/*
* mdadm - manage Linux "md" devices aka RAID arrays.
*
- * Copyright (C) 2001-2006 Neil Brown <neilb@suse.de>
+ * Copyright (C) 2001-2009 Neil Brown <neilb@suse.de>
*
*
* This program is free software; you can redistribute it and/or modify
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*
* Author: Neil Brown
- * Email: <neilb@cse.unsw.edu.au>
- * Paper: Neil Brown
- * School of Computer Science and Engineering
- * The University of New South Wales
- * Sydney, 2052
- * Australia
+ * Email: <neilb@suse.de>
*/
#define HAVE_STDINT_H 1
mdp_super_t *sb = st->sb;
time_t atime;
int d;
+ int delta_extra = 0;
char *c;
printf(" Magic : %08x\n", sb->md_magic);
printf(" Creation Time : %.24s\n", ctime(&atime));
c=map_num(pers, sb->level);
printf(" Raid Level : %s\n", c?c:"-unknown-");
- if ((int)sb->level >= 0) {
- int ddsks=0;
+ if ((int)sb->level > 0) {
+ int ddsks = 0, ddsks_denom = 1;
printf(" Used Dev Size : %d%s\n", sb->size,
human_size((long long)sb->size<<10));
switch(sb->level) {
case 4:
case 5: ddsks = sb->raid_disks-1; break;
case 6: ddsks = sb->raid_disks-2; break;
- case 10: ddsks = sb->raid_disks / (sb->layout&255) / ((sb->layout>>8)&255);
+ case 10: ddsks = sb->raid_disks;
+ ddsks_denom = (sb->layout&255) * ((sb->layout>>8)&255);
+ }
+ if (ddsks) {
+ long long asize = sb->size;
+ asize = (asize << 10) * ddsks / ddsks_denom;
+ printf(" Array Size : %llu%s\n",
+ asize >> 10, human_size(asize));
}
- if (ddsks)
- printf(" Array Size : %llu%s\n", (unsigned long long)ddsks * sb->size,
- human_size(ddsks*(long long)sb->size<<10));
}
printf(" Raid Devices : %d\n", sb->raid_disks);
printf(" Total Devices : %d\n", sb->nr_disks);
printf(" Reshape pos'n : %llu%s\n", (unsigned long long)sb->reshape_position/2, human_size((long long)sb->reshape_position<<9));
if (sb->delta_disks) {
printf(" Delta Devices : %d", sb->delta_disks);
- if (sb->delta_disks)
- printf(" (%d->%d)\n", sb->raid_disks-sb->delta_disks, sb->raid_disks);
- else
- printf(" (%d->%d)\n", sb->raid_disks, sb->raid_disks+sb->delta_disks);
+ printf(" (%d->%d)\n", sb->raid_disks-sb->delta_disks, sb->raid_disks);
+ if (((int)sb->delta_disks) < 0)
+ delta_extra = - sb->delta_disks;
}
if (sb->new_level != sb->level) {
c = map_num(pers, sb->new_level);
c = map_num(r5layout, sb->new_layout);
printf(" New Layout : %s\n", c?c:"-unknown-");
}
+ if (sb->level == 6) {
+ c = map_num(r6layout, sb->new_layout);
+ printf(" New Layout : %s\n", c?c:"-unknown-");
+ }
if (sb->level == 10) {
printf(" New Layout : near=%d, %s=%d\n",
sb->new_layout&255,
c = map_num(r5layout, sb->layout);
printf(" Layout : %s\n", c?c:"-unknown-");
}
+ if (sb->level == 6) {
+ c = map_num(r6layout, sb->layout);
+ printf(" Layout : %s\n", c?c:"-unknown-");
+ }
if (sb->level == 10) {
printf(" Layout :");
print_r10_layout(sb->layout);
}
printf("\n");
printf(" Number Major Minor RaidDevice State\n");
- for (d= -1; d<(signed int)(sb->raid_disks+sb->spare_disks); d++) {
+ for (d= -1; d<(signed int)(sb->raid_disks+delta_extra + sb->spare_disks); d++) {
mdp_disk_t *dp;
char *dv;
char nb[5];
{
mdp_super_t *sb = st->sb;
char buf[20];
- char *hash = sha1_buffer(homehost,
- strlen(homehost),
- buf);
+ char *hash;
+
+ if (!homehost)
+ return 0;
+ hash = sha1_buffer(homehost,
+ strlen(homehost),
+ buf);
return (memcmp(&sb->set_uuid2, hash, 8)==0);
}
}
}
-static void getinfo_super0(struct supertype *st, struct mdinfo *info)
+static void getinfo_super0(struct supertype *st, struct mdinfo *info, char *map)
{
mdp_super_t *sb = st->sb;
int working = 0;
int i;
+ int map_disks = info->array.raid_disks;
+ memset(info, 0, sizeof(*info));
info->array.major_version = sb->major_version;
info->array.minor_version = sb->minor_version;
info->array.patch_version = sb->patch_version;
info->array.state = sb->state;
info->component_size = sb->size*2;
+ if (sb->state & (1<<MD_SB_BITMAP_PRESENT))
+ info->bitmap_offset = 8;
+
info->disk.state = sb->this_disk.state;
info->disk.major = sb->this_disk.major;
info->disk.minor = sb->this_disk.minor;
uuid_from_super0(st, info->uuid);
+ info->recovery_start = MaxSector;
if (sb->minor_version > 90 && (sb->reshape_position+1) != 0) {
info->reshape_active = 1;
info->reshape_progress = sb->reshape_position;
info->delta_disks = sb->delta_disks;
info->new_layout = sb->new_layout;
info->new_chunk = sb->new_chunk;
+ if (info->delta_disks < 0)
+ info->array.raid_disks -= info->delta_disks;
} else
info->reshape_active = 0;
+ info->recovery_blocked = info->reshape_active;
+
sprintf(info->name, "%d", sb->md_minor);
/* work_disks is calculated rather than read directly */
for (i=0; i < MD_SB_DISKS; i++)
if ((sb->disks[i].state & (1<<MD_DISK_SYNC)) &&
- (sb->disks[i].raid_disk < info->array.raid_disks) &&
+ (sb->disks[i].raid_disk < (unsigned)info->array.raid_disks) &&
(sb->disks[i].state & (1<<MD_DISK_ACTIVE)) &&
- !(sb->disks[i].state & (1<<MD_DISK_FAULTY)))
+ !(sb->disks[i].state & (1<<MD_DISK_FAULTY))) {
working ++;
+ if (map && i < map_disks)
+ map[i] = 1;
+ } else if (map && i < map_disks)
+ map[i] = 0;
info->array.working_disks = working;
}
+static struct mdinfo *container_content0(struct supertype *st, char *subarray)
+{
+ struct mdinfo *info;
+
+ if (subarray)
+ return NULL;
+
+ info = malloc(sizeof(*info));
+ getinfo_super0(st, info, NULL);
+ return info;
+}
static int update_super0(struct supertype *st, struct mdinfo *info,
char *update,
char *devname, int verbose,
int uuid_set, char *homehost)
{
- /* NOTE: for 'assemble' and 'force' we need to return non-zero if any change was made.
- * For others, the return value is ignored.
+ /* NOTE: for 'assemble' and 'force' we need to return non-zero
+ * if any change was made. For others, the return value is
+ * ignored.
*/
int rv = 0;
+ int uuid[4];
mdp_super_t *sb = st->sb;
if (strcmp(update, "sparc2.2")==0 ) {
/* 2.2 sparc put the events in the wrong place
if (verbose >= 0)
fprintf (stderr, Name ": adjusting superblock of %s for 2.2/sparc compatability.\n",
devname);
- }
- if (strcmp(update, "super-minor") ==0) {
+ } else if (strcmp(update, "super-minor") ==0) {
sb->md_minor = info->array.md_minor;
if (verbose > 0)
fprintf(stderr, Name ": updating superblock of %s with minor number %d\n",
devname, info->array.md_minor);
- }
- if (strcmp(update, "summaries") == 0) {
- int i;
+ } else if (strcmp(update, "summaries") == 0) {
+ unsigned int i;
/* set nr_disks, active_disks, working_disks,
* failed_disks, spare_disks based on disks[]
* array in superblock.
sb->spare_disks++;
} else if (i >= sb->raid_disks && sb->disks[i].number == 0)
sb->disks[i].state = 0;
- }
- if (strcmp(update, "force-one")==0) {
+ } else if (strcmp(update, "force-one")==0) {
/* Not enough devices for a working array, so
* bring this one up-to-date.
