block-5.10-2020-11-07

-----BEGIN PGP SIGNATURE-----
 
 iQJEBAABCAAuFiEEwPw5LcreJtl1+l5K99NY+ylx4KYFAl+m/y4QHGF4Ym9lQGtl
 cm5lbC5kawAKCRD301j7KXHgplQwEACb6FFsEJRn2xnFkyZUQE1QAIEQJWsSAaV+
 xK0u+srqyy+sXvSSQ8wy6FsT3Fgs0CYU8SFkCW/F+EdnqzY0L03RTX7qU+vVtLXC
 CmAOQACKPKplCLROz9hontPwNIyheXzhkD+sBaFlCvgjMJpyZvQfkJSKGcgTkMy3
 ZgQB8j9+3qhkRhk0yA2gjW+uH2EKHyjYQSPUHDhgxzES+ZrwEhXmNc0eKbBDr74R
 +SYJYs2A5ZcauoCR0IYnPq2XiK3kbv3NMRWIgxVN1LWde7RqNo+iCgP1ps06Ax9i
 W5WPxMq9Y9IFacM/bCHIu+TsJMg8poWmYQ/VlCXT8deH2V1dWZ5L2HnhOPVInh3o
 P//3B+rCyGpMKe5s4JjfHddtDhrlbfvyIWg3LEqUxH5VmXtUYgq9pAVOwqaxZMkO
 U2cMWRUgDUVcta2ulQfcCqaQuIC5Mp+Up4HuYfe4jmYeKWMFCfPlCx0ZsIuR9hkz
 wcgODZLHMbEEMKydFnQDoQjngjIpYqr8TJAMjQ0P1c7sEYlVE0wXmQb/uXL7BPPa
 QU7+RxeWiMXDJTn9ZJAKkeWPvFC++JcE0QfZysSRO4eb/ny2i2xalJI6d8v4xwvU
 Q/VgVQeQnhbeves8hwsjQrjSU/oPBCiiMYwBRQikpGbRG5P2NadcEox0MfzS6ojT
 a/bQAoOR9A==
 =BcVw
 -----END PGP SIGNATURE-----

Merge tag 'block-5.10-2020-11-07' of git://git.kernel.dk/linux-block

Pull block fixes from Jens Axboe:

 - NVMe pull request from Christoph:
    - revert a nvme_queue size optimization (Keith Bush)
    - fabrics timeout races fixes (Chao Leng and Sagi Grimberg)"

 - null_blk zone locking fix (Damien)

* tag 'block-5.10-2020-11-07' of git://git.kernel.dk/linux-block:
  null_blk: Fix scheduling in atomic with zoned mode
  nvme-tcp: avoid repeated request completion
  nvme-rdma: avoid repeated request completion
  nvme-tcp: avoid race between time out and tear down
  nvme-rdma: avoid race between time out and tear down
  nvme: introduce nvme_sync_io_queues
  Revert "nvme-pci: remove last_sq_tail"

drivers/block/null_blk.h

@@ -47,7 +47,7 @@ struct nullb_device {
 	unsigned int nr_zones_closed;
 	struct blk_zone *zones;
 	sector_t zone_size_sects;
-	spinlock_t zone_dev_lock;
+	spinlock_t zone_lock;
 	unsigned long *zone_locks;
 
 	unsigned long size; /* device size in MB */

drivers/block/null_blk_zoned.c

@@ -46,11 +46,20 @@ int null_init_zoned_dev(struct nullb_device *dev, struct request_queue *q)
 	if (!dev->zones)
 		return -ENOMEM;
 
-	spin_lock_init(&dev->zone_dev_lock);
+	/*
-	dev->zone_locks = bitmap_zalloc(dev->nr_zones, GFP_KERNEL);
+	 * With memory backing, the zone_lock spinlock needs to be temporarily
-	if (!dev->zone_locks) {
+	 * released to avoid scheduling in atomic context. To guarantee zone
-		kvfree(dev->zones);
+	 * information protection, use a bitmap to lock zones with
-		return -ENOMEM;
+	 * wait_on_bit_lock_io(). Sleeping on the lock is OK as memory backing
+	 * implies that the queue is marked with BLK_MQ_F_BLOCKING.
+	 */
+	spin_lock_init(&dev->zone_lock);
+	if (dev->memory_backed) {
+		dev->zone_locks = bitmap_zalloc(dev->nr_zones, GFP_KERNEL);
+		if (!dev->zone_locks) {
+			kvfree(dev->zones);
+			return -ENOMEM;
+		}
 	}
 
 	if (dev->zone_nr_conv >= dev->nr_zones) {
@@ -137,12 +146,17 @@ void null_free_zoned_dev(struct nullb_device *dev)
 
 static inline void null_lock_zone(struct nullb_device *dev, unsigned int zno)
 {
-	wait_on_bit_lock_io(dev->zone_locks, zno, TASK_UNINTERRUPTIBLE);
+	if (dev->memory_backed)
+		wait_on_bit_lock_io(dev->zone_locks, zno, TASK_UNINTERRUPTIBLE);
+	spin_lock_irq(&dev->zone_lock);
 }
 
 static inline void null_unlock_zone(struct nullb_device *dev, unsigned int zno)
 {
-	clear_and_wake_up_bit(zno, dev->zone_locks);
+	spin_unlock_irq(&dev->zone_lock);
+
+	if (dev->memory_backed)
+		clear_and_wake_up_bit(zno, dev->zone_locks);
 }
 
 int null_report_zones(struct gendisk *disk, sector_t sector,
@@ -322,7 +336,6 @@ static blk_status_t null_zone_write(struct nullb_cmd *cmd, sector_t sector,
 		return null_process_cmd(cmd, REQ_OP_WRITE, sector, nr_sectors);
 
 	null_lock_zone(dev, zno);
-	spin_lock(&dev->zone_dev_lock);
 
 	switch (zone->cond) {
 	case BLK_ZONE_COND_FULL:
@@ -375,9 +388,17 @@ static blk_status_t null_zone_write(struct nullb_cmd *cmd, sector_t sector,
 	if (zone->cond != BLK_ZONE_COND_EXP_OPEN)
 		zone->cond = BLK_ZONE_COND_IMP_OPEN;
 
-	spin_unlock(&dev->zone_dev_lock);
+	/*
+	 * Memory backing allocation may sleep: release the zone_lock spinlock
+	 * to avoid scheduling in atomic context. Zone operation atomicity is
+	 * still guaranteed through the zone_locks bitmap.
+	 */
+	if (dev->memory_backed)
+		spin_unlock_irq(&dev->zone_lock);
 	ret = null_process_cmd(cmd, REQ_OP_WRITE, sector, nr_sectors);
-	spin_lock(&dev->zone_dev_lock);
+	if (dev->memory_backed)
+		spin_lock_irq(&dev->zone_lock);
+
 	if (ret != BLK_STS_OK)
 		goto unlock;
 
@@ -392,7 +413,6 @@ static blk_status_t null_zone_write(struct nullb_cmd *cmd, sector_t sector,
 	ret = BLK_STS_OK;
 
 unlock:
-	spin_unlock(&dev->zone_dev_lock);
 	null_unlock_zone(dev, zno);
 
 	return ret;
@@ -516,9 +536,7 @@ static blk_status_t null_zone_mgmt(struct nullb_cmd *cmd, enum req_opf op,
 			null_lock_zone(dev, i);
 			zone = &dev->zones[i];
 			if (zone->cond != BLK_ZONE_COND_EMPTY) {
-				spin_lock(&dev->zone_dev_lock);
 				null_reset_zone(dev, zone);
-				spin_unlock(&dev->zone_dev_lock);
 				trace_nullb_zone_op(cmd, i, zone->cond);
 			}
 			null_unlock_zone(dev, i);
@@ -530,7 +548,6 @@ static blk_status_t null_zone_mgmt(struct nullb_cmd *cmd, enum req_opf op,
 	zone = &dev->zones[zone_no];
 
 	null_lock_zone(dev, zone_no);
-	spin_lock(&dev->zone_dev_lock);
 
 	switch (op) {
 	case REQ_OP_ZONE_RESET:
@@ -550,8 +567,6 @@ static blk_status_t null_zone_mgmt(struct nullb_cmd *cmd, enum req_opf op,
 		break;
 	}
 
-	spin_unlock(&dev->zone_dev_lock);
-
 	if (ret == BLK_STS_OK)
 		trace_nullb_zone_op(cmd, zone_no, zone->cond);
 

drivers/nvme/host/core.c

@@ -4582,8 +4582,7 @@ void nvme_start_queues(struct nvme_ctrl *ctrl)
 }
 EXPORT_SYMBOL_GPL(nvme_start_queues);
 
-
+void nvme_sync_io_queues(struct nvme_ctrl *ctrl)
-void nvme_sync_queues(struct nvme_ctrl *ctrl)
 {
 	struct nvme_ns *ns;
 
@@ -4591,7 +4590,12 @@ void nvme_sync_queues(struct nvme_ctrl *ctrl)
 	list_for_each_entry(ns, &ctrl->namespaces, list)
 		blk_sync_queue(ns->queue);
 	up_read(&ctrl->namespaces_rwsem);
+}
+EXPORT_SYMBOL_GPL(nvme_sync_io_queues);
 
+void nvme_sync_queues(struct nvme_ctrl *ctrl)
+{
+	nvme_sync_io_queues(ctrl);
 	if (ctrl->admin_q)
 		blk_sync_queue(ctrl->admin_q);
 }

drivers/nvme/host/nvme.h

@@ -602,6 +602,7 @@ void nvme_stop_queues(struct nvme_ctrl *ctrl);
 void nvme_start_queues(struct nvme_ctrl *ctrl);
 void nvme_kill_queues(struct nvme_ctrl *ctrl);
 void nvme_sync_queues(struct nvme_ctrl *ctrl);
+void nvme_sync_io_queues(struct nvme_ctrl *ctrl);
 void nvme_unfreeze(struct nvme_ctrl *ctrl);
 void nvme_wait_freeze(struct nvme_ctrl *ctrl);
 int nvme_wait_freeze_timeout(struct nvme_ctrl *ctrl, long timeout);

drivers/nvme/host/pci.c

@@ -198,6 +198,7 @@ struct nvme_queue {
 	u32 q_depth;
 	u16 cq_vector;
 	u16 sq_tail;
+	u16 last_sq_tail;
 	u16 cq_head;
 	u16 qid;
 	u8 cq_phase;
@@ -455,11 +456,24 @@ static int nvme_pci_map_queues(struct blk_mq_tag_set *set)
 	return 0;
 }
 
-static inline void nvme_write_sq_db(struct nvme_queue *nvmeq)
+/*
+ * Write sq tail if we are asked to, or if the next command would wrap.
+ */
+static inline void nvme_write_sq_db(struct nvme_queue *nvmeq, bool write_sq)
 {
+	if (!write_sq) {
+		u16 next_tail = nvmeq->sq_tail + 1;
+
+		if (next_tail == nvmeq->q_depth)
+			next_tail = 0;
+		if (next_tail != nvmeq->last_sq_tail)
+			return;
+	}
+
 	if (nvme_dbbuf_update_and_check_event(nvmeq->sq_tail,
 			nvmeq->dbbuf_sq_db, nvmeq->dbbuf_sq_ei))
 		writel(nvmeq->sq_tail, nvmeq->q_db);
+	nvmeq->last_sq_tail = nvmeq->sq_tail;
 }
 
 /**
@@ -476,8 +490,7 @@ static void nvme_submit_cmd(struct nvme_queue *nvmeq, struct nvme_command *cmd,
 	       cmd, sizeof(*cmd));
 	if (++nvmeq->sq_tail == nvmeq->q_depth)
 		nvmeq->sq_tail = 0;
-	if (write_sq)
+	nvme_write_sq_db(nvmeq, write_sq);
-		nvme_write_sq_db(nvmeq);
 	spin_unlock(&nvmeq->sq_lock);
 }
 
@@ -486,7 +499,8 @@ static void nvme_commit_rqs(struct blk_mq_hw_ctx *hctx)
 	struct nvme_queue *nvmeq = hctx->driver_data;
 
 	spin_lock(&nvmeq->sq_lock);
-	nvme_write_sq_db(nvmeq);
+	if (nvmeq->sq_tail != nvmeq->last_sq_tail)
+		nvme_write_sq_db(nvmeq, true);
 	spin_unlock(&nvmeq->sq_lock);
 }
 
