1 #ifndef _LINUX_BLKDEV_H
2 #define _LINUX_BLKDEV_H
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
;
28 struct elevator_queue
;
29 struct request_pm_state
;
34 #define BLKDEV_MIN_RQ 4
35 #define BLKDEV_MAX_RQ 128 /* Default maximum */
38 typedef void (rq_end_io_fn
)(struct request
*, int);
42 * count[], starved[], and wait[] are indexed by
43 * BLK_RW_SYNC/BLK_RW_ASYNC
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 */
63 REQ_TYPE_LINUX_BLOCK
, /* generic block layer message */
65 * for ATA/ATAPI devices. this really doesn't belong here, ide should
66 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
67 * private REQ_LB opcodes to differentiate what type of request this is
69 REQ_TYPE_ATA_TASKFILE
,
74 * For request of type REQ_TYPE_LINUX_BLOCK, rq->cmd[0] is the opcode being
75 * sent down (similar to how REQ_TYPE_BLOCK_PC means that ->cmd[] holds a
78 * 0x00 -> 0x3f are driver private, to be used for whatever purpose they need,
79 * typically to differentiate REQ_TYPE_SPECIAL requests.
83 REQ_LB_OP_EJECT
= 0x40, /* eject request */
84 REQ_LB_OP_FLUSH
= 0x41, /* flush request */
85 REQ_LB_OP_DISCARD
= 0x42, /* discard sectors */
89 * request type modified bits. first four bits match BIO_RW* bits, important
92 __REQ_RW
, /* not set, read. set, write */
93 __REQ_FAILFAST_DEV
, /* no driver retries of device errors */
94 __REQ_FAILFAST_TRANSPORT
, /* no driver retries of transport errors */
95 __REQ_FAILFAST_DRIVER
, /* no driver retries of driver errors */
96 /* above flags must match BIO_RW_* */
97 __REQ_DISCARD
, /* request to discard sectors */
98 __REQ_SORTED
, /* elevator knows about this request */
99 __REQ_SOFTBARRIER
, /* may not be passed by ioscheduler */
100 __REQ_HARDBARRIER
, /* may not be passed by drive either */
101 __REQ_FUA
, /* forced unit access */
102 __REQ_NOMERGE
, /* don't touch this for merging */
103 __REQ_STARTED
, /* drive already may have started this one */
104 __REQ_DONTPREP
, /* don't call prep for this one */
105 __REQ_QUEUED
, /* uses queueing */
106 __REQ_ELVPRIV
, /* elevator private data attached */
107 __REQ_FAILED
, /* set if the request failed */
108 __REQ_QUIET
, /* don't worry about errors */
109 __REQ_PREEMPT
, /* set for "ide_preempt" requests */
110 __REQ_ORDERED_COLOR
, /* is before or after barrier */
111 __REQ_RW_SYNC
, /* request is sync (sync write or read) */
112 __REQ_ALLOCED
, /* request came from our alloc pool */
113 __REQ_RW_META
, /* metadata io request */
114 __REQ_COPY_USER
, /* contains copies of user pages */
115 __REQ_INTEGRITY
, /* integrity metadata has been remapped */
116 __REQ_NOIDLE
, /* Don't anticipate more IO after this one */
117 __REQ_IO_STAT
, /* account I/O stat */
118 __REQ_MIXED_MERGE
, /* merge of different types, fail separately */
119 __REQ_NR_BITS
, /* stops here */
122 #define REQ_RW (1 << __REQ_RW)
123 #define REQ_FAILFAST_DEV (1 << __REQ_FAILFAST_DEV)
124 #define REQ_FAILFAST_TRANSPORT (1 << __REQ_FAILFAST_TRANSPORT)
125 #define REQ_FAILFAST_DRIVER (1 << __REQ_FAILFAST_DRIVER)
126 #define REQ_DISCARD (1 << __REQ_DISCARD)
127 #define REQ_SORTED (1 << __REQ_SORTED)
128 #define REQ_SOFTBARRIER (1 << __REQ_SOFTBARRIER)
129 #define REQ_HARDBARRIER (1 << __REQ_HARDBARRIER)
130 #define REQ_FUA (1 << __REQ_FUA)
131 #define REQ_NOMERGE (1 << __REQ_NOMERGE)
132 #define REQ_STARTED (1 << __REQ_STARTED)
133 #define REQ_DONTPREP (1 << __REQ_DONTPREP)
134 #define REQ_QUEUED (1 << __REQ_QUEUED)
135 #define REQ_ELVPRIV (1 << __REQ_ELVPRIV)
136 #define REQ_FAILED (1 << __REQ_FAILED)
137 #define REQ_QUIET (1 << __REQ_QUIET)
138 #define REQ_PREEMPT (1 << __REQ_PREEMPT)
139 #define REQ_ORDERED_COLOR (1 << __REQ_ORDERED_COLOR)
140 #define REQ_RW_SYNC (1 << __REQ_RW_SYNC)
141 #define REQ_ALLOCED (1 << __REQ_ALLOCED)
142 #define REQ_RW_META (1 << __REQ_RW_META)
143 #define REQ_COPY_USER (1 << __REQ_COPY_USER)
144 #define REQ_INTEGRITY (1 << __REQ_INTEGRITY)
145 #define REQ_NOIDLE (1 << __REQ_NOIDLE)
146 #define REQ_IO_STAT (1 << __REQ_IO_STAT)
147 #define REQ_MIXED_MERGE (1 << __REQ_MIXED_MERGE)
149 #define REQ_FAILFAST_MASK (REQ_FAILFAST_DEV | REQ_FAILFAST_TRANSPORT | \
152 #define BLK_MAX_CDB 16
155 * try to put the fields that are referenced together in the same cacheline.
