]> git.ipfire.org Git - thirdparty/linux.git/commitdiff
tools/memory-model/Documentation: Fix SRCU section in explanation.txt
authorUladzislau Rezki (Sony) <urezki@gmail.com>
Tue, 22 Apr 2025 17:38:30 +0000 (19:38 +0200)
committerPaul E. McKenney <paulmck@kernel.org>
Wed, 23 Apr 2025 19:17:04 +0000 (12:17 -0700)
The SRCU read-side critical sections describes the difference between
srcu_down_read()/srcu_up_read() and srcu_read_lock()/srcu_read_unlock()
in a way that a last pair must occur on the same CPU.

This is not true, the srcu_read_unlock() can happen on any CPU, but it
must be performed by the same task that invoked srcu_read_lock().

Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
tools/memory-model/Documentation/explanation.txt

index 6dc8b3642458e5dcb64090b63ed9050517e32014..34aa3172071bd1f2742c0501729caef257ebcb8b 100644 (file)
@@ -1896,7 +1896,7 @@ following respects:
 
 3.     The srcu_down_read() and srcu_up_read() primitives work
        exactly like srcu_read_lock() and srcu_read_unlock(), except
-       that matching calls don't have to execute on the same CPU.
+       that matching calls don't have to execute within the same context.
        (The names are meant to be suggestive of operations on
        semaphores.)  Since the matching is determined by the domain
        pointer and index value, these primitives make it possible for