you're reading...
Interop Marshaling, Managed Structures, SAFEARRAYs

Marshaling a SAFEARRAY of Managed Structures by P/Invoke Part 3.

1. Introduction.

1.1 In part 1 of this series of articles, I demonstrated how to transfer managed arrays to unmanaged code as SAFEARRAYs. The transfer was single-directional “into” the unmanaged function and the SAFEARRAY that was passed to the unmanaged function is treated as “read-only”.

1.2 Then in part 2, I showed how to return a SAFEARRAY from unmanaged code to managed code as an “out” parameter. The transfer this time was also single-directional but the direction is “outwards” towards the managed code caller. The SAFEARRAY that was returned is used to generate a managed array.

1.3 Here in part 3, I shall explain how to pass a managed array “to and from” unmanaged code as both an “in” and “out” (i.e. by reference) parameter.

1.4 A parameter passed this way is marshaled “into” the unmanaged function when it is called and when the function returns, the same parameter is marshaled “out of” the unmanaged function. The parameter is also subject to modification by the target function.

1.5 As usual, throughout this article, we shall be working only with single-dimensional managed arrays and SAFEARRAYs.

2. TestStructure, CSConsoleApp.tlb Type Library and UnmanagedDll.DLL

2.1 We shall be using the same TestStructure struct that we have developed in part 1.

2.2 We shall also continue to use the CSConsoleApp.tlb type library that was produced from the CSConsoleApp console application solution that was presented in part 1.

2.3 We shall augment UnmanagedDll.dll with some helper functions as well as a new exported API to be called in the CSConsoleApp console application.

2.4 The source codes of the CSConsoleApp console application will also be updated with new test codes.

3. Unmanaged API that References a SAFEARRAY of TestStructure.

3.1 The new exported function that we will expose to C# takes as parameter a double pointer to a SAFEARRAY of TestStructure structures.

3.2 This double pointer will be used to point to a SAFEARRAY of TestStructure UDTs that derives from a managed array. It will also be used to return a SAFEARRAY of TestStructure UDTs to the caller.

3.3 In our case, the caller will be the interop marshaler which will first transform a managed array into a SAFEARRAY in order to pass to the exported function. The interop marshaler will then transform the returned SAFEARRAY back into a managed array.

3.4 Note the protocol on memory ownership : because the interop marshaler is the eventual receiver of the SAFEARRAY which is returned, the interop marshaler owns the SAFEARRAY and is at liberty to destroy it when it no longer needs it.

3.5 The following is a full code listing of this function :

// ModifyArrayOfTestStructure() will modify the contents
// of the input SAFEARRAY. It will also add new elements
// to the SAFEARRAY.
extern "C" __declspec(dllexport) void __stdcall ModifyArrayOfTestStructure
  /*[in, out]*/ SAFEARRAY** ppSafeArrayToModify
	IRecordInfoPtr spIRecordInfo = NULL;

	// We obtain a pointer to the IRecordInfo interface associated
	// with the SAFEARRAY. It is used to clear TestStructure structs.
	SafeArrayGetRecordInfo(*ppSafeArrayToModify, &spIRecordInfo);

	// If we are unable to obtain such an interface,
	if (spIRecordInfo == NULL)

	// First determine how many elements there are inside
	// "ppSafeArrayToModify".
	long LBound = 0;
	long UBound = 0;

	SafeArrayGetLBound(*ppSafeArrayToModify, 1, &LBound);
	SafeArrayGetUBound(*ppSafeArrayToModify, 1, &UBound);

	ULONG ulSafeArraySize = UBound - LBound + 1;

	if (ulSafeArraySize > 0)
		// The SAFEARRAY must have at least one element
		// for us to modify.
		TestStructure test_structure;
		long rgIndices[1];

		// The UDT receiving structure must be cleared
		// before calling on SafeArrayGetElement() to
		// obtain a copy of an element.
		memset(&test_structure, 0, sizeof(TestStructure));

		// We will modify the very first element.
		rgIndices[0] = 0;

			(void FAR*)&test_structure

		// Increment the values of the
		// "m_integer" and "m_double"
		// fields.
		test_structure.m_integer += 100;
		test_structure.m_double += 100;
		// Clear the original BSTR and
		// set it to a different value.
		test_structure.m_string = ::SysAllocString(L"Modified String");

		// Insert "test_structure" into the SAFEARRAY
		// at the same index position (i.e. 0);

		// After "inserting" "test_structure" into
		// the SAFEARRAY, we must remember to clear it.
		spIRecordInfo -> RecordClear((void*)(&test_structure));

	// Now modify the size of the SAFEARRAY.
	// Increase it 3 more elements.
	SAFEARRAYBOUND rgsabound[1];

	rgsabound[0].lLbound = 0;
	rgsabound[0].cElements = ulSafeArraySize + 3;

