geom_io.c revision 106338
1/*- 2 * Copyright (c) 2002 Poul-Henning Kamp 3 * Copyright (c) 2002 Networks Associates Technology, Inc. 4 * All rights reserved. 5 * 6 * This software was developed for the FreeBSD Project by Poul-Henning Kamp 7 * and NAI Labs, the Security Research Division of Network Associates, Inc. 8 * under DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the 9 * DARPA CHATS research program. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. The names of the authors may not be used to endorse or promote 20 * products derived from this software without specific prior written 21 * permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 * 35 * $FreeBSD: head/sys/geom/geom_io.c 106338 2002-11-02 11:08:07Z phk $ 36 */ 37 38 39#include <sys/param.h> 40#include <sys/stdint.h> 41#ifndef _KERNEL 42#include <stdio.h> 43#include <string.h> 44#include <stdlib.h> 45#include <signal.h> 46#include <err.h> 47#include <sched.h> 48#else 49#include <sys/systm.h> 50#include <sys/kernel.h> 51#include <sys/malloc.h> 52#include <sys/bio.h> 53#endif 54 55#include <sys/errno.h> 56#include <geom/geom.h> 57#include <geom/geom_int.h> 58 59static struct g_bioq g_bio_run_down; 60static struct g_bioq g_bio_run_up; 61static struct g_bioq g_bio_idle; 62 63static u_int pace; 64 65#include <machine/atomic.h> 66 67static void 68g_bioq_lock(struct g_bioq *bq) 69{ 70 71 mtx_lock(&bq->bio_queue_lock); 72} 73 74static void 75g_bioq_unlock(struct g_bioq *bq) 76{ 77 78 mtx_unlock(&bq->bio_queue_lock); 79} 80 81#if 0 82static void 83g_bioq_destroy(struct g_bioq *bq) 84{ 85 86 mtx_destroy(&bq->bio_queue_lock); 87} 88#endif 89 90static void 91g_bioq_init(struct g_bioq *bq) 92{ 93 94 TAILQ_INIT(&bq->bio_queue); 95 mtx_init(&bq->bio_queue_lock, "bio queue", NULL, MTX_DEF); 96} 97 98static struct bio * 99g_bioq_first(struct g_bioq *bq) 100{ 101 struct bio *bp; 102 103 g_bioq_lock(bq); 104 bp = TAILQ_FIRST(&bq->bio_queue); 105 if (bp != NULL) { 106 TAILQ_REMOVE(&bq->bio_queue, bp, bio_queue); 107 bq->bio_queue_length--; 108 } 109 g_bioq_unlock(bq); 110 return (bp); 111} 112 113static void 114g_bioq_enqueue_tail(struct bio *bp, struct g_bioq *rq) 115{ 116 117 g_bioq_lock(rq); 118 TAILQ_INSERT_TAIL(&rq->bio_queue, bp, bio_queue); 119 rq->bio_queue_length++; 120 g_bioq_unlock(rq); 121} 122 123struct bio * 124g_new_bio(void) 125{ 126 struct bio *bp; 127 128 bp = g_bioq_first(&g_bio_idle); 129 if (bp == NULL) 130 bp = g_malloc(sizeof *bp, M_NOWAIT | M_ZERO); 131 /* g_trace(G_T_BIO, "g_new_bio() = %p", bp); */ 132 return (bp); 133} 134 135void 136g_destroy_bio(struct bio *bp) 137{ 138 139 /* g_trace(G_T_BIO, "g_destroy_bio(%p)", bp); */ 140 bzero(bp, sizeof *bp); 141 g_bioq_enqueue_tail(bp, &g_bio_idle); 142} 143 144struct bio * 145g_clone_bio(struct bio *bp) 146{ 147 struct bio *bp2; 148 149 bp2 = g_new_bio(); 150 if (bp2 != NULL) { 151 bp2->bio_linkage = bp; 152 bp2->bio_cmd = bp->bio_cmd; 153 bp2->bio_length = bp->bio_length; 154 bp2->bio_offset = bp->bio_offset; 155 bp2->bio_data = bp->bio_data; 156 bp2->bio_attribute = bp->bio_attribute; 157 bp->bio_children++; /* XXX: atomic ? */ 158 } 159 /* g_trace(G_T_BIO, "g_clone_bio(%p) = %p", bp, bp2); */ 160 return(bp2); 161} 162 163void 164g_io_init() 165{ 166 167 g_bioq_init(&g_bio_run_down); 168 g_bioq_init(&g_bio_run_up); 169 g_bioq_init(&g_bio_idle); 170} 171 172int 173g_io_setattr(const char *attr, struct g_consumer *cp, int len, void *ptr) 174{ 175 struct bio *bp; 176 int error; 177 178 g_trace(G_T_BIO, "bio_setattr(%s)", attr); 179 bp = g_new_bio(); 180 bp->bio_cmd = BIO_SETATTR; 181 bp->bio_done = NULL; 182 bp->bio_attribute = attr; 183 bp->bio_length = len; 184 bp->bio_data = ptr; 185 g_io_request(bp, cp); 186 error = biowait(bp, "gsetattr"); 187 g_destroy_bio(bp); 188 return (error); 189} 190 191 192int 193g_io_getattr(const char *attr, struct g_consumer *cp, int *len, void *ptr) 194{ 195 struct bio *bp; 196 int error; 197 198 g_trace(G_T_BIO, "bio_getattr(%s)", attr); 199 bp = g_new_bio(); 200 bp->bio_cmd = BIO_GETATTR; 201 bp->bio_done = NULL; 202 bp->bio_attribute = attr; 203 bp->bio_length = *len; 204 bp->bio_data = ptr; 205 g_io_request(bp, cp); 206 error = biowait(bp, "ggetattr"); 207 *len = bp->bio_completed; 208 g_destroy_bio(bp); 209 return (error); 210} 211 212void 213g_io_request(struct bio *bp, struct g_consumer *cp) 214{ 215 int error; 216 off_t excess; 217 218 KASSERT(cp != NULL, ("NULL cp in g_io_request")); 219 KASSERT(bp != NULL, ("NULL bp in g_io_request")); 220 KASSERT(bp->bio_data != NULL, ("NULL bp->data in g_io_request")); 221 error = 0; 222 bp->bio_from = cp; 223 bp->bio_to = cp->provider; 224 bp->bio_error = 0; 225 bp->bio_completed = 0; 226 227 /* begin_stats(&bp->stats); */ 228 229 atomic_add_int(&cp->biocount, 1); 230 /* Fail on unattached consumers */ 231 if (bp->bio_to == NULL) { 232 g_io_deliver(bp, ENXIO); 233 return; 234 } 235 /* Fail if access doesn't allow operation */ 236 switch(bp->bio_cmd) { 237 case BIO_READ: 238 case BIO_GETATTR: 239 if (cp->acr == 0) { 240 g_io_deliver(bp, EPERM); 241 return; 242 } 243 break; 244 case BIO_WRITE: 245 case BIO_DELETE: 246 if (cp->acw == 0) { 247 g_io_deliver(bp, EPERM); 248 return; 249 } 250 break; 251 case BIO_SETATTR: 252 /* XXX: Should ideally check for (cp->ace == 0) */ 253 if ((cp->acw == 0)) { 254#ifdef DIAGNOSTIC 255 printf("setattr on %s mode (%d,%d,%d)\n", 256 cp->provider->name, 257 cp->acr, cp->acw, cp->ace); 258#endif 259 g_io_deliver(bp, EPERM); 260 return; 261 } 262 break; 263 default: 264 g_io_deliver(bp, EPERM); 265 return; 266 } 267 /* if provider is marked for error, don't disturb. */ 268 if (bp->bio_to->error) { 269 g_io_deliver(bp, bp->bio_to->error); 270 return; 271 } 272 switch(bp->bio_cmd) { 273 case BIO_READ: 274 case BIO_WRITE: 275 case BIO_DELETE: 276 /* Reject requests past the end of media. */ 277 if (bp->bio_offset > bp->bio_to->mediasize) { 278 g_io_deliver(bp, EIO); 279 return; 280 } 281 /* Truncate requests to the end of providers media. */ 282 excess = bp->bio_offset + bp->bio_length; 283 if (excess > bp->bio_to->mediasize) { 284 excess -= bp->bio_to->mediasize; 285 bp->bio_length -= excess; 286 } 287 /* Deliver zero length transfers right here. */ 288 if (bp->bio_length == 0) { 289 g_io_deliver(bp, 0); 290 return; 291 } 292 break; 293 default: 294 break; 295 } 296 /* Pass it on down. */ 297 g_trace(G_T_BIO, "bio_request(%p) from %p(%s) to %p(%s) cmd %d", 298 bp, bp->bio_from, bp->bio_from->geom->name, 299 bp->bio_to, bp->bio_to->name, bp->bio_cmd); 300 g_bioq_enqueue_tail(bp, &g_bio_run_down); 301 wakeup(&g_wait_down); 302} 303 304void 305g_io_deliver(struct bio *bp, int error) 306{ 307 308 KASSERT(bp != NULL, ("NULL bp in g_io_deliver")); 309 KASSERT(bp->bio_from != NULL, ("NULL bio_from in g_io_deliver")); 310 KASSERT(bp->bio_from->geom != NULL, ("NULL bio_from->geom in g_io_deliver")); 311 KASSERT(bp->bio_to != NULL, ("NULL bio_to in g_io_deliver")); 312 g_trace(G_T_BIO, 313 "g_io_deliver(%p) from %p(%s) to %p(%s) cmd %d error %d off %jd len %jd", 314 bp, bp->bio_from, bp->bio_from->geom->name, 315 bp->bio_to, bp->bio_to->name, bp->bio_cmd, error, 316 (intmax_t)bp->bio_offset, (intmax_t)bp->bio_length); 317 /* finish_stats(&bp->stats); */ 318 319 if (error == ENOMEM) { 320 printf("ENOMEM %p on %p(%s)\n", 321 bp, bp->bio_to, bp->bio_to->name); 322 g_io_request(bp, bp->bio_from); 323 pace++; 324 return; 325 } 326 327 bp->bio_error = error; 328 329 g_bioq_enqueue_tail(bp, &g_bio_run_up); 330 331 wakeup(&g_wait_up); 332} 333 334void 335g_io_schedule_down(struct thread *tp __unused) 336{ 337 struct bio *bp; 338 339 for(;;) { 340 bp = g_bioq_first(&g_bio_run_down); 341 if (bp == NULL) 342 break; 343 bp->bio_to->geom->start(bp); 344 if (pace) { 345 pace--; 346 break; 347 } 348 } 349} 350 351void 352g_io_schedule_up(struct thread *tp __unused) 353{ 354 struct bio *bp; 355 struct g_consumer *cp; 356 357 for(;;) { 358 bp = g_bioq_first(&g_bio_run_up); 359 if (bp == NULL) 360 break; 361 362 cp = bp->bio_from; 363 364 atomic_add_int(&cp->biocount, -1); 365 biodone(bp); 366 } 367} 368 369void * 370g_read_data(struct g_consumer *cp, off_t offset, off_t length, int *error) 371{ 372 struct bio *bp; 373 void *ptr; 374 int errorc; 375 376 bp = g_new_bio(); 377 bp->bio_cmd = BIO_READ; 378 bp->bio_done = NULL; 379 bp->bio_offset = offset; 380 bp->bio_length = length; 381 ptr = g_malloc(length, M_WAITOK); 382 bp->bio_data = ptr; 383 g_io_request(bp, cp); 384 errorc = biowait(bp, "gread"); 385 if (error != NULL) 386 *error = errorc; 387 g_destroy_bio(bp); 388 if (errorc) { 389 g_free(ptr); 390 ptr = NULL; 391 } 392 return (ptr); 393} 394 395int 396g_write_data(struct g_consumer *cp, off_t offset, void *ptr, off_t length) 397{ 398 struct bio *bp; 399 int error; 400 401 bp = g_new_bio(); 402 bp->bio_cmd = BIO_WRITE; 403 bp->bio_done = NULL; 404 bp->bio_offset = offset; 405 bp->bio_length = length; 406 bp->bio_data = ptr; 407 g_io_request(bp, cp); 408 error = biowait(bp, "gwrite"); 409 g_destroy_bio(bp); 410 return (error); 411} 412