Home | History | Annotate | Line # | Download | only in kern
subr_disk.c revision 1.132.8.1
      1 /*	$NetBSD: subr_disk.c,v 1.132.8.1 2021/05/31 22:15:20 cjep Exp $	*/
      2 
      3 /*-
      4  * Copyright (c) 1996, 1997, 1999, 2000, 2009 The NetBSD Foundation, Inc.
      5  * All rights reserved.
      6  *
      7  * This code is derived from software contributed to The NetBSD Foundation
      8  * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
      9  * NASA Ames Research Center.
     10  *
     11  * Redistribution and use in source and binary forms, with or without
     12  * modification, are permitted provided that the following conditions
     13  * are met:
     14  * 1. Redistributions of source code must retain the above copyright
     15  *    notice, this list of conditions and the following disclaimer.
     16  * 2. Redistributions in binary form must reproduce the above copyright
     17  *    notice, this list of conditions and the following disclaimer in the
     18  *    documentation and/or other materials provided with the distribution.
     19  *
     20  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     21  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     22  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     23  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     24  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     25  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     26  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     27  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     28  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     29  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     30  * POSSIBILITY OF SUCH DAMAGE.
     31  */
     32 
     33 /*
     34  * Copyright (c) 1982, 1986, 1988, 1993
     35  *	The Regents of the University of California.  All rights reserved.
     36  * (c) UNIX System Laboratories, Inc.
     37  * All or some portions of this file are derived from material licensed
     38  * to the University of California by American Telephone and Telegraph
     39  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
     40  * the permission of UNIX System Laboratories, Inc.
     41  *
     42  * Redistribution and use in source and binary forms, with or without
     43  * modification, are permitted provided that the following conditions
     44  * are met:
     45  * 1. Redistributions of source code must retain the above copyright
     46  *    notice, this list of conditions and the following disclaimer.
     47  * 2. Redistributions in binary form must reproduce the above copyright
     48  *    notice, this list of conditions and the following disclaimer in the
     49  *    documentation and/or other materials provided with the distribution.
     50  * 3. Neither the name of the University nor the names of its contributors
     51  *    may be used to endorse or promote products derived from this software
     52  *    without specific prior written permission.
     53  *
     54  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     55  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     56  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     57  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     58  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     59  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     60  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     61  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     62  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     63  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     64  * SUCH DAMAGE.
     65  *
     66  *	@(#)ufs_disksubr.c	8.5 (Berkeley) 1/21/94
     67  */
     68 
     69 #include <sys/cdefs.h>
     70 __KERNEL_RCSID(0, "$NetBSD: subr_disk.c,v 1.132.8.1 2021/05/31 22:15:20 cjep Exp $");
     71 
     72 #include <sys/param.h>
     73 #include <sys/kernel.h>
     74 #include <sys/kmem.h>
     75 #include <sys/buf.h>
     76 #include <sys/fcntl.h>
     77 #include <sys/syslog.h>
     78 #include <sys/disklabel.h>
     79 #include <sys/disk.h>
     80 #include <sys/sysctl.h>
     81 #include <lib/libkern/libkern.h>
     82 
     83 /*
     84  * Disk error is the preface to plaintive error messages
     85  * about failing disk transfers.  It prints messages of the form
     86 
     87 hp0g: hard error reading fsbn 12345 of 12344-12347 (hp0 bn %d cn %d tn %d sn %d)
     88 
     89  * if the offset of the error in the transfer and a disk label
     90  * are both available.  blkdone should be -1 if the position of the error
     91  * is unknown; the disklabel pointer may be null from drivers that have not
     92  * been converted to use them.  The message is printed with printf
     93  * if pri is LOG_PRINTF, otherwise it uses log at the specified priority.
     94  * The message should be completed (with at least a newline) with printf
     95  * or addlog, respectively.  There is no trailing space.
     96  */
     97 #ifndef PRIdaddr
     98 #define PRIdaddr PRId64
     99 #endif
    100 void
    101 diskerr(const struct buf *bp, const char *dname, const char *what, int pri,
    102     int blkdone, const struct disklabel *lp)
    103 {
    104 	int unit = DISKUNIT(bp->b_dev), part = DISKPART(bp->b_dev);
    105 	void (*pr)(const char *, ...) __printflike(1, 2);
    106 	char partname = 'a' + part;
    107 	daddr_t sn;
    108 
    109 	if (/*CONSTCOND*/0)
    110 		/* Compiler will error this if the format is wrong... */
    111 		printf("%" PRIdaddr, bp->b_blkno);
    112 
    113 	if (pri != LOG_PRINTF) {
    114 		static const char fmt[] = "";
    115 		log(pri, fmt);
    116 		pr = addlog;
    117 	} else
    118 		pr = printf;
    119 	(*pr)("%s%d%c: %s %sing fsbn ", dname, unit, partname, what,
    120 	    bp->b_flags & B_READ ? "read" : "writ");
    121 	sn = bp->b_blkno;
    122 	if (bp->b_bcount <= DEV_BSIZE)
    123 		(*pr)("%" PRIdaddr, sn);
    124 	else {
    125 		if (blkdone >= 0) {
    126 			sn += blkdone;
    127 			(*pr)("%" PRIdaddr " of ", sn);
    128 		}
    129 		(*pr)("%" PRIdaddr "-%" PRIdaddr "", bp->b_blkno,
    130 		    bp->b_blkno + (bp->b_bcount - 1) / DEV_BSIZE);
    131 	}
    132 	if (lp && (blkdone >= 0 || bp->b_bcount <= lp->d_secsize)) {
    133 		sn += lp->d_partitions[part].p_offset;
    134 		(*pr)(" (%s%d bn %" PRIdaddr "; cn %" PRIdaddr "",
    135 		    dname, unit, sn, sn / lp->d_secpercyl);
    136 		sn %= lp->d_secpercyl;
    137 		(*pr)(" tn %" PRIdaddr " sn %" PRIdaddr ")",
    138 		    sn / lp->d_nsectors, sn % lp->d_nsectors);
    139 	}
    140 }
    141 
    142 /*
    143  * Searches the iostatlist for the disk corresponding to the
    144  * name provided.
    145  */
    146 struct disk *
    147 disk_find(const char *name)
    148 {
    149 	struct io_stats *stat;
    150 
    151 	stat = iostat_find(name);
    152 
    153 	if ((stat != NULL) && (stat->io_type == IOSTAT_DISK))
    154 		return stat->io_parent;
    155 
    156 	return (NULL);
    157 }
    158 
    159 void
    160 disk_init(struct disk *diskp, const char *name, const struct dkdriver *driver)
    161 {
    162 	u_int blocksize = DEV_BSIZE;
    163 
    164 	/*
    165 	 * Initialize the wedge-related locks and other fields.
    166 	 */
    167 	mutex_init(&diskp->dk_rawlock, MUTEX_DEFAULT, IPL_NONE);
    168 	mutex_init(&diskp->dk_openlock, MUTEX_DEFAULT, IPL_NONE);
    169 	LIST_INIT(&diskp->dk_wedges);
    170 	diskp->dk_nwedges = 0;
    171 	diskp->dk_labelsector = LABELSECTOR;
    172 	diskp->dk_blkshift = DK_BSIZE2BLKSHIFT(blocksize);
    173 	diskp->dk_byteshift = DK_BSIZE2BYTESHIFT(blocksize);
    174 	diskp->dk_name = name;
    175 	diskp->dk_driver = driver;
    176 }
    177 
    178 /*
    179  * Rename a disk.
    180  */
    181 void
    182 disk_rename(struct disk *diskp, const char *name)
    183 {
    184 
    185 	diskp->dk_name = name;
    186 	iostat_rename(diskp->dk_stats, diskp->dk_name);
    187 }
    188 
    189 /*
    190  * Attach a disk.
    191  */
    192 void
    193 disk_attach(struct disk *diskp)
    194 {
    195 
    196 	/*
    197 	 * Allocate and initialize the disklabel structures.
    198 	 */
    199 	diskp->dk_label = kmem_zalloc(sizeof(struct disklabel), KM_SLEEP);
    200 	diskp->dk_cpulabel = kmem_zalloc(sizeof(struct cpu_disklabel),
    201 	    KM_SLEEP);
    202 
    203 	/*
    204 	 * Set up the stats collection.
    205 	 */
    206 	diskp->dk_stats = iostat_alloc(IOSTAT_DISK, diskp, diskp->dk_name);
    207 }
    208 
    209 int
    210 disk_begindetach(struct disk *dk, int (*lastclose)(device_t),
    211     device_t self, int flags)
    212 {
    213 	int rc;
    214 
    215 	rc = 0;
    216 	mutex_enter(&dk->dk_openlock);
    217 	if (dk->dk_openmask == 0)
    218 		;	/* nothing to do */
    219 	else if ((flags & DETACH_FORCE) == 0)
    220 		rc = EBUSY;
    221 	else if (lastclose != NULL)
    222 		rc = (*lastclose)(self);
    223 	mutex_exit(&dk->dk_openlock);
    224 
    225 	return rc;
    226 }
    227 
    228 /*
    229  * Detach a disk.
    230  */
    231 void
    232 disk_detach(struct disk *diskp)
    233 {
    234 
    235 	/*
    236 	 * Remove from the drivelist.
    237 	 */
    238 	iostat_free(diskp->dk_stats);
    239 
    240 	/*
    241 	 * Release the disk-info dictionary.
    242 	 */
    243 	if (diskp->dk_info) {
    244 		prop_object_release(diskp->dk_info);
    245 		diskp->dk_info = NULL;
    246 	}
    247 
    248 	/*
    249 	 * Free the space used by the disklabel structures.
    250 	 */
    251 	kmem_free(diskp->dk_label, sizeof(*diskp->dk_label));
    252 	kmem_free(diskp->dk_cpulabel, sizeof(*diskp->dk_cpulabel));
    253 }
    254 
    255 void
    256 disk_destroy(struct disk *diskp)
    257 {
    258 
    259 	mutex_destroy(&diskp->dk_openlock);
    260 	mutex_destroy(&diskp->dk_rawlock);
    261 }
    262 
    263 /*
    264  * Mark the disk as having work queued for metrics collection.
    265  */
    266 void
    267 disk_wait(struct disk *diskp)
    268 {
    269 
    270 	iostat_wait(diskp->dk_stats);
    271 }
    272 
    273 /*
    274  * Mark the disk as busy for metrics collection.
    275  */
    276 void
    277 disk_busy(struct disk *diskp)
    278 {
    279 
    280 	iostat_busy(diskp->dk_stats);
    281 }
    282 
    283 /*
    284  * Finished disk operations, gather metrics.
    285  */
    286 void
    287 disk_unbusy(struct disk *diskp, long bcount, int read)
    288 {
    289 
    290 	iostat_unbusy(diskp->dk_stats, bcount, read);
    291 }
    292 
    293 /*
    294  * Return true if disk has an I/O operation in flight.
    295  */
    296 bool
    297 disk_isbusy(struct disk *diskp)
    298 {
    299 
    300 	return iostat_isbusy(diskp->dk_stats);
    301 }
    302 
    303 /*
    304  * Bounds checking against the media size, used for the raw partition.
    305  * secsize, mediasize and b_blkno must all be the same units.
    306  * Possibly this has to be DEV_BSIZE (512).
    307  */
    308 int
    309 bounds_check_with_mediasize(struct buf *bp, int secsize, uint64_t mediasize)
    310 {
    311 	int64_t sz;
    312 
    313 	if (bp->b_blkno < 0) {
    314 		/* Reject negative offsets immediately. */
    315 		bp->b_error = EINVAL;
    316 		return 0;
    317 	}
    318 
    319 	sz = howmany((int64_t)bp->b_bcount, secsize);
    320 
    321 	/*
    322 	 * bp->b_bcount is a 32-bit value, and we rejected a negative
    323 	 * bp->b_blkno already, so "bp->b_blkno + sz" cannot overflow.
    324 	 */
    325 
    326 	if (bp->b_blkno + sz > mediasize) {
    327 		sz = mediasize - bp->b_blkno;
    328 		if (sz == 0) {
    329 			/* If exactly at end of disk, return EOF. */
    330 			bp->b_resid = bp->b_bcount;
    331 			return 0;
    332 		}
    333 		if (sz < 0) {
    334 			/* If past end of disk, return EINVAL. */
    335 			bp->b_error = EINVAL;
    336 			return 0;
    337 		}
    338 		/* Otherwise, truncate request. */
    339 		bp->b_bcount = sz * secsize;
    340 	}
    341 
    342 	return 1;
    343 }
    344 
    345 /*
    346  * Determine the size of the transfer, and make sure it is
    347  * within the boundaries of the partition. Adjust transfer
    348  * if needed, and signal errors or early completion.
    349  */
    350 int
    351 bounds_check_with_label(struct disk *dk, struct buf *bp, int wlabel)
    352 {
    353 	struct disklabel *lp = dk->dk_label;
    354 	struct partition *p = lp->d_partitions + DISKPART(bp->b_dev);
    355 	uint64_t p_size, p_offset, labelsector;
    356 	int64_t sz;
    357 
    358 	if (bp->b_blkno < 0) {
    359 		/* Reject negative offsets immediately. */
    360 		bp->b_error = EINVAL;
    361 		return -1;
    362 	}
    363 
    364 	/* Protect against division by zero. XXX: Should never happen?!?! */
    365 	if ((lp->d_secsize / DEV_BSIZE) == 0 || lp->d_secpercyl == 0) {
    366 		bp->b_error = EINVAL;
    367 		return -1;
    368 	}
    369 
    370 	p_size = (uint64_t)p->p_size << dk->dk_blkshift;
    371 	p_offset = (uint64_t)p->p_offset << dk->dk_blkshift;
    372 #if RAW_PART == 3
    373 	labelsector = lp->d_partitions[2].p_offset;
    374 #else
    375 	labelsector = lp->d_partitions[RAW_PART].p_offset;
    376 #endif
    377 	labelsector = (labelsector + dk->dk_labelsector) << dk->dk_blkshift;
    378 
    379 	sz = howmany((int64_t)bp->b_bcount, DEV_BSIZE);
    380 
    381 	/*
    382 	 * bp->b_bcount is a 32-bit value, and we rejected a negative
    383 	 * bp->b_blkno already, so "bp->b_blkno + sz" cannot overflow.
    384 	 */
    385 
    386 	if (bp->b_blkno + sz > p_size) {
    387 		sz = p_size - bp->b_blkno;
    388 		if (sz == 0) {
    389 			/* If exactly at end of disk, return EOF. */
    390 			bp->b_resid = bp->b_bcount;
    391 			return 0;
    392 		}
    393 		if (sz < 0) {
    394 			/* If past end of disk, return EINVAL. */
    395 			bp->b_error = EINVAL;
    396 			return -1;
    397 		}
    398 		/* Otherwise, truncate request. */
    399 		bp->b_bcount = sz << DEV_BSHIFT;
    400 	}
    401 
    402 	/* Overwriting disk label? */
    403 	if (bp->b_blkno + p_offset <= labelsector &&
    404 	    bp->b_blkno + p_offset + sz > labelsector &&
    405 	    (bp->b_flags & B_READ) == 0 && !wlabel) {
    406 		bp->b_error = EROFS;
    407 		return -1;
    408 	}
    409 
    410 	/* calculate cylinder for disksort to order transfers with */
    411 	bp->b_cylinder = (bp->b_blkno + p->p_offset) /
    412 	    (lp->d_secsize / DEV_BSIZE) / lp->d_secpercyl;
    413 	return 1;
    414 }
    415 
    416 int
    417 disk_read_sectors(void (*strat)(struct buf *), const struct disklabel *lp,
    418     struct buf *bp, unsigned int sector, int count)
    419 {
    420 
    421 	if ((lp->d_secsize / DEV_BSIZE) == 0 || lp->d_secpercyl == 0)
    422 		return EINVAL;
    423 
    424 	bp->b_blkno = btodb((off_t)sector * lp->d_secsize);
    425 	bp->b_bcount = count * lp->d_secsize;
    426 	bp->b_flags = (bp->b_flags & ~B_WRITE) | B_READ;
    427 	bp->b_oflags &= ~BO_DONE;
    428 	bp->b_cylinder = sector / lp->d_secpercyl;
    429 	(*strat)(bp);
    430 	return biowait(bp);
    431 }
    432 
    433 const char *
    434 convertdisklabel(struct disklabel *lp, void (*strat)(struct buf *),
    435     struct buf *bp, uint32_t secperunit)
    436 {
    437 	struct partition rp, *altp, *p;
    438 	int geom_ok;
    439 	const char *str;
    440 
    441 	memset(&rp, 0, sizeof(rp));
    442 	rp.p_size = secperunit;
    443 	rp.p_fstype = FS_UNUSED;
    444 
    445 	/* If we can seek to d_secperunit - 1, believe the disk geometry. */
    446 	if (secperunit != 0 &&
    447 	    disk_read_sectors(strat, lp, bp, secperunit - 1, 1) == 0)
    448 		geom_ok = 1;
    449 	else
    450 		geom_ok = 0;
    451 
    452 #if 0
    453 	printf("%s: secperunit (%" PRIu32 ") %s\n", __func__,
    454 	    secperunit, geom_ok ? "ok" : "not ok");
    455 #endif
    456 
    457 	p = &lp->d_partitions[RAW_PART];
    458 	if (RAW_PART == 'c' - 'a')
    459 		altp = &lp->d_partitions['d' - 'a'];
    460 	else
    461 		altp = &lp->d_partitions['c' - 'a'];
    462 
    463 	if (lp->d_npartitions > RAW_PART && p->p_offset == 0 && p->p_size != 0)
    464 		return NULL;	/* already a raw partition */
    465 	else if (lp->d_npartitions > MAX('c', 'd') - 'a' &&
    466 		 altp->p_offset == 0 && altp->p_size != 0) {
    467 		/* alternate partition ('c' or 'd') is suitable for raw slot,
    468 		 * swap with 'd' or 'c'.
    469 		 */
    470 		rp = *p;
    471 		*p = *altp;
    472 		*altp = rp;
    473 		return NULL;
    474 	} else if (lp->d_npartitions <= RAW_PART &&
    475 	           lp->d_npartitions > 'c' - 'a') {
    476 		/* No raw partition is present, but the alternate is present.
    477 		 * Copy alternate to raw partition.
    478 		 */
    479 		lp->d_npartitions = RAW_PART + 1;
    480 		*p = *altp;
    481 		return NULL;
    482 	} else if (!geom_ok)
    483 		str = "no raw partition and disk reports bad geometry";
    484 	else if (lp->d_npartitions <= RAW_PART) {
    485 		memset(&lp->d_partitions[lp->d_npartitions], 0,
    486 		    sizeof(struct partition) * (RAW_PART - lp->d_npartitions));
    487 		*p = rp;
    488 		lp->d_npartitions = RAW_PART + 1;
    489 		return NULL;
    490 	} else if (lp->d_npartitions < MAXPARTITIONS) {
    491 		memmove(p + 1, p,
    492 		    sizeof(struct partition) * (lp->d_npartitions - RAW_PART));
    493 		*p = rp;
    494 		lp->d_npartitions++;
    495 		return NULL;
    496 	} else
    497 		str = "no raw partition and partition table is full";
    498 #ifdef DIAGNOSTIC
    499 	printf("Bad partition: %s\n", str);
    500 	printf("type = %u, subtype = %u, typename = %s\n",
    501 	    lp->d_type, lp->d_subtype, lp->d_typename);
    502 	printf("secsize = %u, nsectors = %u, ntracks = %u\n",
    503 	    lp->d_secsize, lp->d_nsectors, lp->d_ntracks);
    504 	printf("ncylinders = %u, secpercyl = %u, secperunit = %u\n",
    505 	    lp->d_ncylinders, lp->d_secpercyl, lp->d_secperunit);
    506 	printf("npartitions = %u\n", lp->d_npartitions);
    507 
    508 	for (size_t i = 0; i < MIN(lp->d_npartitions, MAXPARTITIONS); i++) {
    509 		p = &lp->d_partitions[i];
    510 		printf("\t%c: offset = %u size = %u fstype = %u\n",
    511 		    (char)(i + 'a'), p->p_offset, p->p_size, p->p_fstype);
    512 	}
    513 #endif
    514 	return str;
    515 }
    516 
    517 /*
    518  * disk_ioctl --
    519  *	Generic disk ioctl handling.
    520  */
    521 int
    522 disk_ioctl(struct disk *dk, dev_t dev, u_long cmd, void *data, int flag,
    523     struct lwp *l)
    524 {
    525 	struct dkwedge_info *dkw;
    526 	struct partinfo *pi;
    527 	struct partition *dp;
    528 #ifdef __HAVE_OLD_DISKLABEL
    529 	struct disklabel newlabel;
    530 #endif
    531 
    532 	switch (cmd) {
    533 	case DIOCGDISKINFO:
    534 		if (dk->dk_info == NULL)
    535 			return ENOTSUP;
    536 		return prop_dictionary_copyout_ioctl(data, cmd, dk->dk_info);
    537 
    538 	case DIOCGSECTORSIZE:
    539 		*(u_int *)data = dk->dk_geom.dg_secsize;
    540 		return 0;
    541 
    542 	case DIOCGMEDIASIZE:
    543 		*(off_t *)data = (off_t)dk->dk_geom.dg_secsize *
    544 		    dk->dk_geom.dg_secperunit;
    545 		return 0;
    546 	default:
    547 		break;
    548 	}
    549 
    550 	if (dev == NODEV)
    551 		return EPASSTHROUGH;
    552 
    553 	/* The following should be moved to dk_ioctl */
    554 	switch (cmd) {
    555 	case DIOCGDINFO:
    556 		if (dk->dk_label == NULL)
    557 			return EBUSY;
    558 		memcpy(data, dk->dk_label, sizeof (*dk->dk_label));
    559 		return 0;
    560 
    561 #ifdef __HAVE_OLD_DISKLABEL
    562 	case ODIOCGDINFO:
    563 		if (dk->dk_label == NULL)
    564 			return EBUSY;
    565 		memcpy(&newlabel, dk->dk_label, sizeof(newlabel));
    566 		if (newlabel.d_npartitions > OLDMAXPARTITIONS)
    567 			return ENOTTY;
    568 		memcpy(data, &newlabel, sizeof(struct olddisklabel));
    569 		return 0;
    570 #endif
    571 
    572 	case DIOCGPARTINFO:
    573 		pi = data;
    574 		memset(pi, 0, sizeof(*pi));
    575 		pi->pi_secsize = dk->dk_geom.dg_secsize;
    576 		pi->pi_bsize = MAX(BLKDEV_IOSIZE, pi->pi_secsize);
    577 
    578 		if (DISKPART(dev) == RAW_PART) {
    579 			pi->pi_size = dk->dk_geom.dg_secperunit;
    580 			return 0;
    581 		}
    582 
    583 		if (dk->dk_label == NULL)
    584 			return EBUSY;
    585 
    586 		dp = &dk->dk_label->d_partitions[DISKPART(dev)];
    587 		pi->pi_offset = dp->p_offset;
    588 		pi->pi_size = dp->p_size;
    589 
    590 		pi->pi_fstype = dp->p_fstype;
    591 		pi->pi_frag = dp->p_frag;
    592 		pi->pi_fsize = dp->p_fsize;
    593 		pi->pi_cpg = dp->p_cpg;
    594 
    595 		/*
    596 		 * dholland 20130616: XXX this logic should not be
    597 		 * here. It is here because the old buffer cache
    598 		 * demands that all accesses to the same blocks need
    599 		 * to be the same size; but it only works for FFS and
    600 		 * nowadays I think it'll fail silently if the size
    601 		 * info in the disklabel is wrong. (Or missing.) The
    602 		 * buffer cache needs to be smarter; or failing that
    603 		 * we need a reliable way here to get the right block
    604 		 * size; or a reliable way to guarantee that (a) the
    605 		 * fs is not mounted when we get here and (b) any
    606 		 * buffers generated here will get purged when the fs
    607 		 * does get mounted.
