Registrations are now open. Join us today!
There is still a lot of work to do on the wiki yet! More information about editing can be found here.
Already have an account?

Microsoft KB Archive/100635

From BetaArchive Wiki
Knowledge Base

How to dynamically load dynamic-link libraries (DLLs) in Windows NT

Article ID: 100635

Article Last Modified on 12/9/2005


  • Microsoft Visual C++ 1.0 Professional Edition
  • Microsoft Visual C++ 2.0 Professional Edition
  • Microsoft Visual C++ 4.0 Standard Edition
  • Microsoft Visual C++ 5.0 Enterprise Edition
  • Microsoft Visual C++ 6.0 Enterprise Edition
  • Microsoft Visual C++ 5.0 Professional Edition
  • Microsoft Visual C++ 6.0 Professional Edition
  • Microsoft Visual C++ 6.0 Standard Edition
  • Microsoft Visual C++ .NET 2002 Standard Edition
  • Microsoft Visual C++ .NET 2003 Standard Edition
  • Microsoft Visual C++ 2005 Express Edition

This article was previously published under Q100635


When you use the LoadLibrary() function in 16-bit Windows or in OS/2, the operating system loads the specified Dynamic-Link Library (DLL) only once. Therefore, the DLL has the same address in every process. However, dynamic loading of DLLs works differently in Windows NT.

The operating system loads a DLL separately for each process because each application has its own address space in Windows NT; the address space is shared in 16-bit Windows and in OS/2. Because the operating system must map pages into the address space for each process, the DLL may be loaded at different addresses in different processes. The memory manager optimizes loading DLLs such that if two processes share the same pages from the same DLL image, they share the same physical memory.

Each DLL has a preferred base address, specified at link time. If the address space range from the preferred base address to the base address plus the image size is not available, then the operating system loads the DLL somewhere else in memory and applies fixups to its addresses. There is no method to specify the load address at load time.

To summarize, the system performs the following steps at load time:

  1. Examines the image and determines its preferred base address and required size.
  2. Finds the address space required and maps the image, copy-on-write, from the file.
  3. Applies internal fixups if the image is not at its preferred base address.
  4. Fixes up all dynamic link imports by placing the correct address for each imported function into the appropriate entry of the Import Address Table. This table stores 32-bit addresses contiguously; to store up to 1024 imported functions requires it to dirty only one page of memory.


The pages that contain code are shared, using a copy-on-write protection scheme. Copy-on-write means that a page is read-only; however, if a process writes to the page and access violation does not occur. Instead, the memory manager makes a private copy of the page for the application's use and allows the write to proceed. For example, if two processes start from the same .EXE file, each process initially has all pages mapped from the .EXE file copy-on-write. As the two processes proceed to modify pages, each one receives a private copy of the modified pages. The memory manager is free to optimize unmodified pages and to map the same physical memory into the address space of both processes. Modified pages are swapped to and from the page file instead of the .EXE file.

Two types of fixups are available. The first is used for the address of an imported function. According to the Portable Executable specification, this type of fixup is stored in the Import Address Table (IAT), an array of 32-bit function pointers, one for each imported function. The IAT has its own page or pages because it is always modified. A call to an imported function is actually an indirect call through the appropriate entry in the IAT. When an image is loaded at its preferred base address, imported function fixups are the only fixups required.

Note that an optimization is available whereby each import library exports a 32-bit number that corresponds with each function in addition to any name or ordinal number. This serves as a "hint" to speed the fixups performed at load time. If the hints in the application and in the loaded DLL do not match, the loader performs a binary search based on the function name.

The other fixup type is required for references to code or data in the image when the image is loaded somewhere other than at its preferred base address. When the memory manager removes a page from memory, it checks to see of the page has been modified. If not, the page retains its copy-on-write mapping and it can be discarded from memory. Otherwise, it must be written to the page file so the modified page can be recovered from the page file rather than from the executable image file.

Even if an application calls LoadLibrary() for a DLL more than once, the DLL entry point, DllMain(), is called only once and only one DLL_PROCESS_ATTACH entry is created. Similarly, if the application calls FreeLibrary() more than once, DLL_PROCESS_DETACH occurs only for the call in which the DLL reference count returns to zero.

Global instance data for the DLL is stored on a per-process basis (only one data set per process). If it is necessary to store global instance data for each LoadLibrary() call performed in one process, consider using thread local storage (TLS) as an alternative. If you use multiple threads of execution, TLS provides unique data storage for each ThreadID value. This process requires very little overhead for the DLL; it must only create a global TLS index at process initialization. At thread initialization, use GlobalAlloc(), HeapAlloc(), LocalAlloc(), functions in the C run-time library, or another method to allocate a block of memory and call the TlsSetValue() function to store a pointer to the memory using the global TLS index value. Win32 internally stores each thread's pointer indexed by TLS index and ThreadID to provide thread-specific storage.

Keywords: kbhowto kbinfo kbdll kblangc KB100635