Finalization

New and Notable 37 - Some Notable CLR Posts

Sam Gentile's Blog — Sat, 21 Feb 2004 06:57:00 GMT

re: Finalization

Paul Wilson — Sat, 21 Feb 2004 07:20:00 GMT

I'm curious about #6 -- dedicated thread in TS/Citrix environment. I understand the basic fact, but I'm wondering if you could better define the cost or impact?

Thanks, Paul Wilson

Chris Brumme discusses Finalization

Junfeng Zhang's .Net Framework Notes — Sat, 21 Feb 2004 12:02:00 GMT

re: Finalization

Chris Brumme — Sat, 21 Feb 2004 19:11:00 GMT

Paul,

We really need some folks from the OS kernel team to start writing their own blogs

Earlier this morning I posted a blog on hosting that talks about thread overhead from the perspective of load on the OS scheduler. This is certainly one of the reasons for limiting the number of threads per process in Terminal Server scenarios. In addition, I've heard various rumors about the amount of non-paged memory and other scarce resources required per thread. But I'm not an OS dev and I would just be repeating hearsay.

Let's just say that this concern about extra threads in managed processes is a recurring theme at our Performance reviews with senior executives like Jim Allchin. And these guys certainly do understand the underlying OS resource issues.

re: Finalization

Sean — Sat, 21 Feb 2004 20:26:00 GMT

Chris, the MSDN docs tell us to clean up managed and unmanaged resources in a Dispose or Close method. But they also say, "Note that even when you provide explicit control by way of Dispose, you should provide implicit cleanup using the Finalize method." [1]

Do you still agree with this, or is it better to skip on the Finalize code? The reason I ask is not completely based on what you've written in your blog, but if you don't close DataReaders (initialized with CommandBehavior.CloseConnection) enough times you can overload the connection pool. I'm assuming that's because the DataReader gets GCed but doesn't do some other cleanup on the connection.

---Sean

[1] http://msdn.microsoft.com/library/default.asp?url=/library/en-us/cpgenref/html/cpconFinalizeDispose.asp

re: Finalization

Junfeng Zhang — Sun, 22 Feb 2004 00:19:00 GMT

Sean, the reason is your Displose method is not guaranteed to be run. In that case, you want your finalizer to be your last defense.

Of course, you can (and should) call GC.SuppressFinalize(this) in your Displose method.

re: Finalization

Gheorghe Marius — Mon, 23 Feb 2004 11:27:00 GMT

Dude...first of all thank you for your blog.

"high priority Finalizer thread". Why not time critical priority ? After all the finalization thread is "short lived". Seting the time critical priority means that will finish his job faster...and because is short lived it can't really mess with the OS scheduler.

re: Finalization

Chris Brumme — Mon, 23 Feb 2004 18:18:00 GMT

Gheorghe,

I don't think we can use 'time critical' priority for finalization, with good conscience. Keep in mind that this is arbitrary user code. Here is the relevant guidance from MSDN:

You should almost never use REALTIME_PRIORITY_CLASS, because this interrupts system threads that manage mouse input, keyboard input, and background disk flushing. This class can be appropriate for applications that "talk" directly to hardware or that perform brief tasks that should have limited interruptions

Knowing .NET

TrackBack — Mon, 23 Feb 2004 20:38:00 GMT

Knowing .NET

Hard Reading on Garbage Collection

Simply Patrick — Mon, 23 Feb 2004 20:39:00 GMT

??? CLR ? garbage collection ???????: Finalization, Chris Brumme Dispose Idiom, Tristan Cartony Resource management, Brian Harry...

.NET in Terminal Server (or Citrix) Scenarios

Paul Wilson's .NET Blog — Mon, 23 Feb 2004 23:31:00 GMT

So you think you know why finalization is bad?

Jeff Key — Tue, 24 Feb 2004 07:52:00 GMT

So you think you know why finalization is bad?

Jeff Key — Tue, 24 Feb 2004 07:52:00 GMT

Hosting, and Finalization

stevex.Text — Thu, 04 Mar 2004 17:17:00 GMT

Hosting, and Finalization

stevex.Text — Thu, 04 Mar 2004 17:18:00 GMT

re: Finalization

Artyom Kamshilin — Tue, 16 Mar 2004 09:29:00 GMT

Chris, thanks for your entry, it's truly awesome.

About the finalization, it looks like this question was a hard one when building the CLR, but that didn't make it easier to hack through the lots of information about Dispose() inside MSDN. Coming from Delphi background, I remember how easy and straightforward it was made there when it came to destructing an object - one needs to override a virtual method called Destroy(), and provide any custom destruction code there.

