[thirdparty/mdadm.git] / Manage.c

/*
 * mdadm - manage Linux "md" devices aka RAID arrays.
 *
 * Copyright (C) 2001-2006 Neil Brown <neilb@suse.de>
 *
 *
 *    This program is free software; you can redistribute it and/or modify
 *    it under the terms of the GNU General Public License as published by
 *    the Free Software Foundation; either version 2 of the License, or
 *    (at your option) any later version.
 *
 *    This program is distributed in the hope that it will be useful,
 *    but WITHOUT ANY WARRANTY; without even the implied warranty of
 *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *    GNU General Public License for more details.
 *
 *    You should have received a copy of the GNU General Public License
 *    along with this program; if not, write to the Free Software
 *    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 *
 *    Author: Neil Brown
 *    Email: <neilb@cse.unsw.edu.au>
 *    Paper: Neil Brown
 *           School of Computer Science and Engineering
 *           The University of New South Wales
 *           Sydney, 2052
 *           Australia
 */

#include "mdadm.h"
#include "md_u.h"
#include "md_p.h"

#define REGISTER_DEV 		_IO (MD_MAJOR, 1)
#define START_MD     		_IO (MD_MAJOR, 2)
#define STOP_MD      		_IO (MD_MAJOR, 3)

int Manage_ro(char *devname, int fd, int readonly)
{
	/* switch to readonly or rw
	 *
	 * requires >= 0.90.0
	 * first check that array is runing
	 * use RESTART_ARRAY_RW or STOP_ARRAY_RO
	 *
	 */
	mdu_array_info_t array;

	if (md_get_version(fd) < 9000) {
		fprintf(stderr, Name ": need md driver version 0.90.0 or later\n");
		return 1;
	}
	if (ioctl(fd, GET_ARRAY_INFO, &array)) {
		fprintf(stderr, Name ": %s does not appear to be active.\n",
			devname);
		return 1;
	}

	if (readonly>0) {
		if (ioctl(fd, STOP_ARRAY_RO, NULL)) {
			fprintf(stderr, Name ": failed to set readonly for %s: %s\n",
				devname, strerror(errno));
			return 1;
		}
	} else if (readonly < 0) {
		if (ioctl(fd, RESTART_ARRAY_RW, NULL)) {
			fprintf(stderr, Name ": failed to set writable for %s: %s\n",
				devname, strerror(errno));
			return 1;
		}
	}
	return 0;
}

#ifndef MDASSEMBLE

int Manage_runstop(char *devname, int fd, int runstop, int quiet)
{
	/* Run or stop the array. array must already be configured
	 * required >= 0.90.0
	 */
	mdu_param_t param; /* unused */

	if (runstop == -1 && md_get_version(fd) < 9000) {
		if (ioctl(fd, STOP_MD, 0)) {
			if (!quiet) fprintf(stderr, Name ": stopping device %s failed: %s\n",
					    devname, strerror(errno));
			return 1;
		}
	}

	if (md_get_version(fd) < 9000) {
		fprintf(stderr, Name ": need md driver version 0.90.0 or later\n");
		return 1;
	}
	/*
	if (ioctl(fd, GET_ARRAY_INFO, &array)) {
		fprintf(stderr, Name ": %s does not appear to be active.\n",
			devname);
		return 1;
	}
	*/
	if (runstop>0) {
		if (ioctl(fd, RUN_ARRAY, &param)) {
			fprintf(stderr, Name ": failed to run array %s: %s\n",
				devname, strerror(errno));
			return 1;
		}
		if (quiet <= 0)
			fprintf(stderr, Name ": started %s\n", devname);
	} else if (runstop < 0){
		struct map_ent *map = NULL;
		struct stat stb;
		struct mdinfo *mdi;
		/* If this is an mdmon managed array, just write 'inactive'
		 * to the array state and let mdmon clear up.
		 */
		mdi = sysfs_read(fd, -1, GET_LEVEL|GET_VERSION);
		if (mdi &&
		    mdi->array.level > 0 &&
		    mdi->text_version[0] == '/') {
			char *cp;

			/* This is mdmon managed. */
			close(fd);
			if (sysfs_set_str(mdi, NULL,
					  "array_state", "inactive") < 0) {
				if (quiet==0)
					fprintf(stderr, Name
						": fail to stop array %s: %s\n",
						devname, strerror(errno));
				return 1;
			}

			/* Give monitor a chance to act */
			cp = strchr(mdi->text_version+1, '/');
			if (*cp)
				*cp = 0;
			ping_monitor(mdi->text_version+1);

			fd = open(devname, O_RDONLY);
		}
		if (mdi)
			sysfs_free(mdi);

		if (fd >= 0 && ioctl(fd, STOP_ARRAY, NULL)) {
			if (quiet==0)
				fprintf(stderr, Name
					": fail to stop array %s: %s\n",
					devname, strerror(errno));
			return 1;
		}

		if (quiet <= 0)
			fprintf(stderr, Name ": stopped %s\n", devname);
		if (fd >= 0 && fstat(fd, &stb) == 0) {
			int devnum;
			if (major(stb.st_rdev) == MD_MAJOR)
				devnum = minor(stb.st_rdev);
			else
				devnum = -1-(minor(stb.st_rdev)>>6);
			map_delete(&map, devnum);
			map_write(map);
			map_free(map);
		}
	}
	return 0;
}

int Manage_resize(char *devname, int fd, long long size, int raid_disks)
{
	mdu_array_info_t info;
	if (ioctl(fd, GET_ARRAY_INFO, &info) != 0) {
		fprintf(stderr, Name ": Cannot get array information for %s: %s\n",
			devname, strerror(errno));
		return 1;
	}
	if (size >= 0)
		info.size = size;
	if (raid_disks > 0)
		info.raid_disks = raid_disks;
	if (ioctl(fd, SET_ARRAY_INFO, &info) != 0) {
		fprintf(stderr, Name ": Cannot set device size/shape for %s: %s\n",
			devname, strerror(errno));
		return 1;
	}
	return 0;
}

int Manage_reconfig(char *devname, int fd, int layout)
{
	mdu_array_info_t info;
	if (ioctl(fd, GET_ARRAY_INFO, &info) != 0) {
		fprintf(stderr, Name ": Cannot get array information for %s: %s\n",
			devname, strerror(errno));
		return 1;
	}
	info.layout = layout;
	printf("layout set to %d\n", info.layout);
	if (ioctl(fd, SET_ARRAY_INFO, &info) != 0) {
		fprintf(stderr, Name ": Cannot set layout for %s: %s\n",
			devname, strerror(errno));
		return 1;
	}
	return 0;
}

int Manage_subdevs(char *devname, int fd,
		   mddev_dev_t devlist, int verbose)
{
	/* do something to each dev.
	 * devmode can be
	 *  'a' - add the device
	 *	   try HOT_ADD_DISK
	 *         If that fails EINVAL, try ADD_NEW_DISK
	 *  'r' - remove the device HOT_REMOVE_DISK
	 *        device can be 'faulty' or 'detached' in which case all
	 *	  matching devices are removed.
	 *  'f' - set the device faulty SET_DISK_FAULTY
	 *        device can be 'detached' in which case any device that
	 *	  is inaccessible will be marked faulty.
	 */
	mdu_array_info_t array;
	mdu_disk_info_t disc;
	unsigned long long array_size;
	mddev_dev_t dv, next = NULL;
	struct stat stb;
	int j, jnext = 0;
	int tfd;
	struct supertype *st, *tst;
	int duuid[4];
	int ouuid[4];
	int lfd = -1;

	if (ioctl(fd, GET_ARRAY_INFO, &array)) {
		fprintf(stderr, Name ": cannot get array info for %s\n",
			devname);
		return 1;
	}

	/* array.size is only 32 bit and may be truncated.
	 * So read from sysfs if possible, and record number of sectors
	 */

	array_size = get_component_size(fd);
	if (array_size <= 0)
		array_size = array.size * 2;

	tst = super_by_fd(fd);
	if (!tst) {
		fprintf(stderr, Name ": unsupport array - version %d.%d\n",
			array.major_version, array.minor_version);
		return 1;
	}

	for (dv = devlist, j=0 ; dv; dv = next, j = jnext) {
		unsigned long long ldsize;
		char dvname[20];
		char *dnprintable = dv->devname;
		int err;

		next = dv->next;
		jnext = 0;

		if (strcmp(dv->devname, "failed")==0 ||
		    strcmp(dv->devname, "faulty")==0) {
			if (dv->disposition != 'r') {
				fprintf(stderr, Name ": %s only meaningful "
					"with -r, not -%c\n",
					dv->devname, dv->disposition);
				return 1;
			}
			for (; j < array.raid_disks + array.nr_disks ; j++) {
				disc.number = j;
				if (ioctl(fd, GET_DISK_INFO, &disc))
					continue;
				if (disc.major == 0 && disc.minor == 0)
					continue;
				if ((disc.state & 1) == 0) /* faulty */
					continue;
				stb.st_rdev = makedev(disc.major, disc.minor);
				next = dv;
				jnext = j+1;
				sprintf(dvname,"%d:%d", disc.major, disc.minor);
				dnprintable = dvname;
				break;
			}
			if (jnext == 0)
				continue;
		} else if (strcmp(dv->devname, "detached") == 0) {
			if (dv->disposition != 'r' && dv->disposition != 'f') {
				fprintf(stderr, Name ": %s only meaningful "
					"with -r of -f, not -%c\n",
					dv->devname, dv->disposition);
				return 1;
			}
			for (; j < array.raid_disks + array.nr_disks; j++) {
				int sfd;
				disc.number = j;
				if (ioctl(fd, GET_DISK_INFO, &disc))
					continue;
				if (disc.major == 0 && disc.minor == 0)
					continue;
				sprintf(dvname,"%d:%d", disc.major, disc.minor);
				sfd = dev_open(dvname, O_RDONLY);
				if (sfd >= 0) {
					close(sfd);
					continue;
				}
				if (dv->disposition == 'f' &&
				    (disc.state & 1) == 1) /* already faulty */
					continue;
				if (errno != ENXIO)
					continue;
				stb.st_rdev = makedev(disc.major, disc.minor);
				next = dv;
				jnext = j+1;
				dnprintable = dvname;
				break;
			}
			if (jnext == 0)
				continue;
		} else {
			j = 0;

			if (stat(dv->devname, &stb)) {
				fprintf(stderr, Name ": cannot find %s: %s\n",
					dv->devname, strerror(errno));
				return 1;
			}
			if ((stb.st_mode & S_IFMT) != S_IFBLK) {
				fprintf(stderr, Name ": %s is not a "
					"block device.\n",
					dv->devname);
				return 1;
			}
		}
		switch(dv->disposition){
		default:
			fprintf(stderr, Name ": internal error - devmode[%s]=%d\n",
				dv->devname, dv->disposition);
			return 1;
		case 'a':
			/* add the device */
			if (tst->subarray[0]) {
				fprintf(stderr, Name ": Cannot add disks to a"
					" \'member\' array, perform this"
					" operation on the parent container\n");
				return 1;
			}
			/* Make sure it isn't in use (in 2.6 or later) */
			tfd = open(dv->devname, O_RDONLY|O_EXCL|O_DIRECT);
			if (tfd < 0) {
				fprintf(stderr, Name ": Cannot open %s: %s\n",
					dv->devname, strerror(errno));
				return 1;
			}
			remove_partitions(tfd);

			st = dup_super(tst);

			if (array.not_persistent==0)
				st->ss->load_super(st, tfd, NULL);

			if (!get_dev_size(tfd, dv->devname, &ldsize)) {
				close(tfd);
				return 1;
			}
			close(tfd);


			if (!tst->ss->external &&
			    array.major_version == 0 &&
			    md_get_version(fd)%100 < 2) {
				if (ioctl(fd, HOT_ADD_DISK,
					  (unsigned long)stb.st_rdev)==0) {
					if (verbose >= 0)
						fprintf(stderr, Name ": hot added %s\n",
							dv->devname);
					continue;
				}

				fprintf(stderr, Name ": hot add failed for %s: %s\n",
					dv->devname, strerror(errno));
				return 1;
			}

			if (array.not_persistent == 0) {

				/* Make sure device is large enough */
				if (tst->ss->avail_size(tst, ldsize/512) <
				    array_size) {
					fprintf(stderr, Name ": %s not large enough to join array\n",
						dv->devname);
					return 1;
				}

				/* need to find a sample superblock to copy, and
				 * a spare slot to use
				 */
				for (j = 0; j < tst->max_devs; j++) {
					char *dev;
					int dfd;
					disc.number = j;
					if (ioctl(fd, GET_DISK_INFO, &disc))
						continue;
					if (disc.major==0 && disc.minor==0)
						continue;
					if ((disc.state & 4)==0) continue; /* sync */
					/* Looks like a good device to try */
					dev = map_dev(disc.major, disc.minor, 1);
					if (!dev) continue;
					dfd = dev_open(dev, O_RDONLY);
					if (dfd < 0) continue;
					if (tst->ss->load_super(tst, dfd,
								NULL)) {
						close(dfd);
						continue;
					}
					close(dfd);
					break;
				}
				if (!tst->sb) {
					fprintf(stderr, Name ": cannot find valid superblock in this array - HELP\n");
					return 1;
				}
				/* Possibly this device was recently part of the array
				 * and was temporarily removed, and is now being re-added.
				 * If so, we can simply re-add it.
				 */
				tst->ss->uuid_from_super(tst, duuid);

				/* re-add doesn't work for version-1 superblocks
				 * before 2.6.18 :-(
				 */
				if (array.major_version == 1 &&
				    get_linux_version() <= 2006018)
					;
				else if (st->sb) {
					st->ss->uuid_from_super(st, ouuid);
					if (memcmp(duuid, ouuid, sizeof(ouuid))==0) {
						/* looks close enough for now.  Kernel
						 * will worry about whether a bitmap
						 * based reconstruction is possible.
						 */
						struct mdinfo mdi;
						st->ss->getinfo_super(st, &mdi);
						disc.major = major(stb.st_rdev);
						disc.minor = minor(stb.st_rdev);
						disc.number = mdi.disk.number;
						disc.raid_disk = mdi.disk.raid_disk;
						disc.state = mdi.disk.state;
						if (dv->writemostly)
							disc.state |= 1 << MD_DISK_WRITEMOSTLY;
						if (ioctl(fd, ADD_NEW_DISK, &disc) == 0) {
							if (verbose >= 0)
								fprintf(stderr, Name ": re-added %s\n", dv->devname);
							continue;
						}
						/* fall back on normal-add */
					}
				}
			} else {
				/* non-persistent. Must ensure that new drive
				 * is at least array.size big.
				 */
				if (ldsize/512 < array_size) {
					fprintf(stderr, Name ": %s not large enough to join array\n",
						dv->devname);
					return 1;
				}
			}
			/* in 2.6.17 and earlier, version-1 superblocks won't
			 * use the number we write, but will choose a free number.
			 * we must choose the same free number, which requires
			 * starting at 'raid_disks' and counting up
			 */
			for (j = array.raid_disks; j< tst->max_devs; j++) {
				disc.number = j;
				if (ioctl(fd, GET_DISK_INFO, &disc))
					break;
				if (disc.major==0 && disc.minor==0)
					break;
				if (disc.state & 8) /* removed */
					break;
			}
			disc.major = major(stb.st_rdev);
			disc.minor = minor(stb.st_rdev);
			disc.number =j;
			disc.state = 0;
			if (array.not_persistent==0) {
				int dfd;
				if (dv->writemostly)
					disc.state |= 1 << MD_DISK_WRITEMOSTLY;
				dfd = open(dv->devname, O_RDWR | O_EXCL|O_DIRECT);
				tst->ss->add_to_super(tst, &disc, dfd,
						      dv->devname);
				/* write_init_super will close 'dfd' */
				if (tst->ss->write_init_super(tst))
					return 1;
			} else if (dv->re_add) {
				/*  this had better be raid1.
				 * As we are "--re-add"ing we must find a spare slot
				 * to fill.
				 */
				char *used = malloc(array.raid_disks);
				memset(used, 0, array.raid_disks);
				for (j=0; j< tst->max_devs; j++) {
					mdu_disk_info_t disc2;
					disc2.number = j;
					if (ioctl(fd, GET_DISK_INFO, &disc2))
						continue;
					if (disc2.major==0 && disc2.minor==0)
						continue;
					if (disc2.state & 8) /* removed */
						continue;
					if (disc2.raid_disk < 0)
						continue;
					if (disc2.raid_disk > array.raid_disks)
						continue;
					used[disc2.raid_disk] = 1;
				}
				for (j=0 ; j<array.raid_disks; j++)
					if (!used[j]) {
						disc.raid_disk = j;
						disc.state |= (1<<MD_DISK_SYNC);
						break;
					}
			}
			if (dv->writemostly)
				disc.state |= (1 << MD_DISK_WRITEMOSTLY);
			if (ioctl(fd,ADD_NEW_DISK, &disc)) {
				fprintf(stderr, Name ": add new device failed for %s as %d: %s\n",
					dv->devname, j, strerror(errno));
				return 1;
			}
			if (verbose >= 0)
				fprintf(stderr, Name ": added %s\n", dv->devname);
			break;

		case 'r':
			/* hot remove */
			if (tst->subarray[0]) {
				fprintf(stderr, Name ": Cannot remove disks from a"
					" \'member\' array, perform this"
					" operation on the parent container\n");
				return 1;
			}
			if (tst->ss->external) {
				/* To remove a device from a container, we must
				 * check that it isn't in use in an array.
				 * This involves looking in the 'holders'
				 * directory - there must be just one entry,
				 * the container.
				 * To ensure that it doesn't get used as a
				 * hold spare while we are checking, we
				 * get an O_EXCL open on the container
				 */
				int dnum = fd2devnum(fd);
				lfd = open_dev_excl(dnum);
				if (lfd < 0) {
					fprintf(stderr, Name
						": Cannot get exclusive access "
						" to container - odd\n");
					return 1;
				}
				if (!sysfs_unique_holder(dnum, stb.st_rdev)) {
					fprintf(stderr, Name
						": %s is %s, cannot remove.\n",
						dnprintable,
						errno == EEXIST ? "still in use":
						"not a member");
					close(lfd);
					return 1;
				}
			}
			/* FIXME check that it is a current member */
			err = ioctl(fd, HOT_REMOVE_DISK, (unsigned long)stb.st_rdev);
			if (err && errno == ENODEV) {
				/* Old kernels rejected this if no personality
				 * registered */
				struct mdinfo *sra = sysfs_read(fd, 0, GET_DEVS);
				struct mdinfo *dv = NULL;
				if (sra)
					dv = sra->devs;
				for ( ; dv ; dv=dv->next)
					if (dv->disk.major == major(stb.st_rdev) &&
					    dv->disk.minor == minor(stb.st_rdev))
						break;
				if (dv)
					err = sysfs_set_str(sra, dv,
							    "state", "remove");
				else
					err = -1;
				if (sra)
					sysfs_free(sra);
			}
			if (err) {
				fprintf(stderr, Name ": hot remove failed "
					"for %s: %s\n",	dnprintable,
					strerror(errno));
				if (lfd >= 0)
					close(lfd);
				return 1;
			}
			close(lfd);
			if (verbose >= 0)
				fprintf(stderr, Name ": hot removed %s\n",
					dnprintable);
			break;

		case 'f': /* set faulty */
			/* FIXME check current member */
			if (ioctl(fd, SET_DISK_FAULTY, (unsigned long) stb.st_rdev)) {
				fprintf(stderr, Name ": set device faulty failed for %s:  %s\n",
					dnprintable, strerror(errno));
				return 1;
			}
			if (verbose >= 0)
				fprintf(stderr, Name ": set %s faulty in %s\n",
					dnprintable, devname);
			break;
		}
	}
	return 0;

}

int autodetect(void)
{
	/* Open any md device, and issue the RAID_AUTORUN ioctl */
	int rv = 1;
	int fd = dev_open("9:0", O_RDONLY);
	if (fd >= 0) {
		if (ioctl(fd, RAID_AUTORUN, 0) == 0)
			rv = 0;
		close(fd);
	}
	return rv;
}
#endif
Commit	Line	Data
64c4757e	1	/*
9a9dab36	2	* mdadm - manage Linux "md" devices aka RAID arrays.
64c4757e	3	*
4f589ad0	4	* Copyright (C) 2001-2006 Neil Brown <neilb@suse.de>
64c4757e NB	5	*
	6	*
	7	* This program is free software; you can redistribute it and/or modify
	8	* it under the terms of the GNU General Public License as published by
	9	* the Free Software Foundation; either version 2 of the License, or
	10	* (at your option) any later version.
	11	*
	12	* This program is distributed in the hope that it will be useful,
	13	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	14	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	15	* GNU General Public License for more details.
	16	*
	17	* You should have received a copy of the GNU General Public License
	18	* along with this program; if not, write to the Free Software
	19	* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
	20	*
	21	* Author: Neil Brown
	22	* Email: <neilb@cse.unsw.edu.au>
	23	* Paper: Neil Brown
	24	* School of Computer Science and Engineering
	25	* The University of New South Wales
	26	* Sydney, 2052
	27	* Australia
	28	*/
	29
9a9dab36	30	#include "mdadm.h"
682c7051 NB	31	#include "md_u.h"
682c7051 NB	32	#include "md_p.h"
64c4757e	33
82b27616 NB	34	#define REGISTER_DEV _IO (MD_MAJOR, 1)
	35	#define START_MD _IO (MD_MAJOR, 2)
	36	#define STOP_MD _IO (MD_MAJOR, 3)
	37
64c4757e NB	38	int Manage_ro(char *devname, int fd, int readonly)
64c4757e NB	39	{
682c7051 NB	40	/* switch to readonly or rw
	41	*
	42	* requires >= 0.90.0
	43	* first check that array is runing
	44	* use RESTART_ARRAY_RW or STOP_ARRAY_RO
	45	*
	46	*/
	47	mdu_array_info_t array;
aba69144	48
682c7051 NB	49	if (md_get_version(fd) < 9000) {
	50	fprintf(stderr, Name ": need md driver version 0.90.0 or later\n");
	51	return 1;
	52	}
	53	if (ioctl(fd, GET_ARRAY_INFO, &array)) {
	54	fprintf(stderr, Name ": %s does not appear to be active.\n",
	55	devname);
	56	return 1;
	57	}
aba69144	58
682c7051 NB	59	if (readonly>0) {
	60	if (ioctl(fd, STOP_ARRAY_RO, NULL)) {
	61	fprintf(stderr, Name ": failed to set readonly for %s: %s\n",
	62	devname, strerror(errno));
	63	return 1;
	64	}
	65	} else if (readonly < 0) {
	66	if (ioctl(fd, RESTART_ARRAY_RW, NULL)) {
82b27616	67	fprintf(stderr, Name ": failed to set writable for %s: %s\n",
682c7051 NB	68	devname, strerror(errno));
	69	return 1;
	70	}
	71	}
aba69144	72	return 0;
64c4757e NB	73	}
64c4757e NB	74
435d4ebb LB	75	#ifndef MDASSEMBLE
435d4ebb LB	76
91f068bf	77	int Manage_runstop(char *devname, int fd, int runstop, int quiet)
64c4757e	78	{
682c7051 NB	79	/* Run or stop the array. array must already be configured
	80	* required >= 0.90.0
	81	*/
682c7051	82	mdu_param_t param; /* unused */
82b27616 NB	83
	84	if (runstop == -1 && md_get_version(fd) < 9000) {
	85	if (ioctl(fd, STOP_MD, 0)) {
91f068bf NB	86	if (!quiet) fprintf(stderr, Name ": stopping device %s failed: %s\n",
91f068bf NB	87	devname, strerror(errno));
82b27616 NB	88	return 1;
	89	}
	90	}
aba69144	91
682c7051 NB	92	if (md_get_version(fd) < 9000) {
	93	fprintf(stderr, Name ": need md driver version 0.90.0 or later\n");
	94	return 1;
	95	}
82b27616	96	/*
682c7051 NB	97	if (ioctl(fd, GET_ARRAY_INFO, &array)) {
	98	fprintf(stderr, Name ": %s does not appear to be active.\n",
	99	devname);
	100	return 1;
	101	}
82b27616	102	*/
682c7051 NB	103	if (runstop>0) {
	104	if (ioctl(fd, RUN_ARRAY, &param)) {
	105	fprintf(stderr, Name ": failed to run array %s: %s\n",
	106	devname, strerror(errno));
	107	return 1;
	108	}
8382f19b NB	109	if (quiet <= 0)
8382f19b NB	110	fprintf(stderr, Name ": started %s\n", devname);
682c7051	111	} else if (runstop < 0){
8382f19b NB	112	struct map_ent *map = NULL;
8382f19b NB	113	struct stat stb;
daf7a3ce NB	114	struct mdinfo *mdi;
	115	/* If this is an mdmon managed array, just write 'inactive'
	116	* to the array state and let mdmon clear up.
	117	*/
	118	mdi = sysfs_read(fd, -1, GET_LEVEL\|GET_VERSION);
	119	if (mdi &&
	120	mdi->array.level > 0 &&
	121	mdi->text_version[0] == '/') {
	122	char *cp;
	123
	124	/* This is mdmon managed. */
	125	close(fd);
	126	if (sysfs_set_str(mdi, NULL,
	127	"array_state", "inactive") < 0) {
	128	if (quiet==0)
	129	fprintf(stderr, Name
	130	": fail to stop array %s: %s\n",
	131	devname, strerror(errno));
	132	return 1;
	133	}
	134
	135	/* Give monitor a chance to act */
	136	cp = strchr(mdi->text_version+1, '/');
	137	if (*cp)
	138	*cp = 0;
	139	ping_monitor(mdi->text_version+1);
	140
	141	fd = open(devname, O_RDONLY);
	142	}
	143	if (mdi)
	144	sysfs_free(mdi);
	145
	146	if (fd >= 0 && ioctl(fd, STOP_ARRAY, NULL)) {
ab56093f	147	if (quiet==0)
daf7a3ce NB	148	fprintf(stderr, Name
daf7a3ce NB	149	": fail to stop array %s: %s\n",
91f068bf	150	devname, strerror(errno));
682c7051 NB	151	return 1;
682c7051 NB	152	}
daf7a3ce	153
ab56093f NB	154	if (quiet <= 0)
ab56093f NB	155	fprintf(stderr, Name ": stopped %s\n", devname);
daf7a3ce	156	if (fd >= 0 && fstat(fd, &stb) == 0) {
8382f19b NB	157	int devnum;
	158	if (major(stb.st_rdev) == MD_MAJOR)
	159	devnum = minor(stb.st_rdev);
	160	else
	161	devnum = -1-(minor(stb.st_rdev)>>6);
	162	map_delete(&map, devnum);
	163	map_write(map);
	164	map_free(map);
	165	}
682c7051 NB	166	}
682c7051 NB	167	return 0;
64c4757e NB	168	}
64c4757e NB	169
dd0781e5 NB	170	int Manage_resize(char *devname, int fd, long long size, int raid_disks)
	171	{
	172	mdu_array_info_t info;
	173	if (ioctl(fd, GET_ARRAY_INFO, &info) != 0) {
	174	fprintf(stderr, Name ": Cannot get array information for %s: %s\n",
	175	devname, strerror(errno));
	176	return 1;
	177	}
	178	if (size >= 0)
	179	info.size = size;
	180	if (raid_disks > 0)
	181	info.raid_disks = raid_disks;
	182	if (ioctl(fd, SET_ARRAY_INFO, &info) != 0) {
	183	fprintf(stderr, Name ": Cannot set device size/shape for %s: %s\n",
	184	devname, strerror(errno));
	185	return 1;
	186	}
	187	return 0;
	188	}
	189
b5e64645 NB	190	int Manage_reconfig(char *devname, int fd, int layout)
	191	{
	192	mdu_array_info_t info;
	193	if (ioctl(fd, GET_ARRAY_INFO, &info) != 0) {
	194	fprintf(stderr, Name ": Cannot get array information for %s: %s\n",
	195	devname, strerror(errno));
	196	return 1;
	197	}
	198	info.layout = layout;
	199	printf("layout set to %d\n", info.layout);
	200	if (ioctl(fd, SET_ARRAY_INFO, &info) != 0) {
	201	fprintf(stderr, Name ": Cannot set layout for %s: %s\n",
	202	devname, strerror(errno));
	203	return 1;
	204	}
	205	return 0;
	206	}
dd0781e5	207
64c4757e	208	int Manage_subdevs(char *devname, int fd,
dab6685f	209	mddev_dev_t devlist, int verbose)
cd29a5c8	210	{
682c7051 NB	211	/* do something to each dev.
	212	* devmode can be
	213	* 'a' - add the device
	214	* try HOT_ADD_DISK
	215	* If that fails EINVAL, try ADD_NEW_DISK
	216	* 'r' - remove the device HOT_REMOVE_DISK
b80da661 NB	217	* device can be 'faulty' or 'detached' in which case all
b80da661 NB	218	* matching devices are removed.
682c7051	219	* 'f' - set the device faulty SET_DISK_FAULTY
b80da661 NB	220	* device can be 'detached' in which case any device that
b80da661 NB	221	* is inaccessible will be marked faulty.
682c7051 NB	222	*/
	223	mdu_array_info_t array;
	224	mdu_disk_info_t disc;
7a3be72f	225	unsigned long long array_size;
b80da661	226	mddev_dev_t dv, next = NULL;
682c7051	227	struct stat stb;
b80da661	228	int j, jnext = 0;
0fbf459d	229	int tfd;
3da92f27	230	struct supertype st, tst;
fe80f49b NB	231	int duuid[4];
fe80f49b NB	232	int ouuid[4];
f94d52f4	233	int lfd = -1;
682c7051 NB	234
	235	if (ioctl(fd, GET_ARRAY_INFO, &array)) {
	236	fprintf(stderr, Name ": cannot get array info for %s\n",
	237	devname);
	238	return 1;
	239	}
3da92f27	240
7a3be72f NB	241	/* array.size is only 32 bit and may be truncated.
	242	* So read from sysfs if possible, and record number of sectors
	243	*/
	244
	245	array_size = get_component_size(fd);
	246	if (array_size <= 0)
	247	array_size = array.size * 2;
	248
1686dc25	249	tst = super_by_fd(fd);
3da92f27 NB	250	if (!tst) {
	251	fprintf(stderr, Name ": unsupport array - version %d.%d\n",
	252	array.major_version, array.minor_version);
	253	return 1;
	254	}
	255
b80da661	256	for (dv = devlist, j=0 ; dv; dv = next, j = jnext) {
4a39c6f2	257	unsigned long long ldsize;
b80da661 NB	258	char dvname[20];
b80da661 NB	259	char *dnprintable = dv->devname;
f94d52f4	260	int err;
4a39c6f2	261
b80da661 NB	262	next = dv->next;
	263	jnext = 0;
	264
	265	if (strcmp(dv->devname, "failed")==0 \|\|
	266	strcmp(dv->devname, "faulty")==0) {
	267	if (dv->disposition != 'r') {
	268	fprintf(stderr, Name ": %s only meaningful "
	269	"with -r, not -%c\n",
	270	dv->devname, dv->disposition);
	271	return 1;
	272	}
	273	for (; j < array.raid_disks + array.nr_disks ; j++) {
	274	disc.number = j;
	275	if (ioctl(fd, GET_DISK_INFO, &disc))
	276	continue;
	277	if (disc.major == 0 && disc.minor == 0)
	278	continue;
	279	if ((disc.state & 1) == 0) /* faulty */
	280	continue;
	281	stb.st_rdev = makedev(disc.major, disc.minor);
	282	next = dv;
	283	jnext = j+1;
	284	sprintf(dvname,"%d:%d", disc.major, disc.minor);
	285	dnprintable = dvname;
	286	break;
	287	}
	288	if (jnext == 0)
	289	continue;
	290	} else if (strcmp(dv->devname, "detached") == 0) {
	291	if (dv->disposition != 'r' && dv->disposition != 'f') {
	292	fprintf(stderr, Name ": %s only meaningful "
	293	"with -r of -f, not -%c\n",
	294	dv->devname, dv->disposition);
	295	return 1;
	296	}
	297	for (; j < array.raid_disks + array.nr_disks; j++) {
	298	int sfd;
	299	disc.number = j;
	300	if (ioctl(fd, GET_DISK_INFO, &disc))
	301	continue;
	302	if (disc.major == 0 && disc.minor == 0)
	303	continue;
	304	sprintf(dvname,"%d:%d", disc.major, disc.minor);
	305	sfd = dev_open(dvname, O_RDONLY);
	306	if (sfd >= 0) {
	307	close(sfd);
	308	continue;
	309	}
	310	if (dv->disposition == 'f' &&
	311	(disc.state & 1) == 1) /* already faulty */
	312	continue;
	313	if (errno != ENXIO)
	314	continue;
	315	stb.st_rdev = makedev(disc.major, disc.minor);
	316	next = dv;
	317	jnext = j+1;
	318	dnprintable = dvname;
	319	break;
	320	}
	321	if (jnext == 0)
	322	continue;
	323	} else {
	324	j = 0;
	325
326	if (stat(dv->devname, &stb)) {
327	fprintf(stderr, Name ": cannot find %s: %s\n",
328	dv->devname, strerror(errno));
329	return 1;
330	}
331	if ((stb.st_mode & S_IFMT) != S_IFBLK) {
332	fprintf(stderr, Name ": %s is not a "
333	"block device.\n",
334	dv->devname);
335	return 1;
336	}
682c7051	337	}
cd29a5c8	338	switch(dv->disposition){
682c7051	339	default:
c913b90e NB	340	fprintf(stderr, Name ": internal error - devmode[%s]=%d\n",
c913b90e NB	341	dv->devname, dv->disposition);
682c7051 NB	342	return 1;
682c7051 NB	343	case 'a':
4a39c6f2	344	/* add the device */
f7e7067b	345	if (tst->subarray[0]) {
f7dd881f DW	346	fprintf(stderr, Name ": Cannot add disks to a"
	347	" \'member\' array, perform this"
	348	" operation on the parent container\n");
	349	return 1;
f94d52f4	350	}
f277ce36	351	/* Make sure it isn't in use (in 2.6 or later) */
6416d527	352	tfd = open(dv->devname, O_RDONLY\|O_EXCL\|O_DIRECT);
0fbf459d	353	if (tfd < 0) {
d7eaf49f NB	354	fprintf(stderr, Name ": Cannot open %s: %s\n",
	355	dv->devname, strerror(errno));
	356	return 1;
	357	}
25037aa8	358	remove_partitions(tfd);
3da92f27	359
1686dc25	360	st = dup_super(tst);
3da92f27	361
fe80f49b	362	if (array.not_persistent==0)
3da92f27 NB	363	st->ss->load_super(st, tfd, NULL);
3da92f27 NB	364
beae1dfe	365	if (!get_dev_size(tfd, dv->devname, &ldsize)) {
4a39c6f2 NB	366	close(tfd);
4a39c6f2 NB	367	return 1;
4a39c6f2	368	}
0fbf459d	369	close(tfd);
fe80f49b	370
f94d52f4 NB	371
	372	if (!tst->ss->external &&
	373	array.major_version == 0 &&
60d9a174	374	md_get_version(fd)%100 < 2) {
892debc8 NB	375	if (ioctl(fd, HOT_ADD_DISK,
892debc8 NB	376	(unsigned long)stb.st_rdev)==0) {
dab6685f NB	377	if (verbose >= 0)
	378	fprintf(stderr, Name ": hot added %s\n",
	379	dv->devname);
892debc8 NB	380	continue;
	381	}
	382
cd29a5c8	383	fprintf(stderr, Name ": hot add failed for %s: %s\n",
892debc8	384	dv->devname, strerror(errno));
cd29a5c8 NB	385	return 1;
cd29a5c8 NB	386	}
892debc8	387
fe80f49b NB	388	if (array.not_persistent == 0) {
fe80f49b NB	389
4a39c6f2	390	/* Make sure device is large enough */
3da92f27	391	if (tst->ss->avail_size(tst, ldsize/512) <
7a3be72f	392	array_size) {
4a39c6f2 NB	393	fprintf(stderr, Name ": %s not large enough to join array\n",
	394	dv->devname);
	395	return 1;
	396	}
	397
fe80f49b	398	/* need to find a sample superblock to copy, and
aba69144	399	* a spare slot to use
fe80f49b	400	*/
3da92f27	401	for (j = 0; j < tst->max_devs; j++) {
fe80f49b NB	402	char *dev;
	403	int dfd;
	404	disc.number = j;
	405	if (ioctl(fd, GET_DISK_INFO, &disc))
	406	continue;
	407	if (disc.major==0 && disc.minor==0)
	408	continue;
	409	if ((disc.state & 4)==0) continue; /* sync */
	410	/* Looks like a good device to try */
16c6fa80	411	dev = map_dev(disc.major, disc.minor, 1);
fe80f49b	412	if (!dev) continue;
16c6fa80	413	dfd = dev_open(dev, O_RDONLY);
fe80f49b	414	if (dfd < 0) continue;
3da92f27 NB	415	if (tst->ss->load_super(tst, dfd,
3da92f27 NB	416	NULL)) {
fe80f49b NB	417	close(dfd);
	418	continue;
	419	}
892debc8	420	close(dfd);
fe80f49b NB	421	break;
fe80f49b NB	422	}
3da92f27	423	if (!tst->sb) {
fe80f49b NB	424	fprintf(stderr, Name ": cannot find valid superblock in this array - HELP\n");
	425	return 1;
	426	}
	427	/* Possibly this device was recently part of the array
	428	* and was temporarily removed, and is now being re-added.
	429	* If so, we can simply re-add it.
	430	*/
3da92f27	431	tst->ss->uuid_from_super(tst, duuid);
7eae7080 NB	432
	433	/* re-add doesn't work for version-1 superblocks
	434	* before 2.6.18 :-(
	435	*/
	436	if (array.major_version == 1 &&
	437	get_linux_version() <= 2006018)
	438	;
3da92f27 NB	439	else if (st->sb) {
3da92f27 NB	440	st->ss->uuid_from_super(st, ouuid);
fe80f49b	441	if (memcmp(duuid, ouuid, sizeof(ouuid))==0) {
4d20d744 NB	442	/* looks close enough for now. Kernel
	443	* will worry about whether a bitmap
	444	* based reconstruction is possible.
fe80f49b NB	445	*/
fe80f49b NB	446	struct mdinfo mdi;
3da92f27	447	st->ss->getinfo_super(st, &mdi);
fe80f49b NB	448	disc.major = major(stb.st_rdev);
	449	disc.minor = minor(stb.st_rdev);
	450	disc.number = mdi.disk.number;
	451	disc.raid_disk = mdi.disk.raid_disk;
	452	disc.state = mdi.disk.state;
4d20d744 NB	453	if (dv->writemostly)
4d20d744 NB	454	disc.state \|= 1 << MD_DISK_WRITEMOSTLY;
fe80f49b NB	455	if (ioctl(fd, ADD_NEW_DISK, &disc) == 0) {
	456	if (verbose >= 0)
	457	fprintf(stderr, Name ": re-added %s\n", dv->devname);
308e1801	458	continue;
fe80f49b NB	459	}
	460	/* fall back on normal-add */
	461	}
892debc8	462	}
4a39c6f2 NB	463	} else {
	464	/* non-persistent. Must ensure that new drive
	465	* is at least array.size big.
	466	*/
7a3be72f	467	if (ldsize/512 < array_size) {
4a39c6f2 NB	468	fprintf(stderr, Name ": %s not large enough to join array\n",
	469	dv->devname);
	470	return 1;
	471	}
892debc8	472	}
7eae7080 NB	473	/* in 2.6.17 and earlier, version-1 superblocks won't
	474	* use the number we write, but will choose a free number.
	475	* we must choose the same free number, which requires
