CodeProject: boost 2: shared_ptr wraps resource handles. Free source code and programming help

Report Article

Discuss

Send to a friend

39 votes for this Article.

Popularity: 7.28 Rating: 4.58 out of 5

Introduction

Windows resources, like GDI handles, are hard to manage correctly without creating leaks. This article shows how smart pointers can simplify this, and almost eliminate a major source of errors. It continues my previous article, Smart Pointers to boost your code, providing a practical example for using shared_ptr. (If you are not familiar with boost::shared_ptr, you should read it now!)

Update: While the first version was more or less like "concept art", I brushed up the sources a bit to make the more useful. I added a reset() function that sets the HandleRef to 0, brushed up the sources a bit and added a special implementation for HDC's. Also, I fixed some snippets here, and added a few things to the article.

Background will examine the problem in detail, and discuss the common solutions.
Smart Pointers to rescue will show how smart pointers can help here, with a simple example.
Developing a Solution will develop the idea into a set of template classes, that make reusing the idea flexible and allow adding new types. Sample code discusses the drop-in header provided.
Using the Solution - jump here if you want to know how to use it.
CDCRef for Device Contexts (new)
Discussion (new)

Background

Many resource handles encountered in Win32 programming do not fit an object-oriented environment very well. Here is a list of problems:

How you acquire the handle determines if you have to call some cleanup function on it. If you use CreateFont, you have to call DeleteObject when you no longer need it. However, if you got the HFONT from a Window, you must not delete the font.
There is no way to tell from a handle whether we should delete it or have to leave it alone.
There are many handle types, with many different Delete/Release functions, which must be matched exactly.
Handles have "pointer" semantics, i.e., copy constructor and assignment create only a new reference to the actual resource. This is desirable for performance reasons, however, this makes it complicated to use RAII in an object oriented environment.

If you have a function returning such a handle, you must at least specify if and when to release the handle. Things get much more complicated if the handle is a member of the class. Take a look at this innocent code snippet:

class CMessage
{
  protected:
    HFONT   m_font;
  public:

    ~CMessage();
  // ...

};

CMessage CreateMessage(CString const & msgText, 
  LPCTSTR fontName, int fontSize);

Code Snippet (1): Demonstrating the problem

Let me ask one question: Should the destructor of CMessage call DeleteObject(m_font)?

Yes? Then the return value of CreateMessage will give you a corrupted font handle to a font that does not exist anymore.
No? Who will delete the font then?

One way or another, the user of the class is concerned with managing your HFONT resource, or what he can do with your class is severely limited.

A multitude of solutions exist, some are simple, some not:

Prohibiting Copy Constructor, and Assignment operator (you can't have the CreateMessage function, then).
Keeping a "bool deleteTheFont" flag together with the font handle. (I worked with std::pair<HANDLE, bool> for a while, but this is still a pain.)
Count the references to the font. (E.g., the copy constructor would have to increment the reference count.)
Use some kind of internal reference counting. (This works for File, Event, Thread, and many other handles, using DuplicateHandle. However, this is another can of worms opening up. No such luck with GDI handles.)
Always copying the object (complicated, expensive, sometimes not possible).
Wrapping one of the solutions into a CFont class.

The nature of handles makes them prime candidates for a reference counting mechanism:

They act as "references to resources", but the resources themselves cannot/should not be copied.
The same resource is reused in many places, and must not be deleted as long as someone is using it.
It should, however, be released as soon as possible, to "play nice" with the system resources.
These resources discussed here don't reference each other, so circular references should virtually never appear.

Further, we would like to keep the possibility of "unmanaged" handles (i.e., it is not deleted automatically). This flag should be set when the smart pointer is created (close to where we acquire the resource, because this is the place where we know how to treat it).

See below: Why Not MFC?, why the MFC solution doesn't do it for me.

Smart Pointers to Rescue

As seen above, a reference counting smart pointer is ideal to handle Handle.

Why boost::shared_ptr is great for this:

shared_ptr does not put any requirements on the type of the resource. (We don't need to inherit HFONT from some CRefCountable class. Phew!)
shared_ptr allows a custom deleter, which can do the resource-specific cleanup. (DeleteObject for fonts, CloseRegKey for handles, etc.)
The custom deleter also allows to not delete the resource automatically.

Note: I mentioned custom deleters only briefly (if at all) in my previous article. See the boost documentation for more.

Let's look at the example for an HFONT:

// first we need a custom deleter for the font:

void delete_HFONT(HFONT * p) 
{ 
  _ASSERTE(p != NULL);   // boost says you don't need  this, 

                         // but lets play safe...

  DeleteObject(*pFont);  // delete the Windows FONT resource

  delete pFont;          // delete the object itself 

                         // (normally done by default deleter)

};

// typedef for our smart pointer:

typedef boost::shared_ptr<HFONT> CFontPtr;

// and a simple "Create" function:

CFontPtr CreateFontPtr(HFONT font, bool deleteOnRelease)
{
  if (deleteOnRelease) {
    // construct a new FontPtr. the custom deleter 

    // is specified as second argument

   return CFontPtr(new HFONT(font), delete_HFONT);  // (A)

  }
  else {
    // construct a new FontPtr with the default deleter:

    return CFontPtr(new HFONT(font));
  }
}

Code Snippet (2): The initial idea

The line (A) is the "magic" one: here we initialize the CFontPtr as usual, but specify that the object should be deleted using delete_HFONT when the last reference is gone.

Now, we can make heavy use of CFontPtr: we can use it as return value from functions. We can have it as class member, and the default copy constructor, assignment operator, and destructor do exactly what they should.

Do you remember the first rule for using smart pointers? Put the resource into a smart pointer as soon as you get it. This rule holds up here as well: because when we get the font handle, we know exactly if it should be deleted or not. The smart pointer will carry around this flag, and automatically "do what we want".

There are some problems even this solution cannot solve:

Someone could delete the font behind our back.
We could create the font pointer with deleteOnRelease = false, and then forget to delete it ourselves.

But such is life in C++, you are always free to shoot yourself in the foot you like most.

Some problems, however, can be solved better:

Checking for a "null" font is tedious: we have to check both the smart pointer, and the actual font:
```
if (m_fontPtr && *m_fontPtr) // do we have a font?
```
We must dereference the smart pointer any time we want to access the actual object.
To have to write three entities (deleter, smart pointer typedef, Create function) for any other handle type we would like to support.

But this is the question of the next part.

Developing a complete solution

This paragraph shows how to turn the concept into a complete extensible solution. It might be helpful in understanding how template libraries evolve into the complex beasts they are.

The following goals were set:

Simple check for NULL resource
Automatic cast to the handle type
Extensible for other types
Minimize the amount of code to write for new types (only a typedef would be great)
Header-only definitions (no separate .cpp / .lib)
This goal was actually set "on the run", because there is not much that can go into a .cpp, and header-only makes the library easier to reuse.

Encapsulation

First, we encapsulate the details of the smart pointer, solving the first two requirements:

class CFontRef
{
  protected:
    typedef boost::shared_ptr<HFONT> tBoostSP;
    tBoostSP m_ptr;

  public:
    explicit CFontRef(HFONT font, bool deleteOnRelease)
    {
      if (deleteOnRelease)
        m_ptr = tBoostSP(new HFONT(font), delete_HFONT);
      else
        m_ptr = tBoostSP(new HFONT(font));
    }

    operator HFONT() const { return m_ptr ? *m_ptr : NULL; }
};

Code Snippet (3): Moving to a specialized class

You might (or should) notice the following things:

The public interface is reduced to the bare minimum. (A "good thing".)
The CreateFontPtr function has become the constructor.
An automatic cast operator allows both the if (font) test, and using the class in a place where a HFONT is expected.
The deleter remains an associated function (not shown here).
The class was renamed from "Ptr" to "Ref", since it syntactically cast more like a reference than a pointer.
The constructor does not specify a default parameter. This was done since no value is the "obvious default". Further, it makes the construction explicit (so the explicit keyword isn't really needed).
I typedef'ed the boost pointer inside the class. It occurs in quite a few places, the typedef makes the code easier to read, but "outside" the class, no one really needs it.

Another note: Now would be a good time to make the simplification used below. However, I'd like to go the "safe & clean way" a bit further.

Second, we can make both the Handle type and the Deleter a template parameter. It's simple for the Handle, but some compilers can't handle a function as template argument. The standard solution is to turn the Deleter function into a functor - that is, a class that overloads operator().

template <typename HDL, typename DELETER>
class CHandleRefT
{
  protected:
    typedef boost::shared_ptr<HDL> tBoostSP;
    tBoostSP m_ptr; 

  public:
    CHandleRefT(HDL h, bool deleteOnRelease)
    {
      if (deleteOnRelease) m_ptr = tBoostSP(new HDL(h), DELETER());
      else m_ptr = tBoostSP(new HDL(h));
    } 
    operator HDL() const { return m_ptr ? *m_ptr : NULL; }
}; 

// the font deleter, turned into a functor:

struct CDeleterHFONT
{
    void operator()(HFONT * p) { DeleteObject(*p); delete p; }
}; 

// the typedef for our CFontRef:

typedef CHandleRefT<HFONT, CDeleterHFONT> CFontRef;

Code snippet (4): turning the class into a template

DeleteObject is used for many types, so we don't want to write our own deleter for each. However, we would like to keep everything strictly typed, so we make it a template again:

template <typename GDIHANDLE>
struct CDeleter_GDIObject
{
  void operator()(GDIHANDLE * p) { DeleteObject(*p); delete p; }
};

// the typedef now has a nested template:

typedef CHandleRef<HFONT, CDeleter_GDIObject<HFONT> > CFontRef;

Code snippet (5): a helper template

Specialized, more lightweight implementation

There are two things still bugging me:

The deleters still need to remember the "delete p" part.
I would like to avoid the heap copy of the resource handle if possible.

