xref: /dragonfly/sys/kern/subr_diskiocom.c (revision 65cc0652)
1 /*
2  * Copyright (c) 2012 The DragonFly Project.  All rights reserved.
3  *
4  * This code is derived from software contributed to The DragonFly Project
5  * by Matthew Dillon <dillon@backplane.com>
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  *
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in
15  *    the documentation and/or other materials provided with the
16  *    distribution.
17  * 3. Neither the name of The DragonFly Project nor the names of its
18  *    contributors may be used to endorse or promote products derived
19  *    from this software without specific, prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
24  * FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE
25  * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
26  * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
27  * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
28  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
29  * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
30  * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
31  * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  */
34 #include <sys/param.h>
35 #include <sys/systm.h>
36 #include <sys/kernel.h>
37 #include <sys/proc.h>
38 #include <sys/sysctl.h>
39 #include <sys/buf.h>
40 #include <sys/conf.h>
41 #include <sys/disklabel.h>
42 #include <sys/disklabel32.h>
43 #include <sys/disklabel64.h>
44 #include <sys/diskslice.h>
45 #include <sys/diskmbr.h>
46 #include <sys/disk.h>
47 #include <sys/malloc.h>
48 #include <sys/device.h>
49 #include <sys/devfs.h>
50 #include <sys/thread.h>
51 #include <sys/queue.h>
52 #include <sys/lock.h>
53 #include <sys/stat.h>
54 #include <sys/uuid.h>
55 #include <sys/dmsg.h>
56 
57 #include <sys/buf2.h>
58 #include <sys/msgport2.h>
59 #include <sys/thread2.h>
60 
61 struct dios_open {
62 	int	openrd;
63 	int	openwr;
64 };
65 
66 struct dios_io {
67 	int	count;
68 	int	eof;
69 	kdmsg_data_t data;
70 };
71 
72 static MALLOC_DEFINE(M_DMSG_DISK, "dmsg_disk", "disk dmsg");
73 
74 static int blk_active;
75 SYSCTL_INT(_debug, OID_AUTO, blk_active, CTLFLAG_RW, &blk_active, 0,
76            "Number of active iocom IOs");
77 
78 static int disk_iocom_reconnect(struct disk *dp, struct file *fp);
79 static int disk_rcvdmsg(kdmsg_msg_t *msg);
80 
81 static void disk_blk_open(struct disk *dp, kdmsg_msg_t *msg);
82 static void disk_blk_read(struct disk *dp, kdmsg_msg_t *msg);
83 static void disk_blk_write(struct disk *dp, kdmsg_msg_t *msg);
84 static void disk_blk_flush(struct disk *dp, kdmsg_msg_t *msg);
85 static void disk_blk_freeblks(struct disk *dp, kdmsg_msg_t *msg);
86 static void diskiodone(struct bio *bio);
87 
88 void
89 disk_iocom_init(struct disk *dp)
90 {
91 	kdmsg_iocom_init(&dp->d_iocom, dp,
92 			 KDMSG_IOCOMF_AUTOCONN |
93 			 KDMSG_IOCOMF_AUTORXSPAN |
94 			 KDMSG_IOCOMF_AUTOTXSPAN,
95 			 M_DMSG_DISK, disk_rcvdmsg);
96 }
97 
98 void
99 disk_iocom_update(struct disk *dp)
100 {
101 }
102 
103 void
104 disk_iocom_uninit(struct disk *dp)
105 {
106 	kdmsg_iocom_uninit(&dp->d_iocom);
107 }
108 
109 int
110 disk_iocom_ioctl(struct disk *dp, u_long cmd, void *data)
111 {
112 	struct file *fp;
113 	struct disk_ioc_recluster *recl;
114 	int error;
115 
116 	switch(cmd) {
117 	case DIOCRECLUSTER:
118 		recl = data;
119 		fp = holdfp(curproc->p_fd, recl->fd, -1);
120 		if (fp) {
121 			error = disk_iocom_reconnect(dp, fp);
122 		} else {
123 			error = EINVAL;
124 		}
125 		break;
126 	default:
127 		error = EOPNOTSUPP;
128 		break;
129 	}
130 	return error;
131 }
132 
133 static
134 int
135 disk_iocom_reconnect(struct disk *dp, struct file *fp)
136 {
137 	char devname[64];
138 
139 	ksnprintf(devname, sizeof(devname), "%s%d",
140 		  dev_dname(dp->d_rawdev), dkunit(dp->d_rawdev));
141 
142 	kdmsg_iocom_reconnect(&dp->d_iocom, fp, devname);
143 
144 	dp->d_iocom.auto_lnk_conn.proto_version = DMSG_SPAN_PROTO_1;
145 	dp->d_iocom.auto_lnk_conn.peer_type = DMSG_PEER_BLOCK;
146 	dp->d_iocom.auto_lnk_conn.peer_mask = 1LLU << DMSG_PEER_BLOCK;
147 	dp->d_iocom.auto_lnk_conn.peer_mask = (uint64_t)-1;
148 #if 0
149 	if (dp->d_info.d_serialno) {
150 		ksnprintf(dp->d_iocom.auto_lnk_conn.peer_label,
151 			  sizeof(dp->d_iocom.auto_lnk_conn.peer_label),
152 			  "%s/%s", hostname, dp->d_info.d_serialno);
153 	} else {
154 		ksnprintf(dp->d_iocom.auto_lnk_conn.peer_label,
155 			  sizeof(dp->d_iocom.auto_lnk_conn.peer_label),
156 			  "%s/%s", hostname, devname);
157 	}
158 #endif
159 	ksnprintf(dp->d_iocom.auto_lnk_conn.peer_label,
160 		  sizeof(dp->d_iocom.auto_lnk_conn.peer_label),
161 		  "%s/%s", hostname, devname);
162 
163 	dp->d_iocom.auto_lnk_span.proto_version = DMSG_SPAN_PROTO_1;
164 	dp->d_iocom.auto_lnk_span.peer_type = DMSG_PEER_BLOCK;
165 	dp->d_iocom.auto_lnk_span.media.block.bytes =
166 						dp->d_info.d_media_size;
167 	dp->d_iocom.auto_lnk_span.media.block.blksize =
168 						dp->d_info.d_media_blksize;
169 	ksnprintf(dp->d_iocom.auto_lnk_span.peer_label,
170 		  sizeof(dp->d_iocom.auto_lnk_span.peer_label),
171 		  "%s", dp->d_iocom.auto_lnk_conn.peer_label);
172 	if (dp->d_info.d_serialno) {
173 		ksnprintf(dp->d_iocom.auto_lnk_span.pfs_label,
174 			  sizeof(dp->d_iocom.auto_lnk_span.pfs_label),
175 			  "%s", dp->d_info.d_serialno);
176 	} else {
177 		/*
178 		 * If no serial number is available generate a dummy serial
179 		 * number from the host and device name and pray.  This will
180 		 * allow e.g. /dev/vn* to look meaningful on a remote machine.
181 		 */
182 		ksnprintf(dp->d_iocom.auto_lnk_span.pfs_label,
183 			  sizeof(dp->d_iocom.auto_lnk_span.pfs_label),
184 			  "%s.%s", hostname, devname);
185 	}
186 
187 	kdmsg_iocom_autoinitiate(&dp->d_iocom, NULL);
188 
189 	return (0);
190 }
191 
192 static int
193 disk_rcvdmsg(kdmsg_msg_t *msg)
194 {
195 	struct disk *dp = msg->state->iocom->handle;
196 
197 	/*
198 	 * Handle debug messages (these might not be in transactions)
199 	 */
200 	switch(msg->any.head.cmd & DMSGF_CMDSWMASK) {
201 	case DMSG_DBG_SHELL:
202 		/*
203 		 * Execute shell command (not supported atm)
204 		 */
205 		kdmsg_msg_reply(msg, DMSG_ERR_NOSUPP);
206 		return(0);
207 	case DMSG_DBG_SHELL | DMSGF_REPLY:
208 		if (msg->aux_data) {
209 			msg->aux_data[msg->aux_size - 1] = 0;
210 			kprintf("diskiocom: DEBUGMSG: %s\n", msg->aux_data);
211 		}
212 		return(0);
213 	}
214 
215 	/*
216 	 * All remaining messages must be in a transaction.
