/*
* 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"
return (a*1000000)+(b*1000)+c;
}
+#ifndef MDASSEMBLE
+long long parse_size(char *size)
+{
+ /* parse 'size' which should be a number optionally
+ * followed by 'K', 'M', or 'G'.
+ * Without a suffix, K is assumed.
+ * Number returned is in sectors (half-K)
+ */
+ char *c;
+ long long s = strtoll(size, &c, 10);
+ if (s > 0) {
+ switch (*c) {
+ case 'K':
+ c++;
+ default:
+ s *= 2;
+ break;
+ case 'M':
+ c++;
+ s *= 1024 * 2;
+ break;
+ case 'G':
+ c++;
+ s *= 1024 * 1024 * 2;
+ break;
+ }
+ }
+ if (*c)
+ s = 0;
+ return s;
+}
+
+int parse_layout_10(char *layout)
+{
+ int copies, rv;
+ char *cp;
+ /* Parse the layout string for raid10 */
+ /* 'f', 'o' or 'n' followed by a number <= raid_disks */
+ if ((layout[0] != 'n' && layout[0] != 'f' && layout[0] != 'o') ||
+ (copies = strtoul(layout+1, &cp, 10)) < 1 ||
+ copies > 200 ||
+ *cp)
+ return -1;
+ if (layout[0] == 'n')
+ rv = 256 + copies;
+ else if (layout[0] == 'o')
+ rv = 0x10000 + (copies<<8) + 1;
+ else
+ rv = 1 + (copies<<8);
+ return rv;
+}
+
+int parse_layout_faulty(char *layout)
+{
+ /* Parse the layout string for 'faulty' */
+ int ln = strcspn(layout, "0123456789");
+ char *m = strdup(layout);
+ int mode;
+ m[ln] = 0;
+ mode = map_name(faultylayout, m);
+ if (mode == UnSet)
+ return -1;
+
+ return mode | (atoi(layout+ln)<< ModeShift);
+}
+#endif
+
void remove_partitions(int fd)
{
/* remove partitions from this block devices.
} while (first != 0);
return 1;
- case -4:
+ case LEVEL_MULTIPATH:
return avail_disks>= 1;
- case -1:
+ case LEVEL_LINEAR:
case 0:
return avail_disks == raid_disks;
case 1:
}
}
+const int uuid_match_any[4] = { ~0, ~0, ~0, ~0 };
int same_uuid(int a[4], int b[4], int swapuuid)
{
+ if (memcmp(a, uuid_match_any, sizeof(int[4])) == 0 ||
+ memcmp(b, uuid_match_any, sizeof(int[4])) == 0)
+ return 1;
+
if (swapuuid) {
/* parse uuids are hostendian.
* uuid's from some superblocks are big-ending
memcpy(a, b, 16);
}
-char *fname_from_uuid(struct supertype *st, struct mdinfo *info, char *buf)
+char *__fname_from_uuid(int id[4], int swap, char *buf, char sep)
{
- int i;
+ int i, j;
char uuid[16];
char *c = buf;
strcpy(c, "UUID-");
c += strlen(c);
- copy_uuid(uuid, info->uuid, st->ss->swapuuid);
- for (i=0; i<16; i++) {
- if (i && (i&3)==0) {
- strcpy(c, "-");
- c++;
+ copy_uuid(uuid, id, swap);
+ for (i = 0; i < 4; i++) {
+ if (i)
+ *c++ = sep;
+ for (j = 3; j >= 0; j--) {
+ sprintf(c,"%02x", (unsigned char) uuid[j+4*i]);
+ c+= 2;
}
- sprintf(c,"%02x", (unsigned char)uuid[i]);
- c+= 2;
}
return buf;
+
+}
+
+char *fname_from_uuid(struct supertype *st, struct mdinfo *info, char *buf, char sep)
+{
+ return __fname_from_uuid(info->uuid, st->ss->swapuuid, buf, sep);
}
#ifndef MDASSEMBLE
if (strncmp(d, "/d",2)==0)
d += 2, type=1; /* /dev/md/dN{pM} */
else if (strncmp(d, "/md_d", 5)==0)
- d += 5, type=1; /* /dev/md_dNpM */
+ d += 5, type=1; /* /dev/md_dN{pM} */
else if (strncmp(d, "/md", 3)==0)
d += 3, type=-1; /* /dev/mdN */
else if (d-dev > 3 && strncmp(d-2, "md/", 3)==0)
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);
+ if (stat(name, &st) != 0)
+ return 0;
stb = &st;
}
/*
* Find a block device with the right major/minor number.
* If we find multiple names, choose the shortest.
- * If we find a non-standard name, it is probably there
- * deliberately so prefer it over a standard name.
+ * If we find a name in /dev/md/, we prefer that.
* This applies only to names for MD devices.
*/
char *map_dev(int major, int minor, int create)
{
struct devmap *p;
- char *std = NULL, *nonstd=NULL;
+ char *regular = NULL, *preferred=NULL;
int did_check = 0;
if (major == 0 && minor == 0)
for (p=devlist; p; p=p->next)
if (p->major == major &&
p->minor == minor) {
- if (is_standard(p->name, NULL)) {
- if (std == NULL ||
- strlen(p->name) < strlen(std))
- std = p->name;
+ if (strncmp(p->name, "/dev/md/",8) == 0) {
+ if (preferred == NULL ||
+ strlen(p->name) < strlen(preferred))
+ preferred = p->name;
} else {
- if (nonstd == NULL ||
- strlen(p->name) < strlen(nonstd))
- nonstd = p->name;
+ if (regular == NULL ||
+ strlen(p->name) < strlen(regular))
+ regular = p->name;
}
}
- if (!std && !nonstd && !did_check) {
+ if (!regular && !preferred && !did_check) {
devlist_ready = 0;
goto retry;
}
- if (create && !std && !nonstd) {
+ if (create && !regular && !preferred) {
static char buf[30];
snprintf(buf, sizeof(buf), "%d:%d", major, minor);
- nonstd = buf;
+ regular = buf;
}
- return nonstd ? nonstd : std;
+ return preferred ? preferred : regular;
}
unsigned long calc_csum(void *super, int bytes)
);
return buf;
}
+
+void print_r10_layout(int layout)
+{
+ int near = layout & 255;
+ int far = (layout >> 8) & 255;
+ int offset = (layout&0x10000);
+ char *sep = "";
+
+ if (near != 1) {
+ printf("%s near=%d", sep, near);
+ sep = ",";
+ }
+ if (far != 1)
+ printf("%s %s=%d", sep, offset?"offset":"far", far);
+ if (near*far == 1)
+ printf("NO REDUNDANCY");
+}
#endif
unsigned long long calc_array_size(int level, int raid_disks, int layout,
{
int devnum;
for (devnum = 127; devnum != 128;
- devnum = devnum ? devnum-1 : (1<<22)-1) {
+ devnum = devnum ? devnum-1 : (1<<20)-1) {
char *dn;
int _devnum;
return fd;
}
+int open_dev(int devnum)
+{
+ char buf[20];
+
+ sprintf(buf, "%d:%d", dev2major(devnum), dev2minor(devnum));
+ return dev_open(buf, O_RDWR);
+}
+
int open_dev_excl(int devnum)
{
char buf[20];
return -1;
}
+int same_dev(char *one, char *two)
+{
+ struct stat st1, st2;
+ if (stat(one, &st1) != 0)
+ return 0;
+ if (stat(two, &st2) != 0)
+ return 0;
+ if ((st1.st_mode & S_IFMT) != S_IFBLK)
+ return 0;
+ if ((st2.st_mode & S_IFMT) != S_IFBLK)
+ return 0;
+ return st1.st_rdev == st2.st_rdev;
+}
+
+void wait_for(char *dev, int fd)
+{
+ int i;
+ struct stat stb_want;
+
+ if (fstat(fd, &stb_want) != 0 ||
+ (stb_want.st_mode & S_IFMT) != S_IFBLK)
+ return;
+
+ for (i=0 ; i<25 ; i++) {
+ struct stat stb;
+ if (stat(dev, &stb) == 0 &&
+ (stb.st_mode & S_IFMT) == S_IFBLK &&
+ (stb.st_rdev == stb_want.st_rdev))
+ return;
+ usleep(200000);
+ }
+ if (i == 25)
+ dprintf("%s: timeout waiting for %s\n", __func__, dev);
+}
+
struct superswitch *superlist[] = { &super0, &super1, &super_ddf, &super_imsm, NULL };
#if !defined(MDASSEMBLE) || defined(MDASSEMBLE) && defined(MDASSEMBLE_AUTO)
if (sra)
sysfs_free(sra);
sra = sysfs_read(-1, devnum, GET_VERSION);
- verstr = sra->text_version ? : "-no-metadata-";
+ if (sra && sra->text_version[0])
+ verstr = sra->text_version;
+ else
+ verstr = "-no-metadata-";
}
for (i = 0; st == NULL && superlist[i] ; i++)
int rv;
#ifndef MDASSEMBLE
if (st->ss->external) {
- rv = sysfs_add_disk(sra, info);
+ if (info->disk.state & (1<<MD_DISK_SYNC))
+ info->recovery_start = MaxSector;
+ else
+ info->recovery_start = 0;
+ rv = sysfs_add_disk(sra, info, 0);
if (! rv) {
struct mdinfo *sd2;
for (sd2 = sra->devs; sd2; sd2=sd2->next)
return rv;
}
+unsigned long long min_recovery_start(struct mdinfo *array)
+{
+ /* find the minimum recovery_start in an array for metadata
+ * formats that only record per-array recovery progress instead
+ * of per-device
+ */
+ unsigned long long recovery_start = MaxSector;
+ struct mdinfo *d;
+
+ for (d = array->devs; d; d = d->next)
+ recovery_start = min(recovery_start, d->recovery_start);
+
+ return recovery_start;
+}
+
char *devnum2devname(int num)
{
char name[100];
- if (num > 0)
+ if (num >= 0)
sprintf(name, "md%d", num);
else
sprintf(name, "md_d%d", -1-num);
int stat2devnum(struct stat *st)
{
+ char path[30];
+ char link[200];
+ char *cp;
+ int n;
+
if ((S_IFMT & st->st_mode) == S_IFBLK) {
if (major(st->st_rdev) == MD_MAJOR)
return minor(st->st_rdev);
- else
- return -1- (minor(st->st_rdev)>>6);
+ else if (major(st->st_rdev) == get_mdp_major())
+ return -1- (minor(st->st_rdev)>>MdpMinorShift);
+
+ /* must be an extended-minor partition. Look at the
+ * /sys/dev/block/%d:%d link which must look like
+ * ../../block/mdXXX/mdXXXpYY
+ */
+ sprintf(path, "/sys/dev/block/%d:%d", major(st->st_rdev),
+ minor(st->st_rdev));
+ n = readlink(path, link, sizeof(link)-1);
+ if (n <= 0)
+ return NoMdDev;
+ link[n] = 0;
+ cp = strrchr(link, '/');
+ if (cp) *cp = 0;
+ cp = strchr(link, '/');
+ if (cp && strncmp(cp, "/md", 3) == 0)
+ return devname2devnum(cp+1);
}
- return -1;
+ return NoMdDev;
}
struct stat stb;
if (fstat(fd, &stb) == 0)
return stat2devnum(&stb);
- return -1;
+ return NoMdDev;
}
int mdmon_running(int devnum)
NULL
};
- if (env_no_mdmon())
+ if (check_env("MDADM_NO_MDMON"))
return 0;
len = readlink("/proc/self/exe", pathbuf, sizeof(pathbuf));
for (i=0; paths[i]; i++)
if (paths[i][0])
execl(paths[i], "mdmon",
- map_dev(dev2major(devnum),
- dev2minor(devnum),
- 1), NULL);
+ devnum2devname(devnum),
+ NULL);
exit(1);
case -1: fprintf(stderr, Name ": cannot run mdmon. "
"Array remains readonly\n");
return 0;
}
-int env_no_mdmon(void)
+int check_env(char *name)
{
- char *val = getenv("MDADM_NO_MDMON");
+ char *val = getenv(name);
if (val && atoi(val) == 1)
return 1;
return 0;
}
+__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;
+}
+
#ifndef MDASSEMBLE
int flush_metadata_updates(struct supertype *st)
{
projects / thirdparty / mdadm.git / blobdiff