/*
* 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>
*/
#include "mdadm.h"
__u64 utime; /* 40 bits second, 24 btes microseconds */
__u64 events; /* incremented when superblock updated */
__u64 resync_offset; /* data before this offset (from data_offset) known to be in sync */
- __u32 sb_csum; /* checksum upto devs[max_dev] */
- __u32 max_dev; /* size of devs[] array to consider */
+ __u32 sb_csum; /* checksum upto dev_roles[max_dev] */
+ __u32 max_dev; /* size of dev_roles[] array to consider */
__u8 pad3[64-32]; /* set to 0 when writing */
/* device state information. Indexed by dev_number.
__u16 dev_roles[0]; /* role in array, or 0xffff for a spare, or 0xfffe for faulty */
};
+#define MAX_SB_SIZE 4096
+/* bitmap super size is 256, but we round up to a sector for alignment */
+#define BM_SUPER_SIZE 512
+#define MAX_DEVS ((int)(MAX_SB_SIZE - sizeof(struct mdp_superblock_1)) / 2)
+#define SUPER1_SIZE (MAX_SB_SIZE + BM_SUPER_SIZE \
+ + sizeof(struct misc_dev_info))
+
struct misc_dev_info {
__u64 device_size;
};
unsigned long long newcsum;
int size = sizeof(*sb) + __le32_to_cpu(sb->max_dev)*2;
unsigned int *isuper = (unsigned int*)sb;
- int i;
/* make sure I can count... */
if (offsetof(struct mdp_superblock_1,data_offset) != 128 ||
disk_csum = sb->sb_csum;
sb->sb_csum = 0;
newcsum = 0;
- for (i=0; size>=4; size -= 4 ) {
+ for (; size>=4; size -= 4 ) {
newcsum += __le32_to_cpu(*isuper);
isuper++;
}
return __cpu_to_le32(csum);
}
+/*
+ * Information related to file descriptor used for aligned reads/writes.
+ * Cache the block size.
+ */
+struct align_fd {
+ int fd;
+ int blk_sz;
+};
+
+static void init_afd(struct align_fd *afd, int fd)
+{
+ afd->fd = fd;
+
+ if (ioctl(afd->fd, BLKSSZGET, &afd->blk_sz) != 0)
+ afd->blk_sz = 512;
+}
+
+static char abuf[4096+4096];
+static int aread(struct align_fd *afd, void *buf, int len)
+{
+ /* aligned read.
+ * On devices with a 4K sector size, we need to read
+ * the full sector and copy relevant bits into
+ * the buffer
+ */
+ int bsize, iosize;
+ char *b;
+ int n;
+
+ bsize = afd->blk_sz;
+
+ if (!bsize || bsize > 4096 || len > 4096) {
+ if (!bsize)
+ fprintf(stderr, "WARNING - aread() called with "
+ "invalid block size\n");
+ return -1;
+ }
+ b = ROUND_UP_PTR((char *)abuf, 4096);
+
+ for (iosize = 0; iosize < len; iosize += bsize)
+ ;
+ n = read(afd->fd, b, iosize);
+ if (n <= 0)
+ return n;
+ lseek(afd->fd, len - n, 1);
+ if (n > len)
+ n = len;
+ memcpy(buf, b, n);
+ return n;
+}
+
+static int awrite(struct align_fd *afd, void *buf, int len)
+{
+ /* aligned write.
+ * On devices with a 4K sector size, we need to write
+ * the full sector. We pre-read if the sector is larger
+ * than the write.
+ * The address must be sector-aligned.
+ */
+ int bsize, iosize;
+ char *b;
+ int n;
+
+ bsize = afd->blk_sz;
+ if (!bsize || bsize > 4096 || len > 4096) {
+ if (!bsize)
+ fprintf(stderr, "WARNING - awrite() called with "
+ "invalid block size\n");
+ return -1;
+ }
+ b = ROUND_UP_PTR((char *)abuf, 4096);
+
+ for (iosize = 0; iosize < len ; iosize += bsize)
+ ;
+
+ if (len != iosize) {
+ n = read(afd->fd, b, iosize);
+ if (n <= 0)
+ return n;
+ lseek(afd->fd, -n, 1);
+ }
+
+ memcpy(b, buf, len);
+ n = write(afd->fd, b, iosize);
+ if (n <= 0)
+ return n;
+ lseek(afd->fd, len - n, 1);
+ return len;
+}
+
#ifndef MDASSEMBLE
static void examine_super1(struct supertype *st, char *homehost)
{
struct mdp_superblock_1 *sb = st->sb;
time_t atime;
- int d;
- int faulty;
+ unsigned int d;
+ int role;
+ int delta_extra = 0;
int i;
char *c;
int l = homehost ? strlen(homehost) : 0;
printf(" Avail Dev Size : %llu%s\n",
(unsigned long long)__le64_to_cpu(sb->data_size),
human_size(__le64_to_cpu(sb->data_size)<<9));
- if (__le32_to_cpu(sb->level) >= 0) {
- int ddsks=0;
+ if (__le32_to_cpu(sb->level) > 0) {
+ int ddsks = 0, ddsks_denom = 1;
switch(__le32_to_cpu(sb->level)) {
case 1: ddsks=1;break;
case 4:
case 6: ddsks = __le32_to_cpu(sb->raid_disks)-2; break;
case 10:
layout = __le32_to_cpu(sb->layout);
- ddsks = __le32_to_cpu(sb->raid_disks)
- / (layout&255) / ((layout>>8)&255);
+ ddsks = __le32_to_cpu(sb->raid_disks);
+ ddsks_denom = (layout&255) * ((layout>>8)&255);
}
- if (ddsks)
+ if (ddsks) {
+ long long asize = __le64_to_cpu(sb->size);
+ asize = (asize << 9) * ddsks / ddsks_denom;
printf(" Array Size : %llu%s\n",
- ddsks*(unsigned long long)__le64_to_cpu(sb->size),
- human_size(ddsks*__le64_to_cpu(sb->size)<<9));
+ asize >> 10, human_size(asize));
+ }
if (sb->size != sb->data_size)
printf(" Used Dev Size : %llu%s\n",
(unsigned long long)__le64_to_cpu(sb->size),
human_size(__le64_to_cpu(sb->reshape_position)<<9));
if (__le32_to_cpu(sb->delta_disks)) {
printf(" Delta Devices : %d", __le32_to_cpu(sb->delta_disks));
- if (__le32_to_cpu(sb->delta_disks))
- printf(" (%d->%d)\n",
- __le32_to_cpu(sb->raid_disks)-__le32_to_cpu(sb->delta_disks),
- __le32_to_cpu(sb->raid_disks));
- else
- printf(" (%d->%d)\n", __le32_to_cpu(sb->raid_disks),
- __le32_to_cpu(sb->raid_disks)+__le32_to_cpu(sb->delta_disks));
+ printf(" (%d->%d)\n",
+ __le32_to_cpu(sb->raid_disks)-__le32_to_cpu(sb->delta_disks),
+ __le32_to_cpu(sb->raid_disks));
+ if ((int)__le32_to_cpu(sb->delta_disks) < 0)
+ delta_extra = -__le32_to_cpu(sb->delta_disks);
}
if (__le32_to_cpu(sb->new_level) != __le32_to_cpu(sb->level)) {
c = map_num(pers, __le32_to_cpu(sb->new_level));
c = map_num(r5layout, __le32_to_cpu(sb->new_layout));
printf(" New Layout : %s\n", c?c:"-unknown-");
}
+ if (__le32_to_cpu(sb->level) == 6) {
+ c = map_num(r6layout, __le32_to_cpu(sb->new_layout));
+ printf(" New Layout : %s\n", c?c:"-unknown-");
+ }
if (__le32_to_cpu(sb->level) == 10) {
printf(" New Layout :");
print_r10_layout(__le32_to_cpu(sb->new_layout));
printf("\n");
}
if (sb->devflags) {
- printf(" Flags :");
+ printf(" Flags :");
if (sb->devflags & WriteMostly1)
printf(" write-mostly");
printf("\n");
c = map_num(r5layout, __le32_to_cpu(sb->layout));
printf(" Layout : %s\n", c?c:"-unknown-");
}
+ if (__le32_to_cpu(sb->level) == 6) {
+ c = map_num(r6layout, __le32_to_cpu(sb->layout));
+ printf(" Layout : %s\n", c?c:"-unknown-");
+ }
if (__le32_to_cpu(sb->level) == 10) {
int lo = __le32_to_cpu(sb->layout);
printf(" Layout :");
default: break;
}
printf("\n");
+#if 0
+ /* This turns out to just be confusing */
printf(" Array Slot : %d (", __le32_to_cpu(sb->dev_number));
for (i= __le32_to_cpu(sb->max_dev); i> 0 ; i--)
if (__le16_to_cpu(sb->dev_roles[i-1]) != 0xffff)
else printf("%d", role);
}
printf(")\n");
+#endif
+ printf(" Device Role : ");
+ d = __le32_to_cpu(sb->dev_number);
+ if (d < __le32_to_cpu(sb->max_dev))
+ role = __le16_to_cpu(sb->dev_roles[d]);
+ else
+ role = 0xFFFF;
+ if (role >= 0xFFFE)
+ printf("spare\n");
+ else
+ printf("Active device %d\n", role);
+
printf(" Array State : ");
- for (d=0; d<__le32_to_cpu(sb->raid_disks); d++) {
+ for (d=0; d<__le32_to_cpu(sb->raid_disks) + delta_extra; d++) {
int cnt = 0;
- int me = 0;
- int i;
+ unsigned int i;
for (i=0; i< __le32_to_cpu(sb->max_dev); i++) {
- int role = __le16_to_cpu(sb->dev_roles[i]);
- if (role == d) {
- if (i == __le32_to_cpu(sb->dev_number))
- me = 1;
+ unsigned int role = __le16_to_cpu(sb->dev_roles[i]);
+ if (role == d)
cnt++;
- }
}
if (cnt > 1) printf("?");
- else if (cnt == 1 && me) printf("U");
- else if (cnt == 1) printf("u");
- else printf ("_");
+ else if (cnt == 1) printf("A");
+ else printf (".");
}
+#if 0
+ /* This is confusing too */
faulty = 0;
for (i=0; i< __le32_to_cpu(sb->max_dev); i++) {
int role = __le16_to_cpu(sb->dev_roles[i]);
faulty++;
}
if (faulty) printf(" %d failed", faulty);
+#endif
+ printf(" ('A' == active, '.' == missing)");
printf("\n");
}
-static void brief_examine_super1(struct supertype *st)
+static void brief_examine_super1(struct supertype *st, int verbose)
{
struct mdp_superblock_1 *sb = st->sb;
int i;
else
nm = NULL;
- printf("ARRAY%s%s level=%s ",
- nm ? " /dev/md/":"", nm,
- c?c:"-unknown-");
+ printf("ARRAY%s%s", nm ? " /dev/md/":"", nm);
+ if (verbose && c)
+ printf(" level=%s", c);
sb_offset = __le64_to_cpu(sb->super_offset);
if (sb_offset <= 4)
- printf("metadata=1.1 ");
+ printf(" metadata=1.1 ");
else if (sb_offset <= 8)
- printf("metadata=1.2 ");
+ printf(" metadata=1.2 ");
else
- printf("metadata=1.0 ");
- printf("num-devices=%d UUID=", __le32_to_cpu(sb->raid_disks));
+ printf(" metadata=1.0 ");
+ if (verbose)
+ printf("num-devices=%d ", __le32_to_cpu(sb->raid_disks));
+ printf("UUID=");
for (i=0; i<16; i++) {
if ((i&3)==0 && i != 0) printf(":");
printf("%02x", sb->set_uuid[i]);
cuuid[i] = super->set_uuid[i];
}
-static void getinfo_super1(struct supertype *st, struct mdinfo *info)
+static void getinfo_super1(struct supertype *st, struct mdinfo *info, char *map)
{
struct mdp_superblock_1 *sb = st->sb;
int working = 0;
- int i;
- int role;
+ unsigned int i;
+ unsigned int role;
+ unsigned int map_disks = info->array.raid_disks;
+ memset(info, 0, sizeof(*info));
info->array.major_version = 1;
info->array.minor_version = st->minor_version;
info->array.patch_version = 0;
info->array.utime = __le64_to_cpu(sb->utime);
info->array.chunk_size = __le32_to_cpu(sb->chunksize)*512;
info->array.state =
- (__le64_to_cpu(sb->resync_offset) >= __le64_to_cpu(sb->size))
+ (__le64_to_cpu(sb->resync_offset) == MaxSector)
? 1 : 0;
info->data_offset = __le64_to_cpu(sb->data_offset);
info->component_size = __le64_to_cpu(sb->size);
+ if (sb->feature_map & __le32_to_cpu(MD_FEATURE_BITMAP_OFFSET))
+ info->bitmap_offset = (long)__le32_to_cpu(sb->bitmap_offset);
info->disk.major = 0;
info->disk.minor = 0;
info->disk.number = __le32_to_cpu(sb->dev_number);
if (__le32_to_cpu(sb->dev_number) >= __le32_to_cpu(sb->max_dev) ||
- __le32_to_cpu(sb->max_dev) > 512)
+ __le32_to_cpu(sb->dev_number) >= MAX_DEVS)
role = 0xfffe;
else
role = __le16_to_cpu(sb->dev_roles[__le32_to_cpu(sb->dev_number)]);
info->disk.raid_disk = -1;
switch(role) {
case 0xFFFF:
- info->disk.state = 2; /* spare: ACTIVE, not sync, not faulty */
+ info->disk.state = 0; /* spare: not active, not sync, not faulty */
break;
case 0xFFFE:
info->disk.state = 1; /* faulty */
info->disk.state = 6; /* active and in sync */
info->disk.raid_disk = role;
}
+ if (sb->devflags & WriteMostly1)
+ info->disk.state |= (1 << MD_DISK_WRITEMOSTLY);
info->events = __le64_to_cpu(sb->events);
sprintf(info->text_version, "1.%d", st->minor_version);
info->safe_mode_delay = 200;
strncpy(info->name, sb->set_name, 32);
info->name[32] = 0;
+ if (sb->feature_map & __le32_to_cpu(MD_FEATURE_RECOVERY_OFFSET))
+ info->recovery_start = __le32_to_cpu(sb->recovery_offset);
+ else
+ info->recovery_start = MaxSector;
+
if (sb->feature_map & __le32_to_cpu(MD_FEATURE_RESHAPE_ACTIVE)) {
info->reshape_active = 1;
info->reshape_progress = __le64_to_cpu(sb->reshape_position);
info->delta_disks = __le32_to_cpu(sb->delta_disks);
info->new_layout = __le32_to_cpu(sb->new_layout);
info->new_chunk = __le32_to_cpu(sb->new_chunk)<<9;
+ if (info->delta_disks < 0)
+ info->array.raid_disks -= info->delta_disks;
} else
info->reshape_active = 0;
- for (i=0; i< __le32_to_cpu(sb->max_dev); i++) {
+ info->recovery_blocked = info->reshape_active;
+
+ if (map)
+ for (i=0; i<map_disks; i++)
+ map[i] = 0;
+ for (i = 0; i < __le32_to_cpu(sb->max_dev); i++) {
role = __le16_to_cpu(sb->dev_roles[i]);
- if (/*role == 0xFFFF || */role < info->array.raid_disks)
+ if (/*role == 0xFFFF || */role < (unsigned) info->array.raid_disks) {
working++;
+ if (map && role < map_disks)
+ map[role] = 1;
+ }
}
info->array.working_disks = working;
}
+static struct mdinfo *container_content1(struct supertype *st, char *subarray)
+{
+ struct mdinfo *info;
+
+ if (subarray)
+ return NULL;
+
+ info = malloc(sizeof(*info));
+ getinfo_super1(st, info, NULL);
+ return info;
+}
+
static int update_super1(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;
struct mdp_superblock_1 *sb = st->sb;
if (sb->events != __cpu_to_le64(info->events))
rv = 1;
sb->events = __cpu_to_le64(info->events);
- }
- if (strcmp(update, "force-array")==0) {
+ } else if (strcmp(update, "force-array")==0) {
/* Degraded array and 'force' requests to
* maybe need to mark it 'clean'.
*/
switch(__le32_to_cpu(sb->level)) {
case 5: case 4: case 6:
/* need to force clean */
- if (sb->resync_offset != ~0ULL)
+ if (sb->resync_offset != MaxSector)
rv = 1;
- sb->resync_offset = ~0ULL;
+ sb->resync_offset = MaxSector;
}
- }
- if (strcmp(update, "assemble")==0) {
+ } else if (strcmp(update, "assemble")==0) {
int d = info->disk.number;
int want;
if (info->disk.state == 6)
- want = __cpu_to_le32(info->disk.raid_disk);
+ want = info->disk.raid_disk;
else
want = 0xFFFF;
- if (sb->dev_roles[d] != want) {
- sb->dev_roles[d] = want;
+ if (sb->dev_roles[d] != __cpu_to_le16(want)) {
+ sb->dev_roles[d] = __cpu_to_le16(want);
rv = 1;
}
- }
- if (strcmp(update, "linear-grow-new") == 0) {
- int i;
+ if (info->reshape_active &&
+ sb->feature_map & __le32_to_cpu(MD_FEATURE_RESHAPE_ACTIVE) &&
+ info->delta_disks >= 0 &&
+ info->reshape_progress < __le64_to_cpu(sb->reshape_position)) {
+ sb->reshape_position = __cpu_to_le64(info->reshape_progress);
+ rv = 1;
+ }
+ if (info->reshape_active &&
+ sb->feature_map & __le32_to_cpu(MD_FEATURE_RESHAPE_ACTIVE) &&
+ info->delta_disks < 0 &&
+ info->reshape_progress > __le64_to_cpu(sb->reshape_position)) {
+ sb->reshape_position = __cpu_to_le64(info->reshape_progress);
+ rv = 1;
+ }
+ } else if (strcmp(update, "linear-grow-new") == 0) {
+ unsigned int i;
int rfd, fd;
- int max = __le32_to_cpu(sb->max_dev);
+ unsigned int max = __le32_to_cpu(sb->max_dev);
for (i=0 ; i < max ; i++)
if (__le16_to_cpu(sb->dev_roles[i]) >= 0xfffe)
if ((rfd = open("/dev/urandom", O_RDONLY)) < 0 ||
read(rfd, sb->device_uuid, 16) != 16) {
- *(__u32*)(sb->device_uuid) = random();
- *(__u32*)(sb->device_uuid+4) = random();
- *(__u32*)(sb->device_uuid+8) = random();
- *(__u32*)(sb->device_uuid+12) = random();
+ __u32 r[4] = {random(), random(), random(), random()};
+ memcpy(sb->device_uuid, r, 16);
}
+ if (rfd >= 0)
+ close(rfd);
sb->dev_roles[i] =
__cpu_to_le16(info->disk.raid_disk);
ds - __le64_to_cpu(sb->data_offset));
}
}
- }
- if (strcmp(update, "linear-grow-update") == 0) {
+ } else if (strcmp(update, "linear-grow-update") == 0) {
sb->raid_disks = __cpu_to_le32(info->array.raid_disks);
sb->dev_roles[info->disk.number] =
__cpu_to_le16(info->disk.raid_disk);
- }
- if (strcmp(update, "resync") == 0) {
+ } else if (strcmp(update, "resync") == 0) {
/* make sure resync happens */
sb->resync_offset = 0ULL;
- }
- if (strcmp(update, "uuid") == 0) {
+ } else if (strcmp(update, "uuid") == 0) {
copy_uuid(sb->set_uuid, info->uuid, super1.swapuuid);
if (__le32_to_cpu(sb->feature_map)&MD_FEATURE_BITMAP_OFFSET) {
struct bitmap_super_s *bm;
- bm = (struct bitmap_super_s*)(st->sb+1024);
+ bm = (struct bitmap_super_s*)(st->sb+MAX_SB_SIZE);
memcpy(bm->uuid, sb->set_uuid, 16);
}
- }
- if (strcmp(update, "homehost") == 0 &&
- homehost) {
+ } else if (strcmp(update, "no-bitmap") == 0) {
+ sb->feature_map &= ~__cpu_to_le32(MD_FEATURE_BITMAP_OFFSET);
+ } else if (strcmp(update, "homehost") == 0 &&
+ homehost) {
char *c;
update = "name";
c = strchr(sb->set_name, ':');
else
strncpy(info->name, sb->set_name, 32);
info->name[32] = 0;
- }
- if (strcmp(update, "name") == 0) {
+ } else if (strcmp(update, "name") == 0) {
if (info->name[0] == 0)
sprintf(info->name, "%d", info->array.md_minor);
memset(sb->set_name, 0, sizeof(sb->set_name));
strcat(sb->set_name, info->name);
} else
strcpy(sb->set_name, info->name);
- }
- if (strcmp(update, "devicesize") == 0 &&
+ } else if (strcmp(update, "devicesize") == 0 &&
__le64_to_cpu(sb->super_offset) <
__le64_to_cpu(sb->data_offset)) {
/* set data_size to device size less data_offset */
struct misc_dev_info *misc = (struct misc_dev_info*)
- (st->sb + 1024 + 512);
+ (st->sb + MAX_SB_SIZE + BM_SUPER_SIZE);
printf("Size was %llu\n", (unsigned long long)
__le64_to_cpu(sb->data_size));
sb->data_size = __cpu_to_le64(
misc->device_size - __le64_to_cpu(sb->data_offset));
printf("Size is %llu\n", (unsigned long long)
__le64_to_cpu(sb->data_size));
- }
- if (strcmp(update, "_reshape_progress")==0)
+ } else if (strcmp(update, "_reshape_progress")==0)
sb->reshape_position = __cpu_to_le64(info->reshape_progress);
+ else if (strcmp(update, "writemostly")==0)
+ sb->devflags |= WriteMostly1;
+ else if (strcmp(update, "readwrite")==0)
+ sb->devflags &= ~WriteMostly1;
+ else
+ rv = -1;
sb->sb_csum = calc_sb_1_csum(sb);
return rv;
int spares;
int rfd;
char defname[10];
+ int sbsize;
- if (posix_memalign((void**)&sb, 512, (1024 + 512 +
- sizeof(struct misc_dev_info))) != 0) {
+ if (posix_memalign((void**)&sb, 4096, SUPER1_SIZE) != 0) {
fprintf(stderr, Name
": %s could not allocate superblock\n", __func__);
return 0;
}
- memset(sb, 0, 1024);
+ memset(sb, 0, SUPER1_SIZE);
st->sb = sb;
if (info == NULL) {
}
spares = info->working_disks - info->active_disks;
- if (info->raid_disks + spares > 384) {
+ if (info->raid_disks + spares > MAX_DEVS) {
fprintf(stderr, Name ": too many devices requested: %d+%d > %d\n",
- info->raid_disks , spares, 384);
+ info->raid_disks , spares, MAX_DEVS);
return 0;
}
else {
if ((rfd = open("/dev/urandom", O_RDONLY)) < 0 ||
read(rfd, sb->set_uuid, 16) != 16) {
- *(__u32*)(sb->set_uuid) = random();
- *(__u32*)(sb->set_uuid+4) = random();
- *(__u32*)(sb->set_uuid+8) = random();
- *(__u32*)(sb->set_uuid+12) = random();
+ __u32 r[4] = {random(), random(), random(), random()};
+ memcpy(sb->set_uuid, r, 16);
}
if (rfd >= 0) close(rfd);
}
sprintf(defname, "%d", info->md_minor);
name = defname;
}
- memset(sb->set_name, 0, 32);
if (homehost &&
strchr(name, ':')== NULL &&
strlen(homehost)+1+strlen(name) < 32) {
sb->utime = sb->ctime;
sb->events = __cpu_to_le64(1);
if (info->state & (1<<MD_SB_CLEAN))
- sb->resync_offset = ~0ULL;
+ sb->resync_offset = MaxSector;
else
sb->resync_offset = 0;
- sb->max_dev = __cpu_to_le32((1024- sizeof(struct mdp_superblock_1))/
- sizeof(sb->dev_roles[0]));
- memset(sb->pad3, 0, sizeof(sb->pad3));
+ sbsize = sizeof(struct mdp_superblock_1) + 2 * (info->raid_disks + spares);
+ sbsize = ROUND_UP(sbsize, 512);
+ sb->max_dev = __cpu_to_le32((sbsize - sizeof(struct mdp_superblock_1)) / 2);
- memset(sb->dev_roles, 0xff, 1024 - sizeof(struct mdp_superblock_1));
+ memset(sb->dev_roles, 0xff, MAX_SB_SIZE - sizeof(struct mdp_superblock_1));
return 1;
}
else
*rp = 0xfffe;
+ if (dk->number >= (int)__le32_to_cpu(sb->max_dev) &&
+ __le32_to_cpu(sb->max_dev) < MAX_DEVS)
+ sb->max_dev = __cpu_to_le32(dk->number+1);
+
sb->dev_number = __cpu_to_le32(dk->number);
+ sb->devflags = 0; /* don't copy another disks flags */
sb->sb_csum = calc_sb_1_csum(sb);
dip = (struct devinfo **)&st->info;
{
struct mdp_superblock_1 *sb = st->sb;
unsigned long long sb_offset;
+ struct align_fd afd;
int sbsize;
unsigned long long dsize;
if (dsize < 24)
return 2;
+ init_afd(&afd, fd);
+
/*
* Calculate the position of the superblock.
* It is always aligned to a 4K boundary and
if (lseek64(fd, sb_offset << 9, 0)< 0LL)
return 3;
- sbsize = sizeof(*sb) + 2 * __le32_to_cpu(sb->max_dev);
- sbsize = (sbsize+511)&(~511UL);
+ sbsize = ROUND_UP(sizeof(*sb) + 2 * __le32_to_cpu(sb->max_dev), 512);
- if (write(fd, sb, sbsize) != sbsize)
+ if (awrite(&afd, sb, sbsize) != sbsize)
return 4;
if (sb->feature_map & __cpu_to_le32(MD_FEATURE_BITMAP_OFFSET)) {
struct bitmap_super_s *bm = (struct bitmap_super_s*)
- (((char*)sb)+1024);
+ (((char*)sb)+MAX_SB_SIZE);
if (__le32_to_cpu(bm->magic) == BITMAP_MAGIC) {
locate_bitmap1(st, fd);
- if (write(fd, bm, ROUND_UP(sizeof(*bm),512)) !=
- ROUND_UP(sizeof(*bm),512))
- return 5;
+ if (awrite(&afd, bm, sizeof(*bm)) != sizeof(*bm))
+ return 5;
}
}
fsync(fd);
{
/* if the device is bigger than 8Gig, save 64k for bitmap usage,
* if bigger than 200Gig, save 128k
+ * NOTE: result must be multiple of 4K else bad things happen
+ * on 4K-sector devices.
*/
if (devsize < 64*2) return 0;
if (devsize - 64*2 >= 200*1024*1024*2)
return 4*2;
}
+static void free_super1(struct supertype *st);
+
#ifndef MDASSEMBLE
static int write_init_super1(struct supertype *st)
{
struct mdp_superblock_1 *sb = st->sb;
- struct supertype refst;
+ struct supertype *refst;
int rfd;
int rv = 0;
- int bm_space;
+ unsigned long long bm_space;
+ unsigned long long reserved;
struct devinfo *di;
unsigned long long dsize, array_size;
- long long sb_offset;
+ unsigned long long sb_offset, headroom;
for (di = st->info; di && ! rv ; di = di->next) {
if (di->disk.state == 1)
if (di->fd < 0)
continue;
- Kill(di->devname, 0, 1, 1);
- Kill(di->devname, 0, 1, 1);
+ while (Kill(di->devname, NULL, 0, 1, 1) == 0)
+ ;
sb->dev_number = __cpu_to_le32(di->disk.number);
if (di->disk.state & (1<<MD_DISK_WRITEMOSTLY))
- sb->devflags |= __cpu_to_le32(WriteMostly1);
+ sb->devflags |= WriteMostly1;
+ else
+ sb->devflags &= ~WriteMostly1;
if ((rfd = open("/dev/urandom", O_RDONLY)) < 0 ||
read(rfd, sb->device_uuid, 16) != 16) {
- *(__u32*)(sb->device_uuid) = random();
- *(__u32*)(sb->device_uuid+4) = random();
- *(__u32*)(sb->device_uuid+8) = random();
- *(__u32*)(sb->device_uuid+12) = random();
+ __u32 r[4] = {random(), random(), random(), random()};
+ memcpy(sb->device_uuid, r, 16);
}
- if (rfd >= 0) close(rfd);
+ if (rfd >= 0)
+ close(rfd);
+
sb->events = 0;
- refst =*st;
- refst.sb = NULL;
- if (load_super1(&refst, di->fd, NULL)==0) {
- struct mdp_superblock_1 *refsb = refst.sb;
+ refst = dup_super(st);
+ if (load_super1(refst, di->fd, NULL)==0) {
+ struct mdp_superblock_1 *refsb = refst->sb;
memcpy(sb->device_uuid, refsb->device_uuid, 16);
if (memcmp(sb->set_uuid, refsb->set_uuid, 16)==0) {
if (get_linux_version() >= 2006018)
sb->dev_number = refsb->dev_number;
}
- free(refsb);
+ free_super1(refst);
}
+ free(refst);
- if (!get_dev_size(di->fd, NULL, &dsize))
- return 1;
+ if (!get_dev_size(di->fd, NULL, &dsize)) {
+ rv = 1;
+ goto error_out;
+ }
dsize >>= 9;
if (dsize < 24) {
close(di->fd);
- return 2;
+ rv = 2;
+ goto error_out;
}
/* work out how much space we left for a bitmap */
bm_space = choose_bm_space(array_size);
+ /* We try to leave 0.1% at the start for reshape
+ * operations, but limit this to 128Meg (0.1% of 10Gig)
+ * which is plenty for efficient reshapes
+ */
+ headroom = 128 * 1024 * 2;
+ while (headroom << 10 > array_size)
+ headroom >>= 1;
+
switch(st->minor_version) {
case 0:
sb_offset = dsize;
sb_offset &= ~(4*2-1);
sb->super_offset = __cpu_to_le64(sb_offset);
sb->data_offset = __cpu_to_le64(0);
- if (sb_offset - bm_space < array_size)
- bm_space = sb_offset - array_size;
+ if (sb_offset < array_size + bm_space)
+ bm_space = sb_offset - array_size;
sb->data_size = __cpu_to_le64(sb_offset - bm_space);
break;
case 1:
sb->super_offset = __cpu_to_le64(0);
- if (4*2 + bm_space + __le64_to_cpu(sb->size) > dsize)
- bm_space = dsize - __le64_to_cpu(sb->size) -4*2;
- sb->data_offset = __cpu_to_le64(bm_space + 4*2);
- sb->data_size = __cpu_to_le64(dsize - bm_space - 4*2);
+ reserved = bm_space + 4*2;
+ /* Try for multiple of 1Meg so it is nicely aligned */
+ #define ONE_MEG (2*1024)
+ reserved = ((reserved + ONE_MEG-1)/ONE_MEG) * ONE_MEG;
+ if (reserved + __le64_to_cpu(sb->size) > dsize)
+ reserved = dsize - __le64_to_cpu(sb->size);
+ /* force 4K alignment */
+ reserved &= ~7ULL;
+
+ if (reserved < headroom)
+ reserved = headroom;
+
+ sb->data_offset = __cpu_to_le64(reserved);
+ sb->data_size = __cpu_to_le64(dsize - reserved);
break;
case 2:
sb_offset = 4*2;
> dsize)
bm_space = dsize - __le64_to_cpu(sb->size)
- 4*2 - 4*2;
- sb->data_offset = __cpu_to_le64(4*2 + 4*2 + bm_space);
- sb->data_size = __cpu_to_le64(dsize - 4*2 - 4*2
- - bm_space );
+
+ reserved = bm_space + 4*2 + 4*2;
+ /* Try for multiple of 1Meg so it is nicely aligned */
+ #define ONE_MEG (2*1024)
+ reserved = ((reserved + ONE_MEG-1)/ONE_MEG) * ONE_MEG;
+ if (reserved + __le64_to_cpu(sb->size) > dsize)
+ reserved = dsize - __le64_to_cpu(sb->size);
+ /* force 4K alignment */
+ reserved &= ~7ULL;
+
+ if (reserved < headroom)
+ reserved = headroom;
+
+ sb->data_offset = __cpu_to_le64(reserved);
+ sb->data_size = __cpu_to_le64(dsize - reserved);
break;
default:
- return -EINVAL;
+ fprintf(stderr, Name ": Failed to write invalid "
+ "metadata format 1.%i to %s\n",
+ st->minor_version, di->devname);
+ rv = -EINVAL;
+ goto out;
}
sb->sb_csum = calc_sb_1_csum(sb);
rv = store_super1(st, di->fd);
- if (rv)
- fprintf(stderr,
- Name ": failed to write superblock to %s\n",
- di->devname);
-
if (rv == 0 && (__le32_to_cpu(sb->feature_map) & 1))
rv = st->ss->write_bitmap(st, di->fd);
close(di->fd);
di->fd = -1;
}
+error_out:
+ if (rv)
+ fprintf(stderr, Name ": Failed to write metadata to %s\n",
+ di->devname);
+out:
return rv;
}
#endif
return 1;
if (!first) {
- if (posix_memalign((void**)&first, 512,
- 1024 + 512 +
- sizeof(struct misc_dev_info)) != 0) {
+ if (posix_memalign((void**)&first, 4096, SUPER1_SIZE) != 0) {
fprintf(stderr, Name
": %s could not allocate superblock\n", __func__);
return 1;
}
- memcpy(first, second, 1024 + 512 +
- sizeof(struct misc_dev_info));
+ memcpy(first, second, SUPER1_SIZE);
st->sb = first;
return 0;
}
return 0;
}
-static void free_super1(struct supertype *st);
-
static int load_super1(struct supertype *st, int fd, char *devname)
{
unsigned long long dsize;
int uuid[4];
struct bitmap_super_s *bsb;
struct misc_dev_info *misc;
+ struct align_fd afd;
free_super1(st);
- if (st->subarray[0])
- return 1;
+ init_afd(&afd, fd);
if (st->ss == NULL || st->minor_version == -1) {
int bestvers = -1;
int rv;
tst.minor_version = bestvers;
tst.ss = &super1;
- tst.max_devs = 384;
+ tst.max_devs = MAX_DEVS;
rv = load_super1(&tst, fd, devname);
if (rv == 0)
*st = tst;
return 1;
}
- if (posix_memalign((void**)&super, 512,
- 1024 + 512 +
- sizeof(struct misc_dev_info)) != 0) {
+ if (posix_memalign((void**)&super, 4096, SUPER1_SIZE) != 0) {
fprintf(stderr, Name ": %s could not allocate superblock\n",
__func__);
return 1;
}
- if (read(fd, super, 1024) != 1024) {
+ if (aread(&afd, super, MAX_SB_SIZE) != MAX_SB_SIZE) {
if (devname)
fprintf(stderr, Name ": Cannot read superblock on %s\n",
devname);
}
st->sb = super;
- bsb = (struct bitmap_super_s *)(((char*)super)+1024);
+ bsb = (struct bitmap_super_s *)(((char*)super)+MAX_SB_SIZE);
- misc = (struct misc_dev_info*) (((char*)super)+1024+512);
+ misc = (struct misc_dev_info*) (((char*)super)+MAX_SB_SIZE+BM_SUPER_SIZE);
misc->device_size = dsize;
/* Now check on the bitmap superblock */
* should get that written out.
*/
locate_bitmap1(st, fd);
- if (read(fd, ((char*)super)+1024, 512)
- != 512)
+ if (aread(&afd, bsb, 512) != 512)
goto no_bitmap;
uuid_from_super1(st, uuid);
return 0;
no_bitmap:
- super->feature_map = __cpu_to_le32(__le32_to_cpu(super->feature_map) & ~1);
+ super->feature_map = __cpu_to_le32(__le32_to_cpu(super->feature_map)
+ & ~MD_FEATURE_BITMAP_OFFSET);
return 0;
}
static struct supertype *match_metadata_desc1(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 = &super1;
- st->max_devs = 384;
+ st->max_devs = MAX_DEVS;
st->sb = NULL;
/* leading zeros can be safely ignored. --detail generates them. */
while (*arg == '0')
return st;
}
if (strcmp(arg, "1.1") == 0 ||
- strcmp(arg, "1.01") == 0) {
+ strcmp(arg, "1.01") == 0
+ ) {
st->minor_version = 1;
return st;
}
if (strcmp(arg, "1.2") == 0 ||
+#ifndef DEFAULT_OLD_METADATA /* ifdef in super0.c */
+ strcmp(arg, "default") == 0 ||
+#endif /* DEFAULT_OLD_METADATA */
strcmp(arg, "1.02") == 0) {
st->minor_version = 2;
return st;
else if (__le32_to_cpu(super->feature_map)&MD_FEATURE_BITMAP_OFFSET) {
/* hot-add. allow for actual size of bitmap */
struct bitmap_super_s *bsb;
- bsb = (struct bitmap_super_s *)(((char*)super)+1024);
+ bsb = (struct bitmap_super_s *)(((char*)super)+MAX_SB_SIZE);
devsize -= bitmap_sectors(bsb);
}
#endif
+ if (st->minor_version < 0)
+ /* not specified, so time to set default */
+ st->minor_version = 2;
+ if (super == NULL && st->minor_version > 0) {
+ /* haven't committed to a size yet, so allow some
+ * slack for space for reshape.
+ * Limit slack to 128M, but aim for about 0.1%
+ */
+ unsigned long long headroom = 128*1024*2;
+ while ((headroom << 10) > devsize)
+ headroom >>= 1;
+ devsize -= headroom;
+ }
switch(st->minor_version) {
- case -1: /* no specified. Now time to set default */
- st->minor_version = 0;
- /* FALL THROUGH */
case 0:
/* at end */
return ((devsize - 8*2 ) & ~(4*2-1));
int may_change, int major)
{
/*
- * If not may_change, then this is a 'Grow', and the bitmap
- * must fit after the superblock.
- * If may_change, then this is create, and we can put the bitmap
- * before the superblock if we like, or may move the start.
- * If !may_change, the bitmap MUST live at offset of 1K, until
- * we get a sysfs interface.
+ * If not may_change, then this is a 'Grow' without sysfs support for
+ * bitmaps, and the bitmap must fit after the superblock at 1K offset.
+ * If may_change, then this is create or a Grow with sysfs syupport,
+ * and we can put the bitmap wherever we like.
*
* size is in sectors, chunk is in bytes !!!
*/
unsigned long long max_bits;
unsigned long long min_chunk;
long offset;
- int chunk = *chunkp;
+ unsigned long long chunk = *chunkp;
int room = 0;
+ int creating = 0;
struct mdp_superblock_1 *sb = st->sb;
- bitmap_super_t *bms = (bitmap_super_t*)(((char*)sb) + 1024);
+ bitmap_super_t *bms = (bitmap_super_t*)(((char*)sb) + MAX_SB_SIZE);
+ int uuid[4];
+ if (__le64_to_cpu(sb->data_size) == 0)
+ /* Must be creating the array, else data_size would be non-zero */
+ creating = 1;
switch(st->minor_version) {
case 0:
- /* either 3K after the superblock, or some amount of space
- * before.
+ /* either 3K after the superblock (when hot-add),
+ * or some amount of space before.
*/
- if (may_change) {
+ if (creating) {
/* We are creating array, so we *know* how much room has
* been left.
*/
offset = 0;
room = choose_bm_space(__le64_to_cpu(sb->size));
- if (room == 4*2) {
- /* make it 3K after the superblock */
- room = 3*2;
- offset = 2;
- }
} else {
room = __le64_to_cpu(sb->super_offset)
- __le64_to_cpu(sb->data_offset)
- __le64_to_cpu(sb->data_size);
- /* remove '1 ||' when we can set offset via sysfs */
- if (1 || (room < 3*2 &&
+
+ if (!may_change || (room < 3*2 &&
__le32_to_cpu(sb->max_dev) <= 384)) {
room = 3*2;
offset = 1*2;
break;
case 1:
case 2: /* between superblock and data */
- if (may_change) {
+ if (creating) {
offset = 4*2;
room = choose_bm_space(__le64_to_cpu(sb->size));
} else {
room = __le64_to_cpu(sb->data_offset)
- __le64_to_cpu(sb->super_offset);
- if (1 || __le32_to_cpu(sb->max_dev) <= 384) {
- room -= 2;
+ if (!may_change) {
+ room -= 2; /* Leave 1K for superblock */
offset = 2;
} else {
- room -= 4*2;
+ room -= 4*2; /* leave 4K for superblock */
offset = 4*2;
}
}
min_chunk *= 2;
bits = (bits+1)/2;
}
- if (chunk == UnSet)
+ if (chunk == UnSet) {
+ /* For practical purpose, 64Meg is a good
+ * default chunk size for internal bitmaps.
+ */
chunk = min_chunk;
- else if (chunk < min_chunk)
+ if (chunk < 64*1024*1024)
+ chunk = 64*1024*1024;
+ } else if (chunk < min_chunk)
return 0; /* chunk size too small */
if (chunk == 0) /* rounding problem */
return 0;
if (offset == 0) {
+ /* start bitmap on a 4K boundary with enough space for
+ * the bitmap
+ */
bits = (size*512) / chunk + 1;
- room = ((bits+7)/8 + sizeof(bitmap_super_t) +511)/512;
+ room = ((bits+7)/8 + sizeof(bitmap_super_t) +4095)/4096;
+ room *= 8; /* convert 4K blocks to sectors */
offset = -room;
}
- sb->bitmap_offset = __cpu_to_le32(offset);
+ sb->bitmap_offset = (long)__cpu_to_le32(offset);
- sb->feature_map = __cpu_to_le32(__le32_to_cpu(sb->feature_map) | 1);
+ sb->feature_map = __cpu_to_le32(__le32_to_cpu(sb->feature_map)
+ | MD_FEATURE_BITMAP_OFFSET);
memset(bms, 0, sizeof(*bms));
bms->magic = __cpu_to_le32(BITMAP_MAGIC);
bms->version = __cpu_to_le32(major);
- uuid_from_super1(st, (int*)bms->uuid);
+ uuid_from_super1(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);
return 1;
}
-
static void locate_bitmap1(struct supertype *st, int fd)
{
unsigned long long offset;
static int write_bitmap1(struct supertype *st, int fd)
{
struct mdp_superblock_1 *sb = st->sb;
- bitmap_super_t *bms = (bitmap_super_t*)(((char*)sb)+1024);
+ bitmap_super_t *bms = (bitmap_super_t*)(((char*)sb)+MAX_SB_SIZE);
int rv = 0;
-
+ void *buf;
int towrite, n;
- char abuf[4096+512];
- char *buf = (char*)(((long)(abuf+512))&~511UL);
+ struct align_fd afd;
+
+ init_afd(&afd, fd);
locate_bitmap1(st, fd);
+ if (posix_memalign(&buf, 4096, 4096))
+ return -ENOMEM;
+
memset(buf, 0xff, 4096);
- memcpy(buf, ((char*)sb)+1024, sizeof(bitmap_super_t));
+ memcpy(buf, (char *)bms, sizeof(bitmap_super_t));
towrite = __le64_to_cpu(bms->sync_size) / (__le32_to_cpu(bms->chunksize)>>9);
towrite = (towrite+7) >> 3; /* bits to bytes */
n = towrite;
if (n > 4096)
n = 4096;
- n = write(fd, buf, n);
+ n = awrite(&afd, buf, n);
if (n > 0)
towrite -= n;
else
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_geometry1(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;
- if (level == LEVEL_CONTAINER)
+ if (level == LEVEL_CONTAINER) {
+ if (verbose)
+ fprintf(stderr, Name ": 1.x metadata does not support containers\n");
return 0;
+ }
+ if (chunk && *chunk == UnSet)
+ *chunk = DEFAULT_CHUNK;
+
if (!subdev)
return 1;
.match_home = match_home1,
.uuid_from_super = uuid_from_super1,
.getinfo_super = getinfo_super1,
+ .container_content = container_content1,
.update_super = update_super1,
.init_super = init_super1,
.store_super = store_super1,
#else
.swapuuid = 1,
#endif
+ .name = "1.x",
};
projects / thirdparty / mdadm.git / blobdiff