/*
* mdadm - Intel(R) Matrix Storage Manager Support
*
- * Copyright (C) 2002-2007 Intel Corporation
+ * Copyright (C) 2002-2008 Intel Corporation
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
#define MPB_SIG_LEN (strlen(MPB_SIGNATURE))
#define MPB_VERSION_RAID0 "1.0.00"
#define MPB_VERSION_RAID1 "1.1.00"
+#define MPB_VERSION_MANY_VOLUMES_PER_ARRAY "1.2.00"
+#define MPB_VERSION_3OR4_DISK_ARRAY "1.2.01"
#define MPB_VERSION_RAID5 "1.2.02"
+#define MPB_VERSION_5OR6_DISK_ARRAY "1.2.04"
+#define MPB_VERSION_CNG "1.2.06"
+#define MPB_VERSION_ATTRIBS "1.3.00"
#define MAX_SIGNATURE_LENGTH 32
#define MAX_RAID_SERIAL_LEN 16
+
+#define MPB_ATTRIB_CHECKSUM_VERIFY __cpu_to_le32(0x80000000)
+#define MPB_ATTRIB_PM __cpu_to_le32(0x40000000)
+#define MPB_ATTRIB_2TB __cpu_to_le32(0x20000000)
+#define MPB_ATTRIB_RAID0 __cpu_to_le32(0x00000001)
+#define MPB_ATTRIB_RAID1 __cpu_to_le32(0x00000002)
+#define MPB_ATTRIB_RAID10 __cpu_to_le32(0x00000004)
+#define MPB_ATTRIB_RAID1E __cpu_to_le32(0x00000008)
+#define MPB_ATTRIB_RAID5 __cpu_to_le32(0x00000010)
+#define MPB_ATTRIB_RAIDCNG __cpu_to_le32(0x00000020)
+
#define MPB_SECTOR_CNT 418
#define IMSM_RESERVED_SECTORS 4096
__u8 serial[MAX_RAID_SERIAL_LEN];/* 0xD8 - 0xE7 ascii serial number */
__u32 total_blocks; /* 0xE8 - 0xEB total blocks */
__u32 scsi_id; /* 0xEC - 0xEF scsi ID */
+#define SPARE_DISK __cpu_to_le32(0x01) /* Spare */
+#define CONFIGURED_DISK __cpu_to_le32(0x02) /* Member of some RaidDev */
+#define FAILED_DISK __cpu_to_le32(0x04) /* Permanent failure */
+#define USABLE_DISK __cpu_to_le32(0x08) /* Fully usable unless FAILED_DISK is set */
__u32 status; /* 0xF0 - 0xF3 */
-#define SPARE_DISK 0x01 /* Spare */
-#define CONFIGURED_DISK 0x02 /* Member of some RaidDev */
-#define FAILED_DISK 0x04 /* Permanent failure */
-#define USABLE_DISK 0x08 /* Fully usable unless FAILED_DISK is set */
-
-#define IMSM_DISK_FILLERS 5
+ __u32 owner_cfg_num; /* which config 0,1,2... owns this disk */
+#define IMSM_DISK_FILLERS 4
__u32 filler[IMSM_DISK_FILLERS]; /* 0xF4 - 0x107 MPB_DISK_FILLERS for future expansion */
};
__u8 map_state; /* Normal, Uninitialized, Degraded, Failed */
#define IMSM_T_STATE_NORMAL 0
#define IMSM_T_STATE_UNINITIALIZED 1
-#define IMSM_T_STATE_DEGRADED 2 /* FIXME: is this correct? */
-#define IMSM_T_STATE_FAILED 3 /* FIXME: is this correct? */
+#define IMSM_T_STATE_DEGRADED 2
+#define IMSM_T_STATE_FAILED 3
__u8 raid_level;
#define IMSM_T_RAID0 0
#define IMSM_T_RAID1 1
#define IMSM_T_RAID5 5 /* since metadata version 1.2.02 ? */
__u8 num_members; /* number of member disks */
- __u8 reserved[3];
+ __u8 num_domains; /* number of parity domains */
+ __u8 failed_disk_num; /* valid only when state is degraded */
+ __u8 reserved[1];
__u32 filler[7]; /* expansion area */
#define IMSM_ORD_REBUILD (1 << 24)
__u32 disk_ord_tbl[1]; /* disk_ord_tbl[num_members],
} __attribute__ ((packed));
struct imsm_vol {
- __u32 reserved[2];
+ __u32 curr_migr_unit;
+ __u32 checkpoint_id; /* id to access curr_migr_unit */
__u8 migr_state; /* Normal or Migrating */
+#define MIGR_INIT 0
+#define MIGR_REBUILD 1
+#define MIGR_VERIFY 2 /* analagous to echo check > sync_action */
+#define MIGR_GEN_MIGR 3
+#define MIGR_STATE_CHANGE 4
__u8 migr_type; /* Initializing, Rebuilding, ... */
__u8 dirty;
- __u8 fill[1];
- __u32 filler[5];
+ __u8 fs_state; /* fast-sync state for CnG (0xff == disabled) */
+ __u16 verify_errors; /* number of mismatches */
+ __u16 bad_blocks; /* number of bad blocks during verify */
+ __u32 filler[4];
struct imsm_map map[1];
/* here comes another one if migr_state */
} __attribute__ ((packed));
struct imsm_dev {
- __u8 volume[MAX_RAID_SERIAL_LEN];
+ __u8 volume[MAX_RAID_SERIAL_LEN];
__u32 size_low;
__u32 size_high;
+#define DEV_BOOTABLE __cpu_to_le32(0x01)
+#define DEV_BOOT_DEVICE __cpu_to_le32(0x02)
+#define DEV_READ_COALESCING __cpu_to_le32(0x04)
+#define DEV_WRITE_COALESCING __cpu_to_le32(0x08)
+#define DEV_LAST_SHUTDOWN_DIRTY __cpu_to_le32(0x10)
+#define DEV_HIDDEN_AT_BOOT __cpu_to_le32(0x20)
+#define DEV_CURRENTLY_HIDDEN __cpu_to_le32(0x40)
+#define DEV_VERIFY_AND_FIX __cpu_to_le32(0x80)
+#define DEV_MAP_STATE_UNINIT __cpu_to_le32(0x100)
+#define DEV_NO_AUTO_RECOVERY __cpu_to_le32(0x200)
+#define DEV_CLONE_N_GO __cpu_to_le32(0x400)
+#define DEV_CLONE_MAN_SYNC __cpu_to_le32(0x800)
+#define DEV_CNG_MASTER_DISK_NUM __cpu_to_le32(0x1000)
__u32 status; /* Persistent RaidDev status */
__u32 reserved_blocks; /* Reserved blocks at beginning of volume */
-#define IMSM_DEV_FILLERS 12
+ __u8 migr_priority;
+ __u8 num_sub_vols;
+ __u8 tid;
+ __u8 cng_master_disk;
+ __u16 cache_policy;
+ __u8 cng_state;
+ __u8 cng_sub_state;
+#define IMSM_DEV_FILLERS 10
__u32 filler[IMSM_DEV_FILLERS];
struct imsm_vol vol;
} __attribute__ ((packed));
int fd;
} *disks;
struct dl *add; /* list of disks to add while mdmon active */
+ struct dl *missing; /* disks removed while we weren't looking */
struct bbm_log *bbm_log;
};
enum imsm_update_type type;
};
-static int imsm_env_devname_as_serial(void)
-{
- char *val = getenv("IMSM_DEVNAME_AS_SERIAL");
-
- if (val && atoi(val) == 1)
- return 1;
-
- return 0;
-}
-
-
static struct supertype *match_metadata_desc_imsm(char *arg)
{
struct supertype *st;
__u32 *p = (__u32 *) mpb;
__u32 sum = 0;
- while (end--)
- sum += __le32_to_cpu(*p++);
+ while (end--) {
+ sum += __le32_to_cpu(*p);
+ p++;
+ }
return sum - __le32_to_cpu(mpb->check_sum);
}
return map->raid_level;
}
-#ifndef MDASSEMBLE
static int cmp_extent(const void *av, const void *bv)
{
const struct extent *a = av;
struct extent *rv, *e;
int i, j;
int memberships = 0;
+ __u32 reservation = MPB_SECTOR_CNT + IMSM_RESERVED_SECTORS;
for (i = 0; i < super->anchor->num_raid_devs; i++) {
struct imsm_dev *dev = get_imsm_dev(super, i);
}
qsort(rv, memberships, sizeof(*rv), cmp_extent);
- e->start = __le32_to_cpu(dl->disk.total_blocks) -
- (MPB_SECTOR_CNT + IMSM_RESERVED_SECTORS);
+ /* determine the start of the metadata
+ * when no raid devices are defined use the default
+ * ...otherwise allow the metadata to truncate the value
+ * as is the case with older versions of imsm
+ */
+ if (memberships) {
+ struct extent *last = &rv[memberships - 1];
+ __u32 remainder;
+
+ remainder = __le32_to_cpu(dl->disk.total_blocks) -
+ (last->start + last->size);
+ if (reservation > remainder)
+ reservation = remainder;
+ }
+ e->start = __le32_to_cpu(dl->disk.total_blocks) - reservation;
e->size = 0;
return rv;
}
-static void print_imsm_dev(struct imsm_dev *dev, int index)
+/* try to determine how much space is reserved for metadata from
+ * the last get_extents() entry, otherwise fallback to the
+ * default
+ */
+static __u32 imsm_reserved_sectors(struct intel_super *super, struct dl *dl)
+{
+ struct extent *e;
+ int i;
+ __u32 rv;
+
+ /* for spares just return a minimal reservation which will grow
+ * once the spare is picked up by an array
+ */
+ if (dl->index == -1)
+ return MPB_SECTOR_CNT;
+
+ e = get_extents(super, dl);
+ if (!e)
+ return MPB_SECTOR_CNT + IMSM_RESERVED_SECTORS;
+
+ /* scroll to last entry */
+ for (i = 0; e[i].size; i++)
+ continue;
+
+ rv = __le32_to_cpu(dl->disk.total_blocks) - e[i].start;
+
+ free(e);
+
+ return rv;
+}
+
+#ifndef MDASSEMBLE
+static void print_imsm_dev(struct imsm_dev *dev, char *uuid, int disk_idx)
{
__u64 sz;
int slot;
__u32 ord;
printf("\n");
- printf("[%s]:\n", dev->volume);
+ printf("[%.16s]:\n", dev->volume);
+ 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 (index == get_imsm_disk_idx(dev, slot))
+ if (disk_idx== get_imsm_disk_idx(dev, slot))
