diff --git a/build/build_img.sh b/build/build_img.sh
index 391efa28..02f77638 100644
--- a/build/build_img.sh
+++ b/build/build_img.sh
@@ -19,348 +19,888 @@
 #    along with this program.  If not, see <https://www.gnu.org/licenses/>.  #
 #                                                                            #
 # ========================================================================== #
+#!/bin/bash
 
-SRCPATH=/mnt/nas/src
-BOOTFS=/tmp/bootfs
-ROOTFS=/tmp/rootfs
-OUTPUTDIR=/mnt/nas/src/output
-LOOPDEV=/dev/loop10
-DATE=240315
+# --- 配置 ---
+# 允许通过环境变量覆盖默认路径
+SRCPATH="${SRCPATH:-/mnt/nfs/lfs/src}"
+BOOTFS="${BOOTFS:-/tmp/bootfs}"
+ROOTFS="${ROOTFS:-/tmp/rootfs}"
+OUTPUTDIR="${OUTPUTDIR:-/mnt/nfs/lfs/src/output}"
+TMPDIR="${TMPDIR:-$SRCPATH/tmp}"
+DATE=$(date +%y%m%d)
 export LC_ALL=C
 
+# 全局变量
+LOOPDEV=""
+ROOTFS_MOUNTED=0
+BOOTFS_MOUNTED=0
+PROC_MOUNTED=0
+SYS_MOUNTED=0
+DEV_MOUNTED=0
+DOCKER_CONTAINER_NAME="to_build_rootfs_$$" # 使用PID防止并发冲突
+#PREBUILT_DIR="$TMPDIR/prebuilt_binaries"
+PREBUILT_DIR="/tmp/prebuilt_binaries"
+
+# --- 清理函数 ---
+cleanup() {
+    echo "信息：执行清理操作..."
+    # 尝试卸载 chroot 环境下的挂载点
+    if [[ "$DEV_MOUNTED" -eq 1 ]]; then
+        echo "信息：卸载 $ROOTFS/dev ..."
+        sudo umount "$ROOTFS/dev" || echo "警告：卸载 $ROOTFS/dev 失败，可能已被卸载"
+        DEV_MOUNTED=0
+    fi
+    if [[ "$SYS_MOUNTED" -eq 1 ]]; then
+        echo "信息：卸载 $ROOTFS/sys ..."
+        sudo umount "$ROOTFS/sys" || echo "警告：卸载 $ROOTFS/sys 失败，可能已被卸载"
+        SYS_MOUNTED=0
+    fi
+    if [[ "$PROC_MOUNTED" -eq 1 ]]; then
+        echo "信息：卸载 $ROOTFS/proc ..."
+        sudo umount "$ROOTFS/proc" || echo "警告：卸载 $ROOTFS/proc 失败，可能已被卸载"
+        PROC_MOUNTED=0
+    fi
+
+    # 尝试卸载主根文件系统
+    if [[ "$ROOTFS_MOUNTED" -eq 1 && -d "$ROOTFS" ]]; then
+        echo "信息：卸载 $ROOTFS ..."
+        sudo umount "$ROOTFS" || sudo umount -l "$ROOTFS" || echo "警告：卸载 $ROOTFS 失败"
+        ROOTFS_MOUNTED=0
+    fi
+     # 尝试卸载引导文件系统 (如果使用)
+    if [[ "$BOOTFS_MOUNTED" -eq 1 && -d "$BOOTFS" ]]; then
+        echo "信息：卸载 $BOOTFS ..."
+        sudo umount "$BOOTFS" || sudo umount -l "$BOOTFS" || echo "警告：卸载 $BOOTFS 失败"
+        BOOTFS_MOUNTED=0
+    fi
+
+    # 尝试分离 loop 设备
+    if [[ -n "$LOOPDEV" && -b "$LOOPDEV" ]]; then
+        echo "信息：尝试 zerofree $LOOPDEV ..."
+        sudo zerofree "$LOOPDEV" || echo "警告：zerofree $LOOPDEV 失败，可能文件系统不支持或未干净卸载"
+        echo "信息：分离 loop 设备 $LOOPDEV ..."
+        sudo losetup -d "$LOOPDEV" || echo "警告：分离 $LOOPDEV 失败"
+        LOOPDEV=""
+    fi
+
+    # 尝试删除 Docker 容器
+    echo "信息：检查并删除 Docker 容器 $DOCKER_CONTAINER_NAME ..."
+    if sudo docker ps -a --format '{{.Names}}' | grep -q "^${DOCKER_CONTAINER_NAME}$"; then
+        sudo docker rm -f "$DOCKER_CONTAINER_NAME" || echo "警告：删除 Docker 容器 $DOCKER_CONTAINER_NAME 失败"
+    else
+        echo "信息：Docker 容器 $DOCKER_CONTAINER_NAME 不存在或已被删除。"
+    fi
+
+    # 清理临时目录和挂载点目录
+    echo "信息：清理临时文件和目录..."
+    sudo rm -rf "$PREBUILT_DIR"
+    # 注意：不自动删除 $TMPDIR/rootfs.img 等原始或处理中的镜像文件，由各函数管理
+    # 只删除挂载点目录本身
+    if [[ -d "$ROOTFS" ]]; then
+        sudo rmdir "$ROOTFS" || echo "警告：删除目录 $ROOTFS 失败，可能非空"
+    fi
+     if [[ -d "$BOOTFS" ]]; then
+        sudo rmdir "$BOOTFS" || echo "警告：删除目录 $BOOTFS 失败，可能非空"
+    fi
+
+    echo "信息：清理完成。"
+}
+
+# --- 注册清理函数 ---
+# 在脚本退出、收到错误信号、中断信号、终止信号时执行 cleanup
+trap cleanup EXIT ERR INT TERM
+
+# --- 辅助函数 ---
+
+# 查找并设置一个可用的 loop 设备
+find_loop_device() {
+    echo "信息：查找可用的 loop 设备..."
+    # 只使用 --find 来获取设备名
+    LOOPDEV=$(sudo losetup --find)
+    if [[ -z "$LOOPDEV" || ! -e "$LOOPDEV" ]]; then # 检查设备是否存在，而不是 -b (因为此时还未关联)
+        echo "错误：无法找到可用的 loop 设备。请确保 'loop' 内核模块已加载且有可用设备。" >&2
+        # 尝试加载模块以防万一
+        sudo modprobe loop || echo "警告：尝试加载 loop 模块失败。"
+        # 再次尝试查找
+        LOOPDEV=$(sudo losetup --find)
+         if [[ -z "$LOOPDEV" || ! -e "$LOOPDEV" ]]; then
+            echo "错误：再次尝试后仍无法找到可用的 loop 设备。" >&2
+            exit 1
+         fi
+    fi
+    echo "信息：找到可用 loop 设备名：$LOOPDEV"
+}
+
+# 检查并创建目录
+ensure_dir() {
+    if [[ ! -d "$1" ]]; then
+        echo "信息：创建目录 $1 ..."
+        sudo mkdir -p "$1" || { echo "错误：创建目录 $1 失败" >&2; exit 1; }
+    fi
+}
+
+# 执行 chroot 命令
+run_in_chroot() {
+    echo "信息：在 chroot 环境 ($ROOTFS) 中执行命令..."
+    sudo chroot --userspec "root:root" "$ROOTFS" bash -ec "$1" || { echo "错误：在 chroot 环境中执行命令失败" >&2; exit 1; }
+    echo "信息：chroot 命令执行完成。"
+}
+
+
+# --- 核心功能函数 ---
+
 write_meta() {
-    sudo chroot --userspec "root:root" $ROOTFS bash -c "sed -i 's/localhost.localdomain/$1/g' /etc/kvmd/meta.yaml"
+    local hostname="$1"
+    echo "信息：在 chroot 环境中设置主机名/元数据为 $hostname ..."
+    run_in_chroot "sed -i 's/localhost.localdomain/$hostname/g' /etc/kvmd/meta.yaml"
 }
 
 mount_rootfs() {
-    mkdir $ROOTFS $SRCPATH/tmp/rootfs
-    sudo mount $LOOPDEV  $ROOTFS || exit -1
-    sudo mount -t proc proc $ROOTFS/proc || exit -1
-    sudo mount -t sysfs sys $ROOTFS/sys || exit -1
-    sudo mount -o bind /dev $ROOTFS/dev || exit -1
+    echo "信息：挂载根文件系统到 $ROOTFS ..."
+    ensure_dir "$ROOTFS"
+    sudo mount "$LOOPDEV" "$ROOTFS" || { echo "错误：挂载 $LOOPDEV 到 $ROOTFS 失败" >&2; exit 1; }
+    ROOTFS_MOUNTED=1
+
+    echo "信息：挂载 proc, sys, dev 到 chroot 环境..."
+    ensure_dir "$ROOTFS/proc"
+    sudo mount -t proc proc "$ROOTFS/proc" || { echo "错误：挂载 proc 到 $ROOTFS/proc 失败" >&2; exit 1; }
+    PROC_MOUNTED=1
+
+    ensure_dir "$ROOTFS/sys"
+    sudo mount -t sysfs sys "$ROOTFS/sys" || { echo "错误：挂载 sys 到 $ROOTFS/sys 失败" >&2; exit 1; }
+    SYS_MOUNTED=1
+
+    ensure_dir "$ROOTFS/dev"
+    sudo mount -o bind /dev "$ROOTFS/dev" || { echo "错误：绑定挂载 /dev 到 $ROOTFS/dev 失败" >&2; exit 1; }
+    DEV_MOUNTED=1
+    echo "信息：根文件系统及虚拟文件系统挂载完成。"
 }
 
