null_blk.c 18.9 KB
Newer Older
1
#include <linux/module.h>
2

3 4 5 6 7 8 9 10
#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>
11
#include <linux/lightnvm.h>
12 13 14 15 16 17 18 19 20

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;
21
	struct hrtimer timer;
22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
};

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;
37
	struct nvm_dev *ndev;
38
	struct blk_mq_tag_set tag_set;
39 40 41 42 43 44
	struct hrtimer timer;
	unsigned int queue_depth;
	spinlock_t lock;

	struct nullb_queue *queues;
	unsigned int nr_queues;
45
	char disk_name[DISK_NAME_LEN];
46 47 48 49 50 51
};

static LIST_HEAD(nullb_list);
static struct mutex lock;
static int null_major;
static int nullb_indexes;
M
Matias Bjørling 已提交
52
static struct kmem_cache *ppa_cache;
53 54 55 56 57

enum {
	NULL_IRQ_NONE		= 0,
	NULL_IRQ_SOFTIRQ	= 1,
	NULL_IRQ_TIMER		= 2,
58
};
59

60
enum {
61 62 63 64 65
	NULL_Q_BIO		= 0,
	NULL_Q_RQ		= 1,
	NULL_Q_MQ		= 2,
};

66
static int submit_queues;
67 68 69 70 71 72 73 74
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;
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95

static int null_param_store_val(const char *str, int *val, int min, int max)
{
	int ret, new_val;

	ret = kstrtoint(str, 10, &new_val);
	if (ret)
		return -EINVAL;

	if (new_val < min || new_val > max)
		return -EINVAL;

	*val = new_val;
	return 0;
}

static int null_set_queue_mode(const char *str, const struct kernel_param *kp)
{
	return null_param_store_val(str, &queue_mode, NULL_Q_BIO, NULL_Q_MQ);
}

96
static const struct kernel_param_ops null_queue_mode_param_ops = {
97 98 99 100 101
	.set	= null_set_queue_mode,
	.get	= param_get_int,
};

device_param_cb(queue_mode, &null_queue_mode_param_ops, &queue_mode, S_IRUGO);
102
MODULE_PARM_DESC(queue_mode, "Block interface to use (0=bio,1=rq,2=multiqueue)");
103 104 105 106 107 108 109 110 111 112 113 114 115

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");

116 117 118 119
static bool use_lightnvm;
module_param(use_lightnvm, bool, S_IRUGO);
MODULE_PARM_DESC(use_lightnvm, "Register as a LightNVM device");

120
static int irqmode = NULL_IRQ_SOFTIRQ;
121 122 123 124 125 126 127

static int null_set_irqmode(const char *str, const struct kernel_param *kp)
{
	return null_param_store_val(str, &irqmode, NULL_IRQ_NONE,
					NULL_IRQ_TIMER);
}

128
static const struct kernel_param_ops null_irqmode_param_ops = {
129 130 131 132 133
	.set	= null_set_irqmode,
	.get	= param_get_int,
};

device_param_cb(irqmode, &null_irqmode_param_ops, &irqmode, S_IRUGO);
134 135
MODULE_PARM_DESC(irqmode, "IRQ completion handler. 0-none, 1-softirq, 2-timer");

136 137
static unsigned long completion_nsec = 10000;
module_param(completion_nsec, ulong, S_IRUGO);
138 139 140 141 142 143
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");

144
static bool use_per_node_hctx = false;
145
module_param(use_per_node_hctx, bool, S_IRUGO);
146
MODULE_PARM_DESC(use_per_node_hctx, "Use per-node allocation for hardware context queues. Default: false");
147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173

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);
}

174 175
static enum hrtimer_restart null_cmd_timer_expired(struct hrtimer *timer);

176 177 178 179 180 181 182 183 184 185
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;
186 187 188 189 190
		if (irqmode == NULL_IRQ_TIMER) {
			hrtimer_init(&cmd->timer, CLOCK_MONOTONIC,
				     HRTIMER_MODE_REL);
			cmd->timer.function = null_cmd_timer_expired;
		}
191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220
		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)
{
221 222
	struct request_queue *q = NULL;

M
Mike Krinkin 已提交
223 224 225
	if (cmd->rq)
		q = cmd->rq->q;

226 227
	switch (queue_mode)  {
	case NULL_Q_MQ:
228
		blk_mq_end_request(cmd->rq, 0);
229 230 231 232 233 234
		return;
	case NULL_Q_RQ:
		INIT_LIST_HEAD(&cmd->rq->queuelist);
		blk_end_request_all(cmd->rq, 0);
		break;
	case NULL_Q_BIO:
235
		bio_endio(cmd->bio);
236
		break;
237
	}
238

239 240
	free_cmd(cmd);

241
	/* Restart queue if needed, as we are freeing a tag */
242
	if (queue_mode == NULL_Q_RQ && blk_queue_stopped(q)) {
243 244 245
		unsigned long flags;

		spin_lock_irqsave(q->queue_lock, flags);
246
		blk_start_queue_async(q);
247
		spin_unlock_irqrestore(q->queue_lock, flags);
248
	}
249 250 251 252 253
}

static enum hrtimer_restart null_cmd_timer_expired(struct hrtimer *timer)
{
	end_cmd(container_of(timer, struct nullb_cmd, timer));
254 255 256 257 258 259

	return HRTIMER_NORESTART;
}

static void null_cmd_end_timer(struct nullb_cmd *cmd)
{
T
Thomas Gleixner 已提交
260
	ktime_t kt = completion_nsec;
261

262
	hrtimer_start(&cmd->timer, kt, HRTIMER_MODE_REL);
263 264 265 266
}

static void null_softirq_done_fn(struct request *rq)
{
267 268 269 270
	if (queue_mode == NULL_Q_MQ)
		end_cmd(blk_mq_rq_to_pdu(rq));
	else
		end_cmd(rq->special);
271 272 273 274 275 276 277
}

static inline void null_handle_cmd(struct nullb_cmd *cmd)
{
	/* Complete IO by inline, softirq or timer */
	switch (irqmode) {
	case NULL_IRQ_SOFTIRQ:
278 279
		switch (queue_mode)  {
		case NULL_Q_MQ:
280
			blk_mq_complete_request(cmd->rq, cmd->rq->errors);
281 282 283 284 285 286 287 288 289 290 291 292 293
			break;
		case NULL_Q_RQ:
			blk_complete_request(cmd->rq);
			break;
		case NULL_Q_BIO:
			/*
			 * XXX: no proper submitting cpu information available.
			 */
			end_cmd(cmd);
			break;
		}
		break;
	case NULL_IRQ_NONE:
294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311
		end_cmd(cmd);
		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];
}

312
static blk_qc_t null_queue_bio(struct request_queue *q, struct bio *bio)
313 314 315 316 317 318 319 320 321
{
	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);
322
	return BLK_QC_T_NONE;
323 324 325 326 327 328 329 330 331 332 333 334 335 336
}

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;
	}
337
	blk_stop_queue(q);
338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354

	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);
	}
}

355 356
static int null_queue_rq(struct blk_mq_hw_ctx *hctx,
			 const struct blk_mq_queue_data *bd)
357
{
358
	struct nullb_cmd *cmd = blk_mq_rq_to_pdu(bd->rq);
359

360 361 362 363
	if (irqmode == NULL_IRQ_TIMER) {
		hrtimer_init(&cmd->timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
		cmd->timer.function = null_cmd_timer_expired;
	}
364
	cmd->rq = bd->rq;
365 366
	cmd->nq = hctx->driver_data;

367
	blk_mq_start_request(bd->rq);
368

369 370 371 372
	null_handle_cmd(cmd);
	return BLK_MQ_RQ_QUEUE_OK;
}

373 374 375 376 377 378 379 380 381
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;
}

382 383 384 385 386 387 388
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;
389 390
	null_init_queue(nullb, nq);
	nullb->nr_queues++;
391 392 393 394 395 396 397

	return 0;
}

static struct blk_mq_ops null_mq_ops = {
	.queue_rq       = null_queue_rq,
	.init_hctx	= null_init_hctx,
398
	.complete	= null_softirq_done_fn,
399 400
};

401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416
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);
}

417 418 419 420 421 422
#ifdef CONFIG_NVM

static void null_lnvm_end_io(struct request *rq, int error)
{
	struct nvm_rq *rqd = rq->end_io_data;

423
	nvm_end_io(rqd, error);
424 425 426 427

	blk_put_request(rq);
}

428
static int null_lnvm_submit_io(struct nvm_dev *dev, struct nvm_rq *rqd)
429
{
430
	struct request_queue *q = dev->q;
431 432 433
	struct request *rq;
	struct bio *bio = rqd->bio;

434
	rq = blk_mq_alloc_request(q, bio_data_dir(bio), 0);
435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454
	if (IS_ERR(rq))
		return -ENOMEM;

	rq->cmd_type = REQ_TYPE_DRV_PRIV;
	rq->__sector = bio->bi_iter.bi_sector;
	rq->ioprio = bio_prio(bio);

	if (bio_has_data(bio))
		rq->nr_phys_segments = bio_phys_segments(q, bio);

	rq->__data_len = bio->bi_iter.bi_size;
	rq->bio = rq->biotail = bio;

	rq->end_io_data = rqd;

	blk_execute_rq_nowait(q, NULL, rq, 0, null_lnvm_end_io);

	return 0;
}

455
static int null_lnvm_id(struct nvm_dev *dev, struct nvm_id *id)
456 457
{
	sector_t size = gb * 1024 * 1024 * 1024ULL;
458
	sector_t blksize;
459 460 461 462 463
	struct nvm_id_group *grp;

	id->ver_id = 0x1;
	id->vmnt = 0;
	id->cgrps = 1;
464
	id->cap = 0x2;
465
	id->dom = 0x1;
466 467 468 469 470 471 472 473 474 475 476 477 478

	id->ppaf.blk_offset = 0;
	id->ppaf.blk_len = 16;
	id->ppaf.pg_offset = 16;
	id->ppaf.pg_len = 16;
	id->ppaf.sect_offset = 32;
	id->ppaf.sect_len = 8;
	id->ppaf.pln_offset = 40;
	id->ppaf.pln_len = 8;
	id->ppaf.lun_offset = 48;
	id->ppaf.lun_len = 8;
	id->ppaf.ch_offset = 56;
	id->ppaf.ch_len = 8;
479

480 481
	sector_div(size, bs); /* convert size to pages */
	size >>= 8; /* concert size to pgs pr blk */
482 483
	grp = &id->groups[0];
	grp->mtype = 0;
484
	grp->fmtype = 0;
485 486
	grp->num_ch = 1;
	grp->num_pg = 256;
487
	blksize = size;
488
	size >>= 16;
489
	grp->num_lun = size + 1;
490
	sector_div(blksize, grp->num_lun);
491 492 493
	grp->num_blk = blksize;
	grp->num_pln = 1;

494 495 496 497 498 499 500 501 502 503 504 505 506 507
	grp->fpg_sz = bs;
	grp->csecs = bs;
	grp->trdt = 25000;
	grp->trdm = 25000;
	grp->tprt = 500000;
	grp->tprm = 500000;
	grp->tbet = 1500000;
	grp->tbem = 1500000;
	grp->mpos = 0x010101; /* single plane rwe */
	grp->cpar = hw_queue_depth;

	return 0;
}

508
static void *null_lnvm_create_dma_pool(struct nvm_dev *dev, char *name)
509 510 511
{
	mempool_t *virtmem_pool;

M
Matias Bjørling 已提交
512
	virtmem_pool = mempool_create_slab_pool(64, ppa_cache);
513 514 515 516 517 518 519 520 521 522 523 524 525
	if (!virtmem_pool) {
		pr_err("null_blk: Unable to create virtual memory pool\n");
		return NULL;
	}

	return virtmem_pool;
}

static void null_lnvm_destroy_dma_pool(void *pool)
{
	mempool_destroy(pool);
}

526
static void *null_lnvm_dev_dma_alloc(struct nvm_dev *dev, void *pool,
527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549
				gfp_t mem_flags, dma_addr_t *dma_handler)
{
	return mempool_alloc(pool, mem_flags);
}

static void null_lnvm_dev_dma_free(void *pool, void *entry,
							dma_addr_t dma_handler)
{
	mempool_free(entry, pool);
}

static struct nvm_dev_ops null_lnvm_dev_ops = {
	.identity		= null_lnvm_id,
	.submit_io		= null_lnvm_submit_io,

	.create_dma_pool	= null_lnvm_create_dma_pool,
	.destroy_dma_pool	= null_lnvm_destroy_dma_pool,
	.dev_dma_alloc		= null_lnvm_dev_dma_alloc,
	.dev_dma_free		= null_lnvm_dev_dma_free,

	/* Simulate nvme protocol restriction */
	.max_phys_sect		= 64,
};
550 551 552

static int null_nvm_register(struct nullb *nullb)
{
553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570
	struct nvm_dev *dev;
	int rv;

	dev = nvm_alloc_dev(0);
	if (!dev)
		return -ENOMEM;

	dev->q = nullb->q;
	memcpy(dev->name, nullb->disk_name, DISK_NAME_LEN);
	dev->ops = &null_lnvm_dev_ops;

	rv = nvm_register(dev);
	if (rv) {
		kfree(dev);
		return rv;
	}
	nullb->ndev = dev;
	return 0;
571 572 573 574
}

static void null_nvm_unregister(struct nullb *nullb)
{
575
	nvm_unregister(nullb->ndev);
576
}
577
#else
578 579
static int null_nvm_register(struct nullb *nullb)
{
580
	pr_err("null_blk: CONFIG_NVM needs to be enabled for LightNVM\n");
581 582 583
	return -EINVAL;
}
static void null_nvm_unregister(struct nullb *nullb) {}
584 585
#endif /* CONFIG_NVM */

586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602
static void null_del_dev(struct nullb *nullb)
{
	list_del_init(&nullb->list);

	if (use_lightnvm)
		null_nvm_unregister(nullb);
	else
		del_gendisk(nullb->disk);
	blk_cleanup_queue(nullb->q);
	if (queue_mode == NULL_Q_MQ)
		blk_mq_free_tag_set(&nullb->tag_set);
	if (!use_lightnvm)
		put_disk(nullb->disk);
	cleanup_queues(nullb);
	kfree(nullb);
}

603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624
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)
625
		return -ENOMEM;
626 627 628 629 630

	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);
631
		return -ENOMEM;
632 633 634 635 636 637 638 639 640 641 642 643 644 645
	}

	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 int setup_queues(struct nullb *nullb)
{
646 647
	nullb->queues = kzalloc(submit_queues * sizeof(struct nullb_queue),
								GFP_KERNEL);
648
	if (!nullb->queues)
649
		return -ENOMEM;
650 651 652 653

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

654 655 656 657 658 659 660
	return 0;
}

static int init_driver_queues(struct nullb *nullb)
{
	struct nullb_queue *nq;
	int i, ret = 0;
661 662 663

	for (i = 0; i < submit_queues; i++) {
		nq = &nullb->queues[i];
664 665 666 667 668

		null_init_queue(nullb, nq);

		ret = setup_commands(nq);
		if (ret)
669
			return ret;
670 671
		nullb->nr_queues++;
	}
672
	return 0;
673 674
}

675
static int null_gendisk_register(struct nullb *nullb)
676 677 678
{
	struct gendisk *disk;
	sector_t size;
679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700

	disk = nullb->disk = alloc_disk_node(1, home_node);
	if (!disk)
		return -ENOMEM;
	size = gb * 1024 * 1024 * 1024ULL;
	set_capacity(disk, size >> 9);

	disk->flags |= GENHD_FL_EXT_DEVT | GENHD_FL_SUPPRESS_PARTITION_INFO;
	disk->major		= null_major;
	disk->first_minor	= nullb->index;
	disk->fops		= &null_fops;
	disk->private_data	= nullb;
	disk->queue		= nullb->q;
	strncpy(disk->disk_name, nullb->disk_name, DISK_NAME_LEN);

	add_disk(disk);
	return 0;
}

static int null_add_dev(void)
{
	struct nullb *nullb;
701
	int rv;
702 703

	nullb = kzalloc_node(sizeof(*nullb), GFP_KERNEL, home_node);
704 705
	if (!nullb) {
		rv = -ENOMEM;
706
		goto out;
707
	}
708 709 710

	spin_lock_init(&nullb->lock);

711 712 713
	if (queue_mode == NULL_Q_MQ && use_per_node_hctx)
		submit_queues = nr_online_nodes;

714 715
	rv = setup_queues(nullb);
	if (rv)
716
		goto out_free_nullb;
717 718

	if (queue_mode == NULL_Q_MQ) {
719
		nullb->tag_set.ops = &null_mq_ops;
720 721 722 723 724 725 726
		nullb->tag_set.nr_hw_queues = submit_queues;
		nullb->tag_set.queue_depth = hw_queue_depth;
		nullb->tag_set.numa_node = home_node;
		nullb->tag_set.cmd_size	= sizeof(struct nullb_cmd);
		nullb->tag_set.flags = BLK_MQ_F_SHOULD_MERGE;
		nullb->tag_set.driver_data = nullb;

727 728
		rv = blk_mq_alloc_tag_set(&nullb->tag_set);
		if (rv)
729 730 731
			goto out_cleanup_queues;

		nullb->q = blk_mq_init_queue(&nullb->tag_set);
732
		if (IS_ERR(nullb->q)) {
733
			rv = -ENOMEM;
734
			goto out_cleanup_tags;
735
		}
736 737
	} else if (queue_mode == NULL_Q_BIO) {
		nullb->q = blk_alloc_queue_node(GFP_KERNEL, home_node);
738 739
		if (!nullb->q) {
			rv = -ENOMEM;
740
			goto out_cleanup_queues;
741
		}
742
		blk_queue_make_request(nullb->q, null_queue_bio);
743 744 745
		rv = init_driver_queues(nullb);
		if (rv)
			goto out_cleanup_blk_queue;
746 747
	} else {
		nullb->q = blk_init_queue_node(null_request_fn, &nullb->lock, home_node);
748 749
		if (!nullb->q) {
			rv = -ENOMEM;
750
			goto out_cleanup_queues;
751
		}
752
		blk_queue_prep_rq(nullb->q, null_rq_prep_fn);
753
		blk_queue_softirq_done(nullb->q, null_softirq_done_fn);
754 755 756
		rv = init_driver_queues(nullb);
		if (rv)
			goto out_cleanup_blk_queue;
757 758 759 760
	}

	nullb->q->queuedata = nullb;
	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, nullb->q);
761
	queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, nullb->q);
762 763 764 765 766 767 768 769

	mutex_lock(&lock);
	nullb->index = nullb_indexes++;
	mutex_unlock(&lock);

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

770 771
	sprintf(nullb->disk_name, "nullb%d", nullb->index);

772 773 774 775
	if (use_lightnvm)
		rv = null_nvm_register(nullb);
	else
		rv = null_gendisk_register(nullb);
776

777 778
	if (rv)
		goto out_cleanup_blk_queue;
779 780 781 782

	mutex_lock(&lock);
	list_add_tail(&nullb->list, &nullb_list);
	mutex_unlock(&lock);
783

784
	return 0;
785 786 787 788 789 790 791 792 793 794
out_cleanup_blk_queue:
	blk_cleanup_queue(nullb->q);
out_cleanup_tags:
	if (queue_mode == NULL_Q_MQ)
		blk_mq_free_tag_set(&nullb->tag_set);
out_cleanup_queues:
	cleanup_queues(nullb);
out_free_nullb:
	kfree(nullb);
out:
795
	return rv;
796 797 798 799
}

static int __init null_init(void)
{
800
	int ret = 0;
801
	unsigned int i;
802
	struct nullb *nullb;
803

804 805 806 807 808
	if (bs > PAGE_SIZE) {
		pr_warn("null_blk: invalid block size\n");
		pr_warn("null_blk: defaults block size to %lu\n", PAGE_SIZE);
		bs = PAGE_SIZE;
	}
809

M
Matias Bjørling 已提交
810 811 812 813 814 815
	if (use_lightnvm && bs != 4096) {
		pr_warn("null_blk: LightNVM only supports 4k block size\n");
		pr_warn("null_blk: defaults block size to 4k\n");
		bs = 4096;
	}

816 817 818 819 820 821
	if (use_lightnvm && queue_mode != NULL_Q_MQ) {
		pr_warn("null_blk: LightNVM only supported for blk-mq\n");
		pr_warn("null_blk: defaults queue mode to blk-mq\n");
		queue_mode = NULL_Q_MQ;
	}

822
	if (queue_mode == NULL_Q_MQ && use_per_node_hctx) {
823
		if (submit_queues < nr_online_nodes) {
824 825
			pr_warn("null_blk: submit_queues param is set to %u.",
							nr_online_nodes);
826 827
			submit_queues = nr_online_nodes;
		}
828
	} else if (submit_queues > nr_cpu_ids)
829 830 831 832 833 834 835 836 837 838
		submit_queues = nr_cpu_ids;
	else if (!submit_queues)
		submit_queues = 1;

	mutex_init(&lock);

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

M
Matias Bjørling 已提交
839 840 841 842 843
	if (use_lightnvm) {
		ppa_cache = kmem_cache_create("ppa_cache", 64 * sizeof(u64),
								0, 0, NULL);
		if (!ppa_cache) {
			pr_err("null_blk: unable to create ppa cache\n");
844 845
			ret = -ENOMEM;
			goto err_ppa;
M
Matias Bjørling 已提交
846 847 848
		}
	}

849
	for (i = 0; i < nr_devices; i++) {
850 851 852
		ret = null_add_dev();
		if (ret)
			goto err_dev;
853 854 855 856
	}

	pr_info("null: module loaded\n");
	return 0;
857 858 859 860 861 862

err_dev:
	while (!list_empty(&nullb_list)) {
		nullb = list_entry(nullb_list.next, struct nullb, list);
		null_del_dev(nullb);
	}
M
Matias Bjørling 已提交
863
	kmem_cache_destroy(ppa_cache);
864 865 866
err_ppa:
	unregister_blkdev(null_major, "nullb");
	return ret;
867 868 869 870 871 872 873 874 875 876 877 878 879 880
}

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);
M
Matias Bjørling 已提交
881 882

	kmem_cache_destroy(ppa_cache);
883 884 885 886 887 888 889
}

module_init(null_init);
module_exit(null_exit);

MODULE_AUTHOR("Jens Axboe <jaxboe@fusionio.com>");
MODULE_LICENSE("GPL");