	HRESULT hrRetTemp = SafeArrayRedim

	if (SUCCEEDED(hrRetTemp))
		for (ULONG ulIndex = ulSafeArraySize; ulIndex < (ulSafeArraySize + 3); ulIndex++)
			long rgIndices[1];
			TestStructure test_structure;

			rgIndices[0] = ulIndex;

			memset(&test_structure, 0, sizeof(TestStructure));

			// Add simple values to the fields of the new TestStructure
			// structs that are added to the SAFEARRAY.
			test_structure.m_integer = (int)ulIndex;
			test_structure.m_double = (double)ulIndex;
			test_structure.m_string = ::SysAllocString(L"New String");			


			spIRecordInfo -> RecordClear((void*)(&test_structure));

The following is a synopsis of this function :

  • The function first tries to obtain a pointer to the IRecordInfo interface associated with the UDT contained in the SAFEARRAY. This is done by using SafeArrayGetRecordInfo().
  • It then calculates the total number of elements in the SAFEARRAY (using SafeArrayGetLBound() and SafeArrayGetUBound()).
  • Assuming that the SAFEARRAY contains at least one element, this function will modify the field values of the very first element.
  • Note that SafeArrayGetElement() is used to obtain a copy of the target element and then SafeArrayPutElement() is used to replace the target element.
  • When SafeArrayPutElement() is called, it will first clear the fields of the existing element residing at the target index.
  • Then the onus is on the ModifyArrayOfTestStructure() function to clear the TestStructure struct which was used to insert the UDT at the target index. For this, IRecordInfo::RecordClear() is used.
  • After that, the ModifyArrayOfTestStructure() function uses SafeArrayRedim() to resize the SAFEARRAY to a larger size.
  • A loop is then performed to insert new additional UDTs into the SAFEARRAY.
  • Note that as each UDT is inserted into the SAFEARRAY, a copy of the UDT is made and stored in the SAFEARRAY.
  • Hence the original TestStructure UDT must be cleared after each call to SafeArrayPutElement().

4. Example C# Call to ModifyArrayOfTestStructure().

4.1 The following shows how the ModifyArrayOfTestStructure() API should be declared in a C# program :

[DllImport("UnmanagedDll.dll", CallingConvention = CallingConvention.StdCall)]
private static extern void ModifyArrayOfTestStructure
  [In][Out] [MarshalAs(UnmanagedType.SafeArray, SafeArraySubType = VarEnum.VT_RECORD)]
  ref TestStructure[] SafeArrayToModify

Now note the use of the various attributes :

  • The presence of the InAttribute and the OutAttribute indicate that the managed array parameter (i.e. “SafeArrayToModify”) is to be marshaled into and out of the ModifyArrayOfTestStructure() function. The “ref” keyword further indicates this to the C# compiler.
  • These attributes also indicate to the interop marshaler that whatever form the counterpart parameter (i.e. the parameter of the unmanaged function) takes when it is passed to the target function, it may be eventually modified before its return.
  • However, it is nevertheless owned by the caller which is at liberty to destroy it when it is returned from the unmanaged function.
  • The way the MarshalAsAttribute is specified as well as the presence of the OutAttribute indicate to the interop marshaler that the counterpart parameter will take the form of a double pointer to a SAFEARRAY.
  • The “SafeArraySubType” field for the MarshalAsAttribute, being equal to “VarEnum.VT_RECORD”, indicates to the interop marshaler that the SAFEARRAY will contain UDTs.
  • And since the “SafeArrayToModify” parameter is typed as an array of TestStructure, the UDT must be the equivalent of the TestStructure.

4.2 The following is a sample C# function that makes a call to GetArrayOfTestStructure() :

static void DoTest_ModifyArrayOfTestStructure()
    // Define and instantiate a managed array of 3
    // TestStructure structs.
    TestStructure[] SafeArrayOfTestStructure = new TestStructure[3];

    // Assign simple values to the elements of the array.
    for (int i = 0; i < SafeArrayOfTestStructure.Length; i++)
        SafeArrayOfTestStructure[i].m_integer = i;
        SafeArrayOfTestStructure[i].m_double = (double)i;
        SafeArrayOfTestStructure[i].m_string = string.Format("Hello World [{0}]", i);

    // Call on ModifyArrayOfTestStructure() to modify
    // this array.
    ModifyArrayOfTestStructure(ref SafeArrayOfTestStructure);

    // Display the contents of the array.
    for (int i = 0; i < SafeArrayOfTestStructure.Length; i++)
        Console.WriteLine("SafeArrayOfTestStructure[{0}].m_integer : [{1}]", i, SafeArrayOfTestStructure[i].m_integer);
        Console.WriteLine("SafeArrayOfTestStructure[{0}].m_double : [{1}]", i, SafeArrayOfTestStructure[i].m_double);
        Console.WriteLine("SafeArrayOfTestStructure[{0}].m_string : [{1:S}]", i, SafeArrayOfTestStructure[i].m_string);

The following is a synopsis :

  • A managed array of 3 TestStructure structs is defined and instantiated.
  • Simple values are assigned to the field values of each element.
  • The ModifyArrayOfTestStructure() API is then invoked and the managed array is passed as a parameter by reference.
  • Thereafter, we display all the contents of the managed array.

