Real-Time Systems Laboratory
Department of Computer and Information Science
Linköping University, Sweden
The lecture notes are partly based on lecture notes by Simin Nadjm-Tehrani, Jörgen Hansson, Anders Törne.
They also loosely follow Burns’ and Welling book “Real-Time Systems and Programming Languages”. These lecture notes should only be used for internal teaching purposes at Linköping University.
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
31 pages
• The “Simple Process Model” assumes independence among the tasks
– In reality this is seldom the case
• Shared critical resource – a block of code that is shared by several processes but requires mutual exclusion at any point in time
• Blocking – a higher priority process waiting for a lower priority one
• A “priority inversion” occurs
• Obs: preemption – when a lower priority process waits for higher priority one – considered normal
• The blocking time must be
– bounded
– measurable
– Small
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
2 of 31
• When a process waits for a lower priority process the priority inversion is unavoidable
• An avoidable really bad case
– A low priority process (P1) locks the resource
– A high priority process (P2) has to wait on the semaphore (blocked state)
– A medium priority process (P3) preempts P1 and runs to completion before P2!
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
3 of 31
• Q (or V) – execution tick with access to the Q (or V) critical sections which are protected by mutual exclusion
• E – execution tick not accessing critical resources
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
4 of 31
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
5 of 31
• Scheme for eliminating avoidable priority inversions
– Process’ priority is no longer static
• If P1 is suspended due to P2, and if P1 has higher priority, then the priority of P2 is raised to the same priority as P1
• Priority inheritance
– Is transitive
– Guarantees an upper bound of blocking time as P1 is blocked only while P2 is using the resource
– Mutual exclusion guaranteed by protocol
– Deadlocks are still possible
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
6 of 31

• Worst case blocking time calculation
B i
= k
K
= 1 usage ( k , i ) C i
– B i
– the maximum blocking time of process I
– usage(k,i) = 1 if resource k is used by at least one process with a priority less than P i and at least one process with priority greater or equal to P i
, otherwise 0.
– C(k) – worst case execution time of k critical resource
• This is an upper bound
– even higher accuracy formulas might be available
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
7 of 31
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
8 of 31
• Two forms
– Original ceiling priority protocol (OCPP)
– Immediate ceiling priority protocol (ICPP)
• Properties (on a single processor system):
– A high priority process can be blocked at most once during its execution by lower priority processes
• Only in the beginning in the case of ICPP
– Deadlocks are prevented
– Transitive blocking is prevented
• I.e. chain blocking (c is blocked by b, which is blocked by a)
– Mutual exclusive access to resources is ensured
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
9 of 31
• Static default priority
– assigned to each process (by a FPS scheme)
• Static ceiling value
– For each resource
– Is the maximum static priority of the processes that use it
• Dynamic priority
– Assigned at run-time, “real” priority
– Maximum of the process own static priority and any priority it inherits due to it blocking processes with higher priority
• Lock on a resource
– Can be gained by a process only if its dynamic priority is higher than the ceiling of any currently locked resource
• But excluding any resource that it has already locked itself
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
10 of 31
• How is the priority of L1 changing in time?
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
11 of 31
• Static default priority
– assigned to each process (by a FPS scheme)
• Static ceiling value
– For each resource
– Is the maximum static priority of the processes that use it
• Dynamic priority
– Assigned at run-time, “real” priority
– Maximum of its own static priority and the ceiling values of any resources it has locked.
• Consequence
– A process can be blocked only at the beginning of the execution, equivalent with:
– All resources of a process must be free before execution
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
12 of 31

Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
• Worst case blocking for both ceiling protocols
B i
=
K max k = 1 usage ( k , i ) C i
• Since a process can only be blocked once (per activation)
– But more processes will experience this block
13 of 31 Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
14 of 31
• Although the worst-case behaviour of the two ceiling schemes is identical (from a scheduling view point), there are some points of difference
• ICPP is easier to implement than the original (OCPP)
– No need to monitor blocking relationships in ICPP
• ICPP leads to less context switches
– Blocking occurs prior to first execution in ICPP
• ICPP requires more priority movements
– This happens with all resource usages in ICPP
– OCPP only changes priority if an actual block has occurred
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
15 of 31
• Properties (on a single processor system):
– A high priority process can be blocked at most once during its execution by lower priority processes
• Prove …
– Transitive blocking is prevented
• Prove …
– Mutual exclusive access to resources is ensured by the protocol itself
• Prove …
– Deadlocks are prevented
• Let’s prove together
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
16 of 31
• The 4 conditions necessary for a deadlock
– Mutual exclusion
• Only one process at a time can use a resource
– Hold and Wait
• A process may hold allocated resources while awaiting assignment of other resources.
– No forced release
• A resource can be released only voluntarily by the process holding it
– Circular wait
• A closed chain of processes exists, such that each process holds at least one resource needed by the next process in the chain
• Alternatively, each process waits for a messages that has to be sent by the next process in the chain
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
17 of 31
• Deadlock prevention
– At system design. Offline
• Deadlock avoidance
– Dynamically considers the request, decides if safe. Online
• Deadlock detection and treatment
• Analogy avoidance vs. prevention
– Traffic lights vs. circulation agent
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
Resource allocation graph
18 of 31