The first problem could be solved with another helper template, but that wouldturn the actual type of CFontRef into something like boost::shared_ptr<HFONT, GenericDeleter< DeleteObjectDeleter<HFONT> > >. I see no solution for the second in a generic approach.

However, both solve themselves when using some internal knowledge about Windows: all resource handles in Windows can be represented by a void *, and we can use a cast to get the strictly typed handle. This is so deeply rooted in the Win32 API (in fact, if you #undef the STRICT macro, most handle types are declared as void *), that we won't see a change till .NET takes over completely. Further, boost::shared_ptr can use void as template argument.

So, with the following code, we can go:

// a void deleter replaces the default deleter for "unmanaged" handles

struct CDeleter_Void
{
  void operator()(void *) {}
}

template <typename HDL, typename DELETER>
class CHandleRef
{
  protected:
    typedef boost::shared_ptr<void> tBoostSP;
    tBoostSP m_ptr;

  public:
    explicit CHandleRef(HDL h, bool deleteOnRelease)
    {
      if (deleteOnRelease)
        m_ptr = tBoostSP(h, DELETER());
      else
        m_ptr = tBoostSP(h, CDeleter_Void());
    }

    operator HDL() const { return (HDL) m_ptr.get(); }
};

struct CDeleter_GDIObject
{
  void operator()(void * p) { DeleteObject( (HGDIOBJ) p); }
};

typedef CHandleRef<HFONT, CDeleter_GDIObject> CFontRef;

Code snippet (6): the final code

Before, the smart pointer stored a pointer to our resource handle. Now, we store the resource handle directly in the smart pointer (represented as void *).

Using the solution

(I'll repeat a few things here for the impatient who skipped over all that boring explanation stuff)

CHandleRefT is a template class implementing a counted reference to a Windows resource handle. Usage rules are similar to a reference counted smart pointer. It is used to implement a variety of Windows Resource handles, like:

HIMAGELIST ==> CImageListRef
HMENU ==> CMenuRef
HANDLE ==> CHandleRef
HBITMAP ==> CBitmapRef
HBRUSH ==> CBrushRef
HPEN ==> CPenRef

Template parameters:

HDL: type of the resource handle (e.g., HFONT).
DELETER : a functor releasing the resources of Handle of type HDL (e.g., a functor calling DeleteObject). The handle is passed as void *.
A custom deleter can be passed to the constructor. However, DELETER is not a template parameter of the class (which avoids template propagation)

Automatic vs. manual handles

When constructing a HandleRef from a raw handle, you pass a bDeleteOnRelease flag indicating if the handle should be released automatically when the last reference to it goes out of scope.

For an automatic handle, assign the raw handle to a HandleRef immediately after construction, specifying bDeleteOnRelease=true. Then, pass it around only as HandleRef. This guarantees the handle is deleted when it is no longer used.

For a manual handle, that you will delete yourself manually, or that must not be deleted, specify bDeleteOnRelease=false.

The idea here is that you can pass around, copy, store, or return a HandleRef, it remembers its deletion policy.

Guidelines

Utility functions that merely receive and use a resource handle (without storing it) may use the raw handle as argument. However, when the Handle is stored (e.g., as a class member) or used as a return value from a function, a HandleRef is recommended.

As an example, the CImageListRef class has the following members:

Constructor

CImageListRef(HIMAGELIST il, bool deleteOnRelease)

Initializes a new CImageListRef.

il [HIMAGELIST]: Image list to hold
deleteOnRelease [bool]: if true, the image List il is destroyed when the last reference to it (made through the CImageListRef instance) goes out of scope. The correct destroy function (ImageList_Destroy) is used.

operator HIMAGELIST

implicitly casts the CImageListRef to the contained HIMAGELIST.

reset()

Releases the image list.

To add support for a new type:

The type must be convertible to and from void *.

Write a Delete functor for the type:

struct CDeleter_MyType { void operator()(void * p) { MyTypeRelease(p); }

use a typedef

typedef CHandleRef<CMyType, CDeleter_MyType> CMyTypeRef;

Why not MFC?

MFC tried to solve the problem with its own wrappers. Unfortunately, they chose to wrap the first option from the "possible solutions" above: no copy functionality, but they pass around CFont *. This might have been a good design decision back then, but now, it is just a pain. Consider a function returning a CFont *. Now, there are two objects that need to be released correctly: the Windows HFONT, and the CFont C++ object. Should you:

Delete the CFont * when you are done with it, because it was allocated dynamically?
Not delete the CFont *, but use it only until some other class is destroyed (because the other class "holds" the font)?
Not delete the font, but use it only in the current message handler (because it's a temporary MFC object that is deleted in the next OnIdle call)?
Detach the HFONT from the CFont * before deleting it (because you have to get rid of the MFC object, but the Windows resource is still in use somewhere)?

When turning UI code into general routines and classes, I regularly stumble over all these four scenarios. It is not a pleasant experience, I can tell you - so I end up using raw resource handles instead, to have "only the normal problems".

Further, I like UI code not to be dependent on MFC, if possible. The best code IMO is a library that relies solely on the Win32 API interface, and does not put any requirements on the framework.

CDCRef for Device Contexts

The CDCRef class, referencing a HDC, deserves additional discussion.

Device Contexts are the most complicated resource I've come across: There are two cleanup functions: DeleteDC and ReleaseDC, and the latter needs an additional parameter (the HWND the DC was acquired from).

CDCRef is implemented as separate class, since I didn't want the complexity of HDC to "creep" into the Other HandleRef classes. The main difference is the constructor - instead of a flag, you directly pass the deleter:

CDCRef(hdc, CDCRef::Null()) // a CDCRef that doesn't get released automatically

CDCRef(hdc, CDCRef::Delete()) // a CDCRef to be released by DeleteDC

CDCRef(hdc, CDCRef::Release(hwnd))  // a CDCRef to be released by ReleaseDC

For ReleaseDC, the HWND is passed as parameter to the deleter. Again, we associated all information required for cleanup with the object when it is constructed, which is a design principle of HandleRef.

Further, CDCRef provides static member functions that wrap the Win32 API functions that acquire a HDC. For example, CDCRef::GetDC implements the GetDC Win32 API function, but returns a CDCRef.

Discussion

The solution provided here is not new. The idea of the article is to show that the solution becomes very simple when using an appropriate library (about 20 lines of actual code), and how such a solution would be developed from the initial idea to use shared_ptr.

Now, How good is the solution?

You need to know how Win32 Resources are managed
You need to know how smart pointers work
You have a slight overhead over plain Win32 code: a separately allocated object (8 Bytes) for each managed handle, and the cleanup call is through a function pointer.

The second is not an issue: smart pointers are such a fundamental technique that you shouldn't be caught without. The last might be an issue when handling lots of resources. However, such an application likely benefits most from automated resource management, and you can keep performance under control with using a raw handle (or a CHandleRefT<> const &) where appropriate.

The first actually reflects a design choice: CHandleRef does not isolate you from the underlying API, but makes it transparent and easier to use. (MFC, in contrast, isolates you very well in the default case, and fails misderably in all others). Additional advantage: CHandleRef is not only library independent, it also integrates easily with other libraries.

Is a "better" solution possible?

Definitely. You still can do things wrong: release an object while it is still selected into a device context, specify the wrong cleanup policy, or delete the handle while it is still in use. However, a completely safe solution would require to wrap the entire GDI API: all functions creating or using an GDI object.

CDCRef actually show the difference between an open wrapper (allowing access to the underlying API handles) and a close wrapper (denying the same access).

To make CDCRef construction "foolproof" the CDCRef constructor would have to be protected, so they can be constructed only through dedicated functions that initialize correctly. But this requires wrappers for all functions acquiring a DC. Further, to make CDCRef completely foolproof, one also would have to wrap all functions accepting a DC, and remove the operator HDC() from the class. The open wrapper leaves you some responsibility, but you have no problems if I forgot to provide a wrapper (or the original API is extended).

Conclusion and Thank You

Thank you all for the encouraging feedback! It is great to see the time to write the article was well spent.

For a more formal conclusion: We have seen that

Resource Handles, as many other objects, can have a variety of destruction policies, which are not "visible" from the handle itself
The deletion policy should be "attached" to the handle when the handle is acquired
reference counted smart pointers are perfect for wrapping resource handles
boost::shared_ptr provides the features to make an implementation simple
by using some platform-specific knowledge, we can make the solution more efficient, and less complex in terms of code, while keeping the original interface

License

This article has no explicit license attached to it but may contain usage terms in the article text or the download files themselves. If in doubt please contact the author via the discussion board below.

A list of licenses authors might use can be found here

About the Author

peterchen

Supporter

Peter is tired of being called "Mr. Chen", even so certain individuals insist on it. No, he's not chinese.

Peter has seen lots of boxes you youngsters wouldn't even accept as calculators. He is proud of having visited the insides of a 16 Bit Machine.

In his spare time he ponders new ways of turning groceries into biohazards, or tries to coax South American officials to add some stamps to his passport.

He recently started blogging about hisFirst Real C# Project[^] (a.k.a. "get it done"). Musing how this changes development for him or in general, tidbits he discovered, and else.

Beyond these trivialities Peter works for Klippel[^], a small german company that wants to make mankind happier by selling them novel loudspeaker measurement equipment.

