Please enable JavaScript.

Coggle requires JavaScript to display documents.

Deadlock (Conditions for resource deadlock (Hold-and-Wait, No-preemption,…

- - - - One Resource of each Type
        
        Using Resource Graph
        
        Detect Cycle
      - Multiple Resource of each Type
        
        Using matrix
        
        A : Available resource vector
        
        C : Current allocation matrix
        
        E : Existing resource vector
        
        R : Request Matrix
        
        any unmarked processes are known to be deadlocked
    - - Recovery Through Preemtion
        
        Difficult or Impossible
      - Rollback
        
        Check Point
      - Kill process
        
        First Approach
        
        Kill process in the cycle
        
        Check the cycle, if so, go back to first line (1)
        
        Repeat until the cycle is broken
        
        Second Approach
        
        Kill a process not in the cycle to release its resource
        
        The killed process is holding some resource that processes in cycle need it.
        
        Victim process have ability to re-run from beginning without any ill effect.
    - - System does not attempt to prevent deadlock from occurring
      - It let deadlock occur, tries to detect when it happen
      - Take some action to recover after the fact
  - - - Safe
        
        if there exists a safe sequence
      - Unsafe
    - - Single Resource
      - Multiple Resource
    - - $dropped image link$
    - - If certain information is available in advance
      - There is an algorithm that can always avoid deadlock by making the right choice all the time
      - Resources are allocated carefully
  - - - Prevent process that holds resource from waiting for more resources or
      - Must guarantee that whenever a process requests a resource, it does not hold any other resources, or
      - :arrow_right: All processes need to request all their resources before starting execution.
      - require a process requesting a resource to first temporarily release all the resource it currently holds, then it tries to get everything it needs all at one.
    - - Preempted resources are added to the list of resources for which the process is waiting.
      - Process will be restarted only when it can regain its old resources, as well as the new ones that it is requesting.
      - If a process that is holding some resources requests another resource then the resource cannot be immediately allocated to it but all resources currently being held are released.
      - Some recourse can be virtualized to avoid the first condition such as spooling and daemon.
    - - Try to make sure that as few processes as possible may actually claim (thình cầu, xin) the resource.
      - Avoiding assigning a resource when that is not absolutely necessary, and
    - - First Approach
        
        If it needs a second one, it must release the first one.
        
        :arrow_right: If it needs a second one, it must release the first one.
        
        A process is entitled (cho quyền) only to a single resource at the moment.
      - Second Approach
        
        Process can request resources whenever they want to, but all requests must be made in numerical order.
        
        :arrow_right: The resource allocation graph can never have cycles but the numerically ordering the resources may be impossible to find an ordering that satisfies every ones.
        
        Provide a global numbering of the resources.