*/
if (sb->events_hi != ehi ||
sb->events_lo != elo)
rv = 1;
- }
- if (strcmp(update, "force-array")==0) {
+ } else if (strcmp(update, "force-array")==0) {
/* degraded array and 'force' requested, so
* maybe need to mark it 'clean'
*/
sb->state |= (1 << MD_SB_CLEAN);
rv = 1;
}
- }
- if (strcmp(update, "assemble")==0) {
+ } else if (strcmp(update, "assemble")==0) {
int d = info->disk.number;
int wonly = sb->disks[d].state & (1<<MD_DISK_WRITEMOSTLY);
- if ((sb->disks[d].state & ~(1<<MD_DISK_WRITEMOSTLY))
- != info->disk.state) {
+ int mask = (1<<MD_DISK_WRITEMOSTLY);
+ int add = 0;
+ if (sb->minor_version >= 91)
+ /* During reshape we don't insist on everything
+ * being marked 'sync'
+ */
+ add = (1<<MD_DISK_SYNC);
+ if (((sb->disks[d].state & ~mask) | add)
+ != (unsigned)info->disk.state) {
sb->disks[d].state = info->disk.state | wonly;
rv = 1;
}
- }
- if (strcmp(update, "linear-grow-new") == 0) {
+ if (info->reshape_active &&
+ sb->minor_version > 90 && (sb->reshape_position+1) != 0 &&
+ info->delta_disks >= 0 &&
+ info->reshape_progress < sb->reshape_position) {
+ sb->reshape_position = info->reshape_progress;
+ rv = 1;
+ }
+ if (info->reshape_active &&
+ sb->minor_version > 90 && (sb->reshape_position+1) != 0 &&
+ info->delta_disks < 0 &&
+ info->reshape_progress > sb->reshape_position) {
+ sb->reshape_position = info->reshape_progress;
+ rv = 1;
+ }
+ } else if (strcmp(update, "linear-grow-new") == 0) {
memset(&sb->disks[info->disk.number], 0, sizeof(sb->disks[0]));
sb->disks[info->disk.number].number = info->disk.number;
sb->disks[info->disk.number].major = info->disk.major;
sb->disks[info->disk.number].raid_disk = info->disk.raid_disk;
sb->disks[info->disk.number].state = info->disk.state;
sb->this_disk = sb->disks[info->disk.number];
- }
- if (strcmp(update, "linear-grow-update") == 0) {
+ } else if (strcmp(update, "linear-grow-update") == 0) {
sb->raid_disks = info->array.raid_disks;
sb->nr_disks = info->array.nr_disks;
sb->active_disks = info->array.active_disks;
sb->disks[info->disk.number].minor = info->disk.minor;
sb->disks[info->disk.number].raid_disk = info->disk.raid_disk;
sb->disks[info->disk.number].state = info->disk.state;
- }
- if (strcmp(update, "resync") == 0) {
+ } else if (strcmp(update, "resync") == 0) {
/* make sure resync happens */
sb->state &= ~(1<<MD_SB_CLEAN);
sb->recovery_cp = 0;
- }
- if (strcmp(update, "homehost") == 0 &&
- homehost) {
+ } else if (strcmp(update, "homehost") == 0 &&
+ homehost) {
uuid_set = 0;
update = "uuid";
info->uuid[0] = sb->set_uuid0;
info->uuid[1] = sb->set_uuid1;
- }
- if (strcmp(update, "uuid") == 0) {
+ } else if (strcmp(update, "uuid") == 0) {
if (!uuid_set && homehost) {
char buf[20];
char *hash = sha1_buffer(homehost,
if (sb->state & (1<<MD_SB_BITMAP_PRESENT)) {
struct bitmap_super_s *bm;
bm = (struct bitmap_super_s*)(sb+1);
- uuid_from_super0(st, (int*)bm->uuid);
+ uuid_from_super0(st, uuid);
+ memcpy(bm->uuid, uuid, 16);
}
- }
- if (strcmp(update, "_reshape_progress")==0)
+ } else if (strcmp(update, "no-bitmap") == 0) {
+ sb->state &= ~(1<<MD_SB_BITMAP_PRESENT);
+ } else if (strcmp(update, "_reshape_progress")==0)
sb->reshape_position = info->reshape_progress;
+ else if (strcmp(update, "writemostly")==0)
+ sb->state |= (1<<MD_DISK_WRITEMOSTLY);
+ else if (strcmp(update, "readwrite")==0)
+ sb->state &= ~(1<<MD_DISK_WRITEMOSTLY);
+ else
+ rv = -1;
sb->sb_csum = calc_sb0_csum(sb);
return rv;
sb->gvalid_words = 0; /* ignored */
sb->ctime = time(0);
sb->level = info->level;
- if (size != info->size)
+ if (size != (unsigned long long)info->size)
return 0;
sb->size = info->size;
sb->nr_disks = info->nr_disks;
dk->minor = dinfo->minor;
dk->raid_disk = dinfo->raid_disk;
dk->state = dinfo->state;
+ /* In case our source disk was writemostly, don't copy that bit */
+ dk->state &= ~(1<<MD_DISK_WRITEMOSTLY);
sb->this_disk = sb->disks[dinfo->number];
sb->sb_csum = calc_sb0_csum(sb);
continue;
if (di->fd == -1)
continue;
- Kill(di->devname, 0, 1, 1);
- Kill(di->devname, 0, 1, 1);
+ while (Kill(di->devname, NULL, 0, 1, 1) == 0)
+ ;
sb->disks[di->disk.number].state &= ~(1<<MD_DISK_FAULTY);
fprintf(stderr,
Name ": failed to write superblock to %s\n",
di->devname);
- close(di->fd);
- di->fd = -1;
}
return rv;
}
free_super0(st);
- if (st->subarray[0])
- return 1;
-
if (!get_dev_size(fd, devname, &dsize))
return 1;
static struct supertype *match_metadata_desc0(char *arg)
{
- struct supertype *st = malloc(sizeof(*st));
- if (!st) return st;
+ struct supertype *st = calloc(1, sizeof(*st));
+ if (!st)
+ return st;
- memset(st, 0, sizeof(*st));
+ st->container_dev = NoMdDev;
st->ss = &super0;
st->info = NULL;
st->minor_version = 90;
while (arg[0] == '0' && arg[1] == '0')
arg++;
if (strcmp(arg, "0") == 0 ||
- strcmp(arg, "0.90") == 0 ||
+#ifdef DEFAULT_OLD_METADATA /* ifndef in super1.c */
strcmp(arg, "default") == 0 ||
- strcmp(arg, "") == 0 /* no metadata */
+#endif /* DEFAULT_OLD_METADATA */
+ strcmp(arg, "0.90") == 0 ||
+ strcmp(arg, "") == 0 /* no metadata - i.e. non_persistent */
)
return st;
* size is in sectors, chunk is in bytes !!!
