VEYE CS Camera source for Jetson/zh
英伟达Jetson平台上VEYE摄像头驱动源码编译指南
1 概述
NVIDIA为每个Jetpack版本都发布了Linux开发文档。比如L4T 35.1和32.5。本章节是以官方的Kernel Customization章节为基础的,不够详尽之处,请参考官方文档。
我们使用安装Ubuntu的host pc和Jetson板子作开发环境。
- Jetpack4.x
驱动与linux内核编译到一起,在host pc上进行。dtb的编译也在host pc进行。
- Jetpack5.x
驱动独立于linux内核,编译为modules,在Jetson板上进行。dtb的编译在host pc进行。
1.1 版本对应关系
- Jetpack4.2.2,L4T版本r32.2.1
- Jetpack4.3,L4T版本r32.3.1
- Jetpack4.4,L4T版本r32.4.3
- Jetpack4.4.1,L4T版本r32.4.4
- Jetpack4.5,L4T版本r32.5
- Jetpack4.5.1,L4T版本r32.5.1
- Jetpack4.6,L4T版本r32.6.1
- Jetpack4.6.1,L4T版本r32.7.1
- Jetpack4.6.2,L4T版本r32.7.2
- Jetpack5.0.1DP,L4T版本r34.1.1
- Jetpack5.0.2, L4T版本r35.1
- Jetpack5.1, L4T版本r35.2.1
1.2 查看当前L4T版本
首先查看当前L4T版本,尽量使用同一版本进行替换。
cat /etc/nv_tegra_release
如显示:
# R32 (release), REVISION: 4.3......
则表示当前版本为32.4.3。
2 HOST开发环境准备
以下操作均为在HOST PC进行。
2.1 编译工具链
- Jetpack4.x
请参考这个链接安装编译工具链。
- Jetpack5.x
请参考这个链接安装编译工具链。
2.2 L4T源码和Rootfs
首先使用SDK Manager在HOST PC上面下载对应的Linux_for_Tegra开发包。
可以通过SDK Manager或者直接下载两种方式来获得源码。具体参考官方文档,此处不再赘述。我推荐直接下载的方式。
使用SDK Manager时同步代码采用如下指令:
- Jetpack4.2.2
./source_sync.sh -t tegra-l4t-r32.2.1
- Jetpack4.3
./source_sync.sh -t tegra-l4t-r32.3.1
- Jetpack4.4
./source_sync.sh -t tegra-l4t-r32.4.3
- Jetpack4.4.1
./source_sync.sh -t tegra-l4t-r32.4.4
- Jetpack4.5
./source_sync.sh -t tegra-l4t-r32.5
- Jetpack4.5.1
./source_sync.sh -t tegra-l4t-r32.5.1
- Jetpack4.6
./source_sync.sh -t tegra-l4t-r32.6.1
- Jetpack4.6.1
./source_sync.sh -t tegra-l4t-r32.7.1
- Jetpack4.6.2
- Jetpack5.0.1 DP
./source_sync.sh -t jetson_34.1.1
- Jetpack5.0.2
./source_sync.sh -t jetson_35.1
- Jetpack5.1
./source_sync.sh -t jetson_35.2.1
2.3 配置开发环境
以下步骤假设SDK安装目录为<TOPDIR>,源代码在$L4T_DIR/sources目录。
export TOP_DIR=<absolute path to top dir accroding to your jetpack version and board name;for example /home/xumm/nvidia/nvidia_sdk/JetPack_4.5_Linux_JETSON_XAVIER_NX_DEVKIT/>
- 通用部分
export L4T_DIR=$TOP_DIR/Linux_for_Tegra
export LOCALVERSION=-tegra
export LDK_ROOTFS_DIR=$TOP_DIR/Linux_for_Tegra/rootfs
export ARCH=arm64
export TEGRA_KERNEL_OUT=$L4T_DIR/sources/kernel/out_kernel
export KERNEL_PATH=$L4T_DIR/sources/kernel/out_kernel
