#define CQ_SIZE(depth) (depth * sizeof(struct nvme_completion))
#define ADMIN_TIMEOUT (admin_timeout * HZ)
#define SHUTDOWN_TIMEOUT (shutdown_timeout * HZ)
-#define IOD_TIMEOUT (retry_time * HZ)
static unsigned char admin_timeout = 60;
module_param(admin_timeout, byte, 0644);
module_param_named(io_timeout, nvme_io_timeout, byte, 0644);
MODULE_PARM_DESC(io_timeout, "timeout in seconds for I/O");
-static unsigned char retry_time = 30;
-module_param(retry_time, byte, 0644);
-MODULE_PARM_DESC(retry_time, "time in seconds to retry failed I/O");
-
static unsigned char shutdown_timeout = 5;
module_param(shutdown_timeout, byte, 0644);
MODULE_PARM_DESC(shutdown_timeout, "timeout in seconds for controller shutdown");
static struct task_struct *nvme_thread;
static struct workqueue_struct *nvme_workq;
static wait_queue_head_t nvme_kthread_wait;
-static struct notifier_block nvme_nb;
static struct class *nvme_class;
* commands and one for I/O commands).
*/
struct nvme_queue {
- struct llist_node node;
struct device *q_dmadev;
struct nvme_dev *dev;
char irqname[24]; /* nvme4294967295-65535\0 */
struct nvme_cmd_info *cmd = blk_mq_rq_to_pdu(req);
struct nvme_queue *nvmeq = cmd->nvmeq;
- /*
- * The aborted req will be completed on receiving the abort req.
- * We enable the timer again. If hit twice, it'll cause a device reset,
- * as the device then is in a faulty state.
- */
- int ret = BLK_EH_RESET_TIMER;
-
dev_warn(nvmeq->q_dmadev, "Timeout I/O %d QID %d\n", req->tag,
nvmeq->qid);
-
spin_lock_irq(&nvmeq->q_lock);
- if (!nvmeq->dev->initialized) {
- /*
- * Force cancelled command frees the request, which requires we
- * return BLK_EH_NOT_HANDLED.
- */
- nvme_cancel_queue_ios(nvmeq->hctx, req, nvmeq, reserved);
- ret = BLK_EH_NOT_HANDLED;
- } else
- nvme_abort_req(req);
+ nvme_abort_req(req);
spin_unlock_irq(&nvmeq->q_lock);
- return ret;
+ /*
+ * The aborted req will be completed on receiving the abort req.
+ * We enable the timer again. If hit twice, it'll cause a device reset,
+ * as the device then is in a faulty state.
+ */
+ return BLK_EH_RESET_TIMER;
}
static void nvme_free_queue(struct nvme_queue *nvmeq)
struct blk_mq_hw_ctx *hctx = nvmeq->hctx;
spin_lock_irq(&nvmeq->q_lock);
- nvme_process_cq(nvmeq);
if (hctx && hctx->tags)
blk_mq_tag_busy_iter(hctx, nvme_cancel_queue_ios, nvmeq);
spin_unlock_irq(&nvmeq->q_lock);
}
if (!qid && dev->admin_q)
blk_mq_freeze_queue_start(dev->admin_q);
- nvme_clear_queue(nvmeq);
+
+ spin_lock_irq(&nvmeq->q_lock);
+ nvme_process_cq(nvmeq);
+ spin_unlock_irq(&nvmeq->q_lock);
}
static struct nvme_queue *nvme_alloc_queue(struct nvme_dev *dev, int qid,
spin_lock(&dev_list_lock);
list_for_each_entry_safe(dev, next, &dev_list, node) {
int i;
- if (readl(&dev->bar->csts) & NVME_CSTS_CFS &&
- dev->initialized) {
+ if (readl(&dev->bar->csts) & NVME_CSTS_CFS) {
if (work_busy(&dev->reset_work))
continue;
list_del_init(&dev->node);
static void nvme_del_queue_end(struct nvme_queue *nvmeq)
{
struct nvme_delq_ctx *dq = nvmeq->cmdinfo.ctx;
-
- nvme_clear_queue(nvmeq);
nvme_put_dq(dq);
}
int i;
u32 csts = -1;
- dev->initialized = 0;
nvme_dev_list_remove(dev);
if (dev->bar) {
for (i = dev->queue_count - 1; i >= 0; i--) {
struct nvme_queue *nvmeq = dev->queues[i];
nvme_suspend_queue(nvmeq);
- nvme_clear_queue(nvmeq);
}
} else {
nvme_disable_io_queues(dev);
nvme_disable_queue(dev, 0);
}
nvme_dev_unmap(dev);
+
+ for (i = dev->queue_count - 1; i >= 0; i--)
+ nvme_clear_queue(dev->queues[i]);
}
static void nvme_dev_remove(struct nvme_dev *dev)
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;
nvme_unfreeze_queues(dev);
nvme_set_irq_hints(dev);
}
- dev->initialized = 1;
return 0;
}
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;
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);
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);
+ return;
+ reset:
+ if (!work_busy(&dev->reset_work)) {
+ 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);
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);
static void __exit nvme_exit(void)
{
pci_unregister_driver(&nvme_driver);
- unregister_hotcpu_notifier(&nvme_nb);
unregister_blkdev(nvme_major, "nvme");
destroy_workqueue(nvme_workq);
class_destroy(nvme_class);