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
);
54 void __blk_mq_unfreeze_queue(struct request_queue
*q
, bool force_atomic
);
55 void blk_queue_start_drain(struct request_queue
*q
);
57 #define BIO_INLINE_VECS 4
58 struct bio_vec
*bvec_alloc(mempool_t
*pool
, unsigned short *nr_vecs
,
60 void bvec_free(mempool_t
*pool
, struct bio_vec
*bv
, unsigned short nr_vecs
);
62 static inline bool biovec_phys_mergeable(struct request_queue
*q
,
63 struct bio_vec
*vec1
, struct bio_vec
*vec2
)
65 unsigned long mask
= queue_segment_boundary(q
);
66 phys_addr_t addr1
= page_to_phys(vec1
->bv_page
) + vec1
->bv_offset
;
67 phys_addr_t addr2
= page_to_phys(vec2
->bv_page
) + vec2
->bv_offset
;
69 if (addr1
+ vec1
->bv_len
!= addr2
)
71 if (xen_domain() && !xen_biovec_phys_mergeable(vec1
, vec2
->bv_page
))
73 if ((addr1
| mask
) != ((addr2
+ vec2
->bv_len
- 1) | mask
))
78 static inline bool __bvec_gap_to_prev(struct request_queue
*q
,
79 struct bio_vec
*bprv
, unsigned int offset
)
81 return (offset
& queue_virt_boundary(q
)) ||
82 ((bprv
->bv_offset
+ bprv
->bv_len
) & queue_virt_boundary(q
));
86 * Check if adding a bio_vec after bprv with offset would create a gap in
87 * the SG list. Most drivers don't care about this, but some do.
89 static inline bool bvec_gap_to_prev(struct request_queue
*q
,
90 struct bio_vec
*bprv
, unsigned int offset
)
92 if (!queue_virt_boundary(q
))
94 return __bvec_gap_to_prev(q
, bprv
, offset
);
97 #ifdef CONFIG_BLK_DEV_INTEGRITY
98 void blk_flush_integrity(void);
99 bool __bio_integrity_endio(struct bio
*);
100 void bio_integrity_free(struct bio
*bio
);
101 static inline bool bio_integrity_endio(struct bio
*bio
)
103 if (bio_integrity(bio
))
104 return __bio_integrity_endio(bio
);
108 bool blk_integrity_merge_rq(struct request_queue
*, struct request
*,
110 bool blk_integrity_merge_bio(struct request_queue
*, struct request
*,
113 static inline bool integrity_req_gap_back_merge(struct request
*req
,
116 struct bio_integrity_payload
*bip
= bio_integrity(req
->bio
);
117 struct bio_integrity_payload
*bip_next
= bio_integrity(next
);
119 return bvec_gap_to_prev(req
->q
, &bip
->bip_vec
[bip
->bip_vcnt
- 1],
120 bip_next
->bip_vec
[0].bv_offset
);
123 static inline bool integrity_req_gap_front_merge(struct request
*req
,
126 struct bio_integrity_payload
*bip
= bio_integrity(bio
);
127 struct bio_integrity_payload
*bip_next
= bio_integrity(req
->bio
);
129 return bvec_gap_to_prev(req
->q
, &bip
->bip_vec
[bip
->bip_vcnt
- 1],
130 bip_next
->bip_vec
[0].bv_offset
);
133 int blk_integrity_add(struct gendisk
*disk
);
134 void blk_integrity_del(struct gendisk
*);
135 #else /* CONFIG_BLK_DEV_INTEGRITY */
136 static inline bool blk_integrity_merge_rq(struct request_queue
*rq
,
137 struct request
*r1
, struct request
*r2
)
141 static inline bool blk_integrity_merge_bio(struct request_queue
*rq
,
142 struct request
*r
, struct bio
*b
)
146 static inline bool integrity_req_gap_back_merge(struct request
*req
,
151 static inline bool integrity_req_gap_front_merge(struct request
*req
,
157 static inline void blk_flush_integrity(void)
160 static inline bool bio_integrity_endio(struct bio
*bio
)
164 static inline void bio_integrity_free(struct bio
*bio
)
167 static inline int blk_integrity_add(struct gendisk
*disk
)
171 static inline void blk_integrity_del(struct gendisk
*disk
)
174 #endif /* CONFIG_BLK_DEV_INTEGRITY */
176 unsigned long blk_rq_timeout(unsigned long timeout
);
177 void blk_add_timer(struct request
*req
);
179 bool blk_attempt_plug_merge(struct request_queue
*q
, struct bio
*bio
,
180 unsigned int nr_segs
, struct request
**same_queue_rq
);
181 bool blk_bio_list_merge(struct request_queue
*q
, struct list_head
*list
,
182 struct bio
*bio
, unsigned int nr_segs
);
184 void blk_account_io_start(struct request
*req
);
185 void blk_account_io_done(struct request
*req
, u64 now
);
188 * Internal elevator interface
190 #define ELV_ON_HASH(rq) ((rq)->rq_flags & RQF_HASHED)
192 void blk_insert_flush(struct request
*rq
);
194 int elevator_switch_mq(struct request_queue
*q
,
195 struct elevator_type
*new_e
);
196 void __elevator_exit(struct request_queue
*, struct elevator_queue
*);
197 int elv_register_queue(struct request_queue
*q
, bool uevent
);
198 void elv_unregister_queue(struct request_queue
*q
);
200 static inline void elevator_exit(struct request_queue
*q
,
201 struct elevator_queue
*e
)
203 lockdep_assert_held(&q
->sysfs_lock
);
205 blk_mq_sched_free_requests(q
);
206 __elevator_exit(q
, e
);
209 ssize_t
part_size_show(struct device
*dev
, struct device_attribute
*attr
,
211 ssize_t
part_stat_show(struct device
*dev
, struct device_attribute
*attr
,
213 ssize_t
part_inflight_show(struct device
*dev
, struct device_attribute
*attr
,
215 ssize_t
part_fail_show(struct device
*dev
, struct device_attribute
*attr
,
217 ssize_t
part_fail_store(struct device
*dev
, struct device_attribute
*attr
,
218 const char *buf
, size_t count
);
219 ssize_t
part_timeout_show(struct device
*, struct device_attribute
*, char *);
220 ssize_t
part_timeout_store(struct device
*, struct device_attribute
*,
221 const char *, size_t);
223 void __blk_queue_split(struct bio
**bio
, unsigned int *nr_segs
);
224 int ll_back_merge_fn(struct request
*req
, struct bio
*bio
,
225 unsigned int nr_segs
);
226 bool blk_attempt_req_merge(struct request_queue
*q
, struct request
*rq
,
227 struct request
*next
);
228 unsigned int blk_recalc_rq_segments(struct request
*rq
);
229 void blk_rq_set_mixed_merge(struct request
*rq
);
230 bool blk_rq_merge_ok(struct request
*rq
, struct bio
*bio
);
231 enum elv_merge
blk_try_merge(struct request
*rq
, struct bio
*bio
);
233 int blk_dev_init(void);
236 * Contribute to IO statistics IFF:
238 * a) it's attached to a gendisk, and
239 * b) the queue had IO stats enabled when this request was started
241 static inline bool blk_do_io_stat(struct request
*rq
)
243 return rq
->rq_disk
&& (rq
->rq_flags
& RQF_IO_STAT
);
246 static inline void req_set_nomerge(struct request_queue
*q
, struct request
*req
)
248 req
->cmd_flags
|= REQ_NOMERGE
;
249 if (req
== q
->last_merge
)
250 q
->last_merge
= NULL
;
254 * The max size one bio can handle is UINT_MAX becasue bvec_iter.bi_size
255 * is defined as 'unsigned int', meantime it has to aligned to with logical
256 * block size which is the minimum accepted unit by hardware.
