nvme fixes for 5.10

- rdma error handling fixes (Chao Leng)
  - fc error handling and reconnect fixes (James Smart)
  - fix the qid displace when tracing ioctl command (Keith Busch)
  - don't use BLK_MQ_REQ_NOWAIT for passthru (Chaitanya Kulkarni)
  - fix MTDT for passthru (Logan Gunthorpe)
  - blacklist Write Same on more devices (Kai-Heng Feng)
  - fix an uninitialized work struct (zhenwei pi)
 -----BEGIN PGP SIGNATURE-----
 
 iQI/BAABCgApFiEEgdbnc3r/njty3Iq9D55TZVIEUYMFAl+StyILHGhjaEBsc3Qu
 ZGUACgkQD55TZVIEUYPkMg//RdykKGgAPLFiFDsWpZCQEUiiZjWKndtEpX4orT25
 o7QE6TunecQFNZPIobow164ZbeeaKP5syCnPvtjgYMVvmIwZMZcHhA+pAZQAu9/j
 tsZ9wquTa1QV4qJbDmY4ptQGIqXODf8rLZybqMhw/e0l+E0iI5ngkn8TCnNQTqP5
 S0YFzdoVDefubDx9uzuLIbCuzBW+zv8bpxRaYWxKDI8AmezDn6PV8ZeFZOgxsmkI
 0nBSHfwbSPvxSWHAw/pzGoQ0g5aOe7zxHtslz/g3gOYVto/6JsGyCVAXICym5ZbD
 59GQ5PPZggfw+GVtdoEA/A21hMusf0aGnk7lTpNnLJ1IXmVNUI61/1AaD4LBAXjM
 Q1VfgP4YrXl+XpYhbGcAYB0OTNsntM8O9xsX1ZQpW2VRnKgjPiohtKq2QAEdWCKH
 mZR9f+XIe/BRnRRl0A37IyUDeo4/EFiQ66wKW+GAL1R/x/rJUFdY7sM4u1j0XGgN
 WZMHdpLyoSlXYcV280I0w8f7wASwAZ78A87w5W7pcuBKH0aCnssreRQ2vOGPHb0x
 7cqNQ+lyrp/Rhb9NHj0jJWEEJ6tHa8xH4/Tz+hsktx0vkK454YmS3Rbrll3D9wtC
 7LZZkluy9r3pg24xJNUo6V0DyVCF1Nl063TZ5YR8/N4lXRlk/6BE4k1IsK/Jcl8g
 4T4=
 =gjYy
 -----END PGP SIGNATURE-----

Merge tag 'nvme-5.10-2020-10-23' of git://git.infradead.org/nvme into block-5.10

Pull NVMe fixes from Christoph:

"nvme fixes for 5.10

 - rdma error handling fixes (Chao Leng)
 - fc error handling and reconnect fixes (James Smart)
 - fix the qid displace when tracing ioctl command (Keith Busch)
 - don't use BLK_MQ_REQ_NOWAIT for passthru (Chaitanya Kulkarni)
 - fix MTDT for passthru (Logan Gunthorpe)
 - blacklist Write Same on more devices (Kai-Heng Feng)
 - fix an uninitialized work struct (zhenwei pi)"

* tag 'nvme-5.10-2020-10-23' of git://git.infradead.org/nvme:
  nvme-fc: shorten reconnect delay if possible for FC
  nvme-fc: wait for queues to freeze before calling update_hr_hw_queues
  nvme-fc: fix error loop in create_hw_io_queues
  nvme-fc: fix io timeout to abort I/O
  nvmet: don't use BLK_MQ_REQ_NOWAIT for passthru
  nvmet: cleanup nvmet_passthru_map_sg()
  nvmet: limit passthru MTDS by BIO_MAX_PAGES
  nvmet: fix uninitialized work for zero kato
  nvme-pci: disable Write Zeroes on Sandisk Skyhawk
  nvme: use queuedata for nvme_req_qid
  nvme-rdma: fix crash due to incorrect cqe
  nvme-rdma: fix crash when connect rejected

drivers/nvme/host/fc.c

@@ -26,6 +26,10 @@ enum nvme_fc_queue_flags {
 };
 
 #define NVME_FC_DEFAULT_DEV_LOSS_TMO	60	/* seconds */
+#define NVME_FC_DEFAULT_RECONNECT_TMO	2	/* delay between reconnects
+						 * when connected and a
+						 * connection failure.
+						 */
 
 struct nvme_fc_queue {
 	struct nvme_fc_ctrl	*ctrl;
@@ -1837,8 +1841,10 @@ __nvme_fc_abort_op(struct nvme_fc_ctrl *ctrl, struct nvme_fc_fcp_op *op)
 	opstate = atomic_xchg(&op->state, FCPOP_STATE_ABORTED);
 	if (opstate != FCPOP_STATE_ACTIVE)
 		atomic_set(&op->state, opstate);
-	else if (test_bit(FCCTRL_TERMIO, &ctrl->flags))
+	else if (test_bit(FCCTRL_TERMIO, &ctrl->flags)) {
+		op->flags |= FCOP_FLAGS_TERMIO;
 		ctrl->iocnt++;
+	}
 	spin_unlock_irqrestore(&ctrl->lock, flags);
 
 	if (opstate != FCPOP_STATE_ACTIVE)
@@ -1874,7 +1880,8 @@ __nvme_fc_fcpop_chk_teardowns(struct nvme_fc_ctrl *ctrl,
 
 	if (opstate == FCPOP_STATE_ABORTED) {
 		spin_lock_irqsave(&ctrl->lock, flags);
-		if (test_bit(FCCTRL_TERMIO, &ctrl->flags)) {
+		if (test_bit(FCCTRL_TERMIO, &ctrl->flags) &&
+		    op->flags & FCOP_FLAGS_TERMIO) {
 			if (!--ctrl->iocnt)
 				wake_up(&ctrl->ioabort_wait);
 		}
@@ -2314,7 +2321,7 @@ nvme_fc_create_hw_io_queues(struct nvme_fc_ctrl *ctrl, u16 qsize)
 	return 0;
 
 delete_queues:
-	for (; i >= 0; i--)
+	for (; i > 0; i--)
 		__nvme_fc_delete_hw_queue(ctrl, &ctrl->queues[i], i);
 	return ret;
 }
@@ -2433,7 +2440,7 @@ nvme_fc_error_recovery(struct nvme_fc_ctrl *ctrl, char *errmsg)
 		return;
 
 	dev_warn(ctrl->ctrl.device,
-		"NVME-FC{%d}: transport association error detected: %s\n",
+		"NVME-FC{%d}: transport association event: %s\n",
 		ctrl->cnum, errmsg);
 	dev_warn(ctrl->ctrl.device,
 		"NVME-FC{%d}: resetting controller\n", ctrl->cnum);
@@ -2446,15 +2453,20 @@ nvme_fc_timeout(struct request *rq, bool reserved)
 {
 	struct nvme_fc_fcp_op *op = blk_mq_rq_to_pdu(rq);
 	struct nvme_fc_ctrl *ctrl = op->ctrl;
+	struct nvme_fc_cmd_iu *cmdiu = &op->cmd_iu;
+	struct nvme_command *sqe = &cmdiu->sqe;
 
 	/*
-	 * we can't individually ABTS an io without affecting the queue,
+	 * Attempt to abort the offending command. Command completion
-	 * thus killing the queue, and thus the association.
+	 * will detect the aborted io and will fail the connection.
-	 * So resolve by performing a controller reset, which will stop
-	 * the host/io stack, terminate the association on the link,
-	 * and recreate an association on the link.
 	 */
-	nvme_fc_error_recovery(ctrl, "io timeout error");
+	dev_info(ctrl->ctrl.device,
+		"NVME-FC{%d.%d}: io timeout: opcode %d fctype %d w10/11: "
+		"x%08x/x%08x\n",
+		ctrl->cnum, op->queue->qnum, sqe->common.opcode,
+		sqe->connect.fctype, sqe->common.cdw10, sqe->common.cdw11);
+	if (__nvme_fc_abort_op(ctrl, op))
+		nvme_fc_error_recovery(ctrl, "io timeout abort failed");
 
 	/*
 	 * the io abort has been initiated. Have the reset timer
@@ -2726,6 +2738,7 @@ nvme_fc_complete_rq(struct request *rq)
 	struct nvme_fc_ctrl *ctrl = op->ctrl;
 
 	atomic_set(&op->state, FCPOP_STATE_IDLE);
+	op->flags &= ~FCOP_FLAGS_TERMIO;
 
 	nvme_fc_unmap_data(ctrl, rq, op);
 	nvme_complete_rq(rq);
@@ -2876,11 +2889,14 @@ nvme_fc_recreate_io_queues(struct nvme_fc_ctrl *ctrl)
 	if (ret)
 		goto out_delete_hw_queues;
 
-	if (prior_ioq_cnt != nr_io_queues)
+	if (prior_ioq_cnt != nr_io_queues) {
 		dev_info(ctrl->ctrl.device,
 			"reconnect: revising io queue count from %d to %d\n",
 			prior_ioq_cnt, nr_io_queues);
+		nvme_wait_freeze(&ctrl->ctrl);
 		blk_mq_update_nr_hw_queues(&ctrl->tag_set, nr_io_queues);
+		nvme_unfreeze(&ctrl->ctrl);
+	}
 
 	return 0;
 
@@ -3090,6 +3106,61 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 	return ret;
 }
 
+
+/*
+ * This routine runs through all outstanding commands on the association
+ * and aborts them.  This routine is typically be called by the
+ * delete_association routine. It is also called due to an error during
+ * reconnect. In that scenario, it is most likely a command that initializes
+ * the controller, including fabric Connect commands on io queues, that
+ * may have timed out or failed thus the io must be killed for the connect
+ * thread to see the error.
+ */
+static void
+__nvme_fc_abort_outstanding_ios(struct nvme_fc_ctrl *ctrl, bool start_queues)
+{
+	/*
+	 * If io queues are present, stop them and terminate all outstanding
+	 * ios on them. As FC allocates FC exchange for each io, the
+	 * transport must contact the LLDD to terminate the exchange,
+	 * thus releasing the FC exchange. We use blk_mq_tagset_busy_itr()
+	 * to tell us what io's are busy and invoke a transport routine
+	 * to kill them with the LLDD.  After terminating the exchange
+	 * the LLDD will call the transport's normal io done path, but it
+	 * will have an aborted status. The done path will return the
+	 * io requests back to the block layer as part of normal completions
+	 * (but with error status).
+	 */
+	if (ctrl->ctrl.queue_count > 1) {
+		nvme_stop_queues(&ctrl->ctrl);
+		blk_mq_tagset_busy_iter(&ctrl->tag_set,
+				nvme_fc_terminate_exchange, &ctrl->ctrl);
+		blk_mq_tagset_wait_completed_request(&ctrl->tag_set);
+		if (start_queues)
+			nvme_start_queues(&ctrl->ctrl);
+	}
+
+	/*
+	 * Other transports, which don't have link-level contexts bound
+	 * to sqe's, would try to gracefully shutdown the controller by
+	 * writing the registers for shutdown and polling (call
+	 * nvme_shutdown_ctrl()). Given a bunch of i/o was potentially
+	 * just aborted and we will wait on those contexts, and given
+	 * there was no indication of how live the controlelr is on the
+	 * link, don't send more io to create more contexts for the
+	 * shutdown. Let the controller fail via keepalive failure if
+	 * its still present.
+	 */
+
+	/*
+	 * clean up the admin queue. Same thing as above.
+	 */
+	blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
+	blk_mq_tagset_busy_iter(&ctrl->admin_tag_set,
+				nvme_fc_terminate_exchange, &ctrl->ctrl);
+	blk_mq_tagset_wait_completed_request(&ctrl->admin_tag_set);
+}
+
 /*
  * This routine stops operation of the controller on the host side.
  * On the host os stack side: Admin and IO queues are stopped,
@@ -3110,46 +3181,7 @@ nvme_fc_delete_association(struct nvme_fc_ctrl *ctrl)
 	ctrl->iocnt = 0;
 	spin_unlock_irqrestore(&ctrl->lock, flags);
 
-	/*
+	__nvme_fc_abort_outstanding_ios(ctrl, false);
-	 * If io queues are present, stop them and terminate all outstanding
-	 * ios on them. As FC allocates FC exchange for each io, the
-	 * transport must contact the LLDD to terminate the exchange,
-	 * thus releasing the FC exchange. We use blk_mq_tagset_busy_itr()
-	 * to tell us what io's are busy and invoke a transport routine
-	 * to kill them with the LLDD.  After terminating the exchange
-	 * the LLDD will call the transport's normal io done path, but it
-	 * will have an aborted status. The done path will return the
-	 * io requests back to the block layer as part of normal completions
-	 * (but with error status).
-	 */
-	if (ctrl->ctrl.queue_count > 1) {
-		nvme_stop_queues(&ctrl->ctrl);
-		blk_mq_tagset_busy_iter(&ctrl->tag_set,
-				nvme_fc_terminate_exchange, &ctrl->ctrl);
-		blk_mq_tagset_wait_completed_request(&ctrl->tag_set);
-	}
-
-	/*
-	 * Other transports, which don't have link-level contexts bound
-	 * to sqe's, would try to gracefully shutdown the controller by
-	 * writing the registers for shutdown and polling (call
-	 * nvme_shutdown_ctrl()). Given a bunch of i/o was potentially
-	 * just aborted and we will wait on those contexts, and given
-	 * there was no indication of how live the controlelr is on the
-	 * link, don't send more io to create more contexts for the
-	 * shutdown. Let the controller fail via keepalive failure if
-	 * its still present.
-	 */
-
-	/*
-	 * clean up the admin queue. Same thing as above.
-	 * use blk_mq_tagset_busy_itr() and the transport routine to
-	 * terminate the exchanges.
-	 */
-	blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
-	blk_mq_tagset_busy_iter(&ctrl->admin_tag_set,
-				nvme_fc_terminate_exchange, &ctrl->ctrl);
-	blk_mq_tagset_wait_completed_request(&ctrl->admin_tag_set);
 
 	/* kill the aens as they are a separate path */
 	nvme_fc_abort_aen_ops(ctrl);
@@ -3263,21 +3295,26 @@ static void
 __nvme_fc_terminate_io(struct nvme_fc_ctrl *ctrl)
 {
 	/*
-	 * if state is connecting - the error occurred as part of a
+	 * if state is CONNECTING - the error occurred as part of a
-	 * reconnect attempt. The create_association error paths will
+	 * reconnect attempt. Abort any ios on the association and
-	 * clean up any outstanding io.
+	 * let the create_association error paths resolve things.
-	 *
-	 * if it's a different state - ensure all pending io is
-	 * terminated. Given this can delay while waiting for the
-	 * aborted io to return, we recheck adapter state below
-	 * before changing state.
 	 */
-	if (ctrl->ctrl.state != NVME_CTRL_CONNECTING) {
+	if (ctrl->ctrl.state == NVME_CTRL_CONNECTING) {
+		__nvme_fc_abort_outstanding_ios(ctrl, true);
+		return;
+	}
+
+	/*
+	 * For any other state, kill the association. As this routine
+	 * is a common io abort routine for resetting and such, after
+	 * the association is terminated, ensure that the state is set
+	 * to CONNECTING.
+	 */
+
 	nvme_stop_keep_alive(&ctrl->ctrl);
 
 	/* will block will waiting for io to terminate */
 	nvme_fc_delete_association(ctrl);
-	}
 
 	if (ctrl->ctrl.state != NVME_CTRL_CONNECTING &&
 	    !nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_CONNECTING))
@@ -3403,7 +3440,7 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 {
 	struct nvme_fc_ctrl *ctrl;
 	unsigned long flags;
-	int ret, idx;
+	int ret, idx, ctrl_loss_tmo;
 
 	if (!(rport->remoteport.port_role &
 	    (FC_PORT_ROLE_NVME_DISCOVERY | FC_PORT_ROLE_NVME_TARGET))) {
@@ -3429,6 +3466,19 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 		goto out_free_ctrl;
 	}
 
+	/*
+	 * if ctrl_loss_tmo is being enforced and the default reconnect delay
+	 * is being used, change to a shorter reconnect delay for FC.
+	 */
+	if (opts->max_reconnects != -1 &&
+	    opts->reconnect_delay == NVMF_DEF_RECONNECT_DELAY &&
+	    opts->reconnect_delay > NVME_FC_DEFAULT_RECONNECT_TMO) {
+		ctrl_loss_tmo = opts->max_reconnects * opts->reconnect_delay;
+		opts->reconnect_delay = NVME_FC_DEFAULT_RECONNECT_TMO;
+		opts->max_reconnects = DIV_ROUND_UP(ctrl_loss_tmo,
+						opts->reconnect_delay);
+	}
+
 	ctrl->ctrl.opts = opts;
 	ctrl->ctrl.nr_reconnects = 0;
 	if (lport->dev)

drivers/nvme/host/nvme.h

@@ -176,7 +176,7 @@ static inline struct nvme_request *nvme_req(struct request *req)
 
 static inline u16 nvme_req_qid(struct request *req)
 {
-	if (!req->rq_disk)
+	if (!req->q->queuedata)
 		return 0;
 	return blk_mq_unique_tag_to_hwq(blk_mq_unique_tag(req)) + 1;
 }

drivers/nvme/host/pci.c

@@ -3185,6 +3185,8 @@ static const struct pci_device_id nvme_id_table[] = {
 				NVME_QUIRK_IGNORE_DEV_SUBNQN, },
 	{ PCI_DEVICE(0x1c5c, 0x1504),   /* SK Hynix PC400 */
 		.driver_data = NVME_QUIRK_DISABLE_WRITE_ZEROES, },
+	{ PCI_DEVICE(0x15b7, 0x2001),   /*  Sandisk Skyhawk */
+		.driver_data = NVME_QUIRK_DISABLE_WRITE_ZEROES, },
 	{ PCI_DEVICE(PCI_VENDOR_ID_APPLE, 0x2001),
 		.driver_data = NVME_QUIRK_SINGLE_VECTOR },
 	{ PCI_DEVICE(PCI_VENDOR_ID_APPLE, 0x2003) },

drivers/nvme/host/rdma.c

@@ -1730,10 +1730,11 @@ static void nvme_rdma_process_nvme_rsp(struct nvme_rdma_queue *queue,
 	req->result = cqe->result;
 
 	if (wc->wc_flags & IB_WC_WITH_INVALIDATE) {
-		if (unlikely(wc->ex.invalidate_rkey != req->mr->rkey)) {
+		if (unlikely(!req->mr ||
+			     wc->ex.invalidate_rkey != req->mr->rkey)) {
 			dev_err(queue->ctrl->ctrl.device,
 				"Bogus remote invalidation for rkey %#x\n",
-				req->mr->rkey);
+				req->mr ? req->mr->rkey : 0);
 			nvme_rdma_error_recovery(queue->ctrl);
 		}
 	} else if (req->mr) {
@@ -1926,7 +1927,6 @@ static int nvme_rdma_cm_handler(struct rdma_cm_id *cm_id,
 		complete(&queue->cm_done);
 		return 0;
 	case RDMA_CM_EVENT_REJECTED:
-		nvme_rdma_destroy_queue_ib(queue);
 		cm_error = nvme_rdma_conn_rejected(queue, ev);
 		break;
 	case RDMA_CM_EVENT_ROUTE_ERROR:

drivers/nvme/target/core.c

@@ -1126,6 +1126,7 @@ static void nvmet_start_ctrl(struct nvmet_ctrl *ctrl)
 	 * in case a host died before it enabled the controller.  Hence, simply
 	 * reset the keep alive timer when the controller is enabled.
 	 */
+	if (ctrl->kato)
 		mod_delayed_work(system_wq, &ctrl->ka_work, ctrl->kato * HZ);
 }
 

drivers/nvme/target/passthru.c

@@ -26,7 +26,7 @@ static u16 nvmet_passthru_override_id_ctrl(struct nvmet_req *req)
 	struct nvme_ctrl *pctrl = ctrl->subsys->passthru_ctrl;
 	u16 status = NVME_SC_SUCCESS;
 	struct nvme_id_ctrl *id;
-	u32 max_hw_sectors;
+	int max_hw_sectors;
 	int page_shift;
 
 	id = kzalloc(sizeof(*id), GFP_KERNEL);
@@ -48,6 +48,13 @@ static u16 nvmet_passthru_override_id_ctrl(struct nvmet_req *req)
 	max_hw_sectors = min_not_zero(pctrl->max_segments << (PAGE_SHIFT - 9),
 				      pctrl->max_hw_sectors);
 
+	/*
+	 * nvmet_passthru_map_sg is limitted to using a single bio so limit
+	 * the mdts based on BIO_MAX_PAGES as well
+	 */
+	max_hw_sectors = min_not_zero(BIO_MAX_PAGES << (PAGE_SHIFT - 9),
+				      max_hw_sectors);
+
 	page_shift = NVME_CAP_MPSMIN(ctrl->cap) + 12;
 
 	id->mdts = ilog2(max_hw_sectors) + 9 - page_shift;
@@ -180,18 +187,20 @@ static void nvmet_passthru_req_done(struct request *rq,
 
 static int nvmet_passthru_map_sg(struct nvmet_req *req, struct request *rq)
 {
-	int sg_cnt = req->sg_cnt;
 	struct scatterlist *sg;
 	int op_flags = 0;
 	struct bio *bio;
 	int i, ret;
 
+	if (req->sg_cnt > BIO_MAX_PAGES)
+		return -EINVAL;
+
 	if (req->cmd->common.opcode == nvme_cmd_flush)
 		op_flags = REQ_FUA;
 	else if (nvme_is_write(req->cmd))
 		op_flags = REQ_SYNC | REQ_IDLE;
 
-	bio = bio_alloc(GFP_KERNEL, min(sg_cnt, BIO_MAX_PAGES));
+	bio = bio_alloc(GFP_KERNEL, req->sg_cnt);
 	bio->bi_end_io = bio_put;
 	bio->bi_opf = req_op(rq) | op_flags;
 
@@ -201,7 +210,6 @@ static int nvmet_passthru_map_sg(struct nvmet_req *req, struct request *rq)
 			bio_put(bio);
 			return -EINVAL;
 		}
-		sg_cnt--;
 	}
 
 	ret = blk_rq_append_bio(rq, &bio);
@@ -236,7 +244,7 @@ static void nvmet_passthru_execute_cmd(struct nvmet_req *req)
 		q = ns->queue;
 	}
 
-	rq = nvme_alloc_request(q, req->cmd, BLK_MQ_REQ_NOWAIT, NVME_QID_ANY);
+	rq = nvme_alloc_request(q, req->cmd, 0, NVME_QID_ANY);
 	if (IS_ERR(rq)) {
 		status = NVME_SC_INTERNAL;
 		goto out_put_ns;