156 * if you modify this structure, be sure to check block/blk-core.c:rq_init()
160 struct list_head queuelist
;
161 struct call_single_data csd
;
164 struct request_queue
*q
;
166 unsigned int cmd_flags
;
167 enum rq_cmd_type_bits cmd_type
;
168 unsigned long atomic_flags
;
170 /* the following two fields are internal, NEVER access directly */
171 sector_t __sector
; /* sector cursor */
172 unsigned int __data_len
; /* total data len */
177 struct hlist_node hash
; /* merge hash */
179 * The rb_node is only used inside the io scheduler, requests
180 * are pruned when moved to the dispatch queue. So let the
181 * completion_data share space with the rb_node.
184 struct rb_node rb_node
; /* sort/lookup */
185 void *completion_data
;
189 * two pointers are available for the IO schedulers, if they need
190 * more they have to dynamically allocate it.
192 void *elevator_private
;
193 void *elevator_private2
;
195 struct gendisk
*rq_disk
;
196 unsigned long start_time
;
198 /* Number of scatter-gather DMA addr+len pairs after
199 * physical address coalescing is performed.
201 unsigned short nr_phys_segments
;
203 unsigned short ioprio
;
205 void *special
; /* opaque pointer available for LLD use */
206 char *buffer
; /* kaddr of the current segment if available */
214 * when request is used as a packet command carrier
216 unsigned short cmd_len
;
217 unsigned char __cmd
[BLK_MAX_CDB
];
220 unsigned int extra_len
; /* length of alignment and padding */
221 unsigned int sense_len
;
222 unsigned int resid_len
; /* residual count */
225 unsigned long deadline
;
226 struct list_head timeout_list
;
227 unsigned int timeout
;
231 * completion callback.
233 rq_end_io_fn
*end_io
;
237 struct request
*next_rq
;
240 static inline unsigned short req_get_ioprio(struct request
*req
)
246 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
247 * requests. Some step values could eventually be made generic.
249 struct request_pm_state
251 /* PM state machine step value, currently driver specific */
253 /* requested PM state value (S1, S2, S3, S4, ...) */
255 void* data
; /* for driver use */
258 #include <linux/elevator.h>
260 typedef void (request_fn_proc
) (struct request_queue
*q
);
261 typedef int (make_request_fn
) (struct request_queue
*q
, struct bio
*bio
);
262 typedef int (prep_rq_fn
) (struct request_queue
*, struct request
*);
263 typedef void (unplug_fn
) (struct request_queue
*);
264 typedef int (prepare_discard_fn
) (struct request_queue
*, struct request
*);
267 struct bvec_merge_data
{
268 struct block_device
*bi_bdev
;
273 typedef int (merge_bvec_fn
) (struct request_queue
*, struct bvec_merge_data
*,
275 typedef void (prepare_flush_fn
) (struct request_queue
*, struct request
*);
276 typedef void (softirq_done_fn
)(struct request
*);
277 typedef int (dma_drain_needed_fn
)(struct request
*);
278 typedef int (lld_busy_fn
) (struct request_queue
*q
);
280 enum blk_eh_timer_return
{
286 typedef enum blk_eh_timer_return (rq_timed_out_fn
)(struct request
*);
288 enum blk_queue_state
{
293 struct blk_queue_tag
{
294 struct request
**tag_index
; /* map of busy tags */
295 unsigned long *tag_map
; /* bit map of free/busy tags */
296 int busy
; /* current depth */
297 int max_depth
; /* what we will send to device */
298 int real_max_depth
; /* what the array can hold */
299 atomic_t refcnt
; /* map can be shared */
302 #define BLK_SCSI_MAX_CMDS (256)
303 #define BLK_SCSI_CMD_PER_LONG (BLK_SCSI_MAX_CMDS / (sizeof(long) * 8))
305 struct queue_limits
{
306 unsigned long bounce_pfn
;
307 unsigned long seg_boundary_mask
;
309 unsigned int max_hw_sectors
;
310 unsigned int max_sectors
;
311 unsigned int max_segment_size
;
312 unsigned int physical_block_size
;
313 unsigned int alignment_offset
;
317 unsigned short logical_block_size
;
318 unsigned short max_hw_segments
;
319 unsigned short max_phys_segments
;
321 unsigned char misaligned
;
322 unsigned char no_cluster
;
328 * Together with queue_head for cacheline sharing
330 struct list_head queue_head
;
331 struct request
*last_merge
;
332 struct elevator_queue
*elevator
;
335 * the queue request freelist, one for reads and one for writes
337 struct request_list rq
;
339 request_fn_proc
*request_fn
;
340 make_request_fn
*make_request_fn
;
341 prep_rq_fn
*prep_rq_fn
;
342 unplug_fn
*unplug_fn
;
343 prepare_discard_fn
*prepare_discard_fn
;
344 merge_bvec_fn
*merge_bvec_fn
;
345 prepare_flush_fn
*prepare_flush_fn
;
346 softirq_done_fn
*softirq_done_fn
;
347 rq_timed_out_fn
*rq_timed_out_fn
;
348 dma_drain_needed_fn
*dma_drain_needed
;
349 lld_busy_fn
*lld_busy_fn
;
352 * Dispatch queue sorting
355 struct request
*boundary_rq
;
358 * Auto-unplugging state
360 struct timer_list unplug_timer
;
361 int unplug_thresh
; /* After this many requests */
362 unsigned long unplug_delay
; /* After this many jiffies */
363 struct work_struct unplug_work
;
365 struct backing_dev_info backing_dev_info
;
368 * The queue owner gets to use this for whatever they like.
369 * ll_rw_blk doesn't touch it.
374 * queue needs bounce pages for pages above this limit
379 * various queue flags, see QUEUE_* below
381 unsigned long queue_flags
;
384 * protects queue structures from reentrancy. ->__queue_lock should
385 * _never_ be used directly, it is queue private. always use
388 spinlock_t __queue_lock
;
389 spinlock_t
*queue_lock
;
399 unsigned long nr_requests
; /* Max # of requests */
400 unsigned int nr_congestion_on
;
401 unsigned int nr_congestion_off
;
402 unsigned int nr_batching
;
404 void *dma_drain_buffer
;
405 unsigned int dma_drain_size
;
406 unsigned int dma_pad_mask
;
407 unsigned int dma_alignment
;
409 struct blk_queue_tag
*queue_tags
;
410 struct list_head tag_busy_list
;
412 unsigned int nr_sorted
;
413 unsigned int in_flight
[2];
415 unsigned int rq_timeout
;
416 struct timer_list timeout
;
417 struct list_head timeout_list
;
419 struct queue_limits limits
;
424 unsigned int sg_timeout
;
425 unsigned int sg_reserved_size
;
427 #ifdef CONFIG_BLK_DEV_IO_TRACE
428 struct blk_trace
*blk_trace
;
431 * reserved for flush operations
433 unsigned int ordered
, next_ordered
, ordseq
;
434 int orderr
, ordcolor
;
435 struct request pre_flush_rq
, bar_rq
, post_flush_rq
;
436 struct request
*orig_bar_rq
;
438 struct mutex sysfs_lock
;
440 #if defined(CONFIG_BLK_DEV_BSG)
441 struct bsg_class_device bsg_dev
;
445 #define QUEUE_FLAG_CLUSTER 0 /* cluster several segments into 1 */
446 #define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
447 #define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
448 #define QUEUE_FLAG_SYNCFULL 3 /* read queue has been filled */
449 #define QUEUE_FLAG_ASYNCFULL 4 /* write queue has been filled */
450 #define QUEUE_FLAG_DEAD 5 /* queue being torn down */
451 #define QUEUE_FLAG_REENTER 6 /* Re-entrancy avoidance */
452 #define QUEUE_FLAG_PLUGGED 7 /* queue is plugged */
453 #define QUEUE_FLAG_ELVSWITCH 8 /* don't use elevator, just do FIFO */
454 #define QUEUE_FLAG_BIDI 9 /* queue supports bidi requests */
455 #define QUEUE_FLAG_NOMERGES 10 /* disable merge attempts */
456 #define QUEUE_FLAG_SAME_COMP 11 /* force complete on same CPU */
457 #define QUEUE_FLAG_FAIL_IO 12 /* fake timeout */
458 #define QUEUE_FLAG_STACKABLE 13 /* supports request stacking */
459 #define QUEUE_FLAG_NONROT 14 /* non-rotational device (SSD) */
460 #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */
461 #define QUEUE_FLAG_IO_STAT 15 /* do IO stats */
462 #define QUEUE_FLAG_CQ 16 /* hardware does queuing */
464 #define QUEUE_FLAG_DEFAULT ((1 << QUEUE_FLAG_IO_STAT) | \
465 (1 << QUEUE_FLAG_CLUSTER) | \
466 (1 << QUEUE_FLAG_STACKABLE) | \
467 (1 << QUEUE_FLAG_SAME_COMP))
469 static inline int queue_is_locked(struct request_queue
*q
)
472 spinlock_t
*lock
= q
->queue_lock
;
473 return lock
&& spin_is_locked(lock
);
479 static inline void queue_flag_set_unlocked(unsigned int flag
,
480 struct request_queue
*q
)
482 __set_bit(flag
, &q
->queue_flags
);
485 static inline int queue_flag_test_and_clear(unsigned int flag
,
486 struct request_queue
*q
)
488 WARN_ON_ONCE(!queue_is_locked(q
));
490 if (test_bit(flag
, &q
->queue_flags
)) {
491 __clear_bit(flag
, &q
->queue_flags
);
498 static inline int queue_flag_test_and_set(unsigned int flag
,
499 struct request_queue
*q
)
501 WARN_ON_ONCE(!queue_is_locked(q
));
503 if (!test_bit(flag
, &q
->queue_flags
)) {
504 __set_bit(flag
, &q
->queue_flags
);
511 static inline void queue_flag_set(unsigned int flag
, struct request_queue
*q
)
513 WARN_ON_ONCE(!queue_is_locked(q
));
514 __set_bit(flag
, &q
->queue_flags
);
517 static inline void queue_flag_clear_unlocked(unsigned int flag
,
518 struct request_queue
*q
)
520 __clear_bit(flag
, &q
->queue_flags
);
523 static inline int queue_in_flight(struct request_queue
*q
)
525 return q
->in_flight
[0] + q
->in_flight
[1];
528 static inline void queue_flag_clear(unsigned int flag
, struct request_queue
*q
)
530 WARN_ON_ONCE(!queue_is_locked(q
));
531 __clear_bit(flag
, &q
->queue_flags
);
536 * Hardbarrier is supported with one of the following methods.
538 * NONE : hardbarrier unsupported
539 * DRAIN : ordering by draining is enough
540 * DRAIN_FLUSH : ordering by draining w/ pre and post flushes
541 * DRAIN_FUA : ordering by draining w/ pre flush and FUA write
542 * TAG : ordering by tag is enough
543 * TAG_FLUSH : ordering by tag w/ pre and post flushes
544 * TAG_FUA : ordering by tag w/ pre flush and FUA write
546 QUEUE_ORDERED_BY_DRAIN
= 0x01,
547 QUEUE_ORDERED_BY_TAG
= 0x02,
548 QUEUE_ORDERED_DO_PREFLUSH
= 0x10,
549 QUEUE_ORDERED_DO_BAR
= 0x20,
550 QUEUE_ORDERED_DO_POSTFLUSH
= 0x40,
551 QUEUE_ORDERED_DO_FUA
= 0x80,
553 QUEUE_ORDERED_NONE
= 0x00,
555 QUEUE_ORDERED_DRAIN
= QUEUE_ORDERED_BY_DRAIN
|
556 QUEUE_ORDERED_DO_BAR
,
557 QUEUE_ORDERED_DRAIN_FLUSH
= QUEUE_ORDERED_DRAIN
|
558 QUEUE_ORDERED_DO_PREFLUSH
|
559 QUEUE_ORDERED_DO_POSTFLUSH
,
560 QUEUE_ORDERED_DRAIN_FUA
= QUEUE_ORDERED_DRAIN
|
561 QUEUE_ORDERED_DO_PREFLUSH
|
562 QUEUE_ORDERED_DO_FUA
,
564 QUEUE_ORDERED_TAG
= QUEUE_ORDERED_BY_TAG
|
565 QUEUE_ORDERED_DO_BAR
,
566 QUEUE_ORDERED_TAG_FLUSH
= QUEUE_ORDERED_TAG
|
567 QUEUE_ORDERED_DO_PREFLUSH
|
568 QUEUE_ORDERED_DO_POSTFLUSH
,
569 QUEUE_ORDERED_TAG_FUA
= QUEUE_ORDERED_TAG
|
570 QUEUE_ORDERED_DO_PREFLUSH
|
571 QUEUE_ORDERED_DO_FUA
,
574 * Ordered operation sequence
576 QUEUE_ORDSEQ_STARTED
= 0x01, /* flushing in progress */
577 QUEUE_ORDSEQ_DRAIN
= 0x02, /* waiting for the queue to be drained */
578 QUEUE_ORDSEQ_PREFLUSH
= 0x04, /* pre-flushing in progress */
579 QUEUE_ORDSEQ_BAR
= 0x08, /* original barrier req in progress */
580 QUEUE_ORDSEQ_POSTFLUSH
= 0x10, /* post-flushing in progress */
581 QUEUE_ORDSEQ_DONE
= 0x20,
584 #define blk_queue_plugged(q) test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags)
585 #define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
586 #define blk_queue_queuing(q) test_bit(QUEUE_FLAG_CQ, &(q)->queue_flags)
587 #define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
588 #define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
589 #define blk_queue_nonrot(q) test_bit(QUEUE_FLAG_NONROT, &(q)->queue_flags)
590 #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags)
591 #define blk_queue_flushing(q) ((q)->ordseq)
592 #define blk_queue_stackable(q) \
593 test_bit(QUEUE_FLAG_STACKABLE, &(q)->queue_flags)
595 #define blk_fs_request(rq) ((rq)->cmd_type == REQ_TYPE_FS)
596 #define blk_pc_request(rq) ((rq)->cmd_type == REQ_TYPE_BLOCK_PC)
597 #define blk_special_request(rq) ((rq)->cmd_type == REQ_TYPE_SPECIAL)
598 #define blk_sense_request(rq) ((rq)->cmd_type == REQ_TYPE_SENSE)
600 #define blk_failfast_dev(rq) ((rq)->cmd_flags & REQ_FAILFAST_DEV)
601 #define blk_failfast_transport(rq) ((rq)->cmd_flags & REQ_FAILFAST_TRANSPORT)
602 #define blk_failfast_driver(rq) ((rq)->cmd_flags & REQ_FAILFAST_DRIVER)
603 #define blk_noretry_request(rq) (blk_failfast_dev(rq) || \
604 blk_failfast_transport(rq) || \
605 blk_failfast_driver(rq))
606 #define blk_rq_started(rq) ((rq)->cmd_flags & REQ_STARTED)
607 #define blk_rq_io_stat(rq) ((rq)->cmd_flags & REQ_IO_STAT)
608 #define blk_rq_quiet(rq) ((rq)->cmd_flags & REQ_QUIET)
610 #define blk_account_rq(rq) (blk_rq_started(rq) && (blk_fs_request(rq) || blk_discard_rq(rq)))
612 #define blk_pm_suspend_request(rq) ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND)
613 #define blk_pm_resume_request(rq) ((rq)->cmd_type == REQ_TYPE_PM_RESUME)
614 #define blk_pm_request(rq) \
615 (blk_pm_suspend_request(rq) || blk_pm_resume_request(rq))
617 #define blk_rq_cpu_valid(rq) ((rq)->cpu != -1)
618 #define blk_sorted_rq(rq) ((rq)->cmd_flags & REQ_SORTED)
619 #define blk_barrier_rq(rq) ((rq)->cmd_flags & REQ_HARDBARRIER)
620 #define blk_fua_rq(rq) ((rq)->cmd_flags & REQ_FUA)
621 #define blk_discard_rq(rq) ((rq)->cmd_flags & REQ_DISCARD)
622 #define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
623 /* rq->queuelist of dequeued request must be list_empty() */
624 #define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
626 #define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
628 #define rq_data_dir(rq) ((rq)->cmd_flags & 1)
631 * We regard a request as sync, if either a read or a sync write
633 static inline bool rw_is_sync(unsigned int rw_flags
)
635 return !(rw_flags
& REQ_RW
) || (rw_flags
& REQ_RW_SYNC
);
638 static inline bool rq_is_sync(struct request
*rq
)
640 return rw_is_sync(rq
->cmd_flags
);
643 #define rq_is_meta(rq) ((rq)->cmd_flags & REQ_RW_META)
644 #define rq_noidle(rq) ((rq)->cmd_flags & REQ_NOIDLE)
646 static inline int blk_queue_full(struct request_queue
*q
, int sync
)
649 return test_bit(QUEUE_FLAG_SYNCFULL
, &q
->queue_flags
);
650 return test_bit(QUEUE_FLAG_ASYNCFULL
, &q
->queue_flags
);
653 static inline void blk_set_queue_full(struct request_queue
*q
, int sync
)
656 queue_flag_set(QUEUE_FLAG_SYNCFULL
, q
);
658 queue_flag_set(QUEUE_FLAG_ASYNCFULL
, q
);
661 static inline void blk_clear_queue_full(struct request_queue
*q
, int sync
)
664 queue_flag_clear(QUEUE_FLAG_SYNCFULL
, q
);
666 queue_flag_clear(QUEUE_FLAG_ASYNCFULL
, q
);
671 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
672 * it already be started by driver.
674 #define RQ_NOMERGE_FLAGS \
675 (REQ_NOMERGE | REQ_STARTED | REQ_HARDBARRIER | REQ_SOFTBARRIER)
676 #define rq_mergeable(rq) \
677 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && \
678 (blk_discard_rq(rq) || blk_fs_request((rq))))
681 * q->prep_rq_fn return values
683 #define BLKPREP_OK 0 /* serve it */
684 #define BLKPREP_KILL 1 /* fatal error, kill */
685 #define BLKPREP_DEFER 2 /* leave on queue */
687 extern unsigned long blk_max_low_pfn
, blk_max_pfn
;
690 * standard bounce addresses:
692 * BLK_BOUNCE_HIGH : bounce all highmem pages
693 * BLK_BOUNCE_ANY : don't bounce anything
694 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
697 #if BITS_PER_LONG == 32
698 #define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
700 #define BLK_BOUNCE_HIGH -1ULL
702 #define BLK_BOUNCE_ANY (-1ULL)
703 #define BLK_BOUNCE_ISA (ISA_DMA_THRESHOLD)
706 * default timeout for SG_IO if none specified
708 #define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
709 #define BLK_MIN_SG_TIMEOUT (7 * HZ)
712 extern int init_emergency_isa_pool(void);
713 extern void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
);
715 static inline int init_emergency_isa_pool(void)
719 static inline void blk_queue_bounce(struct request_queue
*q
, struct bio
**bio
)
722 #endif /* CONFIG_MMU */
728 unsigned long offset
;
733 struct req_iterator
{
738 /* This should not be used directly - use rq_for_each_segment */
739 #define for_each_bio(_bio) \
740 for (; _bio; _bio = _bio->bi_next)
741 #define __rq_for_each_bio(_bio, rq) \
743 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
745 #define rq_for_each_segment(bvl, _rq, _iter) \
746 __rq_for_each_bio(_iter.bio, _rq) \
747 bio_for_each_segment(bvl, _iter.bio, _iter.i)
749 #define rq_iter_last(rq, _iter) \
750 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
752 extern int blk_register_queue(struct gendisk
*disk
);
753 extern void blk_unregister_queue(struct gendisk
*disk
);
754 extern void register_disk(struct gendisk
*dev
);
755 extern void generic_make_request(struct bio
*bio
);
756 extern void blk_rq_init(struct request_queue
*q
, struct request
*rq
);
757 extern void blk_put_request(struct request
*);
758 extern void __blk_put_request(struct request_queue
*, struct request
*);
759 extern struct request
*blk_get_request(struct request_queue
*, int, gfp_t
);
760 extern struct request
*blk_make_request(struct request_queue
*, struct bio
*,
762 extern void blk_insert_request(struct request_queue
*, struct request
*, int, void *);
763 extern void blk_requeue_request(struct request_queue
*, struct request
*);
764 extern int blk_rq_check_limits(struct request_queue
*q
, struct request
*rq
);
765 extern int blk_lld_busy(struct request_queue
*q
);
766 extern int blk_rq_prep_clone(struct request
*rq
, struct request
*rq_src
,
767 struct bio_set
*bs
, gfp_t gfp_mask
,
768 int (*bio_ctr
)(struct bio
*, struct bio
*, void *),
770 extern void blk_rq_unprep_clone(struct request
*rq
);
771 extern int blk_insert_cloned_request(struct request_queue
*q
,
773 extern void blk_plug_device(struct request_queue
*);
774 extern void blk_plug_device_unlocked(struct request_queue
*);
775 extern int blk_remove_plug(struct request_queue
*);
776 extern void blk_recount_segments(struct request_queue
*, struct bio
*);
777 extern int scsi_cmd_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
778 unsigned int, void __user
*);
779 extern int sg_scsi_ioctl(struct request_queue
*, struct gendisk
*, fmode_t
,
780 struct scsi_ioctl_command __user
*);
783 * A queue has just exitted congestion. Note this in the global counter of
784 * congested queues, and wake up anyone who was waiting for requests to be
787 static inline void blk_clear_queue_congested(struct request_queue
*q
, int sync
)
789 clear_bdi_congested(&q
->backing_dev_info
, sync
);
793 * A queue has just entered congestion. Flag that in the queue's VM-visible
794 * state flags and increment the global gounter of congested queues.
796 static inline void blk_set_queue_congested(struct request_queue
*q
, int sync
)
798 set_bdi_congested(&q
->backing_dev_info
, sync
);
801 extern void blk_start_queue(struct request_queue
*q
);
802 extern void blk_stop_queue(struct request_queue
*q
);
803 extern void blk_sync_queue(struct request_queue
*q
);
804 extern void __blk_stop_queue(struct request_queue
*q
);
805 extern void __blk_run_queue(struct request_queue
*);
806 extern void blk_run_queue(struct request_queue
*);
807 extern int blk_rq_map_user(struct request_queue
*, struct request
*,
808 struct rq_map_data
*, void __user
*, unsigned long,
810 extern int blk_rq_unmap_user(struct bio
*);
811 extern int blk_rq_map_kern(struct request_queue
*, struct request
*, void *, unsigned int, gfp_t
);
812 extern int blk_rq_map_user_iov(struct request_queue
*, struct request
*,
813 struct rq_map_data
*, struct sg_iovec
*, int,
814 unsigned int, gfp_t
);
815 extern int blk_execute_rq(struct request_queue
*, struct gendisk
*,
816 struct request
*, int);
817 extern void blk_execute_rq_nowait(struct request_queue
*, struct gendisk
*,
818 struct request
*, int, rq_end_io_fn
*);
819 extern void blk_unplug(struct request_queue
*q
);
821 static inline struct request_queue
*bdev_get_queue(struct block_device
*bdev
)
823 return bdev
->bd_disk
->queue
;
826 static inline void blk_run_backing_dev(struct backing_dev_info
*bdi
,
829 if (bdi
&& bdi
->unplug_io_fn
)
830 bdi
->unplug_io_fn(bdi
, page
);
833 static inline void blk_run_address_space(struct address_space
*mapping
)
836 blk_run_backing_dev(mapping
->backing_dev_info
, NULL
);
840 * blk_rq_pos() : the current sector
841 * blk_rq_bytes() : bytes left in the entire request
842 * blk_rq_cur_bytes() : bytes left in the current segment
843 * blk_rq_err_bytes() : bytes left till the next error boundary
844 * blk_rq_sectors() : sectors left in the entire request
845 * blk_rq_cur_sectors() : sectors left in the current segment
846 * blk_rq_err_sectors() : sectors left till the next error boundary
848 static inline sector_t
blk_rq_pos(const struct request
*rq
)
853 static inline unsigned int blk_rq_bytes(const struct request
*rq
)
855 return rq
->__data_len
;
858 static inline int blk_rq_cur_bytes(const struct request
*rq
)
860 return rq
->bio
? bio_cur_bytes(rq
->bio
) : 0;
863 extern unsigned int blk_rq_err_bytes(const struct request
*rq
);
865 static inline unsigned int blk_rq_sectors(const struct request
*rq
)
867 return blk_rq_bytes(rq
) >> 9;
870 static inline unsigned int blk_rq_cur_sectors(const struct request
*rq
)
872 return blk_rq_cur_bytes(rq
) >> 9;
875 static inline unsigned int blk_rq_err_sectors(const struct request
*rq
)
877 return blk_rq_err_bytes(rq
) >> 9;
881 * Request issue related functions.
883 extern struct request
*blk_peek_request(struct request_queue
*q
);
884 extern void blk_start_request(struct request
*rq
);
885 extern struct request
*blk_fetch_request(struct request_queue
*q
);
888 * Request completion related functions.
890 * blk_update_request() completes given number of bytes and updates
891 * the request without completing it.
893 * blk_end_request() and friends. __blk_end_request() must be called
894 * with the request queue spinlock acquired.
896 * Several drivers define their own end_request and call
897 * blk_end_request() for parts of the original function.
898 * This prevents code duplication in drivers.
900 extern bool blk_update_request(struct request
*rq
, int error
,
901 unsigned int nr_bytes
);
902 extern bool blk_end_request(struct request
*rq
, int error
,
903 unsigned int nr_bytes
);
904 extern void blk_end_request_all(struct request
*rq
, int error
);
905 extern bool blk_end_request_cur(struct request
*rq
, int error
);
906 extern bool blk_end_request_err(struct request
*rq
, int error
);
907 extern bool __blk_end_request(struct request
*rq
, int error
,
908 unsigned int nr_bytes
);
909 extern void __blk_end_request_all(struct request
*rq
, int error
);
910 extern bool __blk_end_request_cur(struct request
*rq
, int error
);
911 extern bool __blk_end_request_err(struct request
*rq
, int error
);
913 extern void blk_complete_request(struct request
*);
914 extern void __blk_complete_request(struct request
*);
915 extern void blk_abort_request(struct request
*);
916 extern void blk_abort_queue(struct request_queue
*);
919 * Access functions for manipulating queue properties
921 extern struct request_queue
*blk_init_queue_node(request_fn_proc
*rfn
,
922 spinlock_t
*lock
, int node_id
);
923 extern struct request_queue
*blk_init_queue(request_fn_proc
*, spinlock_t
*);
924 extern void blk_cleanup_queue(struct request_queue
*);
925 extern void blk_queue_make_request(struct request_queue
*, make_request_fn
*);
926 extern void blk_queue_bounce_limit(struct request_queue
*, u64
);
927 extern void blk_queue_max_sectors(struct request_queue
*, unsigned int);
928 extern void blk_queue_max_hw_sectors(struct request_queue
*, unsigned int);
929 extern void blk_queue_max_phys_segments(struct request_queue
*, unsigned short);
930 extern void blk_queue_max_hw_segments(struct request_queue
*, unsigned short);
931 extern void blk_queue_max_segment_size(struct request_queue
*, unsigned int);
932 extern void blk_queue_logical_block_size(struct request_queue
*, unsigned short);
933 extern void blk_queue_physical_block_size(struct request_queue
*, unsigned short);
934 extern void blk_queue_alignment_offset(struct request_queue
*q
,
935 unsigned int alignment
);
936 extern void blk_limits_io_min(struct queue_limits
*limits
, unsigned int min
);
937 extern void blk_queue_io_min(struct request_queue
*q
, unsigned int min
);
938 extern void blk_limits_io_opt(struct queue_limits
*limits
, unsigned int opt
);
939 extern void blk_queue_io_opt(struct request_queue
*q
, unsigned int opt
);
940 extern void blk_set_default_limits(struct queue_limits
*lim
);
941 extern int blk_stack_limits(struct queue_limits
*t
, struct queue_limits
*b
,
943 extern void disk_stack_limits(struct gendisk
*disk
, struct block_device
*bdev
,
945 extern void blk_queue_stack_limits(struct request_queue
*t
, struct request_queue
*b
);
946 extern void blk_queue_dma_pad(struct request_queue
*, unsigned int);
947 extern void blk_queue_update_dma_pad(struct request_queue
*, unsigned int);
948 extern int blk_queue_dma_drain(struct request_queue
*q
,
949 dma_drain_needed_fn
*dma_drain_needed
,
950 void *buf
, unsigned int size
);
951 extern void blk_queue_lld_busy(struct request_queue
*q
, lld_busy_fn
*fn
);
952 extern void blk_queue_segment_boundary(struct request_queue
*, unsigned long);
953 extern void blk_queue_prep_rq(struct request_queue
*, prep_rq_fn
*pfn
);
954 extern void blk_queue_merge_bvec(struct request_queue
*, merge_bvec_fn
*);
955 extern void blk_queue_dma_alignment(struct request_queue
*, int);
956 extern void blk_queue_update_dma_alignment(struct request_queue
*, int);
957 extern void blk_queue_softirq_done(struct request_queue
*, softirq_done_fn
*);
958 extern void blk_queue_set_discard(struct request_queue
*, prepare_discard_fn
*);
959 extern void blk_queue_rq_timed_out(struct request_queue
*, rq_timed_out_fn
*);
960 extern void blk_queue_rq_timeout(struct request_queue
*, unsigned int);
961 extern struct backing_dev_info
*blk_get_backing_dev_info(struct block_device
*bdev
);
962 extern int blk_queue_ordered(struct request_queue
*, unsigned, prepare_flush_fn
*);
963 extern bool blk_do_ordered(struct request_queue
*, struct request
**);
964 extern unsigned blk_ordered_cur_seq(struct request_queue
*);
965 extern unsigned blk_ordered_req_seq(struct request
*);
966 extern bool blk_ordered_complete_seq(struct request_queue
*, unsigned, int);
968 extern int blk_rq_map_sg(struct request_queue
*, struct request
*, struct scatterlist
*);
969 extern void blk_dump_rq_flags(struct request
*, char *);
970 extern void generic_unplug_device(struct request_queue
*);
971 extern long nr_blockdev_pages(void);
973 int blk_get_queue(struct request_queue
*);
974 struct request_queue
*blk_alloc_queue(gfp_t
);
975 struct request_queue
*blk_alloc_queue_node(gfp_t
, int);
976 extern void blk_put_queue(struct request_queue
*);
981 #define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
982 extern int blk_queue_start_tag(struct request_queue
*, struct request
*);
983 extern struct request
*blk_queue_find_tag(struct request_queue
*, int);
984 extern void blk_queue_end_tag(struct request_queue
*, struct request
*);
985 extern int blk_queue_init_tags(struct request_queue
*, int, struct blk_queue_tag
*);
986 extern void blk_queue_free_tags(struct request_queue
*);
987 extern int blk_queue_resize_tags(struct request_queue
*, int);
988 extern void blk_queue_invalidate_tags(struct request_queue
*);
989 extern struct blk_queue_tag
*blk_init_tags(int);
990 extern void blk_free_tags(struct blk_queue_tag
*);
992 static inline struct request
*blk_map_queue_find_tag(struct blk_queue_tag
*bqt
,
995 if (unlikely(bqt
== NULL
|| tag
>= bqt
->real_max_depth
))
997 return bqt
->tag_index
[tag
];
1000 extern int blkdev_issue_flush(struct block_device
*, sector_t
*);
1001 #define DISCARD_FL_WAIT 0x01 /* wait for completion */
1002 #define DISCARD_FL_BARRIER 0x02 /* issue DISCARD_BARRIER request */
1003 extern int blkdev_issue_discard(struct block_device
*, sector_t sector
,
1004 sector_t nr_sects
, gfp_t
, int flags
);
1006 static inline int sb_issue_discard(struct super_block
*sb
,
1007 sector_t block
, sector_t nr_blocks
)
1009 block
<<= (sb
->s_blocksize_bits
- 9);
1010 nr_blocks
<<= (sb
->s_blocksize_bits
- 9);
1011 return blkdev_issue_discard(sb
->s_bdev
, block
, nr_blocks
, GFP_KERNEL
,
1012 DISCARD_FL_BARRIER
);
1015 extern int blk_verify_command(unsigned char *cmd
, fmode_t has_write_perm
);
1017 #define MAX_PHYS_SEGMENTS 128
1018 #define MAX_HW_SEGMENTS 128
1019 #define SAFE_MAX_SECTORS 255
1020 #define BLK_DEF_MAX_SECTORS 1024
1022 #define MAX_SEGMENT_SIZE 65536
