• 安装
  参考本小节参考文档中的安装指南
  
  首先是 UEFI 还是 BIOS 的问题，因为这个可能会设计到后面分区的问题 BIOS+GPT+GRUB 大约 1M 就够了，通常我们会给 boot 分大约 200M，
  但是如果是 UEFI 推荐引导就超过 256M ，因此最好先看一看自己系统是不是只支持 UEFI ，像苹果那种，我之前就踩了这个坑。
  
  有两条建议，还有就是在进行安装的时候建议在安装阶段就把必要的像网络管理工具和编辑工具安装好，比如说我就非常不适应 vi 因此还是先把 vim 装上吧。
  另外就是可以升级了之后再退出安装。这样会避免一些不必要的麻烦。
• 下载/升级源
  
  搜索可用源 https://archlinux.org/mirrorlist/
  
  官方维护的源状态 https://archlinux.org/mirrors/status/
• 安装问题1
  
  描述：如果你在安装或者升级的时候遇到类似这种错误，error: PKGNAME: signature from “USERNAME <EMAIL_NAME>” is unknown trust
  
  如：error: libspeech: signature from “Alexander Epaneshikov <email@alex19ep.me>” is unknown trust
  
  八成是因为你本地密钥库破坏，或者很久没有更新导致丢失某些密钥导致，网上有推荐的一种方法是设置不对包进行任何签名校验，很显然这样有非常大的安全风险。
  
  如果你不愿意一个一个慢慢的修复，一个方法是执行下面步骤中的 step4, 但是偶尔也有失败的情况，因此一步到位的方法是将下列 step1 – step4 都操作一遍。
  
  step1: 删除本地缓存签名：sudo rm -rf /etc/pacman.d/gnupg/*
  
  step2: 重新初始化密钥库： sudo pacman-key --init
  
  step3: 重新添加密钥：sudo pacman-key --populate archlinux
  
  step4: 再次刷新一遍: sudo pacman-key --refresh-keys, 这一步骤可能会花费你几分钟到几十分钟的时间
• 如何登陆 Windows 远程管理界面
  
  安装 freerdp sudo pacman -S freerdp （ ubuntu 安装 sudo apt install freerdp2-x11 ）
  
  运行以下命令 xfreerdp /v:1.2.3.4 /u:Administrator /p:PASSWORD
• 安装 deepin-screenshot-copy
  
  因为最新的 deepin-screen-recorder 出现了一些不兼容的情况，请参考 https://wiki.archlinux.org/index.php/Screen_capture，
  因此在这里我们需要安装 https://aur.archlinux.org/packages/deepin-screenshot-copy-patch/
  
  在命令行执行以下脚本安装

  cd /tmp/; git clone https://aur.archlinux.org/deepin-screenshot-copy-patch.git; cd /tmp/deepin-screenshot-copy-patch; makepkg -si
      

1. Getting and installing Arch(其中介绍了怎么制作安装介质)
2. Arch Linux 安装指南[2016.01]
3. 文件系统，以及支持文件系统所需要安装的包

• 查找软件安装位置: dpkg -S
• 以及安装了那些软件: dpkg -L
• ubuntu 包管理

  查看当前安装的包dpkg -l,卸载软件时卸载之依赖包（如果此依赖包还被其他依赖，将不被卸载）,请使用 aptitude 命令（其中 remove 是删除软件以及其依赖，purge，删除软件，以及其依赖，以及其配置文件）, apt-get 的 autoremove 参数也可以用来卸载当前指定包以及其依赖。但真心不好用，参考一下 aptitude , pacman 就知道。不过 ubuntu 上的包资源很全面。

• 升级更新
  同 ubuntu apt-get update; apt-get upgrade
• Kali 源
  
  在 /etc/apt/source-list 中添加以下两行

  deb http://mirrors.ustc.edu.cn/kali kali-rolling main non-free contrib
  deb-src http://mirrors.ustc.edu.cn/kali kali-rolling main non-free contrib
      

• virtualbox 全屏问题
  执行以下代码修复 apt-get update; apt-get install -y virtualbox-guest-x11; reboot, 参考 此处
• 分区规划
  
  如果跟分区与 home 分区分开，建议至少保证根分区大于 11G 否则后续更新会比较吃力，我自己的是 20G, 安装在 virtualbox 上，应付日常开发足矣。

• 官网

• 清华源( Debian 10( buster ) )

  # 编辑 `/etc/apt/sources.list` 文件，删除原文件所有内容，用以下内容取代：
  deb http://mirrors.tuna.tsinghua.edu.cn/raspbian/raspbian/ buster main non-free contrib rpi
  deb-src http://mirrors.tuna.tsinghua.edu.cn/raspbian/raspbian/ buster main non-free contrib rpi
  
