CMAKE error when trying to build project inside of a docker container: cant find cudart lib

Question

Using an nvidia jetson tx2 running ubuntu18.04 with docker,nvidia-docker2 and l4t-cuda installed on the host system. Main Error when compiling:

CMake Error at /usr/local/lib/python3.6/dist-packages/cmake/data/share/cmake-3.25/Modules/FindPackageHandleStandardArgs.cmake:230 (message):
  Could NOT find CUDA (missing: CUDA_CUDART_LIBRARY) (found suitable version
  "10.2", minimum required is "10.2")
Call Stack (most recent call first):
  /usr/local/lib/python3.6/dist-packages/cmake/data/share/cmake-3.25/Modules/FindPackageHandleStandardArgs.cmake:600 (_FPHSA_FAILURE_MESSAGE)
  /usr/local/lib/python3.6/dist-packages/cmake/data/share/cmake-3.25/Modules/FindCUDA.cmake:1266 (find_package_handle_standard_args)
  CMakeLists.txt:17 (find_package)

CMakeLists.txt:

cmake_minimum_required (VERSION 3.5)

project(vision)
enable_testing()

# Variables scopes follow standard rules
# Variables defined here will carry over to its children, ergo subdirectories

# Setup ZED libs
find_package(ZED 3 REQUIRED)
include_directories(${ZED_INCLUDE_DIRS})
link_directories(${ZED_LIBRARY_DIR})

# Setup CUDA libs for zed and ai modules
find_package(CUDA ${ZED_CUDA_VERSION} REQUIRED)
include_directories(${CUDA_INCLUDE_DIRS})
link_directories(${CUDA_LIBRARY_DIRS})

# Setup OpenCV libs
find_package(OpenCV REQUIRED)
include_directories(${OpenCV_INLCUDE_DIRS})
# Check if OpenMP is installed 
find_package(OpenMP)
checkPackage("OpenMP" "OpenMP not found, please install it to improve performances: 'sudo apt install libomp-dev'")

# TensorRT
set(TENSORRT_ROOT /usr/src/tensorrt/)
find_path(TENSORRT_INCLUDE_DIR NvInfer.h
        HINTS ${TENSORRT_ROOT} PATH_SUFFIXES include/)
message(STATUS "Found TensorRT headers at ${TENSORRT_INCLUDE_DIR}")

set(MODEL_INCLUDE ../code/includes)
set(MODEL_LIB_DIR libs)
set(YAML_INCLUDE ../depends/yaml-cpp/include)
set(YAML_LIB_DIR ../depends/yaml-cpp/libs)

include_directories(${MODEL_INCLUDE} ${YAML_INCLUDE})
link_directories(${MODEL_LIB_DIR} ${YAML_LIB_DIR})

# Setup Darknet libs
#find_library(DARKNET_LIBRARY NAMES dark libdark.so libdarknet.so)
#find_package(dark REQUIRED)

# Setup HTTP libs
find_package(httplib REQUIRED)
find_package(nlohmann_json 3.2.0 REQUIRED)

# System libs
SET(SPECIAL_OS_LIBS "pthread")
link_libraries(stdc++fs)

# Optional definitions
add_definitions(-std=c++17 -g -O3)

# Add sub directories
add_subdirectory(zed_module)
add_subdirectory(ai_module)
add_subdirectory(http_api_module)
add_subdirectory(executable_module)

option(RUN_TESTS "Build the tests" off)

if (RUN_TESTS OR CMAKE_BUILD_TYPE MATCHES Debug)
    add_subdirectory(test)

steps that fail in Dockerfile using image stereolabs/zed:3.7-devel-jetson-jp4.6:


WORKDIR /opt

RUN git clone https://github.com/Cruiz102/Vision-Module
WORKDIR /opt/Vision-Module 
RUN mkdir build-debug && cd build-debug
RUN pwd
WORKDIR /opt/Vision-Module/build-debug
RUN cmake -DCMAKE_BUILD_TYPE=Release ..

contents of /etc/docker/daemon.json

{
    "runtimes": {
        "nvidia": {
            "path": "/usr/bin/nvidia-container-runtime",
            "runtimeArgs": []
        }
    },
    "default-runtime": "nvidia"
}

Using the jetson, I've tried using flags to set the toolkit dir as well as editing daemon.json reinstalling dependencies, changing docker images, installing and reinstalling cudaart on host, changing flags and finishing the build in interactive mode However I always get the same error.

score 1 · Answer 1 · answered Dec 14 '22 at 14:58

1

I have looked into Docker some time ago, so I'm not an expert, put as far as I remember, Docker and Docker containers are like virtual machines. It doesn't matter whether your pc has any cuda support or whether the libraries are installed. They are not part of your Docker VM. And since Docker runs without GUI this stuff will not be installed right away.

I don't see any code to install it in your container. Are you using one that has cuda support? If not you need to install it using your Dockerfile, not on your host

answered Dec 14 '22 at 14:58

Thorsten Schmitz

542
7
19

Hi, the image stereolabs/zed:3.7-devel-jetson-jp4.6 already comes with cuda support oob I confirmed that since inside the container I can find /usr/local/cuda-10.2 dir. My main goal is to get the source files from the project to compile with said cuda installation. Another problem is that the host has to have nvidia docker installed in order to bassically pass through the gpu to the vm, I followed https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html – atlas7 Dec 14 '22 at 17:25
Ok. Did you try adding the folder manually to CMakes search paths? – Thorsten Schmitz Dec 14 '22 at 17:35
Hey sorry for taking so long to respond I tried adding another include_directories with the path to /usr/local/cuda-10.2/lib64 to no avail still get the same error. – atlas7 Dec 16 '22 at 22:18
I didn't mean include_directories. If you look at the find_package doc https://cmake.org/cmake/help/latest/command/find_package.html you see there is a variable specifically for paths to searchs in addition to the standard paths https://cmake.org/cmake/help/latest/variable/CMAKE_MODULE_PATH.html#variable:CMAKE_MODULE_PATH – Thorsten Schmitz Dec 17 '22 at 08:31

CMAKE error when trying to build project inside of a docker container: cant find cudart lib

1 Answers1