Tx2nx cross compile

/opt/gcc-linaro-7.5.0-2019.12-x86_64_aarch64-linux-gnu/bin/…/lib/gcc/aarch64-linux-gnu/7.5.0/…/…/…/…/aarch64-linux-gnu/bin/ld: 当搜索用于 /media/alfred/nfs///usr/local/cuda/lib64/libnvjpeg.so 时跳过不兼容的 -lnvjpeg
/opt/gcc-linaro-7.5.0-2019.12-x86_64_aarch64-linux-gnu/bin/…/lib/gcc/aarch64-linux-gnu/7.5.0/…/…/…/…/aarch64-linux-gnu/bin/ld: 当搜索用于 /media/alfred/nfs///usr/local/cuda/lib64/libcudart.so 时跳过不兼容的 -lcudart
/opt/gcc-linaro-7.5.0-2019.12-x86_64_aarch64-linux-gnu/bin/…/lib/gcc/aarch64-linux-gnu/7.5.0/…/…/…/…/aarch64-linux-gnu/bin/ld: 找不到 -lcudart

It’s the error at last. I have mounted the tx2nx:/.
make TARGET_ROOTFS=/media/alfred/nfs/ CROSS_COMPILE=aarch64-linux-gnu- TEGRA_ARMABI=aarch64-linux-gnu
What’s the matter?


To compile a Jetson binary on the host, you will need a CUDA cross-compile package.

Have you installed it?
If not, you can find the package from the SDK manager directly.


How to install CUDA cross-compile package? It seems that all components have been installed.


YES. It seem that all the packages are well-installed in your environment.

Could you share more about your use case?
Do you want to cross-compile a Jetson NX binary on an x86 host?
Or cross-compile a NX binary on another Jetson board, ex. TX2?


Please try to generate img.raw, mount it to your host PC and try again.

Please refer to command of running flash.sh to generate img.raw in
Jetson-multimedia-api sample cross compile instructions - #9 by DaneLLL

@DaneLLL @AastaLLL
I want to cross-compile a TX2 NX binary on an x86 host
I mount it to the board.mount -o nolock tx2nxip:/ /nfs
The tx2nx works well. I can find the directory on the board.

The guidance is in document:

So it does not work if you follow the steps? We have tried and it works fine. It seems like you don’s put the device into recovery mode and generate img.raw. Please flash system image and SDK components to TX2NX developer kit and close the image.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.

Hi AlfredNG,
Please followed DaneLLL’s guide of Cross-platform support, or just simply comment out cuda related , if you actually not use them during compiling.