Deadlocks (Class #16 (contd.

))
 Outline:
 Deadlock problem
 System Model
 Deadlock Characterization
 Methods for Handling Deadlocks
 Deadlock Prevention
 Deadlock Avoidance
 Deadlock Detection
 Recovery from Deadlock
 Combined Approach to Deadlock Handling

8.1
 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.
 P1 and P2 each hold one tape drive and each needs another one.
 Example
 semaphores A and B, initialized to 1

P0 P1
wait (A); wait(B)
wait (B); wait(A)

8.2
 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.

8.3
 System Model
 A set of processes is in a deadlock state when every
process in the set is waiting for an event that can be
caused by another process in the set.
 Events:
 Resource acquisition and release.
 Resource types R , R , . . ., R
1 2 n
 Physical resources:CPU cycles, memory space, I/O devices
 Logical resources: files, semaphores, and monitors
 Each resource type R has W instances.
i i
 Each process utilizes a resource as follows:
 Request: if the request is not granted , then it must wait.
 Use: The process can operate on the resource.
 Release: The process releases the resources.
 Multi-threaded programs are good candidates for
deadlock because multiple threads compete for shared
resources.
8.4
 Deadlock Characterization
Deadlock can arise if four conditions hold simultaneously.
 Mutual exclusion: only one process at a time can use at least
one resource. Resources are non sharable.
 Hold and wait: a process holding at least one resource is
waiting to acquire additional resources held by other processes.
 No preemption: Resources can not be preempted; that is 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 {P0, P1, …Pn-1,Pn, P0} of
waiting processes such that P0 is waiting for a resource that is
held by P1, P1 is waiting for a resource that is held by
P2, …, Pn–1 is waiting for a resource that is held by
Pn, and Pn is waiting for a resource that is held by P0.

ALL FOUR CONDITIONS MUST HOLD.

8.5
 Resource-Allocation Graph
A tool to describe the deadlock.

A set of vertices V and a set of edges E.

 V is partitioned into two types:
 P = {P1, P2, …, Pn}, the set consisting of all the processes in
the system.

 R = {R1, R2, …, Rm}, the set consisting of all resource types

in the system.
 request edge – directed edge P1  Rj
 assignment edge – directed edge Rj  Pi

8.6
 Resource-Allocation Graph (Cont.)

 Process

 Resource Type with 4 instances

 Pi requests instance of Rj
Pi
Rj

 Pi is holding an instance of Rj

Pi
Rj

8.7
Example of a Resource Allocation Graph

8.8
Resource Allocation Graph With A Deadlock

8.9
Resource Allocation Graph With A Cycle But No Deadlock

8.10
 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.

8.11
 Methods for Handling Deadlocks

 Three ways
 Ensure that the system will never enter a
deadlock state.

 Allow the system to enter a deadlock

state and then recover.

 Ignore the problem and pretend that

deadlocks never occur in the system;
used by most operating systems,
including UNIX.

8.12
 Methods for Handling Deadlocks
 To ensure that deadlocks never occur, the system can
use either deadlock prevention and deadlock-avoidance
schemes.
 Deadlock prevention: set of methods for ensuring that at least
one of the necessary conditions can not hold.
 Deadlock avoidance: Requires that operating system be given
in advance additional information concerning which resources
a process will request and use during its life time.
 Deadlock detection: examines the state of the system to
determine whether a deadlock has occurred or not.
 Alternatively, assume that deadlock would not occur.
Resort to manual recovery when performance
degrades due to deadlock.

8.13
 Deadlock Prevention

 Deadlock prevention is a set of

methods for ensuring that at least
one of the necessary conditions can
not hold.

