struct intel_dev {
struct imsm_dev *dev;
struct intel_dev *next;
- int index;
+ unsigned index;
};
/* internal representation of IMSM metadata */
void *next_buf; /* for realloc'ing buf from the manager */
size_t next_len;
int updates_pending; /* count of pending updates for mdmon */
- 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' */
__u32 random; /* random data for seeding new family numbers */
enum imsm_update_type {
update_activate_spare,
update_create_array,
+ update_kill_array,
+ update_rename_array,
update_add_disk,
};
struct imsm_dev dev;
};
+struct imsm_update_kill_array {
+ enum imsm_update_type type;
+ int dev_idx;
+};
+
+struct imsm_update_rename_array {
+ enum imsm_update_type type;
+ __u8 name[MAX_RAID_SERIAL_LEN];
+ int dev_idx;
+};
+
struct imsm_update_add_disk {
enum imsm_update_type type;
};
return &mpb->disk[index];
}
-#ifndef MDASSEMBLE
/* retrieve a disk from the parsed metadata */
static struct imsm_disk *get_imsm_disk(struct intel_super *super, __u8 index)
{
return NULL;
}
-#endif
/* generate a checksum directly from the anchor when the anchor is known to be
* up-to-date, currently only at load or write_super after coalescing
map->disk_ord_tbl[slot] = __cpu_to_le32(ord);
}
-static int get_imsm_disk_slot(struct imsm_map *map, int idx)
+static int get_imsm_disk_slot(struct imsm_map *map, unsigned idx)
{
int slot;
__u32 ord;
*/
remainder &= ~1UL;
/* make sure remainder is still sane */
- if (remainder < ROUND_UP(super->len, 512) >> 9)
+ if (remainder < (unsigned)ROUND_UP(super->len, 512) >> 9)
remainder = ROUND_UP(super->len, 512) >> 9;
if (reservation > remainder)
reservation = remainder;
}
#ifndef MDASSEMBLE
+static __u64 blocks_per_migr_unit(struct imsm_dev *dev);
+
static void print_imsm_dev(struct imsm_dev *dev, char *uuid, int disk_idx)
{
__u64 sz;
- int slot;
+ int slot, i;
struct imsm_map *map = get_imsm_map(dev, 0);
__u32 ord;
printf(" UUID : %s\n", uuid);
printf(" RAID Level : %d\n", get_imsm_raid_level(map));
printf(" Members : %d\n", map->num_members);
+ printf(" Slots : [");
+ for (i = 0; i < map->num_members; i++) {
+ ord = get_imsm_ord_tbl_ent(dev, i);
+ printf("%s", ord & IMSM_ORD_REBUILD ? "_" : "U");
+ }
+ printf("]\n");
slot = get_imsm_disk_slot(map, disk_idx);
if (slot >= 0) {
ord = get_imsm_ord_tbl_ent(dev, slot);
printf(" Chunk Size : %u KiB\n",
__le16_to_cpu(map->blocks_per_strip) / 2);
printf(" Reserved : %d\n", __le32_to_cpu(dev->reserved_blocks));
- printf(" Migrate State : %s", dev->vol.migr_state ? "migrating" : "idle\n");
+ printf(" Migrate State : ");
if (dev->vol.migr_state) {
if (migr_type(dev) == MIGR_INIT)
- printf(": initializing\n");
+ printf("initialize\n");
else if (migr_type(dev) == MIGR_REBUILD)
- printf(": rebuilding\n");
+ printf("rebuild\n");
else if (migr_type(dev) == MIGR_VERIFY)
- printf(": check\n");
+ printf("check\n");
else if (migr_type(dev) == MIGR_GEN_MIGR)
- printf(": general migration\n");
+ printf("general migration\n");
else if (migr_type(dev) == MIGR_STATE_CHANGE)
- printf(": state change\n");
+ printf("state change\n");
else if (migr_type(dev) == MIGR_REPAIR)
- printf(": repair\n");
+ printf("repair\n");
else
- printf(": <unknown:%d>\n", migr_type(dev));
- }
+ printf("<unknown:%d>\n", migr_type(dev));
+ } else
+ printf("idle\n");
printf(" Map State : %s", map_state_str[map->map_state]);
if (dev->vol.migr_state) {
struct imsm_map *map = get_imsm_map(dev, 1);
+
printf(" <-- %s", map_state_str[map->map_state]);
+ printf("\n Checkpoint : %u (%llu)",
+ __le32_to_cpu(dev->vol.curr_migr_unit),
+ (unsigned long long)blocks_per_migr_unit(dev));
}
printf("\n");
printf(" Dirty State : %s\n", dev->vol.dirty ? "dirty" : "clean");
int err = 0;
unsigned long port_mask = (1 << port_count) - 1;
- if (port_count > sizeof(port_mask) * 8) {
+ if (port_count > (int)sizeof(port_mask) * 8) {
if (verbose)
fprintf(stderr, Name ": port_count %d out of range\n", port_count);
return 2;
}
#endif
+static __u32 migr_strip_blocks_resync(struct imsm_dev *dev)
+{
+ /* migr_strip_size when repairing or initializing parity */
+ struct imsm_map *map = get_imsm_map(dev, 0);
+ __u32 chunk = __le32_to_cpu(map->blocks_per_strip);
+
+ switch (get_imsm_raid_level(map)) {
+ case 5:
+ case 10:
+ return chunk;
+ default:
+ return 128*1024 >> 9;
+ }
+}
+
+static __u32 migr_strip_blocks_rebuild(struct imsm_dev *dev)
+{
+ /* migr_strip_size when rebuilding a degraded disk, no idea why
+ * this is different than migr_strip_size_resync(), but it's good
+ * to be compatible
+ */
+ struct imsm_map *map = get_imsm_map(dev, 1);
+ __u32 chunk = __le32_to_cpu(map->blocks_per_strip);
+
+ switch (get_imsm_raid_level(map)) {
+ case 1:
+ case 10:
+ if (map->num_members % map->num_domains == 0)
+ return 128*1024 >> 9;
+ else
+ return chunk;
+ case 5:
+ return max((__u32) 64*1024 >> 9, chunk);
+ default:
+ return 128*1024 >> 9;
+ }
+}
+
+static __u32 num_stripes_per_unit_resync(struct imsm_dev *dev)
+{
+ struct imsm_map *lo = get_imsm_map(dev, 0);
+ struct imsm_map *hi = get_imsm_map(dev, 1);
+ __u32 lo_chunk = __le32_to_cpu(lo->blocks_per_strip);
+ __u32 hi_chunk = __le32_to_cpu(hi->blocks_per_strip);
+
+ return max((__u32) 1, hi_chunk / lo_chunk);
+}
+
+static __u32 num_stripes_per_unit_rebuild(struct imsm_dev *dev)
+{
+ struct imsm_map *lo = get_imsm_map(dev, 0);
+ int level = get_imsm_raid_level(lo);
+
+ if (level == 1 || level == 10) {
+ struct imsm_map *hi = get_imsm_map(dev, 1);
+
+ return hi->num_domains;
+ } else
+ return num_stripes_per_unit_resync(dev);
+}
+
+static __u8 imsm_num_data_members(struct imsm_dev *dev)
+{
+ /* named 'imsm_' because raid0, raid1 and raid10
+ * counter-intuitively have the same number of data disks
+ */
+ struct imsm_map *map = get_imsm_map(dev, 0);
+
+ switch (get_imsm_raid_level(map)) {
+ case 0:
+ case 1:
+ case 10:
+ return map->num_members;
+ case 5:
+ return map->num_members - 1;
+ default:
+ dprintf("%s: unsupported raid level\n", __func__);
+ return 0;
+ }
+}
+
+static __u32 parity_segment_depth(struct imsm_dev *dev)
+{
+ struct imsm_map *map = get_imsm_map(dev, 0);
+ __u32 chunk = __le32_to_cpu(map->blocks_per_strip);
+
+ switch(get_imsm_raid_level(map)) {
+ case 1:
+ case 10:
+ return chunk * map->num_domains;
+ case 5:
+ return chunk * map->num_members;
+ default:
+ return chunk;
+ }
+}
+
+static __u32 map_migr_block(struct imsm_dev *dev, __u32 block)
+{
+ struct imsm_map *map = get_imsm_map(dev, 1);
+ __u32 chunk = __le32_to_cpu(map->blocks_per_strip);
+ __u32 strip = block / chunk;
+
+ switch (get_imsm_raid_level(map)) {
+ case 1:
+ case 10: {
+ __u32 vol_strip = (strip * map->num_domains) + 1;
+ __u32 vol_stripe = vol_strip / map->num_members;
+
+ return vol_stripe * chunk + block % chunk;
+ } case 5: {
+ __u32 stripe = strip / (map->num_members - 1);
+
+ return stripe * chunk + block % chunk;
+ }
+ default:
+ return 0;
+ }
+}
+
+static __u64 blocks_per_migr_unit(struct imsm_dev *dev)
+{
+ /* calculate the conversion factor between per member 'blocks'
+ * (md/{resync,rebuild}_start) and imsm migration units, return
+ * 0 for the 'not migrating' and 'unsupported migration' cases
+ */
+ if (!dev->vol.migr_state)
+ return 0;
+
+ switch (migr_type(dev)) {
+ case MIGR_VERIFY:
+ case MIGR_REPAIR:
+ case MIGR_INIT: {
+ struct imsm_map *map = get_imsm_map(dev, 0);
+ __u32 stripes_per_unit;
+ __u32 blocks_per_unit;
+ __u32 parity_depth;
+ __u32 migr_chunk;
+ __u32 block_map;
+ __u32 block_rel;
+ __u32 segment;
+ __u32 stripe;
+ __u8 disks;
+
+ /* yes, this is really the translation of migr_units to
+ * per-member blocks in the 'resync' case
+ */
+ stripes_per_unit = num_stripes_per_unit_resync(dev);
+ migr_chunk = migr_strip_blocks_resync(dev);
+ disks = imsm_num_data_members(dev);
+ blocks_per_unit = stripes_per_unit * migr_chunk * disks;
+ stripe = __le32_to_cpu(map->blocks_per_strip) * disks;
+ segment = blocks_per_unit / stripe;
+ block_rel = blocks_per_unit - segment * stripe;
+ parity_depth = parity_segment_depth(dev);
+ block_map = map_migr_block(dev, block_rel);
+ return block_map + parity_depth * segment;
+ }
+ case MIGR_REBUILD: {
+ __u32 stripes_per_unit;
+ __u32 migr_chunk;
+
+ stripes_per_unit = num_stripes_per_unit_rebuild(dev);
+ migr_chunk = migr_strip_blocks_rebuild(dev);
+ return migr_chunk * stripes_per_unit;
+ }
+ case MIGR_GEN_MIGR:
+ case MIGR_STATE_CHANGE:
+ default:
+ return 0;
+ }
+}
+
static int imsm_level_to_layout(int level)
{
switch (level) {
struct imsm_dev *dev = get_imsm_dev(super, super->current_vol);
struct imsm_map *map = get_imsm_map(dev, 0);
struct dl *dl;
+ char *devname;
for (dl = super->disks; dl; dl = dl->next)
if (dl->raiddisk == info->disk.raid_disk)
info->data_offset = __le32_to_cpu(map->pba_of_lba0);
info->component_size = __le32_to_cpu(map->blocks_per_member);
memset(info->uuid, 0, sizeof(info->uuid));
+ info->recovery_start = MaxSector;
+ info->reshape_active = 0;
- if (map->map_state == IMSM_T_STATE_UNINITIALIZED || dev->vol.dirty)
- info->resync_start = 0;
- else if (dev->vol.migr_state)
- /* FIXME add curr_migr_unit to resync_start conversion */
+ if (map->map_state == IMSM_T_STATE_UNINITIALIZED || dev->vol.dirty) {
info->resync_start = 0;
- else
- info->resync_start = ~0ULL;
+ } else if (dev->vol.migr_state) {
+ switch (migr_type(dev)) {
+ case MIGR_REPAIR:
+ case MIGR_INIT: {
+ __u64 blocks_per_unit = blocks_per_migr_unit(dev);
+ __u64 units = __le32_to_cpu(dev->vol.curr_migr_unit);
+
+ info->resync_start = blocks_per_unit * units;
+ break;
+ }
+ case MIGR_VERIFY:
