ctl_backend_block.c revision 288800
1/*- 2 * Copyright (c) 2003 Silicon Graphics International Corp. 3 * Copyright (c) 2009-2011 Spectra Logic Corporation 4 * Copyright (c) 2012 The FreeBSD Foundation 5 * All rights reserved. 6 * 7 * Portions of this software were developed by Edward Tomasz Napierala 8 * under sponsorship from the FreeBSD Foundation. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions, and the following disclaimer, 15 * without modification. 16 * 2. Redistributions in binary form must reproduce at minimum a disclaimer 17 * substantially similar to the "NO WARRANTY" disclaimer below 18 * ("Disclaimer") and any redistribution must be conditioned upon 19 * including a substantially similar Disclaimer requirement for further 20 * binary redistribution. 21 * 22 * NO WARRANTY 23 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 24 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 25 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR 26 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 27 * HOLDERS OR CONTRIBUTORS BE LIABLE FOR 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, 31 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING 32 * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 33 * POSSIBILITY OF SUCH DAMAGES. 34 * 35 * $Id: //depot/users/kenm/FreeBSD-test2/sys/cam/ctl/ctl_backend_block.c#5 $ 36 */ 37/* 38 * CAM Target Layer driver backend for block devices. 39 * 40 * Author: Ken Merry <ken@FreeBSD.org> 41 */ 42#include <sys/cdefs.h> 43__FBSDID("$FreeBSD: stable/10/sys/cam/ctl/ctl_backend_block.c 288800 2015-10-05 11:03:24Z mav $"); 44 45#include <opt_kdtrace.h> 46 47#include <sys/param.h> 48#include <sys/systm.h> 49#include <sys/kernel.h> 50#include <sys/types.h> 51#include <sys/kthread.h> 52#include <sys/bio.h> 53#include <sys/fcntl.h> 54#include <sys/limits.h> 55#include <sys/lock.h> 56#include <sys/mutex.h> 57#include <sys/condvar.h> 58#include <sys/malloc.h> 59#include <sys/conf.h> 60#include <sys/ioccom.h> 61#include <sys/queue.h> 62#include <sys/sbuf.h> 63#include <sys/endian.h> 64#include <sys/uio.h> 65#include <sys/buf.h> 66#include <sys/taskqueue.h> 67#include <sys/vnode.h> 68#include <sys/namei.h> 69#include <sys/mount.h> 70#include <sys/disk.h> 71#include <sys/fcntl.h> 72#include <sys/filedesc.h> 73#include <sys/filio.h> 74#include <sys/proc.h> 75#include <sys/pcpu.h> 76#include <sys/module.h> 77#include <sys/sdt.h> 78#include <sys/devicestat.h> 79#include <sys/sysctl.h> 80 81#include <geom/geom.h> 82 83#include <cam/cam.h> 84#include <cam/scsi/scsi_all.h> 85#include <cam/scsi/scsi_da.h> 86#include <cam/ctl/ctl_io.h> 87#include <cam/ctl/ctl.h> 88#include <cam/ctl/ctl_backend.h> 89#include <cam/ctl/ctl_ioctl.h> 90#include <cam/ctl/ctl_ha.h> 91#include <cam/ctl/ctl_scsi_all.h> 92#include <cam/ctl/ctl_private.h> 93#include <cam/ctl/ctl_error.h> 94 95/* 96 * The idea here is that we'll allocate enough S/G space to hold a 1MB 97 * I/O. If we get an I/O larger than that, we'll split it. 98 */ 99#define CTLBLK_HALF_IO_SIZE (512 * 1024) 100#define CTLBLK_MAX_IO_SIZE (CTLBLK_HALF_IO_SIZE * 2) 101#define CTLBLK_MAX_SEG MAXPHYS 102#define CTLBLK_HALF_SEGS MAX(CTLBLK_HALF_IO_SIZE / CTLBLK_MAX_SEG, 1) 103#define CTLBLK_MAX_SEGS (CTLBLK_HALF_SEGS * 2) 104 105#ifdef CTLBLK_DEBUG 106#define DPRINTF(fmt, args...) \ 107 printf("cbb(%s:%d): " fmt, __FUNCTION__, __LINE__, ##args) 108#else 109#define DPRINTF(fmt, args...) do {} while(0) 110#endif 111 112#define PRIV(io) \ 113 ((struct ctl_ptr_len_flags *)&(io)->io_hdr.ctl_private[CTL_PRIV_BACKEND]) 114#define ARGS(io) \ 115 ((struct ctl_lba_len_flags *)&(io)->io_hdr.ctl_private[CTL_PRIV_LBA_LEN]) 116 117SDT_PROVIDER_DEFINE(cbb); 118 119typedef enum { 120 CTL_BE_BLOCK_LUN_UNCONFIGURED = 0x01, 121 CTL_BE_BLOCK_LUN_CONFIG_ERR = 0x02, 122 CTL_BE_BLOCK_LUN_WAITING = 0x04, 123} ctl_be_block_lun_flags; 124 125typedef enum { 126 CTL_BE_BLOCK_NONE, 127 CTL_BE_BLOCK_DEV, 128 CTL_BE_BLOCK_FILE 129} ctl_be_block_type; 130 131struct ctl_be_block_filedata { 132 struct ucred *cred; 133}; 134 135union ctl_be_block_bedata { 136 struct ctl_be_block_filedata file; 137}; 138 139struct ctl_be_block_io; 140struct ctl_be_block_lun; 141 142typedef void (*cbb_dispatch_t)(struct ctl_be_block_lun *be_lun, 143 struct ctl_be_block_io *beio); 144typedef uint64_t (*cbb_getattr_t)(struct ctl_be_block_lun *be_lun, 145 const char *attrname); 146 147/* 148 * Backend LUN structure. There is a 1:1 mapping between a block device 149 * and a backend block LUN, and between a backend block LUN and a CTL LUN. 150 */ 151struct ctl_be_block_lun { 152 struct ctl_lun_create_params params; 153 char lunname[32]; 154 char *dev_path; 155 ctl_be_block_type dev_type; 156 struct vnode *vn; 157 union ctl_be_block_bedata backend; 158 cbb_dispatch_t dispatch; 159 cbb_dispatch_t lun_flush; 160 cbb_dispatch_t unmap; 161 cbb_dispatch_t get_lba_status; 162 cbb_getattr_t getattr; 163 uma_zone_t lun_zone; 164 uint64_t size_blocks; 165 uint64_t size_bytes; 166 struct ctl_be_block_softc *softc; 167 struct devstat *disk_stats; 168 ctl_be_block_lun_flags flags; 169 STAILQ_ENTRY(ctl_be_block_lun) links; 170 struct ctl_be_lun cbe_lun; 171 struct taskqueue *io_taskqueue; 172 struct task io_task; 173 int num_threads; 174 STAILQ_HEAD(, ctl_io_hdr) input_queue; 175 STAILQ_HEAD(, ctl_io_hdr) config_read_queue; 176 STAILQ_HEAD(, ctl_io_hdr) config_write_queue; 177 STAILQ_HEAD(, ctl_io_hdr) datamove_queue; 178 struct mtx_padalign io_lock; 179 struct mtx_padalign queue_lock; 180}; 181 182/* 183 * Overall softc structure for the block backend module. 184 */ 185struct ctl_be_block_softc { 186 struct mtx lock; 187 int num_luns; 188 STAILQ_HEAD(, ctl_be_block_lun) lun_list; 189}; 190 191static struct ctl_be_block_softc backend_block_softc; 192 193/* 194 * Per-I/O information. 195 */ 196struct ctl_be_block_io { 197 union ctl_io *io; 198 struct ctl_sg_entry sg_segs[CTLBLK_MAX_SEGS]; 199 struct iovec xiovecs[CTLBLK_MAX_SEGS]; 200 int bio_cmd; 201 int num_segs; 202 int num_bios_sent; 203 int num_bios_done; 204 int send_complete; 205 int num_errors; 206 struct bintime ds_t0; 207 devstat_tag_type ds_tag_type; 208 devstat_trans_flags ds_trans_type; 209 uint64_t io_len; 210 uint64_t io_offset; 211 int io_arg; 212 struct ctl_be_block_softc *softc; 213 struct ctl_be_block_lun *lun; 214 void (*beio_cont)(struct ctl_be_block_io *beio); /* to continue processing */ 215}; 216 217extern struct ctl_softc *control_softc; 218 219static int cbb_num_threads = 14; 220TUNABLE_INT("kern.cam.ctl.block.num_threads", &cbb_num_threads); 221SYSCTL_NODE(_kern_cam_ctl, OID_AUTO, block, CTLFLAG_RD, 0, 222 "CAM Target Layer Block Backend"); 223SYSCTL_INT(_kern_cam_ctl_block, OID_AUTO, num_threads, CTLFLAG_RW, 224 &cbb_num_threads, 0, "Number of threads per backing file"); 225 226static struct ctl_be_block_io *ctl_alloc_beio(struct ctl_be_block_softc *softc); 227static void ctl_free_beio(struct ctl_be_block_io *beio); 228static void ctl_complete_beio(struct ctl_be_block_io *beio); 229static int ctl_be_block_move_done(union ctl_io *io); 230static void ctl_be_block_biodone(struct bio *bio); 231static void ctl_be_block_flush_file(struct ctl_be_block_lun *be_lun, 232 struct ctl_be_block_io *beio); 233static void ctl_be_block_dispatch_file(struct ctl_be_block_lun *be_lun, 234 struct ctl_be_block_io *beio); 235static void ctl_be_block_gls_file(struct ctl_be_block_lun *be_lun, 236 struct ctl_be_block_io *beio); 237static uint64_t ctl_be_block_getattr_file(struct ctl_be_block_lun *be_lun, 238 const char *attrname); 239static void ctl_be_block_flush_dev(struct ctl_be_block_lun *be_lun, 240 struct ctl_be_block_io *beio); 241static void ctl_be_block_unmap_dev(struct ctl_be_block_lun *be_lun, 242 struct ctl_be_block_io *beio); 243static void ctl_be_block_dispatch_dev(struct ctl_be_block_lun *be_lun, 244 struct ctl_be_block_io *beio); 245static uint64_t ctl_be_block_getattr_dev(struct ctl_be_block_lun *be_lun, 246 const char *attrname); 247static void ctl_be_block_cr_dispatch(struct ctl_be_block_lun *be_lun, 248 union ctl_io *io); 249static void ctl_be_block_cw_dispatch(struct ctl_be_block_lun *be_lun, 250 union ctl_io *io); 251static void ctl_be_block_dispatch(struct ctl_be_block_lun *be_lun, 252 union ctl_io *io); 253static void ctl_be_block_worker(void *context, int pending); 254static int ctl_be_block_submit(union ctl_io *io); 255static int ctl_be_block_ioctl(struct cdev *dev, u_long cmd, caddr_t addr, 256 int flag, struct thread *td); 257static int ctl_be_block_open_file(struct ctl_be_block_lun *be_lun, 258 struct ctl_lun_req *req); 259static int ctl_be_block_open_dev(struct ctl_be_block_lun *be_lun, 260 struct ctl_lun_req *req); 261static int ctl_be_block_close(struct ctl_be_block_lun *be_lun); 262static int ctl_be_block_open(struct ctl_be_block_softc *softc, 263 struct ctl_be_block_lun *be_lun, 264 struct ctl_lun_req *req); 265static int ctl_be_block_create(struct ctl_be_block_softc *softc, 266 struct ctl_lun_req *req); 267static int ctl_be_block_rm(struct ctl_be_block_softc *softc, 268 struct ctl_lun_req *req); 269static int ctl_be_block_modify(struct ctl_be_block_softc *softc, 270 struct ctl_lun_req *req); 271static void ctl_be_block_lun_shutdown(void *be_lun); 272static void ctl_be_block_lun_config_status(void *be_lun, 273 ctl_lun_config_status status); 274static int ctl_be_block_config_write(union ctl_io *io); 275static int ctl_be_block_config_read(union ctl_io *io); 276static int ctl_be_block_lun_info(void *be_lun, struct sbuf *sb); 277static uint64_t ctl_be_block_lun_attr(void *be_lun, const char *attrname); 278int ctl_be_block_init(void); 279 280static struct ctl_backend_driver ctl_be_block_driver = 281{ 282 .name = "block", 283 .flags = CTL_BE_FLAG_HAS_CONFIG, 284 .init = ctl_be_block_init, 285 .data_submit = ctl_be_block_submit, 286 .data_move_done = ctl_be_block_move_done, 287 .config_read = ctl_be_block_config_read, 288 .config_write = ctl_be_block_config_write, 289 .ioctl = ctl_be_block_ioctl, 290 .lun_info = ctl_be_block_lun_info, 291 .lun_attr = ctl_be_block_lun_attr 292}; 293 294MALLOC_DEFINE(M_CTLBLK, "ctlblk", "Memory used for CTL block backend"); 295CTL_BACKEND_DECLARE(cbb, ctl_be_block_driver); 296 297static uma_zone_t beio_zone; 298 299static struct ctl_be_block_io * 300ctl_alloc_beio(struct ctl_be_block_softc *softc) 301{ 302 struct ctl_be_block_io *beio; 303 304 beio = uma_zalloc(beio_zone, M_WAITOK | M_ZERO); 305 beio->softc = softc; 306 return (beio); 307} 308 309static void 310ctl_free_beio(struct ctl_be_block_io *beio) 311{ 312 int duplicate_free; 313 int i; 314 315 duplicate_free = 0; 316 317 for (i = 0; i < beio->num_segs; i++) { 318 if (beio->sg_segs[i].addr == NULL) 319 duplicate_free++; 320 321 uma_zfree(beio->lun->lun_zone, beio->sg_segs[i].addr); 322 beio->sg_segs[i].addr = NULL; 323 324 /* For compare we had two equal S/G lists. */ 325 if (ARGS(beio->io)->flags & CTL_LLF_COMPARE) { 326 uma_zfree(beio->lun->lun_zone, 327 beio->sg_segs[i + CTLBLK_HALF_SEGS].addr); 328 beio->sg_segs[i + CTLBLK_HALF_SEGS].addr = NULL; 329 } 330 } 331 332 if (duplicate_free > 0) { 333 printf("%s: %d duplicate frees out of %d segments\n", __func__, 334 duplicate_free, beio->num_segs); 335 } 336 337 uma_zfree(beio_zone, beio); 338} 339 340static void 341ctl_complete_beio(struct ctl_be_block_io *beio) 342{ 343 union ctl_io *io = beio->io; 344 345 if (beio->beio_cont != NULL) { 346 beio->beio_cont(beio); 347 } else { 348 ctl_free_beio(beio); 349 ctl_data_submit_done(io); 350 } 351} 352 353static size_t 354cmp(uint8_t *a, uint8_t *b, size_t size) 355{ 356 size_t i; 357 358 for (i = 0; i < size; i++) { 359 if (a[i] != b[i]) 360 break; 361 } 362 return (i); 363} 364 365static void 366ctl_be_block_compare(union ctl_io *io) 367{ 368 struct ctl_be_block_io *beio; 369 uint64_t off, res; 370 int i; 371 uint8_t info[8]; 372 373 beio = (struct ctl_be_block_io *)PRIV(io)->ptr; 374 off = 0; 375 for (i = 0; i < beio->num_segs; i++) { 376 res = cmp(beio->sg_segs[i].addr, 377 beio->sg_segs[i + CTLBLK_HALF_SEGS].addr, 378 beio->sg_segs[i].len); 379 off += res; 380 if (res < beio->sg_segs[i].len) 381 break; 382 } 383 if (i < beio->num_segs) { 384 scsi_u64to8b(off, info); 385 ctl_set_sense(&io->scsiio, /*current_error*/ 1, 386 /*sense_key*/ SSD_KEY_MISCOMPARE, 387 /*asc*/ 0x1D, /*ascq*/ 0x00, 388 /*type*/ SSD_ELEM_INFO, 389 /*size*/ sizeof(info), /*data*/ &info, 390 /*type*/ SSD_ELEM_NONE); 391 } else 392 ctl_set_success(&io->scsiio); 393} 394 395static int 396ctl_be_block_move_done(union ctl_io *io) 397{ 398 struct ctl_be_block_io *beio; 399 struct ctl_be_block_lun *be_lun; 400 struct ctl_lba_len_flags *lbalen; 401#ifdef CTL_TIME_IO 402 struct bintime cur_bt; 403#endif 404 405 beio = (struct ctl_be_block_io *)PRIV(io)->ptr; 406 be_lun = beio->lun; 407 408 DPRINTF("entered\n"); 409 410#ifdef CTL_TIME_IO 411 getbinuptime(&cur_bt); 412 bintime_sub(&cur_bt, &io->io_hdr.dma_start_bt); 413 bintime_add(&io->io_hdr.dma_bt, &cur_bt); 414#endif 415 io->io_hdr.num_dmas++; 416 io->scsiio.kern_rel_offset += io->scsiio.kern_data_len; 417 418 /* 419 * We set status at this point for read commands, and write 420 * commands with errors. 421 */ 422 if (io->io_hdr.flags & CTL_FLAG_ABORT) { 423 ; 424 } else if ((io->io_hdr.port_status == 0) && 425 ((io->io_hdr.status & CTL_STATUS_MASK) == CTL_STATUS_NONE)) { 426 lbalen = ARGS(beio->io); 427 if (lbalen->flags & CTL_LLF_READ) { 428 ctl_set_success(&io->scsiio); 429 } else if (lbalen->flags & CTL_LLF_COMPARE) { 430 /* We have two data blocks ready for comparison. */ 431 ctl_be_block_compare(io); 432 } 433 } else if ((io->io_hdr.port_status != 0) && 434 ((io->io_hdr.status & CTL_STATUS_MASK) == CTL_STATUS_NONE || 435 (io->io_hdr.status & CTL_STATUS_MASK) == CTL_SUCCESS)) { 436 /* 437 * For hardware error sense keys, the sense key 438 * specific value is defined to be a retry count, 439 * but we use it to pass back an internal FETD 440 * error code. XXX KDM Hopefully the FETD is only 441 * using 16 bits for an error code, since that's 442 * all the space we have in the sks field. 443 */ 444 ctl_set_internal_failure(&io->scsiio, 445 /*sks_valid*/ 1, 446 /*retry_count*/ 447 io->io_hdr.port_status); 448 } 449 450 /* 451 * If this is a read, or a write with errors, it is done. 452 */ 453 if ((beio->bio_cmd == BIO_READ) 454 || ((io->io_hdr.flags & CTL_FLAG_ABORT) != 0) 455 || ((io->io_hdr.status & CTL_STATUS_MASK) != CTL_STATUS_NONE)) { 456 ctl_complete_beio(beio); 457 return (0); 458 } 459 460 /* 461 * At this point, we have a write and the DMA completed 462 * successfully. We now have to queue it to the task queue to 463 * execute the backend I/O. That is because we do blocking 464 * memory allocations, and in the file backing case, blocking I/O. 465 * This move done routine is generally called in the SIM's 466 * interrupt context, and therefore we cannot block. 467 */ 468 mtx_lock(&be_lun->queue_lock); 469 STAILQ_INSERT_TAIL(&be_lun->datamove_queue, &io->io_hdr, links); 470 mtx_unlock(&be_lun->queue_lock); 471 taskqueue_enqueue(be_lun->io_taskqueue, &be_lun->io_task); 472 473 return (0); 474} 475 476static void 477ctl_be_block_biodone(struct bio *bio) 478{ 479 struct ctl_be_block_io *beio; 480 struct ctl_be_block_lun *be_lun; 481 union ctl_io *io; 482 int error; 483 484 beio = bio->bio_caller1; 485 be_lun = beio->lun; 486 io = beio->io; 487 488 DPRINTF("entered\n"); 489 490 error = bio->bio_error; 491 mtx_lock(&be_lun->io_lock); 492 if (error != 0) 493 beio->num_errors++; 494 495 beio->num_bios_done++; 496 497 /* 498 * XXX KDM will this cause WITNESS to complain? Holding a lock 499 * during the free might cause it to complain. 500 */ 501 g_destroy_bio(bio); 502 503 /* 504 * If the send complete bit isn't set, or we aren't the last I/O to 505 * complete, then we're done. 506 */ 507 if ((beio->send_complete == 0) 508 || (beio->num_bios_done < beio->num_bios_sent)) { 509 mtx_unlock(&be_lun->io_lock); 510 return; 511 } 512 513 /* 514 * At this point, we've verified that we are the last I/O to 515 * complete, so it's safe to drop the lock. 516 */ 517 devstat_end_transaction(beio->lun->disk_stats, beio->io_len, 518 beio->ds_tag_type, beio->ds_trans_type, 519 /*now*/ NULL, /*then*/&beio->ds_t0); 520 mtx_unlock(&be_lun->io_lock); 521 522 /* 523 * If there are any errors from the backing device, we fail the 524 * entire I/O with a medium error. 525 */ 526 if (beio->num_errors > 0) { 527 if (error == EOPNOTSUPP) { 528 ctl_set_invalid_opcode(&io->scsiio); 529 } else if (error == ENOSPC || error == EDQUOT) { 530 ctl_set_space_alloc_fail(&io->scsiio); 531 } else if (error == EROFS || error == EACCES) { 532 ctl_set_hw_write_protected(&io->scsiio); 533 } else if (beio->bio_cmd == BIO_FLUSH) { 534 /* XXX KDM is there is a better error here? */ 535 ctl_set_internal_failure(&io->scsiio, 536 /*sks_valid*/ 1, 537 /*retry_count*/ 0xbad2); 538 } else { 539 ctl_set_medium_error(&io->scsiio, 540 beio->bio_cmd == BIO_READ); 541 } 542 ctl_complete_beio(beio); 543 return; 544 } 545 546 /* 547 * If this is a write, a flush, a delete or verify, we're all done. 548 * If this is a read, we can now send the data to the user. 549 */ 550 if ((beio->bio_cmd == BIO_WRITE) 551 || (beio->bio_cmd == BIO_FLUSH) 552 || (beio->bio_cmd == BIO_DELETE) 553 || (ARGS(io)->flags & CTL_LLF_VERIFY)) { 554 ctl_set_success(&io->scsiio); 555 ctl_complete_beio(beio); 556 } else { 557 if ((ARGS(io)->flags & CTL_LLF_READ) && 558 beio->beio_cont == NULL) { 559 ctl_set_success(&io->scsiio); 560 ctl_serseq_done(io); 561 } 562#ifdef CTL_TIME_IO 563 getbinuptime(&io->io_hdr.dma_start_bt); 564#endif 565 ctl_datamove(io); 566 } 567} 568 569static void 570ctl_be_block_flush_file(struct ctl_be_block_lun *be_lun, 571 struct ctl_be_block_io *beio) 572{ 573 union ctl_io *io = beio->io; 574 struct mount *mountpoint; 575 int error, lock_flags; 576 577 DPRINTF("entered\n"); 578 579 binuptime(&beio->ds_t0); 580 mtx_lock(&be_lun->io_lock); 581 devstat_start_transaction(beio->lun->disk_stats, &beio->ds_t0); 582 mtx_unlock(&be_lun->io_lock); 583 584 (void) vn_start_write(be_lun->vn, &mountpoint, V_WAIT); 585 586 if (MNT_SHARED_WRITES(mountpoint) || 587 ((mountpoint == NULL) && MNT_SHARED_WRITES(be_lun->vn->v_mount))) 588 lock_flags = LK_SHARED; 589 else 590 lock_flags = LK_EXCLUSIVE; 591 vn_lock(be_lun->vn, lock_flags | LK_RETRY); 592 error = VOP_FSYNC(be_lun->vn, beio->io_arg ? MNT_NOWAIT : MNT_WAIT, 593 curthread); 594 VOP_UNLOCK(be_lun->vn, 0); 595 596 vn_finished_write(mountpoint); 597 598 mtx_lock(&be_lun->io_lock); 599 devstat_end_transaction(beio->lun->disk_stats, beio->io_len, 600 beio->ds_tag_type, beio->ds_trans_type, 601 /*now*/ NULL, /*then*/&beio->ds_t0); 602 mtx_unlock(&be_lun->io_lock); 603 604 if (error == 0) 605 ctl_set_success(&io->scsiio); 606 else { 607 /* XXX KDM is there is a better error here? */ 608 ctl_set_internal_failure(&io->scsiio, 609 /*sks_valid*/ 1, 610 /*retry_count*/ 0xbad1); 611 } 612 613 ctl_complete_beio(beio); 614} 615 616SDT_PROBE_DEFINE1(cbb, kernel, read, file_start, "uint64_t"); 617SDT_PROBE_DEFINE1(cbb, kernel, write, file_start, "uint64_t"); 618SDT_PROBE_DEFINE1(cbb, kernel, read, file_done,"uint64_t"); 619SDT_PROBE_DEFINE1(cbb, kernel, write, file_done, "uint64_t"); 620 621static void 622ctl_be_block_dispatch_file(struct ctl_be_block_lun *be_lun, 623 struct ctl_be_block_io *beio) 624{ 625 struct ctl_be_block_filedata *file_data; 626 union ctl_io *io; 627 struct uio xuio; 628 struct iovec *xiovec; 629 size_t s; 630 int error, flags, i; 631 632 DPRINTF("entered\n"); 633 634 file_data = &be_lun->backend.file; 635 io = beio->io; 636 flags = 0; 637 if (ARGS(io)->flags & CTL_LLF_DPO) 638 flags |= IO_DIRECT; 639 if (beio->bio_cmd == BIO_WRITE && ARGS(io)->flags & CTL_LLF_FUA) 640 flags |= IO_SYNC; 641 642 bzero(&xuio, sizeof(xuio)); 643 if (beio->bio_cmd == BIO_READ) { 644 SDT_PROBE(cbb, kernel, read, file_start, 0, 0, 0, 0, 0); 645 xuio.uio_rw = UIO_READ; 646 } else { 647 SDT_PROBE(cbb, kernel, write, file_start, 0, 0, 0, 0, 0); 648 xuio.uio_rw = UIO_WRITE; 649 } 650 xuio.uio_offset = beio->io_offset; 651 xuio.uio_resid = beio->io_len; 652 xuio.uio_segflg = UIO_SYSSPACE; 653 xuio.uio_iov = beio->xiovecs; 654 xuio.uio_iovcnt = beio->num_segs; 655 xuio.uio_td = curthread; 656 657 for (i = 0, xiovec = xuio.uio_iov; i < xuio.uio_iovcnt; i++, xiovec++) { 658 xiovec->iov_base = beio->sg_segs[i].addr; 659 xiovec->iov_len = beio->sg_segs[i].len; 660 } 661 662 binuptime(&beio->ds_t0); 663 mtx_lock(&be_lun->io_lock); 664 devstat_start_transaction(beio->lun->disk_stats, &beio->ds_t0); 665 mtx_unlock(&be_lun->io_lock); 666 667 if (beio->bio_cmd == BIO_READ) { 668 vn_lock(be_lun->vn, LK_SHARED | LK_RETRY); 669 670 /* 671 * UFS pays attention to IO_DIRECT for reads. If the 672 * DIRECTIO option is configured into the kernel, it calls 673 * ffs_rawread(). But that only works for single-segment 674 * uios with user space addresses. In our case, with a 675 * kernel uio, it still reads into the buffer cache, but it 676 * will just try to release the buffer from the cache later 677 * on in ffs_read(). 678 * 679 * ZFS does not pay attention to IO_DIRECT for reads. 680 * 681 * UFS does not pay attention to IO_SYNC for reads. 682 * 683 * ZFS pays attention to IO_SYNC (which translates into the 684 * Solaris define FRSYNC for zfs_read()) for reads. It 685 * attempts to sync the file before reading. 686 */ 687 error = VOP_READ(be_lun->vn, &xuio, flags, file_data->cred); 688 689 VOP_UNLOCK(be_lun->vn, 0); 690 SDT_PROBE(cbb, kernel, read, file_done, 0, 0, 0, 0, 0); 691 if (error == 0 && xuio.uio_resid > 0) { 692 /* 693 * If we red less then requested (EOF), then 694 * we should clean the rest of the buffer. 695 */ 696 s = beio->io_len - xuio.uio_resid; 697 for (i = 0; i < beio->num_segs; i++) { 698 if (s >= beio->sg_segs[i].len) { 699 s -= beio->sg_segs[i].len; 700 continue; 701 } 702 bzero((uint8_t *)beio->sg_segs[i].addr + s, 703 beio->sg_segs[i].len - s); 704 s = 0; 705 } 706 } 707 } else { 708 struct mount *mountpoint; 709 int lock_flags; 710 711 (void)vn_start_write(be_lun->vn, &mountpoint, V_WAIT); 712 713 if (MNT_SHARED_WRITES(mountpoint) || ((mountpoint == NULL) 714 && MNT_SHARED_WRITES(be_lun->vn->v_mount))) 715 lock_flags = LK_SHARED; 716 else 717 lock_flags = LK_EXCLUSIVE; 718 vn_lock(be_lun->vn, lock_flags | LK_RETRY); 719 720 /* 721 * UFS pays attention to IO_DIRECT for writes. The write 722 * is done asynchronously. (Normally the write would just 723 * get put into cache. 724 * 725 * UFS pays attention to IO_SYNC for writes. It will 726 * attempt to write the buffer out synchronously if that 727 * flag is set. 728 * 729 * ZFS does not pay attention to IO_DIRECT for writes. 730 * 731 * ZFS pays attention to IO_SYNC (a.k.a. FSYNC or FRSYNC) 732 * for writes. It will flush the transaction from the 733 * cache before returning. 734 */ 735 error = VOP_WRITE(be_lun->vn, &xuio, flags, file_data->cred); 736 VOP_UNLOCK(be_lun->vn, 0); 737 738 vn_finished_write(mountpoint); 739 SDT_PROBE(cbb, kernel, write, file_done, 0, 0, 0, 0, 0); 740 } 741 742 mtx_lock(&be_lun->io_lock); 743 devstat_end_transaction(beio->lun->disk_stats, beio->io_len, 744 beio->ds_tag_type, beio->ds_trans_type, 745 /*now*/ NULL, /*then*/&beio->ds_t0); 746 mtx_unlock(&be_lun->io_lock); 747 748 /* 749 * If we got an error, set the sense data to "MEDIUM ERROR" and 750 * return the I/O to the user. 751 */ 752 if (error != 0) { 753 if (error == ENOSPC || error == EDQUOT) { 754 ctl_set_space_alloc_fail(&io->scsiio); 755 } else if (error == EROFS || error == EACCES) { 756 ctl_set_hw_write_protected(&io->scsiio); 757 } else { 758 ctl_set_medium_error(&io->scsiio, 759 beio->bio_cmd == BIO_READ); 760 } 761 ctl_complete_beio(beio); 762 return; 763 } 764 765 /* 766 * If this is a write or a verify, we're all done. 767 * If this is a read, we can now send the data to the user. 768 */ 769 if ((beio->bio_cmd == BIO_WRITE) || 770 (ARGS(io)->flags & CTL_LLF_VERIFY)) { 771 ctl_set_success(&io->scsiio); 772 ctl_complete_beio(beio); 773 } else { 774 if ((ARGS(io)->flags & CTL_LLF_READ) && 775 beio->beio_cont == NULL) { 776 ctl_set_success(&io->scsiio); 777 ctl_serseq_done(io); 778 } 779#ifdef CTL_TIME_IO 780 getbinuptime(&io->io_hdr.dma_start_bt); 781#endif 782 ctl_datamove(io); 783 } 784} 785 786static void 787ctl_be_block_gls_file(struct ctl_be_block_lun *be_lun, 788 struct ctl_be_block_io *beio) 789{ 790 union ctl_io *io = beio->io; 791 struct ctl_lba_len_flags *lbalen = ARGS(io); 792 struct scsi_get_lba_status_data *data; 793 off_t roff, off; 794 int error, status; 795 796 DPRINTF("entered\n"); 797 798 off = roff = ((off_t)lbalen->lba) * be_lun->cbe_lun.blocksize; 799 vn_lock(be_lun->vn, LK_SHARED | LK_RETRY); 800 error = VOP_IOCTL(be_lun->vn, FIOSEEKHOLE, &off, 801 0, curthread->td_ucred, curthread); 802 if (error == 0 && off > roff) 803 status = 0; /* mapped up to off */ 804 else { 805 error = VOP_IOCTL(be_lun->vn, FIOSEEKDATA, &off, 806 0, curthread->td_ucred, curthread); 807 if (error == 0 && off > roff) 808 status = 1; /* deallocated up to off */ 809 else { 810 status = 0; /* unknown up to the end */ 811 off = be_lun->size_bytes; 812 } 813 } 814 VOP_UNLOCK(be_lun->vn, 0); 815 816 data = (struct scsi_get_lba_status_data *)io->scsiio.kern_data_ptr; 817 scsi_u64to8b(lbalen->lba, data->descr[0].addr); 818 scsi_ulto4b(MIN(UINT32_MAX, off / be_lun->cbe_lun.blocksize - 819 lbalen->lba), data->descr[0].length); 820 data->descr[0].status = status; 821 822 ctl_complete_beio(beio); 823} 824 825static uint64_t 826ctl_be_block_getattr_file(struct ctl_be_block_lun *be_lun, const char *attrname) 827{ 828 struct vattr vattr; 829 struct statfs statfs; 830 uint64_t val; 831 int error; 832 833 val = UINT64_MAX; 834 if (be_lun->vn == NULL) 835 return (val); 836 vn_lock(be_lun->vn, LK_SHARED | LK_RETRY); 837 if (strcmp(attrname, "blocksused") == 0) { 838 error = VOP_GETATTR(be_lun->vn, &vattr, curthread->td_ucred); 839 if (error == 0) 840 val = vattr.va_bytes / be_lun->cbe_lun.blocksize; 841 } 842 if (strcmp(attrname, "blocksavail") == 0 && 843 (be_lun->vn->v_iflag & VI_DOOMED) == 0) { 844 error = VFS_STATFS(be_lun->vn->v_mount, &statfs); 845 if (error == 0) 846 val = statfs.f_bavail * statfs.f_bsize / 847 be_lun->cbe_lun.blocksize; 848 } 849 VOP_UNLOCK(be_lun->vn, 0); 850 return (val); 851} 852 853static void 854ctl_be_block_dispatch_zvol(struct ctl_be_block_lun *be_lun, 855 struct ctl_be_block_io *beio) 856{ 857 union ctl_io *io; 858 struct cdevsw *csw; 859 struct cdev *dev; 860 struct uio xuio; 861 struct iovec *xiovec; 862 int error, flags, i, ref; 863 864 DPRINTF("entered\n"); 865 866 io = beio->io; 867 flags = 0; 868 if (ARGS(io)->flags & CTL_LLF_DPO) 869 flags |= IO_DIRECT; 870 if (beio->bio_cmd == BIO_WRITE && ARGS(io)->flags & CTL_LLF_FUA) 871 flags |= IO_SYNC; 872 873 bzero(&xuio, sizeof(xuio)); 874 if (beio->bio_cmd == BIO_READ) { 875 SDT_PROBE(cbb, kernel, read, file_start, 0, 0, 0, 0, 0); 876 xuio.uio_rw = UIO_READ; 877 } else { 878 SDT_PROBE(cbb, kernel, write, file_start, 0, 0, 0, 0, 0); 879 xuio.uio_rw = UIO_WRITE; 880 } 881 xuio.uio_offset = beio->io_offset; 882 xuio.uio_resid = beio->io_len; 883 xuio.uio_segflg = UIO_SYSSPACE; 884 xuio.uio_iov = beio->xiovecs; 885 xuio.uio_iovcnt = beio->num_segs; 886 xuio.uio_td = curthread; 887 888 for (i = 0, xiovec = xuio.uio_iov; i < xuio.uio_iovcnt; i++, xiovec++) { 889 xiovec->iov_base = beio->sg_segs[i].addr; 890 xiovec->iov_len = beio->sg_segs[i].len; 891 } 892 893 binuptime(&beio->ds_t0); 894 mtx_lock(&be_lun->io_lock); 895 devstat_start_transaction(beio->lun->disk_stats, &beio->ds_t0); 896 mtx_unlock(&be_lun->io_lock); 897 898 csw = devvn_refthread(be_lun->vn, &dev, &ref); 899 if (csw) { 900 if (beio->bio_cmd == BIO_READ) 901 error = csw->d_read(dev, &xuio, flags); 902 else 903 error = csw->d_write(dev, &xuio, flags); 904 dev_relthread(dev, ref); 905 } else 906 error = ENXIO; 907 908 if (beio->bio_cmd == BIO_READ) 909 SDT_PROBE(cbb, kernel, read, file_done, 0, 0, 0, 0, 0); 910 else 911 SDT_PROBE(cbb, kernel, write, file_done, 0, 0, 0, 0, 0); 912 913 mtx_lock(&be_lun->io_lock); 914 devstat_end_transaction(beio->lun->disk_stats, beio->io_len, 915 beio->ds_tag_type, beio->ds_trans_type, 916 /*now*/ NULL, /*then*/&beio->ds_t0); 917 mtx_unlock(&be_lun->io_lock); 918 919 /* 920 * If we got an error, set the sense data to "MEDIUM ERROR" and 921 * return the I/O to the user. 922 */ 923 if (error != 0) { 924 if (error == ENOSPC || error == EDQUOT) { 925 ctl_set_space_alloc_fail(&io->scsiio); 926 } else if (error == EROFS || error == EACCES) { 927 ctl_set_hw_write_protected(&io->scsiio); 928 } else { 929 ctl_set_medium_error(&io->scsiio, 930 beio->bio_cmd == BIO_READ); 931 } 932 ctl_complete_beio(beio); 933 return; 934 } 935 936 /* 937 * If this is a write or a verify, we're all done. 938 * If this is a read, we can now send the data to the user. 939 */ 940 if ((beio->bio_cmd == BIO_WRITE) || 941 (ARGS(io)->flags & CTL_LLF_VERIFY)) { 942 ctl_set_success(&io->scsiio); 943 ctl_complete_beio(beio); 944 } else { 945 if ((ARGS(io)->flags & CTL_LLF_READ) && 946 beio->beio_cont == NULL) { 947 ctl_set_success(&io->scsiio); 948 ctl_serseq_done(io); 949 } 950#ifdef CTL_TIME_IO 951 getbinuptime(&io->io_hdr.dma_start_bt); 952#endif 953 ctl_datamove(io); 954 } 955} 956 957static void 958ctl_be_block_gls_zvol(struct ctl_be_block_lun *be_lun, 959 struct ctl_be_block_io *beio) 960{ 961 union ctl_io *io = beio->io; 962 struct cdevsw *csw; 963 struct cdev *dev; 964 struct ctl_lba_len_flags *lbalen = ARGS(io); 965 struct scsi_get_lba_status_data *data; 966 off_t roff, off; 967 int error, ref, status; 968 969 DPRINTF("entered\n"); 970 971 csw = devvn_refthread(be_lun->vn, &dev, &ref); 972 if (csw == NULL) { 973 status = 0; /* unknown up to the end */ 974 off = be_lun->size_bytes; 975 goto done; 976 } 977 off = roff = ((off_t)lbalen->lba) * be_lun->cbe_lun.blocksize; 978 error = csw->d_ioctl(dev, FIOSEEKHOLE, (caddr_t)&off, FREAD, 979 curthread); 980 if (error == 0 && off > roff) 981 status = 0; /* mapped up to off */ 982 else { 983 error = csw->d_ioctl(dev, FIOSEEKDATA, (caddr_t)&off, FREAD, 984 curthread); 985 if (error == 0 && off > roff) 986 status = 1; /* deallocated up to off */ 987 else { 988 status = 0; /* unknown up to the end */ 989 off = be_lun->size_bytes; 990 } 991 } 992 dev_relthread(dev, ref); 993 994done: 995 data = (struct scsi_get_lba_status_data *)io->scsiio.kern_data_ptr; 996 scsi_u64to8b(lbalen->lba, data->descr[0].addr); 997 scsi_ulto4b(MIN(UINT32_MAX, off / be_lun->cbe_lun.blocksize - 998 lbalen->lba), data->descr[0].length); 999 data->descr[0].status = status; 1000 1001 ctl_complete_beio(beio); 1002} 1003 1004static void 1005ctl_be_block_flush_dev(struct ctl_be_block_lun *be_lun, 1006 struct ctl_be_block_io *beio) 1007{ 1008 struct bio *bio; 1009 struct cdevsw *csw; 1010 struct cdev *dev; 1011 int ref; 1012 1013 DPRINTF("entered\n"); 1014 1015 /* This can't fail, it's a blocking allocation. */ 1016 bio = g_alloc_bio(); 1017 1018 bio->bio_cmd = BIO_FLUSH; 1019 bio->bio_offset = 0; 1020 bio->bio_data = 0; 1021 bio->bio_done = ctl_be_block_biodone; 1022 bio->bio_caller1 = beio; 1023 bio->bio_pblkno = 0; 1024 1025 /* 1026 * We don't need to acquire the LUN lock here, because we are only 1027 * sending one bio, and so there is no other context to synchronize 1028 * with. 1029 */ 1030 beio->num_bios_sent = 1; 1031 beio->send_complete = 1; 1032 1033 binuptime(&beio->ds_t0); 1034 mtx_lock(&be_lun->io_lock); 1035 devstat_start_transaction(be_lun->disk_stats, &beio->ds_t0); 1036 mtx_unlock(&be_lun->io_lock); 1037 1038 csw = devvn_refthread(be_lun->vn, &dev, &ref); 1039 if (csw) { 1040 bio->bio_dev = dev; 1041 csw->d_strategy(bio); 1042 dev_relthread(dev, ref); 1043 } else { 1044 bio->bio_error = ENXIO; 1045 ctl_be_block_biodone(bio); 1046 } 1047} 1048 1049static void 1050ctl_be_block_unmap_dev_range(struct ctl_be_block_lun *be_lun, 1051 struct ctl_be_block_io *beio, 1052 uint64_t off, uint64_t len, int last) 1053{ 1054 struct bio *bio; 1055 uint64_t maxlen; 1056 struct cdevsw *csw; 1057 struct cdev *dev; 1058 int ref; 1059 1060 csw = devvn_refthread(be_lun->vn, &dev, &ref); 1061 maxlen = LONG_MAX - (LONG_MAX % be_lun->cbe_lun.blocksize); 1062 while (len > 0) { 1063 bio = g_alloc_bio(); 1064 bio->bio_cmd = BIO_DELETE; 1065 bio->bio_dev = dev; 1066 bio->bio_offset = off; 1067 bio->bio_length = MIN(len, maxlen); 1068 bio->bio_data = 0; 1069 bio->bio_done = ctl_be_block_biodone; 1070 bio->bio_caller1 = beio; 1071 bio->bio_pblkno = off / be_lun->cbe_lun.blocksize; 1072 1073 off += bio->bio_length; 1074 len -= bio->bio_length; 1075 1076 mtx_lock(&be_lun->io_lock); 1077 beio->num_bios_sent++; 1078 if (last && len == 0) 1079 beio->send_complete = 1; 1080 mtx_unlock(&be_lun->io_lock); 1081 1082 if (csw) { 1083 csw->d_strategy(bio); 1084 } else { 1085 bio->bio_error = ENXIO; 1086 ctl_be_block_biodone(bio); 1087 } 1088 } 1089 if (csw) 1090 dev_relthread(dev, ref); 1091} 1092 1093static void 1094ctl_be_block_unmap_dev(struct ctl_be_block_lun *be_lun, 1095 struct ctl_be_block_io *beio) 1096{ 1097 union ctl_io *io; 1098 struct ctl_ptr_len_flags *ptrlen; 1099 struct scsi_unmap_desc *buf, *end; 1100 uint64_t len; 1101 1102 io = beio->io; 1103 1104 DPRINTF("entered\n"); 1105 1106 binuptime(&beio->ds_t0); 1107 mtx_lock(&be_lun->io_lock); 1108 devstat_start_transaction(be_lun->disk_stats, &beio->ds_t0); 1109 mtx_unlock(&be_lun->io_lock); 1110 1111 if (beio->io_offset == -1) { 1112 beio->io_len = 0; 1113 ptrlen = (struct ctl_ptr_len_flags *)&io->io_hdr.ctl_private[CTL_PRIV_LBA_LEN]; 1114 buf = (struct scsi_unmap_desc *)ptrlen->ptr; 1115 end = buf + ptrlen->len / sizeof(*buf); 1116 for (; buf < end; buf++) { 1117 len = (uint64_t)scsi_4btoul(buf->length) * 1118 be_lun->cbe_lun.blocksize; 1119 beio->io_len += len; 1120 ctl_be_block_unmap_dev_range(be_lun, beio, 1121 scsi_8btou64(buf->lba) * be_lun->cbe_lun.blocksize, 1122 len, (end - buf < 2) ? TRUE : FALSE); 1123 } 1124 } else 1125 ctl_be_block_unmap_dev_range(be_lun, beio, 1126 beio->io_offset, beio->io_len, TRUE); 1127} 1128 1129static void 1130ctl_be_block_dispatch_dev(struct ctl_be_block_lun *be_lun, 1131 struct ctl_be_block_io *beio) 1132{ 1133 TAILQ_HEAD(, bio) queue = TAILQ_HEAD_INITIALIZER(queue); 1134 struct bio *bio; 1135 struct cdevsw *csw; 1136 struct cdev *dev; 1137 off_t cur_offset; 1138 int i, max_iosize, ref; 1139 1140 DPRINTF("entered\n"); 1141 csw = devvn_refthread(be_lun->vn, &dev, &ref); 1142 1143 /* 1144 * We have to limit our I/O size to the maximum supported by the 1145 * backend device. Hopefully it is MAXPHYS. If the driver doesn't 1146 * set it properly, use DFLTPHYS. 1147 */ 1148 if (csw) { 1149 max_iosize = dev->si_iosize_max; 1150 if (max_iosize < PAGE_SIZE) 1151 max_iosize = DFLTPHYS; 1152 } else 1153 max_iosize = DFLTPHYS; 1154 1155 cur_offset = beio->io_offset; 1156 for (i = 0; i < beio->num_segs; i++) { 1157 size_t cur_size; 1158 uint8_t *cur_ptr; 1159 1160 cur_size = beio->sg_segs[i].len; 1161 cur_ptr = beio->sg_segs[i].addr; 1162 1163 while (cur_size > 0) { 1164 /* This can't fail, it's a blocking allocation. */ 1165 bio = g_alloc_bio(); 1166 1167 KASSERT(bio != NULL, ("g_alloc_bio() failed!\n")); 1168 1169 bio->bio_cmd = beio->bio_cmd; 1170 bio->bio_dev = dev; 1171 bio->bio_caller1 = beio; 1172 bio->bio_length = min(cur_size, max_iosize); 1173 bio->bio_offset = cur_offset; 1174 bio->bio_data = cur_ptr; 1175 bio->bio_done = ctl_be_block_biodone; 1176 bio->bio_pblkno = cur_offset / be_lun->cbe_lun.blocksize; 1177 1178 cur_offset += bio->bio_length; 1179 cur_ptr += bio->bio_length; 1180 cur_size -= bio->bio_length; 1181 1182 TAILQ_INSERT_TAIL(&queue, bio, bio_queue); 1183 beio->num_bios_sent++; 1184 } 1185 } 1186 binuptime(&beio->ds_t0); 1187 mtx_lock(&be_lun->io_lock); 1188 devstat_start_transaction(be_lun->disk_stats, &beio->ds_t0); 1189 beio->send_complete = 1; 1190 mtx_unlock(&be_lun->io_lock); 1191 1192 /* 1193 * Fire off all allocated requests! 1194 */ 1195 while ((bio = TAILQ_FIRST(&queue)) != NULL) { 1196 TAILQ_REMOVE(&queue, bio, bio_queue); 1197 if (csw) 1198 csw->d_strategy(bio); 1199 else { 1200 bio->bio_error = ENXIO; 1201 ctl_be_block_biodone(bio); 1202 } 1203 } 1204 if (csw) 1205 dev_relthread(dev, ref); 1206} 1207 1208static uint64_t 1209ctl_be_block_getattr_dev(struct ctl_be_block_lun *be_lun, const char *attrname) 1210{ 1211 struct diocgattr_arg arg; 1212 struct cdevsw *csw; 1213 struct cdev *dev; 1214 int error, ref; 1215 1216 csw = devvn_refthread(be_lun->vn, &dev, &ref); 1217 if (csw == NULL) 1218 return (UINT64_MAX); 1219 strlcpy(arg.name, attrname, sizeof(arg.name)); 1220 arg.len = sizeof(arg.value.off); 1221 if (csw->d_ioctl) { 1222 error = csw->d_ioctl(dev, DIOCGATTR, (caddr_t)&arg, FREAD, 1223 curthread); 1224 } else 1225 error = ENODEV; 1226 dev_relthread(dev, ref); 1227 if (error != 0) 1228 return (UINT64_MAX); 1229 return (arg.value.off); 1230} 1231 1232static void 1233ctl_be_block_cw_dispatch_sync(struct ctl_be_block_lun *be_lun, 1234 union ctl_io *io) 1235{ 1236 struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun; 1237 struct ctl_be_block_io *beio; 1238 struct ctl_lba_len_flags *lbalen; 1239 1240 DPRINTF("entered\n"); 1241 beio = (struct ctl_be_block_io *)PRIV(io)->ptr; 1242 lbalen = (struct ctl_lba_len_flags *)&io->io_hdr.ctl_private[CTL_PRIV_LBA_LEN]; 1243 1244 beio->io_len = lbalen->len * cbe_lun->blocksize; 1245 beio->io_offset = lbalen->lba * cbe_lun->blocksize; 1246 beio->io_arg = (lbalen->flags & SSC_IMMED) != 0; 1247 beio->bio_cmd = BIO_FLUSH; 1248 beio->ds_trans_type = DEVSTAT_NO_DATA; 1249 DPRINTF("SYNC\n"); 1250 be_lun->lun_flush(be_lun, beio); 1251} 1252 1253static void 1254ctl_be_block_cw_done_ws(struct ctl_be_block_io *beio) 1255{ 1256 union ctl_io *io; 1257 1258 io = beio->io; 1259 ctl_free_beio(beio); 1260 if ((io->io_hdr.flags & CTL_FLAG_ABORT) || 1261 ((io->io_hdr.status & CTL_STATUS_MASK) != CTL_STATUS_NONE && 1262 (io->io_hdr.status & CTL_STATUS_MASK) != CTL_SUCCESS)) { 1263 ctl_config_write_done(io); 1264 return; 1265 } 1266 1267 ctl_be_block_config_write(io); 1268} 1269 1270static void 1271ctl_be_block_cw_dispatch_ws(struct ctl_be_block_lun *be_lun, 1272 union ctl_io *io) 1273{ 1274 struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun; 1275 struct ctl_be_block_io *beio; 1276 struct ctl_lba_len_flags *lbalen; 1277 uint64_t len_left, lba; 1278 uint32_t pb, pbo, adj; 1279 int i, seglen; 1280 uint8_t *buf, *end; 1281 1282 DPRINTF("entered\n"); 1283 1284 beio = (struct ctl_be_block_io *)PRIV(io)->ptr; 1285 lbalen = ARGS(beio->io); 1286 1287 if (lbalen->flags & ~(SWS_LBDATA | SWS_UNMAP | SWS_ANCHOR | SWS_NDOB) || 1288 (lbalen->flags & (SWS_UNMAP | SWS_ANCHOR) && be_lun->unmap == NULL)) { 1289 ctl_free_beio(beio); 1290 ctl_set_invalid_field(&io->scsiio, 1291 /*sks_valid*/ 1, 1292 /*command*/ 1, 1293 /*field*/ 1, 1294 /*bit_valid*/ 0, 1295 /*bit*/ 0); 1296 ctl_config_write_done(io); 1297 return; 1298 } 1299 1300 if (lbalen->flags & (SWS_UNMAP | SWS_ANCHOR)) { 1301 beio->io_offset = lbalen->lba * cbe_lun->blocksize; 1302 beio->io_len = (uint64_t)lbalen->len * cbe_lun->blocksize; 1303 beio->bio_cmd = BIO_DELETE; 1304 beio->ds_trans_type = DEVSTAT_FREE; 1305 1306 be_lun->unmap(be_lun, beio); 1307 return; 1308 } 1309 1310 beio->bio_cmd = BIO_WRITE; 1311 beio->ds_trans_type = DEVSTAT_WRITE; 1312 1313 DPRINTF("WRITE SAME at LBA %jx len %u\n", 1314 (uintmax_t)lbalen->lba, lbalen->len); 1315 1316 pb = cbe_lun->blocksize << be_lun->cbe_lun.pblockexp; 1317 if (be_lun->cbe_lun.pblockoff > 0) 1318 pbo = pb - cbe_lun->blocksize * be_lun->cbe_lun.pblockoff; 1319 else 1320 pbo = 0; 1321 len_left = (uint64_t)lbalen->len * cbe_lun->blocksize; 1322 for (i = 0, lba = 0; i < CTLBLK_MAX_SEGS && len_left > 0; i++) { 1323 1324 /* 1325 * Setup the S/G entry for this chunk. 1326 */ 1327 seglen = MIN(CTLBLK_MAX_SEG, len_left); 1328 if (pb > cbe_lun->blocksize) { 1329 adj = ((lbalen->lba + lba) * cbe_lun->blocksize + 1330 seglen - pbo) % pb; 1331 if (seglen > adj) 1332 seglen -= adj; 1333 else 1334 seglen -= seglen % cbe_lun->blocksize; 1335 } else 1336 seglen -= seglen % cbe_lun->blocksize; 1337 beio->sg_segs[i].len = seglen; 1338 beio->sg_segs[i].addr = uma_zalloc(be_lun->lun_zone, M_WAITOK); 1339 1340 DPRINTF("segment %d addr %p len %zd\n", i, 1341 beio->sg_segs[i].addr, beio->sg_segs[i].len); 1342 1343 beio->num_segs++; 1344 len_left -= seglen; 1345 1346 buf = beio->sg_segs[i].addr; 1347 end = buf + seglen; 1348 for (; buf < end; buf += cbe_lun->blocksize) { 1349 if (lbalen->flags & SWS_NDOB) { 1350 memset(buf, 0, cbe_lun->blocksize); 1351 } else { 1352 memcpy(buf, io->scsiio.kern_data_ptr, 1353 cbe_lun->blocksize); 1354 } 1355 if (lbalen->flags & SWS_LBDATA) 1356 scsi_ulto4b(lbalen->lba + lba, buf); 1357 lba++; 1358 } 1359 } 1360 1361 beio->io_offset = lbalen->lba * cbe_lun->blocksize; 1362 beio->io_len = lba * cbe_lun->blocksize; 1363 1364 /* We can not do all in one run. Correct and schedule rerun. */ 1365 if (len_left > 0) { 1366 lbalen->lba += lba; 1367 lbalen->len -= lba; 1368 beio->beio_cont = ctl_be_block_cw_done_ws; 1369 } 1370 1371 be_lun->dispatch(be_lun, beio); 1372} 1373 1374static void 1375ctl_be_block_cw_dispatch_unmap(struct ctl_be_block_lun *be_lun, 1376 union ctl_io *io) 1377{ 1378 struct ctl_be_block_io *beio; 1379 struct ctl_ptr_len_flags *ptrlen; 1380 1381 DPRINTF("entered\n"); 1382 1383 beio = (struct ctl_be_block_io *)PRIV(io)->ptr; 1384 ptrlen = (struct ctl_ptr_len_flags *)&io->io_hdr.ctl_private[CTL_PRIV_LBA_LEN]; 1385 1386 if ((ptrlen->flags & ~SU_ANCHOR) != 0 || be_lun->unmap == NULL) { 1387 ctl_free_beio(beio); 1388 ctl_set_invalid_field(&io->scsiio, 1389 /*sks_valid*/ 0, 1390 /*command*/ 1, 1391 /*field*/ 0, 1392 /*bit_valid*/ 0, 1393 /*bit*/ 0); 1394 ctl_config_write_done(io); 1395 return; 1396 } 1397 1398 beio->io_len = 0; 1399 beio->io_offset = -1; 1400 beio->bio_cmd = BIO_DELETE; 1401 beio->ds_trans_type = DEVSTAT_FREE; 1402 DPRINTF("UNMAP\n"); 1403 be_lun->unmap(be_lun, beio); 1404} 1405 1406static void 1407ctl_be_block_cr_done(struct ctl_be_block_io *beio) 1408{ 1409 union ctl_io *io; 1410 1411 io = beio->io; 1412 ctl_free_beio(beio); 1413 ctl_config_read_done(io); 1414} 1415 1416static void 1417ctl_be_block_cr_dispatch(struct ctl_be_block_lun *be_lun, 1418 union ctl_io *io) 1419{ 1420 struct ctl_be_block_io *beio; 1421 struct ctl_be_block_softc *softc; 1422 1423 DPRINTF("entered\n"); 1424 1425 softc = be_lun->softc; 1426 beio = ctl_alloc_beio(softc); 1427 beio->io = io; 1428 beio->lun = be_lun; 1429 beio->beio_cont = ctl_be_block_cr_done; 1430 PRIV(io)->ptr = (void *)beio; 1431 1432 switch (io->scsiio.cdb[0]) { 1433 case SERVICE_ACTION_IN: /* GET LBA STATUS */ 1434 beio->bio_cmd = -1; 1435 beio->ds_trans_type = DEVSTAT_NO_DATA; 1436 beio->ds_tag_type = DEVSTAT_TAG_ORDERED; 1437 beio->io_len = 0; 1438 if (be_lun->get_lba_status) 1439 be_lun->get_lba_status(be_lun, beio); 1440 else 1441 ctl_be_block_cr_done(beio); 1442 break; 1443 default: 1444 panic("Unhandled CDB type %#x", io->scsiio.cdb[0]); 1445 break; 1446 } 1447} 1448 1449static void 1450ctl_be_block_cw_done(struct ctl_be_block_io *beio) 1451{ 1452 union ctl_io *io; 1453 1454 io = beio->io; 1455 ctl_free_beio(beio); 1456 ctl_config_write_done(io); 1457} 1458 1459static void 1460ctl_be_block_cw_dispatch(struct ctl_be_block_lun *be_lun, 1461 union ctl_io *io) 1462{ 1463 struct ctl_be_block_io *beio; 1464 struct ctl_be_block_softc *softc; 1465 1466 DPRINTF("entered\n"); 1467 1468 softc = be_lun->softc; 1469 beio = ctl_alloc_beio(softc); 1470 beio->io = io; 1471 beio->lun = be_lun; 1472 beio->beio_cont = ctl_be_block_cw_done; 1473 switch (io->scsiio.tag_type) { 1474 case CTL_TAG_ORDERED: 1475 beio->ds_tag_type = DEVSTAT_TAG_ORDERED; 1476 break; 1477 case CTL_TAG_HEAD_OF_QUEUE: 1478 beio->ds_tag_type = DEVSTAT_TAG_HEAD; 1479 break; 1480 case CTL_TAG_UNTAGGED: 1481 case CTL_TAG_SIMPLE: 1482 case CTL_TAG_ACA: 1483 default: 1484 beio->ds_tag_type = DEVSTAT_TAG_SIMPLE; 1485 break; 1486 } 1487 PRIV(io)->ptr = (void *)beio; 1488 1489 switch (io->scsiio.cdb[0]) { 1490 case SYNCHRONIZE_CACHE: 1491 case SYNCHRONIZE_CACHE_16: 1492 ctl_be_block_cw_dispatch_sync(be_lun, io); 1493 break; 1494 case WRITE_SAME_10: 1495 case WRITE_SAME_16: 1496 ctl_be_block_cw_dispatch_ws(be_lun, io); 1497 break; 1498 case UNMAP: 1499 ctl_be_block_cw_dispatch_unmap(be_lun, io); 1500 break; 1501 default: 1502 panic("Unhandled CDB type %#x", io->scsiio.cdb[0]); 1503 break; 1504 } 1505} 1506 1507SDT_PROBE_DEFINE1(cbb, kernel, read, start, "uint64_t"); 1508SDT_PROBE_DEFINE1(cbb, kernel, write, start, "uint64_t"); 1509SDT_PROBE_DEFINE1(cbb, kernel, read, alloc_done, "uint64_t"); 1510SDT_PROBE_DEFINE1(cbb, kernel, write, alloc_done, "uint64_t"); 1511 1512static void 1513ctl_be_block_next(struct ctl_be_block_io *beio) 1514{ 1515 struct ctl_be_block_lun *be_lun; 1516 union ctl_io *io; 1517 1518 io = beio->io; 1519 be_lun = beio->lun; 1520 ctl_free_beio(beio); 1521 if ((io->io_hdr.flags & CTL_FLAG_ABORT) || 1522 ((io->io_hdr.status & CTL_STATUS_MASK) != CTL_STATUS_NONE && 1523 (io->io_hdr.status & CTL_STATUS_MASK) != CTL_SUCCESS)) { 1524 ctl_data_submit_done(io); 1525 return; 1526 } 1527 1528 io->io_hdr.status &= ~CTL_STATUS_MASK; 1529 io->io_hdr.status |= CTL_STATUS_NONE; 1530 1531 mtx_lock(&be_lun->queue_lock); 1532 STAILQ_INSERT_TAIL(&be_lun->input_queue, &io->io_hdr, links); 1533 mtx_unlock(&be_lun->queue_lock); 1534 taskqueue_enqueue(be_lun->io_taskqueue, &be_lun->io_task); 1535} 1536 1537static void 1538ctl_be_block_dispatch(struct ctl_be_block_lun *be_lun, 1539 union ctl_io *io) 1540{ 1541 struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun; 1542 struct ctl_be_block_io *beio; 1543 struct ctl_be_block_softc *softc; 1544 struct ctl_lba_len_flags *lbalen; 1545 struct ctl_ptr_len_flags *bptrlen; 1546 uint64_t len_left, lbas; 1547 int i; 1548 1549 softc = be_lun->softc; 1550 1551 DPRINTF("entered\n"); 1552 1553 lbalen = ARGS(io); 1554 if (lbalen->flags & CTL_LLF_WRITE) { 1555 SDT_PROBE(cbb, kernel, write, start, 0, 0, 0, 0, 0); 1556 } else { 1557 SDT_PROBE(cbb, kernel, read, start, 0, 0, 0, 0, 0); 1558 } 1559 1560 beio = ctl_alloc_beio(softc); 1561 beio->io = io; 1562 beio->lun = be_lun; 1563 bptrlen = PRIV(io); 1564 bptrlen->ptr = (void *)beio; 1565 1566 switch (io->scsiio.tag_type) { 1567 case CTL_TAG_ORDERED: 1568 beio->ds_tag_type = DEVSTAT_TAG_ORDERED; 1569 break; 1570 case CTL_TAG_HEAD_OF_QUEUE: 1571 beio->ds_tag_type = DEVSTAT_TAG_HEAD; 1572 break; 1573 case CTL_TAG_UNTAGGED: 1574 case CTL_TAG_SIMPLE: 1575 case CTL_TAG_ACA: 1576 default: 1577 beio->ds_tag_type = DEVSTAT_TAG_SIMPLE; 1578 break; 1579 } 1580 1581 if (lbalen->flags & CTL_LLF_WRITE) { 1582 beio->bio_cmd = BIO_WRITE; 1583 beio->ds_trans_type = DEVSTAT_WRITE; 1584 } else { 1585 beio->bio_cmd = BIO_READ; 1586 beio->ds_trans_type = DEVSTAT_READ; 1587 } 1588 1589 DPRINTF("%s at LBA %jx len %u @%ju\n", 1590 (beio->bio_cmd == BIO_READ) ? "READ" : "WRITE", 1591 (uintmax_t)lbalen->lba, lbalen->len, bptrlen->len); 1592 if (lbalen->flags & CTL_LLF_COMPARE) 1593 lbas = CTLBLK_HALF_IO_SIZE; 1594 else 1595 lbas = CTLBLK_MAX_IO_SIZE; 1596 lbas = MIN(lbalen->len - bptrlen->len, lbas / cbe_lun->blocksize); 1597 beio->io_offset = (lbalen->lba + bptrlen->len) * cbe_lun->blocksize; 1598 beio->io_len = lbas * cbe_lun->blocksize; 1599 bptrlen->len += lbas; 1600 1601 for (i = 0, len_left = beio->io_len; len_left > 0; i++) { 1602 KASSERT(i < CTLBLK_MAX_SEGS, ("Too many segs (%d >= %d)", 1603 i, CTLBLK_MAX_SEGS)); 1604 1605 /* 1606 * Setup the S/G entry for this chunk. 1607 */ 1608 beio->sg_segs[i].len = min(CTLBLK_MAX_SEG, len_left); 1609 beio->sg_segs[i].addr = uma_zalloc(be_lun->lun_zone, M_WAITOK); 1610 1611 DPRINTF("segment %d addr %p len %zd\n", i, 1612 beio->sg_segs[i].addr, beio->sg_segs[i].len); 1613 1614 /* Set up second segment for compare operation. */ 1615 if (lbalen->flags & CTL_LLF_COMPARE) { 1616 beio->sg_segs[i + CTLBLK_HALF_SEGS].len = 1617 beio->sg_segs[i].len; 1618 beio->sg_segs[i + CTLBLK_HALF_SEGS].addr = 1619 uma_zalloc(be_lun->lun_zone, M_WAITOK); 1620 } 1621 1622 beio->num_segs++; 1623 len_left -= beio->sg_segs[i].len; 1624 } 1625 if (bptrlen->len < lbalen->len) 1626 beio->beio_cont = ctl_be_block_next; 1627 io->scsiio.be_move_done = ctl_be_block_move_done; 1628 /* For compare we have separate S/G lists for read and datamove. */ 1629 if (lbalen->flags & CTL_LLF_COMPARE) 1630 io->scsiio.kern_data_ptr = (uint8_t *)&beio->sg_segs[CTLBLK_HALF_SEGS]; 1631 else 1632 io->scsiio.kern_data_ptr = (uint8_t *)beio->sg_segs; 1633 io->scsiio.kern_data_len = beio->io_len; 1634 io->scsiio.kern_data_resid = 0; 1635 io->scsiio.kern_sg_entries = beio->num_segs; 1636 io->io_hdr.flags |= CTL_FLAG_ALLOCATED; 1637 1638 /* 1639 * For the read case, we need to read the data into our buffers and 1640 * then we can send it back to the user. For the write case, we 1641 * need to get the data from the user first. 1642 */ 1643 if (beio->bio_cmd == BIO_READ) { 1644 SDT_PROBE(cbb, kernel, read, alloc_done, 0, 0, 0, 0, 0); 1645 be_lun->dispatch(be_lun, beio); 1646 } else { 1647 SDT_PROBE(cbb, kernel, write, alloc_done, 0, 0, 0, 0, 0); 1648#ifdef CTL_TIME_IO 1649 getbinuptime(&io->io_hdr.dma_start_bt); 1650#endif 1651 ctl_datamove(io); 1652 } 1653} 1654 1655static void 1656ctl_be_block_worker(void *context, int pending) 1657{ 1658 struct ctl_be_block_lun *be_lun = (struct ctl_be_block_lun *)context; 1659 struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun; 1660 union ctl_io *io; 1661 struct ctl_be_block_io *beio; 1662 1663 DPRINTF("entered\n"); 1664 /* 1665 * Fetch and process I/Os from all queues. If we detect LUN 1666 * CTL_LUN_FLAG_OFFLINE status here -- it is result of a race, 1667 * so make response maximally opaque to not confuse initiator. 1668 */ 1669 for (;;) { 1670 mtx_lock(&be_lun->queue_lock); 1671 io = (union ctl_io *)STAILQ_FIRST(&be_lun->datamove_queue); 1672 if (io != NULL) { 1673 DPRINTF("datamove queue\n"); 1674 STAILQ_REMOVE(&be_lun->datamove_queue, &io->io_hdr, 1675 ctl_io_hdr, links); 1676 mtx_unlock(&be_lun->queue_lock); 1677 beio = (struct ctl_be_block_io *)PRIV(io)->ptr; 1678 if (cbe_lun->flags & CTL_LUN_FLAG_OFFLINE) { 1679 ctl_set_busy(&io->scsiio); 1680 ctl_complete_beio(beio); 1681 return; 1682 } 1683 be_lun->dispatch(be_lun, beio); 1684 continue; 1685 } 1686 io = (union ctl_io *)STAILQ_FIRST(&be_lun->config_write_queue); 1687 if (io != NULL) { 1688 DPRINTF("config write queue\n"); 1689 STAILQ_REMOVE(&be_lun->config_write_queue, &io->io_hdr, 1690 ctl_io_hdr, links); 1691 mtx_unlock(&be_lun->queue_lock); 1692 if (cbe_lun->flags & CTL_LUN_FLAG_OFFLINE) { 1693 ctl_set_busy(&io->scsiio); 1694 ctl_config_write_done(io); 1695 return; 1696 } 1697 ctl_be_block_cw_dispatch(be_lun, io); 1698 continue; 1699 } 1700 io = (union ctl_io *)STAILQ_FIRST(&be_lun->config_read_queue); 1701 if (io != NULL) { 1702 DPRINTF("config read queue\n"); 1703 STAILQ_REMOVE(&be_lun->config_read_queue, &io->io_hdr, 1704 ctl_io_hdr, links); 1705 mtx_unlock(&be_lun->queue_lock); 1706 if (cbe_lun->flags & CTL_LUN_FLAG_OFFLINE) { 1707 ctl_set_busy(&io->scsiio); 1708 ctl_config_read_done(io); 1709 return; 1710 } 1711 ctl_be_block_cr_dispatch(be_lun, io); 1712 continue; 1713 } 1714 io = (union ctl_io *)STAILQ_FIRST(&be_lun->input_queue); 1715 if (io != NULL) { 1716 DPRINTF("input queue\n"); 1717 STAILQ_REMOVE(&be_lun->input_queue, &io->io_hdr, 1718 ctl_io_hdr, links); 1719 mtx_unlock(&be_lun->queue_lock); 1720 if (cbe_lun->flags & CTL_LUN_FLAG_OFFLINE) { 1721 ctl_set_busy(&io->scsiio); 1722 ctl_data_submit_done(io); 1723 return; 1724 } 1725 ctl_be_block_dispatch(be_lun, io); 1726 continue; 1727 } 1728 1729 /* 1730 * If we get here, there is no work left in the queues, so 1731 * just break out and let the task queue go to sleep. 1732 */ 1733 mtx_unlock(&be_lun->queue_lock); 1734 break; 1735 } 1736} 1737 1738/* 1739 * Entry point from CTL to the backend for I/O. We queue everything to a 1740 * work thread, so this just puts the I/O on a queue and wakes up the 1741 * thread. 1742 */ 1743static int 1744ctl_be_block_submit(union ctl_io *io) 1745{ 1746 struct ctl_be_block_lun *be_lun; 1747 struct ctl_be_lun *cbe_lun; 1748 1749 DPRINTF("entered\n"); 1750 1751 cbe_lun = (struct ctl_be_lun *)io->io_hdr.ctl_private[ 1752 CTL_PRIV_BACKEND_LUN].ptr; 1753 be_lun = (struct ctl_be_block_lun *)cbe_lun->be_lun; 1754 1755 /* 1756 * Make sure we only get SCSI I/O. 1757 */ 1758 KASSERT(io->io_hdr.io_type == CTL_IO_SCSI, ("Non-SCSI I/O (type " 1759 "%#x) encountered", io->io_hdr.io_type)); 1760 1761 PRIV(io)->len = 0; 1762 1763 mtx_lock(&be_lun->queue_lock); 1764 STAILQ_INSERT_TAIL(&be_lun->input_queue, &io->io_hdr, links); 1765 mtx_unlock(&be_lun->queue_lock); 1766 taskqueue_enqueue(be_lun->io_taskqueue, &be_lun->io_task); 1767 1768 return (CTL_RETVAL_COMPLETE); 1769} 1770 1771static int 1772ctl_be_block_ioctl(struct cdev *dev, u_long cmd, caddr_t addr, 1773 int flag, struct thread *td) 1774{ 1775 struct ctl_be_block_softc *softc; 1776 int error; 1777 1778 softc = &backend_block_softc; 1779 1780 error = 0; 1781 1782 switch (cmd) { 1783 case CTL_LUN_REQ: { 1784 struct ctl_lun_req *lun_req; 1785 1786 lun_req = (struct ctl_lun_req *)addr; 1787 1788 switch (lun_req->reqtype) { 1789 case CTL_LUNREQ_CREATE: 1790 error = ctl_be_block_create(softc, lun_req); 1791 break; 1792 case CTL_LUNREQ_RM: 1793 error = ctl_be_block_rm(softc, lun_req); 1794 break; 1795 case CTL_LUNREQ_MODIFY: 1796 error = ctl_be_block_modify(softc, lun_req); 1797 break; 1798 default: 1799 lun_req->status = CTL_LUN_ERROR; 1800 snprintf(lun_req->error_str, sizeof(lun_req->error_str), 1801 "invalid LUN request type %d", 1802 lun_req->reqtype); 1803 break; 1804 } 1805 break; 1806 } 1807 default: 1808 error = ENOTTY; 1809 break; 1810 } 1811 1812 return (error); 1813} 1814 1815static int 1816ctl_be_block_open_file(struct ctl_be_block_lun *be_lun, struct ctl_lun_req *req) 1817{ 1818 struct ctl_be_lun *cbe_lun; 1819 struct ctl_be_block_filedata *file_data; 1820 struct ctl_lun_create_params *params; 1821 char *value; 1822 struct vattr vattr; 1823 off_t ps, pss, po, pos, us, uss, uo, uos; 1824 int error; 1825 1826 cbe_lun = &be_lun->cbe_lun; 1827 file_data = &be_lun->backend.file; 1828 params = &be_lun->params; 1829 1830 be_lun->dev_type = CTL_BE_BLOCK_FILE; 1831 be_lun->dispatch = ctl_be_block_dispatch_file; 1832 be_lun->lun_flush = ctl_be_block_flush_file; 1833 be_lun->get_lba_status = ctl_be_block_gls_file; 1834 be_lun->getattr = ctl_be_block_getattr_file; 1835 be_lun->unmap = NULL; 1836 cbe_lun->flags &= ~CTL_LUN_FLAG_UNMAP; 1837 1838 error = VOP_GETATTR(be_lun->vn, &vattr, curthread->td_ucred); 1839 if (error != 0) { 1840 snprintf(req->error_str, sizeof(req->error_str), 1841 "error calling VOP_GETATTR() for file %s", 1842 be_lun->dev_path); 1843 return (error); 1844 } 1845 1846 /* 1847 * Verify that we have the ability to upgrade to exclusive 1848 * access on this file so we can trap errors at open instead 1849 * of reporting them during first access. 1850 */ 1851 if (VOP_ISLOCKED(be_lun->vn) != LK_EXCLUSIVE) { 1852 vn_lock(be_lun->vn, LK_UPGRADE | LK_RETRY); 1853 if (be_lun->vn->v_iflag & VI_DOOMED) { 1854 error = EBADF; 1855 snprintf(req->error_str, sizeof(req->error_str), 1856 "error locking file %s", be_lun->dev_path); 1857 return (error); 1858 } 1859 } 1860 1861 file_data->cred = crhold(curthread->td_ucred); 1862 if (params->lun_size_bytes != 0) 1863 be_lun->size_bytes = params->lun_size_bytes; 1864 else 1865 be_lun->size_bytes = vattr.va_size; 1866 1867 /* 1868 * For files we can use any logical block size. Prefer 512 bytes 1869 * for compatibility reasons. If file's vattr.va_blocksize 1870 * (preferred I/O block size) is bigger and multiple to chosen 1871 * logical block size -- report it as physical block size. 1872 */ 1873 if (params->blocksize_bytes != 0) 1874 cbe_lun->blocksize = params->blocksize_bytes; 1875 else 1876 cbe_lun->blocksize = 512; 1877 be_lun->size_blocks = be_lun->size_bytes / cbe_lun->blocksize; 1878 cbe_lun->maxlba = (be_lun->size_blocks == 0) ? 1879 0 : (be_lun->size_blocks - 1); 1880 1881 us = ps = vattr.va_blocksize; 1882 uo = po = 0; 1883 1884 value = ctl_get_opt(&cbe_lun->options, "pblocksize"); 1885 if (value != NULL) 1886 ctl_expand_number(value, &ps); 1887 value = ctl_get_opt(&cbe_lun->options, "pblockoffset"); 1888 if (value != NULL) 1889 ctl_expand_number(value, &po); 1890 pss = ps / cbe_lun->blocksize; 1891 pos = po / cbe_lun->blocksize; 1892 if ((pss > 0) && (pss * cbe_lun->blocksize == ps) && (pss >= pos) && 1893 ((pss & (pss - 1)) == 0) && (pos * cbe_lun->blocksize == po)) { 1894 cbe_lun->pblockexp = fls(pss) - 1; 1895 cbe_lun->pblockoff = (pss - pos) % pss; 1896 } 1897 1898 value = ctl_get_opt(&cbe_lun->options, "ublocksize"); 1899 if (value != NULL) 1900 ctl_expand_number(value, &us); 1901 value = ctl_get_opt(&cbe_lun->options, "ublockoffset"); 1902 if (value != NULL) 1903 ctl_expand_number(value, &uo); 1904 uss = us / cbe_lun->blocksize; 1905 uos = uo / cbe_lun->blocksize; 1906 if ((uss > 0) && (uss * cbe_lun->blocksize == us) && (uss >= uos) && 1907 ((uss & (uss - 1)) == 0) && (uos * cbe_lun->blocksize == uo)) { 1908 cbe_lun->ublockexp = fls(uss) - 1; 1909 cbe_lun->ublockoff = (uss - uos) % uss; 1910 } 1911 1912 /* 1913 * Sanity check. The media size has to be at least one 1914 * sector long. 1915 */ 1916 if (be_lun->size_bytes < cbe_lun->blocksize) { 1917 error = EINVAL; 1918 snprintf(req->error_str, sizeof(req->error_str), 1919 "file %s size %ju < block size %u", be_lun->dev_path, 1920 (uintmax_t)be_lun->size_bytes, cbe_lun->blocksize); 1921 } 1922 1923 cbe_lun->opttxferlen = CTLBLK_MAX_IO_SIZE / cbe_lun->blocksize; 1924 return (error); 1925} 1926 1927static int 1928ctl_be_block_open_dev(struct ctl_be_block_lun *be_lun, struct ctl_lun_req *req) 1929{ 1930 struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun; 1931 struct ctl_lun_create_params *params; 1932 struct cdevsw *csw; 1933 struct cdev *dev; 1934 char *value; 1935 int error, atomic, maxio, ref, unmap, tmp; 1936 off_t ps, pss, po, pos, us, uss, uo, uos, otmp; 1937 1938 params = &be_lun->params; 1939 1940 be_lun->dev_type = CTL_BE_BLOCK_DEV; 1941 csw = devvn_refthread(be_lun->vn, &dev, &ref); 1942 if (csw == NULL) 1943 return (ENXIO); 1944 if (strcmp(csw->d_name, "zvol") == 0) { 1945 be_lun->dispatch = ctl_be_block_dispatch_zvol; 1946 be_lun->get_lba_status = ctl_be_block_gls_zvol; 1947 atomic = maxio = CTLBLK_MAX_IO_SIZE; 1948 } else { 1949 be_lun->dispatch = ctl_be_block_dispatch_dev; 1950 be_lun->get_lba_status = NULL; 1951 atomic = 0; 1952 maxio = dev->si_iosize_max; 1953 if (maxio <= 0) 1954 maxio = DFLTPHYS; 1955 if (maxio > CTLBLK_MAX_IO_SIZE) 1956 maxio = CTLBLK_MAX_IO_SIZE; 1957 } 1958 be_lun->lun_flush = ctl_be_block_flush_dev; 1959 be_lun->getattr = ctl_be_block_getattr_dev; 1960 be_lun->unmap = ctl_be_block_unmap_dev; 1961 1962 if (!csw->d_ioctl) { 1963 dev_relthread(dev, ref); 1964 snprintf(req->error_str, sizeof(req->error_str), 1965 "no d_ioctl for device %s!", be_lun->dev_path); 1966 return (ENODEV); 1967 } 1968 1969 error = csw->d_ioctl(dev, DIOCGSECTORSIZE, (caddr_t)&tmp, FREAD, 1970 curthread); 1971 if (error) { 1972 dev_relthread(dev, ref); 1973 snprintf(req->error_str, sizeof(req->error_str), 1974 "error %d returned for DIOCGSECTORSIZE ioctl " 1975 "on %s!", error, be_lun->dev_path); 1976 return (error); 1977 } 1978 1979 /* 1980 * If the user has asked for a blocksize that is greater than the 1981 * backing device's blocksize, we can do it only if the blocksize 1982 * the user is asking for is an even multiple of the underlying 1983 * device's blocksize. 1984 */ 1985 if ((params->blocksize_bytes != 0) && 1986 (params->blocksize_bytes >= tmp)) { 1987 if (params->blocksize_bytes % tmp == 0) { 1988 cbe_lun->blocksize = params->blocksize_bytes; 1989 } else { 1990 dev_relthread(dev, ref); 1991 snprintf(req->error_str, sizeof(req->error_str), 1992 "requested blocksize %u is not an even " 1993 "multiple of backing device blocksize %u", 1994 params->blocksize_bytes, tmp); 1995 return (EINVAL); 1996 } 1997 } else if (params->blocksize_bytes != 0) { 1998 dev_relthread(dev, ref); 1999 snprintf(req->error_str, sizeof(req->error_str), 2000 "requested blocksize %u < backing device " 2001 "blocksize %u", params->blocksize_bytes, tmp); 2002 return (EINVAL); 2003 } else 2004 cbe_lun->blocksize = tmp; 2005 2006 error = csw->d_ioctl(dev, DIOCGMEDIASIZE, (caddr_t)&otmp, FREAD, 2007 curthread); 2008 if (error) { 2009 dev_relthread(dev, ref); 2010 snprintf(req->error_str, sizeof(req->error_str), 2011 "error %d returned for DIOCGMEDIASIZE " 2012 " ioctl on %s!", error, 2013 be_lun->dev_path); 2014 return (error); 2015 } 2016 2017 if (params->lun_size_bytes != 0) { 2018 if (params->lun_size_bytes > otmp) { 2019 dev_relthread(dev, ref); 2020 snprintf(req->error_str, sizeof(req->error_str), 2021 "requested LUN size %ju > backing device " 2022 "size %ju", 2023 (uintmax_t)params->lun_size_bytes, 2024 (uintmax_t)otmp); 2025 return (EINVAL); 2026 } 2027 2028 be_lun->size_bytes = params->lun_size_bytes; 2029 } else 2030 be_lun->size_bytes = otmp; 2031 be_lun->size_blocks = be_lun->size_bytes / cbe_lun->blocksize; 2032 cbe_lun->maxlba = (be_lun->size_blocks == 0) ? 2033 0 : (be_lun->size_blocks - 1); 2034 2035 error = csw->d_ioctl(dev, DIOCGSTRIPESIZE, (caddr_t)&ps, FREAD, 2036 curthread); 2037 if (error) 2038 ps = po = 0; 2039 else { 2040 error = csw->d_ioctl(dev, DIOCGSTRIPEOFFSET, (caddr_t)&po, 2041 FREAD, curthread); 2042 if (error) 2043 po = 0; 2044 } 2045 us = ps; 2046 uo = po; 2047 2048 value = ctl_get_opt(&cbe_lun->options, "pblocksize"); 2049 if (value != NULL) 2050 ctl_expand_number(value, &ps); 2051 value = ctl_get_opt(&cbe_lun->options, "pblockoffset"); 2052 if (value != NULL) 2053 ctl_expand_number(value, &po); 2054 pss = ps / cbe_lun->blocksize; 2055 pos = po / cbe_lun->blocksize; 2056 if ((pss > 0) && (pss * cbe_lun->blocksize == ps) && (pss >= pos) && 2057 ((pss & (pss - 1)) == 0) && (pos * cbe_lun->blocksize == po)) { 2058 cbe_lun->pblockexp = fls(pss) - 1; 2059 cbe_lun->pblockoff = (pss - pos) % pss; 2060 } 2061 2062 value = ctl_get_opt(&cbe_lun->options, "ublocksize"); 2063 if (value != NULL) 2064 ctl_expand_number(value, &us); 2065 value = ctl_get_opt(&cbe_lun->options, "ublockoffset"); 2066 if (value != NULL) 2067 ctl_expand_number(value, &uo); 2068 uss = us / cbe_lun->blocksize; 2069 uos = uo / cbe_lun->blocksize; 2070 if ((uss > 0) && (uss * cbe_lun->blocksize == us) && (uss >= uos) && 2071 ((uss & (uss - 1)) == 0) && (uos * cbe_lun->blocksize == uo)) { 2072 cbe_lun->ublockexp = fls(uss) - 1; 2073 cbe_lun->ublockoff = (uss - uos) % uss; 2074 } 2075 2076 cbe_lun->atomicblock = atomic / cbe_lun->blocksize; 2077 cbe_lun->opttxferlen = maxio / cbe_lun->blocksize; 2078 2079 if (be_lun->dispatch == ctl_be_block_dispatch_zvol) { 2080 unmap = 1; 2081 } else { 2082 struct diocgattr_arg arg; 2083 2084 strlcpy(arg.name, "GEOM::candelete", sizeof(arg.name)); 2085 arg.len = sizeof(arg.value.i); 2086 error = csw->d_ioctl(dev, DIOCGATTR, (caddr_t)&arg, FREAD, 2087 curthread); 2088 unmap = (error == 0) ? arg.value.i : 0; 2089 } 2090 value = ctl_get_opt(&cbe_lun->options, "unmap"); 2091 if (value != NULL) 2092 unmap = (strcmp(value, "on") == 0); 2093 if (unmap) 2094 cbe_lun->flags |= CTL_LUN_FLAG_UNMAP; 2095 else 2096 cbe_lun->flags &= ~CTL_LUN_FLAG_UNMAP; 2097 2098 dev_relthread(dev, ref); 2099 return (0); 2100} 2101 2102static int 2103ctl_be_block_close(struct ctl_be_block_lun *be_lun) 2104{ 2105 struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun; 2106 int flags; 2107 2108 if (be_lun->vn) { 2109 flags = FREAD; 2110 if ((cbe_lun->flags & CTL_LUN_FLAG_READONLY) == 0) 2111 flags |= FWRITE; 2112 (void)vn_close(be_lun->vn, flags, NOCRED, curthread); 2113 be_lun->vn = NULL; 2114 2115 switch (be_lun->dev_type) { 2116 case CTL_BE_BLOCK_DEV: 2117 break; 2118 case CTL_BE_BLOCK_FILE: 2119 if (be_lun->backend.file.cred != NULL) { 2120 crfree(be_lun->backend.file.cred); 2121 be_lun->backend.file.cred = NULL; 2122 } 2123 break; 2124 case CTL_BE_BLOCK_NONE: 2125 break; 2126 default: 2127 panic("Unexpected backend type."); 2128 break; 2129 } 2130 be_lun->dev_type = CTL_BE_BLOCK_NONE; 2131 } 2132 return (0); 2133} 2134 2135static int 2136ctl_be_block_open(struct ctl_be_block_softc *softc, 2137 struct ctl_be_block_lun *be_lun, struct ctl_lun_req *req) 2138{ 2139 struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun; 2140 struct nameidata nd; 2141 char *value; 2142 int error, flags; 2143 2144 error = 0; 2145 if (rootvnode == NULL) { 2146 snprintf(req->error_str, sizeof(req->error_str), 2147 "Root filesystem is not mounted"); 2148 return (1); 2149 } 2150 if (!curthread->td_proc->p_fd->fd_cdir) { 2151 curthread->td_proc->p_fd->fd_cdir = rootvnode; 2152 VREF(rootvnode); 2153 } 2154 if (!curthread->td_proc->p_fd->fd_rdir) { 2155 curthread->td_proc->p_fd->fd_rdir = rootvnode; 2156 VREF(rootvnode); 2157 } 2158 if (!curthread->td_proc->p_fd->fd_jdir) { 2159 curthread->td_proc->p_fd->fd_jdir = rootvnode; 2160 VREF(rootvnode); 2161 } 2162 2163 value = ctl_get_opt(&cbe_lun->options, "file"); 2164 if (value == NULL) { 2165 snprintf(req->error_str, sizeof(req->error_str), 2166 "no file argument specified"); 2167 return (1); 2168 } 2169 free(be_lun->dev_path, M_CTLBLK); 2170 be_lun->dev_path = strdup(value, M_CTLBLK); 2171 2172 flags = FREAD; 2173 value = ctl_get_opt(&cbe_lun->options, "readonly"); 2174 if (value == NULL || strcmp(value, "on") != 0) 2175 flags |= FWRITE; 2176 2177again: 2178 NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, be_lun->dev_path, curthread); 2179 error = vn_open(&nd, &flags, 0, NULL); 2180 if ((error == EROFS || error == EACCES) && (flags & FWRITE)) { 2181 flags &= ~FWRITE; 2182 goto again; 2183 } 2184 if (error) { 2185 /* 2186 * This is the only reasonable guess we can make as far as 2187 * path if the user doesn't give us a fully qualified path. 2188 * If they want to specify a file, they need to specify the 2189 * full path. 2190 */ 2191 if (be_lun->dev_path[0] != '/') { 2192 char *dev_name; 2193 2194 asprintf(&dev_name, M_CTLBLK, "/dev/%s", 2195 be_lun->dev_path); 2196 free(be_lun->dev_path, M_CTLBLK); 2197 be_lun->dev_path = dev_name; 2198 goto again; 2199 } 2200 snprintf(req->error_str, sizeof(req->error_str), 2201 "error opening %s: %d", be_lun->dev_path, error); 2202 return (error); 2203 } 2204 if (flags & FWRITE) 2205 cbe_lun->flags &= ~CTL_LUN_FLAG_READONLY; 2206 else 2207 cbe_lun->flags |= CTL_LUN_FLAG_READONLY; 2208 2209 NDFREE(&nd, NDF_ONLY_PNBUF); 2210 be_lun->vn = nd.ni_vp; 2211 2212 /* We only support disks and files. */ 2213 if (vn_isdisk(be_lun->vn, &error)) { 2214 error = ctl_be_block_open_dev(be_lun, req); 2215 } else if (be_lun->vn->v_type == VREG) { 2216 error = ctl_be_block_open_file(be_lun, req); 2217 } else { 2218 error = EINVAL; 2219 snprintf(req->error_str, sizeof(req->error_str), 2220 "%s is not a disk or plain file", be_lun->dev_path); 2221 } 2222 VOP_UNLOCK(be_lun->vn, 0); 2223 2224 if (error != 0) 2225 ctl_be_block_close(be_lun); 2226 cbe_lun->serseq = CTL_LUN_SERSEQ_OFF; 2227 if (be_lun->dispatch != ctl_be_block_dispatch_dev) 2228 cbe_lun->serseq = CTL_LUN_SERSEQ_READ; 2229 value = ctl_get_opt(&cbe_lun->options, "serseq"); 2230 if (value != NULL && strcmp(value, "on") == 0) 2231 cbe_lun->serseq = CTL_LUN_SERSEQ_ON; 2232 else if (value != NULL && strcmp(value, "read") == 0) 2233 cbe_lun->serseq = CTL_LUN_SERSEQ_READ; 2234 else if (value != NULL && strcmp(value, "off") == 0) 2235 cbe_lun->serseq = CTL_LUN_SERSEQ_OFF; 2236 return (0); 2237} 2238 2239static int 2240ctl_be_block_create(struct ctl_be_block_softc *softc, struct ctl_lun_req *req) 2241{ 2242 struct ctl_be_lun *cbe_lun; 2243 struct ctl_be_block_lun *be_lun; 2244 struct ctl_lun_create_params *params; 2245 char num_thread_str[16]; 2246 char tmpstr[32]; 2247 char *value; 2248 int retval, num_threads; 2249 int tmp_num_threads; 2250 2251 params = &req->reqdata.create; 2252 retval = 0; 2253 req->status = CTL_LUN_OK; 2254 2255 be_lun = malloc(sizeof(*be_lun), M_CTLBLK, M_ZERO | M_WAITOK); 2256 cbe_lun = &be_lun->cbe_lun; 2257 cbe_lun->be_lun = be_lun; 2258 be_lun->params = req->reqdata.create; 2259 be_lun->softc = softc; 2260 STAILQ_INIT(&be_lun->input_queue); 2261 STAILQ_INIT(&be_lun->config_read_queue); 2262 STAILQ_INIT(&be_lun->config_write_queue); 2263 STAILQ_INIT(&be_lun->datamove_queue); 2264 sprintf(be_lun->lunname, "cblk%d", softc->num_luns); 2265 mtx_init(&be_lun->io_lock, "cblk io lock", NULL, MTX_DEF); 2266 mtx_init(&be_lun->queue_lock, "cblk queue lock", NULL, MTX_DEF); 2267 ctl_init_opts(&cbe_lun->options, 2268 req->num_be_args, req->kern_be_args); 2269 be_lun->lun_zone = uma_zcreate(be_lun->lunname, CTLBLK_MAX_SEG, 2270 NULL, NULL, NULL, NULL, /*align*/ 0, /*flags*/0); 2271 if (be_lun->lun_zone == NULL) { 2272 snprintf(req->error_str, sizeof(req->error_str), 2273 "error allocating UMA zone"); 2274 goto bailout_error; 2275 } 2276 2277 if (params->flags & CTL_LUN_FLAG_DEV_TYPE) 2278 cbe_lun->lun_type = params->device_type; 2279 else 2280 cbe_lun->lun_type = T_DIRECT; 2281 be_lun->flags = CTL_BE_BLOCK_LUN_UNCONFIGURED; 2282 cbe_lun->flags = 0; 2283 value = ctl_get_opt(&cbe_lun->options, "ha_role"); 2284 if (value != NULL) { 2285 if (strcmp(value, "primary") == 0) 2286 cbe_lun->flags |= CTL_LUN_FLAG_PRIMARY; 2287 } else if (control_softc->flags & CTL_FLAG_ACTIVE_SHELF) 2288 cbe_lun->flags |= CTL_LUN_FLAG_PRIMARY; 2289 2290 if (cbe_lun->lun_type == T_DIRECT) { 2291 be_lun->size_bytes = params->lun_size_bytes; 2292 if (params->blocksize_bytes != 0) 2293 cbe_lun->blocksize = params->blocksize_bytes; 2294 else 2295 cbe_lun->blocksize = 512; 2296 be_lun->size_blocks = be_lun->size_bytes / cbe_lun->blocksize; 2297 cbe_lun->maxlba = (be_lun->size_blocks == 0) ? 2298 0 : (be_lun->size_blocks - 1); 2299 2300 if ((cbe_lun->flags & CTL_LUN_FLAG_PRIMARY) || 2301 control_softc->ha_mode == CTL_HA_MODE_SER_ONLY) { 2302 retval = ctl_be_block_open(softc, be_lun, req); 2303 if (retval != 0) { 2304 retval = 0; 2305 req->status = CTL_LUN_WARNING; 2306 } 2307 } 2308 num_threads = cbb_num_threads; 2309 } else { 2310 num_threads = 1; 2311 } 2312 2313 value = ctl_get_opt(&cbe_lun->options, "num_threads"); 2314 if (value != NULL) { 2315 tmp_num_threads = strtol(value, NULL, 0); 2316 2317 /* 2318 * We don't let the user specify less than one 2319 * thread, but hope he's clueful enough not to 2320 * specify 1000 threads. 2321 */ 2322 if (tmp_num_threads < 1) { 2323 snprintf(req->error_str, sizeof(req->error_str), 2324 "invalid number of threads %s", 2325 num_thread_str); 2326 goto bailout_error; 2327 } 2328 num_threads = tmp_num_threads; 2329 } 2330 2331 if (be_lun->vn == NULL) 2332 cbe_lun->flags |= CTL_LUN_FLAG_OFFLINE; 2333 /* Tell the user the blocksize we ended up using */ 2334 params->lun_size_bytes = be_lun->size_bytes; 2335 params->blocksize_bytes = cbe_lun->blocksize; 2336 if (params->flags & CTL_LUN_FLAG_ID_REQ) { 2337 cbe_lun->req_lun_id = params->req_lun_id; 2338 cbe_lun->flags |= CTL_LUN_FLAG_ID_REQ; 2339 } else 2340 cbe_lun->req_lun_id = 0; 2341 2342 cbe_lun->lun_shutdown = ctl_be_block_lun_shutdown; 2343 cbe_lun->lun_config_status = ctl_be_block_lun_config_status; 2344 cbe_lun->be = &ctl_be_block_driver; 2345 2346 if ((params->flags & CTL_LUN_FLAG_SERIAL_NUM) == 0) { 2347 snprintf(tmpstr, sizeof(tmpstr), "MYSERIAL%4d", 2348 softc->num_luns); 2349 strncpy((char *)cbe_lun->serial_num, tmpstr, 2350 MIN(sizeof(cbe_lun->serial_num), sizeof(tmpstr))); 2351 2352 /* Tell the user what we used for a serial number */ 2353 strncpy((char *)params->serial_num, tmpstr, 2354 MIN(sizeof(params->serial_num), sizeof(tmpstr))); 2355 } else { 2356 strncpy((char *)cbe_lun->serial_num, params->serial_num, 2357 MIN(sizeof(cbe_lun->serial_num), 2358 sizeof(params->serial_num))); 2359 } 2360 if ((params->flags & CTL_LUN_FLAG_DEVID) == 0) { 2361 snprintf(tmpstr, sizeof(tmpstr), "MYDEVID%4d", softc->num_luns); 2362 strncpy((char *)cbe_lun->device_id, tmpstr, 2363 MIN(sizeof(cbe_lun->device_id), sizeof(tmpstr))); 2364 2365 /* Tell the user what we used for a device ID */ 2366 strncpy((char *)params->device_id, tmpstr, 2367 MIN(sizeof(params->device_id), sizeof(tmpstr))); 2368 } else { 2369 strncpy((char *)cbe_lun->device_id, params->device_id, 2370 MIN(sizeof(cbe_lun->device_id), 2371 sizeof(params->device_id))); 2372 } 2373 2374 TASK_INIT(&be_lun->io_task, /*priority*/0, ctl_be_block_worker, be_lun); 2375 2376 be_lun->io_taskqueue = taskqueue_create(be_lun->lunname, M_WAITOK, 2377 taskqueue_thread_enqueue, /*context*/&be_lun->io_taskqueue); 2378 2379 if (be_lun->io_taskqueue == NULL) { 2380 snprintf(req->error_str, sizeof(req->error_str), 2381 "unable to create taskqueue"); 2382 goto bailout_error; 2383 } 2384 2385 /* 2386 * Note that we start the same number of threads by default for 2387 * both the file case and the block device case. For the file 2388 * case, we need multiple threads to allow concurrency, because the 2389 * vnode interface is designed to be a blocking interface. For the 2390 * block device case, ZFS zvols at least will block the caller's 2391 * context in many instances, and so we need multiple threads to 2392 * overcome that problem. Other block devices don't need as many 2393 * threads, but they shouldn't cause too many problems. 2394 * 2395 * If the user wants to just have a single thread for a block 2396 * device, he can specify that when the LUN is created, or change 2397 * the tunable/sysctl to alter the default number of threads. 2398 */ 2399 retval = taskqueue_start_threads(&be_lun->io_taskqueue, 2400 /*num threads*/num_threads, 2401 /*priority*/PWAIT, 2402 /*thread name*/ 2403 "%s taskq", be_lun->lunname); 2404 2405 if (retval != 0) 2406 goto bailout_error; 2407 2408 be_lun->num_threads = num_threads; 2409 2410 mtx_lock(&softc->lock); 2411 softc->num_luns++; 2412 STAILQ_INSERT_TAIL(&softc->lun_list, be_lun, links); 2413 2414 mtx_unlock(&softc->lock); 2415 2416 retval = ctl_add_lun(&be_lun->cbe_lun); 2417 if (retval != 0) { 2418 mtx_lock(&softc->lock); 2419 STAILQ_REMOVE(&softc->lun_list, be_lun, ctl_be_block_lun, 2420 links); 2421 softc->num_luns--; 2422 mtx_unlock(&softc->lock); 2423 snprintf(req->error_str, sizeof(req->error_str), 2424 "ctl_add_lun() returned error %d, see dmesg for " 2425 "details", retval); 2426 retval = 0; 2427 goto bailout_error; 2428 } 2429 2430 mtx_lock(&softc->lock); 2431 2432 /* 2433 * Tell the config_status routine that we're waiting so it won't 2434 * clean up the LUN in the event of an error. 2435 */ 2436 be_lun->flags |= CTL_BE_BLOCK_LUN_WAITING; 2437 2438 while (be_lun->flags & CTL_BE_BLOCK_LUN_UNCONFIGURED) { 2439 retval = msleep(be_lun, &softc->lock, PCATCH, "ctlblk", 0); 2440 if (retval == EINTR) 2441 break; 2442 } 2443 be_lun->flags &= ~CTL_BE_BLOCK_LUN_WAITING; 2444 2445 if (be_lun->flags & CTL_BE_BLOCK_LUN_CONFIG_ERR) { 2446 snprintf(req->error_str, sizeof(req->error_str), 2447 "LUN configuration error, see dmesg for details"); 2448 STAILQ_REMOVE(&softc->lun_list, be_lun, ctl_be_block_lun, 2449 links); 2450 softc->num_luns--; 2451 mtx_unlock(&softc->lock); 2452 goto bailout_error; 2453 } else { 2454 params->req_lun_id = cbe_lun->lun_id; 2455 } 2456 2457 mtx_unlock(&softc->lock); 2458 2459 be_lun->disk_stats = devstat_new_entry("cbb", params->req_lun_id, 2460 cbe_lun->blocksize, 2461 DEVSTAT_ALL_SUPPORTED, 2462 cbe_lun->lun_type 2463 | DEVSTAT_TYPE_IF_OTHER, 2464 DEVSTAT_PRIORITY_OTHER); 2465 2466 return (retval); 2467 2468bailout_error: 2469 req->status = CTL_LUN_ERROR; 2470 2471 if (be_lun->io_taskqueue != NULL) 2472 taskqueue_free(be_lun->io_taskqueue); 2473 ctl_be_block_close(be_lun); 2474 if (be_lun->dev_path != NULL) 2475 free(be_lun->dev_path, M_CTLBLK); 2476 if (be_lun->lun_zone != NULL) 2477 uma_zdestroy(be_lun->lun_zone); 2478 ctl_free_opts(&cbe_lun->options); 2479 mtx_destroy(&be_lun->queue_lock); 2480 mtx_destroy(&be_lun->io_lock); 2481 free(be_lun, M_CTLBLK); 2482 2483 return (retval); 2484} 2485 2486static int 2487ctl_be_block_rm(struct ctl_be_block_softc *softc, struct ctl_lun_req *req) 2488{ 2489 struct ctl_lun_rm_params *params; 2490 struct ctl_be_block_lun *be_lun; 2491 struct ctl_be_lun *cbe_lun; 2492 int retval; 2493 2494 params = &req->reqdata.rm; 2495 2496 mtx_lock(&softc->lock); 2497 STAILQ_FOREACH(be_lun, &softc->lun_list, links) { 2498 if (be_lun->cbe_lun.lun_id == params->lun_id) 2499 break; 