Skip to content

Conversation

@tongyuantongyu
Copy link
Member

@tongyuantongyu tongyuantongyu commented Sep 16, 2025

Summary by CodeRabbit

  • Bug Fixes
    • Improved reliability of loading shared libraries on Linux, reducing “cannot open shared object file” errors when importing or running modules.
    • Better handling of dependencies (including NCCL) across varied install locations, such as system prefixes and Python site-packages.
  • Chores
    • Modernized Linux runtime library search configuration for native and Python bindings to use standard mechanisms.
    • No changes on Windows.

Description

Add RPATH to libraries so they can find necessary libraries on non-system Python installation:

For non-system Python installation (e.g. conda environments), native libraries are installed in $PREFIX/lib while Python packages are installed in $PREFIX/lib/python${PYVER}/site-packages. ADD RPATH to help our libraries find dependencies under $PREFIX/lib.

Fix #6102.

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.

@coderabbitai
Copy link
Contributor

coderabbitai bot commented Sep 16, 2025

📝 Walkthrough

Walkthrough

Updates CMake RPATH handling on Linux across core, nanobind, and pybind targets. Moves rpath configuration from LINK_FLAGS to BUILD_RPATH for module targets and adds additional rpath entries, including a path four levels up and NCCL libs. Windows logic remains unchanged.

Changes

Cohort / File(s) Summary
Core shared library RPATH (Linux)
cpp/tensorrt_llm/CMakeLists.txt
Expands runtime search path from "$ORIGIN" to multi-entry path including "$ORIGIN/../../../.."; adds comments about install prefix locations. No Windows changes.
Module RPATH refactor (Linux)
cpp/tensorrt_llm/nanobind/CMakeLists.txt, cpp/tensorrt_llm/pybind/CMakeLists.txt
Replaces rpath in LINK_FLAGS with BUILD_RPATH using new rpath lists. Adds entries: "$ORIGIN/libs", "$ORIGIN/../../..", and "$ORIGIN/../nvidia/nccl/lib". Preserves AS_NEEDED_FLAG and UNDEFINED_FLAG via LINK_FLAGS. Windows unaffected.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Pre-merge checks and finishing touches

✅ Passed checks (3 passed)
Check name Status Explanation
Docstring Coverage ✅ Passed No functions found in the changes. Docstring coverage check skipped.
Title Check ✅ Passed The PR title "[#6102][fix] support non-system python installation" succinctly captures the primary change—adding support for non-system Python installations by adjusting RPATHs. It includes the issue reference and the correct "[fix]" type token from the repository conventions. The phrasing is concise and clear for reviewers scanning history.
Description Check ✅ Passed The PR description explains the problem and solution clearly, stating that RPATH entries are added so libraries can find dependencies in non-system Python installs and referencing Fix #6102. It uses the repository template and includes the required headings, but the Test Coverage section is empty and no specific tests or reproduction steps are provided. Overall the description is sufficient to understand intent but lacks concrete verification details.
✨ Finishing touches
  • 📝 Generate Docstrings
🧪 Generate unit tests
  • Create PR with unit tests
  • Post copyable unit tests in a comment

Tip

👮 Agentic pre-merge checks are now available in preview!

Pro plan users can now enable pre-merge checks in their settings to enforce checklists before merging PRs.

  • Built-in checks – Quickly apply ready-made checks to enforce title conventions, require pull request descriptions that follow templates, validate linked issues for compliance, and more.
  • Custom agentic checks – Define your own rules using CodeRabbit’s advanced agentic capabilities to enforce organization-specific policies and workflows. For example, you can instruct CodeRabbit’s agent to verify that API documentation is updated whenever API schema files are modified in a PR. Note: Upto 5 custom checks are currently allowed during the preview period. Pricing for this feature will be announced in a few weeks.

Please see the documentation for more information.

Example:

reviews:
  pre_merge_checks:
    custom_checks:
      - name: "Undocumented Breaking Changes"
        mode: "warning"
        instructions: |
          Pass/fail criteria: All breaking changes to public APIs, CLI flags, environment variables, configuration keys, database schemas, or HTTP/GraphQL endpoints must be documented in the "Breaking Change" section of the PR description and in CHANGELOG.md. Exclude purely internal or private changes (e.g., code not exported from package entry points or explicitly marked as internal).

