[None][fix] fix EPLB init hang

[dongxuy04]

Summary by CodeRabbit

• Bug Fixes
  • Fixed expert token count initialization in mixture-of-experts layers to ensure proper execution state handling during model inference.

Description

Fix EPLB init hang.

Test Coverage

PR Checklist

Please review the following before submitting your PR:

• PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.

• PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.

• Test cases are provided for new code paths (see test instructions)

• Any new dependencies have been scanned for license and vulnerabilities

• CODEOWNERS updated if ownership changes

• Documentation updated as needed

• The reviewers assigned automatically/manually are appropriate for the PR.

• Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

/bot [-h] ['run', 'kill', 'skip', 'reuse-pipeline'] ...

Provide a user friendly way for developers to interact with a Jenkins server.

Run /bot [-h|--help] to print this help message.

See details below for each supported subcommand.

run [--reuse-test (optional)pipeline-id --disable-fail-fast --skip-test --stage-list "A10-PyTorch-1, xxx" --gpu-type "A30, H100_PCIe" --test-backend "pytorch, cpp" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" --detailed-log --debug(experimental)]

Launch build/test pipelines. All previously running jobs will be killed.

--reuse-test (optional)pipeline-id (OPTIONAL) : Allow the new pipeline to reuse build artifacts and skip successful test stages from a specified pipeline or the last pipeline if no pipeline-id is indicated. If the Git commit ID has changed, this option will be always ignored. The DEFAULT behavior of the bot is to reuse build artifacts and successful test results from the last pipeline.

--disable-reuse-test (OPTIONAL) : Explicitly prevent the pipeline from reusing build artifacts and skipping successful test stages from a previous pipeline. Ensure that all builds and tests are run regardless of previous successes.

--disable-fail-fast (OPTIONAL) : Disable fail fast on build/tests/infra failures.

--skip-test (OPTIONAL) : Skip all test stages, but still run build stages, package stages and sanity check stages. Note: Does NOT update GitHub check status.

--stage-list "A10-PyTorch-1, xxx" (OPTIONAL) : Only run the specified test stages. Examples: "A10-PyTorch-1, xxx". Note: Does NOT update GitHub check status.

--gpu-type "A30, H100_PCIe" (OPTIONAL) : Only run the test stages on the specified GPU types. Examples: "A30, H100_PCIe". Note: Does NOT update GitHub check status.

--test-backend "pytorch, cpp" (OPTIONAL) : Skip test stages which don't match the specified backends. Only support [pytorch, cpp, tensorrt, triton]. Examples: "pytorch, cpp" (does not run test stages with tensorrt or triton backend). Note: Does NOT update GitHub pipeline status.

--only-multi-gpu-test (OPTIONAL) : Only run the multi-GPU tests. Note: Does NOT update GitHub check status.

--disable-multi-gpu-test (OPTIONAL) : Disable the multi-GPU tests. Note: Does NOT update GitHub check status.

--add-multi-gpu-test (OPTIONAL) : Force run the multi-GPU tests in addition to running L0 pre-merge pipeline.

--post-merge (OPTIONAL) : Run the L0 post-merge pipeline instead of the ordinary L0 pre-merge pipeline.

--extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" (OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx".

--detailed-log (OPTIONAL) : Enable flushing out all logs to the Jenkins console. This will significantly increase the log volume and may slow down the job.

--debug (OPTIONAL) : Experimental feature. Enable access to the CI container for debugging purpose. Note: Specify exactly one stage in the stage-list parameter to access the appropriate container environment. Note: Does NOT update GitHub check status.

For guidance on mapping tests to stage names, see docs/source/reference/ci-overview.md
and the scripts/test_to_stage_mapping.py helper.

kill

kill

Kill all running builds associated with pull request.

skip

skip --comment COMMENT

Skip testing for latest commit on pull request. --comment "Reason for skipping build/test" is required. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

reuse-pipeline

reuse-pipeline

Reuse a previous pipeline to validate current commit. This action will also kill all currently running builds associated with the pull request. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

[dongxuy04]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #22432 [ run ] triggered by Bot. Commit: f14f596

[coderabbitai]

📝 Walkthrough

Walkthrough

Removed the early exit condition in zeroExpertTokenCountKernel when the enabled parameter is zero. The kernel now unconditionally executes and initializes and writes expert token counts, writing zeros when disabled instead of skipping execution.