  # 编辑 `/etc/apt/sources.list.d/raspi.list` 文件，删除原文件所有内容，用以下内容取代：
  deb http://mirrors.tuna.tsinghua.edu.cn/raspberrypi/ buster main ui
  

• 使能 Camera
  在终端运行下面这条命令: sudo raspi-config，然后顺着下面的步骤设置就可以了： Select : 5 Interfacing Options Configure connections to peripherals –> Select : P1 Camera Enable/Disable connection to the Raspberry Pi Camera –> Select yes –> sudo reboot
  
  检查是否安装成功执行以下命令： raspistill -o image.jpg, 该命令会在当前目录保存一个从摄像头撷取的一张图片，并保存到 image.jpg 中。
• 安装了 Camera 之后 无法发现 /dev/video 节点
  
  如果你按照上面的步骤打开了树梅派的 Camera 功能，那么你可能在 /dev/目录下找不到 video[x] 节点。我查到的原因是树莓派中的camera module是放在/boot/目录下以固件的形式加载的，不是一个标准的v4l2的摄像头ko驱动。所以加载起来之后会找不到/dev/video[x] 的设备节点，这是因为这个驱动是在底层的，v4l2这个驱动框架还没有加载，所以要在 /etc/modules-load.d/modules.conf 里面添加一行 bcm2835-v4l2，然后重启，就可以了。（这句话意思是在系统启动之后会加载这个文件中模块名，加载模块会在 /lib/modules 中。）

  # sudo vim /etc/modules-load.d/modules.conf
  
      # 在文尾添加下面这行
      bcm2835-v4l2
  
  # find /lib/modules/ -name "*bcm2835-v4l2*"
  
      /lib/modules/4.9.35+/kernel/drivers/media/platform/bcm2835/bcm2835-v4l2.ko
      /lib/modules/4.9.35-v7+/kernel/drivers/media/platform/bcm2835/bcm2835-v4l2.ko
  
  

• 对 exfat 支持
  安装 apt-get install exfat-fuse
• 对 ntfs 支持
  
  安装 apt-get install ntfs-3g
• raspberry pi 发射热点
  参考： https://www.raspberrypi.org/documentation/configuration/wireless/access-point-routed.md
• 引脚图
  
  Ref: https://pinout.xyz/
  
  
• GPIO 电器特性
  
  Ref: https://www.tomshardware.com/reviews/raspberry-pi-gpio-pinout,6122.html
  
  The best thing about any Raspberry Pi, including the new Raspberry Pi 4, is that you can use it to build all kinds of awesome contraptions, from robots to retro gaming consoles and fart detectors. Most of the sensors, motors, lights and other peripherals that make these projects possible connect to the Pi’s set of GPIO (General Purpose Input Output) pins. These pins offer a direct connection to the System on Chip (SoC) at the heart of the Pi, enabling the Pi to communicate with external components. Every Pi model since the Raspberry Pi B+ has had 40 GPIO pins, though on the Pi Zero and Zero W, you have 40 holes that you can solder pins or wires into.
  
  This guide has been updated to reflect the new capabilities of the Raspberry Pi 4, which still comes with 40 GPIO pins, but has a few extra I2C, SPI and UART connections available.
  
  No matter what you’re building, you need to know the Raspberry Pi GPIO pinout, the map and explanation of what each pin can do. While some pins provide electricity, others are grounds and still others connect to different kinds of interfaces, all of which we explain below.
  • General Purpose Input Output (GPIO) Pins
    
    The GPIO is the most basic, yet accessible aspect of the Raspberry Pi. GPIO pins are digital which means they can have two states, off or on. They can have a direction to receive or send current (input, output respectively) and we can control the state and direction of the pins using programming languages such as Python, JavaScript, node-RED etc.
    
    The operating voltage of the GPIO pins is 3.3v with a maximum current draw of 16mA. This means that we can safely power one or two LEDs (Light Emitting Diodes) from a single GPIO pin, via a resistor. But for anything requiring more current, a DC motor for example, we will need to use external components to ensure that we do not damage the GPIO.
    
    Controlling a GPIO pin with Python is accomplished by first importing a library of pre-written code. The most common library is RPi.GPIO (https://pypi.org/project/RPi.GPIO/) and it has been used to create thousands of projects since the early days of the Raspberry Pi. In more recent times a new library called GPIO Zero (https://pypi.org/project/gpiozero/)has been introduced, offering an easier entry for those new to Python and basic electronics. Both of these libraries come pre-installed with the Raspbian operating system.
    