8.14
 Deadlock Prevention
Restrain the ways request can be made.
 Mutual Exclusion – It is not possible to prevent
deadlocks through ME, as some resources are
intrinsically non-sharable.
 Hold and Wait – must guarantee that whenever a
process requests a resource, it does not hold any other
resources.
 Protocol 1: Requires a process to request and be
allocated all its resources before it begins execution.
 System calls requesting resources for a process
precede all other system calls.
 Protocol 2: Allow process to request resources only
when the process has none.
 A process can request some resources and use
them. Before, it can request additional resources, it
must release all the resources that it is currently
allocated.
8.15
 Deadlock Prevention
 Hold and Wait – Example
 Consider a process that copies data from tape drive to a disk file and then prints
the results to the printer.
 Protocol1: If all the resources are requested at the beginning of a process, the
process must request the tape drive, disk file, and printer.
 It will the printer during entire execution, even though it needs at the end.
 Protocol 2: It copies data from tape drive to disk file and releases them. It gain
requests disk file and printer.
 Disadvantages:
 Resource utilization is very low
 Resource may be allocated but unused for a long time.
 Starvation is possible
 A process that needs several resources may have to wait indefinitely, at
least one of the resources that it needs is always allocated to another
process.
 Inefficient
 Delays process initiation
 Future resource requirements must be known

8.16
 Deadlock Prevention (Cont.)
 No Preemption –
 Protocol: 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.
 Possible for CPU registers and memory space whose state can be
restored later, but not possible for printers and tape drives.

8.17
 Deadlock Prevention (Cont.)

 Circular Wait –
 F:RN is a one-to-one function, where N is a set of natural
numbers.
 Protocol 1: impose a total ordering of all resource types,
and require that each process requests resources in an
increasing order of enumeration.
 A process initially can request any number of instances
of type Ri. Then the process can request instances of a
resource type Rj if and only if F(Rj) >F(Ri).
 Protocol 2: Whenever a process requests an instance of
resource type Rj, it has released any other resources Ri
such that F(Ri) >=F(Rj).
 Ordering

8.18
 Deadlock prevention
 Circular Wait:
 “witness” system call is implemented in BSD UNIX for lock order
verifier
 “witness” uses mutual-exclusion locks to protect critical regions and
maintains the relationship of lock orders in the system
 Suppose thread_one aquires locks in the order first_mutex and
second_mutex. Then witness records that first_mutex should be
obtained before second_mutex. If thread_two later aquires locks out
of order, witness generates a warning message.

thread_one thread_two:
{ {
lock(first_mutex); lock(second_mutex);
lock(second_mutex); lock (first_mutex);
. .
. .
} }
8.19
 Deadlock Avoidance
 Deadlock prevention algorithms
 Low device utilization, and reduced system
throughput.
 Alternative method is get additional
information about the processes
 Resources currently available, resources
currently allocated to a process, and future
requests and releases of each process.
 Various algorithms differ about amount and
type of information required.

8.20
 Deadlock Avoidance (Class #18)

Requires that the system has some additional a priori information

available.
 Simplest and most useful model 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.

8.21
 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 <P , P , …, P > is safe if for each P , the resources that

1 2 n i
Pi can still request can be satisfied by currently available resources
+ resources held by all the Pj, with j<I.
 If P resource needs are not immediately available, then P can wait
i i
until all Pj have finished.
 When P is finished, P can obtain needed resources, execute, return
j i
allocated resources, and terminate.
 When P terminates, P can obtain its needed resources, and so on.
i i+1

8.22
 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.

8.23
Safe, Unsafe , Deadlock State

8.24
 Example
 Consider a system with 12 magnetic tape drives and three processes
P0,P1,P2.
 P0 requires 10 tape drives, P1 may need 4, and P2 may need up to 9
tape drives.
 Suppose at time t0, P0 is holding 5 tape drives, P1 holding 2, and P2 is
holding 2 tape drives.
 There are 3 free tape drives

Maximum needs allocated current needs

P0 10 5 5
P1 4 2 2
P2 9 2 7

 At t0, system is in safe condition.

