]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - include/linux/blkdev.h
Fix off-by-one error in iov_iter_advance()
[mirror_ubuntu-artful-kernel.git] / include / linux / blkdev.h
CommitLineData
1da177e4
LT
1#ifndef _LINUX_BLKDEV_H
2#define _LINUX_BLKDEV_H
3
f5ff8422
JA
4#ifdef CONFIG_BLOCK
5
bcfd8d36 6#include <linux/sched.h>
1da177e4
LT
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>
d351af01 19#include <linux/bsg.h>
1da177e4
LT
20
21#include <asm/scatterlist.h>
22
21b2f0c8
CH
23struct scsi_ioctl_command;
24
1da177e4 25struct request_queue;
1da177e4
LT
26struct elevator_queue;
27typedef struct elevator_queue elevator_t;
28struct request_pm_state;
2056a782 29struct blk_trace;
3d6392cf
JA
30struct request;
31struct sg_io_hdr;
1da177e4
LT
32
33#define BLKDEV_MIN_RQ 4
34#define BLKDEV_MAX_RQ 128 /* Default maximum */
35
1da177e4 36struct request;
8ffdc655 37typedef void (rq_end_io_fn)(struct request *, int);
1da177e4
LT
38
39struct request_list {
40 int count[2];
41 int starved[2];
cb98fc8b 42 int elvpriv;
1da177e4
LT
43 mempool_t *rq_pool;
44 wait_queue_head_t wait[2];
1da177e4
LT
45};
46
4aff5e23
JA
47/*
48 * request command types
49 */
50enum rq_cmd_type_bits {
51 REQ_TYPE_FS = 1, /* fs request */
52 REQ_TYPE_BLOCK_PC, /* scsi command */
53 REQ_TYPE_SENSE, /* sense request */
54 REQ_TYPE_PM_SUSPEND, /* suspend request */
55 REQ_TYPE_PM_RESUME, /* resume request */
56 REQ_TYPE_PM_SHUTDOWN, /* shutdown request */
57 REQ_TYPE_FLUSH, /* flush request */
58 REQ_TYPE_SPECIAL, /* driver defined type */
59 REQ_TYPE_LINUX_BLOCK, /* generic block layer message */
60 /*
61 * for ATA/ATAPI devices. this really doesn't belong here, ide should
62 * use REQ_TYPE_SPECIAL and use rq->cmd[0] with the range of driver
63 * private REQ_LB opcodes to differentiate what type of request this is
64 */
4aff5e23 65 REQ_TYPE_ATA_TASKFILE,
cea2885a 66 REQ_TYPE_ATA_PC,
4aff5e23
JA
67};
68
69/*
70 * For request of type REQ_TYPE_LINUX_BLOCK, rq->cmd[0] is the opcode being
71 * sent down (similar to how REQ_TYPE_BLOCK_PC means that ->cmd[] holds a
72 * SCSI cdb.
73 *
74 * 0x00 -> 0x3f are driver private, to be used for whatever purpose they need,
75 * typically to differentiate REQ_TYPE_SPECIAL requests.
76 *
77 */
78enum {
79 /*
80 * just examples for now
81 */
82 REQ_LB_OP_EJECT = 0x40, /* eject request */
83 REQ_LB_OP_FLUSH = 0x41, /* flush device */
84};
85
86/*
87 * request type modified bits. first three bits match BIO_RW* bits, important
88 */
89enum rq_flag_bits {
90 __REQ_RW, /* not set, read. set, write */
91 __REQ_FAILFAST, /* no low level driver retries */
92 __REQ_SORTED, /* elevator knows about this request */
93 __REQ_SOFTBARRIER, /* may not be passed by ioscheduler */
94 __REQ_HARDBARRIER, /* may not be passed by drive either */
95 __REQ_FUA, /* forced unit access */
96 __REQ_NOMERGE, /* don't touch this for merging */
97 __REQ_STARTED, /* drive already may have started this one */
98 __REQ_DONTPREP, /* don't call prep for this one */
99 __REQ_QUEUED, /* uses queueing */
100 __REQ_ELVPRIV, /* elevator private data attached */
101 __REQ_FAILED, /* set if the request failed */
102 __REQ_QUIET, /* don't worry about errors */
103 __REQ_PREEMPT, /* set for "ide_preempt" requests */
104 __REQ_ORDERED_COLOR, /* is before or after barrier */
105 __REQ_RW_SYNC, /* request is sync (O_DIRECT) */
49171e5c 106 __REQ_ALLOCED, /* request came from our alloc pool */
5404bc7a 107 __REQ_RW_META, /* metadata io request */
f18573ab 108 __REQ_COPY_USER, /* contains copies of user pages */
7ba1ba12 109 __REQ_INTEGRITY, /* integrity metadata has been remapped */
4aff5e23
JA
110 __REQ_NR_BITS, /* stops here */
111};
112
113#define REQ_RW (1 << __REQ_RW)
114#define REQ_FAILFAST (1 << __REQ_FAILFAST)
115#define REQ_SORTED (1 << __REQ_SORTED)
116#define REQ_SOFTBARRIER (1 << __REQ_SOFTBARRIER)
117#define REQ_HARDBARRIER (1 << __REQ_HARDBARRIER)
118#define REQ_FUA (1 << __REQ_FUA)
119#define REQ_NOMERGE (1 << __REQ_NOMERGE)
120#define REQ_STARTED (1 << __REQ_STARTED)
121#define REQ_DONTPREP (1 << __REQ_DONTPREP)
122#define REQ_QUEUED (1 << __REQ_QUEUED)
123#define REQ_ELVPRIV (1 << __REQ_ELVPRIV)
124#define REQ_FAILED (1 << __REQ_FAILED)
125#define REQ_QUIET (1 << __REQ_QUIET)
126#define REQ_PREEMPT (1 << __REQ_PREEMPT)
127#define REQ_ORDERED_COLOR (1 << __REQ_ORDERED_COLOR)
128#define REQ_RW_SYNC (1 << __REQ_RW_SYNC)
49171e5c 129#define REQ_ALLOCED (1 << __REQ_ALLOCED)
5404bc7a 130#define REQ_RW_META (1 << __REQ_RW_META)
f18573ab 131#define REQ_COPY_USER (1 << __REQ_COPY_USER)
7ba1ba12 132#define REQ_INTEGRITY (1 << __REQ_INTEGRITY)
4aff5e23 133
1da177e4
LT
134#define BLK_MAX_CDB 16
135
136/*
63a71386
JA
137 * try to put the fields that are referenced together in the same cacheline.
138 * if you modify this structure, be sure to check block/blk-core.c:rq_init()
139 * as well!
1da177e4
LT
140 */
141struct request {
ff856bad
JA
142 struct list_head queuelist;
143 struct list_head donelist;
144
165125e1 145 struct request_queue *q;
e6a1c874 146
4aff5e23
JA
147 unsigned int cmd_flags;
148 enum rq_cmd_type_bits cmd_type;
1da177e4
LT
149
150 /* Maintain bio traversal state for part by part I/O submission.
151 * hard_* are block layer internals, no driver should touch them!
152 */
153
154 sector_t sector; /* next sector to submit */
e6a1c874 155 sector_t hard_sector; /* next sector to complete */
1da177e4 156 unsigned long nr_sectors; /* no. of sectors left to submit */
e6a1c874 157 unsigned long hard_nr_sectors; /* no. of sectors left to complete */
1da177e4
LT
158 /* no. of sectors left to submit in the current segment */
159 unsigned int current_nr_sectors;
160
1da177e4
LT
161 /* no. of sectors left to complete in the current segment */
162 unsigned int hard_cur_sectors;
163
164 struct bio *bio;
165 struct bio *biotail;
166
9817064b 167 struct hlist_node hash; /* merge hash */
e6a1c874
JA
168 /*
169 * The rb_node is only used inside the io scheduler, requests
170 * are pruned when moved to the dispatch queue. So let the
171 * completion_data share space with the rb_node.
172 */
173 union {
174 struct rb_node rb_node; /* sort/lookup */
175 void *completion_data;
176 };
9817064b 177
ff7d145f
JA
178 /*
179 * two pointers are available for the IO schedulers, if they need
180 * more they have to dynamically allocate it.
181 */
1da177e4 182 void *elevator_private;
ff7d145f
JA
183 void *elevator_private2;
184
8f34ee75 185 struct gendisk *rq_disk;
1da177e4
LT
186 unsigned long start_time;
187
188 /* Number of scatter-gather DMA addr+len pairs after
189 * physical address coalescing is performed.
190 */
191 unsigned short nr_phys_segments;
192
193 /* Number of scatter-gather addr+len pairs after
194 * physical and DMA remapping hardware coalescing is performed.
195 * This is the number of scatter-gather entries the driver
196 * will actually have to deal with after DMA mapping is done.
197 */
198 unsigned short nr_hw_segments;
199
8f34ee75
JA
200 unsigned short ioprio;
201
1da177e4 202 void *special;
8f34ee75 203 char *buffer;
1da177e4 204
cdd60262
JA
205 int tag;
206 int errors;
207
208 int ref_count;
209
1da177e4
LT
210 /*
211 * when request is used as a packet command carrier
212 */
d7e3c324
FT
213 unsigned short cmd_len;
214 unsigned char __cmd[BLK_MAX_CDB];
215 unsigned char *cmd;
1da177e4
LT
216
217 unsigned int data_len;
7a85f889 218 unsigned int extra_len; /* length of alignment and padding */
1da177e4 219 unsigned int sense_len;
8f34ee75 220 void *data;
1da177e4
LT
221 void *sense;
222
223 unsigned int timeout;
17e01f21 224 int retries;
1da177e4 225
1da177e4 226 /*
c00895ab 227 * completion callback.
1da177e4
LT
228 */
229 rq_end_io_fn *end_io;
230 void *end_io_data;
abae1fde
FT
231
232 /* for bidi */
233 struct request *next_rq;
1da177e4
LT
234};
235
236/*
4aff5e23 237 * State information carried for REQ_TYPE_PM_SUSPEND and REQ_TYPE_PM_RESUME
1da177e4
LT
238 * requests. Some step values could eventually be made generic.
239 */
240struct request_pm_state
241{
242 /* PM state machine step value, currently driver specific */
243 int pm_step;
244 /* requested PM state value (S1, S2, S3, S4, ...) */
245 u32 pm_state;
246 void* data; /* for driver use */
247};
248
249#include <linux/elevator.h>
250
165125e1
JA
251typedef void (request_fn_proc) (struct request_queue *q);
252typedef int (make_request_fn) (struct request_queue *q, struct bio *bio);
253typedef int (prep_rq_fn) (struct request_queue *, struct request *);
254typedef void (unplug_fn) (struct request_queue *);
1da177e4
LT
255
256struct bio_vec;
cc371e66
AK
257struct bvec_merge_data {
258 struct block_device *bi_bdev;
259 sector_t bi_sector;
260 unsigned bi_size;
261 unsigned long bi_rw;
262};
263typedef int (merge_bvec_fn) (struct request_queue *, struct bvec_merge_data *,
264 struct bio_vec *);
165125e1 265typedef void (prepare_flush_fn) (struct request_queue *, struct request *);
ff856bad 266typedef void (softirq_done_fn)(struct request *);
2fb98e84 267typedef int (dma_drain_needed_fn)(struct request *);
1da177e4
LT
268
269enum blk_queue_state {
270 Queue_down,
271 Queue_up,
272};
273
1da177e4
LT
274struct blk_queue_tag {
275 struct request **tag_index; /* map of busy tags */
276 unsigned long *tag_map; /* bit map of free/busy tags */
1da177e4
LT
277 int busy; /* current depth */
278 int max_depth; /* what we will send to device */
ba025082 279 int real_max_depth; /* what the array can hold */
1da177e4
LT
280 atomic_t refcnt; /* map can be shared */
281};
282
283struct request_queue
284{
285 /*
286 * Together with queue_head for cacheline sharing
287 */
288 struct list_head queue_head;
289 struct request *last_merge;
290 elevator_t *elevator;
291
292 /*
293 * the queue request freelist, one for reads and one for writes
294 */
295 struct request_list rq;
296
297 request_fn_proc *request_fn;
1da177e4
LT
298 make_request_fn *make_request_fn;
299 prep_rq_fn *prep_rq_fn;
300 unplug_fn *unplug_fn;
301 merge_bvec_fn *merge_bvec_fn;
1da177e4 302 prepare_flush_fn *prepare_flush_fn;
ff856bad 303 softirq_done_fn *softirq_done_fn;
2fb98e84 304 dma_drain_needed_fn *dma_drain_needed;
1da177e4 305
8922e16c
TH
306 /*
307 * Dispatch queue sorting
308 */
1b47f531 309 sector_t end_sector;
8922e16c 310 struct request *boundary_rq;
8922e16c 311
1da177e4
LT
312 /*
313 * Auto-unplugging state
314 */
315 struct timer_list unplug_timer;
316 int unplug_thresh; /* After this many requests */
317 unsigned long unplug_delay; /* After this many jiffies */
318 struct work_struct unplug_work;
319
320 struct backing_dev_info backing_dev_info;
321
322 /*
323 * The queue owner gets to use this for whatever they like.
324 * ll_rw_blk doesn't touch it.
325 */
326 void *queuedata;
327
1da177e4
LT
328 /*
329 * queue needs bounce pages for pages above this limit
330 */
331 unsigned long bounce_pfn;
8267e268 332 gfp_t bounce_gfp;
1da177e4
LT
333
334 /*
335 * various queue flags, see QUEUE_* below
336 */
337 unsigned long queue_flags;
338
339 /*
152587de
JA
340 * protects queue structures from reentrancy. ->__queue_lock should
341 * _never_ be used directly, it is queue private. always use
342 * ->queue_lock.
1da177e4 343 */
152587de 344 spinlock_t __queue_lock;
1da177e4
LT
345 spinlock_t *queue_lock;
346
347 /*
348 * queue kobject
349 */
350 struct kobject kobj;
351
352 /*
353 * queue settings
354 */
355 unsigned long nr_requests; /* Max # of requests */
356 unsigned int nr_congestion_on;
357 unsigned int nr_congestion_off;
358 unsigned int nr_batching;
359
2cb2e147
JA
360 unsigned int max_sectors;
361 unsigned int max_hw_sectors;
1da177e4
LT
362 unsigned short max_phys_segments;
363 unsigned short max_hw_segments;
364 unsigned short hardsect_size;
365 unsigned int max_segment_size;
366
367 unsigned long seg_boundary_mask;
fa0ccd83
JB
368 void *dma_drain_buffer;
369 unsigned int dma_drain_size;
e3790c7d 370 unsigned int dma_pad_mask;
1da177e4
LT
371 unsigned int dma_alignment;
372
373 struct blk_queue_tag *queue_tags;
6eca9004 374 struct list_head tag_busy_list;
1da177e4 375
15853af9 376 unsigned int nr_sorted;
1da177e4
LT
377 unsigned int in_flight;
378
379 /*
380 * sg stuff
381 */
382 unsigned int sg_timeout;
383 unsigned int sg_reserved_size;
1946089a 384 int node;
6c5c9341 385#ifdef CONFIG_BLK_DEV_IO_TRACE
2056a782 386 struct blk_trace *blk_trace;
6c5c9341 387#endif
1da177e4
LT
388 /*
389 * reserved for flush operations
390 */
797e7dbb
TH
391 unsigned int ordered, next_ordered, ordseq;
392 int orderr, ordcolor;
393 struct request pre_flush_rq, bar_rq, post_flush_rq;
394 struct request *orig_bar_rq;
483f4afc
AV
395
396 struct mutex sysfs_lock;
d351af01
FT
397
398#if defined(CONFIG_BLK_DEV_BSG)
399 struct bsg_class_device bsg_dev;
400#endif
1da177e4
LT
401};
402
1da177e4
LT
403#define QUEUE_FLAG_CLUSTER 0 /* cluster several segments into 1 */
404#define QUEUE_FLAG_QUEUED 1 /* uses generic tag queueing */
405#define QUEUE_FLAG_STOPPED 2 /* queue is stopped */
4e97182a
QY
406#define QUEUE_FLAG_READFULL 3 /* read queue has been filled */
407#define QUEUE_FLAG_WRITEFULL 4 /* write queue has been filled */
1da177e4
LT
408#define QUEUE_FLAG_DEAD 5 /* queue being torn down */
409#define QUEUE_FLAG_REENTER 6 /* Re-entrancy avoidance */
410#define QUEUE_FLAG_PLUGGED 7 /* queue is plugged */
64521d1a 411#define QUEUE_FLAG_ELVSWITCH 8 /* don't use elevator, just do FIFO */
abae1fde 412#define QUEUE_FLAG_BIDI 9 /* queue supports bidi requests */
ac9fafa1 413#define QUEUE_FLAG_NOMERGES 10 /* disable merge attempts */
797e7dbb 414
8f45c1a5
LT
415static inline int queue_is_locked(struct request_queue *q)
416{
7663c1e2 417#ifdef CONFIG_SMP
8f45c1a5
LT
418 spinlock_t *lock = q->queue_lock;
419 return lock && spin_is_locked(lock);
7663c1e2
JA
420#else
421 return 1;
422#endif
8f45c1a5
LT
423}
424
75ad23bc
NP
425static inline void queue_flag_set_unlocked(unsigned int flag,
426 struct request_queue *q)
427{
428 __set_bit(flag, &q->queue_flags);
429}
430
e48ec690
JA
431static inline int queue_flag_test_and_clear(unsigned int flag,
432 struct request_queue *q)
433{
434 WARN_ON_ONCE(!queue_is_locked(q));
435
436 if (test_bit(flag, &q->queue_flags)) {
437 __clear_bit(flag, &q->queue_flags);
438 return 1;
439 }
440
441 return 0;
442}
443
444static inline int queue_flag_test_and_set(unsigned int flag,
445 struct request_queue *q)
446{
447 WARN_ON_ONCE(!queue_is_locked(q));
448
449 if (!test_bit(flag, &q->queue_flags)) {
450 __set_bit(flag, &q->queue_flags);
451 return 0;
452 }
453
454 return 1;
455}
456
75ad23bc
NP
457static inline void queue_flag_set(unsigned int flag, struct request_queue *q)
458{
8f45c1a5 459 WARN_ON_ONCE(!queue_is_locked(q));
75ad23bc
NP
460 __set_bit(flag, &q->queue_flags);
461}
462
463static inline void queue_flag_clear_unlocked(unsigned int flag,
464 struct request_queue *q)
465{
466 __clear_bit(flag, &q->queue_flags);
467}
468
469static inline void queue_flag_clear(unsigned int flag, struct request_queue *q)
470{
8f45c1a5 471 WARN_ON_ONCE(!queue_is_locked(q));
75ad23bc
NP
472 __clear_bit(flag, &q->queue_flags);
473}
474
797e7dbb
TH
475enum {
476 /*
477 * Hardbarrier is supported with one of the following methods.
478 *
479 * NONE : hardbarrier unsupported
480 * DRAIN : ordering by draining is enough
481 * DRAIN_FLUSH : ordering by draining w/ pre and post flushes
482 * DRAIN_FUA : ordering by draining w/ pre flush and FUA write
483 * TAG : ordering by tag is enough
484 * TAG_FLUSH : ordering by tag w/ pre and post flushes
485 * TAG_FUA : ordering by tag w/ pre flush and FUA write
486 */
487 QUEUE_ORDERED_NONE = 0x00,
488 QUEUE_ORDERED_DRAIN = 0x01,
489 QUEUE_ORDERED_TAG = 0x02,
490
491 QUEUE_ORDERED_PREFLUSH = 0x10,
492 QUEUE_ORDERED_POSTFLUSH = 0x20,
493 QUEUE_ORDERED_FUA = 0x40,
494
495 QUEUE_ORDERED_DRAIN_FLUSH = QUEUE_ORDERED_DRAIN |
496 QUEUE_ORDERED_PREFLUSH | QUEUE_ORDERED_POSTFLUSH,
497 QUEUE_ORDERED_DRAIN_FUA = QUEUE_ORDERED_DRAIN |
498 QUEUE_ORDERED_PREFLUSH | QUEUE_ORDERED_FUA,
499 QUEUE_ORDERED_TAG_FLUSH = QUEUE_ORDERED_TAG |
500 QUEUE_ORDERED_PREFLUSH | QUEUE_ORDERED_POSTFLUSH,
501 QUEUE_ORDERED_TAG_FUA = QUEUE_ORDERED_TAG |
502 QUEUE_ORDERED_PREFLUSH | QUEUE_ORDERED_FUA,
503
504 /*
505 * Ordered operation sequence
506 */
507 QUEUE_ORDSEQ_STARTED = 0x01, /* flushing in progress */
508 QUEUE_ORDSEQ_DRAIN = 0x02, /* waiting for the queue to be drained */
509 QUEUE_ORDSEQ_PREFLUSH = 0x04, /* pre-flushing in progress */
510 QUEUE_ORDSEQ_BAR = 0x08, /* original barrier req in progress */
511 QUEUE_ORDSEQ_POSTFLUSH = 0x10, /* post-flushing in progress */
512 QUEUE_ORDSEQ_DONE = 0x20,
513};
1da177e4
LT
514
515#define blk_queue_plugged(q) test_bit(QUEUE_FLAG_PLUGGED, &(q)->queue_flags)
516#define blk_queue_tagged(q) test_bit(QUEUE_FLAG_QUEUED, &(q)->queue_flags)
517#define blk_queue_stopped(q) test_bit(QUEUE_FLAG_STOPPED, &(q)->queue_flags)
ac9fafa1 518#define blk_queue_nomerges(q) test_bit(QUEUE_FLAG_NOMERGES, &(q)->queue_flags)
797e7dbb 519#define blk_queue_flushing(q) ((q)->ordseq)
1da177e4 520
4aff5e23
JA
521#define blk_fs_request(rq) ((rq)->cmd_type == REQ_TYPE_FS)
522#define blk_pc_request(rq) ((rq)->cmd_type == REQ_TYPE_BLOCK_PC)
523#define blk_special_request(rq) ((rq)->cmd_type == REQ_TYPE_SPECIAL)
524#define blk_sense_request(rq) ((rq)->cmd_type == REQ_TYPE_SENSE)
525
526#define blk_noretry_request(rq) ((rq)->cmd_flags & REQ_FAILFAST)
527#define blk_rq_started(rq) ((rq)->cmd_flags & REQ_STARTED)
1da177e4
LT
528
529#define blk_account_rq(rq) (blk_rq_started(rq) && blk_fs_request(rq))
530
4aff5e23
JA
531#define blk_pm_suspend_request(rq) ((rq)->cmd_type == REQ_TYPE_PM_SUSPEND)
532#define blk_pm_resume_request(rq) ((rq)->cmd_type == REQ_TYPE_PM_RESUME)
1da177e4 533#define blk_pm_request(rq) \
4aff5e23 534 (blk_pm_suspend_request(rq) || blk_pm_resume_request(rq))
1da177e4 535
4aff5e23
JA
536#define blk_sorted_rq(rq) ((rq)->cmd_flags & REQ_SORTED)
537#define blk_barrier_rq(rq) ((rq)->cmd_flags & REQ_HARDBARRIER)
538#define blk_fua_rq(rq) ((rq)->cmd_flags & REQ_FUA)
abae1fde 539#define blk_bidi_rq(rq) ((rq)->next_rq != NULL)
bf2de6f5 540#define blk_empty_barrier(rq) (blk_barrier_rq(rq) && blk_fs_request(rq) && !(rq)->hard_nr_sectors)
336cdb40
KU
541/* rq->queuelist of dequeued request must be list_empty() */
542#define blk_queued_rq(rq) (!list_empty(&(rq)->queuelist))
1da177e4
LT
543
544#define list_entry_rq(ptr) list_entry((ptr), struct request, queuelist)
545
4aff5e23 546#define rq_data_dir(rq) ((rq)->cmd_flags & 1)
1da177e4 547
9e2585a8
JA
548/*
549 * We regard a request as sync, if it's a READ or a SYNC write.
550 */
551#define rq_is_sync(rq) (rq_data_dir((rq)) == READ || (rq)->cmd_flags & REQ_RW_SYNC)
5404bc7a 552#define rq_is_meta(rq) ((rq)->cmd_flags & REQ_RW_META)
9e2585a8 553
1da177e4
LT
554static inline int blk_queue_full(struct request_queue *q, int rw)
555{
556 if (rw == READ)
557 return test_bit(QUEUE_FLAG_READFULL, &q->queue_flags);
558 return test_bit(QUEUE_FLAG_WRITEFULL, &q->queue_flags);
559}
560
561static inline void blk_set_queue_full(struct request_queue *q, int rw)
562{
563 if (rw == READ)
75ad23bc 564 queue_flag_set(QUEUE_FLAG_READFULL, q);
1da177e4 565 else
75ad23bc 566 queue_flag_set(QUEUE_FLAG_WRITEFULL, q);
1da177e4
LT
567}
568
569static inline void blk_clear_queue_full(struct request_queue *q, int rw)
570{
571 if (rw == READ)
75ad23bc 572 queue_flag_clear(QUEUE_FLAG_READFULL, q);
1da177e4 573 else
75ad23bc 574 queue_flag_clear(QUEUE_FLAG_WRITEFULL, q);
1da177e4
LT
575}
576
577
578/*
579 * mergeable request must not have _NOMERGE or _BARRIER bit set, nor may
580 * it already be started by driver.
581 */
582#define RQ_NOMERGE_FLAGS \
583 (REQ_NOMERGE | REQ_STARTED | REQ_HARDBARRIER | REQ_SOFTBARRIER)
584#define rq_mergeable(rq) \
4aff5e23 585 (!((rq)->cmd_flags & RQ_NOMERGE_FLAGS) && blk_fs_request((rq)))
1da177e4 586
1da177e4
LT
587/*
588 * q->prep_rq_fn return values
589 */
590#define BLKPREP_OK 0 /* serve it */
591#define BLKPREP_KILL 1 /* fatal error, kill */
592#define BLKPREP_DEFER 2 /* leave on queue */
593
594extern unsigned long blk_max_low_pfn, blk_max_pfn;
595
596/*
597 * standard bounce addresses:
598 *
599 * BLK_BOUNCE_HIGH : bounce all highmem pages
600 * BLK_BOUNCE_ANY : don't bounce anything
601 * BLK_BOUNCE_ISA : bounce pages above ISA DMA boundary
602 */
2472892a
AK
603
604#if BITS_PER_LONG == 32
1da177e4 605#define BLK_BOUNCE_HIGH ((u64)blk_max_low_pfn << PAGE_SHIFT)
2472892a
AK
606#else
607#define BLK_BOUNCE_HIGH -1ULL
608#endif
609#define BLK_BOUNCE_ANY (-1ULL)
1da177e4
LT
610#define BLK_BOUNCE_ISA (ISA_DMA_THRESHOLD)
611
3d6392cf
JA
612/*
613 * default timeout for SG_IO if none specified
614 */
615#define BLK_DEFAULT_SG_TIMEOUT (60 * HZ)
616
2a7326b5 617#ifdef CONFIG_BOUNCE
1da177e4 618extern int init_emergency_isa_pool(void);
165125e1 619extern void blk_queue_bounce(struct request_queue *q, struct bio **bio);
1da177e4
LT
620#else
621static inline int init_emergency_isa_pool(void)
622{
623 return 0;
624}
165125e1 625static inline void blk_queue_bounce(struct request_queue *q, struct bio **bio)
1da177e4
LT
626{
627}
628#endif /* CONFIG_MMU */
629
5705f702
N
630struct req_iterator {
631 int i;
632 struct bio *bio;
633};
634
635/* This should not be used directly - use rq_for_each_segment */
636#define __rq_for_each_bio(_bio, rq) \
1da177e4
LT
637 if ((rq->bio)) \
638 for (_bio = (rq)->bio; _bio; _bio = _bio->bi_next)
639
5705f702
N
640#define rq_for_each_segment(bvl, _rq, _iter) \
641 __rq_for_each_bio(_iter.bio, _rq) \
642 bio_for_each_segment(bvl, _iter.bio, _iter.i)
643
644#define rq_iter_last(rq, _iter) \
645 (_iter.bio->bi_next == NULL && _iter.i == _iter.bio->bi_vcnt-1)
646
1da177e4
LT
647extern int blk_register_queue(struct gendisk *disk);
648extern void blk_unregister_queue(struct gendisk *disk);
649extern void register_disk(struct gendisk *dev);
650extern void generic_make_request(struct bio *bio);
2a4aa30c 651extern void blk_rq_init(struct request_queue *q, struct request *rq);
1da177e4 652extern void blk_put_request(struct request *);
165125e1 653extern void __blk_put_request(struct request_queue *, struct request *);
165125e1
JA
654extern struct request *blk_get_request(struct request_queue *, int, gfp_t);
655extern void blk_insert_request(struct request_queue *, struct request *, int, void *);
656extern void blk_requeue_request(struct request_queue *, struct request *);
657extern void blk_plug_device(struct request_queue *);
658extern int blk_remove_plug(struct request_queue *);
659extern void blk_recount_segments(struct request_queue *, struct bio *);
45e79a3a
FT
660extern int scsi_cmd_ioctl(struct file *, struct request_queue *,
661 struct gendisk *, unsigned int, void __user *);
21b2f0c8
CH
662extern int sg_scsi_ioctl(struct file *, struct request_queue *,
663 struct gendisk *, struct scsi_ioctl_command __user *);
3fcfab16 664
1aa4f24f
JA
665/*
666 * Temporary export, until SCSI gets fixed up.
667 */
3001ca77
N
668extern int blk_rq_append_bio(struct request_queue *q, struct request *rq,
669 struct bio *bio);
1aa4f24f 670
3fcfab16
AM
671/*
672 * A queue has just exitted congestion. Note this in the global counter of
673 * congested queues, and wake up anyone who was waiting for requests to be
674 * put back.
675 */
165125e1 676static inline void blk_clear_queue_congested(struct request_queue *q, int rw)
3fcfab16
AM
677{
678 clear_bdi_congested(&q->backing_dev_info, rw);
679}
680
681/*
682 * A queue has just entered congestion. Flag that in the queue's VM-visible
683 * state flags and increment the global gounter of congested queues.
684 */
165125e1 685static inline void blk_set_queue_congested(struct request_queue *q, int rw)
3fcfab16
AM
686{
687 set_bdi_congested(&q->backing_dev_info, rw);
688}
689
165125e1
JA
690extern void blk_start_queue(struct request_queue *q);
691extern void blk_stop_queue(struct request_queue *q);
1da177e4 692extern void blk_sync_queue(struct request_queue *q);
165125e1 693extern void __blk_stop_queue(struct request_queue *q);
75ad23bc 694extern void __blk_run_queue(struct request_queue *);
165125e1
JA
695extern void blk_run_queue(struct request_queue *);
696extern void blk_start_queueing(struct request_queue *);
697extern int blk_rq_map_user(struct request_queue *, struct request *, void __user *, unsigned long);
8e5cfc45 698extern int blk_rq_unmap_user(struct bio *);
165125e1
JA
699extern int blk_rq_map_kern(struct request_queue *, struct request *, void *, unsigned int, gfp_t);
700extern int blk_rq_map_user_iov(struct request_queue *, struct request *,
0e75f906 701 struct sg_iovec *, int, unsigned int);
165125e1 702extern int blk_execute_rq(struct request_queue *, struct gendisk *,
994ca9a1 703 struct request *, int);
165125e1 704extern void blk_execute_rq_nowait(struct request_queue *, struct gendisk *,
15fc858a 705 struct request *, int, rq_end_io_fn *);
2ad8b1ef 706extern void blk_unplug(struct request_queue *q);
6e39b69e 707
165125e1 708static inline struct request_queue *bdev_get_queue(struct block_device *bdev)
1da177e4
LT
709{
710 return bdev->bd_disk->queue;
711}
712
713static inline void blk_run_backing_dev(struct backing_dev_info *bdi,
714 struct page *page)
715{
716 if (bdi && bdi->unplug_io_fn)
717 bdi->unplug_io_fn(bdi, page);
718}
719
720static inline void blk_run_address_space(struct address_space *mapping)
721{
722 if (mapping)
723 blk_run_backing_dev(mapping->backing_dev_info, NULL);
724}
725
726/*
3bcddeac
KU
727 * blk_end_request() and friends.
728 * __blk_end_request() and end_request() must be called with
729 * the request queue spinlock acquired.
1da177e4
LT
730 *
731 * Several drivers define their own end_request and call
3bcddeac
KU
732 * blk_end_request() for parts of the original function.
733 * This prevents code duplication in drivers.
1da177e4 734 */
22b13210
JA
735extern int blk_end_request(struct request *rq, int error,
736 unsigned int nr_bytes);
737extern int __blk_end_request(struct request *rq, int error,
738 unsigned int nr_bytes);
739extern int blk_end_bidi_request(struct request *rq, int error,
740 unsigned int nr_bytes, unsigned int bidi_bytes);
a0cd1285
JA
741extern void end_request(struct request *, int);
742extern void end_queued_request(struct request *, int);
743extern void end_dequeued_request(struct request *, int);
22b13210
JA
744extern int blk_end_request_callback(struct request *rq, int error,
745 unsigned int nr_bytes,
746 int (drv_callback)(struct request *));
ff856bad
JA
747extern void blk_complete_request(struct request *);
748
1da177e4 749/*
3b11313a
KU
750 * blk_end_request() takes bytes instead of sectors as a complete size.
751 * blk_rq_bytes() returns bytes left to complete in the entire request.
752 * blk_rq_cur_bytes() returns bytes left to complete in the current segment.
1da177e4 753 */
3b11313a
KU
754extern unsigned int blk_rq_bytes(struct request *rq);
755extern unsigned int blk_rq_cur_bytes(struct request *rq);
1da177e4
LT
756
757static inline void blkdev_dequeue_request(struct request *req)
758{
8922e16c 759 elv_dequeue_request(req->q, req);
1da177e4 760}
1da177e4 761
1da177e4
LT
762/*
763 * Access functions for manipulating queue properties
764 */
165125e1 765extern struct request_queue *blk_init_queue_node(request_fn_proc *rfn,
1946089a 766 spinlock_t *lock, int node_id);
165125e1
JA
767extern struct request_queue *blk_init_queue(request_fn_proc *, spinlock_t *);
768extern void blk_cleanup_queue(struct request_queue *);
769extern void blk_queue_make_request(struct request_queue *, make_request_fn *);
770extern void blk_queue_bounce_limit(struct request_queue *, u64);
771extern void blk_queue_max_sectors(struct request_queue *, unsigned int);
772extern void blk_queue_max_phys_segments(struct request_queue *, unsigned short);
773extern void blk_queue_max_hw_segments(struct request_queue *, unsigned short);
774extern void blk_queue_max_segment_size(struct request_queue *, unsigned int);
775extern void blk_queue_hardsect_size(struct request_queue *, unsigned short);
776extern void blk_queue_stack_limits(struct request_queue *t, struct request_queue *b);
e3790c7d 777extern void blk_queue_dma_pad(struct request_queue *, unsigned int);
27f8221a 778extern void blk_queue_update_dma_pad(struct request_queue *, unsigned int);
2fb98e84
TH
779extern int blk_queue_dma_drain(struct request_queue *q,
780 dma_drain_needed_fn *dma_drain_needed,
781 void *buf, unsigned int size);
165125e1
JA
782extern void blk_queue_segment_boundary(struct request_queue *, unsigned long);
783extern void blk_queue_prep_rq(struct request_queue *, prep_rq_fn *pfn);
784extern void blk_queue_merge_bvec(struct request_queue *, merge_bvec_fn *);
785extern void blk_queue_dma_alignment(struct request_queue *, int);
11c3e689 786extern void blk_queue_update_dma_alignment(struct request_queue *, int);
165125e1 787extern void blk_queue_softirq_done(struct request_queue *, softirq_done_fn *);
1da177e4 788extern struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev);
165125e1 789extern int blk_queue_ordered(struct request_queue *, unsigned, prepare_flush_fn *);
165125e1
JA
790extern int blk_do_ordered(struct request_queue *, struct request **);
791extern unsigned blk_ordered_cur_seq(struct request_queue *);
797e7dbb 792extern unsigned blk_ordered_req_seq(struct request *);
165125e1 793extern void blk_ordered_complete_seq(struct request_queue *, unsigned, int);
1da177e4 794
165125e1 795extern int blk_rq_map_sg(struct request_queue *, struct request *, struct scatterlist *);
1da177e4 796extern void blk_dump_rq_flags(struct request *, char *);
165125e1
JA
797extern void generic_unplug_device(struct request_queue *);
798extern void __generic_unplug_device(struct request_queue *);
1da177e4 799extern long nr_blockdev_pages(void);
1da177e4 800
165125e1
JA
801int blk_get_queue(struct request_queue *);
802struct request_queue *blk_alloc_queue(gfp_t);
803struct request_queue *blk_alloc_queue_node(gfp_t, int);
804extern void blk_put_queue(struct request_queue *);
1da177e4
LT
805
806/*
807 * tag stuff
808 */
809#define blk_queue_tag_depth(q) ((q)->queue_tags->busy)
810#define blk_queue_tag_queue(q) ((q)->queue_tags->busy < (q)->queue_tags->max_depth)
4aff5e23 811#define blk_rq_tagged(rq) ((rq)->cmd_flags & REQ_QUEUED)
165125e1
JA
812extern int blk_queue_start_tag(struct request_queue *, struct request *);
813extern struct request *blk_queue_find_tag(struct request_queue *, int);
814extern void blk_queue_end_tag(struct request_queue *, struct request *);
815extern int blk_queue_init_tags(struct request_queue *, int, struct blk_queue_tag *);
816extern void blk_queue_free_tags(struct request_queue *);
817extern int blk_queue_resize_tags(struct request_queue *, int);
818extern void blk_queue_invalidate_tags(struct request_queue *);
492dfb48
JB
819extern struct blk_queue_tag *blk_init_tags(int);
820extern void blk_free_tags(struct blk_queue_tag *);
1da177e4 821
f583f492
DS
822static inline struct request *blk_map_queue_find_tag(struct blk_queue_tag *bqt,
823 int tag)
824{
825 if (unlikely(bqt == NULL || tag >= bqt->real_max_depth))
826 return NULL;
827 return bqt->tag_index[tag];
828}
829
1da177e4
LT
830extern int blkdev_issue_flush(struct block_device *, sector_t *);
831
0b07de85
AG
832/*
833* command filter functions
834*/
835extern int blk_verify_command(struct file *file, unsigned char *cmd);
836extern int blk_cmd_filter_verify_command(struct blk_scsi_cmd_filter *filter,
837 unsigned char *cmd, mode_t *f_mode);
838extern int blk_register_filter(struct gendisk *disk);
839extern void blk_unregister_filter(struct gendisk *disk);
840
1da177e4
LT
841#define MAX_PHYS_SEGMENTS 128
842#define MAX_HW_SEGMENTS 128
defd94b7
MC
843#define SAFE_MAX_SECTORS 255
844#define BLK_DEF_MAX_SECTORS 1024