Please share your feedback with us on this Discord post.


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

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

Copy link
Contributor

@coderabbitai coderabbitai bot left a comment

Choose a reason for hiding this comment

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

Actionable comments posted: 3

🧹 Nitpick comments (3)
cpp/tensorrt_llm/CMakeLists.txt (1)

297-301: Scope NOT WIN32 to Linux or add macOS handling.

NOT WIN32 also matches Apple; $ORIGIN does not apply on macOS (@loader_path is needed). If macOS isn’t supported, restrict the condition; otherwise add an Apple branch.

Suggested tweak:

-if(NOT WIN32)
+if(UNIX AND NOT APPLE)
   ...
-endif()
+elseif(APPLE)
+  # set_target_properties(${SHARED_TARGET} PROPERTIES BUILD_RPATH "@loader_path;@loader_path/../../../..")
+endif()
cpp/tensorrt_llm/pybind/CMakeLists.txt (1)

58-70: Guard for Apple if needed.

Same rationale as core: NOT WIN32 matches Apple; $ORIGIN is Linux-specific.

-if(NOT WIN32)
+if(UNIX AND NOT APPLE)
   ...
-endif()
+elseif(APPLE)
+  # set_target_properties(${TRTLLM_PYBIND_MODULE} PROPERTIES BUILD_RPATH "@loader_path/libs;@loader_path/../../..")
+endif()
cpp/tensorrt_llm/nanobind/CMakeLists.txt (1)

56-67: Consider Linux-only guard or add macOS branch.

Mirror the platform handling recommendation from pybind/core.

📜 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 c6ab207 and bd31f93.

📒 Files selected for processing (3)
  • cpp/tensorrt_llm/CMakeLists.txt (1 hunks)
  • cpp/tensorrt_llm/nanobind/CMakeLists.txt (1 hunks)
  • cpp/tensorrt_llm/pybind/CMakeLists.txt (1 hunks)
🧰 Additional context used
🧠 Learnings (2)
📓 Common learnings
Learnt from: tongyuantongyu
PR: NVIDIA/TensorRT-LLM#6984
File: cpp/tensorrt_llm/CMakeLists.txt:297-299
Timestamp: 2025-08-18T09:08:07.687Z
Learning: In the TensorRT-LLM project, artifacts are manually copied rather than installed via `cmake --install`, so INSTALL_RPATH properties are not needed - only BUILD_RPATH affects the final artifacts.
📚 Learning: 2025-08-18T09:08:07.687Z
Learnt from: tongyuantongyu
PR: NVIDIA/TensorRT-LLM#6984
File: cpp/tensorrt_llm/CMakeLists.txt:297-299
Timestamp: 2025-08-18T09:08:07.687Z
Learning: In the TensorRT-LLM project, artifacts are manually copied rather than installed via `cmake --install`, so INSTALL_RPATH properties are not needed - only BUILD_RPATH affects the final artifacts.

Applied to files:

  • cpp/tensorrt_llm/CMakeLists.txt
  • cpp/tensorrt_llm/pybind/CMakeLists.txt
  • cpp/tensorrt_llm/nanobind/CMakeLists.txt
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
  • GitHub Check: Pre-commit Check
🔇 Additional comments (3)
cpp/tensorrt_llm/CMakeLists.txt (1)

297-301: Good move to BUILD_RPATH; matches how artifacts are copied in this repo.

Using BUILD_RPATH (and not INSTALL_RPATH) aligns with the project’s packaging flow where artifacts are manually copied, not installed.

cpp/tensorrt_llm/nanobind/CMakeLists.txt (1)

57-63: RPATH entries are appropriate for non-system Python.

Matches the pybind layout and covers $PREFIX/lib and Python-wheel NCCL.

cpp/tensorrt_llm/pybind/CMakeLists.txt (1)

59-66: RPATH list looks correct for non-system Python layouts — please run local verification

  • $ORIGIN/libs → tensorrt_llm/libs
  • $ORIGIN/../../.. → $PREFIX/lib
  • $ORIGIN/../nvidia/nccl/lib → site-packages/nvidia/nccl/lib

Should cover venv/Conda deployments.

