X-Git-Url: http://git.ipfire.org/?p=thirdparty%2Fmdadm.git;a=blobdiff_plain;f=util.c;h=4cecd6d0cfdc1f0a357475624ac72195ed871be8;hp=ab606ab12fa08f4861eb6fef549e89050f6f4131;hb=d2ca644994d642c31b41242140e1fe819711c8f7;hpb=0a416ec30712b201fbbf300e032cfea02109cd23 diff --git a/util.c b/util.c index ab606ab1..4cecd6d0 100644 --- a/util.c +++ b/util.c @@ -1,7 +1,7 @@ /* * mdadm - manage Linux "md" devices aka RAID arrays. * - * Copyright (C) 2001-2002 Neil Brown + * Copyright (C) 2001-2006 Neil Brown * * * This program is free software; you can redistribute it and/or modify @@ -31,6 +31,41 @@ #include "md_p.h" #include #include +#include +#include + +/* + * 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 @@ -40,33 +75,32 @@ */ 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; + 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; + 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; } @@ -99,7 +133,6 @@ int md_get_version(int fd) return -1; } - int get_linux_version() { struct utsname name; @@ -118,7 +151,26 @@ int get_linux_version() return (a*1000000)+(b*1000)+c; } -int enough(int level, int raid_disks, int layout, +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; @@ -127,7 +179,7 @@ int enough(int level, int raid_disks, int layout, /* This is the tricky one - we need to check * which actual disks are present. */ - copies = (layout&255)* (layout>>8); + copies = (layout&255)* ((layout>>8) & 255); first=0; do { /* there must be one of the 'copies' form 'first' */ @@ -153,9 +205,15 @@ int enough(int level, int raid_disks, int layout, 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; } @@ -166,7 +224,7 @@ int same_uuid(int a[4], 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.. + * if there is a difference, we need to swap.. */ unsigned char *ac = (unsigned char *)a; unsigned char *bc = (unsigned char *)b; @@ -188,7 +246,27 @@ int same_uuid(int a[4], int b[4], int swapuuid) 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) { /* @@ -240,25 +318,24 @@ int check_reiser(int fd, char *name) 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 = %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; level = map_num(pers, info.array.level); if (!level) level = "-unknown-"; @@ -286,6 +363,7 @@ int ask(char *mesg) fprintf(stderr, Name ": assuming 'no'\n"); return 0; } +#endif /* MDASSEMBLE */ char *map_num(mapping_t *map, int num) { @@ -312,7 +390,10 @@ int is_standard(char *dev, int *nump) { /* 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 + * Returns 1 if a partitionable standard, + * -1 if non-partitonable, + * 0 if not a standard name. */ char *d = strrchr(dev, '/'); int type=0; @@ -354,48 +435,47 @@ struct devmap { } *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 (strncmp(n, "/dev/.", 6)==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; + return 0; } +#endif /* HAVE_FTW */ +#endif /* HAVE_NFTW */ /* * Find a block device with the right major/minor number. @@ -404,18 +484,31 @@ int add_dev(const char *name, const struct stat *stb, int flag, struct FTW *s) * 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) { 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) @@ -431,11 +524,19 @@ char *map_dev(int major, int minor) 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; @@ -447,7 +548,7 @@ unsigned long calc_csum(void *super, int bytes) 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). @@ -458,6 +559,7 @@ unsigned long calc_csum(void *super, int bytes) return csum; } +#ifndef MDASSEMBLE char *human_size(long long bytes) { static char buf[30]; @@ -492,7 +594,6 @@ char *human_size(long long bytes) char *human_size_brief(long long bytes) { static char buf[30]; - if (bytes < 5000*1024) snprintf(buf, sizeof(buf), "%ld.%02ldKiB", @@ -510,7 +611,26 @@ char *human_size_brief(long long bytes) ); return buf; } +#endif + +unsigned long long calc_array_size(int level, int raid_disks, int layout, + int chunksize, unsigned long long devsize) +{ + int data_disks = 0; + switch (level) { + case 0: data_disks = raid_disks; break; + case 1: data_disks = 1; break; + case 4: + case 5: data_disks = raid_disks - 1; break; + case 6: data_disks = raid_disks - 2; break; + case 10: data_disks = raid_disks / (layout & 255) / ((layout>>8)&255); + break; + } + devsize &= ~(unsigned long long)((chunksize>>9)-1); + return data_disks * devsize; +} +#if !defined(MDASSEMBLE) || defined(MDASSEMBLE) && defined(MDASSEMBLE_AUTO) int get_mdp_major(void) { static int mdp_major = -1; @@ -573,7 +693,7 @@ char *get_md_name(int dev) && (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); @@ -595,6 +715,30 @@ void put_md_name(char *name) unlink(name); } +int find_free_devnum(int use_partitions) +{ + int devnum; + for (devnum = 127; devnum != 128; + devnum = devnum ? devnum-1 : (1<<22)-1) { + char *dn; + int _devnum; + + _devnum = use_partitions ? (-1-devnum) : devnum; + if (mddev_busy(_devnum)) + continue; + /* make sure it is new to /dev too, at least as a + * non-standard */ + dn = map_dev(dev2major(_devnum), dev2minor(_devnum), 0); + if (dn && ! is_standard(dn, NULL)) + continue; + break; + } + if (devnum == 128) + return NoMdDev; + return use_partitions ? (-1-devnum) : devnum; +} +#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 @@ -603,12 +747,17 @@ int dev_open(char *dev, int flags) char *e; int fd = -1; char devname[32]; - int major = strtoul(dev, &e, 0); + 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); + snprintf(devname, sizeof(devname), "/dev/.tmp.md.%d:%d:%d", + (int)getpid(), major, minor); if (mknod(devname, S_IFBLK|0600, makedev(major, minor))==0) { fd = open(devname, flags); unlink(devname); @@ -618,22 +767,109 @@ int dev_open(char *dev, int flags) return fd; } -struct superswitch *superlist[] = { &super0, &super1, NULL }; +int open_dev_excl(int devnum) +{ + char buf[20]; + int i; -struct supertype *super_by_version(int vers, int minor) + sprintf(buf, "%d:%d", dev2major(devnum), dev2minor(devnum)); + for (i=0 ; i<25 ; i++) { + int fd = dev_open(buf, O_RDWR|O_EXCL); + if (fd >= 0) + return fd; + if (errno != EBUSY) + return fd; + usleep(200000); + } + return -1; +} + +struct superswitch *superlist[] = { &super0, &super1, &super_ddf, &super_imsm, NULL }; + +#if !defined(MDASSEMBLE) || defined(MDASSEMBLE) && defined(MDASSEMBLE_AUTO) + +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; + char *subarray = NULL; + + 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) { + sprintf(version, "%d.%d", vers, minor); + verstr = version; + } + if (minor == -2 && verstr[0] == '/') { + char *dev = verstr+1; + subarray = strchr(dev, '/'); + int devnum; + if (subarray) + *subarray++ = '\0'; + if (strncmp(dev, "md_d", 4) == 0) + devnum = -1-atoi(dev+4); + else + devnum = atoi(dev+2); + subarray = strdup(subarray); + if (sra) + sysfs_free(sra); + sra = sysfs_read(-1, devnum, GET_VERSION); + verstr = sra->text_version ? : "-no-metadata-"; } - if (vers == 1) { - st->ss = &super1; - st->max_devs = 384; + for (i = 0; st == NULL && superlist[i] ; i++) + st = superlist[i]->match_metadata_desc(verstr); + + if (sra) + sysfs_free(sra); + if (st) { + st->sb = NULL; + if (subarray) { + strncpy(st->subarray, subarray, 32); + st->subarray[31] = 0; + free(subarray); + } else + st->subarray[0] = 0; } - st->minor_version = minor; + return st; +} +#endif /* !defined(MDASSEMBLE) || defined(MDASSEMBLE) && defined(MDASSEMBLE_AUTO) */ + + +struct supertype *dup_super(struct supertype *orig) +{ + struct supertype *st; + + if (!orig) + return orig; + st = malloc(sizeof(*st)); + if (!st) + return st; + memset(st, 0, sizeof(*st)); + st->ss = orig->ss; + st->max_devs = orig->max_devs; + st->minor_version = orig->minor_version; + strcpy(st->subarray, orig->subarray); + st->sb = NULL; + st->info = NULL; return st; } @@ -646,34 +882,31 @@ struct supertype *guess_super(int fd) struct supertype *st; unsigned long besttime = 0; int bestsuper = -1; - - void *sbp = NULL; int i; st = malloc(sizeof(*st)); - memset(st, 0, sizeof(*st)); for (i=0 ; superlist[i]; i++) { int rv; ss = superlist[i]; - st->ss = NULL; - rv = ss->load_super(st, fd, &sbp, NULL); + memset(st, 0, sizeof(*st)); + 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; } - free(sbp); + ss->free_super(st); } } if (bestsuper != -1) { int rv; - st->ss = NULL; - rv = superlist[bestsuper]->load_super(st, fd, &sbp, NULL); + memset(st, 0, sizeof(*st)); + rv = superlist[bestsuper]->load_super(st, fd, NULL); if (rv == 0) { - free(sbp); + superlist[bestsuper]->free_super(st); return st; } } @@ -681,6 +914,157 @@ struct supertype *guess_super(int fd) 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; +} + +int open_container(int fd) +{ + /* 'fd' is a block device. Find out if it is in use + * by a container, and return an open fd on that container. + */ + char path[256]; + char *e; + DIR *dir; + struct dirent *de; + int dfd, n; + char buf[200]; + int major, minor; + struct stat st; + + if (fstat(fd, &st) != 0) + return -1; + sprintf(path, "/sys/dev/block/%d:%d/holders", + (int)major(st.st_rdev), (int)minor(st.st_rdev)); + e = path + strlen(path); + + dir = opendir(path); + if (!dir) + return -1; + while ((de = readdir(dir))) { + if (de->d_ino == 0) + continue; + if (de->d_name[0] == '.') + continue; + sprintf(e, "/%s/dev", de->d_name); + dfd = open(path, O_RDONLY); + if (dfd < 0) + continue; + n = read(dfd, buf, sizeof(buf)); + close(dfd); + if (n <= 0 || n >= sizeof(buf)) + continue; + buf[n] = 0; + if (sscanf(buf, "%d:%d", &major, &minor) != 2) + continue; + sprintf(buf, "%d:%d", major, minor); + dfd = dev_open(buf, O_RDONLY); + if (dfd >= 0) { + closedir(dir); + return dfd; + } + } + closedir(dir); + return -1; +} + +char *devnum2devname(int num) +{ + char name[100]; + if (num > 0) + sprintf(name, "md%d", num); + else + sprintf(name, "md_d%d", -1-num); + return strdup(name); +} + +int fd2devnum(int fd) +{ + struct stat stb; + if (fstat(fd, &stb) == 0 && + (S_IFMT&stb.st_mode)==S_IFBLK) { + if (major(stb.st_rdev) == MD_MAJOR) + return minor(stb.st_rdev); + else + return -1- (minor(stb.st_rdev)>>6); + } + return -1; +} + +int mdmon_running(int devnum) +{ + char path[100]; + char pid[10]; + int fd; + int n; + sprintf(path, "/var/run/mdadm/%s.pid", devnum2devname(devnum)); + fd = open(path, O_RDONLY, 0); + + if (fd < 0) + return 0; + n = read(fd, pid, 9); + close(fd); + if (n <= 0) + return 0; + if (kill(atoi(pid), 0) == 0) + return 1; + return 0; +} + +int signal_mdmon(int devnum) +{ + char path[100]; + char pid[10]; + int fd; + int n; + sprintf(path, "/var/run/mdadm/%s.pid", devnum2devname(devnum)); + fd = open(path, O_RDONLY, 0); + + if (fd < 0) + return 0; + n = read(fd, pid, 9); + close(fd); + if (n <= 0) + return 0; + if (kill(atoi(pid), SIGUSR1) == 0) + return 1; + return 0; +} + + #ifdef __TINYC__ /* tinyc doesn't optimize this check in ioctl.h out ... */