    GPIO pins have multiple names; the first most obvious reference is their “physical” location on the GPIO. Starting at the top left of the GPIO, and by that we mean the pin nearest to where the micro SD card is inserted, we have physical pin 1 which provides 3v3 power. To the right of that pin is physical pin 2 which provides 5v power. The pin numbers then increase as we move down each column, with pin 1 going to pin 3, 5,7 etc until we reach pin 39. You will quickly see that each pin from 1 to 39 in this column follows an odd number sequence. And for the column starting with pin 2 it will go 4,6,8 etc until it reaches 40. Following an even number sequence. Physical pin numbering is the most basic way to locate a pin, but many of the tutorials written for the Raspberry Pi follow a different numbering sequence.
    
    Broadcom (BCM) pin numbering (aka GPIO pin numbering) seems to be chaotic to the average user. With GPIO17, 22 and 27 following on from each other with little thought to logical numbering. The BCM pin mapping refers to the GPIO pins that have been directly connected to the System on a Chip (SoC) of the Raspberry Pi. In essence we have direct links to the brain of our Pi to connect sensors and components for use in our projects.
    
    You will see the majority of Raspberry Pi tutorials using this reference and that is because it is the officially supported pin numbering scheme from the Raspberry Pi Foundation. So it is best practice to start using and learning the BCM pin numbering scheme as it will become second nature to you over time. Also note that BCM and GPIO pin numbering refer to the same scheme. So for example GPIO17 is the same as BCM17.
    
    Certain GPIO pins also have alternate functions that allow them to interface with different kinds of devices that use the I2C, SPI or UART protocols. For example GPIO3 and GPIO 4 are also SDA and SCL I2C pins used to connect devices using the I2C protocol. To use these pins with these protocols we need to enable the interfaces using the Raspberry Pi Configuration application found in the Raspbian OS, Preferences menu.
  • I2C, SPI and UART: Which Do You Use?
    
    We’ll get into the specific differences between I2C, SPI and UART below, but if you’re wondering which one you need to use to connect to given device, the short answer is to check the spec sheet. For example, one tiny LED screen might require SPI and another might use I2C (almost nothing uses UART). If you read the documentation that comes with a product (provided it has some), it will usually tell you which Pi pins to use.
    
    For Raspberry Pi 4 users note that there are now many more I2C, SPI and UART pins available to you. These extra interfaces are activated using device tree overlays and can provide four extra SPI, I2C and UART connections.
  • I2C – Inter-Integrated Circuit
    
    I2C is a low speed two wire serial protocol to connect devices using the I2C standard. Devices using the I2C standard have a master slave relationship. There can be more than one master, but each slave device requires a unique address, obtained by the manufacturer from NXP, formerly known as Philips Semiconductors. This means that we can talk to multiple devices on a single I2C connection as each device is unique and discoverable by the user and the computer using Linux commands such as i2cdetect.
    
    As mentioned earlier I2C has two connections: SDA and SCL. They work by sending data to and from the SDA connection, with the speed controlled via the SCL pin. I2C is a quick and easy way to add many different components, such as LCD / OLED screens, temperature sensors and analog to digital converters for use with photoresistors etc to you project. While proving to be a little more tricky to understand than standard GPIO pins, the knowledge gained from learning I2C will serve you well as you will understand how to connect higher precision sensors for use in the field.
    
    The Raspberry Pi has two I2C connections at GPIO 2 and 3 (SDA and SCL) are for I2C0 (master) and physical pins 27 and 28 are I2C pins that enable the Pi to talk to compatible HAT (Hardware Attached on Top) add on boards.
  • SPI – Serial Peripheral Interface
    
    SPI is another protocol for connecting compatible devices to your Raspberry Pi. It is similar to I2C in that there is a master slave relationship between the Raspberry Pi and the devices connected to it.
    
    Typically SPI is used to send data over short distances between microcontrollers and components such as shift registers, sensors and even an SD card. Data is synchronised using a clock (SCLK at GPIO11) from the master (our Pi) and the data is sent from the Pi to our SPI component using the MOSI (GPIO GPIO10) pin. MOSI stands for Master Out Slave In. If the component needs to reply to our Pi, then it will send data back using the MISO pin (GPIO9) which stands for Master In Slave Out.
  • UART – Universal Asynchronous Receiver / Transmitter
    
    Commonly known as “Serial,” the UART pins (Transmit GPIO14, Receive GPIO15) provide a console / terminal login for headless setup, which means connecting to the Pi without a keyboard or pointing device. Normally, the easiest way to do a headless Raspberry Pi setup is simply to control the Pi over a network or direct USB connection (in the case of Pi Zero).
    