 <P1,P0,P2> satisfies satisfy safety condition.
 Suppose, at t1, P2 requests and is allocated 1 more tape drive.
 The system is no longer in safe state.
 The mistake was in granting the request of P2 for one more tape drive.
 The main idea of avoidance algorithm is to ensure the system is always
in a safe state.
8.25
 Two algorithms for Deadlock
Avoidance
 Resource-Allocation Graph algoritm
 Bankers algoritm

8.26
 Resource-Allocation Graph Algorithm
 If we have a resource-allocation (RA) system with one instance of
each resource type, we can use this approach.
 Claim edge Pi  Rj indicated that process Pj may request resource Rj;
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.
 Before the process starts executing, all its claim edges must appear
in the RA graph.
 Suppose a process Pi requests a resource Rj. The request can be
granted if converting the request edge Pi  Rj to an assignment
edge Rj Pi does not result in the formation of a cycle in the RA
graph.

8.27
 Resource-Allocation Graph Algorithm
 Cycle detection algorithm is used to detect
a cycle.
 If there are n processes detecting a cycle
requires an order of n*n operations.

8.28
 Resource-Allocation Graph For Deadlock Avoidance

Suppose P2 requests R2

8.29
 Unsafe State In Resource-Allocation Graph

Cycle in RA graph! So system is in an unsafe state. So we can

not allocate R2 to P2.
8.30
 Banker’s Algorithm
 Bank never allocates its available cash such
that it no longer satisfy the needs of all
customers.

 Multiple instances.

 Each process must a priori claim 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.

8.31
 Data Structures for the Banker’s Algorithm
Let n = number of processes, and m = number of resources types.
 Available: Vector of length m. If available [j] = k, there are k
instances of resource type Rj available.
 Max: n x m matrix. If Max [i,j] = k, then process Pi may request at
most k instances of resource type Rj.
 Allocation: n x m matrix. If Allocation[i,j] = k then Pi is currently
allocated k instances of Rj.
 Need: n x m matrix. If Need[i,j] = k, then Pi may need k more
instances of Rj to complete its task.

Need [i,j] = Max[i,j] – Allocation [i,j].

 Notation:
 Let X and Y be vectors of length n.
 We say X <= Y if and only if X[i] <= Y[i] for all i=1,2,…,n.
 For example, if X=(1,7,3,2) and Y=(0,3,2,1) then Y<= X.
 Y < X if Y<=X and Y # X.
8.32
 Resource-Request Algorithm for Process Pi
Request = request vector for process Pi. If Requesti [j] =
k then process Pi wants k instances of resource type Rj.
1. If Requesti  Needi go to step 2. Otherwise, raise error
condition, since process has exceeded its maximum claim.
2. If Requesti  Available, go to step 3. Otherwise Pi must wait,
since resources are not available.
3. Pretend to allocate requested resources to Pi by modifying the
state as follows:
Available = Available - Requesti;
Allocationi = Allocationi + Requesti;
Needi = Needi – Requesti;;
• If safe  the resources are allocated to Pi.
• If unsafe  Pi must wait, and the old resource-allocation
state is restored

8.33
 Safety Algorithm
1. Let Work and Finish be vectors of length m and n,
respectively. Initialize:
Work = Available
Finish [i] = false for i - 1,3, …, n.
2. Find and i such that both:
(a) Finish [i] = false
(b) Needi  Work
If no such i exists, go to step 4.
3. Work = Work + Allocationi
Finish[i] = true
go to step 2.
4. If Finish [i] == true for all i, then the system is in a safe
state.

• Safety algorithm may require an order of m*n2 operations.

8.34
 Example of Banker’s Algorithm

 5 processes P0 through P4; 3 resource types A

(10 instances), B (5instances, and C (7 instances).
 Snapshot at time T0:
Allocation Max Available
ABC ABC ABC
P0 010 753 332
P1 200 322
P2 302 902
P3 211 222
P4 002 433

8.35
 Example (Cont.)

 The content of the matrix. Need is defined to be Max –

Allocation.
Need
ABC
P0 743
P1 122
P2 600
P3 011
P4 431
 The system is in a safe state since the sequence < P1, P3, P4,
P2, P0> satisfies safety criteria.

8.36
 Example P1 Request (1,0,2) (Cont.)
 Check that Request  Available (that is, (1,0,2)  (3,3,2) 
true.
Allocation Need Available
ABC ABC ABC
P0 0 1 0 743 230
P1 302 020
P2 301 600
P3 211 011
P4 002 431
 Executing safety algorithm shows that sequence <P1, P3, P4,
P0, P2> satisfies safety requirement.
 Can request for (3,3,0) by P4 be granted?
 Can request for (0,2,0) by P0 be granted?

