---
hackpadID: gnQyhcXcjtT
hackpadWorkspace: tossug
tags: hackpad-import, tossug
---
# Linux 讀書會 - 第 13 週 10/28/2014
###### tags: `TOSSUG 讀書會`
總目錄 [edX Introduction to Linux](https://tossug.hackpad.com/dVX1LvoCcii)

## 課程筆記

**Section 1**

Process: 執行中的程式

Thread: 輕量級 process

* Scheduler在分配cpu time的時候是以thread為基本單位，每個process至少會有一個thread ？!
* 是的，每個process在kernel scheduler的認知上只是一組用相同address space的threads，應該說kernel scheduler根本不用知道single-thread process和它的thread的分別，它只要排task

[TCB](https://en.wikipedia.org/wiki/Thread_control_block) 和 [PCB](https://en.wikipedia.org/wiki/Process_control_block) 儲存 thread 和 process 的必要資訊。

<undefined>* **Scheduling**</undefined>

[](https://en.wikipedia.org/wiki/Scheduling_(computing))[https://en.wikipedia.org/wiki/Scheduling_(computing](https://en.wikipedia.org/wiki/Scheduling_%28computing)[)](https://en.wikipedia.org/wiki/Scheduling_%28computing%29)

[](http://mmdays.com/2007/09/28/cpu_scheduling/)[http://mmdays.com/2007/09/28/cpu_scheduling/](http://mmdays.com/2007/09/28/cpu_scheduling/)

[](http://www.tutorialspoint.com/operating_system/os_process_scheduling_algorithms.htm)[http://www.tutorialspoint.com/operating_system/os_process_scheduling_algorithms.htm](http://www.tutorialspoint.com/operating_system/os_process_scheduling_algorithms.htm)

<undefined>* **Process scheduler**</undefined>

* long-term sheduling
* mid-term sheduling
* short-term sheduling

<undefined>* **Scheduling algorithm**</undefined>

對 process 進行分工，決定每個 process 切成幾份 task，每個 task 分到多少 CPU time。

* First-Come-First-Served (FCFS)
* Shortest-Job-First (SJF)
* 不可能實作，因為若事先知道時間，就不用排程演算法了。
* Priority
* Each process is assigned a priority. Process with highest priority is to be executed first and so on.
* Processes with same priority are executed on first come first serve basis.
* Priority can be decided based on memory requirements, time requirements or any other resource requirement.
* Round Robin (RR)
* 透過 quantum 值決定每個 task 分到多少時間。
* Multi Queue
* 依用途分成不同的 queue，可各自套用不同的演算法。

Linux kernel 預設使用 [CFS](https://en.wikipedia.org/wiki/Completely_Fair_Scheduler) 排程演算法，透過[紅黑樹](https://en.wikipedia.org/wiki/Red–black_tree)儲存記錄 task，時間複雜度為 O(log n)。

Ref: [](http://en.wikipedia.org/wiki/Completely_Fair_Scheduler)http://en.wikipedia.org/wiki/Completely_Fair_Scheduler

<undefined>* **基本觀念複習**</undefined>

*

nice value, niceness:

好人值，數值愈高，優先權愈低。

Resource starvation

永遠有 process 執行不到。

[](https://en.wikipedia.org/wiki/Resource_starvation)[https://en.wikipedia.org/wiki/Resource_starvation](https://en.wikipedia.org/wiki/Resource_starvation)

**Section 2**

ps aux

ps axo

top 以外的監控工具：

* htop: [](http://hisham.hm/htop/)[http://hisham.hm/htop/](http://hisham.hm/htop/)
* atop（OS X 不支援）: [](http://www.atoptool.nl/)[http://www.atoptool.nl/](http://www.atoptool.nl/)
* nmon（OS X 不支援）: [](http://nmon.sourceforge.net/pmwiki.php)[http://nmon.sourceforge.net/pmwiki.php](http://nmon.sourceforge.net/pmwiki.php)
* vtop: [](http://parall.ax/vtop)[http://parall.ax/vtop](http://parall.ax/vtop)
* iotop: [](http://guichaz.free.fr/iotop/)[http://guichaz.free.fr/iotop/](http://guichaz.free.fr/iotop/)
* iftop: [](http://www.ex-parrot.com/pdw/iftop/)[http://www.ex-parrot.com/pdw/iftop/](http://www.ex-parrot.com/pdw/iftop/)

top -Hp pid -> 監控某個process下所有thread的CPU Usage (Application can use it to tune performance)

* 非常實用！

kernel space

掛載 rootfs 前，切出一塊記憶體空間，用來識別和載入硬體資源 (driver, kernel modules, etc...)，也就是進入系統後看到的 `/proc`。FreeBSD 和 Mac OS X 不建立 `/proc`，具體行為還沒研究。

user space

不屬於 kernel space，獨立一塊供使用者使用的記憶體空間。

## 本週作業

Chapter 17: Processes

* Section 3 - Section 4

## 活動簽到

[Carl Su](/ep/profile/n5euV0AaWLn)

[P Fisher](/ep/profile/oTOWRrYfPRk)

[Sam Lu](/ep/profile/zi86w39M6U6)

[violetson](/ep/profile/oJusv72f72w)

[Manuel Stallman](/ep/profile/GgkcGJEol5r)

[Kommodore Kezza](/ep/profile/CvxduB5FRRT)