+ /* we could emulate the checkpointing of
+ * 'sync_action=check' migrations, but for now
+ * we just immediately complete them
+ */
+ case MIGR_REBUILD:
+ /* this is handled by container_content_imsm() */
+ case MIGR_GEN_MIGR:
+ case MIGR_STATE_CHANGE:
+ /* FIXME handle other migrations */
+ default:
+ /* we are not dirty, so... */
+ info->resync_start = MaxSector;
+ }
+ } else
+ info->resync_start = MaxSector;
strncpy(info->name, (char *) dev->volume, MAX_RAID_SERIAL_LEN);
info->name[MAX_RAID_SERIAL_LEN] = 0;
info->array.major_version = -1;
info->array.minor_version = -2;
- sprintf(info->text_version, "/%s/%d",
- devnum2devname(st->container_dev),
- info->container_member);
+ devname = devnum2devname(st->container_dev);
+ *info->text_version = '\0';
+ if (devname)
+ sprintf(info->text_version, "/%s/%d", devname, info->container_member);
+ free(devname);
info->safe_mode_delay = 4000; /* 4 secs like the Matrix driver */
uuid_from_super_imsm(st, info->uuid);
}
}
}
+
+static __u8 imsm_check_degraded(struct intel_super *super, struct imsm_dev *dev, int failed);
+static int imsm_count_failed(struct intel_super *super, struct imsm_dev *dev);
+
+static struct imsm_disk *get_imsm_missing(struct intel_super *super, __u8 index)
+{
+ struct dl *d;
+
+ for (d = super->missing; d; d = d->next)
+ if (d->index == index)
+ return &d->disk;
+ return NULL;
+}
+
static void getinfo_super_imsm(struct supertype *st, struct mdinfo *info)
{
struct intel_super *super = st->sb;
info->disk.number = -1;
info->disk.state = 0;
info->name[0] = 0;
+ info->recovery_start = MaxSector;
+
+ /* do we have the all the insync disks that we expect? */
+ if (st->loaded_container) {
+ struct imsm_super *mpb = super->anchor;
+ int max_enough = -1, i;
+
+ for (i = 0; i < mpb->num_raid_devs; i++) {
+ struct imsm_dev *dev = get_imsm_dev(super, i);
+ int failed, enough, j, missing = 0;
+ struct imsm_map *map;
+ __u8 state;
+
+ failed = imsm_count_failed(super, dev);
+ state = imsm_check_degraded(super, dev, failed);
+ map = get_imsm_map(dev, dev->vol.migr_state);
+
+ /* any newly missing disks?
+ * (catches single-degraded vs double-degraded)
+ */
+ for (j = 0; j < map->num_members; j++) {
+ __u32 ord = get_imsm_ord_tbl_ent(dev, i);
+ __u32 idx = ord_to_idx(ord);
+
+ if (!(ord & IMSM_ORD_REBUILD) &&
+ get_imsm_missing(super, idx)) {
+ missing = 1;
+ break;
+ }
+ }
+
+ if (state == IMSM_T_STATE_FAILED)
+ enough = -1;
+ else if (state == IMSM_T_STATE_DEGRADED &&
+ (state != map->map_state || missing))
+ enough = 0;
+ else /* we're normal, or already degraded */
+ enough = 1;
+
+ /* in the missing/failed disk case check to see
+ * if at least one array is runnable
+ */
+ max_enough = max(max_enough, enough);
+ }
+ dprintf("%s: enough: %d\n", __func__, max_enough);
+ info->container_enough = max_enough;
+ } else
+ info->container_enough = -1;
if (super->disks) {
__u32 reserved = imsm_reserved_sectors(super, super->disks);
first->anchor->num_raid_devs = sec->anchor->num_raid_devs;
first->anchor->orig_family_num = sec->anchor->orig_family_num;
first->anchor->family_num = sec->anchor->family_num;
+ memcpy(first->anchor->sig, sec->anchor->sig, MAX_SIGNATURE_LENGTH);
for (i = 0; i < sec->anchor->num_raid_devs; i++)
imsm_copy_dev(get_imsm_dev(first, i), get_imsm_dev(sec, i));
}
{
struct stat st;
char path[256];
- char dname[100];
+ char dname[PATH_MAX];
char *nm;
int rv;
__u32 check_sum;
get_dev_size(fd, NULL, &dsize);
+ if (dsize < 1024) {
+ if (devname)
+ fprintf(stderr,
+ Name ": %s: device to small for imsm\n",
+ devname);
+ return 1;
+ }
if (lseek64(fd, dsize - (512 * 2), SEEK_SET) < 0) {
if (devname)
return 1;
}
- if (read(fd, super->buf + 512, super->len - 512) != super->len - 512) {
+ if ((unsigned)read(fd, super->buf + 512, super->len - 512) != super->len - 512) {
if (devname)
fprintf(stderr,
Name ": Cannot read extended mpb on %s: %s\n",
st->sb = NULL;
}
-static struct intel_super *alloc_super(int creating_imsm)
+static struct intel_super *alloc_super(void)
{
struct intel_super *super = malloc(sizeof(*super));
if (super) {
memset(super, 0, sizeof(*super));
- super->creating_imsm = creating_imsm;
super->current_vol = -1;
super->create_offset = ~((__u32 ) 0);
if (!check_env("IMSM_NO_PLATFORM"))
struct intel_disk *idisk;
idisk = disk_list_get(dl->serial, disk_list);
- if (is_spare(&idisk->disk) &&
+ if (idisk && is_spare(&idisk->disk) &&
!is_failed(&idisk->disk) && !is_configured(&idisk->disk))
dl->index = -1;
else {
int retry;
int err = 0;
int i;
- enum sysfs_read_flags flags;
-
- 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);
+ sra = sysfs_read(fd, 0, GET_LEVEL|GET_VERSION|GET_DEVS|GET_STATE);
if (!sra)
return 1;
if (sra->array.major_version != -1 ||
sra->array.minor_version != -2 ||
