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 short bi_phys_segments;
80 /* Number of segments after physical and DMA remapping
81 * hardware coalescing is performed.
83 unsigned short bi_hw_segments;
85 unsigned int bi_size; /* residual I/O count */
88 * To keep track of the max hw size, we account for the
89 * sizes of the first and last virtually mergeable segments
92 unsigned int bi_hw_front_size;
93 unsigned int bi_hw_back_size;
95 unsigned int bi_max_vecs; /* max bvl_vecs we can hold */
97 struct bio_vec *bi_io_vec; /* the actual vec list */
99 bio_end_io_t *bi_end_io;
100 atomic_t bi_cnt; /* pin count */
103 #if defined(CONFIG_BLK_DEV_INTEGRITY)
104 struct bio_integrity_payload *bi_integrity; /* data integrity */
107 bio_destructor_t *bi_destructor; /* destructor */
113 #define BIO_UPTODATE 0 /* ok after I/O completion */
114 #define BIO_RW_BLOCK 1 /* RW_AHEAD set, and read/write would block */
115 #define BIO_EOF 2 /* out-out-bounds error */
116 #define BIO_SEG_VALID 3 /* nr_hw_seg valid */
117 #define BIO_CLONED 4 /* doesn't own data */
118 #define BIO_BOUNCED 5 /* bio is a bounce bio */
119 #define BIO_USER_MAPPED 6 /* contains user pages */
120 #define BIO_EOPNOTSUPP 7 /* not supported */
121 #define bio_flagged(bio, flag) ((bio)->bi_flags & (1 << (flag)))
124 * top 4 bits of bio flags indicate the pool this bio came from
126 #define BIO_POOL_BITS (4)
127 #define BIO_POOL_OFFSET (BITS_PER_LONG - BIO_POOL_BITS)
128 #define BIO_POOL_MASK (1UL << BIO_POOL_OFFSET)
129 #define BIO_POOL_IDX(bio) ((bio)->bi_flags >> BIO_POOL_OFFSET)
134 * bit 0 -- read (not set) or write (set)
135 * bit 1 -- rw-ahead when set
137 * bit 3 -- fail fast, don't want low level driver retries
138 * bit 4 -- synchronous I/O hint: the block layer will unplug immediately
139 * bit 5 -- metadata request
140 * bit 6 -- discard sectors
142 #define BIO_RW 0 /* Must match RW in req flags (blkdev.h) */
143 #define BIO_RW_AHEAD 1 /* Must match FAILFAST in req flags */
144 #define BIO_RW_BARRIER 2
145 #define BIO_RW_FAILFAST 3
146 #define BIO_RW_SYNC 4
147 #define BIO_RW_META 5
148 #define BIO_RW_DISCARD 6
151 * upper 16 bits of bi_rw define the io priority of this bio
153 #define BIO_PRIO_SHIFT (8 * sizeof(unsigned long) - IOPRIO_BITS)
154 #define bio_prio(bio) ((bio)->bi_rw >> BIO_PRIO_SHIFT)
155 #define bio_prio_valid(bio) ioprio_valid(bio_prio(bio))
157 #define bio_set_prio(bio, prio) do { \
158 WARN_ON(prio >= (1 << IOPRIO_BITS)); \
159 (bio)->bi_rw &= ((1UL << BIO_PRIO_SHIFT) - 1); \
160 (bio)->bi_rw |= ((unsigned long) (prio) << BIO_PRIO_SHIFT); \
164 * various member access, note that bio_data should of course not be used
165 * on highmem page vectors
167 #define bio_iovec_idx(bio, idx) (&((bio)->bi_io_vec[(idx)]))
168 #define bio_iovec(bio) bio_iovec_idx((bio), (bio)->bi_idx)
169 #define bio_page(bio) bio_iovec((bio))->bv_page
170 #define bio_offset(bio) bio_iovec((bio))->bv_offset
171 #define bio_segments(bio) ((bio)->bi_vcnt - (bio)->bi_idx)
172 #define bio_sectors(bio) ((bio)->bi_size >> 9)
173 #define bio_barrier(bio) ((bio)->bi_rw & (1 << BIO_RW_BARRIER))
174 #define bio_sync(bio) ((bio)->bi_rw & (1 << BIO_RW_SYNC))
175 #define bio_failfast(bio) ((bio)->bi_rw & (1 << BIO_RW_FAILFAST))
176 #define bio_rw_ahead(bio) ((bio)->bi_rw & (1 << BIO_RW_AHEAD))
177 #define bio_rw_meta(bio) ((bio)->bi_rw & (1 << BIO_RW_META))
178 #define bio_discard(bio) ((bio)->bi_rw & (1 << BIO_RW_DISCARD))
179 #define bio_empty_barrier(bio) (bio_barrier(bio) && !bio_has_data(bio) && !bio_discard(bio))
181 static inline unsigned int bio_cur_sectors(struct bio *bio)
184 return bio_iovec(bio)->bv_len >> 9;
185 else /* dataless requests such as discard */
186 return bio->bi_size >> 9;
189 static inline void *bio_data(struct bio *bio)
192 return page_address(bio_page(bio)) + bio_offset(bio);
200 #define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
201 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
204 * queues that have highmem support enabled may still need to revert to
205 * PIO transfers occasionally and thus map high pages temporarily. For
206 * permanent PIO fall back, user is probably better off disabling highmem
207 * I/O completely on that queue (see ide-dma for example)
209 #define __bio_kmap_atomic(bio, idx, kmtype) \
210 (kmap_atomic(bio_iovec_idx((bio), (idx))->bv_page, kmtype) + \
211 bio_iovec_idx((bio), (idx))->bv_offset)
213 #define __bio_kunmap_atomic(addr, kmtype) kunmap_atomic(addr, kmtype)
219 #define __BVEC_END(bio) bio_iovec_idx((bio), (bio)->bi_vcnt - 1)
220 #define __BVEC_START(bio) bio_iovec_idx((bio), (bio)->bi_idx)
223 * allow arch override, for eg virtualized architectures (put in asm/io.h)
225 #ifndef BIOVEC_PHYS_MERGEABLE
226 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
227 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
230 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
231 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
232 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
233 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, (q)->seg_boundary_mask)
234 #define BIO_SEG_BOUNDARY(q, b1, b2) \
235 BIOVEC_SEG_BOUNDARY((q), __BVEC_END((b1)), __BVEC_START((b2)))
237 #define bio_io_error(bio) bio_endio((bio), -EIO)
240 * drivers should not use the __ version unless they _really_ want to
241 * run through the entire bio and not just pending pieces
243 #define __bio_for_each_segment(bvl, bio, i, start_idx) \
244 for (bvl = bio_iovec_idx((bio), (start_idx)), i = (start_idx); \
245 i < (bio)->bi_vcnt; \
248 #define bio_for_each_segment(bvl, bio, i) \
249 __bio_for_each_segment(bvl, bio, i, (bio)->bi_idx)
252 * get a reference to a bio, so it won't disappear. the intended use is
256 * submit_bio(rw, bio);
257 * if (bio->bi_flags ...)
261 * without the bio_get(), it could potentially complete I/O before submit_bio
262 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
265 #define bio_get(bio) atomic_inc(&(bio)->bi_cnt)
267 #if defined(CONFIG_BLK_DEV_INTEGRITY)
269 * bio integrity payload
271 struct bio_integrity_payload {
272 struct bio *bip_bio; /* parent bio */
273 struct bio_vec *bip_vec; /* integrity data vector */
275 sector_t bip_sector; /* virtual start sector */
277 void *bip_buf; /* generated integrity data */
278 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
280 int bip_error; /* saved I/O error */
281 unsigned int bip_size;
283 unsigned short bip_pool; /* pool the ivec came from */
284 unsigned short bip_vcnt; /* # of integrity bio_vecs */
285 unsigned short bip_idx; /* current bip_vec index */
287 struct work_struct bip_work; /* I/O completion */
289 #endif /* CONFIG_BLK_DEV_INTEGRITY */
292 * A bio_pair is used when we need to split a bio.
