| 1 | /* |
| 2 | * Copyright (c) 2009 The DragonFly Project. All rights reserved. |
| 3 | * |
| 4 | * This code is derived from software contributed to The DragonFly Project |
| 5 | * by Alex Hornung <ahornung@gmail.com> |
| 6 | * |
| 7 | * Redistribution and use in source and binary forms, with or without |
| 8 | * modification, are permitted provided that the following conditions |
| 9 | * are met: |
| 10 | * |
| 11 | * 1. Redistributions of source code must retain the above copyright |
| 12 | * notice, this list of conditions and the following disclaimer. |
| 13 | * 2. Redistributions in binary form must reproduce the above copyright |
| 14 | * notice, this list of conditions and the following disclaimer in |
| 15 | * the documentation and/or other materials provided with the |
| 16 | * distribution. |
| 17 | * 3. Neither the name of The DragonFly Project nor the names of its |
| 18 | * contributors may be used to endorse or promote products derived |
| 19 | * from this software without specific, prior written permission. |
| 20 | * |
| 21 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
| 22 | * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
| 23 | * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS |
| 24 | * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE |
| 25 | * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, |
| 26 | * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING, |
| 27 | * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
| 28 | * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED |
| 29 | * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, |
| 30 | * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT |
| 31 | * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 32 | * SUCH DAMAGE. |
| 33 | */ |
| 34 | #ifndef _SYS_DSCHED_H_ |
| 35 | #define _SYS_DSCHED_H_ |
| 36 | |
| 37 | #if defined(_KERNEL) |
| 38 | |
| 39 | #ifndef _SYS_QUEUE_H_ |
| 40 | #include <sys/queue.h> |
| 41 | #endif |
| 42 | #ifndef _SYS_BIO_H_ |
| 43 | #include <sys/bio.h> |
| 44 | #endif |
| 45 | #ifndef _SYS_BIOTRACK_H_ |
| 46 | #include <sys/biotrack.h> |
| 47 | #endif |
| 48 | #ifndef _SYS_LOCK_H_ |
| 49 | #include <sys/lock.h> |
| 50 | #endif |
| 51 | #ifndef _SYS_CONF_H_ |
| 52 | #include <sys/conf.h> |
| 53 | #endif |
| 54 | #ifndef _SYS_MSGPORT_H_ |
| 55 | #include <sys/msgport.h> |
| 56 | #endif |
| 57 | #ifndef _SYS_SYSCTL_H_ |
| 58 | #include <sys/sysctl.h> |
| 59 | #endif |
| 60 | #ifndef SYS_DISK_H_ |
| 61 | #include <sys/disk.h> |
| 62 | #endif |
| 63 | |
| 64 | #define DSCHED_POLICY_NAME_LENGTH 64 |
| 65 | |
| 66 | #define dsched_set_disk_priv(dp, x) ((dp)->d_dsched_priv1 = (x)) |
| 67 | #define dsched_get_disk_priv(dp) ((dp)?((dp)->d_dsched_priv1):NULL) |
| 68 | #define dsched_set_proc_priv(pp, x) ((pp)->p_dsched_priv1 = (x)) |
| 69 | #define dsched_get_proc_priv(pp) ((pp)?((pp)->p_dsched_priv1):NULL) |
| 70 | |
| 71 | #define dsched_set_thread_priv(td, x) ((td)->td_dsched_priv1 = (x)) |
| 72 | #define dsched_get_thread_priv(td) ((td)?((td)->td_dsched_priv1):NULL) |
| 73 | |
| 74 | #define dsched_set_buf_priv(bp, x) ((bp)->b_iosched = (x)) |
| 75 | #define dsched_get_buf_priv(bp) ((bp)?((bp)->b_iosched):NULL) |
| 76 | #define dsched_clr_buf_priv(bp) ((bp)->b_iosched = NULL) |
| 77 | #define dsched_is_clear_buf_priv(bp) ((bp)->b_iosched == NULL) |
| 78 | |
| 79 | |
| 80 | #define dsched_set_bio_dp(bio, x) ((bio)->bio_caller_info1.ptr = (x)) |
| 81 | #define dsched_get_bio_dp(bio) ((bio)?((bio)->bio_caller_info1.ptr):NULL) |
| 82 | #define dsched_set_bio_priv(bio, x) ((bio)->bio_caller_info2.ptr = (x)) |
| 83 | #define dsched_get_bio_priv(bio) ((bio)?((bio)->bio_caller_info2.ptr):NULL) |
| 84 | #define dsched_set_bio_stime(bio, x) ((bio)->bio_caller_info3.lvalue = (x)) |
| 85 | #define dsched_get_bio_stime(bio) ((bio)?((bio)->bio_caller_info3.lvalue):0) |
| 86 | #define dsched_set_bio_tdio(bio, x) ((bio)->bio_caller_info3.ptr = (x)) |
| 87 | #define dsched_get_bio_tdio(bio) ((bio)?((bio)->bio_caller_info3.ptr):0) |
| 88 | |
| 89 | |
| 90 | struct dsched_thread_ctx { |
| 91 | TAILQ_ENTRY(dsched_thread_ctx) link; |
| 92 | |
| 93 | TAILQ_HEAD(, dsched_thread_io) tdio_list; /* list of thread_io */ |
| 94 | struct lock lock; |
| 95 | |
| 96 | int32_t refcount; |
| 97 | |
| 98 | struct proc *p; |
| 99 | struct thread *td; |
| 100 | int32_t dead; |
| 101 | }; |
| 102 | |
| 103 | struct dsched_disk_ctx { |
| 104 | TAILQ_ENTRY(dsched_disk_ctx) link; |
| 105 | |
| 106 | TAILQ_HEAD(, dsched_thread_io) tdio_list; /* list of thread_io of disk */ |
| 107 | struct lock lock; |
| 108 | |
| 109 | int32_t refcount; |
| 110 | int32_t flags; |
| 111 | |
| 112 | int max_tag_queue_depth; /* estimated max tag queue depth */ |
| 113 | int current_tag_queue_depth; /* estimated current tag queue depth */ |
| 114 | |
| 115 | struct disk *dp; /* back pointer to disk struct */ |
| 116 | |
| 117 | struct sysctl_ctx_list sysctl_ctx; |
| 118 | }; |
| 119 | |
| 120 | struct dsched_policy; |
| 121 | |
| 122 | struct dsched_thread_io { |
| 123 | TAILQ_ENTRY(dsched_thread_io) link; |
| 124 | TAILQ_ENTRY(dsched_thread_io) dlink; |
| 125 | |
| 126 | TAILQ_HEAD(, bio) queue; /* IO queue (bio) */ |
| 127 | struct lock lock; |
| 128 | int32_t qlength;/* IO queue length */ |
| 129 | |
| 130 | int32_t refcount; |
| 131 | |
| 132 | int32_t flags; |
| 133 | |
| 134 | struct disk *dp; |
| 135 | struct dsched_disk_ctx *diskctx; |
| 136 | struct dsched_thread_ctx *tdctx; |
| 137 | struct proc *p; |
| 138 | struct dsched_policy *debug_policy; |
| 139 | int debug_inited; |
| 140 | int debug_priv; |
| 141 | }; |
| 142 | |
| 143 | typedef int dsched_prepare_t(struct dsched_disk_ctx *diskctx); |
| 144 | typedef void dsched_teardown_t(struct dsched_disk_ctx *diskctx); |
| 145 | typedef void dsched_cancel_t(struct dsched_disk_ctx *diskctx); |
| 146 | typedef int dsched_queue_t(struct dsched_disk_ctx *diskctx, |
| 147 | struct dsched_thread_io *tdio, struct bio *bio); |
| 148 | typedef void dsched_dequeue_t(struct dsched_disk_ctx *diskctx); |
| 149 | |
| 150 | typedef void dsched_new_tdio_t(struct dsched_thread_io *tdio); |
| 151 | typedef void dsched_new_diskctx_t(struct dsched_disk_ctx *diskctx); |
| 152 | typedef void dsched_destroy_tdio_t(struct dsched_thread_io *tdio); |
| 153 | typedef void dsched_destroy_diskctx_t(struct dsched_disk_ctx *diskctx); |
| 154 | typedef void dsched_bio_done_t(struct bio *bio); |
| 155 | typedef void dsched_polling_func_t(struct dsched_disk_ctx *diskctx); |
| 156 | |
| 157 | struct dsched_policy { |
| 158 | char name[DSCHED_POLICY_NAME_LENGTH]; |
| 159 | uint64_t uniq_id; |
| 160 | int ref_count; |
| 161 | |
| 162 | TAILQ_ENTRY(dsched_policy) link; |
| 163 | |
| 164 | dsched_prepare_t *prepare; |
| 165 | dsched_teardown_t *teardown; |
| 166 | dsched_cancel_t *cancel_all; |
| 167 | dsched_queue_t *bio_queue; |
| 168 | |
| 169 | dsched_new_tdio_t *new_tdio; |
| 170 | dsched_new_diskctx_t *new_diskctx; |
| 171 | dsched_destroy_tdio_t *destroy_tdio; |
| 172 | dsched_destroy_diskctx_t *destroy_diskctx; |
| 173 | |
| 174 | dsched_bio_done_t *bio_done; /* call back when a bio dispatched by dsched_strategy_request_polling() is done */ |
| 175 | dsched_polling_func_t *polling_func; /* it gets called when the disk is idle or about to idle */ |
| 176 | }; |
| 177 | |
| 178 | TAILQ_HEAD(dsched_policy_head, dsched_policy); |
| 179 | |
| 180 | |
| 181 | #define DSCHED_THREAD_IO_LOCKINIT(x) \ |
| 182 | lockinit(&(x)->lock, "tdiobioq", 0, LK_CANRECURSE) |
| 183 | |
| 184 | #define DSCHED_THREAD_IO_LOCK(x) do { \ |
| 185 | dsched_thread_io_ref((x)); \ |
| 186 | lockmgr(&(x)->lock, LK_EXCLUSIVE); \ |
| 187 | } while(0) |
| 188 | |
| 189 | #define DSCHED_THREAD_IO_UNLOCK(x) do { \ |
| 190 | lockmgr(&(x)->lock, LK_RELEASE); \ |
| 191 | dsched_thread_io_unref((x)); \ |
| 192 | } while(0) |
| 193 | |
| 194 | #define DSCHED_DISK_CTX_LOCKINIT(x) \ |
| 195 | lockinit(&(x)->lock, "tdiodiskq", 0, LK_CANRECURSE) |
| 196 | |
| 197 | #define DSCHED_DISK_CTX_LOCK(x) do { \ |
| 198 | dsched_disk_ctx_ref((x)); \ |
| 199 | lockmgr(&(x)->lock, LK_EXCLUSIVE); \ |
| 200 | } while(0) |
| 201 | |
| 202 | #define DSCHED_DISK_CTX_UNLOCK(x) do { \ |
| 203 | lockmgr(&(x)->lock, LK_RELEASE); \ |
| 204 | dsched_disk_ctx_unref((x)); \ |
| 205 | } while(0) |
| 206 | |
| 207 | #define DSCHED_DISK_CTX_LOCK_ASSERT(x) \ |
| 208 | KKASSERT(lockstatus(&(x)->lock, curthread) == LK_EXCLUSIVE) |
| 209 | |
| 210 | #define DSCHED_GLOBAL_THREAD_CTX_LOCKINIT(x) \ |
| 211 | lockinit(&dsched_tdctx_lock, "tdctxglob", 0, LK_CANRECURSE) |
| 212 | #define DSCHED_GLOBAL_THREAD_CTX_LOCK(x) \ |
| 213 | lockmgr(&dsched_tdctx_lock, LK_EXCLUSIVE) |
| 214 | #define DSCHED_GLOBAL_THREAD_CTX_UNLOCK(x) \ |
| 215 | lockmgr(&dsched_tdctx_lock, LK_RELEASE) |
| 216 | |
| 217 | #define DSCHED_THREAD_CTX_LOCKINIT(x) \ |
| 218 | lockinit(&(x)->lock, "tdctx", 0, LK_CANRECURSE) |
| 219 | |
| 220 | #define DSCHED_THREAD_CTX_LOCK(x) do { \ |
| 221 | dsched_thread_ctx_ref((x)); \ |
| 222 | lockmgr(&(x)->lock, LK_EXCLUSIVE); \ |
| 223 | } while(0) |
| 224 | |
| 225 | #define DSCHED_THREAD_CTX_UNLOCK(x) do { \ |
| 226 | lockmgr(&(x)->lock, LK_RELEASE); \ |
| 227 | dsched_thread_ctx_unref((x)); \ |
| 228 | } while(0) |
| 229 | |
| 230 | /* flags for thread_io */ |
| 231 | #define DSCHED_LINKED_DISK_CTX 0x01 |
| 232 | #define DSCHED_LINKED_THREAD_CTX 0x02 |
| 233 | /* flags for disk_ctx */ |
| 234 | #define DSCHED_SYSCTL_CTX_INITED 0x01 |
| 235 | |
| 236 | #define DSCHED_THREAD_CTX_MAX_SZ sizeof(struct dsched_thread_ctx) |
| 237 | #define DSCHED_THREAD_IO_MAX_SZ 256 |
| 238 | #define DSCHED_DISK_CTX_MAX_SZ 512 |
| 239 | |
| 240 | #define DSCHED_POLICY_MODULE(name, evh) \ |
| 241 | static moduledata_t name##_mod = { \ |
| 242 | #name, \ |
| 243 | evh, \ |
| 244 | NULL \ |
| 245 | }; \ |
| 246 | DECLARE_MODULE(name, name##_mod, SI_SUB_PRE_DRIVERS, SI_ORDER_MIDDLE) |
| 247 | |
| 248 | void dsched_disk_create_callback(struct disk *dp, const char *head_name, int unit); |
| 249 | void dsched_disk_update_callback(struct disk *dp, struct disk_info *info); |
| 250 | void dsched_disk_destroy_callback(struct disk *dp); |
| 251 | void dsched_queue(struct disk *dp, struct bio *bio); |
| 252 | int dsched_register(struct dsched_policy *d_policy); |
| 253 | int dsched_unregister(struct dsched_policy *d_policy); |
| 254 | int dsched_switch(struct disk *dp, struct dsched_policy *new_policy); |
| 255 | void dsched_set_policy(struct disk *dp, struct dsched_policy *new_policy); |
| 256 | struct dsched_policy *dsched_find_policy(char *search); |
| 257 | struct disk *dsched_find_disk(char *search); |
| 258 | struct dsched_policy *dsched_policy_enumerate(struct dsched_policy *pol); |
| 259 | struct disk *dsched_disk_enumerate(struct disk *dp, struct dsched_policy *policy); |
| 260 | void dsched_cancel_bio(struct bio *bp); |
| 261 | void dsched_strategy_raw(struct disk *dp, struct bio *bp); |
| 262 | void dsched_strategy_sync(struct disk *dp, struct bio *bp); |
| 263 | void dsched_strategy_async(struct disk *dp, struct bio *bp, biodone_t *done, void *priv); |
| 264 | void dsched_strategy_request_polling(struct disk *bp, struct bio *bio, struct dsched_disk_ctx *diskctx); |
| 265 | int dsched_debug(int level, char *fmt, ...) __printflike(2, 3); |
| 266 | |
| 267 | void policy_new(struct disk *dp, struct dsched_policy *pol); |
| 268 | void policy_destroy(struct disk *dp); |
| 269 | |
| 270 | void dsched_disk_ctx_ref(struct dsched_disk_ctx *diskctx); |
| 271 | void dsched_thread_io_ref(struct dsched_thread_io *tdio); |
| 272 | void dsched_thread_ctx_ref(struct dsched_thread_ctx *tdctx); |
| 273 | void dsched_disk_ctx_unref(struct dsched_disk_ctx *diskctx); |
| 274 | void dsched_thread_io_unref(struct dsched_thread_io *tdio); |
| 275 | void dsched_thread_ctx_unref(struct dsched_thread_ctx *tdctx); |
| 276 | |
| 277 | struct dsched_thread_io *dsched_new_policy_thread_tdio(struct dsched_disk_ctx *diskctx, |
| 278 | struct dsched_policy *pol); |
| 279 | struct dsched_thread_io *dsched_thread_io_alloc(struct disk *dp, |
| 280 | struct dsched_thread_ctx *tdctx, struct dsched_policy *pol); |
| 281 | struct dsched_disk_ctx *dsched_disk_ctx_alloc(struct disk *dp, |
| 282 | struct dsched_policy *pol); |
| 283 | struct dsched_thread_ctx *dsched_thread_ctx_alloc(struct proc *p); |
| 284 | |
| 285 | typedef void dsched_new_buf_t(struct buf *bp); |
| 286 | typedef void dsched_new_proc_t(struct proc *p); |
| 287 | typedef void dsched_new_thread_t(struct thread *td); |
| 288 | typedef void dsched_exit_buf_t(struct buf *bp); |
| 289 | typedef void dsched_exit_proc_t(struct proc *p); |
| 290 | typedef void dsched_exit_thread_t(struct thread *td); |
| 291 | |
| 292 | dsched_new_buf_t dsched_new_buf; |
| 293 | dsched_new_proc_t dsched_new_proc; |
| 294 | dsched_new_thread_t dsched_new_thread; |
| 295 | dsched_exit_buf_t dsched_exit_buf; |
| 296 | dsched_exit_proc_t dsched_exit_proc; |
| 297 | dsched_exit_thread_t dsched_exit_thread; |
| 298 | |
| 299 | #endif /* _KERNEL */ |
| 300 | |
| 301 | |
| 302 | #define DSCHED_NAME_LENGTH 64 |
| 303 | #define DSCHED_SET_DEVICE_POLICY _IOWR('d', 1, struct dsched_ioctl) |
| 304 | #define DSCHED_LIST_DISKS _IOWR('d', 2, struct dsched_ioctl) |
| 305 | #define DSCHED_LIST_DISK _IOWR('d', 3, struct dsched_ioctl) |
| 306 | #define DSCHED_LIST_POLICIES _IOWR('d', 4, struct dsched_ioctl) |
| 307 | |
| 308 | struct dsched_ioctl { |
| 309 | uint16_t num_elem; |
| 310 | char dev_name[DSCHED_NAME_LENGTH]; |
| 311 | char pol_name[DSCHED_NAME_LENGTH]; |
| 312 | }; |
| 313 | |
| 314 | struct dsched_stats { |
| 315 | int32_t tdctx_allocations; |
| 316 | int32_t tdio_allocations; |
| 317 | int32_t diskctx_allocations; |
| 318 | |
| 319 | int32_t no_tdctx; |
| 320 | |
| 321 | int32_t nthreads; |
| 322 | int32_t nprocs; |
| 323 | }; |
| 324 | |
| 325 | #endif /* _SYS_DSCHED_H_ */ |