-umount_rootfs() {
-    sudo umount  $ROOTFS/sys
-    sudo umount  $ROOTFS/dev
-    sudo umount  $ROOTFS/proc
-    sudo umount $ROOTFS
-	sudo zerofree $LOOPDEV
-    sudo losetup -d $LOOPDEV
-	sudo docker rm to_build_rootfs
-	sudo rm -rf $SRCPATH/tmp/rootfs/*
+# umount_rootfs 函数不再需要，清理工作由 trap -> cleanup 完成
+
+prepare_dns_and_mirrors() {
+    echo "信息：在 chroot 环境中准备 DNS 和更换软件源..."
+    run_in_chroot "
+    mkdir -p /run/systemd/resolve/ \\
+    && touch /run/systemd/resolve/stub-resolv.conf \\
+    && printf '%s\\n' 'nameserver 1.1.1.1' 'nameserver 1.0.0.1' > /etc/resolv.conf \\
+    && echo '信息：尝试更换镜像源...' \\
+    && bash <(curl -sSL https://gitee.com/SuperManito/LinuxMirrors/raw/main/ChangeMirrors.sh) \\
+        --source mirrors.tuna.tsinghua.edu.cn --upgrade-software false --web-protocol http || echo '警告：更换镜像源脚本执行失败，可能网络不通或脚本已更改'
+    "
 }
 
-parpare_dns() {
-    sudo chroot --userspec "root:root" $ROOTFS bash -c " \
-    mkdir -p /run/systemd/resolve/ \
-    && touch /run/systemd/resolve/stub-resolv.conf \
-    && printf '%s\n' 'nameserver 1.1.1.1' 'nameserver 1.0.0.1' > /etc/resolv.conf \
-    && bash <(curl -sSL https://gitee.com/SuperManito/LinuxMirrors/raw/main/ChangeMirrors.sh) \
-        --source mirrors.tuna.tsinghua.edu.cn --updata-software false --web-protocol http "
+# (可选) 如果需要强制使用特定 Armbian 源
+delete_armbian_verify(){
+    echo "信息：在 chroot 环境中修改 Armbian 软件源..."
+    run_in_chroot "echo 'deb http://mirrors.ustc.edu.cn/armbian bullseye main bullseye-utils bullseye-desktop' > /etc/apt/sources.list.d/armbian.list"
 }
 
-delete_armbain_verify(){
-    sudo chroot --userspec "root:root" $ROOTFS bash -c "echo 'deb  http://mirrors.ustc.edu.cn/armbian bullseye main bullseye-utils bullseye-desktop' > /etc/apt/sources.list.d/armbian.list "
+prepare_external_binaries() {
+    local platform="$1" # linux/arm or linux/amd64 or linux/aarch64
+    local docker_image="registry.cn-hangzhou.aliyuncs.com/silentwind/kvmd-stage-0"
+
+    echo "信息：准备外部预编译二进制文件 (平台: $platform)..."
+    ensure_dir "$PREBUILT_DIR"
+
+    echo "信息：拉取 Docker 镜像 $docker_image (平台: $platform)..."
+    sudo docker pull --platform "$platform" "$docker_image" || { echo "错误：拉取 Docker 镜像 $docker_image 失败" >&2; exit 1; }
+
+    echo "信息：创建 Docker 容器 $DOCKER_CONTAINER_NAME ..."
+    sudo docker create --name "$DOCKER_CONTAINER_NAME" "$docker_image" || { echo "错误：创建 Docker 容器 $DOCKER_CONTAINER_NAME 失败" >&2; exit 1; }
+
+    echo "信息：从 Docker 容器导出文件到 $PREBUILT_DIR ..."
+    # 不直接解压到 $ROOTFS，先解压到临时目录
+    sudo docker export "$DOCKER_CONTAINER_NAME" | sudo tar -xf - -C "$PREBUILT_DIR" || { echo "错误：导出并解压 Docker 容器内容失败" >&2; exit 1; }
+
+    # Docker 容器使命完成，可以删除了 (也可以等到 trap cleanup)
+    # sudo docker rm "$DOCKER_CONTAINER_NAME"
+    # DOCKER_CONTAINER_NAME="" # 清空变量避免 trap 重复删除
+
+    echo "信息：预编译二进制文件准备完成，存放于 $PREBUILT_DIR"
 }
 
-config_file() {
-	
-    sudo mkdir -p $ROOTFS/etc/kvmd/override.d $ROOTFS/etc/kvmd/vnc $ROOTFS/var/lib/kvmd/msd $ROOTFS/opt/vc/bin $ROOTFS/usr/share/kvmd $ROOTFS/One-KVM \
-        $ROOTFS/usr/share/janus/javascript $ROOTFS/usr/lib/ustreamer/janus $ROOTFS/run/kvmd $ROOTFS/var/lib/kvmd/msd/images $ROOTFS/var/lib/kvmd/msd/meta \
-		$ROOTFS/tmp/wheel/ $ROOTFS/usr/lib/janus/transports/  $ROOTFS/usr/lib/janus/loggers
-    sudo rsync -a  --exclude={src,.github} . $ROOTFS/One-KVM
-    sudo cp -r configs/kvmd/* configs/nginx configs/janus $ROOTFS/etc/kvmd
-    sudo cp -r web extras contrib/keymaps $ROOTFS/usr/share/kvmd
-    sudo cp testenv/fakes/vcgencmd $ROOTFS/usr/bin/
-    sudo cp -r testenv/js/* $ROOTFS/usr/share/janus/javascript/
-    sudo cp build/platform/$1 $ROOTFS/usr/share/kvmd/platform
-    if [ -f "$SRCPATH/image/$1/rc.local" ]; then
-        sudo cp $SRCPATH/image/$1/rc.local $ROOTFS/etc/
+config_base_files() {
+    local platform_id="$1" # e.g., "onecloud", "cumebox2"
+    echo "信息：配置基础文件和目录结构 ($platform_id)..."
+
+    echo "信息：创建 KVMD 相关目录..."
+    ensure_dir "$ROOTFS/etc/kvmd/override.d"
+    ensure_dir "$ROOTFS/etc/kvmd/vnc"
+    ensure_dir "$ROOTFS/var/lib/kvmd/msd/images"
+    ensure_dir "$ROOTFS/var/lib/kvmd/msd/meta"
+    ensure_dir "$ROOTFS/opt/vc/bin" # 即使是假的 vcgencmd 也需要目录
+    ensure_dir "$ROOTFS/usr/share/kvmd"
+    ensure_dir "$ROOTFS/One-KVM" # 源码目录
+    ensure_dir "$ROOTFS/usr/share/janus/javascript"
+    ensure_dir "$ROOTFS/usr/lib/ustreamer/janus"
+    ensure_dir "$ROOTFS/run/kvmd"
+    ensure_dir "$ROOTFS/tmp/wheel/"
+    ensure_dir "$ROOTFS/usr/lib/janus/transports/"
+    ensure_dir "$ROOTFS/usr/lib/janus/loggers"
+
+    echo "信息：复制 One-KVM 源码..."
+    # 假设当前目录就是 One-KVM 源码根目录
+    sudo rsync -a --exclude={.git,.github,output,tmp} . "$ROOTFS/One-KVM/" || { echo "错误：复制 One-KVM 源码失败" >&2; exit 1; }
+
+    echo "信息：复制配置文件..."
+    sudo cp -r configs/kvmd/* configs/nginx configs/janus "$ROOTFS/etc/kvmd/"
+    sudo cp -r web extras contrib/keymaps "$ROOTFS/usr/share/kvmd/"
+    sudo cp testenv/fakes/vcgencmd "$ROOTFS/usr/bin/" # Fake vcgencmd
+    sudo cp -r testenv/js/* "$ROOTFS/usr/share/janus/javascript/"
+    sudo cp "build/platform/$platform_id" "$ROOTFS/usr/share/kvmd/platform" || { echo "错误：复制平台文件 build/platform/$platform_id 失败" >&2; exit 1; }
+
+    # 复制特定设备的 rc.local (如果存在)
+    if [ -f "$SRCPATH/image/$platform_id/rc.local" ]; then
+        echo "信息：复制设备特定的 rc.local 文件..."
+        sudo cp "$SRCPATH/image/$platform_id/rc.local" "$ROOTFS/etc/"
     fi
 
-	sudo docker pull --platform linux/$2 registry.cn-hangzhou.aliyuncs.com/silentwind/kvmd-stage-0
-	sudo docker create --name to_build_rootfs registry.cn-hangzhou.aliyuncs.com/silentwind/kvmd-stage-0
-	sudo docker export to_build_rootfs  | sudo tar -xvf - -C $SRCPATH/tmp/rootfs
-	sudo cp $SRCPATH/tmp/rootfs/tmp/lib/*  $ROOTFS/lib/*-linux-*/
-	sudo cp $SRCPATH/tmp/rootfs/tmp/ustreamer/ustreamer $SRCPATH/tmp/rootfs/tmp/ustreamer/ustreamer-dump $SRCPATH/tmp/rootfs/usr/bin/janus $ROOTFS/usr/bin/
-	sudo cp $SRCPATH/tmp/rootfs/tmp/ustreamer/janus/libjanus_ustreamer.so $ROOTFS/usr/lib/ustreamer/janus/
-	sudo cp $SRCPATH/tmp/rootfs/tmp/wheel/*.whl $ROOTFS/tmp/wheel/
-	sudo cp $SRCPATH/tmp/rootfs/usr/lib/janus/transports/* $ROOTFS/usr/lib/janus/transports/
+    echo "信息：从预编译目录复制二进制文件和库..."
+    # 从 prepare_external_binaries 准备好的目录复制
+    # 注意：这里的路径需要根据 docker export 出来的实际结构调整
+    sudo cp "$PREBUILT_DIR/tmp/lib/"* "$ROOTFS/lib/"*-linux-*/ || echo "警告：复制 /tmp/lib/* 失败，可能源目录或目标目录不存在或不匹配"
+    sudo cp "$PREBUILT_DIR/tmp/ustreamer/ustreamer" "$PREBUILT_DIR/tmp/ustreamer/ustreamer-dump" "$PREBUILT_DIR/usr/bin/janus" "$ROOTFS/usr/bin/" || { echo "错误：复制 ustreamer/janus 二进制文件失败" >&2; exit 1; }
+	sudo cp "$PREBUILT_DIR/tmp/ustreamer/janus/libjanus_ustreamer.so" "$ROOTFS/usr/lib/ustreamer/janus/" || { echo "错误：复制 libjanus_ustreamer.so 失败" >&2; exit 1; }
+	sudo cp "$PREBUILT_DIR/tmp/wheel/"*.whl "$ROOTFS/tmp/wheel/" || { echo "错误：复制 Python wheel 文件失败" >&2; exit 1; }
+	sudo cp "$PREBUILT_DIR/usr/lib/janus/transports/"* "$ROOTFS/usr/lib/janus/transports/" || { echo "错误：复制 Janus transports 失败" >&2; exit 1; }
 
-	sudo mv $ROOTFS/etc/apt/apt.conf.d/50apt-file.conf{,.disabled}
+	# 禁用 apt-file (如果不需要)
+ 	if [ -f "$ROOTFS/etc/apt/apt.conf.d/50apt-file.conf" ]; then
+        echo "信息：禁用 apt-file 配置..."
+        sudo mv "$ROOTFS/etc/apt/apt.conf.d/50apt-file.conf" "$ROOTFS/etc/apt/apt.conf.d/50apt-file.conf.disabled"
+    fi
+    echo "信息：基础文件配置完成。"
 }
 
+
+# --- KVMD 安装与配置 (拆分后) ---
+
+install_base_packages() {
+    echo "信息：在 chroot 环境中更新源并安装基础软件包..."
+    run_in_chroot "
+    apt-get update && \\
+    apt install -y --no-install-recommends \\
+        libxkbcommon-x11-0 nginx tesseract-ocr tesseract-ocr-eng tesseract-ocr-chi-sim \\
+        iptables network-manager curl kmod libmicrohttpd12 libjansson4 libssl3 \\
+        libsofia-sip-ua0 libglib2.0-0 libopus0 libogg0 libcurl4 libconfig9 \\
+        python3-pip net-tools && \\
+    apt clean && \\
+    rm -rf /var/lib/apt/lists/*
+    "
+}
+
+configure_network() {
+    local network_type="$1" # "systemd-networkd" or others (default network-manager)
+    if [ "$network_type" = "systemd-networkd" ]; then
+        echo "信息：在 chroot 环境中配置 systemd-networkd..."
+        run_in_chroot "
+        echo -e '[Match]\\nName=eth0\\n\\n[Network]\\nDHCP=yes\\n\\n[Link]\\nMACAddress=B6:AE:B3:21:42:0C' > /etc/systemd/network/99-eth0.network && \\
+        systemctl mask NetworkManager && \\
+        systemctl unmask systemd-networkd && \\
+        systemctl enable systemd-networkd systemd-resolved
+        "
+    else
+        echo "信息：使用默认的网络管理器 (NetworkManager)..."
+        # 可能需要确保 NetworkManager 是启用的 (通常默认是)
+        run_in_chroot "systemctl enable NetworkManager"
+    fi
+}
+
+install_python_deps() {
+    echo "信息：在 chroot 环境中安装 Python 依赖 (wheels)..."
+    run_in_chroot "
+    pip3 install --no-cache-dir --break-system-packages /tmp/wheel/*.whl && \\
+    pip3 cache purge && \\
+    rm -rf /tmp/wheel
+    "
+}
+
+configure_kvmd_core() {
+     echo "信息：在 chroot 环境中安装和配置 KVMD 核心..."
+     run_in_chroot "
+     cd /One-KVM && \\
+     python3 setup.py install && \\
+     bash scripts/kvmd-gencert --do-the-thing && \\
+     bash scripts/kvmd-gencert --do-the-thing --vnc && \\
+     kvmd-nginx-mkconf /etc/kvmd/nginx/nginx.conf.mako /etc/kvmd/nginx/nginx.conf && \\
+     kvmd -m
+     "
+}
+
+configure_system() {
+    echo "信息：在 chroot 环境中配置系统级设置 (sudoers, udev, services)..."
+    run_in_chroot "
+    cat /One-KVM/configs/os/sudoers/v2-hdmiusb >> /etc/sudoers && \\
+    cat /One-KVM/configs/os/udev/v2-hdmiusb-rpi4.rules > /etc/udev/rules.d/99-kvmd.rules && \\
+    echo 'libcomposite' >> /etc/modules && \\
+    mv /usr/local/bin/kvmd* /usr/bin/ || echo '信息：/usr/local/bin/kvmd* 未找到或移动失败，可能已在/usr/bin' && \\
+    cp /One-KVM/configs/os/services/* /etc/systemd/system/ && \\
+    cp /One-KVM/configs/os/tmpfiles.conf /usr/lib/tmpfiles.d/ && \\
+    mv /etc/kvmd/supervisord.conf /etc/supervisord.conf && \\
+    chmod +x /etc/update-motd.d/* || echo '警告：chmod /etc/update-motd.d/* 失败' && \\
+    echo 'kvmd ALL=(ALL) NOPASSWD: /etc/kvmd/custom_atx/gpio.sh' >> /etc/sudoers && \\
+    echo 'kvmd ALL=(ALL) NOPASSWD: /etc/kvmd/custom_atx/usbrelay_hid.sh' >> /etc/sudoers && \\
+    systemd-sysusers /One-KVM/configs/os/sysusers.conf && \\
+    systemd-sysusers /One-KVM/configs/os/kvmd-webterm.conf && \\
+    ln -sf /usr/share/tesseract-ocr/*/tessdata /usr/share/tessdata || echo '警告：创建 tesseract 链接失败' && \\
+    sed -i 's/8080/80/g' /etc/kvmd/override.yaml && \\
+    sed -i 's/4430/443/g' /etc/kvmd/override.yaml && \\
+    chown kvmd -R /var/lib/kvmd/msd/ && \\
+    systemctl enable kvmd kvmd-otg kvmd-nginx kvmd-vnc kvmd-ipmi kvmd-webterm kvmd-janus kvmd-media && \\
+    systemctl disable nginx && \\
+    rm -rf /One-KVM
+    "
+}
+
+install_webterm() {
+    local arch="$1" # armhf, aarch64, x86_64
+    local ttyd_arch="$arch"
+    # ttyd release 可能没有 armhf，需要确认使用的架构名
+    if [ "$arch" = "armhf" ]; then
+        # ttyd 可能标记为 armv7, arm, 需要确认
+        echo "警告：ttyd 可能没有 armhf 的 release，尝试使用 armv7 或 arm。请检查 ttyd release 页面！"
+        # 假设用 armv7
+        ttyd_arch="armv7"
+    elif [ "$arch" = "amd64" ]; then
+         ttyd_arch="x86_64" # ttyd 通常用 x86_64
+    fi
+
+    echo "信息：在 chroot 环境中下载并安装 ttyd ($ttyd_arch)..."
+    run_in_chroot "
+    curl -L https://gh.llkk.cc/https://github.com/tsl0922/ttyd/releases/download/1.7.7/ttyd.${ttyd_arch} -o /usr/bin/ttyd && \\
+    chmod +x /usr/bin/ttyd && \\
+    mkdir -p /home/kvmd-webterm && \\
+    chown kvmd-webterm /home/kvmd-webterm
+    "
+}
+
+apply_kvmd_tweaks() {
+    local arch="$1" # armhf, aarch64, x86_64
+    local device_type="$2" # "gpio" or "video1" or other
+    local atx_setting=""
+    local hid_setting=""
+
+    echo "信息：根据架构 ($arch) 和设备类型 ($device_type) 调整 KVMD 配置..."
+
+    if [ "$arch" = "x86_64" ] || [ "$arch" = "amd64" ]; then
+        echo "信息：检测到 x86_64/amd64 架构，禁用 OTG，设置 ATX 为 USBRELAY_HID..."
+        run_in_chroot "
+        systemctl disable kvmd-otg && \\
+        sed -i 's/^ATX=.*/ATX=USBRELAY_HID/' /etc/kvmd/atx.sh && \\
+        sed -i 's/device: \/dev\/ttyUSB0/device: \/dev\/kvmd-hid/g' /etc/kvmd/override.yaml
+        "
+    else
+        echo "信息：检测到 ARM 架构 ($arch)..."
+        # ARM 架构，配置 HID 为 OTG
+        hid_setting="otg"
+        run_in_chroot "
+        sed -i 's/#type: otg/type: otg/g' /etc/kvmd/override.yaml && \\
+        sed -i 's/device: \/dev\/ttyUSB0/#device: \/dev\/ttyUSB0/g' /etc/kvmd/override.yaml # 注释掉 ttyUSB0
+        "
+        echo "信息：设置 HID 为 $hid_setting"
+        run_in_chroot "sed -i 's/type: ch9329/type: $hid_setting/g' /etc/kvmd/override.yaml"
+
+
+        # 根据 device_type 配置 ATX
+        if [ "$device_type" = "gpio" ]; then
+            echo "信息：设备类型为 gpio，设置 ATX 为 GPIO 并配置引脚..."
+            atx_setting="GPIO"
+             run_in_chroot "
+             sed -i 's/^ATX=.*/ATX=GPIO/' /etc/kvmd/atx.sh && \\
+             sed -i 's/SHUTDOWNPIN/gpiochip1 7/g' /etc/kvmd/custom_atx/gpio.sh && \\
+             sed -i 's/REBOOTPIN/gpiochip0 11/g' /etc/kvmd/custom_atx/gpio.sh
+             "
+        else
+            echo "信息：设备类型不是 gpio ($device_type)，设置 ATX 为 USBRELAY_HID..."
+            atx_setting="USBRELAY_HID"
+             run_in_chroot "sed -i 's/^ATX=.*/ATX=USBRELAY_HID/' /etc/kvmd/atx.sh"
+        fi
+
+        # 配置视频设备
+        if [ "$device_type" = "video1" ]; then
+            echo "信息：设备类型为 video1，设置视频设备为 /dev/video1..."
+            run_in_chroot "sed -i 's|/dev/video0|/dev/video1|g' /etc/kvmd/override.yaml"
+        else
+             echo "信息：使用默认视频设备 /dev/video0..."
+        fi
+    fi
+    echo "信息：KVMD 配置调整完成。"
+}
+
+
+# --- 整体安装流程 ---
+install_and_configure_kvmd() {
+    local arch="$1"         # 架构: armhf, aarch64, x86_64/amd64
+    local device_type="$2"  # 设备特性: "gpio", "video1", "" (空或其他)
+    local network_type="$3" # 网络配置: "systemd-networkd", "" (默认 network-manager)
+    local host_arch=""      # Docker 平台架构: arm, aarch64, amd64
+
+    # 映射架构名称
+    case "$arch" in
+        armhf) host_arch="arm" ;;
+        aarch64) host_arch="arm64" ;; # docker aarch64 平台名是 arm64
+        x86_64|amd64) host_arch="amd64"; arch="x86_64" ;; # 统一内部使用 x86_64
+        *) echo "错误：不支持的架构 $arch"; exit 1 ;;
+    esac
+
+
+    prepare_external_binaries "linux/$host_arch"
+    config_base_files "$TARGET_DEVICE_NAME" # 使用全局变量传递设备名
+
+    # 特定设备的额外文件配置 (如果存在)
+    if declare -f "config_${TARGET_DEVICE_NAME}_files" > /dev/null; then
+        echo "信息：执行特定设备的文件配置函数 config_${TARGET_DEVICE_NAME}_files ..."
+        "config_${TARGET_DEVICE_NAME}_files"
+    fi
+
+    # 某些镜像可能需要准备DNS和换源
+    if [[ "$NEED_PREPARE_DNS" = true ]]; then
+        prepare_dns_and_mirrors
+    fi
+    # 可选：强制使用特定armbian源
+    # delete_armbian_verify
+
+    # 执行安装步骤
+    install_base_packages
+    configure_network "$network_type"
+    install_python_deps
+    configure_kvmd_core
+    configure_system
+    install_webterm "$arch" # 传递原始架构名给ttyd下载
+    apply_kvmd_tweaks "$arch" "$device_type"
+
+    run_in_chroot "df -h" # 显示最终磁盘使用情况
+    echo "信息：One-KVM 安装和配置完成。"
+}
+
+# --- 打包函数 ---
+
 pack_img() {
-    sudo mv $SRCPATH/tmp/rootfs.img  $OUTPUTDIR/One-KVM_by-SilentWind_$1_$DATE.img
-    if [ "$1" = "Vm" ]; then
-        sudo qemu-img convert -f raw -O vmdk $OUTPUTDIR/One-KVM_by-SilentWind_Vm_$DATE.img $OUTPUTDIR/One-KVM_by-SilentWind_Vmare-uefi_$DATE.vmdk
-        sudo qemu-img convert -f raw -O vdi $OUTPUTDIR/One-KVM_by-SilentWind_Vm_$DATE.img $OUTPUTDIR/One-KVM_by-SilentWind_Virtualbox-uefi_$DATE.vdi
+    local device_name_friendly="$1" # e.g., "Vm", "Cumebox2"
+    local target_img_name="One-KVM_by-SilentWind_${device_name_friendly}_${DATE}.img"
+    local source_img="$TMPDIR/rootfs.img"
+
+    echo "信息：开始打包镜像 ($device_name_friendly)..."
+    ensure_dir "$OUTPUTDIR"
+
+    echo "信息：移动镜像文件 $source_img 到 $OUTPUTDIR/$target_img_name ..."
+    sudo mv "$source_img" "$OUTPUTDIR/$target_img_name" || { echo "错误：移动镜像文件失败" >&2; exit 1; }
+
+    if [ "$device_name_friendly" = "Vm" ]; then
+        echo "信息：为 Vm 目标转换镜像格式 (vmdk, vdi)..."
+        local raw_img="$OUTPUTDIR/$target_img_name"
+        local vmdk_img="$OUTPUTDIR/One-KVM_by-SilentWind_Vmare-uefi_${DATE}.vmdk"
+        local vdi_img="$OUTPUTDIR/One-KVM_by-SilentWind_Virtualbox-uefi_${DATE}.vdi"
+
+        echo "信息：转换为 VMDK..."
+        sudo qemu-img convert -f raw -O vmdk "$raw_img" "$vmdk_img" || echo "警告：转换为 VMDK 失败"
+        echo "信息：转换为 VDI..."
+        sudo qemu-img convert -f raw -O vdi "$raw_img" "$vdi_img" || echo "警告：转换为 VDI 失败"
     fi
-}
-
-onecloud_rootfs() {
-    $SRCPATH/image/onecloud/AmlImg_v0.3.1_linux_amd64 unpack $SRCPATH/image/onecloud/Armbian_by-SilentWind_24.5.0-trunk_Onecloud_bookworm_legacy_5.9.0-rc7_minimal.burn.img $SRCPATH/tmp
-    simg2img $SRCPATH/tmp/6.boot.PARTITION.sparse $SRCPATH/tmp/bootfs.img
-    simg2img $SRCPATH/tmp/7.rootfs.PARTITION.sparse $SRCPATH/tmp/rootfs.img
-    mkdir $BOOTFS
-    sudo losetup $LOOPDEV $SRCPATH/tmp/bootfs.img  || exit -1
-    sudo mount $LOOPDEV $BOOTFS
-    sudo cp $SRCPATH/image/onecloud/meson8b-onecloud-fix.dtb $BOOTFS/dtb/meson8b-onecloud.dtb
-    sudo umount $BOOTFS
-    sudo losetup -d $LOOPDEV
-    dd if=/dev/zero of=/tmp/add.img bs=1M count=256 && cat /tmp/add.img >> $SRCPATH/tmp/rootfs.img && rm /tmp/add.img
-    e2fsck -f $SRCPATH/tmp/rootfs.img && resize2fs $SRCPATH/tmp/rootfs.img
-    sudo losetup $LOOPDEV $SRCPATH/tmp/rootfs.img
-}
-
-cumebox2_rootfs() {
-    cp $SRCPATH/image/cumebox2/Armbian_25.2.2_Khadas-vim1_bookworm_current_6.12.17_minimal.img $SRCPATH/tmp/rootfs.img
-    sudo parted -s $SRCPATH/tmp/rootfs.img resizepart 1 100% || exit -1
-    sudo losetup --offset $((8192*512)) $LOOPDEV $SRCPATH/tmp/rootfs.img  || exit -1
-    sudo e2fsck -f $LOOPDEV && sudo resize2fs $LOOPDEV
-}
-
-chainedbox_rootfs_and_fix_dtb() {
-    cp $SRCPATH/image/chainedbox/Armbian_24.11.0_rockchip_chainedbox_bookworm_6.1.112_server_2024.10.02_add800m.img $SRCPATH/tmp/rootfs.img
-    mkdir $BOOTFS
-    sudo losetup --offset $((32768*512)) $LOOPDEV $SRCPATH/tmp/rootfs.img  || exit -1
-    sudo mount $LOOPDEV $BOOTFS
-    sudo cp $SRCPATH/image/chainedbox/rk3328-l1pro-1296mhz-fix.dtb $BOOTFS/dtb/rockchip/rk3328-l1pro-1296mhz.dtb
-    sudo umount $BOOTFS
-    sudo losetup -d $LOOPDEV
-    sudo losetup --offset $((1081344*512)) $LOOPDEV $SRCPATH/tmp/rootfs.img
-}
-
-vm_rootfs() {
-    cp $SRCPATH/image/vm/Armbian_24.8.1_Uefi-x86_bookworm_current_6.6.47_minimal_add1g.img $SRCPATH/tmp/rootfs.img
-    sudo losetup --offset $((540672*512)) $LOOPDEV $SRCPATH/tmp/rootfs.img  || exit -1
-}
-
-e900v22c_rootfs() {
-    cp $SRCPATH/image/e900v22c/Armbian_23.08.0_amlogic_s905l3a_bookworm_5.15.123_server_2023.08.01.img $SRCPATH/tmp/rootfs.img
-    dd if=/dev/zero of=/tmp/add.img bs=1M count=400 && cat /tmp/add.img >> $SRCPATH/tmp/rootfs.img && rm /tmp/add.img
-    sudo parted -s $SRCPATH/tmp/rootfs.img resizepart 2 100% || exit -1
-    sudo losetup --offset $((532480*512)) $LOOPDEV $SRCPATH/tmp/rootfs.img  || exit -1
-    sudo e2fsck -f $LOOPDEV && sudo resize2fs $LOOPDEV
-}
-
-
-octopus-flanet_rootfs() {
-    cp $SRCPATH/image/octopus-flanet/Armbian_24.11.0_amlogic_s912_bookworm_6.1.114_server_2024.11.01.img $SRCPATH/tmp/rootfs.img
-    mkdir $BOOTFS
-    sudo losetup --offset $((8192*512)) $LOOPDEV $SRCPATH/tmp/rootfs.img  || exit -1
-    sudo mount $LOOPDEV $BOOTFS
-    sudo sed -i "s/meson-gxm-octopus-planet.dtb/meson-gxm-khadas-vim2.dtb/g" $BOOTFS/uEnv.txt
-    sudo umount $BOOTFS
-    sudo losetup -d $LOOPDEV
-    dd if=/dev/zero of=/tmp/add.img bs=1M count=400 && cat /tmp/add.img >> $SRCPATH/tmp/rootfs.img && rm /tmp/add.img
-    sudo parted -s $SRCPATH/tmp/rootfs.img resizepart 2 100% || exit -1
-    sudo losetup --offset $((1056768*512)) $LOOPDEV $SRCPATH/tmp/rootfs.img  || exit -1
-    sudo e2fsck -f $LOOPDEV && sudo resize2fs $LOOPDEV
-}
-
-
-config_cumebox2_file() {
-    sudo mkdir $ROOTFS/etc/oled
-    sudo cp $SRCPATH/image/cumebox2/v-fix.dtb $ROOTFS/boot/dtb/amlogic/meson-gxl-s905x-khadas-vim.dtb
-    sudo cp $SRCPATH/image/cumebox2/ssd $ROOTFS/usr/bin/
-	sudo chmod +x $ROOTFS/usr/bin/ssd
-    sudo cp $SRCPATH/image/cumebox2/config.json $ROOTFS/etc/oled/config.json
-}
-
-config_octopus-flanet_file() {
-    sudo cp $SRCPATH/image/octopus-flanet/model_database.conf $ROOTFS/etc/model_database.conf
-}
-
-instal_one-kvm() {
-    #$1 arch; $2 deivce: "gpio" or "video1"; $3 network: "systemd-networkd",default is network-manager
-    sudo chroot --userspec "root:root" $ROOTFS bash -c " \
-        df -h \
-        && apt-get update \
-        && apt install -y --no-install-recommends libxkbcommon-x11-0 nginx tesseract-ocr tesseract-ocr-eng tesseract-ocr-chi-sim iptables network-manager \
-        	curl kmod libmicrohttpd12 libjansson4 libssl3 libsofia-sip-ua0 libglib2.0-0 libopus0 libogg0 libcurl4 libconfig9 python3-pip net-tools \
-        && apt clean \
-		&& rm -rf /var/lib/apt/lists/* "
-
-    if [ "$3" = "systemd-networkd" ]; then 
-        sudo chroot --userspec "root:root" $ROOTFS bash -c " \
-            echo -e '[Match]\nName=eth0\n\n[Network]\nDHCP=yes\n\n[Link]\nMACAddress=B6:AE:B3:21:42:0C' > /etc/systemd/network/99-eth0.network \
-            && systemctl mask NetworkManager \
-            && systemctl unmask systemd-networkd \
-            && systemctl enable systemd-networkd systemd-resolved "        
-    fi
-    sudo chroot --userspec "root:root" $ROOTFS bash -c " \
-        pip3 install --no-cache-dir --break-system-packages /tmp/wheel/*.whl \
-        && pip3 cache purge \
-		&& rm -r /tmp/wheel "
-
-    sudo chroot --userspec "root:root" $ROOTFS bash -c " \
-        cd /One-KVM \
-        && python3 setup.py install \
-        && bash scripts/kvmd-gencert --do-the-thing \
-        && bash scripts/kvmd-gencert --do-the-thing --vnc \
-        && kvmd-nginx-mkconf /etc/kvmd/nginx/nginx.conf.mako /etc/kvmd/nginx/nginx.conf \
-        && kvmd -m "
-
-    sudo chroot --userspec "root:root" $ROOTFS bash -c " \
-        cat /One-KVM/configs/os/sudoers/v2-hdmiusb >> /etc/sudoers \
-        && cat /One-KVM/configs/os/udev/v2-hdmiusb-rpi4.rules > /etc/udev/rules.d/99-kvmd.rules \
-        && echo 'libcomposite' >> /etc/modules \
-        && mv /usr/local/bin/kvmd* /usr/bin \
-        && cp /One-KVM/configs/os/services/* /etc/systemd/system/ \
-        && cp /One-KVM/configs/os/tmpfiles.conf /usr/lib/tmpfiles.d/ \
-		&& mv /etc/kvmd/supervisord.conf /etc/supervisord.conf \
-        && chmod +x /etc/update-motd.d/* \
-        && echo 'kvmd ALL=(ALL) NOPASSWD: /etc/kvmd/custom_atx/gpio.sh' >>  /etc/sudoers \
-        && echo 'kvmd ALL=(ALL) NOPASSWD: /etc/kvmd/custom_atx/usbrelay_hid.sh' >>  /etc/sudoers \
-        && systemd-sysusers /One-KVM/configs/os/sysusers.conf \
-        && systemd-sysusers /One-KVM/configs/os/kvmd-webterm.conf \
-        && ln -sf /usr/share/tesseract-ocr/*/tessdata /usr/share/tessdata \
-        && sed -i 's/8080/80/g' /etc/kvmd/override.yaml \
-        && sed -i 's/4430/443/g' /etc/kvmd/override.yaml \
-        && chown kvmd -R /var/lib/kvmd/msd/ \
-        && systemctl enable kvmd kvmd-otg kvmd-nginx kvmd-vnc kvmd-ipmi kvmd-webterm kvmd-janus kvmd-media \
-        && systemctl disable nginx \
-        && rm -r /One-KVM "
-
-    sudo chroot --userspec "root:root" $ROOTFS bash -c " \
-        curl https://gh.llkk.cc/https://github.com/tsl0922/ttyd/releases/download/1.7.7/ttyd.$1 -L -o /usr/bin/ttyd \
-        && chmod +x /usr/bin/ttyd \
-        && mkdir -p /home/kvmd-webterm \
-        && chown kvmd-webterm /home/kvmd-webterm "
-
-    if [ "$1" = "x86_64" ]; then
-        sudo chroot --userspec "root:root" $ROOTFS bash -c " \
-            systemctl disable kvmd-otg \
-            && sed -i '2c ATX=USBRELAY_HID' /etc/kvmd/atx.sh \
-            && sed -i 's/device: \/dev\/ttyUSB0/device: \/dev\/kvmd-hid/g' /etc/kvmd/override.yaml "
-    else
-        if [ "$2" = "gpio" ]; then
-            sudo chroot --userspec "root:root" $ROOTFS bash -c " \
-                sed -i '2c ATX=GPIO' /etc/kvmd/atx.sh \
-                && sed -i 's/SHUTDOWNPIN/gpiochip1 7/g' /etc/kvmd/custom_atx/gpio.sh \
-                && sed -i 's/REBOOTPIN/gpiochip0 11/g' /etc/kvmd/custom_atx/gpio.sh "
-        else
-            sudo chroot --userspec "root:root" $ROOTFS sed -i '2c ATX=USBRELAY_HID' /etc/kvmd/atx.sh
-            
-        fi
-        if [ "$2" = "video1" ]; then
-            sudo chroot --userspec "root:root" $ROOTFS sed -i 's/\/dev\/video0/\/dev\/video1/g' /etc/kvmd/override.yaml
-        fi
-        sudo chroot --userspec "root:root" $ROOTFS bash -c " \
-            sed -i 's/ch9329/otg/g' /etc/kvmd/override.yaml \
-            && sed -i 's/device: \/dev\/ttyUSB0//g' /etc/kvmd/override.yaml \
-            && sed -i 's/#type: otg/type: otg/g' /etc/kvmd/override.yaml "
-    fi
-
-	sudo chroot --userspec "root:root" $ROOTFS bash -c "df -h"
+    echo "信息：镜像打包完成: $OUTPUTDIR/$target_img_name"
 }
 
