Skip to content

fix: Ensure that Python stub generation works against libnvidia-ml stubs #6188

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
Open

fix: Ensure that Python stub generation works against libnvidia-ml stubs #6188

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a6f21ae
fix: Ensure that Python stub generation works against libnvidia-ml stubs
MartinMarciniszyn Jul 18, 2025
44c545c
Update docker images.
MartinMarciniszyn Jul 18, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 7 additions & 2 deletions docker/Dockerfile.multi
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,8 +71,9 @@ RUN bash ./install_pytorch.sh $TORCH_INSTALL_TYPE && rm install_pytorch.sh
ENV PYTORCH_CUDA_ALLOC_CONF="garbage_collection_threshold:0.99999"

# Install OpenCV with FFMPEG support
RUN pip3 uninstall -y opencv && rm -rf /usr/local/lib/python3*/dist-packages/cv2/
RUN pip3 install opencv-python-headless --force-reinstall --no-deps --no-cache-dir
RUN pip3 uninstall -y opencv && \
rm -rf /usr/local/lib/python3*/dist-packages/cv2/ && \
pip3 install opencv-python-headless --force-reinstall --no-deps --no-cache-dir

# WARs against security issues inherited from pytorch:25.04
# * https://github.com/advisories/GHSA-vqfr-h8mv-ghfj
Expand All @@ -87,6 +88,10 @@ RUN pip3 install --upgrade --no-cache-dir \
"protobuf>=4.25.8" \
"jupyter-core>=5.8.1"

# WAR: build containers only contain stub version of libnvidia-ml.so and not the real version.
# Need to create symbolic link to prevent errors for stub generation.
RUN ln -s libnvidia-ml.so $(find /usr -name *libnvidia-ml.so).1

FROM ${TRITON_IMAGE}:${TRITON_BASE_TAG} AS triton

FROM devel AS tritondevel
Expand Down
8 changes: 4 additions & 4 deletions jenkins/current_image_tags.properties
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,7 @@
#
# NB: Typically, the suffix indicates the PR whose CI pipeline generated the images. In case that
# images are adopted from PostMerge pipelines, the abbreviated commit hash is used instead.
LLM_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:pytorch-25.05-py3-x86_64-ubuntu24.04-trt10.11.0.33-skip-tritondevel-202507162011-ec3ebae
LLM_SBSA_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:pytorch-25.05-py3-aarch64-ubuntu24.04-trt10.11.0.33-skip-tritondevel-202507162011-ec3ebae
LLM_ROCKYLINUX8_PY310_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:cuda-12.9.0-devel-rocky8-x86_64-rocky8-py310-trt10.11.0.33-skip-tritondevel-202507162011-ec3ebae
LLM_ROCKYLINUX8_PY312_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:cuda-12.9.0-devel-rocky8-x86_64-rocky8-py312-trt10.11.0.33-skip-tritondevel-202507162011-ec3ebae
LLM_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:pytorch-25.05-py3-x86_64-ubuntu24.04-trt10.11.0.33-skip-tritondevel-202507181716-6188
LLM_SBSA_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:pytorch-25.05-py3-aarch64-ubuntu24.04-trt10.11.0.33-skip-tritondevel-202507181716-6188
LLM_ROCKYLINUX8_PY310_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:cuda-12.9.0-devel-rocky8-x86_64-rocky8-py310-trt10.11.0.33-skip-tritondevel-202507181716-6188
LLM_ROCKYLINUX8_PY312_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:cuda-12.9.0-devel-rocky8-x86_64-rocky8-py312-trt10.11.0.33-skip-tritondevel-202507181716-6188
19 changes: 0 additions & 19 deletions scripts/build_wheel.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -723,25 +723,6 @@ def get_binding_lib(subdirectory, name):
if 'LD_LIBRARY_PATH' in env_ld:
new_library_path += f":{env_ld['LD_LIBRARY_PATH']}"

result = build_run("find /usr -name *libnvidia-ml.so*",
capture_output=True,
text=True)
assert result.returncode == 0, f"Failed to run find *libnvidia-ml.so*: {result.stderr}"

# Build containers only contain stub version of libnvidia-ml.so and not the real version.
# If real version not in system, we need to create symbolic link to stub version to prevent import errors.
if "libnvidia-ml.so.1" not in result.stdout:
if "libnvidia-ml.so" in result.stdout:
line = result.stdout.splitlines()[0]
path = os.path.dirname(line)
new_library_path += f":{path}"
Copy link
Collaborator

@jiaganc jiaganc Jul 21, 2025 •
edited
Loading

Choose a reason for hiding this comment

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

This line adds the stub lib path to LD_LIBRARY_PATH. Removing this causes the build error. I think we need to add code like:

new_library_path += ":/usr/local/cuda/lib64/stub"

build_run(f"ln -s {line} {path}/libnvidia-ml.so.1")
else:
print(
f"Failed to find libnvidia-ml.so: {result.stderr}",
file=sys.stderr)
exit(1)

env_ld["LD_LIBRARY_PATH"] = new_library_path
if binding_type == "nanobind":
build_run(
Expand Down