Oracle Internals: An Introduction

Latch Gets

When an Oracle process needs to access a data structure protected by a latch, it can request to get the latch in one of two modes—willing-to-wait mode or no-wait mode (also called immediate mode).

Willing-to-Wait Mode

Oracle expects latches to be held briefly and intermittently. So if a process attempts to get a latch in willing-to-wait mode and finds that the latch is not available, it will spin briefly and then try again. When a process spins, it executes a simple series of instructions a number of times, as a way of waiting before trying again. This is sometimes called an active wait because from the operating system’s perspective, the process is still actively consuming CPU cycles, although it is really just waiting a while.

The amount of CPU time that a process will burn before trying to get the latch once again is very small and fixed (although it was tunable in Oracle7 using the _LATCH_SPIN_COUNT parameter). If the next attempt to get the latch fails again, the procedure will be repeated up to the number of times specified by the _SPIN_COUNT parameter. This parameter normally defaults to 2000 iterations in multi-processor environments.

Why spin?

The idea of spinning is that another process executing on another CPU may release the latch, thereby allowing the spinning process to proceed. Of course, it makes no sense to spin on a machine with just one CPU, and so Oracle does not.

The alternative to spinning is to relinquish the CPU and allow another process to use it. ...