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 */
82 unsigned int bi_max_vecs; /* max bvl_vecs we can hold */
84 unsigned int bi_comp_cpu; /* completion CPU */
86 struct bio_vec *bi_io_vec; /* the actual vec list */
88 bio_end_io_t *bi_end_io;
89 atomic_t bi_cnt; /* pin count */
92 #if defined(CONFIG_BLK_DEV_INTEGRITY)
93 struct bio_integrity_payload *bi_integrity; /* data integrity */
96 bio_destructor_t *bi_destructor; /* destructor */
102 #define BIO_UPTODATE 0 /* ok after I/O completion */
103 #define BIO_RW_BLOCK 1 /* RW_AHEAD set, and read/write would block */
104 #define BIO_EOF 2 /* out-out-bounds error */
105 #define BIO_SEG_VALID 3 /* bi_phys_segments valid */
106 #define BIO_CLONED 4 /* doesn't own data */
107 #define BIO_BOUNCED 5 /* bio is a bounce bio */
108 #define BIO_USER_MAPPED 6 /* contains user pages */
109 #define BIO_EOPNOTSUPP 7 /* not supported */
110 #define BIO_CPU_AFFINE 8 /* complete bio on same CPU as submitted */
111 #define BIO_NULL_MAPPED 9 /* contains invalid user pages */
112 #define BIO_FS_INTEGRITY 10 /* fs owns integrity data, not block layer */
113 #define bio_flagged(bio, flag) ((bio)->bi_flags & (1 << (flag)))
116 * top 4 bits of bio flags indicate the pool this bio came from
118 #define BIO_POOL_BITS (4)
119 #define BIO_POOL_OFFSET (BITS_PER_LONG - BIO_POOL_BITS)
120 #define BIO_POOL_MASK (1UL << BIO_POOL_OFFSET)
121 #define BIO_POOL_IDX(bio) ((bio)->bi_flags >> BIO_POOL_OFFSET)
126 * bit 0 -- read (not set) or write (set)
127 * bit 1 -- rw-ahead when set
129 * bit 3 -- fail fast, don't want low level driver retries
130 * bit 4 -- synchronous I/O hint: the block layer will unplug immediately
131 * bit 5 -- metadata request
132 * bit 6 -- discard sectors
134 #define BIO_RW 0 /* Must match RW in req flags (blkdev.h) */
135 #define BIO_RW_AHEAD 1 /* Must match FAILFAST in req flags */
136 #define BIO_RW_BARRIER 2
137 #define BIO_RW_FAILFAST 3
138 #define BIO_RW_SYNC 4
139 #define BIO_RW_META 5
140 #define BIO_RW_DISCARD 6
143 * upper 16 bits of bi_rw define the io priority of this bio
145 #define BIO_PRIO_SHIFT (8 * sizeof(unsigned long) - IOPRIO_BITS)
146 #define bio_prio(bio) ((bio)->bi_rw >> BIO_PRIO_SHIFT)
147 #define bio_prio_valid(bio) ioprio_valid(bio_prio(bio))
149 #define bio_set_prio(bio, prio) do { \
150 WARN_ON(prio >= (1 << IOPRIO_BITS)); \
151 (bio)->bi_rw &= ((1UL << BIO_PRIO_SHIFT) - 1); \
152 (bio)->bi_rw |= ((unsigned long) (prio) << BIO_PRIO_SHIFT); \
156 * various member access, note that bio_data should of course not be used
157 * on highmem page vectors
159 #define bio_iovec_idx(bio, idx) (&((bio)->bi_io_vec[(idx)]))
160 #define bio_iovec(bio) bio_iovec_idx((bio), (bio)->bi_idx)
161 #define bio_page(bio) bio_iovec((bio))->bv_page
162 #define bio_offset(bio) bio_iovec((bio))->bv_offset
163 #define bio_segments(bio) ((bio)->bi_vcnt - (bio)->bi_idx)
164 #define bio_sectors(bio) ((bio)->bi_size >> 9)
165 #define bio_barrier(bio) ((bio)->bi_rw & (1 << BIO_RW_BARRIER))
166 #define bio_sync(bio) ((bio)->bi_rw & (1 << BIO_RW_SYNC))
167 #define bio_failfast(bio) ((bio)->bi_rw & (1 << BIO_RW_FAILFAST))
168 #define bio_rw_ahead(bio) ((bio)->bi_rw & (1 << BIO_RW_AHEAD))
169 #define bio_rw_meta(bio) ((bio)->bi_rw & (1 << BIO_RW_META))
170 #define bio_discard(bio) ((bio)->bi_rw & (1 << BIO_RW_DISCARD))
171 #define bio_empty_barrier(bio) (bio_barrier(bio) && !bio_has_data(bio) && !bio_discard(bio))
173 static inline unsigned int bio_cur_sectors(struct bio *bio)
176 return bio_iovec(bio)->bv_len >> 9;
177 else /* dataless requests such as discard */
178 return bio->bi_size >> 9;
181 static inline void *bio_data(struct bio *bio)
184 return page_address(bio_page(bio)) + bio_offset(bio);
192 #define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
193 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
196 * queues that have highmem support enabled may still need to revert to
197 * PIO transfers occasionally and thus map high pages temporarily. For
198 * permanent PIO fall back, user is probably better off disabling highmem
199 * I/O completely on that queue (see ide-dma for example)
201 #define __bio_kmap_atomic(bio, idx, kmtype) \
202 (kmap_atomic(bio_iovec_idx((bio), (idx))->bv_page, kmtype) + \
203 bio_iovec_idx((bio), (idx))->bv_offset)
205 #define __bio_kunmap_atomic(addr, kmtype) kunmap_atomic(addr, kmtype)
211 #define __BVEC_END(bio) bio_iovec_idx((bio), (bio)->bi_vcnt - 1)
212 #define __BVEC_START(bio) bio_iovec_idx((bio), (bio)->bi_idx)
215 * allow arch override, for eg virtualized architectures (put in asm/io.h)
217 #ifndef BIOVEC_PHYS_MERGEABLE
218 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
219 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
222 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
223 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
224 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
225 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, (q)->seg_boundary_mask)
226 #define BIO_SEG_BOUNDARY(q, b1, b2) \
227 BIOVEC_SEG_BOUNDARY((q), __BVEC_END((b1)), __BVEC_START((b2)))
229 #define bio_io_error(bio) bio_endio((bio), -EIO)
232 * drivers should not use the __ version unless they _really_ want to
233 * run through the entire bio and not just pending pieces
235 #define __bio_for_each_segment(bvl, bio, i, start_idx) \
236 for (bvl = bio_iovec_idx((bio), (start_idx)), i = (start_idx); \
237 i < (bio)->bi_vcnt; \
240 #define bio_for_each_segment(bvl, bio, i) \
241 __bio_for_each_segment(bvl, bio, i, (bio)->bi_idx)
244 * get a reference to a bio, so it won't disappear. the intended use is
248 * submit_bio(rw, bio);
249 * if (bio->bi_flags ...)
253 * without the bio_get(), it could potentially complete I/O before submit_bio
254 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
257 #define bio_get(bio) atomic_inc(&(bio)->bi_cnt)
259 #if defined(CONFIG_BLK_DEV_INTEGRITY)
261 * bio integrity payload
263 struct bio_integrity_payload {
264 struct bio *bip_bio; /* parent bio */
265 struct bio_vec *bip_vec; /* integrity data vector */
267 sector_t bip_sector; /* virtual start sector */
269 void *bip_buf; /* generated integrity data */
270 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
272 int bip_error; /* saved I/O error */
273 unsigned int bip_size;
275 unsigned short bip_pool; /* pool the ivec came from */
276 unsigned short bip_vcnt; /* # of integrity bio_vecs */
277 unsigned short bip_idx; /* current bip_vec index */
279 struct work_struct bip_work; /* I/O completion */
281 #endif /* CONFIG_BLK_DEV_INTEGRITY */
284 * A bio_pair is used when we need to split a bio.
