/*
* 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 0;
}
+static int is_member_busy(char *metadata_version)
+{
+ /* check if the given member array is active */
+ struct mdstat_ent *mdstat = mdstat_read(1, 0);
+ struct mdstat_ent *ent;
+ int busy = 0;
+
+ for (ent = mdstat; ent; ent = ent->next) {
+ if (ent->metadata_version == NULL)
+ continue;
+ if (strncmp(ent->metadata_version, "external:", 9) != 0)
+ continue;
+ if (!is_subarray(&ent->metadata_version[9]))
+ continue;
+ /* Skip first char - it can be '/' or '-' */
+ if (strcmp(&ent->metadata_version[10], metadata_version+1) == 0) {
+ busy = 1;
+ break;
+ }
+ }
+ free_mdstat(mdstat);
+
+ return busy;
+}
+
+static int ident_matches(struct mddev_ident *ident,
+ struct mdinfo *content,
+ struct supertype *tst,
+ char *homehost,
+ char *update, char *devname)
+{
+
+ if (ident->uuid_set && (!update || strcmp(update, "uuid")!= 0) &&
+ same_uuid(content->uuid, ident->uuid, tst->ss->swapuuid)==0) {
+ if (devname)
+ fprintf(stderr, Name ": %s has wrong uuid.\n",
+ devname);
+ return 0;
+ }
+ if (ident->name[0] && (!update || strcmp(update, "name")!= 0) &&
+ name_matches(content->name, ident->name, homehost)==0) {
+ if (devname)
+ fprintf(stderr, Name ": %s has wrong name.\n",
+ devname);
+ return 0;
+ }
+ if (ident->super_minor != UnSet &&
+ ident->super_minor != content->array.md_minor) {
+ if (devname)
+ fprintf(stderr, Name ": %s has wrong super-minor.\n",
+ devname);
+ return 0;
+ }
+ if (ident->level != UnSet &&
+ ident->level != content->array.level) {
+ if (devname)
+ fprintf(stderr, Name ": %s has wrong raid level.\n",
+ devname);
+ return 0;
+ }
+ if (ident->raid_disks != UnSet &&
+ ident->raid_disks!= content->array.raid_disks) {
+ if (devname)
+ fprintf(stderr, Name ": %s requires wrong number of drives.\n",
+ devname);
+ return 0;
+ }
+ if (ident->member && ident->member[0]) {
+ /* content->text_version must match */
+ char *s = strchr(content->text_version+1, '/');
+ if (s == NULL) {
+ if (devname)
+ fprintf(stderr, Name ": %s is not a container and one is required.\n",
+ devname);
+ return 0;
+ } else if (strcmp(ident->member, s+1) != 0) {
+ if (devname)
+ fprintf(stderr, Name ": skipping wrong member %s is %s\n",
+ content->text_version, devname);
+ return 0;
+ }
+ }
+ return 1;
+}
+
+
int Assemble(struct supertype *st, char *mddev,
- mddev_ident_t ident,
- mddev_dev_t devlist, char *backup_file,
+ struct mddev_ident *ident,
+ struct mddev_dev *devlist, char *backup_file,
int readonly, int runstop,
- char *update, char *homehost,
+ char *update, char *homehost, int require_homehost,
int verbose, int force)
{
/*
*/
int mdfd;
int clean;
- int auto_assem = (mddev == NULL);
+ int auto_assem = (mddev == NULL && !ident->uuid_set &&
+ ident->super_minor == UnSet && ident->name[0] == 0
+ && (ident->container == NULL || ident->member == NULL));
int old_linux = 0;
int vers = vers; /* Keep gcc quite - it really is initialised */
struct {
*/
struct mdinfo i;
} *devices;
+ char *devmap;
int *best = NULL; /* indexed by raid_disk */
- unsigned int bestcnt = 0;
+ int bestcnt = 0;
int devcnt = 0;
- unsigned int okcnt, sparecnt;
+ unsigned int okcnt, sparecnt, rebuilding_cnt;
unsigned int req_cnt;
- unsigned int i;
+ int i;
int most_recent = 0;
int chosen_drive;
int change = 0;
int inargv = 0;
+ int report_missmatch;
int bitmap_done;
int start_partial_ok = (runstop >= 0) &&
(force || devlist==NULL || auto_assem);
unsigned int num_devs;
- mddev_dev_t tmpdev;
+ struct mddev_dev *tmpdev;
struct mdinfo info;
- mdu_array_info_t tmp_inf;
+ struct mdinfo *content = NULL;
char *avail;
int nextspare = 0;
- char *name;
+ char *name = NULL;
int trustworthy;
char chosen_name[1024];
- memset(&info, 0, sizeof(info));
-
if (get_linux_version() < 2004000)
old_linux = 1;
if (!devlist &&
ident->uuid_set == 0 &&
- ident->super_minor < 0 &&
+ (ident->super_minor < 0 || ident->super_minor == UnSet) &&
+ ident->name[0] == 0 &&
+ (ident->container == NULL || ident->member == NULL) &&
ident->devices == NULL) {
fprintf(stderr, Name ": No identity information available for %s - cannot assemble.\n",
mddev ? mddev : "further assembly");
return 1;
}
- /* if the configuration specifies a container then we use that to
- * determine the devices and retrieve the array configuration
- */
-#ifndef MDASSEMBLE
- if (ident->container && ident->member) {
- int cfd = open(ident->container, O_RDWR);
- struct mdinfo *mdi;
- struct supertype container;
-
- if (verbose>0)
- fprintf(stderr, Name ": looking to assemble member array %s"
- " inside container %s\n", ident->member, ident->container);
- if (cfd < 0) {
- if (verbose>0)
- fprintf(stderr, Name ": unable to open container %s: %s\n",
- ident->container, strerror(errno));
- return 1;
- }
-
- mdi = sysfs_read(cfd, fd2devnum(cfd), GET_VERSION);
- if (!mdi) {
- close(cfd);
- if (verbose>0)
- fprintf(stderr, Name ": unable to read container %s\n",
- ident->container);
- return 1;
- }
- container.ss = find_metadata_methods(mdi->text_version);
- sysfs_free(mdi);
- if (!container.ss) {
- close(cfd);
- fprintf(stderr, Name ": %s uses unknown metadata: %s\n",
- ident->container, mdi->text_version);
- return 1;
- }
- if (container.ss->load_super(&container, cfd, ident->container)) {
- fprintf(stderr, Name ": Cannot load metadata for container %s\n",
- ident->container);
- return 1;
- }
-
- return Incremental_container(&container, ident->container,
- verbose, runstop, ident->autof,
- LOCAL);
- }
-#endif
if (devlist == NULL)
devlist = conf_get_devs();
else if (mddev)
inargv = 1;
+ report_missmatch = ((inargv && verbose >= 0) || verbose > 0);
try_again:
/* We come back here when doing auto-assembly and attempting some
* set of devices failed. Those are now marked as ->used==2 and
num_devs++;
tmpdev = tmpdev->next;
}
- devices = malloc(num_devs * sizeof(*devices));
if (!st && ident->st) st = ident->st;
*/
for (tmpdev = devlist;
tmpdev;
- tmpdev = tmpdev->next) {
+ tmpdev = tmpdev ? tmpdev->next : NULL) {
char *devname = tmpdev->devname;
int dfd;
struct stat stb;
struct supertype *tst = dup_super(st);
+ struct dev_policy *pol = NULL;
+ int found_container = 0;
if (tmpdev->used > 1) continue;
if (ident->devices &&
!match_oneof(ident->devices, devname)) {
- if ((inargv && verbose>=0) || verbose > 0)
+ if (report_missmatch)
fprintf(stderr, Name ": %s is not one of %s\n", devname, ident->devices);
continue;
}
dfd = dev_open(devname, O_RDONLY|O_EXCL);
if (dfd < 0) {
- if ((inargv && verbose >= 0) || verbose > 0)
+ if (report_missmatch)
fprintf(stderr, Name ": cannot open device %s: %s\n",
devname, strerror(errno));
tmpdev->used = 2;
fprintf(stderr, Name ": %s is not a block device.\n",
devname);
tmpdev->used = 2;
- } else if (!tst && (tst = guess_super(dfd)) == NULL) {
- if ((inargv && verbose >= 0) || verbose > 0)
- fprintf(stderr, Name ": no recogniseable superblock on %s\n",
- devname);
- tmpdev->used = 2;
- } else if (tst->ss->load_super(tst,dfd, NULL)) {
- if ((inargv && verbose >= 0) || verbose > 0)
- fprintf( stderr, Name ": no RAID superblock on %s\n",
- devname);
+ } else if (must_be_container(dfd)) {
+ if (st) {
+ /* already found some components, this cannot
+ * be another one.
+ */
+ if (report_missmatch)
+ fprintf(stderr, Name ": %s is a container, but we are looking for components\n",
+ devname);
+ tmpdev->used = 2;
+ } if (!tst && (tst = super_by_fd(dfd, NULL)) == NULL) {
+ if (report_missmatch)
+ fprintf(stderr, Name ": not a recognisable container: %s\n",
+ devname);
+ tmpdev->used = 2;
+ } else if (tst->ss->load_container(tst, dfd, NULL)) {
+ if (report_missmatch)
+ fprintf(stderr, Name ": no correct container type: %s\n",
+ devname);
+ tmpdev->used = 2;
+ } else if (auto_assem &&
+ !conf_test_metadata(tst->ss->name, (pol = devnum_policy(stb.st_rdev)),
+ tst->ss->match_home(tst, homehost) == 1)) {
+ if (report_missmatch)
+ fprintf(stderr, Name ": %s has metadata type %s for which "
+ "auto-assembly is disabled\n",
+ devname, tst->ss->name);
+ tmpdev->used = 2;
+ } else
+ found_container = 1;
} else {
- tst->ss->getinfo_super(tst, &info);
+ if (!tst && (tst = guess_super(dfd)) == NULL) {
+ if (report_missmatch)
+ fprintf(stderr, Name ": no recogniseable superblock on %s\n",
+ devname);
+ tmpdev->used = 2;
+ } else if (tst->ss->load_super(tst,dfd, NULL)) {
+ if (report_missmatch)
+ fprintf(stderr, Name ": no RAID superblock on %s\n",
+ devname);
+ tmpdev->used = 2;
+ } else if (tst->ss->compare_super == NULL) {
+ if (report_missmatch)
+ fprintf(stderr, Name ": Cannot assemble %s metadata on %s\n",
+ tst->ss->name, devname);
+ tmpdev->used = 2;
+ } else if (auto_assem && st == NULL &&
+ !conf_test_metadata(tst->ss->name, (pol = devnum_policy(stb.st_rdev)),
+ tst->ss->match_home(tst, homehost) == 1)) {
+ if (report_missmatch)
+ fprintf(stderr, Name ": %s has metadata type %s for which "
+ "auto-assembly is disabled\n",
+ devname, tst->ss->name);
+ tmpdev->used = 2;
+ }
}
if (dfd >= 0) close(dfd);
+ if (tmpdev->used == 2) {
+ if (auto_assem)
+ /* Ignore unrecognised devices during auto-assembly */
+ goto loop;
+ if (ident->uuid_set || ident->name[0] ||
+ ident->super_minor != UnSet)
+ /* Ignore unrecognised device if looking for
+ * specific array */
+ goto loop;
+
- if (ident->uuid_set && (!update || strcmp(update, "uuid")!= 0) &&
- (!tst || !tst->sb ||
- same_uuid(info.uuid, ident->uuid, tst->ss->swapuuid)==0)) {
- if ((inargv && verbose >= 0) || verbose > 0)
- fprintf(stderr, Name ": %s has wrong uuid.\n",
- devname);
- goto loop;
- }
- if (ident->name[0] && (!update || strcmp(update, "name")!= 0) &&
- (!tst || !tst->sb ||
- name_matches(info.name, ident->name, homehost)==0)) {
- if ((inargv && verbose >= 0) || verbose > 0)
- fprintf(stderr, Name ": %s has wrong name.\n",
- devname);
- goto loop;
- }
- if (ident->super_minor != UnSet &&
- (!tst || !tst->sb ||
- ident->super_minor != info.array.md_minor)) {
- if ((inargv && verbose >= 0) || verbose > 0)
- fprintf(stderr, Name ": %s has wrong super-minor.\n",
- devname);
- goto loop;
- }
- if (ident->level != UnSet &&
- (!tst || !tst->sb ||
- ident->level != info.array.level)) {
- if ((inargv && verbose >= 0) || verbose > 0)
- fprintf(stderr, Name ": %s has wrong raid level.\n",
- devname);
- goto loop;
- }
- if (ident->raid_disks != UnSet &&
- (!tst || !tst->sb ||
- ident->raid_disks!= info.array.raid_disks)) {
- if ((inargv && verbose >= 0) || verbose > 0)
- fprintf(stderr, Name ": %s requires wrong number of drives.\n",
- devname);
- goto loop;
- }
- if (auto_assem) {
- if (tst == NULL || tst->sb == NULL)
- continue;
- }
- /* If we are this far, then we are nearly commited to this device.
- * If the super_block doesn't exist, or doesn't match others,
- * then we probably cannot continue
- * However if one of the arrays is for the homehost, and
- * the other isn't that can disambiguate.
- */
-
- if (!tst || !tst->sb) {
fprintf(stderr, Name ": %s has no superblock - assembly aborted\n",
devname);
if (st)
st->ss->free_super(st);
+ dev_policy_free(pol);
return 1;
}
- if (st == NULL)
- st = dup_super(tst);
- if (st->minor_version == -1)
- st->minor_version = tst->minor_version;
- if (st->ss != tst->ss ||
- st->minor_version != tst->minor_version ||
- st->ss->compare_super(st, tst) != 0) {
- /* Some mismatch. If exactly one array matches this host,
- * we can resolve on that one.
- * Or, if we are auto assembling, we just ignore the second
- * for now.
+ if (found_container) {
+ /* tmpdev is a container. We need to be either
+ * looking for a member, or auto-assembling
*/
- if (auto_assem)
- goto loop;
- if (homehost) {
- int first = st->ss->match_home(st, homehost);
- int last = tst->ss->match_home(tst, homehost);
- if (first != last &&
- (first == 1 || last == 1)) {
- /* We can do something */
- if (first) {/* just ignore this one */
- if ((inargv && verbose >= 0) || verbose > 0)
- fprintf(stderr, Name ": %s misses out due to wrong homehost\n",
- devname);
- goto loop;
- } else { /* reject all those sofar */
- mddev_dev_t td;
- if ((inargv && verbose >= 0) || verbose > 0)
- fprintf(stderr, Name ": %s overrides previous devices due to good homehost\n",
+
+ if (ident->container) {
+ if (ident->container[0] == '/' &&
+ !same_dev(ident->container, devname)) {
+ if (report_missmatch)
+ fprintf(stderr, Name ": %s is not the container required (%s)\n",
+ devname, ident->container);
+ goto loop;
+ }
+ if (ident->container[0] != '/') {
+ /* we have a uuid */
+ int uuid[4];
+ if (!parse_uuid(ident->container, uuid) ||
+ !same_uuid(content->uuid, uuid, tst->ss->swapuuid)) {
+ if (report_missmatch)
+ fprintf(stderr, Name ": %s has wrong UUID to be required container\n",
devname);
- for (td=devlist; td != tmpdev; td=td->next)
- if (td->used == 1)
- td->used = 0;
- tmpdev->used = 1;
goto loop;
}
}
}
- fprintf(stderr, Name ": superblock on %s doesn't match others - assembly aborted\n",
- devname);
- tst->ss->free_super(tst);
- st->ss->free_super(st);
- return 1;
- }
+ /* It is worth looking inside this container.
+ */
+ if (verbose > 0)
+ fprintf(stderr, Name ": looking in container %s\n",
+ devname);
+
+ for (content = tst->ss->container_content(tst, NULL);
+ content;
+ content = content->next) {
+
+ if (!ident_matches(ident, content, tst,
+ homehost, update,
+ report_missmatch ? devname : NULL))
+ /* message already printed */;
+ else if (is_member_busy(content->text_version)) {
+ if (report_missmatch)
+ fprintf(stderr, Name ": member %s in %s is already assembled\n",
+ content->text_version,
+ devname);
+ } else
+ break;
+ }
+ if (!content) {
+ tmpdev->used = 2;
+ goto loop; /* empty container */
+ }
- tmpdev->used = 1;
+ st = tst; tst = NULL;
+ if (!auto_assem && inargv && tmpdev->next != NULL) {
+ fprintf(stderr, Name ": %s is a container, but is not "
+ "only device given: confused and aborting\n",
+ devname);
+ st->ss->free_super(st);
+ dev_policy_free(pol);
+ return 1;
+ }
+ if (verbose > 0)
+ fprintf(stderr, Name ": found match on member %s in %s\n",
+ content->text_version, devname);
+ /* make sure we finished the loop */
+ tmpdev = NULL;
+ goto loop;
+ } else {
+
+ content = &info;
+ memset(content, 0, sizeof(*content));
+ tst->ss->getinfo_super(tst, content, NULL);
+
+ if (!ident_matches(ident, content, tst,
+ homehost, update,
+ report_missmatch ? devname : NULL))
+ goto loop;
+
+ if (st == NULL)
+ st = dup_super(tst);
+ if (st->minor_version == -1)
+ st->minor_version = tst->minor_version;
+ if (st->ss != tst->ss ||
+ st->minor_version != tst->minor_version ||
+ st->ss->compare_super(st, tst) != 0) {
+ /* Some mismatch. If exactly one array matches this host,
+ * we can resolve on that one.
+ * Or, if we are auto assembling, we just ignore the second
+ * for now.
+ */
+ if (auto_assem)
+ goto loop;
+ if (homehost) {
+ int first = st->ss->match_home(st, homehost);
+ int last = tst->ss->match_home(tst, homehost);
+ if (first != last &&
+ (first == 1 || last == 1)) {
+ /* We can do something */
+ if (first) {/* just ignore this one */
+ if (report_missmatch)
+ fprintf(stderr, Name ": %s misses out due to wrong homehost\n",
+ devname);
+ goto loop;
+ } else { /* reject all those sofar */
+ struct mddev_dev *td;
+ if (report_missmatch)
+ fprintf(stderr, Name ": %s overrides previous devices due to good homehost\n",
+ devname);
+ for (td=devlist; td != tmpdev; td=td->next)
+ if (td->used == 1)
+ td->used = 0;
+ tmpdev->used = 1;
+ goto loop;
+ }
+ }
+ }
+ fprintf(stderr, Name ": superblock on %s doesn't match others - assembly aborted\n",
+ devname);
+ tst->ss->free_super(tst);
+ st->ss->free_super(st);
+ dev_policy_free(pol);
+ return 1;
+ }
+ tmpdev->used = 1;
+ }
loop:
+ dev_policy_free(pol);
+ pol = NULL;
if (tst)
tst->ss->free_super(tst);
}
- if (!st || !st->sb)
+ if (!st || !st->sb || !content)
return 2;
- /* Now need to open array the device. Use create_mddev */
- st->ss->getinfo_super(st, &info);
+ /* Now need to open the array device. Use create_mddev */
+ if (content == &info)
+ st->ss->getinfo_super(st, content, NULL);
trustworthy = FOREIGN;
- switch (st->ss->match_home(st, homehost)) {
- case 0:
- trustworthy = FOREIGN;
- name = info.name;
- break;
+ name = content->name;
+ switch (st->ss->match_home(st, homehost)
+ ?: st->ss->match_home(st, "any")) {
case 1:
trustworthy = LOCAL;
- name = strchr(info.name, ':');
+ name = strchr(content->name, ':');
if (name)
name++;
else
- name = info.name;
- break;
- case -1:
- if (info.name[0] == 0 && info.array.level == LEVEL_CONTAINER) {
- name = info.text_version;
- trustworthy = METADATA;
- } else
- trustworthy = FOREIGN;
+ name = content->name;
break;
}
+ if (!auto_assem)
+ /* If the array is listed in mdadm.conf or on
+ * command line, then we trust the name
+ * even if the array doesn't look local
+ */
+ trustworthy = LOCAL;
+
+ if (name[0] == 0 &&
+ content->array.level == LEVEL_CONTAINER) {
+ name = content->text_version;
+ trustworthy = METADATA;
+ }
+
+ if (name[0] && trustworthy != LOCAL &&
+ ! require_homehost &&
+ conf_name_is_free(name))
+ trustworthy = LOCAL;
+
+ if (trustworthy == LOCAL &&
+ strchr(name, ':'))
+ /* Ignore 'host:' prefix of name */
+ name = strchr(name, ':')+1;
+
mdfd = create_mddev(mddev, name, ident->autof, trustworthy,
chosen_name);
if (mdfd < 0) {
st->ss->free_super(st);
- free(devices);
if (auto_assem)
goto try_again;
return 1;
close(mdfd);
return 1;
}
- if (ioctl(mdfd, GET_ARRAY_INFO, &tmp_inf)==0) {
+ if (mddev_busy(fd2devnum(mdfd))) {
fprintf(stderr, Name ": %s already active, cannot restart it!\n",
mddev);
for (tmpdev = devlist ;
close(mdfd);
mdfd = -3;
st->ss->free_super(st);
- free(devices);
if (auto_assem)
goto try_again;
return 1;
}
ioctl(mdfd, STOP_ARRAY, NULL); /* just incase it was started but has no content */
+#ifndef MDASSEMBLE
+ if (content != &info) {
+ /* This is a member of a container. Try starting the array. */
+ return assemble_container_content(st, mdfd, content, runstop,
+ chosen_name, verbose);
+ }
+#endif
/* Ok, no bad inconsistancy, we can try updating etc */
bitmap_done = 0;
+ content->update_private = NULL;
+ devices = malloc(num_devs * sizeof(*devices));
+ devmap = calloc(num_devs * content->array.raid_disks, 1);
for (tmpdev = devlist; tmpdev; tmpdev=tmpdev->next) if (tmpdev->used == 1) {
char *devname = tmpdev->devname;
struct stat stb;
/* prepare useful information in info structures */
struct stat stb2;
struct supertype *tst;
+ int err;
fstat(mdfd, &stb2);
if (strcmp(update, "uuid")==0 &&
remove_partitions(dfd);
tst = dup_super(st);
- tst->ss->load_super(tst, dfd, NULL);
- tst->ss->getinfo_super(tst, &info);
-
- memcpy(info.uuid, ident->uuid, 16);
- strcpy(info.name, ident->name);
- info.array.md_minor = minor(stb2.st_rdev);
-
- tst->ss->update_super(tst, &info, update,
- devname, verbose,
- ident->uuid_set, homehost);
+ if (dfd < 0 || tst->ss->load_super(tst, dfd, NULL) != 0) {
+ fprintf(stderr, Name ": cannot re-read metadata from %s - aborting\n",
+ devname);
+ if (dfd >= 0)
+ close(dfd);
+ close(mdfd);
+ free(devices);
+ free(devmap);
+ return 1;
+ }
+ tst->ss->getinfo_super(tst, content, devmap + devcnt * content->array.raid_disks);
+
+ memcpy(content->uuid, ident->uuid, 16);
+ strcpy(content->name, ident->name);
+ content->array.md_minor = minor(stb2.st_rdev);
+
+ if (strcmp(update, "byteorder") == 0)
+ err = 0;
+ else
+ err = tst->ss->update_super(tst, content, update,
+ devname, verbose,
+ ident->uuid_set,
+ homehost);
+ if (err < 0) {
+ fprintf(stderr,
+ Name ": --update=%s not understood"
+ " for %s metadata\n",
+ update, tst->ss->name);
+ tst->ss->free_super(tst);
+ free(tst);
+ close(mdfd);
+ close(dfd);
+ free(devices);
+ free(devmap);
+ return 1;
+ }
if (strcmp(update, "uuid")==0 &&
!ident->uuid_set) {
ident->uuid_set = 1;
- memcpy(ident->uuid, info.uuid, 16);
+ memcpy(ident->uuid, content->uuid, 16);
}
- if (dfd < 0)
- fprintf(stderr, Name ": Cannot open %s for superblock update\n",
- devname);
- else if (tst->ss->store_super(tst, dfd))
+ if (tst->ss->store_super(tst, dfd))
fprintf(stderr, Name ": Could not re-write superblock on %s.\n",
devname);
- if (dfd >= 0)
- close(dfd);
+ close(dfd);
if (strcmp(update, "uuid")==0 &&
ident->bitmap_fd >= 0 && !bitmap_done) {
if (bitmap_update_uuid(ident->bitmap_fd,
- info.uuid,
+ content->uuid,
tst->ss->swapuuid) != 0)
fprintf(stderr, Name ": Could not update uuid on external bitmap.\n");
else
remove_partitions(dfd);
- tst->ss->load_super(tst, dfd, NULL);
- tst->ss->getinfo_super(tst, &info);
+ if (dfd < 0 || tst->ss->load_super(tst, dfd, NULL) != 0) {
+ fprintf(stderr, Name ": cannot re-read metadata from %s - aborting\n",
+ devname);
+ if (dfd >= 0)
+ close(dfd);
+ close(mdfd);
+ free(devices);
+ free(devmap);
+ return 1;
+ }
+ tst->ss->getinfo_super(tst, content, devmap + devcnt * content->array.raid_disks);
tst->ss->free_super(tst);
close(dfd);
}
if (verbose > 0)
fprintf(stderr, Name ": %s is identified as a member of %s, slot %d.\n",
- devname, mddev, info.disk.raid_disk);
+ devname, mddev, content->disk.raid_disk);
devices[devcnt].devname = devname;
devices[devcnt].uptodate = 0;
- devices[devcnt].i = info;
+ devices[devcnt].i = *content;
devices[devcnt].i.disk.major = major(stb.st_rdev);
devices[devcnt].i.disk.minor = minor(stb.st_rdev);
if (most_recent < devcnt) {
> devices[most_recent].i.events)
most_recent = devcnt;
}
- if (info.array.level == -4)
+ if (content->array.level == LEVEL_MULTIPATH)
/* with multipath, the raid_disk from the superblock is meaningless */
i = devcnt;
else
i = devices[devcnt].i.disk.raid_disk;
if (i+1 == 0) {
- if (nextspare < info.array.raid_disks)
- nextspare = info.array.raid_disks;
+ if (nextspare < content->array.raid_disks)
+ nextspare = content->array.raid_disks;
i = nextspare++;
} else {
- if (i >= info.array.raid_disks &&
+ if (i >= content->array.raid_disks &&
i >= nextspare)
nextspare = i+1;
}
if (i < 10000) {
if (i >= bestcnt) {
- unsigned int newbestcnt = i+10;
+ int newbestcnt = i+10;
int *newbest = malloc(sizeof(int)*newbestcnt);
- unsigned int c;
+ int c;
for (c=0; c < newbestcnt; c++)
if (c < bestcnt)
newbest[c] = best[c];
&& (devices[best[i]].i.disk.minor
!= devices[devcnt].i.disk.minor)
&& st->ss == &super0
- && info.array.level != LEVEL_MULTIPATH) {
+ && content->array.level != LEVEL_MULTIPATH) {
/* two different devices with identical superblock.
* Could be a mis-detection caused by overlapping
* partitions. fail-safe.
"the\n DEVICE list in mdadm.conf"
);
close(mdfd);
+ free(devices);
+ free(devmap);
return 1;
}
if (best[i] == -1
}
devcnt++;
}
+ free(content->update_private);
+ content->update_private = NULL;
if (devcnt == 0) {
fprintf(stderr, Name ": no devices found for %s\n",
if (st)
st->ss->free_super(st);
close(mdfd);
+ free(devices);
+ free(devmap);
return 1;
}
if (update && strcmp(update, "byteorder")==0)
st->minor_version = 90;
- st->ss->getinfo_super(st, &info);
- clean = info.array.state & 1;
+ st->ss->getinfo_super(st, content, NULL);
+ clean = content->array.state & 1;
/* now we have some devices that might be suitable.
* I wonder how many
*/
- avail = malloc(info.array.raid_disks);
- memset(avail, 0, info.array.raid_disks);
+ avail = malloc(content->array.raid_disks);
+ memset(avail, 0, content->array.raid_disks);
okcnt = 0;
sparecnt=0;
- for (i=0; i< bestcnt ;i++) {
+ rebuilding_cnt=0;
+ for (i=0; i< bestcnt; i++) {
int j = best[i];
int event_margin = 1; /* always allow a difference of '1'
* like the kernel does
/* note: we ignore error flags in multipath arrays
* as they don't make sense
*/
- if (info.array.level != -4)
- if (!(devices[j].i.disk.state & (1<<MD_DISK_SYNC))) {
+ if (content->array.level != LEVEL_MULTIPATH)
+ if (!(devices[j].i.disk.state & (1<<MD_DISK_ACTIVE))) {
if (!(devices[j].i.disk.state
& (1<<MD_DISK_FAULTY)))
sparecnt++;
continue;
}
+ /* If this devices thinks that 'most_recent' has failed, then
+ * we must reject this device.
+ */
+ if (j != most_recent &&
+ content->array.raid_disks > 0 &&
+ devices[most_recent].i.disk.raid_disk >= 0 &&
+ devmap[j * content->array.raid_disks + devices[most_recent].i.disk.raid_disk] == 0) {
+ if (verbose > -1)
+ fprintf(stderr, Name ": ignoring %s as it reports %s as failed\n",
+ devices[j].devname, devices[most_recent].devname);
+ best[i] = -1;
+ continue;
+ }
if (devices[j].i.events+event_margin >=
devices[most_recent].i.events) {
devices[j].uptodate = 1;
- if (i < info.array.raid_disks) {
- okcnt++;
- avail[i]=1;
+ if (i < content->array.raid_disks) {
+ if (devices[j].i.recovery_start == MaxSector) {
+ okcnt++;
+ avail[i]=1;
+ } else
+ rebuilding_cnt++;
} else
sparecnt++;
}
}
- while (force && !enough(info.array.level, info.array.raid_disks,
- info.array.layout, 1,
+ free(devmap);
+ while (force && !enough(content->array.level, content->array.raid_disks,
+ content->array.layout, 1,
avail, okcnt)) {
/* Choose the newest best drive which is
* not up-to-date, update the superblock
*/
int fd;
struct supertype *tst;
- long long current_events;
+ unsigned long long current_events;
chosen_drive = -1;
- for (i=0; i<info.array.raid_disks && i < bestcnt; i++) {
+ for (i = 0; i < content->array.raid_disks && i < bestcnt; i++) {
int j = best[i];
if (j>=0 &&
!devices[j].uptodate &&
- devices[j].i.events > 0 &&
+ devices[j].i.recovery_start == MaxSector &&
(chosen_drive < 0 ||
devices[j].i.events
> devices[chosen_drive].i.events))
devices[chosen_drive].i.events = 0;
continue;
}
- info.events = devices[most_recent].i.events;
- tst->ss->update_super(tst, &info, "force-one",
+ content->events = devices[most_recent].i.events;
+ tst->ss->update_super(tst, content, "force-one",
devices[chosen_drive].devname, verbose,
0, NULL);
/* If there are any other drives of the same vintage,
* add them in as well. We can't lose and we might gain
*/
- for (i=0; i<info.array.raid_disks && i < bestcnt ; i++) {
+ for (i = 0; i < content->array.raid_disks && i < bestcnt ; i++) {
int j = best[i];
if (j >= 0 &&
!devices[j].uptodate &&
- devices[j].i.events > 0 &&
devices[j].i.events == current_events) {
chosen_drive = j;
goto add_another;
fprintf(stderr, Name ": Cannot open %s: %s\n",
devices[j].devname, strerror(errno));
close(mdfd);
+ free(devices);
return 1;
}
if (st->ss->load_super(st,fd, NULL)) {
fprintf(stderr, Name ": RAID superblock has disappeared from %s\n",
devices[j].devname);
close(mdfd);
+ free(devices);
return 1;
}
close(fd);
if (st->sb == NULL) {
fprintf(stderr, Name ": No suitable drives found for %s\n", mddev);
close(mdfd);
+ free(devices);
return 1;
}
- st->ss->getinfo_super(st, &info);
+ st->ss->getinfo_super(st, content, NULL);
#ifndef MDASSEMBLE
- sysfs_init(&info, mdfd, 0);
+ sysfs_init(content, mdfd, 0);
#endif
for (i=0; i<bestcnt; i++) {
int j = best[i];
unsigned int desired_state;
- if (i < info.array.raid_disks)
+ if (i < content->array.raid_disks)
desired_state = (1<<MD_DISK_ACTIVE) | (1<<MD_DISK_SYNC);
else
desired_state = 0;
continue;
devices[j].i.disk.state = desired_state;
+ if (!(devices[j].i.array.state & 1))
+ clean = 0;
if (st->ss->update_super(st, &devices[j].i, "assemble", NULL,
verbose, 0, NULL)) {
#endif
}
if (force && !clean &&
- !enough(info.array.level, info.array.raid_disks,
- info.array.layout, clean,
+ !enough(content->array.level, content->array.raid_disks,
+ content->array.layout, clean,
avail, okcnt)) {
- change += st->ss->update_super(st, &info, "force-array",
+ change += st->ss->update_super(st, content, "force-array",
devices[chosen_drive].devname, verbose,
0, NULL);
clean = 1;
fprintf(stderr, Name ": Could not open %s for write - cannot Assemble array.\n",
devices[chosen_drive].devname);
close(mdfd);
+ free(devices);
return 1;
}
if (st->ss->store_super(st, fd)) {
fprintf(stderr, Name ": Could not re-write superblock on %s\n",
devices[chosen_drive].devname);
close(mdfd);
+ free(devices);
return 1;
}
close(fd);
* The code of doing this lives in Grow.c
*/
#ifndef MDASSEMBLE
- if (info.reshape_active) {
+ if (content->reshape_active) {
int err = 0;
int *fdlist = malloc(sizeof(int)* bestcnt);
+ if (verbose > 0)
+ fprintf(stderr, Name ":%s has an active reshape - checking "
+ "if critical section needs to be restored\n",
+ chosen_name);
for (i=0; i<bestcnt; i++) {
int j = best[i];
if (j >= 0) {
fdlist[i] = -1;
}
if (!err)
- err = Grow_restart(st, &info, fdlist, bestcnt, backup_file);
+ err = Grow_restart(st, content, fdlist, bestcnt, backup_file, verbose > 0);
while (i>0) {
i--;
if (fdlist[i]>=0) close(fdlist[i]);
}
if (err) {
fprintf(stderr, Name ": Failed to restore critical section for reshape, sorry.\n");
+ if (backup_file == NULL)
+ fprintf(stderr," Possibly you needed to specify the --backup-file\n");
close(mdfd);
+ free(devices);
return err;
}
}
/* count number of in-sync devices according to the superblock.
* We must have this number to start the array without -s or -R
*/
- req_cnt = info.array.working_disks;
+ req_cnt = content->array.working_disks;
/* Almost ready to actually *do* something */
if (!old_linux) {
/* First, fill in the map, so that udev can find our name
* as soon as we become active.
*/
- map_update(NULL, fd2devnum(mdfd), info.text_version,
- info.uuid, chosen_name);
+ map_update(NULL, fd2devnum(mdfd), content->text_version,
+ content->uuid, chosen_name);
- rv = set_array_info(mdfd, st, &info);
+ rv = set_array_info(mdfd, st, content);
if (rv) {
fprintf(stderr, Name ": failed to set array info for %s: %s\n",
mddev, strerror(errno));
+ ioctl(mdfd, STOP_ARRAY, NULL);
close(mdfd);
+ free(devices);
return 1;
}
if (ident->bitmap_fd >= 0) {
if (ioctl(mdfd, SET_BITMAP_FILE, ident->bitmap_fd) != 0) {
fprintf(stderr, Name ": SET_BITMAP_FILE failed.\n");
+ ioctl(mdfd, STOP_ARRAY, NULL);
close(mdfd);
+ free(devices);
return 1;
}
} else if (ident->bitmap_file) {
if (bmfd < 0) {
fprintf(stderr, Name ": Could not open bitmap file %s\n",
ident->bitmap_file);
+ ioctl(mdfd, STOP_ARRAY, NULL);
close(mdfd);
+ free(devices);
return 1;
}
if (ioctl(mdfd, SET_BITMAP_FILE, bmfd) != 0) {
fprintf(stderr, Name ": Failed to set bitmapfile for %s\n", mddev);
close(bmfd);
+ ioctl(mdfd, STOP_ARRAY, NULL);
close(mdfd);
+ free(devices);
return 1;
}
close(bmfd);
j = chosen_drive;
if (j >= 0 /* && devices[j].uptodate */) {
- rv = add_disk(mdfd, st, &info, &devices[j].i);
+ rv = add_disk(mdfd, st, content, &devices[j].i);
if (rv) {
fprintf(stderr, Name ": failed to add "
devices[j].devname,
mddev,
strerror(errno));
- if (i < info.array.raid_disks
+ if (i < content->array.raid_disks
|| i == bestcnt)
okcnt--;
else
"to %s as %d\n",
devices[j].devname, mddev,
devices[j].i.disk.raid_disk);
- } else if (verbose > 0 && i < info.array.raid_disks)
+ } else if (verbose > 0 && i < content->array.raid_disks)
fprintf(stderr, Name ": no uptodate device for "
"slot %d of %s\n",
i, mddev);
}
- if (info.array.level == LEVEL_CONTAINER) {
+ if (content->array.level == LEVEL_CONTAINER) {
if (verbose >= 0) {
fprintf(stderr, Name ": Container %s has been "
"assembled with %d drive%s",
mddev, okcnt+sparecnt, okcnt+sparecnt==1?"":"s");
- if (okcnt < info.array.raid_disks)
+ if (okcnt < (unsigned)content->array.raid_disks)
fprintf(stderr, " (out of %d)",
- info.array.raid_disks);
+ content->array.raid_disks);
fprintf(stderr, "\n");
}
+ sysfs_uevent(content, "change");
+ wait_for(chosen_name, mdfd);
close(mdfd);
+ free(devices);
return 0;
}
if (runstop == 1 ||
(runstop <= 0 &&
- ( enough(info.array.level, info.array.raid_disks,
- info.array.layout, clean, avail, okcnt) &&
- (okcnt >= req_cnt || start_partial_ok)
+ ( enough(content->array.level, content->array.raid_disks,
+ content->array.layout, clean, avail, okcnt) &&
+ (okcnt + rebuilding_cnt >= req_cnt || start_partial_ok)
))) {
- if (ioctl(mdfd, RUN_ARRAY, NULL)==0) {
+ /* This array is good-to-go.
+ * If a reshape is in progress then we might need to
+ * continue monitoring it. In that case we start
+ * it read-only and let the grow code make it writable.
+ */
+ int rv;
+#ifndef MDASSEMBLE
+ if (content->reshape_active &&
+ content->delta_disks <= 0)
+ rv = Grow_continue(mdfd, st, content, backup_file);
+ else
+#endif
+ rv = ioctl(mdfd, RUN_ARRAY, NULL);
+ if (rv == 0) {
if (verbose >= 0) {
fprintf(stderr, Name ": %s has been started with %d drive%s",
mddev, okcnt, okcnt==1?"":"s");
- if (okcnt < info.array.raid_disks)
- fprintf(stderr, " (out of %d)", info.array.raid_disks);
+ if (okcnt < (unsigned)content->array.raid_disks)
+ fprintf(stderr, " (out of %d)", content->array.raid_disks);
+ if (rebuilding_cnt)
+ fprintf(stderr, "%s %d rebuilding", sparecnt?",":" and", rebuilding_cnt);
if (sparecnt)
fprintf(stderr, " and %d spare%s", sparecnt, sparecnt==1?"":"s");
fprintf(stderr, ".\n");
}
+ if (content->reshape_active &&
+ content->array.level >= 4 &&
+ content->array.level <= 6) {
+ /* might need to increase the size
+ * of the stripe cache - default is 256
+ */
+ if (256 < 4 * (content->array.chunk_size/4096)) {
+ struct mdinfo *sra = sysfs_read(mdfd, 0, 0);
+ if (sra)
+ sysfs_set_num(sra, NULL,
+ "stripe_cache_size",
+ (4 * content->array.chunk_size / 4096) + 1);
+ }
+ }
+ if (okcnt < (unsigned)content->array.raid_disks) {
+ /* If any devices did not get added
+ * because the kernel rejected them based
+ * on event count, try adding them
+ * again providing the action policy is
+ * 're-add' or greater. The bitmap
+ * might allow them to be included, or
+ * they will become spares.
+ */
+ for (i = 0; i <= bestcnt; i++) {
+ int j = best[i];
+ if (j >= 0 && !devices[j].uptodate) {
+ if (!disk_action_allows(&devices[j].i, st->ss->name, act_re_add))
+ continue;
+ rv = add_disk(mdfd, st, content,
+ &devices[j].i);
+ if (rv == 0 && verbose >= 0)
+ fprintf(stderr,
+ Name ": %s has been re-added.\n",
+ devices[j].devname);
+ }
+ }
+ }
+ wait_for(mddev, mdfd);
close(mdfd);
if (auto_assem) {
int usecs = 1;
usecs <<= 1;
}
}
+ free(devices);
return 0;
}
fprintf(stderr, Name ": failed to RUN_ARRAY %s: %s\n",
mddev, strerror(errno));
- if (!enough(info.array.level, info.array.raid_disks,
- info.array.layout, 1, avail, okcnt))
+ if (!enough(content->array.level, content->array.raid_disks,
+ content->array.layout, 1, avail, okcnt))
fprintf(stderr, Name ": Not enough devices to "
"start the array.\n");
- else if (!enough(info.array.level,
- info.array.raid_disks,
- info.array.layout, clean,
+ else if (!enough(content->array.level,
+ content->array.raid_disks,
+ content->array.layout, clean,
avail, okcnt))
fprintf(stderr, Name ": Not enough devices to "
"start the array while not clean "
if (auto_assem)
ioctl(mdfd, STOP_ARRAY, NULL);
close(mdfd);
+ free(devices);
return 1;
}
if (runstop == -1) {
fprintf(stderr, Name ": %s assembled from %d drive%s",
mddev, okcnt, okcnt==1?"":"s");
- if (okcnt != info.array.raid_disks)
- fprintf(stderr, " (out of %d)", info.array.raid_disks);
+ if (okcnt != (unsigned)content->array.raid_disks)
+ fprintf(stderr, " (out of %d)", content->array.raid_disks);
fprintf(stderr, ", but not started.\n");
close(mdfd);
+ free(devices);
return 0;
}
if (verbose >= -1) {
fprintf(stderr, Name ": %s assembled from %d drive%s", mddev, okcnt, okcnt==1?"":"s");
+ if (rebuilding_cnt)
+ fprintf(stderr, "%s %d rebuilding", sparecnt?", ":" and ", rebuilding_cnt);
if (sparecnt)
fprintf(stderr, " and %d spare%s", sparecnt, sparecnt==1?"":"s");
- if (!enough(info.array.level, info.array.raid_disks,
- info.array.layout, 1, avail, okcnt))
+ if (!enough(content->array.level, content->array.raid_disks,
+ content->array.layout, 1, avail, okcnt))
fprintf(stderr, " - not enough to start the array.\n");
- else if (!enough(info.array.level,
- info.array.raid_disks,
- info.array.layout, clean,
+ else if (!enough(content->array.level,
+ content->array.raid_disks,
+ content->array.layout, clean,
avail, okcnt))
fprintf(stderr, " - not enough to start the "
"array while not clean - consider "
"--force.\n");
else {
- if (req_cnt == info.array.raid_disks)
+ if (req_cnt == (unsigned)content->array.raid_disks)
fprintf(stderr, " - need all %d to start it", req_cnt);
else
- fprintf(stderr, " - need %d of %d to start", req_cnt, info.array.raid_disks);
+ fprintf(stderr, " - need %d of %d to start", req_cnt, content->array.raid_disks);
fprintf(stderr, " (use --run to insist).\n");
}
}
if (auto_assem)
ioctl(mdfd, STOP_ARRAY, NULL);
+ close(mdfd);
+ free(devices);
return 1;
} else {
/* The "chosen_drive" is a good choice, and if necessary, the superblock has
}
close(mdfd);
+ free(devices);
return 0;
}
+
+#ifndef MDASSEMBLE
+int assemble_container_content(struct supertype *st, int mdfd,
+ struct mdinfo *content, int runstop,
+ char *chosen_name, int verbose)
+{
+ struct mdinfo *dev, *sra;
+ int working = 0, preexist = 0;
+ struct map_ent *map = NULL;
+
+ sysfs_init(content, mdfd, 0);
+
+ sra = sysfs_read(mdfd, 0, GET_VERSION);
+ if (sra == NULL || strcmp(sra->text_version, content->text_version) != 0)
+ if (sysfs_set_array(content, md_get_version(mdfd)) != 0) {
+ close(mdfd);
+ return 1;
+ }
+ if (sra)
+ sysfs_free(sra);
+
+ for (dev = content->devs; dev; dev = dev->next)
+ if (sysfs_add_disk(content, dev, 1) == 0)
+ working++;
+ else if (errno == EEXIST)
+ preexist++;
+ if (working == 0) {
+ close(mdfd);
+ return 1;/* Nothing new, don't try to start */
+ }
+
+ map_update(&map, fd2devnum(mdfd),
+ content->text_version,
+ content->uuid, chosen_name);
+
+ if (runstop > 0 ||
+ (working + preexist) >= content->array.working_disks) {
+ int err;
+
+ switch(content->array.level) {
+ case LEVEL_LINEAR:
+ case LEVEL_MULTIPATH:
+ case 0:
+ err = sysfs_set_str(content, NULL, "array_state",
+ "active");
+ break;
+ default:
+ err = sysfs_set_str(content, NULL, "array_state",
+ "readonly");
+ /* start mdmon if needed. */
+ if (!err) {
+ if (!mdmon_running(st->container_dev))
+ start_mdmon(st->container_dev);
+ ping_monitor(devnum2devname(st->container_dev));
+ }
+ break;
+ }
+ if (!err)
+ sysfs_set_safemode(content, content->safe_mode_delay);
+ if (verbose >= 0) {
+ if (err)
+ fprintf(stderr, Name
+ ": array %s now has %d devices",
+ chosen_name, working + preexist);
+ else
+ fprintf(stderr, Name
+ ": Started %s with %d devices",
+ chosen_name, working + preexist);
+ if (preexist)
+ fprintf(stderr, " (%d new)", working);
+ fprintf(stderr, "\n");
+ }
+ if (!err)
+ wait_for(chosen_name, mdfd);
+ close(mdfd);
+ return 0;
+ /* FIXME should have an O_EXCL and wait for read-auto */
+ } else {
+ if (verbose >= 0)
+ fprintf(stderr, Name
+ ": %s assembled with %d devices but "
+ "not started\n",
+ chosen_name, working);
+ close(mdfd);
+ return 1;
+ }
+}
+#endif
+