Flash tab — Inspect Flash 烧录面板 — Inspect

fastboot

ready就绪

The connected agent advertises fastboot and no job is running. 已连接的 agent 报告了 fastboot 能力,当前没有作业在跑。

agent bench-01
last job flash cfg · ok · 0.14s
uart COM27 @ 115200
readers 1

image

The file is uploaded to the hub workspace first, then flashed from there — so re-writing the same image costs no second upload. 文件先传到 hub 的工作区,再从那里烧 —— 同一个镜像重烧一次不用再传一遍。

Drop a file here 把文件拖到这里

vendor_cfg.bin 1.2 KB · sha256 3f9a…c1

write

The agent verifies the digest before it touches the device — a damaged transfer costs a retry, not a half-written partition. agent 会在动设备之前先核对摘要 —— 传坏了的代价是重传一次,不是写了一半的分区。

Writing vendor_cfg cannot be undone from here, and the board may not come back to report it. Click flash again to confirm. 写 vendor_cfg 在这里撤不回来,板子也可能起不来告诉你结果。再点一次"烧录"确认。

leave fastboot

A board in fastboot has dropped off adb. This is the way back — without it, recovering needs someone at the machine. 进了 fastboot 的板子已经从 adb 里消失。这是回去的路 —— 没有它,恢复就得有人到机器跟前。

bench USB

bench-01 · 7 devices

What the bench host itself sees on USB, and which mode each board is in. When fastboot lists nothing, this tells a missing driver from a cable that is not on this host. 台架那台机器自己在 USB 上看到了什么,以及每块板子处在什么模式。fastboot 列表为空时,靠这里分清是没装驱动,还是线根本没插在这台机器上。

fastboot 18d1:d00d · serial 0123456789 · Android no driver没装驱动 A board is in fastboot but no driver is bound on the bench host — install the Android bootloader (WinUSB) driver there; fastboot cannot see it until then. 有块板子在 fastboot,但台架机器上没有绑定驱动 —— 在那台机器上装 Android bootloader(WinUSB)驱动,装好之前 fastboot 看不到它。

6 other devices (hub, keyboard, serial adapter …) — show all 另有 6 个设备(集线器、键盘、串口转接器……)—— 全部显示

package

flash plan

Flash a whole package by its manifest. The plan is checked as a whole first — any problem and not a single row runs — and what runs is exactly the plan you confirmed: an image changed since is refused before it is written. 按烧录清单烧整包。先整份检查——有任何问题就一行都不跑;跑的就是你确认的那份计划:确认之后镜像被换掉,会在写入之前被拦下。

31 rows行 28 run执行 3 skipped跳过 2.36 GB fingerprint指纹 0361bb78…

#rowdecision决定 bytesmd5status状态
1setvar model board-xskip (setvar)—
2flash abl_a ← abl.elfskip (abl_a)—
4reboot-bootloaderrunok · 9.8s
13flash boot_a ← boot.imgrun100663296314f71f4…ok · 3.1s
23flash super ← super.imgrun22360214489c02e1aa…1.2 / 2.2 GB
30erase miscrun…
+ 25 more rows另外 25 行
This runs 28 rows on the board, userdata included. It cannot be undone from here. Click "flash package" again to confirm. 这会在板子上执行 28 行,包括 userdata。在这里撤不回来。再点一次「烧整包」确认。

board config

/dev/block/by-name/<cfg> · 2097152 bytes

What the partition holds right now, read back from the booted board. Flashing reports that fastboot believed it wrote; this is the only place that shows the bytes. The partition is found by content — the one whose head parses as KEY="VALUE" — not by a name, which differs per product. Needs root, and the board in Android. 板子上这个分区现在实际是什么,从已启动的系统读回来的。烧录报告的是 fastboot 认为它写了;只有这里能看到字节。分区靠内容认——头部能解析成 KEY="VALUE" 的那个——不靠名字,名字每个产品都不一样。需要 root,且板子已进系统。

keyvalue
MAIN_BOARDV03
PORT_BOARDV03
BASEBAND81
DUAL_SIMTRUE
PNBOARD-X-0AW-RF0-11AU

timeline

workspace/devices/…/flash/flash-vendor_cfg-2026-08-07T19-22-04/

Job progress and the board's UART on one clock — the point of watching is being able to line them up. 作业进度和板子的 UART 用同一个时钟 —— 看 UART 的意义就在于能把两者对上。

wrote cfg ← cfg-RESTORE-V03.ini (941 bytes) 0.14s

fastboot reported the write succeeded. The partition was not read back — if the contents matter, check on the board after it boots. fastboot 报告写入成功。没有回读校验——内容要紧的话,等板子起来后去板上验一下。

0.00jobaccepted — receiving 1248 bytes
0.14jobtransfer 1248/1248 bytes (100%)
0.21jobstarting fastboot flash vendor_cfg
0.68uartfastboot: processing commands
1.02jobSending 'vendor_cfg' (1 KB)
1.44uartmmc0: writing partition vendor_cfg
1.90jobWriting 'vendor_cfg' OKAY