NVMe: Asynchronous controller probe

This performs the longest parts of nvme device probe in scheduled work.
This speeds up probe significantly when multiple devices are in use.
Signed-off-by: Keith Busch <keith.busch@intel.com>

drivers/block/nvme-core.c

@@ -2800,6 +2800,10 @@ static int nvme_dev_open(struct inode *inode, struct file *f)
 	spin_lock(&dev_list_lock);
 	list_for_each_entry(dev, &dev_list, node) {
 		if (dev->instance == instance) {
+			if (!dev->admin_q) {
+				ret = -EWOULDBLOCK;
+				break;
+			}
 			if (!kref_get_unless_zero(&dev->kref))
 				break;
 			f->private_data = dev;
@@ -2982,6 +2986,7 @@ static void nvme_reset_workfn(struct work_struct *work)
 	dev->reset_workfn(work);
 }
 
+static void nvme_async_probe(struct work_struct *work);
 static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 {
 	int node, result = -ENOMEM;
@@ -3017,34 +3022,20 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 		goto release;
 
 	kref_init(&dev->kref);
-	result = nvme_dev_start(dev);
-	if (result)
-		goto release_pools;
-
 	dev->device = device_create(nvme_class, &pdev->dev,
 				MKDEV(nvme_char_major, dev->instance),
 				dev, "nvme%d", dev->instance);
 	if (IS_ERR(dev->device)) {
 		result = PTR_ERR(dev->device);
-		goto shutdown;
+		goto release_pools;
 	}
 	get_device(dev->device);
 
-	if (dev->online_queues > 1)
-		result = nvme_dev_add(dev);
-	if (result)
-		goto device_del;
-
-	nvme_set_irq_hints(dev);
-	dev->initialized = 1;
+	INIT_WORK(&dev->probe_work, nvme_async_probe);
+	schedule_work(&dev->probe_work);
 	return 0;
 
- device_del:
-	device_destroy(nvme_class, MKDEV(nvme_char_major, dev->instance));
- shutdown:
-	nvme_dev_shutdown(dev);
  release_pools:
-	nvme_free_queues(dev, 0);
 	nvme_release_prp_pools(dev);
  release:
 	nvme_release_instance(dev);
@@ -3057,6 +3048,28 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	return result;
 }
 
+static void nvme_async_probe(struct work_struct *work)
+{
+	struct nvme_dev *dev = container_of(work, struct nvme_dev, probe_work);
+	int result;
+
+	result = nvme_dev_start(dev);
+	if (result)
+		goto reset;
+
+	if (dev->online_queues > 1)
+		result = nvme_dev_add(dev);
+	if (result)
+		goto reset;
+
+	nvme_set_irq_hints(dev);
+	dev->initialized = 1;
+	return;
+ reset:
+	dev->reset_workfn = nvme_reset_failed_dev;
+	queue_work(nvme_workq, &dev->reset_work);
+}
+
 static void nvme_reset_notify(struct pci_dev *pdev, bool prepare)
 {
 	struct nvme_dev *dev = pci_get_drvdata(pdev);
@@ -3082,6 +3095,7 @@ static void nvme_remove(struct pci_dev *pdev)
 	spin_unlock(&dev_list_lock);
 
 	pci_set_drvdata(pdev, NULL);
+	flush_work(&dev->probe_work);
 	flush_work(&dev->reset_work);
 	nvme_dev_shutdown(dev);
 	nvme_dev_remove(dev);

include/linux/nvme.h

@@ -91,6 +91,7 @@ struct nvme_dev {
 	struct device *device;
 	work_func_t reset_workfn;
 	struct work_struct reset_work;
+	struct work_struct probe_work;
 	char name[12];
 	char serial[20];
 	char model[40];