Commit 989bcd62 authored by Uday Shankar's avatar Uday Shankar Committed by Jens Axboe
Browse files

ublk: remove io_cmds list in ublk_queue 



The current I/O dispatch mechanism - queueing I/O by adding it to the
io_cmds list (and poking task_work as needed), then dispatching it in
ublk server task context by reversing io_cmds and completing the
io_uring command associated to each one - was introduced by commit
7d4a9317 ("ublk_drv: don't forward io commands in reserve order")
to ensure that the ublk server received I/O in the same order that the
block layer submitted it to ublk_drv. This mechanism was only needed for
the "raw" task_work submission mechanism, since the io_uring task work
wrapper maintains FIFO ordering (using quite a similar mechanism in
fact). The "raw" task_work submission mechanism is no longer supported
in ublk_drv as of commit 29dc5d06 ("ublk: kill queuing request by
task_work_add"), so the explicit llist/reversal is no longer needed - it
just duplicates logic already present in the underlying io_uring APIs.
Remove it.

Signed-off-by: default avatarUday Shankar <ushankar@purestorage.com>
Reviewed-by: default avatarMing Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20250318-ublk_io_cmds-v1-1-c1bb74798fef@purestorage.com


Signed-off-by: default avatarJens Axboe <axboe@kernel.dk>
parent e1a0202c

drivers/block/ublk_drv.c

+11 −35
Original line number Diff line number Diff line
@@ -74,8 +74,6 @@ 
	 UBLK_PARAM_TYPE_DMA_ALIGN)



struct ublk_rq_data {

	struct llist_node node;



	struct kref ref;

};
@@ -142,8 +140,6 @@ struct ublk_queue { 
	struct task_struct	*ubq_daemon;

	char *io_cmd_buf;



	struct llist_head	io_cmds;



	unsigned long io_addr;	/* mapped vm address */

	unsigned int max_io_sz;

	bool force_abort;
@@ -1108,7 +1104,7 @@ static void ublk_complete_rq(struct kref *ref) 
}



/*

 * Since __ublk_rq_task_work always fails requests immediately during

 * Since ublk_rq_task_work_cb always fails requests immediately during

 * exiting, __ublk_fail_req() is only called from abort context during

 * exiting. So lock is unnecessary.

 *
@@ -1154,11 +1150,14 @@ static inline void __ublk_abort_rq(struct ublk_queue *ubq, 
		blk_mq_end_request(rq, BLK_STS_IOERR);

}



static inline void __ublk_rq_task_work(struct request *req,

				       unsigned issue_flags)

static void ublk_rq_task_work_cb(struct io_uring_cmd *cmd,

				 unsigned int issue_flags)

{

	struct ublk_queue *ubq = req->mq_hctx->driver_data;

	int tag = req->tag;

	struct ublk_uring_cmd_pdu *pdu = ublk_get_uring_cmd_pdu(cmd);

	struct ublk_queue *ubq = pdu->ubq;

	int tag = pdu->tag;

	struct request *req = blk_mq_tag_to_rq(

		ubq->dev->tag_set.tags[ubq->q_id], tag);

	struct ublk_io *io = &ubq->ios[tag];

	unsigned int mapped_bytes;
@@ -1233,35 +1232,12 @@ static inline void __ublk_rq_task_work(struct request *req, 
	ubq_complete_io_cmd(io, UBLK_IO_RES_OK, issue_flags);

}



static inline void ublk_forward_io_cmds(struct ublk_queue *ubq,

					unsigned issue_flags)

{

	struct llist_node *io_cmds = llist_del_all(&ubq->io_cmds);

	struct ublk_rq_data *data, *tmp;



	io_cmds = llist_reverse_order(io_cmds);

	llist_for_each_entry_safe(data, tmp, io_cmds, node)

		__ublk_rq_task_work(blk_mq_rq_from_pdu(data), issue_flags);

}



static void ublk_rq_task_work_cb(struct io_uring_cmd *cmd, unsigned issue_flags)

{

	struct ublk_uring_cmd_pdu *pdu = ublk_get_uring_cmd_pdu(cmd);

	struct ublk_queue *ubq = pdu->ubq;



	ublk_forward_io_cmds(ubq, issue_flags);

}



static void ublk_queue_cmd(struct ublk_queue *ubq, struct request *rq)

{

	struct ublk_rq_data *data = blk_mq_rq_to_pdu(rq);



	if (llist_add(&data->node, &ubq->io_cmds)) {

	struct ublk_io *io = &ubq->ios[rq->tag];



	io_uring_cmd_complete_in_task(io->cmd, ublk_rq_task_work_cb);

}

}



static enum blk_eh_timer_return ublk_timeout(struct request *rq)

{
@@ -1453,7 +1429,7 @@ static void ublk_abort_queue(struct ublk_device *ub, struct ublk_queue *ubq) 
			struct request *rq;



			/*

			 * Either we fail the request or ublk_rq_task_work_fn

			 * Either we fail the request or ublk_rq_task_work_cb

			 * will do it

			 */

			rq = blk_mq_tag_to_rq(ub->tag_set.tags[ubq->q_id], i);