Dear @lizhensheng,
To run docker services on the Target, developers must follow the steps to rebuild the kernel and filesystem per the instructions in Compiling the Kernel (Kernel 5.15) | NVIDIA Docs
with the following modifications:
- For step 4, include
libncurses-dev
andvim
in theapt-get install
command. - For step 6, use
defconfig
. - For step 7, after starting the menuconfig with
make -C kernel O=${PWD}/out-linux menuconfig
, press ENTER to select General setup. Navigate to Namespaces support and press the space bar to enable. Press ENTER to select Namespaces support. Navigate to User namespace and press the space bar to enable. Save the configuration and exit the menuconfig. Save the menuconfig with themake -C kernel O=${PWD}/out-linux savedefconfig
command. When copying the updated defconfig, usecp ${PWD}/out-linux/defconfig kernel/arch/arm64/configs/tegra_defconfig
. - For step 11.d, replace instances of
<kernel_version>
in the code snippet with5.15.98-rt-tegra
before executing the commands. - For step 12.c.i, replace the
<fstype>
string withstandard
before saving the newupdate_rfs.CONFIG.json
file. Before running Build-FS per #12.c.i.2, execute the following snippet to remove all YAML entries for kernel modules for theupdates/dkms
path
python3 -B - << END
import yaml
with open('/drive/drive-linux/filesystem/copytarget/manifest/copytarget-kernel-modules.yaml', 'r') as f:
data = yaml.safe_load(f)
data['fileList'] = [data_ for data_ in data['fileList'] if data_['destination'].find('updates/dkms') == -1]
with open('/drive/drive-linux/filesystem/copytarget/manifest/copytarget-kernel-modules.yaml', 'w') as f:
yaml.dump(data, f)
END
Let me know if it works for you.