[IR][Transform] Implement SequentialNode::ResolveDependency()

[Dayuxiaoshui]

This commit implements the ResolveDependency() method for SequentialNode to handle pass dependency resolution. The implementation:

1. Collects all enabled passes from the current list
2. Recursively collects required passes from the global registry
3. Builds a dependency graph using adjacency lists
4. Performs topological sort using Kahn's algorithm
5. Detects and handles circular dependencies with warnings
6. Updates the passes list with the sorted order

Key features:

• Handles transitive dependencies correctly
• Respects PassContext to filter passes by opt_level
• Gracefully handles unresolvable dependencies
• Warns about circular dependencies

Also includes fixes for Relax/Torch from_exported_program crash with FakeTensor and lifted tensors.

Tested:

• All 3 tests pass
• Code compiles successfully
• Lint checks pass

[gemini-code-assist]

Summary of Changes

Hello @Dayuxiaoshui, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request significantly enhances the IR transformation framework by introducing robust pass dependency resolution within SequentialNode. This ensures that passes are executed in the correct order, respecting their prerequisites and opt_level configurations, which is critical for compiler optimizations. Additionally, it addresses a stability issue in the PyTorch frontend, preventing crashes when translating models that contain FakeTensor or lifted tensors.

Highlights

• Pass Dependency Resolution: Implemented SequentialNode::ResolveDependency() to automatically order passes based on their dependencies. This mechanism uses Kahn's algorithm for topological sorting, handling transitive dependencies and respecting PassContext opt_level filtering.
• Circular Dependency Detection: Introduced detection and warning for circular dependencies within the pass sequence, ensuring graceful handling and logging of such scenarios.
• Relax/Torch Frontend Fix: Fixed a crash in Relax/Torch from_exported_program when encountering FakeTensor or lifted tensors by gracefully skipping them during parameter binding and logging a warning.
• New Test Coverage: Added comprehensive unit tests for pass dependency resolution and the FakeTensor handling in the PyTorch frontend.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request introduces a new dependency resolution mechanism for Sequential passes using topological sort, which is a great improvement over the existing simple dependency handling. It also includes a fix for crashes when importing PyTorch ExportedProgram with FakeTensor.

The implementation of ResolveDependency in C++ is well-structured. However, my main concern is that this new method doesn't seem to be called anywhere, and the existing Sequential::operator() still uses the old, simpler dependency resolution logic. This means the core improvement of this PR might not be active.

I've also added some comments on the C++ implementation regarding exception handling and a small simplification. Additionally, the new Python tests for dependency resolution seem to only cover basic cases and don't validate the new topological sort logic. It would be great to add more comprehensive tests for different dependency graphs.

The fix for the PyTorch frontend looks good and the accompanying test is a valuable addition.