[deliverable/linux.git] / drivers / block / null_blk.c

#include <linux/module.h>

#include <linux/moduleparam.h>
#include <linux/sched.h>
#include <linux/fs.h>
#include <linux/blkdev.h>
#include <linux/init.h>
#include <linux/slab.h>
#include <linux/blk-mq.h>
#include <linux/hrtimer.h>

struct nullb_cmd {
	struct list_head list;
	struct llist_node ll_list;
	struct call_single_data csd;
	struct request *rq;
	struct bio *bio;
	unsigned int tag;
	struct nullb_queue *nq;
};

struct nullb_queue {
	unsigned long *tag_map;
	wait_queue_head_t wait;
	unsigned int queue_depth;

	struct nullb_cmd *cmds;
};

struct nullb {
	struct list_head list;
	unsigned int index;
	struct request_queue *q;
	struct gendisk *disk;
	struct hrtimer timer;
	unsigned int queue_depth;
	spinlock_t lock;

	struct nullb_queue *queues;
	unsigned int nr_queues;
};

static LIST_HEAD(nullb_list);
static struct mutex lock;
static int null_major;
static int nullb_indexes;

struct completion_queue {
	struct llist_head list;
	struct hrtimer timer;
};

/*
 * These are per-cpu for now, they will need to be configured by the
 * complete_queues parameter and appropriately mapped.
 */
static DEFINE_PER_CPU(struct completion_queue, completion_queues);

enum {
	NULL_IRQ_NONE		= 0,
	NULL_IRQ_SOFTIRQ	= 1,
	NULL_IRQ_TIMER		= 2,

	NULL_Q_BIO		= 0,
	NULL_Q_RQ		= 1,
	NULL_Q_MQ		= 2,
};

static int submit_queues;
module_param(submit_queues, int, S_IRUGO);
MODULE_PARM_DESC(submit_queues, "Number of submission queues");

static int home_node = NUMA_NO_NODE;
module_param(home_node, int, S_IRUGO);
MODULE_PARM_DESC(home_node, "Home node for the device");

static int queue_mode = NULL_Q_MQ;
module_param(queue_mode, int, S_IRUGO);
MODULE_PARM_DESC(use_mq, "Use blk-mq interface (0=bio,1=rq,2=multiqueue)");

static int gb = 250;
module_param(gb, int, S_IRUGO);
MODULE_PARM_DESC(gb, "Size in GB");

static int bs = 512;
module_param(bs, int, S_IRUGO);
MODULE_PARM_DESC(bs, "Block size (in bytes)");

static int nr_devices = 2;
module_param(nr_devices, int, S_IRUGO);
MODULE_PARM_DESC(nr_devices, "Number of devices to register");

static int irqmode = NULL_IRQ_SOFTIRQ;
module_param(irqmode, int, S_IRUGO);
MODULE_PARM_DESC(irqmode, "IRQ completion handler. 0-none, 1-softirq, 2-timer");

static int completion_nsec = 10000;
module_param(completion_nsec, int, S_IRUGO);
MODULE_PARM_DESC(completion_nsec, "Time in ns to complete a request in hardware. Default: 10,000ns");

static int hw_queue_depth = 64;
module_param(hw_queue_depth, int, S_IRUGO);
MODULE_PARM_DESC(hw_queue_depth, "Queue depth for each hardware queue. Default: 64");

static bool use_per_node_hctx = false;
module_param(use_per_node_hctx, bool, S_IRUGO);
MODULE_PARM_DESC(use_per_node_hctx, "Use per-node allocation for hardware context queues. Default: false");

static void put_tag(struct nullb_queue *nq, unsigned int tag)
{
	clear_bit_unlock(tag, nq->tag_map);

	if (waitqueue_active(&nq->wait))
		wake_up(&nq->wait);
}

static unsigned int get_tag(struct nullb_queue *nq)
{
	unsigned int tag;

	do {
		tag = find_first_zero_bit(nq->tag_map, nq->queue_depth);
		if (tag >= nq->queue_depth)
			return -1U;
	} while (test_and_set_bit_lock(tag, nq->tag_map));

	return tag;
}

static void free_cmd(struct nullb_cmd *cmd)
{
	put_tag(cmd->nq, cmd->tag);
}

static struct nullb_cmd *__alloc_cmd(struct nullb_queue *nq)
{
	struct nullb_cmd *cmd;
	unsigned int tag;

	tag = get_tag(nq);
	if (tag != -1U) {
		cmd = &nq->cmds[tag];
		cmd->tag = tag;
		cmd->nq = nq;
		return cmd;
	}

	return NULL;
}

static struct nullb_cmd *alloc_cmd(struct nullb_queue *nq, int can_wait)
{
	struct nullb_cmd *cmd;
	DEFINE_WAIT(wait);

	cmd = __alloc_cmd(nq);
	if (cmd || !can_wait)
		return cmd;

	do {
		prepare_to_wait(&nq->wait, &wait, TASK_UNINTERRUPTIBLE);
		cmd = __alloc_cmd(nq);
		if (cmd)
			break;

		io_schedule();
	} while (1);

	finish_wait(&nq->wait, &wait);
	return cmd;
}

static void end_cmd(struct nullb_cmd *cmd)
{
	if (cmd->rq) {
		if (queue_mode == NULL_Q_MQ)
			blk_mq_end_io(cmd->rq, 0);
		else {
			INIT_LIST_HEAD(&cmd->rq->queuelist);
			blk_end_request_all(cmd->rq, 0);
		}
	} else if (cmd->bio)
		bio_endio(cmd->bio, 0);

	if (queue_mode != NULL_Q_MQ)
		free_cmd(cmd);
}

static enum hrtimer_restart null_cmd_timer_expired(struct hrtimer *timer)
{
	struct completion_queue *cq;
	struct llist_node *entry;
	struct nullb_cmd *cmd;

	cq = &per_cpu(completion_queues, smp_processor_id());

	while ((entry = llist_del_all(&cq->list)) != NULL) {
		do {
			cmd = container_of(entry, struct nullb_cmd, ll_list);
			end_cmd(cmd);
			entry = entry->next;
		} while (entry);
	}

	return HRTIMER_NORESTART;
}

static void null_cmd_end_timer(struct nullb_cmd *cmd)
{
	struct completion_queue *cq = &per_cpu(completion_queues, get_cpu());

	cmd->ll_list.next = NULL;
	if (llist_add(&cmd->ll_list, &cq->list)) {
		ktime_t kt = ktime_set(0, completion_nsec);

		hrtimer_start(&cq->timer, kt, HRTIMER_MODE_REL);
	}

	put_cpu();
}

static void null_softirq_done_fn(struct request *rq)
{
	blk_end_request_all(rq, 0);
}

#ifdef CONFIG_SMP

static void null_ipi_cmd_end_io(void *data)
{
	struct completion_queue *cq;
	struct llist_node *entry, *next;
	struct nullb_cmd *cmd;

	cq = &per_cpu(completion_queues, smp_processor_id());

	entry = llist_del_all(&cq->list);

	while (entry) {
		next = entry->next;
		cmd = llist_entry(entry, struct nullb_cmd, ll_list);
		end_cmd(cmd);
		entry = next;
	}
}

static void null_cmd_end_ipi(struct nullb_cmd *cmd)
{
	struct call_single_data *data = &cmd->csd;
	int cpu = get_cpu();
	struct completion_queue *cq = &per_cpu(completion_queues, cpu);

	cmd->ll_list.next = NULL;

	if (llist_add(&cmd->ll_list, &cq->list)) {
		data->func = null_ipi_cmd_end_io;
		data->flags = 0;
		__smp_call_function_single(cpu, data, 0);
	}

	put_cpu();
}

#endif /* CONFIG_SMP */

static inline void null_handle_cmd(struct nullb_cmd *cmd)
{
	/* Complete IO by inline, softirq or timer */
	switch (irqmode) {
	case NULL_IRQ_NONE:
		end_cmd(cmd);
		break;
	case NULL_IRQ_SOFTIRQ:
#ifdef CONFIG_SMP
		null_cmd_end_ipi(cmd);
#else
		end_cmd(cmd);
#endif
		break;
	case NULL_IRQ_TIMER:
		null_cmd_end_timer(cmd);
		break;
	}
}

static struct nullb_queue *nullb_to_queue(struct nullb *nullb)
{
	int index = 0;

	if (nullb->nr_queues != 1)
		index = raw_smp_processor_id() / ((nr_cpu_ids + nullb->nr_queues - 1) / nullb->nr_queues);

	return &nullb->queues[index];
}

static void null_queue_bio(struct request_queue *q, struct bio *bio)
{
	struct nullb *nullb = q->queuedata;
	struct nullb_queue *nq = nullb_to_queue(nullb);
	struct nullb_cmd *cmd;

	cmd = alloc_cmd(nq, 1);
	cmd->bio = bio;

	null_handle_cmd(cmd);
}

static int null_rq_prep_fn(struct request_queue *q, struct request *req)
{
	struct nullb *nullb = q->queuedata;
	struct nullb_queue *nq = nullb_to_queue(nullb);
	struct nullb_cmd *cmd;

	cmd = alloc_cmd(nq, 0);
	if (cmd) {
		cmd->rq = req;
		req->special = cmd;
		return BLKPREP_OK;
	}

	return BLKPREP_DEFER;
}

static void null_request_fn(struct request_queue *q)
{
	struct request *rq;

	while ((rq = blk_fetch_request(q)) != NULL) {
		struct nullb_cmd *cmd = rq->special;

		spin_unlock_irq(q->queue_lock);
		null_handle_cmd(cmd);
		spin_lock_irq(q->queue_lock);
	}
}

static int null_queue_rq(struct blk_mq_hw_ctx *hctx, struct request *rq)
{
	struct nullb_cmd *cmd = rq->special;

	cmd->rq = rq;
	cmd->nq = hctx->driver_data;

	null_handle_cmd(cmd);
	return BLK_MQ_RQ_QUEUE_OK;
}

static struct blk_mq_hw_ctx *null_alloc_hctx(struct blk_mq_reg *reg, unsigned int hctx_index)
{
	int b_size = DIV_ROUND_UP(reg->nr_hw_queues, nr_online_nodes);
	int tip = (reg->nr_hw_queues % nr_online_nodes);
	int node = 0, i, n;

	/*
	 * Split submit queues evenly wrt to the number of nodes. If uneven,
	 * fill the first buckets with one extra, until the rest is filled with
	 * no extra.
	 */
	for (i = 0, n = 1; i < hctx_index; i++, n++) {
		if (n % b_size == 0) {
			n = 0;
			node++;

			tip--;
			if (!tip)
				b_size = reg->nr_hw_queues / nr_online_nodes;
		}
	}

	/*
	 * A node might not be online, therefore map the relative node id to the
	 * real node id.
	 */
	for_each_online_node(n) {
		if (!node)
			break;
		node--;
	}

	return kzalloc_node(sizeof(struct blk_mq_hw_ctx), GFP_KERNEL, n);
}

static void null_free_hctx(struct blk_mq_hw_ctx *hctx, unsigned int hctx_index)
{
	kfree(hctx);
}

static void null_init_queue(struct nullb *nullb, struct nullb_queue *nq)
{
	BUG_ON(!nullb);
	BUG_ON(!nq);

	init_waitqueue_head(&nq->wait);
	nq->queue_depth = nullb->queue_depth;
}

static int null_init_hctx(struct blk_mq_hw_ctx *hctx, void *data,
			  unsigned int index)
{
	struct nullb *nullb = data;
	struct nullb_queue *nq = &nullb->queues[index];

	hctx->driver_data = nq;
	null_init_queue(nullb, nq);
	nullb->nr_queues++;

	return 0;
}

static struct blk_mq_ops null_mq_ops = {
	.queue_rq       = null_queue_rq,
	.map_queue      = blk_mq_map_queue,
	.init_hctx	= null_init_hctx,
};

static struct blk_mq_reg null_mq_reg = {
	.ops		= &null_mq_ops,
	.queue_depth	= 64,
	.cmd_size	= sizeof(struct nullb_cmd),
	.flags		= BLK_MQ_F_SHOULD_MERGE,
};

static void null_del_dev(struct nullb *nullb)
{
	list_del_init(&nullb->list);

	del_gendisk(nullb->disk);
	if (queue_mode == NULL_Q_MQ)
		blk_mq_free_queue(nullb->q);
	else
		blk_cleanup_queue(nullb->q);
	put_disk(nullb->disk);
	kfree(nullb);
}

static int null_open(struct block_device *bdev, fmode_t mode)
{
	return 0;
}

static void null_release(struct gendisk *disk, fmode_t mode)
{
}

static const struct block_device_operations null_fops = {
	.owner =	THIS_MODULE,
	.open =		null_open,
	.release =	null_release,
};

static int setup_commands(struct nullb_queue *nq)
{
	struct nullb_cmd *cmd;
	int i, tag_size;

	nq->cmds = kzalloc(nq->queue_depth * sizeof(*cmd), GFP_KERNEL);
	if (!nq->cmds)
		return -ENOMEM;

	tag_size = ALIGN(nq->queue_depth, BITS_PER_LONG) / BITS_PER_LONG;
	nq->tag_map = kzalloc(tag_size * sizeof(unsigned long), GFP_KERNEL);
	if (!nq->tag_map) {
		kfree(nq->cmds);
		return -ENOMEM;
	}

	for (i = 0; i < nq->queue_depth; i++) {
		cmd = &nq->cmds[i];
		INIT_LIST_HEAD(&cmd->list);
		cmd->ll_list.next = NULL;
		cmd->tag = -1U;
	}

	return 0;
}

static void cleanup_queue(struct nullb_queue *nq)
{
	kfree(nq->tag_map);
	kfree(nq->cmds);
}

static void cleanup_queues(struct nullb *nullb)
{
	int i;

	for (i = 0; i < nullb->nr_queues; i++)
		cleanup_queue(&nullb->queues[i]);

	kfree(nullb->queues);
}

static int setup_queues(struct nullb *nullb)
{
	nullb->queues = kzalloc(submit_queues * sizeof(struct nullb_queue),
								GFP_KERNEL);
	if (!nullb->queues)
		return -ENOMEM;

	nullb->nr_queues = 0;
	nullb->queue_depth = hw_queue_depth;

	return 0;
}

static int init_driver_queues(struct nullb *nullb)
{
	struct nullb_queue *nq;
	int i, ret = 0;

	for (i = 0; i < submit_queues; i++) {
		nq = &nullb->queues[i];

		null_init_queue(nullb, nq);

		ret = setup_commands(nq);
		if (ret)
			goto err_queue;
		nullb->nr_queues++;
	}

	return 0;
err_queue:
	cleanup_queues(nullb);
	return ret;
}

static int null_add_dev(void)
{
	struct gendisk *disk;
	struct nullb *nullb;
	sector_t size;

	nullb = kzalloc_node(sizeof(*nullb), GFP_KERNEL, home_node);
	if (!nullb)
		return -ENOMEM;

	spin_lock_init(&nullb->lock);

	if (queue_mode == NULL_Q_MQ && use_per_node_hctx)
		submit_queues = nr_online_nodes;

	if (setup_queues(nullb))
		goto err;

	if (queue_mode == NULL_Q_MQ) {
		null_mq_reg.numa_node = home_node;
		null_mq_reg.queue_depth = hw_queue_depth;
		null_mq_reg.nr_hw_queues = submit_queues;

		if (use_per_node_hctx) {
			null_mq_reg.ops->alloc_hctx = null_alloc_hctx;
			null_mq_reg.ops->free_hctx = null_free_hctx;
		} else {
			null_mq_reg.ops->alloc_hctx = blk_mq_alloc_single_hw_queue;
			null_mq_reg.ops->free_hctx = blk_mq_free_single_hw_queue;
		}

		nullb->q = blk_mq_init_queue(&null_mq_reg, nullb);
	} else if (queue_mode == NULL_Q_BIO) {
		nullb->q = blk_alloc_queue_node(GFP_KERNEL, home_node);
		blk_queue_make_request(nullb->q, null_queue_bio);
		init_driver_queues(nullb);
	} else {
		nullb->q = blk_init_queue_node(null_request_fn, &nullb->lock, home_node);
		blk_queue_prep_rq(nullb->q, null_rq_prep_fn);
		if (nullb->q)
			blk_queue_softirq_done(nullb->q, null_softirq_done_fn);
		init_driver_queues(nullb);
	}

	if (!nullb->q)
		goto queue_fail;

	nullb->q->queuedata = nullb;
	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, nullb->q);

	disk = nullb->disk = alloc_disk_node(1, home_node);
	if (!disk) {
queue_fail:
		if (queue_mode == NULL_Q_MQ)
			blk_mq_free_queue(nullb->q);
		else
			blk_cleanup_queue(nullb->q);
		cleanup_queues(nullb);
err:
		kfree(nullb);
		return -ENOMEM;
	}

	mutex_lock(&lock);
	list_add_tail(&nullb->list, &nullb_list);
	nullb->index = nullb_indexes++;
	mutex_unlock(&lock);

	blk_queue_logical_block_size(nullb->q, bs);
	blk_queue_physical_block_size(nullb->q, bs);

	size = gb * 1024 * 1024 * 1024ULL;
	sector_div(size, bs);
	set_capacity(disk, size);

	disk->flags |= GENHD_FL_EXT_DEVT;
	disk->major		= null_major;
	disk->first_minor	= nullb->index;
	disk->fops		= &null_fops;
	disk->private_data	= nullb;
	disk->queue		= nullb->q;
	sprintf(disk->disk_name, "nullb%d", nullb->index);
	add_disk(disk);
	return 0;
}

static int __init null_init(void)
{
	unsigned int i;

#if !defined(CONFIG_SMP)
	if (irqmode == NULL_IRQ_SOFTIRQ) {
		pr_warn("null_blk: softirq completions not available.\n");
		pr_warn("null_blk: using direct completions.\n");
		irqmode = NULL_IRQ_NONE;
	}
#endif

	if (queue_mode == NULL_Q_MQ && use_per_node_hctx) {
		if (submit_queues < nr_online_nodes) {
			pr_warn("null_blk: submit_queues param is set to %u.",
							nr_online_nodes);
			submit_queues = nr_online_nodes;
		}
	} else if (submit_queues > nr_cpu_ids)
		submit_queues = nr_cpu_ids;
	else if (!submit_queues)
		submit_queues = 1;

	mutex_init(&lock);

	/* Initialize a separate list for each CPU for issuing softirqs */
	for_each_possible_cpu(i) {
		struct completion_queue *cq = &per_cpu(completion_queues, i);

		init_llist_head(&cq->list);

		if (irqmode != NULL_IRQ_TIMER)
			continue;

		hrtimer_init(&cq->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
		cq->timer.function = null_cmd_timer_expired;
	}

	null_major = register_blkdev(0, "nullb");
	if (null_major < 0)
		return null_major;

	for (i = 0; i < nr_devices; i++) {
		if (null_add_dev()) {
			unregister_blkdev(null_major, "nullb");
			return -EINVAL;
		}
	}

	pr_info("null: module loaded\n");
	return 0;
}

static void __exit null_exit(void)
{
	struct nullb *nullb;

	unregister_blkdev(null_major, "nullb");

	mutex_lock(&lock);
	while (!list_empty(&nullb_list)) {
		nullb = list_entry(nullb_list.next, struct nullb, list);
		null_del_dev(nullb);
	}
	mutex_unlock(&lock);
}

module_init(null_init);
module_exit(null_exit);

MODULE_AUTHOR("Jens Axboe <jaxboe@fusionio.com>");
MODULE_LICENSE("GPL");
Commit	Line	Data
f2298c04	1	#include <linux/module.h>
fc1bc354	2
f2298c04 JA	3	#include <linux/moduleparam.h>
	4	#include <linux/sched.h>
	5	#include <linux/fs.h>
	6	#include <linux/blkdev.h>
	7	#include <linux/init.h>
	8	#include <linux/slab.h>
	9	#include <linux/blk-mq.h>
	10	#include <linux/hrtimer.h>
	11
	12	struct nullb_cmd {
	13	struct list_head list;
	14	struct llist_node ll_list;
	15	struct call_single_data csd;
	16	struct request *rq;
	17	struct bio *bio;
	18	unsigned int tag;
	19	struct nullb_queue *nq;
	20	};
	21
	22	struct nullb_queue {
	23	unsigned long *tag_map;
	24	wait_queue_head_t wait;
	25	unsigned int queue_depth;
	26
	27	struct nullb_cmd *cmds;
	28	};
	29
	30	struct nullb {
	31	struct list_head list;
	32	unsigned int index;
	33	struct request_queue *q;
	34	struct gendisk *disk;
	35	struct hrtimer timer;
	36	unsigned int queue_depth;
	37	spinlock_t lock;
	38
	39	struct nullb_queue *queues;
	40	unsigned int nr_queues;
	41	};
	42
	43	static LIST_HEAD(nullb_list);
	44	static struct mutex lock;
	45	static int null_major;
	46	static int nullb_indexes;
	47
	48	struct completion_queue {
	49	struct llist_head list;
	50	struct hrtimer timer;
	51	};
	52
	53	/*
	54	* These are per-cpu for now, they will need to be configured by the
	55	* complete_queues parameter and appropriately mapped.
	56	*/
	57	static DEFINE_PER_CPU(struct completion_queue, completion_queues);
	58
	59	enum {
	60	NULL_IRQ_NONE = 0,
	61	NULL_IRQ_SOFTIRQ = 1,
	62	NULL_IRQ_TIMER = 2,
	63
	64	NULL_Q_BIO = 0,
	65	NULL_Q_RQ = 1,
	66	NULL_Q_MQ = 2,
67	};
68
2d263a78	69	static int submit_queues;
f2298c04 JA	70	module_param(submit_queues, int, S_IRUGO);
	71	MODULE_PARM_DESC(submit_queues, "Number of submission queues");
	72
	73	static int home_node = NUMA_NO_NODE;
	74	module_param(home_node, int, S_IRUGO);
	75	MODULE_PARM_DESC(home_node, "Home node for the device");
	76
	77	static int queue_mode = NULL_Q_MQ;
	78	module_param(queue_mode, int, S_IRUGO);
	79	MODULE_PARM_DESC(use_mq, "Use blk-mq interface (0=bio,1=rq,2=multiqueue)");
	80
	81	static int gb = 250;
	82	module_param(gb, int, S_IRUGO);
	83	MODULE_PARM_DESC(gb, "Size in GB");
	84
	85	static int bs = 512;
	86	module_param(bs, int, S_IRUGO);
	87	MODULE_PARM_DESC(bs, "Block size (in bytes)");
	88
	89	static int nr_devices = 2;
	90	module_param(nr_devices, int, S_IRUGO);
	91	MODULE_PARM_DESC(nr_devices, "Number of devices to register");
	92
	93	static int irqmode = NULL_IRQ_SOFTIRQ;
	94	module_param(irqmode, int, S_IRUGO);
	95	MODULE_PARM_DESC(irqmode, "IRQ completion handler. 0-none, 1-softirq, 2-timer");
	96
	97	static int completion_nsec = 10000;
	98	module_param(completion_nsec, int, S_IRUGO);
	99	MODULE_PARM_DESC(completion_nsec, "Time in ns to complete a request in hardware. Default: 10,000ns");
	100
	101	static int hw_queue_depth = 64;
	102	module_param(hw_queue_depth, int, S_IRUGO);
	103	MODULE_PARM_DESC(hw_queue_depth, "Queue depth for each hardware queue. Default: 64");
	104
20005244	105	static bool use_per_node_hctx = false;
f2298c04	106	module_param(use_per_node_hctx, bool, S_IRUGO);
20005244	107	MODULE_PARM_DESC(use_per_node_hctx, "Use per-node allocation for hardware context queues. Default: false");
f2298c04 JA	108
	109	static void put_tag(struct nullb_queue *nq, unsigned int tag)
	110	{
	111	clear_bit_unlock(tag, nq->tag_map);
	112
	113	if (waitqueue_active(&nq->wait))
	114	wake_up(&nq->wait);
	115	}
	116
	117	static unsigned int get_tag(struct nullb_queue *nq)
	118	{
	119	unsigned int tag;
	120
	121	do {
	122	tag = find_first_zero_bit(nq->tag_map, nq->queue_depth);
	123	if (tag >= nq->queue_depth)
	124	return -1U;
	125	} while (test_and_set_bit_lock(tag, nq->tag_map));
	126
	127	return tag;
	128	}
	129
	130	static void free_cmd(struct nullb_cmd *cmd)
	131	{
	132	put_tag(cmd->nq, cmd->tag);
	133	}
	134
	135	static struct nullb_cmd __alloc_cmd(struct nullb_queue nq)
	136	{
	137	struct nullb_cmd *cmd;
	138	unsigned int tag;
	139
	140	tag = get_tag(nq);
	141	if (tag != -1U) {
	142	cmd = &nq->cmds[tag];
	143	cmd->tag = tag;
	144	cmd->nq = nq;
	145	return cmd;
	146	}
	147
	148	return NULL;
	149	}
	150
	151	static struct nullb_cmd alloc_cmd(struct nullb_queue nq, int can_wait)
	152	{
	153	struct nullb_cmd *cmd;
	154	DEFINE_WAIT(wait);
	155
	156	cmd = __alloc_cmd(nq);
	157	if (cmd \|\| !can_wait)
	158	return cmd;
	159
	160	do {
	161	prepare_to_wait(&nq->wait, &wait, TASK_UNINTERRUPTIBLE);
	162	cmd = __alloc_cmd(nq);
	163	if (cmd)
	164	break;
	165
	166	io_schedule();
	167	} while (1);
	168
	169	finish_wait(&nq->wait, &wait);
	170	return cmd;
	171	}
172
173	static void end_cmd(struct nullb_cmd *cmd)
174	{
175	if (cmd->rq) {
176	if (queue_mode == NULL_Q_MQ)
177	blk_mq_end_io(cmd->rq, 0);
178	else {
179	INIT_LIST_HEAD(&cmd->rq->queuelist);
180	blk_end_request_all(cmd->rq, 0);
181	}
182	} else if (cmd->bio)
183	bio_endio(cmd->bio, 0);
184
185	if (queue_mode != NULL_Q_MQ)
186	free_cmd(cmd);
187	}
188
189	static enum hrtimer_restart null_cmd_timer_expired(struct hrtimer *timer)
190	{
191	struct completion_queue *cq;
192	struct llist_node *entry;
193	struct nullb_cmd *cmd;
194
195	cq = &per_cpu(completion_queues, smp_processor_id());
196
197	while ((entry = llist_del_all(&cq->list)) != NULL) {
198	do {
199	cmd = container_of(entry, struct nullb_cmd, ll_list);
200	end_cmd(cmd);
201	entry = entry->next;
202	} while (entry);
203	}
204
205	return HRTIMER_NORESTART;
206	}
207
208	static void null_cmd_end_timer(struct nullb_cmd *cmd)
209	{
210	struct completion_queue *cq = &per_cpu(completion_queues, get_cpu());
211
212	cmd->ll_list.next = NULL;
213	if (llist_add(&cmd->ll_list, &cq->list)) {
214	ktime_t kt = ktime_set(0, completion_nsec);
215
216	hrtimer_start(&cq->timer, kt, HRTIMER_MODE_REL);
217	}
218
219	put_cpu();
220	}
221
222	static void null_softirq_done_fn(struct request *rq)
223	{
224	blk_end_request_all(rq, 0);
225	}
226
044c8d4b	227	#ifdef CONFIG_SMP
f2298c04 JA	228
	229	static void null_ipi_cmd_end_io(void *data)
	230	{
	231	struct completion_queue *cq;
	232	struct llist_node entry, next;
	233	struct nullb_cmd *cmd;
	234
	235	cq = &per_cpu(completion_queues, smp_processor_id());
	236
	237	entry = llist_del_all(&cq->list);
	238
	239	while (entry) {
	240	next = entry->next;
	241	cmd = llist_entry(entry, struct nullb_cmd, ll_list);
	242	end_cmd(cmd);
	243	entry = next;
	244	}
	245	}
	246
	247	static void null_cmd_end_ipi(struct nullb_cmd *cmd)
	248	{
	249	struct call_single_data *data = &cmd->csd;
	250	int cpu = get_cpu();
	251	struct completion_queue *cq = &per_cpu(completion_queues, cpu);
	252
	253	cmd->ll_list.next = NULL;
	254
	255	if (llist_add(&cmd->ll_list, &cq->list)) {
	256	data->func = null_ipi_cmd_end_io;
	257	data->flags = 0;
	258	__smp_call_function_single(cpu, data, 0);
	259	}
	260
	261	put_cpu();
	262	}
	263
044c8d4b	264	#endif /* CONFIG_SMP */
f2298c04 JA	265
	266	static inline void null_handle_cmd(struct nullb_cmd *cmd)
	267	{
	268	/* Complete IO by inline, softirq or timer */
	269	switch (irqmode) {
	270	case NULL_IRQ_NONE:
	271	end_cmd(cmd);
	272	break;
	273	case NULL_IRQ_SOFTIRQ:
044c8d4b	274	#ifdef CONFIG_SMP
f2298c04 JA	275	null_cmd_end_ipi(cmd);
	276	#else
	277	end_cmd(cmd);
	278	#endif
	279	break;
	280	case NULL_IRQ_TIMER:
	281	null_cmd_end_timer(cmd);
	282	break;
	283	}
	284	}
	285
	286	static struct nullb_queue nullb_to_queue(struct nullb nullb)
	287	{
	288	int index = 0;
	289
	290	if (nullb->nr_queues != 1)
	291	index = raw_smp_processor_id() / ((nr_cpu_ids + nullb->nr_queues - 1) / nullb->nr_queues);
	292
	293	return &nullb->queues[index];
	294	}
	295
	296	static void null_queue_bio(struct request_queue q, struct bio bio)
	297	{
	298	struct nullb *nullb = q->queuedata;
	299	struct nullb_queue *nq = nullb_to_queue(nullb);
	300	struct nullb_cmd *cmd;
	301
	302	cmd = alloc_cmd(nq, 1);
	303	cmd->bio = bio;
	304
	305	null_handle_cmd(cmd);
	306	}
	307
	308	static int null_rq_prep_fn(struct request_queue q, struct request req)
	309	{
	310	struct nullb *nullb = q->queuedata;
	311	struct nullb_queue *nq = nullb_to_queue(nullb);
	312	struct nullb_cmd *cmd;
	313
	314	cmd = alloc_cmd(nq, 0);
	315	if (cmd) {
	316	cmd->rq = req;
	317	req->special = cmd;
	318	return BLKPREP_OK;
	319	}
	320
	321	return BLKPREP_DEFER;
	322	}
	323
	324	static void null_request_fn(struct request_queue *q)
	325	{
	326	struct request *rq;
	327
	328	while ((rq = blk_fetch_request(q)) != NULL) {
	329	struct nullb_cmd *cmd = rq->special;
	330
	331	spin_unlock_irq(q->queue_lock);
	332	null_handle_cmd(cmd);
	333	spin_lock_irq(q->queue_lock);
	334	}
	335	}
	336
	337	static int null_queue_rq(struct blk_mq_hw_ctx hctx, struct request rq)
	338	{
339	struct nullb_cmd *cmd = rq->special;
340
341	cmd->rq = rq;
342	cmd->nq = hctx->driver_data;
343
344	null_handle_cmd(cmd);
345	return BLK_MQ_RQ_QUEUE_OK;
346	}
347
348	static struct blk_mq_hw_ctx null_alloc_hctx(struct blk_mq_reg reg, unsigned int hctx_index)
349	{
fc1bc354 MB	350	int b_size = DIV_ROUND_UP(reg->nr_hw_queues, nr_online_nodes);
	351	int tip = (reg->nr_hw_queues % nr_online_nodes);
	352	int node = 0, i, n;
	353
	354	/*
	355	* Split submit queues evenly wrt to the number of nodes. If uneven,
	356	* fill the first buckets with one extra, until the rest is filled with
	357	* no extra.
	358	*/
	359	for (i = 0, n = 1; i < hctx_index; i++, n++) {
	360	if (n % b_size == 0) {
	361	n = 0;
	362	node++;
	363
	364	tip--;
	365	if (!tip)
	366	b_size = reg->nr_hw_queues / nr_online_nodes;
	367	}
	368	}
	369
	370	/*
	371	* A node might not be online, therefore map the relative node id to the
	372	* real node id.
	373	*/
	374	for_each_online_node(n) {
	375	if (!node)
	376	break;
	377	node--;
	378	}
	379
	380	return kzalloc_node(sizeof(struct blk_mq_hw_ctx), GFP_KERNEL, n);
f2298c04 JA	381	}
	382
	383	static void null_free_hctx(struct blk_mq_hw_ctx *hctx, unsigned int hctx_index)
	384	{
	385	kfree(hctx);
	386	}
	387
2d263a78 MB	388	static void null_init_queue(struct nullb nullb, struct nullb_queue nq)
	389	{
	390	BUG_ON(!nullb);
	391	BUG_ON(!nq);
	392
	393	init_waitqueue_head(&nq->wait);
	394	nq->queue_depth = nullb->queue_depth;
	395	}
	396
f2298c04 JA	397	static int null_init_hctx(struct blk_mq_hw_ctx hctx, void data,
	398	unsigned int index)
	399	{
	400	struct nullb *nullb = data;
	401	struct nullb_queue *nq = &nullb->queues[index];
	402
f2298c04	403	hctx->driver_data = nq;
2d263a78 MB	404	null_init_queue(nullb, nq);
2d263a78 MB	405	nullb->nr_queues++;
f2298c04 JA	406
	407	return 0;
	408	}
	409
	410	static struct blk_mq_ops null_mq_ops = {
	411	.queue_rq = null_queue_rq,
	412	.map_queue = blk_mq_map_queue,
	413	.init_hctx = null_init_hctx,
	414	};
	415
	416	static struct blk_mq_reg null_mq_reg = {
	417	.ops = &null_mq_ops,
	418	.queue_depth = 64,
	419	.cmd_size = sizeof(struct nullb_cmd),
	420	.flags = BLK_MQ_F_SHOULD_MERGE,
	421	};
	422
	423	static void null_del_dev(struct nullb *nullb)
	424	{
	425	list_del_init(&nullb->list);
	426
	427	del_gendisk(nullb->disk);
	428	if (queue_mode == NULL_Q_MQ)
	429	blk_mq_free_queue(nullb->q);
	430	else
	431	blk_cleanup_queue(nullb->q);
	432	put_disk(nullb->disk);
	433	kfree(nullb);
	434	}
	435
	436	static int null_open(struct block_device *bdev, fmode_t mode)
	437	{
	438	return 0;
	439	}
	440
	441	static void null_release(struct gendisk *disk, fmode_t mode)
	442	{
	443	}
	444
	445	static const struct block_device_operations null_fops = {
	446	.owner = THIS_MODULE,
	447	.open = null_open,
	448	.release = null_release,
	449	};
	450
	451	static int setup_commands(struct nullb_queue *nq)
	452	{
	453	struct nullb_cmd *cmd;
	454	int i, tag_size;
	455
	456	nq->cmds = kzalloc(nq->queue_depth * sizeof(*cmd), GFP_KERNEL);
	457	if (!nq->cmds)
2d263a78	458	return -ENOMEM;
f2298c04 JA	459
	460	tag_size = ALIGN(nq->queue_depth, BITS_PER_LONG) / BITS_PER_LONG;
	461	nq->tag_map = kzalloc(tag_size * sizeof(unsigned long), GFP_KERNEL);
	462	if (!nq->tag_map) {
	463	kfree(nq->cmds);
2d263a78	464	return -ENOMEM;
f2298c04 JA	465	}
	466
	467	for (i = 0; i < nq->queue_depth; i++) {
	468	cmd = &nq->cmds[i];
	469	INIT_LIST_HEAD(&cmd->list);
	470	cmd->ll_list.next = NULL;
	471	cmd->tag = -1U;
	472	}
	473
	474	return 0;
	475	}
	476
	477	static void cleanup_queue(struct nullb_queue *nq)
	478	{
	479	kfree(nq->tag_map);
	480	kfree(nq->cmds);
	481	}
	482
	483	static void cleanup_queues(struct nullb *nullb)
	484	{
	485	int i;
	486
	487	for (i = 0; i < nullb->nr_queues; i++)
	488	cleanup_queue(&nullb->queues[i]);
	489
	490	kfree(nullb->queues);
	491	}
	492
	493	static int setup_queues(struct nullb *nullb)
	494	{
2d263a78 MB	495	nullb->queues = kzalloc(submit_queues * sizeof(struct nullb_queue),
2d263a78 MB	496	GFP_KERNEL);
f2298c04	497	if (!nullb->queues)
2d263a78	498	return -ENOMEM;
f2298c04 JA	499
	500	nullb->nr_queues = 0;
	501	nullb->queue_depth = hw_queue_depth;
	502
2d263a78 MB	503	return 0;
	504	}
	505
	506	static int init_driver_queues(struct nullb *nullb)
	507	{
	508	struct nullb_queue *nq;
	509	int i, ret = 0;
f2298c04 JA	510
	511	for (i = 0; i < submit_queues; i++) {
	512	nq = &nullb->queues[i];
2d263a78 MB	513
	514	null_init_queue(nullb, nq);
	515
	516	ret = setup_commands(nq);
	517	if (ret)
	518	goto err_queue;
f2298c04 JA	519	nullb->nr_queues++;
	520	}
	521
2d263a78 MB	522	return 0;
2d263a78 MB	523	err_queue:
f2298c04	524	cleanup_queues(nullb);
2d263a78	525	return ret;
f2298c04 JA	526	}
	527
	528	static int null_add_dev(void)
	529	{
	530	struct gendisk *disk;
	531	struct nullb *nullb;
	532	sector_t size;
	533
	534	nullb = kzalloc_node(sizeof(*nullb), GFP_KERNEL, home_node);
	535	if (!nullb)
	536	return -ENOMEM;
	537
	538	spin_lock_init(&nullb->lock);
	539
57053d8c MB	540	if (queue_mode == NULL_Q_MQ && use_per_node_hctx)
	541	submit_queues = nr_online_nodes;
	542
f2298c04 JA	543	if (setup_queues(nullb))
	544	goto err;
	545
	546	if (queue_mode == NULL_Q_MQ) {
	547	null_mq_reg.numa_node = home_node;
	548	null_mq_reg.queue_depth = hw_queue_depth;
57053d8c	549	null_mq_reg.nr_hw_queues = submit_queues;
f2298c04 JA	550
	551	if (use_per_node_hctx) {
	552	null_mq_reg.ops->alloc_hctx = null_alloc_hctx;
	553	null_mq_reg.ops->free_hctx = null_free_hctx;
f2298c04 JA	554	} else {
	555	null_mq_reg.ops->alloc_hctx = blk_mq_alloc_single_hw_queue;
	556	null_mq_reg.ops->free_hctx = blk_mq_free_single_hw_queue;
f2298c04 JA	557	}
	558
	559	nullb->q = blk_mq_init_queue(&null_mq_reg, nullb);
	560	} else if (queue_mode == NULL_Q_BIO) {
	561	nullb->q = blk_alloc_queue_node(GFP_KERNEL, home_node);
	562	blk_queue_make_request(nullb->q, null_queue_bio);
2d263a78	563	init_driver_queues(nullb);
f2298c04 JA	564	} else {
	565	nullb->q = blk_init_queue_node(null_request_fn, &nullb->lock, home_node);
	566	blk_queue_prep_rq(nullb->q, null_rq_prep_fn);
	567	if (nullb->q)
	568	blk_queue_softirq_done(nullb->q, null_softirq_done_fn);
2d263a78	569	init_driver_queues(nullb);
f2298c04 JA	570	}
	571
	572	if (!nullb->q)
	573	goto queue_fail;
	574
	575	nullb->q->queuedata = nullb;
	576	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, nullb->q);
	577
	578	disk = nullb->disk = alloc_disk_node(1, home_node);
	579	if (!disk) {
	580	queue_fail:
	581	if (queue_mode == NULL_Q_MQ)
	582	blk_mq_free_queue(nullb->q);
	583	else
	584	blk_cleanup_queue(nullb->q);
	585	cleanup_queues(nullb);
	586	err:
	587	kfree(nullb);
	588	return -ENOMEM;
	589	}
	590
	591	mutex_lock(&lock);
	592	list_add_tail(&nullb->list, &nullb_list);
	593	nullb->index = nullb_indexes++;
	594	mutex_unlock(&lock);
	595
	596	blk_queue_logical_block_size(nullb->q, bs);
	597	blk_queue_physical_block_size(nullb->q, bs);
	598
	599	size = gb * 1024 * 1024 * 1024ULL;
	600	sector_div(size, bs);
	601	set_capacity(disk, size);
	602
	603	disk->flags \|= GENHD_FL_EXT_DEVT;
	604	disk->major = null_major;
	605	disk->first_minor = nullb->index;
	606	disk->fops = &null_fops;
	607	disk->private_data = nullb;
	608	disk->queue = nullb->q;
	609	sprintf(disk->disk_name, "nullb%d", nullb->index);
	610	add_disk(disk);
	611	return 0;
	612	}
	613
	614	static int __init null_init(void)
	615	{
	616	unsigned int i;
	617
044c8d4b	618	#if !defined(CONFIG_SMP)
f2298c04 JA	619	if (irqmode == NULL_IRQ_SOFTIRQ) {
	620	pr_warn("null_blk: softirq completions not available.\n");
	621	pr_warn("null_blk: using direct completions.\n");
	622	irqmode = NULL_IRQ_NONE;
	623	}
	624	#endif
	625
d15ee6b1	626	if (queue_mode == NULL_Q_MQ && use_per_node_hctx) {
fc1bc354	627	if (submit_queues < nr_online_nodes) {
d15ee6b1 MB	628	pr_warn("null_blk: submit_queues param is set to %u.",
d15ee6b1 MB	629	nr_online_nodes);
fc1bc354 MB	630	submit_queues = nr_online_nodes;
fc1bc354 MB	631	}
d15ee6b1	632	} else if (submit_queues > nr_cpu_ids)
f2298c04 JA	633	submit_queues = nr_cpu_ids;
	634	else if (!submit_queues)
	635	submit_queues = 1;
	636
	637	mutex_init(&lock);
	638
	639	/* Initialize a separate list for each CPU for issuing softirqs */
	640	for_each_possible_cpu(i) {
	641	struct completion_queue *cq = &per_cpu(completion_queues, i);
	642
	643	init_llist_head(&cq->list);
	644
	645	if (irqmode != NULL_IRQ_TIMER)
	646	continue;
	647
	648	hrtimer_init(&cq->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
	649	cq->timer.function = null_cmd_timer_expired;
	650	}
	651
	652	null_major = register_blkdev(0, "nullb");
	653	if (null_major < 0)
	654	return null_major;
	655
	656	for (i = 0; i < nr_devices; i++) {
	657	if (null_add_dev()) {
	658	unregister_blkdev(null_major, "nullb");
	659	return -EINVAL;
	660	}
	661	}
	662
	663	pr_info("null: module loaded\n");
	664	return 0;
	665	}
	666
	667	static void __exit null_exit(void)
	668	{
	669	struct nullb *nullb;
	670
	671	unregister_blkdev(null_major, "nullb");
	672
	673	mutex_lock(&lock);
	674	while (!list_empty(&nullb_list)) {
	675	nullb = list_entry(nullb_list.next, struct nullb, list);
	676	null_del_dev(nullb);
	677	}
	678	mutex_unlock(&lock);
	679	}
	680
	681	module_init(null_init);
	682	module_exit(null_exit);
	683
	684	MODULE_AUTHOR("Jens Axboe <jaxboe@fusionio.com>");
	685	MODULE_LICENSE("GPL");