[PG] force register local memory for P2P memory regions by UNIDY2002 · Pull Request #1690 · kvcache-ai/Mooncake

UNIDY2002 · 2026-03-18T07:17:37Z

Description

Similar to #1656.

Module

Type of Change

How Has This Been Tested?

Checklist

I have performed a self-review of my own code.
I have formatted my own code using ./scripts/code_format.sh before submitting.
I have updated the documentation.
I have added tests to prove my changes are effective.

gemini-code-assist · 2026-03-18T07:17:56Z

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 addresses a bug in the PyTorch Backend (mooncake-pg) by enhancing the P2P memory registration mechanism. It ensures that local memory regions used for P2P communication are registered with a specific location, moving away from a wildcard approach. This change aims to improve the precision and correctness of memory management for P2P operations, aligning with best practices for distributed memory handling.

Highlights

P2P Memory Registration: Introduced a location field to the P2PProxy::Options and P2PProxy class to specify the memory region for P2P operations.
Specific Location Usage: Modified the registerLocalMemory calls within P2PProxy to use the newly added location_ member instead of a generic kWildcardLocation for both CPU and CUDA memory buffers and control regions.

🧠 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.

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 location parameter to force registration of local memory for P2P memory regions with a specific location, instead of using a wildcard. The changes are straightforward and correctly implement the intended logic by passing the location from the MooncakeBackend down to the P2PProxy where it's used for memory registration.

My main feedback is regarding code formatting consistency. While the code was likely auto-formatted, there are some inconsistencies in how function calls are wrapped. I've added specific suggestions to improve readability and maintain a consistent style. Overall, the changes are good.

gemini-code-assist · 2026-03-18T07:20:12Z

mooncake-pg/src/p2p_proxy.cpp

+        int rc = engine_->registerLocalMemory(resources_.send_buffer_,
+                                              kP2PTotalBufferSize, location_);


To improve readability and maintain consistency, please consider reformatting this function call. The formatting of registerLocalMemory calls has become inconsistent in this file. This comment applies to other similar calls in this file.

Suggested change

int rc = engine_->registerLocalMemory(resources_.send_buffer_,

kP2PTotalBufferSize, location_);

int rc = engine_->registerLocalMemory(

resources_.send_buffer_, kP2PTotalBufferSize, location_);

gemini-code-assist · 2026-03-18T07:20:12Z

mooncake-pg/src/p2p_proxy.cpp

+        rc = engine_->registerLocalMemory(resources_.recv_buffer_,
+                                          kP2PTotalBufferSize, location_);


Please reformat for consistency, as mentioned in the previous comment.

Suggested change

rc = engine_->registerLocalMemory(resources_.recv_buffer_,

kP2PTotalBufferSize, location_);

rc = engine_->registerLocalMemory(

resources_.recv_buffer_, kP2PTotalBufferSize, location_);

gemini-code-assist · 2026-03-18T07:20:12Z

mooncake-pg/src/p2p_proxy.cpp

+        int rc = engine_->registerLocalMemory(resources_.send_buffer_,
+                                              kP2PTotalBufferSize, location_);


Please reformat for consistency, as mentioned in the previous comments.

Suggested change

int rc = engine_->registerLocalMemory(resources_.send_buffer_,

kP2PTotalBufferSize, location_);

int rc = engine_->registerLocalMemory(

resources_.send_buffer_, kP2PTotalBufferSize, location_);

gemini-code-assist · 2026-03-18T07:20:12Z

mooncake-pg/src/p2p_proxy.cpp

+        rc = engine_->registerLocalMemory(resources_.recv_buffer_,
+                                          kP2PTotalBufferSize, location_);


Please reformat for consistency, as mentioned in the previous comments.

Suggested change

rc = engine_->registerLocalMemory(resources_.recv_buffer_,

kP2PTotalBufferSize, location_);

rc = engine_->registerLocalMemory(

resources_.recv_buffer_, kP2PTotalBufferSize, location_);

gemini-code-assist · 2026-03-18T07:20:12Z

mooncake-pg/src/p2p_proxy.cpp

    int rc = engine_->registerLocalMemory(resources_.ctrl_send_region_,
                                          kMaxNumRanks * sizeof(P2PControlSlot),
-                                          kWildcardLocation);
+                                          location_);


Please reformat for consistency. This formatting is also inconsistent with other calls in this file (e.g. line 75).

