xref: /freebsd/sys/cam/ctl/ctl_backend_block.c (revision 52c2bb75163559a6e2866ad374a7de67a4ea1273)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3  *
4  * Copyright (c) 2003 Silicon Graphics International Corp.
5  * Copyright (c) 2009-2011 Spectra Logic Corporation
6  * Copyright (c) 2012 The FreeBSD Foundation
7  * Copyright (c) 2014-2015 Alexander Motin <mav@FreeBSD.org>
8  * All rights reserved.
9  *
10  * Portions of this software were developed by Edward Tomasz Napierala
11  * under sponsorship from the FreeBSD Foundation.
12  *
13  * Redistribution and use in source and binary forms, with or without
14  * modification, are permitted provided that the following conditions
15  * are met:
16  * 1. Redistributions of source code must retain the above copyright
17  *    notice, this list of conditions, and the following disclaimer,
18  *    without modification.
19  * 2. Redistributions in binary form must reproduce at minimum a disclaimer
20  *    substantially similar to the "NO WARRANTY" disclaimer below
21  *    ("Disclaimer") and any redistribution must be conditioned upon
22  *    including a substantially similar Disclaimer requirement for further
23  *    binary redistribution.
24  *
25  * NO WARRANTY
26  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
27  * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
28  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR
29  * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
30  * HOLDERS OR CONTRIBUTORS BE LIABLE FOR SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
34  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
35  * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
36  * POSSIBILITY OF SUCH DAMAGES.
37  *
38  * $Id: //depot/users/kenm/FreeBSD-test2/sys/cam/ctl/ctl_backend_block.c#5 $
39  */
40 /*
41  * CAM Target Layer driver backend for block devices.
42  *
43  * Author: Ken Merry <ken@FreeBSD.org>
44  */
45 #include <sys/cdefs.h>
46 __FBSDID("$FreeBSD$");
47 
48 #include <sys/param.h>
49 #include <sys/systm.h>
50 #include <sys/kernel.h>
51 #include <sys/types.h>
52 #include <sys/kthread.h>
53 #include <sys/bio.h>
54 #include <sys/fcntl.h>
55 #include <sys/limits.h>
56 #include <sys/lock.h>
57 #include <sys/mutex.h>
58 #include <sys/condvar.h>
59 #include <sys/malloc.h>
60 #include <sys/conf.h>
61 #include <sys/ioccom.h>
62 #include <sys/queue.h>
63 #include <sys/sbuf.h>
64 #include <sys/endian.h>
65 #include <sys/uio.h>
66 #include <sys/buf.h>
67 #include <sys/taskqueue.h>
68 #include <sys/vnode.h>
69 #include <sys/namei.h>
70 #include <sys/mount.h>
71 #include <sys/disk.h>
72 #include <sys/fcntl.h>
73 #include <sys/filedesc.h>
74 #include <sys/filio.h>
75 #include <sys/proc.h>
76 #include <sys/pcpu.h>
77 #include <sys/module.h>
78 #include <sys/sdt.h>
79 #include <sys/devicestat.h>
80 #include <sys/sysctl.h>
81 #include <sys/nv.h>
82 #include <sys/dnv.h>
83 
84 #include <geom/geom.h>
85 
86 #include <cam/cam.h>
87 #include <cam/scsi/scsi_all.h>
88 #include <cam/scsi/scsi_da.h>
89 #include <cam/ctl/ctl_io.h>
90 #include <cam/ctl/ctl.h>
91 #include <cam/ctl/ctl_backend.h>
92 #include <cam/ctl/ctl_ioctl.h>
93 #include <cam/ctl/ctl_ha.h>
94 #include <cam/ctl/ctl_scsi_all.h>
95 #include <cam/ctl/ctl_private.h>
96 #include <cam/ctl/ctl_error.h>
97 
98 /*
99  * The idea here is that we'll allocate enough S/G space to hold a 1MB
100  * I/O.  If we get an I/O larger than that, we'll split it.
101  */
102 #define	CTLBLK_HALF_IO_SIZE	(512 * 1024)
103 #define	CTLBLK_MAX_IO_SIZE	(CTLBLK_HALF_IO_SIZE * 2)
104 #define	CTLBLK_MAX_SEG		MAXPHYS
105 #define	CTLBLK_HALF_SEGS	MAX(CTLBLK_HALF_IO_SIZE / CTLBLK_MAX_SEG, 1)
106 #define	CTLBLK_MAX_SEGS		(CTLBLK_HALF_SEGS * 2)
107 
108 #ifdef CTLBLK_DEBUG
109 #define DPRINTF(fmt, args...) \
110     printf("cbb(%s:%d): " fmt, __FUNCTION__, __LINE__, ##args)
111 #else
112 #define DPRINTF(fmt, args...) do {} while(0)
113 #endif
114 
115 #define PRIV(io)	\
116     ((struct ctl_ptr_len_flags *)&(io)->io_hdr.ctl_private[CTL_PRIV_BACKEND])
117 #define ARGS(io)	\
118     ((struct ctl_lba_len_flags *)&(io)->io_hdr.ctl_private[CTL_PRIV_LBA_LEN])
119 
120 SDT_PROVIDER_DEFINE(cbb);
121 
122 typedef enum {
123 	CTL_BE_BLOCK_LUN_UNCONFIGURED	= 0x01,
124 	CTL_BE_BLOCK_LUN_CONFIG_ERR	= 0x02,
125 	CTL_BE_BLOCK_LUN_WAITING	= 0x04,
126 } ctl_be_block_lun_flags;
127 
128 typedef enum {
129 	CTL_BE_BLOCK_NONE,
130 	CTL_BE_BLOCK_DEV,
131 	CTL_BE_BLOCK_FILE
132 } ctl_be_block_type;
133 
134 struct ctl_be_block_filedata {
135 	struct ucred *cred;
136 };
137 
138 union ctl_be_block_bedata {
139 	struct ctl_be_block_filedata file;
140 };
141 
142 struct ctl_be_block_io;
143 struct ctl_be_block_lun;
144 
145 typedef void (*cbb_dispatch_t)(struct ctl_be_block_lun *be_lun,
146 			       struct ctl_be_block_io *beio);
147 typedef uint64_t (*cbb_getattr_t)(struct ctl_be_block_lun *be_lun,
148 				  const char *attrname);
149 
150 /*
151  * Backend LUN structure.  There is a 1:1 mapping between a block device
152  * and a backend block LUN, and between a backend block LUN and a CTL LUN.
153  */
154 struct ctl_be_block_lun {
155 	struct ctl_lun_create_params params;
156 	char lunname[32];
157 	char *dev_path;
158 	ctl_be_block_type dev_type;
159 	struct vnode *vn;
160 	union ctl_be_block_bedata backend;
161 	cbb_dispatch_t dispatch;
162 	cbb_dispatch_t lun_flush;
163 	cbb_dispatch_t unmap;
164 	cbb_dispatch_t get_lba_status;
165 	cbb_getattr_t getattr;
166 	uma_zone_t lun_zone;
167 	uint64_t size_blocks;
168 	uint64_t size_bytes;
169 	struct ctl_be_block_softc *softc;
170 	struct devstat *disk_stats;
171 	ctl_be_block_lun_flags flags;
172 	STAILQ_ENTRY(ctl_be_block_lun) links;
173 	struct ctl_be_lun cbe_lun;
174 	struct taskqueue *io_taskqueue;
175 	struct task io_task;
176 	int num_threads;
177 	STAILQ_HEAD(, ctl_io_hdr) input_queue;
178 	STAILQ_HEAD(, ctl_io_hdr) config_read_queue;
179 	STAILQ_HEAD(, ctl_io_hdr) config_write_queue;
180 	STAILQ_HEAD(, ctl_io_hdr) datamove_queue;
181 	struct mtx_padalign io_lock;
182 	struct mtx_padalign queue_lock;
183 };
184 
185 /*
186  * Overall softc structure for the block backend module.
187  */
188 struct ctl_be_block_softc {
189 	struct mtx			 lock;
190 	uma_zone_t			 beio_zone;
191 	int				 num_luns;
192 	STAILQ_HEAD(, ctl_be_block_lun)	 lun_list;
193 };
194 
195 static struct ctl_be_block_softc backend_block_softc;
196 
197 /*
198  * Per-I/O information.
199  */
200 struct ctl_be_block_io {
201 	union ctl_io			*io;
202 	struct ctl_sg_entry		sg_segs[CTLBLK_MAX_SEGS];
203 	struct iovec			xiovecs[CTLBLK_MAX_SEGS];
204 	int				bio_cmd;
205 	int				num_segs;
206 	int				num_bios_sent;
207 	int				num_bios_done;
208 	int				send_complete;
209 	int				first_error;
210 	uint64_t			first_error_offset;
211 	struct bintime			ds_t0;
212 	devstat_tag_type		ds_tag_type;
213 	devstat_trans_flags		ds_trans_type;
214 	uint64_t			io_len;
215 	uint64_t			io_offset;
216 	int				io_arg;
217 	struct ctl_be_block_softc	*softc;
218 	struct ctl_be_block_lun		*lun;
219 	void (*beio_cont)(struct ctl_be_block_io *beio); /* to continue processing */
220 };
221 
222 extern struct ctl_softc *control_softc;
223 
224 static int cbb_num_threads = 14;
225 SYSCTL_NODE(_kern_cam_ctl, OID_AUTO, block, CTLFLAG_RD, 0,
226 	    "CAM Target Layer Block Backend");
227 SYSCTL_INT(_kern_cam_ctl_block, OID_AUTO, num_threads, CTLFLAG_RWTUN,
228            &cbb_num_threads, 0, "Number of threads per backing file");
229 
230 static struct ctl_be_block_io *ctl_alloc_beio(struct ctl_be_block_softc *softc);
231 static void ctl_free_beio(struct ctl_be_block_io *beio);
232 static void ctl_complete_beio(struct ctl_be_block_io *beio);
233 static int ctl_be_block_move_done(union ctl_io *io);
234 static void ctl_be_block_biodone(struct bio *bio);
235 static void ctl_be_block_flush_file(struct ctl_be_block_lun *be_lun,
236 				    struct ctl_be_block_io *beio);
237 static void ctl_be_block_dispatch_file(struct ctl_be_block_lun *be_lun,
238 				       struct ctl_be_block_io *beio);
239 static void ctl_be_block_gls_file(struct ctl_be_block_lun *be_lun,
240 				  struct ctl_be_block_io *beio);
241 static uint64_t ctl_be_block_getattr_file(struct ctl_be_block_lun *be_lun,
242 					 const char *attrname);
243 static void ctl_be_block_flush_dev(struct ctl_be_block_lun *be_lun,
244 				   struct ctl_be_block_io *beio);
245 static void ctl_be_block_unmap_dev(struct ctl_be_block_lun *be_lun,
246 				   struct ctl_be_block_io *beio);
247 static void ctl_be_block_dispatch_dev(struct ctl_be_block_lun *be_lun,
248 				      struct ctl_be_block_io *beio);
249 static uint64_t ctl_be_block_getattr_dev(struct ctl_be_block_lun *be_lun,
250 					 const char *attrname);
251 static void ctl_be_block_cr_dispatch(struct ctl_be_block_lun *be_lun,
252 				    union ctl_io *io);
253 static void ctl_be_block_cw_dispatch(struct ctl_be_block_lun *be_lun,
254 				    union ctl_io *io);
255 static void ctl_be_block_dispatch(struct ctl_be_block_lun *be_lun,
256 				  union ctl_io *io);
257 static void ctl_be_block_worker(void *context, int pending);
258 static int ctl_be_block_submit(union ctl_io *io);
259 static int ctl_be_block_ioctl(struct cdev *dev, u_long cmd, caddr_t addr,
260 				   int flag, struct thread *td);
261 static int ctl_be_block_open_file(struct ctl_be_block_lun *be_lun,
262 				  struct ctl_lun_req *req);
263 static int ctl_be_block_open_dev(struct ctl_be_block_lun *be_lun,
264 				 struct ctl_lun_req *req);
265 static int ctl_be_block_close(struct ctl_be_block_lun *be_lun);
266 static int ctl_be_block_open(struct ctl_be_block_lun *be_lun,
267 			     struct ctl_lun_req *req);
268 static int ctl_be_block_create(struct ctl_be_block_softc *softc,
269 			       struct ctl_lun_req *req);
270 static int ctl_be_block_rm(struct ctl_be_block_softc *softc,
271 			   struct ctl_lun_req *req);
272 static int ctl_be_block_modify(struct ctl_be_block_softc *softc,
273 			   struct ctl_lun_req *req);
274 static void ctl_be_block_lun_shutdown(void *be_lun);
275 static void ctl_be_block_lun_config_status(void *be_lun,
276 					   ctl_lun_config_status status);
277 static int ctl_be_block_config_write(union ctl_io *io);
278 static int ctl_be_block_config_read(union ctl_io *io);
279 static int ctl_be_block_lun_info(void *be_lun, struct sbuf *sb);
280 static uint64_t ctl_be_block_lun_attr(void *be_lun, const char *attrname);
281 static int ctl_be_block_init(void);
282 static int ctl_be_block_shutdown(void);
283 
284 static struct ctl_backend_driver ctl_be_block_driver =
285 {
286 	.name = "block",
287 	.flags = CTL_BE_FLAG_HAS_CONFIG,
288 	.init = ctl_be_block_init,
289 	.shutdown = ctl_be_block_shutdown,
290 	.data_submit = ctl_be_block_submit,
291 	.data_move_done = ctl_be_block_move_done,
292 	.config_read = ctl_be_block_config_read,
293 	.config_write = ctl_be_block_config_write,
294 	.ioctl = ctl_be_block_ioctl,
295 	.lun_info = ctl_be_block_lun_info,
296 	.lun_attr = ctl_be_block_lun_attr
297 };
298 
299 MALLOC_DEFINE(M_CTLBLK, "ctlblk", "Memory used for CTL block backend");
300 CTL_BACKEND_DECLARE(cbb, ctl_be_block_driver);
301 
302 static struct ctl_be_block_io *
303 ctl_alloc_beio(struct ctl_be_block_softc *softc)
304 {
305 	struct ctl_be_block_io *beio;
306 
307 	beio = uma_zalloc(softc->beio_zone, M_WAITOK | M_ZERO);
308 	beio->softc = softc;
309 	return (beio);
310 }
311 
312 static void
313 ctl_free_beio(struct ctl_be_block_io *beio)
314 {
315 	int duplicate_free;
316 	int i;
317 
318 	duplicate_free = 0;
319 
320 	for (i = 0; i < beio->num_segs; i++) {
321 		if (beio->sg_segs[i].addr == NULL)
322 			duplicate_free++;
323 
324 		uma_zfree(beio->lun->lun_zone, beio->sg_segs[i].addr);
325 		beio->sg_segs[i].addr = NULL;
326 
327 		/* For compare we had two equal S/G lists. */
328 		if (ARGS(beio->io)->flags & CTL_LLF_COMPARE) {
329 			uma_zfree(beio->lun->lun_zone,
330 			    beio->sg_segs[i + CTLBLK_HALF_SEGS].addr);
331 			beio->sg_segs[i + CTLBLK_HALF_SEGS].addr = NULL;
332 		}
333 	}
334 
335 	if (duplicate_free > 0) {
336 		printf("%s: %d duplicate frees out of %d segments\n", __func__,
337 		       duplicate_free, beio->num_segs);
338 	}
339 
340 	uma_zfree(beio->softc->beio_zone, beio);
341 }
342 
343 static void
344 ctl_complete_beio(struct ctl_be_block_io *beio)
345 {
346 	union ctl_io *io = beio->io;
347 
348 	if (beio->beio_cont != NULL) {
349 		beio->beio_cont(beio);
350 	} else {
351 		ctl_free_beio(beio);
352 		ctl_data_submit_done(io);
353 	}
354 }
355 
356 static size_t
357 cmp(uint8_t *a, uint8_t *b, size_t size)
358 {
359 	size_t i;
360 
361 	for (i = 0; i < size; i++) {
362 		if (a[i] != b[i])
363 			break;
364 	}
365 	return (i);
366 }
367 
368 static void
369 ctl_be_block_compare(union ctl_io *io)
370 {
371 	struct ctl_be_block_io *beio;
372 	uint64_t off, res;
373 	int i;
374 	uint8_t info[8];
375 
376 	beio = (struct ctl_be_block_io *)PRIV(io)->ptr;
377 	off = 0;
378 	for (i = 0; i < beio->num_segs; i++) {
379 		res = cmp(beio->sg_segs[i].addr,
380 		    beio->sg_segs[i + CTLBLK_HALF_SEGS].addr,
381 		    beio->sg_segs[i].len);
382 		off += res;
383 		if (res < beio->sg_segs[i].len)
384 			break;
385 	}
386 	if (i < beio->num_segs) {
387 		scsi_u64to8b(off, info);
388 		ctl_set_sense(&io->scsiio, /*current_error*/ 1,
389 		    /*sense_key*/ SSD_KEY_MISCOMPARE,
390 		    /*asc*/ 0x1D, /*ascq*/ 0x00,
391 		    /*type*/ SSD_ELEM_INFO,
392 		    /*size*/ sizeof(info), /*data*/ &info,
393 		    /*type*/ SSD_ELEM_NONE);
394 	} else
395 		ctl_set_success(&io->scsiio);
396 }
397 
398 static int
399 ctl_be_block_move_done(union ctl_io *io)
400 {
401 	struct ctl_be_block_io *beio;
402 	struct ctl_be_block_lun *be_lun;
403 	struct ctl_lba_len_flags *lbalen;
404 #ifdef CTL_TIME_IO
405 	struct bintime cur_bt;
406 #endif
407 
408 	beio = (struct ctl_be_block_io *)PRIV(io)->ptr;
409 	be_lun = beio->lun;
410 
411 	DPRINTF("entered\n");
412 
413 #ifdef CTL_TIME_IO
414 	getbinuptime(&cur_bt);
415 	bintime_sub(&cur_bt, &io->io_hdr.dma_start_bt);
416 	bintime_add(&io->io_hdr.dma_bt, &cur_bt);
417 #endif
418 	io->io_hdr.num_dmas++;
419 	io->scsiio.kern_rel_offset += io->scsiio.kern_data_len;
420 
421 	/*
422 	 * We set status at this point for read commands, and write
423 	 * commands with errors.
424 	 */
425 	if (io->io_hdr.flags & CTL_FLAG_ABORT) {
426 		;
427 	} else if ((io->io_hdr.port_status != 0) &&
428 	    ((io->io_hdr.status & CTL_STATUS_MASK) == CTL_STATUS_NONE ||
429 	     (io->io_hdr.status & CTL_STATUS_MASK) == CTL_SUCCESS)) {
430 		ctl_set_internal_failure(&io->scsiio, /*sks_valid*/ 1,
431 		    /*retry_count*/ io->io_hdr.port_status);
432 	} else if (io->scsiio.kern_data_resid != 0 &&
433 	    (io->io_hdr.flags & CTL_FLAG_DATA_MASK) == CTL_FLAG_DATA_OUT &&
434 	    ((io->io_hdr.status & CTL_STATUS_MASK) == CTL_STATUS_NONE ||
435 	     (io->io_hdr.status & CTL_STATUS_MASK) == CTL_SUCCESS)) {
436 		ctl_set_invalid_field_ciu(&io->scsiio);
437 	} else if ((io->io_hdr.port_status == 0) &&
438 	    ((io->io_hdr.status & CTL_STATUS_MASK) == CTL_STATUS_NONE)) {
439 		lbalen = ARGS(beio->io);
440 		if (lbalen->flags & CTL_LLF_READ) {
441 			ctl_set_success(&io->scsiio);
442 		} else if (lbalen->flags & CTL_LLF_COMPARE) {
443 			/* We have two data blocks ready for comparison. */
444 			ctl_be_block_compare(io);
445 		}
446 	}
447 
448 	/*
449 	 * If this is a read, or a write with errors, it is done.
450 	 */
451 	if ((beio->bio_cmd == BIO_READ)
452 	 || ((io->io_hdr.flags & CTL_FLAG_ABORT) != 0)
453 	 || ((io->io_hdr.status & CTL_STATUS_MASK) != CTL_STATUS_NONE)) {
454 		ctl_complete_beio(beio);
455 		return (0);
456 	}
457 
458 	/*
459 	 * At this point, we have a write and the DMA completed
460 	 * successfully.  We now have to queue it to the task queue to
461 	 * execute the backend I/O.  That is because we do blocking
462 	 * memory allocations, and in the file backing case, blocking I/O.
463 	 * This move done routine is generally called in the SIM's
464 	 * interrupt context, and therefore we cannot block.
465 	 */
466 	mtx_lock(&be_lun->queue_lock);
467 	STAILQ_INSERT_TAIL(&be_lun->datamove_queue, &io->io_hdr, links);
468 	mtx_unlock(&be_lun->queue_lock);
469 	taskqueue_enqueue(be_lun->io_taskqueue, &be_lun->io_task);
470 
471 	return (0);
472 }
473 
474 static void
475 ctl_be_block_biodone(struct bio *bio)
476 {
477 	struct ctl_be_block_io *beio;
478 	struct ctl_be_block_lun *be_lun;
479 	union ctl_io *io;
480 	int error;
481 
482 	beio = bio->bio_caller1;
483 	be_lun = beio->lun;
484 	io = beio->io;
485 
486 	DPRINTF("entered\n");
487 
488 	error = bio->bio_error;
489 	mtx_lock(&be_lun->io_lock);
490 	if (error != 0 &&
491 	    (beio->first_error == 0 ||
492 	     bio->bio_offset < beio->first_error_offset)) {
493 		beio->first_error = error;
494 		beio->first_error_offset = bio->bio_offset;
495 	}
496 
497 	beio->num_bios_done++;
498 
499 	/*
500 	 * XXX KDM will this cause WITNESS to complain?  Holding a lock
501 	 * during the free might cause it to complain.
502 	 */
503 	g_destroy_bio(bio);
504 
505 	/*
506 	 * If the send complete bit isn't set, or we aren't the last I/O to
507 	 * complete, then we're done.
508 	 */
509 	if ((beio->send_complete == 0)
510 	 || (beio->num_bios_done < beio->num_bios_sent)) {
511 		mtx_unlock(&be_lun->io_lock);
512 		return;
513 	}
514 
515 	/*
516 	 * At this point, we've verified that we are the last I/O to
517 	 * complete, so it's safe to drop the lock.
518 	 */
519 	devstat_end_transaction(beio->lun->disk_stats, beio->io_len,
520 	    beio->ds_tag_type, beio->ds_trans_type,
521 	    /*now*/ NULL, /*then*/&beio->ds_t0);
522 	mtx_unlock(&be_lun->io_lock);
523 
524 	/*
525 	 * If there are any errors from the backing device, we fail the
526 	 * entire I/O with a medium error.
527 	 */
528 	error = beio->first_error;
529 	if (error != 0) {
530 		if (error == EOPNOTSUPP) {
531 			ctl_set_invalid_opcode(&io->scsiio);
532 		} else if (error == ENOSPC || error == EDQUOT) {
533 			ctl_set_space_alloc_fail(&io->scsiio);
534 		} else if (error == EROFS || error == EACCES) {
535 			ctl_set_hw_write_protected(&io->scsiio);
536 		} else if (beio->bio_cmd == BIO_FLUSH) {
537 			/* XXX KDM is there is a better error here? */
538 			ctl_set_internal_failure(&io->scsiio,
539 						 /*sks_valid*/ 1,
540 						 /*retry_count*/ 0xbad2);
541 		} else {
542 			ctl_set_medium_error(&io->scsiio,
543 			    beio->bio_cmd == BIO_READ);
544 		}
545 		ctl_complete_beio(beio);
546 		return;
547 	}
548 
549 	/*
550 	 * If this is a write, a flush, a delete or verify, we're all done.
551 	 * If this is a read, we can now send the data to the user.
552 	 */
553 	if ((beio->bio_cmd == BIO_WRITE)
554 	 || (beio->bio_cmd == BIO_FLUSH)
555 	 || (beio->bio_cmd == BIO_DELETE)
556 	 || (ARGS(io)->flags & CTL_LLF_VERIFY)) {
557 		ctl_set_success(&io->scsiio);
558 		ctl_complete_beio(beio);
559 	} else {
560 		if ((ARGS(io)->flags & CTL_LLF_READ) &&
561 		    beio->beio_cont == NULL) {
562 			ctl_set_success(&io->scsiio);
563 			ctl_serseq_done(io);
564 		}
565 #ifdef CTL_TIME_IO
566 		getbinuptime(&io->io_hdr.dma_start_bt);
567 #endif
568 		ctl_datamove(io);
569 	}
570 }
571 
572 static void
573 ctl_be_block_flush_file(struct ctl_be_block_lun *be_lun,
574 			struct ctl_be_block_io *beio)
575 {
576 	union ctl_io *io = beio->io;
577 	struct mount *mountpoint;
578 	int error, lock_flags;
579 
580 	DPRINTF("entered\n");
581 
582 	binuptime(&beio->ds_t0);
583 	mtx_lock(&be_lun->io_lock);
584 	devstat_start_transaction(beio->lun->disk_stats, &beio->ds_t0);
585 	mtx_unlock(&be_lun->io_lock);
586 
587 	(void) vn_start_write(be_lun->vn, &mountpoint, V_WAIT);
588 
589 	if (MNT_SHARED_WRITES(mountpoint) ||
590 	    ((mountpoint == NULL) && MNT_SHARED_WRITES(be_lun->vn->v_mount)))
591 		lock_flags = LK_SHARED;
592 	else
593 		lock_flags = LK_EXCLUSIVE;
594 	vn_lock(be_lun->vn, lock_flags | LK_RETRY);
595 	error = VOP_FSYNC(be_lun->vn, beio->io_arg ? MNT_NOWAIT : MNT_WAIT,
596 	    curthread);
597 	VOP_UNLOCK(be_lun->vn, 0);
598 
599 	vn_finished_write(mountpoint);
600 
601 	mtx_lock(&be_lun->io_lock);
602 	devstat_end_transaction(beio->lun->disk_stats, beio->io_len,
603 	    beio->ds_tag_type, beio->ds_trans_type,
604 	    /*now*/ NULL, /*then*/&beio->ds_t0);
605 	mtx_unlock(&be_lun->io_lock);
606 
607 	if (error == 0)
608 		ctl_set_success(&io->scsiio);
609 	else {
610 		/* XXX KDM is there is a better error here? */
611 		ctl_set_internal_failure(&io->scsiio,
612 					 /*sks_valid*/ 1,
613 					 /*retry_count*/ 0xbad1);
614 	}
615 
616 	ctl_complete_beio(beio);
617 }
618 
619 SDT_PROBE_DEFINE1(cbb, , read, file_start, "uint64_t");
620 SDT_PROBE_DEFINE1(cbb, , write, file_start, "uint64_t");
621 SDT_PROBE_DEFINE1(cbb, , read, file_done,"uint64_t");
622 SDT_PROBE_DEFINE1(cbb, , write, file_done, "uint64_t");
623 
624 static void
625 ctl_be_block_dispatch_file(struct ctl_be_block_lun *be_lun,
626 			   struct ctl_be_block_io *beio)
627 {
628 	struct ctl_be_block_filedata *file_data;
629 	union ctl_io *io;
630 	struct uio xuio;
631 	struct iovec *xiovec;
632 	size_t s;
633 	int error, flags, i;
634 
635 	DPRINTF("entered\n");
636 
637 	file_data = &be_lun->backend.file;
638 	io = beio->io;
639 	flags = 0;
640 	if (ARGS(io)->flags & CTL_LLF_DPO)
641 		flags |= IO_DIRECT;
642 	if (beio->bio_cmd == BIO_WRITE && ARGS(io)->flags & CTL_LLF_FUA)
643 		flags |= IO_SYNC;
644 
645 	bzero(&xuio, sizeof(xuio));
646 	if (beio->bio_cmd == BIO_READ) {
647 		SDT_PROBE0(cbb, , read, file_start);
648 		xuio.uio_rw = UIO_READ;
649 	} else {
650 		SDT_PROBE0(cbb, , write, file_start);
651 		xuio.uio_rw = UIO_WRITE;
652 	}
653 	xuio.uio_offset = beio->io_offset;
654 	xuio.uio_resid = beio->io_len;
655 	xuio.uio_segflg = UIO_SYSSPACE;
656 	xuio.uio_iov = beio->xiovecs;
657 	xuio.uio_iovcnt = beio->num_segs;
658 	xuio.uio_td = curthread;
659 
660 	for (i = 0, xiovec = xuio.uio_iov; i < xuio.uio_iovcnt; i++, xiovec++) {
661 		xiovec->iov_base = beio->sg_segs[i].addr;
662 		xiovec->iov_len = beio->sg_segs[i].len;
663 	}
664 
665 	binuptime(&beio->ds_t0);
666 	mtx_lock(&be_lun->io_lock);
667 	devstat_start_transaction(beio->lun->disk_stats, &beio->ds_t0);
668 	mtx_unlock(&be_lun->io_lock);
669 
670 	if (beio->bio_cmd == BIO_READ) {
671 		vn_lock(be_lun->vn, LK_SHARED | LK_RETRY);
672 
673 		/*
674 		 * UFS pays attention to IO_DIRECT for reads.  If the
675 		 * DIRECTIO option is configured into the kernel, it calls
676 		 * ffs_rawread().  But that only works for single-segment
677 		 * uios with user space addresses.  In our case, with a
678 		 * kernel uio, it still reads into the buffer cache, but it
679 		 * will just try to release the buffer from the cache later
680 		 * on in ffs_read().
681 		 *
682 		 * ZFS does not pay attention to IO_DIRECT for reads.
683 		 *
684 		 * UFS does not pay attention to IO_SYNC for reads.
685 		 *
686 		 * ZFS pays attention to IO_SYNC (which translates into the
687 		 * Solaris define FRSYNC for zfs_read()) for reads.  It
688 		 * attempts to sync the file before reading.
689 		 */
690 		error = VOP_READ(be_lun->vn, &xuio, flags, file_data->cred);
691 
692 		VOP_UNLOCK(be_lun->vn, 0);
693 		SDT_PROBE0(cbb, , read, file_done);
694 		if (error == 0 && xuio.uio_resid > 0) {
695 			/*
696 			 * If we red less then requested (EOF), then
697 			 * we should clean the rest of the buffer.
698 			 */
699 			s = beio->io_len - xuio.uio_resid;
700 			for (i = 0; i < beio->num_segs; i++) {
701 				if (s >= beio->sg_segs[i].len) {
702 					s -= beio->sg_segs[i].len;
703 					continue;
704 				}
705 				bzero((uint8_t *)beio->sg_segs[i].addr + s,
706 				    beio->sg_segs[i].len - s);
707 				s = 0;
708 			}
709 		}
710 	} else {
711 		struct mount *mountpoint;
712 		int lock_flags;
713 
714 		(void)vn_start_write(be_lun->vn, &mountpoint, V_WAIT);
715 
716 		if (MNT_SHARED_WRITES(mountpoint) || ((mountpoint == NULL)
717 		  && MNT_SHARED_WRITES(be_lun->vn->v_mount)))
718 			lock_flags = LK_SHARED;
719 		else
720 			lock_flags = LK_EXCLUSIVE;
721 		vn_lock(be_lun->vn, lock_flags | LK_RETRY);
722 
723 		/*
724 		 * UFS pays attention to IO_DIRECT for writes.  The write
725 		 * is done asynchronously.  (Normally the write would just
726 		 * get put into cache.
727 		 *
728 		 * UFS pays attention to IO_SYNC for writes.  It will
729 		 * attempt to write the buffer out synchronously if that
730 		 * flag is set.
731 		 *
732 		 * ZFS does not pay attention to IO_DIRECT for writes.
733 		 *
734 		 * ZFS pays attention to IO_SYNC (a.k.a. FSYNC or FRSYNC)
735 		 * for writes.  It will flush the transaction from the
736 		 * cache before returning.
737 		 */
738 		error = VOP_WRITE(be_lun->vn, &xuio, flags, file_data->cred);
739 		VOP_UNLOCK(be_lun->vn, 0);
740 
741 		vn_finished_write(mountpoint);
742 		SDT_PROBE0(cbb, , write, file_done);
743         }
744 
745 	mtx_lock(&be_lun->io_lock);
746 	devstat_end_transaction(beio->lun->disk_stats, beio->io_len,
747 	    beio->ds_tag_type, beio->ds_trans_type,
748 	    /*now*/ NULL, /*then*/&beio->ds_t0);
749 	mtx_unlock(&be_lun->io_lock);
750 
751 	/*
752 	 * If we got an error, set the sense data to "MEDIUM ERROR" and
753 	 * return the I/O to the user.
754 	 */
755 	if (error != 0) {
756 		if (error == ENOSPC || error == EDQUOT) {
757 			ctl_set_space_alloc_fail(&io->scsiio);
758 		} else if (error == EROFS || error == EACCES) {
759 			ctl_set_hw_write_protected(&io->scsiio);
760 		} else {
761 			ctl_set_medium_error(&io->scsiio,
762 			    beio->bio_cmd == BIO_READ);
763 		}
764 		ctl_complete_beio(beio);
765 		return;
766 	}
767 
768 	/*
769 	 * If this is a write or a verify, we're all done.
770 	 * If this is a read, we can now send the data to the user.
771 	 */
772 	if ((beio->bio_cmd == BIO_WRITE) ||
773 	    (ARGS(io)->flags & CTL_LLF_VERIFY)) {
774 		ctl_set_success(&io->scsiio);
775 		ctl_complete_beio(beio);
776 	} else {
777 		if ((ARGS(io)->flags & CTL_LLF_READ) &&
778 		    beio->beio_cont == NULL) {
779 			ctl_set_success(&io->scsiio);
780 			ctl_serseq_done(io);
781 		}
782 #ifdef CTL_TIME_IO
783 		getbinuptime(&io->io_hdr.dma_start_bt);
784 #endif
785 		ctl_datamove(io);
786 	}
787 }
788 
789 static void
790 ctl_be_block_gls_file(struct ctl_be_block_lun *be_lun,
791 			struct ctl_be_block_io *beio)
792 {
793 	union ctl_io *io = beio->io;
794 	struct ctl_lba_len_flags *lbalen = ARGS(io);
795 	struct scsi_get_lba_status_data *data;
796 	off_t roff, off;
797 	int error, status;
798 
799 	DPRINTF("entered\n");
800 
801 	off = roff = ((off_t)lbalen->lba) * be_lun->cbe_lun.blocksize;
802 	vn_lock(be_lun->vn, LK_SHARED | LK_RETRY);
803 	error = VOP_IOCTL(be_lun->vn, FIOSEEKHOLE, &off,
804 	    0, curthread->td_ucred, curthread);
805 	if (error == 0 && off > roff)
806 		status = 0;	/* mapped up to off */
807 	else {
808 		error = VOP_IOCTL(be_lun->vn, FIOSEEKDATA, &off,
809 		    0, curthread->td_ucred, curthread);
810 		if (error == 0 && off > roff)
811 			status = 1;	/* deallocated up to off */
812 		else {
813 			status = 0;	/* unknown up to the end */
814 			off = be_lun->size_bytes;
815 		}
816 	}
817 	VOP_UNLOCK(be_lun->vn, 0);
818 
819 	data = (struct scsi_get_lba_status_data *)io->scsiio.kern_data_ptr;
820 	scsi_u64to8b(lbalen->lba, data->descr[0].addr);
821 	scsi_ulto4b(MIN(UINT32_MAX, off / be_lun->cbe_lun.blocksize -
822 	    lbalen->lba), data->descr[0].length);
823 	data->descr[0].status = status;
824 
825 	ctl_complete_beio(beio);
826 }
827 
828 static uint64_t
829 ctl_be_block_getattr_file(struct ctl_be_block_lun *be_lun, const char *attrname)
830 {
831 	struct vattr		vattr;
832 	struct statfs		statfs;
833 	uint64_t		val;
834 	int			error;
835 
836 	val = UINT64_MAX;
837 	if (be_lun->vn == NULL)
838 		return (val);
839 	vn_lock(be_lun->vn, LK_SHARED | LK_RETRY);
840 	if (strcmp(attrname, "blocksused") == 0) {
841 		error = VOP_GETATTR(be_lun->vn, &vattr, curthread->td_ucred);
842 		if (error == 0)
843 			val = vattr.va_bytes / be_lun->cbe_lun.blocksize;
844 	}
845 	if (strcmp(attrname, "blocksavail") == 0 &&
846 	    (be_lun->vn->v_iflag & VI_DOOMED) == 0) {
847 		error = VFS_STATFS(be_lun->vn->v_mount, &statfs);
848 		if (error == 0)
849 			val = statfs.f_bavail * statfs.f_bsize /
850 			    be_lun->cbe_lun.blocksize;
851 	}
852 	VOP_UNLOCK(be_lun->vn, 0);
853 	return (val);
854 }
855 
856 static void
857 ctl_be_block_dispatch_zvol(struct ctl_be_block_lun *be_lun,
858 			   struct ctl_be_block_io *beio)
859 {
860 	union ctl_io *io;
861 	struct cdevsw *csw;
862 	struct cdev *dev;
863 	struct uio xuio;
864 	struct iovec *xiovec;
865 	int error, flags, i, ref;
866 
867 	DPRINTF("entered\n");
868 
869 	io = beio->io;
870 	flags = 0;
871 	if (ARGS(io)->flags & CTL_LLF_DPO)
872 		flags |= IO_DIRECT;
873 	if (beio->bio_cmd == BIO_WRITE && ARGS(io)->flags & CTL_LLF_FUA)
874 		flags |= IO_SYNC;
875 
876 	bzero(&xuio, sizeof(xuio));
877 	if (beio->bio_cmd == BIO_READ) {
878 		SDT_PROBE0(cbb, , read, file_start);
879 		xuio.uio_rw = UIO_READ;
880 	} else {
881 		SDT_PROBE0(cbb, , write, file_start);
882 		xuio.uio_rw = UIO_WRITE;
883 	}
884 	xuio.uio_offset = beio->io_offset;
885 	xuio.uio_resid = beio->io_len;
886 	xuio.uio_segflg = UIO_SYSSPACE;
887 	xuio.uio_iov = beio->xiovecs;
888 	xuio.uio_iovcnt = beio->num_segs;
889 	xuio.uio_td = curthread;
890 
891 	for (i = 0, xiovec = xuio.uio_iov; i < xuio.uio_iovcnt; i++, xiovec++) {
892 		xiovec->iov_base = beio->sg_segs[i].addr;
893 		xiovec->iov_len = beio->sg_segs[i].len;
894 	}
895 
896 	binuptime(&beio->ds_t0);
897 	mtx_lock(&be_lun->io_lock);
898 	devstat_start_transaction(beio->lun->disk_stats, &beio->ds_t0);
899 	mtx_unlock(&be_lun->io_lock);
900 
901 	csw = devvn_refthread(be_lun->vn, &dev, &ref);
902 	if (csw) {
903 		if (beio->bio_cmd == BIO_READ)
904 			error = csw->d_read(dev, &xuio, flags);
905 		else
906 			error = csw->d_write(dev, &xuio, flags);
907 		dev_relthread(dev, ref);
908 	} else
909 		error = ENXIO;
910 
911 	if (beio->bio_cmd == BIO_READ)
912 		SDT_PROBE0(cbb, , read, file_done);
913 	else
914 		SDT_PROBE0(cbb, , write, file_done);
915 
916 	mtx_lock(&be_lun->io_lock);
917 	devstat_end_transaction(beio->lun->disk_stats, beio->io_len,
918 	    beio->ds_tag_type, beio->ds_trans_type,
919 	    /*now*/ NULL, /*then*/&beio->ds_t0);
920 	mtx_unlock(&be_lun->io_lock);
921 
922 	/*
923 	 * If we got an error, set the sense data to "MEDIUM ERROR" and
924 	 * return the I/O to the user.
925 	 */
926 	if (error != 0) {
927 		if (error == ENOSPC || error == EDQUOT) {
928 			ctl_set_space_alloc_fail(&io->scsiio);
929 		} else if (error == EROFS || error == EACCES) {
930 			ctl_set_hw_write_protected(&io->scsiio);
931 		} else {
932 			ctl_set_medium_error(&io->scsiio,
933 			    beio->bio_cmd == BIO_READ);
934 		}
935 		ctl_complete_beio(beio);
936 		return;
937 	}
938 
939 	/*
940 	 * If this is a write or a verify, we're all done.
941 	 * If this is a read, we can now send the data to the user.
942 	 */
943 	if ((beio->bio_cmd == BIO_WRITE) ||
944 	    (ARGS(io)->flags & CTL_LLF_VERIFY)) {
945 		ctl_set_success(&io->scsiio);
946 		ctl_complete_beio(beio);
947 	} else {
948 		if ((ARGS(io)->flags & CTL_LLF_READ) &&
949 		    beio->beio_cont == NULL) {
950 			ctl_set_success(&io->scsiio);
951 			ctl_serseq_done(io);
952 		}
953 #ifdef CTL_TIME_IO
954 		getbinuptime(&io->io_hdr.dma_start_bt);
955 #endif
956 		ctl_datamove(io);
957 	}
958 }
959 
960 static void
961 ctl_be_block_gls_zvol(struct ctl_be_block_lun *be_lun,
962 			struct ctl_be_block_io *beio)
963 {
964 	union ctl_io *io = beio->io;
965 	struct cdevsw *csw;
966 	struct cdev *dev;
967 	struct ctl_lba_len_flags *lbalen = ARGS(io);
968 	struct scsi_get_lba_status_data *data;
969 	off_t roff, off;
970 	int error, ref, status;
971 
972 	DPRINTF("entered\n");
973 
974 	csw = devvn_refthread(be_lun->vn, &dev, &ref);
975 	if (csw == NULL) {
976 		status = 0;	/* unknown up to the end */
977 		off = be_lun->size_bytes;
978 		goto done;
979 	}
980 	off = roff = ((off_t)lbalen->lba) * be_lun->cbe_lun.blocksize;
981 	error = csw->d_ioctl(dev, FIOSEEKHOLE, (caddr_t)&off, FREAD,
982 	    curthread);
983 	if (error == 0 && off > roff)
984 		status = 0;	/* mapped up to off */
985 	else {
986 		error = csw->d_ioctl(dev, FIOSEEKDATA, (caddr_t)&off, FREAD,
987 		    curthread);
988 		if (error == 0 && off > roff)
989 			status = 1;	/* deallocated up to off */
990 		else {
991 			status = 0;	/* unknown up to the end */
992 			off = be_lun->size_bytes;
993 		}
994 	}
995 	dev_relthread(dev, ref);
996 
997 done:
998 	data = (struct scsi_get_lba_status_data *)io->scsiio.kern_data_ptr;
999 	scsi_u64to8b(lbalen->lba, data->descr[0].addr);
1000 	scsi_ulto4b(MIN(UINT32_MAX, off / be_lun->cbe_lun.blocksize -
1001 	    lbalen->lba), data->descr[0].length);
1002 	data->descr[0].status = status;
1003 
1004 	ctl_complete_beio(beio);
1005 }
1006 
1007 static void
1008 ctl_be_block_flush_dev(struct ctl_be_block_lun *be_lun,
1009 		       struct ctl_be_block_io *beio)
1010 {
1011 	struct bio *bio;
1012 	struct cdevsw *csw;
1013 	struct cdev *dev;
1014 	int ref;
1015 
1016 	DPRINTF("entered\n");
1017 
1018 	/* This can't fail, it's a blocking allocation. */
1019 	bio = g_alloc_bio();
1020 
1021 	bio->bio_cmd	    = BIO_FLUSH;
1022 	bio->bio_offset	    = 0;
1023 	bio->bio_data	    = 0;
1024 	bio->bio_done	    = ctl_be_block_biodone;
1025 	bio->bio_caller1    = beio;
1026 	bio->bio_pblkno	    = 0;
1027 
1028 	/*
1029 	 * We don't need to acquire the LUN lock here, because we are only
1030 	 * sending one bio, and so there is no other context to synchronize
1031 	 * with.
1032 	 */
1033 	beio->num_bios_sent = 1;
1034 	beio->send_complete = 1;
1035 
1036 	binuptime(&beio->ds_t0);
1037 	mtx_lock(&be_lun->io_lock);
1038 	devstat_start_transaction(be_lun->disk_stats, &beio->ds_t0);
1039 	mtx_unlock(&be_lun->io_lock);
1040 
1041 	csw = devvn_refthread(be_lun->vn, &dev, &ref);
1042 	if (csw) {
1043 		bio->bio_dev = dev;
1044 		csw->d_strategy(bio);
1045 		dev_relthread(dev, ref);
1046 	} else {
1047 		bio->bio_error = ENXIO;
1048 		ctl_be_block_biodone(bio);
1049 	}
1050 }
1051 
1052 static void
1053 ctl_be_block_unmap_dev_range(struct ctl_be_block_lun *be_lun,
1054 		       struct ctl_be_block_io *beio,
1055 		       uint64_t off, uint64_t len, int last)
1056 {
1057 	struct bio *bio;
1058 	uint64_t maxlen;
1059 	struct cdevsw *csw;
1060 	struct cdev *dev;
1061 	int ref;
1062 
1063 	csw = devvn_refthread(be_lun->vn, &dev, &ref);
1064 	maxlen = LONG_MAX - (LONG_MAX % be_lun->cbe_lun.blocksize);
1065 	while (len > 0) {
1066 		bio = g_alloc_bio();
1067 		bio->bio_cmd	    = BIO_DELETE;
1068 		bio->bio_dev	    = dev;
1069 		bio->bio_offset	    = off;
1070 		bio->bio_length	    = MIN(len, maxlen);
1071 		bio->bio_data	    = 0;
1072 		bio->bio_done	    = ctl_be_block_biodone;
1073 		bio->bio_caller1    = beio;
1074 		bio->bio_pblkno     = off / be_lun->cbe_lun.blocksize;
1075 
1076 		off += bio->bio_length;
1077 		len -= bio->bio_length;
1078 
1079 		mtx_lock(&be_lun->io_lock);
1080 		beio->num_bios_sent++;
1081 		if (last && len == 0)
1082 			beio->send_complete = 1;
1083 		mtx_unlock(&be_lun->io_lock);
1084 
1085 		if (csw) {
1086 			csw->d_strategy(bio);
1087 		} else {
1088 			bio->bio_error = ENXIO;
1089 			ctl_be_block_biodone(bio);
1090 		}
1091 	}
1092 	if (csw)
1093 		dev_relthread(dev, ref);
1094 }
1095 
1096 static void
1097 ctl_be_block_unmap_dev(struct ctl_be_block_lun *be_lun,
1098 		       struct ctl_be_block_io *beio)
1099 {
1100 	union ctl_io *io;
1101 	struct ctl_ptr_len_flags *ptrlen;
1102 	struct scsi_unmap_desc *buf, *end;
1103 	uint64_t len;
1104 
1105 	io = beio->io;
1106 
1107 	DPRINTF("entered\n");
1108 
1109 	binuptime(&beio->ds_t0);
1110 	mtx_lock(&be_lun->io_lock);
1111 	devstat_start_transaction(be_lun->disk_stats, &beio->ds_t0);
1112 	mtx_unlock(&be_lun->io_lock);
1113 
1114 	if (beio->io_offset == -1) {
1115 		beio->io_len = 0;
1116 		ptrlen = (struct ctl_ptr_len_flags *)&io->io_hdr.ctl_private[CTL_PRIV_LBA_LEN];
1117 		buf = (struct scsi_unmap_desc *)ptrlen->ptr;
1118 		end = buf + ptrlen->len / sizeof(*buf);
1119 		for (; buf < end; buf++) {
1120 			len = (uint64_t)scsi_4btoul(buf->length) *
1121 			    be_lun->cbe_lun.blocksize;
1122 			beio->io_len += len;
1123 			ctl_be_block_unmap_dev_range(be_lun, beio,
1124 			    scsi_8btou64(buf->lba) * be_lun->cbe_lun.blocksize,
1125 			    len, (end - buf < 2) ? TRUE : FALSE);
1126 		}
1127 	} else
1128 		ctl_be_block_unmap_dev_range(be_lun, beio,
1129 		    beio->io_offset, beio->io_len, TRUE);
1130 }
1131 
1132 static void
1133 ctl_be_block_dispatch_dev(struct ctl_be_block_lun *be_lun,
1134 			  struct ctl_be_block_io *beio)
1135 {
1136 	TAILQ_HEAD(, bio) queue = TAILQ_HEAD_INITIALIZER(queue);
1137 	struct bio *bio;
1138 	struct cdevsw *csw;
1139 	struct cdev *dev;
1140 	off_t cur_offset;
1141 	int i, max_iosize, ref;
1142 
1143 	DPRINTF("entered\n");
1144 	csw = devvn_refthread(be_lun->vn, &dev, &ref);
1145 
1146 	/*
1147 	 * We have to limit our I/O size to the maximum supported by the
1148 	 * backend device.  Hopefully it is MAXPHYS.  If the driver doesn't
1149 	 * set it properly, use DFLTPHYS.
1150 	 */
1151 	if (csw) {
1152 		max_iosize = dev->si_iosize_max;
1153 		if (max_iosize < PAGE_SIZE)
1154 			max_iosize = DFLTPHYS;
1155 	} else
1156 		max_iosize = DFLTPHYS;
1157 
1158 	cur_offset = beio->io_offset;
1159 	for (i = 0; i < beio->num_segs; i++) {
1160 		size_t cur_size;
1161 		uint8_t *cur_ptr;
1162 
1163 		cur_size = beio->sg_segs[i].len;
1164 		cur_ptr = beio->sg_segs[i].addr;
1165 
1166 		while (cur_size > 0) {
1167 			/* This can't fail, it's a blocking allocation. */
1168 			bio = g_alloc_bio();
1169 
1170 			KASSERT(bio != NULL, ("g_alloc_bio() failed!\n"));
1171 
1172 			bio->bio_cmd = beio->bio_cmd;
1173 			bio->bio_dev = dev;
1174 			bio->bio_caller1 = beio;
1175 			bio->bio_length = min(cur_size, max_iosize);
1176 			bio->bio_offset = cur_offset;
1177 			bio->bio_data = cur_ptr;
1178 			bio->bio_done = ctl_be_block_biodone;
1179 			bio->bio_pblkno = cur_offset / be_lun->cbe_lun.blocksize;
1180 
1181 			cur_offset += bio->bio_length;
1182 			cur_ptr += bio->bio_length;
1183 			cur_size -= bio->bio_length;
1184 
1185 			TAILQ_INSERT_TAIL(&queue, bio, bio_queue);
1186 			beio->num_bios_sent++;
1187 		}
1188 	}
1189 	binuptime(&beio->ds_t0);
1190 	mtx_lock(&be_lun->io_lock);
1191 	devstat_start_transaction(be_lun->disk_stats, &beio->ds_t0);
1192 	beio->send_complete = 1;
1193 	mtx_unlock(&be_lun->io_lock);
1194 
1195 	/*
1196 	 * Fire off all allocated requests!
1197 	 */
1198 	while ((bio = TAILQ_FIRST(&queue)) != NULL) {
1199 		TAILQ_REMOVE(&queue, bio, bio_queue);
1200 		if (csw)
1201 			csw->d_strategy(bio);
1202 		else {
1203 			bio->bio_error = ENXIO;
1204 			ctl_be_block_biodone(bio);
1205 		}
1206 	}
1207 	if (csw)
1208 		dev_relthread(dev, ref);
1209 }
1210 
1211 static uint64_t
1212 ctl_be_block_getattr_dev(struct ctl_be_block_lun *be_lun, const char *attrname)
1213 {
1214 	struct diocgattr_arg	arg;
1215 	struct cdevsw *csw;
1216 	struct cdev *dev;
1217 	int error, ref;
1218 
1219 	csw = devvn_refthread(be_lun->vn, &dev, &ref);
1220 	if (csw == NULL)
1221 		return (UINT64_MAX);
1222 	strlcpy(arg.name, attrname, sizeof(arg.name));
1223 	arg.len = sizeof(arg.value.off);
1224 	if (csw->d_ioctl) {
1225 		error = csw->d_ioctl(dev, DIOCGATTR, (caddr_t)&arg, FREAD,
1226 		    curthread);
1227 	} else
1228 		error = ENODEV;
1229 	dev_relthread(dev, ref);
1230 	if (error != 0)
1231 		return (UINT64_MAX);
1232 	return (arg.value.off);
1233 }
1234 
1235 static void
1236 ctl_be_block_cw_dispatch_sync(struct ctl_be_block_lun *be_lun,
1237 			    union ctl_io *io)
1238 {
1239 	struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun;
1240 	struct ctl_be_block_io *beio;
1241 	struct ctl_lba_len_flags *lbalen;
1242 
1243 	DPRINTF("entered\n");
1244 	beio = (struct ctl_be_block_io *)PRIV(io)->ptr;
1245 	lbalen = (struct ctl_lba_len_flags *)&io->io_hdr.ctl_private[CTL_PRIV_LBA_LEN];
1246 
1247 	beio->io_len = lbalen->len * cbe_lun->blocksize;
1248 	beio->io_offset = lbalen->lba * cbe_lun->blocksize;
1249 	beio->io_arg = (lbalen->flags & SSC_IMMED) != 0;
1250 	beio->bio_cmd = BIO_FLUSH;
1251 	beio->ds_trans_type = DEVSTAT_NO_DATA;
1252 	DPRINTF("SYNC\n");
1253 	be_lun->lun_flush(be_lun, beio);
1254 }
1255 
1256 static void
1257 ctl_be_block_cw_done_ws(struct ctl_be_block_io *beio)
1258 {
1259 	union ctl_io *io;
1260 
1261 	io = beio->io;
1262 	ctl_free_beio(beio);
1263 	if ((io->io_hdr.flags & CTL_FLAG_ABORT) ||
1264 	    ((io->io_hdr.status & CTL_STATUS_MASK) != CTL_STATUS_NONE &&
1265 	     (io->io_hdr.status & CTL_STATUS_MASK) != CTL_SUCCESS)) {
1266 		ctl_config_write_done(io);
1267 		return;
1268 	}
1269 
1270 	ctl_be_block_config_write(io);
1271 }
1272 
1273 static void
1274 ctl_be_block_cw_dispatch_ws(struct ctl_be_block_lun *be_lun,
1275 			    union ctl_io *io)
1276 {
1277 	struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun;
1278 	struct ctl_be_block_io *beio;
1279 	struct ctl_lba_len_flags *lbalen;
1280 	uint64_t len_left, lba;
1281 	uint32_t pb, pbo, adj;
1282 	int i, seglen;
1283 	uint8_t *buf, *end;
1284 
1285 	DPRINTF("entered\n");
1286 
1287 	beio = (struct ctl_be_block_io *)PRIV(io)->ptr;
1288 	lbalen = ARGS(beio->io);
1289 
1290 	if (lbalen->flags & ~(SWS_LBDATA | SWS_UNMAP | SWS_ANCHOR | SWS_NDOB) ||
1291 	    (lbalen->flags & (SWS_UNMAP | SWS_ANCHOR) && be_lun->unmap == NULL)) {
1292 		ctl_free_beio(beio);
1293 		ctl_set_invalid_field(&io->scsiio,
1294 				      /*sks_valid*/ 1,
1295 				      /*command*/ 1,
1296 				      /*field*/ 1,
1297 				      /*bit_valid*/ 0,
1298 				      /*bit*/ 0);
1299 		ctl_config_write_done(io);
1300 		return;
1301 	}
1302 
1303 	if (lbalen->flags & (SWS_UNMAP | SWS_ANCHOR)) {
1304 		beio->io_offset = lbalen->lba * cbe_lun->blocksize;
1305 		beio->io_len = (uint64_t)lbalen->len * cbe_lun->blocksize;
1306 		beio->bio_cmd = BIO_DELETE;
1307 		beio->ds_trans_type = DEVSTAT_FREE;
1308 
1309 		be_lun->unmap(be_lun, beio);
1310 		return;
1311 	}
1312 
1313 	beio->bio_cmd = BIO_WRITE;
1314 	beio->ds_trans_type = DEVSTAT_WRITE;
1315 
1316 	DPRINTF("WRITE SAME at LBA %jx len %u\n",
1317 	       (uintmax_t)lbalen->lba, lbalen->len);
1318 
1319 	pb = cbe_lun->blocksize << be_lun->cbe_lun.pblockexp;
1320 	if (be_lun->cbe_lun.pblockoff > 0)
1321 		pbo = pb - cbe_lun->blocksize * be_lun->cbe_lun.pblockoff;
1322 	else
1323 		pbo = 0;
1324 	len_left = (uint64_t)lbalen->len * cbe_lun->blocksize;
1325 	for (i = 0, lba = 0; i < CTLBLK_MAX_SEGS && len_left > 0; i++) {
1326 
1327 		/*
1328 		 * Setup the S/G entry for this chunk.
1329 		 */
1330 		seglen = MIN(CTLBLK_MAX_SEG, len_left);
1331 		if (pb > cbe_lun->blocksize) {
1332 			adj = ((lbalen->lba + lba) * cbe_lun->blocksize +
1333 			    seglen - pbo) % pb;
1334 			if (seglen > adj)
1335 				seglen -= adj;
1336 			else
1337 				seglen -= seglen % cbe_lun->blocksize;
1338 		} else
1339 			seglen -= seglen % cbe_lun->blocksize;
1340 		beio->sg_segs[i].len = seglen;
1341 		beio->sg_segs[i].addr = uma_zalloc(be_lun->lun_zone, M_WAITOK);
1342 
1343 		DPRINTF("segment %d addr %p len %zd\n", i,
1344 			beio->sg_segs[i].addr, beio->sg_segs[i].len);
1345 
1346 		beio->num_segs++;
1347 		len_left -= seglen;
1348 
1349 		buf = beio->sg_segs[i].addr;
1350 		end = buf + seglen;
1351 		for (; buf < end; buf += cbe_lun->blocksize) {
1352 			if (lbalen->flags & SWS_NDOB) {
1353 				memset(buf, 0, cbe_lun->blocksize);
1354 			} else {
1355 				memcpy(buf, io->scsiio.kern_data_ptr,
1356 				    cbe_lun->blocksize);
1357 			}
1358 			if (lbalen->flags & SWS_LBDATA)
1359 				scsi_ulto4b(lbalen->lba + lba, buf);
1360 			lba++;
1361 		}
1362 	}
1363 
1364 	beio->io_offset = lbalen->lba * cbe_lun->blocksize;
1365 	beio->io_len = lba * cbe_lun->blocksize;
1366 
1367 	/* We can not do all in one run. Correct and schedule rerun. */
1368 	if (len_left > 0) {
1369 		lbalen->lba += lba;
1370 		lbalen->len -= lba;
1371 		beio->beio_cont = ctl_be_block_cw_done_ws;
1372 	}
1373 
1374 	be_lun->dispatch(be_lun, beio);
1375 }
1376 
1377 static void
1378 ctl_be_block_cw_dispatch_unmap(struct ctl_be_block_lun *be_lun,
1379 			    union ctl_io *io)
1380 {
1381 	struct ctl_be_block_io *beio;
1382 	struct ctl_ptr_len_flags *ptrlen;
1383 
1384 	DPRINTF("entered\n");
1385 
1386 	beio = (struct ctl_be_block_io *)PRIV(io)->ptr;
1387 	ptrlen = (struct ctl_ptr_len_flags *)&io->io_hdr.ctl_private[CTL_PRIV_LBA_LEN];
1388 
1389 	if ((ptrlen->flags & ~SU_ANCHOR) != 0 || be_lun->unmap == NULL) {
1390 		ctl_free_beio(beio);
1391 		ctl_set_invalid_field(&io->scsiio,
1392 				      /*sks_valid*/ 0,
1393 				      /*command*/ 1,
1394 				      /*field*/ 0,
1395 				      /*bit_valid*/ 0,
1396 				      /*bit*/ 0);
1397 		ctl_config_write_done(io);
1398 		return;
1399 	}
1400 
1401 	beio->io_len = 0;
1402 	beio->io_offset = -1;
1403 	beio->bio_cmd = BIO_DELETE;
1404 	beio->ds_trans_type = DEVSTAT_FREE;
1405 	DPRINTF("UNMAP\n");
1406 	be_lun->unmap(be_lun, beio);
1407 }
1408 
1409 static void
1410 ctl_be_block_cr_done(struct ctl_be_block_io *beio)
1411 {
1412 	union ctl_io *io;
1413 
1414 	io = beio->io;
1415 	ctl_free_beio(beio);
1416 	ctl_config_read_done(io);
1417 }
1418 
1419 static void
1420 ctl_be_block_cr_dispatch(struct ctl_be_block_lun *be_lun,
1421 			 union ctl_io *io)
1422 {
1423 	struct ctl_be_block_io *beio;
1424 	struct ctl_be_block_softc *softc;
1425 
1426 	DPRINTF("entered\n");
1427 
1428 	softc = be_lun->softc;
1429 	beio = ctl_alloc_beio(softc);
1430 	beio->io = io;
1431 	beio->lun = be_lun;
1432 	beio->beio_cont = ctl_be_block_cr_done;
1433 	PRIV(io)->ptr = (void *)beio;
1434 
1435 	switch (io->scsiio.cdb[0]) {
1436 	case SERVICE_ACTION_IN:		/* GET LBA STATUS */
1437 		beio->bio_cmd = -1;
1438 		beio->ds_trans_type = DEVSTAT_NO_DATA;
1439 		beio->ds_tag_type = DEVSTAT_TAG_ORDERED;
1440 		beio->io_len = 0;
1441 		if (be_lun->get_lba_status)
1442 			be_lun->get_lba_status(be_lun, beio);
1443 		else
1444 			ctl_be_block_cr_done(beio);
1445 		break;
1446 	default:
1447 		panic("Unhandled CDB type %#x", io->scsiio.cdb[0]);
1448 		break;
1449 	}
1450 }
1451 
1452 static void
1453 ctl_be_block_cw_done(struct ctl_be_block_io *beio)
1454 {
1455 	union ctl_io *io;
1456 
1457 	io = beio->io;
1458 	ctl_free_beio(beio);
1459 	ctl_config_write_done(io);
1460 }
1461 
1462 static void
1463 ctl_be_block_cw_dispatch(struct ctl_be_block_lun *be_lun,
1464 			 union ctl_io *io)
1465 {
1466 	struct ctl_be_block_io *beio;
1467 	struct ctl_be_block_softc *softc;
1468 
1469 	DPRINTF("entered\n");
1470 
1471 	softc = be_lun->softc;
1472 	beio = ctl_alloc_beio(softc);
1473 	beio->io = io;
1474 	beio->lun = be_lun;
1475 	beio->beio_cont = ctl_be_block_cw_done;
1476 	switch (io->scsiio.tag_type) {
1477 	case CTL_TAG_ORDERED:
1478 		beio->ds_tag_type = DEVSTAT_TAG_ORDERED;
1479 		break;
1480 	case CTL_TAG_HEAD_OF_QUEUE:
1481 		beio->ds_tag_type = DEVSTAT_TAG_HEAD;
1482 		break;
1483 	case CTL_TAG_UNTAGGED:
1484 	case CTL_TAG_SIMPLE:
1485 	case CTL_TAG_ACA:
1486 	default:
1487 		beio->ds_tag_type = DEVSTAT_TAG_SIMPLE;
1488 		break;
1489 	}
1490 	PRIV(io)->ptr = (void *)beio;
1491 
1492 	switch (io->scsiio.cdb[0]) {
1493 	case SYNCHRONIZE_CACHE:
1494 	case SYNCHRONIZE_CACHE_16:
1495 		ctl_be_block_cw_dispatch_sync(be_lun, io);
1496 		break;
1497 	case WRITE_SAME_10:
1498 	case WRITE_SAME_16:
1499 		ctl_be_block_cw_dispatch_ws(be_lun, io);
1500 		break;
1501 	case UNMAP:
1502 		ctl_be_block_cw_dispatch_unmap(be_lun, io);
1503 		break;
1504 	default:
1505 		panic("Unhandled CDB type %#x", io->scsiio.cdb[0]);
1506 		break;
1507 	}
1508 }
1509 
1510 SDT_PROBE_DEFINE1(cbb, , read, start, "uint64_t");
1511 SDT_PROBE_DEFINE1(cbb, , write, start, "uint64_t");
1512 SDT_PROBE_DEFINE1(cbb, , read, alloc_done, "uint64_t");
1513 SDT_PROBE_DEFINE1(cbb, , write, alloc_done, "uint64_t");
1514 
1515 static void
1516 ctl_be_block_next(struct ctl_be_block_io *beio)
1517 {
1518 	struct ctl_be_block_lun *be_lun;
1519 	union ctl_io *io;
1520 
1521 	io = beio->io;
1522 	be_lun = beio->lun;
1523 	ctl_free_beio(beio);
1524 	if ((io->io_hdr.flags & CTL_FLAG_ABORT) ||
1525 	    ((io->io_hdr.status & CTL_STATUS_MASK) != CTL_STATUS_NONE &&
1526 	     (io->io_hdr.status & CTL_STATUS_MASK) != CTL_SUCCESS)) {
1527 		ctl_data_submit_done(io);
1528 		return;
1529 	}
1530 
1531 	io->io_hdr.status &= ~CTL_STATUS_MASK;
1532 	io->io_hdr.status |= CTL_STATUS_NONE;
1533 
1534 	mtx_lock(&be_lun->queue_lock);
1535 	STAILQ_INSERT_TAIL(&be_lun->input_queue, &io->io_hdr, links);
1536 	mtx_unlock(&be_lun->queue_lock);
1537 	taskqueue_enqueue(be_lun->io_taskqueue, &be_lun->io_task);
1538 }
1539 
1540 static void
1541 ctl_be_block_dispatch(struct ctl_be_block_lun *be_lun,
1542 			   union ctl_io *io)
1543 {
1544 	struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun;
1545 	struct ctl_be_block_io *beio;
1546 	struct ctl_be_block_softc *softc;
1547 	struct ctl_lba_len_flags *lbalen;
1548 	struct ctl_ptr_len_flags *bptrlen;
1549 	uint64_t len_left, lbas;
1550 	int i;
1551 
1552 	softc = be_lun->softc;
1553 
1554 	DPRINTF("entered\n");
1555 
1556 	lbalen = ARGS(io);
1557 	if (lbalen->flags & CTL_LLF_WRITE) {
1558 		SDT_PROBE0(cbb, , write, start);
1559 	} else {
1560 		SDT_PROBE0(cbb, , read, start);
1561 	}
1562 
1563 	beio = ctl_alloc_beio(softc);
1564 	beio->io = io;
1565 	beio->lun = be_lun;
1566 	bptrlen = PRIV(io);
1567 	bptrlen->ptr = (void *)beio;
1568 
1569 	switch (io->scsiio.tag_type) {
1570 	case CTL_TAG_ORDERED:
1571 		beio->ds_tag_type = DEVSTAT_TAG_ORDERED;
1572 		break;
1573 	case CTL_TAG_HEAD_OF_QUEUE:
1574 		beio->ds_tag_type = DEVSTAT_TAG_HEAD;
1575 		break;
1576 	case CTL_TAG_UNTAGGED:
1577 	case CTL_TAG_SIMPLE:
1578 	case CTL_TAG_ACA:
1579 	default:
1580 		beio->ds_tag_type = DEVSTAT_TAG_SIMPLE;
1581 		break;
1582 	}
1583 
1584 	if (lbalen->flags & CTL_LLF_WRITE) {
1585 		beio->bio_cmd = BIO_WRITE;
1586 		beio->ds_trans_type = DEVSTAT_WRITE;
1587 	} else {
1588 		beio->bio_cmd = BIO_READ;
1589 		beio->ds_trans_type = DEVSTAT_READ;
1590 	}
1591 
1592 	DPRINTF("%s at LBA %jx len %u @%ju\n",
1593 	       (beio->bio_cmd == BIO_READ) ? "READ" : "WRITE",
1594 	       (uintmax_t)lbalen->lba, lbalen->len, bptrlen->len);
1595 	if (lbalen->flags & CTL_LLF_COMPARE)
1596 		lbas = CTLBLK_HALF_IO_SIZE;
1597 	else
1598 		lbas = CTLBLK_MAX_IO_SIZE;
1599 	lbas = MIN(lbalen->len - bptrlen->len, lbas / cbe_lun->blocksize);
1600 	beio->io_offset = (lbalen->lba + bptrlen->len) * cbe_lun->blocksize;
1601 	beio->io_len = lbas * cbe_lun->blocksize;
1602 	bptrlen->len += lbas;
1603 
1604 	for (i = 0, len_left = beio->io_len; len_left > 0; i++) {
1605 		KASSERT(i < CTLBLK_MAX_SEGS, ("Too many segs (%d >= %d)",
1606 		    i, CTLBLK_MAX_SEGS));
1607 
1608 		/*
1609 		 * Setup the S/G entry for this chunk.
1610 		 */
1611 		beio->sg_segs[i].len = min(CTLBLK_MAX_SEG, len_left);
1612 		beio->sg_segs[i].addr = uma_zalloc(be_lun->lun_zone, M_WAITOK);
1613 
1614 		DPRINTF("segment %d addr %p len %zd\n", i,
1615 			beio->sg_segs[i].addr, beio->sg_segs[i].len);
1616 
1617 		/* Set up second segment for compare operation. */
1618 		if (lbalen->flags & CTL_LLF_COMPARE) {
1619 			beio->sg_segs[i + CTLBLK_HALF_SEGS].len =
1620 			    beio->sg_segs[i].len;
1621 			beio->sg_segs[i + CTLBLK_HALF_SEGS].addr =
1622 			    uma_zalloc(be_lun->lun_zone, M_WAITOK);
1623 		}
1624 
1625 		beio->num_segs++;
1626 		len_left -= beio->sg_segs[i].len;
1627 	}
1628 	if (bptrlen->len < lbalen->len)
1629 		beio->beio_cont = ctl_be_block_next;
1630 	io->scsiio.be_move_done = ctl_be_block_move_done;
1631 	/* For compare we have separate S/G lists for read and datamove. */
1632 	if (lbalen->flags & CTL_LLF_COMPARE)
1633 		io->scsiio.kern_data_ptr = (uint8_t *)&beio->sg_segs[CTLBLK_HALF_SEGS];
1634 	else
1635 		io->scsiio.kern_data_ptr = (uint8_t *)beio->sg_segs;
1636 	io->scsiio.kern_data_len = beio->io_len;
1637 	io->scsiio.kern_sg_entries = beio->num_segs;
1638 	io->io_hdr.flags |= CTL_FLAG_ALLOCATED;
1639 
1640 	/*
1641 	 * For the read case, we need to read the data into our buffers and
1642 	 * then we can send it back to the user.  For the write case, we
1643 	 * need to get the data from the user first.
1644 	 */
1645 	if (beio->bio_cmd == BIO_READ) {
1646 		SDT_PROBE0(cbb, , read, alloc_done);
1647 		be_lun->dispatch(be_lun, beio);
1648 	} else {
1649 		SDT_PROBE0(cbb, , write, alloc_done);
1650 #ifdef CTL_TIME_IO
1651 		getbinuptime(&io->io_hdr.dma_start_bt);
1652 #endif
1653 		ctl_datamove(io);
1654 	}
1655 }
1656 
1657 static void
1658 ctl_be_block_worker(void *context, int pending)
1659 {
1660 	struct ctl_be_block_lun *be_lun = (struct ctl_be_block_lun *)context;
1661 	struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun;
1662 	union ctl_io *io;
1663 	struct ctl_be_block_io *beio;
1664 
1665 	DPRINTF("entered\n");
1666 	/*
1667 	 * Fetch and process I/Os from all queues.  If we detect LUN
1668 	 * CTL_LUN_FLAG_NO_MEDIA status here -- it is result of a race,
1669 	 * so make response maximally opaque to not confuse initiator.
1670 	 */
1671 	for (;;) {
1672 		mtx_lock(&be_lun->queue_lock);
1673 		io = (union ctl_io *)STAILQ_FIRST(&be_lun->datamove_queue);
1674 		if (io != NULL) {
1675 			DPRINTF("datamove queue\n");
1676 			STAILQ_REMOVE(&be_lun->datamove_queue, &io->io_hdr,
1677 				      ctl_io_hdr, links);
1678 			mtx_unlock(&be_lun->queue_lock);
1679 			beio = (struct ctl_be_block_io *)PRIV(io)->ptr;
1680 			if (cbe_lun->flags & CTL_LUN_FLAG_NO_MEDIA) {
1681 				ctl_set_busy(&io->scsiio);
1682 				ctl_complete_beio(beio);
1683 				return;
1684 			}
1685 			be_lun->dispatch(be_lun, beio);
1686 			continue;
1687 		}
1688 		io = (union ctl_io *)STAILQ_FIRST(&be_lun->config_write_queue);
1689 		if (io != NULL) {
1690 			DPRINTF("config write queue\n");
1691 			STAILQ_REMOVE(&be_lun->config_write_queue, &io->io_hdr,
1692 				      ctl_io_hdr, links);
1693 			mtx_unlock(&be_lun->queue_lock);
1694 			if (cbe_lun->flags & CTL_LUN_FLAG_NO_MEDIA) {
1695 				ctl_set_busy(&io->scsiio);
1696 				ctl_config_write_done(io);
1697 				return;
1698 			}
1699 			ctl_be_block_cw_dispatch(be_lun, io);
1700 			continue;
1701 		}
1702 		io = (union ctl_io *)STAILQ_FIRST(&be_lun->config_read_queue);
1703 		if (io != NULL) {
1704 			DPRINTF("config read queue\n");
1705 			STAILQ_REMOVE(&be_lun->config_read_queue, &io->io_hdr,
1706 				      ctl_io_hdr, links);
1707 			mtx_unlock(&be_lun->queue_lock);
1708 			if (cbe_lun->flags & CTL_LUN_FLAG_NO_MEDIA) {
1709 				ctl_set_busy(&io->scsiio);
1710 				ctl_config_read_done(io);
1711 				return;
1712 			}
1713 			ctl_be_block_cr_dispatch(be_lun, io);
1714 			continue;
1715 		}
1716 		io = (union ctl_io *)STAILQ_FIRST(&be_lun->input_queue);
1717 		if (io != NULL) {
1718 			DPRINTF("input queue\n");
1719 			STAILQ_REMOVE(&be_lun->input_queue, &io->io_hdr,
1720 				      ctl_io_hdr, links);
1721 			mtx_unlock(&be_lun->queue_lock);
1722 			if (cbe_lun->flags & CTL_LUN_FLAG_NO_MEDIA) {
1723 				ctl_set_busy(&io->scsiio);
1724 				ctl_data_submit_done(io);
1725 				return;
1726 			}
1727 			ctl_be_block_dispatch(be_lun, io);
1728 			continue;
1729 		}
1730 
1731 		/*
1732 		 * If we get here, there is no work left in the queues, so
1733 		 * just break out and let the task queue go to sleep.
1734 		 */
1735 		mtx_unlock(&be_lun->queue_lock);
1736 		break;
1737 	}
1738 }
1739 
1740 /*
1741  * Entry point from CTL to the backend for I/O.  We queue everything to a
1742  * work thread, so this just puts the I/O on a queue and wakes up the
1743  * thread.
1744  */
1745 static int
1746 ctl_be_block_submit(union ctl_io *io)
1747 {
1748 	struct ctl_be_block_lun *be_lun;
1749 	struct ctl_be_lun *cbe_lun;
1750 
1751 	DPRINTF("entered\n");
1752 
1753 	cbe_lun = CTL_BACKEND_LUN(io);
1754 	be_lun = (struct ctl_be_block_lun *)cbe_lun->be_lun;
1755 
1756 	/*
1757 	 * Make sure we only get SCSI I/O.
1758 	 */
1759 	KASSERT(io->io_hdr.io_type == CTL_IO_SCSI, ("Non-SCSI I/O (type "
1760 		"%#x) encountered", io->io_hdr.io_type));
1761 
1762 	PRIV(io)->len = 0;
1763 
1764 	mtx_lock(&be_lun->queue_lock);
1765 	STAILQ_INSERT_TAIL(&be_lun->input_queue, &io->io_hdr, links);
1766 	mtx_unlock(&be_lun->queue_lock);
1767 	taskqueue_enqueue(be_lun->io_taskqueue, &be_lun->io_task);
1768 
1769 	return (CTL_RETVAL_COMPLETE);
1770 }
1771 
1772 static int
1773 ctl_be_block_ioctl(struct cdev *dev, u_long cmd, caddr_t addr,
1774 			int flag, struct thread *td)
1775 {
1776 	struct ctl_be_block_softc *softc;
1777 	int error;
1778 
1779 	softc = &backend_block_softc;
1780 
1781 	error = 0;
1782 
1783 	switch (cmd) {
1784 	case CTL_LUN_REQ: {
1785 		struct ctl_lun_req *lun_req;
1786 
1787 		lun_req = (struct ctl_lun_req *)addr;
1788 
1789 		switch (lun_req->reqtype) {
1790 		case CTL_LUNREQ_CREATE:
1791 			error = ctl_be_block_create(softc, lun_req);
1792 			break;
1793 		case CTL_LUNREQ_RM:
1794 			error = ctl_be_block_rm(softc, lun_req);
1795 			break;
1796 		case CTL_LUNREQ_MODIFY:
1797 			error = ctl_be_block_modify(softc, lun_req);
1798 			break;
1799 		default:
1800 			lun_req->status = CTL_LUN_ERROR;
1801 			snprintf(lun_req->error_str, sizeof(lun_req->error_str),
1802 				 "invalid LUN request type %d",
1803 				 lun_req->reqtype);
1804 			break;
1805 		}
1806 		break;
1807 	}
1808 	default:
1809 		error = ENOTTY;
1810 		break;
1811 	}
1812 
1813 	return (error);
1814 }
1815 
1816 static int
1817 ctl_be_block_open_file(struct ctl_be_block_lun *be_lun, struct ctl_lun_req *req)
1818 {
1819 	struct ctl_be_lun *cbe_lun;
1820 	struct ctl_be_block_filedata *file_data;
1821 	struct ctl_lun_create_params *params;
1822 	const char		     *value;
1823 	struct vattr		      vattr;
1824 	off_t			      ps, pss, po, pos, us, uss, uo, uos;
1825 	int			      error;
1826 
1827 	cbe_lun = &be_lun->cbe_lun;
1828 	file_data = &be_lun->backend.file;
1829 	params = &be_lun->params;
1830 
1831 	be_lun->dev_type = CTL_BE_BLOCK_FILE;
1832 	be_lun->dispatch = ctl_be_block_dispatch_file;
1833 	be_lun->lun_flush = ctl_be_block_flush_file;
1834 	be_lun->get_lba_status = ctl_be_block_gls_file;
1835 	be_lun->getattr = ctl_be_block_getattr_file;
1836 	be_lun->unmap = NULL;
1837 	cbe_lun->flags &= ~CTL_LUN_FLAG_UNMAP;
1838 
1839 	error = VOP_GETATTR(be_lun->vn, &vattr, curthread->td_ucred);
1840 	if (error != 0) {
1841 		snprintf(req->error_str, sizeof(req->error_str),
1842 			 "error calling VOP_GETATTR() for file %s",
1843 			 be_lun->dev_path);
1844 		return (error);
1845 	}
1846 
1847 	file_data->cred = crhold(curthread->td_ucred);
1848 	if (params->lun_size_bytes != 0)
1849 		be_lun->size_bytes = params->lun_size_bytes;
1850 	else
1851 		be_lun->size_bytes = vattr.va_size;
1852 
1853 	/*
1854 	 * For files we can use any logical block size.  Prefer 512 bytes
1855 	 * for compatibility reasons.  If file's vattr.va_blocksize
1856 	 * (preferred I/O block size) is bigger and multiple to chosen
1857 	 * logical block size -- report it as physical block size.
1858 	 */
1859 	if (params->blocksize_bytes != 0)
1860 		cbe_lun->blocksize = params->blocksize_bytes;
1861 	else if (cbe_lun->lun_type == T_CDROM)
1862 		cbe_lun->blocksize = 2048;
1863 	else
1864 		cbe_lun->blocksize = 512;
1865 	be_lun->size_blocks = be_lun->size_bytes / cbe_lun->blocksize;
1866 	cbe_lun->maxlba = (be_lun->size_blocks == 0) ?
1867 	    0 : (be_lun->size_blocks - 1);
1868 
1869 	us = ps = vattr.va_blocksize;
1870 	uo = po = 0;
1871 
1872 	value = dnvlist_get_string(cbe_lun->options, "pblocksize", NULL);
1873 	if (value != NULL)
1874 		ctl_expand_number(value, &ps);
1875 	value = dnvlist_get_string(cbe_lun->options, "pblockoffset", NULL);
1876 	if (value != NULL)
1877 		ctl_expand_number(value, &po);
1878 	pss = ps / cbe_lun->blocksize;
1879 	pos = po / cbe_lun->blocksize;
1880 	if ((pss > 0) && (pss * cbe_lun->blocksize == ps) && (pss >= pos) &&
1881 	    ((pss & (pss - 1)) == 0) && (pos * cbe_lun->blocksize == po)) {
1882 		cbe_lun->pblockexp = fls(pss) - 1;
1883 		cbe_lun->pblockoff = (pss - pos) % pss;
1884 	}
1885 
1886 	value = dnvlist_get_string(cbe_lun->options, "ublocksize", NULL);
1887 	if (value != NULL)
1888 		ctl_expand_number(value, &us);
1889 	value = dnvlist_get_string(cbe_lun->options, "ublockoffset", NULL);
1890 	if (value != NULL)
1891 		ctl_expand_number(value, &uo);
1892 	uss = us / cbe_lun->blocksize;
1893 	uos = uo / cbe_lun->blocksize;
1894 	if ((uss > 0) && (uss * cbe_lun->blocksize == us) && (uss >= uos) &&
1895 	    ((uss & (uss - 1)) == 0) && (uos * cbe_lun->blocksize == uo)) {
1896 		cbe_lun->ublockexp = fls(uss) - 1;
1897 		cbe_lun->ublockoff = (uss - uos) % uss;
1898 	}
1899 
1900 	/*
1901 	 * Sanity check.  The media size has to be at least one
1902 	 * sector long.
1903 	 */
1904 	if (be_lun->size_bytes < cbe_lun->blocksize) {
1905 		error = EINVAL;
1906 		snprintf(req->error_str, sizeof(req->error_str),
1907 			 "file %s size %ju < block size %u", be_lun->dev_path,
1908 			 (uintmax_t)be_lun->size_bytes, cbe_lun->blocksize);
1909 	}
1910 
1911 	cbe_lun->opttxferlen = CTLBLK_MAX_IO_SIZE / cbe_lun->blocksize;
1912 	return (error);
1913 }
1914 
1915 static int
1916 ctl_be_block_open_dev(struct ctl_be_block_lun *be_lun, struct ctl_lun_req *req)
1917 {
1918 	struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun;
1919 	struct ctl_lun_create_params *params;
1920 	struct cdevsw		     *csw;
1921 	struct cdev		     *dev;
1922 	const char		     *value;
1923 	int			      error, atomic, maxio, ref, unmap, tmp;
1924 	off_t			      ps, pss, po, pos, us, uss, uo, uos, otmp;
1925 
1926 	params = &be_lun->params;
1927 
1928 	be_lun->dev_type = CTL_BE_BLOCK_DEV;
1929 	csw = devvn_refthread(be_lun->vn, &dev, &ref);
1930 	if (csw == NULL)
1931 		return (ENXIO);
1932 	if (strcmp(csw->d_name, "zvol") == 0) {
1933 		be_lun->dispatch = ctl_be_block_dispatch_zvol;
1934 		be_lun->get_lba_status = ctl_be_block_gls_zvol;
1935 		atomic = maxio = CTLBLK_MAX_IO_SIZE;
1936 	} else {
1937 		be_lun->dispatch = ctl_be_block_dispatch_dev;
1938 		be_lun->get_lba_status = NULL;
1939 		atomic = 0;
1940 		maxio = dev->si_iosize_max;
1941 		if (maxio <= 0)
1942 			maxio = DFLTPHYS;
1943 		if (maxio > CTLBLK_MAX_IO_SIZE)
1944 			maxio = CTLBLK_MAX_IO_SIZE;
1945 	}
1946 	be_lun->lun_flush = ctl_be_block_flush_dev;
1947 	be_lun->getattr = ctl_be_block_getattr_dev;
1948 	be_lun->unmap = ctl_be_block_unmap_dev;
1949 
1950 	if (!csw->d_ioctl) {
1951 		dev_relthread(dev, ref);
1952 		snprintf(req->error_str, sizeof(req->error_str),
1953 			 "no d_ioctl for device %s!", be_lun->dev_path);
1954 		return (ENODEV);
1955 	}
1956 
1957 	error = csw->d_ioctl(dev, DIOCGSECTORSIZE, (caddr_t)&tmp, FREAD,
1958 			       curthread);
1959 	if (error) {
1960 		dev_relthread(dev, ref);
1961 		snprintf(req->error_str, sizeof(req->error_str),
1962 			 "error %d returned for DIOCGSECTORSIZE ioctl "
1963 			 "on %s!", error, be_lun->dev_path);
1964 		return (error);
1965 	}
1966 
1967 	/*
1968 	 * If the user has asked for a blocksize that is greater than the
1969 	 * backing device's blocksize, we can do it only if the blocksize
1970 	 * the user is asking for is an even multiple of the underlying
1971 	 * device's blocksize.
1972 	 */
1973 	if ((params->blocksize_bytes != 0) &&
1974 	    (params->blocksize_bytes >= tmp)) {
1975 		if (params->blocksize_bytes % tmp == 0) {
1976 			cbe_lun->blocksize = params->blocksize_bytes;
1977 		} else {
1978 			dev_relthread(dev, ref);
1979 			snprintf(req->error_str, sizeof(req->error_str),
1980 				 "requested blocksize %u is not an even "
1981 				 "multiple of backing device blocksize %u",
1982 				 params->blocksize_bytes, tmp);
1983 			return (EINVAL);
1984 		}
1985 	} else if (params->blocksize_bytes != 0) {
1986 		dev_relthread(dev, ref);
1987 		snprintf(req->error_str, sizeof(req->error_str),
1988 			 "requested blocksize %u < backing device "
1989 			 "blocksize %u", params->blocksize_bytes, tmp);
1990 		return (EINVAL);
1991 	} else if (cbe_lun->lun_type == T_CDROM)
1992 		cbe_lun->blocksize = MAX(tmp, 2048);
1993 	else
1994 		cbe_lun->blocksize = tmp;
1995 
1996 	error = csw->d_ioctl(dev, DIOCGMEDIASIZE, (caddr_t)&otmp, FREAD,
1997 			     curthread);
1998 	if (error) {
1999 		dev_relthread(dev, ref);
2000 		snprintf(req->error_str, sizeof(req->error_str),
2001 			 "error %d returned for DIOCGMEDIASIZE "
2002 			 " ioctl on %s!", error,
2003 			 be_lun->dev_path);
2004 		return (error);
2005 	}
2006 
2007 	if (params->lun_size_bytes != 0) {
2008 		if (params->lun_size_bytes > otmp) {
2009 			dev_relthread(dev, ref);
2010 			snprintf(req->error_str, sizeof(req->error_str),
2011 				 "requested LUN size %ju > backing device "
2012 				 "size %ju",
2013 				 (uintmax_t)params->lun_size_bytes,
2014 				 (uintmax_t)otmp);
2015 			return (EINVAL);
2016 		}
2017 
2018 		be_lun->size_bytes = params->lun_size_bytes;
2019 	} else
2020 		be_lun->size_bytes = otmp;
2021 	be_lun->size_blocks = be_lun->size_bytes / cbe_lun->blocksize;
2022 	cbe_lun->maxlba = (be_lun->size_blocks == 0) ?
2023 	    0 : (be_lun->size_blocks - 1);
2024 
2025 	error = csw->d_ioctl(dev, DIOCGSTRIPESIZE, (caddr_t)&ps, FREAD,
2026 	    curthread);
2027 	if (error)
2028 		ps = po = 0;
2029 	else {
2030 		error = csw->d_ioctl(dev, DIOCGSTRIPEOFFSET, (caddr_t)&po,
2031 		    FREAD, curthread);
2032 		if (error)
2033 			po = 0;
2034 	}
2035 	us = ps;
2036 	uo = po;
2037 
2038 	value = dnvlist_get_string(cbe_lun->options, "pblocksize", NULL);
2039 	if (value != NULL)
2040 		ctl_expand_number(value, &ps);
2041 	value = dnvlist_get_string(cbe_lun->options, "pblockoffset", NULL);
2042 	if (value != NULL)
2043 		ctl_expand_number(value, &po);
2044 	pss = ps / cbe_lun->blocksize;
2045 	pos = po / cbe_lun->blocksize;
2046 	if ((pss > 0) && (pss * cbe_lun->blocksize == ps) && (pss >= pos) &&
2047 	    ((pss & (pss - 1)) == 0) && (pos * cbe_lun->blocksize == po)) {
2048 		cbe_lun->pblockexp = fls(pss) - 1;
2049 		cbe_lun->pblockoff = (pss - pos) % pss;
2050 	}
2051 
2052 	value = dnvlist_get_string(cbe_lun->options, "ublocksize", NULL);
2053 	if (value != NULL)
2054 		ctl_expand_number(value, &us);
2055 	value = dnvlist_get_string(cbe_lun->options, "ublockoffset", NULL);
2056 	if (value != NULL)
2057 		ctl_expand_number(value, &uo);
2058 	uss = us / cbe_lun->blocksize;
2059 	uos = uo / cbe_lun->blocksize;
2060 	if ((uss > 0) && (uss * cbe_lun->blocksize == us) && (uss >= uos) &&
2061 	    ((uss & (uss - 1)) == 0) && (uos * cbe_lun->blocksize == uo)) {
2062 		cbe_lun->ublockexp = fls(uss) - 1;
2063 		cbe_lun->ublockoff = (uss - uos) % uss;
2064 	}
2065 
2066 	cbe_lun->atomicblock = atomic / cbe_lun->blocksize;
2067 	cbe_lun->opttxferlen = maxio / cbe_lun->blocksize;
2068 
2069 	if (be_lun->dispatch == ctl_be_block_dispatch_zvol) {
2070 		unmap = 1;
2071 	} else {
2072 		struct diocgattr_arg	arg;
2073 
2074 		strlcpy(arg.name, "GEOM::candelete", sizeof(arg.name));
2075 		arg.len = sizeof(arg.value.i);
2076 		error = csw->d_ioctl(dev, DIOCGATTR, (caddr_t)&arg, FREAD,
2077 		    curthread);
2078 		unmap = (error == 0) ? arg.value.i : 0;
2079 	}
2080 	value = dnvlist_get_string(cbe_lun->options, "unmap", NULL);
2081 	if (value != NULL)
2082 		unmap = (strcmp(value, "on") == 0);
2083 	if (unmap)
2084 		cbe_lun->flags |= CTL_LUN_FLAG_UNMAP;
2085 	else
2086 		cbe_lun->flags &= ~CTL_LUN_FLAG_UNMAP;
2087 
2088 	dev_relthread(dev, ref);
2089 	return (0);
2090 }
2091 
2092 static int
2093 ctl_be_block_close(struct ctl_be_block_lun *be_lun)
2094 {
2095 	struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun;
2096 	int flags;
2097 
2098 	if (be_lun->vn) {
2099 		flags = FREAD;
2100 		if ((cbe_lun->flags & CTL_LUN_FLAG_READONLY) == 0)
2101 			flags |= FWRITE;
2102 		(void)vn_close(be_lun->vn, flags, NOCRED, curthread);
2103 		be_lun->vn = NULL;
2104 
2105 		switch (be_lun->dev_type) {
2106 		case CTL_BE_BLOCK_DEV:
2107 			break;
2108 		case CTL_BE_BLOCK_FILE:
2109 			if (be_lun->backend.file.cred != NULL) {
2110 				crfree(be_lun->backend.file.cred);
2111 				be_lun->backend.file.cred = NULL;
2112 			}
2113 			break;
2114 		case CTL_BE_BLOCK_NONE:
2115 			break;
2116 		default:
2117 			panic("Unexpected backend type %d", be_lun->dev_type);
2118 			break;
2119 		}
2120 		be_lun->dev_type = CTL_BE_BLOCK_NONE;
2121 	}
2122 	return (0);
2123 }
2124 
2125 static int
2126 ctl_be_block_open(struct ctl_be_block_lun *be_lun, struct ctl_lun_req *req)
2127 {
2128 	struct ctl_be_lun *cbe_lun = &be_lun->cbe_lun;
2129 	struct nameidata nd;
2130 	const char	*value;
2131 	int		 error, flags;
2132 
2133 	error = 0;
2134 	if (rootvnode == NULL) {
2135 		snprintf(req->error_str, sizeof(req->error_str),
2136 			 "Root filesystem is not mounted");
2137 		return (1);
2138 	}
2139 	pwd_ensure_dirs();
2140 
2141 	value = dnvlist_get_string(cbe_lun->options, "file", NULL);
2142 	if (value == NULL) {
2143 		snprintf(req->error_str, sizeof(req->error_str),
2144 			 "no file argument specified");
2145 		return (1);
2146 	}
2147 	free(be_lun->dev_path, M_CTLBLK);
2148 	be_lun->dev_path = strdup(value, M_CTLBLK);
2149 
2150 	flags = FREAD;
2151 	value = dnvlist_get_string(cbe_lun->options, "readonly", NULL);
2152 	if (value != NULL) {
2153 		if (strcmp(value, "on") != 0)
2154 			flags |= FWRITE;
2155 	} else if (cbe_lun->lun_type == T_DIRECT)
2156 		flags |= FWRITE;
2157 
2158 again:
2159 	NDINIT(&nd, LOOKUP, FOLLOW, UIO_SYSSPACE, be_lun->dev_path, curthread);
2160 	error = vn_open(&nd, &flags, 0, NULL);
2161 	if ((error == EROFS || error == EACCES) && (flags & FWRITE)) {
2162 		flags &= ~FWRITE;
2163 		goto again;
2164 	}
2165 	if (error) {
2166 		/*
2167 		 * This is the only reasonable guess we can make as far as
2168 		 * path if the user doesn't give us a fully qualified path.