- strcmp(sra->text_version, "imsm") != 0)
- return 1;
-
+ strcmp(sra->text_version, "imsm") != 0) {
+ err = 1;
+ goto error;
+ }
/* load all mpbs */
for (sd = sra->devs, i = 0; sd; sd = sd->next, i++) {
- struct intel_super *s = alloc_super(0);
+ struct intel_super *s = alloc_super();
char nm[32];
int dfd;
}
if (st->subarray[0]) {
- if (atoi(st->subarray) <= super->anchor->num_raid_devs)
- super->current_vol = atoi(st->subarray);
+ unsigned long val;
+ char *ep;
+
+ err = 1;
+ val = strtoul(st->subarray, &ep, 10);
+ if (*ep != '\0') {
+ free_imsm(super);
+ goto error;
+ }
+
+ if (val < super->anchor->num_raid_devs)
+ super->current_vol = val;
else {
free_imsm(super);
- err = 1;
goto error;
}
}
super_list = super_list->next;
free_imsm(s);
}
+ sysfs_free(sra);
if (err)
return err;
return 0;
#endif
+ if (test_partition(fd))
+ /* IMSM not allowed on partitions */
+ return 1;
+
free_super_imsm(st);
- super = alloc_super(0);
+ super = alloc_super();
if (!super) {
fprintf(stderr,
Name ": malloc of %zu failed.\n",
}
if (st->subarray[0]) {
- if (atoi(st->subarray) <= super->anchor->num_raid_devs)
- super->current_vol = atoi(st->subarray);
+ unsigned long val;
+ char *ep;
+
+ val = strtoul(st->subarray, &ep, 10);
+ if (*ep != '\0') {
+ free_imsm(super);
+ return 1;
+ }
+
+ if (val < super->anchor->num_raid_devs)
+ super->current_vol = val;
else {
free_imsm(super);
return 1;
}
}
+static int check_name(struct intel_super *super, char *name, int quiet)
+{
+ struct imsm_super *mpb = super->anchor;
+ char *reason = NULL;
+ int i;
+
+ if (strlen(name) > MAX_RAID_SERIAL_LEN)
+ reason = "must be 16 characters or less";
+
+ for (i = 0; i < mpb->num_raid_devs; i++) {
+ struct imsm_dev *dev = get_imsm_dev(super, i);
+
+ if (strncmp((char *) dev->volume, name, MAX_RAID_SERIAL_LEN) == 0) {
+ reason = "already exists";
+ break;
+ }
+ }
+
+ if (reason && !quiet)
+ fprintf(stderr, Name ": imsm volume name %s\n", reason);
+
+ return !reason;
+}
+
static int init_super_imsm_volume(struct supertype *st, mdu_array_info_t *info,
unsigned long long size, char *name,
char *homehost, int *uuid)
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;
- }
- }
-
+ if (!check_name(super, name, 0))
+ return 0;
sprintf(st->subarray, "%d", idx);
dv = malloc(sizeof(*dv));
if (!dv) {
map->ddf = 1;
if (info->level == 1 && info->raid_disks > 2) {
+ free(dev);
+ free(dv);
fprintf(stderr, Name": imsm does not support more than 2 disks"
"in a raid1 volume\n");
return 0;
map->num_members = info->raid_disks;
for (i = 0; i < map->num_members; i++) {
/* initialized in add_to_super */
- set_imsm_ord_tbl_ent(map, i, 0);
+ set_imsm_ord_tbl_ent(map, i, IMSM_ORD_REBUILD);
}
mpb->num_raid_devs++;
else
mpb_size = 512;
- super = alloc_super(1);
+ super = alloc_super();
if (super && posix_memalign(&super->buf, 512, mpb_size) != 0) {
free(super);
super = NULL;
struct dl *dl;
struct imsm_dev *dev;
struct imsm_map *map;
+ int slot;
dev = get_imsm_dev(super, super->current_vol);
map = get_imsm_map(dev, 0);
dl->index = super->anchor->num_disks;
super->anchor->num_disks++;
}
+ /* Check the device has not already been added */
+ slot = get_imsm_disk_slot(map, dl->index);
+ if (slot >= 0 &&
+ (get_imsm_ord_tbl_ent(dev, slot) & IMSM_ORD_REBUILD) == 0) {
+ fprintf(stderr, Name ": %s has been included in this array twice\n",
+ devname);
+ return 1;
+ }
set_imsm_ord_tbl_ent(map, dk->number, dl->index);
dl->disk.status = CONFIGURED_DISK;
}
return rv;
- } else
+ } else {
+ struct dl *d;
+ for (d = super->disks; d; d = d->next)
+ Kill(d->devname, NULL, 0, 1, 1);
return write_super_imsm(st->sb, 1);
+ }
}
#endif
unsigned long reserve;
if (!e)
- return ~0ULL; /* error */
+ return 0;
/* coalesce and sort all extents. also, check to see if we need to
* reserve space between member arrays
} while (e[i-1].size);
free(e);
+ if (maxsize == 0)
+ return 0;
+
+ /* FIXME assumes volume at offset 0 is the first volume in a
+ * container
+ */
if (start_extent > 0)
reserve = IMSM_RESERVED_SECTORS; /* gap between raid regions */
else
reserve = 0;
if (maxsize < reserve)
- return ~0ULL;
+ return 0;
super->create_offset = ~((__u32) 0);
if (start + reserve > super->create_offset)
- return ~0ULL; /* start overflows create_offset */
+ return 0; /* start overflows create_offset */
super->create_offset = start + reserve;
return maxsize - reserve;
}
#define pr_vrb(fmt, arg...) (void) (verbose && fprintf(stderr, Name fmt, ##arg))
+static int
+validate_geometry_imsm_orom(struct intel_super *super, int level, int layout,
+ int raiddisks, int chunk, int verbose)
+{
+ if (!is_raid_level_supported(super->orom, level, raiddisks)) {
+ pr_vrb(": platform does not support raid%d with %d disk%s\n",
+ level, raiddisks, raiddisks > 1 ? "s" : "");
