* Incremental.c - support --incremental. Part of:
* mdadm - manage Linux "md" devices aka RAID arrays.
*
- * Copyright (C) 2006-2009 Neil Brown <neilb@suse.de>
+ * Copyright (C) 2006-2012 Neil Brown <neilb@suse.de>
*
*
* This program is free software; you can redistribute it and/or modify
*/
#include "mdadm.h"
+#include <dirent.h>
+#include <ctype.h>
-static int count_active(struct supertype *st, int mdfd, char **availp,
+static int count_active(struct supertype *st, struct mdinfo *sra,
+ int mdfd, char **availp,
struct mdinfo *info);
static void find_reject(int mdfd, struct supertype *st, struct mdinfo *sra,
int number, __u64 events, int verbose,
char *array_name);
static int try_spare(char *devname, int *dfdp, struct dev_policy *pol,
+ struct map_ent *target,
struct supertype *st, int verbose);
-int Incremental(char *devname, int verbose, int runstop,
- struct supertype *st, char *homehost, int require_homehost,
- int autof)
+static int Incremental_container(struct supertype *st, char *devname,
+ struct context *c);
+
+int Incremental(char *devname, struct context *c,
+ struct supertype *st)
{
/* Add this device to an array, creating the array if necessary
* and starting the array if sensible or - if runstop>0 - if possible.
*/
struct stat stb;
struct mdinfo info, dinfo;
- struct mddev_ident_s *array_list, *match;
+ struct mdinfo *sra = NULL, *d;
+ struct mddev_ident *match;
char chosen_name[1024];
int rv = 1;
struct map_ent *mp, *map = NULL;
int dfd = -1, mdfd = -1;
- char *avail;
+ char *avail = NULL;
int active_disks;
- int trustworthy = FOREIGN;
+ int trustworthy;
char *name_to_use;
mdu_array_info_t ainf;
struct dev_policy *policy = NULL;
+ struct map_ent target_array;
+ int have_target;
struct createinfo *ci = conf_get_create_info();
+ if (stat(devname, &stb) < 0) {
+ if (c->verbose >= 0)
+ pr_err("stat failed for %s: %s.\n",
+ devname, strerror(errno));
+ return rv;
+ }
+ if ((stb.st_mode & S_IFMT) != S_IFBLK) {
+ if (c->verbose >= 0)
+ pr_err("%s is not a block device.\n",
+ devname);
+ return rv;
+ }
+ dfd = dev_open(devname, O_RDONLY);
+ if (dfd < 0) {
+ if (c->verbose >= 0)
+ pr_err("cannot open %s: %s.\n",
+ devname, strerror(errno));
+ return rv;
+ }
+ /* If the device is a container, we do something very different */
+ if (must_be_container(dfd)) {
+ if (!st)
+ st = super_by_fd(dfd, NULL);
+ if (st)
+ st->ignore_hw_compat = 1;
+ if (st && st->ss->load_container)
+ rv = st->ss->load_container(st, dfd, NULL);
+
+ close(dfd);
+ if (!rv && st->ss->container_content) {
+ if (map_lock(&map))
+ pr_err("failed to get "
+ "exclusive lock on mapfile\n");
+ rv = Incremental_container(st, devname, c);
+ map_unlock(&map);
+ return rv;
+ }
+
+ pr_err("%s is not part of an md array.\n",
+ devname);
+ return rv;
+ }
/* 1/ Check if device is permitted by mdadm.conf */
if (!conf_test_dev(devname)) {
- if (verbose >= 0)
- fprintf(stderr, Name
- ": %s not permitted by mdadm.conf.\n",
- devname);
+ if (c->verbose >= 0)
+ pr_err("%s not permitted by mdadm.conf.\n",
+ devname);
goto out;
}
/* 2/ Find metadata, reject if none appropriate (check
* version/name from args) */
- dfd = dev_open(devname, O_RDONLY|O_EXCL);
- if (dfd < 0) {
- if (verbose >= 0)
- fprintf(stderr, Name ": cannot open %s: %s.\n",
- devname, strerror(errno));
- goto out;
- }
if (fstat(dfd, &stb) < 0) {
- if (verbose >= 0)
- fprintf(stderr, Name ": fstat failed for %s: %s.\n",
+ if (c->verbose >= 0)
+ pr_err("fstat failed for %s: %s.\n",
devname, strerror(errno));
goto out;
}
if ((stb.st_mode & S_IFMT) != S_IFBLK) {
- if (verbose >= 0)
- fprintf(stderr, Name ": %s is not a block device.\n",
+ if (c->verbose >= 0)
+ pr_err("%s is not a block device.\n",
devname);
goto out;
}
dinfo.disk.minor = minor(stb.st_rdev);
policy = disk_policy(&dinfo);
+ have_target = policy_check_path(&dinfo, &target_array);
if (st == NULL && (st = guess_super(dfd)) == NULL) {
- if (verbose >= 0)
- fprintf(stderr, Name
- ": no recognisable superblock on %s.\n",
- devname);
- rv = try_spare(devname, &dfd, policy, st, verbose);
+ if (c->verbose >= 0)
+ pr_err("no recognisable superblock on %s.\n",
+ devname);
+ rv = try_spare(devname, &dfd, policy,
+ have_target ? &target_array : NULL,
+ st, c->verbose);
goto out;
}
+ st->ignore_hw_compat = 1;
if (st->ss->compare_super == NULL ||
st->ss->load_super(st, dfd, NULL)) {
- if (verbose >= 0)
- fprintf(stderr, Name ": no RAID superblock on %s.\n",
+ if (c->verbose >= 0)
+ pr_err("no RAID superblock on %s.\n",
devname);
- rv = try_spare(devname, &dfd, policy, st, verbose);
+ rv = try_spare(devname, &dfd, policy,
+ have_target ? &target_array : NULL,
+ st, c->verbose);
free(st);
goto out;
}
close (dfd); dfd = -1;
- memset(&info, 0, sizeof(info));
- st->ss->getinfo_super(st, &info);
- /* 3/ Check if there is a match in mdadm.conf */
-
- array_list = conf_get_ident(NULL);
- match = NULL;
- for (; array_list; array_list = array_list->next) {
- if (array_list->uuid_set &&
- same_uuid(array_list->uuid, info.uuid, st->ss->swapuuid)
- == 0) {
- if (verbose >= 2 && array_list->devname)
- fprintf(stderr, Name
- ": UUID differs from %s.\n",
- array_list->devname);
- continue;
- }
- if (array_list->name[0] &&
- strcasecmp(array_list->name, info.name) != 0) {
- if (verbose >= 2 && array_list->devname)
- fprintf(stderr, Name
- ": Name differs from %s.\n",
- array_list->devname);
- continue;
- }
- if (array_list->devices &&
- !match_oneof(array_list->devices, devname)) {
- if (verbose >= 2 && array_list->devname)
- fprintf(stderr, Name
- ": Not a listed device for %s.\n",
- array_list->devname);
- continue;
- }
- if (array_list->super_minor != UnSet &&
- array_list->super_minor != info.array.md_minor) {
- if (verbose >= 2 && array_list->devname)
- fprintf(stderr, Name
- ": Different super-minor to %s.\n",
- array_list->devname);
- continue;
- }
- if (!array_list->uuid_set &&
- !array_list->name[0] &&
- !array_list->devices &&
- array_list->super_minor == UnSet) {
- if (verbose >= 2 && array_list->devname)
- fprintf(stderr, Name
- ": %s doesn't have any identifying information.\n",
- array_list->devname);
- continue;
- }
- /* FIXME, should I check raid_disks and level too?? */
+ st->ss->getinfo_super(st, &info, NULL);
- if (match) {
- if (verbose >= 0) {
- if (match->devname && array_list->devname)
- fprintf(stderr, Name
- ": we match both %s and %s - cannot decide which to use.\n",
- match->devname, array_list->devname);
- else
- fprintf(stderr, Name
- ": multiple lines in mdadm.conf match\n");
- }
- rv = 2;
- goto out;
- }
- match = array_list;
- }
+ /* 3/ Check if there is a match in mdadm.conf */
+ match = conf_match(st, &info, devname, c->verbose, &rv);
+ if (!match && rv == 2)
+ goto out;
if (match && match->devname
&& strcasecmp(match->devname, "<ignore>") == 0) {
- if (verbose >= 0)
- fprintf(stderr, Name ": array containing %s is explicitly"
+ if (c->verbose >= 0)
+ pr_err("array containing %s is explicitly"
" ignored by mdadm.conf\n",
devname);
goto out;
* on that. */
if (match)
trustworthy = LOCAL;
- else if (st->ss->match_home(st, homehost) == 1)
+ else if (st->ss->match_home(st, c->homehost) == 1)
trustworthy = LOCAL;
else if (st->ss->match_home(st, "any") == 1)
trustworthy = LOCAL_ANY;
else
trustworthy = FOREIGN;
-
- if (!match && !conf_test_metadata(st->ss->name,
+ if (!match && !conf_test_metadata(st->ss->name, policy,
(trustworthy == LOCAL))) {
- if (verbose >= 1)
- fprintf(stderr, Name
- ": %s has metadata type %s for which "
- "auto-assembly is disabled\n",
- devname, st->ss->name);
+ if (c->verbose >= 1)
+ pr_err("%s has metadata type %s for which "
+ "auto-assembly is disabled\n",
+ devname, st->ss->name);
goto out;
}
if (trustworthy == LOCAL_ANY)
* CREATE.
*/
if (match && match->autof)
- autof = match->autof;
- if (autof == 0)
- autof = ci->autof;
-
- if (st->ss->container_content && st->loaded_container) {
- if ((runstop > 0 && info.container_enough >= 0) ||
- info.container_enough > 0)
- /* pass */;
- else {
- if (verbose)
- fprintf(stderr, Name ": not enough devices to start the container\n");
- rv = 0;
- goto out;
- }
-
- /* This is a pre-built container array, so we do something
- * rather different.
- */
- rv = Incremental_container(st, devname, verbose, runstop,
- autof, trustworthy);
- goto out;
- }
+ c->autof = match->autof;
+ if (c->autof == 0)
+ c->autof = ci->autof;
name_to_use = info.name;
if (name_to_use[0] == 0 &&
- info.array.level == LEVEL_CONTAINER &&
- trustworthy == LOCAL) {
+ info.array.level == LEVEL_CONTAINER) {
name_to_use = info.text_version;
trustworthy = METADATA;
}
if (name_to_use[0] && trustworthy != LOCAL &&
- ! require_homehost &&
+ ! c->require_homehost &&
conf_name_is_free(name_to_use))
trustworthy = LOCAL;
/* 4/ Check if array exists.
*/
if (map_lock(&map))
- fprintf(stderr, Name ": failed to get exclusive lock on "
+ pr_err("failed to get exclusive lock on "
"mapfile\n");
+ /* Now check we can get O_EXCL. If not, probably "mdadm -A" has
+ * taken over
+ */
+ dfd = dev_open(devname, O_RDONLY|O_EXCL);
+ if (dfd < 0) {
+ if (c->verbose >= 0)
+ pr_err("cannot reopen %s: %s.\n",
+ devname, strerror(errno));
+ goto out_unlock;
+ }
+ /* Cannot hold it open while we add the device to the array,
+ * so we must release the O_EXCL and depend on the map_lock()
+ */
+ close(dfd);
+ dfd = -1;
+
mp = map_by_uuid(&map, info.uuid);
if (mp)
mdfd = open_dev(mp->devnum);
mdfd = -1;
if (mdfd < 0) {
- struct mdinfo *sra;
/* Couldn't find an existing array, maybe make a new one */
mdfd = create_mddev(match ? match->devname : NULL,
- name_to_use, autof, trustworthy, chosen_name);
+ name_to_use, c->autof, trustworthy, chosen_name);
if (mdfd < 0)
- goto out;
+ goto out_unlock;
sysfs_init(&info, mdfd, 0);
if (set_array_info(mdfd, st, &info) != 0) {
- fprintf(stderr, Name ": failed to set array info for %s: %s\n",
+ pr_err("failed to set array info for %s: %s\n",
chosen_name, strerror(errno));
rv = 2;
- goto out;
+ goto out_unlock;
}
dinfo = info;
dinfo.disk.major = major(stb.st_rdev);
dinfo.disk.minor = minor(stb.st_rdev);
if (add_disk(mdfd, st, &info, &dinfo) != 0) {
- fprintf(stderr, Name ": failed to add %s to %s: %s.\n",
+ pr_err("failed to add %s to %s: %s.\n",
devname, chosen_name, strerror(errno));
ioctl(mdfd, STOP_ARRAY, 0);
rv = 2;
- goto out;
+ goto out_unlock;
}
- sra = sysfs_read(mdfd, fd2devnum(mdfd), GET_DEVS);
+ sra = sysfs_read(mdfd, -1, (GET_DEVS | GET_STATE |
+ GET_OFFSET | GET_SIZE));
+
if (!sra || !sra->devs || sra->devs->disk.raid_disk >= 0) {
/* It really should be 'none' - must be old buggy
* kernel, and mdadm -I may not be able to complete.
* So reject it.
*/
ioctl(mdfd, STOP_ARRAY, NULL);
- fprintf(stderr, Name
- ": You have an old buggy kernel which cannot support\n"
- " --incremental reliably. Aborting.\n");
- sysfs_free(sra);
+ pr_err("You have an old buggy kernel which cannot support\n"
+ " --incremental reliably. Aborting.\n");
rv = 2;
- goto out;
+ goto out_unlock;
}
info.array.working_disks = 1;
- sysfs_free(sra);
/* 6/ Make sure /var/run/mdadm.map contains this array. */
map_update(&map, fd2devnum(mdfd),
info.text_version,
char dn[20];
int dfd2;
int err;
- struct mdinfo *sra;
struct supertype *st2;
struct mdinfo info2, *d;
+ sra = sysfs_read(mdfd, -1, (GET_DEVS | GET_STATE |
+ GET_OFFSET | GET_SIZE));
+
if (mp->path)
strcpy(chosen_name, mp->path);
else
* array was possibly started early and our best bet is
* to add this anyway.
* Also if action policy is re-add or better we allow
- * re-add
+ * re-add.
+ * This doesn't apply to containers as the 'non-spare'
+ * flag has a different meaning. The test has to happen
+ * at the device level there
*/
- if ((info.disk.state & (1<<MD_DISK_SYNC)) != 0
+ if (!st->ss->external
+ && (info.disk.state & (1<<MD_DISK_SYNC)) != 0
&& ! policy_action_allows(policy, st->ss->name,
act_re_add)
- && runstop < 1) {
- int active = 0;
-
- if (st->ss->external) {
- char *devname = devnum2devname(fd2devnum(mdfd));
-
- active = devname && is_container_active(devname);
- free(devname);
- } else if (ioctl(mdfd, GET_ARRAY_INFO, &ainf) == 0)
- active = 1;
- if (active) {
- fprintf(stderr, Name
- ": not adding %s to active array (without --run) %s\n",
- devname, chosen_name);
+ && c->runstop < 1) {
+ if (ioctl(mdfd, GET_ARRAY_INFO, &ainf) == 0) {
+ pr_err("not adding %s to active array (without --run) %s\n",
+ devname, chosen_name);
rv = 2;
- goto out;
+ goto out_unlock;
}
}
- sra = sysfs_read(mdfd, fd2devnum(mdfd), (GET_DEVS | GET_STATE));
if (!sra) {
rv = 2;
- goto out;
+ goto out_unlock;
}
if (sra->devs) {
sprintf(dn, "%d:%d", sra->devs->disk.major,
sra->devs->disk.minor);
dfd2 = dev_open(dn, O_RDONLY);
+ if (dfd2 < 0) {
+ pr_err("unable to open %s\n", devname);
+ rv = 2;
+ goto out_unlock;
+ }
st2 = dup_super(st);
if (st2->ss->load_super(st2, dfd2, NULL) ||
st->ss->compare_super(st, st2) != 0) {
- fprintf(stderr, Name
- ": metadata mismatch between %s and "
- "chosen array %s\n",
- devname, chosen_name);
+ pr_err("metadata mismatch between %s and "
+ "chosen array %s\n",
+ devname, chosen_name);
close(dfd2);
rv = 2;
- goto out;
+ goto out_unlock;
}
close(dfd2);
- memset(&info2, 0, sizeof(info2));
- st2->ss->getinfo_super(st2, &info2);
+ st2->ss->getinfo_super(st2, &info2, NULL);
st2->ss->free_super(st2);
if (info.array.level != info2.array.level ||
memcmp(info.uuid, info2.uuid, 16) != 0 ||
info.array.raid_disks != info2.array.raid_disks) {
- fprintf(stderr, Name
- ": unexpected difference between %s and %s.\n",
- chosen_name, devname);
+ pr_err("unexpected difference between %s and %s.\n",
+ chosen_name, devname);
rv = 2;
- goto out;
+ goto out_unlock;
}
}
- info2.disk.major = major(stb.st_rdev);
- info2.disk.minor = minor(stb.st_rdev);
+ info.disk.major = major(stb.st_rdev);
+ info.disk.minor = minor(stb.st_rdev);
/* add disk needs to know about containers */
if (st->ss->external)
sra->array.level = LEVEL_CONTAINER;
- err = add_disk(mdfd, st, sra, &info2);
+ err = add_disk(mdfd, st, sra, &info);
if (err < 0 && errno == EBUSY) {
/* could be another device present with the same
* disk.number. Find and reject any such
*/
find_reject(mdfd, st, sra, info.disk.number,
- info.events, verbose, chosen_name);
- err = add_disk(mdfd, st, sra, &info2);
+ info.events, c->verbose, chosen_name);
+ err = add_disk(mdfd, st, sra, &info);
}
if (err < 0) {
- fprintf(stderr, Name ": failed to add %s to %s: %s.\n",
+ pr_err("failed to add %s to %s: %s.\n",
devname, chosen_name, strerror(errno));
rv = 2;
- goto out;
+ goto out_unlock;
}
info.array.working_disks = 0;
for (d = sra->devs; d; d=d->next)
info.array.working_disks ++;
-
+
}
/* 7/ Is there enough devices to possibly start the array? */
/* 7a/ if not, finish with success. */
if (info.array.level == LEVEL_CONTAINER) {
+ int devnum = devnum; /* defined and used iff ->external */
/* Try to assemble within the container */
- map_unlock(&map);
- sysfs_uevent(&info, "change");
- if (verbose >= 0)
- fprintf(stderr, Name
- ": container %s now has %d devices\n",
- chosen_name, info.array.working_disks);
+ sysfs_uevent(sra, "change");
+ if (c->verbose >= 0)
+ pr_err("container %s now has %d device%s\n",
+ chosen_name, info.array.working_disks,
+ info.array.working_disks == 1?"":"s");
wait_for(chosen_name, mdfd);
+ if (st->ss->external)
+ devnum = fd2devnum(mdfd);
+ if (st->ss->load_container)
+ rv = st->ss->load_container(st, mdfd, NULL);
close(mdfd);
- rv = Incremental(chosen_name, verbose, runstop,
- NULL, homehost, require_homehost, autof);
+ sysfs_free(sra);
+ if (!rv)
+ rv = Incremental_container(st, chosen_name, c);
+ map_unlock(&map);
if (rv == 1)
/* Don't fail the whole -I if a subarray didn't
* have enough devices to start yet
*/
rv = 0;
+ /* after spare is added, ping monitor for external metadata
+ * so that it can eg. try to rebuild degraded array */
+ if (st->ss->external)
+ ping_monitor_by_id(devnum);
return rv;
}
- avail = NULL;
- active_disks = count_active(st, mdfd, &avail, &info);
+
+ /* We have added something to the array, so need to re-read the
+ * state. Eventually this state should be kept up-to-date as
+ * things change.
+ */
+ sysfs_free(sra);
+ sra = sysfs_read(mdfd, -1, (GET_DEVS | GET_STATE |
+ GET_OFFSET | GET_SIZE));
+ active_disks = count_active(st, sra, mdfd, &avail, &info);
if (enough(info.array.level, info.array.raid_disks,
info.array.layout, info.array.state & 1,
- avail, active_disks) == 0) {
- free(avail);
- if (verbose >= 0)
- fprintf(stderr, Name
- ": %s attached to %s, not enough to start (%d).\n",
- devname, chosen_name, active_disks);
- map_unlock(&map);
+ avail) == 0) {
+ if (c->verbose >= 0)
+ pr_err("%s attached to %s, not enough to start (%d).\n",
+ devname, chosen_name, active_disks);
rv = 0;
- goto out;
+ goto out_unlock;
}
- free(avail);
/* 7b/ if yes, */
/* - if number of OK devices match expected, or -R and there */
/* + start the array (auto-readonly). */
if (ioctl(mdfd, GET_ARRAY_INFO, &ainf) == 0) {
- if (verbose >= 0)
- fprintf(stderr, Name
- ": %s attached to %s which is already active.\n",
- devname, chosen_name);
- map_unlock(&map);
+ if (c->verbose >= 0)
+ pr_err("%s attached to %s which is already active.\n",
+ devname, chosen_name);
rv = 0;
- goto out;
+ goto out_unlock;
}
map_unlock(&map);
- if (runstop > 0 || active_disks >= info.array.working_disks) {
- struct mdinfo *sra, *dsk;
+ if (c->runstop > 0 || active_disks >= info.array.working_disks) {
+ struct mdinfo *dsk;
/* Let's try to start it */
+
+ if (info.reshape_active && !(info.reshape_active & RESHAPE_NO_BACKUP)) {
+ fprintf(stderr, Name
+ ": %s: This array is being reshaped and cannot be started\n"
+ " by --incremental. Please use --assemble\n",
+ chosen_name);
+ goto out;
+ }
if (match && match->bitmap_file) {
int bmfd = open(match->bitmap_file, O_RDWR);
if (bmfd < 0) {
- fprintf(stderr, Name
- ": Could not open bitmap file %s.\n",
- match->bitmap_file);
+ pr_err("Could not open bitmap file %s.\n",
+ match->bitmap_file);
goto out;
}
if (ioctl(mdfd, SET_BITMAP_FILE, bmfd) != 0) {
close(bmfd);
- fprintf(stderr, Name
- ": Failed to set bitmapfile for %s.\n",
- chosen_name);
+ pr_err("Failed to set bitmapfile for %s.\n",
+ chosen_name);
goto out;
}
close(bmfd);
}
- /* GET_* needed so add_disk works below */
- sra = sysfs_read(mdfd, fd2devnum(mdfd),
- GET_DEVS|GET_OFFSET|GET_SIZE|GET_STATE);
+ /* Need to remove from the array any devices which
+ * 'count_active' discerned were too old or inappropriate
+ */
+ for (d = sra ? sra->devs : NULL ; d ; d = d->next)
+ if (d->disk.state & (1<<MD_DISK_REMOVED))
+ remove_disk(mdfd, st, sra, d);
+
if ((sra == NULL || active_disks >= info.array.working_disks)
&& trustworthy != FOREIGN)
rv = ioctl(mdfd, RUN_ARRAY, NULL);
rv = sysfs_set_str(sra, NULL,
"array_state", "read-auto");
if (rv == 0) {
- if (verbose >= 0)
- fprintf(stderr, Name
- ": %s attached to %s, which has been started.\n",
- devname, chosen_name);
+ if (c->verbose >= 0)
+ pr_err("%s attached to %s, which has been started.\n",
+ devname, chosen_name);
rv = 0;
wait_for(chosen_name, mdfd);
/* We just started the array, so some devices
for (dsk = sra->devs; dsk ; dsk = dsk->next) {
if (disk_action_allows(dsk, st->ss->name, act_re_add) &&
add_disk(mdfd, st, sra, dsk) == 0)
- fprintf(stderr, Name
- ": %s re-added to %s\n",
- dsk->sys_name, chosen_name);
+ pr_err("%s re-added to %s\n",
+ dsk->sys_name, chosen_name);
}
} else {
- fprintf(stderr, Name
- ": %s attached to %s, but failed to start: %s.\n",
- devname, chosen_name, strerror(errno));
+ pr_err("%s attached to %s, but failed to start: %s.\n",
+ devname, chosen_name, strerror(errno));
rv = 1;
}
} else {
- if (verbose >= 0)
- fprintf(stderr, Name
- ": %s attached to %s, not enough to start safely.\n",
- devname, chosen_name);
+ if (c->verbose >= 0)
+ pr_err("%s attached to %s, not enough to start safely.\n",
+ devname, chosen_name);
rv = 0;
}
out:
+ free(avail);
if (dfd >= 0)
close(dfd);
if (mdfd >= 0)
close(mdfd);
if (policy)
dev_policy_free(policy);
+ if (sra)
+ sysfs_free(sra);
return rv;
+out_unlock:
+ map_unlock(&map);
+ goto out;
}
static void find_reject(int mdfd, struct supertype *st, struct mdinfo *sra,
close(dfd);
continue;
}
- st->ss->getinfo_super(st, &info);
+ st->ss->getinfo_super(st, &info, NULL);
st->ss->free_super(st);
close(dfd);
sysfs_set_str(sra, d, "slot", "none");
if (sysfs_set_str(sra, d, "state", "remove") == 0)
if (verbose >= 0)
- fprintf(stderr, Name
- ": removing old device %s from %s\n",
- d->sys_name+4, array_name);
+ pr_err("removing old device %s from %s\n",
+ d->sys_name+4, array_name);
}
}
-static int count_active(struct supertype *st, int mdfd, char **availp,
+static int count_active(struct supertype *st, struct mdinfo *sra,
+ int mdfd, char **availp,
struct mdinfo *bestinfo)
{
/* count how many devices in sra think they are active */
struct mdinfo *d;
- int cnt = 0, cnt1 = 0;
+ int cnt = 0;
+ int replcnt = 0;
__u64 max_events = 0;
- struct mdinfo *sra = sysfs_read(mdfd, -1, GET_DEVS | GET_STATE);
char *avail = NULL;
+ int *best = NULL;
+ char *devmap = NULL;
+ int numdevs = 0;
+ int devnum;
+ int b, i;
+ int raid_disks = 0;
if (!sra)
return 0;
- for (d = sra->devs ; d ; d = d->next) {
+ for (d = sra->devs ; d ; d = d->next)
+ numdevs++;
+ for (d = sra->devs, devnum = 0 ; d ; d = d->next, devnum++) {
char dn[30];
int dfd;
int ok;
close(dfd);
if (ok != 0)
continue;
- st->ss->getinfo_super(st, &info);
+ info.array.raid_disks = raid_disks;
+ st->ss->getinfo_super(st, &info, devmap + raid_disks * devnum);
if (!avail) {
- avail = malloc(info.array.raid_disks);
- if (!avail) {
- fprintf(stderr, Name ": out of memory.\n");
- exit(1);
- }
- memset(avail, 0, info.array.raid_disks);
+ raid_disks = info.array.raid_disks;
+ avail = xcalloc(raid_disks, 1);
*availp = avail;
+
+ best = xcalloc(raid_disks, sizeof(int));
+ devmap = xcalloc(raid_disks, numdevs);
+
+ st->ss->getinfo_super(st, &info, devmap);
}
if (info.disk.state & (1<<MD_DISK_SYNC))
cnt++;
max_events = info.events;
avail[info.disk.raid_disk] = 2;
- st->ss->getinfo_super(st, bestinfo);
+ best[info.disk.raid_disk] = devnum;
+ st->ss->getinfo_super(st, bestinfo, NULL);
} else if (info.events == max_events) {
- cnt++;
avail[info.disk.raid_disk] = 2;
+ best[info.disk.raid_disk] = devnum;
} else if (info.events == max_events-1) {
- cnt1++;
- avail[info.disk.raid_disk] = 1;
+ if (avail[info.disk.raid_disk] == 0) {
+ avail[info.disk.raid_disk] = 1;
+ best[info.disk.raid_disk] = devnum;
+ }
} else if (info.events < max_events - 1)
;
else if (info.events == max_events+1) {
int i;
- cnt1 = cnt;
- cnt = 1;
max_events = info.events;
- for (i=0; i<info.array.raid_disks; i++)
+ for (i = 0; i < raid_disks; i++)
if (avail[i])
avail[i]--;
avail[info.disk.raid_disk] = 2;
- st->ss->getinfo_super(st, bestinfo);
+ best[info.disk.raid_disk] = devnum;
+ st->ss->getinfo_super(st, bestinfo, NULL);
} else { /* info.events much bigger */
- cnt = 1; cnt1 = 0;
- memset(avail, 0, info.disk.raid_disk);
+ memset(avail, 0, raid_disks);
max_events = info.events;
- st->ss->getinfo_super(st, bestinfo);
+ avail[info.disk.raid_disk] = 2;
+ best[info.disk.raid_disk] = devnum;
+ st->ss->getinfo_super(st, bestinfo, NULL);
}
- }
+ } else if (info.disk.state & (1<<MD_DISK_REPLACEMENT))
+ replcnt++;
st->ss->free_super(st);
}
- return cnt + cnt1;
+ if (!avail)
+ return 0;
+ /* We need to reject any device that thinks the best device is
+ * failed or missing */
+ for (b = 0; b < raid_disks; b++)
+ if (avail[b] == 2)
+ break;
+ cnt = 0;
+ for (i = 0 ; i < raid_disks ; i++) {
+ if (i != b && avail[i])
+ if (devmap[raid_disks * best[i] + b] == 0) {
+ /* This device thinks 'b' is failed -
+ * don't use it */
+ devnum = best[i];
+ for (d=sra->devs ; devnum; d = d->next)
+ devnum--;
+ d->disk.state |= (1 << MD_DISK_REMOVED);
+ avail[i] = 0;
+ }
+ if (avail[i])
+ cnt++;
+ }
+ free(best);
+ free(devmap);
+ return cnt + replcnt;
+}
+
+/* test if container has degraded member(s) */
+static int container_members_max_degradation(struct map_ent *map, struct map_ent *me)
+{
+ mdu_array_info_t array;
+ int afd;
+ int max_degraded = 0;
+
+ for(; map; map = map->next) {
+ if (!is_subarray(map->metadata) ||
+ devname2devnum(map->metadata+1) != me->devnum)
+ continue;
+ afd = open_dev(map->devnum);
+ if (afd < 0)
+ continue;
+ /* most accurate information regarding array degradation */
+ if (ioctl(afd, GET_ARRAY_INFO, &array) >= 0) {
+ int degraded = array.raid_disks - array.active_disks -
+ array.spare_disks;
+ if (degraded > max_degraded)
+ max_degraded = degraded;
+ }
+ close(afd);
+ }
+ return (max_degraded);
}
static int array_try_spare(char *devname, int *dfdp, struct dev_policy *pol,
+ struct map_ent *target, int bare,
struct supertype *st, int verbose)
{
/* This device doesn't have any md metadata
- * If it is 'bare' and theh device policy allows 'spare' look for
- * an array or container to attach it to.
+ * The device policy allows 'spare' and if !bare, it allows spare-same-slot.
+ * If 'st' is not set, then we only know that some metadata allows this,
+ * others possibly don't.
+ * So look for a container or array to attach the device to.
+ * Prefer 'target' if that is set and the array is found.
+ *
* If st is set, then only arrays of that type are considered
* Return 0 on success, or some exit code on failure, probably 1.
*/
- int rv = -1;
+ int rv = 1;
struct stat stb;
struct map_ent *mp, *map = NULL;
struct mdinfo *chosen = NULL;
*/
if (map_lock(&map)) {
- fprintf(stderr, Name ": failed to get exclusive lock on "
+ pr_err("failed to get exclusive lock on "
"mapfile\n");
return 1;
}
struct domainlist *dl = NULL;
struct mdinfo *sra;
unsigned long long devsize;
+ unsigned long long component_size = 0;
if (is_subarray(mp->metadata))
continue;
(st->minor_version >= 0 &&
st->minor_version != st2->minor_version)) {
if (verbose > 1)
- fprintf(stderr, Name ": not adding %s to %s as metadata type doesn't match\n",
+ pr_err("not adding %s to %s as metadata type doesn't match\n",
devname, mp->path);
free(st2);
continue;
GET_DEVS|GET_OFFSET|GET_SIZE|GET_STATE|
GET_COMPONENT|GET_VERSION);
if (sra)
- sra->array.failed_disks = 0;
+ sra->array.failed_disks = -1;
}
if (!sra)
continue;
if (st == NULL) {
int i;
st2 = NULL;
- for(i=0; !st2 && superlist[i]; i++)
+ for(i = 0; !st2 && superlist[i]; i++)
st2 = superlist[i]->match_metadata_desc(
sra->text_version);
+ if (!st2) {
+ if (verbose > 1)
+ pr_err("not adding %s to %s"
+ " as metadata not recognised.\n",
+ devname, mp->path);
+ goto next;
+ }
+ /* Need to double check the 'act_spare' permissions applies
+ * to this metadata.
+ */
+ if (!policy_action_allows(pol, st2->ss->name, act_spare))
+ goto next;
+ if (!bare && !policy_action_allows(pol, st2->ss->name,
+ act_spare_same_slot))
+ goto next;
} else
st2 = st;
+ /* update number of failed disks for mostly degraded
+ * container member */
+ if (sra->array.failed_disks == -1)
+ sra->array.failed_disks = container_members_max_degradation(map, mp);
+
get_dev_size(dfd, NULL, &devsize);
- if (st2->ss->avail_size(st2, devsize) < sra->component_size) {
+ if (sra->component_size == 0) {
+ /* true for containers, here we must read superblock
+ * to obtain minimum spare size */
+ struct supertype *st3 = dup_super(st2);
+ int mdfd = open_dev(mp->devnum);
+ if (mdfd < 0) {
+ free(st3);
+ goto next;
+ }
+ if (st3->ss->load_container &&
+ !st3->ss->load_container(st3, mdfd, mp->path)) {
+ component_size = st3->ss->min_acceptable_spare_size(st3);
+ st3->ss->free_super(st3);
+ }
+ free(st3);
+ close(mdfd);
+ }
+ if ((sra->component_size > 0 &&
+ st2->ss->avail_size(st2, devsize,
+ sra->devs
+ ? sra->devs->data_offset
+ : INVALID_SECTORS)
+ < sra->component_size)
+ ||
+ (sra->component_size == 0 && devsize < component_size)) {
if (verbose > 1)
- fprintf(stderr, Name ": not adding %s to %s as it is too small\n",
+ pr_err("not adding %s to %s as it is too small\n",
devname, mp->path);
goto next;
}
+ /* test against target.
+ * If 'target' is set and 'bare' is false, we only accept
+ * arrays/containers that match 'target'.
+ * If 'target' is set and 'bare' is true, we prefer the
+ * array which matches 'target'.
+ * target is considered only if we deal with degraded array
+ */
+ if (target && policy_action_allows(pol, st2->ss->name,
+ act_spare_same_slot)) {
+ if (strcmp(target->metadata, mp->metadata) == 0 &&
+ memcmp(target->uuid, mp->uuid,
+ sizeof(target->uuid)) == 0 &&
+ sra->array.failed_disks > 0) {
+ /* This is our target!! */
+ if (chosen)
+ sysfs_free(chosen);
+ chosen = sra;
+ sra = NULL;
+ /* skip to end so we don't check any more */
+ while (mp->next)
+ mp = mp->next;
+ goto next;
+ }
+ /* not our target */
+ if (!bare)
+ goto next;
+ }
+
dl = domain_from_array(sra, st2->ss->name);
- if (!domain_test(dl, pol, st2->ss->name)) {
+ if (domain_test(dl, pol, st2->ss->name) != 1) {
/* domain test fails */
if (verbose > 1)
- fprintf(stderr, Name ": not adding %s to %s as it is not in a compatible domain\n",
+ pr_err("not adding %s to %s as"
+ " it is not in a compatible domain\n",
devname, mp->path);
goto next;
/* add current device to chosen array as a spare */
int mdfd = open_dev(devname2devnum(chosen->sys_name));
if (mdfd >= 0) {
- struct mddev_dev_s devlist;
+ struct mddev_dev devlist;
char devname[20];
devlist.next = NULL;
devlist.used = 0;
- devlist.re_add = 0;
devlist.writemostly = 0;
devlist.devname = devname;
sprintf(devname, "%d:%d", major(stb.st_rdev),
close(dfd);
*dfdp = -1;
rv = Manage_subdevs(chosen->sys_name, mdfd, &devlist,
- -1, 0);
+ -1, 0, NULL, 0);
close(mdfd);
}
if (verbose > 0) {
if (rv == 0)
- fprintf(stderr, Name ": added %s as spare for %s\n",
+ pr_err("added %s as spare for %s\n",
devname, chosen->sys_name);
else
- fprintf(stderr, Name ": failed to add %s as spare for %s\n",
+ pr_err("failed to add %s as spare for %s\n",
devname, chosen->sys_name);
}
sysfs_free(chosen);
}
- return rv ? 0 : 1;
+ map_unlock(&map);
+ return rv;
}
static int partition_try_spare(char *devname, int *dfdp, struct dev_policy *pol,
struct supertype *st, int verbose)
{
- return 1;
+ /* we know that at least one partition virtual-metadata is
+ * allowed to incorporate spares like this device. We need to
+ * find a suitable device to copy partition information from.
+ *
+ * Getting a list of all disk (not partition) devices is
+ * slightly non-trivial. We could look at /sys/block, but
+ * that is theoretically due to be removed. Maybe best to use
+ * /dev/disk/by-path/?* and ignore names ending '-partNN' as
+ * we depend on this directory of 'path' info. But that fails
+ * to find loop devices and probably others. Maybe don't
+ * worry about that, they aren't the real target.
+ *
+ * So: check things in /dev/disk/by-path to see if they are in
+ * a compatible domain, then load the partition table and see
+ * if it is OK for the new device, and choose the largest
+ * partition table that fits.
+ */
+ DIR *dir;
+ struct dirent *de;
+ char *chosen = NULL;
+ unsigned long long chosen_size = 0;
+ struct supertype *chosen_st = NULL;
+ int fd;
+
+ dir = opendir("/dev/disk/by-path");
+ if (!dir)
+ return 1;
+ while ((de = readdir(dir)) != NULL) {
+ char *ep;
+ struct dev_policy *pol2 = NULL;
+ struct domainlist *domlist = NULL;
+ int fd = -1;
+ struct mdinfo info;
+ struct supertype *st2 = NULL;
+ char *devname = NULL;
+ unsigned long long devsectors;
+
+ if (de->d_ino == 0 ||
+ de->d_name[0] == '.' ||
+ (de->d_type != DT_LNK && de->d_type != DT_UNKNOWN))
+ goto next;
+
+ ep = de->d_name + strlen(de->d_name);
+ while (ep > de->d_name &&
+ isdigit(ep[-1]))
+ ep--;
+ if (ep > de->d_name + 5 &&
+ strncmp(ep-5, "-part", 5) == 0)
+ /* This is a partition - skip it */
+ goto next;
+
+ pol2 = path_policy(de->d_name, type_disk);
+
+ domain_merge(&domlist, pol2, st ? st->ss->name : NULL);
+ if (domain_test(domlist, pol, st ? st->ss->name : NULL) != 1)
+ /* new device is incompatible with this device. */
+ goto next;
+
+ domain_free(domlist);
+ domlist = NULL;
+
+ if (asprintf(&devname, "/dev/disk/by-path/%s", de->d_name) != 1) {
+ devname = NULL;
+ goto next;
+ }
+ fd = open(devname, O_RDONLY);
+ if (fd < 0)
+ goto next;
+ if (get_dev_size(fd, devname, &devsectors) == 0)
+ goto next;
+ devsectors >>= 9;
+
+ if (st)
+ st2 = dup_super(st);
+ else
+ st2 = guess_super_type(fd, guess_partitions);
+ if (st2 == NULL ||
+ st2->ss->load_super(st2, fd, NULL) < 0)
+ goto next;
+
+ if (!st) {
+ /* Check domain policy again, this time referring to metadata */
+ domain_merge(&domlist, pol2, st2->ss->name);
+ if (domain_test(domlist, pol, st2->ss->name) != 1)
+ /* Incompatible devices for this metadata type */
+ goto next;
+ if (!policy_action_allows(pol, st2->ss->name, act_spare))
+ /* Some partition types allow sparing, but not
+ * this one.
+ */
+ goto next;
+ }
+
+ st2->ss->getinfo_super(st2, &info, NULL);
+ if (info.component_size > devsectors)
+ /* This partitioning doesn't fit in the device */
+ goto next;
+
+ /* This is an acceptable device to copy partition
+ * metadata from. We could just stop here, but I
+ * think I want to keep looking incase a larger
+ * metadata which makes better use of the device can
+ * be found.
+ */
+ if (chosen == NULL ||
+ chosen_size < info.component_size) {
+ chosen_size = info.component_size;
+ free(chosen);
+ chosen = devname;
+ devname = NULL;
+ if (chosen_st) {
+ chosen_st->ss->free_super(chosen_st);
+ free(chosen_st);
+ }
+ chosen_st = st2;
+ st2 = NULL;
+ }
+
+ next:
+ free(devname);
+ domain_free(domlist);
+ dev_policy_free(pol2);
+ if (st2)
+ st2->ss->free_super(st2);
+ free(st2);
+
+ if (fd >= 0)
+ close(fd);
+ }
+
+ closedir(dir);
+
+ if (!chosen)
+ return 1;
+
+ /* 'chosen' is the best device we can find. Let's write its
+ * metadata to devname dfd is read-only so don't use that
+ */
+ fd = open(devname, O_RDWR);
+ if (fd >= 0) {
+ chosen_st->ss->store_super(chosen_st, fd);
+ close(fd);
+ }
+ free(chosen);
+ chosen_st->ss->free_super(chosen_st);
+ free(chosen_st);
+ return 0;
}
+static int is_bare(int dfd)
+{
+ unsigned long long size = 0;
+ char bufpad[4096 + 4096];
+ char *buf = (char*)(((long)bufpad + 4096) & ~4095);
+
+ if (lseek(dfd, 0, SEEK_SET) != 0 ||
+ read(dfd, buf, 4096) != 4096)
+ return 0;
+
+ if (buf[0] != '\0' && buf[0] != '\x5a' && buf[0] != '\xff')
+ return 0;
+ if (memcmp(buf, buf+1, 4095) != 0)
+ return 0;
+
+ /* OK, first 4K appear blank, try the end. */
+ get_dev_size(dfd, NULL, &size);
+ if (lseek(dfd, size-4096, SEEK_SET) < 0 ||
+ read(dfd, buf, 4096) != 4096)
+ return 0;
+
+ if (buf[0] != '\0' && buf[0] != '\x5a' && buf[0] != '\xff')
+ return 0;
+ if (memcmp(buf, buf+1, 4095) != 0)
+ return 0;
+
+ return 1;
+}
/* adding a spare to a regular array is quite different from adding one to
* a set-of-partitions virtual array.
* Arrays are given priority over partitions.
*/
static int try_spare(char *devname, int *dfdp, struct dev_policy *pol,
+ struct map_ent *target,
struct supertype *st, int verbose)
{
int i;
int rv;
int arrays_ok = 0;
int partitions_ok = 0;
- char bufpad[4096 + 4096];
- char *buf = (char*)(((long)bufpad + 4096) & ~4095);
int dfd = *dfdp;
+ int bare;
- /* Can only add a spare if device has at least one domains */
+ /* Can only add a spare if device has at least one domain */
if (pol_find(pol, pol_domain) == NULL)
return 1;
/* And only if some action allows spares */
if (!policy_action_allows(pol, st?st->ss->name:NULL, act_spare))
return 1;
- /* Now check if the device is bare - we don't add non-bare devices
- * yet even if action=-spare
+ /* Now check if the device is bare.
+ * bare devices can always be added as a spare
+ * non-bare devices can only be added if spare-same-slot is permitted,
+ * and this device is replacing a previous device - in which case 'target'
+ * will be set.
*/
+ if (!is_bare(dfd)) {
+ /* Must have a target and allow same_slot */
+ /* Later - may allow force_spare without target */
+ if (!target ||
+ !policy_action_allows(pol, st?st->ss->name:NULL,
+ act_spare_same_slot)) {
+ if (verbose > 1)
+ pr_err("%s is not bare, so not "
+ "considering as a spare\n",
+ devname);
+ return 1;
+ }
+ bare = 0;
+ } else
+ bare = 1;
- if (lseek(dfd, 0, SEEK_SET) != 0 ||
- read(dfd, buf, 4096) != 4096) {
- not_bare:
- if (verbose > 1)
- fprintf(stderr, Name ": %s is not bare, so not considering as a spare\n",
- devname);
- return 1;
- }
- if (buf[0] != '\0' && buf[0] != '\x5a' && buf[0] != '\xff')
- goto not_bare;
- if (memcmp(buf, buf+1, 4095) != 0)
- goto not_bare;
-
- /* OK, first 4K appear blank, try the end. */
- if (lseek(dfd, -4096, SEEK_END) < 0 ||
- read(dfd, buf, 4096) != 4096)
- goto not_bare;
-
- if (buf[0] != '\0' && buf[0] != '\x5a' && buf[0] != '\xff')
- goto not_bare;
- if (memcmp(buf, buf+1, 4095) != 0)
- goto not_bare;
-
- /* This device passes our test for 'is bare'.
- * Let's see what policy allows for such things.
+ /* It might be OK to add this device to an array - need to see
+ * what arrays might be candidates.
*/
if (st) {
/* just try try 'array' or 'partition' based on this metadata */
if (st->ss->add_to_super)
- return array_try_spare(devname, dfdp, pol,
+ return array_try_spare(devname, dfdp, pol, target, bare,
st, verbose);
else
return partition_try_spare(devname, dfdp, pol,
st, verbose);
}
- /* Now see which metadata type support spare */
+ /* No metadata was specified or found so options are open.
+ * Check for whether any array metadata, or any partition metadata
+ * might allow adding the spare. This check is just help to avoid
+ * a more costly scan of all arrays when we can be sure that will
+ * fail.
+ */
for (i = 0; (!arrays_ok || !partitions_ok) && superlist[i] ; i++) {
if (superlist[i]->add_to_super && !arrays_ok &&
policy_action_allows(pol, superlist[i]->name, act_spare))
policy_action_allows(pol, superlist[i]->name, act_spare))
partitions_ok = 1;
}
- rv = 0;
+ rv = 1;
if (arrays_ok)
- rv = array_try_spare(devname, dfdp, pol, st, verbose);
- if (rv == 0 && partitions_ok)
+ rv = array_try_spare(devname, dfdp, pol, target, bare,
+ st, verbose);
+ if (rv != 0 && partitions_ok)
rv = partition_try_spare(devname, dfdp, pol, st, verbose);
return rv;
}
*/
struct map_ent *mapl = NULL;
struct map_ent *me;
- mddev_ident_t devs, mddev;
+ struct mddev_ident *devs, *mddev;
int rv = 0;
map_read(&mapl);
}
if (verbose >= 0) {
if (added == 0)
- fprintf(stderr, Name
- ": Added bitmap %s to %s\n",
- mddev->bitmap_file, me->path);
+ pr_err("Added bitmap %s to %s\n",
+ mddev->bitmap_file, me->path);
else if (errno != EEXIST)
- fprintf(stderr, Name
- ": Failed to add bitmap to %s: %s\n",
- me->path, strerror(errno));
+ pr_err("Failed to add bitmap to %s: %s\n",
+ me->path, strerror(errno));
}
}
+ /* FIXME check for reshape_active and consider not
+ * starting array.
+ */
sra = sysfs_read(mdfd, 0, 0);
if (sra) {
if (sysfs_set_str(sra, NULL,
"array_state", "read-auto") == 0) {
if (verbose >= 0)
- fprintf(stderr, Name
- ": started array %s\n",
- me->path ?: devnum2devname(me->devnum));
+ pr_err("started array %s\n",
+ me->path ?: devnum2devname(me->devnum));
} else {
- fprintf(stderr, Name
- ": failed to start array %s: %s\n",
- me->path ?: devnum2devname(me->devnum),
- strerror(errno));
+ pr_err("failed to start array %s: %s\n",
+ me->path ?: devnum2devname(me->devnum),
+ strerror(errno));
rv = 1;
}
+ sysfs_free(sra);
}
}
return rv;
} else {
int uuid[4];
struct map_ent *mp, *map = NULL;
-
+
if (!parse_uuid(devname, uuid))
return mdname;
mp = map_by_uuid(&map, uuid);
return mdname;
}
-int Incremental_container(struct supertype *st, char *devname, int verbose,
- int runstop, int autof, int trustworthy)
+static int Incremental_container(struct supertype *st, char *devname,
+ struct context *c)
{
/* Collect the contents of this container and for each
* array, choose a device name and assemble the array.