Where are you from?[^]

Please, if you are using one of my articles for anything, just leave me a comment. Seeing that this stuff is actually useful to someone is what keeps me posting and updating them.
Should you happen to not like it, tell me, too

Occupation:	Web Developer
Location:	Germany

Other popular STL articles:

An In-Depth Study of the STL Deque Container
This article presents an in-depth analysis of std::deque and offers guidance as to when to prefer using it as opposed to std::vector, by taking into consideration memory allocation and container performance.
An STL-like bidirectional map
A template container implementing a bidirectional map that blends well with STL.
Smart Pointers to boost your code
A beginner's introduction to the smart pointers provided by the boost library.
Upgrading an STL-based application to use Unicode.
Problems that developers will face when upgrading an STL-based application to use Unicode and how to solve them.
Can search_n be more efficient?
This article is discussing the efficiency of the most popular search_n implementations. Furthermore, it is introducing a new search_n specialization for random access iterators, which outruns by far the most commonly used implementations.

Article Top

You must Sign In to use this message board.

Msgs 1 to 25 of 42 (Total in Forum: 42) (Refresh)

FirstPrevNext

WTL classes to manage GDI Handles etc...

operandx

22:55 29 Oct '07

Nice article.

Several years passed since this article was posted - just some heads-up to the readers who are unfamiliar with WTL - an originally MS extension to ATL.

WTL has many wrapper-like classes to standard window handles, such as HFONT, HDC, HBRUSH, HWND and many many more.

For example:
typedef CMenuT CMenuHandle;
typedef CMenuT CMenu;
typedef CPenT CPenHandle;
typedef CPenT CPen;

atlgdi.h:
// CPenT
// CBrushT
// CLogFont
// CFontT
// CBitmapT
// CPaletteT
// CRgnT
// CDCT

atluser.h:
// CMenuItemInfo
// CMenuT
// CAcceleratorT
// CIconT
// CCursorT

----

Cheers

/Zeev

Re: WTL classes to manage GDI Handles etc...

peterchen

11:05 30 Oct '07

operandx wrote:
Several years passed since this article was posted

Scary, somehow.

I've said that before some posts below - my problem is that the "delete it or not" flag is a template parameter, meaning I can't assign a CMenuT<true> to a CMenuT<false> - so I can't decide about policy only when I create the handle. They work very well as "scope guards" - i.e. making sure some handle is destroyed when the wrapper leaves scope. I aimed at a bit more than that.

Anyway, thanks for your feedback Big Grin

As mos feedback shows, this is only a partial and/or incomplete solution - this is also my experience using them.

We are a big screwed up dysfunctional psychotic happy family - some more screwed up, others more happy, but everybody's psychotic joint venture definition of CP
My first real C# project | Linkify!|FoldWithUs! | sighist

Handles already use Reference counts

Zac Howland

9:26 17 May '06

The kernel has a reference count for handles (all handles) that you ask it to create. The problem is that many people don't pass handles correctly. If you are going to do anything that you think should require an increment to the reference count, you should use DuplicateHandle (and a subsequent call to the appropriate release method). Adding another reference count to it will just decrease the performance of your applications.

As an aside, MFC knows about the kernel reference counting and hence its wrappers handle them properly (in most cases).

If you decide to become a software engineer, you are signing up to have a 1/2" piece of silicon tell you exactly how stupid you really are for 8 hours a day, 5 days a week

Zac

3.40/5 (2 votes)

Re: Handles already use Reference counts

peterchen

10:47 17 May '06

Important point. Still, I'm not convinced Smile

Zac Howland wrote:
many people don't pass handles correctly

The problem I try to solve is not passing, but returning and storing.

I see the following advantagesof this solution:

(1) Not all "Handle-Like things" have an DuplicateHandle.
One example: Registry keys (not technically kernel handles, you can open another handle to the same key, but you lose the access rights of the original HKEY)

(2) The solution is generic for all pointer-sized handles. Less code, less test cases, less things to know. A custom implementation for fonts would be very expensive (HFONT -> LOGFONT -> HFONT, or is there a better one?)

(3) The kernel reference count isn't available (directly or indirectly) in user mode.

E.g. DuplicateHandle has two drawbacks:
It returns a new handle value (I tried XP SP2, file handle, DUPLICATE_SAME_ACCESS). So we can safely assume using DuplicateHandle takes longer and eats additional resources per copy.
Further, the copy operation could fail, which must be handled.

using CHandleRef in turn has the following drawbacks:
- additional tracking object allocated on the heap for each handle (but not for each copy). That makes CHandleRef unsuitable if you rapidly create and destroy thousands of handles. It's the one major drawback I admit freely Wink

- twice the size. No real problem
- pass by value is bit more expensive than passing a handle by value.
- additional dependency

In the end I think it's 90% abot taste and style. Only with the remaining 10% I'd insist that class/library interfaces using handle wrappers like this are more elegant and usable, and once "reference counted" us understood, I can safely forget about so many things, which is my personal holy grail of software design.

At the sides, MFC in most cases opts to make the wrappers uncopyable. A proper decision design wise, but damn inconvenient.