1da177e4
LT
845
846#define MAX_SEGMENT_SIZE 65536
847
848#define blkdev_entry_to_request(entry) list_entry((entry), struct request, queuelist)
849
165125e1 850static inline int queue_hardsect_size(struct request_queue *q)
1da177e4
LT
851{
852 int retval = 512;
853
854 if (q && q->hardsect_size)
855 retval = q->hardsect_size;
856
857 return retval;
858}
859
860static inline int bdev_hardsect_size(struct block_device *bdev)
861{
862 return queue_hardsect_size(bdev_get_queue(bdev));
863}
864
165125e1 865static inline int queue_dma_alignment(struct request_queue *q)
1da177e4 866{
482eb689 867 return q ? q->dma_alignment : 511;
1da177e4
LT
868}
869
1da177e4
LT
870/* assumes size > 256 */
871static inline unsigned int blksize_bits(unsigned int size)
872{
873 unsigned int bits = 8;
874 do {
875 bits++;
876 size >>= 1;
877 } while (size > 256);
878 return bits;
879}
880
2befb9e3 881static inline unsigned int block_size(struct block_device *bdev)
1da177e4
LT
882{
883 return bdev->bd_block_size;
884}
885
886typedef struct {struct page *v;} Sector;
887
888unsigned char *read_dev_sector(struct block_device *, sector_t, Sector *);
889
890static inline void put_dev_sector(Sector p)
891{
892 page_cache_release(p.v);
893}
894
895struct work_struct;
896int kblockd_schedule_work(struct work_struct *work);
19a75d83 897void kblockd_flush_work(struct work_struct *work);
1da177e4 898
1da177e4
LT
899#define MODULE_ALIAS_BLOCKDEV(major,minor) \
900 MODULE_ALIAS("block-major-" __stringify(major) "-" __stringify(minor))
901#define MODULE_ALIAS_BLOCKDEV_MAJOR(major) \
902 MODULE_ALIAS("block-major-" __stringify(major) "-*")
903
7ba1ba12
MP
904#if defined(CONFIG_BLK_DEV_INTEGRITY)
905
b24498d4
JA
906#define INTEGRITY_FLAG_READ 2 /* verify data integrity on read */
907#define INTEGRITY_FLAG_WRITE 4 /* generate data integrity on write */
7ba1ba12
MP
908
909struct blk_integrity_exchg {
910 void *prot_buf;
911 void *data_buf;
912 sector_t sector;
913 unsigned int data_size;
914 unsigned short sector_size;
915 const char *disk_name;
916};
917
918typedef void (integrity_gen_fn) (struct blk_integrity_exchg *);
919typedef int (integrity_vrfy_fn) (struct blk_integrity_exchg *);
920typedef void (integrity_set_tag_fn) (void *, void *, unsigned int);
921typedef void (integrity_get_tag_fn) (void *, void *, unsigned int);
922
923struct blk_integrity {
924 integrity_gen_fn *generate_fn;
925 integrity_vrfy_fn *verify_fn;
926 integrity_set_tag_fn *set_tag_fn;
927 integrity_get_tag_fn *get_tag_fn;
928
929 unsigned short flags;
930 unsigned short tuple_size;
931 unsigned short sector_size;
932 unsigned short tag_size;
933
934 const char *name;
935
936 struct kobject kobj;
937};
938
939extern int blk_integrity_register(struct gendisk *, struct blk_integrity *);
940extern void blk_integrity_unregister(struct gendisk *);
941extern int blk_integrity_compare(struct block_device *, struct block_device *);
942extern int blk_rq_map_integrity_sg(struct request *, struct scatterlist *);
943extern int blk_rq_count_integrity_sg(struct request *);
944
945static inline unsigned short blk_integrity_tuple_size(struct blk_integrity *bi)
946{
947 if (bi)
948 return bi->tuple_size;
949
950 return 0;
951}
952
953static inline struct blk_integrity *bdev_get_integrity(struct block_device *bdev)
954{
955 return bdev->bd_disk->integrity;
956}
957
958static inline unsigned int bdev_get_tag_size(struct block_device *bdev)
959{
960 struct blk_integrity *bi = bdev_get_integrity(bdev);
961
962 if (bi)
963 return bi->tag_size;
964
965 return 0;
966}
967
968static inline int bdev_integrity_enabled(struct block_device *bdev, int rw)
969{
970 struct blk_integrity *bi = bdev_get_integrity(bdev);
971
972 if (bi == NULL)
973 return 0;
974
975 if (rw == READ && bi->verify_fn != NULL &&
b24498d4 976 (bi->flags & INTEGRITY_FLAG_READ))
7ba1ba12
MP
977 return 1;
978
979 if (rw == WRITE && bi->generate_fn != NULL &&
b24498d4 980 (bi->flags & INTEGRITY_FLAG_WRITE))
7ba1ba12
MP
981 return 1;
982
983 return 0;
984}
985
986static inline int blk_integrity_rq(struct request *rq)
987{
d442cc44
MP
988 if (rq->bio == NULL)
989 return 0;
990
7ba1ba12
MP
991 return bio_integrity(rq->bio);
992}
993
994#else /* CONFIG_BLK_DEV_INTEGRITY */
995
996#define blk_integrity_rq(rq) (0)
997#define blk_rq_count_integrity_sg(a) (0)
998#define blk_rq_map_integrity_sg(a, b) (0)
999#define bdev_get_integrity(a) (0)
1000#define bdev_get_tag_size(a) (0)
1001#define blk_integrity_compare(a, b) (0)
1002#define blk_integrity_register(a, b) (0)
1003#define blk_integrity_unregister(a) do { } while (0);
1004
1005#endif /* CONFIG_BLK_DEV_INTEGRITY */
1006
9361401e
DH
1007#else /* CONFIG_BLOCK */
1008/*
1009 * stubs for when the block layer is configured out
1010 */
1011#define buffer_heads_over_limit 0
1012
9361401e
DH
1013static inline long nr_blockdev_pages(void)
1014{
1015 return 0;
1016}
1017
9361401e
DH
1018#endif /* CONFIG_BLOCK */
1019
1da177e4 1020#endif