break;
if (slot < map->num_members) {
ord = get_imsm_ord_tbl_ent(dev, slot);
printf(" Dirty State : %s\n", dev->vol.dirty ? "dirty" : "clean");
}
-static void print_imsm_disk(struct imsm_super *mpb, int index)
+static void print_imsm_disk(struct imsm_super *mpb, int index, __u32 reserved)
{
struct imsm_disk *disk = __get_imsm_disk(mpb, index);
char str[MAX_RAID_SERIAL_LEN + 1];
printf("\n");
snprintf(str, MAX_RAID_SERIAL_LEN + 1, "%s", disk->serial);
printf(" Disk%02d Serial : %s\n", index, str);
- s = __le32_to_cpu(disk->status);
+ s = disk->status;
printf(" State :%s%s%s%s\n", s&SPARE_DISK ? " spare" : "",
s&CONFIGURED_DISK ? " active" : "",
s&FAILED_DISK ? " failed" : "",
s&USABLE_DISK ? " usable" : "");
printf(" Id : %08x\n", __le32_to_cpu(disk->scsi_id));
- sz = __le32_to_cpu(disk->total_blocks) -
- (MPB_SECTOR_CNT + IMSM_RESERVED_SECTORS * mpb->num_raid_devs);
+ sz = __le32_to_cpu(disk->total_blocks) - reserved;
printf(" Usable Size : %llu%s\n", (unsigned long long)sz,
human_size(sz * 512));
}
+static void getinfo_super_imsm(struct supertype *st, struct mdinfo *info);
+
static void examine_super_imsm(struct supertype *st, char *homehost)
{
struct intel_super *super = st->sb;
struct imsm_super *mpb = super->anchor;
char str[MAX_SIGNATURE_LENGTH];
int i;
+ struct mdinfo info;
+ char nbuf[64];
__u32 sum;
+ __u32 reserved = imsm_reserved_sectors(super, super->disks);
+
snprintf(str, MPB_SIG_LEN, "%s", mpb->sig);
printf(" Magic : %s\n", str);
printf(" Version : %s\n", get_imsm_version(mpb));
printf(" Family : %08x\n", __le32_to_cpu(mpb->family_num));
printf(" Generation : %08x\n", __le32_to_cpu(mpb->generation_num));
+ getinfo_super_imsm(st, &info);
+ fname_from_uuid(st, &info, nbuf,'-');
+ printf(" UUID : %s\n", nbuf + 5);
sum = __le32_to_cpu(mpb->check_sum);
printf(" Checksum : %08x %s\n", sum,
__gen_imsm_checksum(mpb) == sum ? "correct" : "incorrect");
printf(" MPB Sectors : %d\n", mpb_sectors(mpb));
printf(" Disks : %d\n", mpb->num_disks);
printf(" RAID Devices : %d\n", mpb->num_raid_devs);
- print_imsm_disk(mpb, super->disks->index);
+ print_imsm_disk(mpb, super->disks->index, reserved);
if (super->bbm_log) {
struct bbm_log *log = super->bbm_log;
printf(" Spare Blocks : %d\n", __le32_to_cpu(log->reserved_spare_block_count));
printf(" First Spare : %llx\n", __le64_to_cpu(log->first_spare_lba));
}
- for (i = 0; i < mpb->num_raid_devs; i++)
- print_imsm_dev(__get_imsm_dev(mpb, i), super->disks->index);
+ for (i = 0; i < mpb->num_raid_devs; i++) {
+ struct mdinfo info;
+ struct imsm_dev *dev = __get_imsm_dev(mpb, i);
+
+ super->current_vol = i;
+ getinfo_super_imsm(st, &info);
+ fname_from_uuid(st, &info, nbuf, '-');
+ print_imsm_dev(dev, nbuf + 5, super->disks->index);
+ }
for (i = 0; i < mpb->num_disks; i++) {
if (i == super->disks->index)
continue;
- print_imsm_disk(mpb, i);
+ print_imsm_disk(mpb, i, reserved);
}
}
-static void getinfo_super_imsm(struct supertype *st, struct mdinfo *info);
-
static void brief_examine_super_imsm(struct supertype *st)
{
- /* We just write a generic DDF ARRAY entry
- */
+ /* We just write a generic IMSM ARRAY entry */
struct mdinfo info;
char nbuf[64];
+ char nbuf1[64];
+ struct intel_super *super = st->sb;
+ int i;
+
+ if (!super->anchor->num_raid_devs)
+ return;
getinfo_super_imsm(st, &info);
fname_from_uuid(st, &info, nbuf,'-');
- printf("ARRAY /dev/imsm metadata=imsm UUID=%s\n", nbuf + 5);
+ printf("ARRAY metadata=imsm auto=md UUID=%s\n", nbuf + 5);
+ for (i = 0; i < super->anchor->num_raid_devs; i++) {
+ struct imsm_dev *dev = get_imsm_dev(super, i);
+
+ super->current_vol = i;
+ getinfo_super_imsm(st, &info);
+ fname_from_uuid(st, &info, nbuf1,'-');
+ printf("ARRAY /dev/md/%.16s container=%s\n"
+ " member=%d auto=mdp UUID=%s\n",