    But, if there’s no network connection, you can also control a headless Pi using a serial cable or USB to serial board from a computer running a terminal console. UART is exceptionally reliable and provides access to a Pi without the need for extra equipment. Just remember to enable the Serial Console in the Raspberry Pi Configuration application. Chances are that you won’t want to do this, but the UART support is there if you need it.
  • Ground (gnd)
    
    Ground is commonly referred to as GND, gnd or – but they all mean the same thing. GND is where all voltages can be measured from and it also completes an electrical circuit. It is our zero point and by connecting a component, such as an LED to a power source and ground the component becomes part of the circuit and current will flow through the LED and produce light.
    
    When building circuits it is always wise to make your ground connections first before applying any power as it will prevent any issues with sensitive components. The Raspberry Pi has eight ground connections along the GPIO and each of these ground pins connects to one single ground connection. So the choice of which ground pin to use is determined by personal preference, or convenience when connecting components.
  • 5v
    
    The 5v pins give direct access to the 5v supply coming from your mains adaptor, less power than used by the Raspberry Pi itself. A Pi can be powered directly from these pins, and it can also power other 5v devices. When using these pins directly, be careful and check your voltages before making a connection because they bypass any safety features, such as the voltage regulator and fuse which are there to protect your Pi. Bypass these with a higher voltage and you could render your Pi inoperable.
  • 3v3
    
    The 3v pin is there to offer a stable 3.3v supply to power components and to test LEDs. In reality, it will be rare that you factor this pin into a build, but it does have a special use. When connecting an LED to the GPIO, we first need to make sure that the LED is wired up correctly and that it lights up. By connecting the long leg of the LED, the anode to the 3.3v pin via a resistor, and the shorter leg, the cathode to any of the Ground (gnd) pins we can check that our LED lights up and is working. This eliminates a hardware fault from the project and enables us to start building our project with confidence.

• [1]raspberry pi 应用程序仓库
• [2]树莓派raspbian系统下，如何开启摄像头/camera模块，并使用简单的命令
• [3]Camkit ( 使用C语言写成，包含了从：图像采集–>色彩转换–>H264编码–>RTP打包–>网络发送的全套接口 )
• 清华大学镜像站
• Raspberry Pi schema 下载
• Raspberry Pi 2/3/4 相关 datesheet layout 下载

• init system 历史
  
  参考： https://blog.darknedgy.net/technology/2015/09/05/0/
• systemctl 一些基本的使用方法
  
  service_name 代指服务的名称，其中 service 可是是具体的 demon （*.service），也可以是挂载点 （*.mount），套接口 （*.socket）
  • systemctl start service_name –> 开启服务
  • systemctl stop service_name –> 关闭服务
  • systemctl restart service_name –> 重启服务
  • systemctl status service_name –> 查看服务的运行状态
  • systemctl enable service_name –> 将服务设置为启动
  • systemctl disable service_name –> 将服务取消开机启动
  • systemctl is-enabled( is-active, is-failed ） service_name –> 查看服务是否启用（是否为激活/运行状态，是否出现错误）
  • systemctl show service_name –> 查看服务的配置详细信息
  • systemctl list-units –> 列出所有运行中单元
  • systemctl list-unit-files –> 列举所有(包括运行和未运行的)的单元
  • systemctl --failed –> 列举所有运行失败的单元
  • systemctl mask( unmask ) service_name –> 屏蔽（显示）服务
  • systemctl kill service_name –> 杀死服务
  • systemctl show -p CPUShares service_name –> 获取当前服务的 cpu 分配额
  • systemctl set-property service_name CPUShares=2000 –> 将某个服务的 CPU 分配额设置成 2000 （一般默认是 1024）
  • systemctl list-dependencies service_name –> 查看某个服务的依赖性
  • systemd-analyze critical-chain sevice_name –> 查看某个服务的关键链（如果不知道这是什么最好的办法就是动手测试，并 google ）
  • systemd-cgls –> 按照等级，列举出控制组
  • systemd-cgtop –> 按CPU、内存、输入和输出列出控制组
  • systemctl rescue –> 进入系统救援模式
  • systemctl emergency –> 进入系统紧急模式
  • systemctl get-default –> 列举出当前的运行等级
  • systemctl isolate runlevel5.target –> 启动运行等级（5，图形模式）（PS：以此类推）
  • systemctl set-default runlevel(x:3,5...).target –> 设置默认运行等级
  • systemctl reboot –> 重启
  • systemctl halt –> 停止
  • systemctl suspend –> 挂起
  • systemctl hibernate –> 休眠（PS：休眠和混合睡眠都要求系统拥有 swap 分区，而且大小与内存一样大）
  • systemctl hybrid-sleep –> 混合睡眠
• 一些其他的资料
  采访 Systemd 和 PulseAudio 创始人 Lennart
  systemd、upstart和system V
  SystemV和BSD的区别 V

