return sector_count(__le32_to_cpu(mpb->mpb_size));
}
+struct intel_dev {
+ struct imsm_dev *dev;
+ struct intel_dev *next;
+ int index;
+};
+
/* internal representation of IMSM metadata */
struct intel_super {
union {
int creating_imsm; /* flag to indicate container creation */
int current_vol; /* index of raid device undergoing creation */
__u32 create_offset; /* common start for 'current_vol' */
- #define IMSM_MAX_RAID_DEVS 2
- struct imsm_dev *dev_tbl[IMSM_MAX_RAID_DEVS];
+ struct intel_dev *devlist;
struct dl {
struct dl *next;
int index;
static struct imsm_dev *get_imsm_dev(struct intel_super *super, __u8 index)
{
+ struct intel_dev *dv;
+
if (index >= super->anchor->num_raid_devs)
return NULL;
- return super->dev_tbl[index];
+ for (dv = super->devlist; dv; dv = dv->next)
+ if (dv->index == index)
+ return dv->dev;
+ return NULL;
}
static __u32 get_imsm_ord_tbl_ent(struct imsm_dev *dev, int slot)
map->disk_ord_tbl[slot] = __cpu_to_le32(ord);
}
+static int get_imsm_disk_slot(struct imsm_map *map, int idx)
+{
+ int slot;
+ __u32 ord;
+
+ for (slot = 0; slot < map->num_members; slot++) {
+ ord = __le32_to_cpu(map->disk_ord_tbl[slot]);
+ if (ord_to_idx(ord) == idx)
+ return slot;
+ }
+
+ return -1;
+}
+
static int get_imsm_raid_level(struct imsm_map *map)
{
if (map->raid_level == 1) {
static int count_memberships(struct dl *dl, struct intel_super *super)
{
int memberships = 0;
- int i, j;
+ int i;
for (i = 0; i < super->anchor->num_raid_devs; i++) {
struct imsm_dev *dev = get_imsm_dev(super, i);
struct imsm_map *map = get_imsm_map(dev, 0);
- for (j = 0; j < map->num_members; j++) {
- __u32 index = get_imsm_disk_idx(dev, j);
-
- if (index == dl->index)
- memberships++;
- }
+ if (get_imsm_disk_slot(map, dl->index) >= 0)
+ memberships++;
}
return memberships;
{
/* find a list of used extents on the given physical device */
struct extent *rv, *e;
- int i, j;
+ int i;
int memberships = count_memberships(dl, super);
__u32 reservation = MPB_SECTOR_CNT + IMSM_RESERVED_SECTORS;
struct imsm_dev *dev = get_imsm_dev(super, i);
struct imsm_map *map = get_imsm_map(dev, 0);
- for (j = 0; j < map->num_members; j++) {
- __u32 index = get_imsm_disk_idx(dev, j);
-
- if (index == dl->index) {
- e->start = __le32_to_cpu(map->pba_of_lba0);
- e->size = __le32_to_cpu(map->blocks_per_member);
- e++;
- }
+ if (get_imsm_disk_slot(map, dl->index) >= 0) {
+ e->start = __le32_to_cpu(map->pba_of_lba0);
+ e->size = __le32_to_cpu(map->blocks_per_member);
+ e++;
}
}
qsort(rv, memberships, sizeof(*rv), cmp_extent);
printf(" UUID : %s\n", uuid);
printf(" RAID Level : %d\n", get_imsm_raid_level(map));
printf(" Members : %d\n", map->num_members);
- for (slot = 0; slot < map->num_members; slot++)
- if (disk_idx== get_imsm_disk_idx(dev, slot))
- break;
- if (slot < map->num_members) {
+ slot = get_imsm_disk_slot(map, disk_idx);
+ if (slot >= 0) {
ord = get_imsm_ord_tbl_ent(dev, slot);
printf(" This Slot : %d%s\n", slot,
ord & IMSM_ORD_REBUILD ? " (out-of-sync)" : "");
return err;
}
-static int detail_platform_imsm(int verbose)
+static int detail_platform_imsm(int verbose, int enumerate_only)
{
/* There are two components to imsm platform support, the ahci SATA
* controller and the option-rom. To find the SATA controller we
int host_base = 0;
int port_count = 0;
+ if (enumerate_only) {
+ if (check_env("IMSM_NO_PLATFORM") || find_imsm_orom())
+ return 0;
+ return 2;
+ }
+
list = find_driver_devices("pci", "ahci");
for (hba = list; hba; hba = hba->next)
if (devpath_to_vendor(hba->path) == 0x8086)
case 10:
return 0x102;
}
- return -1;
+ return UnSet;
}
static void getinfo_super_imsm_volume(struct supertype *st, struct mdinfo *info)
uuid_from_super_imsm(st, info->uuid);
}
+/* check the config file to see if we can return a real uuid for this spare */
+static void fixup_container_spare_uuid(struct mdinfo *inf)
+{
+ struct mddev_ident_s *array_list;
+
+ if (inf->array.level != LEVEL_CONTAINER ||
+ memcmp(inf->uuid, uuid_match_any, sizeof(int[4])) != 0)
+ return;
+
+ array_list = conf_get_ident(NULL);
+
+ for (; array_list; array_list = array_list->next) {
+ if (array_list->uuid_set) {
+ struct supertype *_sst; /* spare supertype */
+ struct supertype *_cst; /* container supertype */
+
+ _cst = array_list->st;
+ _sst = _cst->ss->match_metadata_desc(inf->text_version);
+ if (_sst) {
+ memcpy(inf->uuid, array_list->uuid, sizeof(int[4]));
+ free(_sst);
+ break;
+ }
+ }
+ }
+}
static void getinfo_super_imsm(struct supertype *st, struct mdinfo *info)
{
info->component_size = reserved;
s = disk->status;
info->disk.state = s & CONFIGURED_DISK ? (1 << MD_DISK_ACTIVE) : 0;
+ /* we don't change info->disk.raid_disk here because
+ * this state will be finalized in mdmon after we have
+ * found the 'most fresh' version of the metadata
+ */
info->disk.state |= s & FAILED_DISK ? (1 << MD_DISK_FAULTY) : 0;
info->disk.state |= s & SPARE_DISK ? 0 : (1 << MD_DISK_SYNC);
}
*/
if (info->disk.state & (1 << MD_DISK_SYNC) || super->anchor->num_raid_devs)
uuid_from_super_imsm(st, info->uuid);
- else
+ else {
memcpy(info->uuid, uuid_match_any, sizeof(int[4]));
+ fixup_container_spare_uuid(info);
+ }
}
static int update_super_imsm(struct supertype *st, struct mdinfo *info,
return devsize - (MPB_SECTOR_CNT + IMSM_RESERVED_SECTORS);
}
+static void free_devlist(struct intel_super *super)
+{
+ struct intel_dev *dv;
+
+ while (super->devlist) {
+ dv = super->devlist->next;
+ free(super->devlist->dev);
+ free(super->devlist);
+ super->devlist = dv;
+ }
+}
+
+static void imsm_copy_dev(struct imsm_dev *dest, struct imsm_dev *src)
+{
+ memcpy(dest, src, sizeof_imsm_dev(src, 0));
+}
+
static int compare_super_imsm(struct supertype *st, struct supertype *tst)
{
/*
if (first->anchor->num_raid_devs == 0 &&
sec->anchor->num_raid_devs > 0) {
int i;
+ struct intel_dev *dv;
+ struct imsm_dev *dev;
/* we need to copy raid device info from sec if an allocation
* fails here we don't associate the spare
*/
for (i = 0; i < sec->anchor->num_raid_devs; i++) {
- first->dev_tbl[i] = malloc(sizeof(struct imsm_dev));
- if (!first->dev_tbl) {
- while (--i >= 0) {
- free(first->dev_tbl[i]);
- first->dev_tbl[i] = NULL;
- }
- fprintf(stderr, "imsm: failed to associate spare\n");
- return 3;
+ dv = malloc(sizeof(*dv));
+ if (!dv)
+ break;
+ dev = malloc(sizeof_imsm_dev(get_imsm_dev(sec, i), 1));
+ if (!dev) {
+ free(dv);
+ break;
}
- *first->dev_tbl[i] = *sec->dev_tbl[i];
+ dv->dev = dev;
+ dv->index = i;
+ dv->next = first->devlist;
+ first->devlist = dv;
+ }
+ if (i <= sec->anchor->num_raid_devs) {
+ /* allocation failure */
+ free_devlist(first);
+ fprintf(stderr, "imsm: failed to associate spare\n");
+ return 3;
}
+ for (i = 0; i < sec->anchor->num_raid_devs; i++)
+ imsm_copy_dev(get_imsm_dev(first, i), get_imsm_dev(sec, i));
first->anchor->num_raid_devs = sec->anchor->num_raid_devs;
first->anchor->family_num = sec->anchor->family_num;
/* trim leading whitespace */
rsp_len = scsi_serial[3];
+ if (!rsp_len) {
+ if (devname)
+ fprintf(stderr,
+ Name ": Failed to retrieve serial for %s\n",
+ devname);
+ return 2;
+ }
rsp_buf = (char *) &scsi_serial[4];
c = rsp_buf;
while (isspace(*c))
return 0;
}
-static void imsm_copy_dev(struct imsm_dev *dest, struct imsm_dev *src)
-{
- memcpy(dest, src, sizeof_imsm_dev(src, 0));
-}
-
#ifndef MDASSEMBLE
/* When migrating map0 contains the 'destination' state while map1
* contains the current state. When not migrating map0 contains the
* 4/ Rebuild (migr_state=1 migr_type=MIGR_REBUILD map0state=normal
* map1state=degraded)
*/
-static void migrate(struct imsm_dev *dev, __u8 to_state, int rebuild_resync)
+static void migrate(struct imsm_dev *dev, __u8 to_state, int migr_type)
{
struct imsm_map *dest;
struct imsm_map *src = get_imsm_map(dev, 0);
dev->vol.migr_state = 1;
- dev->vol.migr_type = rebuild_resync;
+ dev->vol.migr_type = migr_type;
dev->vol.curr_migr_unit = 0;
dest = get_imsm_map(dev, 1);
+ /* duplicate and then set the target end state in map[0] */
memcpy(dest, src, sizeof_imsm_map(src));
+ if (migr_type == MIGR_REBUILD) {
+ __u32 ord;
+ int i;
+
+ for (i = 0; i < src->num_members; i++) {
+ ord = __le32_to_cpu(src->disk_ord_tbl[i]);
+ set_imsm_ord_tbl_ent(src, i, ord_to_idx(ord));
+ }
+ }
+
src->map_state = to_state;
}
static void end_migration(struct imsm_dev *dev, __u8 map_state)
{
struct imsm_map *map = get_imsm_map(dev, 0);
+ struct imsm_map *prev = get_imsm_map(dev, dev->vol.migr_state);
+ int i;
+
+ /* merge any IMSM_ORD_REBUILD bits that were not successfully
+ * completed in the last migration.
+ *
+ * FIXME add support for online capacity expansion and
+ * raid-level-migration
+ */
+ for (i = 0; i < prev->num_members; i++)
+ map->disk_ord_tbl[i] |= prev->disk_ord_tbl[i];
dev->vol.migr_state = 0;
dev->vol.curr_migr_unit = 0;
for (i = 0; i < super->anchor->num_raid_devs; i++) {
struct imsm_dev *dev_iter = __get_imsm_dev(super->anchor, i);
+ struct intel_dev *dv;
len = sizeof_imsm_dev(dev_iter, 0);
len_migr = sizeof_imsm_dev(dev_iter, 1);
if (len_migr > len)
space_needed += len_migr - len;
+ dv = malloc(sizeof(*dv));
+ if (!dv)
+ return 1;
dev_new = malloc(len_migr);
- if (!dev_new)
+ if (!dev_new) {
+ free(dv);
return 1;
+ }
imsm_copy_dev(dev_new, dev_iter);
- super->dev_tbl[i] = dev_new;
+ dv->dev = dev_new;
+ dv->index = i;
+ dv->next = super->devlist;
+ super->devlist = dv;
}
/* ensure that super->buf is large enough when all raid devices
sectors = mpb_sectors(anchor) - 1;
free(anchor);
if (!sectors) {
+ check_sum = __gen_imsm_checksum(super->anchor);
+ if (check_sum != __le32_to_cpu(super->anchor->check_sum)) {
+ if (devname)
+ fprintf(stderr,
+ Name ": IMSM checksum %x != %x on %s\n",
+ check_sum,
+ __le32_to_cpu(super->anchor->check_sum),
+ devname);
+ return 2;
+ }
+
rc = load_imsm_disk(fd, super, devname, 0);
if (rc == 0)
rc = parse_raid_devices(super);
Name ": IMSM checksum %x != %x on %s\n",
check_sum, __le32_to_cpu(super->anchor->check_sum),
devname);
- return 2;
+ return 3;
}
/* FIXME the BBM log is disk specific so we cannot use this global
/* free all the pieces hanging off of a super pointer */
static void __free_imsm(struct intel_super *super, int free_disks)
{
- int i;
-
if (super->buf) {
free(super->buf);
super->buf = NULL;
}
if (free_disks)
free_imsm_disks(super);
- for (i = 0; i < IMSM_MAX_RAID_DEVS; i++)
- if (super->dev_tbl[i]) {
- free(super->dev_tbl[i]);
- super->dev_tbl[i] = NULL;
- }
+ free_devlist(super);
if (super->hba) {
free((void *) super->hba);
super->hba = NULL;
super->create_offset = ~((__u32 ) 0);
if (!check_env("IMSM_NO_PLATFORM"))
super->orom = find_imsm_orom();
- if (super->orom) {
+ if (super->orom && !check_env("IMSM_TEST_OROM")) {
struct sys_dev *list, *ent;
/* find the first intel ahci controller */
dl = serial_to_dl(disk->serial, super);
if (dl)
continue;
- /* ok we have a 'disk' without a live entry in
- * super->disks
- */
- if (disk->status & FAILED_DISK || !(disk->status & USABLE_DISK))
- continue; /* never mind, already marked */
dl = malloc(sizeof(*dl));
if (!dl)
dl->index = i;
serialcpy(dl->serial, disk->serial);
dl->disk = *disk;
+ dl->e = NULL;
dl->next = super->missing;
super->missing = dl;
}
char nm[20];
int dfd;
int rv;
+ int devnum = fd2devnum(fd);
+ int retry;
+ enum sysfs_read_flags flags;
- /* check if this disk is a member of an active array */
- sra = sysfs_read(fd, 0, GET_LEVEL|GET_VERSION|GET_DEVS|GET_STATE);
+ flags = GET_LEVEL|GET_VERSION|GET_DEVS|GET_STATE;
+ if (mdmon_running(devnum))
+ flags |= SKIP_GONE_DEVS;
+
+ /* check if 'fd' an opened container */
+ sra = sysfs_read(fd, 0, flags);
if (!sra)
return 1;
for (sd = sra->devs; sd; sd = sd->next) {
sprintf(nm, "%d:%d", sd->disk.major, sd->disk.minor);
dfd = dev_open(nm, keep_fd ? O_RDWR : O_RDONLY);
- if (!dfd) {
+ if (dfd < 0) {
free_imsm(super);
return 2;
}
rv = load_imsm_mpb(dfd, super, NULL);
+
+ /* retry the load if we might have raced against mdmon */
+ if (rv == 3 && mdmon_running(devnum))
+ for (retry = 0; retry < 3; retry++) {
+ usleep(3000);
+ rv = load_imsm_mpb(dfd, super, NULL);
+ if (rv != 3)
+ break;
+ }
if (!keep_fd)
close(dfd);
if (rv == 0) {
}
} else {
free_imsm(super);
- return 2;
+ return rv;
}
}
/* load the most up to date anchor */
sprintf(nm, "%d:%d", best->disk.major, best->disk.minor);
dfd = dev_open(nm, O_RDONLY);
- if (!dfd) {
+ if (dfd < 0) {
free_imsm(super);
return 1;
}
for (sd = sra->devs ; sd ; sd = sd->next) {
sprintf(nm, "%d:%d", sd->disk.major, sd->disk.minor);
dfd = dev_open(nm, keep_fd? O_RDWR : O_RDONLY);
- if (!dfd) {
+ if (dfd < 0) {
free_imsm(super);
return 2;
}
}
*sbp = super;
- st->container_dev = fd2devnum(fd);
+ st->container_dev = devnum;
if (st->ss == NULL) {
st->ss = &super_imsm;
st->minor_version = 0;
static __u32 info_to_blocks_per_member(mdu_array_info_t *info)
{
- return (info->size * 2) & ~(info_to_blocks_per_strip(info) - 1);
+ if (info->level == 1)
+ return info->size * 2;
+ else
+ return (info->size * 2) & ~(info_to_blocks_per_strip(info) - 1);
}
static void imsm_update_version_info(struct intel_super *super)
*/
struct intel_super *super = st->sb;
struct imsm_super *mpb = super->anchor;
+ struct intel_dev *dv;
struct imsm_dev *dev;
struct imsm_vol *vol;
struct imsm_map *map;
*/
if (super->current_vol == 0)
mpb->num_disks = 0;
+
+ for (i = 0; i < super->current_vol; i++) {
+ dev = get_imsm_dev(super, i);
+ if (strncmp((char *) dev->volume, name,
+ MAX_RAID_SERIAL_LEN) == 0) {
+ fprintf(stderr, Name": '%s' is already defined for this container\n",
+ name);
+ return 0;
+ }
+ }
+
sprintf(st->subarray, "%d", idx);
+ dv = malloc(sizeof(*dv));
+ if (!dv) {
+ fprintf(stderr, Name ": failed to allocate device list entry\n");
+ return 0;
+ }
dev = malloc(sizeof(*dev) + sizeof(__u32) * (info->raid_disks - 1));
if (!dev) {
+ free(dv);
fprintf(stderr, Name": could not allocate raid device\n");
return 0;
}
map->blocks_per_member = __cpu_to_le32(info_to_blocks_per_member(info));
map->blocks_per_strip = __cpu_to_le16(info_to_blocks_per_strip(info));
map->num_data_stripes = __cpu_to_le32(info_to_num_data_stripes(info));
+ map->failed_disk_num = ~0;
map->map_state = info->level ? IMSM_T_STATE_UNINITIALIZED :
IMSM_T_STATE_NORMAL;
set_imsm_ord_tbl_ent(map, i, 0);
}
mpb->num_raid_devs++;
- super->dev_tbl[super->current_vol] = dev;
+
+ dv->dev = dev;
+ dv->index = super->current_vol;
+ dv->next = super->devlist;
+ super->devlist = dv;
imsm_update_version_info(super);
dd->index = -1;
dd->devname = devname ? strdup(devname) : NULL;
dd->fd = fd;
+ dd->e = NULL;
rv = imsm_read_serial(fd, devname, dd->serial);
if (rv) {
fprintf(stderr,
for (i = 0; i < mpb->num_raid_devs; i++) {
struct imsm_dev *dev = __get_imsm_dev(mpb, i);
- imsm_copy_dev(dev, super->dev_tbl[i]);
+ imsm_copy_dev(dev, get_imsm_dev(super, i));
mpb_size += sizeof_imsm_dev(dev, 0);
}
mpb_size += __le32_to_cpu(mpb->bbm_log_size);
case 1:
if (raiddisks > 2)
return imsm_orom_has_raid1e(orom);
- else
- return imsm_orom_has_raid1(orom);
- case 10: return imsm_orom_has_raid10(orom);
- case 5: return imsm_orom_has_raid5(orom);
+ return imsm_orom_has_raid1(orom) && raiddisks == 2;
+ case 10: return imsm_orom_has_raid10(orom) && raiddisks == 4;
+ case 5: return imsm_orom_has_raid5(orom) && raiddisks > 2;
}
else
return 1; /* not on an Intel RAID platform so anything goes */
return 0;
}
-#define vprintf(fmt, arg...) (void) (verbose && fprintf(stderr, Name fmt, ##arg))
+#define pr_vrb(fmt, arg...) (void) (verbose && fprintf(stderr, Name fmt, ##arg))
/* validate_geometry_imsm_volume - lifted from validate_geometry_ddf_bvd
* FIX ME add ahci details
*/
{
struct stat stb;
struct intel_super *super = st->sb;
+ struct imsm_super *mpb = super->anchor;
struct dl *dl;
unsigned long long pos = 0;
unsigned long long maxsize;
return 0;
if (!is_raid_level_supported(super->orom, level, raiddisks)) {
- vprintf(": platform does not support raid level: %d\n", level);
+ pr_vrb(": platform does not support raid%d with %d disk%s\n",
+ level, raiddisks, raiddisks > 1 ? "s" : "");
return 0;
}
- if (super->orom && !imsm_orom_has_chunk(super->orom, chunk)) {
- vprintf(": platform does not support a chunk size of: %d\n", chunk);
+ if (super->orom && level != 1 &&
+ !imsm_orom_has_chunk(super->orom, chunk)) {
+ pr_vrb(": platform does not support a chunk size of: %d\n", chunk);
return 0;
}
if (layout != imsm_level_to_layout(level)) {
if (level == 5)
- vprintf(": imsm raid 5 only supports the left-asymmetric layout\n");
+ pr_vrb(": imsm raid 5 only supports the left-asymmetric layout\n");
else if (level == 10)
- vprintf(": imsm raid 10 only supports the n2 layout\n");
+ pr_vrb(": imsm raid 10 only supports the n2 layout\n");
else
- vprintf(": imsm unknown layout %#x for this raid level %d\n",
+ pr_vrb(": imsm unknown layout %#x for this raid level %d\n",
layout, level);
return 0;
}
* 'raiddisks' device extents of size 'size' at a given
* offset
*/
- unsigned long long minsize = size*2 /* convert to blocks */;
+ unsigned long long minsize = size;
unsigned long long start_offset = ~0ULL;
int dcnt = 0;
if (minsize == 0)
fprintf(stderr, Name ": %s is not in the "
"same imsm set\n", dev);
return 0;
+ } else if (super->orom && dl->index < 0 && mpb->num_raid_devs) {
+ /* If a volume is present then the current creation attempt
+ * cannot incorporate new spares because the orom may not
+ * understand this configuration (all member disks must be
+ * members of each array in the container).
+ */
+ fprintf(stderr, Name ": %s is a spare and a volume"
+ " is already defined for this container\n", dev);
+ fprintf(stderr, Name ": The option-rom requires all member"
+ " disks to be a member of all volumes\n");
+ return 0;
}
/* retrieve the largest free space block */
verbose);
}
+ if (!dev) {
+ if (st->sb && freesize) {
+ /* Should do auto-layout here */
+ fprintf(stderr, Name ": IMSM does not support auto-layout yet\n");
+ return 0;
+ }
+ return 1;
+ }
if (st->sb) {
/* creating in a given container */
return validate_geometry_imsm_volume(st, level, layout,
int failed = 0;
struct imsm_disk *disk;
struct imsm_map *map = get_imsm_map(dev, 0);
+ struct imsm_map *prev = get_imsm_map(dev, dev->vol.migr_state);
+ __u32 ord;
+ int idx;
- for (i = 0; i < map->num_members; i++) {
- __u32 ord = get_imsm_ord_tbl_ent(dev, i);
- int idx = ord_to_idx(ord);
+ /* at the beginning of migration we set IMSM_ORD_REBUILD on
+ * disks that are being rebuilt. New failures are recorded to
+ * map[0]. So we look through all the disks we started with and
+ * see if any failures are still present, or if any new ones
+ * have arrived
+ *
+ * FIXME add support for online capacity expansion and
+ * raid-level-migration
+ */
+ for (i = 0; i < prev->num_members; i++) {
+ ord = __le32_to_cpu(prev->disk_ord_tbl[i]);
+ ord |= __le32_to_cpu(map->disk_ord_tbl[i]);
+ idx = ord_to_idx(ord);
disk = get_imsm_disk(super, idx);
if (!disk || disk->status & FAILED_DISK ||
return 0;
}
-static void mark_failure(struct imsm_disk *disk)
+/* return true if we recorded new information */
+static int mark_failure(struct imsm_dev *dev, struct imsm_disk *disk, int idx)
{
- if (disk->status & FAILED_DISK)
- return;
+ __u32 ord;
+ int slot;
+ struct imsm_map *map;
+
+ /* new failures are always set in map[0] */
+ map = get_imsm_map(dev, 0);
+
+ slot = get_imsm_disk_slot(map, idx);
+ if (slot < 0)
+ return 0;
+
+ ord = __le32_to_cpu(map->disk_ord_tbl[slot]);
+ if ((disk->status & FAILED_DISK) && (ord & IMSM_ORD_REBUILD))
+ return 0;
+
disk->status |= FAILED_DISK;
+ set_imsm_ord_tbl_ent(map, slot, idx | IMSM_ORD_REBUILD);
+ if (map->failed_disk_num == ~0)
+ map->failed_disk_num = slot;
+ return 1;
+}
+
+static void mark_missing(struct imsm_dev *dev, struct imsm_disk *disk, int idx)
+{
+ mark_failure(dev, disk, idx);
+
+ if (disk->scsi_id == __cpu_to_le32(~(__u32)0))
+ return;
+
disk->scsi_id = __cpu_to_le32(~(__u32)0);
memmove(&disk->serial[0], &disk->serial[1], MAX_RAID_SERIAL_LEN - 1);
}
dprintf("imsm: mark missing\n");
end_migration(dev, map_state);
for (dl = super->missing; dl; dl = dl->next)
- mark_failure(&dl->disk);
+ mark_missing(dev, &dl->disk, dl->index);
super->updates_pending++;
}
if (is_resync_complete(a)) {
/* complete intialization / resync,
- * recovery is completed in ->set_disk
+ * recovery and interrupted recovery is completed in
+ * ->set_disk
*/
if (is_resyncing(dev)) {
dprintf("imsm: mark resync done\n");
disk = get_imsm_disk(super, ord_to_idx(ord));
/* check for new failures */
- if ((state & DS_FAULTY) && !(disk->status & FAILED_DISK)) {
- mark_failure(disk);
- super->updates_pending++;
+ if (state & DS_FAULTY) {
+ if (mark_failure(dev, disk, ord_to_idx(ord)))
+ super->updates_pending++;
}
/* check if in_sync */
- if (state & DS_INSYNC && ord & IMSM_ORD_REBUILD) {
+ if (state & DS_INSYNC && ord & IMSM_ORD_REBUILD && is_rebuilding(dev)) {
struct imsm_map *migr_map = get_imsm_map(dev, 1);
set_imsm_ord_tbl_ent(migr_map, n, ord_to_idx(ord));
/* check if recovery complete, newly degraded, or failed */
