From: Willy Tarreau <w@1wt.eu>
Date: Mon, 14 Nov 2022 16:54:07 +0000 (+0100)
Subject: OPTIM: stick-table: avoid atomic ops in stktable_requeue_exp() when possible
X-Git-Tag: v2.7-dev9~70
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=3238f79d12871937c55ff05f2853679a0d511420;p=thirdparty%2Fhaproxy.git

OPTIM: stick-table: avoid atomic ops in stktable_requeue_exp() when possible

Since the task's time resolution is the millisecond we know there
will not be more than 1000 useful updates per second, so there's no
point in doing a CAS and a task_queue() for each call, better first
check if we're going to change the date. Now we're certain not to
perform such operations more than 1000 times a second for a given
table.

The loop was modified because this improvement will also be used to fix a
bug later.
---

diff --git a/src/stick_table.c b/src/stick_table.c
index 61e63b8ae6..1820da3d9d 100644
--- a/src/stick_table.c
+++ b/src/stick_table.c
@@ -550,11 +550,17 @@ void stktable_requeue_exp(struct stktable *t, const struct stksess *ts)
 
 	/* set the task's expire to the newest expiration date. */
 	old_exp = HA_ATOMIC_LOAD(&t->exp_task->expire);
-	do {
+	new_exp = tick_first(expire, old_exp);
+
+	/* let's not go further if we're already up to date */
+	if (new_exp == old_exp)
+		return;
+
+	while (new_exp != old_exp &&
+	       !HA_ATOMIC_CAS(&t->exp_task->expire, &old_exp, new_exp)) {
+		__ha_cpu_relax();
 		new_exp = tick_first(expire, old_exp);
-	} while (new_exp != old_exp &&
-		 !HA_ATOMIC_CAS(&t->exp_task->expire, &old_exp, new_exp) &&
-		 __ha_cpu_relax());
+	}
 
 	task_queue(t->exp_task);
 }