    608 		 */
    609 		if (dp->p_fstype == FS_BSDFFS &&
    610 		    dp->p_frag != 0 && dp->p_fsize != 0)
    611 			pi->pi_bsize = dp->p_frag * dp->p_fsize;
    612 		return 0;
    613 
    614 	case DIOCAWEDGE:
    615 		if ((flag & FWRITE) == 0)
    616 			return EBADF;
    617 
    618 		dkw = data;
    619 		strlcpy(dkw->dkw_parent, dk->dk_name, sizeof(dkw->dkw_parent));
    620 		return dkwedge_add(dkw);
    621 
    622 	case DIOCDWEDGE:
    623 		if ((flag & FWRITE) == 0)
    624 			return EBADF;
    625 
    626 		dkw = data;
    627 		strlcpy(dkw->dkw_parent, dk->dk_name, sizeof(dkw->dkw_parent));
    628 		return dkwedge_del(dkw);
    629 
    630 	case DIOCLWEDGES:
    631 		return dkwedge_list(dk, data, l);
    632 
    633 	case DIOCMWEDGES:
    634 		if ((flag & FWRITE) == 0)
    635 			return EBADF;
    636 
    637 		dkwedge_discover(dk);
    638 		return 0;
    639 
    640 	case DIOCRMWEDGES:
    641 		if ((flag & FWRITE) == 0)
    642 			return EBADF;
    643 
    644 		dkwedge_delall(dk);
    645 		return 0;
    646 
    647 	default:
    648 		return EPASSTHROUGH;
    649 	}
    650 }
    651 
    652 void
    653 disk_set_info(device_t dev, struct disk *dk, const char *type)
    654 {
    655 	struct disk_geom *dg = &dk->dk_geom;
    656 
    657 	if (dg->dg_secsize == 0) {
    658 #ifdef DIAGNOSTIC
    659 		printf("%s: fixing 0 sector size\n", dk->dk_name);
    660 #endif
    661 		dg->dg_secsize = DEV_BSIZE;
    662 	}
    663 
    664 	dk->dk_blkshift = DK_BSIZE2BLKSHIFT(dg->dg_secsize);
    665 	dk->dk_byteshift = DK_BSIZE2BYTESHIFT(dg->dg_secsize);
    666 
    667 	if (dg->dg_secperunit == 0) {
    668 #ifdef DIAGNOSTIC
    669 		if (dg->dg_ncylinders == 0) {
    670 			printf("%s: secperunit and ncylinders are zero\n",
    671 			    dk->dk_name);
    672 		}
    673 		if (dg->dg_nsectors == 0 || dg->dg_ntracks == 0) {
    674 			printf("%s: secperunit and (sectors or tracks) "
    675 			    "are zero\n", dk->dk_name);
    676 		}
    677 #endif
    678 		dg->dg_secperunit = (int64_t) dg->dg_nsectors *
    679 		    dg->dg_ntracks * dg->dg_ncylinders;
    680 	}
    681 
    682 	if (dg->dg_ncylinders == 0) {
    683 		if (dg->dg_ntracks && dg->dg_nsectors)
    684 			dg->dg_ncylinders = dg->dg_secperunit /
    685 			    (dg->dg_ntracks * dg->dg_nsectors);
    686 	}
    687 
    688 	prop_dictionary_t disk_info, odisk_info, geom;
    689 
    690 	disk_info = prop_dictionary_create();
    691 	geom = prop_dictionary_create();
    692 
    693 	prop_dictionary_set_uint64(geom, "sectors-per-unit",
    694 	    dg->dg_secperunit);
    695 
    696 	prop_dictionary_set_uint32(geom, "sector-size", dg->dg_secsize);
    697 
    698 	if (dg->dg_nsectors)
    699 		prop_dictionary_set_uint16(geom, "sectors-per-track",
    700 		    dg->dg_nsectors);
    701 
    702 	if (dg->dg_ntracks)
    703 		prop_dictionary_set_uint16(geom, "tracks-per-cylinder",
    704 		    dg->dg_ntracks);
    705 
    706 	if (dg->dg_ncylinders)
    707 		prop_dictionary_set_uint64(geom, "cylinders-per-unit",
    708 		    dg->dg_ncylinders);
    709 
    710 	prop_dictionary_set(disk_info, "geometry", geom);
    711 
    712 	if (type)
    713 		prop_dictionary_set_string_nocopy(disk_info, "type", type);
    714 
    715 	prop_object_release(geom);
    716 
    717 	odisk_info = dk->dk_info;
    718 	dk->dk_info = disk_info;
    719 
    720 	if (dev)
    721 		prop_dictionary_set(device_properties(dev), "disk-info",
    722 		    disk_info);
    723 
    724 	/*
    725 	 * Don't release disk_info here; we keep a reference to it.
    726 	 * disk_detach() will release it when we go away.
    727 	 */
    728 	if (odisk_info)
    729 		prop_object_release(odisk_info);
    730 }
    731