From 3701da40df87a204094ef0087f4e43df67d05e0a Mon Sep 17 00:00:00 2001 From: liufumingren Date: Fri, 25 Sep 2026 17:05:42 +0800 Subject: [PATCH] =?UTF-8?q?chore(scripts):=20=E6=96=B0=E5=A2=9E=E7=8E=B0?= =?UTF-8?q?=E5=9C=BA=E5=8F=96=E8=AF=81=E4=B8=8E=E7=94=B5=E6=BA=90=E5=8E=8B?= =?UTF-8?q?=E6=B5=8B=E8=84=9A=E6=9C=AC?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 现场"扫描枪收不到按键"需要区分"按键没到内核"与"到了内核但上层丢了", 以及区分"供电不足/外设取电"与"枪线材问题",补三个可直接复用的工具: - scripts/scan-forensics.ps1:故障当场一键取证(设备/root/uptime、/proc/bus/input/devices、 dmesg usb/hid、App 侧 logcat(WMSScan/WMSPage/UniConsole)、dumpsys input、内核 getevent 抓窗), 并自动输出判据:内核有按键而 App keys=0 -> 上层(Android 分发/WebView);内核无按键 -> 枪/线/供电。 - scripts/power-probe.sh:采样 SARADC / 温度 / 扫码枪是否仍在总线上(RK818 不上报电流, 这是唯一可用的侧面遥测)。 - scripts/power-load-test.sh:固定负载(屏幕最亮 + N 核忙循环)压测并在结束后自动杀掉负载。 实测结论(X9R 12V/2A 24W 临界、音响插平板 USB 口会冲掉 Hub + 枪)见 README 机型注意一节。 --- scripts/power-load-test.sh | 42 +++++++++++++ scripts/power-probe.sh | 30 ++++++++++ scripts/scan-forensics.ps1 | 120 +++++++++++++++++++++++++++++++++++++ 3 files changed, 192 insertions(+) create mode 100644 scripts/power-load-test.sh create mode 100644 scripts/power-probe.sh create mode 100644 scripts/scan-forensics.ps1 diff --git a/scripts/power-load-test.sh b/scripts/power-load-test.sh new file mode 100644 index 0000000..78a4781 --- /dev/null +++ b/scripts/power-load-test.sh @@ -0,0 +1,42 @@ +#!/system/bin/sh +# power-load-test.sh - put the X9R under a reproducible load while sampling the power picture. +# +# Load = screen brightness max + N CPU busy loops (+ whatever the app/scanner does concurrently). +# Sampling = SARADC channels / thermals / CPU freq / whether the USB scanner is still enumerated. +# The busy loops are killed at the end (never left running) and brightness is restored by the caller. +# +# usage: power-load-test.sh [seconds] [cpus] [outfile] +SECS=${1:-120} +CPUS=${2:-4} +OUT=${3:-/data/local/tmp/pwr_load.txt} +DEV=/sys/bus/iio/devices/iio:device0 + +PIDS="" +i=0 +while [ "$i" -lt "$CPUS" ]; do + (while :; do :; done) & + PIDS="$PIDS $!" + i=$((i + 1)) +done + +: > "$OUT" +n=0 +while [ "$n" -lt "$SECS" ]; do + u=$(cat /proc/uptime); u=${u%% *} + a0=$(cat $DEV/in_voltage0_raw 2>/dev/null) + a1=$(cat $DEV/in_voltage1_raw 2>/dev/null) + a2=$(cat $DEV/in_voltage2_raw 2>/dev/null) + t0=$(cat /sys/class/thermal/thermal_zone0/temp 2>/dev/null) + t1=$(cat /sys/class/thermal/thermal_zone1/temp 2>/dev/null) + f0=$(cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_cur_freq 2>/dev/null) + f3=$(cat /sys/devices/system/cpu/cpu3/cpufreq/scaling_cur_freq 2>/dev/null) + gun=$(grep -c Newtologic /proc/bus/input/devices 2>/dev/null) + echo "$u $a0 $a1 $a2 $t0 $t1 $f0 $f3 $gun" >> "$OUT" + n=$((n + 1)) + sleep 0.5 +done + +for p in $PIDS; do kill -9 "$p" 2>/dev/null; done +sleep 1 +echo "done $SECS samples -> $OUT" +echo "load pids killed:$PIDS" diff --git a/scripts/power-probe.sh b/scripts/power-probe.sh new file mode 100644 index 0000000..870ab9f --- /dev/null +++ b/scripts/power-probe.sh @@ -0,0 +1,30 @@ +#!