So after some thinking I came up with a class that implements IDisposable, and provides two methods that can be overriden in inheriting classes. Here is the code:

using System;

namespace Arkamax.Framework
{
/// <summary>
/// Provides a convenient way to manage resources disposal.
///
/// To use, inherit a class from SmartDisposer and
/// override DisposeManaged() to dispose managed objects,
/// and DisposeUnmanaged() to dispose unmanaged ones.
///
/// </summary>
public class SmartDisposer: IDisposable
{
private bool disposed;

private Object disposalSync;

public bool Disposed
{
get { lock(disposalSync) { return disposed; } }
}

public SmartDisposer()
{
disposalSync = new Object();
}

protected virtual void DisposeManaged()
{
}

protected virtual void DisposeUnmanaged()
{
}

public void Dispose()
{
Dispose(true);

GC.SuppressFinalize(this);
}

protected virtual void Dispose(bool FromUserMode)
{
// this has been put here in an attempt to
// protect from multiple threads calling Dispose() same time

lock (disposalSync)
{
// Check to see if Dispose has already been called.
if(!disposed)
{
// If FromUserMode equals true, dispose all managed
// and unmanaged resources.

if(FromUserMode)
{
// Dispose managed resources
DisposeManaged();
}

// Release unmanaged resources. If FromUserMode is false,
// only the following code is executed.

DisposeUnmanaged();
}

disposed = true;
}
}

// Use C# destructor syntax for finalization code.
// This destructor will run only if the Dispose method
// does not get called.
// It gives your base class the opportunity to finalize.
// Do not provide destructors in types derived from this class.

~SmartDisposer()
{
Dispose(false);
}

}
}

Thus, by inheriting from this class and overriding DisposeManaged() and DisposeUnmanaged(), it becomes quite easy to handle this twisted issue.

Hope this helps somebody.

Regards,
Artyom

re: Finalization

Chris Brumme — Thu, 18 Mar 2004 19:11:00 GMT

That looks good.

If there are issues with using the resource and simultaneously calling Dispose on the resource from a different thread, then that requires additional synchronization. For most resources, that extra synchronization is unnecessary. For some resources, like file handles, it is quite important.

It's interesting that the person who was responsible for Delphi's destruction model was also heavily involved in the .NET IDisposable interface.

re: Finalization

Artyom Kamshilin — Sat, 20 Mar 2004 09:33:00 GMT

So it looks like we could then declare disposalSync this way:

protected Object disposalSync;

and do a lock(disposalSync) in the inheriting classes whenever we do some sensitive operation (like file handling).

> It's interesting that the person who was responsible
> for Delphi's destruction model was also heavily
> involved in the .NET IDisposable interface.

Wow, that becomes more and more interesting :) When I heard that Anders Hejlsberg (the original author of Turbo Pascal, and the father of Delphi) is supervising .NET development, I felt quite warmer :), but when it came to comparing Delphi and .NET destruction models, I couldn't even think it was the same person - these models are quite different in nature. Could you let us know who that person was, does he have a blog? I'm heavily interested in leaving no white space on .NET basics like disposal, so any bit of information might turn to be invaluable.

re: Finalization

Chris Brumme — Mon, 22 Mar 2004 16:42:00 GMT

The person is Anders.

To ReleaseComObject or Not To ReleaseComObject?

Andrew Whitechapel — Sat, 17 Apr 2004 12:10:00 GMT

or Would You, Could You in the Rain? I thought Id try to round off the discussion with a simple summary Two questions: 1 Under what circumstances would you (want/need to) RCO? 2 Under what circumstances could you (safely) RCO? Some Answers: 1 Under what circumstances would you want/need to RCO? 1.1 When your RCW is proxying a COM object thats holding onto a lot of memory or other resource that you need to release eagerly (for performance or contention reasons). 1.2 When you believe a circular reference has deadlocked and you want to break the cycle. 1.3 When your...

Chris Brumme Updates His Hosting and Finalization

Sam Gentile's Blog — Tue, 27 Apr 2004 01:14:00 GMT

re: Finalization

RichB — Tue, 27 Apr 2004 09:22:00 GMT

Now this has been updated with post-CTP info, the XXXX marks formerly hid "SafeHandles, Critical Finalizers and Constrained Execution Regions" and all the information about them.

Finalization and Hosting in the .NET CLR

Marcus Mac Innes' Blog — Tue, 27 Apr 2004 11:04:00 GMT

The Final(ization) word?

