linux-kernel - [RFC PATCH] kernel: locking: rwsem optimization for rtlinux

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <1652257281-11494-1-git-send-email-zhaoyang.huang@unisoc.com>
Date:   Wed, 11 May 2022 16:21:21 +0800
From:   "zhaoyang.huang" <zhaoyang.huang@...soc.com>
To:     Thomas Gleixner <tglx@...utronix.de>,
        Sebastian Andrzej Siewior <bigeasy@...utronix.de>,
        Zhaoyang Huang <huangzhaoyang@...il.com>,
        <linux-kernel@...r.kernel.org>, <xuewen.yan@...soc.com>,
        <ke.wang@...soc.com>, <ben.dai@...soc.com>,
        <shian.wang@...soc.com>, <linux-rt-users@...r.kernel.org>
Subject: [RFC PATCH] kernel: locking: rwsem optimization for rtlinux

From: Zhaoyang Huang <zhaoyang.huang@...soc.com>

With regard to rtlinux, releasing the lock to RT waiter could be helpful to
increase its responding time even if cfs writer listed at the head of the
wait_list.

Signed-off-by: Zhaoyang Huang <zhaoyang.huang@...soc.com>
---
 kernel/locking/rwsem-xadd.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/kernel/locking/rwsem-xadd.c b/kernel/locking/rwsem-xadd.c
index e795908..1b3f836 100644
--- a/kernel/locking/rwsem-xadd.c
+++ b/kernel/locking/rwsem-xadd.c
@@ -138,6 +138,18 @@ static void __rwsem_mark_wake(struct rw_semaphore *sem,
 	waiter = list_first_entry(&sem->wait_list, struct rwsem_waiter, list);
 
 	if (waiter->type == RWSEM_WAITING_FOR_WRITE) {
+		if ((waiter->task->policy != SCHED_NORMAL)
+			&&(waiter->task->policy != SCHED_BATCH))
+			;
+		else {
+			list_for_each_entry(tmp, sem->wait_list.next, list) {
+				if ((tmp->task->policy != SCHED_NORMAL)
+					&&(tmp->task->policy != SCHED_BATCH)) {
+					wake_q_add(wake_q, tmp->task);
+					return;
+				}
+			}
+		}
 		if (wake_type == RWSEM_WAKE_ANY) {
 			/*
 			 * Mark writer at the front of the queue for wakeup.
-- 
1.9.1