Building the Jetson Linux Kernel（36.3）

[Olivia798]

Black screen after kernel recompilation with new Jetson Linux 36.3 - Jetson & Embedded Systems / Jetson AGX Orin - NVIDIA Developer Forums我按照如下进行的，进行到make ARCH=arm64 LOCALVERSION=-davey-tegra -j8 Image这一步，

Under Linux_for_Tegra/source/kernel/kernel-jammy-src/:

make ARCH=arm64 LOCALVERSION=-davey-tegra defconfig
make ARCH=arm64 LOCALVERSION=-davey-tegra menuconfig
make ARCH=arm64 LOCALVERSION=-davey-tegra -j8 Image
make ARCH=arm64 LOCALVERSION=-davey-tegra -j8 modules
sudo make modules_install INSTALL_MOD_PATH=/Linux_for_Tegra/rootfs
(Manually copy Image into Linux_for_Tegra/kernel/Image and Linux_for_Tegra/rootfs/boot/Image)

Under Linux_for_Tegra/source/:
export KERNEL_HEADERS=$PWD/kernel/kernel-jammy-src
make modules
export INSTALL_MOD_PATH=Linux_for_Tegra/rootfs
sudo -E make modules_install

Under Linux_for_Tegra/:
sudo ./tools/l4t_update_initrd.sh

Then flash the device.

出现错误提示，提示为

arch/arm64/Makefile:36: Detected assembler with broken .inst; disassembly will be unreliable
  SYNC    include/config/auto.conf.cmd
arch/arm64/Makefile:36: Detected assembler with broken .inst; disassembly will be unreliable
warning: ld does not support --fix-cortex-a53-843419; kernel may be susceptible to erratum
warning: LSE atomics not supported by binutils
  CC      scripts/mod/empty.o
  CC      scripts/mod/devicetable-offsets.s
gcc: error: unrecognized command-line option ‘-mlittle-endian’
make[1]: *** [scripts/Makefile.build:295: scripts/mod/empty.o] Error 1
make[1]: *** Waiting for unfinished jobs....
gcc: error: unrecognized command-line option ‘-mlittle-endian’
make[1]: *** [scripts/Makefile.build:122: scripts/mod/devicetable-offsets.s] Error 1
make: *** [Makefile:1265: prepare0] Error 2

[DaveYYY]

你的編譯器版本不對所以有些選項會報錯

[Olivia798]

[DaneLLL]

Hi,
Please refer to steps in developer guide:

Kernel Customization — NVIDIA Jetson Linux Developer Guide 1 documentation

You may be missing this setting:

$ export CROSS_COMPILE=<toolchain-path>/bin/aarch64-buildroot-linux-gnu-

[Olivia798]

sudo ./tools/l4t_update_initrd.sh执行完， 还需要执行./apply_binaries.sh 这个吗？

[DaveYYY]

不用
不然你剛更新完的initrd image又會被BSP裡預設的那一包覆蓋過去

[Olivia798]

生成的新内核， 直接替换到设备 就生效了吧？

[DaveYYY]

kernel image, kernel modules, initrd image全部都要一起裝

[Olivia798]

Run the following commands to install the kernel and in-tree modules:

$ export INSTALL_MOD_PATH=<install-path>/Linux_for_Tegra/rootfs/
$ sudo -E make install -C kernel
$ cp kernel/kernel-jammy-src/arch/arm64/boot/Image \
  <install-path>/Linux_for_Tegra/kernel/Image

这一步也执行完成了， 接下来就是对设备重新刷机，就可以完成了？

[DaveYYY]

kernel modules有分in-tree和OOT (out-of-tree)
兩個都要做
https://docs.nvidia.com/jetson/archives/r36.3/DeveloperGuide/SD/Kernel/KernelCustomization.html#building-the-nvidia-out-of-tree-modules

[Olivia798]

