4 * Copyright (C) 2001 Jens Axboe <axboe@suse.de>
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 as
8 * published by the Free Software Foundation.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
16 * You should have received a copy of the GNU General Public Licens
17 * along with this program; if not, write to the Free Software
18 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-
23 #include <linux/highmem.h>
24 #include <linux/mempool.h>
25 #include <linux/ioprio.h>
34 #define BIO_BUG_ON BUG_ON
39 #define BIO_MAX_PAGES 256
40 #define BIO_MAX_SIZE (BIO_MAX_PAGES << PAGE_CACHE_SHIFT)
41 #define BIO_MAX_SECTORS (BIO_MAX_SIZE >> 9)
44 * was unsigned short, but we might as well be ready for > 64kB I/O pages
49 unsigned int bv_offset
;
54 struct bio_integrity_payload
;
55 typedef void (bio_end_io_t
) (struct bio
*, int);
56 typedef void (bio_destructor_t
) (struct bio
*);
59 * main unit of I/O for the block layer and lower layers (ie drivers and
63 sector_t bi_sector
; /* device address in 512 byte
65 struct bio
*bi_next
; /* request queue link */
66 struct block_device
*bi_bdev
;
67 unsigned long bi_flags
; /* status, command, etc */
68 unsigned long bi_rw
; /* bottom bits READ/WRITE,
72 unsigned short bi_vcnt
; /* how many bio_vec's */
73 unsigned short bi_idx
; /* current index into bvl_vec */
75 /* Number of segments in this BIO after
76 * physical address coalescing is performed.
78 unsigned int bi_phys_segments
;
80 unsigned int bi_size
; /* residual I/O count */
83 * To keep track of the max segment size, we account for the
84 * sizes of the first and last mergeable segments in this bio.
86 unsigned int bi_seg_front_size
;
87 unsigned int bi_seg_back_size
;
89 unsigned int bi_max_vecs
; /* max bvl_vecs we can hold */
91 unsigned int bi_comp_cpu
; /* completion CPU */
93 atomic_t bi_cnt
; /* pin count */
95 struct bio_vec
*bi_io_vec
; /* the actual vec list */
97 bio_end_io_t
*bi_end_io
;
100 #if defined(CONFIG_BLK_DEV_INTEGRITY)
101 struct bio_integrity_payload
*bi_integrity
; /* data integrity */
104 bio_destructor_t
*bi_destructor
; /* destructor */
110 #define BIO_UPTODATE 0 /* ok after I/O completion */
111 #define BIO_RW_BLOCK 1 /* RW_AHEAD set, and read/write would block */
112 #define BIO_EOF 2 /* out-out-bounds error */
113 #define BIO_SEG_VALID 3 /* bi_phys_segments valid */
114 #define BIO_CLONED 4 /* doesn't own data */
115 #define BIO_BOUNCED 5 /* bio is a bounce bio */
116 #define BIO_USER_MAPPED 6 /* contains user pages */
117 #define BIO_EOPNOTSUPP 7 /* not supported */
118 #define BIO_CPU_AFFINE 8 /* complete bio on same CPU as submitted */
119 #define BIO_NULL_MAPPED 9 /* contains invalid user pages */
120 #define BIO_FS_INTEGRITY 10 /* fs owns integrity data, not block layer */
121 #define BIO_QUIET 11 /* Make BIO Quiet */
122 #define bio_flagged(bio, flag) ((bio)->bi_flags & (1 << (flag)))
125 * top 4 bits of bio flags indicate the pool this bio came from
127 #define BIO_POOL_BITS (4)
128 #define BIO_POOL_OFFSET (BITS_PER_LONG - BIO_POOL_BITS)
129 #define BIO_POOL_MASK (1UL << BIO_POOL_OFFSET)
130 #define BIO_POOL_IDX(bio) ((bio)->bi_flags >> BIO_POOL_OFFSET)
135 * bit 0 -- data direction
136 * If not set, bio is a read from device. If set, it's a write to device.
137 * bit 1 -- rw-ahead when set
139 * Insert a serialization point in the IO queue, forcing previously
140 * submitted IO to be completed before this oen is issued.
