Insights-compatible memory tracking (#214)

This change introduces support for tracing of memory allocation activity. The code is ported from UE5, and Unreal Insights can be used to analyze the output. This is currently only fully supported on Windows, but will be extended to Mac/Linux in the near future.

To activate full memory tracking, pass `--trace=memory` on the commandline alongside `--tracehost=<ip>` or `-tracefile=<path>`. For more control over how much detail is traced you can instead pass some combination of `callstack`, `memtag`, `memalloc` instead. In practice, `--trace=memory` is an alias for `--trace=callstack,memtag,memalloc`). For convenience we also support `--trace=memory_light` which omits call stacks.

This change also introduces multiple memory allocators, which may be selected via command-line option `--malloc=<allocator>`:

* `mimalloc` - mimalloc (default, same as before)
* `rpmalloc` - rpmalloc is another high performance allocator for multithreaded applications which may be a better option than mimalloc (to be evaluated). Due to toolchain limitations this is currently only supported on Windows.
* `stomp` - an allocator intended to be used during development/debugging to help track down memory issues such as use-after-free or out-of-bounds access. Currently only supported on Windows.
* `ansi` - fallback to default system allocator

1 files changed, 189 insertions, 0 deletions

diff --git a/src/zencore/memory/mallocrpmalloc.cpp b/src/zencore/memory/mallocrpmalloc.cpp
new file mode 100644
index 000000000..ffced27c9
--- /dev/null
+++ b/src/zencore/memory/mallocrpmalloc.cpp
@@ -0,0 +1,189 @@
+// Copyright Epic Games, Inc. All Rights Reserved.
+
+#include <zencore/intmath.h>
+#include <zencore/memory/align.h>
+#include <zencore/memory/mallocrpmalloc.h>
+
+#if ZEN_RPMALLOC_ENABLED
+
+#	include "rpmalloc.h"
+
+/** Value we fill a memory block with after it is free, in UE_BUILD_DEBUG **/
+#	define DEBUG_FILL_FREED (0xdd)
+
+/** Value we fill a new memory block with, in UE_BUILD_DEBUG **/
+#	define DEBUG_FILL_NEW (0xcd)
+
+#	define ZEN_ENABLE_DEBUG_FILL 1
+
+namespace zen {
+
+FMallocRpmalloc::FMallocRpmalloc()
+{
+	rpmalloc_initialize(nullptr);
+}
+
+FMallocRpmalloc::~FMallocRpmalloc()
+{
+	rpmalloc_finalize();
+}
+
+void*
+FMallocRpmalloc::TryMalloc(size_t Size, uint32_t Alignment)
+{
+	void* NewPtr = nullptr;
+
+	if (Alignment != DEFAULT_ALIGNMENT)
+	{
+		Alignment = Max(uint32_t(Size >= 16 ? 16 : 8), Alignment);
+		NewPtr	  = rpaligned_alloc(Alignment, Size);
+	}
+	else
+	{
+		NewPtr = rpaligned_alloc(uint32_t(Size >= 16 ? 16 : 8), Size);
+	}
+
+#	if ZEN_BUILD_DEBUG && ZEN_ENABLE_DEBUG_FILL
+	if (Size && NewPtr != nullptr)
+	{
+		memset(NewPtr, DEBUG_FILL_NEW, rpmalloc_usable_size(NewPtr));
+	}
+#	endif
+
+	return NewPtr;
+}
+
+void*