+ return 0;
+ }
+ 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)
+ pr_vrb(": imsm raid 5 only supports the left-asymmetric layout\n");
+ else if (level == 10)
+ pr_vrb(": imsm raid 10 only supports the n2 layout\n");
+ else
+ pr_vrb(": imsm unknown layout %#x for this raid level %d\n",
+ layout, level);
+ return 0;
+ }
+
+ return 1;
+}
+
/* 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;
+ struct imsm_super *mpb = super->anchor;
struct dl *dl;
unsigned long long pos = 0;
unsigned long long maxsize;
/* We must have the container info already read in. */
if (!super)
return 0;
- mpb = super->anchor;
- if (!is_raid_level_supported(super->orom, level, raiddisks)) {
- pr_vrb(": platform does not support raid%d with %d disk%s\n",
- level, raiddisks, raiddisks > 1 ? "s" : "");
- return 0;
- }
- 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)
- pr_vrb(": imsm raid 5 only supports the left-asymmetric layout\n");
- else if (level == 10)
- pr_vrb(": imsm raid 10 only supports the n2 layout\n");
- else
- pr_vrb(": imsm unknown layout %#x for this raid level %d\n",
- layout, level);
+ if (!validate_geometry_imsm_orom(super, level, layout, raiddisks, chunk, verbose))
return 0;
- }
if (!dev) {
/* General test: make sure there is space for
* offset
*/
unsigned long long minsize = size;
- unsigned long long start_offset = ~0ULL;
+ unsigned long long start_offset = MaxSector;
int dcnt = 0;
if (minsize == 0)
minsize = MPB_SECTOR_CNT + IMSM_RESERVED_SECTORS;
esize = e[i].start - pos;
if (esize >= minsize)
found = 1;
- if (found && start_offset == ~0ULL) {
+ if (found && start_offset == MaxSector) {
start_offset = pos;
break;
} else if (found && pos != start_offset) {
if ((S_IFMT & stb.st_mode) != S_IFBLK)
return 0;
for (dl = super->disks ; dl ; dl = dl->next) {
- if (dl->major == major(stb.st_rdev) &&
- dl->minor == minor(stb.st_rdev))
+ if (dl->major == (int)major(stb.st_rdev) &&
+ dl->minor == (int)minor(stb.st_rdev))
break;
}
if (!dl) {
i += dl->extent_cnt;
maxsize = merge_extents(super, i);
- if (maxsize < size) {
+ if (maxsize < size || maxsize == 0) {
if (verbose)
fprintf(stderr, Name ": not enough space after merge (%llu < %llu)\n",
maxsize, size);
return 0;
- } else if (maxsize == ~0ULL) {
- if (verbose)
- fprintf(stderr, Name ": failed to merge %d extents\n", i);
- return 0;
}
*freesize = maxsize;
if (cnt < raiddisks ||
(super->orom && used && used != raiddisks) ||
- maxsize < minsize) {
+ maxsize < minsize ||
+ maxsize == 0) {
fprintf(stderr, Name ": not enough devices with space to create array.\n");
return 0; /* No enough free spaces large enough */
}
* created. add_to_super and getinfo_super
* detect when autolayout is in progress.
*/
+ if (!validate_geometry_imsm_orom(st->sb, level, layout,
+ raiddisks, chunk,
+ verbose))
+ return 0;
return reserve_space(st, raiddisks, size, chunk, freesize);
}
return 1;
dev, freesize, verbose);
}
- /* limit creation to the following levels */
- if (!dev)
- switch (level) {
- case 0:
- case 1:
- case 10:
- case 5:
- return 0;
- default:
- if (verbose)
- fprintf(stderr, Name
- ": IMSM only supports levels 0,1,5,10\n");
- return 1;
- }
-
/* This device needs to be a device in an 'imsm' container */
fd = open(dev, O_RDONLY|O_EXCL, 0);
if (fd >= 0) {
close(cfd);
return 0;
}
+
+static int default_chunk_imsm(struct supertype *st)
+{
+ struct intel_super *super = st->sb;
+
+ if (!super->orom)
+ return 0;
+
+ return imsm_orom_default_chunk(super->orom);
+}
+
+static void handle_missing(struct intel_super *super, struct imsm_dev *dev);
+
+static int kill_subarray_imsm(struct supertype *st)
+{
+ /* remove the subarray currently referenced by ->current_vol */
+ __u8 i;
+ struct intel_dev **dp;
+ struct intel_super *super = st->sb;
+ __u8 current_vol = super->current_vol;
+ struct imsm_super *mpb = super->anchor;
+
+ if (super->current_vol < 0)
+ return 2;
+ super->current_vol = -1; /* invalidate subarray cursor */
+
+ /* block deletions that would change the uuid of active subarrays
+ *
+ * FIXME when immutable ids are available, but note that we'll
+ * also need to fixup the invalidated/active subarray indexes in
+ * mdstat
+ */
+ for (i = 0; i < mpb->num_raid_devs; i++) {
+ char subarray[4];
+
+ if (i < current_vol)
+ continue;
+ sprintf(subarray, "%u", i);
+ if (is_subarray_active(subarray, st->devname)) {
+ fprintf(stderr,
+ Name ": deleting subarray-%d would change the UUID of active subarray-%d, aborting\n",
+ current_vol, i);
+
+ return 2;
+ }
+ }
+
+ if (st->update_tail) {
+ struct imsm_update_kill_array *u = malloc(sizeof(*u));
+
+ if (!u)
+ return 2;
+ u->type = update_kill_array;
+ u->dev_idx = current_vol;
+ append_metadata_update(st, u, sizeof(*u));
+
+ return 0;
+ }
+
+ for (dp = &super->devlist; *dp;)
