Interview_Process/Thread/Deadlock/Paging/Semaphone/Mutex/Segmentation/Virtual Memory

Process 🆚 Thread

• Process:
  • program running on memory
  • has own address, resource

• Thread: work unit in process
  • does not have independent address, resource
  • only has independent stack

💡 https://soheeparklee.github.io/posts/OS-2process-thread/

✅ Why use multi threading instead of multi processing?

System call to create process and allocate resource decrease

• communication between thread is costless than communication between process(IPC)
  
• However, need to take synchronization into account for multi threading
  

✅ Context Switching?

하나의 프로세스가 CPU를 사용 중인 상태에서 다른 프로세스가 CPU를 사용하도록 하기 위해, 이전의 프로세스 상태를 보관하고 새로운 프로세스의 상태를 적재하는 작업
한 프로세스의 문맥은 그 프로세스의 PCB에 기록됨

💡 https://soheeparklee.github.io/posts/OS-6pcb/

Semaphore 🆚 Mutex

• Semaphore:

  • 공유자원에 접근할 수 있는 최대 허용치만큼만 동시에 사용자 접근 가능
  • 스레드들은 리소스 접근 요청
  • 세마포어는 카운트가 하나씩 줄어들게 되며 리소스가 모두 사용중인 경우(카운트=0) 다음 작업은 대기

• Mutex:

  • 제어되는 섹션에 하나의 스레드만 허용
  • 해당 섹션에 접근하려는 다른 스레드들을 강제적으로 막음
  • 첫번재 스레드가 해당 섹션을 빠져나올 때까지 기다려야 함

💡 https://soheeparklee.github.io/posts/OS-11semapore/

✅ What is deadlock?

What is deadlock?

When process cannot get the resource, and cannot work
occurs when several threads are trying to gain access to shared resource

Four conditions of deadlock

• Mutual exclusion: shared resouce requires exclusive control, inshareable
  
• Resource holding: thread is waiting for another resource to be released while holding onto an allocated resource
  
• circular wait: each process waiting for a resource held by the next process in the chain
  
• no preemption: thread cannot get a resource until the resource is finished by another thread
  
  

How to avoid deadlock?

• prevention: eliminate condition
  (for no-preemtion: let go of the holding resource)
  
• avoidance: run algorithm on requests to check for a safe state
  
• detection
  > > recovery
  

💡 https://soheeparklee.github.io/posts/OS-9deadlock/

✅ Memory Hierchy

• Register
• Cache
• RAM
• Hard Disk

💡 https://soheeparklee.github.io/posts/4-WEB_os_architecture/

✅ Memory Allocation Algorithm

• First fit:

  • search memory from start
  • allocate the first space that has enough storage

• Next fit:

  • begin search from the last allocated place

• Best fit:

  • search all the memory to find the best place to allocate

💡 https://soheeparklee.github.io/posts/OS-12paging/

✅ Paging, Segmentation

✔️ Paging

• optimize memory usage
  
• divide memory into fixed-size pages
  
  

✔️ Segmentation

• divide memory into various size, more flexible
  

💡 https://soheeparklee.github.io/posts/OS-12paging/

✅ Page Fault, Virtual Memory

✔️ Page Fault

• when a running program accesses a memory page that is mapped into virtual address space
  
• but not loaded on physical memory
  
• OS has to replace one of the existing page with newly needed page
  
• thus, the need for page replacement algorithm
  
  

✔️ Virtual Memory

• memory management technique
  
• create an illusion of memory using RAM + disk storage
  
• 실제 메모리 안에 공간이 부족하면, 현재 사용하고 있지 않은 데이터를 빼내어 가상 메모리에 저장해두고, 실제 메모리에선 처리만 하게 하는 것이 가상 메모리의 역할이다.
  
• 즉, 실제 메모리에 놀고 있는 공간이 없게 계속 일을 시키는 것. 이를 도와주는 것이 ‘가상 메모리’
  
  

✅ Page Replacement Algoirthm

• FIFO: keep memory in queue, oldest page is in front
• LRU: Least recently used 최근 가장 오랫동안 사용하지 않은 페이지
• LFU: Least frequently used 사용 빈도 적은 페이지
• MRU: Most recently used

⭐️ Key words in process, thread

• multi threading, process
• synchronization

⭐️ Key words in CPU, thread

• CPU ➡️ thread allocate
• CPU scheduling
• preemptive: Round Robin, SRTF
• non-preemptive: FCFS, SJF, HRN
• starvation
• aging

⭐️ Key words in Memory, page

• Memory ➡️ page
• paging
• segmentation
• internal
• external segmentation
• compaction

• best fit/next fit/first fit

• page replacement
• page fault
• FIFO/LRU/MRU/NUR/LFU/Optimal Place Replacement