2500 } 2501 mtx_unlock(&softc->lock); 2502 if (be_lun == NULL) { 2503 snprintf(req->error_str, sizeof(req->error_str), 2504 "LUN %u is not managed by the block backend", 2505 params->lun_id); 2506 goto bailout_error; 2507 } 2508 cbe_lun = &be_lun->cbe_lun; 2509 2510 retval = ctl_disable_lun(cbe_lun); 2511 if (retval != 0) { 2512 snprintf(req->error_str, sizeof(req->error_str), 2513 "error %d returned from ctl_disable_lun() for " 2514 "LUN %d", retval, params->lun_id); 2515 goto bailout_error; 2516 } 2517 2518 if (be_lun->vn != NULL) { 2519 cbe_lun->flags |= CTL_LUN_FLAG_OFFLINE; 2520 ctl_lun_offline(cbe_lun); 2521 taskqueue_drain_all(be_lun->io_taskqueue); 2522 ctl_be_block_close(be_lun); 2523 } 2524 2525 retval = ctl_invalidate_lun(cbe_lun); 2526 if (retval != 0) { 2527 snprintf(req->error_str, sizeof(req->error_str), 2528 "error %d returned from ctl_invalidate_lun() for " 2529 "LUN %d", retval, params->lun_id); 2530 goto bailout_error; 2531 } 2532 2533 mtx_lock(&softc->lock); 2534 be_lun->flags |= CTL_BE_BLOCK_LUN_WAITING; 2535 while ((be_lun->flags & CTL_BE_BLOCK_LUN_UNCONFIGURED) == 0) { 2536 retval = msleep(be_lun, &softc->lock, PCATCH, "ctlblk", 0); 2537 if (retval == EINTR) 2538 break; 2539 } 2540 be_lun->flags &= ~CTL_BE_BLOCK_LUN_WAITING; 2541 2542 if ((be_lun->flags & CTL_BE_BLOCK_LUN_UNCONFIGURED) == 0) { 2543 snprintf(req->error_str, sizeof(req->error_str), 2544 "interrupted waiting for LUN to be freed"); 2545 mtx_unlock(&softc->lock); 2546 goto bailout_error; 2547 } 2548 2549 STAILQ_REMOVE(&softc->lun_list, be_lun, ctl_be_block_lun, links); 2550 2551 softc->num_luns--; 2552 mtx_unlock(&softc->lock); 2553 2554 taskqueue_drain_all(be_lun->io_taskqueue); 2555 taskqueue_free(be_lun->io_taskqueue); 2556 2557 if (be_lun->disk_stats != NULL) 2558 devstat_remove_entry(be_lun->disk_stats); 2559 2560 uma_zdestroy(be_lun->lun_zone); 2561 2562 ctl_free_opts(&cbe_lun->options); 2563 free(be_lun->dev_path, M_CTLBLK); 2564 mtx_destroy(&be_lun->queue_lock); 2565 mtx_destroy(&be_lun->io_lock); 2566 free(be_lun, M_CTLBLK); 2567 2568 req->status = CTL_LUN_OK; 2569 return (0); 2570 2571bailout_error: 2572 req->status = CTL_LUN_ERROR; 2573 return (0); 2574} 2575 2576static int 2577ctl_be_block_modify(struct ctl_be_block_softc *softc, struct ctl_lun_req *req) 2578{ 2579 struct ctl_lun_modify_params *params; 2580 struct ctl_be_block_lun *be_lun; 2581 struct ctl_be_lun *cbe_lun; 2582 char *value; 2583 uint64_t oldsize; 2584 int error, wasprim; 2585 2586 params = &req->reqdata.modify; 2587 2588 mtx_lock(&softc->lock); 2589 STAILQ_FOREACH(be_lun, &softc->lun_list, links) { 2590 if (be_lun->cbe_lun.lun_id == params->lun_id) 2591 break; 2592 } 2593 mtx_unlock(&softc->lock); 2594 if (be_lun == NULL) { 2595 snprintf(req->error_str, sizeof(req->error_str), 2596 "LUN %u is not managed by the block backend", 2597 params->lun_id); 2598 goto bailout_error; 2599 } 2600 cbe_lun = &be_lun->cbe_lun; 2601 2602 if (params->lun_size_bytes != 0) 2603 be_lun->params.lun_size_bytes = params->lun_size_bytes; 2604 ctl_update_opts(&cbe_lun->options, req->num_be_args, req->kern_be_args); 2605 2606 wasprim = (cbe_lun->flags & CTL_LUN_FLAG_PRIMARY); 2607 value = ctl_get_opt(&cbe_lun->options, "ha_role"); 2608 if (value != NULL) { 2609 if (strcmp(value, "primary") == 0) 2610 cbe_lun->flags |= CTL_LUN_FLAG_PRIMARY; 2611 else 2612 cbe_lun->flags &= ~CTL_LUN_FLAG_PRIMARY; 2613 } else if (control_softc->flags & CTL_FLAG_ACTIVE_SHELF) 2614 cbe_lun->flags |= CTL_LUN_FLAG_PRIMARY; 2615 else 2616 cbe_lun->flags &= ~CTL_LUN_FLAG_PRIMARY; 2617 if (wasprim != (cbe_lun->flags & CTL_LUN_FLAG_PRIMARY)) { 2618 if (cbe_lun->flags & CTL_LUN_FLAG_PRIMARY) 2619 ctl_lun_primary(cbe_lun); 2620 else 2621 ctl_lun_secondary(cbe_lun); 2622 } 2623 2624 oldsize = be_lun->size_blocks; 2625 if ((cbe_lun->flags & CTL_LUN_FLAG_PRIMARY) || 2626 control_softc->ha_mode == CTL_HA_MODE_SER_ONLY) { 2627 if (be_lun->vn == NULL) 2628 error = ctl_be_block_open(softc, be_lun, req); 2629 else if (vn_isdisk(be_lun->vn, &error)) 2630 error = ctl_be_block_open_dev(be_lun, req); 2631 else if (be_lun->vn->v_type == VREG) 2632 error = ctl_be_block_open_file(be_lun, req); 2633 else 2634 error = EINVAL; 2635 if ((cbe_lun->flags & CTL_LUN_FLAG_OFFLINE) && 2636 be_lun->vn != NULL) { 2637 cbe_lun->flags &= ~CTL_LUN_FLAG_OFFLINE; 2638 ctl_lun_online(cbe_lun); 2639 } 2640 } else { 2641 if (be_lun->vn != NULL) { 2642 cbe_lun->flags |= CTL_LUN_FLAG_OFFLINE; 2643 ctl_lun_offline(cbe_lun); 2644 taskqueue_drain_all(be_lun->io_taskqueue); 2645 error = ctl_be_block_close(be_lun); 2646 } else 2647 error = 0; 2648 } 2649 if (be_lun->size_blocks != oldsize) 2650 ctl_lun_capacity_changed(cbe_lun); 2651 2652 /* Tell the user the exact size we ended up using */ 2653 params->lun_size_bytes = be_lun->size_bytes; 2654 2655 req->status = error ? CTL_LUN_WARNING : CTL_LUN_OK; 2656 return (0); 2657 2658bailout_error: 2659 req->status = CTL_LUN_ERROR; 2660 return (0); 2661} 2662 2663static void 2664ctl_be_block_lun_shutdown(void *be_lun) 2665{ 2666 struct ctl_be_block_lun *lun; 2667 struct ctl_be_block_softc *softc; 2668 2669 lun = (struct ctl_be_block_lun *)be_lun; 2670 softc = lun->softc; 2671 2672 mtx_lock(&softc->lock); 2673 lun->flags |= CTL_BE_BLOCK_LUN_UNCONFIGURED; 2674 if (lun->flags & CTL_BE_BLOCK_LUN_WAITING) 2675 wakeup(lun); 2676 mtx_unlock(&softc->lock); 2677} 2678 2679static void 2680ctl_be_block_lun_config_status(void *be_lun, ctl_lun_config_status status) 2681{ 2682 struct ctl_be_block_lun *lun; 2683 struct ctl_be_block_softc *softc; 2684 2685 lun = (struct ctl_be_block_lun *)be_lun; 2686 softc = lun->softc; 2687 2688 if (status == CTL_LUN_CONFIG_OK) { 2689 mtx_lock(&softc->lock); 2690 lun->flags &= ~CTL_BE_BLOCK_LUN_UNCONFIGURED; 2691 if (lun->flags & CTL_BE_BLOCK_LUN_WAITING) 2692 wakeup(lun); 2693 mtx_unlock(&softc->lock); 2694 2695 /* 2696 * We successfully added the LUN, attempt to enable it. 2697 */ 2698 if (ctl_enable_lun(&lun->cbe_lun) != 0) { 2699 printf("%s: ctl_enable_lun() failed!\n", __func__); 2700 if (ctl_invalidate_lun(&lun->cbe_lun) != 0) { 2701 printf("%s: ctl_invalidate_lun() failed!\n", 2702 __func__); 2703 } 2704 } 2705 2706 return; 2707 } 2708 2709 2710 mtx_lock(&softc->lock); 2711 lun->flags &= ~CTL_BE_BLOCK_LUN_UNCONFIGURED; 2712 lun->flags |= CTL_BE_BLOCK_LUN_CONFIG_ERR; 2713 wakeup(lun); 2714 mtx_unlock(&softc->lock); 2715} 2716 2717 2718static int 2719ctl_be_block_config_write(union ctl_io *io) 2720{ 2721 struct ctl_be_block_lun *be_lun; 2722 struct ctl_be_lun *cbe_lun; 2723 int retval; 2724 2725 DPRINTF("entered\n"); 2726 2727 cbe_lun = (struct ctl_be_lun *)io->io_hdr.ctl_private[ 2728 CTL_PRIV_BACKEND_LUN].ptr; 2729 be_lun = (struct ctl_be_block_lun *)cbe_lun->be_lun; 2730 2731 retval = 0; 2732 switch (io->scsiio.cdb[0]) { 2733 case SYNCHRONIZE_CACHE: 2734 case SYNCHRONIZE_CACHE_16: 2735 case WRITE_SAME_10: 2736 case WRITE_SAME_16: 2737 case UNMAP: 2738 /* 2739 * The upper level CTL code will filter out any CDBs with 2740 * the immediate bit set and return the proper error. 2741 * 2742 * We don't really need to worry about what LBA range the 2743 * user asked to be synced out. When they issue a sync 2744 * cache command, we'll sync out the whole thing. 2745 */ 2746 mtx_lock(&be_lun->queue_lock); 2747 STAILQ_INSERT_TAIL(&be_lun->config_write_queue, &io->io_hdr, 2748 links); 2749 mtx_unlock(&be_lun->queue_lock); 2750 taskqueue_enqueue(be_lun->io_taskqueue, &be_lun->io_task); 2751 break; 2752 case START_STOP_UNIT: { 2753 struct scsi_start_stop_unit *cdb; 2754 2755 cdb = (struct scsi_start_stop_unit *)io->scsiio.cdb; 2756 2757 if (cdb->how & SSS_START) 2758 retval = ctl_start_lun(cbe_lun); 2759 else { 2760 retval = ctl_stop_lun(cbe_lun); 2761 /* 2762 * XXX KDM Copan-specific offline behavior. 2763 * Figure out a reasonable way to port this? 2764 */ 2765#ifdef NEEDTOPORT 2766 if ((retval == 0) 2767 && (cdb->byte2 & SSS_ONOFFLINE)) 2768 retval = ctl_lun_offline(cbe_lun); 2769#endif 2770 } 2771 2772 /* 2773 * In general, the above routines should not fail. They 2774 * just set state for the LUN. So we've got something 2775 * pretty wrong here if we can't start or stop the LUN. 2776 */ 2777 if (retval != 0) { 2778 ctl_set_internal_failure(&io->scsiio, 2779 /*sks_valid*/ 1, 2780 /*retry_count*/ 0xf051); 2781 retval = CTL_RETVAL_COMPLETE; 2782 } else { 2783 ctl_set_success(&io->scsiio); 2784 } 2785 ctl_config_write_done(io); 2786 break; 2787 } 2788 default: 2789 ctl_set_invalid_opcode(&io->scsiio); 2790 ctl_config_write_done(io); 2791 retval = CTL_RETVAL_COMPLETE; 2792 break; 2793 } 2794 2795 return (retval); 2796} 2797 2798static int 2799ctl_be_block_config_read(union ctl_io *io) 2800{ 2801 struct ctl_be_block_lun *be_lun; 2802 struct ctl_be_lun *cbe_lun; 2803 int retval = 0; 2804 2805 DPRINTF("entered\n"); 2806 2807 cbe_lun = (struct ctl_be_lun *)io->io_hdr.ctl_private[ 2808 CTL_PRIV_BACKEND_LUN].ptr; 2809 be_lun = (struct ctl_be_block_lun *)cbe_lun->be_lun; 2810 2811 switch (io->scsiio.cdb[0]) { 2812 case SERVICE_ACTION_IN: 2813 if (io->scsiio.cdb[1] == SGLS_SERVICE_ACTION) { 2814 mtx_lock(&be_lun->queue_lock); 2815 STAILQ_INSERT_TAIL(&be_lun->config_read_queue, 2816 &io->io_hdr, links); 2817 mtx_unlock(&be_lun->queue_lock); 2818 taskqueue_enqueue(be_lun->io_taskqueue, 2819 &be_lun->io_task); 2820 retval = CTL_RETVAL_QUEUED; 2821 break; 2822 } 2823 ctl_set_invalid_field(&io->scsiio, 2824 /*sks_valid*/ 1, 2825 /*command*/ 1, 2826 /*field*/ 1, 2827 /*bit_valid*/ 1, 2828 /*bit*/ 4); 2829 ctl_config_read_done(io); 2830 retval = CTL_RETVAL_COMPLETE; 2831 break; 2832 default: 2833 ctl_set_invalid_opcode(&io->scsiio); 2834 ctl_config_read_done(io); 2835 retval = CTL_RETVAL_COMPLETE; 2836 break; 2837 } 2838 2839 return (retval); 2840} 2841 2842static int 2843ctl_be_block_lun_info(void *be_lun, struct sbuf *sb) 2844{ 2845 struct ctl_be_block_lun *lun; 2846 int retval; 2847 2848 lun = (struct ctl_be_block_lun *)be_lun; 2849 2850 retval = sbuf_printf(sb, "\t<num_threads>"); 2851 if (retval != 0) 2852 goto bailout; 2853 retval = sbuf_printf(sb, "%d", lun->num_threads); 2854 if (retval != 0) 2855 goto bailout; 2856 retval = sbuf_printf(sb, "</num_threads>\n"); 2857 2858bailout: 2859 return (retval); 2860} 2861 2862static uint64_t 2863ctl_be_block_lun_attr(void *be_lun, const char *attrname) 2864{ 2865 struct ctl_be_block_lun *lun = (struct ctl_be_block_lun *)be_lun; 2866 2867 if (lun->getattr == NULL) 2868 return (UINT64_MAX); 2869 return (lun->getattr(lun, attrname)); 2870} 2871 2872int 2873ctl_be_block_init(void) 2874{ 2875 struct ctl_be_block_softc *softc; 2876 int retval; 2877 2878 softc = &backend_block_softc; 2879 retval = 0; 2880 2881 mtx_init(&softc->lock, "ctlblock", NULL, MTX_DEF); 2882 beio_zone = uma_zcreate("beio", sizeof(struct ctl_be_block_io), 2883 NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); 2884 STAILQ_INIT(&softc->lun_list); 2885 2886 return (retval); 2887} 2888