217 	 *
218 	 * NOTE!  We currently don't care if the transaction is just
219 	 *	  the span transaction (for disk probes) or if it is the
220 	 *	  BLK_OPEN transaction.
221 	 *
222 	 * NOTE!  We are switching on the first message's command.  The
223 	 *	  actual message command within the transaction may be
224 	 *	  different (if streaming within a transaction).
225 	 */
226 	if (msg->state == &msg->state->iocom->state0) {
227 		kdmsg_msg_reply(msg, DMSG_ERR_NOSUPP);
228 		return(0);
229 	}
230 
231 	switch(msg->state->rxcmd & DMSGF_CMDSWMASK) {
232 	case DMSG_BLK_OPEN:
233 		disk_blk_open(dp, msg);
234 		break;
235 	case DMSG_BLK_READ:
236 		/*
237 		 * not reached normally but leave in for completeness
238 		 */
239 		disk_blk_read(dp, msg);
240 		break;
241 	case DMSG_BLK_WRITE:
242 		disk_blk_write(dp, msg);
243 		break;
244 	case DMSG_BLK_FLUSH:
245 		disk_blk_flush(dp, msg);
246 		break;
247 	case DMSG_BLK_FREEBLKS:
248 		disk_blk_freeblks(dp, msg);
249 		break;
250 	default:
251 		if ((msg->any.head.cmd & DMSGF_REPLY) == 0) {
252 			if (msg->any.head.cmd & DMSGF_DELETE)
253 				kdmsg_msg_reply(msg, DMSG_ERR_NOSUPP);
254 			else
255 				kdmsg_msg_result(msg, DMSG_ERR_NOSUPP);
256 		}
257 		break;
258 	}
259 	return (0);
260 }
261 
262 static
263 void
264 disk_blk_open(struct disk *dp, kdmsg_msg_t *msg)
265 {
266 	struct dios_open *openst;
267 	int error = DMSG_ERR_NOSUPP;
268 	int fflags;
269 
270 	openst = msg->state->any.any;
271 	if ((msg->any.head.cmd & DMSGF_CMDSWMASK) == DMSG_BLK_OPEN) {
272 		if (openst == NULL) {
273 			openst = kmalloc(sizeof(*openst), M_DEVBUF,
274 						M_WAITOK | M_ZERO);
275 			msg->state->any.any = openst;
276 		}
277 		fflags = 0;
278 		if (msg->any.blk_open.modes & DMSG_BLKOPEN_RD)
279 			fflags = FREAD;
280 		if (msg->any.blk_open.modes & DMSG_BLKOPEN_WR)
281 			fflags |= FWRITE;
282 		error = dev_dopen(dp->d_rawdev, fflags, S_IFCHR, proc0.p_ucred, NULL);
283 		if (error) {
284 			error = DMSG_ERR_IO;
285 		} else {
286 			if (msg->any.blk_open.modes & DMSG_BLKOPEN_RD)
287 				++openst->openrd;
288 			if (msg->any.blk_open.modes & DMSG_BLKOPEN_WR)
289 				++openst->openwr;
290 		}
291 	}
292 #if 0
293 	if ((msg->any.head.cmd & DMSGF_CMDSWMASK) == DMSG_BLK_CLOSE &&
294 	    openst) {
295 		fflags = 0;
296 		if ((msg->any.blk_open.modes & DMSG_BLKOPEN_RD) &&
297 		    openst->openrd) {
298 			fflags = FREAD;
299 		}
300 		if ((msg->any.blk_open.modes & DMSG_BLKOPEN_WR) &&
301 		    openst->openwr) {
302 			fflags |= FWRITE;
303 		}
304 		error = dev_dclose(dp->d_rawdev, fflags, S_IFCHR, NULL);
305 		if (error) {
306 			error = DMSG_ERR_IO;
307 		} else {
308 			if (msg->any.blk_open.modes & DMSG_BLKOPEN_RD)
309 				--openst->openrd;
310 			if (msg->any.blk_open.modes & DMSG_BLKOPEN_WR)
311 				--openst->openwr;
312 		}
313 	}
314 #endif
315 	if (msg->any.head.cmd & DMSGF_DELETE) {
316 		if (openst) {
317 			while (openst->openrd && openst->openwr) {
318 				--openst->openrd;
319 				--openst->openwr;
320 				dev_dclose(dp->d_rawdev, FREAD|FWRITE, S_IFCHR, NULL);
321 			}
322 			while (openst->openrd) {
323 				--openst->openrd;
324 				dev_dclose(dp->d_rawdev, FREAD, S_IFCHR, NULL);
325 			}
326 			while (openst->openwr) {
327 				--openst->openwr;
328 				dev_dclose(dp->d_rawdev, FWRITE, S_IFCHR, NULL);
329 			}
330 			kfree(openst, M_DEVBUF);
331 			msg->state->any.any = NULL;
332 		}
333 		kdmsg_msg_reply(msg, error);
334 	} else {
335 		kdmsg_msg_result(msg, error);
336 	}
337 }
338 
339 static
340 void
341 disk_blk_read(struct disk *dp, kdmsg_msg_t *msg)
342 {
343 	struct dios_io *iost;
344 	struct buf *bp;
345 	struct bio *bio;
346 	int error = DMSG_ERR_NOSUPP;
347 	int reterr = 1;
348 
349 	/*
350 	 * Only DMSG_BLK_READ commands imply read ops.
351 	 */
352 	iost = msg->state->any.any;
353 	if ((msg->any.head.cmd & DMSGF_CMDSWMASK) == DMSG_BLK_READ) {
354 		if (msg->any.blk_read.bytes < DEV_BSIZE ||
355 		    msg->any.blk_read.bytes > MAXPHYS) {
356 			error = DMSG_ERR_PARAM;
357 			goto done;
358 		}
359 		if (iost == NULL) {
360 			iost = kmalloc(sizeof(*iost), M_DEVBUF,
361 				       M_WAITOK | M_ZERO);
362 			msg->state->any.any = iost;
363 		}
364 		reterr = 0;
365 		bp = getpbuf_mem(NULL);
366 		KKASSERT(msg->any.blk_read.bytes <= bp->b_bufsize);
367 		bio = &bp->b_bio1;
368 		bp->b_cmd = BUF_CMD_READ;
369 		bp->b_bcount = msg->any.blk_read.bytes;
370 		bp->b_resid = bp->b_bcount;
371 		bio->bio_offset = msg->any.blk_read.offset;
372 		bio->bio_caller_info1.ptr = msg->state;
373 		bio->bio_done = diskiodone;
374 
375 		/* kdmsg_state_hold(msg->state); */
376 		atomic_add_int(&blk_active, 1);
377 		atomic_add_int(&iost->count, 1);
378 		if (msg->any.head.cmd & DMSGF_DELETE)
379 			iost->eof = 1;
380 		BUF_KERNPROC(bp);
381 		dev_dstrategy(dp->d_rawdev, bio);
382 	}
383 done:
384 	if (reterr) {
385 		if (msg->any.head.cmd & DMSGF_DELETE) {
386 			if (iost && iost->count == 0) {
387 				kfree(iost, M_DEVBUF);
388 				msg->state->any.any = NULL;
389 			}
390 			kdmsg_msg_reply(msg, error);
391 		} else {
392 			kdmsg_msg_result(msg, error);
393 		}
394 	}
395 }
396 
397 static
398 void
399 disk_blk_write(struct disk *dp, kdmsg_msg_t *msg)
400 {
401 	struct dios_io *iost;
402 	struct buf *bp;
403 	struct bio *bio;
404 	int error = DMSG_ERR_NOSUPP;
405 	int reterr = 1;
406 
407 	/*
408 	 * Only DMSG_BLK_WRITE commands imply read ops.
409 	 */
410 	iost = msg->state->any.any;
411 	if ((msg->any.head.cmd & DMSGF_CMDSWMASK) == DMSG_BLK_WRITE) {
412 		if (msg->any.blk_write.bytes < DEV_BSIZE ||
413 		    msg->any.blk_write.bytes > MAXPHYS) {
414 			error = DMSG_ERR_PARAM;
415 			goto done;
416 		}
417 		if (iost == NULL) {
418 			iost = kmalloc(sizeof(*iost), M_DEVBUF,
419 				       M_WAITOK | M_ZERO);
420 			msg->state->any.any = iost;
421 		}
422 
423 		/*
424 		 * Issue WRITE.  Short data implies zeros.  Try to optimize
425 		 * the buffer cache buffer for the case where we can just
426 		 * use the message's data pointer.
427 		 */
428 		reterr = 0;
429 		if (msg->aux_size >= msg->any.blk_write.bytes)
430 			bp = getpbuf(NULL);
431 		else
432 			bp = getpbuf_mem(NULL);
433 		KKASSERT(msg->any.blk_write.bytes <= bp->b_bufsize);
434 		bio = &bp->b_bio1;
435 		bp->b_cmd = BUF_CMD_WRITE;
436 		bp->b_bcount = msg->any.blk_write.bytes;
437 		bp->b_resid = bp->b_bcount;
438 		if (msg->aux_size >= msg->any.blk_write.bytes) {
439 			bp->b_data = msg->aux_data;
440 			kdmsg_detach_aux_data(msg, &iost->data);
441 		} else {
442 			bcopy(msg->aux_data, bp->b_data, msg->aux_size);
443 			bzero(bp->b_data + msg->aux_size,
444 			      msg->any.blk_write.bytes - msg->aux_size);
445 			bzero(&iost->data, sizeof(iost->data));
446 		}
447 		bio->bio_offset = msg->any.blk_write.offset;
448 		bio->bio_caller_info1.ptr = msg->state;
449 		bio->bio_done = diskiodone;
450 
451 		/* kdmsg_state_hold(msg->state); */
452 		atomic_add_int(&blk_active, 1);
453 		atomic_add_int(&iost->count, 1);
454 		if (msg->any.head.cmd & DMSGF_DELETE)
455 			iost->eof = 1;
456 		BUF_KERNPROC(bp);
457 		dev_dstrategy(dp->d_rawdev, bio);
458 	}
459 done:
460 	if (reterr) {
461 		if (msg->any.head.cmd & DMSGF_DELETE) {
462 			if (iost && iost->count == 0) {
463 				kfree(iost, M_DEVBUF);
464 				msg->state->any.any = NULL;
465 			}
466 			kdmsg_msg_reply(msg, error);
467 		} else {
468 			kdmsg_msg_result(msg, error);
469 		}
470 	}
471 }
472 
473 static
474 void
475 disk_blk_flush(struct disk *dp, kdmsg_msg_t *msg)
476 {
477 	struct dios_io *iost;
478 	struct buf *bp;
479 	struct bio *bio;
480 	int error = DMSG_ERR_NOSUPP;
481 	int reterr = 1;
482 
483 	/*
484 	 * Only DMSG_BLK_FLUSH commands imply read ops.
485 	 */
486 	iost = msg->state->any.any;
487 	if ((msg->any.head.cmd & DMSGF_CMDSWMASK) == DMSG_BLK_FLUSH) {
488 		if (iost == NULL) {
489 			iost = kmalloc(sizeof(*iost), M_DEVBUF,
490 				       M_WAITOK | M_ZERO);
491 			msg->state->any.any = iost;
492 		}
493 		reterr = 0;
494 		bp = getpbuf(NULL);
495 		bio = &bp->b_bio1;
496 		bp->b_cmd = BUF_CMD_FLUSH;
497 		bp->b_bcount = msg->any.blk_flush.bytes;
498 		bp->b_resid = 0;
499 		bio->bio_offset = msg->any.blk_flush.offset;
500 		bio->bio_caller_info1.ptr = msg->state;
501 		bio->bio_done = diskiodone;
502 
503 		/* kdmsg_state_hold(msg->state); */
504 		atomic_add_int(&blk_active, 1);
505 		atomic_add_int(&iost->count, 1);
506 		if (msg->any.head.cmd & DMSGF_DELETE)
507 			iost->eof = 1;
508 		BUF_KERNPROC(bp);
509 		dev_dstrategy(dp->d_rawdev, bio);
510 	}
511 	if (reterr) {
512 		if (msg->any.head.cmd & DMSGF_DELETE) {
513 			if (iost && iost->count == 0) {
514 				kfree(iost, M_DEVBUF);
515 				msg->state->any.any = NULL;
516 			}
517 			kdmsg_msg_reply(msg, error);
518 		} else {
519 			kdmsg_msg_result(msg, error);
520 		}
521 	}
522 }
523 
524 static
525 void
526 disk_blk_freeblks(struct disk *dp, kdmsg_msg_t *msg)
527 {
528 	struct dios_io *iost;
529 	struct buf *bp;
530 	struct bio *bio;
531 	int error = DMSG_ERR_NOSUPP;
532 	int reterr = 1;
533 
534 	/*
535 	 * Only DMSG_BLK_FREEBLKS commands imply read ops.
536 	 */
537 	iost = msg->state->any.any;
538 	if ((msg->any.head.cmd & DMSGF_CMDSWMASK) == DMSG_BLK_FREEBLKS) {
539 		if (iost == NULL) {
540 			iost = kmalloc(sizeof(*iost), M_DEVBUF,
541 				       M_WAITOK | M_ZERO);
542 			msg->state->any.any = iost;
543 		}
544 		reterr = 0;
545 		bp = getpbuf(NULL);
546 		bio = &bp->b_bio1;
547 		bp->b_cmd = BUF_CMD_FREEBLKS;
548 		bp->b_bcount = msg->any.blk_freeblks.bytes;
549 		bp->b_resid = 0;
550 		bio->bio_offset = msg->any.blk_freeblks.offset;
551 		bio->bio_caller_info1.ptr = msg->state;
552 		bio->bio_done = diskiodone;
553 
554 		/* kdmsg_state_hold(msg->state); */
555 		atomic_add_int(&blk_active, 1);
556 		atomic_add_int(&iost->count, 1);
557 		if (msg->any.head.cmd & DMSGF_DELETE)
558 			iost->eof = 1;
559 		BUF_KERNPROC(bp);
560 		dev_dstrategy(dp->d_rawdev, bio);
561 	}
562 	if (reterr) {
563 		if (msg->any.head.cmd & DMSGF_DELETE) {
564 			if (iost && iost->count == 0) {
565 				kfree(iost, M_DEVBUF);
566 				msg->state->any.any = NULL;
567 			}
568 			kdmsg_msg_reply(msg, error);
569 		} else {
570 			kdmsg_msg_result(msg, error);
571 		}
572 	}
573 }
574 
575 static
576 void
577 diskiodone(struct bio *bio)
