Skip to content

Fix missing clone in hicache#20130

Merged
ispobock merged 1 commit intomainfrom
fix-missing-clone
Mar 8, 2026
Merged

Fix missing clone in hicache#20130
ispobock merged 1 commit intomainfrom
fix-missing-clone

Conversation

@ispobock
Copy link
Copy Markdown
Collaborator

@ispobock ispobock commented Mar 8, 2026

Motivation

follow up: #17400

Modifications

Accuracy Tests

Benchmarking and Profiling

Checklist

Review Process

  1. Ping Merge Oncalls to start the PR flow. See the PR Merge Process.
  2. Get approvals from CODEOWNERS and other reviewers.
  3. Trigger CI tests with comments or contact authorized users to do so.
    • /tag-run-ci-label, /rerun-failed-ci, /tag-and-rerun-ci
  4. After green CI and required approvals, ask Merge Oncalls to merge.

@gemini-code-assist
Copy link
Copy Markdown
Contributor

Summary of Changes

Hello, 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 resolves a critical issue within the hiradix_cache module where tensor assignments were creating references instead of independent copies. By introducing explicit .clone() calls, the change guarantees that cache nodes hold their own distinct data, thereby preventing unexpected modifications when the original tensor data changes. This enhancement improves the robustness and reliability of the caching mechanism, especially in complex scenarios like KV cache recomputation.

Highlights

  • Data Integrity: Ensured that tensor values assigned to cache nodes are explicitly cloned, preventing unintended shared references and potential data corruption.
  • Cache Consistency: Addressed a subtle bug in the hiradix_cache module related to how tensor slices were handled during load_back and insert operations, particularly for evicted nodes and KV cache recomputation.

🧠 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.

Changelog
  • python/sglang/srt/mem_cache/hiradix_cache.py
    • Added .clone() to the assignment of node.value in the load_back function to ensure a deep copy of device indices.
    • Added .clone() to the assignment of node.value in the insert function when handling evicted nodes to prevent shared references.
Activity
  • No human activity has been recorded on this pull request yet.
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.

@ispobock
Copy link
Copy Markdown
Collaborator Author

ispobock commented Mar 8, 2026

/tag-and-rerun-ci

@github-actions github-actions Bot added the run-ci label Mar 8, 2026
Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request adds .clone() calls during tensor assignments in python/sglang/srt/mem_cache/hiradix_cache.py. Specifically, in the load_back function, slices of device_indices are cloned, and in the insert function, slices of the value tensor are cloned when updating evicted nodes. This ensures that each node.value becomes an independent tensor with its own memory, rather than a view into a shared underlying storage. This change prevents potential memory corruption issues and aligns with the memory management strategy seen elsewhere in the cache logic, such as in the _split_node function.

@ispobock ispobock merged commit 07359ef into main Mar 8, 2026
155 of 174 checks passed
@ispobock ispobock deleted the fix-missing-clone branch March 8, 2026 15:21
liubiyongge pushed a commit to liubiyongge/sglang that referenced this pull request Mar 13, 2026
Wangzheee pushed a commit to Wangzheee/sglang that referenced this pull request Mar 21, 2026
JustinTong0323 pushed a commit to JustinTong0323/sglang that referenced this pull request Apr 7, 2026
yhyang201 pushed a commit to yhyang201/sglang that referenced this pull request Apr 22, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants