• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * Zoned block device handling
4  *
5  * Copyright (c) 2015, Hannes Reinecke
6  * Copyright (c) 2015, SUSE Linux GmbH
7  *
8  * Copyright (c) 2016, Damien Le Moal
9  * Copyright (c) 2016, Western Digital
10  */
11 
12 #include <linux/kernel.h>
13 #include <linux/module.h>
14 #include <linux/rbtree.h>
15 #include <linux/blkdev.h>
16 #include <linux/blk-mq.h>
17 #include <linux/mm.h>
18 #include <linux/vmalloc.h>
19 #include <linux/sched/mm.h>
20 
21 #include "blk.h"
22 
23 #define ZONE_COND_NAME(name) [BLK_ZONE_COND_##name] = #name
24 static const char *const zone_cond_name[] = {
25 	ZONE_COND_NAME(NOT_WP),
26 	ZONE_COND_NAME(EMPTY),
27 	ZONE_COND_NAME(IMP_OPEN),
28 	ZONE_COND_NAME(EXP_OPEN),
29 	ZONE_COND_NAME(CLOSED),
30 	ZONE_COND_NAME(READONLY),
31 	ZONE_COND_NAME(FULL),
32 	ZONE_COND_NAME(OFFLINE),
33 };
34 #undef ZONE_COND_NAME
35 
36 /**
37  * blk_zone_cond_str - Return string XXX in BLK_ZONE_COND_XXX.
38  * @zone_cond: BLK_ZONE_COND_XXX.
39  *
40  * Description: Centralize block layer function to convert BLK_ZONE_COND_XXX
41  * into string format. Useful in the debugging and tracing zone conditions. For
42  * invalid BLK_ZONE_COND_XXX it returns string "UNKNOWN".
43  */
blk_zone_cond_str(enum blk_zone_cond zone_cond)44 const char *blk_zone_cond_str(enum blk_zone_cond zone_cond)
45 {
46 	static const char *zone_cond_str = "UNKNOWN";
47 
48 	if (zone_cond < ARRAY_SIZE(zone_cond_name) && zone_cond_name[zone_cond])
49 		zone_cond_str = zone_cond_name[zone_cond];
50 
51 	return zone_cond_str;
52 }
53 EXPORT_SYMBOL_GPL(blk_zone_cond_str);
54 
55 /*
56  * Return true if a request is a write requests that needs zone write locking.
57  */
blk_req_needs_zone_write_lock(struct request * rq)58 bool blk_req_needs_zone_write_lock(struct request *rq)
59 {
60 	return rq->q->seq_zones_wlock && blk_rq_is_seq_zoned_write(rq);
61 }
62 EXPORT_SYMBOL_GPL(blk_req_needs_zone_write_lock);
63 
blk_req_zone_write_trylock(struct request * rq)64 bool blk_req_zone_write_trylock(struct request *rq)
65 {
66 	unsigned int zno = blk_rq_zone_no(rq);
67 
68 	if (test_and_set_bit(zno, rq->q->seq_zones_wlock))
69 		return false;
70 
71 	WARN_ON_ONCE(rq->rq_flags & RQF_ZONE_WRITE_LOCKED);
72 	rq->rq_flags |= RQF_ZONE_WRITE_LOCKED;
73 
74 	return true;
75 }
76 EXPORT_SYMBOL_GPL(blk_req_zone_write_trylock);
77 
__blk_req_zone_write_lock(struct request * rq)78 void __blk_req_zone_write_lock(struct request *rq)
79 {
80 	if (WARN_ON_ONCE(test_and_set_bit(blk_rq_zone_no(rq),
81 					  rq->q->seq_zones_wlock)))
82 		return;
83 
84 	WARN_ON_ONCE(rq->rq_flags & RQF_ZONE_WRITE_LOCKED);
85 	rq->rq_flags |= RQF_ZONE_WRITE_LOCKED;
86 }
87 EXPORT_SYMBOL_GPL(__blk_req_zone_write_lock);
88 
__blk_req_zone_write_unlock(struct request * rq)89 void __blk_req_zone_write_unlock(struct request *rq)
90 {
91 	rq->rq_flags &= ~RQF_ZONE_WRITE_LOCKED;
92 	if (rq->q->seq_zones_wlock)
93 		WARN_ON_ONCE(!test_and_clear_bit(blk_rq_zone_no(rq),
94 						 rq->q->seq_zones_wlock));
95 }
96 EXPORT_SYMBOL_GPL(__blk_req_zone_write_unlock);
97 
98 /**
99  * blkdev_nr_zones - Get number of zones
100  * @disk:	Target gendisk
101  *
102  * Return the total number of zones of a zoned block device, including the
103  * eventual small last zone if present. For a block device without zone
104  * capabilities, the number of zones is always 0.
105  */
blkdev_nr_zones(struct gendisk * disk)106 unsigned int blkdev_nr_zones(struct gendisk *disk)
107 {
108 	sector_t zone_sectors = blk_queue_zone_sectors(disk->queue);
109 	sector_t capacity = get_capacity(disk);
110 
111 	if (!blk_queue_is_zoned(disk->queue))
112 		return 0;
113 
114 	if (is_power_of_2(zone_sectors))
115 		return (capacity + zone_sectors - 1) >> ilog2(zone_sectors);
116 
117 	return DIV_ROUND_UP_SECTOR_T(capacity, zone_sectors);
118 }
119 EXPORT_SYMBOL_GPL(blkdev_nr_zones);
120 
121 /**
122  * blkdev_report_zones - Get zones information
123  * @bdev:	Target block device
124  * @sector:	Sector from which to report zones
125  * @nr_zones:	Maximum number of zones to report
126  * @cb:		Callback function called for each reported zone
127  * @data:	Private data for the callback
128  *
129  * Description:
130  *    Get zone information starting from the zone containing @sector for at most
131  *    @nr_zones, and call @cb for each zone reported by the device.
132  *    To report all zones in a device starting from @sector, the BLK_ALL_ZONES
133  *    constant can be passed to @nr_zones.
134  *    Returns the number of zones reported by the device, or a negative errno
135  *    value in case of failure.
136  *
137  *    Note: The caller must use memalloc_noXX_save/restore() calls to control
138  *    memory allocations done within this function.
139  */
blkdev_report_zones(struct block_device * bdev,sector_t sector,unsigned int nr_zones,report_zones_cb cb,void * data)140 int blkdev_report_zones(struct block_device *bdev, sector_t sector,
141 			unsigned int nr_zones, report_zones_cb cb, void *data)
142 {
143 	struct gendisk *disk = bdev->bd_disk;
144 	sector_t capacity = get_capacity(disk);
145 
146 	if (!blk_queue_is_zoned(bdev_get_queue(bdev)) ||
147 	    WARN_ON_ONCE(!disk->fops->report_zones))
148 		return -EOPNOTSUPP;
149 
150 	if (!nr_zones || sector >= capacity)
151 		return 0;
152 
153 	return disk->fops->report_zones(disk, sector, nr_zones, cb, data);
154 }
155 EXPORT_SYMBOL_GPL(blkdev_report_zones);
156 
blk_alloc_zone_bitmap(int node,unsigned int nr_zones)157 static inline unsigned long *blk_alloc_zone_bitmap(int node,
158 						   unsigned int nr_zones)
159 {
160 	return kcalloc_node(BITS_TO_LONGS(nr_zones), sizeof(unsigned long),
161 			    GFP_NOIO, node);
162 }
163 
blk_zone_need_reset_cb(struct blk_zone * zone,unsigned int idx,void * data)164 static int blk_zone_need_reset_cb(struct blk_zone *zone, unsigned int idx,
165 				  void *data)
166 {
167 	/*
168 	 * For an all-zones reset, ignore conventional, empty, read-only
169 	 * and offline zones.
170 	 */
171 	switch (zone->cond) {
172 	case BLK_ZONE_COND_NOT_WP:
173 	case BLK_ZONE_COND_EMPTY:
174 	case BLK_ZONE_COND_READONLY:
175 	case BLK_ZONE_COND_OFFLINE:
176 		return 0;
177 	default:
178 		set_bit(idx, (unsigned long *)data);
179 		return 0;
180 	}
181 }
182 
blkdev_zone_reset_all_emulated(struct block_device * bdev,gfp_t gfp_mask)183 static int blkdev_zone_reset_all_emulated(struct block_device *bdev,
184 					  gfp_t gfp_mask)
185 {
186 	struct request_queue *q = bdev_get_queue(bdev);
187 	sector_t capacity = get_capacity(bdev->bd_disk);
188 	sector_t zone_sectors = blk_queue_zone_sectors(q);
189 	unsigned long *need_reset;
190 	struct bio *bio = NULL;
191 	sector_t sector = 0;
192 	int ret;
193 
194 	need_reset = blk_alloc_zone_bitmap(q->node, q->nr_zones);
195 	if (!need_reset)
196 		return -ENOMEM;
197 
198 	ret = bdev->bd_disk->fops->report_zones(bdev->bd_disk, 0,
199 				q->nr_zones, blk_zone_need_reset_cb,
200 				need_reset);
201 	if (ret < 0)
202 		goto out_free_need_reset;
203 
204 	ret = 0;
205 	while (sector < capacity) {
206 		if (!test_bit(blk_queue_zone_no(q, sector), need_reset)) {
207 			sector += zone_sectors;
208 			continue;
209 		}
210 
211 		bio = blk_next_bio(bio, 0, gfp_mask);
212 		bio_set_dev(bio, bdev);
213 		bio->bi_opf = REQ_OP_ZONE_RESET | REQ_SYNC;
214 		bio->bi_iter.bi_sector = sector;
215 		sector += zone_sectors;
216 
217 		/* This may take a while, so be nice to others */
218 		cond_resched();
219 	}
220 
221 	if (bio) {
222 		ret = submit_bio_wait(bio);
223 		bio_put(bio);
224 	}
225 
226 out_free_need_reset:
227 	kfree(need_reset);
228 	return ret;
229 }
230 
blkdev_zone_reset_all(struct block_device * bdev,gfp_t gfp_mask)231 static int blkdev_zone_reset_all(struct block_device *bdev, gfp_t gfp_mask)
232 {
233 	struct bio bio;
234 
235 	bio_init(&bio, NULL, 0);
236 	bio_set_dev(&bio, bdev);
237 	bio.bi_opf = REQ_OP_ZONE_RESET_ALL | REQ_SYNC;
238 
239 	return submit_bio_wait(&bio);
240 }
241 
242 /**
243  * blkdev_zone_mgmt - Execute a zone management operation on a range of zones
244  * @bdev:	Target block device
245  * @op:		Operation to be performed on the zones
246  * @sector:	Start sector of the first zone to operate on
247  * @nr_sectors:	Number of sectors, should be at least the length of one zone and
248  *		must be zone size aligned.
249  * @gfp_mask:	Memory allocation flags (for bio_alloc)
250  *
251  * Description:
252  *    Perform the specified operation on the range of zones specified by
253  *    @sector..@sector+@nr_sectors. Specifying the entire disk sector range
254  *    is valid, but the specified range should not contain conventional zones.
255  *    The operation to execute on each zone can be a zone reset, open, close
256  *    or finish request.
257  */
blkdev_zone_mgmt(struct block_device * bdev,enum req_opf op,sector_t sector,sector_t nr_sectors,gfp_t gfp_mask)258 int blkdev_zone_mgmt(struct block_device *bdev, enum req_opf op,
259 		     sector_t sector, sector_t nr_sectors,
260 		     gfp_t gfp_mask)
261 {
262 	struct request_queue *q = bdev_get_queue(bdev);
263 	sector_t zone_sectors = blk_queue_zone_sectors(q);
264 	sector_t capacity = get_capacity(bdev->bd_disk);
265 	sector_t end_sector = sector + nr_sectors;
266 	struct bio *bio = NULL;
267 	int ret = 0;
268 
269 	if (!blk_queue_is_zoned(q))
270 		return -EOPNOTSUPP;
271 
272 	if (bdev_read_only(bdev))
273 		return -EPERM;
274 
275 	if (!op_is_zone_mgmt(op))
276 		return -EOPNOTSUPP;
277 
278 	if (end_sector <= sector || end_sector > capacity)
279 		/* Out of range */
280 		return -EINVAL;
281 
282 	/* Check alignment (handle eventual smaller last zone) */
283 	if (!bdev_is_zone_start(bdev, sector))
284 		return -EINVAL;
285 
286 	if (!bdev_is_zone_start(bdev, nr_sectors) && end_sector != capacity)
287 		return -EINVAL;
288 
289 	/*
290 	 * In the case of a zone reset operation over all zones,
291 	 * REQ_OP_ZONE_RESET_ALL can be used with devices supporting this
292 	 * command. For other devices, we emulate this command behavior by
293 	 * identifying the zones needing a reset.
294 	 */
295 	if (op == REQ_OP_ZONE_RESET && sector == 0 && nr_sectors == capacity) {
296 		if (!blk_queue_zone_resetall(q))
297 			return blkdev_zone_reset_all_emulated(bdev, gfp_mask);
298 		return blkdev_zone_reset_all(bdev, gfp_mask);
299 	}
300 
301 	while (sector < end_sector) {
302 		bio = blk_next_bio(bio, 0, gfp_mask);
303 		bio_set_dev(bio, bdev);
304 		bio->bi_opf = op | REQ_SYNC;
305 		bio->bi_iter.bi_sector = sector;
306 		sector += zone_sectors;
307 
308 		/* This may take a while, so be nice to others */
309 		cond_resched();
310 	}
311 
312 	ret = submit_bio_wait(bio);
313 	bio_put(bio);
314 
315 	return ret;
316 }
317 EXPORT_SYMBOL_GPL(blkdev_zone_mgmt);
318 
319 struct zone_report_args {
320 	struct blk_zone __user *zones;
321 };
322 
blkdev_copy_zone_to_user(struct blk_zone * zone,unsigned int idx,void * data)323 static int blkdev_copy_zone_to_user(struct blk_zone *zone, unsigned int idx,
324 				    void *data)
325 {
326 	struct zone_report_args *args = data;
327 
328 	if (copy_to_user(&args->zones[idx], zone, sizeof(struct blk_zone)))
329 		return -EFAULT;
330 	return 0;
331 }
332 
333 /*
334  * BLKREPORTZONE ioctl processing.
335  * Called from blkdev_ioctl.
336  */
blkdev_report_zones_ioctl(struct block_device * bdev,fmode_t mode,unsigned int cmd,unsigned long arg)337 int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode,
338 			      unsigned int cmd, unsigned long arg)
339 {
340 	void __user *argp = (void __user *)arg;
341 	struct zone_report_args args;
342 	struct request_queue *q;
343 	struct blk_zone_report rep;
344 	int ret;
345 
346 	if (!argp)
347 		return -EINVAL;
348 
349 	q = bdev_get_queue(bdev);
350 	if (!q)
351 		return -ENXIO;
352 
353 	if (!blk_queue_is_zoned(q))
354 		return -ENOTTY;
355 
356 	if (copy_from_user(&rep, argp, sizeof(struct blk_zone_report)))
357 		return -EFAULT;
358 
359 	if (!rep.nr_zones)
360 		return -EINVAL;
361 
362 	args.zones = argp + sizeof(struct blk_zone_report);
363 	ret = blkdev_report_zones(bdev, rep.sector, rep.nr_zones,
364 				  blkdev_copy_zone_to_user, &args);
365 	if (ret < 0)
366 		return ret;
367 
368 	rep.nr_zones = ret;
369 	rep.flags = BLK_ZONE_REP_CAPACITY;
370 	if (copy_to_user(argp, &rep, sizeof(struct blk_zone_report)))
371 		return -EFAULT;
372 	return 0;
373 }
374 
blkdev_truncate_zone_range(struct block_device * bdev,fmode_t mode,const struct blk_zone_range * zrange)375 static int blkdev_truncate_zone_range(struct block_device *bdev, fmode_t mode,
376 				      const struct blk_zone_range *zrange)
377 {
378 	loff_t start, end;
379 
380 	if (zrange->sector + zrange->nr_sectors <= zrange->sector ||
381 	    zrange->sector + zrange->nr_sectors > get_capacity(bdev->bd_disk))
382 		/* Out of range */
383 		return -EINVAL;
384 
385 	start = zrange->sector << SECTOR_SHIFT;
386 	end = ((zrange->sector + zrange->nr_sectors) << SECTOR_SHIFT) - 1;
387 
388 	return truncate_bdev_range(bdev, mode, start, end);
389 }
390 
391 /*
392  * BLKRESETZONE, BLKOPENZONE, BLKCLOSEZONE and BLKFINISHZONE ioctl processing.
393  * Called from blkdev_ioctl.
394  */
blkdev_zone_mgmt_ioctl(struct block_device * bdev,fmode_t mode,unsigned int cmd,unsigned long arg)395 int blkdev_zone_mgmt_ioctl(struct block_device *bdev, fmode_t mode,
396 			   unsigned int cmd, unsigned long arg)
397 {
398 	void __user *argp = (void __user *)arg;
399 	struct request_queue *q;
400 	struct blk_zone_range zrange;
401 	enum req_opf op;
402 	int ret;
403 
404 	if (!argp)
405 		return -EINVAL;
406 
407 	q = bdev_get_queue(bdev);
408 	if (!q)
409 		return -ENXIO;
410 
411 	if (!blk_queue_is_zoned(q))
412 		return -ENOTTY;
413 
414 	if (!(mode & FMODE_WRITE))
415 		return -EBADF;
416 
417 	if (copy_from_user(&zrange, argp, sizeof(struct blk_zone_range)))
418 		return -EFAULT;
419 
420 	switch (cmd) {
421 	case BLKRESETZONE:
422 		op = REQ_OP_ZONE_RESET;
423 
424 		/* Invalidate the page cache, including dirty pages. */
425 		filemap_invalidate_lock(bdev->bd_inode->i_mapping);
426 		ret = blkdev_truncate_zone_range(bdev, mode, &zrange);
427 		if (ret)
428 			goto fail;
429 		break;
430 	case BLKOPENZONE:
431 		op = REQ_OP_ZONE_OPEN;
432 		break;
433 	case BLKCLOSEZONE:
434 		op = REQ_OP_ZONE_CLOSE;
435 		break;
436 	case BLKFINISHZONE:
437 		op = REQ_OP_ZONE_FINISH;
438 		break;
439 	default:
440 		return -ENOTTY;
441 	}
442 
443 	ret = blkdev_zone_mgmt(bdev, op, zrange.sector, zrange.nr_sectors,
444 			       GFP_KERNEL);
445 
446 fail:
447 	if (cmd == BLKRESETZONE)
448 		filemap_invalidate_unlock(bdev->bd_inode->i_mapping);
449 
450 	return ret;
451 }
452 
blk_queue_free_zone_bitmaps(struct request_queue * q)453 void blk_queue_free_zone_bitmaps(struct request_queue *q)
454 {
455 	kfree(q->conv_zones_bitmap);
456 	q->conv_zones_bitmap = NULL;
457 	kfree(q->seq_zones_wlock);
458 	q->seq_zones_wlock = NULL;
459 }
460 
461 struct blk_revalidate_zone_args {
462 	struct gendisk	*disk;
463 	unsigned long	*conv_zones_bitmap;
464 	unsigned long	*seq_zones_wlock;
465 	unsigned int	nr_zones;
466 	sector_t	zone_sectors;
467 	sector_t	sector;
468 };
469 
470 /*
471  * Helper function to check the validity of zones of a zoned block device.
472  */
blk_revalidate_zone_cb(struct blk_zone * zone,unsigned int idx,void * data)473 static int blk_revalidate_zone_cb(struct blk_zone *zone, unsigned int idx,
474 				  void *data)
475 {
476 	struct blk_revalidate_zone_args *args = data;
477 	struct gendisk *disk = args->disk;
478 	struct request_queue *q = disk->queue;
479 	sector_t capacity = get_capacity(disk);
480 
481 	/*
482 	 * All zones must have the same size, with the exception on an eventual
483 	 * smaller last zone.
484 	 */
485 	if (zone->start == 0) {
486 		if (zone->len == 0) {
487 			pr_warn("%s: Invalid zero zone size", disk->disk_name);
488 			return -ENODEV;
489 		}
490 
491 		/*
492 		 * Non power-of-2 zone size support was added to remove the
493 		 * gap between zone capacity and zone size. Though it is technically
494 		 * possible to have gaps in a non power-of-2 device, Linux requires
495 		 * the zone size to be equal to zone capacity for non power-of-2
496 		 * zoned devices.
497 		 */
498 		if (!is_power_of_2(zone->len) && zone->capacity < zone->len) {
499 			pr_err("%s: Invalid zone capacity %lld with non power-of-2 zone size %lld",
500 			       disk->disk_name, zone->capacity, zone->len);
501 			return -ENODEV;
502 		}
503 
504 		args->zone_sectors = zone->len;
505 		args->nr_zones = div64_u64(capacity + zone->len - 1, zone->len);
506 	} else if (zone->start + args->zone_sectors < capacity) {
507 		if (zone->len != args->zone_sectors) {
508 			pr_warn("%s: Invalid zoned device with non constant zone size\n",
509 				disk->disk_name);
510 			return -ENODEV;
511 		}
512 	} else {
513 		if (zone->len > args->zone_sectors) {
514 			pr_warn("%s: Invalid zoned device with larger last zone size\n",
515 				disk->disk_name);
516 			return -ENODEV;
517 		}
518 	}
519 
520 	/* Check for holes in the zone report */
521 	if (zone->start != args->sector) {
522 		pr_warn("%s: Zone gap at sectors %llu..%llu\n",
523 			disk->disk_name, args->sector, zone->start);
524 		return -ENODEV;
525 	}
526 
527 	/* Check zone type */
528 	switch (zone->type) {
529 	case BLK_ZONE_TYPE_CONVENTIONAL:
530 		if (!args->conv_zones_bitmap) {
531 			args->conv_zones_bitmap =
532 				blk_alloc_zone_bitmap(q->node, args->nr_zones);
533 			if (!args->conv_zones_bitmap)
534 				return -ENOMEM;
535 		}
536 		set_bit(idx, args->conv_zones_bitmap);
537 		break;
538 	case BLK_ZONE_TYPE_SEQWRITE_REQ:
539 	case BLK_ZONE_TYPE_SEQWRITE_PREF:
540 		if (!args->seq_zones_wlock) {
541 			args->seq_zones_wlock =
542 				blk_alloc_zone_bitmap(q->node, args->nr_zones);
543 			if (!args->seq_zones_wlock)
544 				return -ENOMEM;
545 		}
546 		break;
547 	default:
548 		pr_warn("%s: Invalid zone type 0x%x at sectors %llu\n",
549 			disk->disk_name, (int)zone->type, zone->start);
550 		return -ENODEV;
551 	}
552 
553 	args->sector += zone->len;
554 	return 0;
555 }
556 
557 /**
558  * blk_revalidate_disk_zones - (re)allocate and initialize zone bitmaps
559  * @disk:	Target disk
560  * @update_driver_data:	Callback to update driver data on the frozen disk
561  *
562  * Helper function for low-level device drivers to (re) allocate and initialize
563  * a disk request queue zone bitmaps. This functions should normally be called
564  * within the disk ->revalidate method for blk-mq based drivers.  For BIO based
565  * drivers only q->nr_zones needs to be updated so that the sysfs exposed value
566  * is correct.
