arno/linux-insides
1
0
Fork 0
mirror of https://github.com/0xAX/linux-insides.git synced 2024-12-22 14:48:08 +00:00
Code Issues Releases Wiki Activity

Merge pull request #663 from tlkh/patch-1

Browse Source 
Proposed improvements to explanation of spinlock
This commit is contained in:
0xAX 2019-03-18 14:15:38 +06:00 committed by GitHub
commit 58ee66cb11
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
SyncPrim/linux-sync-3.md
View File

@ -15,9 +15,9 @@ Introduction to the semaphores in the Linux kernel
So, what is it `semaphore`? As you may guess - `semaphore` is yet another mechanism for support of thread or process synchronization. The Linux kernel already provides implementation of one synchronization mechanism - `spinlocks`, why do we need in yet another one? To answer on this question we need to know details of both of these mechanisms. We already familiar with the `spinlocks`, so let's start from this mechanism. So, what is it `semaphore`? As you may guess - `semaphore` is yet another mechanism for support of thread or process synchronization. The Linux kernel already provides implementation of one synchronization mechanism - `spinlocks`, why do we need in yet another one? To answer on this question we need to know details of both of these mechanisms. We already familiar with the `spinlocks`, so let's start from this mechanism.
The main idea behind `spinlock` concept is a lock which will be acquired for a very short time. We can't sleep when a lock acquired by a process or thread, because other processes wait us. [Context switch](https://en.wikipedia.org/wiki/Context_switch) is not not allowed because [preemption](https://en.wikipedia.org/wiki/Preemption_%28computing%29) is disabled to avoid [deadlocks](https://en.wikipedia.org/wiki/Deadlock). `spinlock` creates a lock which will be acquired to protect a shared resource from being modified by more than one process. As a result, other processes that try to acquire the current lock get stopped (aka "spin-in-place" or busy waiting). [Context switch](https://en.wikipedia.org/wiki/Context_switch) is not allowed because [preemption](https://en.wikipedia.org/wiki/Preemption_%28computing%29) is disabled to avoid [deadlocks](https://en.wikipedia.org/wiki/Deadlock). As a result, `spinlock` should only be used if the lock will only be acquired for a very short period of time, otherwise amount of busy waiting accumulated by other processes results in extremely inefficient operation. For locks that need to be acquired for a relatively long period of time, we turn to `semaphore`.
In this way, [semaphores](https://en.wikipedia.org/wiki/Semaphore_%28programming%29) is a good solution for locks which may be acquired for a long time. In other way this mechanism is not optimal for locks that acquired for a short time. To understand this, we need to know what is `semaphore`. [semaphores](https://en.wikipedia.org/wiki/Semaphore_%28programming%29) is a good solution for locks which may be acquired for a long time. In other way this mechanism is not optimal for locks that acquired for a short time. To understand this, we need to know what is `semaphore`.
As usual synchronization primitive, a `semaphore` is based on a variable. This variable may be incremented or decremented and it's state will represent ability to acquire lock. Notice that value of the variable is not limited to `0` and `1`. There are two types of `semaphores`: As usual synchronization primitive, a `semaphore` is based on a variable. This variable may be incremented or decremented and it's state will represent ability to acquire lock. Notice that value of the variable is not limited to `0` and `1`. There are two types of `semaphores`: