X-Git-Url: http://git.ipfire.org/?p=thirdparty%2Fmdadm.git;a=blobdiff_plain;f=Manage.c;h=0c6b59cb06ed3ef2b56a45b9ca20e7b59e0693b1;hp=9e643163cf273e4dff84897b7f683f3a9b6d0910;hb=daf7a3ce96e517360b0456c3a9cca0771b94462d;hpb=16c6fa807c1e43a1a12bbeb7668b4166d14b2cc8

diff --git a/Manage.c b/Manage.c
index 9e643163..0c6b59cb 100644
--- a/Manage.c
+++ b/Manage.c
@@ -1,7 +1,7 @@
 /*
  * mdadm - manage Linux "md" devices aka RAID arrays.
  *
- * Copyright (C) 2001-2002 Neil Brown <neilb@cse.unsw.edu.au>
+ * Copyright (C) 2001-2006 Neil Brown <neilb@suse.de>
  *
  *
  *    This program is free software; you can redistribute it and/or modify
@@ -45,7 +45,7 @@ int Manage_ro(char *devname, int fd, int readonly)
 	 *
 	 */
 	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;
@@ -55,7 +55,7 @@ int Manage_ro(char *devname, int fd, int readonly)
 			devname);
 		return 1;
 	}
-	
+
 	if (readonly>0) {
 		if (ioctl(fd, STOP_ARRAY_RO, NULL)) {
 			fprintf(stderr, Name ": failed to set readonly for %s: %s\n",
@@ -69,9 +69,11 @@ int Manage_ro(char *devname, int fd, int readonly)
 			return 1;
 		}
 	}
-	return 0;			
+	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
@@ -86,7 +88,7 @@ int Manage_runstop(char *devname, int fd, int runstop, int quiet)
 			return 1;
 		}
 	}
-	
+
 	if (md_get_version(fd) < 9000) {
 		fprintf(stderr, Name ": need md driver version 0.90.0 or later\n");
 		return 1;
@@ -104,13 +106,63 @@ int Manage_runstop(char *devname, int fd, int runstop, int quiet)
 				devname, strerror(errno));
 			return 1;
 		}
+		if (quiet <= 0)
+			fprintf(stderr, Name ": started %s\n", devname);
 	} else if (runstop < 0){
-		if (ioctl(fd, STOP_ARRAY, NULL)) {
-			if (!quiet)
-				fprintf(stderr, Name ": fail to stop array %s: %s\n",
+		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;
 }
@@ -162,35 +214,126 @@ int Manage_subdevs(char *devname, int fd,
 	 *	   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;
-	mddev_dev_t dv;
+	unsigned long long array_size;
+	mddev_dev_t dv, next = NULL;
 	struct stat stb;
-	int j;
+	int j, jnext = 0;
 	int tfd;
-	struct supertype *st;
-	void *dsuper = NULL;
-	void *osuper = NULL; /* original super */
+	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;
 	}
-	for (dv = devlist ; dv; dv=dv->next) {
-		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;
+
+	/* 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:
@@ -198,28 +341,36 @@ int Manage_subdevs(char *devname, int fd,
 				dv->devname, dv->disposition);
 			return 1;
 		case 'a':
-			/* add the device - hot or cold */
-			st = super_by_version(array.major_version,
-					      array.minor_version);
-			if (!st) {
-				fprintf(stderr, Name ": unsupport array - version %d.%d\n",
-					array.major_version, array.minor_version);
+			/* 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);
+			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, &osuper, NULL);
-			/* will use osuper later */
+				st->ss->load_super(st, tfd, NULL);
+
+			if (!get_dev_size(tfd, dv->devname, &ldsize)) {
+				close(tfd);
+				return 1;
+			}
 			close(tfd);
 
-			if (array.major_version == 0 &&
+
+			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) {
@@ -236,10 +387,18 @@ int Manage_subdevs(char *devname, int fd,
 
 			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 
+				 * a spare slot to use
 				 */
-				for (j=0; j<st->max_devs; j++) {
+				for (j = 0; j < tst->max_devs; j++) {
 					char *dev;
 					int dfd;
 					disc.number = j;
@@ -253,14 +412,15 @@ int Manage_subdevs(char *devname, int fd,
 					if (!dev) continue;
 					dfd = dev_open(dev, O_RDONLY);
 					if (dfd < 0) continue;
-					if (st->ss->load_super(st, dfd, &dsuper, NULL)) {
+					if (tst->ss->load_super(tst, dfd,
+								NULL)) {
 						close(dfd);
 						continue;
 					}
 					close(dfd);
 					break;
 				}
-				if (!dsuper) {
+				if (!tst->sb) {
 					fprintf(stderr, Name ": cannot find valid superblock in this array - HELP\n");
 					return 1;
 				}
@@ -268,22 +428,30 @@ int Manage_subdevs(char *devname, int fd,
 				 * and was temporarily removed, and is now being re-added.
 				 * If so, we can simply re-add it.
 				 */
-				st->ss->uuid_from_super(duuid, dsuper);
-			
-				if (osuper) {
-					st->ss->uuid_from_super(ouuid, osuper);
+				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) {
-						/* look close enough for now.  Kernel
-						 * will worry about where a bitmap
-						 * based reconstruct is possible
+						/* looks close enough for now.  Kernel
+						 * will worry about whether a bitmap
+						 * based reconstruction is possible.
 						 */
 						struct mdinfo mdi;
-						st->ss->getinfo_super(&mdi, osuper);
+						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);
@@ -292,8 +460,22 @@ int Manage_subdevs(char *devname, int fd,
 						/* 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;
+				}
 			}
-			for (j=0; j< st->max_devs; j++) {
+			/* 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;
@@ -307,10 +489,14 @@ int Manage_subdevs(char *devname, int fd,
 			disc.number =j;
 			disc.state = 0;
 			if (array.not_persistent==0) {
+				int dfd;
 				if (dv->writemostly)
 					disc.state |= 1 << MD_DISK_WRITEMOSTLY;
-				st->ss->add_to_super(dsuper, &disc);
-				if (st->ss->write_init_super(st, dsuper, &disc, dv->devname))
+				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.
@@ -319,7 +505,7 @@ int Manage_subdevs(char *devname, int fd,
 				 */
 				char *used = malloc(array.raid_disks);
 				memset(used, 0, array.raid_disks);
-				for (j=0; j< st->max_devs; j++) {
+				for (j=0; j< tst->max_devs; j++) {
 					mdu_disk_info_t disc2;
 					disc2.number = j;
 					if (ioctl(fd, GET_DISK_INFO, &disc2))
@@ -341,6 +527,8 @@ int Manage_subdevs(char *devname, int fd,
 						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));
@@ -352,29 +540,102 @@ int Manage_subdevs(char *devname, int fd,
 
 		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 */
-			if (ioctl(fd, HOT_REMOVE_DISK, (unsigned long)stb.st_rdev)) {
-				fprintf(stderr, Name ": hot remove failed for %s: %s\n",
-					dv->devname, strerror(errno));
+			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", dv->devname);
+				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",
-					dv->devname, strerror(errno));
+					dnprintable, strerror(errno));
 				return 1;
 			}
 			if (verbose >= 0)
 				fprintf(stderr, Name ": set %s faulty in %s\n",
-					dv->devname, devname);
+					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