Commit | Line | Data |
---|---|---|
b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
1da177e4 LT |
2 | #ifndef _LINUX_BLKDEV_H |
3 | #define _LINUX_BLKDEV_H | |
4 | ||
85fd0bc9 | 5 | #include <linux/sched.h> |
e6017571 | 6 | #include <linux/sched/clock.h> |
85fd0bc9 | 7 | |
f5ff8422 JA |
8 | #ifdef CONFIG_BLOCK |
9 | ||
1da177e4 LT |
10 | #include <linux/major.h> |
11 | #include <linux/genhd.h> | |
12 | #include <linux/list.h> | |
320ae51f | 13 | #include <linux/llist.h> |
1da177e4 LT |
14 | #include <linux/timer.h> |
15 | #include <linux/workqueue.h> | |
16 | #include <linux/pagemap.h> | |
66114cad | 17 | #include <linux/backing-dev-defs.h> |
1da177e4 LT |
18 | #include <linux/wait.h> |
19 | #include <linux/mempool.h> | |
34c0fd54 | 20 | #include <linux/pfn.h> |
1da177e4 | 21 | #include <linux/bio.h> |
1da177e4 | 22 | #include <linux/stringify.h> |
3e6053d7 | 23 | #include <linux/gfp.h> |
d351af01 | 24 | #include <linux/bsg.h> |
c7c22e4d | 25 | #include <linux/smp.h> |
548bc8e1 | 26 | #include <linux/rcupdate.h> |
add703fd | 27 | #include <linux/percpu-refcount.h> |
84be456f | 28 | #include <linux/scatterlist.h> |
6a0cb1bc | 29 | #include <linux/blkzoned.h> |
1da177e4 | 30 | |
de477254 | 31 | struct module; |
21b2f0c8 CH |
32 | struct scsi_ioctl_command; |
33 | ||
1da177e4 | 34 | struct request_queue; |
1da177e4 | 35 | struct elevator_queue; |
2056a782 | 36 | struct blk_trace; |
3d6392cf JA |
37 | struct request; |
38 | struct sg_io_hdr; | |
aa387cc8 | 39 | struct bsg_job; |
3c798398 | 40 | struct blkcg_gq; |
7c94e1c1 | 41 | struct blk_flush_queue; |
bbd3e064 | 42 | struct pr_ops; |
a7905043 | 43 | struct rq_qos; |
34dbad5d OS |
44 | struct blk_queue_stats; |
45 | struct blk_stat_callback; | |
1da177e4 LT |
46 | |
47 | #define BLKDEV_MIN_RQ 4 | |
48 | #define BLKDEV_MAX_RQ 128 /* Default maximum */ | |
49 | ||
096392e0 | 50 | /* Must be consistent with blk_mq_poll_stats_bkt() */ |
0206319f SB |
51 | #define BLK_MQ_POLL_STATS_BKTS 16 |
52 | ||
29ece8b4 YY |
53 | /* Doing classic polling */ |
54 | #define BLK_MQ_POLL_CLASSIC -1 | |
55 | ||
8bd435b3 TH |
56 | /* |
57 | * Maximum number of blkcg policies allowed to be registered concurrently. | |
58 | * Defined here to simplify include dependency. | |
59 | */ | |
01c5f85a | 60 | #define BLKCG_MAX_POLS 5 |
8bd435b3 | 61 | |
2a842aca | 62 | typedef void (rq_end_io_fn)(struct request *, blk_status_t); |
1da177e4 | 63 | |
e8064021 CH |
64 | /* |
65 | * request flags */ | |
66 | typedef __u32 __bitwise req_flags_t; | |
67 | ||
68 | /* elevator knows about this request */ | |
69 | #define RQF_SORTED ((__force req_flags_t)(1 << 0)) | |
70 | /* drive already may have started this one */ | |
71 | #define RQF_STARTED ((__force req_flags_t)(1 << 1)) | |
e8064021 CH |
72 | /* may not be passed by ioscheduler */ |
73 | #define RQF_SOFTBARRIER ((__force req_flags_t)(1 << 3)) | |
74 | /* request for flush sequence */ | |
75 | #define RQF_FLUSH_SEQ ((__force req_flags_t)(1 << 4)) | |
76 | /* merge of different types, fail separately */ | |
77 | #define RQF_MIXED_MERGE ((__force req_flags_t)(1 << 5)) | |
78 | /* track inflight for MQ */ | |
79 | #define RQF_MQ_INFLIGHT ((__force req_flags_t)(1 << 6)) | |
80 | /* don't call prep for this one */ | |
81 | #define RQF_DONTPREP ((__force req_flags_t)(1 << 7)) | |
82 | /* set for "ide_preempt" requests and also for requests for which the SCSI | |
83 | "quiesce" state must be ignored. */ | |
84 | #define RQF_PREEMPT ((__force req_flags_t)(1 << 8)) | |
85 | /* contains copies of user pages */ | |
86 | #define RQF_COPY_USER ((__force req_flags_t)(1 << 9)) | |
87 | /* vaguely specified driver internal error. Ignored by the block layer */ | |
88 | #define RQF_FAILED ((__force req_flags_t)(1 << 10)) | |
89 | /* don't warn about errors */ | |
90 | #define RQF_QUIET ((__force req_flags_t)(1 << 11)) | |
91 | /* elevator private data attached */ | |
92 | #define RQF_ELVPRIV ((__force req_flags_t)(1 << 12)) | |
4822e902 | 93 | /* account into disk and partition IO statistics */ |
e8064021 CH |
94 | #define RQF_IO_STAT ((__force req_flags_t)(1 << 13)) |
95 | /* request came from our alloc pool */ | |
96 | #define RQF_ALLOCED ((__force req_flags_t)(1 << 14)) | |
97 | /* runtime pm request */ | |
98 | #define RQF_PM ((__force req_flags_t)(1 << 15)) | |
99 | /* on IO scheduler merge hash */ | |
100 | #define RQF_HASHED ((__force req_flags_t)(1 << 16)) | |
4822e902 | 101 | /* track IO completion time */ |
cf43e6be | 102 | #define RQF_STATS ((__force req_flags_t)(1 << 17)) |
f9d03f96 CH |
103 | /* Look at ->special_vec for the actual data payload instead of the |
104 | bio chain. */ | |
105 | #define RQF_SPECIAL_PAYLOAD ((__force req_flags_t)(1 << 18)) | |
6cc77e9c CH |
106 | /* The per-zone write lock is held for this request */ |
107 | #define RQF_ZONE_WRITE_LOCKED ((__force req_flags_t)(1 << 19)) | |
76a86f9d | 108 | /* already slept for hybrid poll */ |
12f5b931 | 109 | #define RQF_MQ_POLL_SLEPT ((__force req_flags_t)(1 << 20)) |
da661267 CH |
110 | /* ->timeout has been called, don't expire again */ |
111 | #define RQF_TIMED_OUT ((__force req_flags_t)(1 << 21)) | |
e8064021 CH |
112 | |
113 | /* flags that prevent us from merging requests: */ | |
114 | #define RQF_NOMERGE_FLAGS \ | |
f9d03f96 | 115 | (RQF_STARTED | RQF_SOFTBARRIER | RQF_FLUSH_SEQ | RQF_SPECIAL_PAYLOAD) |
e8064021 | 116 | |
12f5b931 KB |
117 | /* |
118 | * Request state for blk-mq. | |
119 | */ | |
120 | enum mq_rq_state { | |
121 | MQ_RQ_IDLE = 0, | |
122 | MQ_RQ_IN_FLIGHT = 1, | |
123 | MQ_RQ_COMPLETE = 2, | |
124 | }; | |
125 | ||
1da177e4 | 126 | /* |
af76e555 CH |
127 | * Try to put the fields that are referenced together in the same cacheline. |
128 | * | |
129 | * If you modify this structure, make sure to update blk_rq_init() and | |
130 | * especially blk_mq_rq_ctx_init() to take care of the added fields. | |
1da177e4 LT |
131 | */ |
132 | struct request { | |
165125e1 | 133 | struct request_queue *q; |
320ae51f | 134 | struct blk_mq_ctx *mq_ctx; |
ea4f995e | 135 | struct blk_mq_hw_ctx *mq_hctx; |
e6a1c874 | 136 | |
ef295ecf | 137 | unsigned int cmd_flags; /* op and common flags */ |
e8064021 | 138 | req_flags_t rq_flags; |
d486f1f2 | 139 | |
2f578aaf | 140 | int tag; |
d486f1f2 JA |
141 | int internal_tag; |
142 | ||
a2dec7b3 | 143 | /* the following two fields are internal, NEVER access directly */ |
a2dec7b3 | 144 | unsigned int __data_len; /* total data len */ |
181fdde3 | 145 | sector_t __sector; /* sector cursor */ |
1da177e4 LT |
146 | |
147 | struct bio *bio; | |
148 | struct bio *biotail; | |
149 | ||
7c3fb70f JA |
150 | struct list_head queuelist; |
151 | ||
360f92c2 JA |
152 | /* |
153 | * The hash is used inside the scheduler, and killed once the | |
154 | * request reaches the dispatch list. The ipi_list is only used | |
155 | * to queue the request for softirq completion, which is long | |
156 | * after the request has been unhashed (and even removed from | |
157 | * the dispatch list). | |
158 | */ | |
159 | union { | |
160 | struct hlist_node hash; /* merge hash */ | |
161 | struct list_head ipi_list; | |
162 | }; | |
163 | ||
e6a1c874 JA |
164 | /* |
165 | * The rb_node is only used inside the io scheduler, requests | |
166 | * are pruned when moved to the dispatch queue. So let the | |
c186794d | 167 | * completion_data share space with the rb_node. |
e6a1c874 JA |
168 | */ |
169 | union { | |
170 | struct rb_node rb_node; /* sort/lookup */ | |
f9d03f96 | 171 | struct bio_vec special_vec; |
c186794d | 172 | void *completion_data; |
e26738e0 | 173 | int error_count; /* for legacy drivers, don't use */ |
e6a1c874 | 174 | }; |
9817064b | 175 | |
ff7d145f | 176 | /* |
7f1dc8a2 | 177 | * Three pointers are available for the IO schedulers, if they need |
c186794d MS |
178 | * more they have to dynamically allocate it. Flush requests are |
179 | * never put on the IO scheduler. So let the flush fields share | |
a612fddf | 180 | * space with the elevator data. |
ff7d145f | 181 | */ |
c186794d | 182 | union { |
a612fddf TH |
183 | struct { |
184 | struct io_cq *icq; | |
185 | void *priv[2]; | |
186 | } elv; | |
187 | ||
c186794d MS |
188 | struct { |
189 | unsigned int seq; | |
190 | struct list_head list; | |
4853abaa | 191 | rq_end_io_fn *saved_end_io; |
c186794d MS |
192 | } flush; |
193 | }; | |
ff7d145f | 194 | |
8f34ee75 | 195 | struct gendisk *rq_disk; |
09e099d4 | 196 | struct hd_struct *part; |
522a7775 OS |
197 | /* Time that I/O was submitted to the kernel. */ |
198 | u64 start_time_ns; | |
544ccc8d OS |
199 | /* Time that I/O was submitted to the device. */ |
200 | u64 io_start_time_ns; | |
201 | ||
202 | #ifdef CONFIG_BLK_WBT | |
203 | unsigned short wbt_flags; | |
204 | #endif | |
205 | #ifdef CONFIG_BLK_DEV_THROTTLING_LOW | |
206 | unsigned short throtl_size; | |
207 | #endif | |
208 | ||
209 | /* | |
210 | * Number of scatter-gather DMA addr+len pairs after | |
1da177e4 LT |
211 | * physical address coalescing is performed. |
212 | */ | |
213 | unsigned short nr_phys_segments; | |
7c3fb70f | 214 | |
13f05c8d MP |
215 | #if defined(CONFIG_BLK_DEV_INTEGRITY) |
216 | unsigned short nr_integrity_segments; | |
217 | #endif | |
1da177e4 | 218 | |
7c3fb70f | 219 | unsigned short write_hint; |
8f34ee75 JA |
220 | unsigned short ioprio; |
221 | ||
7a85f889 | 222 | unsigned int extra_len; /* length of alignment and padding */ |
1da177e4 | 223 | |
12f5b931 KB |
224 | enum mq_rq_state state; |
225 | refcount_t ref; | |
1d9bd516 | 226 | |
0b7576d8 | 227 | unsigned int timeout; |
079076b3 | 228 | unsigned long deadline; |
cb6934f8 | 229 | |
7c3fb70f | 230 | union { |
0a4b6e2f | 231 | struct __call_single_data csd; |
7c3fb70f JA |
232 | u64 fifo_time; |
233 | }; | |
234 | ||
1da177e4 | 235 | /* |
c00895ab | 236 | * completion callback. |
1da177e4 LT |
237 | */ |
238 | rq_end_io_fn *end_io; | |
239 | void *end_io_data; | |
240 | }; | |
241 | ||
14cb0dc6 ML |
242 | static inline bool blk_op_is_scsi(unsigned int op) |
243 | { | |
244 | return op == REQ_OP_SCSI_IN || op == REQ_OP_SCSI_OUT; | |
245 | } | |
246 | ||
247 | static inline bool blk_op_is_private(unsigned int op) | |
248 | { | |
249 | return op == REQ_OP_DRV_IN || op == REQ_OP_DRV_OUT; | |
250 | } | |
251 | ||
aebf526b CH |
252 | static inline bool blk_rq_is_scsi(struct request *rq) |
253 | { | |
14cb0dc6 | 254 | return blk_op_is_scsi(req_op(rq)); |
aebf526b CH |
255 | } |
256 | ||
257 | static inline bool blk_rq_is_private(struct request *rq) | |
258 | { | |
14cb0dc6 | 259 | return blk_op_is_private(req_op(rq)); |
aebf526b CH |
260 | } |
261 | ||
57292b58 CH |
262 | static inline bool blk_rq_is_passthrough(struct request *rq) |
263 | { | |
aebf526b | 264 | return blk_rq_is_scsi(rq) || blk_rq_is_private(rq); |
57292b58 CH |
265 | } |
266 | ||
14cb0dc6 ML |
267 | static inline bool bio_is_passthrough(struct bio *bio) |
268 | { | |
269 | unsigned op = bio_op(bio); | |
270 | ||
271 | return blk_op_is_scsi(op) || blk_op_is_private(op); | |
272 | } | |
273 | ||
766ca442 FLVC |
274 | static inline unsigned short req_get_ioprio(struct request *req) |
275 | { | |
276 | return req->ioprio; | |
277 | } | |
278 | ||
1da177e4 LT |
279 | #include <linux/elevator.h> |
280 | ||
320ae51f JA |
281 | struct blk_queue_ctx; |
282 | ||
dece1635 | 283 | typedef blk_qc_t (make_request_fn) (struct request_queue *q, struct bio *bio); |
1da177e4 LT |
284 | |
285 | struct bio_vec; | |
2fb98e84 | 286 | typedef int (dma_drain_needed_fn)(struct request *); |
1da177e4 | 287 | |
242f9dcb | 288 | enum blk_eh_timer_return { |
88b0cfad CH |
289 | BLK_EH_DONE, /* drivers has completed the command */ |
290 | BLK_EH_RESET_TIMER, /* reset timer and try again */ | |
242f9dcb JA |
291 | }; |
292 | ||
1da177e4 LT |
293 | enum blk_queue_state { |
294 | Queue_down, | |
295 | Queue_up, | |
296 | }; | |
297 | ||
ee1b6f7a SL |
298 | #define BLK_TAG_ALLOC_FIFO 0 /* allocate starting from 0 */ |
299 | #define BLK_TAG_ALLOC_RR 1 /* allocate starting from last allocated tag */ | |
1da177e4 | 300 | |
abf54393 FT |
301 | #define BLK_SCSI_MAX_CMDS (256) |
302 | #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8)) | |
303 | ||
797476b8 DLM |
304 | /* |
305 | * Zoned block device models (zoned limit). | |
306 | */ | |
307 | enum blk_zoned_model { | |
308 | BLK_ZONED_NONE, /* Regular block device */ | |
309 | BLK_ZONED_HA, /* Host-aware zoned block device */ | |
310 | BLK_ZONED_HM, /* Host-managed zoned block device */ | |
311 | }; | |
312 | ||
025146e1 MP |
313 | struct queue_limits { |
314 | unsigned long bounce_pfn; | |
315 | unsigned long seg_boundary_mask; | |
03100aad | 316 | unsigned long virt_boundary_mask; |
025146e1 MP |
317 | |
318 | unsigned int max_hw_sectors; | |
ca369d51 | 319 | unsigned int max_dev_sectors; |
762380ad | 320 | unsigned int chunk_sectors; |
025146e1 MP |
321 | unsigned int max_sectors; |
322 | unsigned int max_segment_size; | |
c72758f3 MP |
323 | unsigned int physical_block_size; |
324 | unsigned int alignment_offset; | |
325 | unsigned int io_min; | |
326 | unsigned int io_opt; | |
67efc925 | 327 | unsigned int max_discard_sectors; |
0034af03 | 328 | unsigned int max_hw_discard_sectors; |
4363ac7c | 329 | unsigned int max_write_same_sectors; |
a6f0788e | 330 | unsigned int max_write_zeroes_sectors; |
86b37281 MP |
331 | unsigned int discard_granularity; |
332 | unsigned int discard_alignment; | |
025146e1 MP |
333 | |
334 | unsigned short logical_block_size; | |
8a78362c | 335 | unsigned short max_segments; |
13f05c8d | 336 | unsigned short max_integrity_segments; |
1e739730 | 337 | unsigned short max_discard_segments; |
025146e1 | 338 | |
c72758f3 | 339 | unsigned char misaligned; |
86b37281 | 340 | unsigned char discard_misaligned; |
c78afc62 | 341 | unsigned char raid_partial_stripes_expensive; |
797476b8 | 342 | enum blk_zoned_model zoned; |
025146e1 MP |
343 | }; |
344 | ||
6a0cb1bc HR |
345 | #ifdef CONFIG_BLK_DEV_ZONED |
346 | ||
a91e1380 | 347 | extern unsigned int blkdev_nr_zones(struct block_device *bdev); |
6a0cb1bc HR |
348 | extern int blkdev_report_zones(struct block_device *bdev, |
349 | sector_t sector, struct blk_zone *zones, | |
350 | unsigned int *nr_zones, gfp_t gfp_mask); | |
351 | extern int blkdev_reset_zones(struct block_device *bdev, sector_t sectors, | |
352 | sector_t nr_sectors, gfp_t gfp_mask); | |
bf505456 | 353 | extern int blk_revalidate_disk_zones(struct gendisk *disk); |
6a0cb1bc | 354 | |
3ed05a98 ST |
355 | extern int blkdev_report_zones_ioctl(struct block_device *bdev, fmode_t mode, |
356 | unsigned int cmd, unsigned long arg); | |
357 | extern int blkdev_reset_zones_ioctl(struct block_device *bdev, fmode_t mode, | |
358 | unsigned int cmd, unsigned long arg); | |
359 | ||
360 | #else /* CONFIG_BLK_DEV_ZONED */ | |
361 | ||
a91e1380 DLM |
362 | static inline unsigned int blkdev_nr_zones(struct block_device *bdev) |
363 | { | |
364 | return 0; | |
365 | } | |
bf505456 DLM |
366 | |
367 | static inline int blk_revalidate_disk_zones(struct gendisk *disk) | |
368 | { | |
369 | return 0; | |
370 | } | |
371 | ||
3ed05a98 ST |
372 | static inline int blkdev_report_zones_ioctl(struct block_device *bdev, |
373 | fmode_t mode, unsigned int cmd, | |
374 | unsigned long arg) | |
375 | { | |
376 | return -ENOTTY; | |
377 | } | |
378 | ||
379 | static inline int blkdev_reset_zones_ioctl(struct block_device *bdev, | |
380 | fmode_t mode, unsigned int cmd, | |
381 | unsigned long arg) | |
382 | { | |
383 | return -ENOTTY; | |
384 | } | |
385 | ||
6a0cb1bc HR |
386 | #endif /* CONFIG_BLK_DEV_ZONED */ |
387 | ||
d7b76301 | 388 | struct request_queue { |
1da177e4 LT |
389 | /* |
390 | * Together with queue_head for cacheline sharing | |
391 | */ | |
392 | struct list_head queue_head; | |
393 | struct request *last_merge; | |
b374d18a | 394 | struct elevator_queue *elevator; |
1da177e4 | 395 | |
34dbad5d | 396 | struct blk_queue_stats *stats; |
a7905043 | 397 | struct rq_qos *rq_qos; |
87760e5e | 398 | |
1da177e4 | 399 | make_request_fn *make_request_fn; |
2fb98e84 | 400 | dma_drain_needed_fn *dma_drain_needed; |
1da177e4 | 401 | |
f8a5b122 | 402 | const struct blk_mq_ops *mq_ops; |
320ae51f | 403 | |
320ae51f | 404 | /* sw queues */ |
e6cdb092 | 405 | struct blk_mq_ctx __percpu *queue_ctx; |
320ae51f JA |
406 | unsigned int nr_queues; |
407 | ||
d278d4a8 JA |
408 | unsigned int queue_depth; |
409 | ||
320ae51f JA |
410 | /* hw dispatch queues */ |
411 | struct blk_mq_hw_ctx **queue_hw_ctx; | |
412 | unsigned int nr_hw_queues; | |
413 | ||
dc3b17cc | 414 | struct backing_dev_info *backing_dev_info; |
1da177e4 LT |
415 | |
416 | /* | |
417 | * The queue owner gets to use this for whatever they like. | |
418 | * ll_rw_blk doesn't touch it. | |
419 | */ | |
420 | void *queuedata; | |
421 | ||
1da177e4 | 422 | /* |
d7b76301 | 423 | * various queue flags, see QUEUE_* below |
1da177e4 | 424 | */ |
d7b76301 | 425 | unsigned long queue_flags; |
cd84a62e BVA |
426 | /* |
427 | * Number of contexts that have called blk_set_pm_only(). If this | |
428 | * counter is above zero then only RQF_PM and RQF_PREEMPT requests are | |
429 | * processed. | |
430 | */ | |
431 | atomic_t pm_only; | |
1da177e4 | 432 | |
a73f730d TH |
433 | /* |
434 | * ida allocated id for this queue. Used to index queues from | |
435 | * ioctx. | |
436 | */ | |
437 | int id; | |
438 | ||
1da177e4 | 439 | /* |
d7b76301 | 440 | * queue needs bounce pages for pages above this limit |
1da177e4 | 441 | */ |
d7b76301 | 442 | gfp_t bounce_gfp; |
1da177e4 | 443 | |
0d945c1f | 444 | spinlock_t queue_lock; |
1da177e4 LT |
445 | |
446 | /* | |
447 | * queue kobject | |
448 | */ | |
449 | struct kobject kobj; | |
450 | ||
320ae51f JA |
451 | /* |
452 | * mq queue kobject | |
453 | */ | |
1db4909e | 454 | struct kobject *mq_kobj; |
320ae51f | 455 | |
ac6fc48c DW |
456 | #ifdef CONFIG_BLK_DEV_INTEGRITY |
457 | struct blk_integrity integrity; | |
458 | #endif /* CONFIG_BLK_DEV_INTEGRITY */ | |
459 | ||
47fafbc7 | 460 | #ifdef CONFIG_PM |
6c954667 LM |
461 | struct device *dev; |
462 | int rpm_status; | |
463 | unsigned int nr_pending; | |
464 | #endif | |
465 | ||
1da177e4 LT |
466 | /* |
467 | * queue settings | |
468 | */ | |
469 | unsigned long nr_requests; /* Max # of requests */ | |
1da177e4 | 470 | |
fa0ccd83 | 471 | unsigned int dma_drain_size; |
d7b76301 | 472 | void *dma_drain_buffer; |
e3790c7d | 473 | unsigned int dma_pad_mask; |
1da177e4 LT |
474 | unsigned int dma_alignment; |
475 | ||
242f9dcb | 476 | unsigned int rq_timeout; |
64f1c21e | 477 | int poll_nsec; |
34dbad5d OS |
478 | |
479 | struct blk_stat_callback *poll_cb; | |
0206319f | 480 | struct blk_rq_stat poll_stat[BLK_MQ_POLL_STATS_BKTS]; |
34dbad5d | 481 | |
242f9dcb | 482 | struct timer_list timeout; |
287922eb | 483 | struct work_struct timeout_work; |
242f9dcb | 484 | |
a612fddf | 485 | struct list_head icq_list; |
4eef3049 | 486 | #ifdef CONFIG_BLK_CGROUP |
a2b1693b | 487 | DECLARE_BITMAP (blkcg_pols, BLKCG_MAX_POLS); |
3c798398 | 488 | struct blkcg_gq *root_blkg; |
03aa264a | 489 | struct list_head blkg_list; |
4eef3049 | 490 | #endif |
a612fddf | 491 | |
025146e1 MP |
492 | struct queue_limits limits; |
493 | ||
6a5ac984 | 494 | #ifdef CONFIG_BLK_DEV_ZONED |
6cc77e9c CH |
495 | /* |
496 | * Zoned block device information for request dispatch control. | |
497 | * nr_zones is the total number of zones of the device. This is always | |
498 | * 0 for regular block devices. seq_zones_bitmap is a bitmap of nr_zones | |
499 | * bits which indicates if a zone is conventional (bit clear) or | |
500 | * sequential (bit set). seq_zones_wlock is a bitmap of nr_zones | |
501 | * bits which indicates if a zone is write locked, that is, if a write | |
502 | * request targeting the zone was dispatched. All three fields are | |
503 | * initialized by the low level device driver (e.g. scsi/sd.c). | |
504 | * Stacking drivers (device mappers) may or may not initialize | |
505 | * these fields. | |
ccce20fc BVA |
506 | * |
507 | * Reads of this information must be protected with blk_queue_enter() / | |
508 | * blk_queue_exit(). Modifying this information is only allowed while | |
509 | * no requests are being processed. See also blk_mq_freeze_queue() and | |
510 | * blk_mq_unfreeze_queue(). | |
6cc77e9c CH |
511 | */ |
512 | unsigned int nr_zones; | |
513 | unsigned long *seq_zones_bitmap; | |
514 | unsigned long *seq_zones_wlock; | |
6a5ac984 | 515 | #endif /* CONFIG_BLK_DEV_ZONED */ |
6cc77e9c | 516 | |
1da177e4 LT |
517 | /* |
518 | * sg stuff | |
519 | */ | |
520 | unsigned int sg_timeout; | |
521 | unsigned int sg_reserved_size; | |
1946089a | 522 | int node; |
6c5c9341 | 523 | #ifdef CONFIG_BLK_DEV_IO_TRACE |
2056a782 | 524 | struct blk_trace *blk_trace; |
5acb3cc2 | 525 | struct mutex blk_trace_mutex; |
6c5c9341 | 526 | #endif |
1da177e4 | 527 | /* |
4913efe4 | 528 | * for flush operations |
1da177e4 | 529 | */ |
7c94e1c1 | 530 | struct blk_flush_queue *fq; |
483f4afc | 531 | |
6fca6a61 CH |
532 | struct list_head requeue_list; |
533 | spinlock_t requeue_lock; | |
2849450a | 534 | struct delayed_work requeue_work; |
6fca6a61 | 535 | |
483f4afc | 536 | struct mutex sysfs_lock; |
d351af01 | 537 | |
2f8f1336 ML |
538 | /* |
539 | * for reusing dead hctx instance in case of updating | |
540 | * nr_hw_queues | |
541 | */ | |
542 | struct list_head unused_hctx_list; | |
543 | spinlock_t unused_hctx_lock; | |
544 | ||
7996a8b5 | 545 | int mq_freeze_depth; |
d732580b | 546 | |
d351af01 FT |
547 | #if defined(CONFIG_BLK_DEV_BSG) |
548 | struct bsg_class_device bsg_dev; | |
549 | #endif | |
e43473b7 VG |
550 | |
551 | #ifdef CONFIG_BLK_DEV_THROTTLING | |
552 | /* Throttle data */ | |
553 | struct throtl_data *td; | |
554 | #endif | |
548bc8e1 | 555 | struct rcu_head rcu_head; |
320ae51f | 556 | wait_queue_head_t mq_freeze_wq; |
7996a8b5 BL |
557 | /* |
558 | * Protect concurrent access to q_usage_counter by | |
559 | * percpu_ref_kill() and percpu_ref_reinit(). | |
560 | */ | |
561 | struct mutex mq_freeze_lock; | |
3ef28e83 | 562 | struct percpu_ref q_usage_counter; |
0d2602ca JA |
563 | |
564 | struct blk_mq_tag_set *tag_set; | |
565 | struct list_head tag_set_list; | |
338aa96d | 566 | struct bio_set bio_split; |
4593fdbe | 567 | |
03796c14 | 568 | #ifdef CONFIG_BLK_DEBUG_FS |
07e4fead | 569 | struct dentry *debugfs_dir; |
d332ce09 | 570 | struct dentry *sched_debugfs_dir; |
cc56694f | 571 | struct dentry *rqos_debugfs_dir; |
07e4fead OS |
572 | #endif |
573 | ||
4593fdbe | 574 | bool mq_sysfs_init_done; |
6d247d7f CH |
575 | |
576 | size_t cmd_size; | |
dc9edc44 BVA |
577 | |
578 | struct work_struct release_work; | |
f793dfd3 JA |
579 | |
580 | #define BLK_MAX_WRITE_HINTS 5 | |
581 | u64 write_hints[BLK_MAX_WRITE_HINTS]; | |
1da177e4 LT |
582 | }; |
583 | ||
eca7abf3 JA |
584 | #define QUEUE_FLAG_STOPPED 0 /* queue is stopped */ |
585 | #define QUEUE_FLAG_DYING 1 /* queue being torn down */ | |
eca7abf3 JA |
586 | #define QUEUE_FLAG_NOMERGES 3 /* disable merge attempts */ |
587 | #define QUEUE_FLAG_SAME_COMP 4 /* complete on same CPU-group */ | |
588 | #define QUEUE_FLAG_FAIL_IO 5 /* fake timeout */ | |
589 | #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */ | |
590 | #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */ | |
591 | #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */ | |
592 | #define QUEUE_FLAG_DISCARD 8 /* supports DISCARD */ | |
593 | #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */ | |
594 | #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */ | |
595 | #define QUEUE_FLAG_SECERASE 11 /* supports secure erase */ | |
596 | #define QUEUE_FLAG_SAME_FORCE 12 /* force complete on same CPU */ | |
597 | #define QUEUE_FLAG_DEAD 13 /* queue tear-down finished */ | |
598 | #define QUEUE_FLAG_INIT_DONE 14 /* queue is initialized */ | |
eca7abf3 JA |
599 | #define QUEUE_FLAG_POLL 16 /* IO polling enabled if set */ |
600 | #define QUEUE_FLAG_WC 17 /* Write back caching */ | |
601 | #define QUEUE_FLAG_FUA 18 /* device supports FUA writes */ | |
602 | #define QUEUE_FLAG_DAX 19 /* device supports DAX */ | |
603 | #define QUEUE_FLAG_STATS 20 /* track IO start and completion times */ | |
604 | #define QUEUE_FLAG_POLL_STATS 21 /* collecting stats for hybrid polling */ | |
605 | #define QUEUE_FLAG_REGISTERED 22 /* queue has been registered to a disk */ | |
606 | #define QUEUE_FLAG_SCSI_PASSTHROUGH 23 /* queue supports SCSI commands */ | |
607 | #define QUEUE_FLAG_QUIESCED 24 /* queue has been quiesced */ | |
608 | #define QUEUE_FLAG_PCI_P2PDMA 25 /* device supports PCI p2p requests */ | |
797e7dbb | 609 | |
94eddfbe | 610 | #define QUEUE_FLAG_MQ_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \ |
6e0de611 | 611 | (1 << QUEUE_FLAG_SAME_COMP)) |
94eddfbe | 612 | |
8814ce8a BVA |
613 | void blk_queue_flag_set(unsigned int flag, struct request_queue *q); |
614 | void blk_queue_flag_clear(unsigned int flag, struct request_queue *q); | |
615 | bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q); | |
8814ce8a | 616 | |
1da177e4 | 617 | #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags) |
3f3299d5 | 618 | #define blk_queue_dying(q) test_bit(QUEUE_FLAG_DYING, &(q)->queue_flags) |
c246e80d | 619 | #define blk_queue_dead(q) test_bit(QUEUE_FLAG_DEAD, &(q)->queue_flags) |
320ae51f | 620 | #define blk_queue_init_done(q) test_bit(QUEUE_FLAG_INIT_DONE, &(q)->queue_flags) |
ac9fafa1 | 621 | #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags) |
488991e2 AB |
622 | #define blk_queue_noxmerges(q) \ |
623 | test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags) | |
a68bbddb | 624 | #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags) |
bc58ba94 | 625 | #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags) |
e2e1a148 | 626 | #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags) |
c15227de | 627 | #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags) |
288dab8a CH |
628 | #define blk_queue_secure_erase(q) \ |
629 | (test_bit(QUEUE_FLAG_SECERASE, &(q)->queue_flags)) | |
163d4baa | 630 | #define blk_queue_dax(q) test_bit(QUEUE_FLAG_DAX, &(q)->queue_flags) |
9efc160f BVA |
631 | #define blk_queue_scsi_passthrough(q) \ |
632 | test_bit(QUEUE_FLAG_SCSI_PASSTHROUGH, &(q)->queue_flags) | |
49d92c0d LG |
633 | #define blk_queue_pci_p2pdma(q) \ |
634 | test_bit(QUEUE_FLAG_PCI_P2PDMA, &(q)->queue_flags) | |
1da177e4 | 635 | |
33659ebb CH |
636 | #define blk_noretry_request(rq) \ |
637 | ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \ | |
638 | REQ_FAILFAST_DRIVER)) | |
f4560ffe | 639 | #define blk_queue_quiesced(q) test_bit(QUEUE_FLAG_QUIESCED, &(q)->queue_flags) |
cd84a62e | 640 | #define blk_queue_pm_only(q) atomic_read(&(q)->pm_only) |
0ce91444 | 641 | #define blk_queue_fua(q) test_bit(QUEUE_FLAG_FUA, &(q)->queue_flags) |
c9254f2d | 642 | |
cd84a62e BVA |
643 | extern void blk_set_pm_only(struct request_queue *q); |
644 | extern void blk_clear_pm_only(struct request_queue *q); | |
33659ebb | 645 | |
57292b58 CH |
646 | static inline bool blk_account_rq(struct request *rq) |
647 | { | |
648 | return (rq->rq_flags & RQF_STARTED) && !blk_rq_is_passthrough(rq); | |
649 | } | |
33659ebb | 650 | |
1da177e4 LT |
651 | #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist) |
652 | ||
4e1b2d52 | 653 | #define rq_data_dir(rq) (op_is_write(req_op(rq)) ? WRITE : READ) |
1da177e4 | 654 | |
9d9de535 CH |
655 | #define rq_dma_dir(rq) \ |
656 | (op_is_write(req_op(rq)) ? DMA_TO_DEVICE : DMA_FROM_DEVICE) | |
657 | ||
3ab3a031 CH |
658 | #define dma_map_bvec(dev, bv, dir, attrs) \ |
659 | dma_map_page_attrs(dev, (bv)->bv_page, (bv)->bv_offset, (bv)->bv_len, \ | |
660 | (dir), (attrs)) | |
661 | ||
344e9ffc | 662 | static inline bool queue_is_mq(struct request_queue *q) |
49fd524f | 663 | { |
a1ce35fa | 664 | return q->mq_ops; |
49fd524f JA |
665 | } |
666 | ||
797476b8 DLM |
667 | static inline enum blk_zoned_model |
668 | blk_queue_zoned_model(struct request_queue *q) | |
669 | { | |
670 | return q->limits.zoned; | |
671 | } | |
672 | ||
673 | static inline bool blk_queue_is_zoned(struct request_queue *q) | |
674 | { | |
675 | switch (blk_queue_zoned_model(q)) { | |
676 | case BLK_ZONED_HA: | |
677 | case BLK_ZONED_HM: | |
678 | return true; | |
679 | default: | |
680 | return false; | |
681 | } | |
682 | } | |
683 | ||
f99e8648 | 684 | static inline unsigned int blk_queue_zone_sectors(struct request_queue *q) |
6a0cb1bc HR |
685 | { |
686 | return blk_queue_is_zoned(q) ? q->limits.chunk_sectors : 0; | |
687 | } | |
688 | ||
6a5ac984 | 689 | #ifdef CONFIG_BLK_DEV_ZONED |
965b652e DLM |
690 | static inline unsigned int blk_queue_nr_zones(struct request_queue *q) |
691 | { | |
692 | return blk_queue_is_zoned(q) ? q->nr_zones : 0; | |
693 | } | |
694 | ||
6cc77e9c CH |
695 | static inline unsigned int blk_queue_zone_no(struct request_queue *q, |
696 | sector_t sector) | |
697 | { | |
698 | if (!blk_queue_is_zoned(q)) | |
699 | return 0; | |
700 | return sector >> ilog2(q->limits.chunk_sectors); | |
701 | } | |
702 | ||
703 | static inline bool blk_queue_zone_is_seq(struct request_queue *q, | |
704 | sector_t sector) | |
705 | { | |
706 | if (!blk_queue_is_zoned(q) || !q->seq_zones_bitmap) | |
707 | return false; | |
708 | return test_bit(blk_queue_zone_no(q, sector), q->seq_zones_bitmap); | |
709 | } | |
965b652e DLM |
710 | #else /* CONFIG_BLK_DEV_ZONED */ |
711 | static inline unsigned int blk_queue_nr_zones(struct request_queue *q) | |
712 | { | |
713 | return 0; | |
714 | } | |
6a5ac984 | 715 | #endif /* CONFIG_BLK_DEV_ZONED */ |
6cc77e9c | 716 | |
1faa16d2 JA |
717 | static inline bool rq_is_sync(struct request *rq) |
718 | { | |
ef295ecf | 719 | return op_is_sync(rq->cmd_flags); |
1faa16d2 JA |
720 | } |
721 | ||
e2a60da7 MP |
722 | static inline bool rq_mergeable(struct request *rq) |
723 | { | |
57292b58 | 724 | if (blk_rq_is_passthrough(rq)) |
e2a60da7 | 725 | return false; |
1da177e4 | 726 | |
3a5e02ce MC |
727 | if (req_op(rq) == REQ_OP_FLUSH) |
728 | return false; | |
729 | ||
a6f0788e CK |
730 | if (req_op(rq) == REQ_OP_WRITE_ZEROES) |
731 | return false; | |
732 | ||
e2a60da7 | 733 | if (rq->cmd_flags & REQ_NOMERGE_FLAGS) |
e8064021 CH |
734 | return false; |
735 | if (rq->rq_flags & RQF_NOMERGE_FLAGS) | |
e2a60da7 MP |
736 | return false; |
737 | ||
738 | return true; | |
739 | } | |
1da177e4 | 740 | |
4363ac7c MP |
741 | static inline bool blk_write_same_mergeable(struct bio *a, struct bio *b) |
742 | { | |
efbeccdb CH |
743 | if (bio_page(a) == bio_page(b) && |
744 | bio_offset(a) == bio_offset(b)) | |
4363ac7c MP |
745 | return true; |
746 | ||
747 | return false; | |
748 | } | |
749 | ||
d278d4a8 JA |
750 | static inline unsigned int blk_queue_depth(struct request_queue *q) |
751 | { | |
752 | if (q->queue_depth) | |
753 | return q->queue_depth; | |
754 | ||
755 | return q->nr_requests; | |
756 | } | |
757 | ||
1da177e4 LT |
758 | extern unsigned long blk_max_low_pfn, blk_max_pfn; |
759 | ||
760 | /* | |
761 | * standard bounce addresses: | |
762 | * | |
763 | * BLK_BOUNCE_HIGH : bounce all highmem pages | |
764 | * BLK_BOUNCE_ANY : don't bounce anything | |
765 | * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary | |
766 | */ | |
2472892a AK |
767 | |
768 | #if BITS_PER_LONG == 32 | |
1da177e4 | 769 | #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT) |
2472892a AK |
770 | #else |
771 | #define BLK_BOUNCE_HIGH -1ULL | |
772 | #endif | |
773 | #define BLK_BOUNCE_ANY (-1ULL) | |
bfe17231 | 774 | #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24)) |
1da177e4 | 775 | |
3d6392cf JA |
776 | /* |
777 | * default timeout for SG_IO if none specified | |
778 | */ | |
779 | #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ) | |
f2f1fa78 | 780 | #define BLK_MIN_SG_TIMEOUT (7 * HZ) |
3d6392cf | 781 | |
152e283f FT |
782 | struct rq_map_data { |
783 | struct page **pages; | |
784 | int page_order; | |
785 | int nr_entries; | |
56c451f4 | 786 | unsigned long offset; |
97ae77a1 | 787 | int null_mapped; |
ecb554a8 | 788 | int from_user; |
152e283f FT |
789 | }; |
790 | ||
5705f702 | 791 | struct req_iterator { |
7988613b | 792 | struct bvec_iter iter; |
5705f702 N |
793 | struct bio *bio; |
794 | }; | |
795 | ||
796 | /* This should not be used directly - use rq_for_each_segment */ | |
1e428079 JA |
797 | #define for_each_bio(_bio) \ |
798 | for (; _bio; _bio = _bio->bi_next) | |
5705f702 | 799 | #define __rq_for_each_bio(_bio, rq) \ |
1da177e4 LT |
800 | if ((rq->bio)) \ |
801 | for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next) | |
802 | ||
5705f702 N |
803 | #define rq_for_each_segment(bvl, _rq, _iter) \ |
804 | __rq_for_each_bio(_iter.bio, _rq) \ | |
7988613b | 805 | bio_for_each_segment(bvl, _iter.bio, _iter.iter) |
5705f702 | 806 | |
d18d9174 ML |
807 | #define rq_for_each_bvec(bvl, _rq, _iter) \ |
808 | __rq_for_each_bio(_iter.bio, _rq) \ | |
809 | bio_for_each_bvec(bvl, _iter.bio, _iter.iter) | |
810 | ||
4550dd6c | 811 | #define rq_iter_last(bvec, _iter) \ |
7988613b | 812 | (_iter.bio->bi_next == NULL && \ |
4550dd6c | 813 | bio_iter_last(bvec, _iter.iter)) |
5705f702 | 814 | |
2d4dc890 IL |
815 | #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE |
816 | # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform" | |
817 | #endif | |
818 | #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE | |
819 | extern void rq_flush_dcache_pages(struct request *rq); | |
820 | #else | |
821 | static inline void rq_flush_dcache_pages(struct request *rq) | |
822 | { | |
823 | } | |
824 | #endif | |
825 | ||
1da177e4 LT |
826 | extern int blk_register_queue(struct gendisk *disk); |
827 | extern void blk_unregister_queue(struct gendisk *disk); | |
dece1635 | 828 | extern blk_qc_t generic_make_request(struct bio *bio); |
f421e1d9 | 829 | extern blk_qc_t direct_make_request(struct bio *bio); |
2a4aa30c | 830 | extern void blk_rq_init(struct request_queue *q, struct request *rq); |
1da177e4 | 831 | extern void blk_put_request(struct request *); |
cd6ce148 | 832 | extern struct request *blk_get_request(struct request_queue *, unsigned int op, |
ff005a06 | 833 | blk_mq_req_flags_t flags); |
ef9e3fac | 834 | extern int blk_lld_busy(struct request_queue *q); |
78d8e58a MS |
835 | extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src, |
836 | struct bio_set *bs, gfp_t gfp_mask, | |
837 | int (*bio_ctr)(struct bio *, struct bio *, void *), | |
838 | void *data); | |
839 | extern void blk_rq_unprep_clone(struct request *rq); | |
2a842aca | 840 | extern blk_status_t blk_insert_cloned_request(struct request_queue *q, |
82124d60 | 841 | struct request *rq); |
0abc2a10 | 842 | extern int blk_rq_append_bio(struct request *rq, struct bio **bio); |
af67c31f | 843 | extern void blk_queue_split(struct request_queue *, struct bio **); |
0bfc96cb | 844 | extern int scsi_verify_blk_ioctl(struct block_device *, unsigned int); |
577ebb37 PB |
845 | extern int scsi_cmd_blk_ioctl(struct block_device *, fmode_t, |
846 | unsigned int, void __user *); | |
74f3c8af AV |
847 | extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t, |
848 | unsigned int, void __user *); | |
e915e872 AV |
849 | extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t, |
850 | struct scsi_ioctl_command __user *); | |
3fcfab16 | 851 | |
9a95e4ef | 852 | extern int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags); |
2e6edc95 | 853 | extern void blk_queue_exit(struct request_queue *q); |
1da177e4 | 854 | extern void blk_sync_queue(struct request_queue *q); |
a3bce90e | 855 | extern int blk_rq_map_user(struct request_queue *, struct request *, |
152e283f FT |
856 | struct rq_map_data *, void __user *, unsigned long, |
857 | gfp_t); | |
8e5cfc45 | 858 | extern int blk_rq_unmap_user(struct bio *); |
165125e1 JA |
859 | extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t); |
860 | extern int blk_rq_map_user_iov(struct request_queue *, struct request *, | |
26e49cfc KO |
861 | struct rq_map_data *, const struct iov_iter *, |
862 | gfp_t); | |
b7819b92 | 863 | extern void blk_execute_rq(struct request_queue *, struct gendisk *, |
994ca9a1 | 864 | struct request *, int); |
165125e1 | 865 | extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *, |
15fc858a | 866 | struct request *, int, rq_end_io_fn *); |
6e39b69e | 867 | |
e47bc4ed CK |
868 | /* Helper to convert REQ_OP_XXX to its string format XXX */ |
869 | extern const char *blk_op_str(unsigned int op); | |
870 | ||
2a842aca CH |
871 | int blk_status_to_errno(blk_status_t status); |
872 | blk_status_t errno_to_blk_status(int errno); | |
873 | ||
0a1b8b87 | 874 | int blk_poll(struct request_queue *q, blk_qc_t cookie, bool spin); |
05229bee | 875 | |
165125e1 | 876 | static inline struct request_queue *bdev_get_queue(struct block_device *bdev) |
1da177e4 | 877 | { |
ff9ea323 | 878 | return bdev->bd_disk->queue; /* this is never NULL */ |
1da177e4 LT |
879 | } |
880 | ||
233bde21 BVA |
881 | /* |
882 | * The basic unit of block I/O is a sector. It is used in a number of contexts | |
883 | * in Linux (blk, bio, genhd). The size of one sector is 512 = 2**9 | |
884 | * bytes. Variables of type sector_t represent an offset or size that is a | |
885 | * multiple of 512 bytes. Hence these two constants. | |
886 | */ | |
887 | #ifndef SECTOR_SHIFT | |
888 | #define SECTOR_SHIFT 9 | |
889 | #endif | |
890 | #ifndef SECTOR_SIZE | |
891 | #define SECTOR_SIZE (1 << SECTOR_SHIFT) | |
892 | #endif | |
893 | ||
5efccd17 | 894 | /* |
80a761fd TH |
895 | * blk_rq_pos() : the current sector |
896 | * blk_rq_bytes() : bytes left in the entire request | |
897 | * blk_rq_cur_bytes() : bytes left in the current segment | |
898 | * blk_rq_err_bytes() : bytes left till the next error boundary | |
899 | * blk_rq_sectors() : sectors left in the entire request | |
900 | * blk_rq_cur_sectors() : sectors left in the current segment | |
5efccd17 | 901 | */ |
5b93629b TH |
902 | static inline sector_t blk_rq_pos(const struct request *rq) |
903 | { | |
a2dec7b3 | 904 | return rq->__sector; |
2e46e8b2 TH |
905 | } |
906 | ||
907 | static inline unsigned int blk_rq_bytes(const struct request *rq) | |
908 | { | |
a2dec7b3 | 909 | return rq->__data_len; |
5b93629b TH |
910 | } |
911 | ||
2e46e8b2 TH |
912 | static inline int blk_rq_cur_bytes(const struct request *rq) |
913 | { | |
914 | return rq->bio ? bio_cur_bytes(rq->bio) : 0; | |
915 | } | |
5efccd17 | 916 | |
80a761fd TH |
917 | extern unsigned int blk_rq_err_bytes(const struct request *rq); |
918 | ||
5b93629b TH |
919 | static inline unsigned int blk_rq_sectors(const struct request *rq) |
920 | { | |
233bde21 | 921 | return blk_rq_bytes(rq) >> SECTOR_SHIFT; |
5b93629b TH |
922 | } |
923 | ||
924 | static inline unsigned int blk_rq_cur_sectors(const struct request *rq) | |
925 | { | |
233bde21 | 926 | return blk_rq_cur_bytes(rq) >> SECTOR_SHIFT; |
5b93629b TH |
927 | } |
928 | ||
6a5ac984 | 929 | #ifdef CONFIG_BLK_DEV_ZONED |
6cc77e9c CH |
930 | static inline unsigned int blk_rq_zone_no(struct request *rq) |
931 | { | |
932 | return blk_queue_zone_no(rq->q, blk_rq_pos(rq)); | |
933 | } | |
934 | ||
935 | static inline unsigned int blk_rq_zone_is_seq(struct request *rq) | |
936 | { | |
937 | return blk_queue_zone_is_seq(rq->q, blk_rq_pos(rq)); | |
938 | } | |
6a5ac984 | 939 | #endif /* CONFIG_BLK_DEV_ZONED */ |
6cc77e9c | 940 | |
2e3258ec CH |
941 | /* |
942 | * Some commands like WRITE SAME have a payload or data transfer size which | |
943 | * is different from the size of the request. Any driver that supports such | |
944 | * commands using the RQF_SPECIAL_PAYLOAD flag needs to use this helper to | |
945 | * calculate the data transfer size. | |
946 | */ | |
947 | static inline unsigned int blk_rq_payload_bytes(struct request *rq) | |
948 | { | |
949 | if (rq->rq_flags & RQF_SPECIAL_PAYLOAD) | |
950 | return rq->special_vec.bv_len; | |
951 | return blk_rq_bytes(rq); | |
952 | } | |
953 | ||
3aef3cae CH |
954 | /* |
955 | * Return the first full biovec in the request. The caller needs to check that | |
956 | * there are any bvecs before calling this helper. | |
957 | */ | |
958 | static inline struct bio_vec req_bvec(struct request *rq) | |
959 | { | |
960 | if (rq->rq_flags & RQF_SPECIAL_PAYLOAD) | |
961 | return rq->special_vec; | |
962 | return mp_bvec_iter_bvec(rq->bio->bi_io_vec, rq->bio->bi_iter); | |
963 | } | |
964 | ||
f31dc1cd | 965 | static inline unsigned int blk_queue_get_max_sectors(struct request_queue *q, |
8fe0d473 | 966 | int op) |
f31dc1cd | 967 | { |
7afafc8a | 968 | if (unlikely(op == REQ_OP_DISCARD || op == REQ_OP_SECURE_ERASE)) |
233bde21 BVA |
969 | return min(q->limits.max_discard_sectors, |
970 | UINT_MAX >> SECTOR_SHIFT); | |
f31dc1cd | 971 | |
8fe0d473 | 972 | if (unlikely(op == REQ_OP_WRITE_SAME)) |
4363ac7c MP |
973 | return q->limits.max_write_same_sectors; |
974 | ||
a6f0788e CK |
975 | if (unlikely(op == REQ_OP_WRITE_ZEROES)) |
976 | return q->limits.max_write_zeroes_sectors; | |
977 | ||
f31dc1cd MP |
978 | return q->limits.max_sectors; |
979 | } | |
980 | ||
762380ad JA |
981 | /* |
982 | * Return maximum size of a request at given offset. Only valid for | |
983 | * file system requests. | |
984 | */ | |
985 | static inline unsigned int blk_max_size_offset(struct request_queue *q, | |
986 | sector_t offset) | |
987 | { | |
988 | if (!q->limits.chunk_sectors) | |
736ed4de | 989 | return q->limits.max_sectors; |
762380ad | 990 | |
15bfd21f KB |
991 | return min(q->limits.max_sectors, (unsigned int)(q->limits.chunk_sectors - |
992 | (offset & (q->limits.chunk_sectors - 1)))); | |
762380ad JA |
993 | } |
994 | ||
17007f39 DLM |
995 | static inline unsigned int blk_rq_get_max_sectors(struct request *rq, |
996 | sector_t offset) | |
f31dc1cd MP |
997 | { |
998 | struct request_queue *q = rq->q; | |
999 | ||
57292b58 | 1000 | if (blk_rq_is_passthrough(rq)) |
f31dc1cd MP |
1001 | return q->limits.max_hw_sectors; |
1002 | ||
7afafc8a AH |
1003 | if (!q->limits.chunk_sectors || |
1004 | req_op(rq) == REQ_OP_DISCARD || | |
1005 | req_op(rq) == REQ_OP_SECURE_ERASE) | |
8fe0d473 | 1006 | return blk_queue_get_max_sectors(q, req_op(rq)); |
762380ad | 1007 | |
17007f39 | 1008 | return min(blk_max_size_offset(q, offset), |
8fe0d473 | 1009 | blk_queue_get_max_sectors(q, req_op(rq))); |
f31dc1cd MP |
1010 | } |
1011 | ||
75afb352 JN |
1012 | static inline unsigned int blk_rq_count_bios(struct request *rq) |
1013 | { | |
1014 | unsigned int nr_bios = 0; | |
1015 | struct bio *bio; | |
1016 | ||
1017 | __rq_for_each_bio(bio, rq) | |
1018 | nr_bios++; | |
1019 | ||
1020 | return nr_bios; | |
1021 | } | |
1022 | ||
ef71de8b CH |
1023 | void blk_steal_bios(struct bio_list *list, struct request *rq); |
1024 | ||
1da177e4 | 1025 | /* |
2e60e022 TH |
1026 | * Request completion related functions. |
1027 | * | |
1028 | * blk_update_request() completes given number of bytes and updates | |
1029 | * the request without completing it. | |
1da177e4 | 1030 | */ |
2a842aca | 1031 | extern bool blk_update_request(struct request *rq, blk_status_t error, |
2e60e022 | 1032 | unsigned int nr_bytes); |
2e60e022 | 1033 | |
242f9dcb JA |
1034 | extern void __blk_complete_request(struct request *); |
1035 | extern void blk_abort_request(struct request *); | |
ff856bad | 1036 | |
1da177e4 LT |
1037 | /* |
1038 | * Access functions for manipulating queue properties | |
1039 | */ | |
165125e1 JA |
1040 | extern void blk_cleanup_queue(struct request_queue *); |
1041 | extern void blk_queue_make_request(struct request_queue *, make_request_fn *); | |
1042 | extern void blk_queue_bounce_limit(struct request_queue *, u64); | |
086fa5ff | 1043 | extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int); |
762380ad | 1044 | extern void blk_queue_chunk_sectors(struct request_queue *, unsigned int); |
8a78362c | 1045 | extern void blk_queue_max_segments(struct request_queue *, unsigned short); |
1e739730 CH |
1046 | extern void blk_queue_max_discard_segments(struct request_queue *, |
1047 | unsigned short); | |
165125e1 | 1048 | extern void blk_queue_max_segment_size(struct request_queue *, unsigned int); |
67efc925 CH |
1049 | extern void blk_queue_max_discard_sectors(struct request_queue *q, |
1050 | unsigned int max_discard_sectors); | |
4363ac7c MP |
1051 | extern void blk_queue_max_write_same_sectors(struct request_queue *q, |
1052 | unsigned int max_write_same_sectors); | |
a6f0788e CK |
1053 | extern void blk_queue_max_write_zeroes_sectors(struct request_queue *q, |
1054 | unsigned int max_write_same_sectors); | |
e1defc4f | 1055 | extern void blk_queue_logical_block_size(struct request_queue *, unsigned short); |
892b6f90 | 1056 | extern void blk_queue_physical_block_size(struct request_queue *, unsigned int); |
c72758f3 MP |
1057 | extern void blk_queue_alignment_offset(struct request_queue *q, |
1058 | unsigned int alignment); | |
7c958e32 | 1059 | extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min); |
c72758f3 | 1060 | extern void blk_queue_io_min(struct request_queue *q, unsigned int min); |
3c5820c7 | 1061 | extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt); |
c72758f3 | 1062 | extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt); |
d278d4a8 | 1063 | extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth); |
e475bba2 | 1064 | extern void blk_set_default_limits(struct queue_limits *lim); |
b1bd055d | 1065 | extern void blk_set_stacking_limits(struct queue_limits *lim); |
c72758f3 MP |
1066 | extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b, |
1067 | sector_t offset); | |
17be8c24 MP |
1068 | extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev, |
1069 | sector_t offset); | |
c72758f3 MP |
1070 | extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev, |
1071 | sector_t offset); | |
165125e1 | 1072 | extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b); |
27f8221a | 1073 | extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int); |
2fb98e84 TH |
1074 | extern int blk_queue_dma_drain(struct request_queue *q, |
1075 | dma_drain_needed_fn *dma_drain_needed, | |
1076 | void *buf, unsigned int size); | |
165125e1 | 1077 | extern void blk_queue_segment_boundary(struct request_queue *, unsigned long); |
03100aad | 1078 | extern void blk_queue_virt_boundary(struct request_queue *, unsigned long); |
165125e1 | 1079 | extern void blk_queue_dma_alignment(struct request_queue *, int); |
11c3e689 | 1080 | extern void blk_queue_update_dma_alignment(struct request_queue *, int); |
242f9dcb | 1081 | extern void blk_queue_rq_timeout(struct request_queue *, unsigned int); |
93e9d8e8 | 1082 | extern void blk_queue_write_cache(struct request_queue *q, bool enabled, bool fua); |
1da177e4 | 1083 | |
1e739730 CH |
1084 | /* |
1085 | * Number of physical segments as sent to the device. | |
1086 | * | |
1087 | * Normally this is the number of discontiguous data segments sent by the | |
1088 | * submitter. But for data-less command like discard we might have no | |
1089 | * actual data segments submitted, but the driver might have to add it's | |
1090 | * own special payload. In that case we still return 1 here so that this | |
1091 | * special payload will be mapped. | |
1092 | */ | |
f9d03f96 CH |
1093 | static inline unsigned short blk_rq_nr_phys_segments(struct request *rq) |
1094 | { | |
1095 | if (rq->rq_flags & RQF_SPECIAL_PAYLOAD) | |
1096 | return 1; | |
1097 | return rq->nr_phys_segments; | |
1098 | } | |
1099 | ||
1e739730 CH |
1100 | /* |
1101 | * Number of discard segments (or ranges) the driver needs to fill in. | |
1102 | * Each discard bio merged into a request is counted as one segment. | |
1103 | */ | |
1104 | static inline unsigned short blk_rq_nr_discard_segments(struct request *rq) | |
1105 | { | |
1106 | return max_t(unsigned short, rq->nr_phys_segments, 1); | |
1107 | } | |
1108 | ||
165125e1 | 1109 | extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *); |
1da177e4 | 1110 | extern void blk_dump_rq_flags(struct request *, char *); |
1da177e4 | 1111 | extern long nr_blockdev_pages(void); |
1da177e4 | 1112 | |
09ac46c4 | 1113 | bool __must_check blk_get_queue(struct request_queue *); |
165125e1 | 1114 | struct request_queue *blk_alloc_queue(gfp_t); |
6d469642 | 1115 | struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id); |
165125e1 | 1116 | extern void blk_put_queue(struct request_queue *); |
3f21c265 | 1117 | extern void blk_set_queue_dying(struct request_queue *); |
1da177e4 | 1118 | |
316cc67d | 1119 | /* |
75df7136 SJ |
1120 | * blk_plug permits building a queue of related requests by holding the I/O |
1121 | * fragments for a short period. This allows merging of sequential requests | |
1122 | * into single larger request. As the requests are moved from a per-task list to | |
1123 | * the device's request_queue in a batch, this results in improved scalability | |
1124 | * as the lock contention for request_queue lock is reduced. | |
1125 | * | |
1126 | * It is ok not to disable preemption when adding the request to the plug list | |
1127 | * or when attempting a merge, because blk_schedule_flush_list() will only flush | |
1128 | * the plug list when the task sleeps by itself. For details, please see | |
1129 | * schedule() where blk_schedule_flush_plug() is called. | |
316cc67d | 1130 | */ |
73c10101 | 1131 | struct blk_plug { |
320ae51f | 1132 | struct list_head mq_list; /* blk-mq requests */ |
75df7136 | 1133 | struct list_head cb_list; /* md requires an unplug callback */ |
5f0ed774 | 1134 | unsigned short rq_count; |
ce5b009c | 1135 | bool multiple_queues; |
73c10101 | 1136 | }; |
55c022bb | 1137 | #define BLK_MAX_REQUEST_COUNT 16 |
50d24c34 | 1138 | #define BLK_PLUG_FLUSH_SIZE (128 * 1024) |
55c022bb | 1139 | |
9cbb1750 | 1140 | struct blk_plug_cb; |
74018dc3 | 1141 | typedef void (*blk_plug_cb_fn)(struct blk_plug_cb *, bool); |
048c9374 N |
1142 | struct blk_plug_cb { |
1143 | struct list_head list; | |
9cbb1750 N |
1144 | blk_plug_cb_fn callback; |
1145 | void *data; | |
048c9374 | 1146 | }; |
9cbb1750 N |
1147 | extern struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug, |
1148 | void *data, int size); | |
73c10101 JA |
1149 | extern void blk_start_plug(struct blk_plug *); |
1150 | extern void blk_finish_plug(struct blk_plug *); | |
f6603783 | 1151 | extern void blk_flush_plug_list(struct blk_plug *, bool); |
73c10101 JA |
1152 | |
1153 | static inline void blk_flush_plug(struct task_struct *tsk) | |
1154 | { | |
1155 | struct blk_plug *plug = tsk->plug; | |
1156 | ||
a237c1c5 JA |
1157 | if (plug) |
1158 | blk_flush_plug_list(plug, false); | |
1159 | } | |
1160 | ||
1161 | static inline void blk_schedule_flush_plug(struct task_struct *tsk) | |
1162 | { | |
1163 | struct blk_plug *plug = tsk->plug; | |
1164 | ||
88b996cd | 1165 | if (plug) |
f6603783 | 1166 | blk_flush_plug_list(plug, true); |
73c10101 JA |
1167 | } |
1168 | ||
1169 | static inline bool blk_needs_flush_plug(struct task_struct *tsk) | |
1170 | { | |
1171 | struct blk_plug *plug = tsk->plug; | |
1172 | ||
320ae51f | 1173 | return plug && |
a1ce35fa | 1174 | (!list_empty(&plug->mq_list) || |
320ae51f | 1175 | !list_empty(&plug->cb_list)); |
73c10101 JA |
1176 | } |
1177 | ||
ee472d83 CH |
1178 | extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *); |
1179 | extern int blkdev_issue_write_same(struct block_device *bdev, sector_t sector, | |
1180 | sector_t nr_sects, gfp_t gfp_mask, struct page *page); | |
e950fdf7 CH |
1181 | |
1182 | #define BLKDEV_DISCARD_SECURE (1 << 0) /* issue a secure erase */ | |
dd3932ed | 1183 | |
fbd9b09a DM |
1184 | extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector, |
1185 | sector_t nr_sects, gfp_t gfp_mask, unsigned long flags); | |
38f25255 | 1186 | extern int __blkdev_issue_discard(struct block_device *bdev, sector_t sector, |
288dab8a | 1187 | sector_t nr_sects, gfp_t gfp_mask, int flags, |
469e3216 | 1188 | struct bio **biop); |
ee472d83 CH |
1189 | |
1190 | #define BLKDEV_ZERO_NOUNMAP (1 << 0) /* do not free blocks */ | |
cb365b96 | 1191 | #define BLKDEV_ZERO_NOFALLBACK (1 << 1) /* don't write explicit zeroes */ |
ee472d83 | 1192 | |
e73c23ff CK |
1193 | extern int __blkdev_issue_zeroout(struct block_device *bdev, sector_t sector, |
1194 | sector_t nr_sects, gfp_t gfp_mask, struct bio **biop, | |
ee472d83 | 1195 | unsigned flags); |
3f14d792 | 1196 | extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector, |
ee472d83 CH |
1197 | sector_t nr_sects, gfp_t gfp_mask, unsigned flags); |
1198 | ||
2cf6d26a CH |
1199 | static inline int sb_issue_discard(struct super_block *sb, sector_t block, |
1200 | sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags) | |
fb2dce86 | 1201 | { |
233bde21 BVA |
1202 | return blkdev_issue_discard(sb->s_bdev, |
1203 | block << (sb->s_blocksize_bits - | |
1204 | SECTOR_SHIFT), | |
1205 | nr_blocks << (sb->s_blocksize_bits - | |
1206 | SECTOR_SHIFT), | |
2cf6d26a | 1207 | gfp_mask, flags); |
fb2dce86 | 1208 | } |
e6fa0be6 | 1209 | static inline int sb_issue_zeroout(struct super_block *sb, sector_t block, |
a107e5a3 | 1210 | sector_t nr_blocks, gfp_t gfp_mask) |
e6fa0be6 LC |
1211 | { |
1212 | return blkdev_issue_zeroout(sb->s_bdev, | |
233bde21 BVA |
1213 | block << (sb->s_blocksize_bits - |
1214 | SECTOR_SHIFT), | |
1215 | nr_blocks << (sb->s_blocksize_bits - | |
1216 | SECTOR_SHIFT), | |
ee472d83 | 1217 | gfp_mask, 0); |
e6fa0be6 | 1218 | } |
1da177e4 | 1219 | |
f00c4d80 | 1220 | extern int blk_verify_command(unsigned char *cmd, fmode_t mode); |
0b07de85 | 1221 | |
eb28d31b MP |
1222 | enum blk_default_limits { |
1223 | BLK_MAX_SEGMENTS = 128, | |
1224 | BLK_SAFE_MAX_SECTORS = 255, | |
d2be537c | 1225 | BLK_DEF_MAX_SECTORS = 2560, |
eb28d31b MP |
1226 | BLK_MAX_SEGMENT_SIZE = 65536, |
1227 | BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL, | |
1228 | }; | |
0e435ac2 | 1229 | |
ae03bf63 MP |
1230 | static inline unsigned long queue_segment_boundary(struct request_queue *q) |
1231 | { | |
025146e1 | 1232 | return q->limits.seg_boundary_mask; |
ae03bf63 MP |
1233 | } |
1234 | ||
03100aad KB |
1235 | static inline unsigned long queue_virt_boundary(struct request_queue *q) |
1236 | { | |
1237 | return q->limits.virt_boundary_mask; | |
1238 | } | |
1239 | ||
ae03bf63 MP |
1240 | static inline unsigned int queue_max_sectors(struct request_queue *q) |
1241 | { | |
025146e1 | 1242 | return q->limits.max_sectors; |
ae03bf63 MP |
1243 | } |
1244 | ||
1245 | static inline unsigned int queue_max_hw_sectors(struct request_queue *q) | |
1246 | { | |
025146e1 | 1247 | return q->limits.max_hw_sectors; |
ae03bf63 MP |
1248 | } |
1249 | ||
8a78362c | 1250 | static inline unsigned short queue_max_segments(struct request_queue *q) |
ae03bf63 | 1251 | { |
8a78362c | 1252 | return q->limits.max_segments; |
ae03bf63 MP |
1253 | } |
1254 | ||
1e739730 CH |
1255 | static inline unsigned short queue_max_discard_segments(struct request_queue *q) |
1256 | { | |
1257 | return q->limits.max_discard_segments; | |
1258 | } | |
1259 | ||
ae03bf63 MP |
1260 | static inline unsigned int queue_max_segment_size(struct request_queue *q) |
1261 | { | |
025146e1 | 1262 | return q->limits.max_segment_size; |
ae03bf63 MP |
1263 | } |
1264 | ||
e1defc4f | 1265 | static inline unsigned short queue_logical_block_size(struct request_queue *q) |
1da177e4 LT |
1266 | { |
1267 | int retval = 512; | |
1268 | ||
025146e1 MP |
1269 | if (q && q->limits.logical_block_size) |
1270 | retval = q->limits.logical_block_size; | |
1da177e4 LT |
1271 | |
1272 | return retval; | |
1273 | } | |
1274 | ||
e1defc4f | 1275 | static inline unsigned short bdev_logical_block_size(struct block_device *bdev) |
1da177e4 | 1276 | { |
e1defc4f | 1277 | return queue_logical_block_size(bdev_get_queue(bdev)); |
1da177e4 LT |
1278 | } |
1279 | ||
c72758f3 MP |
1280 | static inline unsigned int queue_physical_block_size(struct request_queue *q) |
1281 | { | |
1282 | return q->limits.physical_block_size; | |
1283 | } | |
1284 | ||
892b6f90 | 1285 | static inline unsigned int bdev_physical_block_size(struct block_device *bdev) |
ac481c20 MP |
1286 | { |
1287 | return queue_physical_block_size(bdev_get_queue(bdev)); | |
1288 | } | |
1289 | ||
c72758f3 MP |
1290 | static inline unsigned int queue_io_min(struct request_queue *q) |
1291 | { | |
1292 | return q->limits.io_min; | |
1293 | } | |
1294 | ||
ac481c20 MP |
1295 | static inline int bdev_io_min(struct block_device *bdev) |
1296 | { | |
1297 | return queue_io_min(bdev_get_queue(bdev)); | |
1298 | } | |
1299 | ||
c72758f3 MP |
1300 | static inline unsigned int queue_io_opt(struct request_queue *q) |
1301 | { | |
1302 | return q->limits.io_opt; | |
1303 | } | |
1304 | ||
ac481c20 MP |
1305 | static inline int bdev_io_opt(struct block_device *bdev) |
1306 | { | |
1307 | return queue_io_opt(bdev_get_queue(bdev)); | |
1308 | } | |
1309 | ||
c72758f3 MP |
1310 | static inline int queue_alignment_offset(struct request_queue *q) |
1311 | { | |
ac481c20 | 1312 | if (q->limits.misaligned) |
c72758f3 MP |
1313 | return -1; |
1314 | ||
ac481c20 | 1315 | return q->limits.alignment_offset; |
c72758f3 MP |
1316 | } |
1317 | ||
e03a72e1 | 1318 | static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector) |
81744ee4 MP |
1319 | { |
1320 | unsigned int granularity = max(lim->physical_block_size, lim->io_min); | |
233bde21 BVA |
1321 | unsigned int alignment = sector_div(sector, granularity >> SECTOR_SHIFT) |
1322 | << SECTOR_SHIFT; | |
81744ee4 | 1323 | |
b8839b8c | 1324 | return (granularity + lim->alignment_offset - alignment) % granularity; |
c72758f3 MP |
1325 | } |
1326 | ||
ac481c20 MP |
1327 | static inline int bdev_alignment_offset(struct block_device *bdev) |
1328 | { | |
1329 | struct request_queue *q = bdev_get_queue(bdev); | |
1330 | ||
1331 | if (q->limits.misaligned) | |
1332 | return -1; | |
1333 | ||
1334 | if (bdev != bdev->bd_contains) | |
1335 | return bdev->bd_part->alignment_offset; | |
1336 | ||
1337 | return q->limits.alignment_offset; | |
1338 | } | |
1339 | ||
86b37281 MP |
1340 | static inline int queue_discard_alignment(struct request_queue *q) |
1341 | { | |
1342 | if (q->limits.discard_misaligned) | |
1343 | return -1; | |
1344 | ||
1345 | return q->limits.discard_alignment; | |
1346 | } | |
1347 | ||
e03a72e1 | 1348 | static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector) |
86b37281 | 1349 | { |
59771079 | 1350 | unsigned int alignment, granularity, offset; |
dd3d145d | 1351 | |
a934a00a MP |
1352 | if (!lim->max_discard_sectors) |
1353 | return 0; | |
1354 | ||
59771079 | 1355 | /* Why are these in bytes, not sectors? */ |
233bde21 BVA |
1356 | alignment = lim->discard_alignment >> SECTOR_SHIFT; |
1357 | granularity = lim->discard_granularity >> SECTOR_SHIFT; | |
59771079 LT |
1358 | if (!granularity) |
1359 | return 0; | |
1360 | ||
1361 | /* Offset of the partition start in 'granularity' sectors */ | |
1362 | offset = sector_div(sector, granularity); | |
1363 | ||
1364 | /* And why do we do this modulus *again* in blkdev_issue_discard()? */ | |
1365 | offset = (granularity + alignment - offset) % granularity; | |
1366 | ||
1367 | /* Turn it back into bytes, gaah */ | |
233bde21 | 1368 | return offset << SECTOR_SHIFT; |
86b37281 MP |
1369 | } |
1370 | ||
c6e66634 PB |
1371 | static inline int bdev_discard_alignment(struct block_device *bdev) |
1372 | { | |
1373 | struct request_queue *q = bdev_get_queue(bdev); | |
1374 | ||
1375 | if (bdev != bdev->bd_contains) | |
1376 | return bdev->bd_part->discard_alignment; | |
1377 | ||
1378 | return q->limits.discard_alignment; | |
1379 | } | |
1380 | ||
4363ac7c MP |
1381 | static inline unsigned int bdev_write_same(struct block_device *bdev) |
1382 | { | |
1383 | struct request_queue *q = bdev_get_queue(bdev); | |
1384 | ||
1385 | if (q) | |
1386 | return q->limits.max_write_same_sectors; | |
1387 | ||
1388 | return 0; | |
1389 | } | |
1390 | ||
a6f0788e CK |
1391 | static inline unsigned int bdev_write_zeroes_sectors(struct block_device *bdev) |
1392 | { | |
1393 | struct request_queue *q = bdev_get_queue(bdev); | |
1394 | ||
1395 | if (q) | |
1396 | return q->limits.max_write_zeroes_sectors; | |
1397 | ||
1398 | return 0; | |
1399 | } | |
1400 | ||
797476b8 DLM |
1401 | static inline enum blk_zoned_model bdev_zoned_model(struct block_device *bdev) |
1402 | { | |
1403 | struct request_queue *q = bdev_get_queue(bdev); | |
1404 | ||
1405 | if (q) | |
1406 | return blk_queue_zoned_model(q); | |
1407 | ||
1408 | return BLK_ZONED_NONE; | |
1409 | } | |
1410 | ||
1411 | static inline bool bdev_is_zoned(struct block_device *bdev) | |
1412 | { | |
1413 | struct request_queue *q = bdev_get_queue(bdev); | |
1414 | ||
1415 | if (q) | |
1416 | return blk_queue_is_zoned(q); | |
1417 | ||
1418 | return false; | |
1419 | } | |
1420 | ||
f99e8648 | 1421 | static inline unsigned int bdev_zone_sectors(struct block_device *bdev) |
6a0cb1bc HR |
1422 | { |
1423 | struct request_queue *q = bdev_get_queue(bdev); | |
1424 | ||
1425 | if (q) | |
f99e8648 | 1426 | return blk_queue_zone_sectors(q); |
6cc77e9c CH |
1427 | return 0; |
1428 | } | |
6a0cb1bc | 1429 | |
165125e1 | 1430 | static inline int queue_dma_alignment(struct request_queue *q) |
1da177e4 | 1431 | { |
482eb689 | 1432 | return q ? q->dma_alignment : 511; |
1da177e4 LT |
1433 | } |
1434 | ||
14417799 | 1435 | static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr, |
87904074 FT |
1436 | unsigned int len) |
1437 | { | |
1438 | unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask; | |
14417799 | 1439 | return !(addr & alignment) && !(len & alignment); |
87904074 FT |
1440 | } |
1441 | ||
1da177e4 LT |
1442 | /* assumes size > 256 */ |
1443 | static inline unsigned int blksize_bits(unsigned int size) | |
1444 | { | |
1445 | unsigned int bits = 8; | |
1446 | do { | |
1447 | bits++; | |
1448 | size >>= 1; | |
1449 | } while (size > 256); | |
1450 | return bits; | |
1451 | } | |
1452 | ||
2befb9e3 | 1453 | static inline unsigned int block_size(struct block_device *bdev) |
1da177e4 LT |
1454 | { |
1455 | return bdev->bd_block_size; | |
1456 | } | |
1457 | ||
1458 | typedef struct {struct page *v;} Sector; | |
1459 | ||
1460 | unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *); | |
1461 | ||
1462 | static inline void put_dev_sector(Sector p) | |
1463 | { | |
09cbfeaf | 1464 | put_page(p.v); |
1da177e4 LT |
1465 | } |
1466 | ||
59c3d45e | 1467 | int kblockd_schedule_work(struct work_struct *work); |
ee63cfa7 | 1468 | int kblockd_schedule_work_on(int cpu, struct work_struct *work); |
818cd1cb | 1469 | int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork, unsigned long delay); |
1da177e4 | 1470 | |
1da177e4 LT |
1471 | #define MODULE_ALIAS_BLOCKDEV(major,minor) \ |
1472 | MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor)) | |
1473 | #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \ | |
1474 | MODULE_ALIAS("block-major-" __stringify(major) "-*") | |
1475 | ||
7ba1ba12 MP |
1476 | #if defined(CONFIG_BLK_DEV_INTEGRITY) |
1477 | ||
8288f496 MP |
1478 | enum blk_integrity_flags { |
1479 | BLK_INTEGRITY_VERIFY = 1 << 0, | |
1480 | BLK_INTEGRITY_GENERATE = 1 << 1, | |
3aec2f41 | 1481 | BLK_INTEGRITY_DEVICE_CAPABLE = 1 << 2, |
aae7df50 | 1482 | BLK_INTEGRITY_IP_CHECKSUM = 1 << 3, |
8288f496 | 1483 | }; |
7ba1ba12 | 1484 | |
18593088 | 1485 | struct blk_integrity_iter { |
7ba1ba12 MP |
1486 | void *prot_buf; |
1487 | void *data_buf; | |
3be91c4a | 1488 | sector_t seed; |
7ba1ba12 | 1489 | unsigned int data_size; |
3be91c4a | 1490 | unsigned short interval; |
7ba1ba12 MP |
1491 | const char *disk_name; |
1492 | }; | |
1493 | ||
4e4cbee9 | 1494 | typedef blk_status_t (integrity_processing_fn) (struct blk_integrity_iter *); |
7ba1ba12 | 1495 | |
0f8087ec MP |
1496 | struct blk_integrity_profile { |
1497 | integrity_processing_fn *generate_fn; | |
1498 | integrity_processing_fn *verify_fn; | |
1499 | const char *name; | |
1500 | }; | |
7ba1ba12 | 1501 | |
25520d55 | 1502 | extern void blk_integrity_register(struct gendisk *, struct blk_integrity *); |
7ba1ba12 | 1503 | extern void blk_integrity_unregister(struct gendisk *); |
ad7fce93 | 1504 | extern int blk_integrity_compare(struct gendisk *, struct gendisk *); |
13f05c8d MP |
1505 | extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *, |
1506 | struct scatterlist *); | |
1507 | extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *); | |
4eaf99be MP |
1508 | extern bool blk_integrity_merge_rq(struct request_queue *, struct request *, |
1509 | struct request *); | |
1510 | extern bool blk_integrity_merge_bio(struct request_queue *, struct request *, | |
1511 | struct bio *); | |
7ba1ba12 | 1512 | |
25520d55 | 1513 | static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk) |
b04accc4 | 1514 | { |
ac6fc48c | 1515 | struct blk_integrity *bi = &disk->queue->integrity; |
25520d55 MP |
1516 | |
1517 | if (!bi->profile) | |
1518 | return NULL; | |
1519 | ||
1520 | return bi; | |
b04accc4 JA |
1521 | } |
1522 | ||
25520d55 MP |
1523 | static inline |
1524 | struct blk_integrity *bdev_get_integrity(struct block_device *bdev) | |
b02739b0 | 1525 | { |
25520d55 | 1526 | return blk_get_integrity(bdev->bd_disk); |
b02739b0 MP |
1527 | } |
1528 | ||
180b2f95 | 1529 | static inline bool blk_integrity_rq(struct request *rq) |
7ba1ba12 | 1530 | { |
180b2f95 | 1531 | return rq->cmd_flags & REQ_INTEGRITY; |
7ba1ba12 MP |
1532 | } |
1533 | ||
13f05c8d MP |
1534 | static inline void blk_queue_max_integrity_segments(struct request_queue *q, |
1535 | unsigned int segs) | |
1536 | { | |
1537 | q->limits.max_integrity_segments = segs; | |
1538 | } | |
1539 | ||
1540 | static inline unsigned short | |
1541 | queue_max_integrity_segments(struct request_queue *q) | |
1542 | { | |
1543 | return q->limits.max_integrity_segments; | |
1544 | } | |
1545 | ||
359f6427 GE |
1546 | /** |
1547 | * bio_integrity_intervals - Return number of integrity intervals for a bio | |