+ dev->volume, nbuf + 5, i, nbuf1 + 5);
+ }
}
static void detail_super_imsm(struct supertype *st, char *homehost)
{
- printf("%s\n", __FUNCTION__);
+ struct mdinfo info;
+ char nbuf[64];
+
+ getinfo_super_imsm(st, &info);
+ fname_from_uuid(st, &info, nbuf,'-');
+ printf("\n UUID : %s\n", nbuf + 5);
}
static void brief_detail_super_imsm(struct supertype *st)
static int match_home_imsm(struct supertype *st, char *homehost)
{
- printf("%s\n", __FUNCTION__);
+ /* the imsm metadata format does not specify any host
+ * identification information. We return -1 since we can never
+ * confirm nor deny whether a given array is "meant" for this
+ * host. We rely on compare_super and the 'family_num' field to
+ * exclude member disks that do not belong, and we rely on
+ * mdadm.conf to specify the arrays that should be assembled.
+ * Auto-assembly may still pick up "foreign" arrays.
+ */
return -1;
}
struct imsm_dev *dev = NULL;
sha1_init_ctx(&ctx);
- sha1_process_bytes(super->anchor->sig, MAX_SIGNATURE_LENGTH, &ctx);
+ sha1_process_bytes(super->anchor->sig, MPB_SIG_LEN, &ctx);
sha1_process_bytes(&super->anchor->family_num, sizeof(__u32), &ctx);
if (super->current_vol >= 0)
dev = get_imsm_dev(super, super->current_vol);
info->component_size = __le32_to_cpu(map->blocks_per_member);
memset(info->uuid, 0, sizeof(info->uuid));
- if (map->map_state == IMSM_T_STATE_UNINITIALIZED ||
- dev->vol.dirty || dev->vol.migr_state)
+ if (map->map_state == IMSM_T_STATE_UNINITIALIZED || dev->vol.dirty)
info->resync_start = 0;
+ else if (dev->vol.migr_state)
+ info->resync_start = __le32_to_cpu(dev->vol.curr_migr_unit);
else
info->resync_start = ~0ULL;
info->name[0] = 0;
if (super->disks) {
+ __u32 reserved = imsm_reserved_sectors(super, super->disks);
+
disk = &super->disks->disk;
- info->disk.number = super->disks->index;
- info->disk.raid_disk = super->disks->index;
- info->data_offset = __le32_to_cpu(disk->total_blocks) -
- (MPB_SECTOR_CNT + IMSM_RESERVED_SECTORS);
- info->component_size = MPB_SECTOR_CNT + IMSM_RESERVED_SECTORS;
- s = __le32_to_cpu(disk->status);
+ info->data_offset = __le32_to_cpu(disk->total_blocks) - reserved;
+ info->component_size = reserved;
+ s = disk->status;
info->disk.state = s & CONFIGURED_DISK ? (1 << MD_DISK_ACTIVE) : 0;
info->disk.state |= s & FAILED_DISK ? (1 << MD_DISK_FAULTY) : 0;
- info->disk.state |= s & USABLE_DISK ? (1 << MD_DISK_SYNC) : 0;
+ info->disk.state |= s & SPARE_DISK ? 0 : (1 << MD_DISK_SYNC);
}
- uuid_from_super_imsm(st, info->uuid);
+
+ /* only call uuid_from_super_imsm when this disk is part of a populated container,
+ * ->compare_super may have updated the 'num_raid_devs' field for spares
+ */
+ if (info->disk.state & (1 << MD_DISK_SYNC) || super->anchor->num_raid_devs)
+ uuid_from_super_imsm(st, info->uuid);
+ else
+ memcpy(info->uuid, uuid_match_any, sizeof(int[4]));
}
static int update_super_imsm(struct supertype *st, struct mdinfo *info,
*/
if (first->anchor->num_raid_devs == 0 &&
sec->anchor->num_raid_devs > 0) {
+ int i;
+
+ /* 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;
+ }
+ *first->dev_tbl[i] = *sec->dev_tbl[i];
+ }
+
first->anchor->num_raid_devs = sec->anchor->num_raid_devs;
first->anchor->family_num = sec->anchor->family_num;
}
rv = scsi_get_serial(fd, scsi_serial, sizeof(scsi_serial));
- if (rv && imsm_env_devname_as_serial()) {
+ if (rv && check_env("IMSM_DEVNAME_AS_SERIAL")) {
memset(serial, 0, MAX_RAID_SERIAL_LEN);
fd2devname(fd, (char *) serial);
return 0;
disk_iter = __get_imsm_disk(super->anchor, i);
if (serialcmp(disk_iter->serial, dl->serial) == 0) {
- __u32 status;
-
dl->disk = *disk_iter;
- status = __le32_to_cpu(dl->disk.status);
/* only set index on disks that are a member of a
* populated contianer, i.e. one with raid_devs