8.37
 Deadlock Detection
 Allow system to enter deadlock
state

 Detection algorithm
Resource allocation graph algorithm
Detection algorithm (similar to
Banker’s algoritm)
 Recovery scheme

8.38
 Single Instance of Each Resource
Graph based approach
 Maintain wait-for graph
 Nodes are processes.
 Pi  Pj if Pi is waiting for Pj.

 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 n2 operations, where n
is the number of vertices in the graph.

8.39
Resource-Allocation Graph and Wait-for Graph

Resource-Allocation Graph Corresponding wait-for graph

8.40
 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
[ij] = k, then process Pi is requesting k more
instances of resource type. Rj.

8.41
 Detection Algorithm
1.Let Work and Finish be vectors of length
m and n, respectively Initialize:
(a) Work = Available
(b) For i = 1,2, …, n, if Allocationi  0,
then
Finish[i] = false;otherwise, Finish[i] = true.
2.Find an index i such that both:
(a) Finish[i] == false
(b) Requesti  Work

If no such i exists, go to step 4.

8.42
 Detection Algorithm (Cont.)

3. Work = Work + Allocationi

Finish[i] = true
go to step 2.

4. If Finish[i] == false, for some i, 1  i  n, then

the system is in deadlock state. Moreover, if
Finish[i] == false, then Pi is deadlocked.

Algorithm requires an order of O(m x n2)

operations to detect whether the system is in
deadlocked state.

8.43
 Example of Detection Algorithm

 Five processes P0 through P4; three resource types

A (7 instances), B (2 instances), and C (6 instances).
 Snapshot at time T0:
Allocation Request Available
ABC ABC ABC
P0 0 1 0 000 000
P1 200 202
P2 303 000
P3 211 100
P4 002 002
 Sequence <P0, P2, P3, P1, P4> will result in Finish[i] = true for all
i.

8.44
 Example (Cont.)

 P2 requests an additional instance of type C.

Request
ABC
P0 0 0 0
P1 201
P2 001
P3 100
P4 002
 State of system?
 Can reclaim resources held by process P0, but insufficient
resources to fulfill other processes; requests.
 Deadlock exists, consisting of processes P1, P2, P3, and P4.

8.45
 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.
 Invoking for every request is expensive
 Invoke once per hour; when CPU utilization
drops below 40 %.
8.46
 Recovery from Deadlock: Process Termination

 Several alternatives exist.

Inform the operator
Automatic: system will recover from
the deadlock automatically
Two options exist for automatically
breaking a deadlock
Process termination
Resource preemption

8.47
 Recovery from Deadlock: Process Termination
 Abort all deadlocked processes.
 More expensive
 Abort one process at a time until the deadlock cycle is eliminated.
 Overhead: after the abort of each process detection algorithm have to be
invoked.
 Aborting a process is not easy.
 Printing or updating a file
 In which order should we choose to abort? (should incur minimum cost)
 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?

8.48
 Recovery from Deadlock: Resource Preemption
 We preempt some resources from one process and give
it to other processes. Until the deadlock cycle is broken.
 For preemption three issues are to be addressed
 Selecting a victim – minimize cost.
 Number of resources the deadlocked process is
holding, time consumed by it for execution.
 Rollback – return to some safe state, restart process
for that state.
 Rollback as far as necessary to break the deadlock.
 Starvation – same process may always be picked as
victim,
 Solution: include number of rollbacks in cost factor.

8.49
 Deadlock Handling: summary

 Three basic approaches

 Prevention: Use some protocol to prevent deadlocks,
ensuring the system will never enter a deadlock state.
 Low resource utilization
 Avoidance: Use a priori information; do not allow the
system to enter unsafe state.
 Needs a priori information about future requests.
 Detection: Allow the system to enter deadlock state:
then recover.
 Select the victim based on the cost factors.

8.50