4.3 The following is what happened under the covers :

  • When ModifyArrayOfTestStructure() is called, the interop marshaler will internally prepare a SAFEARRAY and fill it with the (unmanaged) UDT equivalent of each of the (managed) TestStructure structs contained in the “SafeArrayOfTestStructure” array.
  • Then, sensing that the “SafeArrayToModify” parameter of the ModifyArrayOfTestStructure() function has been designated as an “in” and “out” parameter, the interop marshaler passes a double pointer to the SAFEARRAY as parameter when it calls the ModifyArrayOfTestStructure() function.
  • The ModifyArrayOfTestStructure() function is free to modify the SAFEARRAY in whatever way it deems fit.
  • When ModifyArrayOfTestStructure() returns, the interop marshaler will use the returned SAFEARRAY to internally modify the managed array.
  • However, very likely, the interop marshaler will simply delete the entire original managed array and re-create a new one from the latest contents of the SAFEARRAY.
  • When the managed array of TestStructure is finally modified/re-created, the returned SAFEARRAY which was shared by both the interop marshaler and the ModifyArrayOfTestStructure() API will be destroyed. Each TestStructure contained inside the SAFEARRAY will be destroyed by calling on the RecordDestroy() method using the IRecordInfo pointer which is already contained within the SAFEARRAY.

4.4 At runtime, the C# function DoTest_GetArrayOfTestStructure() will produce the following expected output :

SafeArrayOfTestStructure[0].m_integer : [100]
SafeArrayOfTestStructure[0].m_double : [100]
SafeArrayOfTestStructure[0].m_string : [Modified String]
SafeArrayOfTestStructure[1].m_integer : [1]
SafeArrayOfTestStructure[1].m_double : [1]
SafeArrayOfTestStructure[1].m_string : [Hello World [1]]
SafeArrayOfTestStructure[2].m_integer : [2]
SafeArrayOfTestStructure[2].m_double : [2]
SafeArrayOfTestStructure[2].m_string : [Hello World [2]]
SafeArrayOfTestStructure[3].m_integer : [3]
SafeArrayOfTestStructure[3].m_double : [3]
SafeArrayOfTestStructure[3].m_string : [New String]
SafeArrayOfTestStructure[4].m_integer : [4]
SafeArrayOfTestStructure[4].m_double : [4]
SafeArrayOfTestStructure[4].m_string : [New String]
SafeArrayOfTestStructure[5].m_integer : [5]
SafeArrayOfTestStructure[5].m_double : [5]
SafeArrayOfTestStructure[5].m_string : [New String]

The following is a general summary for the results shown above :

  • At the start of the DoTest_ModifyArrayOfTestStructure() function, a managed array of 3 TestStructure structs is created.
  • Then the ModifyArrayOfTestStructure() function modified the very first element to contain values 100, 100 and “Modified String” for its “m_integer”, “m_double” and “m_string” field values respectively.
  • The ModifyArrayOfTestStructure() function also added 3 new TestStructure UDTs with numeric values that correspond with their SAFEARRAY index positions and a standard string value of “New String”.
  • Hence eventually, the “SafeArrayOfTestStructure” array contained 6 elements with values as shown.

5. In Conclusion.

5.1 Here in part 3, I have demonstrated sending and returning a SAFEARRAY to and from an unmanaged function in managed code.

5.2 Various SAFEARRAY APIs are used to demonstrate ways to modify both the contents of SAFEARRAYs as well as the size of SAFEARRAYs.

5.3 I have also shown that the modified contents of a SAFEARRAY is used by the interop marshaler to modify/re-create the original managed array.

5.4 Again I emphasized that with memory ownership, the interop marshaler is at liberty to destroy the returned SAFEARRAY.

5.5 This part marks the end of my basic treatment of marshaling a managed array to and from unmanaged functions by way of SAFEARRAYs.

5.6 Further parts of this series of articles are not required reading for those who are not interested in more complex containment of managed arrays as UDTs in SAFEARRAYs.



About Lim Bio Liong

I've been in software development for nearly 20 years specializing in C , COM and C#. It's truly an exicting time we live in, with so much resources at our disposal to gain and share knowledge. I hope my blog will serve a small part in this global knowledge sharing network. For many years now I've been deeply involved with C development work. However since circa 2010, my current work has required me to use more and more on C# with a particular focus on COM interop. I've also written several articles for CodeProject. However, in recent years I've concentrated my time more on helping others in the MSDN forums. Please feel free to leave a comment whenever you have any constructive criticism over any of my blog posts.


One thought on “Marshaling a SAFEARRAY of Managed Structures by P/Invoke Part 3.

  1. Heya i am for the first time here. I found this board and I
    in finding It truly useful & it helped me out a lot.
    I’m hoping to give one thing back and aid others like you helped me.

    Posted by seo company | April 24, 2013, 3:53 am

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: