[Support] Use malloc instead of non-throwing new (#92157)

When allocating a memory buffer, we use a non-throwing new so that we
can explicitly handle memory buffers that are too large to fit into
memory. However, when exceptions are disabled, LLVM installs a custom
new handler

(https://github.com/llvm/llvm-project/blob/90109d444839683b09f0aafdc50b749cb4b3203b/llvm/lib/Support/InitLLVM.cpp#L61)
that explicitly crashes when we run out of memory

(https://github.com/llvm/llvm-project/blob/de14b749fee41d4ded711e771e43043ae3100cb3/llvm/lib/Support/ErrorHandling.cpp#L188)
and that means this particular out-of-memory situation cannot be
gracefully handled.

This was discovered while working on #embed
(https://github.com/llvm/llvm-project/pull/68620) on Windows and
resulted in a crash rather than the preprocessor issuing a diagnostic as
expected.

This patch switches away from the non-throwing new to a call to malloc
(and free), which will return a null pointer without calling a custom
new handler. It is the only instance in Clang or LLVM that I could find
which used a non-throwing new, so I did not think we would need anything
more involved than this change.

Testing this would be highly platform dependent and so it does not come
with test coverage. And because it doesn't change behavior that users
are likely to be able to observe, it does not come with a release note.

1 files changed, 9 insertions, 2 deletions

diff --git a/llvm/lib/Support/MemoryBuffer.cpp b/llvm/lib/Support/MemoryBuffer.cpp
index 4cc4fe0..50308bd 100644
--- a/llvm/lib/Support/MemoryBuffer.cpp
+++ b/llvm/lib/Support/MemoryBuffer.cpp
@@ -98,7 +98,7 @@ public:
 
   /// Disable sized deallocation for MemoryBufferMem, because it has
   /// tail-allocated data.
-  void operator delete(void *p) { ::operator delete(p); }
+  void operator delete(void *p) { std::free(p); }
 
   StringRef getBufferIdentifier() const override {
     // The name is stored after the class itself.
@@ -315,7 +315,14 @@ WritableMemoryBuffer::getNewUninitMemBuffer(size_t Size,
   size_t RealLen = StringLen + Size + 1 + BufAlign.value();
   if (RealLen <= Size) // Check for rollover.
     return nullptr;
-  char *Mem = static_cast<char*>(operator new(RealLen, std::nothrow));
+  // We use a call to malloc() rather than a call to a non-throwing operator
+  // new() because LLVM unconditionally installs an out of memory new handler
+  // when exceptions are disabled. This new handler intentionally crashes to
+  // aid with debugging, but that makes non-throwing new calls unhelpful.
+  // See MemoryBufferMem::operator delete() for the paired call to free(), and
+  // llvm::install_out_of_memory_new_handler() for the installation of the
+  // custom new handler.
+  char *Mem = static_cast<char *>(std::malloc(RealLen));
   if (!Mem)
     return nullptr;