578 {
579 	struct buf *bp = bio->bio_buf;
580 	kdmsg_state_t *state = bio->bio_caller_info1.ptr;
581 	kdmsg_msg_t *rmsg;
582 	struct dios_io *iost = state->any.any;
583 	int error;
584 	int resid = 0;
585 	int bytes;
586 	uint32_t cmd;
587 	void *data;
588 
589 	cmd = DMSG_LNK_ERROR;
590 	data = NULL;
591 	bytes = 0;
592 
593 	switch(bp->b_cmd) {
594 	case BUF_CMD_READ:
595 		cmd = DMSG_LNK_ERROR;
596 		data = bp->b_data;
597 		bytes = bp->b_bcount;
598 		/* fall through */
599 	case BUF_CMD_WRITE:
600 		if (bp->b_flags & B_ERROR) {
601 			error = bp->b_error;
602 		} else {
603 			error = 0;
604 			resid = bp->b_resid;
605 		}
606 		kdmsg_free_aux_data(&iost->data);
607 		break;
608 	case BUF_CMD_FLUSH:
609 	case BUF_CMD_FREEBLKS:
610 		if (bp->b_flags & B_ERROR)
611 			error = bp->b_error;
612 		else
613 			error = 0;
614 		break;
615 	default:
616 		panic("diskiodone: Unknown bio cmd = %d\n",
617 		      bio->bio_buf->b_cmd);
618 		error = 0;	/* avoid compiler warning */
619 		break;		/* NOT REACHED */
620 	}
621 
622 	/*
623 	 * Convert error to DMSG_ERR_* code.
624 	 */
625 	if (error)
626 		error = DMSG_ERR_IO;
627 
628 	/*
629 	 * Convert LNK_ERROR or BLK_ERROR if non-zero resid.  READS will
630 	 * have already converted cmd to BLK_ERROR and set up data to return.
631 	 */
632 	if (resid && cmd == DMSG_LNK_ERROR)
633 		cmd = DMSG_BLK_ERROR;
634 	/* XXX txcmd is delayed so this won't work for streaming */
635 	if ((state->txcmd & DMSGF_CREATE) == 0)	/* assume serialized */
636 		cmd |= DMSGF_CREATE;
637 	if (iost->eof) {
638 		if (atomic_fetchadd_int(&iost->count, -1) == 1)
639 			cmd |= DMSGF_DELETE;
640 	} else {
641 		atomic_add_int(&iost->count, -1);
642 	}
643 	atomic_add_int(&blk_active, -1);
644 	cmd |= DMSGF_REPLY;
645 
646 	/*
647 	 * Allocate a basic or extended reply.  Be careful not to populate
648 	 * extended header fields unless we allocated an extended reply.
649 	 */
650 	rmsg = kdmsg_msg_alloc(state, cmd, NULL, 0);
651 	if (data) {
652 		rmsg->aux_data = kmalloc(bytes, state->iocom->mmsg, M_INTWAIT);
653 		rmsg->aux_size = bytes;
654 		rmsg->flags |= KDMSG_FLAG_AUXALLOC;
655 		bcopy(data, rmsg->aux_data, bytes);
656 	}
657 	rmsg->any.blk_error.head.error = error;
658 	if ((cmd & DMSGF_BASECMDMASK) == DMSG_BLK_ERROR)
659 		rmsg->any.blk_error.resid = resid;
660 	bio->bio_caller_info1.ptr = NULL;
661 	/* kdmsg_state_drop(state); */
662 	kdmsg_msg_write(rmsg);
663 	if (bp->b_flags & B_PAGING) {
664 		relpbuf(bp, NULL);
665 	} else {
666 		bp->b_flags |= B_INVAL | B_AGE;
667 		relpbuf(bp, NULL);
668 	}
669 }
670