virtio: PCI ERS permanent failure teardown for virtio-blk

drivers/block/virtio_blk.c

@@ -435,6 +435,12 @@ static blk_status_t virtio_queue_rq(struct blk_mq_hw_ctx *hctx,
 	blk_status_t status;
 	int err;
 
+	/* Fail fast if ERS frozen tore down VQs or the disk was marked dead. */
+	if (unlikely(!disk_live(vblk->disk) || !vblk->vqs || !vblk->vdev)) {
+		blk_mq_start_request(req);
+		return BLK_STS_IOERR;
+	}
+
 	status = virtblk_prep_rq(hctx, vblk, req, vbr);
 	if (unlikely(status))
 		return status;
@@ -1559,6 +1565,29 @@ static int virtblk_probe(struct virtio_device *vdev)
 	return err;
 }
 
+/* Stop I/O and mark the gendisk dead (ERS perm_failure or system shutdown). */
+static void virtblk_shutdown(struct virtio_device *vdev)
+{
+	struct virtio_blk *vblk = vdev->priv;
+	struct request_queue *q;
+	unsigned int memflags;
+
+	if (!vblk || !vblk->disk)
+		return;
+
+	flush_work(&vblk->config_work);
+	virtio_break_device(vdev);
+
+	q = vblk->disk->queue;
+	memflags = blk_mq_freeze_queue(q);
+	blk_mq_quiesce_queue_nowait(q);
+
+	blk_mark_disk_dead(vblk->disk);
+
+	blk_mq_unquiesce_queue(q);
+	blk_mq_unfreeze_queue(q, memflags);
+}
+
 static void virtblk_remove(struct virtio_device *vdev)
 {
 	struct virtio_blk *vblk = vdev->priv;
@@ -1682,6 +1711,7 @@ static struct virtio_driver virtio_blk = {
 	.probe				= virtblk_probe,
 	.remove				= virtblk_remove,
 	.config_changed			= virtblk_config_changed,
+	.shutdown			= virtblk_shutdown,
 #ifdef CONFIG_PM_SLEEP
 	.freeze				= virtblk_freeze,
 	.restore			= virtblk_restore,

drivers/virtio/virtio_pci_common.c

@@ -826,7 +826,45 @@ static void virtio_pci_reset_done(struct pci_dev *pci_dev)
 		dev_warn(&pci_dev->dev, "Reset done failure: %d", ret);
 }
 
+static pci_ers_result_t virtio_pci_error_detected(struct pci_dev *pci_dev,
+						  pci_channel_state_t state)
+{
+	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
+
+	/*
+	 * PCI ERS error_detected: quiesce on frozen before bus reset; on
+	 * permanent failure ask the virtio driver to shut down (virtio-blk
+	 * marks the disk dead in its .shutdown handler).
+	 */
+	switch (state) {
+	case pci_channel_io_normal:
+		return PCI_ERS_RESULT_CAN_RECOVER;
+	case pci_channel_io_frozen:
+		pci_info(pci_dev, "frozen error detected, quiesce device\n");
+		if (virtio_device_reset_prepare(&vp_dev->vdev))
+			dev_warn(&pci_dev->dev, "frozen: reset prepare failed\n");
+		return PCI_ERS_RESULT_NEED_RESET;
+	case pci_channel_io_perm_failure:
+		dev_warn(&pci_dev->dev,
+			 "permanent failure, disconnecting device\n");
+		{
+			struct virtio_driver *drv =
+				drv_to_virtio(vp_dev->vdev.dev.driver);
+
+			if (drv && drv->shutdown)
+				drv->shutdown(&vp_dev->vdev);
+			else
+				virtio_break_device(&vp_dev->vdev);
+		}
+		return PCI_ERS_RESULT_DISCONNECT;
+	default:
+		break;
+	}
+	return PCI_ERS_RESULT_NEED_RESET;
+}
+
 static const struct pci_error_handlers virtio_pci_err_handler = {
+	.error_detected = virtio_pci_error_detected,
 	.reset_prepare  = virtio_pci_reset_prepare,
 	.reset_done     = virtio_pci_reset_done,
 };