diff --git a/bsp/README.md b/bsp/README.md
index feb1e45c193..87a9b13a62e 100644
--- a/bsp/README.md
+++ b/bsp/README.md
@@ -247,9 +247,9 @@ This document is based on the RT-Thread mainline repository and categorizes the
 
 #### 🟢 QEMU Virt64 AArch64 (RT-Smart) (SMP)
 
-| BSP Name | GPIO | UART | RTC | VirtIO-BLK | VirtIO-NET | VirtIO-Console | VirtIO-GPU | VirtIO-Input |
-|----------|------|------|-----|-----------|-----------|---------------|-----------|--------------|
-| [qemu-virt64-aarch64](qemu-virt64-aarch64) | ✅ | ✅ | ✅ | ✅ | - | ✅ | ✅ | ✅ |
+| BSP Name | GPIO | UART | RTC | WDT | PCI | Storage | Flash | LCD | Input | Camera | Sound | Crypto | RNG | Net | Console | RPMSG |
+|----------|------|------|-----|-----|-----|---------|-------|-----|-------|--------|-------|--------|-----|-----|---------|-------|
+| [qemu-virt64-aarch64](qemu-virt64-aarch64) | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
 
 #### 🟡 Phytium (RT-Smart) (SMP)
 
diff --git a/bsp/qemu-virt64-aarch64/README.md b/bsp/qemu-virt64-aarch64/README.md
index 6811bf30623..8ac4012c1ca 100644
--- a/bsp/qemu-virt64-aarch64/README.md
+++ b/bsp/qemu-virt64-aarch64/README.md
@@ -110,13 +110,13 @@ msh />
 
 | Option | Default | Description |
 | ------ | ------- | ----------- |
-| `-gic` | `2` | GIC version (`2` / `3` / `max`); auto-switches to GICv3 when `-smp` > 8 |
-| `-smp` | `RT_CPUS_NR` (4) | Number of CPU cores |
+| `-gic` | `2` | GIC version (`2` or `3`); auto-switches to GICv3 when `-smp` > 8; with `-el 2` and GICv3, internally uses `gic-version=max` (not passable via `-gic max`) |
+| `-smp` | `RT_CPUS_NR` (4) | Number of CPU cores; `-dtbo` adds 1 automatically when `amp_soc` is present |
 | `-mem` | `128` | Memory size (MB) |
 | `-el` | `1` | Exception level: `1` default; `2` virtualization on; `3` secure on |
 | `-bootargs` | see below | Kernel boot arguments |
 | `-initrd` | — | Path to initrd image |
-| `-graphic` | — | Graphics mode (ramfb + VirtIO keyboard/tablet), replaces `-nographic` |
+| `-graphic` | — | Graphics mode (ramfb + `virtio-gpu-device` + VirtIO keyboard/tablet), replaces `-nographic` |
 | `-gl` | — | In graphics mode, attach `virtio-gpu-gl-pci` and enable QEMU display OpenGL (e.g. `-gl gtk`); requires QEMU built with `--enable-opengl` and `--enable-virglrenderer`; must be used with `-graphic` |
 | `-debug` | — | Enable GDB debugging (`-S -s`) |
 | `-dumpdtb` | — | Export runtime DTB and convert to DTS |
@@ -133,8 +133,8 @@ msh />
 | `-tap` | — | Use TAP networking (replaces user mode) |
 | `-ssh` | `12055` | SSH port forwarding in user mode (`hostfwd=tcp::PORT-:22`) |
 | `-9p` | — | VirtIO 9P shared directory path (enable VirtIO 9P in menuconfig) |
-| `-camera` | — | Enable vfio-user camera (Linux only, with `qemu-device-camera`) |
-| `-dtbo` | — | AMP mode: DTBO overlay file (e.g. `amp.dtsi`); **requires Smart kernel** (see below) |
+| `-camera` | — | Host V4L2 device path (e.g. `/dev/video0`); Linux only; read access to the device is required (usually add your user to the `video` group); compiles and runs `qemu-device-camera.c` in the background, then attaches a vfio-user PCI camera (requires **libvfio-user**) |
+| `-dtbo` | — | AMP mode: DTBO overlay file (e.g. `amp.dtsi`); auto-increments `-smp` by 1; **requires Smart kernel** and sufficient `-mem` (see below) |
 
 Default `-bootargs`:
 
@@ -156,7 +156,7 @@ QEMU OpenGL display (`-gl`):
 ./qemu.py -graphic -gl gtk
 ```
 
-> **Note**: `-gl` enables OpenGL only on the QEMU / host display path (`virtio-gpu-gl-pci` + VirGL). This BSP includes only the VirtIO GPU 2D framebuffer driver by default — **no OpenGL / VirGL user-space or kernel implementation**. `-gl` does not provide OpenGL acceleration inside RT-Thread. For graphics demos, use `-graphic` (`virtio-gpu-device`) with framebuffer / HMI tests.
+> **Note**: `-gl` enables OpenGL only on the QEMU / host display path (`virtio-gpu-gl-pci` + VirGL). This BSP includes only the VirtIO GPU 2D framebuffer driver by default — **no OpenGL / VirGL user-space or kernel implementation**. `-gl` does not provide OpenGL acceleration inside RT-Thread. For graphics demos, use `-graphic` (`virtio-gpu-device`) with the framebuffer stack.
 
 GDB debugging:
 
@@ -194,14 +194,16 @@ AMP heterogeneous multi-core (RPMSG slave, **requires Smart version**):
 ./qemu.py -dtbo amp.dtsi -mem 258
 ```
 
-> **Note**: The AMP demo requires the **Smart version** (`RT_USING_SMART`). The standard kernel must run in a region where virtual addresses equal physical addresses. The slave image is loaded by QEMU `loader` to fixed physical addresses specified in `amp.dtsi` (e.g. `kernel-entry = 0x48480000`), which differs from the primary MMU layout; the standard kernel cannot boot correctly in that address space. Enable Smart in menuconfig and rebuild.
+> **Note**: The AMP demo requires the **Smart version** (`RT_USING_SMART`) and **`-mem 258`** (memory layout in `amp.dtsi` needs ~258 MB; the default 128 MB is insufficient). `qemu.py` also increments `-smp` by 1 when `amp_soc` is detected (e.g. `RT_CPUS_NR=4` runs as `-smp 5`). QEMU `loader` pre-loads `amp.dtb` and `rtthread.bin` to fixed physical addresses in `amp.dtsi` (e.g. `kernel-entry = 0x48480000`); the primary core then starts the slave via PSCI. The standard kernel must run in a region where virtual addresses equal physical addresses and cannot be used for the slave address space. Enable Smart in menuconfig and rebuild.
 
-Camera demo:
+Camera demo (requires `-graphic` for framebuffer overlay in `camera.c`):
 
 ```
-./qemu.py -graphic -camera
+./qemu.py -graphic -camera /dev/video0
 ```
 
+> **Note**: `-camera` requires a host V4L2 device path. On Linux, install [libvfio-user](https://github.com/nutanix/libvfio-user) first (see comments in `qemu-device-camera.c`), and ensure your user can read the device node (e.g. `sudo usermod -aG video $USER` then log in again, or check with `ls -l /dev/video0`). Open will also fail if another program is using the camera. The script compiles `qemu-device-camera.c`, waits for `/tmp/qemu-device-camera.sock`, starts QEMU, and stops the helper process when QEMU exits.
+
 ### 3.6 Console and Telnet
 
 | Port | Device | Description |
@@ -231,7 +233,7 @@ TAP requires bridge setup on Linux; support on Windows / WSL is limited. SSH por
 
 ## 4. Default QEMU Devices
 
-`qemu.py` attaches the following devices by default (matching the default driver configuration):
+`qemu.py` attaches the following devices by default (matching the default driver configuration). UFS is included only when QEMU ≥ 8.2.0; optional devices are listed with their triggering option.
 
 | QEMU device | Image file | RT-Thread side |
 | ----------- | ---------- | -------------- |
@@ -242,7 +244,7 @@ TAP requires bridge setup on Linux; support on Windows / WSL is limited. SSH por
 | `sdhci-pci` + `sd-card` | `emmc.qcow2` | SDHCI eMMC (enable SDIO) |
 | `nvme` + `nvme-ns` | `nvme.qcow2` | NVMe |
 | `ahci` + `ide-hd` | `ahci.qcow2` | AHCI SATA |
-| `ufs` + `ufs-lu` | `ufs.qcow2` | UFS (QEMU ≥ 8.2.0) |
+| `ufs` + `ufs-lu` (QEMU ≥ 8.2.0 only) | `ufs.qcow2` | UFS |
 | `virtio-net-device` | — | VirtIO Net (enable network stack) |
 | `virtio-rng-device` | — | VirtIO RNG |
 | `virtio-crypto-device` | — | VirtIO Crypto |
@@ -256,7 +258,7 @@ TAP requires bridge setup on Linux; support on Windows / WSL is limited. SSH por
 | `virtio-keyboard-device` / `virtio-tablet-device` | — | VirtIO Input |
 | `virtio-9p-device` (`-9p`) | — | VirtIO 9P (enable driver) |
 | `virtio-sound-pci` (`-sound virtio`) | — | VirtIO Sound |
-| vfio-user camera (`-camera`) | — | PCI camera (BSP `drv_camera`) |
+| vfio-user camera (`-camera <path>`) | host V4L2 device | PCI camera (BSP `drv_camera`) |
 
 ## 5. Driver Support
 
@@ -268,7 +270,6 @@ TAP requires bridge setup on Linux; support on Windows / WSL is limited. SSH por
 | Serial | 8250 PCI | PCI serial, Telnet port 4322 |
 | RTC | PL031 | Real-time clock |
 | GPIO | PL061 | GPIO controller |
-| Clock | ARM Architected Timer | System tick |
 | Watchdog | I6300ESB | PCI watchdog |
 | Storage | VirtIO Block | Block device, `vda` |
 | Storage | VirtIO SCSI | SD / CD-ROM |
@@ -278,31 +279,19 @@ TAP requires bridge setup on Linux; support on Windows / WSL is limited. SSH por
 | Storage | MTD NOR CFI | pflash flash |
 | Storage | BLK partition | DFS / EFI partition table |
 | Graphics | VirtIO GPU | 2D framebuffer (`-graphic`); no OpenGL / VirGL |
-| Graphics | Framebuffer | ROMFB 800×600, HMI / graphics tests |
+| Graphics | Framebuffer | RAMFB 800×600 (`RT_GRAPHIC_ROMFB_WIDTH` / `HEIGHT`, `drv_romfb.c`) |
 | Graphics | LCD | LCD device framework |
 | Input | VirtIO Input | Keyboard, mouse, tablet (`-graphic`) |
-| Input | Touchscreen | Touchscreen framework |
-| Input | Keyboard GPIO | GPIO keyboard |
-| Input | Power | Power key |
 | Audio | Intel HDA | Default `-sound hda` |
 | Audio | VirtIO Sound | `-sound virtio` |
-| Crypto | HwCrypto | AES / DES / 3DES / RC4 / RNG |
 | Crypto | VirtIO Crypto | VirtIO crypto backend |
 | Random | VirtIO RNG | Hardware random number |
 | Comm | VirtIO Console | Telnet 4321 |
-| Comm | VirtIO RPROC Serial | Remoteproc serial link |
 | Comm | RPMSG | VirtIO RPMSG character device |
 | Comm | Mailbox PIC | Mailbox interrupt controller |
 | Firmware | QEMU FW_CFG | Firmware configuration interface |
 | Bus | PCI | ECAM + Generic Host, MSI / MSI-X |
-| Bus | VirtIO | MMIO + PCI transport |
-| IRQ | GICv2 / GICv3 / ITS | Including GICv2m MSI |
-| Devicetree | OFW / FDT | Dynamic devicetree parsing |
-| Other | DMA | DMA framework |
-| Other | MFD EDU / SYSCON | Multi-function devices |
-| Other | Regulator / Reset | Power and reset framework |
-| Filesystem | DFS v2 | FatFs, devfs, romfs |
-| POSIX | Partial APIs | poll / select / termios / pipe, etc. |
+| Other | MFD EDU | Multi-function device |
 
 ### 5.2 Not Enabled by Default but Available via menuconfig
 
@@ -313,16 +302,6 @@ This BSP enables the PCI bus and VirtIO PCI transport; QEMU also attaches the co
 | VirtIO Net | `RT_USING_SAL` + `RT_USING_LWIP` + `RT_USING_ETHERNET` | `virtio-net-device` | Network stack not enabled by default |
 | VirtIO 9P | `RT_VIRTIO_9P` | `virtio-9p-device` | Use with `-9p` shared directory |
 | SDHCI eMMC | `RT_USING_SDIO` + `RT_SDIO_SDHCI_PCI` | `sdhci-pci` | eMMC storage |
-| SDIO DW MMC PCI | `RT_USING_SDIO` + `RT_SDIO_DW_MMC_PCI` | — | DesignWare MMC |
-| USB | `RT_USING_USB_HOST` / `RT_USING_USB_DEVICE` | Add USB controller manually | Not a default QEMU device |
-| CherryUSB | `RT_USING_CHERRYUSB` | Add USB controller manually | — |
-| I2C / SPI | `RT_USING_I2C` / `RT_USING_SPI` | Depends on PCI device | Requires matching PCI controller driver |
-| CAN | `RT_USING_CAN` | Depends on PCI device | — |
-| Sensor | `RT_USING_SENSOR` | — | — |
-| WiFi | `RT_USING_WIFI` | — | — |
-| PM | `RT_USING_PM` | — | — |
-| Pinctrl | `RT_USING_PINCTRL` | — | — |
-| SCMI firmware | `RT_FIRMWARE_ARM_SCMI` + `RT_VIRTIO_SCMI` | — | — |
 
 > **Note**: Any PCI driver marked `depends on RT_USING_PCI` can be enabled on this BSP via menuconfig. If QEMU does not attach the device by default, add the corresponding `-device` argument in `qemu.py`.
 
@@ -330,9 +309,8 @@ This BSP enables the PCI bus and VirtIO PCI transport; QEMU also attaches the co
 
 | Feature | Config | Startup |
 | ------- | ------ | ------- |
-| HMI graphics test | `RT_GRAPHIC_HMI_TEST` | `-graphic` (embedded in `drv_romfb.c`) |
-| Graphics draw test | `RT_GRAPHIC_GRAPHIC_TEST` | `-graphic` (embedded in `drv_romfb.c`) |
-| Camera driver | `RT_SOC_CAMERA` | `-graphic -camera` |
+| RAMFB resolution | `RT_GRAPHIC_ROMFB_WIDTH` / `RT_GRAPHIC_ROMFB_HEIGHT` | `-graphic` (`drv_romfb.c` supplies size to RAMFB) |
+| Camera driver | `RT_SOC_CAMERA` | `-camera <v4l2-path>`; use `-graphic` for `camera.c` preview |
 
 ### 5.4 Sample Applications (`applications/`)
 
@@ -348,9 +326,9 @@ Default application. Prints `hello rt-thread` after boot, then enters msh.
 
 **`amp.c` (AMP + RPMSG)**
 
-The primary core boots an additional CPU running RT-Thread as slave via the `amp_soc` devicetree node and PSCI. Primary and slave communicate through VirtIO RPMSG and Mailbox. The primary writes VirtIO Block `vdb` resource info into the slave DTB before starting the slave.
+QEMU `loader` pre-loads `amp.dtb` and `rtthread.bin` to addresses in `amp.dtsi`; the primary core then boots the slave CPU via the `amp_soc` devicetree node and PSCI. Primary and slave communicate through VirtIO RPMSG and Mailbox. The primary writes VirtIO Block `vdb` resource info into the slave DTB before starting the slave. `qemu.py` increments `-smp` by 1 for the extra CPU.
 
-**Version requirement**: **Smart version** required (`RT_USING_SMART` in menuconfig). The standard kernel must run with virtual addresses equal to physical addresses. The AMP slave is loaded by QEMU into a separate physical memory region (see `amp_memory`, `kernel-entry` in `amp.dtsi`), which violates that constraint — the standard kernel cannot be used for AMP demos.
+**Version requirement**: **Smart version** required (`RT_USING_SMART` in menuconfig), plus **`-mem 258`**. The standard kernel must run with virtual addresses equal to physical addresses. The AMP slave is loaded into a separate physical memory region (see `amp_memory`, `kernel-entry` in `amp.dtsi`), which violates that constraint — the standard kernel cannot be used for AMP demos.
 
 Startup:
 
@@ -374,10 +352,10 @@ An RPMSG endpoint is created automatically at boot (`rpmsg_char0`, endpoint name
 
 Centers and overlays the vfio-user PCI camera feed onto the framebuffer. Default devices `camera0` → `fb0`, runs for about 10 seconds (300 frames).
 
-Startup:
+Startup (Linux host with libvfio-user and a V4L2 camera):
 
 ```
-./qemu.py -graphic -camera
+./qemu.py -graphic -camera /dev/video0
 ```
 
 msh command:
@@ -414,9 +392,10 @@ Optional VirtIO device arguments (add in `qemu.py` as needed):
 
 ## 8. Notes After Configuration Changes
 
-- After changing CPU count, adjust `qemu.py -smp` or rebuild so `RT_CPUS_NR` matches.
-- GIC auto-switches to v3 when exceeding 8 cores; or specify `-gic 3` manually.
+- After changing CPU count, rebuild so `RT_CPUS_NR` matches (or override with `qemu.py -smp`). In `-dtbo` mode, `qemu.py` adds 1 to the CPU count automatically.
+- GIC auto-switches to v3 when exceeding 8 cores; or specify `-gic 3` manually. `gic-version=max` is set only internally when using `-el 2` with GICv3.
 - 9P directory sharing requires QEMU built with virtfs (`--enable-virtfs`).
 - `-gl` requires QEMU built with OpenGL / VirGL (`--enable-opengl`, `--enable-virglrenderer`); this is independent of RT-Thread OpenGL support — the default software stack is 2D graphics only.
-- AMP (`-dtbo`) requires Smart kernel (`RT_USING_SMART`); the standard kernel requires VA=PA and cannot be used with slave images loaded at fixed physical addresses.
+- AMP (`-dtbo`) requires Smart kernel (`RT_USING_SMART`), `-mem 258`, and auto-increments CPU count; the standard kernel requires VA=PA and cannot be used with slave images loaded at fixed physical addresses.
+- `-camera` requires a host V4L2 device path, Linux, and libvfio-user; your user must have read access to the device (typically via the `video` group); `-graphic` is needed only for the `camera.c` framebuffer preview demo.
 - Override root device or boot parameters via `-bootargs`.
diff --git a/bsp/qemu-virt64-aarch64/README_zh.md b/bsp/qemu-virt64-aarch64/README_zh.md
index 7cdcaaf73ca..0b623627d83 100644
--- a/bsp/qemu-virt64-aarch64/README_zh.md
+++ b/bsp/qemu-virt64-aarch64/README_zh.md
@@ -106,13 +106,13 @@ msh />
 
 | 参数 | 默认值 | 说明 |
 | ---- | ------ | ---- |
-| `-gic` | `2` | GIC 版本（`2` / `3` / `max`）；`-smp` 大于 8 时自动切换为 GICv3 |
-| `-smp` | `RT_CPUS_NR`（4） | CPU 核心数 |
+| `-gic` | `2` | GIC 版本（`2` 或 `3`）；`-smp` 大于 8 时自动切换为 GICv3；`-el 2` 且 GICv3 时内部使用 `gic-version=max`（不可通过 `-gic max` 手动指定） |
+| `-smp` | `RT_CPUS_NR`（4） | CPU 核心数；`-dtbo` 检测到 `amp_soc` 时自动加 1 |
 | `-mem` | `128` | 内存大小（MB） |
 | `-el` | `1` | 异常级别：`1` 默认；`2` 开启虚拟化；`3` 开启 secure |
 | `-bootargs` | 见下文 | 内核启动参数 |
 | `-initrd` | — | 指定 initrd 镜像路径 |
-| `-graphic` | — | 图形模式（ramfb + VirtIO 键盘/平板），替代 `-nographic` |
+| `-graphic` | — | 图形模式（ramfb + `virtio-gpu-device` + VirtIO 键盘/平板），替代 `-nographic` |
 | `-gl` | — | 图形模式下挂载 `virtio-gpu-gl-pci` 并开启 QEMU 显示端 OpenGL（如 `-gl gtk`）；需 QEMU 带 `--enable-opengl`、`--enable-virglrenderer`，且须与 `-graphic` 合用 |
 | `-debug` | — | 启用 GDB 调试（`-S -s`） |
 | `-dumpdtb` | — | 导出运行时 DTB 并转为 DTS |
@@ -129,8 +129,8 @@ msh />
 | `-tap` | — | 使用 TAP 网卡（替代 user 模式） |
 | `-ssh` | `12055` | user 模式下 SSH 端口转发（`hostfwd=tcp::PORT-:22`） |
 | `-9p` | — | VirtIO 9P 共享目录路径（需在 menuconfig 中启用 VirtIO 9P） |
-| `-camera` | — | 启用 vfio-user 摄像头（仅 Linux，配合 `qemu-device-camera`） |
-| `-dtbo` | — | AMP 模式：指定 DTBO overlay 文件（如 `amp.dtsi`）；**须 Smart 版本内核**（见下文说明） |
+| `-camera` | — | 宿主机 V4L2 设备路径（如 `/dev/video0`）；仅 Linux；需对设备有读权限（通常加入 `video` 组）；后台编译运行 `qemu-device-camera.c` 并通过 vfio-user 挂载 PCI 摄像头（需安装 **libvfio-user**） |
+| `-dtbo` | — | AMP 模式：指定 DTBO overlay 文件（如 `amp.dtsi`）；自动将 `-smp` 加 1；**须 Smart 版本内核**且配置足够 `-mem`（见下文） |
 
 默认 `-bootargs`：
 
@@ -152,7 +152,7 @@ QEMU OpenGL 显示（`-gl`）：
 ./qemu.py -graphic -gl gtk
 ```
 
-> **说明**：`-gl` 仅在 QEMU / 宿主机显示链路侧启用 OpenGL（`virtio-gpu-gl-pci` + VirGL）。本 BSP 默认仅包含 VirtIO GPU 2D 帧缓冲驱动，**没有 OpenGL / VirGL 用户态与内核侧实现**，因此 `-gl` 不会带来 RT-Thread 内的 OpenGL 加速；图形演示请使用 `-graphic`（`virtio-gpu-device`）配合 framebuffer / HMI 测试。
+> **说明**：`-gl` 仅在 QEMU / 宿主机显示链路侧启用 OpenGL（`virtio-gpu-gl-pci` + VirGL）。本 BSP 默认仅包含 VirtIO GPU 2D 帧缓冲驱动，**没有 OpenGL / VirGL 用户态与内核侧实现**，因此 `-gl` 不会带来 RT-Thread 内的 OpenGL 加速；图形演示请使用 `-graphic`（`virtio-gpu-device`）配合 framebuffer 栈。
 
 GDB 调试：
 
@@ -190,14 +190,16 @@ AMP 异构多核（RPMSG 从核，**须 Smart 版本**）：
 ./qemu.py -dtbo amp.dtsi -mem 258
 ```
 
-> **说明**：AMP 示例要求启用 **Smart 版本**（`RT_USING_SMART`）。标准版内核要求在虚拟地址与物理地址相等的区域运行；从核镜像由 QEMU `loader` 加载到 `amp.dtsi` 指定的固定物理地址（如 `kernel-entry = 0x48480000`），与主核 MMU 映射布局不同，标准版无法在该地址空间正常启动。请在 menuconfig 中开启 Smart 后重新编译。
+> **说明**：AMP 示例要求启用 **Smart 版本**（`RT_USING_SMART`），且必须使用 **`-mem 258`**（`amp.dtsi` 内存布局约需 258 MB，默认 128 MB 不足）。`qemu.py` 检测到 `amp_soc` 时会将 `-smp` 自动加 1（例如 `RT_CPUS_NR=4` 时实际以 `-smp 5` 启动）。QEMU `loader` 会预先将 `amp.dtb` 与 `rtthread.bin` 加载到 `amp.dtsi` 指定的固定物理地址（如 `kernel-entry = 0x48480000`），主核再通过 PSCI 拉起从核。标准版内核要求在虚拟地址与物理地址相等的区域运行，无法用于从核地址空间。请在 menuconfig 中开启 Smart 后重新编译。
 
-摄像头演示：
+摄像头演示（`camera.c` 叠加画面到 framebuffer，需配合 `-graphic`）：
 
 ```
-./qemu.py -graphic -camera
+./qemu.py -graphic -camera /dev/video0
 ```
 
+> **说明**：`-camera` 必须传入宿主机 V4L2 设备路径。Linux 下需先安装 [libvfio-user](https://github.com/nutanix/libvfio-user)（详见 `qemu-device-camera.c` 注释），并确保当前用户对设备节点有读权限（例如 `sudo usermod -aG video $USER` 后重新登录，或用 `ls -l /dev/video0` 确认权限）。若设备正被其他程序占用，也会打开失败。脚本会编译 `qemu-device-camera.c`、等待 `/tmp/qemu-device-camera.sock` 就绪后启动 QEMU，并在 QEMU 退出后结束辅助进程。
+
 ### 3.6 控制台与 Telnet
 
 | 端口 | 设备 | 说明 |
@@ -227,7 +229,7 @@ Linux 下需预先配置 TAP 网桥，Windows / WSL 下 TAP 支持有限。TAP 
 
 ## 4. QEMU 默认挂载设备
 
-`qemu.py` 默认向虚拟机挂载以下设备（与驱动默认配置对应）：
+`qemu.py` 默认向虚拟机挂载以下设备（与驱动默认配置对应）。UFS 仅在 QEMU ≥ 8.2.0 时挂载；带触发条件的可选设备在表中标注。
 
 | QEMU 设备 | 镜像文件 | RT-Thread 侧 |
 | --------- | -------- | ------------ |
@@ -238,7 +240,7 @@ Linux 下需预先配置 TAP 网桥，Windows / WSL 下 TAP 支持有限。TAP 
 | `sdhci-pci` + `sd-card` | `emmc.qcow2` | SDHCI eMMC（需启用 SDIO） |
 | `nvme` + `nvme-ns` | `nvme.qcow2` | NVMe |
 | `ahci` + `ide-hd` | `ahci.qcow2` | AHCI SATA |
-| `ufs` + `ufs-lu` | `ufs.qcow2` | UFS（QEMU ≥ 8.2.0） |
+| `ufs` + `ufs-lu`（仅 QEMU ≥ 8.2.0） | `ufs.qcow2` | UFS |
 | `virtio-net-device` | — | VirtIO Net（需启用网络协议栈） |
 | `virtio-rng-device` | — | VirtIO RNG |
 | `virtio-crypto-device` | — | VirtIO Crypto |
@@ -252,7 +254,7 @@ Linux 下需预先配置 TAP 网桥，Windows / WSL 下 TAP 支持有限。TAP 
 | `virtio-keyboard-device` / `virtio-tablet-device` | — | VirtIO Input |
 | `virtio-9p-device`（`-9p`） | — | VirtIO 9P（需启用驱动） |
 | `virtio-sound-pci`（`-sound virtio`） | — | VirtIO Sound |
-| vfio-user 摄像头（`-camera`） | — | PCI 摄像头（BSP `drv_camera`） |
+| vfio-user 摄像头（`-camera <路径>`） | 宿主机 V4L2 设备 | PCI 摄像头（BSP `drv_camera`） |
 
 ## 5. 驱动支持情况
 
@@ -264,7 +266,6 @@ Linux 下需预先配置 TAP 网桥，Windows / WSL 下 TAP 支持有限。TAP 
 | 串口 | 8250 PCI | PCI 串口，Telnet 端口 4322 |
 | RTC | PL031 | 实时时钟 |
 | GPIO | PL061 | GPIO 控制器 |
-| 时钟 | ARM Architected Timer | 系统 tick |
 | 看门狗 | I6300ESB | PCI 看门狗 |
 | 存储 | VirtIO Block | 块设备，`vda` |
 | 存储 | VirtIO SCSI | SD / CD-ROM |
@@ -274,31 +275,19 @@ Linux 下需预先配置 TAP 网桥，Windows / WSL 下 TAP 支持有限。TAP 
 | 存储 | MTD NOR CFI | pflash 闪存 |
 | 存储 | BLK 分区 | DFS / EFI 分区表 |
 | 图形 | VirtIO GPU | 2D 帧缓冲（`-graphic`）；不含 OpenGL / VirGL |
-| 图形 | Framebuffer | ROMFB 800×600，含 HMI / 图形测试 |
+| 图形 | Framebuffer | RAMFB 800×600（`RT_GRAPHIC_ROMFB_WIDTH` / `HEIGHT`，`drv_romfb.c`） |
 | 图形 | LCD | LCD 设备框架 |
 | 输入 | VirtIO Input | 键盘、鼠标、平板（`-graphic`） |
-| 输入 | Touchscreen | 触摸屏框架 |
-| 输入 | Keyboard GPIO | GPIO 键盘 |
-| 输入 | Power | 电源键 |
 | 音频 | Intel HDA | 默认 `-sound hda` |
 | 音频 | VirtIO Sound | `-sound virtio` |
-| 加密 | HwCrypto | AES / DES / 3DES / RC4 / RNG |
 | 加密 | VirtIO Crypto | VirtIO 加密后端 |
 | 随机数 | VirtIO RNG | 硬件随机数 |
 | 通信 | VirtIO Console | Telnet 4321 |
-| 通信 | VirtIO RPROC Serial | Remoteproc 串口链路 |
 | 通信 | RPMSG | VirtIO RPMSG 字符设备 |
 | 通信 | Mailbox PIC | 邮箱中断控制器 |
 | 固件 | QEMU FW_CFG | 固件配置接口 |
 | 总线 | PCI | ECAM + Generic Host，MSI / MSI-X |
-| 总线 | VirtIO | MMIO + PCI 传输 |
-| 中断 | GICv2 / GICv3 / ITS | 含 GICv2m MSI |
-| 设备树 | OFW / FDT | 动态设备树解析 |
-| 其他 | DMA | DMA 框架 |
-| 其他 | MFD EDU / SYSCON | 多功能设备 |
-| 其他 | Regulator / Reset | 电源与复位框架 |
-| 文件系统 | DFS v2 | FatFs、devfs、romfs |
-| POSIX | 部分接口 | poll / select / termios / pipe 等 |
+| 其他 | MFD EDU | 多功能设备 |
 
 ### 5.2 默认未启用但可通过 menuconfig 开启
 
@@ -309,16 +298,6 @@ Linux 下需预先配置 TAP 网桥，Windows / WSL 下 TAP 支持有限。TAP 
 | VirtIO Net | `RT_USING_SAL` + `RT_USING_LWIP` + `RT_USING_ETHERNET` | `virtio-net-device` | 网络协议栈未默认启用 |
 | VirtIO 9P | `RT_VIRTIO_9P` | `virtio-9p-device` | 配合 `-9p` 共享目录 |
 | SDHCI eMMC | `RT_USING_SDIO` + `RT_SDIO_SDHCI_PCI` | `sdhci-pci` | eMMC 存储 |
-| SDIO DW MMC PCI | `RT_USING_SDIO` + `RT_SDIO_DW_MMC_PCI` | — | DesignWare MMC |
-| USB | `RT_USING_USB_HOST` / `RT_USING_USB_DEVICE` | 需自行添加 USB 控制器 | 非 QEMU 默认设备 |
-| CherryUSB | `RT_USING_CHERRYUSB` | 需自行添加 USB 控制器 | — |
-| I2C / SPI | `RT_USING_I2C` / `RT_USING_SPI` | 视具体 PCI 设备 | 需对应 PCI 控制器驱动 |
-| CAN | `RT_USING_CAN` | 视具体 PCI 设备 | — |
-| Sensor | `RT_USING_SENSOR` | — | — |
-| WiFi | `RT_USING_WIFI` | — | — |
-| PM 电源管理 | `RT_USING_PM` | — | — |
-| Pinctrl | `RT_USING_PINCTRL` | — | — |
-| SCMI 固件 | `RT_FIRMWARE_ARM_SCMI` + `RT_VIRTIO_SCMI` | — | — |
 
 > **说明**：凡标注 `depends on RT_USING_PCI` 的 PCI 类驱动，均可在此 BSP 上通过 menuconfig 启用；QEMU 侧若未挂载对应设备，需自行在 `qemu.py` 中添加 `-device` 参数。
 
@@ -326,9 +305,8 @@ Linux 下需预先配置 TAP 网桥，Windows / WSL 下 TAP 支持有限。TAP 
 
 | 功能 | 配置项 | 启动方式 |
 | ---- | ------ | -------- |
-| HMI 图形测试 | `RT_GRAPHIC_HMI_TEST` | `-graphic`（`drv_romfb.c` 内嵌） |
-| 图形绘制测试 | `RT_GRAPHIC_GRAPHIC_TEST` | `-graphic`（`drv_romfb.c` 内嵌） |
-| 摄像头驱动 | `RT_SOC_CAMERA` | `-graphic -camera` |
+| RAMFB 分辨率 | `RT_GRAPHIC_ROMFB_WIDTH` / `RT_GRAPHIC_ROMFB_HEIGHT` | `-graphic`（`drv_romfb.c` 向 RAMFB 提供尺寸） |
+| 摄像头驱动 | `RT_SOC_CAMERA` | `-camera <v4l2路径>`；`camera.c` 预览需加 `-graphic` |
 
 ### 5.4 示例应用（`applications/`）
 
@@ -344,9 +322,9 @@ Linux 下需预先配置 TAP 网桥，Windows / WSL 下 TAP 支持有限。TAP 
 
 **`amp.c`（AMP + RPMSG）**
 
-主核通过设备树 `amp_soc` 节点，以 PSCI 拉起额外一颗 CPU 运行从核 RT-Thread，主从之间通过 VirtIO RPMSG + Mailbox 通信。主核将 VirtIO Block `vdb` 的资源信息写入从核 DTB 后启动从核。
+QEMU `loader` 预先将 `amp.dtb` 与 `rtthread.bin` 加载到 `amp.dtsi` 指定地址，主核再通过 `amp_soc` 设备树节点与 PSCI 拉起从核 CPU。主从之间通过 VirtIO RPMSG + Mailbox 通信。主核将 VirtIO Block `vdb` 的资源信息写入从核 DTB 后启动从核。`qemu.py` 会为这颗额外 CPU 将 `-smp` 加 1。
 
-**版本要求**：须 **Smart 版本**（menuconfig 中 `RT_USING_SMART`）。标准版内核必须在虚拟地址与物理地址一致的映射下运行；AMP 从核由 QEMU 加载到独立物理内存区域（见 `amp.dtsi` 中 `amp_memory`、`kernel-entry` 等），不满足该约束，因此标准版无法用于 AMP 演示。
+**版本要求**：须 **Smart 版本**（menuconfig 中 `RT_USING_SMART`），且须使用 **`-mem 258`**。标准版内核必须在虚拟地址与物理地址一致的映射下运行；AMP 从核加载到独立物理内存区域（见 `amp.dtsi` 中 `amp_memory`、`kernel-entry` 等），不满足该约束，因此标准版无法用于 AMP 演示。
 
 启动方式：
 
@@ -370,10 +348,10 @@ msh 命令：
 
 将 vfio-user PCI 摄像头画面居中叠加到 framebuffer，默认设备 `camera0` → `fb0`，约 10 秒（300 帧）。
 
-启动方式：
+启动方式（Linux 宿主机，已安装 libvfio-user 且存在 V4L2 摄像头）：
 
 ```
-./qemu.py -graphic -camera
+./qemu.py -graphic -camera /dev/video0
 ```
 
 msh 命令：
@@ -410,9 +388,10 @@ VirtIO 设备可选参数（在 `qemu.py` 中按需添加）：
 
 ## 8. 修改配置后的注意事项
 
-- 修改 CPU 数量后，同步调整 `qemu.py -smp` 或重新编译使 `RT_CPUS_NR` 与之一致。
-- 超过 8 核时 GIC 自动切换为 v3，也可手动指定 `-gic 3`。
+- 修改 CPU 数量后，重新编译使 `RT_CPUS_NR` 与之一致（或用 `qemu.py -smp` 覆盖）。`-dtbo` 模式下 `qemu.py` 会自动将 CPU 数加 1。
+- 超过 8 核时 GIC 自动切换为 v3，也可手动指定 `-gic 3`。`gic-version=max` 仅在 `-el 2` 且 GICv3 时由脚本内部设置。
 - 启用 9P 目录共享需 QEMU 编译时包含 virtfs（`--enable-virtfs`）。
 - 使用 `-gl` 需 QEMU 编译时包含 OpenGL / VirGL（`--enable-opengl`、`--enable-virglrenderer`）；这与 RT-Thread 是否支持 OpenGL 无关，当前软件栈默认仅 2D 图形。
-- AMP（`-dtbo`）须 Smart 版本内核（`RT_USING_SMART`）；标准版要求 VA=PA 运行，无法用于从核固定物理地址加载场景。
+- AMP（`-dtbo`）须 Smart 版本内核（`RT_USING_SMART`）、`-mem 258`，且自动增加 CPU 数；标准版要求 VA=PA 运行，无法用于从核固定物理地址加载场景。
+- `-camera` 须传入宿主机 V4L2 设备路径，仅支持 Linux，且需 libvfio-user；当前用户须能访问该设备（通常加入 `video` 组）；`-graphic` 仅 `camera.c` framebuffer 预览演示需要。
 - 修改根设备或启动参数时，通过 `-bootargs` 覆盖默认参数。