export NVIDIA_PATH=$L4T_DIR/sources/kernel/nvidia
export NANO_DTS_PATH=$L4T_DIR/sources/hardware/nvidia/platform/t210/
export TX2_DTS_PATH=$L4T_DIR/sources/hardware/nvidia/platform/t18x
export XAVIER_DTS_PATH=$L4T_DIR/sources/hardware/nvidia/platform/t19x
- Jetpack 4.x
export CROSS_COMPILE=aarch64-linux-gnu-
export CROSS32CC=arm-linux-gnueabihf-gcc
export COMMON_DTS_PATH=$TEGRA_KERNEL_OUT/arch/arm64/boot/dts/
- Jetpack 5.x
export CROSS_COMPILE_AARCH64_PATH=$HOME/l4t-gcc
export CROSS_COMPILE_AARCH64=$HOME/l4t-gcc/bin/aarch64-buildroot-linux-gnu-
export CROSS_COMPILE=$CROSS_COMPILE_AARCH64
export COMMON_DTS_PATH=$TEGRA_KERNEL_OUT/arch/arm64/boot/dts/nvidia
3 下载我们的bsp包
cd $L4T_DIR
git clone https://github.com/veyeimaging/nvidia_jetson_veye_bsp.git
export RELEASE_PACK_DIR=$L4T_DIR/nvidia_jetson_veye_bsp
4 编译kernel
对于Jetpack4.x版本以及在Jetpack5.x版本下使用MV系列相机的,需要编译kernel。
4.1 MV系列相机的patch包
对于MV系列相机,除了在linux系统中增加了摄像头的驱动,我们还增加了一个内核补丁——veye_mv_l4t_[version].patch。
这个patch具备两个功能:
- 增加了黑白相机的Y10和Y12两种数据格式的支持。
- 增加了trigger模式的支持。
这个补丁包对于VEYE系列和CS系列相机,可以不使用。
cd $L4T_DIR/sources/kernel/
patch -p1 < [path to]veye_mv_l4t_[version].patch
4.2 对于Jetpack4.x,将camera驱动编译进Image
4.2.1 patch代码
cp $RELEASE_PACK_DIR/drivers_source/cam_drv_src/* $NVIDIA_PATH/drivers/media/i2c/
cp $RELEASE_PACK_DIR/drivers_source/kernel_veyecam_config_<l4t_version> $L4T_DIR/sources/kernel/kernel-4.9/arch/arm64/configs/tegra_veyecam_defconfig
4.2.2 修改Kconfig和Makefile
cp $RELEASE_PACK_DIR/drivers_source/cam_drv_src/Kconfig_<ver> $NVIDIA_PATH/drivers/media/i2c/
cp $RELEASE_PACK_DIR/drivers_source/cam_drv_src/Makefile_<ver> $NVIDIA_PATH/drivers/media/i2c/
4.2.3 编译
cd $L4T_DIR/sources/kernel/kernel-4.9/
make ARCH=arm64 O=$TEGRA_KERNEL_OUT tegra_veyecam_defconfig
make ARCH=arm64 O=$TEGRA_KERNEL_OUT Image -j4
4.3 对于Jetpack5.x,无需将camera驱动编译进Image
4.3.1 编译
cd $L4T_DIR/sources/
./nvbuild.sh -o $TEGRA_KERNEL_OUT
如果没有找到此脚本:
nvbuild.sh
is only included in the kernel_src.tbz2
tar file that is part of public_sources.tbz2
. If the kernel sources were synced from the Git server, you need to manually copy nvbuild.sh
from the tar file.