执行到sudo -E make modules_install这一步，结尾如下，没出现successful，

make[1]: Leaving directory '/home/nvidia/36.3.0/Linux_for_Tegra/source/kernel/kernel-jammy-src'
if [ ! -d "/home/nvidia/36.3.0/Linux_for_Tegra/source/nvgpu" ] ; then \
        echo "Directory nvgpu is not found, exiting.."; \
        false; \
fi
================================================================================
make modules_install - nvgpu ...
================================================================================
make -j 4 ARCH=arm64 \
        -C /home/nvidia/36.3.0/Linux_for_Tegra/source/kernel/kernel-jammy-src \
        M=/home/nvidia/36.3.0/Linux_for_Tegra/source/nvgpu/drivers/gpu/nvgpu \
        CONFIG_TEGRA_OOT_MODULE=m \
        srctree.nvidia=/home/nvidia/36.3.0/Linux_for_Tegra/source/nvidia-oot \
        srctree.nvidia-oot=/home/nvidia/36.3.0/Linux_for_Tegra/source/nvidia-oot \
        srctree.nvconftest=/home/nvidia/36.3.0/Linux_for_Tegra/source/out/nvidia-conftest \
        KBUILD_EXTRA_SYMBOLS=/home/nvidia/36.3.0/Linux_for_Tegra/source/nvidia-oot/Module.symvers \
        modules_install
make[1]: Entering directory '/home/nvidia/36.3.0/Linux_for_Tegra/source/kernel/kernel-jammy-src'
  INSTALL /home/nvidia/36.3.0/Linux_for_Tegra/rootfs//lib/modules/5.15.136-tegra/updates/nvgpu.ko
  SIGN    /home/nvidia/36.3.0/Linux_for_Tegra/rootfs//lib/modules/5.15.136-tegra/updates/nvgpu.ko
  DEPMOD  /home/nvidia/36.3.0/Linux_for_Tegra/rootfs//lib/modules/5.15.136-tegra
make[1]: Leaving directory '/home/nvidia/36.3.0/Linux_for_Tegra/source/kernel/kernel-jammy-src'
================================================================================
make modules_install - nvidia-display ...
================================================================================
make -C /home/nvidia/36.3.0/Linux_for_Tegra/source/out/nvidia-linux-header \
        M=/home/nvidia/36.3.0/Linux_for_Tegra/source/nvdisplay/kernel-open modules_install
make[1]: Entering directory '/home/nvidia/36.3.0/Linux_for_Tegra/source/out/nvidia-linux-header'
  INSTALL /home/nvidia/36.3.0/Linux_for_Tegra/rootfs//lib/modules/5.15.136-tegra/updates/nvidia-drm.ko
  SIGN    /home/nvidia/36.3.0/Linux_for_Tegra/rootfs//lib/modules/5.15.136-tegra/updates/nvidia-drm.ko
  INSTALL /home/nvidia/36.3.0/Linux_for_Tegra/rootfs//lib/modules/5.15.136-tegra/updates/nvidia-modeset.ko
  SIGN    /home/nvidia/36.3.0/Linux_for_Tegra/rootfs//lib/modules/5.15.136-tegra/updates/nvidia-modeset.ko
  INSTALL /home/nvidia/36.3.0/Linux_for_Tegra/rootfs//lib/modules/5.15.136-tegra/updates/nvidia.ko
  SIGN    /home/nvidia/36.3.0/Linux_for_Tegra/rootfs//lib/modules/5.15.136-tegra/updates/nvidia.ko
  DEPMOD  /home/nvidia/36.3.0/Linux_for_Tegra/rootfs//lib/modules/5.15.136-tegra
make[1]: Leaving directory '/home/nvidia/36.3.0/Linux_for_Tegra/source/out/nvidia-linux-header'

[Olivia798]

OOT这一步也完成了， 接下来的Building the DTBs 这一步 需要做吗， 还是 可以进行刷机了？

[DaveYYY]

你沒有改device tree就不用

[Olivia798]

我再问一下，什么情况下，替换Image就可以，不用重新刷机？ 还是 都得需要重新刷机才能完成？

[DaveYYY]

基本上都要重新燒板子 不然drivers和kernel image版本對不上 就算能開機也是什麼都不能用