Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Nicholas Piggin <npiggin@gmail.com>
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     Nicholas Piggin <npiggin@gmail.com>,
        Ingo Molnar <mingo@redhat.com>, Will Deacon <will@kernel.org>,
        Waiman Long <longman@redhat.com>,
        Boqun Feng <boqun.feng@gmail.com>, linux-kernel@vger.kernel.org
Subject: [PATCH 10/13] locking/qspinlock: paravirt use simple trylock in case idx overflows
Date:   Tue,  5 Jul 2022 00:38:17 +1000
Message-Id: <20220704143820.3071004-11-npiggin@gmail.com>
In-Reply-To: <20220704143820.3071004-1-npiggin@gmail.com>
References: <20220704143820.3071004-1-npiggin@gmail.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

The mcs node overflow fallback locking case does not have to use
pv_hybrid_queued_unfair_trylock as the trylock, which reduces the
differences between pv and !pv cases.

This was likely an artifact of function renaming making it use the pv
trylock, which is not a bug but it is unexpected.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 kernel/locking/qspinlock.c | 9 ++-------
 1 file changed, 2 insertions(+), 7 deletions(-)

diff --git a/kernel/locking/qspinlock.c b/kernel/locking/qspinlock.c
index 1d5b3443772c..cef0ca7d94e1 100644
--- a/kernel/locking/qspinlock.c
+++ b/kernel/locking/qspinlock.c
@@ -798,13 +798,8 @@ static inline void queued_spin_lock_mcs_queue(struct qspinlock *lock, bool parav
 	 */
 	if (unlikely(idx >= MAX_NODES)) {
 		lockevent_inc(lock_no_node);
-		if (paravirt) {
-			while (!pv_hybrid_queued_unfair_trylock(lock))
-				cpu_relax();
-		} else {
-			while (!queued_spin_trylock(lock))
-				cpu_relax();
-		}
+		while (!queued_spin_trylock(lock))
+			cpu_relax();
 		goto release;
 	}
 
-- 
2.35.1