4.4 编译完成
编译完成的Image文件在:
$TEGRA_KERNEL_OUT/arch/arm64/boot/Image
可以用于整体烧写,也可以用于动态升级。
以下用于整体烧写
sudo cp $TEGRA_KERNEL_OUT/arch/arm64/boot/Image $L4T_DIR/kernel/ -f
- 对于xavier和Orin,需要进行签名
cd $L4T_DIR
./l4t_sign_image.sh --file kernel/Image --chip 0x19
5 编译modules(Jetpack5.x)
Jetpack5.x以后的版本,可以直接以modules方式编译驱动,得到ko文件。推荐直接在Jetson Board上面编译驱动。
cd $RELEASE_PACK_DIR/drivers_source/cam_drv_src/
make
6 编译DTS
首先需要patch代码。
6.1 Nano
cp $RELEASE_PACK_DIR/Nano/JetPack_<ver>_Linux_JETSON_NANO_DEVKIT/dts\ dtb/common/t210/* -r $NANO_DTS_PATH/
For <camera model>
cp $RELEASE_PACK_DIR/Nano/JetPack_<ver>_Linux_JETSON_NANO_DEVKIT/dts\ dtb/<camera model>/tegra210-porg-plugin-manager.dtsi -r $NANO_DTS_PATH/porg/kernel-dts/porg-plugin-manager
6.2 TX2
cp $RELEASE_PACK_DIR/TX2/JetPack_<ver>_Linux_JETSON_TX2/dts\ dtb/common/t18x/* -r $TX2_DTS_PATH/
For <camera model>
cp $RELEASE_PACK_DIR/TX2/JetPack_<ver>_Linux_JETSON_TX2/dts\ dtb/<camera model>/tegra186-quill-p3310-1000-a00-00-base.dts $TX2_DTS_PATH/quill/kernel-dts/
cp $RELEASE_PACK_DIR/TX2/JetPack_<ver>_Linux_JETSON_TX2/dts\ dtb/<camera model>/tegra186-p3636-0001-p3509-0000-a01.dts $TX2_DTS_PATH/lanai/kernel-dts
6.3 AGX Xavier
cp $RELEASE_PACK_DIR/AGX-Xaviar/JetPack_<ver>_Linux_JETSON_AGX_XAVIER/dts\ dtb/common/t19x/* -r $XAVIER_DTS_PATH/
For <camera model>
cp $RELEASE_PACK_DIR/AGX-Xaviar/JetPack_<ver>_Linux_JETSON_AGX_XAVIER/dts\ dtb/<camera model>/tegra194-p2888-0001-p2822-0000.dts $XAVIER_DTS_PATH/galen/kernel-dts/
6.4 Xavier NX
cp $RELEASE_PACK_DIR/Xavier-NX/JetPack_<ver>_Linux_JETSON_XAVIER_NX_DEVKIT/dts\ dtb/common/t19x/* -r $XAVIER_DTS_PATH/
For <camera model>
cp $RELEASE_PACK_DIR/Xavier-NX/JetPack_<ver>_Linux_JETSON_XAVIER_NX_DEVKIT/dts\ dtb/<camera model>/tegra194-p3509-0000-a00.dtsi $XAVIER_DTS_PATH/jakku/kernel-dts/common/
6.5 编译
- For Jetpack4.X
export COMMON_DTS_PATH=$TEGRA_KERNEL_OUT/arch/arm64/boot/dts
cd $L4T_DIR/sources/kernel/kernel-4.9/
- For Jetpack5.X
export COMMON_DTS_PATH=$TEGRA_KERNEL_OUT/arch/arm64/boot/dts/nvidia
cd $L4T_DIR/sources/kernel/kernel-5.10/
make ARCH=arm64 O=$TEGRA_KERNEL_OUT dtbs
You can get the compiled DTB file in the following path.
6.5.1 Nano
$COMMON_DTS_PATH/tegra210-p3448-0000-p3449-0000-a02.dtb
$COMMON_DTS_PATH/tegra210-p3448-0000-p3449-0000-b00.dtb
$COMMON_DTS_PATH/tegra210-p3448-0003-p3542-0000.dtb
6.5.2 TX2
$COMMON_DTS_PATH/tegra186-quill-p3310-1000-c03-00-base.dtb
6.5.3 TX2 NX(p3509-0000 carrier board)
$COMMON_DTS_PATH/tegra186-p3636-0001-p3509-0000-a01.dtb
6.5.4 AGX XAVIER
$COMMON_DTS_PATH/tegra194-p2888-0001-p2822-0000.dtb
6.5.5 XAVIER NX
$COMMON_DTS_PATH/tegra194-p3668-all-p3509-0000.dtb
6.5.6 AGX Orin
$COMMON_DTS_PATH/tegra234-p3701-0000-p3737-0000.dtb