Received-SPF: pass (google.com: domain of linux-kernel+bounces-180821-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
From: Wu Bo <bo.wu@vivo.com>
To: linux-kernel@vger.kernel.org
Cc: Jens Axboe <axboe@kernel.dk>,
	Bart Van Assche <bvanassche@acm.org>,
	linux-block@vger.kernel.org,
	Wu Bo <wubo.oduw@gmail.com>,
	Wu Bo <bo.wu@vivo.com>,
	stable@vger.kernel.org
Subject: [PATCH stable] block/mq-deadline: fix different priority request on the same zone
Date: Thu, 16 May 2024 03:28:38 -0600
Message-Id: <20240516092838.1790674-1-bo.wu@vivo.com>
Content-Transfer-Encoding: 8bit
Content-Type: text/plain
Precedence: bulk
MIME-Version: 1.0
X-MS-Exchange-AntiSpam-MessageData-ChunkCount: 1
X-MS-Exchange-AntiSpam-MessageData-0:
	=?us-ascii?Q?dCqWOKGRePjkRa8c8gEbL4sR2tnqylsKvUgevYEvhrQMS6qdfyy84zwTYPnw?=
 =?us-ascii?Q?X/EnOD6ORDPtEUHada5HH2fMCirQPAvcQAFA/a/1FxRHc9BXd0lbLKkQLR69?=
 =?us-ascii?Q?MPiuHLOebBRo0x/F8483f9Tk48g1lI+nX9FK1u1zQCaH83zyNWL8+gOhPSCp?=
 =?us-ascii?Q?r/fRhukYgszVrFA1wYQptlzgU4A7j0aVrKcWmSlwgg/71yDn6KxA7EDUimah?=
 =?us-ascii?Q?x994TwWwBmYyAaMK8RWdXHU1w27VC5x+hhDn7In62JcZunSznXnAJntlbJ0Z?=
 =?us-ascii?Q?67C3iWTXqkYTfbDAQt8z5M3beTaN6LB2Xtp8yjdfwYg0yOYYMcaTq68c+s0S?=
 =?us-ascii?Q?oOAVAooT2bkqRWm42PiWVuSOgusc4QLtEQjNyWdQsAFqhwYg+GiV4JQF5TRL?=
 =?us-ascii?Q?d0VAR6v51uOxW3ACYjhnA3d9niKQWqODa2zk6NNHOPbTAuLZJm2HeM3wRswG?=
 =?us-ascii?Q?aB21ocFum6cM9Fg+NieUvsgWKTWAasIP2K8OXNQH9gmxrj9ySzqHybzcPUfP?=
 =?us-ascii?Q?SC1dG6mxh/5RhcUlN+y9BMraoZH/jFW8HiPZGXpYkdJVENkmxBbofPtIvUkE?=
 =?us-ascii?Q?84hfTzRcNt7AuLiRhTEPvmdxNjVdkWlr3ANn4PyY5Zw5oB80IKK8/6dVf2tf?=
 =?us-ascii?Q?cM8Aw05x2x0l0yf9yftJSAW1OKS86b+o4pprImZOEZ88uS0aX61HC1wsca2c?=
 =?us-ascii?Q?EPZu4Sy4UikwVft9BSsJHZqyQi/3EhZ5k7znSlUKh2YAA2yOp5YNaVEAoeHb?=
 =?us-ascii?Q?Zw36yQIQEqDuAQglZ1wl2hfpMiPF7QhILMX+ks3R6WiJQd5Tp4MYcWoiAp7a?=
 =?us-ascii?Q?BPybStFUBK+h15PhfUp2VbLm+yoKYRxvoG79Tkm234WO+v+PBXx4iX7owzzX?=
 =?us-ascii?Q?AhplPA2F9NaAYcH4BD2CS11qBAHJJb7AW+fJbvQN2GCvfAmhmZ5fJOJ7Vq+s?=
 =?us-ascii?Q?wDfY7RX/211v3ZmULrw79MsAqHj54dZ1W5ULrYCNIaXwTXKNmqAXkKjn8EH6?=
 =?us-ascii?Q?dInnDccwGBlFPOwfiVvjX/b7oKes+6MVuJJ1KGr8JnzpRWHPjxoTbYLD1jMb?=
 =?us-ascii?Q?39Kzg45ruPOqpJobqA7RH2vR3DWN6NVsXOEi9B7hHhBDqU3eW4Ey2rHQKVs0?=
 =?us-ascii?Q?vzf4tFnV1xMvJ0QPv4Fvofl+8SbFL1asev6XBLeofipXdOm5da8rEbIYQf5o?=
 =?us-ascii?Q?FBtBhaA1gndqW/RknOqdhg46iyqr2WhPUISmrghEX2kMMsK31XcPeOD3PWrG?=
 =?us-ascii?Q?/28Vd6YlJHZL3jWJFHaNcQcLtE7x6D0AxAlNbLNvcB2rSFIEJLIYw5tL2p+Y?=
 =?us-ascii?Q?5WJlvFPheb5+5BiDAebXfh/yq9/Ulf0llfpjt4OEOKEQBdD7q60gr8CRgAmj?=
 =?us-ascii?Q?/oD0jjoSTJ5jm5W3SXqJEdDfE9F8TrXDQFhRexbGb3fPd8bA3tlpM3ma2cCk?=
 =?us-ascii?Q?w1RaMjgwe+kZ6+nbxWY62yZL8Szp9Tk/udWosiKhZgs2YU4NsT3k0KvhB7b9?=
 =?us-ascii?Q?rgtVVe1veYy1T1xybL0pLi3+X25VHWpNJmSYmvyS6tPhLuF+RmxFJn1jQLJB?=
 =?us-ascii?Q?BmD8UVupeTmBkz2DmyHktLa5GHl4Jlpub1Je1q82?=
X-OriginatorOrg: vivo.com
X-MS-Exchange-CrossTenant-Network-Message-Id: 9daf06ce-4d92-405d-2850-08dc7588b36c
X-MS-Exchange-CrossTenant-AuthSource: PSAPR06MB4486.apcprd06.prod.outlook.com
X-MS-Exchange-CrossTenant-AuthAs: Internal
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 16 May 2024 09:15:16.8082
 (UTC)
X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted
X-MS-Exchange-CrossTenant-Id: 923e42dc-48d5-4cbe-b582-1a797a6412ed
X-MS-Exchange-CrossTenant-MailboxType: HOSTED
X-MS-Exchange-CrossTenant-UserPrincipalName: 0Rs4zL32OrUMYsjpjYxiv55ByBV6UrOh4p7CDJZlvadj+xgMCMhvSjKJzhjjkoRuVzlThivf3OhizFiUMZRu3w==
X-MS-Exchange-Transport-CrossTenantHeadersStamped: KL1PR06MB6259

Zoned devices request sequential writing on the same zone. That means
if 2 requests on the saem zone, the lower pos request need to dispatch
to device first.
While different priority has it's own tree & list, request with high
priority will be disptch first.
So if requestA & requestB are on the same zone. RequestA is BE and pos
is X+0. ReqeustB is RT and pos is X+1. RequestB will be disptched before
requestA, which got an ERROR from zoned device.

This is found in a practice scenario when using F2FS on zoned device.
And it is very easy to reproduce:
1. Use fsstress to run 8 test processes
2. Use ionice to change 4/8 processes to RT priority

Fixes: c807ab520fc3 ("block/mq-deadline: Add I/O priority support")
Cc: <stable@vger.kernel.org>
Signed-off-by: Wu Bo <bo.wu@vivo.com>
---
 block/mq-deadline.c    | 31 +++++++++++++++++++++++++++++++
 include/linux/blk-mq.h | 15 +++++++++++++++
 2 files changed, 46 insertions(+)

diff --git a/block/mq-deadline.c b/block/mq-deadline.c
index 02a916ba62ee..6a05dd86e8ca 100644
--- a/block/mq-deadline.c
+++ b/block/mq-deadline.c
@@ -539,6 +539,37 @@ static struct request *__dd_dispatch_request(struct deadline_data *dd,
 	if (started_after(dd, rq, latest_start))
 		return NULL;
 
+	if (!blk_rq_is_seq_zoned_write(rq))
+		goto skip_check;
+	/*
+	 * To ensure sequential writing, check the lower priority class to see
+	 * if there is a request on the same zone and need to be dispatched
+	 * first
+	 */
+	ioprio_class = dd_rq_ioclass(rq);
+	prio = ioprio_class_to_prio[ioprio_class];
+	prio++;
+	for (; prio <= DD_PRIO_MAX; prio++) {
+		struct request *temp_rq;
+		unsigned long flags;
+		bool can_dispatch;
+
+		if (!dd_queued(dd, prio))
+			continue;
+
+		temp_rq = deadline_from_pos(&dd->per_prio[prio], data_dir, blk_rq_pos(rq));
+		if (temp_rq && blk_req_zone_in_one(temp_rq, rq) &&
+				blk_rq_pos(temp_rq) < blk_rq_pos(rq)) {
+			spin_lock_irqsave(&dd->zone_lock, flags);
+			can_dispatch = blk_req_can_dispatch_to_zone(temp_rq);
+			spin_unlock_irqrestore(&dd->zone_lock, flags);
+			if (!can_dispatch)
+				return NULL;
+			rq = temp_rq;
+			per_prio = &dd->per_prio[prio];
+		}
+	}
+skip_check:
 	/*
 	 * rq is the selected appropriate request.
 	 */
diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h
index d3d8fd8e229b..bca1e639e0f3 100644
--- a/include/linux/blk-mq.h
+++ b/include/linux/blk-mq.h
@@ -1202,6 +1202,15 @@ static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
 		return true;
 	return !blk_req_zone_is_write_locked(rq);
 }
+
+static inline bool blk_req_zone_in_one(struct request *rq_a,
+		struct request *rq_b)
+{
+	unsigned int zone_sectors = rq_a->q->limits.chunk_sectors;
+
+	return round_down(blk_rq_pos(rq_a), zone_sectors) ==
+		round_down(blk_rq_pos(rq_b), zone_sectors);
+}
 #else /* CONFIG_BLK_DEV_ZONED */
 static inline bool blk_rq_is_seq_zoned_write(struct request *rq)
 {
@@ -1229,6 +1238,12 @@ static inline bool blk_req_can_dispatch_to_zone(struct request *rq)
 {
 	return true;
 }
+
+static inline bool blk_req_zone_in_one(struct request *rq_a,
+		struct request *rq_b)
+{
+	return false;
+}
 #endif /* CONFIG_BLK_DEV_ZONED */
 
 #endif /* BLK_MQ_H */
-- 
2.35.3