Deadlocks

If two or more processes are waiting for happening of some event, which never happens, then we say that these processes are involved in a state that is called deadlock.
Two processes mutually waiting for an event, which never happens, this state is called deadlock.

4 necessary conditions for deadlock

Mutual Exclusion.
No Pre-emption.
Hold & Wait.
Circular Wait.

Resource Allocation Graph(RAG)

To represent how the resources are allocated to processes, assigned multiple resources, RAG is used. Two types - Single Instance RAG & Multiple Instance RAG.

Methods to handle deadlocks

Deadlock Ignorance(Ostrich method)
Trade off between deadlock ignorance and performance happens in this method. We just ignore the deadlock and do not write any code to prevent or manage it, so that it increases the performance of the OS. This is because the code for deadlock detection, prevention or correction is quite complex. Programmers tend to ignore it because deadlocks happen very rarely and they do not want to compromise on the performance of the OS because of this extra functionality.
Deadlock Prevention
Prevent the 4 necessary conditions of deadlock formation from occurring, or atleast try to remove one of the condition, effectively preventing a deadlock. Prevent any four or all four.
Deadlock Avoidance(Banker's Algorithm)
Every time we allocate resources to a process, we check wether it is safe or unsafe to do it, and wether it will cause a deadlock. Banker's algorithm solves this problem.
Deadlock Detection & Recovery
We detect wether there is a deadlock, and if a deadlock is found, we try to recover the system from it.