Suggested change

int rc = engine_->registerLocalMemory(resources_.ctrl_send_region_,

kMaxNumRanks * sizeof(P2PControlSlot),

kWildcardLocation);

location_);

int rc = engine_->registerLocalMemory(

resources_.ctrl_send_region_,

kMaxNumRanks * sizeof(P2PControlSlot),

location_);

gemini-code-assist · 2026-03-18T07:20:12Z

mooncake-pg/src/p2p_proxy.cpp

    rc = engine_->registerLocalMemory(resources_.ctrl_recv_region_,
                                      kMaxNumRanks * sizeof(P2PControlSlot),
-                                      kWildcardLocation);
+                                      location_);


Please reformat for consistency, as mentioned in the previous comments.

Suggested change

rc = engine_->registerLocalMemory(resources_.ctrl_recv_region_,

kMaxNumRanks * sizeof(P2PControlSlot),

kWildcardLocation);

location_);

rc = engine_->registerLocalMemory(

resources_.ctrl_recv_region_,

kMaxNumRanks * sizeof(P2PControlSlot),

location_);

codecov-commenter · 2026-03-18T07:51:09Z

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

ympcMark

LGTM

) Similar to kvcache-ai#1656.

[PG] force register local memory for P2P memory regions

d37ecea

UNIDY2002 requested review from ympcMark and yuechen-sys as code owners March 18, 2026 07:17

github-actions bot added run-ci PyTorch Backend labels Mar 18, 2026

gemini-code-assist bot reviewed Mar 18, 2026

View reviewed changes

ympcMark approved these changes Mar 18, 2026

View reviewed changes

UNIDY2002 merged commit 47b016d into main Mar 18, 2026
18 of 19 checks passed

UNIDY2002 deleted the sunxun/fix branch March 18, 2026 08:20

caozhanhao mentioned this pull request Mar 24, 2026

[TE] Fix simultaneous open handshake in RdmaEndpoint #1733

Merged

21 tasks

KMSorSMS mentioned this pull request Mar 26, 2026

[PG][TENT] Fix first-collective hangs on NVLink/MNNVL bootstrap #1755

Merged

21 tasks

whn09 pushed a commit to whn09/Mooncake that referenced this pull request Apr 4, 2026

[PG] force register local memory for P2P memory regions (kvcache-ai#1690

f4225cb

) Similar to kvcache-ai#1656.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[PG] force register local memory for P2P memory regions#1690

[PG] force register local memory for P2P memory regions#1690
UNIDY2002 merged 1 commit intomainfrom
sunxun/fix

UNIDY2002 commented Mar 18, 2026

Uh oh!

gemini-code-assist bot commented Mar 18, 2026

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

gemini-code-assist bot Mar 18, 2026

Uh oh!

gemini-code-assist bot Mar 18, 2026

Uh oh!

gemini-code-assist bot Mar 18, 2026

Uh oh!

gemini-code-assist bot Mar 18, 2026

Uh oh!

gemini-code-assist bot Mar 18, 2026

Uh oh!

gemini-code-assist bot Mar 18, 2026

Uh oh!

codecov-commenter commented Mar 18, 2026

Uh oh!

ympcMark left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

		int rc = engine_->registerLocalMemory(resources_.send_buffer_,
		kP2PTotalBufferSize, location_);

		rc = engine_->registerLocalMemory(resources_.recv_buffer_,
		kP2PTotalBufferSize, location_);

Conversation

UNIDY2002 commented Mar 18, 2026

Description

Module

Type of Change

How Has This Been Tested?

Checklist

Uh oh!

gemini-code-assist bot commented Mar 18, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist bot Mar 18, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Mar 18, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Mar 18, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Mar 18, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Mar 18, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Mar 18, 2026

Choose a reason for hiding this comment

Uh oh!

codecov-commenter commented Mar 18, 2026

Codecov Report

Uh oh!

ympcMark left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants