mdadm: add device to a container

Adding a device updates the container and then mdmon takes action upon
noticing a change in devices.  This reuses the container version of
add_to_super to create a new record for the device.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: NeilBrown <neilb@suse.de>

Manage.c

@@ -404,8 +404,7 @@ int Manage_subdevs(char *devname, int fd,
 				return 1;
 			}
 
-			if (array.not_persistent == 0) {
+			if (array.not_persistent == 0 || tst->ss->external) {
-
 				/* Make sure device is large enough */
 				if (tst->ss->avail_size(tst, ldsize/512) <
 				    array_size) {
@@ -415,9 +414,13 @@ int Manage_subdevs(char *devname, int fd,
 				}
 
 				/* need to find a sample superblock to copy, and
-				 * a spare slot to use
+				 * a spare slot to use.
+				 * For 'external' array (well, container based),
+				 * We can just load the metadata for the array.
 				 */
-				for (j = 0; j < tst->max_devs; j++) {
+				if (tst->ss->external) {
+					tst->ss->load_super(tst, fd, NULL);
+				} else for (j = 0; j < tst->max_devs; j++) {
 					char *dev;
 					int dfd;
 					disc.number = j;
@@ -439,6 +442,7 @@ int Manage_subdevs(char *devname, int fd,
 					close(dfd);
 					break;
 				}
+				/* FIXME this is a bad test to be using */
 				if (!tst->sb) {
 					fprintf(stderr, Name ": cannot find valid superblock in this array - HELP\n");
 					return 1;
@@ -507,7 +511,7 @@ int Manage_subdevs(char *devname, int fd,
 			disc.minor = minor(stb.st_rdev);
 			disc.number =j;
 			disc.state = 0;
-			if (array.not_persistent==0) {
+			if (array.not_persistent==0 || tst->ss->external) {
 				int dfd;
 				if (dv->writemostly)
 					disc.state |= 1 << MD_DISK_WRITEMOSTLY;
@@ -515,7 +519,10 @@ int Manage_subdevs(char *devname, int fd,
 				tst->ss->add_to_super(tst, &disc, dfd,
 						      dv->devname);
 				/* write_init_super will close 'dfd' */
-				if (tst->ss->write_init_super(tst))
+				if (tst->ss->external)
+					/* mdmon will write the metadata */
+					close(dfd);
+				else if (tst->ss->write_init_super(tst))
 					return 1;
 			} else if (dv->re_add) {
 				/*  this had better be raid1.
@@ -548,7 +555,52 @@ int Manage_subdevs(char *devname, int fd,
 			}
 			if (dv->writemostly)
 				disc.state |= (1 << MD_DISK_WRITEMOSTLY);
-			if (ioctl(fd,ADD_NEW_DISK, &disc)) {
+			if (tst->ss->external) {
+				/* add a disk to an external metadata container
+				 * only if mdmon is around to see it
+				 */
+				struct mdinfo new_mdi;
+				struct mdinfo *sra;
+				int container_fd;
+				int devnum = fd2devnum(fd);
+
+				container_fd = open_dev_excl(devnum);
+				if (container_fd < 0) {
+					fprintf(stderr, Name ": add failed for %s:"
+						" could not get exclusive access to container\n",
+						dv->devname);
+					return 1;
+				}
+
+				if (!mdmon_running(devnum)) {
+					fprintf(stderr, Name ": add failed for %s: mdmon not running\n",
+						dv->devname);
+					close(container_fd);
+					return 1;
+				}
+
+				sra = sysfs_read(container_fd, -1, 0);
+				if (!sra) {
+					fprintf(stderr, Name ": add failed for %s: sysfs_read failed\n",
+						dv->devname);
+					close(container_fd);
+					return 1;
+				}
+				sra->array.level = LEVEL_CONTAINER;
+				/* Need to set data_offset and component_size */
+				tst->ss->getinfo_super(tst, &new_mdi);
+				new_mdi.disk.major = disc.major;
+				new_mdi.disk.minor = disc.minor;
+				if (sysfs_add_disk(sra, &new_mdi) != 0) {
+					fprintf(stderr, Name ": add new device to external metadata"
+						" failed for %s\n", dv->devname);
+					close(container_fd);
+					return 1;
+				}
+				ping_monitor(devnum2devname(devnum));
+				sysfs_free(sra);
+				close(container_fd);
+			} else 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;

managemon.c

@@ -218,6 +218,34 @@ static void queue_metadata_update(struct metadata_update *mu)
 	*qp = mu;
 }
 
+static void add_disk_to_container(struct supertype *st, struct mdinfo *sd)
+{
+	int dfd;
+	char nm[20];
+	struct metadata_update *update = NULL;
+	mdu_disk_info_t dk = {
+		.number = -1,
+		.major = sd->disk.major,
+		.minor = sd->disk.minor,
+		.raid_disk = -1,
+		.state = 0,
+	};
+
+	dprintf("%s: add %d:%d to container\n",
+		__func__, sd->disk.major, sd->disk.minor);
+
+	sprintf(nm, "%d:%d", sd->disk.major, sd->disk.minor);
+	dfd = dev_open(nm, O_RDWR);
+	if (dfd < 0)
+		return;
+
+	st->update_tail = &update;
+	st->ss->add_to_super(st, &dk, dfd, NULL);
+	st->ss->write_init_super(st);
+	queue_metadata_update(update);
+	st->update_tail = NULL;
+}
+
 static void manage_container(struct mdstat_ent *mdstat,
 			     struct supertype *container)
 {
@@ -256,6 +284,16 @@ static void manage_container(struct mdstat_ent *mdstat,
 			} else
 				cdp = &(*cdp)->next;
 		}
+
+		/* check for additions */
+		for (di = mdi->devs; di; di = di->next) {
+			for (cd = container->devs; cd; cd = cd->next)
+				if (di->disk.major == cd->disk.major &&
+				    di->disk.minor == cd->disk.minor)
+					break;
+			if (!cd)
+				add_disk_to_container(container, di);
+		}
 		sysfs_free(mdi);
 		container->devcnt = mdstat->devcnt;
 	}

super-intel.c

@@ -174,7 +174,8 @@ struct intel_super {
 		struct imsm_disk disk;
 		int fd;
 	} *disks;
-    struct bbm_log *bbm_log;
+	struct dl *add; /* list of disks to add while mdmon active */
+	struct bbm_log *bbm_log;
 };
 
 struct extent {
@@ -185,6 +186,7 @@ struct extent {
 enum imsm_update_type {
 	update_activate_spare,
 	update_create_array,
+	update_add_disk,
 };
 
 struct imsm_update_activate_spare {
@@ -201,6 +203,10 @@ struct imsm_update_create_array {
 	struct imsm_dev dev;
 };
 
+struct imsm_update_add_disk {
+	enum imsm_update_type type;
+};
+
 static int imsm_env_devname_as_serial(void)
 {
 	char *val = getenv("IMSM_DEVNAME_AS_SERIAL");
@@ -564,7 +570,13 @@ static int match_home_imsm(struct supertype *st, char *homehost)
 
 static void uuid_from_super_imsm(struct supertype *st, int uuid[4])
 {
-	printf("%s\n", __FUNCTION__);
+	/* imsm does not track uuid's so just make sure we never return
+	 * the same value twice to break uuid matching in Manage_subdevs
+	 * FIXME what about the use of uuid's with bitmap's?
+	 */
+	static int dummy_id = 0;
+
+	uuid[0] = dummy_id++;
 }
 
 #if 0
@@ -1264,11 +1276,11 @@ static int load_super_imsm_all(struct supertype *st, int fd, void **sbp,
 	}
 
 	*sbp = super;
+	st->container_dev = fd2devnum(fd);
 	if (st->ss == NULL) {
 		st->ss = &super_imsm;
 		st->minor_version = 0;
 		st->max_devs = IMSM_MAX_DEVICES;
-		st->container_dev = fd2devnum(fd);
 	}
 
 	return 0;
@@ -1556,7 +1568,6 @@ static void add_to_super_imsm(struct supertype *st, mdu_disk_info_t *dk,
 	dd->minor = minor(stb.st_rdev);
 	dd->index = -1;
 	dd->devname = devname ? strdup(devname) : NULL;
-	dd->next = super->disks;
 	dd->fd = fd;
 	rv = imsm_read_serial(fd, devname, dd->serial);
 	if (rv) {
@@ -1576,7 +1587,14 @@ static void add_to_super_imsm(struct supertype *st, mdu_disk_info_t *dk,
 		dd->disk.scsi_id = __cpu_to_le32(id);
 	else
 		dd->disk.scsi_id = __cpu_to_le32(0);
-	super->disks = dd;
+
+	if (st->update_tail) {
+		dd->next = super->add;
+		super->add = dd;
+	} else {
+		dd->next = super->disks;
+		super->disks = dd;
+	}
 }
 
 static int store_imsm_mpb(int fd, struct intel_super *super);
@@ -1688,43 +1706,76 @@ static int write_super_imsm(struct intel_super *super, int doclose)
 	return 0;
 }
 
+static int create_array(struct supertype *st)
+{
+	size_t len;
+	struct imsm_update_create_array *u;
+	struct intel_super *super = st->sb;
+	struct imsm_dev *dev = get_imsm_dev(super, super->current_vol);
+
+	len = sizeof(*u) - sizeof(*dev) + sizeof_imsm_dev(dev, 0);
+	u = malloc(len);
+	if (!u) {
+		fprintf(stderr, "%s: failed to allocate update buffer\n",
+			__func__);
+		return 1;
+	}
+
+	u->type = update_create_array;
+	u->dev_idx = super->current_vol;
+	imsm_copy_dev(&u->dev, dev);
+	append_metadata_update(st, u, len);
+
+	return 0;
+}
+
+static int add_disk(struct supertype *st)
+{
+	struct intel_super *super = st->sb;
+	size_t len;
+	struct imsm_update_add_disk *u;
+
+	if (!super->add)
+		return 0;
+
+	len = sizeof(*u);
+	u = malloc(len);
+	if (!u) {
+		fprintf(stderr, "%s: failed to allocate update buffer\n",
+			__func__);
+		return 1;
+	}
+
+	u->type = update_add_disk;
+	append_metadata_update(st, u, len);
+
+	return 0;
+}
+
 static int write_init_super_imsm(struct supertype *st)
 {
 	if (st->update_tail) {
-		/* queue the recently created array as a metadata update */
+		/* queue the recently created array / added disk
-		size_t len;
+		 * as a metadata update */
-		struct imsm_update_create_array *u;
 		struct intel_super *super = st->sb;
-		struct imsm_dev *dev;
 		struct dl *d;
+		int rv;
 
-		if (super->current_vol < 0 ||
+		/* determine if we are creating a volume or adding a disk */
-		    !(dev = get_imsm_dev(super, super->current_vol))) {
+		if (super->current_vol < 0) {
-			fprintf(stderr, "%s: could not determine sub-array\n",
+			/* in the add disk case we are running in mdmon
-				__func__);
+			 * context, so don't close fd's
-			return 1;
+			 */
-		}
+			return add_disk(st);
-
+		} else
-
+			rv = create_array(st);
-		len = sizeof(*u) - sizeof(*dev) + sizeof_imsm_dev(dev, 0);
-		u = malloc(len);
-		if (!u) {
-			fprintf(stderr, "%s: failed to allocate update buffer\n",
-				__func__);
-			return 1;
-		}
-
-		u->type = update_create_array;
-		u->dev_idx = super->current_vol;
-		imsm_copy_dev(&u->dev, dev);
-		append_metadata_update(st, u, len);
 
 		for (d = super->disks; d ; d = d->next) {
 			close(d->fd);
 			d->fd = -1;
 		}
 
-		return 0;
+		return rv;
 	} else
 		return write_super_imsm(st->sb, 1);
 }
@@ -2779,6 +2830,36 @@ static void imsm_process_update(struct supertype *st,
 		}
 		break;
 	}
+	case update_add_disk:
+
+		/* we may be able to repair some arrays if disks are
+		 * being added */
+		if (super->add) {
+			struct active_array *a;
+ 			for (a = st->arrays; a; a = a->next)
+				a->check_degraded = 1;
+		}
+		/* check if we can add / replace some disks in the
+		 * metadata */
+		while (super->add) {
+			struct dl **dlp, *dl, *al;
+			al = super->add;
+			super->add = al->next;
+			for (dlp = &super->disks; *dlp ; ) {
+				if (memcmp(al->serial, (*dlp)->serial,
+					   MAX_RAID_SERIAL_LEN) == 0) {
+					dl = *dlp;
+					*dlp = (*dlp)->next;
+					__free_imsm_disk(dl);
+					break;
+				} else
+					dlp = &(*dlp)->next;
+			}
+			al->next = super->disks;
+			super->disks = al;
+		}
+
+		break;
 	}
 }