blk-throttle: Do not use kblockd workqueue for throtl work
[linux-2.6/linux-acpi-2.6/ibm-acpi-2.6.git] / include / linux / blkdev.h
blob678c440cfaadb72df771f68b0c5ed89c1298194f
1 #ifndef _LINUX_BLKDEV_H
2 #define _LINUX_BLKDEV_H
4 #ifdef CONFIG_BLOCK
6 #include <linux/sched.h>
7 #include <linux/major.h>
8 #include <linux/genhd.h>
9 #include <linux/list.h>
10 #include <linux/timer.h>
11 #include <linux/workqueue.h>
12 #include <linux/pagemap.h>
13 #include <linux/backing-dev.h>
14 #include <linux/wait.h>
15 #include <linux/mempool.h>
16 #include <linux/bio.h>
17 #include <linux/module.h>
18 #include <linux/stringify.h>
19 #include <linux/gfp.h>
20 #include <linux/bsg.h>
21 #include <linux/smp.h>
23 #include <asm/scatterlist.h>
25 struct scsi_ioctl_command;
27 struct request_queue;
28 struct elevator_queue;
29 struct request_pm_state;
30 struct blk_trace;
31 struct request;
32 struct sg_io_hdr;
34 #define BLKDEV_MIN_RQ 4
35 #define BLKDEV_MAX_RQ 128 /* Default maximum */
37 struct request;
38 typedef void (rq_end_io_fn)(struct request *, int);
40 struct request_list {
42 * count[], starved[], and wait[] are indexed by
43 * BLK_RW_SYNC/BLK_RW_ASYNC
45 int count[2];
46 int starved[2];
47 int elvpriv;
48 mempool_t *rq_pool;
49 wait_queue_head_t wait[2];
53 * request command types
55 enum rq_cmd_type_bits {
56 REQ_TYPE_FS = 1, /* fs request */
57 REQ_TYPE_BLOCK_PC, /* scsi command */
58 REQ_TYPE_SENSE, /* sense request */
59 REQ_TYPE_PM_SUSPEND, /* suspend request */
60 REQ_TYPE_PM_RESUME, /* resume request */
61 REQ_TYPE_PM_SHUTDOWN, /* shutdown request */
62 REQ_TYPE_SPECIAL, /* driver defined type */
64 * for ATA/ATAPI devices. this really doesn't belong here, ide should
65 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
66 * private REQ_LB opcodes to differentiate what type of request this is
68 REQ_TYPE_ATA_TASKFILE,
69 REQ_TYPE_ATA_PC,
72 #define BLK_MAX_CDB 16
75 * try to put the fields that are referenced together in the same cacheline.
76 * if you modify this structure, be sure to check block/blk-core.c:rq_init()
77 * as well!
79 struct request {
80 struct list_head queuelist;
81 struct call_single_data csd;
83 struct request_queue *q;
85 unsigned int cmd_flags;
86 enum rq_cmd_type_bits cmd_type;
87 unsigned long atomic_flags;
89 int cpu;
91 /* the following two fields are internal, NEVER access directly */
92 unsigned int __data_len; /* total data len */
93 sector_t __sector; /* sector cursor */
95 struct bio *bio;
96 struct bio *biotail;
98 struct hlist_node hash; /* merge hash */
100 * The rb_node is only used inside the io scheduler, requests
101 * are pruned when moved to the dispatch queue. So let the
102 * completion_data share space with the rb_node.
104 union {
105 struct rb_node rb_node; /* sort/lookup */
106 void *completion_data;
110 * Three pointers are available for the IO schedulers, if they need
111 * more they have to dynamically allocate it.
113 void *elevator_private;
114 void *elevator_private2;
115 void *elevator_private3;
117 struct gendisk *rq_disk;
118 struct hd_struct *part;
119 unsigned long start_time;
120 #ifdef CONFIG_BLK_CGROUP
121 unsigned long long start_time_ns;
122 unsigned long long io_start_time_ns; /* when passed to hardware */
123 #endif
124 /* Number of scatter-gather DMA addr+len pairs after
125 * physical address coalescing is performed.
127 unsigned short nr_phys_segments;
128 #if defined(CONFIG_BLK_DEV_INTEGRITY)
129 unsigned short nr_integrity_segments;
130 #endif
132 unsigned short ioprio;
134 int ref_count;
136 void *special; /* opaque pointer available for LLD use */
137 char *buffer; /* kaddr of the current segment if available */
139 int tag;
140 int errors;
143 * when request is used as a packet command carrier
145 unsigned char __cmd[BLK_MAX_CDB];
146 unsigned char *cmd;
147 unsigned short cmd_len;
149 unsigned int extra_len; /* length of alignment and padding */
150 unsigned int sense_len;
151 unsigned int resid_len; /* residual count */
152 void *sense;
154 unsigned long deadline;
155 struct list_head timeout_list;
156 unsigned int timeout;
157 int retries;
160 * completion callback.
162 rq_end_io_fn *end_io;
163 void *end_io_data;
165 /* for bidi */
166 struct request *next_rq;
169 static inline unsigned short req_get_ioprio(struct request *req)
171 return req->ioprio;
175 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
176 * requests. Some step values could eventually be made generic.
