1 /* SPDX-License-Identifier: GPL-2.0 */
6 #include <linux/blk-mq.h>
7 #include <linux/part_stat.h>
8 #include <linux/blk-crypto.h>
9 #include <linux/memblock.h> /* for max_pfn/max_low_pfn */
11 #include "blk-crypto-internal.h"
13 #include "blk-mq-sched.h"
15 /* Max future timer expiry for timeouts */
16 #define BLK_MAX_TIMEOUT (5 * HZ)
18 extern struct dentry *blk_debugfs_root;
20 struct blk_flush_queue {
21 unsigned int flush_pending_idx:1;
22 unsigned int flush_running_idx:1;
23 blk_status_t rq_status;
24 unsigned long flush_pending_since;
25 struct list_head flush_queue[2];
26 struct list_head flush_data_in_flight;
27 struct request *flush_rq;
29 spinlock_t mq_flush_lock;
32 extern struct kmem_cache *blk_requestq_cachep;
33 extern struct kobj_type blk_queue_ktype;
34 extern struct ida blk_queue_ida;
36 static inline struct blk_flush_queue *
37 blk_get_flush_queue(struct request_queue *q, struct blk_mq_ctx *ctx)
39 return blk_mq_map_queue(q, REQ_OP_FLUSH, ctx)->fq;
42 static inline void __blk_get_queue(struct request_queue *q)
44 kobject_get(&q->kobj);
47 bool is_flush_rq(struct request *req);
49 struct blk_flush_queue *blk_alloc_flush_queue(int node, int cmd_size,
51 void blk_free_flush_queue(struct blk_flush_queue *q);
53 void blk_freeze_queue(struct request_queue *q);
55 #define BIO_INLINE_VECS 4
56 struct bio_vec *bvec_alloc(mempool_t *pool, unsigned short *nr_vecs,
58 void bvec_free(mempool_t *pool, struct bio_vec *bv, unsigned short nr_vecs);
60 static inline bool biovec_phys_mergeable(struct request_queue *q,
61 struct bio_vec *vec1, struct bio_vec *vec2)
63 unsigned long mask = queue_segment_boundary(q);
64 phys_addr_t addr1 = page_to_phys(vec1->bv_page) + vec1->bv_offset;
65 phys_addr_t addr2 = page_to_phys(vec2->bv_page) + vec2->bv_offset;
67 if (addr1 + vec1->bv_len != addr2)
69 if (xen_domain() && !xen_biovec_phys_mergeable(vec1, vec2->bv_page))
71 if ((addr1 | mask) != ((addr2 + vec2->bv_len - 1) | mask))
76 static inline bool __bvec_gap_to_prev(struct request_queue *q,
77 struct bio_vec *bprv, unsigned int offset)
79 return (offset & queue_virt_boundary(q)) ||
80 ((bprv->bv_offset + bprv->bv_len) & queue_virt_boundary(q));
84 * Check if adding a bio_vec after bprv with offset would create a gap in
85 * the SG list. Most drivers don't care about this, but some do.
87 static inline bool bvec_gap_to_prev(struct request_queue *q,
88 struct bio_vec *bprv, unsigned int offset)
90 if (!queue_virt_boundary(q))
92 return __bvec_gap_to_prev(q, bprv, offset);
95 #ifdef CONFIG_BLK_DEV_INTEGRITY
96 void blk_flush_integrity(void);
97 bool __bio_integrity_endio(struct bio *);
98 void bio_integrity_free(struct bio *bio);
99 static inline bool bio_integrity_endio(struct bio *bio)
101 if (bio_integrity(bio))
102 return __bio_integrity_endio(bio);
106 bool blk_integrity_merge_rq(struct request_queue *, struct request *,
108 bool blk_integrity_merge_bio(struct request_queue *, struct request *,
111 static inline bool integrity_req_gap_back_merge(struct request *req,
114 struct bio_integrity_payload *bip = bio_integrity(req->bio);
115 struct bio_integrity_payload *bip_next = bio_integrity(next);
117 return bvec_gap_to_prev(req->q, &bip->bip_vec[bip->bip_vcnt - 1],
118 bip_next->bip_vec[0].bv_offset);
121 static inline bool integrity_req_gap_front_merge(struct request *req,
124 struct bio_integrity_payload *bip = bio_integrity(bio);
125 struct bio_integrity_payload *bip_next = bio_integrity(req->bio);
127 return bvec_gap_to_prev(req->q, &bip->bip_vec[bip->bip_vcnt - 1],
128 bip_next->bip_vec[0].bv_offset);
131 int blk_integrity_add(struct gendisk *disk);
132 void blk_integrity_del(struct gendisk *);