if (map_state == IMSM_T_STATE_NORMAL && is_rebuilding(dev)) {
end_migration(dev, map_state);
+ map = get_imsm_map(dev, 0);
+ map->failed_disk_num = ~0;
super->updates_pending++;
} else if (map_state == IMSM_T_STATE_DEGRADED &&
map->map_state != map_state &&
return dl;
}
-static struct dl *imsm_add_spare(struct intel_super *super, int slot, struct active_array *a)
+static struct dl *imsm_add_spare(struct intel_super *super, int slot,
+ struct active_array *a, int activate_new)
{
struct imsm_dev *dev = get_imsm_dev(super, a->info.container_member);
int idx = get_imsm_disk_idx(dev, slot);
- struct imsm_map *map = get_imsm_map(dev, 0);
+ struct imsm_super *mpb = super->anchor;
+ struct imsm_map *map;
unsigned long long esize;
unsigned long long pos;
struct mdinfo *d;
struct extent *ex;
- int j;
+ int i, j;
int found;
__u32 array_start;
+ __u32 blocks;
struct dl *dl;
for (dl = super->disks; dl; dl = dl->next) {
continue;
/* skip in use or failed drives */
- if (dl->disk.status & FAILED_DISK || idx == dl->index) {
- dprintf("%x:%x status ( %s%s)\n",
- dl->major, dl->minor,
- dl->disk.status & FAILED_DISK ? "failed " : "",
- idx == dl->index ? "in use " : "");
+ if (dl->disk.status & FAILED_DISK || idx == dl->index ||
+ dl->index == -2) {
+ dprintf("%x:%x status (failed: %d index: %d)\n",
+ dl->major, dl->minor,
+ (dl->disk.status & FAILED_DISK) == FAILED_DISK, idx);
continue;
}
+ /* skip pure spares when we are looking for partially
+ * assimilated drives
+ */
+ if (dl->index == -1 && !activate_new)
+ continue;
+
/* Does this unused device have the requisite free space?
- * We need a->info.component_size sectors
+ * It needs to be able to cover all member volumes
*/
ex = get_extents(super, dl);
if (!ex) {
dprintf("cannot get extents\n");
continue;
}
- found = 0;
- j = 0;
- pos = 0;
- array_start = __le32_to_cpu(map->pba_of_lba0);
+ for (i = 0; i < mpb->num_raid_devs; i++) {
+ dev = get_imsm_dev(super, i);
+ map = get_imsm_map(dev, 0);
- do {
- /* check that we can start at pba_of_lba0 with
- * a->info.component_size of space
+ /* check if this disk is already a member of
+ * this array
*/
- esize = ex[j].start - pos;
- if (array_start >= pos &&
- array_start + a->info.component_size < ex[j].start) {
- found = 1;
+ if (get_imsm_disk_slot(map, dl->index) >= 0)
+ continue;
+
+ found = 0;
+ j = 0;
+ pos = 0;
+ array_start = __le32_to_cpu(map->pba_of_lba0);
+ blocks = __le32_to_cpu(map->blocks_per_member);
+
+ do {
+ /* check that we can start at pba_of_lba0 with
+ * blocks_per_member of space
+ */
+ esize = ex[j].start - pos;
+ if (array_start >= pos &&
+ array_start + blocks < ex[j].start) {
+ found = 1;
+ break;
+ }
+ pos = ex[j].start + ex[j].size;
+ j++;
+ } while (ex[j-1].size);
+
+ if (!found)
break;
- }
- pos = ex[j].start + ex[j].size;
- j++;
-
- } while (ex[j-1].size);
+ }
free(ex);
- if (!found) {
- dprintf("%x:%x does not have %llu at %d\n",
+ if (i < mpb->num_raid_devs) {
+ dprintf("%x:%x does not have %u at %u\n",
dl->major, dl->minor,
- a->info.component_size,
- __le32_to_cpu(map->pba_of_lba0));
+ blocks, array_start);
/* No room */
continue;
- } else
- break;
+ }
+ return dl;
}
return dl;
continue;
/*
- * OK, this device needs recovery. Try to re-add the previous
- * occupant of this slot, if this fails add a new spare
+ * OK, this device needs recovery. Try to re-add the
+ * previous occupant of this slot, if this fails see if
+ * we can continue the assimilation of a spare that was
+ * partially assimilated, finally try to activate a new
+ * spare.