Zupancic Perspective — Wed, 28 Apr 2004 06:49:00 GMT

re: Finalization

Stéphane Lajoie — Wed, 28 Apr 2004 16:38:00 GMT

In a comment to that other blog entry about handle recycling attacks, you mention System.Runtime.InteropServices.HandleRef.
From the documentation, it seems to be exactly the same as this new SafeHandle except that it is a value type and it is sort of tied to p/invoking.

But everything involving handles will somehow be tied to p/invoke and value types are more lightweight so what does SafeHandle add to justify its existence? I'm sure I'm missing something...

re: Finalization

Chris Brumme — Wed, 28 Apr 2004 20:30:00 GMT

SafeHandle can be thought of as a "new & improved" HandleRef. From memory, it has the following benefits:

Reduced graph promotion for better performance, because it naturally encourages you to move the finalizable objects into leaf positions of an object graph.

An encapsulating type that prevents code from extracting the underlying OS resource and causing races with the Finalizer. Also, because it has identity rather than value semantics (e.g. HandleRef), there's no risk that passing the encapsulation around will cause refcounting problems with the underlying resource.

PInvoke of course extracts the IntPtr. But it implicitly prevents the race using a technique similar to GC.KeepAlive.

Prevention of handle-recycling attacks and inadvertent recycling bugs, by coordinating usage of the resource with explicit IDisposable::Dispose of the resource.

Coordination with Constrained Execution Regions. Although it takes high trust to define a SafeHandle, it brings the guarantee that the cleanup will occur. This is true even if someone attempts to Abort the thread or perform an orderly shutdown of the process or some other technique that could affect finalization.

Some of these overlap with what HandleRef can do. But SafeHandle is clearly more useful and powerful. I would advise moving over to it.

Various Gubbins.

hulver from Hulver's site — Wed, 05 May 2004 12:34:00 GMT

.net is more complicated than you think. Subtle problems that will bite you in the arse. I thought .net was supposed to fix things like this, not make them worse?<br>
enterfornone :( That .net thing really bites my bag. See, when I code something in Delphi, I know...

The Dirt on Finalization

Delphi Compiler Core — Wed, 05 May 2004 23:03:00 GMT

How .NET finalization relates to Delphi code

re: Finalization

小说 — Sun, 27 Jun 2004 06:10:00 GMT

Thank you for your blog.

re: Finalization

dianying xia zai — Sun, 01 Aug 2004 18:13:00 GMT

http://www.dmoz.net.cn/ wangzhidaquang
http://www.86dmoz.com/ jingpingwangzhi
http://www.kamun.com/ mianfeidianying
http://movie.kamun.com/ dianyingxiazai
http://music.kamun.com/ MP3 free download
http://www.pc530.net/ diannaoaihaozhe
http://www.5icc.com/ duangxingcaixingxiazha
http://www.dianyingxiazai.com/ dianyingxiazai
http://www.yinyuexiazai.com/ yinyuexiazai

SafeHandle

.Net Security Blog — Fri, 13 Aug 2004 20:44:00 GMT

A Few Good GC Links

Chris Lyon's WebLog — Tue, 14 Sep 2004 20:23:00 GMT

Demystifying Dispose

Chris Lyon's WebLog — Tue, 21 Sep 2004 19:47:00 GMT

Demystifying Dispose

Chris Lyon's WebLog — Tue, 21 Sep 2004 19:50:00 GMT

Demystifying Dispose

Chris Lyon's WebLog — Sun, 26 Sep 2004 22:54:00 GMT

A Few Good GC Links

Chris Lyon's WebLog — Tue, 28 Sep 2004 02:25:00 GMT

What the heck is: Finalization

Squawks of the Parrot — Thu, 21 Oct 2004 20:58:00 GMT

Chris Brumme made a blog posting (ages ago--this has been sitting in my pending queue for a while) that reminded me about this, so I thought I'd go into it before I forgot. (I'd recommend reading that link, too--while it deals only with finalization in a .NET environment, and Microsoft's .NET environment specifically (Mono and dotGNU may well have different details) it gives a good overview of some of the more... interesting issues that can be brought up by finalization) Anyway, finalization is the process of letting an object that's now dead have one last shot at cleaning up...

