Building a custom file system or a file-system mini-filter driver for Windows is notoriously complex. Whether you are developing a cloud storage synchronization client (like OneDrive or Dropbox), a Hierarchical Storage Management (HSM) system, or a transparent encryption tool, you will inevitably run into one of the most powerful—and misunderstood—features of the NTFS file system: Reparse Points.

Navigating the intricacies of Windows I/O Request Packets (IRPs) to manage these points from scratch is highly complex and error-prone. Fortunately, the EaseFilter SDK (including the CloudTier Storage Tiering SDK) provides a robust wrapper around the Windows Filter Manager, making it significantly easier to intercept, manage, and utilize reparse points.

Here is a deep dive into handling reparse points for custom file systems using the EaseFilter SDK.

What Are Reparse Points?

Before writing any code, it is crucial to understand what a reparse point actually is. Introduced with NTFS v3.0, a reparse point is an NTFS file system object containing a collection of user-defined data.

When the Windows I/O Manager opens a file or directory and encounters a reparse point, it looks at the Reparse Tag—a unique identifier indicating which file system filter driver should handle the file.

• Standard Uses: Microsoft uses reparse points natively for Symbolic Links, Directory Junctions, and Volume Mount Points.
• Custom Uses: Third-party developers use them to implement "placeholder" files (files that appear to be on disk but actually reside in the cloud) or to redirect file access to completely different locations.

If your filter driver owns the tag, it parses the custom data buffer, performs its logic (like downloading a file from the cloud), and then tells the I/O Manager how to proceed.

The EaseFilter SDK Advantage

Developing a Windows Mini-Filter driver in C/C++ requires a deep understanding of kernel-mode programming. The EaseFilter SDK bridges this gap by providing high-level user-mode APIs (available in C#, C++, and Delphi) that communicate seamlessly with a pre-certified kernel-mode driver.

When handling reparse points, EaseFilter abstracts away the complexities of IRP_MJ_CREATE interception and FSCTL_SET_REPARSE_POINT control codes, allowing you to focus purely on your application's business logic.

Handling Reparse Points: Step-by-Step

Here is how you can effectively manage reparse points using the EaseFilter SDK.

1. Registering Your Custom Reparse Tag

To use custom reparse points, you need a unique Reparse Tag. While Microsoft provides a mechanism to request official third-party tags, you can often use test tags during development. You must configure the EaseFilter driver to listen for files tagged with your specific identifier.

2. Creating a Placeholder File (Stub)

In an HSM or Cloud Sync scenario (such as those built with the CloudTier SDK), you don't want to download a 5GB video file just to show it in Windows Explorer. Instead, you create a 0-byte "stub" file and attach a reparse point to it.

With EaseFilter, you can intercept the file creation process or proactively set the reparse point data using the SDK's file management functions. The data buffer you attach might contain the cloud URL or the backend database ID required to fetch the actual file content later.

3. Intercepting the File Open Request (OnPreCreate)

When a user double-clicks your placeholder file, the Windows I/O Manager initiates a File Open request. The EaseFilter driver intercepts this and triggers an event in your user-mode application.

4. Returning STATUS_REPARSE

This is the magic mechanism. By returning STATUS_REPARSE (or STATUS_REPARSE_DATA_ERROR depending on the context) and supplying a new file path, you are instructing the Windows Object Manager to abandon the current open request and start over using the new path you provided. This happens completely transparently to the user application (like Word or VLC Player), which simply opens the file as if it were there all along.

Best Practices When Using EaseFilter

• Handle Timeouts Gracefully: If fetching the real file data takes too long (e.g., a slow network connection), the calling application might time out. Implement progress callbacks if possible, or ensure your caching mechanism is highly optimized.
• Manage Anti-Virus Scanners: AV software frequently scans newly discovered files, which can accidentally trigger your reparse points and cause massive unintended data downloads from your cloud tier. Use EaseFilter's process-filtering capabilities to ignore requests coming from known AV executables unless explicitly required.
• Clean Up Reparse Data: When a file is "hydrated" (fully downloaded to the local disk), remember to remove the reparse point attribute so subsequent accesses are processed natively by NTFS, saving CPU cycles and avoiding unnecessary round-trips to your user-mode service.

Conclusion

Reparse points are an essential tool for building modern, intelligent file systems. While kernel-mode development can be daunting, the EaseFilter SDK and CloudTier SDK provide a stable, accessible, and powerful framework to implement custom file redirection, tiering, and placeholder logic. By properly managing OnPreCreate events and leveraging STATUS_REPARSE, you can create a seamless experience for your end-users, securely bridging the gap between local storage and your custom backend.