X-Git-Url: http://git.ipfire.org/?a=blobdiff_plain;f=Create.c;h=2bf7ebe2dfac380e2c7abe38ab2c7dce38360d50;hb=61018da020ab3e25b21e0dffc7ecc4f708207a0b;hp=6cf509aa16687c7f57421e140213ca015dd39e37;hpb=3da92f272d017b1817b13f37f41c1ed4d6117291;p=thirdparty%2Fmdadm.git diff --git a/Create.c b/Create.c index 6cf509aa..2bf7ebe2 100644 --- a/Create.c +++ b/Create.c @@ -1,7 +1,7 @@ /* * mdadm - manage Linux "md" devices aka RAID arrays. * - * Copyright (C) 2001-2006 Neil Brown + * Copyright (C) 2001-2009 Neil Brown * * * This program is free software; you can redistribute it and/or modify @@ -19,12 +19,7 @@ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA * * Author: Neil Brown - * Email: - * Paper: Neil Brown - * School of Computer Science and Engineering - * The University of New South Wales - * Sydney, 2052 - * Australia + * Email: */ #include "mdadm.h" @@ -32,12 +27,50 @@ #include "md_p.h" #include -int Create(struct supertype *st, char *mddev, int mdfd, +static int default_layout(struct supertype *st, int level, int verbose) +{ + int layout = UnSet; + + if (st && st->ss->default_layout) + layout = st->ss->default_layout(level); + + if (layout == UnSet) + switch(level) { + default: /* no layout */ + layout = 0; + break; + case 10: + layout = 0x102; /* near=2, far=1 */ + if (verbose > 0) + fprintf(stderr, + Name ": layout defaults to n2\n"); + break; + case 5: + case 6: + layout = map_name(r5layout, "default"); + if (verbose > 0) + fprintf(stderr, + Name ": layout defaults to %s\n", map_num(r5layout, layout)); + break; + case LEVEL_FAULTY: + layout = map_name(faultylayout, "default"); + + if (verbose > 0) + fprintf(stderr, + Name ": layout defaults to %s\n", map_num(faultylayout, layout)); + break; + } + + return layout; +} + + +int Create(struct supertype *st, char *mddev, int chunk, int level, int layout, unsigned long long size, int raiddisks, int sparedisks, char *name, char *homehost, int *uuid, int subdevs, mddev_dev_t devlist, int runstop, int verbose, int force, int assume_clean, - char *bitmap_file, int bitmap_chunk, int write_behind, int delay) + char *bitmap_file, int bitmap_chunk, int write_behind, int delay, int autof) { /* * Create a new raid array. @@ -55,6 +88,7 @@ int Create(struct supertype *st, char *mddev, int mdfd, * if runstop==run, or raiddisks disks were used, * RUN_ARRAY */ + int mdfd; unsigned long long minsize=0, maxsize=0; char *mindisc = NULL; char *maxdisc = NULL; @@ -63,33 +97,38 @@ int Create(struct supertype *st, char *mddev, int mdfd, int fail=0, warn=0; struct stat stb; int first_missing = subdevs * 2; + int second_missing = subdevs * 2; int missing_disks = 0; int insert_point = subdevs * 2; /* where to insert a missing drive */ + int total_slots; int pass; int vers; int rv; int bitmap_fd; + int have_container = 0; + int container_fd = -1; + int need_mdmon = 0; unsigned long long bitmapsize; + struct mdinfo info, *infos; + int did_default = 0; + int do_default_layout = 0; + unsigned long safe_mode_delay = 0; + char chosen_name[1024]; + struct map_ent *map = NULL; + unsigned long long newsize; - mdu_array_info_t array; int major_num = BITMAP_MAJOR_HI; - memset(&array, 0, sizeof(array)); + memset(&info, 0, sizeof(info)); - vers = md_get_version(mdfd); - if (vers < 9000) { - fprintf(stderr, Name ": Create requires md driver version 0.90.0 or later\n"); - return 1; - } else { - mdu_array_info_t inf; - memset(&inf, 0, sizeof(inf)); - ioctl(mdfd, GET_ARRAY_INFO, &inf); - if (inf.working_disks != 0) { - fprintf(stderr, Name ": another array by this name" - " is already running.\n"); - return 1; - } + if (level == UnSet) { + /* "ddf" and "imsm" metadata only supports one level - should possibly + * push this into metadata handler?? + */ + if (st && (st->ss == &super_ddf || st->ss == &super_imsm)) + level = LEVEL_CONTAINER; } + if (level == UnSet) { fprintf(stderr, Name ": a RAID level is needed to create an array.\n"); @@ -115,11 +154,55 @@ int Create(struct supertype *st, char *mddev, int mdfd, Name ": This level does not support spare devices\n"); return 1; } + + if (subdevs == 1 && strcmp(devlist->devname, "missing") != 0) { + /* If given a single device, it might be a container, and we can + * extract a device list from there + */ + mdu_array_info_t inf; + int fd; + + memset(&inf, 0, sizeof(inf)); + fd = open(devlist->devname, O_RDONLY); + if (fd >= 0 && + ioctl(fd, GET_ARRAY_INFO, &inf) == 0 && + inf.raid_disks == 0) { + /* yep, looks like a container */ + if (st) { + rv = st->ss->load_super(st, fd, + devlist->devname); + if (rv == 0) + have_container = 1; + } else { + st = guess_super(fd); + if (st && !(rv = st->ss-> + load_super(st, fd, + devlist->devname))) + have_container = 1; + else + st = NULL; + } + if (have_container) { + subdevs = raiddisks; + first_missing = subdevs * 2; + second_missing = subdevs * 2; + insert_point = subdevs * 2; + } + } + if (fd >= 0) + close(fd); + } + if (st && st->ss->external && sparedisks) { + fprintf(stderr, + Name ": This metadata type does not support " + "spare disks at create time\n"); + return 1; + } if (subdevs > raiddisks+sparedisks) { fprintf(stderr, Name ": You have listed more devices (%d) than are in the array(%d)!\n", subdevs, raiddisks+sparedisks); return 1; } - if (subdevs < raiddisks+sparedisks) { + if (!have_container && subdevs < raiddisks+sparedisks) { fprintf(stderr, Name ": You haven't given enough devices (real or missing) to create this array\n"); return 1; } @@ -130,32 +213,12 @@ int Create(struct supertype *st, char *mddev, int mdfd, } /* now set some defaults */ - if (layout == UnSet) - switch(level) { - default: /* no layout */ - layout = 0; - break; - case 10: - layout = 0x102; /* near=2, far=1 */ - if (verbose > 0) - fprintf(stderr, - Name ": layout defaults to n1\n"); - break; - case 5: - case 6: - layout = map_name(r5layout, "default"); - if (verbose > 0) - fprintf(stderr, - Name ": layout defaults to %s\n", map_num(r5layout, layout)); - break; - case LEVEL_FAULTY: - layout = map_name(faultylayout, "default"); - if (verbose > 0) - fprintf(stderr, - Name ": layout defaults to %s\n", map_num(faultylayout, layout)); - break; - } + + if (layout == UnSet) { + do_default_layout = 1; + layout = default_layout(st, level, verbose); + } if (level == 10) /* check layout fits in array*/ @@ -171,8 +234,19 @@ int Create(struct supertype *st, char *mddev, int mdfd, case 10: case 6: case 0: - case LEVEL_LINEAR: /* linear */ if (chunk == 0) { + if (st && st->ss->default_chunk) + chunk = st->ss->default_chunk(st); + + chunk = chunk ? : 512; + + if (verbose > 0) + fprintf(stderr, Name ": chunk size defaults to %dK\n", chunk); + } + break; + case LEVEL_LINEAR: + /* a chunksize of zero 0s perfectly valid (and preferred) since 2.6.16 */ + if (get_linux_version() < 2006016 && chunk == 0) { chunk = 64; if (verbose > 0) fprintf(stderr, Name ": chunk size defaults to 64K\n"); @@ -181,6 +255,7 @@ int Create(struct supertype *st, char *mddev, int mdfd, case 1: case LEVEL_FAULTY: case LEVEL_MULTIPATH: + case LEVEL_CONTAINER: if (chunk) { chunk = 0; if (verbose > 0) @@ -191,36 +266,38 @@ int Create(struct supertype *st, char *mddev, int mdfd, fprintf(stderr, Name ": unknown level %d\n", level); return 1; } + + if (size && chunk) + size &= ~(unsigned long long)(chunk - 1); + newsize = size * 2; + if (st && ! st->ss->validate_geometry(st, level, layout, raiddisks, + chunk, size*2, NULL, &newsize, verbose>=0)) + return 1; + if (size == 0) { + size = newsize / 2; + if (size && verbose > 0) + fprintf(stderr, Name ": setting size to %lluK\n", + (unsigned long long)size); + } /* now look at the subdevs */ - array.active_disks = 0; - array.working_disks = 0; + info.array.active_disks = 0; + info.array.working_disks = 0; dnum = 0; - for (dv=devlist; dv; dv=dv->next, dnum++) { + for (dv=devlist; dv && !have_container; dv=dv->next, dnum++) { char *dname = dv->devname; - unsigned long long ldsize, freesize; - int fd; + unsigned long long freesize; if (strcasecmp(dname, "missing")==0) { if (first_missing > dnum) first_missing = dnum; + if (second_missing > dnum && dnum > first_missing) + second_missing = dnum; missing_disks ++; continue; } - array.working_disks++; + info.array.working_disks++; if (dnum < raiddisks) - array.active_disks++; - fd = open(dname, O_RDONLY|O_EXCL, 0); - if (fd <0 ) { - fprintf(stderr, Name ": Cannot open %s: %s\n", - dname, strerror(errno)); - fail=1; - continue; - } - if (!get_dev_size(fd, dname, &ldsize)) { - fail = 1; - close(fd); - continue; - } + info.array.active_disks++; if (st == NULL) { struct createinfo *ci = conf_get_create_info(); if (ci) @@ -228,33 +305,46 @@ int Create(struct supertype *st, char *mddev, int mdfd, } if (st == NULL) { /* Need to choose a default metadata, which is different - * depending on the sizes of devices + * depending on geometry of array. */ int i; char *name = "default"; - if (level >= 1 && ldsize > (0x7fffffffULL<<10)) - name = "default/large"; - for(i=0; !st && superlist[i]; i++) + for(i=0; !st && superlist[i]; i++) { st = superlist[i]->match_metadata_desc(name); + if (do_default_layout) + layout = default_layout(st, level, verbose); + if (st && !st->ss->validate_geometry + (st, level, layout, raiddisks, + chunk, size*2, dname, &freesize, + verbose > 0)) + st = NULL; + } if (!st) { - fprintf(stderr, Name ": internal error - no default metadata style\n"); + fprintf(stderr, Name ": device %s not suitable " + "for any style of array\n", + dname); exit(2); } - if (st->ss->major != 0 || + if (st->ss != &super0 || st->minor_version != 90) - fprintf(stderr, Name ": Defaulting to version" - " %d.%d metadata\n", - st->ss->major, - st->minor_version); - } - freesize = st->ss->avail_size(st, ldsize >> 9); - if (freesize == 0) { - fprintf(stderr, Name ": %s is too small: %luK\n", - dname, (unsigned long)(ldsize>>10)); - fail = 1; - close(fd); - continue; + did_default = 1; + } else { + if (do_default_layout) + layout = default_layout(st, level, verbose); + if (!st->ss->validate_geometry(st, level, layout, + raiddisks, + chunk, size*2, dname, + &freesize, + verbose >= 0)) { + + fprintf(stderr, + Name ": %s is not suitable for " + "this array.\n", + dname); + fail = 1; + continue; + } } freesize /= 2; /* convert to K */ @@ -264,10 +354,10 @@ int Create(struct supertype *st, char *mddev, int mdfd, } if (size && freesize < size) { - fprintf(stderr, Name ": %s is smaller that given size." - " %lluK < %lluK + superblock\n", dname, freesize, size); + fprintf(stderr, Name ": %s is smaller than given size." + " %lluK < %lluK + metadata\n", + dname, freesize, size); fail = 1; - close(fd); continue; } if (maxdisc == NULL || (maxdisc && freesize > maxsize)) { @@ -279,24 +369,60 @@ int Create(struct supertype *st, char *mddev, int mdfd, minsize = freesize; } if (runstop != 1 || verbose >= 0) { + int fd = open(dname, O_RDONLY); + if (fd <0 ) { + fprintf(stderr, Name ": Cannot open %s: %s\n", + dname, strerror(errno)); + fail=1; + continue; + } warn |= check_ext2(fd, dname); warn |= check_reiser(fd, dname); warn |= check_raid(fd, dname); + if (strcmp(st->ss->name, "1.x") == 0 && + st->minor_version >= 1) + /* metadata at front */ + warn |= check_partitions(fd, dname, 0); + else if (level == 1 || level == LEVEL_CONTAINER) + /* partitions could be meaningful */ + warn |= check_partitions(fd, dname, freesize*2); + else + /* partitions cannot be meaningful */ + warn |= check_partitions(fd, dname, 0); + if (strcmp(st->ss->name, "1.x") == 0 && + st->minor_version >= 1 && + did_default && + level == 1 && + (warn & 1024) == 0) { + warn |= 1024; + fprintf(stderr, Name ": Note: this array has metadata at the start and\n" + " may not be suitable as a boot device. If you plan to\n" + " store '/boot' on this device please ensure that\n" + " your boot-loader understands md/v1.x metadata, or use\n" + " --metadata=0.90\n"); + } + close(fd); } - close(fd); } + if (have_container) + info.array.working_disks = raiddisks; if (fail) { fprintf(stderr, Name ": create aborted\n"); return 1; } if (size == 0) { - if (mindisc == NULL) { + if (mindisc == NULL && !have_container) { fprintf(stderr, Name ": no size and no drives given - aborting create.\n"); return 1; } - if (level > 0 || level == LEVEL_MULTIPATH || level == LEVEL_FAULTY) { + if (level > 0 || level == LEVEL_MULTIPATH + || level == LEVEL_FAULTY + || st->ss->external ) { /* size is meaningful */ - if (minsize > 0x100000000ULL && st->ss->major == 0) { + if (!st->ss->validate_geometry(st, level, layout, + raiddisks, + chunk, minsize*2, + NULL, NULL, 0)) { fprintf(stderr, Name ": devices too large for RAID level %d\n", level); return 1; } @@ -305,13 +431,21 @@ int Create(struct supertype *st, char *mddev, int mdfd, fprintf(stderr, Name ": size set to %lluK\n", size); } } - if (level > 0 && ((maxsize-size)*100 > maxsize)) { + if (!have_container && level > 0 && ((maxsize-size)*100 > maxsize)) { if (runstop != 1 || verbose >= 0) - fprintf(stderr, Name ": largest drive (%s) exceed size (%lluK) by more than 1%%\n", + fprintf(stderr, Name ": largest drive (%s) exceeds size (%lluK) by more than 1%%\n", maxdisc, size); warn = 1; } + if (st->ss->detail_platform && st->ss->detail_platform(0, 1) != 0) { + if (runstop != 1 || verbose >= 0) + fprintf(stderr, Name ": %s unable to enumerate platform support\n" + " array may not be compatible with hardware/firmware\n", + st->ss->name); + warn = 1; + } + if (warn) { if (runstop!= 1) { if (!ask("Continue creating array? ")) { @@ -328,50 +462,91 @@ int Create(struct supertype *st, char *mddev, int mdfd, * as missing, so that a reconstruct happens (faster than re-parity) * FIX: Can we do this for raid6 as well? */ - if (assume_clean==0 && force == 0 && first_missing >= raiddisks) { + if (st->ss->external == 0 && + assume_clean==0 && force == 0 && first_missing >= raiddisks) { switch ( level ) { case 4: case 5: insert_point = raiddisks-1; sparedisks++; - array.active_disks--; + info.array.active_disks--; missing_disks++; break; default: break; } } + /* For raid6, if creating with 1 missing drive, make a good drive + * into a spare, else the create will fail + */ + if (assume_clean == 0 && force == 0 && first_missing < raiddisks && + st->ss->external == 0 && + second_missing >= raiddisks && level == 6) { + insert_point = raiddisks - 1; + if (insert_point == first_missing) + insert_point--; + sparedisks ++; + info.array.active_disks--; + missing_disks++; + } - if (level <= 0 && first_missing != subdevs * 2) { + if (level <= 0 && first_missing < subdevs * 2) { fprintf(stderr, Name ": This level does not support missing devices\n"); return 1; } + /* We need to create the device */ + map_lock(&map); + mdfd = create_mddev(mddev, name, autof, LOCAL, chosen_name); + if (mdfd < 0) + return 1; + mddev = chosen_name; + + vers = md_get_version(mdfd); + if (vers < 9000) { + fprintf(stderr, Name ": Create requires md driver version 0.90.0 or later\n"); + goto abort; + } else { + mdu_array_info_t inf; + memset(&inf, 0, sizeof(inf)); + ioctl(mdfd, GET_ARRAY_INFO, &inf); + if (inf.working_disks != 0) { + fprintf(stderr, Name ": another array by this name" + " is already running.\n"); + goto abort; + } + } + /* Ok, lets try some ioctls */ - array.level = level; - array.size = size; - array.raid_disks = raiddisks; + info.array.level = level; + info.array.size = size; + info.array.raid_disks = raiddisks; /* The kernel should *know* what md_minor we are dealing * with, but it chooses to trust me instead. Sigh */ - array.md_minor = 0; + info.array.md_minor = 0; if (fstat(mdfd, &stb)==0) - array.md_minor = minor(stb.st_rdev); - array.not_persistent = 0; - /*** FIX: Need to do something about RAID-6 here ***/ + info.array.md_minor = minor(stb.st_rdev); + info.array.not_persistent = 0; + if ( ( (level == 4 || level == 5) && (insert_point < raiddisks || first_missing < raiddisks) ) || - ( level == 6 && missing_disks == 2) + ( level == 6 && (insert_point < raiddisks + || second_missing < raiddisks)) + || + ( level <= 0 ) || assume_clean - ) - array.state = 1; /* clean, but one+ drive will be missing */ - else - array.state = 0; /* not clean, but no errors */ - + ) { + info.array.state = 1; /* clean, but one+ drive will be missing*/ + info.resync_start = MaxSector; + } else { + info.array.state = 0; /* not clean, but no errors */ + info.resync_start = 0; + } if (level == 10) { /* for raid10, the bitmap size is the capacity of the array, * which is array.size * raid_disks / ncopies; @@ -402,12 +577,12 @@ int Create(struct supertype *st, char *mddev, int mdfd, * So for now, we assume that all raid and spare * devices will be given. */ - array.spare_disks=sparedisks; - array.failed_disks=missing_disks; - array.nr_disks = array.working_disks + array.failed_disks; - array.layout = layout; - array.chunk_size = chunk*1024; - array.major_version = st->ss->major; + info.array.spare_disks=sparedisks; + info.array.failed_disks=missing_disks; + info.array.nr_disks = info.array.working_disks + + info.array.failed_disks; + info.array.layout = layout; + info.array.chunk_size = chunk*1024; if (name == NULL || *name == 0) { /* base name on mddev */ @@ -418,6 +593,7 @@ int Create(struct supertype *st, char *mddev, int mdfd, * /dev/md/home -> home * /dev/mdhome -> home */ + /* FIXME compare this with rules in create_mddev */ name = strrchr(mddev, '/'); if (name) { name++; @@ -433,8 +609,38 @@ int Create(struct supertype *st, char *mddev, int mdfd, name += 2; } } - if (!st->ss->init_super(st, &array, size, name, homehost, uuid)) - return 1; + if (!st->ss->init_super(st, &info.array, size, name, homehost, uuid)) + goto abort; + + total_slots = info.array.nr_disks; + sysfs_init(&info, mdfd, 0); + st->ss->getinfo_super(st, &info); + + if (did_default && verbose >= 0) { + if (is_subarray(info.text_version)) { + int dnum = devname2devnum(info.text_version+1); + char *path; + int mdp = get_mdp_major(); + struct mdinfo *mdi; + if (dnum > 0) + path = map_dev(MD_MAJOR, dnum, 1); + else + path = map_dev(mdp, (-1-dnum)<< 6, 1); + + mdi = sysfs_read(-1, dnum, GET_VERSION); + + fprintf(stderr, Name ": Creating array inside " + "%s container %s\n", + mdi?mdi->text_version:"managed", path); + sysfs_free(mdi); + } else + fprintf(stderr, Name ": Defaulting to version" + " %s metadata\n", info.text_version); + } + + map_update(&map, fd2devnum(mdfd), info.text_version, + info.uuid, chosen_name); + map_unlock(&map); if (bitmap_file && vers < 9003) { major_num = BITMAP_MAJOR_HOSTENDIAN; @@ -447,31 +653,60 @@ int Create(struct supertype *st, char *mddev, int mdfd, if (bitmap_file && strcmp(bitmap_file, "internal")==0) { if ((vers%100) < 2) { fprintf(stderr, Name ": internal bitmaps not supported by this kernel.\n"); - return 1; + goto abort; + } + if (!st->ss->add_internal_bitmap) { + fprintf(stderr, Name ": internal bitmaps not supported with %s metadata\n", + st->ss->name); + goto abort; } if (!st->ss->add_internal_bitmap(st, &bitmap_chunk, delay, write_behind, bitmapsize, 1, major_num)) { fprintf(stderr, Name ": Given bitmap chunk size not supported.\n"); - return 1; + goto abort; } bitmap_file = NULL; } + sysfs_init(&info, mdfd, 0); - if ((vers % 100) >= 1) { /* can use different versions */ - mdu_array_info_t inf; - memset(&inf, 0, sizeof(inf)); - inf.major_version = st->ss->major; - inf.minor_version = st->minor_version; - rv = ioctl(mdfd, SET_ARRAY_INFO, &inf); - } else - rv = ioctl(mdfd, SET_ARRAY_INFO, NULL); + if (st->ss->external && st->subarray[0]) { + /* member */ + + /* When creating a member, we need to be careful + * to negotiate with mdmon properly. + * If it is already running, we cannot write to + * the devices and must ask it to do that part. + * If it isn't running, we write to the devices, + * and then start it. + * We hold an exclusive open on the container + * device to make sure mdmon doesn't exit after + * we checked that it is running. + * + * For now, fail if it is already running. + */ + container_fd = open_dev_excl(st->container_dev); + if (container_fd < 0) { + fprintf(stderr, Name ": Cannot get exclusive " + "open on container - weird.\n"); + goto abort; + } + if (mdmon_running(st->container_dev)) { + if (verbose) + fprintf(stderr, Name ": reusing mdmon " + "for %s.\n", + devnum2devname(st->container_dev)); + st->update_tail = &st->updates; + } else + need_mdmon = 1; + } + rv = set_array_info(mdfd, st, &info); if (rv) { - fprintf(stderr, Name ": SET_ARRAY_INFO failed for %s: %s\n", + fprintf(stderr, Name ": failed to set array info for %s: %s\n", mddev, strerror(errno)); - return 1; + goto abort; } if (bitmap_file) { @@ -482,22 +717,22 @@ int Create(struct supertype *st, char *mddev, int mdfd, delay, write_behind, bitmapsize, major_num)) { - return 1; + goto abort; } bitmap_fd = open(bitmap_file, O_RDWR); if (bitmap_fd < 0) { fprintf(stderr, Name ": weird: %s cannot be openned\n", bitmap_file); - return 1; + goto abort; } if (ioctl(mdfd, SET_BITMAP_FILE, bitmap_fd) < 0) { fprintf(stderr, Name ": Cannot set bitmap file for %s: %s\n", mddev, strerror(errno)); - return 1; + goto abort; } } - + infos = malloc(sizeof(*infos) * total_slots); for (pass=1; pass <=2 ; pass++) { mddev_dev_t moved_disk = NULL; /* the disk that was moved out of the insert point */ @@ -506,77 +741,191 @@ int Create(struct supertype *st, char *mddev, int mdfd, dv=(dv->next)?(dv->next):moved_disk, dnum++) { int fd; struct stat stb; - mdu_disk_info_t disk; + struct mdinfo *inf = &infos[dnum]; - disk.number = dnum; + if (dnum >= total_slots) + abort(); if (dnum == insert_point) { moved_disk = dv; + continue; } - disk.raid_disk = disk.number; - if (disk.raid_disk < raiddisks) - disk.state = (1<devname, "missing")==0) + continue; + if (have_container) + moved_disk = NULL; + if (have_container && dnum < info.array.raid_disks - 1) + /* repeatedly use the container */ + moved_disk = dv; + + switch(pass) { + case 1: + *inf = info; + + inf->disk.number = dnum; + inf->disk.raid_disk = dnum; + if (inf->disk.raid_disk < raiddisks) + inf->disk.state = (1<writemostly) - disk.state |= (1<devname, "missing")==0) { - disk.major = 0; - disk.minor = 0; - disk.state = (1<devname, O_RDONLY|O_EXCL, 0); - if (fd < 0) { - fprintf(stderr, Name ": failed to open %s after earlier success - aborting\n", - dv->devname); - return 1; + else + inf->disk.state = 0; + + if (dv->writemostly == 1) + inf->disk.state |= (1<ss->external && st->subarray[0]) + fd = open(dv->devname, O_RDWR); + else + fd = open(dv->devname, O_RDWR|O_EXCL); + + if (fd < 0) { + fprintf(stderr, Name ": failed to open %s " + "after earlier success - aborting\n", + dv->devname); + goto abort; + } + fstat(fd, &stb); + inf->disk.major = major(stb.st_rdev); + inf->disk.minor = minor(stb.st_rdev); + } + if (fd >= 0) + remove_partitions(fd); + if (st->ss->add_to_super(st, &inf->disk, + fd, dv->devname)) { + ioctl(mdfd, STOP_ARRAY, NULL); + goto abort; + } + st->ss->getinfo_super(st, inf); + safe_mode_delay = inf->safe_mode_delay; + + if (have_container && verbose > 0) + fprintf(stderr, Name ": Using %s for device %d\n", + map_dev(inf->disk.major, + inf->disk.minor, + 0), dnum); + + if (!have_container) { + /* getinfo_super might have lost these ... */ + inf->disk.major = major(stb.st_rdev); + inf->disk.minor = minor(stb.st_rdev); } - fstat(fd, &stb); - disk.major = major(stb.st_rdev); - disk.minor = minor(stb.st_rdev); - remove_partitions(fd); - close(fd); - } - switch(pass){ - case 1: - st->ss->add_to_super(st, &disk); break; case 