Some of us walk the memory lane, others plummet into a rabbit hole

Tree in C# || Fold With Us! || sighist

Re: Handles already use Reference counts

Zac Howland

4:43 18 May '06

peterchen wrote:
The problem I try to solve is not passing, but returning and storing.

Returning is a form of passing. And generally speaking, you shouldn't be storing handles. There is a slim chance (depending on the type of handle) that the OS can decide to move it. If you kept the original handle, you will end up with a very random (meaning extremely difficult to debug) crash.

peterchen wrote:
(1) Not all "Handle-Like things" have an DuplicateHandle.
One example: Registry keys (not technically kernel handles, you can open another handle to the same key, but you lose the access rights of the original HKEY)

Anything that is equivalent to HANDLE can be used in DuplicateHandle. In fact, some of the APIs for other types of handles actually make calls to DuplicateHandle. Yes, it will be less efficient than passing a reference to your handle value, but it will also be safer in some cases (actually, almost all cases).

peterchen wrote:
(2) The solution is generic for all pointer-sized handles. Less code, less test cases, less things to know. A custom implementation for fonts would be very expensive (HFONT -> LOGFONT -> HFONT, or is there a better one?)

Your intensions aren't the problem. Unfortunately, when it comes to kernel objects, you can't treat them like any other pointer. If you don't follow the proper steps for creation/deletion for a given type of kernel object, you will easily cause resource leaks that will bog down your entire system (until you reboot). Your choices for things like this are to read the documentation very carefully and wrap those steps into your own custom classes, or to use libraries that already do it for you. MFC does the font stuff fairly well (though, it does take some getting use to).

peterchen wrote:
(3) The kernel reference count isn't available (directly or indirectly) in user mode.

That is how it should be. You shouldn't care what the reference count is. The only time you really need to look at it is when you are debugging, and you can get to it then (through the debugger).

peterchen wrote:
E.g. DuplicateHandle has two drawbacks:
It returns a new handle value (I tried XP SP2, file handle, DUPLICATE_SAME_ACCESS). So we can safely assume using DuplicateHandle takes longer and eats additional resources per copy.
Further, the copy operation could fail, which must be handled.

DuplicateHandle (and its equivalents) basically create a new pointer to the resources. You aren't going to use this method for things like passing the handle into a function to operate on the handle (say, to use it to draw something to the screen). An example of when you want to use it is when you want to have 2 threads drawing to the screen (each thread would have its own copy of the handle).

If the copy operation fails, it means something in the system went very wrong (e.g. you ran out of resources).

I don't question your motives behind this article (I actually like the direction you are thinking), it is just wrapping something that is already done by the OS (the reference counting) creates duplicate functionality and increases complexity (what happens if your reference count gets out of synch with the OS's?). Another way you could wrap handles is to create a method that automatically dupilcates it for you and returns a new wrapper (you probably wouldn't want to do that in the copy-constructor nor the copy-assignment methods though ... that would prevent you from passing the class by value, which you may want to do sometimes).

If you decide to become a software engineer, you are signing up to have a 1/2" piece of silicon tell you exactly how stupid you really are for 8 hours a day, 5 days a week

Zac

Re: Handles already use Reference counts

peterchen

23:23 19 May '06

Zac Howland wrote:
what happens if your reference count gets out of synch with the OS's

The shared_ptr references don't need to be in sync with the kernel. They key is the usgae guide: create, and assign, e.g. I'm using:


CHandleRef OpenLogFile() { ... }

The OpenLogFile method opens the handle and immediately assigns it to a CHandleRef. The raw handle is now "owned" by CHandleRef, meaning I must make sure it is not closed manually, or stored in raw form. As long as I guarantee this, my reference count will be on top of a single reference to the actual file object (I'm not sure how familiar you are with COM, but the same mechanism is used for reference counting across an apartment boundary.)

I agree that this requires some bit of "training", and maybe I find it easy only because I know smart pointers inside out from years of COM. But this pattern (both the idea and the code) is highly reusable, e.g. just two days ago for enumerating firewire devices. Two lines of code for an exception safe HDEVINFO (and another shared_ptr for the dynamic sized structures used in hardware enumeration)

I'm not trying to brush away your comments (even if it may sound like), I often notice that I "learn" something from such a discussion only a few weeks (even months) later.

Zac Howland wrote:
you shouldn't be storing handles.

Sorry, I can't follow you here. The "behavior" of handkles is well defined. Of course I might accidentally close a handle behind the stores back etc. but code can't protect us from stupidity anyway.

What I'm trying to do is making write good code simpler. Most Windows bugs are in resource management. By associating the cleanup function with the handle there's one thing less to remember.

I've always been unhappy with the MFC solution, particulary because it requires you to read lot of MFC documentaiton carefully, and it doesn't interoperate with other libraries very well.

Zac Howland wrote:
If the copy operation fails, it means something in the system went very wrong (e.g. you ran out of resources).

