Windows Defender Application Control (WDAC) and .NET

.NET apps (as written in a high-level language like C#) are compiled to an Intermediate Language (IL). IL is a compact code format that can be supported on any operating system or architecture. Most .NET apps use APIs that are supported in multiple environments, requiring only the .NET runtime to run. IL needs to be compiled to native code in order to execute on a CPU, for example Arm64 or x64. When .NET compiles IL to native image (NI) on a device with a WDAC user mode policy, it first checks whether the original IL file passes the current WDAC policies. If so, .NET sets an NTFS extended attribute (EA) on the generated NI file so that WDAC knows to trust it as well. When the .NET app runs, WDAC sees the EA on the NI file and allows it.

The EA set on the NI file only applies to the currently active WDAC policies. If one of the active WDAC policies is updated or a new policy is applied, the EA on the NI file is invalidated. The next time the app runs, WDAC will block the NI file. .NET handles the block gracefully and falls back to the original IL code. If the IL still passes the latest WDAC policies, then the app runs without any functional impact. Since the IL is now being compiled at runtime, you might notice a slight impact to performance of the app. When .NET must fall back to IL, .NET will also schedule a process to run at the next maintenance window to regenerate all NI files, thus reestablishing the WDAC EA for all code that passes the latest WDAC policies.

In some cases, if an NI file is blocked, you might see a "false positive" block event in the CodeIntegrity - Operational event log as described in WDAC Admin Tips & Known Issues.

To mitigate any performance impact caused when the WDAC EA isn't valid or missing:

  • Avoid updating the WDAC policies often.
  • Run ngen update (on all machine architectures) to force .NET to regenerate all NI files immediately after applying changes to your WDAC policies.
  • Migrate applications to .NET Core (.NET 6 or greater).

WDAC and .NET hardening

Security researchers found that some .NET capabilities that allow apps to load libraries from external sources or generate new code at runtime can be used to circumvent WDAC controls. To address this potential vulnerability, WDAC includes an option called Dynamic Code Security that works with .NET to verify code loaded at runtime.

When the Dynamic Code Security option is enabled, Application Control policy is applied to libraries that .NET loads from external sources. For example, any remote sources, such as the internet or a network share.

Important

.Net dynamic code security hardening is turned on and enforced if any WDAC policy with UMCI enabled has set option 19 Enabled:Dynamic Code Security. There is no audit mode for this feature. You should test your apps with this option set before turning it on across large numbers of devices.

Additionally, it detects tampering in code generated to disk by .NET and blocks loading code that was tampered with.

Dynamic Code Security isn't enabled by default because existing policies might not account for externally loaded libraries. Additionally, a few .NET loading features, including loading unsigned assemblies built with System.Reflection.Emit, aren't currently supported with Dynamic Code Security enabled. Microsoft recommends testing Dynamic Code Security in audit mode before enforcing it to discover whether any new libraries should be included in the policy.

Additionally, customers can precompile for deployment only to prevent an allowed executable from being terminated because it tries to load unsigned dynamically generated code. See the "Precompiling for Deployment Only" section in the ASP.NET Precompilation Overview document for how to fix that.

To enable Dynamic Code Security, add the following option to the <Rules> section of your WDAC policy:

<Rule>
    <Option>Enabled:Dynamic Code Security</Option>
</Rule>