WinDev Coding Questions

Stephen Toub — Wed, 27 Oct 2004 17:12:00 GMT

WinDev Coding Questions

Stephen Toub — Wed, 27 Oct 2004 17:17:00 GMT

WinDev Ask The Experts, .NET trivia and brain-teasers...oh my!

frankred's WebLog — Wed, 27 Oct 2004 19:04:00 GMT

Disposal Anxiety

K. Scott Allen's Blog — Mon, 08 Nov 2004 02:44:00 GMT

Disposal Anxiety

K. Scott Allen's Blog — Mon, 08 Nov 2004 02:45:00 GMT

Destructors vs. GC? Destructors GC!

Sutter's (Online) Mill — Tue, 23 Nov 2004 21:35:00 GMT

Destructors vs. GC? Destructors GC!

Herb — Sun, 12 Dec 2004 20:25:00 GMT

BradA at the Houston .NET UG

.NET Blog - Chris Frazier Style — Thu, 17 Feb 2005 21:38:00 GMT

BradA at the Houston .NET UG

Blue Phoenix — Thu, 17 Feb 2005 21:39:00 GMT

IDisposable Pattern vs Need for Finalize

Chanian, Raj — Wed, 20 Apr 2005 13:31:30 GMT

IDisposable Pattern vs Need for Finalize

re: Best Practices and Member Initialization in C#

Fear and Loathing — Fri, 22 Apr 2005 13:51:41 GMT

SafeHandle...

Eric Gunnerson's C# Compendium — Thu, 04 Aug 2005 20:32:32 GMT

I haven't been paying much attention to Whidbey stuff in the past year or so, but today I came across...

SafeHandle Goodness

Greg Young — Wed, 22 Feb 2006 15:41:07 GMT

SafeHandle Goodness

Greg Young — Wed, 22 Feb 2006 15:57:46 GMT

BradA at the Houston .NET UG

Blue Fenix — Tue, 07 Mar 2006 21:31:48 GMT

BradA at the Houston .NET UG

Blue Fenix — Tue, 07 Mar 2006 21:49:21 GMT

BradA at the Houston .NET UG

Blue Fenix — Tue, 07 Mar 2006 22:00:00 GMT

BradA at the Houston .NET UG

Christopher — Tue, 07 Mar 2006 22:19:30 GMT

re: Husk at rydde op - Desctructor

TrackBack — Mon, 10 Apr 2006 15:48:49 GMT

Critical Finalizers (Reliability Part 1)

Coding Sanity — Thu, 13 Apr 2006 00:55:28 GMT

For those of you who read the article at sadeveloper, apologies. From what I can see, the article got...

Why is my application coughing up a SecurityException after my code stops running?

I may have joined the wrong side — Wed, 28 Jun 2006 21:21:55 GMT

Odd exceptions at odd times
If you apply a PrincipalPermission attribute to a class in order to restrict...

An intelligent backup system for Windows, part 3

Flux and Mutability — Wed, 22 Nov 2006 01:34:45 GMT

An intelligent backup system for Windows, part 3

Why is my application coughing up a SecurityException after my code stops running?

I may have joined the wrong side — Sun, 15 Apr 2007 01:52:07 GMT

Odd exceptions at odd times If you apply a PrincipalPermission attribute to a class in order to restrict

FxCop backlog themes: Disposition and finalization

I may have joined the wrong side — Wed, 16 May 2007 23:07:32 GMT

I skipped ahead a while back with my post on the exceptions theme , and it's time to get back on

Window Stations & Desktops in .NET (1.)

jachymko blog — Mon, 18 Jun 2007 14:31:48 GMT

Prvni z clanku o P/Invoke, window stations a desktopech...

Memorias de Un Aprendiz » Interop with Native Libraries

Memorias de Un Aprendiz » Interop with Native Libraries — Thu, 31 Jan 2008 23:15:44 GMT

PingBack from http://kimerop.exofire.net/wordpress/?p=22

visualstudio microsoft com

visualstudio microsoft com — Sun, 27 Apr 2008 12:47:08 GMT

PingBack from http://meadow.freerealmedia.com/visualstudiomicrosoftcom.html

ASP.NET Tips: Looking at the finalization queue

ASP.NET Debugging — Mon, 28 Apr 2008 19:13:04 GMT

So in a previous post, we talked about Understanding when to use a Finalizer in your .NET class so now

Why C++? « CK Wong’s Microscopic Universe

Why C++? « CK Wong’s Microscopic Universe — Fri, 09 May 2008 14:29:16 GMT

PingBack from http://ckwong.wordpress.com/2008/05/09/why-c/

LINQ to MSI - Part 1 bis - Interop with SafeHandles

B# .NET Blog — Sun, 15 Jun 2008 00:56:22 GMT

A rather unexpected intermezzo in this series. Why? The first rule of blogging is that blog readers are

Quantum Bit Designs » Blog Archive » A Thread-Safe IDisposable Base Class

Thu, 10 Jul 2008 15:00:10 GMT

PingBack from http://blog.quantumbitdesigns.com/2008/07/10/a-thread-safe-idisposable-base-class/

Quantum Bit Designs » Blog Archive » A Thread-Safe IDisposable Base Class

Wed, 23 Jul 2008 06:48:15 GMT

PingBack from http://blog.quantumbitdesigns.com/2008/07/22/a-thread-safe-idisposable-base-class/

Compare oxycontin to oxycodone.

Oxycontin vs. oxycodone. — Sat, 08 Nov 2008 06:56:53 GMT

Oxycodone and oxycontin. Compare oxycontin to oxycodone. Oxycontin vs. oxycodone.

C# COM Object Class Destructor NOT CALLED EVER | keyongtech

C# COM Object Class Destructor NOT CALLED EVER | keyongtech — Thu, 22 Jan 2009 07:37:46 GMT

PingBack from http://www.keyongtech.com/477443-c-com-object-class-destructor

ThreadAbortException

CLR Team Blog — Tue, 28 Apr 2009 21:54:22 GMT

System.Threading.ThreadAbortException is just plain weird. For instance, most exceptions happen because

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Tue, 28 Apr 2009 23:20:14 GMT

PingBack from http://www.codedstyle.com/threadabortexception-6/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Tue, 28 Apr 2009 23:20:15 GMT

PingBack from http://www.codedstyle.com/threadabortexception-3/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Tue, 28 Apr 2009 23:20:15 GMT

PingBack from http://www.codedstyle.com/threadabortexception-12/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Tue, 28 Apr 2009 23:21:19 GMT

PingBack from http://www.codedstyle.com/threadabortexception/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Tue, 28 Apr 2009 23:21:19 GMT

PingBack from http://www.codedstyle.com/threadabortexception-9/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Tue, 28 Apr 2009 23:21:22 GMT

PingBack from http://www.codedstyle.com/threadabortexception-5/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Tue, 28 Apr 2009 23:21:37 GMT

PingBack from http://www.codedstyle.com/threadabortexception-7/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Thu, 07 May 2009 06:38:20 GMT

PingBack from http://www.codedstyle.com/threadabortexception-4/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Thu, 07 May 2009 06:40:20 GMT

PingBack from http://www.codedstyle.com/threadabortexception-10/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Thu, 07 May 2009 06:40:50 GMT

PingBack from http://www.codedstyle.com/threadabortexception-11/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Thu, 07 May 2009 06:41:41 GMT

PingBack from http://www.codedstyle.com/threadabortexception-13/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Thu, 07 May 2009 06:42:47 GMT

PingBack from http://www.codedstyle.com/threadabortexception-16/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Thu, 07 May 2009 06:43:12 GMT

PingBack from http://www.codedstyle.com/threadabortexception-17/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Thu, 07 May 2009 06:43:59 GMT

PingBack from http://www.codedstyle.com/threadabortexception-19/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Thu, 07 May 2009 06:44:49 GMT

PingBack from http://www.codedstyle.com/threadabortexception-20/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Thu, 07 May 2009 06:45:12 GMT

PingBack from http://www.codedstyle.com/threadabortexception-21/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Thu, 07 May 2009 06:45:31 GMT

PingBack from http://www.codedstyle.com/threadabortexception-22/

ThreadAbortException | Coded Style

ThreadAbortException | Coded Style — Thu, 07 May 2009 06:46:34 GMT

PingBack from http://www.codedstyle.com/threadabortexception-24/

cbrumme s WebLog Finalization | Indoor Grills

cbrumme s WebLog Finalization | Indoor Grills — Mon, 01 Jun 2009 22:26:31 GMT

PingBack from http://indoorgrillsrecipes.info/story.php?id=3119

cbrumme s WebLog Finalization | Outdoor Decor

cbrumme s WebLog Finalization | Outdoor Decor — Sun, 14 Jun 2009 01:07:45 GMT

PingBack from http://outdoordecoration.info/story.php?id=1469

cbrumme s WebLog Finalization | pool toys

cbrumme s WebLog Finalization | pool toys — Thu, 18 Jun 2009 11:31:06 GMT

PingBack from http://pooltoysite.info/story.php?id=1114

cbrumme s WebLog Finalization | garden decor

cbrumme s WebLog Finalization | garden decor — Fri, 19 Jun 2009 10:13:37 GMT

PingBack from http://gardendecordesign.info/story.php?id=3936