+ if ((*dp)->index == current_vol) {
+ *dp = (*dp)->next;
+ } else {
+ handle_missing(super, (*dp)->dev);
+ if ((*dp)->index > current_vol)
+ (*dp)->index--;
+ dp = &(*dp)->next;
+ }
+
+ /* no more raid devices, all active components are now spares,
+ * but of course failed are still failed
+ */
+ if (--mpb->num_raid_devs == 0) {
+ struct dl *d;
+
+ for (d = super->disks; d; d = d->next)
+ if (d->index > -2) {
+ d->index = -1;
+ d->disk.status = SPARE_DISK;
+ }
+ }
+
+ super->updates_pending++;
+
+ return 0;
+}
+
+static int update_subarray_imsm(struct supertype *st, char *update, mddev_ident_t ident)
+{
+ /* update the subarray currently referenced by ->current_vol */
+ struct intel_super *super = st->sb;
+ struct imsm_super *mpb = super->anchor;
+
+ if (super->current_vol < 0)
+ return 2;
+
+ if (strcmp(update, "name") == 0) {
+ char *name = ident->name;
+
+ if (is_subarray_active(st->subarray, st->devname)) {
+ fprintf(stderr,
+ Name ": Unable to update name of active subarray\n");
+ return 2;
+ }
+
+ if (!check_name(super, name, 0))
+ return 2;
+
+ if (st->update_tail) {
+ struct imsm_update_rename_array *u = malloc(sizeof(*u));
+
+ if (!u)
+ return 2;
+ u->type = update_rename_array;
+ u->dev_idx = super->current_vol;
+ snprintf((char *) u->name, MAX_RAID_SERIAL_LEN, "%s", name);
+ append_metadata_update(st, u, sizeof(*u));
+ } else {
+ struct imsm_dev *dev;
+ int i;
+
+ dev = get_imsm_dev(super, super->current_vol);
+ snprintf((char *) dev->volume, MAX_RAID_SERIAL_LEN, "%s", name);
+ for (i = 0; i < mpb->num_raid_devs; i++) {
+ dev = get_imsm_dev(super, i);
+ handle_missing(super, dev);
+ }
+ super->updates_pending++;
+ }
+ } else
+ return 2;
+
+ return 0;
+}
#endif /* MDASSEMBLE */
+static int is_rebuilding(struct imsm_dev *dev)
+{
+ struct imsm_map *migr_map;
+
+ if (!dev->vol.migr_state)
+ return 0;
+
+ if (migr_type(dev) != MIGR_REBUILD)
+ return 0;
+
+ migr_map = get_imsm_map(dev, 1);
+
+ if (migr_map->map_state == IMSM_T_STATE_DEGRADED)
+ return 1;
+ else
+ return 0;
+}
+
+static void update_recovery_start(struct imsm_dev *dev, struct mdinfo *array)
+{
+ struct mdinfo *rebuild = NULL;
+ struct mdinfo *d;
+ __u32 units;
+
+ if (!is_rebuilding(dev))
+ return;
+
+ /* Find the rebuild target, but punt on the dual rebuild case */
+ for (d = array->devs; d; d = d->next)
+ if (d->recovery_start == 0) {
+ if (rebuild)
+ return;
+ rebuild = d;
+ }
+
+ if (!rebuild) {
+ /* (?) none of the disks are marked with
+ * IMSM_ORD_REBUILD, so assume they are missing and the
+ * disk_ord_tbl was not correctly updated
+ */
+ dprintf("%s: failed to locate out-of-sync disk\n", __func__);
+ return;
+ }
+
+ units = __le32_to_cpu(dev->vol.curr_migr_unit);
+ rebuild->recovery_start = units * blocks_per_migr_unit(dev);
+}
+
+
static struct mdinfo *container_content_imsm(struct supertype *st)
{
/* Given a container loaded by load_super_imsm_all,
this = malloc(sizeof(*this));
if (!this) {
- fprintf(stderr, Name ": failed to allocate %lu bytes\n",
+ fprintf(stderr, Name ": failed to allocate %zu bytes\n",
sizeof(*this));
break;
}
super->current_vol = i;
getinfo_super_imsm_volume(st, this);
for (slot = 0 ; slot < map->num_members; slot++) {
+ unsigned long long recovery_start;
struct mdinfo *info_d;
struct dl *d;
int idx;
if (d->index == idx)
break;
+ recovery_start = MaxSector;
if (d == NULL)
skip = 1;
if (d && is_failed(&d->disk))
skip = 1;
if (ord & IMSM_ORD_REBUILD)
- skip = 1;
+ recovery_start = 0;
/*
* if we skip some disks the array will be assmebled degraded;
- * reset resync start to avoid a dirty-degraded situation
+ * reset resync start to avoid a dirty-degraded
+ * situation when performing the intial sync
*
* FIXME handle dirty degraded
*/
- if (skip && !dev->vol.dirty)
- this->resync_start = ~0ULL;
+ if ((skip || recovery_start == 0) && !dev->vol.dirty)
+ this->resync_start = MaxSector;
if (skip)
continue;
- info_d = malloc(sizeof(*info_d));
+ info_d = calloc(1, sizeof(*info_d));
if (!info_d) {
fprintf(stderr, Name ": failed to allocate disk"
" for volume %.16s\n", dev->volume);
+ info_d = this->devs;
+ while (info_d) {
+ struct mdinfo *d = info_d->next;
+
+ free(info_d);
+ info_d = d;
+ }
free(this);
this = rest;
break;
}
- memset(info_d, 0, sizeof(*info_d));
info_d->next = this->devs;
this->devs = info_d;
info_d->disk.major = d->major;
info_d->disk.minor = d->minor;
info_d->disk.raid_disk = slot;
+ info_d->recovery_start = recovery_start;
- this->array.working_disks++;
+ if (info_d->recovery_start == MaxSector)
+ this->array.working_disks++;
info_d->events = __le32_to_cpu(mpb->generation_num);
info_d->data_offset = __le32_to_cpu(map->pba_of_lba0);
info_d->component_size = __le32_to_cpu(map->blocks_per_member);
}