@@ -1496,6 +1510,7 @@ static void nvme_init_queue(struct nvme_queue *nvmeq, u16 qid)
 	struct nvme_dev *dev = nvmeq->dev;
 
 	nvmeq->sq_tail = 0;
+	nvmeq->last_sq_tail = 0;
 	nvmeq->cq_head = 0;
 	nvmeq->cq_phase = 1;
 	nvmeq->q_db = &dev->dbs[qid * 2 * dev->db_stride];

drivers/nvme/host/rdma.c

@@ -122,7 +122,6 @@ struct nvme_rdma_ctrl {
 	struct sockaddr_storage src_addr;
 
 	struct nvme_ctrl	ctrl;
-	struct mutex		teardown_lock;
 	bool			use_inline_data;
 	u32			io_queues[HCTX_MAX_TYPES];
 };
@@ -1010,8 +1009,8 @@ out_free_io_queues:
 static void nvme_rdma_teardown_admin_queue(struct nvme_rdma_ctrl *ctrl,
 		bool remove)
 {
-	mutex_lock(&ctrl->teardown_lock);
 	blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
+	blk_sync_queue(ctrl->ctrl.admin_q);
 	nvme_rdma_stop_queue(&ctrl->queues[0]);
 	if (ctrl->ctrl.admin_tagset) {
 		blk_mq_tagset_busy_iter(ctrl->ctrl.admin_tagset,
@@ -1021,16 +1020,15 @@ static void nvme_rdma_teardown_admin_queue(struct nvme_rdma_ctrl *ctrl,
 	if (remove)
 		blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
 	nvme_rdma_destroy_admin_queue(ctrl, remove);
-	mutex_unlock(&ctrl->teardown_lock);
 }
 
 static void nvme_rdma_teardown_io_queues(struct nvme_rdma_ctrl *ctrl,
 		bool remove)
 {
-	mutex_lock(&ctrl->teardown_lock);
 	if (ctrl->ctrl.queue_count > 1) {
 		nvme_start_freeze(&ctrl->ctrl);
 		nvme_stop_queues(&ctrl->ctrl);
+		nvme_sync_io_queues(&ctrl->ctrl);
 		nvme_rdma_stop_io_queues(ctrl);
 		if (ctrl->ctrl.tagset) {
 			blk_mq_tagset_busy_iter(ctrl->ctrl.tagset,
@@ -1041,7 +1039,6 @@ static void nvme_rdma_teardown_io_queues(struct nvme_rdma_ctrl *ctrl,
 			nvme_start_queues(&ctrl->ctrl);
 		nvme_rdma_destroy_io_queues(ctrl, remove);
 	}
-	mutex_unlock(&ctrl->teardown_lock);
 }
 
 static void nvme_rdma_free_ctrl(struct nvme_ctrl *nctrl)
@@ -1976,16 +1973,12 @@ static void nvme_rdma_complete_timed_out(struct request *rq)
 {
 	struct nvme_rdma_request *req = blk_mq_rq_to_pdu(rq);
 	struct nvme_rdma_queue *queue = req->queue;
-	struct nvme_rdma_ctrl *ctrl = queue->ctrl;
 
-	/* fence other contexts that may complete the command */
-	mutex_lock(&ctrl->teardown_lock);
 	nvme_rdma_stop_queue(queue);
-	if (!blk_mq_request_completed(rq)) {
+	if (blk_mq_request_started(rq) && !blk_mq_request_completed(rq)) {
 		nvme_req(rq)->status = NVME_SC_HOST_ABORTED_CMD;
 		blk_mq_complete_request(rq);
 	}
-	mutex_unlock(&ctrl->teardown_lock);
 }
 
 static enum blk_eh_timer_return
@@ -2320,7 +2313,6 @@ static struct nvme_ctrl *nvme_rdma_create_ctrl(struct device *dev,
 		return ERR_PTR(-ENOMEM);
 	ctrl->ctrl.opts = opts;
 	INIT_LIST_HEAD(&ctrl->list);
-	mutex_init(&ctrl->teardown_lock);
 
 	if (!(opts->mask & NVMF_OPT_TRSVCID)) {
 		opts->trsvcid =

drivers/nvme/host/tcp.c

@@ -124,7 +124,6 @@ struct nvme_tcp_ctrl {
 	struct sockaddr_storage src_addr;
 	struct nvme_ctrl	ctrl;
 