293 * This can only happen for a bio that refers to just one
294 * page of data, and in the unusual situation when the
295 * page crosses a chunk/device boundary
297 * The address of the master bio is stored in bio1.bi_private
298 * The address of the pool the pair was allocated from is stored
302 struct bio bio1, bio2;
303 struct bio_vec bv1, bv2;
304 #if defined(CONFIG_BLK_DEV_INTEGRITY)
305 struct bio_integrity_payload bip1, bip2;
306 struct bio_vec iv1, iv2;
311 extern struct bio_pair *bio_split(struct bio *bi, mempool_t *pool,
313 extern mempool_t *bio_split_pool;
314 extern void bio_pair_release(struct bio_pair *dbio);
316 extern struct bio_set *bioset_create(int, int);
317 extern void bioset_free(struct bio_set *);
319 extern struct bio *bio_alloc(gfp_t, int);
320 extern struct bio *bio_alloc_bioset(gfp_t, int, struct bio_set *);
321 extern void bio_put(struct bio *);
322 extern void bio_free(struct bio *, struct bio_set *);
324 extern void bio_endio(struct bio *, int);
325 struct request_queue;
326 extern int bio_phys_segments(struct request_queue *, struct bio *);
327 extern int bio_hw_segments(struct request_queue *, struct bio *);
329 extern void __bio_clone(struct bio *, struct bio *);
330 extern struct bio *bio_clone(struct bio *, gfp_t);
332 extern void bio_init(struct bio *);
334 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
335 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
336 unsigned int, unsigned int);
337 extern int bio_get_nr_vecs(struct block_device *);
338 extern struct bio *bio_map_user(struct request_queue *, struct block_device *,
339 unsigned long, unsigned int, int);
341 extern struct bio *bio_map_user_iov(struct request_queue *,
342 struct block_device *,
343 struct sg_iovec *, int, int);
344 extern void bio_unmap_user(struct bio *);
345 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
347 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
349 extern void bio_set_pages_dirty(struct bio *bio);
350 extern void bio_check_pages_dirty(struct bio *bio);
351 extern struct bio *bio_copy_user(struct request_queue *, unsigned long, unsigned int, int);
352 extern struct bio *bio_copy_user_iov(struct request_queue *, struct sg_iovec *,
354 extern int bio_uncopy_user(struct bio *);
355 void zero_fill_bio(struct bio *bio);
356 extern struct bio_vec *bvec_alloc_bs(gfp_t, int, unsigned long *, struct bio_set *);
357 extern unsigned int bvec_nr_vecs(unsigned short idx);
360 * bio_set is used to allow other portions of the IO system to
361 * allocate their own private memory pools for bio and iovec structures.
362 * These memory pools in turn all allocate from the bio_slab
363 * and the bvec_slabs[].
365 #define BIO_POOL_SIZE 2
366 #define BIOVEC_NR_POOLS 6
370 #if defined(CONFIG_BLK_DEV_INTEGRITY)
371 mempool_t *bio_integrity_pool;
373 mempool_t *bvec_pools[BIOVEC_NR_POOLS];
379 struct kmem_cache *slab;
382 extern struct bio_set *fs_bio_set;
385 * a small number of entries is fine, not going to be performance critical.
386 * basically we just need to survive
388 #define BIO_SPLIT_ENTRIES 2
390 #ifdef CONFIG_HIGHMEM
392 * remember to add offset! and never ever reenable interrupts between a
393 * bvec_kmap_irq and bvec_kunmap_irq!!
395 * This function MUST be inlined - it plays with the CPU interrupt flags.
397 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
402 * might not be a highmem page, but the preempt/irq count
403 * balancing is a lot nicer this way
405 local_irq_save(*flags);
406 addr = (unsigned long) kmap_atomic(bvec->bv_page, KM_BIO_SRC_IRQ);
408 BUG_ON(addr & ~PAGE_MASK);
410 return (char *) addr + bvec->bv_offset;
413 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
415 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
417 kunmap_atomic((void *) ptr, KM_BIO_SRC_IRQ);
418 local_irq_restore(*flags);
422 #define bvec_kmap_irq(bvec, flags) (page_address((bvec)->bv_page) + (bvec)->bv_offset)
423 #define bvec_kunmap_irq(buf, flags) do { *(flags) = 0; } while (0)
426 static inline char *__bio_kmap_irq(struct bio *bio, unsigned short idx,
427 unsigned long *flags)
429 return bvec_kmap_irq(bio_iovec_idx(bio, idx), flags);
431 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
433 #define bio_kmap_irq(bio, flags) \
434 __bio_kmap_irq((bio), (bio)->bi_idx, (flags))
435 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
438 * Check whether this bio carries any data or not. A NULL bio is allowed.
440 static inline int bio_has_data(struct bio *bio)
442 return bio && bio->bi_io_vec != NULL;
445 #if defined(CONFIG_BLK_DEV_INTEGRITY)
447 #define bip_vec_idx(bip, idx) (&(bip->bip_vec[(idx)]))
448 #define bip_vec(bip) bip_vec_idx(bip, 0)
450 #define __bip_for_each_vec(bvl, bip, i, start_idx) \
451 for (bvl = bip_vec_idx((bip), (start_idx)), i = (start_idx); \
452 i < (bip)->bip_vcnt; \
455 #define bip_for_each_vec(bvl, bip, i) \
456 __bip_for_each_vec(bvl, bip, i, (bip)->bip_idx)
458 static inline int bio_integrity(struct bio *bio)
460 #if defined(CONFIG_BLK_DEV_INTEGRITY)
461 return bio->bi_integrity != NULL;
467 extern struct bio_integrity_payload *bio_integrity_alloc_bioset(struct bio *, gfp_t, unsigned int, struct bio_set *);
468 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
469 extern void bio_integrity_free(struct bio *, struct bio_set *);
470 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
471 extern int bio_integrity_enabled(struct bio *bio);
472 extern int bio_integrity_set_tag(struct bio *, void *, unsigned int);
473 extern int bio_integrity_get_tag(struct bio *, void *, unsigned int);
474 extern int bio_integrity_prep(struct bio *);
475 extern void bio_integrity_endio(struct bio *, int);
476 extern void bio_integrity_advance(struct bio *, unsigned int);
477 extern void bio_integrity_trim(struct bio *, unsigned int, unsigned int);
478 extern void bio_integrity_split(struct bio *, struct bio_pair *, int);
479 extern int bio_integrity_clone(struct bio *, struct bio *, struct bio_set *);
480 extern int bioset_integrity_create(struct bio_set *, int);
481 extern void bioset_integrity_free(struct bio_set *);
482 extern void bio_integrity_init_slab(void);
484 #else /* CONFIG_BLK_DEV_INTEGRITY */
486 #define bio_integrity(a) (0)
487 #define bioset_integrity_create(a, b) (0)
488 #define bio_integrity_prep(a) (0)
489 #define bio_integrity_enabled(a) (0)
490 #define bio_integrity_clone(a, b, c) (0)
491 #define bioset_integrity_free(a) do { } while (0)
492 #define bio_integrity_free(a, b) do { } while (0)
493 #define bio_integrity_endio(a, b) do { } while (0)
494 #define bio_integrity_advance(a, b) do { } while (0)
495 #define bio_integrity_trim(a, b, c) do { } while (0)
496 #define bio_integrity_split(a, b, c) do { } while (0)
497 #define bio_integrity_set_tag(a, b, c) do { } while (0)
498 #define bio_integrity_get_tag(a, b, c) do { } while (0)
499 #define bio_integrity_init_slab(a) do { } while (0)
501 #endif /* CONFIG_BLK_DEV_INTEGRITY */
503 #endif /* CONFIG_BLOCK */
504 #endif /* __LINUX_BIO_H */