Sandbox import failed: ModuleNotFoundError: No module named 'xgrammar' — runtime checks couldn't complete here. Please run the verification locally in both a venv and a Conda env and paste outputs (python version, tensorrt_llm import result, readelf RUNPATH/RPATH for the built .so, and LD_LIBRARY_PATH):

#!/bin/bash
set -euo pipefail
python -V
python -c 'import importlib; m=importlib.import_module("tensorrt_llm"); print(m, getattr(m, "__file__", "<package>"))'
# Inspect DT_RUNPATH of the built pybind module
fd -g '**/tensorrt_llm*.so' -a | while read -r so; do
  echo "== $so =="
  command -v readelf >/dev/null && readelf -d "$so" | rg -n 'RUNPATH|RPATH' -n || true
done
# Check NCCL resolution
python - <<'PY'
import os
paths = [os.path.dirname(__import__('tensorrt_llm').__file__)]
print("ORIGINs:", paths)
print("LD_LIBRARY_PATH:", os.environ.get("LD_LIBRARY_PATH"))
PY

@tongyuantongyu tongyuantongyu force-pushed the ytong/non-system-python-install branch 2 times, most recently from 49d0096 to 1e75e41 Compare September 17, 2025 01:54
@tongyuantongyu
Copy link
Member Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #18858 [ run ] triggered by Bot

@tongyuantongyu tongyuantongyu force-pushed the ytong/non-system-python-install branch from 1e75e41 to b00b1c4 Compare September 17, 2025 03:21
@tongyuantongyu
Copy link
Member Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #18878 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #18858 [ run ] completed with state ABORTED
LLM/main/L0_MergeRequest_PR #14137 (Blue Ocean) completed with status: ABORTED

@tensorrt-cicd
Copy link
Collaborator

PR_Github #18878 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #14151 completed with status: 'FAILURE'

@tongyuantongyu tongyuantongyu force-pushed the ytong/non-system-python-install branch from b00b1c4 to 20b3084 Compare September 17, 2025 10:40
@tongyuantongyu
Copy link
Member Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #18986 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #18986 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #14236 completed with status: 'FAILURE'

@tongyuantongyu
Copy link
Member Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19097 [ run ] triggered by Bot

@tongyuantongyu
Copy link
Member Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19140 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19097 [ run ] completed with state ABORTED
/LLM/main/L0_MergeRequest_PR pipeline #14326 completed with status: 'FAILURE'

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19140 [ run ] completed with state FAILURE
/LLM/main/L0_MergeRequest_PR pipeline #14364 completed with status: 'FAILURE'

@tongyuantongyu
Copy link
Member Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19144 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19144 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #14367 completed with status: 'FAILURE'

@tongyuantongyu
Copy link
Member Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19432 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #14600 completed with status: 'FAILURE'

@tongyuantongyu
Copy link
Member Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19471 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19471 [ run ] completed with state FAILURE
/LLM/main/L0_MergeRequest_PR pipeline #14640 completed with status: 'FAILURE'

@tongyuantongyu
Copy link
Member Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19645 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19645 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #14782 completed with status: 'SUCCESS'

@tongyuantongyu tongyuantongyu force-pushed the ytong/non-system-python-install branch from bffcc92 to e0a25ef Compare September 24, 2025 10:06
@tongyuantongyu
Copy link
Member Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19796 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19796 [ run ] completed with state FAILURE
/LLM/main/L0_MergeRequest_PR pipeline #14893 completed with status: 'FAILURE'

@tongyuantongyu
Copy link
Member Author

/bot run

Copy link
Collaborator

@joyang-nv joyang-nv left a comment

Choose a reason for hiding this comment

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

LGTM. Thanks for explanations.

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19904 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19904 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #14981 completed with status: 'FAILURE'

@tongyuantongyu
Copy link
Member Author

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19937 [ run ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19937 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #15007 completed with status: 'SUCCESS'
Pipeline passed with automatic retried tests. Check the rerun report for details.

@tongyuantongyu tongyuantongyu merged commit fae83c3 into NVIDIA:main Sep 26, 2025
5 checks passed
@tongyuantongyu tongyuantongyu deleted the ytong/non-system-python-install branch September 26, 2025 03:32
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

from tensorrt_llm.bindings import DataType, GptJsonConfig ImportError

3 participants