178 struct request_pm_state
180 /* PM state machine step value, currently driver specific */
181 int pm_step;
182 /* requested PM state value (S1, S2, S3, S4, ...) */
183 u32 pm_state;
184 void* data; /* for driver use */
187 #include <linux/elevator.h>
189 typedef void (request_fn_proc) (struct request_queue *q);
190 typedef int (make_request_fn) (struct request_queue *q, struct bio *bio);
191 typedef int (prep_rq_fn) (struct request_queue *, struct request *);
192 typedef void (unprep_rq_fn) (struct request_queue *, struct request *);
193 typedef void (unplug_fn) (struct request_queue *);
195 struct bio_vec;
196 struct bvec_merge_data {
197 struct block_device *bi_bdev;
198 sector_t bi_sector;
199 unsigned bi_size;
200 unsigned long bi_rw;
202 typedef int (merge_bvec_fn) (struct request_queue *, struct bvec_merge_data *,
203 struct bio_vec *);
204 typedef void (softirq_done_fn)(struct request *);
205 typedef int (dma_drain_needed_fn)(struct request *);
206 typedef int (lld_busy_fn) (struct request_queue *q);
208 enum blk_eh_timer_return {
209 BLK_EH_NOT_HANDLED,
210 BLK_EH_HANDLED,
211 BLK_EH_RESET_TIMER,
214 typedef enum blk_eh_timer_return (rq_timed_out_fn)(struct request *);
216 enum blk_queue_state {
217 Queue_down,
218 Queue_up,
221 struct blk_queue_tag {
222 struct request **tag_index; /* map of busy tags */
223 unsigned long *tag_map; /* bit map of free/busy tags */
224 int busy; /* current depth */
225 int max_depth; /* what we will send to device */
226 int real_max_depth; /* what the array can hold */
227 atomic_t refcnt; /* map can be shared */
230 #define BLK_SCSI_MAX_CMDS (256)
231 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
233 struct queue_limits {
234 unsigned long bounce_pfn;
235 unsigned long seg_boundary_mask;
237 unsigned int max_hw_sectors;
238 unsigned int max_sectors;
239 unsigned int max_segment_size;
240 unsigned int physical_block_size;
241 unsigned int alignment_offset;
242 unsigned int io_min;
243 unsigned int io_opt;
244 unsigned int max_discard_sectors;
245 unsigned int discard_granularity;
246 unsigned int discard_alignment;
248 unsigned short logical_block_size;
249 unsigned short max_segments;
250 unsigned short max_integrity_segments;
252 unsigned char misaligned;
253 unsigned char discard_misaligned;
254 unsigned char cluster;
255 signed char discard_zeroes_data;
258 struct request_queue
261 * Together with queue_head for cacheline sharing
263 struct list_head queue_head;
264 struct request *last_merge;
265 struct elevator_queue *elevator;
268 * the queue request freelist, one for reads and one for writes
270 struct request_list rq;
272 request_fn_proc *request_fn;
273 make_request_fn *make_request_fn;
274 prep_rq_fn *prep_rq_fn;
275 unprep_rq_fn *unprep_rq_fn;
276 unplug_fn *unplug_fn;
277 merge_bvec_fn *merge_bvec_fn;
278 softirq_done_fn *softirq_done_fn;
279 rq_timed_out_fn *rq_timed_out_fn;
280 dma_drain_needed_fn *dma_drain_needed;
281 lld_busy_fn *lld_busy_fn;
284 * Dispatch queue sorting
286 sector_t end_sector;
287 struct request *boundary_rq;
290 * Auto-unplugging state
292 struct timer_list unplug_timer;
293 int unplug_thresh; /* After this many requests */
294 unsigned long unplug_delay; /* After this many jiffies */
295 struct work_struct unplug_work;
297 struct backing_dev_info backing_dev_info;
300 * The queue owner gets to use this for whatever they like.
301 * ll_rw_blk doesn't touch it.
303 void *queuedata;
306 * queue needs bounce pages for pages above this limit
308 gfp_t bounce_gfp;
311 * various queue flags, see QUEUE_* below
313 unsigned long queue_flags;
316 * protects queue structures from reentrancy. ->__queue_lock should
317 * _never_ be used directly, it is queue private. always use
318 * ->queue_lock.
320 spinlock_t __queue_lock;
321 spinlock_t *queue_lock;
324 * queue kobject
326 struct kobject kobj;
329 * queue settings
331 unsigned long nr_requests; /* Max # of requests */
332 unsigned int nr_congestion_on;
333 unsigned int nr_congestion_off;
334 unsigned int nr_batching;
336 void *dma_drain_buffer;
337 unsigned int dma_drain_size;
338 unsigned int dma_pad_mask;
339 unsigned int dma_alignment;
341 struct blk_queue_tag *queue_tags;
342 struct list_head tag_busy_list;
344 unsigned int nr_sorted;
345 unsigned int in_flight[2];
347 unsigned int rq_timeout;
348 struct timer_list timeout;
349 struct list_head timeout_list;
351 struct queue_limits limits;
354 * sg stuff
356 unsigned int sg_timeout;
357 unsigned int sg_reserved_size;
358 int node;
359 #ifdef CONFIG_BLK_DEV_IO_TRACE
360 struct blk_trace *blk_trace;
361 #endif
363 * for flush operations
365 unsigned int flush_flags;
366 unsigned int flush_seq;
367 int flush_err;
368 struct request flush_rq;
369 struct request *orig_flush_rq;
370 struct list_head pending_flushes;
372 struct mutex sysfs_lock;
374 #if defined(CONFIG_BLK_DEV_BSG)
375 struct bsg_class_device bsg_dev;
376 #endif
378 #ifdef CONFIG_BLK_DEV_THROTTLING
379 /* Throttle data */
380 struct throtl_data *td;
381 #endif
384 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
385 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
386 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
387 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
388 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
389 #define QUEUE_FLAG_REENTER 6 /* Re-entrancy avoidance */
390 #define QUEUE_FLAG_PLUGGED 7 /* queue is plugged */
391 #define QUEUE_FLAG_ELVSWITCH 8 /* don't use elevator, just do FIFO */
392 #define QUEUE_FLAG_BIDI 9 /* queue supports bidi requests */
393 #define QUEUE_FLAG_NOMERGES 10 /* disable merge attempts */
394 #define QUEUE_FLAG_SAME_COMP 11 /* force complete on same CPU */
395 #define QUEUE_FLAG_FAIL_IO 12 /* fake timeout */
396 #define QUEUE_FLAG_STACKABLE 13 /* supports request stacking */
397 #define QUEUE_FLAG_NONROT 14 /* non-rotational device (SSD) */
398 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
399 #define QUEUE_FLAG_IO_STAT 15 /* do IO stats */
400 #define QUEUE_FLAG_DISCARD 16 /* supports DISCARD */
401 #define QUEUE_FLAG_NOXMERGES 17 /* No extended merges */
402 #define QUEUE_FLAG_ADD_RANDOM 18 /* Contributes to random pool */
403 #define QUEUE_FLAG_SECDISCARD 19 /* supports SECDISCARD */
405 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
406 (1 << QUEUE_FLAG_STACKABLE) | \
407 (1 << QUEUE_FLAG_SAME_COMP) | \
408 (1 << QUEUE_FLAG_ADD_RANDOM))
410 static inline int queue_is_locked(struct request_queue *q)
412 #ifdef CONFIG_SMP
413 spinlock_t *lock = q->queue_lock;
414 return lock && spin_is_locked(lock);
415 #else
416 return 1;
417 #endif
420 static inline void queue_flag_set_unlocked(unsigned int flag,
421 struct request_queue *q)
423 __set_bit(flag, &q->queue_flags);
426 static inline int queue_flag_test_and_clear(unsigned int flag,
427 struct request_queue *q)
429 WARN_ON_ONCE(!queue_is_locked(q));
431 if (test_bit(flag, &q->queue_flags)) {
432 __clear_bit(flag, &q->queue_flags);
433 return 1;
436 return 0;
439 static inline int queue_flag_test_and_set(unsigned int flag,
440 struct request_queue *q)
442 WARN_ON_ONCE(!queue_is_locked(q));
444 if (!test_bit(flag, &q->queue_flags)) {
445 __set_bit(flag, &q->queue_flags);
446 return 0;
449 return 1;
452 static inline void queue_flag_set(unsigned int flag, struct request_queue *q)
454 WARN_ON_ONCE(!queue_is_locked(q));
455 __set_bit(flag, &q->queue_flags);
458 static inline void queue_flag_clear_unlocked(unsigned int flag,
459 struct request_queue *q)
461 __clear_bit(flag, &q->queue_flags);
464 static inline int queue_in_flight(struct request_queue *q)
466 return q->in_flight[0] + q->in_flight[1];
469 static inline void queue_flag_clear(unsigned int flag, struct request_queue *q)
471 WARN_ON_ONCE(!queue_is_locked(q));
472 __clear_bit(flag, &q->queue_flags);
475 #define blk_queue_plugged(q) test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags)
476 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
477 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
478 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
479 #define blk_queue_noxmerges(q) \
480 test_bit(QUEUE_FLAG_NOXMERGES, &(q)->queue_flags)
481 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
482 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
483 #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags)
484 #define blk_queue_stackable(q) \
485 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
486 #define blk_queue_discard(q) test_bit(QUEUE_FLAG_DISCARD, &(q)->queue_flags)
487 #define blk_queue_secdiscard(q) (blk_queue_discard(q) && \
488 test_bit(QUEUE_FLAG_SECDISCARD, &(q)->queue_flags))
490 #define blk_noretry_request(rq) \
491 ((rq)->cmd_flags & (REQ_FAILFAST_DEV|REQ_FAILFAST_TRANSPORT| \
492 REQ_FAILFAST_DRIVER))
494 #define blk_account_rq(rq) \
495 (((rq)->cmd_flags & REQ_STARTED) && \
496 ((rq)->cmd_type == REQ_TYPE_FS || \
497 ((rq)->cmd_flags & REQ_DISCARD)))
499 #define blk_pm_request(rq) \
500 ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND || \
501 (rq)->cmd_type == REQ_TYPE_PM_RESUME)
503 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
504 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
505 /* rq->queuelist of dequeued request must be list_empty() */
506 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
508 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
510 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
512 static inline unsigned int blk_queue_cluster(struct request_queue *q)
514 return q->limits.cluster;
518 * We regard a request as sync, if either a read or a sync write
520 static inline bool rw_is_sync(unsigned int rw_flags)
522 return !(rw_flags & REQ_WRITE) || (rw_flags & REQ_SYNC);
525 static inline bool rq_is_sync(struct request *rq)
527 return rw_is_sync(rq->cmd_flags);
530 static inline int blk_queue_full(struct request_queue *q, int sync)
532 if (sync)
533 return test_bit(QUEUE_FLAG_SYNCFULL, &q->queue_flags);
534 return test_bit(QUEUE_FLAG_ASYNCFULL, &q->queue_flags);
537 static inline void blk_set_queue_full(struct request_queue *q, int sync)
539 if (sync)
540 queue_flag_set(QUEUE_FLAG_SYNCFULL, q);
541 else
542 queue_flag_set(QUEUE_FLAG_ASYNCFULL, q);
545 static inline void blk_clear_queue_full(struct request_queue *q, int sync)
547 if (sync)
548 queue_flag_clear(QUEUE_FLAG_SYNCFULL, q);
549 else
550 queue_flag_clear(QUEUE_FLAG_ASYNCFULL, q);
555 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
556 * it already be started by driver.
558 #define RQ_NOMERGE_FLAGS \
559 (REQ_NOMERGE | REQ_STARTED | REQ_SOFTBARRIER | REQ_FLUSH | REQ_FUA)
560 #define rq_mergeable(rq) \
561 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
562 (((rq)->cmd_flags & REQ_DISCARD) || \
563 (rq)->cmd_type == REQ_TYPE_FS))
566 * q->prep_rq_fn return values
568 #define BLKPREP_OK 0 /* serve it */
569 #define BLKPREP_KILL 1 /* fatal error, kill */
570 #define BLKPREP_DEFER 2 /* leave on queue */
572 extern unsigned long blk_max_low_pfn, blk_max_pfn;
575 * standard bounce addresses:
577 * BLK_BOUNCE_HIGH : bounce all highmem pages
578 * BLK_BOUNCE_ANY : don't bounce anything
579 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
582 #if BITS_PER_LONG == 32
583 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
584 #else
585 #define BLK_BOUNCE_HIGH -1ULL
586 #endif
587 #define BLK_BOUNCE_ANY (-1ULL)
588 #define BLK_BOUNCE_ISA (DMA_BIT_MASK(24))
591 * default timeout for SG_IO if none specified
593 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
594 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
596 #ifdef CONFIG_BOUNCE
597 extern int init_emergency_isa_pool(void);
598 extern void blk_queue_bounce(struct request_queue *q, struct bio **bio);
599 #else
600 static inline int init_emergency_isa_pool(void)
602 return 0;
604 static inline void blk_queue_bounce(struct request_queue *q, struct bio **bio)
607 #endif /* CONFIG_MMU */
609 struct rq_map_data {
610 struct page **pages;
611 int page_order;
612 int nr_entries;
613 unsigned long offset;
614 int null_mapped;
615 int from_user;
618 struct req_iterator {
619 int i;
620 struct bio *bio;
623 /* This should not be used directly - use rq_for_each_segment */
624 #define for_each_bio(_bio) \
625 for (; _bio; _bio = _bio->bi_next)
626 #define __rq_for_each_bio(_bio, rq) \
627 if ((rq->bio)) \
628 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
630 #define rq_for_each_segment(bvl, _rq, _iter) \
631 __rq_for_each_bio(_iter.bio, _rq) \
632 bio_for_each_segment(bvl, _iter.bio, _iter.i)
634 #define rq_iter_last(rq, _iter) \
635 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
637 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
638 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
639 #endif
640 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
641 extern void rq_flush_dcache_pages(struct request *rq);
642 #else
643 static inline void rq_flush_dcache_pages(struct request *rq)
646 #endif
648 extern int blk_register_queue(struct gendisk *disk);
649 extern void blk_unregister_queue(struct gendisk *disk);
650 extern void register_disk(struct gendisk *dev);
651 extern void generic_make_request(struct bio *bio);
652 extern void blk_rq_init(struct request_queue *q, struct request *rq);
653 extern void blk_put_request(struct request *);
654 extern void __blk_put_request(struct request_queue *, struct request *);
655 extern struct request *blk_get_request(struct request_queue *, int, gfp_t);
656 extern struct request *blk_make_request(struct request_queue *, struct bio *,
657 gfp_t);
658 extern void blk_insert_request(struct request_queue *, struct request *, int, void *);
659 extern void blk_requeue_request(struct request_queue *, struct request *);
660 extern void blk_add_request_payload(struct request *rq, struct page *page,
661 unsigned int len);
662 extern int blk_rq_check_limits(struct request_queue *q, struct request *rq);
663 extern int blk_lld_busy(struct request_queue *q);
664 extern int blk_rq_prep_clone(struct request *rq, struct request *rq_src,
665 struct bio_set *bs, gfp_t gfp_mask,
666 int (*bio_ctr)(struct bio *, struct bio *, void *),
667 void *data);
668 extern void blk_rq_unprep_clone(struct request *rq);
669 extern int blk_insert_cloned_request(struct request_queue *q,
670 struct request *rq);
671 extern void blk_plug_device(struct request_queue *);
672 extern void blk_plug_device_unlocked(struct request_queue *);
673 extern int blk_remove_plug(struct request_queue *);
674 extern void blk_recount_segments(struct request_queue *, struct bio *);
675 extern int scsi_cmd_ioctl(struct request_queue *, struct gendisk *, fmode_t,
676 unsigned int, void __user *);
677 extern int sg_scsi_ioctl(struct request_queue *, struct gendisk *, fmode_t,
678 struct scsi_ioctl_command __user *);
681 * A queue has just exitted congestion. Note this in the global counter of
682 * congested queues, and wake up anyone who was waiting for requests to be
683 * put back.
685 static inline void blk_clear_queue_congested(struct request_queue *q, int sync)
687 clear_bdi_congested(&q->backing_dev_info, sync);
691 * A queue has just entered congestion. Flag that in the queue's VM-visible
692 * state flags and increment the global gounter of congested queues.
694 static inline void blk_set_queue_congested(struct request_queue *q, int sync)
696 set_bdi_congested(&q->backing_dev_info, sync);
699 extern void blk_start_queue(struct request_queue *q);
700 extern void blk_stop_queue(struct request_queue *q);
701 extern void blk_sync_queue(struct request_queue *q);
702 extern void __blk_stop_queue(struct request_queue *q);
703 extern void __blk_run_queue(struct request_queue *);
704 extern void blk_run_queue(struct request_queue *);
705 extern int blk_rq_map_user(struct request_queue *, struct request *,
706 struct rq_map_data *, void __user *, unsigned long,
707 gfp_t);
708 extern int blk_rq_unmap_user(struct bio *);
709 extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
710 extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
711 struct rq_map_data *, struct sg_iovec *, int,
712 unsigned int, gfp_t);
713 extern int blk_execute_rq(struct request_queue *, struct gendisk *,
714 struct request *, int);
715 extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
716 struct request *, int, rq_end_io_fn *);
717 extern void blk_unplug(struct request_queue *q);
719 static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
721 return bdev->bd_disk->queue;
725 * blk_rq_pos() : the current sector
726 * blk_rq_bytes() : bytes left in the entire request
727 * blk_rq_cur_bytes() : bytes left in the current segment
728 * blk_rq_err_bytes() : bytes left till the next error boundary
729 * blk_rq_sectors() : sectors left in the entire request
730 * blk_rq_cur_sectors() : sectors left in the current segment
732 static inline sector_t blk_rq_pos(const struct request *rq)
734 return rq->__sector;
737 static inline unsigned int blk_rq_bytes(const struct request *rq)
739 return rq->__data_len;
742 static inline int blk_rq_cur_bytes(const struct request *rq)
744 return rq->bio ? bio_cur_bytes(rq->bio) : 0;
747 extern unsigned int blk_rq_err_bytes(const struct request *rq);
749 static inline unsigned int blk_rq_sectors(const struct request *rq)
751 return blk_rq_bytes(rq) >> 9;
754 static inline unsigned int blk_rq_cur_sectors(const struct request *rq)
756 return blk_rq_cur_bytes(rq) >> 9;
760 * Request issue related functions.
762 extern struct request *blk_peek_request(struct request_queue *q);
763 extern void blk_start_request(struct request *rq);
764 extern struct request *blk_fetch_request(struct request_queue *q);
767 * Request completion related functions.
769 * blk_update_request() completes given number of bytes and updates
770 * the request without completing it.
772 * blk_end_request() and friends. __blk_end_request() must be called
773 * with the request queue spinlock acquired.
775 * Several drivers define their own end_request and call
776 * blk_end_request() for parts of the original function.
777 * This prevents code duplication in drivers.
779 extern bool blk_update_request(struct request *rq, int error,
780 unsigned int nr_bytes);
781 extern bool blk_end_request(struct request *rq, int error,
782 unsigned int nr_bytes);
783 extern void blk_end_request_all(struct request *rq, int error);
784 extern bool blk_end_request_cur(struct request *rq, int error);
785 extern bool blk_end_request_err(struct request *rq, int error);
786 extern bool __blk_end_request(struct request *rq, int error,
787 unsigned int nr_bytes);
788 extern void __blk_end_request_all(struct request *rq, int error);
789 extern bool __blk_end_request_cur(struct request *rq, int error);
790 extern bool __blk_end_request_err(struct request *rq, int error);
792 extern void blk_complete_request(struct request *);
793 extern void __blk_complete_request(struct request *);
794 extern void blk_abort_request(struct request *);
795 extern void blk_abort_queue(struct request_queue *);
796 extern void blk_unprep_request(struct request *);
799 * Access functions for manipulating queue properties
801 extern struct request_queue *blk_init_queue_node(request_fn_proc *rfn,
802 spinlock_t *lock, int node_id);
803 extern struct request_queue *blk_init_allocated_queue_node(struct request_queue *,
804 request_fn_proc *,
805 spinlock_t *, int node_id);
806 extern struct request_queue *blk_init_queue(request_fn_proc *, spinlock_t *);
807 extern struct request_queue *blk_init_allocated_queue(struct request_queue *,
808 request_fn_proc *, spinlock_t *);
809 extern void blk_cleanup_queue(struct request_queue *);
810 extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
811 extern void blk_queue_bounce_limit(struct request_queue *, u64);
812 extern void blk_limits_max_hw_sectors(struct queue_limits *, unsigned int);
813 extern void blk_queue_max_hw_sectors(struct request_queue *, unsigned int);
814 extern void blk_queue_max_segments(struct request_queue *, unsigned short);
815 extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
816 extern void blk_queue_max_discard_sectors(struct request_queue *q,
817 unsigned int max_discard_sectors);
818 extern void blk_queue_logical_block_size(struct request_queue *, unsigned short);
819 extern void blk_queue_physical_block_size(struct request_queue *, unsigned int);
820 extern void blk_queue_alignment_offset(struct request_queue *q,
821 unsigned int alignment);
822 extern void blk_limits_io_min(struct queue_limits *limits, unsigned int min);
823 extern void blk_queue_io_min(struct request_queue *q, unsigned int min);
824 extern void blk_limits_io_opt(struct queue_limits *limits, unsigned int opt);
825 extern void blk_queue_io_opt(struct request_queue *q, unsigned int opt);
826 extern void blk_set_default_limits(struct queue_limits *lim);
827 extern int blk_stack_limits(struct queue_limits *t, struct queue_limits *b,
828 sector_t offset);
829 extern int bdev_stack_limits(struct queue_limits *t, struct block_device *bdev,
830 sector_t offset);
831 extern void disk_stack_limits(struct gendisk *disk, struct block_device *bdev,
832 sector_t offset);
833 extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
834 extern void blk_queue_dma_pad(struct request_queue *, unsigned int);
835 extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
836 extern int blk_queue_dma_drain(struct request_queue *q,
837 dma_drain_needed_fn *dma_drain_needed,
838 void *buf, unsigned int size);
839 extern void blk_queue_lld_busy(struct request_queue *q, lld_busy_fn *fn);
840 extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
841 extern void blk_queue_prep_rq(struct request_queue *, prep_rq_fn *pfn);
842 extern void blk_queue_unprep_rq(struct request_queue *, unprep_rq_fn *ufn);
843 extern void blk_queue_merge_bvec(struct request_queue *, merge_bvec_fn *);
844 extern void blk_queue_dma_alignment(struct request_queue *, int);
845 extern void blk_queue_update_dma_alignment(struct request_queue *, int);
846 extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
847 extern void blk_queue_rq_timed_out(struct request_queue *, rq_timed_out_fn *);
848 extern void blk_queue_rq_timeout(struct request_queue *, unsigned int);
849 extern void blk_queue_flush(struct request_queue *q, unsigned int flush);
850 extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
852 extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
853 extern void blk_dump_rq_flags(struct request *, char *);
854 extern void generic_unplug_device(struct request_queue *);
855 extern long nr_blockdev_pages(void);
857 int blk_get_queue(struct request_queue *);
858 struct request_queue *blk_alloc_queue(gfp_t);
859 struct request_queue *blk_alloc_queue_node(gfp_t, int);
860 extern void blk_put_queue(struct request_queue *);
863 * tag stuff
865 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
866 extern int blk_queue_start_tag(struct request_queue *, struct request *);
867 extern struct request *blk_queue_find_tag(struct request_queue *, int);
868 extern void blk_queue_end_tag(struct request_queue *, struct request *);
869 extern int blk_queue_init_tags(struct request_queue *, int, struct blk_queue_tag *);
870 extern void blk_queue_free_tags(struct request_queue *);
871 extern int blk_queue_resize_tags(struct request_queue *, int);
872 extern void blk_queue_invalidate_tags(struct request_queue *);
873 extern struct blk_queue_tag *blk_init_tags(int);
874 extern void blk_free_tags(struct blk_queue_tag *);
876 static inline struct request *blk_map_queue_find_tag(struct blk_queue_tag *bqt,
877 int tag)
879 if (unlikely(bqt == NULL || tag >= bqt->real_max_depth))
880 return NULL;
881 return bqt->tag_index[tag];
884 #define BLKDEV_DISCARD_SECURE 0x01 /* secure discard */
886 extern int blkdev_issue_flush(struct block_device *, gfp_t, sector_t *);
887 extern int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
888 sector_t nr_sects, gfp_t gfp_mask, unsigned long flags);
889 extern int blkdev_issue_zeroout(struct block_device *bdev, sector_t sector,
890 sector_t nr_sects, gfp_t gfp_mask);
891 static inline int sb_issue_discard(struct super_block *sb, sector_t block,
892 sector_t nr_blocks, gfp_t gfp_mask, unsigned long flags)
894 return blkdev_issue_discard(sb->s_bdev, block << (sb->s_blocksize_bits - 9),
895 nr_blocks << (sb->s_blocksize_bits - 9),
896 gfp_mask, flags);
898 static inline int sb_issue_zeroout(struct super_block *sb, sector_t block,
899 sector_t nr_blocks, gfp_t gfp_mask)
901 return blkdev_issue_zeroout(sb->s_bdev,
902 block << (sb->s_blocksize_bits - 9),
903 nr_blocks << (sb->s_blocksize_bits - 9),
904 gfp_mask);
907 extern int blk_verify_command(unsigned char *cmd, fmode_t has_write_perm);
909 enum blk_default_limits {
910 BLK_MAX_SEGMENTS = 128,
911 BLK_SAFE_MAX_SECTORS = 255,
912 BLK_DEF_MAX_SECTORS = 1024,
913 BLK_MAX_SEGMENT_SIZE = 65536,
914 BLK_SEG_BOUNDARY_MASK = 0xFFFFFFFFUL,
917 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
919 static inline unsigned long queue_bounce_pfn(struct request_queue *q)
921 return q->limits.bounce_pfn;
924 static inline unsigned long queue_segment_boundary(struct request_queue *q)
926 return q->limits.seg_boundary_mask;
929 static inline unsigned int queue_max_sectors(struct request_queue *q)
931 return q->limits.max_sectors;
934 static inline unsigned int queue_max_hw_sectors(struct request_queue *q)
936 return q->limits.max_hw_sectors;
939 static inline unsigned short queue_max_segments(struct request_queue *q)
941 return q->limits.max_segments;
944 static inline unsigned int queue_max_segment_size(struct request_queue *q)
946 return q->limits.max_segment_size;
949 static inline unsigned short queue_logical_block_size(struct request_queue *q)
951 int retval = 512;
953 if (q && q->limits.logical_block_size)
954 retval = q->limits.logical_block_size;
956 return retval;
959 static inline unsigned short bdev_logical_block_size(struct block_device *bdev)
961 return queue_logical_block_size(bdev_get_queue(bdev));
964 static inline unsigned int queue_physical_block_size(struct request_queue *q)
966 return q->limits.physical_block_size;
969 static inline unsigned int bdev_physical_block_size(struct block_device *bdev)
971 return queue_physical_block_size(bdev_get_queue(bdev));
974 static inline unsigned int queue_io_min(struct request_queue *q)
976 return q->limits.io_min;
979 static inline int bdev_io_min(struct block_device *bdev)
981 return queue_io_min(bdev_get_queue(bdev));
984 static inline unsigned int queue_io_opt(struct request_queue *q)
986 return q->limits.io_opt;
989 static inline int bdev_io_opt(struct block_device *bdev)
991 return queue_io_opt(bdev_get_queue(bdev));
994 static inline int queue_alignment_offset(struct request_queue *q)
996 if (q->limits.misaligned)
997 return -1;
999 return q->limits.alignment_offset;
1002 static inline int queue_limit_alignment_offset(struct queue_limits *lim, sector_t sector)
1004 unsigned int granularity = max(lim->physical_block_size, lim->io_min);
1005 unsigned int alignment = (sector << 9) & (granularity - 1);
1007 return (granularity + lim->alignment_offset - alignment)
1008 & (granularity - 1);
1011 static inline int bdev_alignment_offset(struct block_device *bdev)
1013 struct request_queue *q = bdev_get_queue(bdev);
1015 if (q->limits.misaligned)
1016 return -1;
1018 if (bdev != bdev->bd_contains)
1019 return bdev->bd_part->alignment_offset;
1021 return q->limits.alignment_offset;
1024 static inline int queue_discard_alignment(struct request_queue *q)
1026 if (q->limits.discard_misaligned)
1027 return -1;
1029 return q->limits.discard_alignment;
1032 static inline int queue_limit_discard_alignment(struct queue_limits *lim, sector_t sector)
1034 unsigned int alignment = (sector << 9) & (lim->discard_granularity - 1);
1036 return (lim->discard_granularity + lim->discard_alignment - alignment)
1037 & (lim->discard_granularity - 1);
1040 static inline unsigned int queue_discard_zeroes_data(struct request_queue *q)
1042 if (q->limits.discard_zeroes_data == 1)
1043 return 1;
1045 return 0;
1048 static inline unsigned int bdev_discard_zeroes_data(struct block_device *bdev)
1050 return queue_discard_zeroes_data(bdev_get_queue(bdev));
1053 static inline int queue_dma_alignment(struct request_queue *q)
1055 return q ? q->dma_alignment : 511;
1058 static inline int blk_rq_aligned(struct request_queue *q, unsigned long addr,
1059 unsigned int len)
1061 unsigned int alignment = queue_dma_alignment(q) | q->dma_pad_mask;
1062 return !(addr & alignment) && !(len & alignment);
1065 /* assumes size > 256 */
1066 static inline unsigned int blksize_bits(unsigned int size)
1068 unsigned int bits = 8;
1069 do {
1070 bits++;
1071 size >>= 1;
1072 } while (size > 256);
1073 return bits;
1076 static inline unsigned int block_size(struct block_device *bdev)
1078 return bdev->bd_block_size;
1081 typedef struct {struct page *v;} Sector;
1083 unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
1085 static inline void put_dev_sector(Sector p)
1087 page_cache_release(p.v);
1090 struct work_struct;
1091 int kblockd_schedule_work(struct request_queue *q, struct work_struct *work);
1093 #ifdef CONFIG_BLK_CGROUP
1095 * This should not be using sched_clock(). A real patch is in progress
1096 * to fix this up, until that is in place we need to disable preemption
1097 * around sched_clock() in this function and set_io_start_time_ns().
