1
0
mirror of git://sourceware.org/git/lvm2.git synced 2024-12-22 17:35:59 +03:00
lvm2/lib/activate/dev_manager.c

2318 lines
61 KiB
C
Raw Normal View History

/*
2008-01-30 17:00:02 +03:00
* Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
* Copyright (C) 2004-2012 Red Hat, Inc. All rights reserved.
*
2004-03-30 23:35:44 +04:00
* This file is part of LVM2.
*
* This copyrighted material is made available to anyone wishing to use,
* modify, copy, or redistribute it subject to the terms and conditions
* of the GNU Lesser General Public License v.2.1.
2004-03-30 23:35:44 +04:00
*
* You should have received a copy of the GNU Lesser General Public License
2004-03-30 23:35:44 +04:00
* along with this program; if not, write to the Free Software Foundation,
* Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
2002-11-18 17:01:16 +03:00
#include "lib.h"
#include "dev_manager.h"
2002-02-25 19:53:12 +03:00
#include "lvm-string.h"
#include "fs.h"
#include "defaults.h"
2004-09-16 22:40:56 +04:00
#include "segtype.h"
2004-05-05 01:25:57 +04:00
#include "display.h"
2003-11-13 21:47:22 +03:00
#include "toolcontext.h"
2004-05-05 01:25:57 +04:00
#include "targets.h"
#include "config.h"
#include "filter.h"
#include "activate.h"
#include "lvm-exec.h"
#include <limits.h>
#include <dirent.h>
#define MAX_TARGET_PARAMSIZE 50000
2002-03-16 01:59:12 +03:00
typedef enum {
PRELOAD,
2002-03-16 01:59:12 +03:00
ACTIVATE,
DEACTIVATE,
2002-03-16 01:59:12 +03:00
SUSPEND,
SUSPEND_WITH_LOCKFS,
CLEAN
} action_t;
2002-03-16 01:59:12 +03:00
struct dev_manager {
struct dm_pool *mem;
2004-05-05 01:25:57 +04:00
struct cmd_context *cmd;
void *target_state;
uint32_t pvmove_mirror_count;
int flush_required;
unsigned track_pvmove_deps;
char *vg_name;
};
struct lv_layer {
struct logical_volume *lv;
const char *old_name;
};
static const char _thin_layer[] = "tpool";
int read_only_lv(struct logical_volume *lv, struct lv_activate_opts *laopts)
{
return (laopts->read_only || !(lv->vg->status & LVM_WRITE) || !(lv->status & LVM_WRITE));
}
2002-02-25 17:46:57 +03:00
/*
* Low level device-layer operations.
*/
static struct dm_task *_setup_task(const char *name, const char *uuid,
uint32_t *event_nr, int task,
uint32_t major, uint32_t minor)
2002-02-25 17:46:57 +03:00
{
struct dm_task *dmt;
2008-01-30 16:19:47 +03:00
if (!(dmt = dm_task_create(task)))
return_NULL;
2002-02-25 17:46:57 +03:00
2011-01-05 17:03:36 +03:00
if (name && !dm_task_set_name(dmt, name))
goto_out;
2011-01-05 17:03:36 +03:00
if (uuid && *uuid && !dm_task_set_uuid(dmt, uuid))
goto_out;
2011-01-05 17:03:36 +03:00
if (event_nr && !dm_task_set_event_nr(dmt, *event_nr))
goto_out;
2003-04-30 19:26:25 +04:00
2011-01-05 17:03:36 +03:00
if (major && !dm_task_set_major_minor(dmt, major, minor, 1))
goto_out;
if (activation_checks() && !dm_task_enable_checks(dmt))
goto_out;
2002-02-25 17:46:57 +03:00
return dmt;
2011-01-05 17:03:36 +03:00
out:
dm_task_destroy(dmt);
return NULL;
2002-02-25 17:46:57 +03:00
}
2005-10-19 17:59:18 +04:00
static int _info_run(const char *name, const char *dlid, struct dm_info *info,
uint32_t *read_ahead, int mknodes, int with_open_count,
int with_read_ahead, uint32_t major, uint32_t minor)
2002-03-15 00:17:30 +03:00
{
int r = 0;
struct dm_task *dmt;
int dmtask;
2002-03-15 00:17:30 +03:00
dmtask = mknodes ? DM_DEVICE_MKNODES : DM_DEVICE_INFO;
if (!(dmt = _setup_task(mknodes ? name : NULL, dlid, 0, dmtask, major, minor)))
2008-01-30 16:19:47 +03:00
return_0;
2002-03-15 00:17:30 +03:00
if (!with_open_count)
if (!dm_task_no_open_count(dmt))
log_error("Failed to disable open_count");
if (!dm_task_run(dmt))
goto_out;
2002-03-15 00:17:30 +03:00
if (!dm_task_get_info(dmt, info))
goto_out;
2002-03-19 02:25:50 +03:00
if (with_read_ahead && info->exists) {
2007-12-03 21:00:38 +03:00
if (!dm_task_get_read_ahead(dmt, read_ahead))
goto_out;
} else if (read_ahead)
2007-11-29 18:04:12 +03:00
*read_ahead = DM_READ_AHEAD_NONE;
2002-03-15 00:17:30 +03:00
r = 1;
out:
dm_task_destroy(dmt);
return r;
}
2002-02-25 17:46:57 +03:00
int device_is_usable(struct device *dev)
{
struct dm_task *dmt;
struct dm_info info;
const char *name, *uuid;
2008-01-30 17:00:02 +03:00
uint64_t start, length;
char *target_type = NULL;
char *params, *vgname = NULL, *lvname, *layer;
2007-01-26 02:03:48 +03:00
void *next = NULL;
int only_error_target = 1;
int r = 0;
if (!(dmt = dm_task_create(DM_DEVICE_STATUS)))
return_0;
if (!dm_task_set_major_minor(dmt, MAJOR(dev->dev), MINOR(dev->dev), 1))
goto_out;
if (activation_checks() && !dm_task_enable_checks(dmt))
goto_out;
if (!dm_task_run(dmt)) {
log_error("Failed to get state of mapped device");
goto out;
}
if (!dm_task_get_info(dmt, &info))
goto_out;
2010-08-09 18:05:16 +04:00
if (!info.exists)
goto out;
2007-01-26 02:03:48 +03:00
name = dm_task_get_name(dmt);
uuid = dm_task_get_uuid(dmt);
2007-01-26 02:03:48 +03:00
if (!info.target_count) {
log_debug("%s: Empty device %s not usable.", dev_name(dev), name);
goto out;
}
2010-08-09 18:05:16 +04:00
if (info.suspended && ignore_suspended_devices()) {
log_debug("%s: Suspended device %s not usable.", dev_name(dev), name);
goto out;
}
/* FIXME Also check for mirror block_on_error and mpath no paths */
2007-01-26 02:03:48 +03:00
/* For now, we exclude all mirrors */
2008-01-30 17:00:02 +03:00
do {
next = dm_get_next_target(dmt, next, &start, &length,
&target_type, &params);
/* Skip if target type doesn't match */
2010-08-09 18:05:16 +04:00
if (target_type && !strcmp(target_type, "mirror") && ignore_suspended_devices()) {
log_debug("%s: Mirror device %s not usable.", dev_name(dev), name);
2007-01-26 02:03:48 +03:00
goto out;
}
/*
* Snapshot origin could be sitting on top of a mirror which
* could be blocking I/O. Skip snapshot origins entirely for
* now.
*
* FIXME: rather than skipping origin, check if mirror is
* underneath and if the mirror is blocking I/O.
*/
if (target_type && !strcmp(target_type, "snapshot-origin") &&
ignore_suspended_devices()) {
log_debug("%s: Snapshot-origin device %s not usable.",
dev_name(dev), name);
goto out;
}
if (target_type && strcmp(target_type, "error"))
only_error_target = 0;
2008-01-30 17:00:02 +03:00
} while (next);
/* Skip devices consisting entirely of error targets. */
/* FIXME Deal with device stacked above error targets? */
if (only_error_target) {
log_debug("%s: Error device %s not usable.",
dev_name(dev), name);
goto out;
}
/* FIXME Also check dependencies? */
/* Check internal lvm devices */
if (uuid && !strncmp(uuid, UUID_PREFIX, sizeof(UUID_PREFIX) - 1)) {
if (!(vgname = dm_strdup(name)) ||
!dm_split_lvm_name(NULL, NULL, &vgname, &lvname, &layer))
goto_out;
if (lvname && (is_reserved_lvname(lvname) || *layer)) {
log_debug("%s: Reserved internal LV device %s/%s%s%s not usable.",
dev_name(dev), vgname, lvname, *layer ? "-" : "", layer);
goto out;
}
}
r = 1;
out:
dm_free(vgname);
dm_task_destroy(dmt);
return r;
}
static int _info(const char *dlid, int with_open_count, int with_read_ahead,
struct dm_info *info, uint32_t *read_ahead)
{
int r = 0;
if ((r = _info_run(NULL, dlid, info, read_ahead, 0, with_open_count,
with_read_ahead, 0, 0)) && info->exists)
return 1;
else if ((r = _info_run(NULL, dlid + sizeof(UUID_PREFIX) - 1, info,
read_ahead, 0, with_open_count,
with_read_ahead, 0, 0)) && info->exists)
return 1;
return r;
}
static int _info_by_dev(uint32_t major, uint32_t minor, struct dm_info *info)
{
return _info_run(NULL, NULL, info, NULL, 0, 0, 0, major, minor);
}
int dev_manager_info(struct dm_pool *mem, const struct logical_volume *lv,
const char *layer,
int with_open_count, int with_read_ahead,
struct dm_info *info, uint32_t *read_ahead)
{
char *dlid, *name;
int r;
if (!(name = dm_build_dm_name(mem, lv->vg->name, lv->name, layer))) {
log_error("name build failed for %s", lv->name);
return 0;
}
2005-10-19 17:59:18 +04:00
if (!(dlid = build_dm_uuid(mem, lv->lvid.s, layer))) {
