Mi-Jung Choi Dept. of Computer and Science Silberschatz, Galvin and Gagne ©2006 Operating System Principles Chapter 7: Deadlocks

- 2 - Operating System Fall 2009 Ch07 - Deadlocks Chapter 7: Deadlocks  The Deadlock Problem  System Model  Deadlock Characterization  Methods for Handling Deadlocks  Deadlock Prevention  Deadlock Avoidance  Deadlock Detection  Recovery from Deadlock

- 3 - Operating System Fall 2009 Ch07 - Deadlocks Chapter Objectives  To develop a description of deadlocks, which prevent sets of concurrent processes from completing their tasks  To present a number of different methods for preventing or avoiding deadlocks in a computer system.

- 4 - Operating System Fall 2009 Ch07 - Deadlocks The Deadlock Problem  A set of blocked processes each holding a resource and waiting to acquire a resource held by another process in the set.  Example  System has 2 tape drives.  P 1 and P 2 each hold one tape drive and each needs another one.  Example  semaphores A and B, initialized to 1 P 0 P 1 wait (A);wait(B) wait (B);wait(A)

- 5 - Operating System Fall 2009 Ch07 - Deadlocks Bridge Crossing Example  Traffic only in one direction.  Each section of a bridge can be viewed as a resource.  If a deadlock occurs, it can be resolved if one car backs up (preempt resources and rollback).  Several cars may have to be backed up if a deadlock occurs.  Starvation is possible.

- 6 - Operating System Fall 2009 Ch07 - Deadlocks System Model  Resource types R 1, R 2,..., R m CPU cycles, memory space, I/O devices  Each resource type R i has W i instances.  Each process utilizes a resource as follows:  Request: If the request cannot be granted immediately, the requesting process have to wait until it can acquire the resource.  Use: The process operate on the resource.  Release: The process releases the resource.

- 7 - Operating System Fall 2009 Ch07 - Deadlocks Deadlock Characterization  Mutual exclusion: only one process at a time can use a resource.  Hold and wait: a process holding at least one resource is waiting to acquire additional resources held by other processes.  No preemption: a resource can be released only voluntarily by the process holding it, after that process has completed its task.  Circular wait: there exists a set {P 0, P 1, …, P n } of waiting processes such that  P 0 is waiting for a resource that is held by P 1,  P 1 is waiting for a resource that is held by P 2,  …,  P n–1 is waiting for a resource that is held by P n, and  P n is waiting for a resource that is held by P 0. Deadlock can arise if four conditions hold simultaneously.

- 8 - Operating System Fall 2009 Ch07 - Deadlocks Resource-Allocation Graph  V is partitioned into two types:  P = {P 1, P 2, …, P n }, the set consisting of all the processes in the system.  R = {R 1, R 2, …, R m }, the set consisting of all resource types in the system.  E is partitioned into two types:  request edge – directed edge P i  R j  assignment edge – directed edge R j  P i Deadlocks can be described more precisely in terms of a directed graph called a system resource-allocation graph: A set of vertices V and a set of edges E.

- 9 - Operating System Fall 2009 Ch07 - Deadlocks Resource-Allocation Graph (Cont.)  Process  Resource Type with 4 instances  P i requests a instance of R j  P i is holding an instance of R j PiPi PiPi PiPi PiPi RjRj RjRj

Operating System Fall 2009 Ch07 - Deadlocks Example of a Resource Allocation Graph  Deadlock or not?  Mutual Exclusion  R 1, R 2, R 3, R 4  Hold and Wait  P 1, P 2  No Preemption  YES  Circular Wait  NO

Operating System Fall 2009 Ch07 - Deadlocks Resource Allocation Graph With a Deadlock  Deadlock or not?  Mutual Exclusion  R 1, R 2, R 3, R 4  Hold and Wait  P 1, P 2, P 3  No Preemption  YES  Circular Wait  YES

Operating System Fall 2009 Ch07 - Deadlocks Resource Allocation Graph With A Cycle But No Deadlock  Deadlock or not?  Mutual Exclusion  R 1, R 2  Hold and Wait  P 1, P 3  No Preemption  YES  Circular Wait  YES

Operating System Fall 2009 Ch07 - Deadlocks Basic Facts  If graph contains no cycles  no deadlock.  If graph contains a cycle   if only one instance per resource type, then deadlock.  if several instances per resource type, possibility of deadlock.

Operating System Fall 2009 Ch07 - Deadlocks Methods for Handling Deadlocks 1.Ensure that the system will never enter a deadlock state.  Use a protocol to prevent or avoid deadlocks 2.Allow the system to enter a deadlock state and then recover.  deadlock detection, deadlock recovery 3.Ignore the problem and pretend that deadlocks never occur in the system;  used by most operating systems, including UNIX and Windows.  It is then up to the application developer to write programs that handle deadlocks

Operating System Fall 2009 Ch07 - Deadlocks Deadlock Prevention  Mutual Exclusion – not required for sharable resources (read-only file); must hold for nonsharable resources (printer).  Hold and Wait – must guarantee that whenever a process requests a resource, it does not hold any other resources.  require process to request and be allocated all its resources before it begins execution, or  allow process to request resources only when the process has none.  Low resource utilization; starvation possible. By ensuring that at least one of the four conditions for deadlock cannot hold, we can prevent deadlock. Restrain the ways request can be made.

Operating System Fall 2009 Ch07 - Deadlocks Deadlock Prevention (Cont.)  No Preemption –  If a process that is holding some resources requests another resource that cannot be immediately allocated to it, then all resources currently being held are released.  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.  Circular Wait –  impose a total ordering of all resource types, and  require that each process requests resources in an increasing order of enumeration.

Operating System Fall 2009 Ch07 - Deadlocks Deadlock Prevention (Cont.)  Deadlock prevention algorithm  Prevent deadlocks by restraining how requests can be made  The restraints ensures that  at least one of the necessary conditions for deadlock cannot occur and, hence, that deadlocks cannot hold.  Possible side effects deadlock prevention  Low device utilization  Reduced system throughput  Alternative method  Deadlock Avoidance

Operating System Fall 2009 Ch07 - Deadlocks Deadlock Avoidance  Simplest and most useful model for DA requires that  each process declare the maximum number of resources of each type that it may need.  The deadlock-avoidance algorithm dynamically examines the resource-allocation state to ensure that there can never be a circular-wait condition.  Resource-allocation state is defined by the number of available and allocated resources, and the maximum demands of the processes. requires that the system has some additional a priori information available.

Operating System Fall 2009 Ch07 - Deadlocks Resource Allocation State: Safe, Unsafe, Deadlock

Operating System Fall 2009 Ch07 - Deadlocks Safe State  When a process requests an available resource,  system must decide if immediate allocation leaves the system in a safe state.  System is in safe state?  if there exists a safe sequence of all processes.  Sequence is safe if for each P i, the resources that P i can still request can be satisfied by currently available resources + resources held by all the P j, with j<i.  If P i resource needs are not immediately available, then P i can wait until all P j have finished.  When P j is finished, P i can obtain needed resources, execute, return allocated resources, and terminate.  When P i terminates, P i+1 can obtain its needed resources, and so on.

Operating System Fall 2009 Ch07 - Deadlocks Basic Facts  If a system is in safe state  no deadlocks.  If a system is in unsafe state  possibility of deadlock.  Avoidance  ensure that a system will never enter an unsafe state.

Operating System Fall 2009 Ch07 - Deadlocks Resource-Allocation Graph Algorithm  Claim edge P i  R j indicated that process P j may request resource R j ; represented by a dashed line.  Claim edge converts to request edge when a process requests a resource.  When a resource is released by a process, assignment edge reconverts to a claim edge.  Resources must be claimed a priori in the system. PiPi PiPi RjRj PiPi PiPi RjRj PiPi PiPi RjRj

Operating System Fall 2009 Ch07 - Deadlocks Resource-Allocation Graph For Deadlock Avoidance  Find a safe sequence !  P1, P2 → YES  P2, P1 → NO

Operating System Fall 2009 Ch07 - Deadlocks Unsafe State In Resource-Allocation Graph  Find a safe sequence !  P1, P2 → NO  P2, P1 → NO