133 #else /* CONFIG_BLK_DEV_INTEGRITY */
134 static inline bool blk_integrity_merge_rq(struct request_queue *rq,
135 struct request *r1, struct request *r2)
139 static inline bool blk_integrity_merge_bio(struct request_queue *rq,
140 struct request *r, struct bio *b)
144 static inline bool integrity_req_gap_back_merge(struct request *req,
149 static inline bool integrity_req_gap_front_merge(struct request *req,
155 static inline void blk_flush_integrity(void)
158 static inline bool bio_integrity_endio(struct bio *bio)
162 static inline void bio_integrity_free(struct bio *bio)
165 static inline int blk_integrity_add(struct gendisk *disk)
169 static inline void blk_integrity_del(struct gendisk *disk)
172 #endif /* CONFIG_BLK_DEV_INTEGRITY */
174 unsigned long blk_rq_timeout(unsigned long timeout);
175 void blk_add_timer(struct request *req);
177 bool blk_attempt_plug_merge(struct request_queue *q, struct bio *bio,
178 unsigned int nr_segs, struct request **same_queue_rq);
179 bool blk_bio_list_merge(struct request_queue *q, struct list_head *list,
180 struct bio *bio, unsigned int nr_segs);
182 void blk_account_io_start(struct request *req);
183 void blk_account_io_done(struct request *req, u64 now);
186 * Internal elevator interface
188 #define ELV_ON_HASH(rq) ((rq)->rq_flags & RQF_HASHED)
190 void blk_insert_flush(struct request *rq);
192 int elevator_switch_mq(struct request_queue *q,
193 struct elevator_type *new_e);
194 void __elevator_exit(struct request_queue *, struct elevator_queue *);
195 int elv_register_queue(struct request_queue *q, bool uevent);
196 void elv_unregister_queue(struct request_queue *q);
198 static inline void elevator_exit(struct request_queue *q,
199 struct elevator_queue *e)
201 lockdep_assert_held(&q->sysfs_lock);
203 blk_mq_sched_free_requests(q);
204 __elevator_exit(q, e);
207 ssize_t part_size_show(struct device *dev, struct device_attribute *attr,
209 ssize_t part_stat_show(struct device *dev, struct device_attribute *attr,
211 ssize_t part_inflight_show(struct device *dev, struct device_attribute *attr,
213 ssize_t part_fail_show(struct device *dev, struct device_attribute *attr,
215 ssize_t part_fail_store(struct device *dev, struct device_attribute *attr,
216 const char *buf, size_t count);
217 ssize_t part_timeout_show(struct device *, struct device_attribute *, char *);
218 ssize_t part_timeout_store(struct device *, struct device_attribute *,
219 const char *, size_t);
221 void __blk_queue_split(struct bio **bio, unsigned int *nr_segs);
222 int ll_back_merge_fn(struct request *req, struct bio *bio,
223 unsigned int nr_segs);
224 bool blk_attempt_req_merge(struct request_queue *q, struct request *rq,
225 struct request *next);
226 unsigned int blk_recalc_rq_segments(struct request *rq);
227 void blk_rq_set_mixed_merge(struct request *rq);
228 bool blk_rq_merge_ok(struct request *rq, struct bio *bio);
229 enum elv_merge blk_try_merge(struct request *rq, struct bio *bio);
231 int blk_dev_init(void);
234 * Contribute to IO statistics IFF:
236 * a) it's attached to a gendisk, and
237 * b) the queue had IO stats enabled when this request was started
239 static inline bool blk_do_io_stat(struct request *rq)
241 return rq->rq_disk && (rq->rq_flags & RQF_IO_STAT);
244 static inline void req_set_nomerge(struct request_queue *q, struct request *req)
246 req->cmd_flags |= REQ_NOMERGE;
247 if (req == q->last_merge)
248 q->last_merge = NULL;
252 * The max size one bio can handle is UINT_MAX becasue bvec_iter.bi_size
253 * is defined as 'unsigned int', meantime it has to aligned to with logical
254 * block size which is the minimum accepted unit by hardware.