log_error("dlid build failed for %s", name);
2005-10-19 17:59:18 +04:00
return 0;
}
log_debug("Getting device info for %s [%s]", name, dlid);
r = _info(dlid, with_open_count, with_read_ahead, info, read_ahead);
dm_pool_free(mem, name);
return r;
}
static const struct dm_info *_cached_info(struct dm_pool *mem,
const struct logical_volume *lv,
struct dm_tree *dtree)
{
const char *dlid;
struct dm_tree_node *dnode;
const struct dm_info *dinfo;
if (!(dlid = build_dm_uuid(mem, lv->lvid.s, NULL))) {
log_error("dlid build failed for %s", lv->name);
return NULL;
}
/* An activating merging origin won't have a node in the tree yet */
if (!(dnode = dm_tree_find_node_by_uuid(dtree, dlid)))
return NULL;
if (!(dinfo = dm_tree_node_get_info(dnode))) {
log_error("failed to get info from tree node for %s", lv->name);
return NULL;
}
if (!dinfo->exists)
return NULL;
return dinfo;
}
#if 0
2002-05-22 18:03:45 +04:00
/* FIXME Interface must cope with multiple targets */
static int _status_run(const char *name, const char *uuid,
unsigned long long *s, unsigned long long *l,
char **t, uint32_t t_size, char **p, uint32_t p_size)
{
int r = 0;
struct dm_task *dmt;
struct dm_info info;
void *next = NULL;
uint64_t start, length;
char *type = NULL;
char *params = NULL;
if (!(dmt = _setup_task(name, uuid, 0, DM_DEVICE_STATUS, 0, 0)))
2008-01-30 16:19:47 +03:00
return_0;
if (!dm_task_no_open_count(dmt))
log_error("Failed to disable open_count");
if (!dm_task_run(dmt))
goto_out;
if (!dm_task_get_info(dmt, &info) || !info.exists)
goto_out;
do {
next = dm_get_next_target(dmt, next, &start, &length,
&type, &params);
2002-05-22 18:03:45 +04:00
if (type) {
*s = start;
*l = length;
/* Make sure things are null terminated */
strncpy(*t, type, t_size);
2002-05-22 18:03:45 +04:00
(*t)[t_size - 1] = '\0';
strncpy(*p, params, p_size);
2002-05-22 18:03:45 +04:00
(*p)[p_size - 1] = '\0';
r = 1;
2002-05-22 18:03:45 +04:00
/* FIXME Cope with multiple targets! */
break;
}
2002-05-22 18:03:45 +04:00
} while (next);
out:
dm_task_destroy(dmt);
return r;
}
static int _status(const char *name, const char *uuid,
unsigned long long *start, unsigned long long *length,
char **type, uint32_t type_size, char **params,
uint32_t param_size) __attribute__ ((unused));
static int _status(const char *name, const char *uuid,
unsigned long long *start, unsigned long long *length,
char **type, uint32_t type_size, char **params,
uint32_t param_size)
{
if (uuid && *uuid) {
if (_status_run(NULL, uuid, start, length, type,
type_size, params, param_size) &&
*params)
return 1;
else if (_status_run(NULL, uuid + sizeof(UUID_PREFIX) - 1, start,
length, type, type_size, params,
param_size) &&
*params)
return 1;
}
if (name && _status_run(name, NULL, start, length, type, type_size,
2002-05-22 18:03:45 +04:00
params, param_size))
return 1;
2002-05-22 18:03:45 +04:00
return 0;
}
#endif
int lv_has_target_type(struct dm_pool *mem, struct logical_volume *lv,
const char *layer, const char *target_type)
{
int r = 0;
char *dlid;
struct dm_task *dmt;
struct dm_info info;
void *next = NULL;
uint64_t start, length;
char *type = NULL;
char *params = NULL;
if (!(dlid = build_dm_uuid(mem, lv->lvid.s, layer)))
return_0;
if (!(dmt = _setup_task(NULL, dlid, 0,
DM_DEVICE_STATUS, 0, 0)))
goto_bad;
if (!dm_task_no_open_count(dmt))
log_error("Failed to disable open_count");
if (!dm_task_run(dmt))
goto_out;
if (!dm_task_get_info(dmt, &info) || !info.exists)
goto_out;
do {
next = dm_get_next_target(dmt, next, &start, &length,
&type, &params);
if (type && strncmp(type, target_type,
strlen(target_type)) == 0) {
if (info.live_table)
r = 1;
break;
}
} while (next);
out:
dm_task_destroy(dmt);
bad:
dm_pool_free(mem, dlid);
return r;
}
int add_linear_area_to_dtree(struct dm_tree_node *node, uint64_t size, uint32_t extent_size, int use_linear_target, const char *vgname, const char *lvname)
{
uint32_t page_size;
/*
* Use striped or linear target?