1099 static inline void set_start_time_ns(struct request *req)
1101 preempt_disable();
1102 req->start_time_ns = sched_clock();
1103 preempt_enable();
1106 static inline void set_io_start_time_ns(struct request *req)
1108 preempt_disable();
1109 req->io_start_time_ns = sched_clock();
1110 preempt_enable();
1113 static inline uint64_t rq_start_time_ns(struct request *req)
1115 return req->start_time_ns;
1118 static inline uint64_t rq_io_start_time_ns(struct request *req)
1120 return req->io_start_time_ns;
1122 #else
1123 static inline void set_start_time_ns(struct request *req) {}
1124 static inline void set_io_start_time_ns(struct request *req) {}
1125 static inline uint64_t rq_start_time_ns(struct request *req)
1127 return 0;
1129 static inline uint64_t rq_io_start_time_ns(struct request *req)
1131 return 0;
1133 #endif
1135 #ifdef CONFIG_BLK_DEV_THROTTLING
1136 extern int blk_throtl_init(struct request_queue *q);
1137 extern void blk_throtl_exit(struct request_queue *q);
1138 extern int blk_throtl_bio(struct request_queue *q, struct bio **bio);
1139 extern void throtl_shutdown_timer_wq(struct request_queue *q);
1140 #else /* CONFIG_BLK_DEV_THROTTLING */
1141 static inline int blk_throtl_bio(struct request_queue *q, struct bio **bio)
1143 return 0;
1146 static inline int blk_throtl_init(struct request_queue *q) { return 0; }
1147 static inline int blk_throtl_exit(struct request_queue *q) { return 0; }
1148 static inline void throtl_shutdown_timer_wq(struct request_queue *q) {}
1149 #endif /* CONFIG_BLK_DEV_THROTTLING */
1151 #define MODULE_ALIAS_BLOCKDEV(major,minor) \
1152 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
1153 #define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
1154 MODULE_ALIAS("block-major-" __stringify(major) "-*")
1156 #if defined(CONFIG_BLK_DEV_INTEGRITY)
1158 #define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
1159 #define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
1161 struct blk_integrity_exchg {
1162 void *prot_buf;
1163 void *data_buf;
1164 sector_t sector;
1165 unsigned int data_size;
1166 unsigned short sector_size;
1167 const char *disk_name;
1170 typedef void (integrity_gen_fn) (struct blk_integrity_exchg *);
1171 typedef int (integrity_vrfy_fn) (struct blk_integrity_exchg *);
1172 typedef void (integrity_set_tag_fn) (void *, void *, unsigned int);
1173 typedef void (integrity_get_tag_fn) (void *, void *, unsigned int);
1175 struct blk_integrity {
1176 integrity_gen_fn *generate_fn;
1177 integrity_vrfy_fn *verify_fn;
1178 integrity_set_tag_fn *set_tag_fn;
1179 integrity_get_tag_fn *get_tag_fn;
1181 unsigned short flags;
1182 unsigned short tuple_size;
1183 unsigned short sector_size;
1184 unsigned short tag_size;
1186 const char *name;
1188 struct kobject kobj;
1191 extern int blk_integrity_register(struct gendisk *, struct blk_integrity *);
1192 extern void blk_integrity_unregister(struct gendisk *);
1193 extern int blk_integrity_compare(struct gendisk *, struct gendisk *);
1194 extern int blk_rq_map_integrity_sg(struct request_queue *, struct bio *,
1195 struct scatterlist *);
1196 extern int blk_rq_count_integrity_sg(struct request_queue *, struct bio *);
1197 extern int blk_integrity_merge_rq(struct request_queue *, struct request *,
1198 struct request *);
1199 extern int blk_integrity_merge_bio(struct request_queue *, struct request *,
1200 struct bio *);
1202 static inline
1203 struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
1205 return bdev->bd_disk->integrity;
1208 static inline struct blk_integrity *blk_get_integrity(struct gendisk *disk)
1210 return disk->integrity;
1213 static inline int blk_integrity_rq(struct request *rq)
1215 if (rq->bio == NULL)
1216 return 0;
1218 return bio_integrity(rq->bio);
1221 static inline void blk_queue_max_integrity_segments(struct request_queue *q,
1222 unsigned int segs)
1224 q->limits.max_integrity_segments = segs;
1227 static inline unsigned short
1228 queue_max_integrity_segments(struct request_queue *q)
1230 return q->limits.max_integrity_segments;
1233 #else /* CONFIG_BLK_DEV_INTEGRITY */
1235 #define blk_integrity_rq(rq) (0)
1236 #define blk_rq_count_integrity_sg(a, b) (0)
1237 #define blk_rq_map_integrity_sg(a, b, c) (0)
1238 #define bdev_get_integrity(a) (0)
1239 #define blk_get_integrity(a) (0)
1240 #define blk_integrity_compare(a, b) (0)
1241 #define blk_integrity_register(a, b) (0)
1242 #define blk_integrity_unregister(a) do { } while (0);
1243 #define blk_queue_max_integrity_segments(a, b) do { } while (0);
1244 #define queue_max_integrity_segments(a) (0)
1245 #define blk_integrity_merge_rq(a, b, c) (0)
1246 #define blk_integrity_merge_bio(a, b, c) (0)
1248 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1250 struct block_device_operations {
1251 int (*open) (struct block_device *, fmode_t);
1252 int (*release) (struct gendisk *, fmode_t);
1253 int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1254 int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
1255 int (*direct_access) (struct block_device *, sector_t,
1256 void **, unsigned long *);
1257 int (*media_changed) (struct gendisk *);
1258 void (*unlock_native_capacity) (struct gendisk *);
1259 int (*revalidate_disk) (struct gendisk *);
1260 int (*getgeo)(struct block_device *, struct hd_geometry *);
1261 /* this callback is with swap_lock and sometimes page table lock held */
1262 void (*swap_slot_free_notify) (struct block_device *, unsigned long);
1263 struct module *owner;
1266 extern int __blkdev_driver_ioctl(struct block_device *, fmode_t, unsigned int,
1267 unsigned long);
1268 #else /* CONFIG_BLOCK */
1270 * stubs for when the block layer is configured out
1272 #define buffer_heads_over_limit 0
1274 static inline long nr_blockdev_pages(void)
1276 return 0;
1279 #endif /* CONFIG_BLOCK */
1281 #endif