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>
26 #include <linux/bug.h>
32 /* struct bio, bio_vec and BIO_* flags are defined in blk_types.h */
33 #include <linux/blk_types.h>
38 #define BIO_BUG_ON BUG_ON
43 #define BIO_MAX_PAGES 256
44 #define BIO_MAX_SIZE (BIO_MAX_PAGES << PAGE_CACHE_SHIFT)
45 #define BIO_MAX_SECTORS (BIO_MAX_SIZE >> 9)
48 * upper 16 bits of bi_rw define the io priority of this bio
50 #define BIO_PRIO_SHIFT (8 * sizeof(unsigned long) - IOPRIO_BITS)
51 #define bio_prio(bio) ((bio)->bi_rw >> BIO_PRIO_SHIFT)
52 #define bio_prio_valid(bio) ioprio_valid(bio_prio(bio))
54 #define bio_set_prio(bio, prio) do { \
55 WARN_ON(prio >= (1 << IOPRIO_BITS)); \
56 (bio)->bi_rw &= ((1UL << BIO_PRIO_SHIFT) - 1); \
57 (bio)->bi_rw |= ((unsigned long) (prio) << BIO_PRIO_SHIFT); \
61 * various member access, note that bio_data should of course not be used
62 * on highmem page vectors
64 #define bio_iovec_idx(bio, idx) (&((bio)->bi_io_vec[(idx)]))
65 #define bio_iovec(bio) bio_iovec_idx((bio), (bio)->bi_idx)
66 #define bio_page(bio) bio_iovec((bio))->bv_page
67 #define bio_offset(bio) bio_iovec((bio))->bv_offset
68 #define bio_segments(bio) ((bio)->bi_vcnt - (bio)->bi_idx)
69 #define bio_sectors(bio) ((bio)->bi_size >> 9)
71 static inline unsigned int bio_cur_bytes(struct bio *bio)
74 return bio_iovec(bio)->bv_len;
75 else /* dataless requests such as discard */
79 static inline void *bio_data(struct bio *bio)
82 return page_address(bio_page(bio)) + bio_offset(bio);
87 static inline int bio_has_allocated_vec(struct bio *bio)
89 return bio->bi_io_vec && bio->bi_io_vec != bio->bi_inline_vecs;
95 #define bio_to_phys(bio) (page_to_phys(bio_page((bio))) + (unsigned long) bio_offset((bio)))
96 #define bvec_to_phys(bv) (page_to_phys((bv)->bv_page) + (unsigned long) (bv)->bv_offset)
99 * queues that have highmem support enabled may still need to revert to
100 * PIO transfers occasionally and thus map high pages temporarily. For
101 * permanent PIO fall back, user is probably better off disabling highmem
102 * I/O completely on that queue (see ide-dma for example)
104 #define __bio_kmap_atomic(bio, idx, kmtype) \
105 (kmap_atomic(bio_iovec_idx((bio), (idx))->bv_page) + \
106 bio_iovec_idx((bio), (idx))->bv_offset)
108 #define __bio_kunmap_atomic(addr, kmtype) kunmap_atomic(addr)
114 #define __BVEC_END(bio) bio_iovec_idx((bio), (bio)->bi_vcnt - 1)
115 #define __BVEC_START(bio) bio_iovec_idx((bio), (bio)->bi_idx)
117 /* Default implementation of BIOVEC_PHYS_MERGEABLE */
118 #define __BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
119 ((bvec_to_phys((vec1)) + (vec1)->bv_len) == bvec_to_phys((vec2)))
122 * allow arch override, for eg virtualized architectures (put in asm/io.h)
124 #ifndef BIOVEC_PHYS_MERGEABLE
125 #define BIOVEC_PHYS_MERGEABLE(vec1, vec2) \
126 __BIOVEC_PHYS_MERGEABLE(vec1, vec2)
129 #define __BIO_SEG_BOUNDARY(addr1, addr2, mask) \
130 (((addr1) | (mask)) == (((addr2) - 1) | (mask)))
131 #define BIOVEC_SEG_BOUNDARY(q, b1, b2) \
132 __BIO_SEG_BOUNDARY(bvec_to_phys((b1)), bvec_to_phys((b2)) + (b2)->bv_len, queue_segment_boundary((q)))
133 #define BIO_SEG_BOUNDARY(q, b1, b2) \
134 BIOVEC_SEG_BOUNDARY((q), __BVEC_END((b1)), __BVEC_START((b2)))
136 #define bio_io_error(bio) bio_endio((bio), -EIO)
139 * drivers should not use the __ version unless they _really_ want to
140 * run through the entire bio and not just pending pieces
142 #define __bio_for_each_segment(bvl, bio, i, start_idx) \
143 for (bvl = bio_iovec_idx((bio), (start_idx)), i = (start_idx); \
144 i < (bio)->bi_vcnt; \
147 #define bio_for_each_segment(bvl, bio, i) \
148 __bio_for_each_segment(bvl, bio, i, (bio)->bi_idx)
151 * get a reference to a bio, so it won't disappear. the intended use is
155 * submit_bio(rw, bio);
156 * if (bio->bi_flags ...)
