Skip to content

Untrusted data can lead to DoS attack due to hash collisions and stack overflow in MessagePack

Moderate severity GitHub Reviewed Published Jan 31, 2020 in MessagePack-CSharp/MessagePack-CSharp • Updated Oct 16, 2024

Package

nuget MessagePack (NuGet)

Affected versions

< 1.9.11
>= 2.0.0, < 2.1.90

Patched versions

1.9.11
2.1.90
nuget MessagePack.ImmutableCollection (NuGet)
< 1.9.11
>= 2.0.0, < 2.1.90
1.9.11
2.1.90
nuget MessagePack.ReactiveProperty (NuGet)
< 1.9.11
>= 2.0.0, < 2.1.90
1.9.11
2.1.90
nuget MessagePack.Unity (NuGet)
< 1.9.11
>= 2.0.0, < 2.1.90
1.9.11
2.1.90
nuget MessagePack.UnityShims (NuGet)
< 1.9.11
>= 2.0.0, < 2.1.90
1.9.11
2.1.90

Description

Impact

When this library is used to deserialize messagepack data from an untrusted source, there is a risk of a denial of service attack by either of two vectors:

  1. hash collisions - leading to large CPU consumption disproportionate to the size of the data being deserialized.
  2. stack overflow - leading to the deserializing process crashing.

Patches

The following steps are required to mitigate this risk.

  1. Upgrade to a version of the library where a fix is available
  2. Add code to your application to put MessagePack into the defensive UntrustedData mode.
  3. Identify all MessagePack extensions that implement IMessagePackFormatter<T> implementations that do not ship with the MessagePack library to include the security mitigations. This includes those acquired from 3rd party packages and classes included directly into your project. Any AOT formatters generated with the MPC tool must be regenerated with the patched version of mpc.
  4. Review your messagepack-serializable data structures for hash-based collections that use custom or unusual types for the hashed key. See below for details on handling such situations.

Review the MessagePackSecurity class to tweak any settings as necessary to strike the right balance between performance, functionality, and security.

Specialized IEqualityComparer<T> implementations provide the hash collision resistance.
Each type of hashed key may require a specialized implementation of its own.
The patched MessagePack library includes many such implementations for primitive types commonly used as keys in hash-based collections.
If your data structures use custom types as keys in these hash-based collections,
putting MessagePack in UntrustedData mode may lead the deserializer to throw an exception
because no safe IEqualityComparer<T> is available for your custom T type.
You can provide your own safe implementation by deriving from the MessagePackSecurity class
and overriding the GetHashCollisionResistantEqualityComparer<T>() method to return your own
custom implementation when T matches your type, and fallback to return base.GetHashCollisionResistantEqualityComparer<T>(); for types you do not have custom implementations for.

Unrelated to this advisory, but as general security guidance, you should also avoid the Typeless serializer/formatters/resolvers for untrusted data as that opens the door for the untrusted data to potentially deserialize unanticipated types that can compromise security.

MessagePack 1.x users

  1. Upgrade to any 1.9.x version.

  2. When deserializing untrusted data, put MessagePack into a more secure mode with:

    MessagePackSecurity.Active = MessagePackSecurity.UntrustedData;

    In MessagePack v1.x this is a static property and thus the security level is shared by the entire process or AppDomain.
    Use MessagePack v2.1 or later for better control over the security level for your particular use.

  3. Any code produced by mpc should be regenerated with the mpc tool with the matching (patched) version. Such generated code usually is written to a file called Generated.cs. A patched Generated.cs file will typically reference the MessagePackSecurity class.

    Review any custom-written IMessagePackFormatter<T> implementations in your project or that you might use from 3rd party packages to ensure they also utilize the MessagePackSecurity class as required.
    In particular, a formatter that deserializes an object (as opposed to a primitive value) should wrap the deserialization in a using (MessagePackSecurity.DepthStep()) block. For example:

    public MyObject Deserialize(ref MessagePackReader reader, MessagePackSerializerOptions options)
    {
        if (reader.TryReadNil())
        {
            return default;
        }
        else
        {
            using (MessagePackSecurity.DepthStep()) // STACK OVERFLOW MITIGATION
            {
                MyObject o = new MyObject();
                // deserialize members of the object here.
                return o;
            }
        }
    }

    If your custom formatter creates hash-based collections (e.g. Dictionary<K, V> or HashSet<T>) where the hashed key comes from the messagepack data, always instantiate your collection using MessagePackSecurity.Active.GetEqualityComparer<T>() as the equality comparer:

    var collection = new HashSet<T>(MessagePackSecurity.Active.GetEqualityComparer<T>());

    This ensures that when reading untrusted data, you will be using a collision-resistent hash algorithm.