+FMallocRpmalloc::Malloc(size_t Size, uint32_t Alignment)
+{
+	void* Result = TryMalloc(Size, Alignment);
+
+	if (Result == nullptr && Size)
+	{
+		OutOfMemory(Size, Alignment);
+	}
+
+	return Result;
+}
+
+void*
+FMallocRpmalloc::Realloc(void* Ptr, size_t NewSize, uint32_t Alignment)
+{
+	void* Result = TryRealloc(Ptr, NewSize, Alignment);
+
+	if (Result == nullptr && NewSize)
+	{
+		OutOfMemory(NewSize, Alignment);
+	}
+
+	return Result;
+}
+
+void*
+FMallocRpmalloc::TryRealloc(void* Ptr, size_t NewSize, uint32_t Alignment)
+{
+#	if ZEN_BUILD_DEBUG && ZEN_ENABLE_DEBUG_FILL
+	size_t OldSize = 0;
+	if (Ptr)
+	{
+		OldSize = rpmalloc_usable_size(Ptr);
+		if (NewSize < OldSize)
+		{
+			memset((uint8_t*)Ptr + NewSize, DEBUG_FILL_FREED, OldSize - NewSize);
+		}
+	}
+#	endif
+	void* NewPtr = nullptr;
+
+	if (NewSize == 0)
+	{
+		rpfree(Ptr);
+
+		return nullptr;
+	}
+
+#	if ZEN_PLATFORM_MAC
+	// macOS expects all allocations to be aligned to 16 bytes, so on Mac we always have to use mi_realloc_aligned
+	Alignment = AlignArbitrary(Max((uint32_t)16, Alignment), (uint32_t)16);
+	NewPtr	  = rpaligned_realloc(Ptr, Alignment, NewSize, /* OldSize */ 0, /* flags */ 0);
+#	else
+	if (Alignment != DEFAULT_ALIGNMENT)
+	{
+		Alignment = Max(NewSize >= 16 ? (uint32_t)16 : (uint32_t)8, Alignment);
+		NewPtr	  = rpaligned_realloc(Ptr, Alignment, NewSize, /* OldSize */ 0, /* flags */ 0);
+	}
+	else
+	{
+		NewPtr = rprealloc(Ptr, NewSize);
+	}
+#	endif
+
+#	if ZEN_BUILD_DEBUG && ZEN_ENABLE_DEBUG_FILL
+	if (NewPtr && NewSize > OldSize)
+	{
+		memset((uint8_t*)NewPtr + OldSize, DEBUG_FILL_NEW, rpmalloc_usable_size(NewPtr) - OldSize);
+	}
+#	endif
+
+	return NewPtr;
+}
+
+void
+FMallocRpmalloc::Free(void* Ptr)
+{
+	if (!Ptr)
+	{
+		return;
+	}
+
+#	if ZEN_BUILD_DEBUG && ZEN_ENABLE_DEBUG_FILL
+	memset(Ptr, DEBUG_FILL_FREED, rpmalloc_usable_size(Ptr));
+#	endif
+
+	rpfree(Ptr);
+}
+
+void*
+FMallocRpmalloc::MallocZeroed(size_t Size, uint32_t Alignment)
+{
+	void* Result = TryMallocZeroed(Size, Alignment);
+
+	if (Result == nullptr && Size)
+	{
+		OutOfMemory(Size, Alignment);
+	}
+
+	return Result;
+}
+void*
+FMallocRpmalloc::TryMallocZeroed(size_t Size, uint32_t Alignment)
+{
+	void* NewPtr = nullptr;
+
+	if (Alignment != DEFAULT_ALIGNMENT)
+	{
+		Alignment = Max(uint32_t(Size >= 16 ? 16 : 8), Alignment);
+		NewPtr	  = rpaligned_zalloc(Alignment, Size);
+	}
+	else
+	{
+		NewPtr = rpaligned_zalloc(uint32_t(Size >= 16 ? 16 : 8), Size);
+	}
+
+	return NewPtr;
+}
+
+bool
+FMallocRpmalloc::GetAllocationSize(void* Original, size_t& SizeOut)
+{
+	// this is not the same as the allocation size - is that ok?
+	SizeOut = rpmalloc_usable_size(Original);
+	return true;
+}
+void
+FMallocRpmalloc::Trim(bool bTrimThreadCaches)
+{
+	ZEN_UNUSED(bTrimThreadCaches);
+}
+}  // namespace zen
+#endif