On Thu, 09 Jun 2022, Sebastian Andrzej Siewior wrote:

On 2022-05-30 16:15:10 [-0700], Davidlohr Bueso wrote:

quoted

diff --git a/drivers/scsi/ibmvscsi/ibmvfc.c b/drivers/scsi/ibmvscsi/ibmvfc.c
index d0eab5700dc5..31b1900489e7 100644
--- a/drivers/scsi/ibmvscsi/ibmvfc.c
+++ b/drivers/scsi/ibmvscsi/ibmvfc.c

@@ -891,7 +891,7 @@ static void ibmvfc_release_crq_queue(struct ibmvfc_host *vhost)

ibmvfc_dbg(vhost, "Releasing CRQ\n");
free_irq(vdev->irq, vhost);
-	tasklet_kill(&vhost->tasklet);
+        cancel_work_sync(&vhost->work);

s/ {8}/\t/

is there a reason not to use threaded interrupts?

I went with a workqueue here because the resume from suspend also schedules
async processing, so threaded irqs didn't seem like a good fit. This is also
similar to patch 2 (but in that case I overlooked the resume caller which you
pointed out).

Thanks,
Davidlohr