Operating System Fall 2009 Ch07 - Deadlocks Resource-Allocation Graph  The Resource-Allocation Graph is not applicable  to a resource allocation system with multiple instance of each resource type

Operating System Fall 2009 Ch07 - Deadlocks Banker’s Algorithm  Deadlock avoidance algorithm  Supports multiple instances.  Less efficient than the resource-allocation graph scheme  Named because this algorithm could be used in a bank.  Assumption:  Each process must make a priori claim for maximum use.  When a process requests a resource it may have to wait.  When a process gets all its resources it must return them in a finite amount of time.  Two sub-algorithms  Safety Algorithm  Resource-Request Algorithm

Operating System Fall 2009 Ch07 - Deadlocks Data Structures for the Banker’s Algorithm  Available: Vector of length m.  If Available [j] = k, there are k instances of resource type R j available.  Max: n x m matrix.  If Max [i,j] = k, then process P i may request at most k instances of resource type R j.  Allocation: n x m matrix.  If Allocation[i,j] = k then P i is currently allocated k instances of R j.  Need: n x m matrix.  If Need[i,j] = k, then P i may need k more instances of R j to complete its task. Need [i,j] = Max[i,j] – Allocation [i,j]. Let n = number of processes, and m = number of resources types.

Operating System Fall 2009 Ch07 - Deadlocks Safety Algorithm 1.Let Work and Finish be vectors of length m and n, respectively. Initialize: Work = Available Finish [i] = false for i = 0,1, …, n-1. 2.Find an i such that both: (a) Finish [i] == false (b) Need i  Work If no such i exists, go to step 4. 3.Work = Work + Allocation i Finish[i] = true go to step 2. 4.If Finish [i] == true for all i, then the system is in a safe state. Algorithm that find out whether or not a system is in a safe state.  require max m x n 2 operations

Operating System Fall 2009 Ch07 - Deadlocks Resource-Request Algorithm for Process P i Let Request = request vector for process P i. If Request i [j] = k then process P i wants k instances of resource type R j. 1.If Request i  Need i go to step 2. Otherwise, raise error condition, since process has exceeded its maximum claim. 2.If Request i  Available, go to step 3. Otherwise P i must wait, since resources are not available. 3.Pretend to allocate requested resources to P i by modifying the state as follows: Available = Available – Request i ; Allocation i = Allocation i + Request i ; Need i = Need i – Request i ; Algorithm which determines if requests can be safely granted. If safe  the resources are allocated to P i. If unsafe  P i must wait, and the old resource-allocation state is restored

Operating System Fall 2009 Ch07 - Deadlocks Example of Banker’s Algorithm  5 processes P 0 through P 4 ; 3 resources  types A (10 instances), B (5 instances, and C (7 instances).  Snapshot at time T 0 : AllocationMaxAvailable A B CA B C A B C P P P P P

Operating System Fall 2009 Ch07 - Deadlocks Example (Cont.)  The content of the matrix. Need is defined to be Max – Allocation. Need A B C P P P P P  The system is in a safe state  since the sequence satisfies safety criteria.

Operating System Fall 2009 Ch07 - Deadlocks Example P 1 Request (1,0,2) (Cont.)  Check that Request  Available: that is, (1,0,2)  (3,3,2)  true. AllocationNeedAvailable A B CA B CA B C P P P P P  Executing safety algorithm shows that sequence satisfies safety requirement.  Can request for (3,3,0) by P4 be granted?  Can request for (0,2,0) by P0 be granted?

Operating System Fall 2009 Ch07 - Deadlocks Deadlock Detection  Allow system to enter deadlock state  Detection algorithm  Recovery scheme

Operating System Fall 2009 Ch07 - Deadlocks Single Instance of each Resource Type  Maintain wait-for graph  Nodes are processes.  P i  P j if P i is waiting for P j.  Periodically invoke an algorithm that searches for a cycle in the graph.  An algorithm to detect a cycle in a graph requires an order of n 2 operations,  where n is the number of vertices in the graph.

Operating System Fall 2009 Ch07 - Deadlocks Resource-Allocation Graph and Wait-for Graph Resource-Allocation GraphCorresponding Wait-for graph

Operating System Fall 2009 Ch07 - Deadlocks Several Instances of a Resource Type  Available: A vector of length m indicates the number of available resources of each type.  Allocation: An n x m matrix defines the number of resources of each type currently allocated to each process.  Request: An n x m matrix indicates the current request of each process.  If Request [I,j] = k, then process P i is requesting k more instances of resource type R j.

Operating System Fall 2009 Ch07 - Deadlocks Detection Algorithm (Cont.) – modified safety algorithm 1.Let Work and Finish be vectors of length m and n, respectively. Initialize: Work = Available For i=0, 1,.., n-1, if Allocation i ≠ 0, then Finish [i] = false; Otherwise, Finish [i] = true. 2.Find an index i such that both: (a) Finish [i] == false (b) Request i  Work If no such i exists, go to step 4. 3.Work = Work + Allocation i Finish[i] = true go to step 2. 4.If Finish [i] == false, for some i, 0  i  n-1, then the system is in deadlock state. Moreover, if Finish[i] == false, then P i is deadlocked. Algorithm requires an order of O(m x n 2) operations to detect whether the system is in deadlocked state.

Operating System Fall 2009 Ch07 - Deadlocks Example of Detection Algorithm  Five processes P 0 through P 4 ;  three resource types  A (7 instances), B (2 instances), and C (6 instances).  Snapshot at time T 0 : Allocation Request Available A B C A B C A B C P P P P P  Sequence will result in Finish[i] = true for all i.

Operating System Fall 2009 Ch07 - Deadlocks Example (Cont.)  P 2 requests an additional instance of type C. Request A B C P P P P P  State of system?  Can reclaim resources held by process P 0, but insufficient resources to fulfill other processes’ requests.  Deadlock exists, consisting of processes P 1, P 2, P 3, and P 4.

Operating System Fall 2009 Ch07 - Deadlocks Detection-Algorithm Usage  When, and how often, to invoke depends on:  How often a deadlock is likely to occur?  How many processes will need to be rolled back?  one for each disjoint cycle  If detection algorithm is invoked arbitrarily,  there may be many cycles in the resource graph  and so we would not be able to tell which of the many deadlocked processes “caused” the deadlock.

Operating System Fall 2009 Ch07 - Deadlocks Recovery from Deadlock  When a detection algorithm determines that a deadlock exists, several alternatives are available:  To inform the operator that a deadlock has occurred and to let the operator deal with the deadlock manually.  Process Termination  To abort one or more processes to break the circular wait  Resource Preemption.  To preempt some resources from one or more of the deadlocked processes.

Operating System Fall 2009 Ch07 - Deadlocks Recovery from Deadlock: Process Termination 1.Abort all deadlocked processes. 2.Abort one process at a time until the deadlock cycle is eliminated.  In which order should we choose to abort?  Priority of the process.  How long process has computed, and how much longer to completion.  Resources the process has used.  Resources process needs to complete.  How many processes will need to be terminated.  Is process interactive or batch?

Operating System Fall 2009 Ch07 - Deadlocks Recovery from Deadlock: Resource Preemption  Three issues in resource preemption 1.Selecting a victim – minimize cost. 2.Rollback – return to some safe state, restart process from that state. 3.Starvation – same process may always be picked as victim.

Operating System Fall 2009 Ch07 - Deadlocks Summary  A deadlock state occurs  when two or more processes are waiting indefinitely for an event that can be caused only by one of the waiting processes.  Three principle methods for dealing with deadlocks  Deadlock prevention, Deadlock avoidance  Deadlock detection-recovery scheme.  Ignore the deadlock problem  A deadlock requires that four conditions hold simultaneously  Mutual exclusion, hold and wait, no preemption, circular wait  Deadlock prevention ensures that at least one of the necessary conditions never holds.  Deadlock avoidance requires a priori information such as the maximum number of each resource classes.  Bankers algorithm  When deadlock is detected, the system must recover either by terminating some of the deadlocked processes or by preempting resources from some of the deadlocked processes.

Mi-Jung Choi Dept. of Computer and Science Silberschatz, Galvin and Gagne ©2006 Operating System Principles End of Chapter 7