*/
- if (status & FAILED_DISK)
+ if (dl->disk.status & FAILED_DISK)
dl->index = -2;
- else if (status & SPARE_DISK)
+ else if (dl->disk.status & SPARE_DISK)
dl->index = -1;
else
dl->index = i;
}
}
+ /* no match, maybe a stale failed drive */
+ if (i == super->anchor->num_disks && dl->index >= 0) {
+ dl->disk = *__get_imsm_disk(super->anchor, dl->index);
+ if (dl->disk.status & FAILED_DISK)
+ dl->index = -2;
+ }
+
if (alloc)
super->disks = dl;
*
* Migration is indicated by one of the following states
* 1/ Idle (migr_state=0 map0state=normal||unitialized||degraded||failed)
- * 2/ Initialize (migr_state=1 migr_type=0 map0state=normal
+ * 2/ Initialize (migr_state=1 migr_type=MIGR_INIT map0state=normal
* map1state=unitialized)
- * 3/ Verify (Resync) (migr_state=1 migr_type=1 map0state=normal
+ * 3/ Verify (Resync) (migr_state=1 migr_type=MIGR_REBUILD map0state=normal
* map1state=normal)
- * 4/ Rebuild (migr_state=1 migr_type=1 map0state=normal
+ * 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)
dev->vol.migr_state = 1;
dev->vol.migr_type = rebuild_resync;
+ dev->vol.curr_migr_unit = 0;
dest = get_imsm_map(dev, 1);
memcpy(dest, src, sizeof_imsm_map(src));
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);
+
+ dev->vol.migr_state = 0;
+ dev->vol.curr_migr_unit = 0;
+ map->map_state = map_state;
+}
#endif
static int parse_raid_devices(struct intel_super *super)
}
static void free_imsm_disks(struct intel_super *super)
{
- while (super->disks) {
- struct dl *d = super->disks;
+ struct dl *d;
+ while (super->disks) {
+ d = super->disks;
super->disks = d->next;
__free_imsm_disk(d);
}
+ while (super->missing) {
+ d = super->missing;
+ super->missing = d->next;
+ __free_imsm_disk(d);
+ }
+
}
/* free all the pieces hanging off of a super pointer */
}
#ifndef MDASSEMBLE
+/* find_missing - helper routine for load_super_imsm_all that identifies
+ * disks that have disappeared from the system. This routine relies on
+ * the mpb being uptodate, which it is at load time.
+ */
+static int find_missing(struct intel_super *super)
+{
+ int i;
+ struct imsm_super *mpb = super->anchor;
+ struct dl *dl;
+ struct imsm_disk *disk;
+
+ for (i = 0; i < mpb->num_disks; i++) {
+ disk = __get_imsm_disk(mpb, i);
+ for (dl = super->disks; dl; dl = dl->next)
+ if (serialcmp(dl->disk.serial, disk->serial) == 0)
+ break;
+ 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)
+ return 1;
+ dl->major = 0;
+ dl->minor = 0;
+ dl->fd = -1;
+ dl->devname = strdup("missing");
+ dl->index = i;
+ serialcpy(dl->serial, disk->serial);
+ dl->disk = *disk;
+ dl->next = super->missing;
+ super->missing = dl;
+ }
+
+ return 0;
+}
+
static int load_super_imsm_all(struct supertype *st, int fd, void **sbp,
char *devname, int keep_fd)
{
close(dfd);
}
+
+ if (find_missing(super) != 0) {
+ free_imsm(super);
+ return 2;
+ }
+
if (st->subarray[0]) {
if (atoi(st->subarray) <= super->anchor->num_raid_devs)
super->current_vol = atoi(st->subarray);
return (info->size * 2) & ~(info_to_blocks_per_strip(info) - 1);
}
+static void imsm_update_version_info(struct intel_super *super)
+{
+ /* update the version and attributes */
+ struct imsm_super *mpb = super->anchor;
+ char *version;
+ struct imsm_dev *dev;
+ struct imsm_map *map;
+ int i;
+
+ for (i = 0; i < mpb->num_raid_devs; i++) {
+ dev = get_imsm_dev(super, i);
+ map = get_imsm_map(dev, 0);
+ if (__le32_to_cpu(dev->size_high) > 0)
+ mpb->attributes |= MPB_ATTRIB_2TB;
+
+ /* FIXME detect when an array spans a port multiplier */
+ #if 0
+ mpb->attributes |= MPB_ATTRIB_PM;
+ #endif
+
+ if (mpb->num_raid_devs > 1 ||
+ mpb->attributes != MPB_ATTRIB_CHECKSUM_VERIFY) {
+ version = MPB_VERSION_ATTRIBS;
+ switch (get_imsm_raid_level(map)) {
+ case 0: mpb->attributes |= MPB_ATTRIB_RAID0; break;
