Discuss methods for handling deadlocks

Answer 1

• ignore the problem altogether ie. ostrich algorithm it may occur very infrequently, cost of detection/prevention etc may not be worth it.
• detection and recovery
• avoidance by careful resource allocation
• prevention by structurally negating one of the four necessary conditions.

Deadlock PreventionDifference from avoidance is that here, the system itself is build in such a way that there are no deadlocks.

Make sure atleast one of the 4 deadlock conditions is never satisfied.

This may however be even more conservative than deadlock avoidance strategy.

• Attacking Mutex condition

• never grant exclusive access. but this may not be possible for several resources.

• Attacking preemption

• not something you want to do.

• Attacking hold and wait condition

• make a process hold at the most 1 resource at a time.
• make all the requests at the beginning. All or nothing policy. If you feel, retry. eg. 2-phase locking

• Attacking circular wait

• Order all the resources.
• Make sure that the requests are issued in the correct order so that there are no cycles present in the resource graph.

  Resources numbered 1 ... n. Resources can be requested only in increasing order. ie. you cannot request a resource whose no is less than any you may be holding.

Deadlock AvoidanceAvoid actions that may lead to a deadlock.

Think of it as a state machine moving from 1 state to another as each instruction is executed.

Safe State

Safe state is one where

• It is not a deadlocked state
• There is some sequence by which all requests can be satisfied.

To avoid deadlocks, we try to make only those transitions that will take you from one safe state to another. We avoid transitions to unsafe state (a state that is not deadlocked, and is not safe) eg.

Total # of instances of resource = 12

(Max, Allocated, Still Needs)

P0 (10, 5, 5) P1 (4, 2, 2) P2 (9, 2, 7) Free = 3 - Safe

The sequence is a reducible sequence

the first state is safe.

What if P2 requests 1 more and is allocated 1 more instance?

- results in Unsafe state

So do not allow P2's request to be satisfied.

Banker's Algorithm for Deadlock Avoidance

When a request is made, check to see if after the request is satisfied, there is a (atleast one!) sequence of moves that can satisfy all the requests. ie. the new state is safe. If so, satisfy the request, else make the request wait.

How do you find if a state is safen process and m resources

Max[n * m]

Allocated[n * m]

Still_Needs[n * m]

Available[m]

Temp[m]

Done[n]

while () {

Temp[j]=Available[j] for all j

Find an i such that

a) Done[i] = False

b) Still_Needs[i,j] <= Temp[j]

if so {

Temp[j] += Allocated[i,j] for all j

Done[i] = TRUE}

}

else if Done[i] = TRUE for all i then state is safe

else state is unsafe

}

Detection and RecoveryIs there a deadlock currently?

One resource of each type (1 printer, 1 plotter, 1 terminal etc.)

• check if there is a cycle in the resource graph. for each node N in the graph do DFS (depth first search) of the graph with N as the root In the DFS if you come back to a node already traversed, then there is a cycle. }

Multiple resources of each type

• m resources, n processes
• Max resources in existence = [E1, E2, E3, .... Em]
• Current Allocation = C1-n,1-m
• Resources currently Available = [A1, A2, ... Am]
• Request matrix = R1-n,1-m
• Invariant = Sum(Cij) + Aj = Ej
• Define A <= B for 2 vectors, A and B, if Ai <= Bi for all i
• Overview of deadlock detection algorithm,

  Check R matrix, and find a row i such at Ri < A.

  If such a process is found, add Ci to A and remove process i from the system.

  Keep doing this till either you have removed all processes, or you cannot remove any other process.

  Whatever is remaining is deadlocked.

  Basic idea, is that there is atleast 1 execution which will undeadlock the system

Recovery

• through preemption
• rollback
  • keep checkpointing periodically
  • when a deadlock is detected, see which resource is needed.
  • Take away the resource from the process currently having it.
  • Later on, you can restart this process from a check pointed state where it may need to reacquire the resource.
• killing processes
  • where possible, kill a process that can be rerun from the beginning without illeffects