1548 | * @bi: blk_integrity profile for device | |
1549 | * @sectors: Size of the bio in 512-byte sectors | |
1550 | * | |
1551 | * Description: The block layer calculates everything in 512 byte | |
1552 | * sectors but integrity metadata is done in terms of the data integrity | |
1553 | * interval size of the storage device. Convert the block layer sectors | |
1554 | * to the appropriate number of integrity intervals. | |
1555 | */ | |
1556 | static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi, | |
1557 | unsigned int sectors) | |
1558 | { | |
1559 | return sectors >> (bi->interval_exp - 9); | |
1560 | } | |
1561 | ||
1562 | static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi, | |
1563 | unsigned int sectors) | |
1564 | { | |
1565 | return bio_integrity_intervals(bi, sectors) * bi->tuple_size; | |
1566 | } | |
1567 | ||
2a876f5e CH |
1568 | /* |
1569 | * Return the first bvec that contains integrity data. Only drivers that are | |
1570 | * limited to a single integrity segment should use this helper. | |
1571 | */ | |
1572 | static inline struct bio_vec *rq_integrity_vec(struct request *rq) | |
1573 | { | |
1574 | if (WARN_ON_ONCE(queue_max_integrity_segments(rq->q) > 1)) | |
1575 | return NULL; | |
1576 | return rq->bio->bi_integrity->bip_vec; | |
1577 | } | |
1578 | ||
7ba1ba12 MP |
1579 | #else /* CONFIG_BLK_DEV_INTEGRITY */ |
1580 | ||
fd83240a SR |
1581 | struct bio; |
1582 | struct block_device; | |
1583 | struct gendisk; | |
1584 | struct blk_integrity; | |
1585 | ||
1586 | static inline int blk_integrity_rq(struct request *rq) | |
1587 | { | |
1588 | return 0; | |
1589 | } | |
1590 | static inline int blk_rq_count_integrity_sg(struct request_queue *q, | |
1591 | struct bio *b) | |
1592 | { | |
1593 | return 0; | |
1594 | } | |
1595 | static inline int blk_rq_map_integrity_sg(struct request_queue *q, | |
1596 | struct bio *b, | |
1597 | struct scatterlist *s) | |
1598 | { | |
1599 | return 0; | |
1600 | } | |
1601 | static inline struct blk_integrity *bdev_get_integrity(struct block_device *b) | |
1602 | { | |
61a04e5b | 1603 | return NULL; |
fd83240a SR |
1604 | } |
1605 | static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk) | |
1606 | { | |
1607 | return NULL; | |
1608 | } | |
1609 | static inline int blk_integrity_compare(struct gendisk *a, struct gendisk *b) | |
1610 | { | |
1611 | return 0; | |
1612 | } | |
25520d55 | 1613 | static inline void blk_integrity_register(struct gendisk *d, |
fd83240a SR |
1614 | struct blk_integrity *b) |
1615 | { | |
fd83240a SR |
1616 | } |
1617 | static inline void blk_integrity_unregister(struct gendisk *d) | |
1618 | { | |
1619 | } | |
1620 | static inline void blk_queue_max_integrity_segments(struct request_queue *q, | |
1621 | unsigned int segs) | |
1622 | { | |
1623 | } | |
1624 | static inline unsigned short queue_max_integrity_segments(struct request_queue *q) | |
1625 | { | |
1626 | return 0; | |
1627 | } | |
4eaf99be MP |
1628 | static inline bool blk_integrity_merge_rq(struct request_queue *rq, |
1629 | struct request *r1, | |
1630 | struct request *r2) | |
fd83240a | 1631 | { |
cb1a5ab6 | 1632 | return true; |
fd83240a | 1633 | } |
4eaf99be MP |
1634 | static inline bool blk_integrity_merge_bio(struct request_queue *rq, |
1635 | struct request *r, | |
1636 | struct bio *b) | |
fd83240a | 1637 | { |
cb1a5ab6 | 1638 | return true; |
fd83240a | 1639 | } |
25520d55 | 1640 | |
359f6427 GE |
1641 | static inline unsigned int bio_integrity_intervals(struct blk_integrity *bi, |
1642 | unsigned int sectors) | |
1643 | { | |
1644 | return 0; | |
1645 | } | |
1646 | ||
1647 | static inline unsigned int bio_integrity_bytes(struct blk_integrity *bi, | |
1648 | unsigned int sectors) | |
1649 | { | |
1650 | return 0; | |
1651 | } | |
1652 | ||
2a876f5e CH |
1653 | static inline struct bio_vec *rq_integrity_vec(struct request *rq) |
1654 | { | |
1655 | return NULL; | |
1656 | } | |
1657 | ||
7ba1ba12 MP |
1658 | #endif /* CONFIG_BLK_DEV_INTEGRITY */ |
1659 | ||
08f85851 | 1660 | struct block_device_operations { |
d4430d62 | 1661 | int (*open) (struct block_device *, fmode_t); |
db2a144b | 1662 | void (*release) (struct gendisk *, fmode_t); |
3f289dcb | 1663 | int (*rw_page)(struct block_device *, sector_t, struct page *, unsigned int); |
d4430d62 AV |
1664 | int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long); |
1665 | int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long); | |
77ea887e TH |
1666 | unsigned int (*check_events) (struct gendisk *disk, |
1667 | unsigned int clearing); | |
1668 | /* ->media_changed() is DEPRECATED, use ->check_events() instead */ | |
08f85851 | 1669 | int (*media_changed) (struct gendisk *); |
c3e33e04 | 1670 | void (*unlock_native_capacity) (struct gendisk *); |
08f85851 AV |
1671 | int (*revalidate_disk) (struct gendisk *); |
1672 | int (*getgeo)(struct block_device *, struct hd_geometry *); | |
b3a27d05 NG |
1673 | /* this callback is with swap_lock and sometimes page table lock held */ |
1674 | void (*swap_slot_free_notify) (struct block_device *, unsigned long); | |
e76239a3 CH |
1675 | int (*report_zones)(struct gendisk *, sector_t sector, |
1676 | struct blk_zone *zones, unsigned int *nr_zones, | |
1677 | gfp_t gfp_mask); | |
08f85851 | 1678 | struct module *owner; |
bbd3e064 | 1679 | const struct pr_ops *pr_ops; |
08f85851 AV |
1680 | }; |
1681 | ||
633a08b8 AV |
1682 | extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int, |
1683 | unsigned long); | |
47a191fd MW |
1684 | extern int bdev_read_page(struct block_device *, sector_t, struct page *); |
1685 | extern int bdev_write_page(struct block_device *, sector_t, struct page *, | |
1686 | struct writeback_control *); | |
6cc77e9c CH |
1687 | |
1688 | #ifdef CONFIG_BLK_DEV_ZONED | |
1689 | bool blk_req_needs_zone_write_lock(struct request *rq); | |
1690 | void __blk_req_zone_write_lock(struct request *rq); | |
1691 | void __blk_req_zone_write_unlock(struct request *rq); | |
1692 | ||
1693 | static inline void blk_req_zone_write_lock(struct request *rq) | |
1694 | { | |
1695 | if (blk_req_needs_zone_write_lock(rq)) | |
1696 | __blk_req_zone_write_lock(rq); | |
1697 | } | |
1698 | ||
1699 | static inline void blk_req_zone_write_unlock(struct request *rq) | |
1700 | { | |
1701 | if (rq->rq_flags & RQF_ZONE_WRITE_LOCKED) | |
1702 | __blk_req_zone_write_unlock(rq); | |
1703 | } | |
1704 | ||
1705 | static inline bool blk_req_zone_is_write_locked(struct request *rq) | |
1706 | { | |
1707 | return rq->q->seq_zones_wlock && | |
1708 | test_bit(blk_rq_zone_no(rq), rq->q->seq_zones_wlock); | |
1709 | } | |
1710 | ||
1711 | static inline bool blk_req_can_dispatch_to_zone(struct request *rq) | |
1712 | { | |
1713 | if (!blk_req_needs_zone_write_lock(rq)) | |
1714 | return true; | |
1715 | return !blk_req_zone_is_write_locked(rq); | |
1716 | } | |
1717 | #else | |
1718 | static inline bool blk_req_needs_zone_write_lock(struct request *rq) | |
1719 | { | |
1720 | return false; | |
1721 | } | |
1722 | ||
1723 | static inline void blk_req_zone_write_lock(struct request *rq) | |
1724 | { | |
1725 | } | |
1726 | ||
1727 | static inline void blk_req_zone_write_unlock(struct request *rq) | |
1728 | { | |
1729 | } | |
1730 | static inline bool blk_req_zone_is_write_locked(struct request *rq) | |
1731 | { | |
1732 | return false; | |
1733 | } | |
1734 | ||
1735 | static inline bool blk_req_can_dispatch_to_zone(struct request *rq) | |
1736 | { | |
1737 | return true; | |
1738 | } | |
1739 | #endif /* CONFIG_BLK_DEV_ZONED */ | |
1740 | ||
9361401e | 1741 | #else /* CONFIG_BLOCK */ |
ac13a829 FF |
1742 | |
1743 | struct block_device; | |
1744 | ||
9361401e DH |
1745 | /* |
1746 | * stubs for when the block layer is configured out | |
1747 | */ | |
1748 | #define buffer_heads_over_limit 0 | |
1749 | ||
9361401e DH |
1750 | static inline long nr_blockdev_pages(void) |
1751 | { | |
1752 | return 0; | |
1753 | } | |
1754 | ||
1f940bdf JA |
1755 | struct blk_plug { |
1756 | }; | |
1757 | ||
1758 | static inline void blk_start_plug(struct blk_plug *plug) | |
73c10101 JA |
1759 | { |
1760 | } | |
1761 | ||
1f940bdf | 1762 | static inline void blk_finish_plug(struct blk_plug *plug) |
73c10101 JA |
1763 | { |
1764 | } | |
1765 | ||
1f940bdf | 1766 | static inline void blk_flush_plug(struct task_struct *task) |
73c10101 JA |
1767 | { |
1768 | } | |
1769 | ||
a237c1c5 JA |
1770 | static inline void blk_schedule_flush_plug(struct task_struct *task) |
1771 | { | |
1772 | } | |
1773 | ||
1774 | ||
73c10101 JA |
1775 | static inline bool blk_needs_flush_plug(struct task_struct *tsk) |
1776 | { | |
1777 | return false; | |
1778 | } | |
1779 | ||
ac13a829 FF |
1780 | static inline int blkdev_issue_flush(struct block_device *bdev, gfp_t gfp_mask, |
1781 | sector_t *error_sector) | |
1782 | { | |
1783 | return 0; | |
1784 | } | |
1785 | ||
9361401e DH |
1786 | #endif /* CONFIG_BLOCK */ |
1787 | ||
0619317f JA |
1788 | static inline void blk_wake_io_task(struct task_struct *waiter) |
1789 | { | |
1790 | /* | |
1791 | * If we're polling, the task itself is doing the completions. For | |
1792 | * that case, we don't need to signal a wakeup, it's enough to just | |
1793 | * mark us as RUNNING. | |
1794 | */ | |
1795 | if (waiter == current) | |
1796 | __set_current_state(TASK_RUNNING); | |
1797 | else | |
1798 | wake_up_process(waiter); | |
1799 | } | |
1800 | ||
1da177e4 | 1801 | #endif |