1

nvme-loop: flush off pending I/O while shutting down loop controller

While shutting down loop controller, we first quiesce the admin/IO queue,
delete the admin/IO tag-set and then at last destroy the admin/IO queue.
However it's quite possible that during the window between quiescing and
destroying of the admin/IO queue, some admin/IO request might sneak in
and if that happens then we could potentially encounter a hung task
because shutdown operation can't forward progress until any pending I/O
is flushed off.

This commit helps ensure that before destroying the admin/IO queue, we
unquiesce the admin/IO queue so that any outstanding requests, which are
added after the admin/IO queue is quiesced, are now flushed to its
completion.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Nilay Shroff <nilay@linux.ibm.com>
Signed-off-by: Keith Busch <kbusch@kernel.org>
This commit is contained in:
Nilay Shroff 2024-10-16 08:33:14 +05:30 committed by Keith Busch
parent 26bc0a81f6
commit c199fac88f

View File

@ -265,6 +265,13 @@ static void nvme_loop_destroy_admin_queue(struct nvme_loop_ctrl *ctrl)
{
if (!test_and_clear_bit(NVME_LOOP_Q_LIVE, &ctrl->queues[0].flags))
return;
/*
* It's possible that some requests might have been added
* after admin queue is stopped/quiesced. So now start the
* queue to flush these requests to the completion.
*/
nvme_unquiesce_admin_queue(&ctrl->ctrl);
nvmet_sq_destroy(&ctrl->queues[0].nvme_sq);
nvme_remove_admin_tag_set(&ctrl->ctrl);
}
@ -297,6 +304,12 @@ static void nvme_loop_destroy_io_queues(struct nvme_loop_ctrl *ctrl)
nvmet_sq_destroy(&ctrl->queues[i].nvme_sq);
}
ctrl->ctrl.queue_count = 1;
/*
* It's possible that some requests might have been added
* after io queue is stopped/quiesced. So now start the
* queue to flush these requests to the completion.
*/
nvme_unquiesce_io_queues(&ctrl->ctrl);
}
static int nvme_loop_init_io_queues(struct nvme_loop_ctrl *ctrl)