567  * If the @update_driver_data callback function is not NULL, the callback is
568  * executed with the device request queue frozen after all zones have been
569  * checked.
570  */
blk_revalidate_disk_zones(struct gendisk * disk,void (* update_driver_data)(struct gendisk * disk))571 int blk_revalidate_disk_zones(struct gendisk *disk,
572 			      void (*update_driver_data)(struct gendisk *disk))
573 {
574 	struct request_queue *q = disk->queue;
575 	struct blk_revalidate_zone_args args = {
576 		.disk		= disk,
577 	};
578 	unsigned int noio_flag;
579 	int ret;
580 
581 	if (WARN_ON_ONCE(!blk_queue_is_zoned(q)))
582 		return -EIO;
583 	if (WARN_ON_ONCE(!queue_is_mq(q)))
584 		return -EIO;
585 
586 	if (!get_capacity(disk))
587 		return -EIO;
588 
589 	/*
590 	 * Ensure that all memory allocations in this context are done as if
591 	 * GFP_NOIO was specified.
592 	 */
593 	noio_flag = memalloc_noio_save();
594 	ret = disk->fops->report_zones(disk, 0, UINT_MAX,
595 				       blk_revalidate_zone_cb, &args);
596 	if (!ret) {
597 		pr_warn("%s: No zones reported\n", disk->disk_name);
598 		ret = -ENODEV;
599 	}
600 	memalloc_noio_restore(noio_flag);
601 
602 	/*
603 	 * If zones where reported, make sure that the entire disk capacity
604 	 * has been checked.
605 	 */
606 	if (ret > 0 && args.sector != get_capacity(disk)) {
607 		pr_warn("%s: Missing zones from sector %llu\n",
608 			disk->disk_name, args.sector);
609 		ret = -ENODEV;
610 	}
611 
612 	/*
613 	 * Install the new bitmaps and update nr_zones only once the queue is
614 	 * stopped and all I/Os are completed (i.e. a scheduler is not
615 	 * referencing the bitmaps).
616 	 */
617 	blk_mq_freeze_queue(q);
618 	if (ret > 0) {
619 		blk_queue_chunk_sectors(q, args.zone_sectors);
620 		q->nr_zones = args.nr_zones;
621 		swap(q->seq_zones_wlock, args.seq_zones_wlock);
622 		swap(q->conv_zones_bitmap, args.conv_zones_bitmap);
623 		if (update_driver_data)
624 			update_driver_data(disk);
625 		ret = 0;
626 	} else {
627 		pr_warn("%s: failed to revalidate zones\n", disk->disk_name);
628 		blk_queue_free_zone_bitmaps(q);
629 	}
630 	blk_mq_unfreeze_queue(q);
631 
632 	kfree(args.seq_zones_wlock);
633 	kfree(args.conv_zones_bitmap);
634 	return ret;
635 }
636 EXPORT_SYMBOL_GPL(blk_revalidate_disk_zones);
637 
blk_queue_clear_zone_settings(struct request_queue * q)638 void blk_queue_clear_zone_settings(struct request_queue *q)
639 {
640 	blk_mq_freeze_queue(q);
641 
642 	blk_queue_free_zone_bitmaps(q);
643 	blk_queue_flag_clear(QUEUE_FLAG_ZONE_RESETALL, q);
644 	q->required_elevator_features &= ~ELEVATOR_F_ZBD_SEQ_WRITE;
645 	q->nr_zones = 0;
646 	q->max_open_zones = 0;
647 	q->max_active_zones = 0;
648 	q->limits.chunk_sectors = 0;
649 	q->limits.zone_write_granularity = 0;
650 	q->limits.max_zone_append_sectors = 0;
651 
652 	blk_mq_unfreeze_queue(q);
653 }
654