]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blob - include/linux/blk_types.h
block: add scalable completion tracking of requests
[mirror_ubuntu-artful-kernel.git] / include / linux / blk_types.h
1 /*
2 * Block data types and constants. Directly include this file only to
3 * break include dependency loop.
4 */
5 #ifndef __LINUX_BLK_TYPES_H
6 #define __LINUX_BLK_TYPES_H
7
8 #include <linux/types.h>
9 #include <linux/bvec.h>
10
11 struct bio_set;
12 struct bio;
13 struct bio_integrity_payload;
14 struct page;
15 struct block_device;
16 struct io_context;
17 struct cgroup_subsys_state;
18 typedef void (bio_end_io_t) (struct bio *);
19
20 /*
21 * main unit of I/O for the block layer and lower layers (ie drivers and
22 * stacking drivers)
23 */
24 struct bio {
25 struct bio *bi_next; /* request queue link */
26 struct block_device *bi_bdev;
27 int bi_error;
28 unsigned int bi_opf; /* bottom bits req flags,
29 * top bits REQ_OP. Use
30 * accessors.
31 */
32 unsigned short bi_flags; /* status, command, etc */
33 unsigned short bi_ioprio;
34
35 struct bvec_iter bi_iter;
36
37 /* Number of segments in this BIO after
38 * physical address coalescing is performed.
39 */
40 unsigned int bi_phys_segments;
41
42 /*
43 * To keep track of the max segment size, we account for the
44 * sizes of the first and last mergeable segments in this bio.
45 */
46 unsigned int bi_seg_front_size;
47 unsigned int bi_seg_back_size;
48
49 atomic_t __bi_remaining;
50
51 bio_end_io_t *bi_end_io;
52
53 void *bi_private;
54 #ifdef CONFIG_BLK_CGROUP
55 /*
56 * Optional ioc and css associated with this bio. Put on bio
57 * release. Read comment on top of bio_associate_current().
58 */
59 struct io_context *bi_ioc;
60 struct cgroup_subsys_state *bi_css;
61 #endif
62 union {
63 #if defined(CONFIG_BLK_DEV_INTEGRITY)
64 struct bio_integrity_payload *bi_integrity; /* data integrity */
65 #endif
66 };
67
68 unsigned short bi_vcnt; /* how many bio_vec's */
69
70 /*
71 * Everything starting with bi_max_vecs will be preserved by bio_reset()
72 */
73
74 unsigned short bi_max_vecs; /* max bvl_vecs we can hold */
75
76 atomic_t __bi_cnt; /* pin count */
77
78 struct bio_vec *bi_io_vec; /* the actual vec list */
79
80 struct bio_set *bi_pool;
81
82 /*
83 * We can inline a number of vecs at the end of the bio, to avoid
84 * double allocations for a small number of bio_vecs. This member
85 * MUST obviously be kept at the very end of the bio.
86 */
87 struct bio_vec bi_inline_vecs[0];
88 };
89
90 #define BIO_RESET_BYTES offsetof(struct bio, bi_max_vecs)
91
92 /*
93 * bio flags
94 */
95 #define BIO_SEG_VALID 1 /* bi_phys_segments valid */
96 #define BIO_CLONED 2 /* doesn't own data */
97 #define BIO_BOUNCED 3 /* bio is a bounce bio */
98 #define BIO_USER_MAPPED 4 /* contains user pages */
99 #define BIO_NULL_MAPPED 5 /* contains invalid user pages */
100 #define BIO_QUIET 6 /* Make BIO Quiet */
101 #define BIO_CHAIN 7 /* chained bio, ->bi_remaining in effect */
102 #define BIO_REFFED 8 /* bio has elevated ->bi_cnt */
103 #define BIO_THROTTLED 9 /* This bio has already been subjected to
104 * throttling rules. Don't do it again. */
105
106 /*
107 * Flags starting here get preserved by bio_reset() - this includes
108 * BVEC_POOL_IDX()
109 */
110 #define BIO_RESET_BITS 10
111
112 /*
113 * We support 6 different bvec pools, the last one is magic in that it
114 * is backed by a mempool.
115 */
116 #define BVEC_POOL_NR 6
117 #define BVEC_POOL_MAX (BVEC_POOL_NR - 1)
118
119 /*
120 * Top 4 bits of bio flags indicate the pool the bvecs came from. We add
121 * 1 to the actual index so that 0 indicates that there are no bvecs to be
122 * freed.
123 */
124 #define BVEC_POOL_BITS (4)
125 #define BVEC_POOL_OFFSET (16 - BVEC_POOL_BITS)
126 #define BVEC_POOL_IDX(bio) ((bio)->bi_flags >> BVEC_POOL_OFFSET)
127
128 /*
129 * Operations and flags common to the bio and request structures.
130 * We use 8 bits for encoding the operation, and the remaining 24 for flags.
131 *
132 * The least significant bit of the operation number indicates the data
133 * transfer direction:
134 *
135 * - if the least significant bit is set transfers are TO the device
136 * - if the least significant bit is not set transfers are FROM the device
137 *
138 * If a operation does not transfer data the least significant bit has no
139 * meaning.
140 */
141 #define REQ_OP_BITS 8
142 #define REQ_OP_MASK ((1 << REQ_OP_BITS) - 1)
143 #define REQ_FLAG_BITS 24
144
145 enum req_opf {
146 /* read sectors from the device */
147 REQ_OP_READ = 0,
148 /* write sectors to the device */
149 REQ_OP_WRITE = 1,
150 /* flush the volatile write cache */
151 REQ_OP_FLUSH = 2,
152 /* discard sectors */
153 REQ_OP_DISCARD = 3,
154 /* get zone information */
155 REQ_OP_ZONE_REPORT = 4,
156 /* securely erase sectors */
157 REQ_OP_SECURE_ERASE = 5,
158 /* seset a zone write pointer */
159 REQ_OP_ZONE_RESET = 6,
160 /* write the same sector many times */
161 REQ_OP_WRITE_SAME = 7,
162
163 REQ_OP_LAST,
164 };
165
166 enum req_flag_bits {
167 __REQ_FAILFAST_DEV = /* no driver retries of device errors */
168 REQ_OP_BITS,
169 __REQ_FAILFAST_TRANSPORT, /* no driver retries of transport errors */
170 __REQ_FAILFAST_DRIVER, /* no driver retries of driver errors */
171 __REQ_SYNC, /* request is sync (sync write or read) */
172 __REQ_META, /* metadata io request */
173 __REQ_PRIO, /* boost priority in cfq */
174 __REQ_NOMERGE, /* don't touch this for merging */
175 __REQ_IDLE, /* anticipate more IO after this one */
176 __REQ_INTEGRITY, /* I/O includes block integrity payload */
177 __REQ_FUA, /* forced unit access */
178 __REQ_PREFLUSH, /* request for cache flush */
179 __REQ_RAHEAD, /* read ahead, can fail anytime */
180 __REQ_BACKGROUND, /* background IO */
181 __REQ_NR_BITS, /* stops here */
182 };
183
184 #define REQ_FAILFAST_DEV (1ULL << __REQ_FAILFAST_DEV)
185 #define REQ_FAILFAST_TRANSPORT (1ULL << __REQ_FAILFAST_TRANSPORT)
186 #define REQ_FAILFAST_DRIVER (1ULL << __REQ_FAILFAST_DRIVER)
187 #define REQ_SYNC (1ULL << __REQ_SYNC)
188 #define REQ_META (1ULL << __REQ_META)
189 #define REQ_PRIO (1ULL << __REQ_PRIO)
190 #define REQ_NOMERGE (1ULL << __REQ_NOMERGE)
191 #define REQ_IDLE (1ULL << __REQ_IDLE)
192 #define REQ_INTEGRITY (1ULL << __REQ_INTEGRITY)
193 #define REQ_FUA (1ULL << __REQ_FUA)
194 #define REQ_PREFLUSH (1ULL << __REQ_PREFLUSH)
195 #define REQ_RAHEAD (1ULL << __REQ_RAHEAD)
196 #define REQ_BACKGROUND (1ULL << __REQ_BACKGROUND)
197
198 #define REQ_FAILFAST_MASK \
199 (REQ_FAILFAST_DEV | REQ_FAILFAST_TRANSPORT | REQ_FAILFAST_DRIVER)
200
201 #define REQ_NOMERGE_FLAGS \
202 (REQ_NOMERGE | REQ_PREFLUSH | REQ_FUA)
203
204 #define bio_op(bio) \
205 ((bio)->bi_opf & REQ_OP_MASK)
206 #define req_op(req) \
207 ((req)->cmd_flags & REQ_OP_MASK)
208
209 /* obsolete, don't use in new code */
210 #define bio_set_op_attrs(bio, op, op_flags) \
211 ((bio)->bi_opf |= (op | op_flags))
212
213 static inline bool op_is_write(unsigned int op)
214 {
215 return (op & 1);
216 }
217
218 /*
219 * Reads are always treated as synchronous, as are requests with the FUA or
220 * PREFLUSH flag. Other operations may be marked as synchronous using the
221 * REQ_SYNC flag.
222 */
223 static inline bool op_is_sync(unsigned int op)
224 {
225 return (op & REQ_OP_MASK) == REQ_OP_READ ||
226 (op & (REQ_SYNC | REQ_FUA | REQ_PREFLUSH));
227 }
228
229 typedef unsigned int blk_qc_t;
230 #define BLK_QC_T_NONE -1U
231 #define BLK_QC_T_SHIFT 16
232
233 static inline bool blk_qc_t_valid(blk_qc_t cookie)
234 {
235 return cookie != BLK_QC_T_NONE;
236 }
237
238 static inline blk_qc_t blk_tag_to_qc_t(unsigned int tag, unsigned int queue_num)
239 {
240 return tag | (queue_num << BLK_QC_T_SHIFT);
241 }
242
243 static inline unsigned int blk_qc_t_to_queue_num(blk_qc_t cookie)
244 {
245 return cookie >> BLK_QC_T_SHIFT;
246 }
247
248 static inline unsigned int blk_qc_t_to_tag(blk_qc_t cookie)
249 {
250 return cookie & ((1u << BLK_QC_T_SHIFT) - 1);
251 }
252
253 struct blk_issue_stat {
254 u64 time;
255 };
256
257 #define BLK_RQ_STAT_BATCH 64
258
259 struct blk_rq_stat {
260 s64 mean;
261 u64 min;
262 u64 max;
263 s32 nr_samples;
264 s32 nr_batch;
265 u64 batch;
266 s64 time;
267 };
268
269 #endif /* __LINUX_BLK_TYPES_H */