io_uring/uring_cmd: avoid double indirect call in task work dispatch

io_uring task work dispatch makes an indirect call to struct io_kiocb's
io_task_work.func field to allow running arbitrary task work functions.
In the uring_cmd case, this calls io_uring_cmd_work(), which immediately
makes another indirect call to struct io_uring_cmd's task_work_cb field.
Change the uring_cmd task work callbacks to functions whose signatures
match io_req_tw_func_t. Add a function io_uring_cmd_from_tw() to convert
from the task work's struct io_tw_req argument to struct io_uring_cmd *.
Define a constant IO_URING_CMD_TASK_WORK_ISSUE_FLAGS to avoid
manufacturing issue_flags in the uring_cmd task work callbacks. Now
uring_cmd task work dispatch makes a single indirect call to the
uring_cmd implementation's callback. This also allows removing the
task_work_cb field from struct io_uring_cmd, freeing up 8 bytes for
future storage.
Since fuse_uring_send_in_task() now has access to the io_tw_token_t,
check its cancel field directly instead of relying on the
IO_URING_F_TASK_DEAD issue flag.

Signed-off-by: Caleb Sander Mateos <[e-mail]>
Reviewed-by: Christoph Hellwig <[e-mail]>
Signed-off-by: Jens Axboe <[e-mail]>

block/ioctl.c

···
 	bool nowait;
 };
 
-static void blk_cmd_complete(struct io_uring_cmd *cmd, unsigned int issue_flags)
+static void blk_cmd_complete(struct io_tw_req tw_req, io_tw_token_t tw)
 {
+	struct io_uring_cmd *cmd = io_uring_cmd_from_tw(tw_req);
 	struct blk_iou_cmd *bic = io_uring_cmd_to_pdu(cmd, struct blk_iou_cmd);
 
 	if (bic->res == -EAGAIN && bic->nowait)
 		io_uring_cmd_issue_blocking(cmd);
 	else
-		io_uring_cmd_done(cmd, bic->res, issue_flags);
+		io_uring_cmd_done(cmd, bic->res,
+				  IO_URING_CMD_TASK_WORK_ISSUE_FLAGS);
 }
 
 static void bio_cmd_bio_end_io(struct bio *bio)

drivers/block/ublk_drv.c

···
 	return true;
 }
 
-static void ublk_dispatch_req(struct ublk_queue *ubq,
-			      struct request *req,
-			      unsigned int issue_flags)
+static void ublk_dispatch_req(struct ublk_queue *ubq, struct request *req)
 {
+	unsigned int issue_flags = IO_URING_CMD_TASK_WORK_ISSUE_FLAGS;
 	int tag = req->tag;
 	struct ublk_io *io = &ubq->ios[tag];
 
···
 		ublk_complete_io_cmd(io, req, UBLK_IO_RES_OK, issue_flags);
 }
 
-static void ublk_cmd_tw_cb(struct io_uring_cmd *cmd,
-			   unsigned int issue_flags)
+static void ublk_cmd_tw_cb(struct io_tw_req tw_req, io_tw_token_t tw)
 {
+	struct io_uring_cmd *cmd = io_uring_cmd_from_tw(tw_req);
 	struct ublk_uring_cmd_pdu *pdu = ublk_get_uring_cmd_pdu(cmd);
 	struct ublk_queue *ubq = pdu->ubq;
 
-	ublk_dispatch_req(ubq, pdu->req, issue_flags);
+	ublk_dispatch_req(ubq, pdu->req);
 }
 
 static void ublk_queue_cmd(struct ublk_queue *ubq, struct request *rq)
···
 	io_uring_cmd_complete_in_task(cmd, ublk_cmd_tw_cb);
 }
 
-static void ublk_cmd_list_tw_cb(struct io_uring_cmd *cmd,
-		unsigned int issue_flags)
+static void ublk_cmd_list_tw_cb(struct io_tw_req tw_req, io_tw_token_t tw)
 {
+	struct io_uring_cmd *cmd = io_uring_cmd_from_tw(tw_req);
 	struct ublk_uring_cmd_pdu *pdu = ublk_get_uring_cmd_pdu(cmd);
 	struct request *rq = pdu->req_list;
 	struct request *next;
···
 	do {
 		next = rq->rq_next;
 		rq->rq_next = NULL;
-		ublk_dispatch_req(rq->mq_hctx->driver_data, rq, issue_flags);
+		ublk_dispatch_req(rq->mq_hctx->driver_data, rq);
 		rq = next;
 	} while (rq);
 }
···
 	return NULL;
 }
 