Changes

Cohort / File(s)	Summary
CUDA Kernel Control Flow cpp/tensorrt_llm/kernels/moeLoadBalance/moeLoadBalanceKernels.cu	Removed conditional early exit when enabled is 0 in zeroExpertTokenCountKernel; kernel now always executes and writes zeros to per-block expert token count output

Sequence Diagram(s)

sequenceDiagram
    autonumber
    participant Thread as Thread Block
    participant Kernel as zeroExpertTokenCountKernel
    participant Memory as Output Memory

    rect rgb(200, 220, 255)
        Note over Kernel: Previous Behavior (with guard)
        alt enabled != 0
            Thread->>Kernel: invoke with enabled=1
            Kernel->>Memory: write expert token count
        else enabled == 0
            Thread->>Kernel: invoke with enabled=0
            Kernel-->>Kernel: early exit (return)
            Kernel-->>Memory: no write
        end
    end

    rect rgb(220, 255, 200)
        Note over Kernel: New Behavior (unconditional execution)
        Thread->>Kernel: invoke with enabled=0 or enabled=1
        Kernel->>Memory: always write expert token count (zero if disabled)
    end

Loading

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Pre-merge checks and finishing touches

❌ Failed checks (2 warnings)

Check name	Status	Explanation	Resolution
Description Check	⚠️ Warning	The PR description is largely incomplete and fails to meet the template requirements. The Description section only contains a single sentence—"Fix EPLB init hang"—which merely repeats the title and does not explain what the issue is, why it occurs, or how the kernel change addresses it. The Test Coverage section is entirely empty with no test cases listed to validate the change. Additionally, the PR checklist shows most items unchecked, indicating incomplete review of the required checkpoints. While the structural template is present, the substantive content required by the template is missing.	The description should be expanded to explain: (1) what the initialization hang issue was and how it manifested, (2) why removing the early exit condition and ensuring unconditional kernel execution fixes the hang, and (3) the impact of this change on the system. Additionally, the Test Coverage section must list specific tests that validate this change, such as existing MOE load balancing tests or new tests demonstrating the hang is resolved. Finally, the PR author should review and complete the PR checklist items to confirm alignment with TRT-LLM coding guidelines and other requirements.
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	You can run @coderabbitai generate docstrings to improve docstring coverage.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Title Check	✅ Passed	The PR title "[None][fix] fix EPLB init hang" follows the required format from the template with the [None][fix] structure. It is specific and clearly describes the primary change—fixing an initialization hang in the EPLB (Expert Parallel Load Balancing) system. The title aligns well with the changeset, which removes a conditional early exit in the zeroExpertTokenCountKernel and ensures unconditional initialization of expert token counts. The title is concise, free of noise, and provides meaningful context about what is being fixed.

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

📜 Recent review details

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 602b059 and f14f596.

📒 Files selected for processing (1)

• cpp/tensorrt_llm/kernels/moeLoadBalance/moeLoadBalanceKernels.cu (0 hunks)

💤 Files with no reviewable changes (1)

• cpp/tensorrt_llm/kernels/moeLoadBalance/moeLoadBalanceKernels.cu

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[tensorrt-cicd]

PR_Github #22432 [ run ] completed with state SUCCESS. Commit: f14f596
/LLM/main/L0_MergeRequest_PR pipeline #16906 completed with status: 'FAILURE'

[dongxuy04]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #22447 [ run ] triggered by Bot. Commit: f14f596

[tensorrt-cicd]

PR_Github #22447 [ run ] completed with state FAILURE. Commit: f14f596
/LLM/main/L0_MergeRequest_PR pipeline #16917 completed with status: 'FAILURE'

[dongxuy04]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #22496 [ run ] triggered by Bot. Commit: f14f596

[tensorrt-cicd]

PR_Github #22496 [ run ] completed with state FAILURE. Commit: f14f596
/LLM/main/L0_MergeRequest_PR pipeline #16954 completed with status: 'FAILURE'