+ case 1: mpb->attributes |= MPB_ATTRIB_RAID1; break;
+ case 10: mpb->attributes |= MPB_ATTRIB_RAID10; break;
+ case 5: mpb->attributes |= MPB_ATTRIB_RAID5; break;
+ }
+ } else {
+ if (map->num_members >= 5)
+ version = MPB_VERSION_5OR6_DISK_ARRAY;
+ else if (dev->status == DEV_CLONE_N_GO)
+ version = MPB_VERSION_CNG;
+ else if (get_imsm_raid_level(map) == 5)
+ version = MPB_VERSION_RAID5;
+ else if (map->num_members >= 3)
+ version = MPB_VERSION_3OR4_DISK_ARRAY;
+ else if (get_imsm_raid_level(map) == 1)
+ version = MPB_VERSION_RAID1;
+ else
+ version = MPB_VERSION_RAID0;
+ }
+ strcpy(((char *) mpb->sig) + strlen(MPB_SIGNATURE), version);
+ }
+}
+
static int init_super_imsm_volume(struct supertype *st, mdu_array_info_t *info,
unsigned long long size, char *name,
char *homehost, int *uuid)
dev->reserved_blocks = __cpu_to_le32(0);
vol = &dev->vol;
vol->migr_state = 0;
- vol->migr_type = 0;
+ vol->migr_type = MIGR_INIT;
vol->dirty = 0;
+ vol->curr_migr_unit = 0;
for (i = 0; i < idx; i++) {
struct imsm_dev *prev = get_imsm_dev(super, i);
struct imsm_map *pmap = get_imsm_map(prev, 0);
"in a raid1 volume\n");
return 0;
}
- if (info->level == 10)
+ if (info->level == 10) {
map->raid_level = 1;
- else
+ map->num_domains = info->raid_disks / 2;
+ } else {
map->raid_level = info->level;
+ map->num_domains = !!map->raid_level;
+ }
map->num_members = info->raid_disks;
for (i = 0; i < map->num_members; i++) {
mpb->num_raid_devs++;
super->dev_tbl[super->current_vol] = dev;
+ imsm_update_version_info(super);
+
return 1;
}
struct intel_super *super;
struct imsm_super *mpb;
size_t mpb_size;
+ char *version;
if (!info) {
st->sb = NULL;
mpb = super->buf;
memset(mpb, 0, mpb_size);
- memcpy(mpb->sig, MPB_SIGNATURE, strlen(MPB_SIGNATURE));
- memcpy(mpb->sig + strlen(MPB_SIGNATURE), MPB_VERSION_RAID5,
- strlen(MPB_VERSION_RAID5));
+ mpb->attributes = MPB_ATTRIB_CHECKSUM_VERIFY;
+
+ version = (char *) mpb->sig;
+ strcpy(version, MPB_SIGNATURE);
+ version += strlen(MPB_SIGNATURE);
+ strcpy(version, MPB_VERSION_RAID0);
mpb->mpb_size = mpb_size;
st->sb = super;
struct dl *dl;
struct imsm_dev *dev;
struct imsm_map *map;
- __u32 status;
dev = get_imsm_dev(super, super->current_vol);
map = get_imsm_map(dev, 0);
super->anchor->num_disks++;
}
set_imsm_ord_tbl_ent(map, dk->number, dl->index);
- status = CONFIGURED_DISK | USABLE_DISK;
- dl->disk.status = __cpu_to_le32(status);
+ dl->disk.status = CONFIGURED_DISK | USABLE_DISK;
/* if we are creating the first raid device update the family number */
if (super->current_vol == 0) {
struct intel_super *super = st->sb;
struct dl *dd;
unsigned long long size;
- __u32 status, id;
+ __u32 id;
int rv;
struct stat stb;
get_dev_size(fd, NULL, &size);
size /= 512;
- status = USABLE_DISK | SPARE_DISK;
serialcpy(dd->disk.serial, dd->serial);
dd->disk.total_blocks = __cpu_to_le32(size);
- dd->disk.status = __cpu_to_le32(status);
+ dd->disk.status = USABLE_DISK | SPARE_DISK;
if (sysfs_disk_to_scsi_id(fd, &id) == 0)
dd->disk.scsi_id = __cpu_to_le32(id);
else
generation++;
mpb->generation_num = __cpu_to_le32(generation);
+ mpb_size += sizeof(struct imsm_disk) * mpb->num_disks;
for (d = super->disks; d; d = d->next) {
if (d->index == -1)
spares++;
- else {
+ else
mpb->disk[d->index] = d->disk;
- mpb_size += sizeof(struct imsm_disk);
- }
}
+ for (d = super->missing; d; d = d->next)
+ mpb->disk[d->index] = d->disk;
for (i = 0; i < mpb->num_raid_devs; i++) {
struct imsm_dev *dev = __get_imsm_dev(mpb, i);
if (d == NULL)
skip = 1;
- s = d ? __le32_to_cpu(d->disk.status) : 0;
+ s = d ? d->disk.status : 0;
if (s & FAILED_DISK)
skip = 1;
if (!(s & USABLE_DISK))
* slot+1
*/
int i;
- int insync;
+ /* gcc -Os complains that this is unused */
+ int insync = insync;