*/
if (!use_linear_target) {
page_size = lvm_getpagesize() >> SECTOR_SHIFT;
/*
* We'll use the extent size as the stripe size.
* Extent size and page size are always powers of 2.
* The striped target requires that the stripe size is
* divisible by the page size.
*/
if (extent_size >= page_size) {
/* Use striped target */
if (!dm_tree_node_add_striped_target(node, size, extent_size))
return_0;
return 1;
} else
/* Some exotic cases are unsupported by striped. */
log_warn("WARNING: Using linear target for %s/%s: Striped requires extent size (%" PRIu32 " sectors) >= page size (%" PRIu32 ").",
vgname, lvname, extent_size, page_size);
}
/*
* Use linear target.
*/
if (!dm_tree_node_add_linear_target(node, size))
return_0;
return 1;
}
static percent_range_t _combine_percent(percent_t a, percent_t b,
uint32_t numerator, uint32_t denominator)
{
if (a == PERCENT_MERGE_FAILED || b == PERCENT_MERGE_FAILED)
return PERCENT_MERGE_FAILED;
if (a == PERCENT_INVALID || b == PERCENT_INVALID)
return PERCENT_INVALID;
if (a == PERCENT_100 && b == PERCENT_100)
return PERCENT_100;
if (a == PERCENT_0 && b == PERCENT_0)
return PERCENT_0;
return (percent_range_t) make_percent(numerator, denominator);
}
static int _percent_run(struct dev_manager *dm, const char *name,
2005-10-19 17:59:18 +04:00
const char *dlid,
const char *target_type, int wait,
const struct logical_volume *lv, percent_t *overall_percent,
uint32_t *event_nr, int fail_if_percent_unsupported)
{
int r = 0;
struct dm_task *dmt;
2003-04-30 19:26:25 +04:00
struct dm_info info;
void *next = NULL;
uint64_t start, length;
char *type = NULL;
char *params = NULL;
const struct dm_list *segh = lv ? &lv->segments : NULL;
struct lv_segment *seg = NULL;
2004-05-05 01:25:57 +04:00
struct segment_type *segtype;
int first_time = 1;
percent_t percent = PERCENT_INVALID;
uint64_t total_numerator = 0, total_denominator = 0;
*overall_percent = percent;
2005-10-19 17:59:18 +04:00
if (!(dmt = _setup_task(name, dlid, event_nr,
wait ? DM_DEVICE_WAITEVENT : DM_DEVICE_STATUS, 0, 0)))
2008-01-30 16:19:47 +03:00
return_0;
if (!dm_task_no_open_count(dmt))
log_error("Failed to disable open_count");
if (!dm_task_run(dmt))
goto_out;
if (!dm_task_get_info(dmt, &info) || !info.exists)
goto_out;
2003-04-30 19:26:25 +04:00
if (event_nr)
*event_nr = info.event_nr;
do {
next = dm_get_next_target(dmt, next, &start, &length, &type,
&params);
if (lv) {
if (!(segh = dm_list_next(&lv->segments, segh))) {
log_error("Number of segments in active LV %s "
"does not match metadata", lv->name);
goto out;
}
seg = dm_list_item(segh, struct lv_segment);
}
if (!type || !params)
continue;
if (!(segtype = get_segtype_from_string(dm->cmd, target_type)))
2003-04-30 19:26:25 +04:00
continue;
if (strcmp(type, target_type)) {
/* If kernel's type isn't an exact match is it compatible? */
if (!segtype->ops->target_status_compatible ||
!segtype->ops->target_status_compatible(type))
continue;
}
if (!segtype->ops->target_percent)
continue;
if (!segtype->ops->target_percent(&dm->target_state,
&percent, dm->mem,
dm->cmd, seg, params,
2004-05-05 01:25:57 +04:00
&total_numerator,
&total_denominator))
goto_out;
2004-05-05 01:25:57 +04:00
if (first_time) {
*overall_percent = percent;
first_time = 0;
} else
*overall_percent =
_combine_percent(*overall_percent, percent,
total_numerator, total_denominator);
} while (next);
if (lv && dm_list_next(&lv->segments, segh)) {
log_error("Number of segments in active LV %s does not "
"match metadata", lv->name);
goto out;
}
if (first_time) {
/* above ->target_percent() was not executed! */
/* FIXME why return PERCENT_100 et. al. in this case? */
*overall_percent = PERCENT_100;
if (fail_if_percent_unsupported)
goto_out;
}
log_debug("LV percent: %f", percent_to_float(*overall_percent));
r = 1;
out:
dm_task_destroy(dmt);
return r;
}
2005-10-19 17:59:18 +04:00
static int _percent(struct dev_manager *dm, const char *name, const char *dlid,
const char *target_type, int wait,
const struct logical_volume *lv, percent_t *percent,
uint32_t *event_nr, int fail_if_percent_unsupported)
{
if (dlid && *dlid) {
if (_percent_run(dm, NULL, dlid, target_type, wait, lv, percent,
event_nr, fail_if_percent_unsupported))
return 1;
else if (_percent_run(dm, NULL, dlid + sizeof(UUID_PREFIX) - 1,
target_type, wait, lv, percent,
event_nr, fail_if_percent_unsupported))
return 1;
}
if (name && _percent_run(dm, name, NULL, target_type, wait, lv, percent,
event_nr, fail_if_percent_unsupported))
return 1;
return 0;
}
/* FIXME Merge with the percent function */
int dev_manager_transient(struct dev_manager *dm, struct logical_volume *lv)
{
int r = 0;
struct dm_task *dmt;
struct dm_info info;
void *next = NULL;
uint64_t start, length;
char *type = NULL;
char *params = NULL;
char *dlid = NULL;
const char *layer = lv_is_origin(lv) ? "real" : NULL;
const struct dm_list *segh = &lv->segments;
struct lv_segment *seg = NULL;
if (!(dlid = build_dm_uuid(dm->mem, lv->lvid.s, layer)))
return_0;
if (!(dmt = _setup_task(0, dlid, NULL, DM_DEVICE_STATUS, 0, 0)))
return_0;
if (!dm_task_no_open_count(dmt))
log_error("Failed to disable open_count");
if (!dm_task_run(dmt))
goto_out;
if (!dm_task_get_info(dmt, &info) || !info.exists)
goto_out;
do {
next = dm_get_next_target(dmt, next, &start, &length, &type,
&params);
if (!(segh = dm_list_next(&lv->segments, segh))) {
log_error("Number of segments in active LV %s "
"does not match metadata", lv->name);
goto out;
}
seg = dm_list_item(segh, struct lv_segment);
if (!type || !params)
continue;
if (!seg) {
log_error(INTERNAL_ERROR "Segment is not selected.");
goto out;
}
if (seg->segtype->ops->check_transient_status &&
!seg->segtype->ops->check_transient_status(seg, params))
goto_out;
} while (next);
if (dm_list_next(&lv->segments, segh)) {
log_error("Number of segments in active LV %s does not "
"match metadata", lv->name);
goto out;
}
r = 1;
out:
dm_task_destroy(dmt);
return r;
}
/*
* dev_manager implementation.
*/
struct dev_manager *dev_manager_create(struct cmd_context *cmd,
const char *vg_name,
unsigned track_pvmove_deps)
2002-03-19 02:25:50 +03:00
{
struct dm_pool *mem;
struct dev_manager *dm;
2002-03-19 02:25:50 +03:00
2008-01-30 16:19:47 +03:00
if (!(mem = dm_pool_create("dev_manager", 16 * 1024)))
return_NULL;
2002-03-19 02:25:50 +03:00
if (!(dm = dm_pool_zalloc(mem, sizeof(*dm))))
goto_bad;
2002-03-19 02:25:50 +03:00
dm->cmd = cmd;
dm->mem = mem;
if (!(dm->vg_name = dm_pool_strdup(dm->mem, vg_name)))
goto_bad;
/*
* When we manipulate (normally suspend/resume) the PVMOVE
* device directly, there's no need to touch the LVs above.
*/
dm->track_pvmove_deps = track_pvmove_deps;
dm->target_state = NULL;
2009-08-04 19:36:13 +04:00
dm_udev_set_sync_support(cmd->current_settings.udev_sync);
return dm;
bad:
dm_pool_destroy(mem);
return NULL;
}
void dev_manager_destroy(struct dev_manager *dm)
{
dm_pool_destroy(dm->mem);
}
void dev_manager_release(void)
{
dm_lib_release();
}
void dev_manager_exit(void)
{
dm_lib_exit();
}
int dev_manager_snapshot_percent(struct dev_manager *dm,
2006-04-06 18:06:27 +04:00
const struct logical_volume *lv,
percent_t *percent)
{
const struct logical_volume *snap_lv;
char *name;
const char *dlid;
int fail_if_percent_unsupported = 0;
if (lv_is_merging_origin(lv)) {
/*
* Set 'fail_if_percent_unsupported', otherwise passing
* unsupported LV types to _percent will lead to a default
* successful return with percent_range as PERCENT_100.
* - For a merging origin, this will result in a polldaemon
* that runs infinitely (because completion is PERCENT_0)
* - We unfortunately don't yet _know_ if a snapshot-merge
* target is active (activation is deferred if dev is open);
* so we can't short-circuit origin devices based purely on
* existing LVM LV attributes.
*/
fail_if_percent_unsupported = 1;
}
if (lv_is_merging_cow(lv)) {
/* must check percent of origin for a merging snapshot */
snap_lv = origin_from_cow(lv);
} else
snap_lv = lv;
/*
* Build a name for the top layer.
*/
if (!(name = dm_build_dm_name(dm->mem, snap_lv->vg->name, snap_lv->name, NULL)))
return_0;
if (!(dlid = build_dm_uuid(dm->mem, snap_lv->lvid.s, NULL)))
return_0;
/*
* Try and get some info on this device.
*/
log_debug("Getting device status percentage for %s", name);
if (!(_percent(dm, name, dlid, "snapshot", 0, NULL, percent,
NULL, fail_if_percent_unsupported)))
2008-01-30 16:19:47 +03:00
return_0;
/* If the snapshot isn't available, percent will be -1 */
return 1;
}
/* FIXME Merge with snapshot_percent, auto-detecting target type */
/* FIXME Cope with more than one target */
int dev_manager_mirror_percent(struct dev_manager *dm,
const struct logical_volume *lv, int wait,
percent_t *percent, uint32_t *event_nr)
2002-02-25 17:46:57 +03:00
{
char *name;
const char *dlid;
const char *target_type = first_seg(lv)->segtype->name;
const char *layer = (lv_is_origin(lv)) ? "real" : NULL;
2002-02-25 17:46:57 +03:00
2002-03-26 16:41:37 +03:00
/*
* Build a name for the top layer.
2002-03-26 16:41:37 +03:00
*/
if (!(name = dm_build_dm_name(dm->mem, lv->vg->name, lv->name, layer)))
return_0;
if (!(dlid = build_dm_uuid(dm->mem, lv->lvid.s, layer))) {
log_error("dlid build failed for %s", lv->name);
return 0;
}
2002-03-15 00:17:30 +03:00
log_debug("Getting device %s status percentage for %s",
target_type, name);
if (!(_percent(dm, name, dlid, target_type, wait, lv, percent,
event_nr, 0)))
2008-01-30 16:19:47 +03:00
return_0;
2002-03-15 00:17:30 +03:00
return 1;
}
#if 0
log_very_verbose("%s %s", sus ? "Suspending" : "Resuming", name);
log_verbose("Loading %s", dl->name);
log_very_verbose("Activating %s read-only", dl->name);
log_very_verbose("Activated %s %s %03u:%03u", dl->name,
dl->dlid, dl->info.major, dl->info.minor);
2002-03-16 01:59:12 +03:00
if (_get_flag(dl, VISIBLE))
log_verbose("Removing %s", dl->name);
2002-03-16 01:59:12 +03:00
else
log_very_verbose("Removing %s", dl->name);
2002-03-16 01:59:12 +03:00
log_debug("Adding target: %" PRIu64 " %" PRIu64 " %s %s",
extent_size * seg->le, extent_size * seg->len, target, params);
2002-02-25 17:46:57 +03:00
log_debug("Adding target: 0 %" PRIu64 " snapshot-origin %s",
dl->lv->size, params);
log_debug("Adding target: 0 %" PRIu64 " snapshot %s", size, params);
log_debug("Getting device info for %s", dl->name);
/* Rename? */
if ((suffix = strrchr(dl->dlid + sizeof(UUID_PREFIX) - 1, '-')))
suffix++;
new_name = dm_build_dm_name(dm->mem, dm->vg_name, dl->lv->name,
suffix);
2002-03-16 01:59:12 +03:00
static int _belong_to_vg(const char *vgname, const char *name)
{
const char *v = vgname, *n = name;
2002-02-25 17:46:57 +03:00
while (*v) {
if ((*v != *n) || (*v == '-' && *(++n) != '-'))
return 0;
v++, n++;
2002-04-16 18:42:20 +04:00
}
if (*n == '-' && *(n + 1) != '-')
return 1;
else
return 0;
}
2002-04-16 18:42:20 +04:00
2005-04-07 16:39:44 +04:00
if (!(snap_seg = find_cow(lv)))
2002-04-16 18:42:20 +04:00
return 1;
2005-04-07 16:39:44 +04:00
old_origin = snap_seg->origin;
2002-04-16 18:42:20 +04:00
/* Was this the last active snapshot with this origin? */
dm_list_iterate_items(lvl, active_head) {
2005-06-01 20:51:55 +04:00
active = lvl->lv;
2005-04-07 16:39:44 +04:00
if ((snap_seg = find_cow(active)) &&
snap_seg->origin == old_origin) {
2002-04-16 18:42:20 +04:00
return 1;
}
}
#endif
int dev_manager_thin_pool_status(struct dev_manager *dm,
const struct logical_volume *lv,
struct dm_status_thin_pool **status)
{
const char *dlid;
struct dm_task *dmt;
struct dm_info info;
uint64_t start, length;
char *type = NULL;
char *params = NULL;
int r = 0;
/* Build dlid for the thin pool layer */
if (!(dlid = build_dm_uuid(dm->mem, lv->lvid.s, _thin_layer)))
return_0;
log_debug("Getting thin pool device status for %s.", lv->name);
if (!(dmt = _setup_task(NULL, dlid, 0, DM_DEVICE_STATUS, 0, 0)))
return_0;
if (!dm_task_no_open_count(dmt))
log_error("Failed to disable open_count.");
if (!dm_task_run(dmt))
goto_out;
if (!dm_task_get_info(dmt, &info) || !info.exists)
goto_out;
dm_get_next_target(dmt, NULL, &start, &length, &type, &params);
if (!dm_get_status_thin_pool(dm->mem, params, status))
goto_out;
r = 1;
out:
dm_task_destroy(dmt);
return r;
}
int dev_manager_thin_pool_percent(struct dev_manager *dm,
const struct logical_volume *lv,
int metadata, percent_t *percent)
{
char *name;
const char *dlid;
/* Build a name for the top layer */
if (!(name = dm_build_dm_name(dm->mem, lv->vg->name, lv->name,
_thin_layer)))
return_0;
if (!(dlid = build_dm_uuid(dm->mem, lv->lvid.s, _thin_layer)))
return_0;
log_debug("Getting device status percentage for %s", name);
if (!(_percent(dm, name, dlid, "thin-pool", 0,
(metadata) ? lv : NULL, percent, NULL, 1)))
return_0;
return 1;
}
int dev_manager_thin_percent(struct dev_manager *dm,
const struct logical_volume *lv,
int mapped, percent_t *percent)
{
char *name;
const char *dlid;
const char *layer = lv_is_origin(lv) ? "real" : NULL;
/* Build a name for the top layer */
if (!(name = dm_build_dm_name(dm->mem, lv->vg->name, lv->name, layer)))
return_0;
if (!(dlid = build_dm_uuid(dm->mem, lv->lvid.s, layer)))
return_0;
log_debug("Getting device status percentage for %s", name);
if (!(_percent(dm, name, dlid, "thin", 0,
(mapped) ? NULL : lv, percent, NULL, 1)))
return_0;
return 1;
}
/*************************/
/* NEW CODE STARTS HERE */
/*************************/
static int _dev_manager_lv_mknodes(const struct logical_volume *lv)
{
char *name;
if (!(name = dm_build_dm_name(lv->vg->cmd->mem, lv->vg->name,
lv->name, NULL)))
return_0;
return fs_add_lv(lv, name);
}
static int _dev_manager_lv_rmnodes(const struct logical_volume *lv)
{
return fs_del_lv(lv);
}
int dev_manager_mknodes(const struct logical_volume *lv)
{
struct dm_info dminfo;
char *name;
int r = 0;
if (!(name = dm_build_dm_name(lv->vg->cmd->mem, lv->vg->name, lv->name, NULL)))
return_0;
if ((r = _info_run(name, NULL, &dminfo, NULL, 1, 0, 0, 0, 0))) {
if (dminfo.exists) {
if (lv_is_visible(lv))
r = _dev_manager_lv_mknodes(lv);
} else
r = _dev_manager_lv_rmnodes(lv);
}
dm_pool_free(lv->vg->cmd->mem, name);
return r;
}
static uint16_t _get_udev_flags(struct dev_manager *dm, struct logical_volume *lv,
const char *layer)
{
uint16_t udev_flags = 0;
/*
* Instruct also libdevmapper to disable udev
* fallback in accordance to LVM2 settings.
*/
if (!dm->cmd->current_settings.udev_fallback)
udev_flags |= DM_UDEV_DISABLE_LIBRARY_FALLBACK;
/*
* Is this top-level and visible device?
* If not, create just the /dev/mapper content.
*/
2011-10-29 00:34:45 +04:00
/* FIXME: add target's method for this */
if (layer || !lv_is_visible(lv) || lv_is_thin_pool(lv))
udev_flags |= DM_UDEV_DISABLE_SUBSYSTEM_RULES_FLAG |
DM_UDEV_DISABLE_DISK_RULES_FLAG |
DM_UDEV_DISABLE_OTHER_RULES_FLAG;
/*
* There's no need for other udev rules to touch special LVs with
* reserved names. We don't need to populate /dev/disk here either.
* Even if they happen to be visible and top-level.
