Priority Ceiling Protocol

Priority Ceiling Protocol is an advanced task scheduling algorithm, principles of which were presented in the context of the Wireless Embedded Systems MSc.

Transcript

1. Priority Ceiling Protocol
   Konstantinos Goutsos
   EEE8068: Real-Time Computer Systems
   School of Electrical and Electronic Engineering
   Newcastle University
   1
   April 2015
   

   View Slide

2. Presentation Outline
   •Introduction
   • Resources: Shared and Exclusive
   • Semaphores and Critical Sections
   • Task blocking and Deadlocks
   •Priority Inversion Phenomenon
   •Priority Inheritance Protocol
   •Priority Ceiling Protocol
   • Justification
   • Definition
   • Example
   2
   April 2015
   

   View Slide

3. Resources
   • Software structures used by processes
   • Shared resources: can be used by multiple
   processes
   • Exclusive resources: protected against simultaneous
   use by more than one process using critical sections
   3
   April 2015
   

   View Slide

4. Critical Sections
   • Task sections which should not be interrupted
   • Used to control access to exclusive resources
   • Usually implemented with semaphores
   • Semaphores: provided by the OS, bound to an
   exclusive resource
   • Semaphores can be used only by the wait and
   signal primitives
   4
   April 2015
   

   View Slide

5. Task blocking and Deadlocks
   • A task can stay in a waiting state: task blocking
   • The maximum blocking time is very important for real-
   time systems
   • Deadlock: a situation where two or more tasks are
   waiting for each other to complete
   5
   April 2015
   

   View Slide

6. Priority Inversion Phenomenon
   • Appears when using preemption and critical sections
   • A high priority task has to wait for a lower priority one
   to exit a critical section in order to enter its own
   • Happens when a low priority task has entered its
   critical section first
   • May create unbounded blocking as the waiting time
   depends on the execution time of lower priority tasks
   6
   April 2015
   

   View Slide

7. Priority Inversion Phenomenon: Example
   7
   April 2015
   [Buttazzo 2011]
   

   View Slide

8. Priority Inversion Phenomenon: Example
   8
   April 2015
   [Buttazzo 2011]
   1
   > 2
   > 3
   

   View Slide

9. Dealing with the Phenomenon
   • Simple solution: Disable preemption during critical
   sections
   • Complex priority protocols
   • Static priorities: Priority Inheritance, Priority Ceiling
   • Static & dynamic priorities: Stack Resource Policy
   9
   April 2015
   

   View Slide

10. Priority Inheritance Protocol
    • A task causing blocking temporarily assumes the
    priority of the blocked task
    • Plus: Bounds the maximum blocking time (which can
    be calculated)
    • Minus: Blocking time is not reduced and deadlocks are
    not prevented
    10
    April 2015
    

    View Slide

11. Priority Ceiling Protocol
    • Builds on the Priority Inheritance Protocol
    • Adds a rule for granting lock requests on semaphores: Each
    semaphore is assigned a priority ceiling
    • The ceiling is equal to the priority of the highest priority job
    that can lock the semaphore
    • A task can only lock a semaphore if its priority is higher than
    the larger priority ceiling among all locked semaphores
    • Makes sure that once a job enters a critical section it cannot
    be blocked by lower priority jobs
    11
    April 2015
    

    View Slide

12. Priority Ceiling Protocol: Example
    12
    April 2015
    [Buttazzo 2011]
    1
    > 2
    > 3
    

    View Slide

13. References
    G. C. Buttazzo, Hard Real-time Computing Systems: Predictable Scheduling Algorithms and
    Applications. 2011.
    April 2015 13
    

    View Slide