160 * without the bio_get(), it could potentially complete I/O before submit_bio
161 * returns. and then bio would be freed memory when if (bio->bi_flags ...)
164 #define bio_get(bio) atomic_inc(&(bio)->bi_cnt)
166 #if defined(CONFIG_BLK_DEV_INTEGRITY)
168 * bio integrity payload
170 struct bio_integrity_payload {
171 struct bio *bip_bio; /* parent bio */
173 sector_t bip_sector; /* virtual start sector */
175 void *bip_buf; /* generated integrity data */
176 bio_end_io_t *bip_end_io; /* saved I/O completion fn */
178 unsigned int bip_size;
180 unsigned short bip_slab; /* slab the bip came from */
181 unsigned short bip_vcnt; /* # of integrity bio_vecs */
182 unsigned short bip_idx; /* current bip_vec index */
184 struct work_struct bip_work; /* I/O completion */
186 struct bio_vec *bip_vec;
187 struct bio_vec bip_inline_vecs[0];/* embedded bvec array */
189 #endif /* CONFIG_BLK_DEV_INTEGRITY */
192 * A bio_pair is used when we need to split a bio.
193 * This can only happen for a bio that refers to just one
194 * page of data, and in the unusual situation when the
195 * page crosses a chunk/device boundary
197 * The address of the master bio is stored in bio1.bi_private
198 * The address of the pool the pair was allocated from is stored
202 struct bio bio1, bio2;
203 struct bio_vec bv1, bv2;
204 #if defined(CONFIG_BLK_DEV_INTEGRITY)
205 struct bio_integrity_payload bip1, bip2;
206 struct bio_vec iv1, iv2;
211 extern struct bio_pair *bio_split(struct bio *bi, int first_sectors);
212 extern void bio_pair_release(struct bio_pair *dbio);
214 extern struct bio_set *bioset_create(unsigned int, unsigned int);
215 extern void bioset_free(struct bio_set *);
216 extern mempool_t *biovec_create_pool(struct bio_set *bs, int pool_entries);
218 extern struct bio *bio_alloc_bioset(gfp_t, int, struct bio_set *);
219 extern void bio_put(struct bio *);
221 extern void __bio_clone(struct bio *, struct bio *);
222 extern struct bio *bio_clone_bioset(struct bio *, gfp_t, struct bio_set *bs);
224 extern struct bio_set *fs_bio_set;
226 static inline struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
228 return bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set);
231 static inline struct bio *bio_clone(struct bio *bio, gfp_t gfp_mask)
233 return bio_clone_bioset(bio, gfp_mask, fs_bio_set);
236 static inline struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs)
238 return bio_alloc_bioset(gfp_mask, nr_iovecs, NULL);
241 static inline struct bio *bio_clone_kmalloc(struct bio *bio, gfp_t gfp_mask)
243 return bio_clone_bioset(bio, gfp_mask, NULL);
247 extern void bio_endio(struct bio *, int);
248 struct request_queue;
249 extern int bio_phys_segments(struct request_queue *, struct bio *);
251 extern void bio_advance(struct bio *, unsigned);
253 extern void bio_init(struct bio *);
254 extern void bio_reset(struct bio *);
256 extern int bio_add_page(struct bio *, struct page *, unsigned int,unsigned int);
257 extern int bio_add_pc_page(struct request_queue *, struct bio *, struct page *,
258 unsigned int, unsigned int);
259 extern int bio_get_nr_vecs(struct block_device *);
260 extern sector_t bio_sector_offset(struct bio *, unsigned short, unsigned int);
261 extern struct bio *bio_map_user(struct request_queue *, struct block_device *,
262 unsigned long, unsigned int, int, gfp_t);
265 extern struct bio *bio_map_user_iov(struct request_queue *,
266 struct block_device *,
267 struct sg_iovec *, int, int, gfp_t);
268 extern void bio_unmap_user(struct bio *);
269 extern struct bio *bio_map_kern(struct request_queue *, void *, unsigned int,
271 extern struct bio *bio_copy_kern(struct request_queue *, void *, unsigned int,
273 extern void bio_set_pages_dirty(struct bio *bio);
274 extern void bio_check_pages_dirty(struct bio *bio);
276 #ifndef ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
277 # error "You should define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE for your platform"
279 #if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE
280 extern void bio_flush_dcache_pages(struct bio *bi);
282 static inline void bio_flush_dcache_pages(struct bio *bi)
287 extern struct bio *bio_copy_user(struct request_queue *, struct rq_map_data *,
288 unsigned long, unsigned int, int, gfp_t);
289 extern struct bio *bio_copy_user_iov(struct request_queue *,
290 struct rq_map_data *, struct sg_iovec *,
292 extern int bio_uncopy_user(struct bio *);
293 void zero_fill_bio(struct bio *bio);
294 extern struct bio_vec *bvec_alloc(gfp_t, int, unsigned long *, mempool_t *);
295 extern void bvec_free(mempool_t *, struct bio_vec *, unsigned int);
296 extern unsigned int bvec_nr_vecs(unsigned short idx);
298 #ifdef CONFIG_BLK_CGROUP
299 int bio_associate_current(struct bio *bio);
300 void bio_disassociate_task(struct bio *bio);
301 #else /* CONFIG_BLK_CGROUP */
302 static inline int bio_associate_current(struct bio *bio) { return -ENOENT; }
303 static inline void bio_disassociate_task(struct bio *bio) { }
304 #endif /* CONFIG_BLK_CGROUP */
306 #ifdef CONFIG_HIGHMEM
308 * remember never ever reenable interrupts between a bvec_kmap_irq and
311 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
316 * might not be a highmem page, but the preempt/irq count
317 * balancing is a lot nicer this way
319 local_irq_save(*flags);
320 addr = (unsigned long) kmap_atomic(bvec->bv_page);
322 BUG_ON(addr & ~PAGE_MASK);
324 return (char *) addr + bvec->bv_offset;
327 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
329 unsigned long ptr = (unsigned long) buffer & PAGE_MASK;
331 kunmap_atomic((void *) ptr);
332 local_irq_restore(*flags);
336 static inline char *bvec_kmap_irq(struct bio_vec *bvec, unsigned long *flags)
338 return page_address(bvec->bv_page) + bvec->bv_offset;
341 static inline void bvec_kunmap_irq(char *buffer, unsigned long *flags)
347 static inline char *__bio_kmap_irq(struct bio *bio, unsigned short idx,
348 unsigned long *flags)
350 return bvec_kmap_irq(bio_iovec_idx(bio, idx), flags);
352 #define __bio_kunmap_irq(buf, flags) bvec_kunmap_irq(buf, flags)
354 #define bio_kmap_irq(bio, flags) \
355 __bio_kmap_irq((bio), (bio)->bi_idx, (flags))
356 #define bio_kunmap_irq(buf,flags) __bio_kunmap_irq(buf, flags)
359 * Check whether this bio carries any data or not. A NULL bio is allowed.
361 static inline bool bio_has_data(struct bio *bio)
363 if (bio && bio->bi_vcnt)
369 static inline bool bio_is_rw(struct bio *bio)
371 if (!bio_has_data(bio))
374 if (bio->bi_rw & REQ_WRITE_SAME)
380 static inline bool bio_mergeable(struct bio *bio)
382 if (bio->bi_rw & REQ_NOMERGE_FLAGS)
389 * BIO list management for use by remapping drivers (e.g. DM or MD) and loop.
391 * A bio_list anchors a singly-linked list of bios chained through the bi_next
392 * member of the bio. The bio_list also caches the last list member to allow
393 * fast access to the tail.
400 static inline int bio_list_empty(const struct bio_list *bl)
402 return bl->head == NULL;
405 static inline void bio_list_init(struct bio_list *bl)
407 bl->head = bl->tail = NULL;
410 #define bio_list_for_each(bio, bl) \
411 for (bio = (bl)->head; bio; bio = bio->bi_next)
413 static inline unsigned bio_list_size(const struct bio_list *bl)
418 bio_list_for_each(bio, bl)
424 static inline void bio_list_add(struct bio_list *bl, struct bio *bio)
429 bl->tail->bi_next = bio;
436 static inline void bio_list_add_head(struct bio_list *bl, struct bio *bio)
438 bio->bi_next = bl->head;
446 static inline void bio_list_merge(struct bio_list *bl, struct bio_list *bl2)
452 bl->tail->bi_next = bl2->head;
454 bl->head = bl2->head;
456 bl->tail = bl2->tail;
459 static inline void bio_list_merge_head(struct bio_list *bl,
460 struct bio_list *bl2)
466 bl2->tail->bi_next = bl->head;
468 bl->tail = bl2->tail;
470 bl->head = bl2->head;
473 static inline struct bio *bio_list_peek(struct bio_list *bl)
478 static inline struct bio *bio_list_pop(struct bio_list *bl)
480 struct bio *bio = bl->head;
483 bl->head = bl->head->bi_next;
493 static inline struct bio *bio_list_get(struct bio_list *bl)
495 struct bio *bio = bl->head;
497 bl->head = bl->tail = NULL;
503 * bio_set is used to allow other portions of the IO system to
504 * allocate their own private memory pools for bio and iovec structures.
505 * These memory pools in turn all allocate from the bio_slab
506 * and the bvec_slabs[].
508 #define BIO_POOL_SIZE 2
509 #define BIOVEC_NR_POOLS 6
510 #define BIOVEC_MAX_IDX (BIOVEC_NR_POOLS - 1)
513 struct kmem_cache *bio_slab;
514 unsigned int front_pad;
517 mempool_t *bvec_pool;
518 #if defined(CONFIG_BLK_DEV_INTEGRITY)
519 mempool_t *bio_integrity_pool;
520 mempool_t *bvec_integrity_pool;
524 * Deadlock avoidance for stacking block drivers: see comments in
525 * bio_alloc_bioset() for details
527 spinlock_t rescue_lock;
528 struct bio_list rescue_list;
529 struct work_struct rescue_work;
530 struct workqueue_struct *rescue_workqueue;
536 struct kmem_cache *slab;
540 * a small number of entries is fine, not going to be performance critical.
541 * basically we just need to survive
543 #define BIO_SPLIT_ENTRIES 2
545 #if defined(CONFIG_BLK_DEV_INTEGRITY)
547 #define bip_vec_idx(bip, idx) (&(bip->bip_vec[(idx)]))
548 #define bip_vec(bip) bip_vec_idx(bip, 0)
550 #define __bip_for_each_vec(bvl, bip, i, start_idx) \
551 for (bvl = bip_vec_idx((bip), (start_idx)), i = (start_idx); \
552 i < (bip)->bip_vcnt; \
555 #define bip_for_each_vec(bvl, bip, i) \
556 __bip_for_each_vec(bvl, bip, i, (bip)->bip_idx)
558 #define bio_for_each_integrity_vec(_bvl, _bio, _iter) \
560 bip_for_each_vec(_bvl, _bio->bi_integrity, _iter)
562 #define bio_integrity(bio) (bio->bi_integrity != NULL)
564 extern struct bio_integrity_payload *bio_integrity_alloc(struct bio *, gfp_t, unsigned int);
565 extern void bio_integrity_free(struct bio *);
566 extern int bio_integrity_add_page(struct bio *, struct page *, unsigned int, unsigned int);
567 extern int bio_integrity_enabled(struct bio *bio);
568 extern int bio_integrity_set_tag(struct bio *, void *, unsigned int);
569 extern int bio_integrity_get_tag(struct bio *, void *, unsigned int);
570 extern int bio_integrity_prep(struct bio *);
571 extern void bio_integrity_endio(struct bio *, int);
572 extern void bio_integrity_advance(struct bio *, unsigned int);
573 extern void bio_integrity_trim(struct bio *, unsigned int, unsigned int);
574 extern void bio_integrity_split(struct bio *, struct bio_pair *, int);
575 extern int bio_integrity_clone(struct bio *, struct bio *, gfp_t);
576 extern int bioset_integrity_create(struct bio_set *, int);
577 extern void bioset_integrity_free(struct bio_set *);
578 extern void bio_integrity_init(void);
580 #else /* CONFIG_BLK_DEV_INTEGRITY */
582 static inline int bio_integrity(struct bio *bio)
587 static inline int bio_integrity_enabled(struct bio *bio)
592 static inline int bioset_integrity_create(struct bio_set *bs, int pool_size)
597 static inline void bioset_integrity_free (struct bio_set *bs)
602 static inline int bio_integrity_prep(struct bio *bio)
607 static inline void bio_integrity_free(struct bio *bio)
612 static inline int bio_integrity_clone(struct bio *bio, struct bio *bio_src,
618 static inline void bio_integrity_split(struct bio *bio, struct bio_pair *bp,
624 static inline void bio_integrity_advance(struct bio *bio,
625 unsigned int bytes_done)
630 static inline void bio_integrity_trim(struct bio *bio, unsigned int offset,
631 unsigned int sectors)
636 static inline void bio_integrity_init(void)
641 #endif /* CONFIG_BLK_DEV_INTEGRITY */
643 #endif /* CONFIG_BLOCK */
644 #endif /* __LINUX_BIO_H */