Building MAME with Visual Studio 2022 - build errors

[ScottTunstall]

Good evening.

I am having trouble building MAME on Windows 11 using VS 2022 Community edition, version 17.13.6.

I wish to build the entire MAME, not just a subtarget.

I executed the following steps in the order given:

1. Installed MingGW64 from the MINGW64 website (not the MAME dev tools)
2. Used pacman in the MSYS2 MINGW64 shell to install the following: mingw-w64-x86_64-gcc, mingw-w64-x86_64-make, mingw-w64-x86_64-python
3. run mingw32-make vs2022 in the MAME source code root.

This generated a .SLN solution file and projects which load with zero problems in Visual Studio 2022.

When I build in the IDE, most projects build fine but there are some build errors: mame build errors.txt

Has anyone successfully built MAME with VS 2022 recently (as in the past couple of weeks), and if so, what did you change? What SDKs did you use?

I know I can comment out the erroneous code or exclude the failing projects, but I don't fancy doing that every time I pull the latest source. I'd like success with, if not every build, most builds.

Any help?

Thanks in advance.
Scott

[pmackinlay]

Most of those errors require source changes to work around with current MSVC due to compiler bugs/limitations; one or two of them might be avoided by using NOWERROR=1 on your make command line. I think the NTDDI_VERSION error is resolved by setting MODERN_WIN_API as mentioned here.

If you're keen on Visual Studio (and not specifically MSVC), I suggest you consider building with clangcl instead, using these changes. There are still two errors building 3rdparty code (bgfx and flac), however they are both valid and resolved upstream, so they should eventually be merged into MAME master.

[ScottTunstall]

EDIT: Ignore this post, the ioport.h errors seem to have gone away. Strange. Very strange.

Commenting out the broken bgfx code allowed MAME to build correctly. Thanks for your help.

IGNORE THE FOLLOWING

I ran the following to create a VS 2022 solution using clang as compiler:

mingw32-make -j36 vs2022_clang MODERN_WIN_API=1 NOWERROR=1

When I open and build the generated solution, MAME now builds with clangcl as you advised, however I am getting far more build errors with clang than I did with VS 2022/MSBuild: mame build errors.txt

ioport.h seems to be the main culprit this time.

My Visual Studio clang install is 19.1.1.

[pmackinlay]

For what it's worth, the following changes from upstream 3rdparty should be the correct fixes for a clangcl build.

diff --git a/3rdparty/bgfx/src/renderer_d3d12.cpp b/3rdparty/bgfx/src/renderer_d3d12.cpp
index 36d23cb5a91..b22cadec2ea 100644
--- a/3rdparty/bgfx/src/renderer_d3d12.cpp
+++ b/3rdparty/bgfx/src/renderer_d3d12.cpp
@@ -474,13 +474,13 @@ namespace bgfx { namespace d3d12
 
 	static inline D3D12_HEAP_PROPERTIES ID3D12DeviceGetCustomHeapProperties(ID3D12Device *device, uint32_t nodeMask, D3D12_HEAP_TYPE heapType)
 	{
-#if BX_COMPILER_MSVC
+#if BX_COMPILER_MSVC || (BX_COMPILER_CLANG && defined(_MSC_VER))
 		return device->GetCustomHeapProperties(nodeMask, heapType);
 #else
 		D3D12_HEAP_PROPERTIES ret;
 		device->GetCustomHeapProperties(&ret, nodeMask, heapType);
 		return ret;
-#endif // BX_COMPILER_MSVC
+#endif // BX_COMPILER_MSVC || (BX_COMPILER_CLANG && defined(_MSC_VER))
 	}
 
 	static void initHeapProperties(ID3D12Device* _device, D3D12_HEAP_PROPERTIES& _properties)
@@ -525,11 +525,11 @@ namespace bgfx { namespace d3d12
 			void* ptr;
 			DX_CHECK(resource->Map(0, NULL, &ptr) );
 			D3D12_RESOURCE_ALLOCATION_INFO rai;
-#if BX_COMPILER_MSVC
+#if BX_COMPILER_MSVC || (BX_COMPILER_CLANG && defined(_MSC_VER))
 			rai = _device->GetResourceAllocationInfo(1, 1, _resourceDesc);
 #else
 			_device->GetResourceAllocationInfo(&rai, 1, 1, _resourceDesc);
-#endif // BX_COMPILER_MSVC
+#endif // BX_COMPILER_MSVC || (BX_COMPILER_CLANG && defined(_MSC_VER))
 			bx::memSet(ptr, 0, size_t(rai.SizeInBytes) );
 			resource->Unmap(0, NULL);
 		}
@@ -626,35 +626,35 @@ namespace bgfx { namespace d3d12
 