258 static inline unsigned int bio_allowed_max_sectors(struct request_queue
*q
)
260 return round_down(UINT_MAX
, queue_logical_block_size(q
)) >> 9;
264 * The max bio size which is aligned to q->limits.discard_granularity. This
265 * is a hint to split large discard bio in generic block layer, then if device
266 * driver needs to split the discard bio into smaller ones, their bi_size can
267 * be very probably and easily aligned to discard_granularity of the device's
270 static inline unsigned int bio_aligned_discard_max_sectors(
271 struct request_queue
*q
)
273 return round_down(UINT_MAX
, q
->limits
.discard_granularity
) >>
278 * Internal io_context interface
280 void get_io_context(struct io_context
*ioc
);
281 struct io_cq
*ioc_lookup_icq(struct io_context
*ioc
, struct request_queue
*q
);
282 struct io_cq
*ioc_create_icq(struct io_context
*ioc
, struct request_queue
*q
,
284 void ioc_clear_queue(struct request_queue
*q
);
286 int create_task_io_context(struct task_struct
*task
, gfp_t gfp_mask
, int node
);
289 * Internal throttling interface
291 #ifdef CONFIG_BLK_DEV_THROTTLING
292 extern int blk_throtl_init(struct request_queue
*q
);
293 extern void blk_throtl_exit(struct request_queue
*q
);
294 extern void blk_throtl_register_queue(struct request_queue
*q
);
295 extern void blk_throtl_charge_bio_split(struct bio
*bio
);
296 bool blk_throtl_bio(struct bio
*bio
);
297 #else /* CONFIG_BLK_DEV_THROTTLING */
298 static inline int blk_throtl_init(struct request_queue
*q
) { return 0; }
299 static inline void blk_throtl_exit(struct request_queue
*q
) { }
300 static inline void blk_throtl_register_queue(struct request_queue
*q
) { }
301 static inline void blk_throtl_charge_bio_split(struct bio
*bio
) { }
302 static inline bool blk_throtl_bio(struct bio
*bio
) { return false; }
303 #endif /* CONFIG_BLK_DEV_THROTTLING */
304 #ifdef CONFIG_BLK_DEV_THROTTLING_LOW
305 extern ssize_t
blk_throtl_sample_time_show(struct request_queue
*q
, char *page
);
306 extern ssize_t
blk_throtl_sample_time_store(struct request_queue
*q
,
307 const char *page
, size_t count
);
308 extern void blk_throtl_bio_endio(struct bio
*bio
);
309 extern void blk_throtl_stat_add(struct request
*rq
, u64 time
);
311 static inline void blk_throtl_bio_endio(struct bio
*bio
) { }
312 static inline void blk_throtl_stat_add(struct request
*rq
, u64 time
) { }
315 void __blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
317 static inline bool blk_queue_may_bounce(struct request_queue
*q
)
319 return IS_ENABLED(CONFIG_BOUNCE
) &&
320 q
->limits
.bounce
== BLK_BOUNCE_HIGH
&&
321 max_low_pfn
>= max_pfn
;
324 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
326 if (unlikely(blk_queue_may_bounce(q
) && bio_has_data(*bio
)))
327 __blk_queue_bounce(q
, bio
);
330 #ifdef CONFIG_BLK_CGROUP_IOLATENCY
331 extern int blk_iolatency_init(struct request_queue
*q
);
333 static inline int blk_iolatency_init(struct request_queue
*q
) { return 0; }
336 struct bio
*blk_next_bio(struct bio
*bio
, unsigned int nr_pages
, gfp_t gfp
);
338 #ifdef CONFIG_BLK_DEV_ZONED
339 void blk_queue_free_zone_bitmaps(struct request_queue
*q
);
340 void blk_queue_clear_zone_settings(struct request_queue
*q
);
342 static inline void blk_queue_free_zone_bitmaps(struct request_queue
*q
) {}
343 static inline void blk_queue_clear_zone_settings(struct request_queue
*q
) {}
346 int blk_alloc_ext_minor(void);
347 void blk_free_ext_minor(unsigned int minor
);
348 #define ADDPART_FLAG_NONE 0
349 #define ADDPART_FLAG_RAID 1
350 #define ADDPART_FLAG_WHOLEDISK 2
351 int bdev_add_partition(struct gendisk
*disk
, int partno
, sector_t start
,
353 int bdev_del_partition(struct gendisk
*disk
, int partno
);
354 int bdev_resize_partition(struct gendisk
*disk
, int partno
, sector_t start
,
357 int bio_add_hw_page(struct request_queue
*q
, struct bio
*bio
,
358 struct page
*page
, unsigned int len
, unsigned int offset
,
359 unsigned int max_sectors
, bool *same_page
);
361 struct request_queue
*blk_alloc_queue(int node_id
);
363 int disk_alloc_events(struct gendisk
*disk
);
364 void disk_add_events(struct gendisk
*disk
);
365 void disk_del_events(struct gendisk
*disk
);
366 void disk_release_events(struct gendisk
*disk
);
367 extern struct device_attribute dev_attr_events
;
368 extern struct device_attribute dev_attr_events_async
;
369 extern struct device_attribute dev_attr_events_poll_msecs
;
371 static inline void bio_clear_hipri(struct bio
*bio
)
373 /* can't support alloc cache if we turn off polling */
374 bio_clear_flag(bio
, BIO_PERCPU_CACHE
);
375 bio
->bi_opf
&= ~REQ_HIPRI
;
378 extern const struct address_space_operations def_blk_aops
;
380 #endif /* BLK_INTERNAL_H */