Agreed, if the DuplicateHandle faisl, you can probably shut down the app anyway. But error handling is an application aspect that cannot be decided by the library (CHandleRefT doesn't include any error handling. but it makes it much much easier)

Anyway, enjoy the weekend Smile

Some of us walk the memory lane, others plummet into a rabbit hole

Tree in C# || Fold With Us! || sighist

understanding how template libraries evolve into the complex beasts they are

Roland Pibinger

3:37 26 Oct '04

I don't see the advantages of your approach:

- Implementation
Why do you use boost::shared_ptr? Seriously, all you need is a reference counter that indicates if a resource must be destroyed in the destructor.

-Design
Resource handling is 'partly managed' which is error-prone: Ownership is controlled by a flag that has to be set by the user. On the other hand, managed resources are converted automatically to unmnaged handles by operator HDL().

But maybe I have overlooked something?

Best wishes,
Roland Pibinger

Re: understanding how template libraries evolve into the complex beasts they are

peterchen

0:14 27 Oct '04

There are definitely questionable design choices, the following is the motivaiton for taking them, I don't think they are "better" than some other options.

(1)
because boost provides exactly what I need, and reduces the implementation effort. You need:
- A class / struct for the shared resource itself
- The "smart pointer" class (whatever it is)
- A "plugin point" for the custom resource deleter

This takes me to three entities, one of which is visible to the user and needs utmost care in handling (I've written smart pointer like stuff before, it's nothing I wouldn't like to have thorughly tested in production code, and a C++ guru should look for pitfalls)

boost::shared_ptr provides just that: holding the reference count, and the handle, a fully tested reference counting pointer, and a custom deleter that is NOT a template parameter (which is important to avoid template infection of the using application)

(2)
It was not a design goal to completely isolate the client from GDI resource handling. I've pondered it a long time, and concluded it is not possible wihtout wrapping huge chunks of the windows API.

Rather, I wanted:
- a solution that can be integrated with any other library that can work somehow with raw handles.
- SIMPLIFY common resource handling problems
- Move the "How to delete" decision to the point where the resource is acquired
- low amount of code

Using the library can simplify resource handling, if you understand reference counted entities (not too simple, but very common), and understand Win32 GDI resource handling (which is IMO a necessity, unless you wrap the entire GDI API).

implicit cast to HANDLE type:
I was uneasy about the cast to raw handle, it's a tradeoff between comfort and bulletproofness. I opted for comfort, that's all. The alternative would be:
- a cast to "unspecified_bool_type" (as boost does) for the "Handle zero" check
- a HANDLET get() const member
I would prefer the latter if HANDLET could be a complex type. However, we are basically using void * / unspecified_type * here. In this case, I believe, the implicit cast is safe.

Also, since I want to integrate with "anything that takes raw handles", I cannot protect the client from accidentally deleting the raw handle with the wrong function or at the wrong time.

The core implementation is less than 2 dozen lines (not counting curly braces)

I think I stated this before - but the idea of this article was to privde a practical example for my previous one, and see how/if it could solve a long-nagging problem (reference counting for handles).

we are here to help each other get through this thing, whatever it is Vonnegut jr.
boost your code || Fold With Us! || sighist | doxygen

SelectObject

Stewart Tootill

22:23 14 Oct '04

How does this object play with SelectObject, a function which changes the delete requirements of an object.

For Example

CSmartPenHandle pen( ::CreatePen( PS_SOLID, 1, RGB(0,0,0) ) );
// now the pen should be deleted by SelectObject

// I think that ref counting smart pointers are good, but not for DCs,
// there too complex a resource for that. You don't want to leave
// them lying around for sure and ref counting might let that happen
HDC hdc = ::GetDC( hMyWnd );

::SelectObject( hdc, pen.GetWin32Handle() );
// pen now mustn't be deleted. Especially if hdc is a class or window dc.
// if it is neither then typically you'd select the return value back in,
// but if you don't then ReleaseDC deletes the pen, and your destructor
// will fail.

Note that this is just a thought excersise. The example is great as it stands, but i'm simply trying to give an example of why GDI objects in particular don't really lend themselves to smart pointers.

Re: SelectObject

peterchen

5:01 15 Oct '04

Stewart Tootill wrote:
Note that this is just a thought excersise.

Actually, it is not. It is the weak point of my article Wink

I was considering this when writing, but I finally left it out for several reasons.

You could use a "DC Selector" helper (see below), but ultimately, I don't think any amount of code can circumvent all possibilities to shoot yourself in the foot. In the end, you will have to follow some rules.

The basic idea would be to (a) make sure the object is deselected (e.g. using SaveDC/ReleaseDC, possibly with a scope guard to be safe), and (b) hold a strong reference to the object until it is deselected

I wouldn't even use HandleRef for temporary objects created in the WM_PAINT itself, they original idea was for GDI resources you "keep around" instead of reinitalizing them again and again. But the following could help a bit (but it's just a rough outline):



template 
struct CDCSelector
{
  CHandleRef       dc;
  HANDLE                oldHandle;
  CHandleRef    handle;

  CDCSelector(CHandleRef _dc, CHandleRef h) : dc(_dc), handle(h)
  {
    oldHandle = 0;
    if (handle)
        oldHandle = (HANDLE) ::SelectObject(dc, h);
  }

  void Restore()
  {
    if (handle)
    {
      ::SelectObject(dc, oldHandle);
      handle = CHandleRef(); 
    }
  }

  ~CDCSelector() { Restore();    }
};

(It would require special CDC support, since DC's come with two destructors - ReleaseDC and DeleteDC, and you would still need to do the nesting correctly. So it's not perfect either. The only "bulletproof" solutions I see would require a complete wrapping of the DC handle, which I would dislike)

we are here to help each other get through this thing, whatever it is Vonnegut jr.

sighist Fold With Us! || Agile Programming | doxygen

Re: SelectObject

Stewart Tootill

6:01 15 Oct '04

I think that the two destruction methods for DCs (there are more than two if you start to include DirectDraw/3D/Show and GDI+ but that may be taking things two far) are the best reason for not attempting to wrap DCs in classes.

Were you to attempt to do this however, MFCs approach would probably be a good guide. There are three DC classes (plus some other ones for printing that don't matter in this case). CPaintDC gets its DC by calling BeginPaint and releases it with EndPaint and is constructed with a window handle. CClientDC/CWindowDC use the GetDC/GetDCEx functions and ReleaseDC to release them, and straight CDC (which is the base class, but obviously the derived classes override the destruction) which uses DeleteDC to delete the DC which covers most (but not all - see previous comment about DirectX and GDI+) other cases.

Its really a shame that one of the trickiest things to wrap in C++ is also one of the hardest to avoid. I remember starting SDK programming for real on windows having previously used other systems (most notably BeOS) and being thoroughly confused by these DC things. But you can't write even the most trivial GDI application without encountering them.

Re: SelectObject

peterchen

13:31 15 Oct '04

Please allow me to reply in a single post Wink

I don't think virtual is to high of a cost - after all, the bost::shared_ptr solution comes in even higher (and would have made it a no-go for the machines MFC 1.x was intended for)

However, the "battlefield", as I see it, is not virtual vs. templates, but Copy Construction and Assignment.

Since the MFC approach does not provide these, I am forced to pass around CDC * or CDC &, so the wrapper trades in management of one resource (GDI) for another (C++ object lifetime) which is a beast of it's own, and I lose many advantages of scoping.

Reference counting does not plug all holes (SelectObject, as you pointed out, is the gaping barn door, and the template suggested is more a fix than a solution), but it plugs many more holes than it opens, and reduces the "mental overhead" of using GDI resources.

Both the virtual destructor, and the "custom deleter" achieve one thing: associating the destruction policy with the resource at time of acquisition. However, the virtual-only solution prohibits common functionality (passing as argument or return value) and moves the complexity only up one level in the hierarchy.

I would like to emphasize that I don't see the solution presented here as the only one. It is neither perfect, nor the most simple one. But I consider it well balanced, and surprisingly simple once you have "accepted smart pointers in your life".

we are here to help each other get through this thing, whatever it is Vonnegut jr.

sighist Fold With Us! || Agile Programming | doxygen

Re: SelectObject

peterchen

3:42 17 Nov '04

I just updated the article, including a CDCRef (in a separate, but single class). It was more a exercise to see if DC's could be fit into the HandleRef model, but I'm happy with the results. let me know what you think we are here to help each other get through this thing, whatever it is Vonnegut jr. boost your code \|\| Fold With Us! \|\| sighist \| doxygen
Sign In·View Thread·PermaLink

Template Specialisation

Stewart Tootill

22:17 14 Oct '04

I haven't had time to try this, bit I thought you might be able to replace the bool delete on release flag with a template argument and use specialisation to resolve it. Sure it would make the code bigger (but who cares, memory is cheap and hopefully all this stuff gets inlined anyway) but typically you would never pass a non-const value into that field. The type of delete should almost always be known at compile time. And writing a function that can take a dynamic value once you've done it with specialisation is easy. That way the extra cost (small until you start allocation alot of pens with it, or trying to walk thousands of registry keys) is only there for those that need it.

Re: Template Specialisation

peterchen

4:47 15 Oct '04

boost 2: shared_ptr wraps resource handles

Introduction

Contents

Background

Smart Pointers to Rescue

Developing a complete solution

Encapsulation

Specialized, more lightweight implementation

Using the solution

Template parameters:

Automatic vs. manual handles

Guidelines

Constructor

To add support for a new type:

Why not MFC?

CDCRef for Device Contexts

Discussion

Conclusion and Thank You

License

About the Author

Other popular STL articles: