/*
* mdadm - manage Linux "md" devices aka RAID arrays.
*
- * Copyright (C) 2001-2002 Neil Brown <neilb@cse.unsw.edu.au>
+ * Copyright (C) 2001-2006 Neil Brown <neilb@suse.de>
*
*
* This program is free software; you can redistribute it and/or modify
#include <sys/utsname.h>
#include <ctype.h>
+/*
+ * following taken from linux/blkpg.h because they aren't
+ * anywhere else and it isn't safe to #include linux/ * stuff.
+ */
+
+#define BLKPG _IO(0x12,105)
+
+/* The argument structure */
+struct blkpg_ioctl_arg {
+ int op;
+ int flags;
+ int datalen;
+ void *data;
+};
+
+/* The subfunctions (for the op field) */
+#define BLKPG_ADD_PARTITION 1
+#define BLKPG_DEL_PARTITION 2
+
+/* Sizes of name fields. Unused at present. */
+#define BLKPG_DEVNAMELTH 64
+#define BLKPG_VOLNAMELTH 64
+
+/* The data structure for ADD_PARTITION and DEL_PARTITION */
+struct blkpg_partition {
+ long long start; /* starting offset in bytes */
+ long long length; /* length in bytes */
+ int pno; /* partition number */
+ char devname[BLKPG_DEVNAMELTH]; /* partition name, like sda5 or c0d1p2,
+ to be used in kernel messages */
+ char volname[BLKPG_VOLNAMELTH]; /* volume label */
+};
+
/*
* Parse a 128 bit uuid in 4 integers
* format is 32 hexx nibbles with options :.<space> separator
*/
int parse_uuid(char *str, int uuid[4])
{
- int hit = 0; /* number of Hex digIT */
- int i;
- char c;
- for (i=0; i<4; i++) uuid[i]=0;
-
- while ((c= *str++)) {
- int n;
- if (c>='0' && c<='9')
- n = c-'0';
- else if (c>='a' && c <= 'f')
- n = 10 + c - 'a';
- else if (c>='A' && c <= 'F')
- n = 10 + c - 'A';
- else if (strchr(":. -", c))
- continue;
- else return 0;
-
- if (hit<32) {
- uuid[hit/8] <<= 4;
- uuid[hit/8] += n;
+ int hit = 0; /* number of Hex digIT */
+ int i;
+ char c;
+ for (i=0; i<4; i++) uuid[i]=0;
+
+ while ((c= *str++)) {
+ int n;
+ if (c>='0' && c<='9')
+ n = c-'0';
+ else if (c>='a' && c <= 'f')
+ n = 10 + c - 'a';
+ else if (c>='A' && c <= 'F')
+ n = 10 + c - 'A';
+ else if (strchr(":. -", c))
+ continue;
+ else return 0;
+
+ if (hit<32) {
+ uuid[hit/8] <<= 4;
+ uuid[hit/8] += n;
+ }
+ hit++;
}
- hit++;
- }
- if (hit == 32)
- return 1;
- return 0;
-
+ if (hit == 32)
+ return 1;
+ return 0;
}
return -1;
}
-
int get_linux_version()
{
struct utsname name;
return (a*1000000)+(b*1000)+c;
}
-int enough(int level, int raid_disks, int avail_disks)
+void remove_partitions(int fd)
+{
+ /* remove partitions from this block devices.
+ * This is used for components added to an array
+ */
+#ifdef BLKPG_DEL_PARTITION
+ struct blkpg_ioctl_arg a;
+ struct blkpg_partition p;
+
+ a.op = BLKPG_DEL_PARTITION;
+ a.data = (void*)&p;
+ a.datalen = sizeof(p);
+ a.flags = 0;
+ memset(a.data, 0, a.datalen);
+ for (p.pno=0; p.pno < 16; p.pno++)
+ ioctl(fd, BLKPG, &a);
+#endif
+}
+
+int enough(int level, int raid_disks, int layout, int clean,
+ char *avail, int avail_disks)
{
+ int copies, first;
switch (level) {
- case 10: return 1; /* a lie, but it is hard to tell */
+ case 10:
+ /* This is the tricky one - we need to check
+ * which actual disks are present.
+ */
+ copies = (layout&255)* ((layout>>8) & 255);
+ first=0;
+ do {
+ /* there must be one of the 'copies' form 'first' */
+ int n = copies;
+ int cnt=0;
+ while (n--) {
+ if (avail[first])
+ cnt++;
+ first = (first+1) % raid_disks;
+ }
+ if (cnt == 0)
+ return 0;
+
+ } while (first != 0);
+ return 1;
case -4:
return avail_disks>= 1;
return avail_disks >= 1;
case 4:
case 5:
- return avail_disks >= raid_disks-1;
+ if (clean)
+ return avail_disks >= raid_disks-1;
+ else
+ return avail_disks >= raid_disks;
case 6:
- return avail_disks >= raid_disks-2;
+ if (clean)
+ return avail_disks >= raid_disks-2;
+ else
+ return avail_disks >= raid_disks;
default:
return 0;
}
}
-int same_uuid(int a[4], int b[4])
+int same_uuid(int a[4], int b[4], int swapuuid)
{
- if (a[0]==b[0] &&
- a[1]==b[1] &&
- a[2]==b[2] &&
- a[3]==b[3])
- return 1;
- return 0;
+ if (swapuuid) {
+ /* parse uuids are hostendian.
+ * uuid's from some superblocks are big-ending
+ * if there is a difference, we need to swap..
+ */
+ unsigned char *ac = (unsigned char *)a;
+ unsigned char *bc = (unsigned char *)b;
+ int i;
+ for (i=0; i<16; i+= 4) {
+ if (ac[i+0] != bc[i+3] ||
+ ac[i+1] != bc[i+2] ||
+ ac[i+2] != bc[i+1] ||
+ ac[i+3] != bc[i+0])
+ return 0;
+ }
+ return 1;
+ } else {
+ if (a[0]==b[0] &&
+ a[1]==b[1] &&
+ a[2]==b[2] &&
+ a[3]==b[3])
+ return 1;
+ return 0;
+ }
+}
+void copy_uuid(void *a, int b[4], int swapuuid)
+{
+ if (swapuuid) {
+ /* parse uuids are hostendian.
+ * uuid's from some superblocks are big-ending
+ * if there is a difference, we need to swap..
+ */
+ unsigned char *ac = (unsigned char *)a;
+ unsigned char *bc = (unsigned char *)b;
+ int i;
+ for (i=0; i<16; i+= 4) {
+ ac[i+0] = bc[i+3];
+ ac[i+1] = bc[i+2];
+ ac[i+2] = bc[i+1];
+ ac[i+3] = bc[i+0];
+ }
+ } else
+ memcpy(a, b, 16);
}
+#ifndef MDASSEMBLE
int check_ext2(int fd, char *name)
{
/*
*
*/
unsigned char sb[1024];
- int size;
+ unsigned long size;
if (lseek(fd, 64*1024, 0) != 64*1024)
return 0;
if (read(fd, sb, 1024) != 1024)
return 0;
fprintf(stderr, Name ": %s appears to contain a reiserfs file system\n",name);
size = sb[0]|(sb[1]|(sb[2]|sb[3]<<8)<<8)<<8;
- fprintf(stderr, " size = %dK\n", size*4);
-
+ fprintf(stderr, " size = %luK\n", size*4);
+
return 1;
}
int check_raid(int fd, char *name)
{
- void *super;
struct mdinfo info;
time_t crtime;
+ char *level;
struct supertype *st = guess_super(fd);
if (!st) return 0;
- st->ss->load_super(st, fd, &super, name);
+ st->ss->load_super(st, fd, name);
/* Looks like a raid array .. */
fprintf(stderr, Name ": %s appears to be part of a raid array:\n",
name);
- st->ss->getinfo_super(&info, super);
- free(super);
+ st->ss->getinfo_super(st, &info);
+ st->ss->free_super(st);
crtime = info.array.ctime;
- fprintf(stderr, " level=%d devices=%d ctime=%s",
- info.array.level, info.array.raid_disks, ctime(&crtime));
+ level = map_num(pers, info.array.level);
+ if (!level) level = "-unknown-";
+ fprintf(stderr, " level=%s devices=%d ctime=%s",
+ level, info.array.raid_disks, ctime(&crtime));
return 1;
}
fprintf(stderr, Name ": assuming 'no'\n");
return 0;
}
+#endif /* MDASSEMBLE */
char *map_num(mapping_t *map, int num)
{
{
/* tests if dev is a "standard" md dev name.
* i.e if the last component is "/dNN" or "/mdNN",
- * where NN is a string of digits
+ * where NN is a string of digits
*/
char *d = strrchr(dev, '/');
int type=0;
} *devlist = NULL;
int devlist_ready = 0;
-#ifdef UCLIBC
int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s)
{
+ struct stat st;
+ if (S_ISLNK(stb->st_mode)) {
+ stat(name, &st);
+ stb = &st;
+ }
+
+ if ((stb->st_mode&S_IFMT)== S_IFBLK) {
+ char *n = strdup(name);
+ struct devmap *dm = malloc(sizeof(*dm));
+ if (strncmp(n, "/dev/./", 7)==0)
+ strcpy(n+4, name+6);
+ if (dm) {
+ dm->major = major(stb->st_rdev);
+ dm->minor = minor(stb->st_rdev);
+ dm->name = n;
+ dm->next = devlist;
+ devlist = dm;
+ }
+ }
+ return 0;
}
-char *map_dev(int major, int minor)
-{
-#if 0
- fprintf(stderr, "Warning - fail to map %d,%d to a device name\n",
- major, minor);
-#endif
- return NULL;
-}
-#else
-#ifdef __dietlibc__
+#ifndef HAVE_NFTW
+#ifdef HAVE_FTW
int add_dev_1(const char *name, const struct stat *stb, int flag)
{
return add_dev(name, stb, flag, NULL);
}
int nftw(const char *path, int (*han)(const char *name, const struct stat *stb, int flag, struct FTW *s), int nopenfd, int flags)
{
- ftw(path, add_dev_1, nopenfd);
+ return ftw(path, add_dev_1, nopenfd);
}
-#endif
-
-int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s)
+#else
+int nftw(const char *path, int (*han)(const char *name, const struct stat *stb, int flag, struct FTW *s), int nopenfd, int flags)
{
- if ((stb->st_mode&S_IFMT)== S_IFBLK) {
- char *n = strdup(name);
- struct devmap *dm = malloc(sizeof(*dm));
- if (dm) {
- dm->major = major(stb->st_rdev);
- dm->minor = minor(stb->st_rdev);
- dm->name = n;
- dm->next = devlist;
- devlist = dm;
- }
- }
- return 0;
+ return 0;
}
+#endif /* HAVE_FTW */
+#endif /* HAVE_NFTW */
/*
* Find a block device with the right major/minor number.
* deliberately so prefer it over a standard name.
* This applies only to names for MD devices.
*/
-char *map_dev(int major, int minor)
+char *map_dev(int major, int minor, int create)
{
struct devmap *p;
char *std = NULL, *nonstd=NULL;
+ int did_check = 0;
+
+ if (major == 0 && minor == 0)
+ return NULL;
+
+ retry:
if (!devlist_ready) {
- nftw("/dev", add_dev, 10, FTW_PHYS);
+ char *dev = "/dev";
+ struct stat stb;
+ while(devlist) {
+ struct devmap *d = devlist;
+ devlist = d->next;
+ free(d->name);
+ free(d);
+ }
+ if (lstat(dev, &stb)==0 &&
+ S_ISLNK(stb.st_mode))
+ dev = "/dev/.";
+ nftw(dev, add_dev, 10, FTW_PHYS);
devlist_ready=1;
+ did_check = 1;
}
for (p=devlist; p; p=p->next)
nonstd = p->name;
}
}
+ if (!std && !nonstd && !did_check) {
+ devlist_ready = 0;
+ goto retry;
+ }
+ if (create && !std && !nonstd) {
+ static char buf[30];
+ snprintf(buf, sizeof(buf), "%d:%d", major, minor);
+ nonstd = buf;
+ }
+
return nonstd ? nonstd : std;
}
-#endif
-
unsigned long calc_csum(void *super, int bytes)
{
unsigned long long newcsum = 0;
newcsum+= superc[i];
csum = (newcsum& 0xffffffff) + (newcsum>>32);
#ifdef __alpha__
-/* The in-kernel checksum calculation is always 16bit on
+/* The in-kernel checksum calculation is always 16bit on
* the alpha, though it is 32 bit on i386...
* I wonder what it is elsewhere... (it uses and API in
* a way that it shouldn't).
return csum;
}
+#ifndef MDASSEMBLE
char *human_size(long long bytes)
{
static char buf[30];
char *human_size_brief(long long bytes)
{
static char buf[30];
-
if (bytes < 5000*1024)
snprintf(buf, sizeof(buf), "%ld.%02ldKiB",
);
return buf;
}
+#endif
+#if !defined(MDASSEMBLE) || defined(MDASSEMBLE) && defined(MDASSEMBLE_AUTO)
int get_mdp_major(void)
{
static int mdp_major = -1;
&& (stb.st_rdev == rdev))
return devname;
}
- dn = map_dev(major(rdev), minor(rdev));
+ dn = map_dev(major(rdev), minor(rdev), 0);
if (dn)
return dn;
snprintf(devname, sizeof(devname), "/dev/.tmp.md%d", dev);
if (strncmp(name, "/dev/.tmp.md", 12)==0)
unlink(name);
}
+#endif /* !defined(MDASSEMBLE) || defined(MDASSEMBLE) && defined(MDASSEMBLE_AUTO) */
-
-
+int dev_open(char *dev, int flags)
+{
+ /* like 'open', but if 'dev' matches %d:%d, create a temp
+ * block device and open that
+ */
+ char *e;
+ int fd = -1;
+ char devname[32];
+ int major;
+ int minor;
+
+ if (!dev) return -1;
+
+ major = strtoul(dev, &e, 0);
+ if (e > dev && *e == ':' && e[1] &&
+ (minor = strtoul(e+1, &e, 0)) >= 0 &&
+ *e == 0) {
+ snprintf(devname, sizeof(devname), "/dev/.tmp.md.%d:%d", major, minor);
+ if (mknod(devname, S_IFBLK|0600, makedev(major, minor))==0) {
+ fd = open(devname, flags);
+ unlink(devname);
+ }
+ } else
+ fd = open(dev, flags);
+ return fd;
+}
struct superswitch *superlist[] = { &super0, &super1, NULL };
-struct supertype *super_by_version(int vers, int minor)
+struct supertype *super_by_fd(int fd)
{
- struct supertype *st = malloc(sizeof(*st));
- if (!st) return st;
- if (vers == 0) {
- st->ss = &super0;
- st->max_devs = MD_SB_DISKS;
+ mdu_array_info_t array;
+ int vers;
+ int minor;
+ struct supertype *st = NULL;
+ struct mdinfo *sra;
+ char *verstr;
+ char version[20];
+ int i;
+
+ sra = sysfs_read(fd, 0, GET_VERSION);
+
+ if (sra) {
+ vers = sra->array.major_version;
+ minor = sra->array.minor_version;
+ verstr = sra->text_version;
+ } else {
+ if (ioctl(fd, GET_ARRAY_INFO, &array))
+ array.major_version = array.minor_version = 0;
+ vers = array.major_version;
+ minor = array.minor_version;
+ verstr = "";
}
- if (vers == 1) {
- st->ss = &super1;
- st->max_devs = 384;
+ if (vers != -1) {
+ sprintf(version, "%d.%d", vers, minor);
+ verstr = version;
}
- st->minor_version = minor;
+ for (i = 0; st == NULL && superlist[i] ; i++)
+ st = superlist[i]->match_metadata_desc(verstr);
+
+ if (sra)
+ sysfs_free(sra);
+ st->sb = NULL;
return st;
}
+struct supertype *dup_super(struct supertype *st)
+{
+ struct supertype *stnew = NULL;
+ char *verstr = NULL;
+ char version[20];
+ int i;
+
+ if (!st)
+ return st;
+
+ if (st->minor_version == -1)
+ sprintf(version, "%d", st->ss->major);
+ else
+ sprintf(version, "%d.%d", st->ss->major, st->minor_version);
+ verstr = version;
+
+ for (i = 0; stnew == NULL && superlist[i] ; i++)
+ stnew = superlist[i]->match_metadata_desc(verstr);
+
+ stnew->sb = NULL;
+ return stnew;
+}
+
struct supertype *guess_super(int fd)
{
/* try each load_super to find the best match,
struct supertype *st;
unsigned long besttime = 0;
int bestsuper = -1;
-
- void *sbp = NULL;
int i;
st = malloc(sizeof(*st));
for (i=0 ; superlist[i]; i++) {
int rv;
ss = superlist[i];
- rv = ss->load_super(st, fd, &sbp, NULL);
+ st->ss = NULL;
+ rv = ss->load_super(st, fd, NULL);
if (rv == 0) {
struct mdinfo info;
- ss->getinfo_super(&info, sbp);
+ st->ss->getinfo_super(st, &info);
if (bestsuper == -1 ||
besttime < info.array.ctime) {
bestsuper = i;
besttime = info.array.ctime;
}
- st->ss = NULL;
- free(sbp);
+ ss->free_super(st);
}
}
if (bestsuper != -1) {
int rv;
- rv = superlist[bestsuper]->load_super(st, fd, &sbp, NULL);
+ st->ss = NULL;
+ rv = superlist[bestsuper]->load_super(st, fd, NULL);
if (rv == 0) {
- free(sbp);
+ superlist[bestsuper]->free_super(st);
return st;
}
}
free(st);
return NULL;
}
+
+/* Return size of device in bytes */
+int get_dev_size(int fd, char *dname, unsigned long long *sizep)
+{
+ unsigned long long ldsize;
+ struct stat st;
+
+ if (fstat(fd, &st) != -1 && S_ISREG(st.st_mode))
+ ldsize = (unsigned long long)st.st_size;
+ else
+#ifdef BLKGETSIZE64
+ if (ioctl(fd, BLKGETSIZE64, &ldsize) != 0)
+#endif
+ {
+ unsigned long dsize;
+ if (ioctl(fd, BLKGETSIZE, &dsize) == 0) {
+ ldsize = dsize;
+ ldsize <<= 9;
+ } else {
+ if (dname)
+ fprintf(stderr, Name ": Cannot get size of %s: %s\b",
+ dname, strerror(errno));
+ return 0;
+ }
+ }
+ *sizep = ldsize;
+ return 1;
+}
+
+void get_one_disk(int mdfd, mdu_array_info_t *ainf, mdu_disk_info_t *disk)
+{
+ int d;
+ ioctl(mdfd, GET_ARRAY_INFO, ainf);
+ for (d = 0 ; d < ainf->raid_disks + ainf->nr_disks ; d++)
+ if (ioctl(mdfd, GET_DISK_INFO, disk) == 0)
+ return;
+}
+#ifdef __TINYC__
+/* tinyc doesn't optimize this check in ioctl.h out ... */
+unsigned int __invalid_size_argument_for_IOC = 0;
+#endif
+