/*
* mdadm - manage Linux "md" devices aka RAID arrays.
*
- * Copyright (C) 2006-2007 Neil Brown <neilb@suse.de>
+ * Copyright (C) 2006-2009 Neil Brown <neilb@suse.de>
*
*
* This program is free software; you can redistribute it and/or modify
unsigned long long size; /* sectors */
int pdnum; /* index in ->phys */
struct spare_assign *spare;
+ void *mdupdate; /* hold metadata update */
+
+ /* These fields used by auto-layout */
+ int raiddisk; /* slot to fill in autolayout */
+ __u64 esize;
};
};
struct disk_data disk;
- void *mdupdate; /* hold metadata update */
struct vcl *vlist[0]; /* max_part in size */
} *dlist, *add_list;
};
struct stat stb;
char *conf;
int i;
+ int confsec;
int vnum;
int max_virt_disks = __be16_to_cpu(super->active->max_vd_entries);
unsigned long long dsize;
0);
vnum = 0;
- for (i = 0;
- i < __be32_to_cpu(super->active->config_section_length);
- i += super->conf_rec_len) {
+ for (confsec = 0;
+ confsec < __be32_to_cpu(super->active->config_section_length);
+ confsec += super->conf_rec_len) {
struct vd_config *vd =
- (struct vd_config *)((char*)conf + i*512);
+ (struct vd_config *)((char*)conf + confsec*512);
struct vcl *vcl;
if (vd->magic == DDF_SPARE_ASSIGN_MAGIC) {
static int load_super_ddf_all(struct supertype *st, int fd,
void **sbp, char *devname, int keep_fd);
#endif
+
+static void free_super_ddf(struct supertype *st);
+
static int load_super_ddf(struct supertype *st, int fd,
char *devname)
{
/* 32M is a lower bound */
if (dsize <= 32*1024*1024) {
- if (devname) {
+ if (devname)
fprintf(stderr,
Name ": %s is too small for ddf: "
"size is %llu sectors.\n",
devname, dsize>>9);
- return 1;
- }
+ return 1;
}
if (dsize & 511) {
- if (devname) {
+ if (devname)
fprintf(stderr,
Name ": %s is an odd size for ddf: "
"size is %llu bytes.\n",
devname, dsize);
- return 1;
- }
+ return 1;
}
+ free_super_ddf(st);
+
if (posix_memalign((void**)&super, 512, sizeof(*super))!= 0) {
fprintf(stderr, Name ": malloc of %zu failed.\n",
sizeof(*super));
return rv;
}
+ if (st->subarray[0]) {
+ struct vcl *v;
+
+ for (v = super->conflist; v; v = v->next)
+ if (v->vcnum == atoi(st->subarray))
+ super->currentconf = v;
+ if (!super->currentconf) {
+ free(super);
+ return 1;
+ }
+ }
+
/* Should possibly check the sections .... */
st->sb = super;
return map[i].num2;
}
+static int all_ff(char *guid)
+{
+ int i;
+ for (i = 0; i < DDF_GUID_LEN; i++)
+ if (guid[i] != (char)0xff)
+ return 0;
+ return 1;
+}
+
#ifndef MDASSEMBLE
static void print_guid(char *guid, int tstamp)
{
printf("%02X", guid[i]&255);
}
- printf(" (");
+ printf("\n (");
while (l && guid[l-1] == ' ')
l--;
for (i=0 ; i<l ; i++) {
struct vcl *vcl;
for (vcl = sb->conflist ; vcl ; vcl = vcl->next) {
+ int i;
struct vd_config *vc = &vcl->conf;
if (calc_crc(vc, crl*512) != vc->crc)
continue;
/* Ok, we know about this VD, let's give more details */
- printf(" Raid Devices[%d] : %d\n", n,
+ printf(" Raid Devices[%d] : %d (", n,
__be16_to_cpu(vc->prim_elmnt_count));
+ for (i=0; i<__be16_to_cpu(vc->prim_elmnt_count); i++) {
+ int j;
+ int cnt = __be16_to_cpu(sb->phys->used_pdes);
+ for (j=0; j<cnt; j++)
+ if (vc->phys_refnum[i] == sb->phys->entries[j].refnum)
+ break;
+ if (i) printf(" ");
+ if (j < cnt)
+ printf("%d", j);
+ else
+ printf("--");
+ }
+ printf(")\n");
+ if (vc->chunk_shift != 255)
printf(" Chunk Size[%d] : %d sectors\n", n,
1 << vc->chunk_shift);
printf(" Raid Level[%d] : %s\n", n,
map_num(ddf_sec_level, vc->srl) ?: "-unknown-");
}
printf(" Device Size[%d] : %llu\n", n,
- __be64_to_cpu(vc->blocks)/2);
+ (unsigned long long)__be64_to_cpu(vc->blocks)/2);
printf(" Array Size[%d] : %llu\n", n,
- __be64_to_cpu(vc->array_blocks)/2);
+ (unsigned long long)__be64_to_cpu(vc->array_blocks)/2);
}
}
for (i=0; i<cnt; i++) {
struct virtual_entry *ve = &sb->virt->entries[i];
+ printf("\n");
printf(" VD GUID[%d] : ", i); print_guid(ve->guid, 1);
printf("\n");
printf(" unit[%d] : %d\n", i, __be16_to_cpu(ve->unit));
int i;
struct dl *dl;
printf(" Physical Disks : %d\n", cnt);
+ printf(" Number RefNo Size Device Type/State\n");
for (i=0 ; i<cnt ; i++) {
struct phys_disk_entry *pd = &sb->phys->entries[i];
int type = __be16_to_cpu(pd->type);
int state = __be16_to_cpu(pd->state);
- printf(" PD GUID[%d] : ", i); print_guid(pd->guid, 0);
- printf("\n");
- printf(" ref[%d] : %08x\n", i,
+ //printf(" PD GUID[%d] : ", i); print_guid(pd->guid, 0);
+ //printf("\n");
+ printf(" %3d %08x ", i,
__be32_to_cpu(pd->refnum));
- printf(" mode[%d] : %s%s%s%s%s\n", i,
+ printf("%8lluK ",
+ (unsigned long long)__be64_to_cpu(pd->config_size)>>1);
+ for (dl = sb->dlist; dl ; dl = dl->next) {
+ if (dl->disk.refnum == pd->refnum) {
+ char *dv = map_dev(dl->major, dl->minor, 0);
+ if (dv) {
+ printf("%-15s", dv);
+ break;
+ }
+ }
+ }
+ if (!dl)
+ printf("%15s","");
+ printf(" %s%s%s%s%s",
(type&2) ? "active":"",
- (type&4) ? "Global Spare":"",
+ (type&4) ? "Global-Spare":"",
(type&8) ? "spare" : "",
(type&16)? ", foreign" : "",
(type&32)? "pass-through" : "");
- printf(" state[%d] : %s%s%s%s%s%s%s\n", i,
+ printf("/%s%s%s%s%s%s%s",
(state&1)? "Online": "Offline",
(state&2)? ", Failed": "",
(state&4)? ", Rebuilding": "",
(state&8)? ", in-transition": "",
- (state&16)? ", SMART errors": "",
- (state&32)? ", Unrecovered Read Errors": "",
+ (state&16)? ", SMART-errors": "",
+ (state&32)? ", Unrecovered-Read-Errors": "",
(state&64)? ", Missing" : "");
- printf(" Avail Size[%d] : %llu K\n", i,
- __be64_to_cpu(pd->config_size)>>1);
- for (dl = sb->dlist; dl ; dl = dl->next) {
- if (dl->disk.refnum == pd->refnum) {
- char *dv = map_dev(dl->major, dl->minor, 0);
- if (dv)
- printf(" Device[%d] : %s\n",
- i, dv);
- }
- }
printf("\n");
}
}
static void getinfo_super_ddf(struct supertype *st, struct mdinfo *info);
+static void uuid_from_super_ddf(struct supertype *st, int uuid[4]);
-static void brief_examine_super_ddf(struct supertype *st)
+static void brief_examine_super_ddf(struct supertype *st, int verbose)
{
/* We just write a generic DDF ARRAY entry
*/
+ struct ddf_super *ddf = st->sb;
struct mdinfo info;
+ int i;
char nbuf[64];
getinfo_super_ddf(st, &info);
fname_from_uuid(st, &info, nbuf, ':');
+
+ for (i=0; i<__be16_to_cpu(ddf->virt->max_vdes); i++) {
+ struct virtual_entry *ve = &ddf->virt->entries[i];
+ struct vcl vcl;
+ char nbuf1[64];
+ if (all_ff(ve->guid))
+ continue;
+ memcpy(vcl.conf.guid, ve->guid, DDF_GUID_LEN);
+ ddf->currentconf =&vcl;
+ uuid_from_super_ddf(st, info.uuid);
+ fname_from_uuid(st, &info, nbuf1, ':');
+ printf("ARRAY container=%s member=%d UUID=%s\n",
+ nbuf+5, i, nbuf1+5);
+ }
printf("ARRAY metadata=ddf UUID=%s\n", nbuf + 5);
}
+static void export_examine_super_ddf(struct supertype *st)
+{
+ struct mdinfo info;
+ char nbuf[64];
+ getinfo_super_ddf(st, &info);
+ fname_from_uuid(st, &info, nbuf, ':');
+ printf("MD_METADATA=ddf\n");
+ printf("MD_LEVEL=container\n");
+ printf("MD_UUID=%s\n", nbuf+5);
+}
+
+
static void detail_super_ddf(struct supertype *st, char *homehost)
{
/* FIXME later
sha1_init_ctx(&ctx);
sha1_process_bytes(guid, DDF_GUID_LEN, &ctx);
- if (vcl && vcl->conf.sec_elmnt_count > 1)
- sha1_process_bytes(&vcl->conf.sec_elmnt_seq, 1, &ctx);
sha1_finish_ctx(&ctx, buf);
memcpy(uuid, buf, 4*4);
}
info->component_size = ddf->dlist->size - info->data_offset;
} else {
info->disk.number = -1;
+ info->disk.raid_disk = -1;
// info->disk.raid_disk = find refnum in the table and use index;
}
info->disk.state = (1 << MD_DISK_SYNC);
struct vcl *vc = ddf->currentconf;
int cd = ddf->currentdev;
int j;
+ struct dl *dl;
/* FIXME this returns BVD info - what if we want SVD ?? */
__be32_to_cpu(*(__u32*)(vc->conf.guid+16));
info->array.utime = DECADE + __be32_to_cpu(vc->conf.timestamp);
info->array.chunk_size = 512 << vc->conf.chunk_shift;
+ info->custom_array_size = 0;
if (cd >= 0 && cd < ddf->mppe) {
info->data_offset = __be64_to_cpu(vc->lba_offset[cd]);
info->component_size = __be64_to_cpu(vc->conf.blocks);
}
+ for (dl = ddf->dlist; dl ; dl = dl->next)
+ if (dl->raiddisk == info->disk.raid_disk)
+ break;
info->disk.major = 0;
info->disk.minor = 0;
+ if (dl) {
+ info->disk.major = dl->major;
+ info->disk.minor = dl->minor;
+ }
// info->disk.number = __be32_to_cpu(ddf->disk.refnum);
// info->disk.raid_disk = find refnum in the table and use index;
// info->disk.state = ???;
return rv;
}
+__u32 random32(void)
+{
+ __u32 rv;
+ int rfd = open("/dev/urandom", O_RDONLY);
+ if (rfd < 0 || read(rfd, &rv, 4) != 4)
+ rv = random();
+ if (rfd >= 0)
+ close(rfd);
+ return rv;
+}
+
static void make_header_guid(char *guid)
{
__u32 stamp;
- int rfd;
/* Create a DDF Header of Virtual Disk GUID */
/* 24 bytes of fiction required.
memcpy(guid+12, &stamp, 4);
stamp = __cpu_to_be32(time(0) - DECADE);
memcpy(guid+16, &stamp, 4);
- rfd = open("/dev/urandom", O_RDONLY);
- if (rfd < 0 || read(rfd, &stamp, 4) != 4)
- stamp = random();
+ stamp = random32();
memcpy(guid+20, &stamp, 4);
- if (rfd >= 0) close(rfd);
}
static int init_super_ddf_bvd(struct supertype *st,
memcpy(ddf->controller.product_id, "What Is My PID??", 16);
memset(ddf->controller.pad, 0xff, 8);
memset(ddf->controller.vendor_data, 0xff, 448);
+ if (homehost && strlen(homehost) < 440)
+ strcpy((char*)ddf->controller.vendor_data, homehost);
if (posix_memalign((void**)&pd, 512, pdsize) != 0) {
fprintf(stderr, Name ": %s could not allocate pd\n", __func__);
return 1;
}
-static int all_ff(char *guid)
-{
- int i;
- for (i = 0; i < DDF_GUID_LEN; i++)
- if (guid[i] != (char)0xff)
- return 0;
- return 1;
-}
static int chunk_to_shift(int chunksize)
{
return ffs(chunksize/512)-1;
}
break;
case 5:
- case 6:
switch(layout) {
case ALGORITHM_LEFT_ASYMMETRIC:
return DDF_RAID5_N_RESTART;
case ALGORITHM_RIGHT_ASYMMETRIC:
- if (level == 5)
- return DDF_RAID5_0_RESTART;
- else
- return DDF_RAID6_0_RESTART;
+ return DDF_RAID5_0_RESTART;
case ALGORITHM_LEFT_SYMMETRIC:
return DDF_RAID5_N_CONTINUE;
case ALGORITHM_RIGHT_SYMMETRIC:
return -1; /* not mentioned in standard */
}
+ case 6:
+ switch(layout) {
+ case ALGORITHM_ROTATING_N_RESTART:
+ return DDF_RAID5_N_RESTART;
+ case ALGORITHM_ROTATING_ZERO_RESTART:
+ return DDF_RAID6_0_RESTART;
+ case ALGORITHM_ROTATING_N_CONTINUE:
+ return DDF_RAID5_N_CONTINUE;
+ }
}
return -1;
}
case DDF_RAID6:
switch(rlq) {
case DDF_RAID5_N_RESTART:
- return ALGORITHM_LEFT_ASYMMETRIC;
+ return ALGORITHM_ROTATING_N_RESTART;
case DDF_RAID6_0_RESTART:
- return ALGORITHM_RIGHT_ASYMMETRIC;
+ return ALGORITHM_ROTATING_ZERO_RESTART;
case DDF_RAID5_N_CONTINUE:
- return ALGORITHM_LEFT_SYMMETRIC;
+ return ALGORITHM_ROTATING_N_CONTINUE;
default:
return -1;
}
memset(vc->vendor, 0xff, 32);
memset(vc->phys_refnum, 0xff, 4*ddf->mppe);
- memset(vc->phys_refnum+(ddf->mppe * 4), 0x00, 8*ddf->mppe);
+ memset(vc->phys_refnum+ddf->mppe, 0x00, 8*ddf->mppe);
vcl->next = ddf->conflist;
ddf->conflist = vcl;
* the phys_refnum and lba_offset for the newly created vd_config.
* We might also want to update the type in the phys_disk
* section.
+ *
+ * Alternately: fd == -1 and we have already chosen which device to
+ * use and recorded in dlist->raid_disk;
*/
struct dl *dl;
struct ddf_super *ddf = st->sb;
unsigned long long blocks, pos, esize;
struct extent *ex;
- for (dl = ddf->dlist; dl ; dl = dl->next)
- if (dl->major == dk->major &&
- dl->minor == dk->minor)
- break;
+ if (fd == -1) {
+ for (dl = ddf->dlist; dl ; dl = dl->next)
+ if (dl->raiddisk == dk->raid_disk)
+ break;
+ } else {
+ for (dl = ddf->dlist; dl ; dl = dl->next)
+ if (dl->major == dk->major &&
+ dl->minor == dk->minor)
+ break;
+ }
if (!dl || ! (dk->state & (1<<MD_DISK_SYNC)))
return;
return;
dl->vlist[i] = ddf->currentconf;
- dl->fd = fd;
- dl->devname = devname;
+ if (fd >= 0)
+ dl->fd = fd;
+ if (devname)
+ dl->devname = devname;
/* Check how many working raid_disks, and if we can mark
* array as optimal yet
/* add a device to a container, either while creating it or while
* expanding a pre-existing container
*/
-static void add_to_super_ddf(struct supertype *st,
+static int add_to_super_ddf(struct supertype *st,
mdu_disk_info_t *dk, int fd, char *devname)
{
struct ddf_super *ddf = st->sb;
if (ddf->currentconf) {
add_to_super_ddf_bvd(st, dk, fd, devname);
- return;
+ return 0;
}
/* This is device numbered dk->number. We need to create
fprintf(stderr, Name
": %s could allocate buffer for new disk, aborting\n",
__func__);
- abort();
+ return 1;
}
dd->major = major(stb.st_rdev);
dd->minor = minor(stb.st_rdev);
tm = localtime(&now);
sprintf(dd->disk.guid, "%8s%04d%02d%02d",
T10, tm->tm_year+1900, tm->tm_mon+1, tm->tm_mday);
- *(__u32*)(dd->disk.guid + 16) = random();
- *(__u32*)(dd->disk.guid + 20) = random();
+ *(__u32*)(dd->disk.guid + 16) = random32();
+ *(__u32*)(dd->disk.guid + 20) = random32();
do {
/* Cannot be bothered finding a CRC of some irrelevant details*/
- dd->disk.refnum = random();
+ dd->disk.refnum = random32();
for (i = __be16_to_cpu(ddf->active->max_pd_entries) - 1;
i >= 0; i--)
if (ddf->phys->entries[i].refnum == dd->disk.refnum)
ddf->dlist = dd;
ddf->updates_pending = 1;
}
+
+ return 0;
}
/*
}
#ifndef MDASSEMBLE
+
+static int reserve_space(struct supertype *st, int raiddisks,
+ unsigned long long size, int chunk,
+ unsigned long long *freesize)
+{
+ /* Find 'raiddisks' spare extents at least 'size' big (but
+ * only caring about multiples of 'chunk') and remember
+ * them.
+ * If the cannot be found, fail.
+ */
+ struct dl *dl;
+ struct ddf_super *ddf = st->sb;
+ int cnt = 0;
+
+ for (dl = ddf->dlist; dl ; dl=dl->next) {
+ dl->raiddisk = -1;
+ dl->esize = 0;
+ }
+ /* Now find largest extent on each device */
+ for (dl = ddf->dlist ; dl ; dl=dl->next) {
+ struct extent *e = get_extents(ddf, dl);
+ unsigned long long pos = 0;
+ int i = 0;
+ int found = 0;
+ unsigned long long minsize = size;
+
+ if (size == 0)
+ minsize = chunk;
+
+ if (!e)
+ continue;
+ do {
+ unsigned long long esize;
+ esize = e[i].start - pos;
+ if (esize >= minsize) {
+ found = 1;
+ minsize = esize;
+ }
+ pos = e[i].start + e[i].size;
+ i++;
+ } while (e[i-1].size);
+ if (found) {
+ cnt++;
+ dl->esize = minsize;
+ }
+ free(e);
+ }
+ if (cnt < raiddisks) {
+ fprintf(stderr, Name ": not enough devices with space to create array.\n");
+ return 0; /* No enough free spaces large enough */
+ }
+ if (size == 0) {
+ /* choose the largest size of which there are at least 'raiddisk' */
+ for (dl = ddf->dlist ; dl ; dl=dl->next) {
+ struct dl *dl2;
+ if (dl->esize <= size)
+ continue;
+ /* This is bigger than 'size', see if there are enough */
+ cnt = 0;
+ for (dl2 = dl; dl2 ; dl2=dl2->next)
+ if (dl2->esize >= dl->esize)
+ cnt++;
+ if (cnt >= raiddisks)
+ size = dl->esize;
+ }
+ if (chunk) {
+ size = size / chunk;
+ size *= chunk;
+ }
+ *freesize = size;
+ if (size < 32) {
+ fprintf(stderr, Name ": not enough spare devices to create array.\n");
+ return 0;
+ }
+ }
+ /* We have a 'size' of which there are enough spaces.
+ * We simply do a first-fit */
+ cnt = 0;
+ for (dl = ddf->dlist ; dl && cnt < raiddisks ; dl=dl->next) {
+ if (dl->esize < size)
+ continue;
+
+ dl->raiddisk = cnt;
+ cnt++;
+ }
+ return 1;
+}
+
+
+
static int
validate_geometry_ddf_container(struct supertype *st,
int level, int layout, int raiddisks,
verbose);
}
- if (st->sb) {
- /* A container has already been opened, so we are
- * creating in there. Maybe a BVD, maybe an SVD.
- * Should make a distinction one day.
- */
- return validate_geometry_ddf_bvd(st, level, layout, raiddisks,
- chunk, size, dev, freesize,
- verbose);
- }
if (!dev) {
/* Initial sanity check. Exclude illegal levels. */
int i;
if (ddf_level_num[i].num1 == MAXINT)
return 0;
/* Should check layout? etc */
+
+ if (st->sb && freesize) {
+ /* --create was given a container to create in.
+ * So we need to check that there are enough
+ * free spaces and return the amount of space.
+ * We may as well remember which drives were
+ * chosen so that add_to_super/getinfo_super
+ * can return them.
+ */
+ return reserve_space(st, raiddisks, size, chunk, freesize);
+ }
return 1;
}
+ if (st->sb) {
+ /* A container has already been opened, so we are
+ * creating in there. Maybe a BVD, maybe an SVD.
+ * Should make a distinction one day.
+ */
+ return validate_geometry_ddf_bvd(st, level, layout, raiddisks,
+ chunk, size, dev, freesize,
+ verbose);
+ }
/* This is the first device for the array.
* If it is a container, we read it in and do automagic allocations,
* no other devices should be given.
close(fd);
*freesize = avail_size_ddf(st, ldsize >> 9);
+ if (*freesize == 0)
+ return 0;
return 1;
}
int seq;
char nm[20];
int dfd;
+ int devnum = fd2devnum(fd);
+ enum sysfs_read_flags flags;
+
+ flags = GET_LEVEL|GET_VERSION|GET_DEVS|GET_STATE;
+ if (mdmon_running(devnum))
+ flags |= SKIP_GONE_DEVS;
- sra = sysfs_read(fd, 0, GET_LEVEL|GET_VERSION|GET_DEVS|GET_STATE);
+ sra = sysfs_read(fd, 0, flags);
if (!sra)
return 1;
if (sra->array.major_version != -1 ||
if (d->disk.refnum == vc->conf.phys_refnum[i])
break;
if (d == NULL)
- break;
+ /* Haven't found that one yet, maybe there are others */
+ continue;
dev = malloc(sizeof(*dev));
memset(dev, 0, sizeof(*dev));
}
#endif /* MDASSEMBLE */
+static int ddf_level_to_layout(int level)
+{
+ switch(level) {
+ case 0:
+ case 1:
+ return 0;
+ case 5:
+ return ALGORITHM_LEFT_SYMMETRIC;
+ case 6:
+ return ALGORITHM_ROTATING_N_CONTINUE;
+ case 10:
+ return 0x102;
+ default:
+ return UnSet;
+ }
+}
+
struct superswitch super_ddf = {
#ifndef MDASSEMBLE
.examine_super = examine_super_ddf,
.brief_examine_super = brief_examine_super_ddf,
+ .export_examine_super = export_examine_super_ddf,
.detail_super = detail_super_ddf,
.brief_detail_super = brief_detail_super_ddf,
.validate_geometry = validate_geometry_ddf,
.free_super = free_super_ddf,
.match_metadata_desc = match_metadata_desc_ddf,
.container_content = container_content_ddf,
+ .default_layout = ddf_level_to_layout,
.external = 1,
.prepare_update = ddf_prepare_update,
.activate_spare = ddf_activate_spare,
#endif
+ .name = "ddf",
};