Merge tag 'block-6.1-2022-11-05' of git://git.kernel.dk/linux

Pull block fixes from Jens Axboe:

 - Fixes for the ublk driver (Ming)

 - Fixes for error handling memory leaks (Chen Jun, Chen Zhongjin)

 - Explicitly clear the last request in a chain when the plug is
   flushed, as it may have already been issued (Al)

* tag 'block-6.1-2022-11-05' of git://git.kernel.dk/linux:
  block: blk_add_rq_to_plug(): clear stale 'last' after flush
  blk-mq: Fix kmemleak in blk_mq_init_allocated_queue
  block: Fix possible memory leak for rq_wb on add_disk failure
  ublk_drv: add ublk_queue_cmd() for cleanup
  ublk_drv: avoid to touch io_uring cmd in blk_mq io path
  ublk_drv: comment on ublk_driver entry of Kconfig
  ublk_drv: return flag of UBLK_F_URING_CMD_COMP_IN_TASK in case of module

4 files changed, 77 insertions, 50 deletions

diff --git a/block/blk-mq.c b/block/blk-mq.c
index 75c8296b6feb..6a789cda68a5 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1262,6 +1262,7 @@ static void blk_add_rq_to_plug(struct blk_plug *plug, struct request *rq)
 		   (!blk_queue_nomerges(rq->q) &&
 		    blk_rq_bytes(last) >= BLK_PLUG_FLUSH_SIZE)) {
 		blk_mq_flush_plug_list(plug, false);
+		last = NULL;
 		trace_block_plug(rq->q);
 	}
 
@@ -4193,9 +4194,7 @@ int blk_mq_init_allocated_queue(struct blk_mq_tag_set *set,
 	return 0;
 
 err_hctxs:
-	xa_destroy(&q->hctx_table);
-	q->nr_hw_queues = 0;
-	blk_mq_sysfs_deinit(q);
+	blk_mq_release(q);
 err_poll:
 	blk_stat_free_callback(q->poll_cb);
 	q->poll_cb = NULL;
diff --git a/block/genhd.c b/block/genhd.c
index fee90eb98b4a..0f9769db2de8 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -527,6 +527,7 @@ out_unregister_bdi:
 		bdi_unregister(disk->bdi);
 out_unregister_queue:
 	blk_unregister_queue(disk);
+	rq_qos_exit(disk->queue);
 out_put_slave_dir:
 	kobject_put(disk->slave_dir);
 out_put_holder_dir:
diff --git a/drivers/block/Kconfig b/drivers/block/Kconfig
index db1b4b202646..a41145d52de9 100644
--- a/drivers/block/Kconfig
+++ b/drivers/block/Kconfig
@@ -408,6 +408,12 @@ config BLK_DEV_UBLK
 	  definition isn't finalized yet, and might change according to future
 	  requirement, so mark is as experimental now.
 
+	  Say Y if you want to get better performance because task_work_add()
+	  can be used in IO path for replacing io_uring cmd, which will become
+	  shared between IO tasks and ubq daemon, meantime task_work_add() can
+	  can handle batch more effectively, but task_work_add() isn't exported
+	  for module, so ublk has to be built to kernel.
+
 source "drivers/block/rnbd/Kconfig"
 
 endif # BLK_DEV
diff --git a/drivers/block/ublk_drv.c b/drivers/block/ublk_drv.c
index 5afce6ffaadf..f96cb01e9604 100644
--- a/drivers/block/ublk_drv.c
+++ b/drivers/block/ublk_drv.c
@@ -57,11 +57,14 @@
 #define UBLK_PARAM_TYPE_ALL (UBLK_PARAM_TYPE_BASIC | UBLK_PARAM_TYPE_DISCARD)
 
 struct ublk_rq_data {
-	struct callback_head work;
+	union {
+		struct callback_head work;
+		struct llist_node node;
+	};
 };
 
 struct ublk_uring_cmd_pdu {
-	struct request *req;
+	struct ublk_queue *ubq;
 };
 
 /*
@@ -119,6 +122,8 @@ 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;
@@ -764,8 +769,12 @@ static inline void __ublk_rq_task_work(struct request *req)
 static void ublk_rq_task_work_cb(struct io_uring_cmd *cmd)
 {
 	struct ublk_uring_cmd_pdu *pdu = ublk_get_uring_cmd_pdu(cmd);
+	struct ublk_queue *ubq = pdu->ubq;
+	struct llist_node *io_cmds = llist_del_all(&ubq->io_cmds);
+	struct ublk_rq_data *data;
 
-	__ublk_rq_task_work(pdu->req);
+	llist_for_each_entry(data, io_cmds, node)
+		__ublk_rq_task_work(blk_mq_rq_from_pdu(data));
 }
 
 static void ublk_rq_task_work_fn(struct callback_head *work)
@@ -777,6 +786,54 @@ static void ublk_rq_task_work_fn(struct callback_head *work)
 	__ublk_rq_task_work(req);
 }
 
+static void ublk_submit_cmd(struct ublk_queue *ubq, const struct request *rq)
+{
+	struct ublk_io *io = &ubq->ios[rq->tag];
+
+	/*
+	 * If the check pass, we know that this is a re-issued request aborted
+	 * previously in monitor_work because the ubq_daemon(cmd's task) is
+	 * PF_EXITING. We cannot call io_uring_cmd_complete_in_task() anymore
+	 * because this ioucmd's io_uring context may be freed now if no inflight
+	 * ioucmd exists. Otherwise we may cause null-deref in ctx->fallback_work.
+	 *
+	 * Note: monitor_work sets UBLK_IO_FLAG_ABORTED and ends this request(releasing
+	 * the tag). Then the request is re-started(allocating the tag) and we are here.
+	 * Since releasing/allocating a tag implies smp_mb(), finding UBLK_IO_FLAG_ABORTED
+	 * guarantees that here is a re-issued request aborted previously.
+	 */
+	if (unlikely(io->flags & UBLK_IO_FLAG_ABORTED)) {
+		struct llist_node *io_cmds = llist_del_all(&ubq->io_cmds);
+		struct ublk_rq_data *data;
+
+		llist_for_each_entry(data, io_cmds, node)
+			__ublk_abort_rq(ubq, blk_mq_rq_from_pdu(data));
+	} else {
+		struct io_uring_cmd *cmd = io->cmd;
+		struct ublk_uring_cmd_pdu *pdu = ublk_get_uring_cmd_pdu(cmd);
+
+		pdu->ubq = ubq;
+		io_uring_cmd_complete_in_task(cmd, ublk_rq_task_work_cb);
+	}
+}
+
+static void ublk_queue_cmd(struct ublk_queue *ubq, struct request *rq,
+		bool last)
+{
+	struct ublk_rq_data *data = blk_mq_rq_to_pdu(rq);
+
+	if (ublk_can_use_task_work(ubq)) {
+		enum task_work_notify_mode notify_mode = last ?
+			TWA_SIGNAL_NO_IPI : TWA_NONE;
+
+		if (task_work_add(ubq->ubq_daemon, &data->work, notify_mode))
+			__ublk_abort_rq(ubq, rq);
+	} else {
+		if (llist_add(&data->node, &ubq->io_cmds))
+			ublk_submit_cmd(ubq, rq);
+	}
+}
+
 static blk_status_t ublk_queue_rq(struct blk_mq_hw_ctx *hctx,
 		const struct blk_mq_queue_data *bd)
 {
@@ -788,6 +845,7 @@ static blk_status_t ublk_queue_rq(struct blk_mq_hw_ctx *hctx,
 	res = ublk_setup_iod(ubq, rq);
 	if (unlikely(res != BLK_STS_OK))
 		return BLK_STS_IOERR;
+
 	/* With recovery feature enabled, force_abort is set in
 	 * ublk_stop_dev() before calling del_gendisk(). We have to
 	 * abort all requeued and new rqs here to let del_gendisk()
@@ -803,41 +861,11 @@ static blk_status_t ublk_queue_rq(struct blk_mq_hw_ctx *hctx,
 	blk_mq_start_request(bd->rq);
 
 	if (unlikely(ubq_daemon_is_dying(ubq))) {
- fail:
 		__ublk_abort_rq(ubq, rq);
 		return BLK_STS_OK;
 	}
 