+ /* now that the disk list is up-to-date fixup recovery_start */
+ update_recovery_start(dev, this);
rest = this;
}
}
-#ifndef MDASSEMBLE
-static int imsm_open_new(struct supertype *c, struct active_array *a,
- char *inst)
-{
- struct intel_super *super = c->sb;
- struct imsm_super *mpb = super->anchor;
-
- if (atoi(inst) >= mpb->num_raid_devs) {
- fprintf(stderr, "%s: subarry index %d, out of range\n",
- __func__, atoi(inst));
- return -ENODEV;
- }
-
- dprintf("imsm: open_new %s\n", inst);
- a->info.container_member = atoi(inst);
- return 0;
-}
-
static __u8 imsm_check_degraded(struct intel_super *super, struct imsm_dev *dev, int failed)
{
struct imsm_map *map = get_imsm_map(dev, 0);
return failed;
}
+#ifndef MDASSEMBLE
+static int imsm_open_new(struct supertype *c, struct active_array *a,
+ char *inst)
+{
+ struct intel_super *super = c->sb;
+ struct imsm_super *mpb = super->anchor;
+
+ if (atoi(inst) >= mpb->num_raid_devs) {
+ fprintf(stderr, "%s: subarry index %d, out of range\n",
+ __func__, atoi(inst));
+ return -ENODEV;
+ }
+
+ dprintf("imsm: open_new %s\n", inst);
+ a->info.container_member = atoi(inst);
+ return 0;
+}
+
static int is_resyncing(struct imsm_dev *dev)
{
struct imsm_map *migr_map;
return 0;
}
-static int is_rebuilding(struct imsm_dev *dev)
-{
- struct imsm_map *migr_map;
-
- if (!dev->vol.migr_state)
- return 0;
-
- if (migr_type(dev) != MIGR_REBUILD)
- return 0;
-
- migr_map = get_imsm_map(dev, 1);
-
- if (migr_map->map_state == IMSM_T_STATE_DEGRADED)
- return 1;
- else
- return 0;
-}
-
/* return true if we recorded new information */
static int mark_failure(struct imsm_dev *dev, struct imsm_disk *disk, int idx)
{
disk->status |= FAILED_DISK;
disk->status &= ~CONFIGURED_DISK;
set_imsm_ord_tbl_ent(map, slot, idx | IMSM_ORD_REBUILD);
- if (~map->failed_disk_num == 0)
+ if (map->failed_disk_num == 0xff)
map->failed_disk_num = slot;
return 1;
}
memmove(&disk->serial[0], &disk->serial[1], MAX_RAID_SERIAL_LEN - 1);
}
+static void handle_missing(struct intel_super *super, struct imsm_dev *dev)
+{
+ __u8 map_state;
+ struct dl *dl;
+ int failed;
+
+ if (!super->missing)
+ return;
+ failed = imsm_count_failed(super, dev);
+ map_state = imsm_check_degraded(super, dev, failed);
+
+ dprintf("imsm: mark missing\n");
+ end_migration(dev, map_state);
+ for (dl = super->missing; dl; dl = dl->next)
+ mark_missing(dev, &dl->disk, dl->index);
+ super->updates_pending++;
+}
+
/* Handle dirty -> clean transititions and resync. Degraded and rebuild
* states are handled in imsm_set_disk() with one exception, when a
* resync is stopped due to a new failure this routine will set the
struct imsm_map *map = get_imsm_map(dev, 0);
int failed = imsm_count_failed(super, dev);
__u8 map_state = imsm_check_degraded(super, dev, failed);
+ __u32 blocks_per_unit;
/* before we activate this array handle any missing disks */
- if (consistent == 2 && super->missing) {
- struct dl *dl;
+ if (consistent == 2)
+ handle_missing(super, dev);
- dprintf("imsm: mark missing\n");
- end_migration(dev, map_state);
- for (dl = super->missing; dl; dl = dl->next)
- mark_missing(dev, &dl->disk, dl->index);
- super->updates_pending++;
- }
-
if (consistent == 2 &&
- (!is_resync_complete(a) ||
+ (!is_resync_complete(&a->info) ||
map_state != IMSM_T_STATE_NORMAL ||
dev->vol.migr_state))
consistent = 0;
- if (is_resync_complete(a)) {
+ if (is_resync_complete(&a->info)) {
/* complete intialization / resync,
* recovery and interrupted recovery is completed in
* ->set_disk
dprintf("imsm: mark resync done\n");
end_migration(dev, map_state);
super->updates_pending++;
+ a->last_checkpoint = 0;
}
} else if (!is_resyncing(dev) && !failed) {
/* mark the start of the init process if nothing is failed */
- dprintf("imsm: mark resync start (%llu)\n", a->resync_start);
+ dprintf("imsm: mark resync start\n");
if (map->map_state == IMSM_T_STATE_UNINITIALIZED)
migrate(dev, IMSM_T_STATE_NORMAL, MIGR_INIT);
else
super->updates_pending++;
}
- /* FIXME check if we can update curr_migr_unit from resync_start */
+ /* check if we can update curr_migr_unit from resync_start, recovery_start */
+ blocks_per_unit = blocks_per_migr_unit(dev);
+ if (blocks_per_unit) {
+ __u32 units32;
+ __u64 units;
+
+ units = a->last_checkpoint / blocks_per_unit;
+ units32 = units;
+
+ /* check that we did not overflow 32-bits, and that
+ * curr_migr_unit needs updating
+ */
+ if (units32 == units &&
+ __le32_to_cpu(dev->vol.curr_migr_unit) != units32) {
+ dprintf("imsm: mark checkpoint (%u)\n", units32);
+ dev->vol.curr_migr_unit = __cpu_to_le32(units32);
+ super->updates_pending++;
+ }
+ }
/* mark dirty / clean */
if (dev->vol.dirty != !consistent) {
- dprintf("imsm: mark '%s' (%llu)\n",
- consistent ? "clean" : "dirty", a->resync_start);