for (i = 0; i < map->num_members; i++) {
__u32 ord = get_imsm_ord_tbl_ent(dev, i);
insync = 2;
disk = get_imsm_disk(super, idx);
- if (!disk ||
- __le32_to_cpu(disk->status) & FAILED_DISK ||
+ if (!disk || disk->status & FAILED_DISK ||
ord & IMSM_ORD_REBUILD)
insync--;
int idx = ord_to_idx(ord);
disk = get_imsm_disk(super, idx);
- if (!disk ||
- __le32_to_cpu(disk->status) & FAILED_DISK ||
+ if (!disk || disk->status & FAILED_DISK ||
ord & IMSM_ORD_REBUILD)
failed++;
}
if (!dev->vol.migr_state)
return 0;
- if (dev->vol.migr_type == 0)
+ if (dev->vol.migr_type == MIGR_INIT)
return 1;
migr_map = get_imsm_map(dev, 1);
if (!dev->vol.migr_state)
return 0;
- if (dev->vol.migr_type == 0)
+ if (dev->vol.migr_type != MIGR_REBUILD)
return 0;
migr_map = get_imsm_map(dev, 1);
return 0;
}
+static void mark_failure(struct imsm_disk *disk)
+{
+ if (disk->status & FAILED_DISK)
+ return;
+ disk->status |= FAILED_DISK;
+ disk->scsi_id = __cpu_to_le32(~(__u32)0);
+ memmove(&disk->serial[0], &disk->serial[1], MAX_RAID_SERIAL_LEN - 1);
+}
+
/* 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
int failed = imsm_count_failed(super, dev);
__u8 map_state = imsm_check_degraded(super, dev, failed);
+ /* before we activate this array handle any missing disks */
+ if (consistent == 2 && super->missing) {
+ struct dl *dl;
+
+ dprintf("imsm: mark missing\n");
+ end_migration(dev, map_state);
+ for (dl = super->missing; dl; dl = dl->next)
+ mark_failure(&dl->disk);
+ super->updates_pending++;
+ }
+
if (consistent == 2 &&
- (a->resync_start != ~0ULL ||
+ (!is_resync_complete(a) ||
map_state != IMSM_T_STATE_NORMAL ||
dev->vol.migr_state))
consistent = 0;
- if (a->resync_start == ~0ULL) {
+ if (is_resync_complete(a)) {
/* complete intialization / resync,
* recovery is completed in ->set_disk
*/
if (is_resyncing(dev)) {
dprintf("imsm: mark resync done\n");
- dev->vol.migr_state = 0;
- map->map_state = map_state;
+ end_migration(dev, map_state);
super->updates_pending++;
}
} 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);
- map->map_state = map_state;
- migrate(dev, IMSM_T_STATE_NORMAL,
- map->map_state == IMSM_T_STATE_NORMAL);
+ if (map->map_state == IMSM_T_STATE_NORMAL)
+ migrate(dev, IMSM_T_STATE_NORMAL, MIGR_REBUILD);
+ else
+ migrate(dev, IMSM_T_STATE_NORMAL, MIGR_INIT);
+ super->updates_pending++;
+ }
+
+ /* check if we can update the migration checkpoint */
+ if (dev->vol.migr_state &&
+ __le32_to_cpu(dev->vol.curr_migr_unit) != a->resync_start) {
+ dprintf("imsm: checkpoint migration (%llu)\n", a->resync_start);
+ dev->vol.curr_migr_unit = __cpu_to_le32(a->resync_start);
super->updates_pending++;
}
struct imsm_map *map = get_imsm_map(dev, 0);
struct imsm_disk *disk;
int failed;
- __u32 status;
__u32 ord;
__u8 map_state;
disk = get_imsm_disk(super, ord_to_idx(ord));
/* check for new failures */
- status = __le32_to_cpu(disk->status);
- if ((state & DS_FAULTY) && !(status & FAILED_DISK)) {
- status |= FAILED_DISK;
- disk->status = __cpu_to_le32(status);
- disk->scsi_id = __cpu_to_le32(~(__u32)0);
- memmove(&disk->serial[0], &disk->serial[1], MAX_RAID_SERIAL_LEN - 1);
+ if ((state & DS_FAULTY) && !(disk->status & FAILED_DISK)) {
+ mark_failure(disk);
super->updates_pending++;
}
+
/* check if in_sync */
if (state & DS_INSYNC && ord & IMSM_ORD_REBUILD) {
struct imsm_map *migr_map = get_imsm_map(dev, 1);
/* check if recovery complete, newly degraded, or failed */
if (map_state == IMSM_T_STATE_NORMAL && is_rebuilding(dev)) {
- map->map_state = map_state;
- dev->vol.migr_state = 0;
+ end_migration(dev, map_state);
super->updates_pending++;
} else if (map_state == IMSM_T_STATE_DEGRADED &&
map->map_state != map_state &&
} else if (map_state == IMSM_T_STATE_FAILED &&
map->map_state != map_state) {