• By showing deadlock-condition 4. never holds
– Assume a circular wait
– In this “circular wait” there must be two resources that have the highest ceiling
– Now, if one is locked by a process, then no other process can lock the other resource
• As the process that locked the first resource runs with highest priority
– Immediately – ICPP
– When a lock on the second is attempted – OCPP
– Thus no circular wait exists
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
19 of 31
• The response time analysis can include blocking times
R i
= C i
+ B i
+ I i
R i
= C i
+ k
K max
= 1 usage ( k , i ) C i
+
j ∈ hp ( i )
⎡
⎢
R
T j i
⎤
⎥ C j
• Remember
– Response time analysis only for FPS scheduling
– For EDF the worst case response time for a task is not when all tasks start at the same time
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
20 of 31
• Period (T)
– Minimum inter-arrival time
• Important for sporadic tasks with hart RT guarantees
• Undefined for aperiodic tasks
– Average inter-arrival time
• Important for sporadic and aperiodic tasks with soft RT
• Deadline (D)
– Usually D < T
– As sporadic+aperiodic tasks usually respond to an unexpected event (exception handling, warning)
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
21 of 31
• Rule #1
– All processes (both hard & soft) should be schedulable
• Using average execution times
• Using average arrival rates
• With Rule#1 transient overloads are possible
• Rule #2
– All hard RT processes should be schedulable
• Using worst case execution times
• Using worst case arrival rates
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
22 of 31
• Sporadic can be considered as periodic, with worst case inter-arrival rate as the period
• A “server” runs soft RT tasks to not disturb the hard RT ones
– A server usually has a period T s and a capacity C s
– Whenever an aperiodic tasks arrives and capacity is available the task is run until server capacity is depleted
– The capacity is replenished in time
– A server has to be used for aperiodic tasks
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
23 of 31
• Deadline Monotonic
– Fixed Priority Scheduling (FPS)
– Response time analysis supports it
• EDF
– Utilisation analysis not sufficient anymore
– But still optimal scheduling policy
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
24 of 31

Dynamic Scheduling
• on-line
• no a priori knowledge
+ flexibility
+ adaptability
+ few assumptions about workload
- prone to overloads
- decision consumes processor time
only statistical guarantees
Static Scheduling
• off-line
• assumes clairvoyance
• schedules planned for peak load
+ task workloads are periodic => no overloads
+ analyzable
+ cheap
- low utilization
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
25 of 31
• Static Scheduling
– determine maximum execution time
– allocate task to node
– allocate communication slots to LANs
– construct schedules off-line
• Dynamic Scheduling
– “...not necessary to develop a set of detailed plans during the design phase, since the execution schedules are created dynamically as a consequence of the actual demand”
• H. Kopetz
– Things that can be done
• admission control
• overload resolution
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
26 of 31
• Static Scheduling
– Schedules are planned for peak load
– If worst-case exec time >> avg. case exec time => low utilization (usually)
• Dynamic Scheduling
– Better utilization at low or average loads
– Increased online processing time
• Dilemma: to schedule or to execute?
• Higher overhead
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
27 of 31
• Static Scheduling
– Task schedules must be recalculated when
• Maximum execution time changes
• New task is added
• Communication schedules – recalculated for active nodes
• Dynamic Scheduling
– Temporal properties may/should be retested
– A small change in task characteristics brings a small change in performance
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
28 of 31
• Reject tasks
• Abort tasks (i.e., drop)
• Replace tasks
– alternative/contingency actions
– primary/backup
• Partial execution of tasks
– imprecise computation
• Defer execution of tasks
• Migrate tasks
– only in distributed systems
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
29 of 31
• Task migration may increase flexibility
• Communication delays must be bounded
• Remote procedure calls introduce remote blocking
• More about distributed systems it in a future lecture
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
30 of 31

• Chapter 13 in Burns & Wellings
• Chapter 4 in Butazzo
– for the proofs of sufficient condition and optimality
Lecture 3: Sharing resources and dynamic scheduling
Calin Curescu
31 of 31