+ dprintf("imsm: mark '%s'\n", consistent ? "clean" : "dirty");
if (consistent)
dev->vol.dirty = 0;
else
map = get_imsm_map(dev, 0);
map->failed_disk_num = ~0;
super->updates_pending++;
+ a->last_checkpoint = 0;
} else if (map_state == IMSM_T_STATE_DEGRADED &&
map->map_state != map_state &&
!dev->vol.migr_state) {
dprintf("imsm: mark degraded\n");
map->map_state = map_state;
super->updates_pending++;
+ a->last_checkpoint = 0;
} else if (map_state == IMSM_T_STATE_FAILED &&
map->map_state != map_state) {
dprintf("imsm: mark failed\n");
end_migration(dev, map_state);
super->updates_pending++;
+ a->last_checkpoint = 0;
}
}
if (lseek64(fd, dsize - (512 * (2 + sectors)), SEEK_SET) < 0)
return 1;
- if (write(fd, buf + 512, 512 * sectors) != 512 * sectors)
+ if ((unsigned long long)write(fd, buf + 512, 512 * sectors)
+ != 512 * sectors)
return 1;
}
struct extent *ex;
int i, j;
int found;
- __u32 array_start;
- __u32 array_end;
+ __u32 array_start = 0;
+ __u32 array_end = 0;
struct dl *dl;
for (dl = super->disks; dl; dl = dl->next) {
di->disk.major = dl->major;
di->disk.minor = dl->minor;
di->disk.state = 0;
+ di->recovery_start = 0;
di->data_offset = __le32_to_cpu(map->pba_of_lba0);
di->component_size = a->info.component_size;
di->container_member = inst;
return 0;
}
-static void imsm_delete(struct intel_super *super, struct dl **dlp, int index);
+static void imsm_delete(struct intel_super *super, struct dl **dlp, unsigned index);
static void imsm_process_update(struct supertype *st,
struct metadata_update *update)
}
break;
}
+ case update_kill_array: {
+ struct imsm_update_kill_array *u = (void *) update->buf;
+ int victim = u->dev_idx;
+ struct active_array *a;
+ struct intel_dev **dp;
+ struct imsm_dev *dev;
+
+ /* sanity check that we are not affecting the uuid of
+ * active arrays, or deleting an active array
+ *
+ * FIXME when immutable ids are available, but note that
+ * we'll also need to fixup the invalidated/active
+ * subarray indexes in mdstat
+ */
+ for (a = st->arrays; a; a = a->next)
+ if (a->info.container_member >= victim)
+ break;
+ /* by definition if mdmon is running at least one array
+ * is active in the container, so checking
+ * mpb->num_raid_devs is just extra paranoia
+ */
+ dev = get_imsm_dev(super, victim);
+ if (a || !dev || mpb->num_raid_devs == 1) {
+ dprintf("failed to delete subarray-%d\n", victim);
+ break;
+ }
+
+ for (dp = &super->devlist; *dp;)
+ if ((*dp)->index == (unsigned)super->current_vol) {
+ *dp = (*dp)->next;
+ } else {
+ if ((*dp)->index > (unsigned)victim)
+ (*dp)->index--;
+ dp = &(*dp)->next;
+ }
+ mpb->num_raid_devs--;
+ super->updates_pending++;
+ break;
+ }
+ case update_rename_array: {
+ struct imsm_update_rename_array *u = (void *) update->buf;
+ char name[MAX_RAID_SERIAL_LEN+1];
+ int target = u->dev_idx;
+ struct active_array *a;
+ struct imsm_dev *dev;
+
+ /* sanity check that we are not affecting the uuid of
+ * an active array
+ */
+ snprintf(name, MAX_RAID_SERIAL_LEN, "%s", (char *) u->name);
+ name[MAX_RAID_SERIAL_LEN] = '\0';
+ for (a = st->arrays; a; a = a->next)
+ if (a->info.container_member == target)
+ break;
+ dev = get_imsm_dev(super, u->dev_idx);
+ if (a || !dev || !check_name(super, name, 1)) {
+ dprintf("failed to rename subarray-%d\n", target);
+ break;
+ }
+
+ snprintf((char *) dev->volume, MAX_RAID_SERIAL_LEN, "%s", name);
+ super->updates_pending++;
+ break;
+ }
case update_add_disk:
/* we may be able to repair some arrays if disks are
}
/* must be called while manager is quiesced */
-static void imsm_delete(struct intel_super *super, struct dl **dlp, int index)
+static void imsm_delete(struct intel_super *super, struct dl **dlp, unsigned index)
{
struct imsm_super *mpb = super->anchor;
struct dl *iter;
/* shift all indexes down one */
for (iter = super->disks; iter; iter = iter->next)
- if (iter->index > index)
+ if (iter->index > (int)index)
iter->index--;
for (iter = super->missing; iter; iter = iter->next)
- if (iter->index > index)
+ if (iter->index > (int)index)
iter->index--;
for (i = 0; i < mpb->num_raid_devs; i++) {
.brief_detail_super = brief_detail_super_imsm,
.write_init_super = write_init_super_imsm,
.validate_geometry = validate_geometry_imsm,
+ .default_chunk = default_chunk_imsm,
.add_to_super = add_to_super_imsm,
.detail_platform = detail_platform_imsm,
+ .kill_subarray = kill_subarray_imsm,
+ .update_subarray = update_subarray_imsm,
#endif
.match_home = match_home_imsm,
.uuid_from_super= uuid_from_super_imsm,
#ifndef MDASSEMBLE
/* for mdmon */
.open_new = imsm_open_new,
- .load_super = load_super_imsm,
.set_array_state= imsm_set_array_state,
.set_disk = imsm_set_disk,
.sync_metadata = imsm_sync_metadata,
projects / thirdparty / mdadm.git / blobdiff