sys/cam/ctl/ctl_backend_block.c

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