-	if (ublk_can_use_task_work(ubq)) {
-		struct ublk_rq_data *data = blk_mq_rq_to_pdu(rq);
-		enum task_work_notify_mode notify_mode = bd->last ?
-			TWA_SIGNAL_NO_IPI : TWA_NONE;
-
-		if (task_work_add(ubq->ubq_daemon, &data->work, notify_mode))
-			goto fail;
-	} else {
-		struct ublk_io *io = &ubq->ios[rq->tag];
-		struct io_uring_cmd *cmd = io->cmd;
-		struct ublk_uring_cmd_pdu *pdu = ublk_get_uring_cmd_pdu(cmd);
-
-		/*
-		 * If the check pass, we know that this is a re-issued request aborted
-		 * previously in monitor_work because the ubq_daemon(cmd's task) is
-		 * PF_EXITING. We cannot call io_uring_cmd_complete_in_task() anymore
-		 * because this ioucmd's io_uring context may be freed now if no inflight
-		 * ioucmd exists. Otherwise we may cause null-deref in ctx->fallback_work.
-		 *
-		 * Note: monitor_work sets UBLK_IO_FLAG_ABORTED and ends this request(releasing
-		 * the tag). Then the request is re-started(allocating the tag) and we are here.
-		 * Since releasing/allocating a tag implies smp_mb(), finding UBLK_IO_FLAG_ABORTED
-		 * guarantees that here is a re-issued request aborted previously.
-		 */
-		if ((io->flags & UBLK_IO_FLAG_ABORTED))
-			goto fail;
-
-		pdu->req = rq;
-		io_uring_cmd_complete_in_task(cmd, ublk_rq_task_work_cb);
-	}
+	ublk_queue_cmd(ubq, rq, bd->last);
 
 	return BLK_STS_OK;
 }
@@ -1164,22 +1192,12 @@ static void ublk_mark_io_ready(struct ublk_device *ub, struct ublk_queue *ubq)
 }
 
 static void ublk_handle_need_get_data(struct ublk_device *ub, int q_id,
-		int tag, struct io_uring_cmd *cmd)
+		int tag)
 {
 	struct ublk_queue *ubq = ublk_get_queue(ub, q_id);
 	struct request *req = blk_mq_tag_to_rq(ub->tag_set.tags[q_id], tag);
 
-	if (ublk_can_use_task_work(ubq)) {
-		struct ublk_rq_data *data = blk_mq_rq_to_pdu(req);
-
-		/* should not fail since we call it just in ubq->ubq_daemon */
-		task_work_add(ubq->ubq_daemon, &data->work, TWA_SIGNAL_NO_IPI);
-	} else {
-		struct ublk_uring_cmd_pdu *pdu = ublk_get_uring_cmd_pdu(cmd);
-
-		pdu->req = req;
-		io_uring_cmd_complete_in_task(cmd, ublk_rq_task_work_cb);
-	}
+	ublk_queue_cmd(ubq, req, true);
 }
 
 static int ublk_ch_uring_cmd(struct io_uring_cmd *cmd, unsigned int issue_flags)
@@ -1267,7 +1285,7 @@ static int ublk_ch_uring_cmd(struct io_uring_cmd *cmd, unsigned int issue_flags)
 		io->addr = ub_cmd->addr;
 		io->cmd = cmd;
 		io->flags |= UBLK_IO_FLAG_ACTIVE;
-		ublk_handle_need_get_data(ub, ub_cmd->q_id, ub_cmd->tag, cmd);
+		ublk_handle_need_get_data(ub, ub_cmd->q_id, ub_cmd->tag);
 		break;
 	default:
 		goto out;
@@ -1658,6 +1676,9 @@ static int ublk_ctrl_add_dev(struct io_uring_cmd *cmd)
 	 */
 	ub->dev_info.flags &= UBLK_F_ALL;
 
+	if (!IS_BUILTIN(CONFIG_BLK_DEV_UBLK))
+		ub->dev_info.flags |= UBLK_F_URING_CMD_COMP_IN_TASK;
+
 	/* We are not ready to support zero copy */
 	ub->dev_info.flags &= ~UBLK_F_SUPPORT_ZERO_COPY;