-static void ublk_ch_uring_cmd_cb(struct io_uring_cmd *cmd,
-		unsigned int issue_flags)
+static void ublk_ch_uring_cmd_cb(struct io_tw_req tw_req, io_tw_token_t tw)
 {
+	unsigned int issue_flags = IO_URING_CMD_TASK_WORK_ISSUE_FLAGS;
+	struct io_uring_cmd *cmd = io_uring_cmd_from_tw(tw_req);
 	int ret = ublk_ch_uring_cmd_local(cmd, issue_flags);
 
 	if (ret != -EIOCBQUEUED)

drivers/nvme/host/ioctl.c

···
 	return io_uring_cmd_to_pdu(ioucmd, struct nvme_uring_cmd_pdu);
 }
 
-static void nvme_uring_task_cb(struct io_uring_cmd *ioucmd,
-			       unsigned issue_flags)
+static void nvme_uring_task_cb(struct io_tw_req tw_req, io_tw_token_t tw)
 {
+	struct io_uring_cmd *ioucmd = io_uring_cmd_from_tw(tw_req);
 	struct nvme_uring_cmd_pdu *pdu = nvme_uring_cmd_pdu(ioucmd);
 
 	if (pdu->bio)
 		blk_rq_unmap_user(pdu->bio);
-	io_uring_cmd_done32(ioucmd, pdu->status, pdu->result, issue_flags);
+	io_uring_cmd_done32(ioucmd, pdu->status, pdu->result,
+			    IO_URING_CMD_TASK_WORK_ISSUE_FLAGS);
 }
 
 static enum rq_end_io_ret nvme_uring_cmd_end_io(struct request *req,

fs/btrfs/ioctl.c

···
 	struct btrfs_uring_priv *priv;
 };
 
-static void btrfs_uring_read_finished(struct io_uring_cmd *cmd, unsigned int issue_flags)
+static void btrfs_uring_read_finished(struct io_tw_req tw_req, io_tw_token_t tw)
 {
+	struct io_uring_cmd *cmd = io_uring_cmd_from_tw(tw_req);
 	struct io_btrfs_cmd *bc = io_uring_cmd_to_pdu(cmd, struct io_btrfs_cmd);
 	struct btrfs_uring_priv *priv = bc->priv;
 	struct btrfs_inode *inode = BTRFS_I(file_inode(priv->iocb.ki_filp));
···
 	btrfs_unlock_extent(io_tree, priv->start, priv->lockend, &priv->cached_state);
 	btrfs_inode_unlock(inode, BTRFS_ILOCK_SHARED);
 
-	io_uring_cmd_done(cmd, ret, issue_flags);
+	io_uring_cmd_done(cmd, ret, IO_URING_CMD_TASK_WORK_ISSUE_FLAGS);
 	add_rchar(current, ret);
 
 	for (index = 0; index < priv->nr_pages; index++)

fs/fuse/dev_uring.c

···
  * User buffers are not mapped yet - the application does not have permission
  * to write to it - this has to be executed in ring task context.
  */
-static void fuse_uring_send_in_task(struct io_uring_cmd *cmd,
-				    unsigned int issue_flags)
+static void fuse_uring_send_in_task(struct io_tw_req tw_req, io_tw_token_t tw)
 {
+	unsigned int issue_flags = IO_URING_CMD_TASK_WORK_ISSUE_FLAGS;
+	struct io_uring_cmd *cmd = io_uring_cmd_from_tw(tw_req);
 	struct fuse_ring_ent *ent = uring_cmd_to_ring_ent(cmd);
 	struct fuse_ring_queue *queue = ent->queue;
 	int err;
 
-	if (!(issue_flags & IO_URING_F_TASK_DEAD)) {
+	if (!tw.cancel) {
 		err = fuse_uring_prepare_send(ent, ent->fuse_req);
 		if (err) {
 			fuse_uring_next_fuse_req(ent, queue, issue_flags);

include/linux/io_uring/cmd.h

···
 /* io_uring_cmd is being issued again */
 #define IORING_URING_CMD_REISSUE	(1U << 31)
 