*/
- struct mdinfo *list = st->ss->container_content(st);
+ struct mdinfo *list;
struct mdinfo *ra;
struct map_ent *map = NULL;
+ struct mdinfo info;
+ int trustworthy;
+ struct mddev_ident *match;
+ int rv = 0;
+ struct domainlist *domains;
+ struct map_ent *smp;
+ int suuid[4];
+ int sfd;
+ int ra_blocked = 0;
+ int ra_all = 0;
+
+ st->ss->getinfo_super(st, &info, NULL);
+
+ if ((c->runstop > 0 && info.container_enough >= 0) ||
+ info.container_enough > 0)
+ /* pass */;
+ else {
+ if (c->verbose)
+ pr_err("not enough devices to start the container\n");
+ return 0;
+ }
- if (map_lock(&map))
- fprintf(stderr, Name ": failed to get exclusive lock on "
- "mapfile\n");
+ match = conf_match(st, &info, devname, c->verbose, &rv);
+ if (match == NULL && rv == 2)
+ return rv;
+ /* Need to compute 'trustworthy' */
+ if (match)
+ trustworthy = LOCAL;
+ else if (st->ss->match_home(st, c->homehost) == 1)
+ trustworthy = LOCAL;
+ else if (st->ss->match_home(st, "any") == 1)
+ trustworthy = LOCAL;
+ else
+ trustworthy = FOREIGN;
+
+ list = st->ss->container_content(st, NULL);
+ /* when nothing to activate - quit */
+ if (list == NULL)
+ return 0;
for (ra = list ; ra ; ra = ra->next) {
int mdfd;
char chosen_name[1024];
struct map_ent *mp;
- struct mddev_ident_s *match = NULL;
-
+ struct mddev_ident *match = NULL;
+
+ ra_all++;
+ /* do not activate arrays blocked by metadata handler */
+ if (ra->array.state & (1 << MD_SB_BLOCK_VOLUME)) {
+ pr_err("Cannot activate array %s in %s.\n",
+ ra->text_version, devname);
+ ra_blocked++;
+ continue;
+ }
mp = map_by_uuid(&map, ra->uuid);
if (mp) {
* member == ra->text_version after second slash.
*/
char *sub = strchr(ra->text_version+1, '/');
- struct mddev_ident_s *array_list;
+ struct mddev_ident *array_list;
if (sub) {
sub++;
array_list = conf_get_ident(NULL);
free(dn);
/* we have a match */
match = array_list;
- if (verbose>0)
- fprintf(stderr, Name ": match found for member %s\n",
+ if (c->verbose>0)
+ pr_err("match found for member %s\n",
array_list->member);
break;
}
if (match && match->devname &&
strcasecmp(match->devname, "<ignore>") == 0) {
- if (verbose > 0)
- fprintf(stderr, Name ": array %s/%s is "
- "explicitly ignored by mdadm.conf\n",
- match->container, match->member);
+ if (c->verbose > 0)
+ pr_err("array %s/%s is "
+ "explicitly ignored by mdadm.conf\n",
+ match->container, match->member);
return 2;
}
if (match)
mdfd = create_mddev(match ? match->devname : NULL,
ra->name,
- autof,
+ c->autof,
trustworthy,
chosen_name);
}
if (mdfd < 0) {
- fprintf(stderr, Name ": failed to open %s: %s.\n",
+ pr_err("failed to open %s: %s.\n",
chosen_name, strerror(errno));
return 2;
}
- assemble_container_content(st, mdfd, ra, runstop,
- chosen_name, verbose);
+ assemble_container_content(st, mdfd, ra, c,
+ chosen_name);
+ close(mdfd);
}
- map_unlock(&map);
+
+ /* don't move spares to container with volume being activated
+ when all volumes are blocked */
+ if (ra_all == ra_blocked)
+ return 0;
+
+ /* Now move all suitable spares from spare container */
+ domains = domain_from_array(list, st->ss->name);
+ memcpy(suuid, uuid_zero, sizeof(int[4]));
+ if (domains &&
+ (smp = map_by_uuid(&map, suuid)) != NULL &&
+ (sfd = open(smp->path, O_RDONLY)) >= 0) {
+ /* spare container found */
+ struct supertype *sst =
+ super_imsm.match_metadata_desc("imsm");
+ struct mdinfo *sinfo;
+ unsigned long long min_size = 0;
+ if (st->ss->min_acceptable_spare_size)
+ min_size = st->ss->min_acceptable_spare_size(st);
+ if (!sst->ss->load_container(sst, sfd, NULL)) {
+ close(sfd);
+ sinfo = container_choose_spares(sst, min_size,
+ domains, NULL,
+ st->ss->name, 0);
+ sst->ss->free_super(sst);
+ if (sinfo){
+ int count = 0;
+ struct mdinfo *disks = sinfo->devs;
+ while (disks) {
+ /* move spare from spare
+ * container to currently
+ * assembled one
+ */
+ if (move_spare(
+ smp->path,
+ devname,
+ makedev(disks->disk.major,
+ disks->disk.minor)))
+ count++;
+ disks = disks->next;
+ }
+ if (count)
+ pr_err("Added %d spare%s to %s\n",
+ count, count>1?"s":"", devname);
+ }
+ sysfs_free(sinfo);
+ } else
+ close(sfd);
+ }
+ domain_free(domains);
return 0;
}
* raid arrays, and if so first fail (if needed) and then remove the device.
*
* @devname - The device we want to remove
+ * @id_path - name as found in /dev/disk/by-path for this device
*
* Note: the device name must be a kernel name like "sda", so
* that we can find it in /proc/mdstat
*/
-int IncrementalRemove(char *devname, int verbose)
+int IncrementalRemove(char *devname, char *id_path, int verbose)
{
int mdfd;
int rv;
struct mdstat_ent *ent;
- struct mddev_dev_s devlist;
+ struct mddev_dev devlist;
+
+ if (!id_path)
+ dprintf(Name ": incremental removal without --path <id_path> "
+ "lacks the possibility to re-add new device in this "
+ "port\n");
if (strchr(devname, '/')) {
- fprintf(stderr, Name ": incremental removal requires a "
+ pr_err("incremental removal requires a "
"kernel device name, not a file: %s\n", devname);
return 1;
}
ent = mdstat_by_component(devname);
if (!ent) {
- fprintf(stderr, Name ": %s does not appear to be a component "
+ pr_err("%s does not appear to be a component "
"of any array\n", devname);
return 1;
}
mdfd = open_dev(ent->devnum);
if (mdfd < 0) {
- fprintf(stderr, Name ": Cannot open array %s!!\n", ent->dev);
+ pr_err("Cannot open array %s!!\n", ent->dev);
+ free_mdstat(ent);
return 1;
}
+
+ if (id_path) {
+ struct map_ent *map = NULL, *me;
+ me = map_by_devnum(&map, ent->devnum);
+ if (me)
+ policy_save_path(id_path, me);
+ map_free(map);
+ }
+
memset(&devlist, 0, sizeof(devlist));
devlist.devname = devname;
devlist.disposition = 'f';
- Manage_subdevs(ent->dev, mdfd, &devlist, verbose, 0);
+ /* for a container, we must fail each member array */
+ if (ent->metadata_version &&
+ strncmp(ent->metadata_version, "external:", 9) == 0) {
+ struct mdstat_ent *mdstat = mdstat_read(0, 0);
+ struct mdstat_ent *memb;
+ for (memb = mdstat ; memb ; memb = memb->next)
+ if (is_container_member(memb, ent->dev)) {
+ int subfd = open_dev(memb->devnum);
+ if (subfd >= 0) {
+ Manage_subdevs(memb->dev, subfd,
+ &devlist, verbose, 0,
+ NULL, 0);
+ close(subfd);
+ }
+ }
+ free_mdstat(mdstat);
+ } else
+ Manage_subdevs(ent->dev, mdfd, &devlist, verbose, 0, NULL, 0);
devlist.disposition = 'r';
- rv = Manage_subdevs(ent->dev, mdfd, &devlist, verbose, 0);
+ rv = Manage_subdevs(ent->dev, mdfd, &devlist, verbose, 0, NULL, 0);
close(mdfd);
+ free_mdstat(ent);
return rv;
}
projects / thirdparty / mdadm.git / blobdiff