141 * bit 3 -- synchronous I/O hint: the block layer will unplug immediately
142 * Note that this does NOT indicate that the IO itself is sync, just
143 * that the block layer will not postpone issue of this IO by plugging.
144 * bit 4 -- metadata request
145 * Used for tracing to differentiate metadata and data IO. May also
146 * get some preferential treatment in the IO scheduler
147 * bit 5 -- discard sectors
148 * Informs the lower level device that this range of sectors is no longer
149 * used by the file system and may thus be freed by the device. Used
150 * for flash based storage.
151 * bit 6 -- fail fast device errors
152 * bit 7 -- fail fast transport errors
153 * bit 8 -- fail fast driver errors
154 * Don't want driver retries for any fast fail whatever the reason.
156 #define BIO_RW 0 /* Must match RW in req flags (blkdev.h) */
157 #define BIO_RW_AHEAD 1 /* Must match FAILFAST in req flags */
158 #define BIO_RW_BARRIER 2
159 #define BIO_RW_SYNC 3
160 #define BIO_RW_META 4
161 #define BIO_RW_DISCARD 5
162 #define BIO_RW_FAILFAST_DEV 6
163 #define BIO_RW_FAILFAST_TRANSPORT 7
164 #define BIO_RW_FAILFAST_DRIVER 8
167 * upper 16 bits of bi_rw define the io priority of this bio
169 #define BIO_PRIO_SHIFT (8 * sizeof(unsigned long) - IOPRIO_BITS)
170 #define bio_prio(bio) ((bio)->bi_rw >> BIO_PRIO_SHIFT)
171 #define bio_prio_valid(bio) ioprio_valid(bio_prio(bio))
173 #define bio_set_prio(bio, prio) do { \
174 WARN_ON(prio >= (1 << IOPRIO_BITS)); \
175 (bio)->bi_rw &= ((1UL << BIO_PRIO_SHIFT) - 1); \
176 (bio)->bi_rw |= ((unsigned long) (prio) << BIO_PRIO_SHIFT); \
180 * various member access, note that bio_data should of course not be used
181 * on highmem page vectors
183 #define bio_iovec_idx(bio, idx) (&((bio)->bi_io_vec[(idx)]))
184 #define bio_iovec(bio) bio_iovec_idx((bio), (bio)->bi_idx)
185 #define bio_page(bio) bio_iovec((bio))->bv_page
186 #define bio_offset(bio) bio_iovec((bio))->bv_offset
187 #define bio_segments(bio) ((bio)->bi_vcnt - (bio)->bi_idx)
188 #define bio_sectors(bio) ((bio)->bi_size >> 9)
189 #define bio_barrier(bio) ((bio)->bi_rw & (1 << BIO_RW_BARRIER))
190 #define bio_sync(bio) ((bio)->bi_rw & (1 << BIO_RW_SYNC))
191 #define bio_failfast_dev(bio) ((bio)->bi_rw & (1 << BIO_RW_FAILFAST_DEV))
192 #define bio_failfast_transport(bio) \
193 ((bio)->bi_rw & (1 << BIO_RW_FAILFAST_TRANSPORT))
194 #define bio_failfast_driver(bio) ((bio)->bi_rw & (1 << BIO_RW_FAILFAST_DRIVER))
195 #define bio_rw_ahead(bio) ((bio)->bi_rw & (1 << BIO_RW_AHEAD))
196 #define bio_rw_meta(bio) ((bio)->bi_rw & (1 << BIO_RW_META))
197 #define bio_discard(bio) ((bio)->bi_rw & (1 << BIO_RW_DISCARD))
198 #define bio_empty_barrier(bio) (bio_barrier(bio) && !bio_has_data(bio) && !bio_discard(bio))
200 static inline unsigned int bio_cur_sectors(struct bio
*bio
)
203 return bio_iovec(bio
)->bv_len
>> 9;
204 else /* dataless requests such as discard */
205 return bio
->bi_size
>> 9;
208 static inline void *bio_data(struct bio
*bio
)
211 return page_address(bio_page(bio
)) + bio_offset(bio
);
219 #define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
220 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
223 * queues that have highmem support enabled may still need to revert to
224 * PIO transfers occasionally and thus map high pages temporarily. For
225 * permanent PIO fall back, user is probably better off disabling highmem
226 * I/O completely on that queue (see ide-dma for example)
228 #define __bio_kmap_atomic(bio, idx, kmtype) \
229 (kmap_atomic(bio_iovec_idx((bio), (idx))->bv_page, kmtype) + \
230 bio_iovec_idx((bio), (idx))->bv_offset)
232 #define __bio_kunmap_atomic(addr, kmtype) kunmap_atomic(addr, kmtype)
238 #define __BVEC_END(bio) bio_iovec_idx((bio), (bio)->bi_vcnt - 1)
239 #define __BVEC_START(bio) bio_iovec_idx((bio), (bio)->bi_idx)
241 /* Default implementation of BIOVEC_PHYS_MERGEABLE */
242 #define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
243 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
246 * allow arch override, for eg virtualized architectures (put in asm/io.h)
248 #ifndef BIOVEC_PHYS_MERGEABLE
249 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
250 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
253 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
254 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
255 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
256 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, (q)->seg_boundary_mask)
257 #define BIO_SEG_BOUNDARY(q, b1, b2) \
258 BIOVEC_SEG_BOUNDARY((q), __BVEC_END((b1)), __BVEC_START((b2)))
260 #define bio_io_error(bio) bio_endio((bio), -EIO)
263 * drivers should not use the __ version unless they _really_ want to
264 * run through the entire bio and not just pending pieces
266 #define __bio_for_each_segment(bvl, bio, i, start_idx) \
267 for (bvl = bio_iovec_idx((bio), (start_idx)), i = (start_idx); \
268 i < (bio)->bi_vcnt; \
271 #define bio_for_each_segment(bvl, bio, i) \
272 __bio_for_each_segment(bvl, bio, i, (bio)->bi_idx)
275 * get a reference to a bio, so it won't disappear. the intended use is
279 * submit_bio(rw, bio);
280 * if (bio->bi_flags ...)
284 * without the bio_get(), it could potentially complete I/O before submit_bio
285 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
288 #define bio_get(bio) atomic_inc(&(bio)->bi_cnt)
290 #if defined(CONFIG_BLK_DEV_INTEGRITY)
292 * bio integrity payload
294 struct bio_integrity_payload
{
295 struct bio
*bip_bio
; /* parent bio */
296 struct bio_vec
*bip_vec
; /* integrity data vector */
298 sector_t bip_sector
; /* virtual start sector */
300 void *bip_buf
; /* generated integrity data */
301 bio_end_io_t
*bip_end_io
; /* saved I/O completion fn */
303 int bip_error
; /* saved I/O error */
304 unsigned int bip_size
;
306 unsigned short bip_pool
; /* pool the ivec came from */
307 unsigned short bip_vcnt
; /* # of integrity bio_vecs */
308 unsigned short bip_idx
; /* current bip_vec index */
310 struct work_struct bip_work
; /* I/O completion */
312 #endif /* CONFIG_BLK_DEV_INTEGRITY */
315 * A bio_pair is used when we need to split a bio.
316 * This can only happen for a bio that refers to just one
317 * page of data, and in the unusual situation when the
318 * page crosses a chunk/device boundary
320 * The address of the master bio is stored in bio1.bi_private
321 * The address of the pool the pair was allocated from is stored
325 struct bio bio1
, bio2
;
326 struct bio_vec bv1
, bv2
;
327 #if defined(CONFIG_BLK_DEV_INTEGRITY)
328 struct bio_integrity_payload bip1
, bip2
;
329 struct bio_vec iv1
, iv2
;
334 extern struct bio_pair
*bio_split(struct bio
*bi
, int first_sectors
);
335 extern void bio_pair_release(struct bio_pair
*dbio
);
337 extern struct bio_set
*bioset_create(int, int);
338 extern void bioset_free(struct bio_set
*);
340 extern struct bio
*bio_alloc(gfp_t
, int);
341 extern struct bio
*bio_kmalloc(gfp_t
, int);
342 extern struct bio
*bio_alloc_bioset(gfp_t
, int, struct bio_set
*);
343 extern void bio_put(struct bio
*);
344 extern void bio_free(struct bio
*, struct bio_set
*);
346 extern void bio_endio(struct bio
*, int);
347 struct request_queue
;
348 extern int bio_phys_segments(struct request_queue
*, struct bio
*);
350 extern void __bio_clone(struct bio
*, struct bio
*);
351 extern struct bio
*bio_clone(struct bio
*, gfp_t
);
353 extern void bio_init(struct bio
*);
355 extern int bio_add_page(struct bio
*, struct page
*, unsigned int,unsigned int);
356 extern int bio_add_pc_page(struct request_queue
*, struct bio
*, struct page
*,
357 unsigned int, unsigned int);
358 extern int bio_get_nr_vecs(struct block_device
*);
359 extern sector_t
bio_sector_offset(struct bio
*, unsigned short, unsigned int);
360 extern struct bio
*bio_map_user(struct request_queue
*, struct block_device
*,
361 unsigned long, unsigned int, int, gfp_t
);
364 extern struct bio
*bio_map_user_iov(struct request_queue
*,
365 struct block_device
*,
366 struct sg_iovec
*, int, int, gfp_t
);
367 extern void bio_unmap_user(struct bio
*);
368 extern struct bio
*bio_map_kern(struct request_queue
*, void *, unsigned int,
370 extern struct bio
*bio_copy_kern(struct request_queue
*, void *, unsigned int,
372 extern void bio_set_pages_dirty(struct bio
*bio
);
373 extern void bio_check_pages_dirty(struct bio
*bio
);
374 extern struct bio
*bio_copy_user(struct request_queue
*, struct rq_map_data
*,
375 unsigned long, unsigned int, int, gfp_t
);
376 extern struct bio
*bio_copy_user_iov(struct request_queue
*,
377 struct rq_map_data
*, struct sg_iovec
*,
379 extern int bio_uncopy_user(struct bio
*);
380 void zero_fill_bio(struct bio
*bio
);
381 extern struct bio_vec
*bvec_alloc_bs(gfp_t
, int, unsigned long *, struct bio_set
*);
382 extern unsigned int bvec_nr_vecs(unsigned short idx
);
385 * Allow queuer to specify a completion CPU for this bio
387 static inline void bio_set_completion_cpu(struct bio
*bio
, unsigned int cpu
)
389 bio
->bi_comp_cpu
= cpu
;
393 * bio_set is used to allow other portions of the IO system to
394 * allocate their own private memory pools for bio and iovec structures.
395 * These memory pools in turn all allocate from the bio_slab
396 * and the bvec_slabs[].
398 #define BIO_POOL_SIZE 2
399 #define BIOVEC_NR_POOLS 6
400 #define BIOVEC_MAX_IDX (BIOVEC_NR_POOLS - 1)
404 #if defined(CONFIG_BLK_DEV_INTEGRITY)
405 mempool_t
*bio_integrity_pool
;
407 mempool_t
*bvec_pool
;
413 struct kmem_cache
*slab
;
416 extern struct bio_set
*fs_bio_set
;
419 * a small number of entries is fine, not going to be performance critical.
420 * basically we just need to survive
422 #define BIO_SPLIT_ENTRIES 2
424 #ifdef CONFIG_HIGHMEM
426 * remember to add offset! and never ever reenable interrupts between a
427 * bvec_kmap_irq and bvec_kunmap_irq!!
429 * This function MUST be inlined - it plays with the CPU interrupt flags.
431 static inline char *bvec_kmap_irq(struct bio_vec
*bvec
, unsigned long *flags
)
436 * might not be a highmem page, but the preempt/irq count
437 * balancing is a lot nicer this way
439 local_irq_save(*flags
);
440 addr
= (unsigned long) kmap_atomic(bvec
->bv_page
, KM_BIO_SRC_IRQ
);
442 BUG_ON(addr
& ~PAGE_MASK
);
444 return (char *) addr
+ bvec
->bv_offset
;
447 static inline void bvec_kunmap_irq(char *buffer
, unsigned long *flags
)
449 unsigned long ptr
= (unsigned long) buffer
& PAGE_MASK
;
451 kunmap_atomic((void *) ptr
, KM_BIO_SRC_IRQ
);
452 local_irq_restore(*flags
);
456 #define bvec_kmap_irq(bvec, flags) (page_address((bvec)->bv_page) + (bvec)->bv_offset)
457 #define bvec_kunmap_irq(buf, flags) do { *(flags) = 0; } while (0)
460 static inline char *__bio_kmap_irq(struct bio
*bio
, unsigned short idx
,
461 unsigned long *flags
)
463 return bvec_kmap_irq(bio_iovec_idx(bio
, idx
), flags
);
465 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
467 #define bio_kmap_irq(bio, flags) \
468 __bio_kmap_irq((bio), (bio)->bi_idx, (flags))
469 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
472 * Check whether this bio carries any data or not. A NULL bio is allowed.
474 static inline int bio_has_data(struct bio
*bio
)
476 return bio
&& bio
->bi_io_vec
!= NULL
;
479 #if defined(CONFIG_BLK_DEV_INTEGRITY)
481 #define bip_vec_idx(bip, idx) (&(bip->bip_vec[(idx)]))
482 #define bip_vec(bip) bip_vec_idx(bip, 0)
484 #define __bip_for_each_vec(bvl, bip, i, start_idx) \
485 for (bvl = bip_vec_idx((bip), (start_idx)), i = (start_idx); \
486 i < (bip)->bip_vcnt; \
489 #define bip_for_each_vec(bvl, bip, i) \
490 __bip_for_each_vec(bvl, bip, i, (bip)->bip_idx)
492 #define bio_integrity(bio) (bio->bi_integrity != NULL)
494 extern struct bio_integrity_payload
*bio_integrity_alloc_bioset(struct bio
*, gfp_t
, unsigned int, struct bio_set
*);
495 extern struct bio_integrity_payload
*bio_integrity_alloc(struct bio
*, gfp_t
, unsigned int);
496 extern void bio_integrity_free(struct bio
*, struct bio_set
*);
497 extern int bio_integrity_add_page(struct bio
*, struct page
*, unsigned int, unsigned int);
498 extern int bio_integrity_enabled(struct bio
*bio
);
499 extern int bio_integrity_set_tag(struct bio
*, void *, unsigned int);
500 extern int bio_integrity_get_tag(struct bio
*, void *, unsigned int);
501 extern int bio_integrity_prep(struct bio
*);
502 extern void bio_integrity_endio(struct bio
*, int);
503 extern void bio_integrity_advance(struct bio
*, unsigned int);
504 extern void bio_integrity_trim(struct bio
*, unsigned int, unsigned int);
505 extern void bio_integrity_split(struct bio
*, struct bio_pair
*, int);
506 extern int bio_integrity_clone(struct bio
*, struct bio
*, struct bio_set
*);
507 extern int bioset_integrity_create(struct bio_set
*, int);
508 extern void bioset_integrity_free(struct bio_set
*);
509 extern void bio_integrity_init_slab(void);
511 #else /* CONFIG_BLK_DEV_INTEGRITY */
513 #define bio_integrity(a) (0)
514 #define bioset_integrity_create(a, b) (0)
515 #define bio_integrity_prep(a) (0)
516 #define bio_integrity_enabled(a) (0)
517 #define bio_integrity_clone(a, b, c) (0)
518 #define bioset_integrity_free(a) do { } while (0)
519 #define bio_integrity_free(a, b) do { } while (0)
520 #define bio_integrity_endio(a, b) do { } while (0)
521 #define bio_integrity_advance(a, b) do { } while (0)
522 #define bio_integrity_trim(a, b, c) do { } while (0)
523 #define bio_integrity_split(a, b, c) do { } while (0)
524 #define bio_integrity_set_tag(a, b, c) do { } while (0)
525 #define bio_integrity_get_tag(a, b, c) do { } while (0)
526 #define bio_integrity_init_slab(a) do { } while (0)
528 #endif /* CONFIG_BLK_DEV_INTEGRITY */
530 #endif /* CONFIG_BLOCK */
531 #endif /* __LINUX_BIO_H */