256 static inline unsigned int bio_allowed_max_sectors(struct request_queue *q)
258 return round_down(UINT_MAX, queue_logical_block_size(q)) >> 9;
262 * The max bio size which is aligned to q->limits.discard_granularity. This
263 * is a hint to split large discard bio in generic block layer, then if device
264 * driver needs to split the discard bio into smaller ones, their bi_size can
265 * be very probably and easily aligned to discard_granularity of the device's
268 static inline unsigned int bio_aligned_discard_max_sectors(
269 struct request_queue *q)
271 return round_down(UINT_MAX, q->limits.discard_granularity) >>
276 * Internal io_context interface
278 void get_io_context(struct io_context *ioc);
279 struct io_cq *ioc_lookup_icq(struct io_context *ioc, struct request_queue *q);
280 struct io_cq *ioc_create_icq(struct io_context *ioc, struct request_queue *q,
282 void ioc_clear_queue(struct request_queue *q);
284 int create_task_io_context(struct task_struct *task, gfp_t gfp_mask, int node);
287 * Internal throttling interface
289 #ifdef CONFIG_BLK_DEV_THROTTLING
290 extern int blk_throtl_init(struct request_queue *q);
291 extern void blk_throtl_exit(struct request_queue *q);
292 extern void blk_throtl_register_queue(struct request_queue *q);
293 extern void blk_throtl_charge_bio_split(struct bio *bio);
294 bool blk_throtl_bio(struct bio *bio);
295 #else /* CONFIG_BLK_DEV_THROTTLING */
296 static inline int blk_throtl_init(struct request_queue *q) { return 0; }
297 static inline void blk_throtl_exit(struct request_queue *q) { }
298 static inline void blk_throtl_register_queue(struct request_queue *q) { }
299 static inline void blk_throtl_charge_bio_split(struct bio *bio) { }
300 static inline bool blk_throtl_bio(struct bio *bio) { return false; }
301 #endif /* CONFIG_BLK_DEV_THROTTLING */
302 #ifdef CONFIG_BLK_DEV_THROTTLING_LOW
303 extern ssize_t blk_throtl_sample_time_show(struct request_queue *q, char *page);
304 extern ssize_t blk_throtl_sample_time_store(struct request_queue *q,
305 const char *page, size_t count);
306 extern void blk_throtl_bio_endio(struct bio *bio);
307 extern void blk_throtl_stat_add(struct request *rq, u64 time);
309 static inline void blk_throtl_bio_endio(struct bio *bio) { }
310 static inline void blk_throtl_stat_add(struct request *rq, u64 time) { }
313 void __blk_queue_bounce(struct request_queue *q, struct bio **bio);
315 static inline bool blk_queue_may_bounce(struct request_queue *q)
317 return IS_ENABLED(CONFIG_BOUNCE) &&
318 q->limits.bounce == BLK_BOUNCE_HIGH &&
319 max_low_pfn >= max_pfn;
322 static inline void blk_queue_bounce(struct request_queue *q, struct bio **bio)
324 if (unlikely(blk_queue_may_bounce(q) && bio_has_data(*bio)))
325 __blk_queue_bounce(q, bio);
328 #ifdef CONFIG_BLK_CGROUP_IOLATENCY
329 extern int blk_iolatency_init(struct request_queue *q);
331 static inline int blk_iolatency_init(struct request_queue *q) { return 0; }
334 struct bio *blk_next_bio(struct bio *bio, unsigned int nr_pages, gfp_t gfp);
336 #ifdef CONFIG_BLK_DEV_ZONED
337 void blk_queue_free_zone_bitmaps(struct request_queue *q);
338 void blk_queue_clear_zone_settings(struct request_queue *q);
340 static inline void blk_queue_free_zone_bitmaps(struct request_queue *q) {}
341 static inline void blk_queue_clear_zone_settings(struct request_queue *q) {}
344 int blk_alloc_ext_minor(void);
345 void blk_free_ext_minor(unsigned int minor);
346 #define ADDPART_FLAG_NONE 0
347 #define ADDPART_FLAG_RAID 1
348 #define ADDPART_FLAG_WHOLEDISK 2
349 int bdev_add_partition(struct gendisk *disk, int partno, sector_t start,
351 int bdev_del_partition(struct gendisk *disk, int partno);
352 int bdev_resize_partition(struct gendisk *disk, int partno, sector_t start,
355 int bio_add_hw_page(struct request_queue *q, struct bio *bio,
356 struct page *page, unsigned int len, unsigned int offset,
357 unsigned int max_sectors, bool *same_page);
359 struct request_queue *blk_alloc_queue(int node_id);
361 int disk_alloc_events(struct gendisk *disk);
362 void disk_add_events(struct gendisk *disk);
363 void disk_del_events(struct gendisk *disk);
364 void disk_release_events(struct gendisk *disk);
365 extern struct device_attribute dev_attr_events;
366 extern struct device_attribute dev_attr_events_async;
367 extern struct device_attribute dev_attr_events_poll_msecs;
369 static inline void bio_clear_hipri(struct bio *bio)
371 /* can't support alloc cache if we turn off polling */
372 bio_clear_flag(bio, BIO_PERCPU_CACHE);
373 bio->bi_opf &= ~REQ_HIPRI;
376 extern const struct address_space_operations def_blk_aops;
378 #endif /* BLK_INTERNAL_H */