下面介绍一些简单的方法，可以在特定情况下提升命令行的体验，尤其是在服务器多人协作场景下：

• 防止某些 samba 客户端终端开太多连接
  
  使用以下方法限制用户连接数

  /etc/samba/smb.conf
  
  [share]
     comment = A Shared Directory
     path = /home/share
     valid users = share
     public = no
     writable = yes
     max connections = 3
  

• 限制 ssh 客户端开太多连接
  可在文件 /etc/security/limits.conf 中添加类似如下限制：

  限制 User1 的并发为 10
  User1      hard    maxlogins>      5
  
  例子限制所有用户并发为 5
  *      hard    maxlogins>      5
  

• 提升 ssh 终端常规命令优先级，并限制某些特殊程序优先级
  这里我们可能需要创建一个 systemd 的配置来帮助我们实现
  
  首先我们实现一个监控脚本：nicemonitor.sh

  #!/bin/bash
  
  start_server(){
  while true
  do
          # 将特殊线程，以及编译线程优先级调低
          ps -aux | grep -E "vscode|netdata|make|cmake|build|javac" | awk '{print $2}' | while read P
          do
                  renice +5 ${P};
          done
  
          # 把 ssh 服务，以及用户常见长时应用优先级提高
          ps -aux | grep -E "ssh" | awk '{print $2}' | while read P
          do
                  renice -5 ${P};
          done
  
          if [ -f /tmp/xxxxxxroottask ]; then
                  rm /tmp/xxxxxxroottask && sync
                  break;
          fi
          sleep 10
  done
  }
  
  if [ "$1" = "start" ]; then
          start_server
  elif [ "$1" = "shutdown" ]; then
          touch /tmp/xxxxxxroottask
  fi
  

  准备 systend 配置脚本：/etc/systemd/system/nicemonitor.service

  [Unit]
  Description=Ensure system stable runtime
  After=network.target
  
  [Service]
  User=root
  Group=root
  ExecStart=/root/bin/roottask.sh start
  ExecStop=/root/bin/roottask.sh shutdown
  Restart=always
  
  [Install]
  WantedBy=multi-user.target
  

以下是我司方便给用户开账户的脚本，实际使用中用户密码不应该这么简陋， 谨慎参考 ：

#!/bin/bash

echo "Please input user name:"
read NAME

IS_EXIST=$(cat /etc/passwd | grep "${NAME}")

if [ "${IS_EXIST}" = "" ]; then
    echo "USERNAME: ${NAME}"
else
    echo "This man is exist, please select another one"
    echo "/etc/passwd -> ${IS_EXIST}"
    exit -1
fi

echo "Please input select user base home dir:"
echo "1. /home"
echo "2. /home2"
echo "3. /home3"
echo "4. /home4"
read HOME_DIR

case ${HOME_DIR} in
1|/home)
    HOME_DIR="/home"
    ;;
2|/home2)
    HOME_DIR="/home2"
    ;;
3|/home3)
    HOME_DIR="/home3"
    ;;
4|/home4)
    HOME_DIR="/home4"
    ;;
*)
    echo "Not this option"
    exit -1
    ;;
esac

SMB_CONFIG_CONTENT="[${NAME}]\n   comment = A Shared Directory\n   path = ${HOME_DIR}/${NAME}\n   valid users = ${NAME}\n   public = no\n   writable = yes\n   max connections = 10\n"

echo "---"
echo "USER --> ${NAME}"
echo "PASSWORD --> ${NAME}"
echo "HOME_DIR --> ${HOME_DIR}"
echo "SAMBA PASSWORD --> ${NAME}"

echo "--- SMB CONFIG"
echo -e "${SMB_CONFIG_CONTENT}"

useradd -d ${HOME_DIR}/${NAME} -m -p $(openssl passwd -1 ${NAME}) -s /bin/bash ${NAME}
smbpasswd -a ${NAME} -w ${NAME}
echo -e "${SMB_CONFIG_CONTENT}" >> /etc/samba/smb.conf
systemctl  restart smbd

解决 mount nfs 的时候出现 mount: wrong fs type, bad option, bad superblock on **** 的问题。

安装 nfs-utils 。

1. 安装截屏工具 pacman -S scrot
2. 安装 xev 或者 evtest pacman -S xorg-xev
3. 运行 xev 并且按 PrtSrc 你可能会收到如下信息, Print 则是该按键的名字

   KeyPress event, serial 33, synthetic NO, window 0x2400001,
   root 0x1a1, subw 0x0, time 58322539, (691,875), root:(692,916),
   state 0x0, keycode 107 (keysym 0xff61, Print), same_screen YES,
   XLookupString gives 0 bytes: 
   XmbLookupString gives 0 bytes: 
   XFilterEvent returns: False
       

4. 如果你有上述报错信息，则在 awesome 的配置文件( ~/.config/awesome/rc.lua )中配置该功能即可:

   263 globalkeys = gears.table.join(
   ...
   383     awful.key({ }, "Print", function (),
   384         awful.util.spawn("scrot -e 'mv $f ~/Pictures/ 2> /dev/null'") end)
   ...
       