-	struct mutex		teardown_lock;
 	struct work_struct	err_work;
 	struct delayed_work	connect_work;
 	struct nvme_tcp_request async_req;
@@ -1886,8 +1885,8 @@ out_free_queue:
 static void nvme_tcp_teardown_admin_queue(struct nvme_ctrl *ctrl,
 		bool remove)
 {
-	mutex_lock(&to_tcp_ctrl(ctrl)->teardown_lock);
 	blk_mq_quiesce_queue(ctrl->admin_q);
+	blk_sync_queue(ctrl->admin_q);
 	nvme_tcp_stop_queue(ctrl, 0);
 	if (ctrl->admin_tagset) {
 		blk_mq_tagset_busy_iter(ctrl->admin_tagset,
@@ -1897,18 +1896,17 @@ static void nvme_tcp_teardown_admin_queue(struct nvme_ctrl *ctrl,
 	if (remove)
 		blk_mq_unquiesce_queue(ctrl->admin_q);
 	nvme_tcp_destroy_admin_queue(ctrl, remove);
-	mutex_unlock(&to_tcp_ctrl(ctrl)->teardown_lock);
 }
 
 static void nvme_tcp_teardown_io_queues(struct nvme_ctrl *ctrl,
 		bool remove)
 {
-	mutex_lock(&to_tcp_ctrl(ctrl)->teardown_lock);
 	if (ctrl->queue_count <= 1)
-		goto out;
+		return;
 	blk_mq_quiesce_queue(ctrl->admin_q);
 	nvme_start_freeze(ctrl);
 	nvme_stop_queues(ctrl);
+	nvme_sync_io_queues(ctrl);
 	nvme_tcp_stop_io_queues(ctrl);
 	if (ctrl->tagset) {
 		blk_mq_tagset_busy_iter(ctrl->tagset,
@@ -1918,8 +1916,6 @@ static void nvme_tcp_teardown_io_queues(struct nvme_ctrl *ctrl,
 	if (remove)
 		nvme_start_queues(ctrl);
 	nvme_tcp_destroy_io_queues(ctrl, remove);
-out:
-	mutex_unlock(&to_tcp_ctrl(ctrl)->teardown_lock);
 }
 
 static void nvme_tcp_reconnect_or_remove(struct nvme_ctrl *ctrl)
@@ -2171,14 +2167,11 @@ static void nvme_tcp_complete_timed_out(struct request *rq)
 	struct nvme_tcp_request *req = blk_mq_rq_to_pdu(rq);
 	struct nvme_ctrl *ctrl = &req->queue->ctrl->ctrl;
 
-	/* fence other contexts that may complete the command */
-	mutex_lock(&to_tcp_ctrl(ctrl)->teardown_lock);
 	nvme_tcp_stop_queue(ctrl, nvme_tcp_queue_id(req->queue));
-	if (!blk_mq_request_completed(rq)) {
+	if (blk_mq_request_started(rq) && !blk_mq_request_completed(rq)) {
 		nvme_req(rq)->status = NVME_SC_HOST_ABORTED_CMD;
 		blk_mq_complete_request(rq);
 	}
-	mutex_unlock(&to_tcp_ctrl(ctrl)->teardown_lock);
 }
 
 static enum blk_eh_timer_return
@@ -2455,7 +2448,6 @@ static struct nvme_ctrl *nvme_tcp_create_ctrl(struct device *dev,
 			nvme_tcp_reconnect_ctrl_work);
 	INIT_WORK(&ctrl->err_work, nvme_tcp_error_recovery_work);
 	INIT_WORK(&ctrl->ctrl.reset_work, nvme_reset_ctrl_work);
-	mutex_init(&ctrl->teardown_lock);
 
 	if (!(opts->mask & NVMF_OPT_TRSVCID)) {
 		opts->trsvcid =