*/
dl = imsm_readd(super, i, a);
if (!dl)
- dl = imsm_add_spare(super, i, a);
+ dl = imsm_add_spare(super, i, a, 0);
+ if (!dl)
+ dl = imsm_add_spare(super, i, a, 1);
if (!dl)
continue;
found = 0;
for (a = st->arrays; a ; a = a->next) {
dev = get_imsm_dev(super, a->info.container_member);
- for (i = 0; i < map->num_members; i++)
- if (victim == get_imsm_disk_idx(dev, i))
- found++;
+ map = get_imsm_map(dev, 0);
+
+ if (get_imsm_disk_slot(map, victim) >= 0)
+ found++;
}
/* delete the victim if it is no longer being
* (FIX ME) notice that its update did not take hold.
*/
struct imsm_update_create_array *u = (void *) update->buf;
+ struct intel_dev *dv;
struct imsm_dev *dev;
struct imsm_map *map, *new_map;
unsigned long long start, end;
if (u->dev_idx < mpb->num_raid_devs) {
dprintf("%s: subarray %d already defined\n",
__func__, u->dev_idx);
- return;
+ goto create_error;
}
/* check update is next in sequence */
if (u->dev_idx != mpb->num_raid_devs) {
dprintf("%s: can not create array %d expected index %d\n",
__func__, u->dev_idx, mpb->num_raid_devs);
- return;
+ goto create_error;
}
new_map = get_imsm_map(&u->dev, 0);
if (disks_overlap(super, i, u)) {
dprintf("%s: arrays overlap\n", __func__);
- return;
+ goto create_error;
}
}
/* check that prepare update was successful */
if (!update->space) {
dprintf("%s: prepare update failed\n", __func__);
- return;
+ goto create_error;
}
/* check that all disks are still active before committing
dl = serial_to_dl(inf[i].serial, super);
if (!dl) {
dprintf("%s: disk disappeared\n", __func__);
- return;
+ goto create_error;
}
}
set_imsm_ord_tbl_ent(new_map, i, dl->index);
}
- dev = update->space;
+ dv = update->space;
+ dev = dv->dev;
update->space = NULL;
imsm_copy_dev(dev, &u->dev);
- super->dev_tbl[u->dev_idx] = dev;
+ dv->index = u->dev_idx;
+ dv->next = super->devlist;
+ super->devlist = dv;
mpb->num_raid_devs++;
imsm_update_version_info(super);
break;
+ create_error:
+ /* mdmon knows how to release update->space, but not
+ * ((struct intel_dev *) update->space)->dev
+ */
+ if (update->space) {
+ dv = update->space;
+ free(dv->dev);
+ }
+ break;
}
case update_add_disk:
switch (type) {
case update_create_array: {
struct imsm_update_create_array *u = (void *) update->buf;
+ struct intel_dev *dv;
struct imsm_dev *dev = &u->dev;
struct imsm_map *map = get_imsm_map(dev, 0);
struct dl *dl;
inf = get_disk_info(u);
len = sizeof_imsm_dev(dev, 1);
- /* allocate a new super->dev_tbl entry */
- update->space = malloc(len);
+ /* allocate a new super->devlist entry */
+ dv = malloc(sizeof(*dv));
+ if (dv) {
+ dv->dev = malloc(len);
+ if (dv->dev)
+ update->space = dv;
+ else {
+ free(dv);
+ update->space = NULL;
+ }
+ }
/* count how many spares will be converted to members */
for (i = 0; i < map->num_members; i++) {
free(super->next_buf);
super->next_len = buf_len;
- if (posix_memalign(&super->next_buf, buf_len, 512) != 0)
+ if (posix_memalign(&super->next_buf, 512, buf_len) != 0)
super->next_buf = NULL;
}
}
.free_super = free_super_imsm,
.match_metadata_desc = match_metadata_desc_imsm,
.container_content = container_content_imsm,
+ .default_layout = imsm_level_to_layout,
.external = 1,
.name = "imsm",
projects / thirdparty / mdadm.git / blobdiff