From git@z Thu Jan  1 00:00:00 1970
Subject: [PATCH v2] net: make napi_disable() symmetric with enable
From: Jakub Kicinski <kuba@kernel.org>
Date: Fri, 24 Sep 2021 13:24:53 -0700
Message-Id: <20210924202453.1051687-1-kuba@kernel.org>
To: davem@davemloft.net
Cc: netdev@vger.kernel.org, eric.dumazet@gmail.com, weiwan@google.com, xuanzhuo@linux.alibaba.com, Jakub Kicinski <kuba@kernel.org>
List-Id: <netdev.vger.kernel.org>
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit

Commit 3765996e4f0b ("napi: fix race inside napi_enable") fixed
an ordering bug in napi_enable() and made the napi_enable() diverge
from napi_disable(). The state transitions done on disable are
not symmetric to enable.

There is no known bug in napi_disable() this is just refactoring.

Eric suggests we can also replace msleep(1) with a more opportunistic
usleep_range().

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 net/core/dev.c | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)

diff --git a/net/core/dev.c b/net/core/dev.c
index f24c3a9..f0a556a 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -6386,18 +6386,25 @@ EXPORT_SYMBOL(netif_napi_add);
 
 void napi_disable(struct napi_struct *n)
 {
+	unsigned long val, new;
+
 	might_sleep();
 	set_bit(NAPI_STATE_DISABLE, &n->state);
 
-	while (test_and_set_bit(NAPI_STATE_SCHED, &n->state))
-		msleep(1);
-	while (test_and_set_bit(NAPI_STATE_NPSVC, &n->state))
-		msleep(1);
+	do {
+		val = READ_ONCE(n->state);
+		if (val & (NAPIF_STATE_SCHED | NAPIF_STATE_NPSVC)) {
+			usleep_range(20, 200);
+			continue;
+		}
+
+		new = val | NAPIF_STATE_SCHED | NAPIF_STATE_NPSVC;
+		new &= ~(NAPIF_STATE_THREADED);
+	} while (cmpxchg(&n->state, val, new) != val);
 
 	hrtimer_cancel(&n->timer);
 
 	clear_bit(NAPI_STATE_DISABLE, &n->state);
-	clear_bit(NAPI_STATE_THREADED, &n->state);
 }
 EXPORT_SYMBOL(napi_disable);
 
-- 
2.31.1