[PATCHv2] nvme: ensure disabling pairs with unquiesce

[Keith Busch]

From: Keith Busch <kbusch at kernel.org>

If any error handling that disables the controller fails to queue the
reset work, like if the state changed to disconnected inbetween, then
the failed teardown needs to unquiesce the queues since it's no longer
paired with reset_work. Just make sure that the controller can be put
into a resetting state prior to starting the disable so that no other
handling can change with the queue states while recovery is happening.

Reported-by: Ming Lei <ming.lei at redhat.com>
Signed-off-by: Keith Busch <kbusch at kernel.org>
---
v1->v2:

  Don't wait for a resetting state as that could deadlock. Just abort
  the disabling if the error handling can't set the state to resetting.

 drivers/nvme/host/pci.c | 25 +++++++++++++++++--------
 1 file changed, 17 insertions(+), 8 deletions(-)

diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index 8754b4a5c6844..8e7dbe0ab8904 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -1298,9 +1298,7 @@ static enum blk_eh_timer_return nvme_timeout(struct request *req)
 	 */
 	if (nvme_should_reset(dev, csts)) {
 		nvme_warn_reset(dev, csts);
-		nvme_dev_disable(dev, false);
-		nvme_reset_ctrl(&dev->ctrl);
-		return BLK_EH_DONE;
+		goto disable;
 	}
 
 	/*
@@ -1351,10 +1349,7 @@ static enum blk_eh_timer_return nvme_timeout(struct request *req)
 			 "I/O %d QID %d timeout, reset controller\n",
 			 req->tag, nvmeq->qid);
 		nvme_req(req)->flags |= NVME_REQ_CANCELLED;
-		nvme_dev_disable(dev, false);
-		nvme_reset_ctrl(&dev->ctrl);
-
-		return BLK_EH_DONE;
+		goto disable;
 	}
 
 	if (atomic_dec_return(&dev->ctrl.abort_limit) < 0) {
@@ -1391,6 +1386,15 @@ static enum blk_eh_timer_return nvme_timeout(struct request *req)
 	 * as the device then is in a faulty state.
 	 */
 	return BLK_EH_RESET_TIMER;
+
+disable:
+	if (!nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_RESETTING))
+		return BLK_EH_DONE;
+
+	nvme_dev_disable(dev, false);
+	if (nvme_try_sched_reset(&dev->ctrl))
+		nvme_unquiesce_io_queues(&dev->ctrl);
+	return BLK_EH_DONE;
 }
 
 static void nvme_free_queue(struct nvme_queue *nvmeq)
@@ -3278,6 +3282,10 @@ static pci_ers_result_t nvme_error_detected(struct pci_dev *pdev,
 	case pci_channel_io_frozen:
 		dev_warn(dev->ctrl.device,
 			"frozen state error detected, reset controller\n");
+		if (!nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_RESETTING)) {
+			nvme_dev_disable(dev, true);
+			return PCI_ERS_RESULT_DISCONNECT;
+		}
 		nvme_dev_disable(dev, false);
 		return PCI_ERS_RESULT_NEED_RESET;
 	case pci_channel_io_perm_failure:
@@ -3294,7 +3302,8 @@ static pci_ers_result_t nvme_slot_reset(struct pci_dev *pdev)
 
 	dev_info(dev->ctrl.device, "restart after slot reset\n");
 	pci_restore_state(pdev);
-	nvme_reset_ctrl(&dev->ctrl);
+	if (!nvme_try_sched_reset(&dev->ctrl))
+		nvme_unquiesce_io_queues(&dev->ctrl);
 	return PCI_ERS_RESULT_RECOVERED;
 }
 
-- 
2.34.1