以下是我的 ～/.zshrc 的配置

export ZSH=/home/mojies/.oh-my-zsh
ZSH_THEME=mikeh
plugins=(git)
source $ZSH/oh-my-zsh.sh

alias vi='vim -O'
alias CP='rsync -avh --progress -v --links'
alias gitd='git difftool'
alias rm.origin='/usr/bin/rm'
alias rm='trash-put'
alias rm.ls='trash-list'
alias rm.rm='trash-rm'
alias rm.recovery='trash-restore'
alias rm.empty='trash-empty'
alias g++11='g++ -std=c++11'
alias g++14='g++ -std=c++14'
alias g++17='g++ -std=c++17'
alias g++20='g++ -std=c++2a'
alias top='glances'
alias AS='android-studio'
alias df='duf'

# awsome need pull up startx first
isrunning startx; if [ $? -ne 0 ]; then startx;  fi

export WORKON_HOME='~/.virtualenvs'
source /usr/bin/virtualenvwrapper.sh

一个在命令行自动补全命令的工具, 建议终端开 256 色

借用他们的原话是：
Powerlevel10k is a theme for Zsh. It emphasizes speed, flexibility and out-of-the-box experience.

给大家展示一下效果图：

1. oh-my-zsh 仓库地址
2. oh-my-zsh 官网
3. zsh-autosuggestions

make menuconfig之后修改下列内容。

    File systems  --->
        [*] Network File Systems  --->
            [*]   NFS client support
            [ ]     NFS client support for NFS version 3
            [ ]     NFS client support for NFS version 4
            [*]   Root file system on NFS
    [*] Networking support  --->
        Networking options  --->
        [*]   IP: kernel level autoconfiguration
        [*]     IP: DHCP support
        [*]     IP: BOOTP support
        [*]     IP: RARP support

1. 系统调用
   系统调用调用号设置表 ./arch/x86/entry/syscalls/syscall_64.tbl
   
   系统函数原型声明 include/linux/syscalls.h
   
   定义函数原型的时候可以在大部分内核文件中，但是一般会放在 kernel 目录下，如 ./kernel/sys.c
   
   其中用于辅助申明系统调用的宏也在 include/linux/syscalls.h 中，如 #define SYSCALL_DEFINE1(name, ...) SYSCALL_DEFINEx(1, _##name, __VA_ARGS__)
2. 触发进程调度的原因
   
   其一，为了保证所有进程可以得到公平调度，CPU时间被划分为一段段的时间片，这些时间片再被轮流分配给各个进程。这样，当某个进程的时间片耗尽了，就会被系统挂起，切换到其它正在等待CPU的进程运行。
   
   其二，进程在系统资源不足（比如内存不足）时，要等到资源满足后才可以运行，这个时候进程也会被挂起，并由系统调度其他进程运行。
   
   其三，当进程通过睡眠函数sleep这样的方法将自己主动挂起时，自然也会重新调度。
   
   其四，当有优先级更高的进程运行时，为了保证高优先级进程的运行，当前进程会被挂起，由高优先级进程来运行。
   
   最后一个，发生硬件中断时，CPU上的进程会被中断挂起，转而执行内核中的中断服务程序。
   
   vmstat 中的 in interrupt 字段为周期内全局中断次数，而 cs context switch 则为周期内上下文切换的次数（包括进程和线程）
   
   如果想观察进程的上下文切换次数则可以使用 pidstat -w， 其中 cswch 表示自愿上下文切换次数（voluntary context switches）
   而 nvcswch 为非自愿上下文切换（non voluntary context switches）的次数。

3. 配置内核

       Device Drivers  --->
       Character devices-->
       [*] Legacy (BSD) PTY support
       (256) Maximum number of legacy PTY in use
   

4. 配置 busybox 1.23.0

       Busybox Settings  --->
           General Configuration  --->
               [*] Use the devpts filesystem for Unix98 PTYs
   
       Login/Password Management Utilities  --->
           [*] login
           [*]   Run logged in session in a child process
           [*]   Support for login scripts
           [*]   Support for /etc/nologin
           [*]   Support for /etc/securetty
   
       Networking Utilities  --->
           [*] inetd
           [*]   Support echo service
           [*]   Support discard service
           [*]   Support time service
           [*]   Support daytime service
           [*]   Support chargen service
           [*]   Support RPC services
   