 pack_img_onecloud() {
-    sudo rm $SRCPATH/tmp/7.rootfs.PARTITION.sparse
-    sudo img2simg $SRCPATH/tmp/rootfs.img $SRCPATH/tmp/7.rootfs.PARTITION.sparse
-    sudo $SRCPATH/image/onecloud/AmlImg_v0.3.1_linux_amd64 pack $OUTPUTDIR/One-KVM_by-SilentWind_Onecloud_$DATE.burn.img $SRCPATH/tmp/
-    sudo rm $SRCPATH/tmp/*
+    local target_img_name="One-KVM_by-SilentWind_Onecloud_${DATE}.burn.img"
+    local rootfs_raw_img="$TMPDIR/rootfs.img"
+    local rootfs_sparse_img="$TMPDIR/7.rootfs.PARTITION.sparse"
+    local aml_packer="$SRCPATH/image/onecloud/AmlImg_v0.3.1_linux_amd64"
+
+    echo "信息：开始为 Onecloud 打包 burn 镜像..."
+    ensure_dir "$OUTPUTDIR"
+
+    echo "信息：将 raw rootfs 转换为 sparse image..."
+    # 先删除可能存在的旧 sparse 文件
+    sudo rm -f "$rootfs_sparse_img"
+    sudo img2simg "$rootfs_raw_img" "$rootfs_sparse_img" || { echo "错误：img2simg 转换失败" >&2; exit 1; }
+    sudo rm "$rootfs_raw_img" # 删除 raw 文件，因为它已被转换
+
+    echo "信息：使用 AmlImg 工具打包..."
+    # 确保 AmlImg 工具可执行
+    sudo chmod +x "$aml_packer"
+    sudo "$aml_packer" pack "$OUTPUTDIR/$target_img_name" "$TMPDIR/" || { echo "错误：AmlImg 打包失败" >&2; exit 1; }
+
+    echo "信息：清理 Onecloud 临时文件..."
+    sudo rm -f "$TMPDIR/6.boot.PARTITION.sparse" "$TMPDIR/7.rootfs.PARTITION.sparse" "$TMPDIR/dts.img" # 清理所有已知部件
+    # 注意：TMPDIR 下可能还有其他由 unpack 生成的文件，按需清理
+
+    echo "信息：Onecloud burn 镜像打包完成: $OUTPUTDIR/$target_img_name"
 }
 
-#build function
+# --- 设备特定的 Rootfs 准备函数 ---
 
-onecloud() {
-    onecloud_rootfs
-    mount_rootfs
-    config_file "onecloud" "arm"
-    instal_one-kvm armhf gpio systemd-networkd
-    write_meta "onecloud"
-    umount_rootfs
-    pack_img_onecloud
+onecloud_rootfs() {
+    local unpacker="$SRCPATH/image/onecloud/AmlImg_v0.3.1_linux_amd64"
+    local source_image="$SRCPATH/image/onecloud/Armbian_by-SilentWind_24.5.0-trunk_Onecloud_bookworm_legacy_5.9.0-rc7_minimal.burn.img"
+    local bootfs_img="$TMPDIR/bootfs.img"
+    local rootfs_img="$TMPDIR/rootfs.img"
+    local bootfs_sparse="$TMPDIR/6.boot.PARTITION.sparse"
+    local rootfs_sparse="$TMPDIR/7.rootfs.PARTITION.sparse"
+    local bootfs_loopdev="" # 存储 bootfs 使用的 loop 设备
+
+    echo "信息：准备 Onecloud Rootfs..."
+    ensure_dir "$TMPDIR"
+    ensure_dir "$BOOTFS"
+
+    echo "信息：解包 Onecloud burn 镜像..."
+    # ... (unpacker logic) ...
+
+    echo "信息：转换 bootfs 和 rootfs sparse 镜像到 raw 格式..."
+    sudo simg2img "$bootfs_sparse" "$bootfs_img" || { echo "错误：转换 bootfs sparse 镜像失败" >&2; exit 1; }
+    sudo simg2img "$rootfs_sparse" "$rootfs_img" || { echo "错误：转换 rootfs sparse 镜像失败" >&2; exit 1; }
+
+    echo "信息：挂载 bootfs 并修复 DTB..."
+    find_loop_device # 查找一个 loop 设备给 bootfs
+    bootfs_loopdev="$LOOPDEV" # 保存这个设备名
+    echo "信息：将 $bootfs_img 关联到 $bootfs_loopdev..."
+    sudo losetup "$bootfs_loopdev" "$bootfs_img" || { echo "错误：关联 bootfs 镜像到 $bootfs_loopdev 失败" >&2; exit 1; }
+    sudo mount "$bootfs_loopdev" "$BOOTFS" || { echo "错误：挂载 bootfs ($bootfs_loopdev) 失败" >&2; exit 1; }
+    BOOTFS_MOUNTED=1
+    sudo cp "$SRCPATH/image/onecloud/meson8b-onecloud-fix.dtb" "$BOOTFS/dtb/meson8b-onecloud.dtb" || { echo "错误：复制修复后的 DTB 文件失败" >&2; exit 1; }
+    sudo umount "$BOOTFS" || { echo "警告：卸载 bootfs ($BOOTFS) 失败" >&2; BOOTFS_MOUNTED=0; } # 卸载失败不应中断流程
+    BOOTFS_MOUNTED=0
+    echo "信息：分离 bootfs loop 设备 $bootfs_loopdev..."
+    sudo losetup -d "$bootfs_loopdev" || { echo "警告：分离 bootfs loop 设备 $bootfs_loopdev 失败" >&2; }
+    # bootfs_loopdev 对应的设备现在是空闲的
+
+    echo "信息：扩展 rootfs 镜像..."
+    # ... (dd, cat, rm add.img) ...
+
+    echo "信息：检查并调整 rootfs 文件系统大小 (在文件上)..."
+    # 注意：e2fsck/resize2fs 现在直接操作镜像文件，而不是 loop 设备
+    sudo e2fsck -f -y "$rootfs_img" || { echo "警告：e2fsck 检查 rootfs 镜像文件失败" >&2; exit 1; }
+    sudo resize2fs "$rootfs_img" || { echo "错误：resize2fs 调整 rootfs 镜像文件大小失败" >&2; exit 1; }
+
+    echo "信息：设置 rootfs loop 设备..."
+    find_loop_device # 重新查找一个可用的 loop 设备 (可能是刚才释放的那个)
+    echo "信息：将 $rootfs_img 关联到 $LOOPDEV..."
+    sudo losetup "$LOOPDEV" "$rootfs_img" || { echo "错误：关联 rootfs 镜像到 $LOOPDEV 失败" >&2; exit 1; }
+
+    echo "信息：Onecloud Rootfs 准备完成。 Loop 设备 $LOOPDEV 已关联 $rootfs_img"
 }
 
-cumebox2() {
-    cumebox2_rootfs
-    mount_rootfs
-    config_file "cumebox2" "aarch64"
-    config_cumebox2_file
-    parpare_dns
-    instal_one-kvm aarch64 video1
-    write_meta "cumebox2"
-    umount_rootfs
-    pack_img "Cumebox2"
+cumebox2_rootfs() {
+    local source_image="$SRCPATH/image/cumebox2/Armbian_25.2.2_Khadas-vim1_bookworm_current_6.12.17_minimal.img"
+    local target_image="$TMPDIR/rootfs.img"
+    local offset=$((8192 * 512))
+
+    echo "信息：准备 Cumebox2 Rootfs..."
+    ensure_dir "$TMPDIR"
+    cp "$source_image" "$target_image" || { echo "错误：复制 Cumebox2 原始镜像失败" >&2; exit 1; }
+
+    echo "信息：调整镜像分区大小..."
+    sudo parted -s "$target_image" resizepart 1 100% || { echo "错误：使用 parted 调整分区大小失败" >&2; exit 1; }
+
+    echo "信息：设置带偏移量的 loop 设备..."
+    find_loop_device # 查找设备名
+    echo "信息：将 $target_image (偏移 $offset) 关联到 $LOOPDEV..."
+    sudo losetup --offset "$offset" "$LOOPDEV" "$target_image" || { echo "错误：设置带偏移量的 loop 设备 $LOOPDEV 失败" >&2; exit 1; }
+
+    echo "信息：检查并调整文件系统大小 (在 loop 设备上)..."
+    sudo e2fsck -f -y "$LOOPDEV" || { echo "警告：e2fsck 检查 $LOOPDEV 失败" >&2; exit 1; }
+    sudo resize2fs "$LOOPDEV" || { echo "错误：resize2fs 调整 $LOOPDEV 大小失败" >&2; exit 1; }
+
+    echo "信息：Cumebox2 Rootfs 准备完成，loop 设备 $LOOPDEV 已就绪。"
 }
 
-chainedbox() {
-    chainedbox_rootfs_and_fix_dtb
-    mount_rootfs
-    config_file "chainedbox" "aarch64"
-    parpare_dns
-    instal_one-kvm aarch64 video1
-    write_meta "chainedbox"
-    umount_rootfs
-    pack_img "Chainedbox"
+chainedbox_rootfs_and_fix_dtb() {
+    local source_image="$SRCPATH/image/chainedbox/Armbian_24.11.0_rockchip_chainedbox_bookworm_6.1.112_server_2024.10.02_add800m.img"
+    local target_image="$TMPDIR/rootfs.img"
+    local boot_offset=$((32768 * 512))
+    local rootfs_offset=$((1081344 * 512))
+    local bootfs_loopdev=""
+
+    echo "信息：准备 Chainedbox Rootfs 并修复 DTB..."
+    ensure_dir "$TMPDIR"; ensure_dir "$BOOTFS"
+    cp "$source_image" "$target_image" || { echo "错误：复制 Chainedbox 原始镜像失败" >&2; exit 1; }
+
+    echo "信息：挂载 boot 分区并修复 DTB..."
+    find_loop_device # 找 loop 给 boot
+    bootfs_loopdev="$LOOPDEV"
+    echo "信息：将 $target_image (偏移 $boot_offset) 关联到 $bootfs_loopdev..."
+    sudo losetup --offset "$boot_offset" "$bootfs_loopdev" "$target_image" || { echo "错误：设置 boot 分区 loop 设备 $bootfs_loopdev 失败" >&2; exit 1; }
+    sudo mount "$bootfs_loopdev" "$BOOTFS" || { echo "错误：挂载 boot 分区 ($bootfs_loopdev) 失败" >&2; exit 1; }
+    BOOTFS_MOUNTED=1
+    sudo cp "$SRCPATH/image/chainedbox/rk3328-l1pro-1296mhz-fix.dtb" "$BOOTFS/dtb/rockchip/rk3328-l1pro-1296mhz.dtb" || { echo "错误：复制修复后的 DTB 文件失败" >&2; exit 1; }
+    sudo umount "$BOOTFS" || { echo "警告：卸载 boot 分区 ($BOOTFS) 失败" >&2; BOOTFS_MOUNTED=0; }
+    BOOTFS_MOUNTED=0
+    echo "信息：分离 boot loop 设备 $bootfs_loopdev..."
+    sudo losetup -d "$bootfs_loopdev" || { echo "警告：分离 boot 分区 loop 设备 $bootfs_loopdev 失败" >&2; }
+
+    echo "信息：设置 rootfs 分区的 loop 设备..."
+    find_loop_device # 找 loop 给 rootfs
+    echo "信息：将 $target_image (偏移 $rootfs_offset) 关联到 $LOOPDEV..."
+    sudo losetup --offset "$rootfs_offset" "$LOOPDEV" "$target_image" || { echo "错误：设置 rootfs 分区 loop 设备 $LOOPDEV 失败" >&2; exit 1; }
+
+    # 如果需要 resize rootfs 分区，可以在这里操作 $LOOPDEV
+    # echo "信息：检查并调整文件系统大小 (在 loop 设备上)..."
+    # sudo e2fsck -f -y "$LOOPDEV" || { echo "警告：e2fsck 检查 $LOOPDEV 失败" >&2; exit 1; }
+    # sudo resize2fs "$LOOPDEV" || { echo "错误：resize2fs 调整 $LOOPDEV 大小失败" >&2; exit 1; }
+
+    echo "信息：Chainedbox Rootfs 准备完成，loop 设备 $LOOPDEV 已就绪。"
 }
 
-vm() {
-    vm_rootfs
-    mount_rootfs
-    config_file "vm" "amd64"
-    parpare_dns
-    instal_one-kvm x86_64
-    write_meta "vm"
-    umount_rootfs
-    pack_img "Vm"
+
+vm_rootfs() {
+    local source_image="$SRCPATH/image/vm/Armbian_25.2.1_Uefi-x86_bookworm_current_6.12.13_minimal.img"
+    local target_image="$TMPDIR/rootfs.img"
+    local offset=$((540672 * 512))
+
+    echo "信息：准备 Vm Rootfs..."
+    ensure_dir "$TMPDIR"
+    cp "$source_image" "$target_image" || { echo "错误：复制 Vm 原始镜像失败" >&2; exit 1; }
+
+    echo "信息：设置带偏移量的 loop 设备..."
+    find_loop_device # 查找设备名
+    echo "信息：将 $target_image (偏移 $offset) 关联到 $LOOPDEV..."
+    sudo losetup --offset "$offset" "$LOOPDEV" "$target_image" || { echo "错误：设置带偏移量的 loop 设备 $LOOPDEV 失败" >&2; exit 1; }
+
+    # Optional resize on $LOOPDEV here if needed
+
+    echo "信息：Vm Rootfs 准备完成，loop 设备 $LOOPDEV 已就绪。"
 }
 
-e900v22c() {
-    e900v22c_rootfs
-    mount_rootfs
-    config_file "e900v22c" "aarch64"
-    instal_one-kvm aarch64 video1
-    write_meta "e900v22c"
-    umount_rootfs
-    pack_img "E900v22c"
+e900v22c_rootfs() {
+    local source_image="$SRCPATH/image/e900v22c/Armbian_23.08.0_amlogic_s905l3a_bookworm_5.15.123_server_2023.08.01.img"
+    local target_image="$TMPDIR/rootfs.img"
+    local offset=$((532480 * 512))
+    local add_size_mb=400
+
+    echo "信息：准备 E900V22C Rootfs..."
+    ensure_dir "$TMPDIR"
+    cp "$source_image" "$target_image" || { echo "错误：复制 E900V22C 原始镜像失败" >&2; exit 1; }
+
+    echo "信息：扩展镜像文件 (${add_size_mb}MB)..."
+    # ... (dd, cat, rm add.img logic) ...
+
+    echo "信息：调整镜像分区大小 (分区 2)..."
+    sudo parted -s "$target_image" resizepart 2 100% || { echo "错误：使用 parted 调整分区 2 大小失败" >&2; exit 1; }
+
+    echo "信息：设置带偏移量的 loop 设备..."
+    find_loop_device # 查找设备名
+    echo "信息：将 $target_image (偏移 $offset) 关联到 $LOOPDEV..."
+    sudo losetup --offset "$offset" "$LOOPDEV" "$target_image" || { echo "错误：设置带偏移量的 loop 设备 $LOOPDEV 失败" >&2; exit 1; }
+
+    echo "信息：检查并调整文件系统大小 (在 loop 设备上)..."
+    sudo e2fsck -f -y "$LOOPDEV" || { echo "警告：e2fsck 检查 $LOOPDEV 失败" >&2; exit 1; }
+    sudo resize2fs "$LOOPDEV" || { echo "错误：resize2fs 调整 $LOOPDEV 大小失败" >&2; exit 1; }
+
+    echo "信息：E900V22C Rootfs 准备完成，loop 设备 $LOOPDEV 已就绪。"
 }
 
-octopus_flanet() {
-    octopus-flanet_rootfs
-    mount_rootfs
-    config_file "octopus-flanet" "aarch64"
-    config_octopus-flanet_file
-    parpare_dns
-    instal_one-kvm aarch64 video1
-    write_meta "octopus-flanet"
-    umount_rootfs
-    pack_img "Octopus-Flanet"
+octopus_flanet_rootfs() {
+    local source_image="$SRCPATH/image/octopus-flanet/Armbian_24.11.0_amlogic_s912_bookworm_6.1.114_server_2024.11.01.img"
+    local target_image="$TMPDIR/rootfs.img"
+    local boot_offset=$((8192 * 512))
+    local rootfs_offset=$((1056768 * 512))
+    local add_size_mb=400
+    local bootfs_loopdev=""
+
+    echo "信息：准备 Octopus-Planet Rootfs..."
+    ensure_dir "$TMPDIR"; ensure_dir "$BOOTFS"
+    cp "$source_image" "$target_image" || { echo "错误：复制 Octopus-Planet 原始镜像失败" >&2; exit 1; }
+
+    echo "信息：挂载 boot 分区并修改 uEnv.txt (使用 VIM2 DTB)..."
+    find_loop_device # 找 loop 给 boot
+    bootfs_loopdev="$LOOPDEV"
+    echo "信息：将 $target_image (偏移 $boot_offset) 关联到 $bootfs_loopdev..."
+    sudo losetup --offset "$boot_offset" "$bootfs_loopdev" "$target_image" || { echo "错误：设置 boot 分区 loop 设备 $bootfs_loopdev 失败" >&2; exit 1; }
+    sudo mount "$bootfs_loopdev" "$BOOTFS" || { echo "错误：挂载 boot 分区 ($bootfs_loopdev) 失败" >&2; exit 1; }
+    BOOTFS_MOUNTED=1
+    sudo sed -i "s/meson-gxm-octopus-planet.dtb/meson-gxm-khadas-vim2.dtb/g" "$BOOTFS/uEnv.txt" || { echo "错误：修改 uEnv.txt 失败" >&2; exit 1; }
+    sudo umount "$BOOTFS" || { echo "警告：卸载 boot 分区 ($BOOTFS) 失败" >&2; BOOTFS_MOUNTED=0; }
+    BOOTFS_MOUNTED=0
+    echo "信息：分离 boot loop 设备 $bootfs_loopdev..."
+    sudo losetup -d "$bootfs_loopdev" || { echo "警告：分离 boot 分区 loop 设备 $bootfs_loopdev 失败" >&2; }
+
+    echo "信息：扩展镜像文件 (${add_size_mb}MB)..."
+    # ... (dd, cat, rm add.img logic) ...
+
+    echo "信息：调整镜像分区大小 (分区 2)..."
+    sudo parted -s "$target_image" resizepart 2 100% || { echo "错误：使用 parted 调整分区 2 大小失败" >&2; exit 1; }
+
+    echo "信息：设置 rootfs 分区的 loop 设备..."
+    find_loop_device # 找 loop 给 rootfs
+    echo "信息：将 $target_image (偏移 $rootfs_offset) 关联到 $LOOPDEV..."
+    sudo losetup --offset "$rootfs_offset" "$LOOPDEV" "$target_image" || { echo "错误：设置 rootfs 分区 loop 设备 $LOOPDEV 失败" >&2; exit 1; }
+
+    echo "信息：检查并调整文件系统大小 (在 loop 设备上)..."
+    sudo e2fsck -f -y "$LOOPDEV" || { echo "警告：e2fsck 检查 $LOOPDEV 失败" >&2; exit 1; }
+    sudo resize2fs "$LOOPDEV" || { echo "错误：resize2fs 调整 $LOOPDEV 大小失败" >&2; exit 1; }
+
+    echo "信息：Octopus-Planet Rootfs 准备完成，loop 设备 $LOOPDEV 已就绪。"
 }
 
-if [ "$1" = "all" ]; then
-    onecloud
-    cumebox2
-    chainedbox
-    vm
-    e900v22c
-    octopus_flanet
-else
-    case $1 in  
-        onecloud)  
-            onecloud
+# --- 特定设备的文件配置函数 ---
+
+config_cumebox2_files() {
+    echo "信息：为 Cumebox2 配置特定文件 (OLED, DTB)..."
+    ensure_dir "$ROOTFS/etc/oled"
+    # 注意 DTB 路径可能需要根据实际 Armbian 版本调整
+    sudo cp "$SRCPATH/image/cumebox2/v-fix.dtb" "$ROOTFS/boot/dtb/amlogic/meson-gxl-s905x-khadas-vim.dtb" || echo "警告：复制 Cumebox2 DTB 失败"
+    sudo cp "$SRCPATH/image/cumebox2/ssd" "$ROOTFS/usr/bin/" || echo "警告：复制 Cumebox2 ssd 脚本失败"
+	sudo chmod +x "$ROOTFS/usr/bin/ssd" || echo "警告：设置 ssd 脚本执行权限失败"
+    sudo cp "$SRCPATH/image/cumebox2/config.json" "$ROOTFS/etc/oled/config.json" || echo "警告：复制 OLED 配置文件失败"
+}
+
+config_octopus_flanet_files() {
+    echo "信息：为 Octopus-Planet 配置特定文件 (model_database.conf)..."
+    sudo cp "$SRCPATH/image/octopus-flanet/model_database.conf" "$ROOTFS/etc/model_database.conf" || echo "警告：复制 model_database.conf 失败"
+}
+
+
+# --- 构建流程函数 ---
+
+build_target() {
+    local target="$1"
+    echo "=================================================="
+    echo "信息：开始构建目标: $target"
+    echo "=================================================="
+
+    # 设置全局变量，供后续函数使用
+    TARGET_DEVICE_NAME="$target"
+    NEED_PREPARE_DNS=false # 默认不需要准备 DNS
+
+    # 1. 准备 Rootfs
+    case "$target" in
+        onecloud)
+            onecloud_rootfs
+            local arch="armhf"
+            local device_type="gpio"
+            local network_type="systemd-networkd"
             ;;
-        cumebox2)  
-            cumebox2
+        cumebox2)
+            cumebox2_rootfs
+            local arch="aarch64"
+            local device_type="video1"
+            local network_type="" # 默认 NetworkManager
+            NEED_PREPARE_DNS=true
             ;;
-        chainedbox) 
-            chainedbox
+        chainedbox)
+            chainedbox_rootfs_and_fix_dtb
+            local arch="aarch64"
+            local device_type="video1"
+            local network_type=""
+            NEED_PREPARE_DNS=true
             ;;
-        vm)  
-            vm
+        vm)
+            vm_rootfs
+            local arch="amd64" # 使用 amd64 触发 x86_64 逻辑
+            local device_type=""
+            local network_type=""
+            NEED_PREPARE_DNS=true
             ;;
-        e900v22c)  
-            e900v22c
+        e900v22c)
+            e900v22c_rootfs
+            local arch="aarch64"
+            local device_type="video1"
+            local network_type=""
+            # E900V22C 原始镜像可能需要换源/DNS
+            NEED_PREPARE_DNS=true
             ;;
-        octopus-flanet)  
-            octopus_flanet
+        octopus-flanet)
+            octopus_flanet_rootfs
+            local arch="aarch64"
+            local device_type="video1"
+            local network_type=""
+            NEED_PREPARE_DNS=true
             ;;
-        *)  
-            echo "Do no thing." 
+        *)
+            echo "错误：未知或不支持的目标 '$target'" >&2
+            exit 1
             ;;
     esac
-fi
\ No newline at end of file
+
+    # 2. 挂载 Rootfs
+    mount_rootfs
+
+    # 3. 安装和配置 KVMD
+    install_and_configure_kvmd "$arch" "$device_type" "$network_type"
+
+    # 4. 写入元数据/主机名
+    write_meta "$target"
+
+    # 5. 清理挂载 (由 trap -> cleanup 自动完成)
+    # 手动触发一次 unmount (如果需要立即 unmount 而不是等脚本结束)
+    # umount_rootfs # 不再需要显式调用
+
+    # 6. 打包镜像
+    case "$target" in
+        onecloud)
+            pack_img_onecloud
+            ;;
+        vm)
+            pack_img "Vm"
+            ;;
+        cumebox2)
+            pack_img "Cumebox2"
+            ;;
+        chainedbox)
+            pack_img "Chainedbox"
+            ;;
+         e900v22c)
+            pack_img "E900v22c"
+            ;;
+         octopus-flanet)
+            pack_img "Octopus-Flanet"
+            ;;
+        *)
+            echo "错误：未知的打包类型 for '$target'" >&2
+            # 不退出，允许后续清理
+            ;;
+    esac
+
+    echo "=================================================="
+    echo "信息：目标 $target 构建完成！"
+    echo "=================================================="
+}
+
+
+# --- 主逻辑 ---
+
+# 检查是否提供了目标参数
+if [ -z "$1" ]; then
+    echo "用法: $0 <target|all>"
+    echo "可用目标: onecloud, cumebox2, chainedbox, vm, e900v22c, octopus-flanet"
+    exit 1
+fi
+
+# 设置脚本立即退出模式
+set -eo pipefail
+
+# 检查必要的外部工具
+for cmd in sudo docker losetup mount umount parted e2fsck resize2fs qemu-img curl tar python3 pip3 rsync git simg2img img2simg dd cat rm mkdir mv cp sed chmod chown ln grep printf id; do
+    if ! command -v "$cmd" &> /dev/null; then
+        echo "错误：必需的命令 '$cmd' 未找到。请安装相应软件包。" >&2
+        exit 1
+    fi
+done
+# 检查特定工具 (如果脚本中使用了)
+if ! command -v "$SRCPATH/image/onecloud/AmlImg_v0.3.1_linux_amd64" &> /dev/null && [[ "$1" == "onecloud" || "$1" == "all" ]]; then
+     if [ -f "$SRCPATH/image/onecloud/AmlImg_v0.3.1_linux_amd64" ]; then
+        echo "信息：找到 AmlImg 工具，尝试设置执行权限..."
+        sudo chmod +x "$SRCPATH/image/onecloud/AmlImg_v0.3.1_linux_amd64" || echo "警告：设置 AmlImg 执行权限失败"
+     else
+        echo "错误：构建 onecloud 需要 '$SRCPATH/image/onecloud/AmlImg_v0.3.1_linux_amd64'，但未找到。" >&2
+        # 不退出，如果是 all 模式，可能其他目标还能构建
+     fi
+fi
+
+
+# 执行构建
+if [ "$1" = "all" ]; then
+    echo "信息：开始构建所有目标..."
+    build_target "onecloud"
+    build_target "cumebox2"
+    build_target "chainedbox"
+    build_target "vm"
+    build_target "e900v22c"
+    build_target "octopus-flanet"
+    echo "信息：所有目标构建完成。"
+else
+    build_target "$1"
+fi
+
+exit 0 # 显式成功退出
\ No newline at end of file