I have an out-of-process COM server written in C++, which is called by some C# client code. A method on one of the server\'s interfaces returns a large BSTR to the client, a
I guess you need to destroy request
with ::SysFreeString()
. This memory is allocated on server side.
Also, OLE2A
may allocate memory due to conversion (take a look). You do not free it also.
anelson has covered this pretty well, but I wanted to add a couple of points;
CoTaskMemAlloc is not the only COM-friendly allocator -- BSTRs are recognized by the default marshaller, and will be freed/re-allocated using SysAllocString & friends.
Avoiding USES_CONVERSION (due to stack overflow risks -- see anelson's answer), your full code should be something like this [1]
(note that A2BSTR is safe to use, as it calls SysAllocString after conversion, and doesn't use dynamic stack allocation. Also, use array-delete (delete[]) as BuildResponse likely allocates an array of chars)
[1]
HRESULT MyClass::ProcessRequest(BSTR request, BSTR* pResponse)
{
char* pszResponse = BuildResponse(CW2A(request));
*pResponse = A2BSTR(pszResponse);
delete[] pszResponse;
return S_OK;
}
I don't see an obvious problem with your code. Suggest you modify the ProcessRequest method to rule out COM interop as the source of the leak:
HRESULT MyClass::ProcessRequest(BSTR request, BSTR* pResponse)
{
*psResponse = ::SysAllocStringLen(L"[suitably long string here]");
return S_OK;
}
I suspect that won't leak, in which case you've narrowed the leak to your code.
I'd also note the OLE2A allocates memory on the stack, so not only should you not delete pszRequest, but you shouldn't use OLE2A at all, due to the possibility of stack overflow. See this article for safer alternatives.
I'd also suggest you replace A2BSTR with ::SysAllocString(CA2W(pszResponse))