dc8ac0bae3bc80340e1e9cf430e1c1c9638443001118e16de2b3e9036ac4b98105427667b0386d97831415170b68c432438dcad9ef8052de7",
         "find_args": "CONFIG",
         "options": [
-            "CPP_JWT_BUILD_EXAMPLES OFF",
-            "CPP_JWT_BUILD_TESTS OFF",
             "CPP_JWT_USE_VENDORED_NLOHMANN_JSON OFF"
-        ],
-        "patches": [
-            "0001-no-install.patch",
-            "0002-missing-decl.patch"
         ]
     },
     "vulkan-utility-headers": {
@@ -107,12 +98,6 @@
         "repo": "martinus/unordered_dense",
         "sha": "73f3cbb237",
         "hash": "c08c03063938339d61392b687562909c1a92615b6ef39ec8df19ea472aa6b6478e70d7d5e33d4a27b5d23f7806daf57fe1bacb8124c8a945c918c7663a9e8532",
-        "find_args": "CONFIG",
-        "options": [
-            "UNORDERED_DENSE_INSTALL OFF"
-        ],
-        "patches": [
-            "0001-cmake.patch"
-        ]
+        "find_args": "CONFIG"
     }
 }

externals/sse2neon/sse2neon.h

@@ -183,7 +183,7 @@
     }
 
 /* Compiler barrier */
-#if defined(_MSC_VER)
+#if defined(_MSC_VER) && !defined(__clang__)
 #define SSE2NEON_BARRIER() _ReadWriteBarrier()
 #else
 #define SSE2NEON_BARRIER()                     \
@@ -859,7 +859,7 @@ FORCE_INLINE uint64x2_t _sse2neon_vmull_p64(uint64x1_t _a, uint64x1_t _b)
 {
     poly64_t a = vget_lane_p64(vreinterpret_p64_u64(_a), 0);
     poly64_t b = vget_lane_p64(vreinterpret_p64_u64(_b), 0);
-#if defined(_MSC_VER)
+#if defined(_MSC_VER) && !defined(__clang__)
     __n64 a1 = {a}, b1 = {b};
     return vreinterpretq_u64_p128(vmull_p64(a1, b1));
 #else
@@ -1770,7 +1770,7 @@ FORCE_INLINE void _mm_free(void *addr)
 FORCE_INLINE uint64_t _sse2neon_get_fpcr(void)
 {
     uint64_t value;
-#if defined(_MSC_VER)
+#if defined(_MSC_VER) && !defined(__clang__)
     value = _ReadStatusReg(ARM64_FPCR);
 #else
     __asm__ __volatile__("mrs %0, FPCR" : "=r"(value)); /* read */
@@ -1780,7 +1780,7 @@ FORCE_INLINE uint64_t _sse2neon_get_fpcr(void)
 
 FORCE_INLINE void _sse2neon_set_fpcr(uint64_t value)
 {
-#if defined(_MSC_VER)
+#if defined(_MSC_VER) && !defined(__clang__)
     _WriteStatusReg(ARM64_FPCR, value);
 #else
     __asm__ __volatile__("msr FPCR, %0" ::"r"(value));  /* write */
@@ -2249,7 +2249,7 @@ FORCE_INLINE __m128 _mm_or_ps(__m128 a, __m128 b)
 FORCE_INLINE void _mm_prefetch(char const *p, int i)
 {
     (void) i;
-#if defined(_MSC_VER)
+#if defined(_MSC_VER) && !defined(__clang__)
     switch (i) {
     case _MM_HINT_NTA:
         __prefetch2(p, 1);
@@ -4820,7 +4820,7 @@ FORCE_INLINE __m128i _mm_packus_epi16(const __m128i a, const __m128i b)
 // https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html#text=_mm_pause
 FORCE_INLINE void _mm_pause(void)
 {
-#if defined(_MSC_VER)
+#if defined(_MSC_VER) && !defined(__clang__)
     __isb(_ARM64_BARRIER_SY);
 #else
     __asm__ __volatile__("isb\n");
@@ -5716,7 +5716,7 @@ FORCE_INLINE __m128d _mm_undefined_pd(void)
 #pragma GCC diagnostic ignored "-Wuninitialized"
 #endif
     __m128d a;
-#if defined(_MSC_VER)
+#if defined(_MSC_VER) && !defined(__clang__)
     a = _mm_setzero_pd();
 #endif
     return a;
@@ -8130,7 +8130,7 @@ FORCE_INLINE int _sse2neon_sido_negative(int res, int lb, int imm8, int bound)
 
 FORCE_INLINE int _sse2neon_clz(unsigned int x)
 {
-#ifdef _MSC_VER
+#if defined(_MSC_VER) && !defined(__clang__)
     unsigned long cnt = 0;
     if (_BitScanReverse(&cnt, x))
         return 31 - cnt;
@@ -8142,7 +8142,7 @@ FORCE_INLINE int _sse2neon_clz(unsigned int x)
 
 FORCE_INLINE int _sse2neon_ctz(unsigned int x)
 {
-#ifdef _MSC_VER
+#if defined(_MSC_VER) && !defined(__clang__)
     unsigned long cnt = 0;
     if (_BitScanForward(&cnt, x))
         return cnt;
@@ -9058,7 +9058,7 @@ FORCE_INLINE __m128i _mm_aeskeygenassist_si128(__m128i a, const int rcon)
     // AESE does ShiftRows and SubBytes on A
     uint8x16_t u8 = vaeseq_u8(vreinterpretq_u8_m128i(a), vdupq_n_u8(0));
 
-#ifndef _MSC_VER
+#if !defined(_MSC_VER) || defined(__clang__)
     uint8x16_t dest = {
         // Undo ShiftRows step from AESE and extract X1 and X3
         u8[0x4], u8[0x1], u8[0xE], u8[0xB],  // SubBytes(X1)
@@ -9245,7 +9245,7 @@ FORCE_INLINE uint64_t _rdtsc(void)
      * bits wide and it is attributed with the flag 'cap_user_time_short'
      * is true.
      */
-#if defined(_MSC_VER)
+#if defined(_MSC_VER) && !defined(__clang__)
     val = _ReadStatusReg(ARM64_SYSREG(3, 3, 14, 0, 2));
 #else
     __asm__ __volatile__("mrs %0, cntvct_el0" : "=r"(val));