           [*] telnet
           [*]   Pass TERM type to remote host
           [*]   Pass USER type to remote host
           [*] telnetd
           [*]   Support standalone telnetd (not inetd only)
           [*]     Support -w SEC option (inetd wait mode)
   

5. 在 /etc/init.d/rcS 中加入下面这几行

       echo -e '\npts/0\npts/1\npts/2\npts/3\npts/4\npts/5\npts/6\npts/7\n' > /etc/securetty; cat /etc/securetty
       echo 'telnet 23/tcp' > /etc/services
       echo 'telnet stream tcp nowait root /sbin/telnetd' > telnetd
       telnetd
   

6. Q&A and smoe tips, notes:

       注意：（一下某些注意的项因博主在配置时已经存在，或者不需要，因此没有给出详细的配置方法，需要自行查找解决方法）
       1. 取消busybox telnet 登录时的密码问题:
       配置 busybox
       Login/Password Management Utilities  --->
       [ ] login
   
       2. /etc 目录下下面的文件必不可少
       fstab
       service
       inetd.conf
       passwd
   
       3. fstab 中需要自动挂载 /dev/pts
       none /dev/pts devpts mode=0622 0 0
   
       4. /dev/ptmx 设备节点需要存在
   
       5. 遇到配置不过的时候建议借助 syslogd
       先在 busybox 配置支持
       然后运行 syslogd -n -m 0 -L & 即可将log 信息打印到串口终端
   

1. 打开内核中的两个宏：

       step one:
       # meke menuconfig
       General setup  --->
           -*- Configure standard kernel features (expert users)  --->
               [*]   Enable ELF core dumps
   
       step two:
       # make menuconfig
       Executable file formats  --->
           [*] Kernel support for ELF binaries
           [*] Write ELF core dumps with partial segments
   

2. 在启动的时候需要执行 ulimit -c unlimited 以接触内核创建文件大小的限制。
3. 制定 coredump 生成文件路径

   一般来说 core 这个生成文件与可执行文件在同一个位置（使用 cat /proc/sys/kernel/core_pattern 可以查看/设置 coredump 文件的生成位置），但是当程序中用到了 chdir 的时候，core 文件会转移到 change 的那个目录去，因此我们最好设置一下文件生成到固定的位置。如我们想将生成的 coredump 文件放到 /tmp/core 这个文件中，则可以在系统启动的时候执行，echo "/tmp/core" > /proc/sys/kernel/core_pattern。但这样又会存在一个问题，如果多个程序，或者一个程序生成多个 coredump 文件，那么这个文件将被覆盖，下面介绍了一些转意符，在生成文件的时候使用这些转意符可以区分不同时期生成的 coredump 文件。比如如果你想将生成的 coredump 文件放到 /tmp/[进程名]-[进程 id].coredump 这个文件中可以在启动的时候执行这条指令： echo /tmp/%e.%p.coredump > /proc/sys/kernel/core_pattern

       %% 单个%字符
       %p 所dump进程的进程ID
       %u 所dump进程的实际用户ID
       %g 所dump进程的实际组ID
       %s 导致本次core dump的信号
       %t core dump的时间 (由1970年1月1日计起的秒数)
       %h 主机名
       %e 程序文件名
   

1. ID Pin 低电平为 OTG (host) 模式，模块内部已经将 ID PIN 和 PD(2) 关联起来了，如果设备要支持热插拔，那么这个引脚应该使能。在 board.h 中，但是如果你的设备自始至终都只有一个 USB 设备，比如声卡，无线网卡，那么在开机的时候在驱动中在 PD(2) 模拟一个插入 USB 的事件即可。（也就是设置成输出，然后设置成低电平）
2. 首先在 github 上将 PPP 下载下来
3. 然后在驱动中将下面这些项目打开

       1. Device Drivers --->
           USB support --->
               USB Serial Converter support --->
                   USB Generic Serial Driver
                   USB driver for GSM and CDMA modems
       2. Device Drivers --->
           Network device support --->
               PPP (point-to-point protocol) support
               PPP multilink support
               PPP filtering
               PPP support for async serial ports
               PPP support for sync tty ports
       3. Device Drivers --->
           Network device support --->
               USB Network Adapters --->
                   Multi-purpose USB Networking Framework
   
       CONFIG_USB_SERIAL=y
       CONFIG_USB_SERIAL_GENERIC=y
       CONFIG_USB_SERIAL_OPTION=y
       CONFIG_PPP=y
       CONFIG_PPP_ MULTILINK=y
       CONFIG_PPP_FILTER=y
       CONFIG_PPP_ASYNC=y
       CONFIG_PPP_SYNC_TTY=y
       CONFIG_USB_NET_CDCETHER=y (For modem with ECM/NDIS interface)
   

4. 配置设备 ID

   编译内核的时候可能出现内核版本太老不支持 USBID 那么此时你应该找到无线网卡的 ID (你在使用其他 USB 的时候也是一样) lsusb 可以查询到插入设备的 ID。如下：

       # lsusb
       Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
       ...
       

   其中 1d6b:0003 就是 ID 你在内核地 driver/usb/serial/options.h 这个文件中要加入你添加的设备 ID，如中兴的 ZM8620 的 ID 可能是 19d2:0396 那么你在 option_ids 这个数组中应该在结尾添加上如下这行

       static const struct usb_device_id option_ids[] = {
           ...
           // or { USB_DEVICE_AND_INTERFACE_INFO(ZTE_VENDOR_ID, 0x0396, 0xff, 0xff, 0xff) }
           {USB_DEVICE( 0x19d2， 0x0396 )},
           {},
       }
       

5. 然后将下面的这几个文件放到一个文件夹中, 编译出来的文件 pppoe-discovery, chat, pppd, pppstats, pppdump 需要自己编辑 zte_options, ppp-on, pppoe-discovery, disconnect

   以下拿联通的举例:

           ppp-on
   
               #!/bin/sh
               set -e
               RELATIVE_DIR=`dirname $0`
               CUR_DIR=`cd ${RELATIVE_DIR} && pwd && cd -`
               PPPD=${CUR_DIR}/pppd
               CHAT=${CUR_DIR}/chat
   
               OPTION_FILE="zte_options"
               DIALER_SCRIPT="$(pwd)/zte_ppp_dialer"
               exec ${PPPD} file $OPTION_FILE connect "${CHAT} -v -f ${DIALER_SCRIPT}"
   
               disconnect
   
                   #!/bin/sh
               killall pppd
   
   
           zte_options：
   
               /dev/ttyUSB2
               115200
               crtscts
               modem
               persist
               lock
               noauth
               noipdefault
               debug
               nodetach
               user Anyname
               password Anypassword
               ipcp-accept-local
               ipcp-accept-remote
               defaultroute
               usepeerdns
               noccp
               nobsdcomp
               novj
               dump
   
           zte_ppp_dialer
   
               ABORT       "NO CARRIER"
               ABORT       "ERROR"
               TIMEOUT     120
               ""          ATE
               SAY         "ATE"
               ECHO        ON
               OK          ATH
               OK          ATP
               OK          "AT+CGDCONT=1,\"IP\", \"APN\""
               OK          ATD*98#
               CONNECT
       

• 平台设备注册

  // include/linux/platform_device.h
  #define module_platform_driver(__platform_driver) \
      module_driver(__platform_driver, platform_driver_register, \
          platform_driver_unregister)
  

• 内核打印
  在调试内核时可用 printf 的孪生哥哥 printk，printk 的使用语法与 printf 相差不大，但可在格式化字符串前加上打印等级，如:printk( KERN_DEBUG "hello, world!" );

   

      #define KERN_EMERG      "<0>"    /* 系统不可使用 */
      #define KERN_ALERT      "<1>"    /* 需要立即采取行动 */
      #define KERN_CRIT       "<2>"    /* 严重情况 */
      #define KERN_ERR        "<3>"    /* 错误情况 */
      #define KERN_WARNING    "<4>"    /* 警告情况 */
      #define KERN_NOTICE     "<5>"    /* 正常情况, 但是值得注意 */
      #define KERN_INFO       "<6>"    /* 信息型消息 */
      #define KERN_DEBUG      "<7>"    /* 调试级别的信息 */
      /* 使用默认内核日志级别 */
      #define KERN_DEFAULT    ""
      /*
      * 标注为一个“连续”的日志打印输出行(只能用于一个
      * 没有用 \n封闭的行之后). 只能用于启动初期的 core/arch 代码
      * (否则续行是非SMP的安全).
      */
      #define KERN_CONT       ""
  

• ubuntu 下 buildroot 编译过依赖，不完全列表
  sudo apt update; sudo apt upgrade -y; sudo apt install -y virtualbox-guest-additions-iso net-tools htop vim tree git gitk bc unzip python bison flex libncurses5-dev libncursesw5-dev device-tree-compiler expect cmake
• Buildroot 交叉编译
  1. make download V=s — 下载所有依赖项

• Linux音频驱动-OSS和ALSA声音系统简介及其比较
• Linux音频驱动-ALSA概述
• alsa架构分析

• Linux 的仓库地址以及历史 release
• 该网站提供大量的交叉编译工具
• GNU Arm Embedded Toolchain Downloadsarm eabi toolchain 下载地址
• GNU-A Downloadsarm aarch64 toolchain 下载地址