X-Git-Url: http://git.ipfire.org/?a=blobdiff_plain;f=super0.c;h=f3d0c07cc74750363c9a1b353d1a0f3a39b83053;hb=83366b33529567fb13a26773196a03a35b872c52;hp=60c51454bf80ae03a756491d19b6270ccaa229ef;hpb=fe77a154b1be42c65f15f0dad363f100de923489;p=thirdparty%2Fmdadm.git diff --git a/super0.c b/super0.c index 60c51454..f3d0c07c 100644 --- a/super0.c +++ b/super0.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: */ #define HAVE_STDINT_H 1 @@ -90,6 +85,7 @@ static void examine_super0(struct supertype *st, char *homehost) mdp_super_t *sb = st->sb; time_t atime; int d; + int delta_extra = 0; char *c; printf(" Magic : %08x\n", sb->md_magic); @@ -117,7 +113,7 @@ static void examine_super0(struct supertype *st, char *homehost) printf(" Creation Time : %.24s\n", ctime(&atime)); c=map_num(pers, sb->level); printf(" Raid Level : %s\n", c?c:"-unknown-"); - if ((int)sb->level >= 0) { + if ((int)sb->level > 0) { int ddsks=0; printf(" Used Dev Size : %d%s\n", sb->size, human_size((long long)sb->size<<10)); @@ -140,10 +136,9 @@ static void examine_super0(struct supertype *st, char *homehost) printf(" Reshape pos'n : %llu%s\n", (unsigned long long)sb->reshape_position/2, human_size((long long)sb->reshape_position<<9)); if (sb->delta_disks) { printf(" Delta Devices : %d", sb->delta_disks); - if (sb->delta_disks) - printf(" (%d->%d)\n", sb->raid_disks-sb->delta_disks, sb->raid_disks); - else - printf(" (%d->%d)\n", sb->raid_disks, sb->raid_disks+sb->delta_disks); + printf(" (%d->%d)\n", sb->raid_disks-sb->delta_disks, sb->raid_disks); + if (((int)sb->delta_disks) < 0) + delta_extra = - sb->delta_disks; } if (sb->new_level != sb->level) { c = map_num(pers, sb->new_level); @@ -215,7 +210,7 @@ static void examine_super0(struct supertype *st, char *homehost) } printf("\n"); printf(" Number Major Minor RaidDevice State\n"); - for (d= -1; d<(signed int)(sb->raid_disks+sb->spare_disks); d++) { + for (d= -1; d<(signed int)(sb->raid_disks+delta_extra + sb->spare_disks); d++) { mdp_disk_t *dp; char *dv; char nb[5]; @@ -318,9 +313,13 @@ static int match_home0(struct supertype *st, char *homehost) { mdp_super_t *sb = st->sb; char buf[20]; - char *hash = sha1_buffer(homehost, - strlen(homehost), - buf); + char *hash; + + if (!homehost) + return 0; + hash = sha1_buffer(homehost, + strlen(homehost), + buf); return (memcmp(&sb->set_uuid2, hash, 8)==0); } @@ -340,12 +339,14 @@ static void uuid_from_super0(struct supertype *st, int uuid[4]) } } -static void getinfo_super0(struct supertype *st, struct mdinfo *info) +static void getinfo_super0(struct supertype *st, struct mdinfo *info, char *map) { mdp_super_t *sb = st->sb; int working = 0; int i; + int map_disks = info->array.raid_disks; + memset(info, 0, sizeof(*info)); info->array.major_version = sb->major_version; info->array.minor_version = sb->minor_version; info->array.patch_version = sb->patch_version; @@ -373,6 +374,7 @@ static void getinfo_super0(struct supertype *st, struct mdinfo *info) uuid_from_super0(st, info->uuid); + info->recovery_start = MaxSector; if (sb->minor_version > 90 && (sb->reshape_position+1) != 0) { info->reshape_active = 1; info->reshape_progress = sb->reshape_position; @@ -380,30 +382,51 @@ static void getinfo_super0(struct supertype *st, struct mdinfo *info) info->delta_disks = sb->delta_disks; info->new_layout = sb->new_layout; info->new_chunk = sb->new_chunk; + if (info->delta_disks < 0) + info->array.raid_disks -= info->delta_disks; } else info->reshape_active = 0; + info->recovery_blocked = info->reshape_active; + sprintf(info->name, "%d", sb->md_minor); /* work_disks is calculated rather than read directly */ for (i=0; i < MD_SB_DISKS; i++) if ((sb->disks[i].state & (1<disks[i].raid_disk < info->array.raid_disks) && + (sb->disks[i].raid_disk < (unsigned)info->array.raid_disks) && (sb->disks[i].state & (1<disks[i].state & (1<disks[i].state & (1<array.working_disks = working; } +static struct mdinfo *container_content0(struct supertype *st, char *subarray) +{ + struct mdinfo *info; + + if (subarray) + return NULL; + + info = malloc(sizeof(*info)); + getinfo_super0(st, info, NULL); + return info; +} static int update_super0(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; + int uuid[4]; mdp_super_t *sb = st->sb; if (strcmp(update, "sparc2.2")==0 ) { /* 2.2 sparc put the events in the wrong place @@ -417,15 +440,13 @@ static int update_super0(struct supertype *st, struct mdinfo *info, if (verbose >= 0) fprintf (stderr, Name ": adjusting superblock of %s for 2.2/sparc compatability.\n", devname); - } - if (strcmp(update, "super-minor") ==0) { + } else if (strcmp(update, "super-minor") ==0) { sb->md_minor = info->array.md_minor; if (verbose > 0) fprintf(stderr, Name ": updating superblock of %s with minor number %d\n", devname, info->array.md_minor); - } - if (strcmp(update, "summaries") == 0) { - int i; + } else if (strcmp(update, "summaries") == 0) { + unsigned int i; /* set nr_disks, active_disks, working_disks, * failed_disks, spare_disks based on disks[] * array in superblock. @@ -451,8 +472,7 @@ static int update_super0(struct supertype *st, struct mdinfo *info, sb->spare_disks++; } else if (i >= sb->raid_disks && sb->disks[i].number == 0) sb->disks[i].state = 0; - } - if (strcmp(update, "force-one")==0) { + } else if (strcmp(update, "force-one")==0) { /* Not enough devices for a working array, so * bring this one up-to-date. */ @@ -462,8 +482,7 @@ static int update_super0(struct supertype *st, struct mdinfo *info, if (sb->events_hi != ehi || sb->events_lo != elo) rv = 1; - } - if (strcmp(update, "force-array")==0) { + } else if (strcmp(update, "force-array")==0) { /* degraded array and 'force' requested, so * maybe need to mark it 'clean' */ @@ -473,17 +492,36 @@ static int update_super0(struct supertype *st, struct mdinfo *info, sb->state |= (1 << MD_SB_CLEAN); rv = 1; } - } - if (strcmp(update, "assemble")==0) { + } else if (strcmp(update, "assemble")==0) { int d = info->disk.number; int wonly = sb->disks[d].state & (1<disks[d].state & ~(1<disk.state) { + int mask = (1<minor_version >= 91) + /* During reshape we don't insist on everything + * being marked 'sync' + */ + add = (1<disks[d].state & ~mask) | add) + != (unsigned)info->disk.state) { sb->disks[d].state = info->disk.state | wonly; rv = 1; } - } - if (strcmp(update, "linear-grow-new") == 0) { + if (info->reshape_active && + sb->minor_version > 90 && (sb->reshape_position+1) != 0 && + info->delta_disks >= 0 && + info->reshape_progress < sb->reshape_position) { + sb->reshape_position = info->reshape_progress; + rv = 1; + } + if (info->reshape_active && + sb->minor_version > 90 && (sb->reshape_position+1) != 0 && + info->delta_disks < 0 && + info->reshape_progress > sb->reshape_position) { + sb->reshape_position = info->reshape_progress; + rv = 1; + } + } else if (strcmp(update, "linear-grow-new") == 0) { memset(&sb->disks[info->disk.number], 0, sizeof(sb->disks[0])); sb->disks[info->disk.number].number = info->disk.number; sb->disks[info->disk.number].major = info->disk.major; @@ -491,8 +529,7 @@ static int update_super0(struct supertype *st, struct mdinfo *info, sb->disks[info->disk.number].raid_disk = info->disk.raid_disk; sb->disks[info->disk.number].state = info->disk.state; sb->this_disk = sb->disks[info->disk.number]; - } - if (strcmp(update, "linear-grow-update") == 0) { + } else if (strcmp(update, "linear-grow-update") == 0) { sb->raid_disks = info->array.raid_disks; sb->nr_disks = info->array.nr_disks; sb->active_disks = info->array.active_disks; @@ -503,20 +540,17 @@ static int update_super0(struct supertype *st, struct mdinfo *info, sb->disks[info->disk.number].minor = info->disk.minor; sb->disks[info->disk.number].raid_disk = info->disk.raid_disk; sb->disks[info->disk.number].state = info->disk.state; - } - if (strcmp(update, "resync") == 0) { + } else if (strcmp(update, "resync") == 0) { /* make sure resync happens */ sb->state &= ~(1<recovery_cp = 0; - } - if (strcmp(update, "homehost") == 0 && - homehost) { + } else if (strcmp(update, "homehost") == 0 && + homehost) { uuid_set = 0; update = "uuid"; info->uuid[0] = sb->set_uuid0; info->uuid[1] = sb->set_uuid1; - } - if (strcmp(update, "uuid") == 0) { + } else if (strcmp(update, "uuid") == 0) { if (!uuid_set && homehost) { char buf[20]; char *hash = sha1_buffer(homehost, @@ -531,11 +565,19 @@ static int update_super0(struct supertype *st, struct mdinfo *info, if (sb->state & (1<uuid); + uuid_from_super0(st, uuid); + memcpy(bm->uuid, uuid, 16); } - } - if (strcmp(update, "_reshape_progress")==0) + } else if (strcmp(update, "no-bitmap") == 0) { + sb->state &= ~(1<reshape_position = info->reshape_progress; + else if (strcmp(update, "writemostly")==0) + sb->state |= (1<state &= ~(1<sb_csum = calc_sb0_csum(sb); return rv; @@ -584,7 +626,7 @@ static int init_super0(struct supertype *st, mdu_array_info_t *info, sb->gvalid_words = 0; /* ignored */ sb->ctime = time(0); sb->level = info->level; - if (size != info->size) + if (size != (unsigned long long)info->size) return 0; sb->size = info->size; sb->nr_disks = info->nr_disks; @@ -652,6 +694,8 @@ static int add_to_super0(struct supertype *st, mdu_disk_info_t *dinfo, dk->minor = dinfo->minor; dk->raid_disk = dinfo->raid_disk; dk->state = dinfo->state; + /* In case our source disk was writemostly, don't copy that bit */ + dk->state &= ~(1<this_disk = sb->disks[dinfo->number]; sb->sb_csum = calc_sb0_csum(sb); @@ -717,8 +761,8 @@ static int write_init_super0(struct supertype *st) continue; if (di->fd == -1) continue; - Kill(di->devname, 0, 1, 1); - Kill(di->devname, 0, 1, 1); + while (Kill(di->devname, NULL, 0, 1, 1) == 0) + ; sb->disks[di->disk.number].state &= ~(1<devname); - close(di->fd); - di->fd = -1; } return rv; } @@ -804,9 +846,6 @@ static int load_super0(struct supertype *st, int fd, char *devname) free_super0(st); - if (st->subarray[0]) - return 1; - if (!get_dev_size(fd, devname, &dsize)) return 1; @@ -904,6 +943,7 @@ static struct supertype *match_metadata_desc0(char *arg) if (!st) return st; memset(st, 0, sizeof(*st)); + st->container_dev = NoMdDev; st->ss = &super0; st->info = NULL; st->minor_version = 90; @@ -913,9 +953,11 @@ static struct supertype *match_metadata_desc0(char *arg) while (arg[0] == '0' && arg[1] == '0') arg++; if (strcmp(arg, "0") == 0 || - strcmp(arg, "0.90") == 0 || +#ifdef DEFAULT_OLD_METADATA /* ifndef in super1.c */ strcmp(arg, "default") == 0 || - strcmp(arg, "") == 0 /* no metadata */ +#endif /* DEFAULT_OLD_METADATA */ + strcmp(arg, "0.90") == 0 || + strcmp(arg, "") == 0 /* no metadata - i.e. non_persistent */ ) return st; @@ -951,11 +993,12 @@ static int add_internal_bitmap0(struct supertype *st, int *chunkp, * size is in sectors, chunk is in bytes !!! */ unsigned long long bits; - unsigned long long max_bits = 60*1024*8; + unsigned long long max_bits = (60*1024 - sizeof(bitmap_super_t))*8; unsigned long long min_chunk; int chunk = *chunkp; mdp_super_t *sb = st->sb; bitmap_super_t *bms = (bitmap_super_t*)(((char*)sb) + MD_SB_BYTES); + int uuid[4]; min_chunk = 4096; /* sub-page chunks don't work yet.. */ @@ -964,9 +1007,14 @@ static int add_internal_bitmap0(struct supertype *st, int *chunkp, min_chunk *= 2; bits = (bits+1)/2; } - if (chunk == UnSet) + if (chunk == UnSet) { + /* A chunk size less than a few Megabytes gives poor + * performance without increasing resync noticeably + */ chunk = min_chunk; - else if (chunk < min_chunk) + if (chunk < 64*1024*1024) + chunk = 64*1024*1024; + } else if ((unsigned long long)chunk < min_chunk) return 0; /* chunk size too small */ sb->state |= (1<magic = __cpu_to_le32(BITMAP_MAGIC); bms->version = __cpu_to_le32(major); - uuid_from_super0(st, (int*)bms->uuid); + uuid_from_super0(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); @@ -1032,7 +1081,7 @@ static int write_bitmap0(struct supertype *st, int fd) memset(buf, 0xff, 4096); memcpy(buf, ((char*)sb)+MD_SB_BYTES, sizeof(bitmap_super_t)); - towrite = 64*1024; + towrite = 60*1024; while (towrite > 0) { n = towrite; if (n > 4096) @@ -1055,25 +1104,53 @@ static void free_super0(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_geometry0(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; + unsigned int tbmax = 4; - if (level == LEVEL_CONTAINER) + /* prior to linux 3.1, a but limits usable device size to 2TB. + * It was introduced in 2.6.29, but we won't worry about that detail + */ + if (get_linux_version() < 3001000) + tbmax = 2; + + if (level == LEVEL_CONTAINER) { + if (verbose) + fprintf(stderr, Name ": 0.90 metadata does not support containers\n"); return 0; - if (raiddisks > MD_SB_DISKS) + } + if (raiddisks > MD_SB_DISKS) { + if (verbose) + fprintf(stderr, Name ": 0.90 metadata supports at most %d devices per array\n", + MD_SB_DISKS); return 0; - if (size > (0x7fffffffULL<<9)) + } + if (size >= tbmax * 2ULL*1024*1024*1024) { + if (verbose) + fprintf(stderr, Name ": 0.90 metadata supports at most " + "%d terabytes per device\n", tbmax); return 0; + } + if (chunk && *chunk == UnSet) + *chunk = DEFAULT_CHUNK; + if (!subdev) return 1; @@ -1093,8 +1170,6 @@ static int validate_geometry0(struct supertype *st, int level, if (ldsize < MD_RESERVED_SECTORS * 512) return 0; - if (size > (0x7fffffffULL<<9)) - return 0; *freesize = MD_NEW_SIZE_SECTORS(ldsize >> 9); return 1; } @@ -1114,6 +1189,7 @@ struct superswitch super0 = { .match_home = match_home0, .uuid_from_super = uuid_from_super0, .getinfo_super = getinfo_super0, + .container_content = container_content0, .update_super = update_super0, .init_super = init_super0, .store_super = store_super0,