X-Git-Url: http://git.ipfire.org/?a=blobdiff_plain;f=super1.c;h=867aa58c0cf7875ce1d8526f19e54eafd7411710;hb=da827518c1f062e7d49433691d33e103525f9d6a;hp=f209775351ae480255b6533a5e15742b5bdfac62;hpb=4cce4069597491f5714fd81605f212ad2da62e06;p=thirdparty%2Fmdadm.git diff --git a/super1.c b/super1.c index f2097753..867aa58c 100644 --- a/super1.c +++ b/super1.c @@ -1,7 +1,7 @@ /* * mdadm - manage Linux "md" devices aka RAID arrays. * - * Copyright (C) 2001-2006 Neil Brown + * Copyright (C) 2001-2009 Neil Brown * * * This program is free software; you can redistribute it and/or modify @@ -19,12 +19,7 @@ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA * * Author: Neil Brown - * Email: - * Paper: Neil Brown - * School of Computer Science and Engineering - * The University of New South Wales - * Sydney, 2052 - * Australia + * Email: */ #include "mdadm.h" @@ -81,8 +76,8 @@ struct mdp_superblock_1 { __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. @@ -116,7 +111,6 @@ static unsigned int calc_sb_1_csum(struct mdp_superblock_1 * sb) 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 || @@ -128,7 +122,7 @@ static unsigned int calc_sb_1_csum(struct mdp_superblock_1 * sb) 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++; } @@ -141,13 +135,72 @@ static unsigned int calc_sb_1_csum(struct mdp_superblock_1 * sb) return __cpu_to_le32(csum); } +static char abuf[4096+4096]; +static int aread(int fd, 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; + char *b; + int n; + if (ioctl(fd, BLKSSZGET, &bsize) != 0 || + bsize <= len) + return read(fd, buf, len); + if (bsize > 4096) + return -1; + b = (char*)(((long)(abuf+4096))&~4095UL); + + n = read(fd, b, bsize); + if (n <= 0) + return n; + lseek(fd, len - n, 1); + if (n > len) + n = len; + memcpy(buf, b, n); + return n; +} + +static int awrite(int fd, 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; + char *b; + int n; + if (ioctl(fd, BLKSSZGET, &bsize) != 0 || + bsize <= len) + return write(fd, buf, len); + if (bsize > 4096) + return -1; + b = (char*)(((long)(abuf+4096))&~4095UL); + + n = read(fd, b, bsize); + if (n <= 0) + return n; + lseek(fd, -n, 1); + memcpy(b, buf, len); + n = write(fd, b, bsize); + if (n <= 0) + return n; + lseek(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; @@ -185,7 +238,7 @@ static void examine_super1(struct supertype *st, char *homehost) 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) { + if (__le32_to_cpu(sb->level) > 0) { int ddsks=0; switch(__le32_to_cpu(sb->level)) { case 1: ddsks=1;break; @@ -230,13 +283,11 @@ static void examine_super1(struct supertype *st, char *homehost) 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)); @@ -247,6 +298,10 @@ static void examine_super1(struct supertype *st, char *homehost) 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)); @@ -258,7 +313,7 @@ static void examine_super1(struct supertype *st, char *homehost) printf("\n"); } if (sb->devflags) { - printf(" Flags :"); + printf(" Flags :"); if (sb->devflags & WriteMostly1) printf(" write-mostly"); printf("\n"); @@ -278,6 +333,10 @@ static void examine_super1(struct supertype *st, char *homehost) 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 :"); @@ -298,6 +357,8 @@ static void examine_super1(struct supertype *st, char *homehost) 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) @@ -310,24 +371,33 @@ static void examine_super1(struct supertype *st, char *homehost) 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]); @@ -335,11 +405,13 @@ static void examine_super1(struct supertype *st, char *homehost) 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; @@ -355,17 +427,19 @@ static void brief_examine_super1(struct supertype *st) 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]); @@ -479,13 +553,15 @@ static void uuid_from_super1(struct supertype *st, int uuid[4]) 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; @@ -502,6 +578,8 @@ static void getinfo_super1(struct supertype *st, struct mdinfo *info) 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 = __le32_to_cpu(sb->bitmap_offset); info->disk.major = 0; info->disk.minor = 0; @@ -515,7 +593,7 @@ static void getinfo_super1(struct supertype *st, struct mdinfo *info) 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 */ @@ -524,6 +602,8 @@ static void getinfo_super1(struct supertype *st, struct mdinfo *info) 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; @@ -533,6 +613,11 @@ static void getinfo_super1(struct supertype *st, struct mdinfo *info) 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); @@ -540,25 +625,48 @@ static void getinfo_super1(struct supertype *st, struct mdinfo *info) 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; imax_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; @@ -570,35 +678,46 @@ static int update_super1(struct supertype *st, struct mdinfo *info, 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) @@ -610,11 +729,11 @@ static int update_super1(struct supertype *st, struct mdinfo *info, 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); @@ -637,17 +756,14 @@ static int update_super1(struct supertype *st, struct mdinfo *info, 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) { @@ -655,9 +771,10 @@ static int update_super1(struct supertype *st, struct mdinfo *info, bm = (struct bitmap_super_s*)(st->sb+1024); 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, ':'); @@ -666,8 +783,7 @@ static int update_super1(struct supertype *st, struct mdinfo *info, 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)); @@ -679,8 +795,7 @@ static int update_super1(struct supertype *st, struct mdinfo *info, 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 */ @@ -692,9 +807,14 @@ static int update_super1(struct supertype *st, struct mdinfo *info, 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; @@ -739,10 +859,8 @@ static int init_super1(struct supertype *st, mdu_array_info_t *info, 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); } @@ -776,7 +894,7 @@ static int init_super1(struct supertype *st, mdu_array_info_t *info, sb->utime = sb->ctime; sb->events = __cpu_to_le64(1); if (info->state & (1<resync_offset = ~0ULL; + sb->resync_offset = MaxSector; else sb->resync_offset = 0; sb->max_dev = __cpu_to_le32((1024- sizeof(struct mdp_superblock_1))/ @@ -810,7 +928,12 @@ static int add_to_super1(struct supertype *st, mdu_disk_info_t *dk, else *rp = 0xfffe; + if (dk->number >= (int)__le32_to_cpu(sb->max_dev) && + __le32_to_cpu(sb->max_dev) < 384) + 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; @@ -883,7 +1006,7 @@ static int store_super1(struct supertype *st, int fd) sbsize = sizeof(*sb) + 2 * __le32_to_cpu(sb->max_dev); sbsize = (sbsize+511)&(~511UL); - if (write(fd, sb, sbsize) != sbsize) + if (awrite(fd, sb, sbsize) != sbsize) return 4; if (sb->feature_map & __cpu_to_le32(MD_FEATURE_BITMAP_OFFSET)) { @@ -891,8 +1014,8 @@ static int store_super1(struct supertype *st, int fd) (((char*)sb)+1024); 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)) + if (awrite(fd, bm, sizeof(*bm)) != + sizeof(*bm)) return 5; } } @@ -906,6 +1029,8 @@ static unsigned long choose_bm_space(unsigned long devsize) { /* 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) @@ -915,17 +1040,20 @@ static unsigned long choose_bm_space(unsigned long devsize) 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; for (di = st->info; di && ! rv ; di = di->next) { if (di->disk.state == 1) @@ -933,27 +1061,28 @@ static int write_init_super1(struct supertype *st) 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<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) { @@ -966,8 +1095,9 @@ static int write_init_super1(struct supertype *st) 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; @@ -1000,16 +1130,23 @@ static int write_init_super1(struct supertype *st) 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; + + sb->data_offset = __cpu_to_le64(reserved); + sb->data_size = __cpu_to_le64(dsize - reserved); break; case 2: sb_offset = 4*2; @@ -1018,9 +1155,18 @@ static int write_init_super1(struct supertype *st) > 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; + + sb->data_offset = __cpu_to_le64(reserved); + sb->data_size = __cpu_to_le64(dsize - reserved); break; default: return -EINVAL; @@ -1086,8 +1232,6 @@ static int compare_super1(struct supertype *st, struct supertype *tst) return 0; } -static void free_super1(struct supertype *st); - static int load_super1(struct supertype *st, int fd, char *devname) { unsigned long long dsize; @@ -1099,9 +1243,6 @@ static int load_super1(struct supertype *st, int fd, char *devname) free_super1(st); - if (st->subarray[0]) - return 1; - if (st->ss == NULL || st->minor_version == -1) { int bestvers = -1; struct supertype tst; @@ -1189,7 +1330,7 @@ static int load_super1(struct supertype *st, int fd, char *devname) return 1; } - if (read(fd, super, 1024) != 1024) { + if (aread(fd, super, 1024) != 1024) { if (devname) fprintf(stderr, Name ": Cannot read superblock on %s\n", devname); @@ -1234,7 +1375,7 @@ static int load_super1(struct supertype *st, int fd, char *devname) * should get that written out. */ locate_bitmap1(st, fd); - if (read(fd, ((char*)super)+1024, 512) + if (aread(fd, ((char*)super)+1024, 512) != 512) goto no_bitmap; @@ -1245,7 +1386,8 @@ static int load_super1(struct supertype *st, int fd, char *devname) 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; } @@ -1256,6 +1398,7 @@ static struct supertype *match_metadata_desc1(char *arg) if (!st) return st; memset(st, 0, sizeof(*st)); + st->container_dev = NoMdDev; st->ss = &super1; st->max_devs = 384; st->sb = NULL; @@ -1268,11 +1411,15 @@ static struct supertype *match_metadata_desc1(char *arg) 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; @@ -1309,10 +1456,19 @@ static __u64 avail_size1(struct supertype *st, __u64 devsize) } #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 alignment of data_offset. + * We haven't access to device details so allow + * 1 Meg if bigger than 1Gig + */ + if (devsize > 1024*1024*2) + devsize -= 1024*2; + } 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)); @@ -1333,12 +1489,10 @@ add_internal_bitmap1(struct supertype *st, 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 !!! */ @@ -1347,33 +1501,33 @@ add_internal_bitmap1(struct supertype *st, 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); + 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; @@ -1384,17 +1538,17 @@ add_internal_bitmap1(struct supertype *st, 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; } } @@ -1415,26 +1569,37 @@ add_internal_bitmap1(struct supertype *st, 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->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); @@ -1444,7 +1609,6 @@ add_internal_bitmap1(struct supertype *st, return 1; } - static void locate_bitmap1(struct supertype *st, int fd) { unsigned long long offset; @@ -1472,8 +1636,7 @@ static int write_bitmap1(struct supertype *st, int fd) int rv = 0; int towrite, n; - char abuf[4096+512]; - char *buf = (char*)(((long)(abuf+512))&~511UL); + char buf[4096]; locate_bitmap1(st, fd); @@ -1488,7 +1651,7 @@ static int write_bitmap1(struct supertype *st, int fd) n = towrite; if (n > 4096) n = 4096; - n = write(fd, buf, n); + n = awrite(fd, buf, n); if (n > 0) towrite -= n; else @@ -1506,21 +1669,34 @@ static void free_super1(struct supertype *st) { 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; @@ -1558,6 +1734,7 @@ struct superswitch super1 = { .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, @@ -1574,5 +1751,5 @@ struct superswitch super1 = { #else .swapuuid = 1, #endif - .name = "1.0", + .name = "1.x", };