R – OutOfMemory, but no gcroots for many objects

garbage-collectionnetout-of-memorywindbgwinforms

We are developing a rather large Windows Forms application. In several customers' computers it often crashes with OutOfMemory exception. After obtaining full memory dump of the application moments after the exception (clrdump invoked from UnhandledException handler) I analyzed it with ".NET Memory Profiler" and windbg.

The Memory Profiler has shown only 130MB in live object instances. What's interesting is that for many object types is has shown a very large number of unreachable instances (e.g. 22000 unreachable Byte[] instances). In native memory statistics it totals 127MB in all heaps for Data (which is ok), but indicates unreachable 133MB in gen #2 heap and 640MB in large heap(not ok!).

When analyzing the dump with windbg, the above stats are confirmed:

!dumpheap -stat
..... acceptable object sizes...
79330a00   467216     30638712 System.String
0016d488     4804    221756612      Free
79333470    27089    574278304 System.Byte[]

The application does use large number of short buffers through its run time, but does not leak them. Testing many of the Byte[] instances with !gcroot ends up with no roots. Obviously most of those arrays are unreachable as indicated by the memory profiler.

Just to ensure all is fine, !finalizequeue shows no objects are waiting to be finalized

generation 0 has 138 finalizable objects (18bd1938->18bd1b60)
generation 1 has 182 finalizable objects (18bd1660->18bd1938)
generation 2 has 75372 finalizable objects (18b87cb0->18bd1660)
Ready for finalization 0 objects (18bd1b60->18bd1b60)

And also check for native finalizer thread stack trace shows it is not blocked.

At the moment I don't how to diagnose why the GC doesn't collect the data (and I believe it would love to since the process ran out of memory..)

edit: Based in input below I read some more on Large Object Heap fragmentation and it seems that this could be the case.

I have seen some advices to allocate bigger blocks of memory for this kind of data (various byte[] in my case) and manage the memory in this area by myself, but this seems like a rather hackish solution, not the one I would expect to resolve a problem with not-so-special desktop application.

The fragmentation issue is caused by the fact (At least that is what many people from Microsoft state in blogs) that objects on LOH are not relocated during existence, which is understandable, but it seems logical that once some memory pressure is reached, such as a threat of getting OOM, relocation should be performed.

The only thing that worries me before fully trusting that fragmentation is the cause, is that so many object on the LOH are without gcroot references – is this because even for LOH garbage collection is performed only partially?

I'll be happy for pointing me to any interesting solution as at the moment the only one that I know of is custom management of some preallocated memory block.

Any ideas are welcome.
Thanks.

Best Answer

The LOH is subject to fragmentation. This article provides an analysis and the basic directions to work around it.
Maybe you could post some code showing a 'typical' usage of those byte[] buffers?

Related Solutions

C# – Deep cloning objects

Whereas one approach is to implement the ICloneable interface (described here, so I won't regurgitate), here's a nice deep clone object copier I found on The Code Project a while ago and incorporated it into our code. As mentioned elsewhere, it requires your objects to be serializable.

using System;
using System.IO;
using System.Runtime.Serialization;
using System.Runtime.Serialization.Formatters.Binary;

/// <summary>
/// Reference Article http://www.codeproject.com/KB/tips/SerializedObjectCloner.aspx
/// Provides a method for performing a deep copy of an object.
/// Binary Serialization is used to perform the copy.
/// </summary>
public static class ObjectCopier
{
    /// <summary>
    /// Perform a deep copy of the object via serialization.
    /// </summary>
    /// <typeparam name="T">The type of object being copied.</typeparam>
    /// <param name="source">The object instance to copy.</param>
    /// <returns>A deep copy of the object.</returns>
    public static T Clone<T>(T source)
    {
        if (!typeof(T).IsSerializable)
        {
            throw new ArgumentException("The type must be serializable.", nameof(source));
        }

        // Don't serialize a null object, simply return the default for that object
        if (ReferenceEquals(source, null)) return default;

        using var Stream stream = new MemoryStream();
        IFormatter formatter = new BinaryFormatter();
        formatter.Serialize(stream, source);
        stream.Seek(0, SeekOrigin.Begin);
        return (T)formatter.Deserialize(stream);
    }
}

The idea is that it serializes your object and then deserializes it into a fresh object. The benefit is that you don't have to concern yourself about cloning everything when an object gets too complex.

In case of you prefer to use the new extension methods of C# 3.0, change the method to have the following signature:

public static T Clone<T>(this T source)
{
   // ...
}

Now the method call simply becomes objectBeingCloned.Clone();.

EDIT (January 10 2015) Thought I'd revisit this, to mention I recently started using (Newtonsoft) Json to do this, it should be lighter, and avoids the overhead of [Serializable] tags. (NB @atconway has pointed out in the comments that private members are not cloned using the JSON method)

/// <summary>
/// Perform a deep Copy of the object, using Json as a serialization method. NOTE: Private members are not cloned using this method.
/// </summary>
/// <typeparam name="T">The type of object being copied.</typeparam>
/// <param name="source">The object instance to copy.</param>
/// <returns>The copied object.</returns>
public static T CloneJson<T>(this T source)
{            
    // Don't serialize a null object, simply return the default for that object
    if (ReferenceEquals(source, null)) return default;

    // initialize inner objects individually
    // for example in default constructor some list property initialized with some values,
    // but in 'source' these items are cleaned -
    // without ObjectCreationHandling.Replace default constructor values will be added to result
    var deserializeSettings = new JsonSerializerSettings {ObjectCreationHandling = ObjectCreationHandling.Replace};

    return JsonConvert.DeserializeObject<T>(JsonConvert.SerializeObject(source), deserializeSettings);
}

C# – What are the correct version numbers for C#

C# language version history:

These are the versions of C# known about at the time of this writing:

C# 1.0 released with .NET 1.0 and VS2002 (January 2002)
C# 1.2 (bizarrely enough); released with .NET 1.1 and VS2003 (April 2003). First version to call Dispose on IEnumerators which implemented IDisposable. A few other small features.
C# 2.0 released with .NET 2.0 and VS2005 (November 2005). Major new features: generics, anonymous methods, nullable types, and iterator blocks
C# 3.0 released with .NET 3.5 and VS2008 (November 2007). Major new features: lambda expressions, extension methods, expression trees, anonymous types, implicit typing (var), and query expressions
C# 4.0 released with .NET 4 and VS2010 (April 2010). Major new features: late binding (dynamic), delegate and interface generic variance, more COM support, named arguments, tuple data type and optional parameters
C# 5.0 released with .NET 4.5 and VS2012 (August 2012). Major features: async programming, and caller info attributes. Breaking change: loop variable closure.
C# 6.0 released with .NET 4.6 and VS2015 (July 2015). Implemented by Roslyn. Features: initializers for automatically implemented properties, using directives to import static members, exception filters, element initializers, await in catch and finally, extension Add methods in collection initializers.
C# 7.0 released with .NET 4.7 and VS2017 (March 2017). Major new features: tuples, ref locals and ref return, pattern matching (including pattern-based switch statements), inline out parameter declarations, local functions, binary literals, digit separators, and arbitrary async returns.
C# 7.1 released with VS2017 v15.3 (August 2017). New features: async main, tuple member name inference, default expression, and pattern matching with generics.
C# 7.2 released with VS2017 v15.5 (November 2017). New features: private protected access modifier, Span<T>, aka interior pointer, aka stackonly struct, and everything else.
C# 7.3 released with VS2017 v15.7 (May 2018). New features: enum, delegate and unmanaged generic type constraints. ref reassignment. Unsafe improvements: stackalloc initialization, unpinned indexed fixed buffers, custom fixed statements. Improved overloading resolution. Expression variables in initializers and queries. == and != defined for tuples. Auto-properties' backing fields can now be targeted by attributes.
C# 8.0 released with .NET Core 3.0 and VS2019 v16.3 (September 2019). Major new features: nullable reference-types, asynchronous streams, indices and ranges, readonly members, using declarations, default interface methods, static local functions, and enhancement of interpolated verbatim strings.
C# 9.0 released with .NET 5.0 and VS2019 v16.8 (November 2020). Major new features: init-only properties, records, with-expressions, data classes, positional records, top-level programs, improved pattern matching (simple type patterns, relational patterns, logical patterns), improved target typing (target-type new expressions, target typed ?? and ?), and covariant returns. Minor features: relax ordering of ref and partial modifiers, parameter null checking, lambda discard parameters, native ints, attributes on local functions, function pointers, static lambdas, extension GetEnumerator, module initializers, and extending partial.

In response to the OP's question:

What are the correct version numbers for C#? What came out when? Why can't I find any answers about C# 3.5?

There is no such thing as C# 3.5 - the cause of confusion here is that the C# 3.0 is present in .NET 3.5. The language and framework are versioned independently, however - as is the CLR, which is at version 2.0 for .NET 2.0 through 3.5, .NET 4 introducing CLR 4.0, service packs notwithstanding. The CLR in .NET 4.5 has various improvements, but the versioning is unclear: in some places it may be referred to as CLR 4.5 (this MSDN page used to refer to it that way, for example), but the Environment.Version property still reports 4.0.xxx.

As of May 3, 2017, the C# Language Team created a history of C# versions and features on their GitHub repository: Features Added in C# Language Versions. There is also a page that tracks upcoming and recently implemented language features.