285 * This can only happen for a bio that refers to just one
286 * page of data, and in the unusual situation when the
287 * page crosses a chunk/device boundary
289 * The address of the master bio is stored in bio1.bi_private
290 * The address of the pool the pair was allocated from is stored
294 struct bio bio1, bio2;
295 struct bio_vec bv1, bv2;
296 #if defined(CONFIG_BLK_DEV_INTEGRITY)
297 struct bio_integrity_payload bip1, bip2;
298 struct bio_vec iv1, iv2;
303 extern struct bio_pair *bio_split(struct bio *bi, int first_sectors);
304 extern void bio_pair_release(struct bio_pair *dbio);
306 extern struct bio_set *bioset_create(int, int);
307 extern void bioset_free(struct bio_set *);
309 extern struct bio *bio_alloc(gfp_t, int);
310 extern struct bio *bio_kmalloc(gfp_t, int);
311 extern struct bio *bio_alloc_bioset(gfp_t, int, struct bio_set *);
312 extern void bio_put(struct bio *);
313 extern void bio_free(struct bio *, struct bio_set *);
315 extern void bio_endio(struct bio *, int);
316 struct request_queue;
317 extern int bio_phys_segments(struct request_queue *, struct bio *);
319 extern void __bio_clone(struct bio *, struct bio *);
320 extern struct bio *bio_clone(struct bio *, gfp_t);
322 extern void bio_init(struct bio *);
324 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
325 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
326 unsigned int, unsigned int);
327 extern int bio_get_nr_vecs(struct block_device *);
328 extern sector_t bio_sector_offset(struct bio *, unsigned short, unsigned int);
329 extern struct bio *bio_map_user(struct request_queue *, struct block_device *,
330 unsigned long, unsigned int, int, gfp_t);
333 extern struct bio *bio_map_user_iov(struct request_queue *,
334 struct block_device *,
335 struct sg_iovec *, int, int, gfp_t);
336 extern void bio_unmap_user(struct bio *);
337 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
339 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
341 extern void bio_set_pages_dirty(struct bio *bio);
342 extern void bio_check_pages_dirty(struct bio *bio);
343 extern struct bio *bio_copy_user(struct request_queue *, struct rq_map_data *,
344 unsigned long, unsigned int, int, gfp_t);
345 extern struct bio *bio_copy_user_iov(struct request_queue *,
346 struct rq_map_data *, struct sg_iovec *,
348 extern int bio_uncopy_user(struct bio *);
349 void zero_fill_bio(struct bio *bio);
350 extern struct bio_vec *bvec_alloc_bs(gfp_t, int, unsigned long *, struct bio_set *);
351 extern unsigned int bvec_nr_vecs(unsigned short idx);
354 * Allow queuer to specify a completion CPU for this bio
356 static inline void bio_set_completion_cpu(struct bio *bio, unsigned int cpu)
358 bio->bi_comp_cpu = cpu;
362 * bio_set is used to allow other portions of the IO system to
363 * allocate their own private memory pools for bio and iovec structures.
364 * These memory pools in turn all allocate from the bio_slab
365 * and the bvec_slabs[].
367 #define BIO_POOL_SIZE 2
368 #define BIOVEC_NR_POOLS 6
372 #if defined(CONFIG_BLK_DEV_INTEGRITY)
373 mempool_t *bio_integrity_pool;
375 mempool_t *bvec_pools[BIOVEC_NR_POOLS];
381 struct kmem_cache *slab;
384 extern struct bio_set *fs_bio_set;
387 * a small number of entries is fine, not going to be performance critical.
388 * basically we just need to survive
390 #define BIO_SPLIT_ENTRIES 2
392 #ifdef CONFIG_HIGHMEM
394 * remember to add offset! and never ever reenable interrupts between a
395 * bvec_kmap_irq and bvec_kunmap_irq!!
397 * This function MUST be inlined - it plays with the CPU interrupt flags.
399 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
404 * might not be a highmem page, but the preempt/irq count
405 * balancing is a lot nicer this way
407 local_irq_save(*flags);
408 addr = (unsigned long) kmap_atomic(bvec->bv_page, KM_BIO_SRC_IRQ);
410 BUG_ON(addr & ~PAGE_MASK);
412 return (char *) addr + bvec->bv_offset;
415 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
417 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
419 kunmap_atomic((void *) ptr, KM_BIO_SRC_IRQ);
420 local_irq_restore(*flags);
424 #define bvec_kmap_irq(bvec, flags) (page_address((bvec)->bv_page) + (bvec)->bv_offset)
425 #define bvec_kunmap_irq(buf, flags) do { *(flags) = 0; } while (0)
428 static inline char *__bio_kmap_irq(struct bio *bio, unsigned short idx,
429 unsigned long *flags)
431 return bvec_kmap_irq(bio_iovec_idx(bio, idx), flags);
433 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
435 #define bio_kmap_irq(bio, flags) \
436 __bio_kmap_irq((bio), (bio)->bi_idx, (flags))
437 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
440 * Check whether this bio carries any data or not. A NULL bio is allowed.
442 static inline int bio_has_data(struct bio *bio)
444 return bio && bio->bi_io_vec != NULL;
447 #if defined(CONFIG_BLK_DEV_INTEGRITY)
449 #define bip_vec_idx(bip, idx) (&(bip->bip_vec[(idx)]))
450 #define bip_vec(bip) bip_vec_idx(bip, 0)
452 #define __bip_for_each_vec(bvl, bip, i, start_idx) \
453 for (bvl = bip_vec_idx((bip), (start_idx)), i = (start_idx); \
454 i < (bip)->bip_vcnt; \
457 #define bip_for_each_vec(bvl, bip, i) \
458 __bip_for_each_vec(bvl, bip, i, (bip)->bip_idx)
460 #define bio_integrity(bio) (bio->bi_integrity != NULL)
462 extern struct bio_integrity_payload *bio_integrity_alloc_bioset(struct bio *, gfp_t, unsigned int, struct bio_set *);
463 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
464 extern void bio_integrity_free(struct bio *, struct bio_set *);
465 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
466 extern int bio_integrity_enabled(struct bio *bio);
467 extern int bio_integrity_set_tag(struct bio *, void *, unsigned int);
468 extern int bio_integrity_get_tag(struct bio *, void *, unsigned int);
469 extern int bio_integrity_prep(struct bio *);
470 extern void bio_integrity_endio(struct bio *, int);
471 extern void bio_integrity_advance(struct bio *, unsigned int);
472 extern void bio_integrity_trim(struct bio *, unsigned int, unsigned int);
473 extern void bio_integrity_split(struct bio *, struct bio_pair *, int);
474 extern int bio_integrity_clone(struct bio *, struct bio *, struct bio_set *);
475 extern int bioset_integrity_create(struct bio_set *, int);
476 extern void bioset_integrity_free(struct bio_set *);
477 extern void bio_integrity_init_slab(void);
479 #else /* CONFIG_BLK_DEV_INTEGRITY */
481 #define bio_integrity(a) (0)
482 #define bioset_integrity_create(a, b) (0)
483 #define bio_integrity_prep(a) (0)
484 #define bio_integrity_enabled(a) (0)
485 #define bio_integrity_clone(a, b, c) (0)
486 #define bioset_integrity_free(a) do { } while (0)
487 #define bio_integrity_free(a, b) do { } while (0)
488 #define bio_integrity_endio(a, b) do { } while (0)
489 #define bio_integrity_advance(a, b) do { } while (0)
490 #define bio_integrity_trim(a, b, c) do { } while (0)
491 #define bio_integrity_split(a, b, c) do { } while (0)
492 #define bio_integrity_set_tag(a, b, c) do { } while (0)
493 #define bio_integrity_get_tag(a, b, c) do { } while (0)
494 #define bio_integrity_init_slab(a) do { } while (0)
496 #endif /* CONFIG_BLK_DEV_INTEGRITY */
498 #endif /* CONFIG_BLOCK */
499 #endif /* __LINUX_BIO_H */