当出现系统memory不足导致高优先级进程被kill等异常时，需要抓取一段时间内的
memory使用情况.
一. 在使用Mali GPU平台上(MT6572 MT6582 MT6592)使用如下脚本抓取.
具体测试手法如下：
1. 先将手机时间设置成和PC一样，清除mtklog，开启mobilelog.
2. 双击此脚本。
3. 出现问题后，停止脚本(Ctrl+C)将mobilelog和D:\MemLog一起打包
脚本内容：
@echo off
set logpath1=D:\MemLog\procrank.txt
set logpath2=D:\MemLog\GPU_memory_usage.txt
set logpath4=D:\MemLog\meminfo.txt
set logpath5=D:\MemLog\librank.txt
set logdir= D:\MemLog
del %logdir%
:loop
echo %date% %time%
echo %date% %time% >> %logpath1%
adb shell procrank -u >> %logpath1%
echo %date% %time% >> %logpath2%
adb shell cat /proc/mali/memory_usage >> %logpath2%
echo %date% %time% >> %logpath4%
adb shell cat proc/meminfo >> %logpath4%
echo %date% %time% >> %logpath5%
adb shell librank >> %logpath5%
echo "wait 5s to next catch"
ping -n 5 127.0.0.1>nul
goto loop
二. 在非Mali GPU平台上使用如下脚本：
@echo off
set logpath1=D:\MemLog\procrank.txt
set logpath4=D:\MemLog\meminfo.txt
set logdir= D:\MemLog
del %logdir%
:loop
echo %date% %time%
echo %date% %time% >> %logpath1%
adb shell procrank -u >> %logpath1%
echo %date% %time% >> %logpath4%
adb shell cat proc/meminfo >> %logpath4%
echo "wait 5s to next catch"
ping -n 5 127.0.0.1>nul
goto loop
三. 当确认是某个AP进程占用memory多，需要再监控对应进程时.
在Mali GPU平台使用如下脚本。
双击后输入对应进程的PID后开始抓取
@echo "please input the target pid :"
@echo off
adb shell ps
set processid=
set /p processid=Please Input process id to catch:
@echo off
set logpath1=D:\MemLog\procrank.txt
set logpath2=D:\MemLog\GPU_memory_usage.txt
set logpath3=D:\MemLog\Mem-%processid%.txt
set logpath4=D:\MemLog\meminfo.txt
set logpath5=D:\MemLog\librank.txt
set logdir= D:\MemLog
del %logdir%
:loop
echo %date% %time%
echo %date% %time% >> %logpath1%
adb shell procrank -u >> %logpath1%
echo %date% %time% >> %logpath2%
adb shell cat /proc/mali/memory_usage >> %logpath2%
echo %date% %time% >> %logpath3%
adb shell dumpsys meminfo %processid% >> %logpath3%
echo %date% %time% >> %logpath4%
adb shell cat proc/meminfo >> %logpath4%
echo %date% %time% >> %logpath5%
adb shell librank >> %logpath5%
echo "wait 5s to next catch"
ping -n 5 127.0.0.1>nul
goto loop
四. 当确认是某个AP进程占用memory多时，需要再监控对应进程时.
在非Mali GPU平台使用如下脚本。
双击后输入对应进程的PID后开始抓取
@echo "please input the target pid :"
@echo off
adb shell ps
set processid=
set /p processid=Please Input process id to catch:
@echo off
set logpath1=D:\MemLog\procrank.txt
set logpath3=D:\MemLog\Mem-%processid%.txt
set logpath4=D:\MemLog\meminfo.txt
set logdir= D:\MemLog
del %logdir%
:loop
echo %date% %time%
echo %date% %time% >> %logpath1%
adb shell procrank -u >> %logpath1%
echo %date% %time% >> %logpath3%
adb shell dumpsys meminfo %processid% >> %logpath3%
echo %date% %time% >> %logpath4%
adb shell cat proc/meminfo >> %logpath4%
echo "wait 5s to next catch"
ping -n 5 127.0.0.1>nul
goto loop
原文：http://bbs.16rd.com/thread-54279-1-148.html