Merge tag 'nvme-5.12-20210319' of git://git.infradead.org/nvme into block-5.12

Pull NVMe updates from Christoph:

"nvme fixes for 5.12

 - fix tag allocation for keep alive
 - fix a unit mismatch for the Write Zeroes limits
 - various TCP transport fixes (Sagi Grimberg, Elad Grupi)
 - fix iosqes and iocqes validation for discovery controllers (Sagi Grimberg)"

* tag 'nvme-5.12-20210319' of git://git.infradead.org/nvme:
  nvmet-tcp: fix kmap leak when data digest in use
  nvmet: don't check iosqes,iocqes for discovery controllers
  nvme-rdma: fix possible hang when failing to set io queues
  nvme-tcp: fix possible hang when failing to set io queues
  nvme-tcp: fix misuse of __smp_processor_id with preemption enabled
  nvme-tcp: fix a NULL deref when receiving a 0-length r2t PDU
  nvme: fix Write Zeroes limitations
  nvme: allocate the keep alive request using BLK_MQ_REQ_NOWAIT
  nvme: merge nvme_keep_alive into nvme_keep_alive_work
  nvme-fabrics: only reserve a single tag

drivers/nvme/host/core.c

@@ -1226,28 +1226,12 @@ static void nvme_keep_alive_end_io(struct request *rq, blk_status_t status)
 		queue_delayed_work(nvme_wq, &ctrl->ka_work, ctrl->kato * HZ);
 }
 
-static int nvme_keep_alive(struct nvme_ctrl *ctrl)
-{
-	struct request *rq;
-
-	rq = nvme_alloc_request(ctrl->admin_q, &ctrl->ka_cmd,
-			BLK_MQ_REQ_RESERVED);
-	if (IS_ERR(rq))
-		return PTR_ERR(rq);
-
-	rq->timeout = ctrl->kato * HZ;
-	rq->end_io_data = ctrl;
-
-	blk_execute_rq_nowait(NULL, rq, 0, nvme_keep_alive_end_io);
-
-	return 0;
-}
-
 static void nvme_keep_alive_work(struct work_struct *work)
 {
 	struct nvme_ctrl *ctrl = container_of(to_delayed_work(work),
 			struct nvme_ctrl, ka_work);
 	bool comp_seen = ctrl->comp_seen;
+	struct request *rq;
 
 	if ((ctrl->ctratt & NVME_CTRL_ATTR_TBKAS) && comp_seen) {
 		dev_dbg(ctrl->device,
@@ -1257,12 +1241,18 @@ static void nvme_keep_alive_work(struct work_struct *work)
 		return;
 	}
 
-	if (nvme_keep_alive(ctrl)) {
+	rq = nvme_alloc_request(ctrl->admin_q, &ctrl->ka_cmd,
+				BLK_MQ_REQ_RESERVED | BLK_MQ_REQ_NOWAIT);
+	if (IS_ERR(rq)) {
 		/* allocation failure, reset the controller */
-		dev_err(ctrl->device, "keep-alive failed\n");
+		dev_err(ctrl->device, "keep-alive failed: %ld\n", PTR_ERR(rq));
 		nvme_reset_ctrl(ctrl);
 		return;
 	}
+
+	rq->timeout = ctrl->kato * HZ;
+	rq->end_io_data = ctrl;
+	blk_execute_rq_nowait(NULL, rq, 0, nvme_keep_alive_end_io);
 }
 
 static void nvme_start_keep_alive(struct nvme_ctrl *ctrl)
@@ -1964,30 +1954,18 @@ static void nvme_config_discard(struct gendisk *disk, struct nvme_ns *ns)
 		blk_queue_max_write_zeroes_sectors(queue, UINT_MAX);
 }
 
-static void nvme_config_write_zeroes(struct gendisk *disk, struct nvme_ns *ns)
+/*
+ * Even though NVMe spec explicitly states that MDTS is not applicable to the
+ * write-zeroes, we are cautious and limit the size to the controllers
+ * max_hw_sectors value, which is based on the MDTS field and possibly other
+ * limiting factors.
+ */
+static void nvme_config_write_zeroes(struct request_queue *q,
+		struct nvme_ctrl *ctrl)
 {
-	u64 max_blocks;
-
-	if (!(ns->ctrl->oncs & NVME_CTRL_ONCS_WRITE_ZEROES) ||
-	    (ns->ctrl->quirks & NVME_QUIRK_DISABLE_WRITE_ZEROES))
-		return;
-	/*
-	 * Even though NVMe spec explicitly states that MDTS is not
-	 * applicable to the write-zeroes:- "The restriction does not apply to
-	 * commands that do not transfer data between the host and the
-	 * controller (e.g., Write Uncorrectable ro Write Zeroes command).".
-	 * In order to be more cautious use controller's max_hw_sectors value
-	 * to configure the maximum sectors for the write-zeroes which is
-	 * configured based on the controller's MDTS field in the
-	 * nvme_init_identify() if available.
-	 */
-	if (ns->ctrl->max_hw_sectors == UINT_MAX)
-		max_blocks = (u64)USHRT_MAX + 1;
-	else
-		max_blocks = ns->ctrl->max_hw_sectors + 1;
-
-	blk_queue_max_write_zeroes_sectors(disk->queue,
-					   nvme_lba_to_sect(ns, max_blocks));
+	if ((ctrl->oncs & NVME_CTRL_ONCS_WRITE_ZEROES) &&
+	    !(ctrl->quirks & NVME_QUIRK_DISABLE_WRITE_ZEROES))
+		blk_queue_max_write_zeroes_sectors(q, ctrl->max_hw_sectors);
 }
 
 static bool nvme_ns_ids_valid(struct nvme_ns_ids *ids)
@@ -2159,7 +2137,7 @@ static void nvme_update_disk_info(struct gendisk *disk,
 	set_capacity_and_notify(disk, capacity);
 
 	nvme_config_discard(disk, ns);
-	nvme_config_write_zeroes(disk, ns);
+	nvme_config_write_zeroes(disk->queue, ns->ctrl);
 
 	set_disk_ro(disk, (id->nsattr & NVME_NS_ATTR_RO) ||
 		test_bit(NVME_NS_FORCE_RO, &ns->flags));

drivers/nvme/host/fabrics.h

@@ -18,6 +18,13 @@
 /* default is -1: the fail fast mechanism is disabled  */
 #define NVMF_DEF_FAIL_FAST_TMO		-1
 
+/*
+ * Reserved one command for internal usage.  This command is used for sending
+ * the connect command, as well as for the keep alive command on the admin
+ * queue once live.
+ */
+#define NVMF_RESERVED_TAGS	1
+
 /*
  * Define a host as seen by the target.  We allocate one at boot, but also
  * allow the override it when creating controllers.  This is both to provide

drivers/nvme/host/fc.c

@@ -2863,7 +2863,7 @@ nvme_fc_create_io_queues(struct nvme_fc_ctrl *ctrl)
 	memset(&ctrl->tag_set, 0, sizeof(ctrl->tag_set));
 	ctrl->tag_set.ops = &nvme_fc_mq_ops;
 	ctrl->tag_set.queue_depth = ctrl->ctrl.opts->queue_size;
-	ctrl->tag_set.reserved_tags = 1; /* fabric connect */
+	ctrl->tag_set.reserved_tags = NVMF_RESERVED_TAGS;
 	ctrl->tag_set.numa_node = ctrl->ctrl.numa_node;
 	ctrl->tag_set.flags = BLK_MQ_F_SHOULD_MERGE;
 	ctrl->tag_set.cmd_size =
@@ -3485,7 +3485,7 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 	memset(&ctrl->admin_tag_set, 0, sizeof(ctrl->admin_tag_set));
 	ctrl->admin_tag_set.ops = &nvme_fc_admin_mq_ops;
 	ctrl->admin_tag_set.queue_depth = NVME_AQ_MQ_TAG_DEPTH;
-	ctrl->admin_tag_set.reserved_tags = 2; /* fabric connect + Keep-Alive */
+	ctrl->admin_tag_set.reserved_tags = NVMF_RESERVED_TAGS;
 	ctrl->admin_tag_set.numa_node = ctrl->ctrl.numa_node;
 	ctrl->admin_tag_set.cmd_size =
 		struct_size((struct nvme_fcp_op_w_sgl *)NULL, priv,

drivers/nvme/host/rdma.c

@@ -736,8 +736,11 @@ static int nvme_rdma_alloc_io_queues(struct nvme_rdma_ctrl *ctrl)
 		return ret;
 
 	ctrl->ctrl.queue_count = nr_io_queues + 1;
-	if (ctrl->ctrl.queue_count < 2)
-		return 0;
+	if (ctrl->ctrl.queue_count < 2) {
+		dev_err(ctrl->ctrl.device,
+			"unable to set any I/O queues\n");
+		return -ENOMEM;
+	}
 
 	dev_info(ctrl->ctrl.device,
 		"creating %d I/O queues.\n", nr_io_queues);
@@ -798,7 +801,7 @@ static struct blk_mq_tag_set *nvme_rdma_alloc_tagset(struct nvme_ctrl *nctrl,
 		memset(set, 0, sizeof(*set));
 		set->ops = &nvme_rdma_admin_mq_ops;
 		set->queue_depth = NVME_AQ_MQ_TAG_DEPTH;
-		set->reserved_tags = 2; /* connect + keep-alive */
+		set->reserved_tags = NVMF_RESERVED_TAGS;
 		set->numa_node = nctrl->numa_node;
 		set->cmd_size = sizeof(struct nvme_rdma_request) +
 				NVME_RDMA_DATA_SGL_SIZE;
@@ -811,7 +814,7 @@ static struct blk_mq_tag_set *nvme_rdma_alloc_tagset(struct nvme_ctrl *nctrl,
 		memset(set, 0, sizeof(*set));
 		set->ops = &nvme_rdma_mq_ops;
 		set->queue_depth = nctrl->sqsize + 1;
-		set->reserved_tags = 1; /* fabric connect */
+		set->reserved_tags = NVMF_RESERVED_TAGS;
 		set->numa_node = nctrl->numa_node;
 		set->flags = BLK_MQ_F_SHOULD_MERGE;
 		set->cmd_size = sizeof(struct nvme_rdma_request) +

drivers/nvme/host/tcp.c

@@ -287,7 +287,7 @@ static inline void nvme_tcp_queue_request(struct nvme_tcp_request *req,
 	 * directly, otherwise queue io_work. Also, only do that if we
 	 * are on the same cpu, so we don't introduce contention.
 	 */
-	if (queue->io_cpu == __smp_processor_id() &&
+	if (queue->io_cpu == raw_smp_processor_id() &&
 	    sync && empty && mutex_trylock(&queue->send_mutex)) {
 		queue->more_requests = !last;
 		nvme_tcp_send_all(queue);
@@ -568,6 +568,13 @@ static int nvme_tcp_setup_h2c_data_pdu(struct nvme_tcp_request *req,
 	req->pdu_len = le32_to_cpu(pdu->r2t_length);
 	req->pdu_sent = 0;
 
+	if (unlikely(!req->pdu_len)) {
+		dev_err(queue->ctrl->ctrl.device,
+			"req %d r2t len is %u, probably a bug...\n",
+			rq->tag, req->pdu_len);
+		return -EPROTO;
+	}
+
 	if (unlikely(req->data_sent + req->pdu_len > req->data_len)) {
 		dev_err(queue->ctrl->ctrl.device,
 			"req %d r2t len %u exceeded data len %u (%zu sent)\n",
@@ -1575,7 +1582,7 @@ static struct blk_mq_tag_set *nvme_tcp_alloc_tagset(struct nvme_ctrl *nctrl,
 		memset(set, 0, sizeof(*set));
 		set->ops = &nvme_tcp_admin_mq_ops;
 		set->queue_depth = NVME_AQ_MQ_TAG_DEPTH;
-		set->reserved_tags = 2; /* connect + keep-alive */
+		set->reserved_tags = NVMF_RESERVED_TAGS;
 		set->numa_node = nctrl->numa_node;
 		set->flags = BLK_MQ_F_BLOCKING;
 		set->cmd_size = sizeof(struct nvme_tcp_request);
@@ -1587,7 +1594,7 @@ static struct blk_mq_tag_set *nvme_tcp_alloc_tagset(struct nvme_ctrl *nctrl,
 		memset(set, 0, sizeof(*set));
 		set->ops = &nvme_tcp_mq_ops;
 		set->queue_depth = nctrl->sqsize + 1;
-		set->reserved_tags = 1; /* fabric connect */
+		set->reserved_tags = NVMF_RESERVED_TAGS;
 		set->numa_node = nctrl->numa_node;
 		set->flags = BLK_MQ_F_SHOULD_MERGE | BLK_MQ_F_BLOCKING;
 		set->cmd_size = sizeof(struct nvme_tcp_request);
@@ -1745,8 +1752,11 @@ static int nvme_tcp_alloc_io_queues(struct nvme_ctrl *ctrl)
 		return ret;
 
 	ctrl->queue_count = nr_io_queues + 1;
-	if (ctrl->queue_count < 2)
-		return 0;
+	if (ctrl->queue_count < 2) {
+		dev_err(ctrl->device,
+			"unable to set any I/O queues\n");
+		return -ENOMEM;
+	}
 
 	dev_info(ctrl->device,
 		"creating %d I/O queues.\n", nr_io_queues);

drivers/nvme/target/core.c

@@ -1118,9 +1118,20 @@ static void nvmet_start_ctrl(struct nvmet_ctrl *ctrl)
 {
 	lockdep_assert_held(&ctrl->lock);
 
-	if (nvmet_cc_iosqes(ctrl->cc) != NVME_NVM_IOSQES ||
-	    nvmet_cc_iocqes(ctrl->cc) != NVME_NVM_IOCQES ||
-	    nvmet_cc_mps(ctrl->cc) != 0 ||
+	/*
+	 * Only I/O controllers should verify iosqes,iocqes.
+	 * Strictly speaking, the spec says a discovery controller
+	 * should verify iosqes,iocqes are zeroed, however that
+	 * would break backwards compatibility, so don't enforce it.
+	 */
+	if (ctrl->subsys->type != NVME_NQN_DISC &&
+	    (nvmet_cc_iosqes(ctrl->cc) != NVME_NVM_IOSQES ||
+	     nvmet_cc_iocqes(ctrl->cc) != NVME_NVM_IOCQES)) {
+		ctrl->csts = NVME_CSTS_CFS;
+		return;
+	}
+
+	if (nvmet_cc_mps(ctrl->cc) != 0 ||
 	    nvmet_cc_ams(ctrl->cc) != 0 ||
 	    nvmet_cc_css(ctrl->cc) != 0) {
 		ctrl->csts = NVME_CSTS_CFS;

drivers/nvme/target/loop.c

@@ -349,7 +349,7 @@ static int nvme_loop_configure_admin_queue(struct nvme_loop_ctrl *ctrl)
 	memset(&ctrl->admin_tag_set, 0, sizeof(ctrl->admin_tag_set));
 	ctrl->admin_tag_set.ops = &nvme_loop_admin_mq_ops;
 	ctrl->admin_tag_set.queue_depth = NVME_AQ_MQ_TAG_DEPTH;
-	ctrl->admin_tag_set.reserved_tags = 2; /* connect + keep-alive */
+	ctrl->admin_tag_set.reserved_tags = NVMF_RESERVED_TAGS;
 	ctrl->admin_tag_set.numa_node = ctrl->ctrl.numa_node;
 	ctrl->admin_tag_set.cmd_size = sizeof(struct nvme_loop_iod) +
 		NVME_INLINE_SG_CNT * sizeof(struct scatterlist);
@@ -520,7 +520,7 @@ static int nvme_loop_create_io_queues(struct nvme_loop_ctrl *ctrl)
 	memset(&ctrl->tag_set, 0, sizeof(ctrl->tag_set));
 	ctrl->tag_set.ops = &nvme_loop_mq_ops;
 	ctrl->tag_set.queue_depth = ctrl->ctrl.opts->queue_size;
-	ctrl->tag_set.reserved_tags = 1; /* fabric connect */
+	ctrl->tag_set.reserved_tags = NVMF_RESERVED_TAGS;
 	ctrl->tag_set.numa_node = ctrl->ctrl.numa_node;
 	ctrl->tag_set.flags = BLK_MQ_F_SHOULD_MERGE;
 	ctrl->tag_set.cmd_size = sizeof(struct nvme_loop_iod) +

drivers/nvme/target/tcp.c

@@ -1098,11 +1098,11 @@ static int nvmet_tcp_try_recv_data(struct nvmet_tcp_queue *queue)
 		cmd->rbytes_done += ret;
 	}
 
+	nvmet_tcp_unmap_pdu_iovec(cmd);
 	if (queue->data_digest) {
 		nvmet_tcp_prep_recv_ddgst(cmd);
 		return 0;
 	}
-	nvmet_tcp_unmap_pdu_iovec(cmd);
 
 	if (!(cmd->flags & NVMET_TCP_F_INIT_FAILED) &&
 	    cmd->rbytes_done == cmd->req.transfer_len) {