/system/bin/sh +# power-probe.sh - sample the power picture on the RK3288 X9R (needs adb root) +# +# Why: on 2026-09-25 the field unit behaved badly on a 12V/2A adapter and fine on 12V/2.5A, +# which points at input brown-out. The OS has no "rated power" field and the RK818 driver +# reports 0mA for every rail, so the only usable telemetry is: +# - SARADC channels (ff100000.saradc, 10-bit, 1.7578125 mV/LSB) +# - SoC / GPU thermal +# - whether the USB scanner is still enumerated (brown-out can drop the HID device) +# Run it once on each adapter and compare (min/max/mean + gun=0 count). +# +# usage: pwr_sample.sh [samples] [outfile] (0.5 s per sample) +N=${1:-120} +OUT=${2:-/data/local/tmp/pwr_sample.txt} +DEV=/sys/bus/iio/devices/iio:device0 +i=0 +: > "$OUT" +while [ "$i" -lt "$N" ]; do + u=$(cat /proc/uptime); u=${u%% *} + a0=$(cat $DEV/in_voltage0_raw 2>/dev/null) + a1=$(cat $DEV/in_voltage1_raw 2>/dev/null) + a2=$(cat $DEV/in_voltage2_raw 2>/dev/null) + t0=$(cat /sys/class/thermal/thermal_zone0/temp 2>/dev/null) + t1=$(cat /sys/class/thermal/thermal_zone1/temp 2>/dev/null) + gun=$(grep -c Newtologic /proc/bus/input/devices 2>/dev/null) + echo "$u $a0 $a1 $a2 $t0 $t1 $gun" >> "$OUT" + i=$((i + 1)) + sleep 0.5 +done +echo "done $N samples -> $OUT" diff --git a/scripts/scan-forensics.ps1 b/scripts/scan-forensics.ps1 new file mode 100644 index 0000000..d80a59b --- /dev/null +++ b/scripts/scan-forensics.ps1 @@ -0,0 +1,120 @@ +# scan-forensics.ps1 +# One-shot evidence capture for the "scanner stops working" fault (WMS X9R tablet). +# +# Why: on 2026-09-25 the field log proved the break is BELOW the JS layer +# (60s heartbeat keys=0 while every JS gate was green). The only thing JS cannot see is +# whether the kernel/Activity still gets the key events. This script captures that layer. +# +# Usage (tablet plugged in, USB debugging on): +# powershell -NoProfile -ExecutionPolicy Bypass -File scripts\scan-forensics.ps1 +# powershell -NoProfile -ExecutionPolicy Bypass -File scripts\scan-forensics.ps1 -Seconds 30 +# +# Run it WHILE the fault is present, and scan the gun 3-5 times during the capture window. +# Everything is kept (no cleanup) under -OutDir. +# +# Verdict rules: +# kernel key DOWN > 0 but app heartbeat keys=0 -> Android dispatch / WebView / IME layer +# kernel key DOWN = 0 -> gun / cable / USB port (hardware) or scan service +# +# ASCII only on purpose: `powershell -File` reads non-ASCII scripts as ANSI on this machine. + +param( + [int]$Seconds = 20, + [string]$OutDir = '' +) + +$ErrorActionPreference = 'Continue' + +function Find-Adb { + $candidates = @( + (Join-Path $env:LOCALAPPDATA 'Android\Sdk\platform-tools\adb.exe'), + 'C:\Android\Sdk\platform-tools\adb.exe', + 'D:\Android\Sdk\platform-tools\adb.exe' + ) + foreach ($c in $candidates) { + if (Test-Path $c) { return $c } + } + $cmd = Get-Command adb -ErrorAction SilentlyContinue + if ($cmd) { return $cmd.Source } + throw 'adb.exe not found. Set PATH or edit Find-Adb in this script.' +} + +if (-not $OutDir) { + $OutDir = Join-Path $env:TEMP ('scan-forensics\' + (Get-Date -Format 'yyyyMMdd-HHmmss')) +} +New-Item -ItemType Directory -Force -Path $OutDir | Out-Null + +$adb = Find-Adb +Write-Host "adb = $adb" +Write-Host "out dir = $OutDir" +Write-Host '' + +function Sh($cmd) { & $adb shell $cmd 2>&1 } +function Save($name, $lines) { + $path = Join-Path $OutDir $name + $lines | Out-File -FilePath $path -Encoding utf8 + Write-Host (" saved {0}" -f $name) +} + +# 1) device present? +$devices = & $adb devices 2>&1 +$serial = ($devices | Select-String -Pattern '^\S+\s+device$' | ForEach-Object { $_.Line.Split()[0] }) | Select-Object -First 1 +if (-not $serial) { + Write-Host '!! no adb device. Check: USB debugging on, try another USB port (this box only works on some ports), replug.' -ForegroundColor Red + Write-Host ($devices -join "`n") + exit 2 +} +Write-Host "device = $serial" +& $adb root 2>&1 | Out-Null +& $adb wait-for-device 2>&1 | Out-Null + +# 2) static environment +Save '00-device.txt' @( + "serial = $serial", + "captured = $(Get-Date -Format 'yyyy-MM-dd HH:mm:ss')", + "id = $(Sh 'id')", + "model = $(Sh 'getprop ro.product.model')", + "android = $(Sh 'getprop ro.build.version.release') build=$(Sh 'getprop ro.build.type') selinux=$(Sh 'getenforce')", + "uptime = $(Sh 'cat /proc/uptime') (seconds since boot - tells whether a reboot just happened)", + "pkg = $(Sh 'dumpsys window | grep -E ''mCurrentFocus|mFocusedApp''')" +) + +Save '01-input-devices.txt' (Sh 'cat /proc/bus/input/devices') +Save '02-dmesg-usb-input.txt' (Sh "dmesg | grep -iE 'usb|newtologic|hid|input:' | tail -n 120") +Save '03-app-logcat.txt' (Sh "logcat -d -s WMSScan:V WMSPage:V UniConsole:V UniViewConsole:V | tail -n 400") +Save '04-dumpsys-input.txt' (Sh 'dumpsys input | head -n 250') + +# 3) kernel key stream (scan now!) +$gevPath = Join-Path $OutDir '05-getevent.txt' +Write-Host '' +Write-Host ("==> SCAN THE GUN 3-5 TIMES NOW (you have {0} seconds) <==" -f $Seconds) -ForegroundColor Yellow +& $adb shell "timeout $Seconds getevent -lt" 2>&1 | Out-File -FilePath $gevPath -Encoding utf8 +Write-Host (" saved 05-getevent.txt") + +# 4) verdict +$gev = Get-Content $gevPath -ErrorAction SilentlyContinue +$keyDown = @($gev | Select-String -Pattern 'EV_KEY\s+(KEY_[0-9A-Z_]+)\s+DOWN' -AllMatches) +$enterDown = @($gev | Select-String -Pattern 'EV_KEY\s+KEY_ENTER\s+DOWN') +$gunLines = @($gev | Select-String -Pattern 'event\d+.*EV_KEY') +$logcat = Get-Content (Join-Path $OutDir '03-app-logcat.txt') -ErrorAction SilentlyContinue +$health = @($logcat | Select-String -Pattern 'health.*keys=' | Select-Object -Last 1) + +Write-Host '' +Write-Host '================ VERDICT ================' +Write-Host ("kernel : key DOWN events = {0} (ENTER = {1})" -f $keyDown.Count, $enterDown.Count) +Write-Host (" gun/kernel EV_KEY lines = {0}" -f $gunLines.Count) +if ($health.Count -gt 0) { + Write-Host ("app : last heartbeat -> {0}" -f $health[0].Line.Trim()) +} else { + Write-Host 'app : no WMSScan heartbeat in logcat buffer (buffer may have wrapped)' +} +Write-Host '' +if ($keyDown.Count -ge 3) { + Write-Host 'kernel HAS key events -> the break is ABOVE the kernel (Android dispatch / WebView / IME / app).' -ForegroundColor Green + Write-Host ' Next: compare with the app heartbeat keys= above. keys=0 here means the app never got them.' +} else { + Write-Host 'kernel has NO (or almost no) key events -> gun / cable / USB port / scan service (hardware).' -ForegroundColor Red + Write-Host ' Next: swap cable+port, watch dmesg in 02-dmesg-usb-input.txt for disconnect/reconnect.' +} +Write-Host '' +Write-Host 'Keep this folder for the vendor/customer report. Nothing was deleted.'