Learn more about best security practices when reading untrusted data with MessagePack 1.x.

MessagePack 2.x users

  1. Upgrade to any 2.1.x or later version.

  2. When deserializing untrusted data, put MessagePack into a more secure mode by configuring your MessagePackSerializerOptions.Security property:

    var options = MessagePackSerializerOptions.Standard
        .WithSecurity(MessagePackSecurity.UntrustedData);
    
    // Pass the options explicitly for the greatest control.
    T object = MessagePackSerializer.Deserialize<T>(data, options);
    
    // Or set the security level as the default.
    MessagePackSerializer.DefaultOptions = options;
  3. Any code produced by mpc should be regenerated with the mpc tool with the matching (patched) version. Such generated code usually is written to a file called Generated.cs. A patched Generated.cs file will typically reference the Security member on the MessagePackSerializerOptions parameter.

    Review any custom-written IMessagePackFormatter<T> implementations in your project or that you might use from 3rd party packages to ensure they also utilize the MessagePackSecurity class as required.
    In particular, a formatter that deserializes an object (as opposed to a primitive value) should call options.Security.DepthStep(ref reader); before deserializing the object's members, and be sure to revert the depth step with reader.Depth--; before exiting the method. For example:

    public MyObject Deserialize(ref MessagePackReader reader, MessagePackSerializerOptions options)
    {
        if (reader.TryReadNil())
        {
            return default;
        }
        else
        {
            options.Security.DepthStep(ref reader); // STACK OVERFLOW MITIGATION, line 1
            try
            {
                MyObject o = new MyObject();
                // deserialize members of the object here.
                return o;
            }
            finally
            {
                reader.Depth--; // STACK OVERFLOW MITIGATION, line 2
            }
        }
    }

    If your custom formatter creates hash-based collections (e.g. Dictionary<K, V> or HashSet<T>) where the hashed key comes from the messagepack data, always instantiate your collection using options.Security.GetEqualityComparer<TKey>() as the equality comparer:

    var collection = new HashSet<T>(options.Security.GetEqualityComparer<T>());

    This ensures that when reading untrusted data, you will be using a collision-resistent hash algorithm.

Learn more about best security practices when reading untrusted data with MessagePack 2.x.

Workarounds

The security vulnerabilities are in the formatters.
Avoiding the built-in formatters entirely in favor of reading messagepack primitive data directly
or relying on carefully written custom formatters can provide a workaround.

MessagePack v1.x users may utilize the MessagePackBinary static class directly to read the data they expect.
MessagePack v2.x users may utilize the MessagePackReader struct directly to read the data they expect.

References

Learn more about best security practices when reading untrusted data with MessagePack 1.x or MessagePack 2.x.

For more information

If you have any questions or comments about this advisory:

References

Reviewed Jan 31, 2020
Published to the GitHub Advisory Database Jan 31, 2020
Last updated Oct 16, 2024

Severity

Moderate

CVSS overall score

This score calculates overall vulnerability severity from 0 to 10 and is based on the Common Vulnerability Scoring System (CVSS).
/ 10

CVSS v3 base metrics

Attack vector
Network
Attack complexity
High
Privileges required
Low
User interaction
Required
Scope
Unchanged
Confidentiality
None
Integrity
None
Availability
High

CVSS v3 base metrics

Attack vector: More severe the more the remote (logically and physically) an attacker can be in order to exploit the vulnerability.
Attack complexity: More severe for the least complex attacks.
Privileges required: More severe if no privileges are required.
User interaction: More severe when no user interaction is required.
Scope: More severe when a scope change occurs, e.g. one vulnerable component impacts resources in components beyond its security scope.
Confidentiality: More severe when loss of data confidentiality is highest, measuring the level of data access available to an unauthorized user.
Integrity: More severe when loss of data integrity is the highest, measuring the consequence of data modification possible by an unauthorized user.
Availability: More severe when the loss of impacted component availability is highest.
CVSS:3.1/AV:N/AC:H/PR:L/UI:R/S:U/C:N/I:N/A:H

EPSS score

0.248%
(64th percentile)

Weaknesses

CVE ID

CVE-2020-5234

GHSA ID

GHSA-7q36-4xx7-xcxf
Loading Checking history
See something to contribute? Suggest improvements for this vulnerability.