-typedef void (*io_uring_cmd_tw_t)(struct io_uring_cmd *cmd,
-				  unsigned issue_flags);
-
 struct io_uring_cmd {
 	struct file	*file;
 	const struct io_uring_sqe *sqe;
-	/* callback to defer completions to task context */
-	io_uring_cmd_tw_t task_work_cb;
 	u32		cmd_op;
 	u32		flags;
 	u8		pdu[32]; /* available inline for free use */
+	u8		unused[8];
 };
 
 static inline const void *io_uring_sqe_cmd(const struct io_uring_sqe *sqe)
···
 			 unsigned issue_flags, bool is_cqe32);
 
 void __io_uring_cmd_do_in_task(struct io_uring_cmd *ioucmd,
-			    io_uring_cmd_tw_t task_work_cb,
+			    io_req_tw_func_t task_work_cb,
 			    unsigned flags);
 
 /*
···
 {
 }
 static inline void __io_uring_cmd_do_in_task(struct io_uring_cmd *ioucmd,
-			    io_uring_cmd_tw_t task_work_cb, unsigned flags)
+			    io_req_tw_func_t task_work_cb, unsigned flags)
 {
 }
 static inline void io_uring_cmd_mark_cancelable(struct io_uring_cmd *cmd,
···
 }
 #endif
 
+static inline struct io_uring_cmd *io_uring_cmd_from_tw(struct io_tw_req tw_req)
+{
+	return io_kiocb_to_cmd(tw_req.req, struct io_uring_cmd);
+}
+
+/* task_work executor checks the deferred list completion */
+#define IO_URING_CMD_TASK_WORK_ISSUE_FLAGS IO_URING_F_COMPLETE_DEFER
+
 /* users must follow the IOU_F_TWQ_LAZY_WAKE semantics */
 static inline void io_uring_cmd_do_in_task_lazy(struct io_uring_cmd *ioucmd,
-			io_uring_cmd_tw_t task_work_cb)
+			io_req_tw_func_t task_work_cb)
 {
 	__io_uring_cmd_do_in_task(ioucmd, task_work_cb, IOU_F_TWQ_LAZY_WAKE);
 }
 
 static inline void io_uring_cmd_complete_in_task(struct io_uring_cmd *ioucmd,
-			io_uring_cmd_tw_t task_work_cb)
+			io_req_tw_func_t task_work_cb)
 {
 	__io_uring_cmd_do_in_task(ioucmd, task_work_cb, 0);
 }

include/linux/io_uring_types.h

···
 	/* set when uring wants to cancel a previously issued command */
 	IO_URING_F_CANCEL		= (1 << 11),
 	IO_URING_F_COMPAT		= (1 << 12),
-	IO_URING_F_TASK_DEAD		= (1 << 13),
 };
 
 struct io_wq_work_node {

io_uring/uring_cmd.c

···
 }
 EXPORT_SYMBOL_GPL(io_uring_cmd_mark_cancelable);
 
-static void io_uring_cmd_work(struct io_tw_req tw_req, io_tw_token_t tw)
-{
-	struct io_kiocb *req = tw_req.req;
-	struct io_uring_cmd *ioucmd = io_kiocb_to_cmd(req, struct io_uring_cmd);
-	unsigned int flags = IO_URING_F_COMPLETE_DEFER;
-
-	if (unlikely(tw.cancel))
-		flags |= IO_URING_F_TASK_DEAD;
-
-	/* task_work executor checks the deffered list completion */
-	ioucmd->task_work_cb(ioucmd, flags);
-}
-
 void __io_uring_cmd_do_in_task(struct io_uring_cmd *ioucmd,
-			io_uring_cmd_tw_t task_work_cb,
+			io_req_tw_func_t task_work_cb,
 			unsigned flags)
 {
 	struct io_kiocb *req = cmd_to_io_kiocb(ioucmd);
···
 	if (WARN_ON_ONCE(req->flags & REQ_F_APOLL_MULTISHOT))
 		return;
 
-	ioucmd->task_work_cb = task_work_cb;
-	req->io_task_work.func = io_uring_cmd_work;
+	req->io_task_work.func = task_work_cb;
 	__io_req_task_work_add(req, flags);
 }
 EXPORT_SYMBOL_GPL(__io_uring_cmd_do_in_task);