2169 		 * If they want to specify a file, they need to specify the
2170 		 * full path.
2171 		 */
2172 		if (be_lun->dev_path[0] != '/') {
2173 			char *dev_name;
2174 
2175 			asprintf(&dev_name, M_CTLBLK, "/dev/%s",
2176 				be_lun->dev_path);
2177 			free(be_lun->dev_path, M_CTLBLK);
2178 			be_lun->dev_path = dev_name;
2179 			goto again;
2180 		}
2181 		snprintf(req->error_str, sizeof(req->error_str),
2182 		    "error opening %s: %d", be_lun->dev_path, error);
2183 		return (error);
2184 	}
2185 	if (flags & FWRITE)
2186 		cbe_lun->flags &= ~CTL_LUN_FLAG_READONLY;
2187 	else
2188 		cbe_lun->flags |= CTL_LUN_FLAG_READONLY;
2189 
2190 	NDFREE(&nd, NDF_ONLY_PNBUF);
2191 	be_lun->vn = nd.ni_vp;
2192 
2193 	/* We only support disks and files. */
2194 	if (vn_isdisk(be_lun->vn, &error)) {
2195 		error = ctl_be_block_open_dev(be_lun, req);
2196 	} else if (be_lun->vn->v_type == VREG) {
2197 		error = ctl_be_block_open_file(be_lun, req);
2198 	} else {
2199 		error = EINVAL;
2200 		snprintf(req->error_str, sizeof(req->error_str),
2201 			 "%s is not a disk or plain file", be_lun->dev_path);
2202 	}
2203 	VOP_UNLOCK(be_lun->vn, 0);
2204 
2205 	if (error != 0)
2206 		ctl_be_block_close(be_lun);
2207 	cbe_lun->serseq = CTL_LUN_SERSEQ_OFF;
2208 	if (be_lun->dispatch != ctl_be_block_dispatch_dev)
2209 		cbe_lun->serseq = CTL_LUN_SERSEQ_READ;
2210 	value = dnvlist_get_string(cbe_lun->options, "serseq", NULL);
2211 	if (value != NULL && strcmp(value, "on") == 0)
2212 		cbe_lun->serseq = CTL_LUN_SERSEQ_ON;
2213 	else if (value != NULL && strcmp(value, "read") == 0)
2214 		cbe_lun->serseq = CTL_LUN_SERSEQ_READ;
2215 	else if (value != NULL && strcmp(value, "off") == 0)
2216 		cbe_lun->serseq = CTL_LUN_SERSEQ_OFF;
2217 	return (0);
2218 }
2219 
2220 static int
2221 ctl_be_block_create(struct ctl_be_block_softc *softc, struct ctl_lun_req *req)
2222 {
2223 	struct ctl_be_lun *cbe_lun;
2224 	struct ctl_be_block_lun *be_lun;
2225 	struct ctl_lun_create_params *params;
2226 	char num_thread_str[16];
2227 	char tmpstr[32];
2228 	const char *value;
2229 	int retval, num_threads;
2230 	int tmp_num_threads;
2231 
2232 	params = &req->reqdata.create;
2233 	retval = 0;
2234 	req->status = CTL_LUN_OK;
2235 
2236 	be_lun = malloc(sizeof(*be_lun), M_CTLBLK, M_ZERO | M_WAITOK);
2237 	cbe_lun = &be_lun->cbe_lun;
2238 	cbe_lun->be_lun = be_lun;
2239 	be_lun->params = req->reqdata.create;
2240 	be_lun->softc = softc;
2241 	STAILQ_INIT(&be_lun->input_queue);
2242 	STAILQ_INIT(&be_lun->config_read_queue);
2243 	STAILQ_INIT(&be_lun->config_write_queue);
2244 	STAILQ_INIT(&be_lun->datamove_queue);
2245 	sprintf(be_lun->lunname, "cblk%d", softc->num_luns);
2246 	mtx_init(&be_lun->io_lock, "cblk io lock", NULL, MTX_DEF);
2247 	mtx_init(&be_lun->queue_lock, "cblk queue lock", NULL, MTX_DEF);
2248 	cbe_lun->options = nvlist_clone(req->args_nvl);
2249 	be_lun->lun_zone = uma_zcreate(be_lun->lunname, CTLBLK_MAX_SEG,
2250 	    NULL, NULL, NULL, NULL, /*align*/ 0, /*flags*/0);
2251 	if (be_lun->lun_zone == NULL) {
2252 		snprintf(req->error_str, sizeof(req->error_str),
2253 			 "error allocating UMA zone");
2254 		goto bailout_error;
2255 	}
2256 
2257 	if (params->flags & CTL_LUN_FLAG_DEV_TYPE)
2258 		cbe_lun->lun_type = params->device_type;
2259 	else
2260 		cbe_lun->lun_type = T_DIRECT;
2261 	be_lun->flags = CTL_BE_BLOCK_LUN_UNCONFIGURED;
2262 	cbe_lun->flags = 0;
2263 	value = dnvlist_get_string(cbe_lun->options, "ha_role", NULL);
2264 	if (value != NULL) {
2265 		if (strcmp(value, "primary") == 0)
2266 			cbe_lun->flags |= CTL_LUN_FLAG_PRIMARY;
2267 	} else if (control_softc->flags & CTL_FLAG_ACTIVE_SHELF)
2268 		cbe_lun->flags |= CTL_LUN_FLAG_PRIMARY;
2269 
2270 	if (cbe_lun->lun_type == T_DIRECT ||
2271 	    cbe_lun->lun_type == T_CDROM) {
2272 		be_lun->size_bytes = params->lun_size_bytes;
2273 		if (params->blocksize_bytes != 0)
2274 			cbe_lun->blocksize = params->blocksize_bytes;
2275 		else if (cbe_lun->lun_type == T_CDROM)
2276 			cbe_lun->blocksize = 2048;
2277 		else
2278 			cbe_lun->blocksize = 512;
2279 		be_lun->size_blocks = be_lun->size_bytes / cbe_lun->blocksize;
2280 		cbe_lun->maxlba = (be_lun->size_blocks == 0) ?
2281 		    0 : (be_lun->size_blocks - 1);
2282 
2283 		if ((cbe_lun->flags & CTL_LUN_FLAG_PRIMARY) ||
2284 		    control_softc->ha_mode == CTL_HA_MODE_SER_ONLY) {
2285 			retval = ctl_be_block_open(be_lun, req);
2286 			if (retval != 0) {
2287 				retval = 0;
2288 				req->status = CTL_LUN_WARNING;
2289 			}
2290 		}
2291 		num_threads = cbb_num_threads;
2292 	} else {
2293 		num_threads = 1;
2294 	}
2295 
2296 	value = dnvlist_get_string(cbe_lun->options, "num_threads", NULL);
2297 	if (value != NULL) {
2298 		tmp_num_threads = strtol(value, NULL, 0);
2299 
2300 		/*
2301 		 * We don't let the user specify less than one
2302 		 * thread, but hope he's clueful enough not to
2303 		 * specify 1000 threads.
2304 		 */
2305 		if (tmp_num_threads < 1) {
2306 			snprintf(req->error_str, sizeof(req->error_str),
2307 				 "invalid number of threads %s",
2308 				 num_thread_str);
2309 			goto bailout_error;
2310 		}
2311 		num_threads = tmp_num_threads;
2312 	}
2313 
2314 	if (be_lun->vn == NULL)
2315 		cbe_lun->flags |= CTL_LUN_FLAG_NO_MEDIA;
2316 	/* Tell the user the blocksize we ended up using */
2317 	params->lun_size_bytes = be_lun->size_bytes;
2318 	params->blocksize_bytes = cbe_lun->blocksize;
2319 	if (params->flags & CTL_LUN_FLAG_ID_REQ) {
2320 		cbe_lun->req_lun_id = params->req_lun_id;
2321 		cbe_lun->flags |= CTL_LUN_FLAG_ID_REQ;
2322 	} else
2323 		cbe_lun->req_lun_id = 0;
2324 
2325 	cbe_lun->lun_shutdown = ctl_be_block_lun_shutdown;
2326 	cbe_lun->lun_config_status = ctl_be_block_lun_config_status;
2327 	cbe_lun->be = &ctl_be_block_driver;
2328 
2329 	if ((params->flags & CTL_LUN_FLAG_SERIAL_NUM) == 0) {
2330 		snprintf(tmpstr, sizeof(tmpstr), "MYSERIAL%04d",
2331 			 softc->num_luns);
2332 		strncpy((char *)cbe_lun->serial_num, tmpstr,
2333 			MIN(sizeof(cbe_lun->serial_num), sizeof(tmpstr)));
2334 
2335 		/* Tell the user what we used for a serial number */
2336 		strncpy((char *)params->serial_num, tmpstr,
2337 			MIN(sizeof(params->serial_num), sizeof(tmpstr)));
2338 	} else {
2339 		strncpy((char *)cbe_lun->serial_num, params->serial_num,
2340 			MIN(sizeof(cbe_lun->serial_num),
2341 			sizeof(params->serial_num)));
2342 	}
2343 	if ((params->flags & CTL_LUN_FLAG_DEVID) == 0) {
2344 		snprintf(tmpstr, sizeof(tmpstr), "MYDEVID%04d", softc->num_luns);
2345 		strncpy((char *)cbe_lun->device_id, tmpstr,
2346 			MIN(sizeof(cbe_lun->device_id), sizeof(tmpstr)));
2347 
2348 		/* Tell the user what we used for a device ID */
2349 		strncpy((char *)params->device_id, tmpstr,
2350 			MIN(sizeof(params->device_id), sizeof(tmpstr)));
2351 	} else {
2352 		strncpy((char *)cbe_lun->device_id, params->device_id,
2353 			MIN(sizeof(cbe_lun->device_id),
2354 			    sizeof(params->device_id)));
2355 	}
2356 
2357 	TASK_INIT(&be_lun->io_task, /*priority*/0, ctl_be_block_worker, be_lun);
2358 
2359 	be_lun->io_taskqueue = taskqueue_create(be_lun->lunname, M_WAITOK,
2360 	    taskqueue_thread_enqueue, /*context*/&be_lun->io_taskqueue);
2361 
2362 	if (be_lun->io_taskqueue == NULL) {
2363 		snprintf(req->error_str, sizeof(req->error_str),
2364 			 "unable to create taskqueue");
2365 		goto bailout_error;
2366 	}
2367 
2368 	/*
2369 	 * Note that we start the same number of threads by default for
2370 	 * both the file case and the block device case.  For the file
2371 	 * case, we need multiple threads to allow concurrency, because the
2372 	 * vnode interface is designed to be a blocking interface.  For the
2373 	 * block device case, ZFS zvols at least will block the caller's
2374 	 * context in many instances, and so we need multiple threads to
2375 	 * overcome that problem.  Other block devices don't need as many
2376 	 * threads, but they shouldn't cause too many problems.
2377 	 *
2378 	 * If the user wants to just have a single thread for a block
2379 	 * device, he can specify that when the LUN is created, or change
2380 	 * the tunable/sysctl to alter the default number of threads.
2381 	 */
2382 	retval = taskqueue_start_threads(&be_lun->io_taskqueue,
2383 					 /*num threads*/num_threads,
2384 					 /*priority*/PUSER,
2385 					 /*thread name*/
2386 					 "%s taskq", be_lun->lunname);
2387 
2388 	if (retval != 0)
2389 		goto bailout_error;
2390 
2391 	be_lun->num_threads = num_threads;
2392 
2393 	mtx_lock(&softc->lock);
2394 	softc->num_luns++;
2395 	STAILQ_INSERT_TAIL(&softc->lun_list, be_lun, links);
2396 
2397 	mtx_unlock(&softc->lock);
2398 
2399 	retval = ctl_add_lun(&be_lun->cbe_lun);
2400 	if (retval != 0) {
2401 		mtx_lock(&softc->lock);
2402 		STAILQ_REMOVE(&softc->lun_list, be_lun, ctl_be_block_lun,
2403 			      links);
2404 		softc->num_luns--;
2405 		mtx_unlock(&softc->lock);
2406 		snprintf(req->error_str, sizeof(req->error_str),
2407 			 "ctl_add_lun() returned error %d, see dmesg for "
2408 			 "details", retval);
2409 		retval = 0;
2410 		goto bailout_error;
2411 	}
2412 
2413 	mtx_lock(&softc->lock);
2414 
2415 	/*
2416 	 * Tell the config_status routine that we're waiting so it won't
2417 	 * clean up the LUN in the event of an error.
2418 	 */
2419 	be_lun->flags |= CTL_BE_BLOCK_LUN_WAITING;
2420 
2421 	while (be_lun->flags & CTL_BE_BLOCK_LUN_UNCONFIGURED) {
2422 		retval = msleep(be_lun, &softc->lock, PCATCH, "ctlblk", 0);
2423 		if (retval == EINTR)
2424 			break;
2425 	}
2426 	be_lun->flags &= ~CTL_BE_BLOCK_LUN_WAITING;
2427 
2428 	if (be_lun->flags & CTL_BE_BLOCK_LUN_CONFIG_ERR) {
2429 		snprintf(req->error_str, sizeof(req->error_str),
2430 			 "LUN configuration error, see dmesg for details");
2431 		STAILQ_REMOVE(&softc->lun_list, be_lun, ctl_be_block_lun,
2432 			      links);
2433 		softc->num_luns--;
2434 		mtx_unlock(&softc->lock);
2435 		goto bailout_error;
2436 	} else {
2437 		params->req_lun_id = cbe_lun->lun_id;
2438 	}
2439 
2440 	mtx_unlock(&softc->lock);
2441 
2442 	be_lun->disk_stats = devstat_new_entry("cbb", params->req_lun_id,
2443 					       cbe_lun->blocksize,
2444 					       DEVSTAT_ALL_SUPPORTED,
2445 					       cbe_lun->lun_type
2446 					       | DEVSTAT_TYPE_IF_OTHER,
2447 					       DEVSTAT_PRIORITY_OTHER);
2448 
2449 	return (retval);
2450 
2451 bailout_error:
2452 	req->status = CTL_LUN_ERROR;
2453 
2454 	if (be_lun->io_taskqueue != NULL)
2455 		taskqueue_free(be_lun->io_taskqueue);
2456 	ctl_be_block_close(be_lun);
2457 	if (be_lun->dev_path != NULL)
2458 		free(be_lun->dev_path, M_CTLBLK);
2459 	if (be_lun->lun_zone != NULL)
2460 		uma_zdestroy(be_lun->lun_zone);
2461 	nvlist_destroy(cbe_lun->options);
2462 	mtx_destroy(&be_lun->queue_lock);
2463 	mtx_destroy(&be_lun->io_lock);
2464 	free(be_lun, M_CTLBLK);
2465 
2466 	return (retval);
2467 }
2468 
2469 static int
2470 ctl_be_block_rm(struct ctl_be_block_softc *softc, struct ctl_lun_req *req)
2471 {
2472 	struct ctl_lun_rm_params *params;
2473 	struct ctl_be_block_lun *be_lun;
2474 	struct ctl_be_lun *cbe_lun;
2475 	int retval;
2476 
2477 	params = &req->reqdata.rm;
2478 
2479 	mtx_lock(&softc->lock);
2480 	STAILQ_FOREACH(be_lun, &softc->lun_list, links) {
2481 		if (be_lun->cbe_lun.lun_id == params->lun_id)
2482 			break;
2483 	}
2484 	mtx_unlock(&softc->lock);
2485 	if (be_lun == NULL) {
2486 		snprintf(req->error_str, sizeof(req->error_str),
2487 			 "LUN %u is not managed by the block backend",
2488 			 params->lun_id);
2489 		goto bailout_error;
2490 	}
2491 	cbe_lun = &be_lun->cbe_lun;
2492 
2493 	retval = ctl_disable_lun(cbe_lun);
2494 	if (retval != 0) {
2495 		snprintf(req->error_str, sizeof(req->error_str),
2496 			 "error %d returned from ctl_disable_lun() for "
2497 			 "LUN %d", retval, params->lun_id);
2498 		goto bailout_error;
2499 	}
2500 
2501 	if (be_lun->vn != NULL) {
2502 		cbe_lun->flags |= CTL_LUN_FLAG_NO_MEDIA;
2503 		ctl_lun_no_media(cbe_lun);
2504 		taskqueue_drain_all(be_lun->io_taskqueue);
2505 		ctl_be_block_close(be_lun);
2506 	}
2507 
2508 	retval = ctl_invalidate_lun(cbe_lun);
2509 	if (retval != 0) {
2510 		snprintf(req->error_str, sizeof(req->error_str),
2511 			 "error %d returned from ctl_invalidate_lun() for "
2512 			 "LUN %d", retval, params->lun_id);
2513 		goto bailout_error;
2514 	}
2515 
2516 	mtx_lock(&softc->lock);
2517 	be_lun->flags |= CTL_BE_BLOCK_LUN_WAITING;
2518 	while ((be_lun->flags & CTL_BE_BLOCK_LUN_UNCONFIGURED) == 0) {
2519                 retval = msleep(be_lun, &softc->lock, PCATCH, "ctlblk", 0);
2520                 if (retval == EINTR)
2521                         break;
2522         }
2523 	be_lun->flags &= ~CTL_BE_BLOCK_LUN_WAITING;
2524 
2525 	if ((be_lun->flags & CTL_BE_BLOCK_LUN_UNCONFIGURED) == 0) {
2526 		snprintf(req->error_str, sizeof(req->error_str),
2527 			 "interrupted waiting for LUN to be freed");
2528 		mtx_unlock(&softc->lock);
2529 		goto bailout_error;
2530 	}
2531 
2532 	STAILQ_REMOVE(&softc->lun_list, be_lun, ctl_be_block_lun, links);
2533 
2534 	softc->num_luns--;
2535 	mtx_unlock(&softc->lock);
2536 
2537 	taskqueue_drain_all(be_lun->io_taskqueue);
2538 	taskqueue_free(be_lun->io_taskqueue);
2539 
2540 	if (be_lun->disk_stats != NULL)
2541 		devstat_remove_entry(be_lun->disk_stats);
2542 
2543 	uma_zdestroy(be_lun->lun_zone);
2544 
2545 	nvlist_destroy(cbe_lun->options);
2546 	free(be_lun->dev_path, M_CTLBLK);
2547 	mtx_destroy(&be_lun->queue_lock);
2548 	mtx_destroy(&be_lun->io_lock);
2549 	free(be_lun, M_CTLBLK);
2550 
2551 	req->status = CTL_LUN_OK;
2552 	return (0);
2553 
2554 bailout_error:
2555 	req->status = CTL_LUN_ERROR;
2556 	return (0);
2557 }
2558 
2559 static int
2560 ctl_be_block_modify(struct ctl_be_block_softc *softc, struct ctl_lun_req *req)
2561 {
2562 	struct ctl_lun_modify_params *params;
2563 	struct ctl_be_block_lun *be_lun;
2564 	struct ctl_be_lun *cbe_lun;
2565 	const char *value;
2566 	uint64_t oldsize;
2567 	int error, wasprim;
2568 
2569 	params = &req->reqdata.modify;
2570 
2571 	mtx_lock(&softc->lock);
2572 	STAILQ_FOREACH(be_lun, &softc->lun_list, links) {
2573 		if (be_lun->cbe_lun.lun_id == params->lun_id)
2574 			break;
2575 	}
2576 	mtx_unlock(&softc->lock);
2577 	if (be_lun == NULL) {
2578 		snprintf(req->error_str, sizeof(req->error_str),
2579 			 "LUN %u is not managed by the block backend",
2580 			 params->lun_id);
2581 		goto bailout_error;
2582 	}
2583 	cbe_lun = &be_lun->cbe_lun;
2584 
2585 	if (params->lun_size_bytes != 0)
2586 		be_lun->params.lun_size_bytes = params->lun_size_bytes;
2587 
2588 	nvlist_destroy(cbe_lun->options);
2589 	cbe_lun->options = nvlist_clone(req->args_nvl);
2590 
2591 	wasprim = (cbe_lun->flags & CTL_LUN_FLAG_PRIMARY);
2592 	value = dnvlist_get_string(cbe_lun->options, "ha_role", NULL);
2593 	if (value != NULL) {
2594 		if (strcmp(value, "primary") == 0)
2595 			cbe_lun->flags |= CTL_LUN_FLAG_PRIMARY;
2596 		else
2597 			cbe_lun->flags &= ~CTL_LUN_FLAG_PRIMARY;
2598 	} else if (control_softc->flags & CTL_FLAG_ACTIVE_SHELF)
2599 		cbe_lun->flags |= CTL_LUN_FLAG_PRIMARY;
2600 	else
2601 		cbe_lun->flags &= ~CTL_LUN_FLAG_PRIMARY;
2602 	if (wasprim != (cbe_lun->flags & CTL_LUN_FLAG_PRIMARY)) {
2603 		if (cbe_lun->flags & CTL_LUN_FLAG_PRIMARY)
2604 			ctl_lun_primary(cbe_lun);
2605 		else
2606 			ctl_lun_secondary(cbe_lun);
2607 	}
2608 
2609 	oldsize = be_lun->size_blocks;
2610 	if ((cbe_lun->flags & CTL_LUN_FLAG_PRIMARY) ||
2611 	    control_softc->ha_mode == CTL_HA_MODE_SER_ONLY) {
2612 		if (be_lun->vn == NULL)
2613 			error = ctl_be_block_open(be_lun, req);
2614 		else if (vn_isdisk(be_lun->vn, &error))
2615 			error = ctl_be_block_open_dev(be_lun, req);
2616 		else if (be_lun->vn->v_type == VREG) {
2617 			vn_lock(be_lun->vn, LK_SHARED | LK_RETRY);
2618 			error = ctl_be_block_open_file(be_lun, req);
2619 			VOP_UNLOCK(be_lun->vn, 0);
2620 		} else
2621 			error = EINVAL;
2622 		if ((cbe_lun->flags & CTL_LUN_FLAG_NO_MEDIA) &&
2623 		    be_lun->vn != NULL) {
2624 			cbe_lun->flags &= ~CTL_LUN_FLAG_NO_MEDIA;
2625 			ctl_lun_has_media(cbe_lun);
2626 		} else if ((cbe_lun->flags & CTL_LUN_FLAG_NO_MEDIA) == 0 &&
2627 		    be_lun->vn == NULL) {
2628 			cbe_lun->flags |= CTL_LUN_FLAG_NO_MEDIA;
2629 			ctl_lun_no_media(cbe_lun);
2630 		}
2631 		cbe_lun->flags &= ~CTL_LUN_FLAG_EJECTED;
2632 	} else {
2633 		if (be_lun->vn != NULL) {
2634 			cbe_lun->flags |= CTL_LUN_FLAG_NO_MEDIA;
2635 			ctl_lun_no_media(cbe_lun);
2636 			taskqueue_drain_all(be_lun->io_taskqueue);
2637 			error = ctl_be_block_close(be_lun);
2638 		} else
2639 			error = 0;
2640 	}
2641 	if (be_lun->size_blocks != oldsize)
2642 		ctl_lun_capacity_changed(cbe_lun);
2643 
2644 	/* Tell the user the exact size we ended up using */
2645 	params->lun_size_bytes = be_lun->size_bytes;
2646 
2647 	req->status = error ? CTL_LUN_WARNING : CTL_LUN_OK;
2648 	return (0);
2649 
2650 bailout_error:
2651 	req->status = CTL_LUN_ERROR;
2652 	return (0);
2653 }
2654 
2655 static void
2656 ctl_be_block_lun_shutdown(void *be_lun)
2657 {
2658 	struct ctl_be_block_lun *lun = be_lun;
2659 	struct ctl_be_block_softc *softc = lun->softc;
2660 
2661 	mtx_lock(&softc->lock);
2662 	lun->flags |= CTL_BE_BLOCK_LUN_UNCONFIGURED;
2663 	if (lun->flags & CTL_BE_BLOCK_LUN_WAITING)
2664 		wakeup(lun);
2665 	mtx_unlock(&softc->lock);
2666 }
2667 
2668 static void
2669 ctl_be_block_lun_config_status(void *be_lun, ctl_lun_config_status status)
2670 {
2671 	struct ctl_be_block_lun *lun;
2672 	struct ctl_be_block_softc *softc;
2673 
2674 	lun = (struct ctl_be_block_lun *)be_lun;
2675 	softc = lun->softc;
2676 
2677 	if (status == CTL_LUN_CONFIG_OK) {
2678 		mtx_lock(&softc->lock);
2679 		lun->flags &= ~CTL_BE_BLOCK_LUN_UNCONFIGURED;
2680 		if (lun->flags & CTL_BE_BLOCK_LUN_WAITING)
2681 			wakeup(lun);
2682 		mtx_unlock(&softc->lock);
2683 
2684 		/*
2685 		 * We successfully added the LUN, attempt to enable it.
2686 		 */
2687 		if (ctl_enable_lun(&lun->cbe_lun) != 0) {
2688 			printf("%s: ctl_enable_lun() failed!\n", __func__);
2689 			if (ctl_invalidate_lun(&lun->cbe_lun) != 0) {
2690 				printf("%s: ctl_invalidate_lun() failed!\n",
2691 				       __func__);
2692 			}
2693 		}
2694 
2695 		return;
2696 	}
2697 
2698 
2699 	mtx_lock(&softc->lock);
2700 	lun->flags &= ~CTL_BE_BLOCK_LUN_UNCONFIGURED;
2701 	lun->flags |= CTL_BE_BLOCK_LUN_CONFIG_ERR;
2702 	wakeup(lun);
2703 	mtx_unlock(&softc->lock);
2704 }
2705 
2706 
2707 static int
2708 ctl_be_block_config_write(union ctl_io *io)
2709 {
2710 	struct ctl_be_block_lun *be_lun;
2711 	struct ctl_be_lun *cbe_lun;
2712 	int retval;
2713 
2714 	DPRINTF("entered\n");
2715 
2716 	cbe_lun = CTL_BACKEND_LUN(io);
2717 	be_lun = (struct ctl_be_block_lun *)cbe_lun->be_lun;
2718 
2719 	retval = 0;
2720 	switch (io->scsiio.cdb[0]) {
2721 	case SYNCHRONIZE_CACHE:
2722 	case SYNCHRONIZE_CACHE_16:
2723 	case WRITE_SAME_10:
2724 	case WRITE_SAME_16:
2725 	case UNMAP:
2726 		/*
2727 		 * The upper level CTL code will filter out any CDBs with
2728 		 * the immediate bit set and return the proper error.
2729 		 *
2730 		 * We don't really need to worry about what LBA range the
2731 		 * user asked to be synced out.  When they issue a sync
2732 		 * cache command, we'll sync out the whole thing.
2733 		 */
2734 		mtx_lock(&be_lun->queue_lock);
2735 		STAILQ_INSERT_TAIL(&be_lun->config_write_queue, &io->io_hdr,
2736 				   links);
2737 		mtx_unlock(&be_lun->queue_lock);
2738 		taskqueue_enqueue(be_lun->io_taskqueue, &be_lun->io_task);
2739 		break;
2740 	case START_STOP_UNIT: {
2741 		struct scsi_start_stop_unit *cdb;
2742 		struct ctl_lun_req req;
2743 
2744 		cdb = (struct scsi_start_stop_unit *)io->scsiio.cdb;
2745 		if ((cdb->how & SSS_PC_MASK) != 0) {
2746 			ctl_set_success(&io->scsiio);
2747 			ctl_config_write_done(io);
2748 			break;
2749 		}
2750 		if (cdb->how & SSS_START) {
2751 			if ((cdb->how & SSS_LOEJ) && be_lun->vn == NULL) {
2752 				retval = ctl_be_block_open(be_lun, &req);
2753 				cbe_lun->flags &= ~CTL_LUN_FLAG_EJECTED;
2754 				if (retval == 0) {
2755 					cbe_lun->flags &= ~CTL_LUN_FLAG_NO_MEDIA;
2756 					ctl_lun_has_media(cbe_lun);
2757 				} else {
2758 					cbe_lun->flags |= CTL_LUN_FLAG_NO_MEDIA;
2759 					ctl_lun_no_media(cbe_lun);
2760 				}
2761 			}
2762 			ctl_start_lun(cbe_lun);
2763 		} else {
2764 			ctl_stop_lun(cbe_lun);
2765 			if (cdb->how & SSS_LOEJ) {
2766 				cbe_lun->flags |= CTL_LUN_FLAG_NO_MEDIA;
2767 				cbe_lun->flags |= CTL_LUN_FLAG_EJECTED;
2768 				ctl_lun_ejected(cbe_lun);
2769 				if (be_lun->vn != NULL)
2770 					ctl_be_block_close(be_lun);
2771 			}
2772 		}
2773 
2774 		ctl_set_success(&io->scsiio);
2775 		ctl_config_write_done(io);
2776 		break;
2777 	}
2778 	case PREVENT_ALLOW:
2779 		ctl_set_success(&io->scsiio);
2780 		ctl_config_write_done(io);
2781 		break;
2782 	default:
2783 		ctl_set_invalid_opcode(&io->scsiio);
2784 		ctl_config_write_done(io);
2785 		retval = CTL_RETVAL_COMPLETE;
2786 		break;
2787 	}
2788 
2789 	return (retval);
2790 }
2791 
2792 static int
2793 ctl_be_block_config_read(union ctl_io *io)
2794 {
2795 	struct ctl_be_block_lun *be_lun;
2796 	struct ctl_be_lun *cbe_lun;
2797 	int retval = 0;
2798 
2799 	DPRINTF("entered\n");
2800 
2801 	cbe_lun = CTL_BACKEND_LUN(io);
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 static int
2866 ctl_be_block_init(void)
2867 {
2868 	struct ctl_be_block_softc *softc = &backend_block_softc;
2869 
2870 	mtx_init(&softc->lock, "ctlblock", NULL, MTX_DEF);
2871 	softc->beio_zone = uma_zcreate("beio", sizeof(struct ctl_be_block_io),
2872 	    NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0);
2873 	STAILQ_INIT(&softc->lun_list);
2874 	return (0);
2875 }
2876 
2877 
2878 static int
2879 ctl_be_block_shutdown(void)
2880 {
2881 	struct ctl_be_block_softc *softc = &backend_block_softc;
2882 	struct ctl_be_block_lun *lun, *next_lun;
2883 
2884 	mtx_lock(&softc->lock);
2885 	STAILQ_FOREACH_SAFE(lun, &softc->lun_list, links, next_lun) {
2886 		/*
2887 		 * Drop our lock here.  Since ctl_invalidate_lun() can call
2888 		 * back into us, this could potentially lead to a recursive
2889 		 * lock of the same mutex, which would cause a hang.
2890 		 */
2891 		mtx_unlock(&softc->lock);
2892 		ctl_disable_lun(&lun->cbe_lun);
2893 		ctl_invalidate_lun(&lun->cbe_lun);
2894 		mtx_lock(&softc->lock);
2895 	}
2896 	mtx_unlock(&softc->lock);
2897 
2898 	uma_zdestroy(softc->beio_zone);
2899 	mtx_destroy(&softc->lock);
2900 	return (0);
2901 }
2902