2: - if (disk.state == 1) break; - Kill(dv->devname, 0, 1); /* Just be sure it is clean */ - Kill(dv->devname, 0, 1); /* and again, there could be two superblocks */ - st->ss->write_init_super(st, &disk, - dv->devname); - - if (ioctl(mdfd, ADD_NEW_DISK, &disk)) { - fprintf(stderr, Name ": ADD_NEW_DISK for %s failed: %s\n", + inf->errors = 0; + rv = 0; + + rv = add_disk(mdfd, st, &info, inf); + + if (rv) { + fprintf(stderr, + Name ": ADD_NEW_DISK for %s " + "failed: %s\n", dv->devname, strerror(errno)); st->ss->free_super(st); - return 1; + goto abort; } - break; } - if (dv == moved_disk && dnum != insert_point) break; + if (!have_container && + dv == moved_disk && dnum != insert_point) break; + } + if (pass == 1) { + struct mdinfo info_new; + struct map_ent *me = NULL; + + /* check to see if the uuid has changed due to these + * metadata changes, and if so update the member array + * and container uuid. Note ->write_init_super clears + * the subarray cursor such that ->getinfo_super once + * again returns container info. + */ + map_lock(&map); + st->ss->getinfo_super(st, &info_new); + if (st->ss->external && level != LEVEL_CONTAINER && + !same_uuid(info_new.uuid, info.uuid, 0)) { + map_update(&map, fd2devnum(mdfd), + info_new.text_version, + info_new.uuid, chosen_name); + me = map_by_devnum(&map, st->container_dev); + } + + st->ss->write_init_super(st); + + /* update parent container uuid */ + if (me) { + char *path = strdup(me->path); + + st->ss->getinfo_super(st, &info_new); + map_update(&map, st->container_dev, + info_new.text_version, + info_new.uuid, path); + free(path); + } + map_unlock(&map); + + flush_metadata_updates(st); } } + free(infos); st->ss->free_super(st); - /* param is not actually used */ - if (runstop == 1 || subdevs >= raiddisks) { - mdu_param_t param; - if (ioctl(mdfd, RUN_ARRAY, ¶m)) { - fprintf(stderr, Name ": RUN_ARRAY failed: %s\n", - strerror(errno)); - Manage_runstop(mddev, mdfd, -1, 0); - return 1; + if (level == LEVEL_CONTAINER) { + /* No need to start. But we should signal udev to + * create links */ + sysfs_uevent(&info, "change"); + if (verbose >= 0) + fprintf(stderr, Name ": container %s prepared.\n", mddev); + wait_for(chosen_name, mdfd); + } else if (runstop == 1 || subdevs >= raiddisks) { + if (st->ss->external) { + switch(level) { + case LEVEL_LINEAR: + case LEVEL_MULTIPATH: + case 0: + sysfs_set_str(&info, NULL, "array_state", + "active"); + need_mdmon = 0; + break; + default: + sysfs_set_str(&info, NULL, "array_state", + "readonly"); + break; + } + sysfs_set_safemode(&info, safe_mode_delay); + } else { + /* param is not actually used */ + mdu_param_t param; + if (ioctl(mdfd, RUN_ARRAY, ¶m)) { + fprintf(stderr, Name ": RUN_ARRAY failed: %s\n", + strerror(errno)); + ioctl(mdfd, STOP_ARRAY, NULL); + goto abort; + } } if (verbose >= 0) fprintf(stderr, Name ": array %s started.\n", mddev); + if (st->ss->external && st->subarray[0]) { + if (need_mdmon) + start_mdmon(st->container_dev); + + ping_monitor(devnum2devname(st->container_dev)); + close(container_fd); + } + wait_for(chosen_name, mdfd); } else { fprintf(stderr, Name ": not starting array - not enough devices.\n"); } + close(mdfd); return 0; + + abort: + map_lock(&map); + map_remove(&map, fd2devnum(mdfd)); + map_unlock(&map); + + if (mdfd >= 0) + close(mdfd); + return 1; }