*/
unsigned long long bits;
- unsigned long long max_bits = 60*1024*8;
+ unsigned long long max_bits = (60*1024 - sizeof(bitmap_super_t))*8;
unsigned long long min_chunk;
int chunk = *chunkp;
mdp_super_t *sb = st->sb;
bitmap_super_t *bms = (bitmap_super_t*)(((char*)sb) + MD_SB_BYTES);
+ int uuid[4];
min_chunk = 4096; /* sub-page chunks don't work yet.. */
min_chunk *= 2;
bits = (bits+1)/2;
}
- if (chunk == UnSet)
+ if (chunk == UnSet) {
+ /* A chunk size less than a few Megabytes gives poor
+ * performance without increasing resync noticeably
+ */
chunk = min_chunk;
- else if (chunk < min_chunk)
+ if (chunk < 64*1024*1024)
+ chunk = 64*1024*1024;
+ } else if ((unsigned long long)chunk < min_chunk)
return 0; /* chunk size too small */
sb->state |= (1<<MD_SB_BITMAP_PRESENT);
memset(bms, 0, sizeof(*bms));
bms->magic = __cpu_to_le32(BITMAP_MAGIC);
bms->version = __cpu_to_le32(major);
- uuid_from_super0(st, (int*)bms->uuid);
+ uuid_from_super0(st, uuid);
+ memcpy(bms->uuid, uuid, 16);
bms->chunksize = __cpu_to_le32(chunk);
bms->daemon_sleep = __cpu_to_le32(delay);
bms->sync_size = __cpu_to_le64(size);
int rv = 0;
int towrite, n;
- char abuf[4096+4096];
- char *buf = (char*)(((long)(abuf+4096))&~4095L);
+ void *buf;
if (!get_dev_size(fd, NULL, &dsize))
return 1;
-
if (dsize < MD_RESERVED_SECTORS*512)
return -1;
if (lseek64(fd, offset + 4096, 0)< 0LL)
return 3;
+ if (posix_memalign(&buf, 4096, 4096))
+ return -ENOMEM;
+
memset(buf, 0xff, 4096);
memcpy(buf, ((char*)sb)+MD_SB_BYTES, sizeof(bitmap_super_t));
- towrite = 64*1024;
+ towrite = 60*1024;
while (towrite > 0) {
n = towrite;
if (n > 4096)
if (towrite)
rv = -2;
+ free(buf);
return rv;
}
{
if (st->sb)
free(st->sb);
+ while (st->info) {
+ struct devinfo *di = st->info;
+ st->info = di->next;
+ if (di->fd >= 0)
+ close(di->fd);
+ free(di);
+ }
st->sb = NULL;
}
#ifndef MDASSEMBLE
static int validate_geometry0(struct supertype *st, int level,
int layout, int raiddisks,
- int chunk, unsigned long long size,
+ int *chunk, unsigned long long size,
char *subdev, unsigned long long *freesize,
int verbose)
{
unsigned long long ldsize;
int fd;
+ unsigned int tbmax = 4;
- if (level == LEVEL_CONTAINER)
+ /* prior to linux 3.1, a but limits usable device size to 2TB.
+ * It was introduced in 2.6.29, but we won't worry about that detail
+ */
+ if (get_linux_version() < 3001000)
+ tbmax = 2;
+
+ if (level == LEVEL_CONTAINER) {
+ if (verbose)
+ fprintf(stderr, Name ": 0.90 metadata does not support containers\n");
return 0;
- if (raiddisks > MD_SB_DISKS)
+ }
+ if (raiddisks > MD_SB_DISKS) {
+ if (verbose)
+ fprintf(stderr, Name ": 0.90 metadata supports at most %d devices per array\n",
+ MD_SB_DISKS);
return 0;
- if (size > (0x7fffffffULL<<9))
+ }
+ if (size >= tbmax * 2ULL*1024*1024*1024) {
+ if (verbose)
+ fprintf(stderr, Name ": 0.90 metadata supports at most "
+ "%d terabytes per device\n", tbmax);
return 0;
+ }
+ if (chunk && *chunk == UnSet)
+ *chunk = DEFAULT_CHUNK;
+
if (!subdev)
return 1;
if (ldsize < MD_RESERVED_SECTORS * 512)
return 0;
- if (size > (0x7fffffffULL<<9))
- return 0;
*freesize = MD_NEW_SIZE_SECTORS(ldsize >> 9);
return 1;
}
.match_home = match_home0,
.uuid_from_super = uuid_from_super0,
.getinfo_super = getinfo_super0,
+ .container_content = container_content0,
.update_super = update_super0,
.init_super = init_super0,
.store_super = store_super0,