1024 #define BLK_SEG_BOUNDARY_MASK 0xFFFFFFFFUL
1026 #define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
1028 static inline unsigned long queue_bounce_pfn(struct request_queue
*q
)
1030 return q
->limits
.bounce_pfn
;
1033 static inline unsigned long queue_segment_boundary(struct request_queue
*q
)
1035 return q
->limits
.seg_boundary_mask
;
1038 static inline unsigned int queue_max_sectors(struct request_queue
*q
)
1040 return q
->limits
.max_sectors
;
1043 static inline unsigned int queue_max_hw_sectors(struct request_queue
*q
)
1045 return q
->limits
.max_hw_sectors
;
1048 static inline unsigned short queue_max_hw_segments(struct request_queue
*q
)
1050 return q
->limits
.max_hw_segments
;
1053 static inline unsigned short queue_max_phys_segments(struct request_queue
*q
)
1055 return q
->limits
.max_phys_segments
;
1058 static inline unsigned int queue_max_segment_size(struct request_queue
*q
)
1060 return q
->limits
.max_segment_size
;
1063 static inline unsigned short queue_logical_block_size(struct request_queue
*q
)
1067 if (q
&& q
->limits
.logical_block_size
)
1068 retval
= q
->limits
.logical_block_size
;
1073 static inline unsigned short bdev_logical_block_size(struct block_device
*bdev
)
1075 return queue_logical_block_size(bdev_get_queue(bdev
));
1078 static inline unsigned int queue_physical_block_size(struct request_queue
*q
)
1080 return q
->limits
.physical_block_size
;
1083 static inline unsigned int queue_io_min(struct request_queue
*q
)
1085 return q
->limits
.io_min
;
1088 static inline unsigned int queue_io_opt(struct request_queue
*q
)
1090 return q
->limits
.io_opt
;
1093 static inline int queue_alignment_offset(struct request_queue
*q
)
1095 if (q
&& q
->limits
.misaligned
)
1098 if (q
&& q
->limits
.alignment_offset
)
1099 return q
->limits
.alignment_offset
;
1104 static inline int queue_sector_alignment_offset(struct request_queue
*q
,
1107 return ((sector
<< 9) - q
->limits
.alignment_offset
)
1108 & (q
->limits
.io_min
- 1);
1111 static inline int queue_dma_alignment(struct request_queue
*q
)
1113 return q
? q
->dma_alignment
: 511;
1116 static inline int blk_rq_aligned(struct request_queue
*q
, void *addr
,
1119 unsigned int alignment
= queue_dma_alignment(q
) | q
->dma_pad_mask
;
1120 return !((unsigned long)addr
& alignment
) && !(len
& alignment
);
1123 /* assumes size > 256 */
1124 static inline unsigned int blksize_bits(unsigned int size
)
1126 unsigned int bits
= 8;
1130 } while (size
> 256);
1134 static inline unsigned int block_size(struct block_device
*bdev
)
1136 return bdev
->bd_block_size
;
1139 typedef struct {struct page
*v
;} Sector
;
1141 unsigned char *read_dev_sector(struct block_device
*, sector_t
, Sector
*);
1143 static inline void put_dev_sector(Sector p
)
1145 page_cache_release(p
.v
);
1149 int kblockd_schedule_work(struct request_queue
*q
, struct work_struct
*work
);
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
{
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
;
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
*, struct scatterlist
*);
1195 extern int blk_rq_count_integrity_sg(struct request
*);
1198 struct blk_integrity
*bdev_get_integrity(struct block_device
*bdev
)
1200 return bdev
->bd_disk
->integrity
;
1203 static inline struct blk_integrity
*blk_get_integrity(struct gendisk
*disk
)
1205 return disk
->integrity
;
1208 static inline int blk_integrity_rq(struct request
*rq
)
1210 if (rq
->bio
== NULL
)
1213 return bio_integrity(rq
->bio
);
1216 #else /* CONFIG_BLK_DEV_INTEGRITY */
1218 #define blk_integrity_rq(rq) (0)
1219 #define blk_rq_count_integrity_sg(a) (0)
1220 #define blk_rq_map_integrity_sg(a, b) (0)
1221 #define bdev_get_integrity(a) (0)
1222 #define blk_get_integrity(a) (0)
1223 #define blk_integrity_compare(a, b) (0)
1224 #define blk_integrity_register(a, b) (0)
1225 #define blk_integrity_unregister(a) do { } while (0);
1227 #endif /* CONFIG_BLK_DEV_INTEGRITY */
1229 struct block_device_operations
{
1230 int (*open
) (struct block_device
*, fmode_t
);
1231 int (*release
) (struct gendisk
*, fmode_t
);
1232 int (*locked_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1233 int (*ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1234 int (*compat_ioctl
) (struct block_device
*, fmode_t
, unsigned, unsigned long);
1235 int (*direct_access
) (struct block_device
*, sector_t
,
1236 void **, unsigned long *);
1237 int (*media_changed
) (struct gendisk
*);
1238 unsigned long long (*set_capacity
) (struct gendisk
*,
1239 unsigned long long);
1240 int (*revalidate_disk
) (struct gendisk
*);
1241 int (*getgeo
)(struct block_device
*, struct hd_geometry
*);
1242 struct module
*owner
;
1245 extern int __blkdev_driver_ioctl(struct block_device
*, fmode_t
, unsigned int,
1247 #else /* CONFIG_BLOCK */
1249 * stubs for when the block layer is configured out
1251 #define buffer_heads_over_limit 0
1253 static inline long nr_blockdev_pages(void)
1258 #endif /* CONFIG_BLOCK */