 	inline D3D12_CPU_DESCRIPTOR_HANDLE getCPUHandleHeapStart(ID3D12DescriptorHeap* _heap)
 	{
-#if BX_COMPILER_MSVC
+#if BX_COMPILER_MSVC || (BX_COMPILER_CLANG && defined(_MSC_VER))
 		return _heap->GetCPUDescriptorHandleForHeapStart();
 #else
 		D3D12_CPU_DESCRIPTOR_HANDLE handle;
 		_heap->GetCPUDescriptorHandleForHeapStart(&handle);
 		return handle;
-#endif // BX_COMPILER_MSVC
+#endif // BX_COMPILER_MSVC || (BX_COMPILER_CLANG && defined(_MSC_VER))
 	}
 
 	inline D3D12_GPU_DESCRIPTOR_HANDLE getGPUHandleHeapStart(ID3D12DescriptorHeap* _heap)
 	{
-#if BX_COMPILER_MSVC
+#if BX_COMPILER_MSVC || (BX_COMPILER_CLANG && defined(_MSC_VER))
 		return _heap->GetGPUDescriptorHandleForHeapStart();
 #else
 		D3D12_GPU_DESCRIPTOR_HANDLE handle;
 		_heap->GetGPUDescriptorHandleForHeapStart(&handle);
 		return handle;
-#endif // BX_COMPILER_MSVC
+#endif // BX_COMPILER_MSVC || (BX_COMPILER_CLANG && defined(_MSC_VER))
 	}
 
 	inline D3D12_RESOURCE_DESC getResourceDesc(ID3D12Resource* _resource)
 	{
-#if BX_COMPILER_MSVC
+#if BX_COMPILER_MSVC || (BX_COMPILER_CLANG && defined(_MSC_VER))
 		return _resource->GetDesc();
 #else
 		D3D12_RESOURCE_DESC desc;
 		_resource->GetDesc(&desc);
 		return desc;
-#endif // BX_COMPILER_MSVC
+#endif // BX_COMPILER_MSVC || (BX_COMPILER_CLANG && defined(_MSC_VER))
 	}
 
 #if BGFX_CONFIG_DEBUG_ANNOTATION && (BX_PLATFORM_WINDOWS || BX_PLATFORM_WINRT)
diff --git a/3rdparty/flac/src/libFLAC/include/private/bitmath.h b/3rdparty/flac/src/libFLAC/include/private/bitmath.h
index 12e062f88c5..5241462f378 100644
--- a/3rdparty/flac/src/libFLAC/include/private/bitmath.h
+++ b/3rdparty/flac/src/libFLAC/include/private/bitmath.h
@@ -1,6 +1,6 @@
 /* libFLAC - Free Lossless Audio Codec library
  * Copyright (C) 2001-2009  Josh Coalson
- * Copyright (C) 2011-2023  Xiph.Org Foundation
+ * Copyright (C) 2011-2025  Xiph.Org Foundation
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
@@ -82,7 +82,7 @@ static inline uint32_t FLAC__clz_uint32(FLAC__uint32 v)
 	return __builtin_clz(v);
 #elif defined(_MSC_VER)
 	{
-		uint32_t idx;
+		unsigned long idx;
 		_BitScanReverse(&idx, v);
 		return idx ^ 31U;
 	}
@@ -106,7 +106,7 @@ static inline uint32_t FLAC__clz_uint64(FLAC__uint64 v)
 	return __builtin_clzll(v);
 #elif (defined(__INTEL_COMPILER) || defined(_MSC_VER)) && (defined(_M_IA64) || defined(_M_X64))
 	{
-		uint32_t idx;
+		unsigned long idx;
 		_BitScanReverse64(&idx, v);
 		return idx ^ 63U;
 	}
@@ -160,7 +160,7 @@ static inline uint32_t FLAC__bitmath_ilog2(FLAC__uint32 v)
 	return _bit_scan_reverse(v);
 #elif defined(_MSC_VER)
 	{
-		uint32_t idx;
+		unsigned long idx;
 		_BitScanReverse(&idx, v);
 		return idx;
 	}
@@ -177,7 +177,7 @@ static inline uint32_t FLAC__bitmath_ilog2_wide(FLAC__uint64 v)
 /* Sorry, only supported in x64/Itanium.. and both have fast FPU which makes integer-only encoder pointless */
 #elif (defined(__INTEL_COMPILER) || defined(_MSC_VER)) && (defined(_M_IA64) || defined(_M_X64))
 	{
-		uint32_t idx;
+		unsigned long idx;
 		_BitScanReverse64(&idx, v);
 		return idx;
 	}
@@ -206,5 +206,6 @@ static inline uint32_t FLAC__bitmath_ilog2_wide(FLAC__uint64 v)
 }
 
 uint32_t FLAC__bitmath_silog2(FLAC__int64 v);
+uint32_t FLAC__bitmath_extra_mulbits_unsigned(FLAC__uint32 v);
 
 #endif