dprintf("imsm: mark failed\n");
- dev->vol.migr_state = 0;
- map->map_state = map_state;
+ end_migration(dev, map_state);
super->updates_pending++;
}
}
if (dl->index == i)
break;
- if (dl && __le32_to_cpu(dl->disk.status) & FAILED_DISK)
+ if (dl && dl->disk.status & FAILED_DISK)
dl = NULL;
if (dl)
int j;
int found;
__u32 array_start;
- __u32 status;
struct dl *dl;
for (dl = super->disks; dl; dl = dl->next) {
continue;
/* skip in use or failed drives */
- status = __le32_to_cpu(dl->disk.status);
- if (status & FAILED_DISK || idx == dl->index) {
+ if (dl->disk.status & FAILED_DISK || idx == dl->index) {
dprintf("%x:%x status ( %s%s)\n",
dl->major, dl->minor,
- status & FAILED_DISK ? "failed " : "",
+ dl->disk.status & FAILED_DISK ? "failed " : "",
idx == dl->index ? "in use " : "");
continue;
}
/* found a usable disk with enough space */
di = malloc(sizeof(*di));
+ if (!di)
+ continue;
memset(di, 0, sizeof(*di));
/* dl->index will be -1 in the case we are activating a
* disk_ord_tbl for the array
*/
mu = malloc(sizeof(*mu));
- mu->buf = malloc(sizeof(struct imsm_update_activate_spare) * num_spares);
+ if (mu) {
+ mu->buf = malloc(sizeof(struct imsm_update_activate_spare) * num_spares);
+ if (mu->buf == NULL) {
+ free(mu);
+ mu = NULL;
+ }
+ }
+ if (!mu) {
+ while (rv) {
+ struct mdinfo *n = rv->next;
+
+ free(rv);
+ rv = n;
+ }
+ return NULL;
+ }
+
mu->space = NULL;
mu->len = sizeof(struct imsm_update_activate_spare) * num_spares;
mu->next = *updates;
struct imsm_map *migr_map;
struct active_array *a;
struct imsm_disk *disk;
- __u32 status;
__u8 to_state;
struct dl *dl;
unsigned int found;
if (i == u->slot)
continue;
disk = get_imsm_disk(super, get_imsm_disk_idx(dev, i));
- if (!disk ||
- __le32_to_cpu(disk->status) & FAILED_DISK)
+ if (!disk || disk->status & FAILED_DISK)
failed++;
}
super->anchor->num_disks++;
}
disk = &dl->disk;
- status = __le32_to_cpu(disk->status);
- status |= CONFIGURED_DISK;
- status &= ~SPARE_DISK;
- disk->status = __cpu_to_le32(status);
+ disk->status |= CONFIGURED_DISK;
+ disk->status &= ~SPARE_DISK;
/* mark rebuild */
to_state = imsm_check_degraded(super, dev, failed);
map->map_state = IMSM_T_STATE_DEGRADED;
- migrate(dev, to_state, 1);
+ migrate(dev, to_state, MIGR_REBUILD);
migr_map = get_imsm_map(dev, 1);
set_imsm_ord_tbl_ent(map, u->slot, dl->index);
set_imsm_ord_tbl_ent(migr_map, u->slot, dl->index | IMSM_ORD_REBUILD);
if (!found) {
struct dl **dlp;
+ /* We know that 'manager' isn't touching anything,
+ * so it is safe to delete
+ */
for (dlp = &super->disks; *dlp; dlp = &(*dlp)->next)
if ((*dlp)->index == victim)
break;
- /* We know that 'manager' isn't touching anything,
- * so it is safe to:
- */
+
+ /* victim may be on the missing list */
+ if (!*dlp)
+ for (dlp = &super->missing; *dlp; dlp = &(*dlp)->next)
+ if ((*dlp)->index == victim)
+ break;
imsm_delete(super, dlp, victim);
}
break;
/* fix up flags */
for (i = 0; i < map->num_members; i++) {
struct imsm_disk *disk;
- __u32 status;
disk = get_imsm_disk(super, get_imsm_disk_idx(dev, i));
- status = __le32_to_cpu(disk->status);
- status |= CONFIGURED_DISK;
- status &= ~SPARE_DISK;
- disk->status = __cpu_to_le32(status);
+ disk->status |= CONFIGURED_DISK;
+ disk->status &= ~SPARE_DISK;
}
+
+ imsm_update_version_info(super);
+
break;
}
case update_add_disk:
* being added */
if (super->add) {
struct active_array *a;
+
+ super->updates_pending++;
for (a = st->arrays; a; a = a->next)
a->check_degraded = 1;
}
for (iter = super->disks; iter; iter = iter->next)
if (iter->index > index)
iter->index--;
+ for (iter = super->missing; iter; iter = iter->next)
+ if (iter->index > index)
+ iter->index--;
for (i = 0; i < mpb->num_raid_devs; i++) {
dev = get_imsm_dev(super, i);
projects / thirdparty / mdadm.git / blobdiff