blob: 7d3c82431909dd8120322e2360ce32cbd93f87e5
1 | Percpu rw semaphores |
2 | -------------------- |
3 | |
4 | Percpu rw semaphores is a new read-write semaphore design that is |
5 | optimized for locking for reading. |
6 | |
7 | The problem with traditional read-write semaphores is that when multiple |
8 | cores take the lock for reading, the cache line containing the semaphore |
9 | is bouncing between L1 caches of the cores, causing performance |
10 | degradation. |
11 | |
12 | Locking for reading is very fast, it uses RCU and it avoids any atomic |
13 | instruction in the lock and unlock path. On the other hand, locking for |
14 | writing is very expensive, it calls synchronize_rcu() that can take |
15 | hundreds of milliseconds. |
16 | |
17 | The lock is declared with "struct percpu_rw_semaphore" type. |
18 | The lock is initialized percpu_init_rwsem, it returns 0 on success and |
19 | -ENOMEM on allocation failure. |
20 | The lock must be freed with percpu_free_rwsem to avoid memory leak. |
21 | |
22 | The lock is locked for read with percpu_down_read, percpu_up_read and |
23 | for write with percpu_down_write, percpu_up_write. |
24 | |
25 | The idea of using RCU for optimized rw-lock was introduced by |
26 | Eric Dumazet <eric.dumazet@gmail.com>. |
27 | The code was written by Mikulas Patocka <mpatocka@redhat.com> |
28 |