	476	* starting at 'raid_disks' and counting up
	477	*/
3da92f27	478	for (j = array.raid_disks; j< tst->max_devs; j++) {
aa88f531	479	disc.number = j;
682c7051 NB	480	if (ioctl(fd, GET_DISK_INFO, &disc))
	481	break;
	482	if (disc.major==0 && disc.minor==0)
	483	break;
	484	if (disc.state & 8) /* removed */
	485	break;
	486	}
0df46c2a NB	487	disc.major = major(stb.st_rdev);
0df46c2a NB	488	disc.minor = minor(stb.st_rdev);
892debc8 NB	489	disc.number =j;
892debc8 NB	490	disc.state = 0;
fe80f49b	491	if (array.not_persistent==0) {
111d01fc	492	int dfd;
fe80f49b NB	493	if (dv->writemostly)
fe80f49b NB	494	disc.state \|= 1 << MD_DISK_WRITEMOSTLY;
6416d527	495	dfd = open(dv->devname, O_RDWR \| O_EXCL\|O_DIRECT);
111d01fc NB	496	tst->ss->add_to_super(tst, &disc, dfd,
	497	dv->devname);
	498	/* write_init_super will close 'dfd' */
	499	if (tst->ss->write_init_super(tst))
fe80f49b NB	500	return 1;
	501	} else if (dv->re_add) {
	502	/* this had better be raid1.
	503	* As we are "--re-add"ing we must find a spare slot
	504	* to fill.
	505	*/
	506	char *used = malloc(array.raid_disks);
	507	memset(used, 0, array.raid_disks);
3da92f27	508	for (j=0; j< tst->max_devs; j++) {
fe80f49b NB	509	mdu_disk_info_t disc2;
	510	disc2.number = j;
	511	if (ioctl(fd, GET_DISK_INFO, &disc2))
	512	continue;
	513	if (disc2.major==0 && disc2.minor==0)
	514	continue;
	515	if (disc2.state & 8) /* removed */
	516	continue;
	517	if (disc2.raid_disk < 0)
	518	continue;
	519	if (disc2.raid_disk > array.raid_disks)
	520	continue;
	521	used[disc2.raid_disk] = 1;
	522	}
	523	for (j=0 ; j<array.raid_disks; j++)
	524	if (!used[j]) {
	525	disc.raid_disk = j;
	526	disc.state \|= (1<<MD_DISK_SYNC);
	527	break;
	528	}
	529	}
d8def61c NB	530	if (dv->writemostly)
d8def61c NB	531	disc.state \|= (1 << MD_DISK_WRITEMOSTLY);
682c7051	532	if (ioctl(fd,ADD_NEW_DISK, &disc)) {
f277ce36 NB	533	fprintf(stderr, Name ": add new device failed for %s as %d: %s\n",
f277ce36 NB	534	dv->devname, j, strerror(errno));
682c7051 NB	535	return 1;
682c7051 NB	536	}
dab6685f NB	537	if (verbose >= 0)
dab6685f NB	538	fprintf(stderr, Name ": added %s\n", dv->devname);
682c7051 NB	539	break;
	540
	541	case 'r':
	542	/* hot remove */
f7e7067b	543	if (tst->subarray[0]) {
f7dd881f DW	544	fprintf(stderr, Name ": Cannot remove disks from a"
	545	" \'member\' array, perform this"
	546	" operation on the parent container\n");
	547	return 1;
f94d52f4 NB	548	}
	549	if (tst->ss->external) {
	550	/* To remove a device from a container, we must
	551	* check that it isn't in use in an array.
	552	* This involves looking in the 'holders'
	553	* directory - there must be just one entry,
	554	* the container.
	555	* To ensure that it doesn't get used as a
	556	* hold spare while we are checking, we
	557	* get an O_EXCL open on the container
	558	*/
	559	int dnum = fd2devnum(fd);
	560	lfd = open_dev_excl(dnum);
	561	if (lfd < 0) {
	562	fprintf(stderr, Name
	563	": Cannot get exclusive access "
	564	" to container - odd\n");
	565	return 1;
	566	}
	567	if (!sysfs_unique_holder(dnum, stb.st_rdev)) {
	568	fprintf(stderr, Name
	569	": %s is %s, cannot remove.\n",
	570	dnprintable,
	571	errno == EEXIST ? "still in use":
	572	"not a member");
	573	close(lfd);
	574	return 1;
	575	}
	576	}
82b27616	577	/* FIXME check that it is a current member */
f94d52f4 NB	578	err = ioctl(fd, HOT_REMOVE_DISK, (unsigned long)stb.st_rdev);
	579	if (err && errno == ENODEV) {
	580	/* Old kernels rejected this if no personality
	581	* registered */
	582	struct mdinfo *sra = sysfs_read(fd, 0, GET_DEVS);
	583	struct mdinfo *dv = NULL;
	584	if (sra)
	585	dv = sra->devs;
	586	for ( ; dv ; dv=dv->next)
	587	if (dv->disk.major == major(stb.st_rdev) &&
	588	dv->disk.minor == minor(stb.st_rdev))
	589	break;
	590	if (dv)
	591	err = sysfs_set_str(sra, dv,
	592	"state", "remove");
	593	else
	594	err = -1;
	595	if (sra)
	596	sysfs_free(sra);
	597	}
	598	if (err) {
b80da661 NB	599	fprintf(stderr, Name ": hot remove failed "
	600	"for %s: %s\n", dnprintable,
	601	strerror(errno));
f94d52f4 NB	602	if (lfd >= 0)
f94d52f4 NB	603	close(lfd);
682c7051 NB	604	return 1;
682c7051 NB	605	}
f94d52f4	606	close(lfd);
dab6685f	607	if (verbose >= 0)
b80da661 NB	608	fprintf(stderr, Name ": hot removed %s\n",
b80da661 NB	609	dnprintable);
682c7051 NB	610	break;
	611
	612	case 'f': /* set faulty */
	613	/* FIXME check current member */
cd29a5c8	614	if (ioctl(fd, SET_DISK_FAULTY, (unsigned long) stb.st_rdev)) {
b83d95f3	615	fprintf(stderr, Name ": set device faulty failed for %s: %s\n",
b80da661	616	dnprintable, strerror(errno));
682c7051 NB	617	return 1;
682c7051 NB	618	}
dab6685f NB	619	if (verbose >= 0)
dab6685f NB	620	fprintf(stderr, Name ": set %s faulty in %s\n",
b80da661	621	dnprintable, devname);
682c7051 NB	622	break;
	623	}
	624	}
	625	return 0;
aba69144	626
64c4757e	627	}
1f48664b NB	628
	629	int autodetect(void)
	630	{
	631	/* Open any md device, and issue the RAID_AUTORUN ioctl */
	632	int rv = 1;
	633	int fd = dev_open("9:0", O_RDONLY);
	634	if (fd >= 0) {
	635	if (ioctl(fd, RAID_AUTORUN, 0) == 0)
	636	rv = 0;
	637	close(fd);
	638	}
	639	return rv;
	640	}
435d4ebb	641	#endif