*/
else if (is_reserved_lvname(lv->name))
udev_flags |= DM_UDEV_DISABLE_DISK_RULES_FLAG |
DM_UDEV_DISABLE_OTHER_RULES_FLAG;
/*
* Snapshots and origins could have the same rule applied that will
* give symlinks exactly the same name (e.g. a name based on
* filesystem UUID). We give preference to origins to make such
* naming deterministic (e.g. symlinks in /dev/disk/by-uuid).
*/
if (lv_is_cow(lv))
udev_flags |= DM_UDEV_LOW_PRIORITY_FLAG;
/*
* Finally, add flags to disable /dev/mapper and /dev/<vgname> content
* to be created by udev if it is requested by user's configuration.
* This is basically an explicit fallback to old node/symlink creation
* without udev.
*/
if (!dm->cmd->current_settings.udev_rules)
udev_flags |= DM_UDEV_DISABLE_DM_RULES_FLAG |
DM_UDEV_DISABLE_SUBSYSTEM_RULES_FLAG;
return udev_flags;
}
2005-11-09 16:08:41 +03:00
static int _add_dev_to_dtree(struct dev_manager *dm, struct dm_tree *dtree,
struct logical_volume *lv, const char *layer)
{
char *dlid, *name;
struct dm_info info, info2;
if (!(name = dm_build_dm_name(dm->mem, lv->vg->name, lv->name, layer)))
return_0;
if (!(dlid = build_dm_uuid(dm->mem, lv->lvid.s, layer)))
return_0;
2008-01-30 17:00:02 +03:00
log_debug("Getting device info for %s [%s]", name, dlid);
if (!_info(dlid, 1, 0, &info, NULL)) {
2008-01-30 17:00:02 +03:00
log_error("Failed to get info for %s [%s].", name, dlid);
return 0;
}
/*
* For top level volumes verify that existing device match
* requested major/minor and that major/minor pair is available for use
*/
if (!layer && lv->major != -1 && lv->minor != -1) {
/*
* FIXME compare info.major with lv->major if multiple major support
*/
if (info.exists && (info.minor != lv->minor)) {
log_error("Volume %s (%" PRIu32 ":%" PRIu32")"
" differs from already active device "
"(%" PRIu32 ":%" PRIu32")",
lv->name, lv->major, lv->minor, info.major, info.minor);
return 0;
}
if (!info.exists && _info_by_dev(lv->major, lv->minor, &info2) &&
info2.exists) {
log_error("The requested major:minor pair "
"(%" PRIu32 ":%" PRIu32") is already used",
lv->major, lv->minor);
return 0;
}
}
if (info.exists && !dm_tree_add_dev_with_udev_flags(dtree, info.major, info.minor,
_get_udev_flags(dm, lv, layer))) {
2005-11-09 16:08:41 +03:00
log_error("Failed to add device (%" PRIu32 ":%" PRIu32") to dtree",
info.major, info.minor);
return 0;
}
return 1;
}
/*
* Add replicator devices
*
* Using _add_dev_to_dtree() directly instead of _add_lv_to_dtree()
* to avoid extra checks with extensions.
*/
static int _add_partial_replicator_to_dtree(struct dev_manager *dm,
struct dm_tree *dtree,
struct logical_volume *lv)
{
struct logical_volume *rlv = first_seg(lv)->replicator;
struct replicator_device *rdev;
struct replicator_site *rsite;
struct dm_tree_node *rep_node, *rdev_node;
const char *uuid;
if (!lv_is_active_replicator_dev(lv)) {
if (!_add_dev_to_dtree(dm, dtree, lv->rdevice->lv,
NULL))
return_0;
return 1;
}
/* Add _rlog and replicator device */
if (!_add_dev_to_dtree(dm, dtree, first_seg(rlv)->rlog_lv, NULL))
return_0;
if (!_add_dev_to_dtree(dm, dtree, rlv, NULL))
return_0;
if (!(uuid = build_dm_uuid(dm->mem, rlv->lvid.s, NULL)))
return_0;
rep_node = dm_tree_find_node_by_uuid(dtree, uuid);
/* Add all related devices for replicator */
dm_list_iterate_items(rsite, &rlv->rsites)
dm_list_iterate_items(rdev, &rsite->rdevices) {
if (rsite->state == REPLICATOR_STATE_ACTIVE) {
/* Add _rimage LV */
if (!_add_dev_to_dtree(dm, dtree, rdev->lv, NULL))
return_0;
/* Add replicator-dev LV, except of the already added one */
if ((lv != rdev->replicator_dev->lv) &&
!_add_dev_to_dtree(dm, dtree,
rdev->replicator_dev->lv, NULL))
return_0;
/* If replicator exists - try connect existing heads */
if (rep_node) {
uuid = build_dm_uuid(dm->mem,
rdev->replicator_dev->lv->lvid.s,
NULL);
if (!uuid)
return_0;
rdev_node = dm_tree_find_node_by_uuid(dtree, uuid);
if (rdev_node)
dm_tree_node_set_presuspend_node(rdev_node,
rep_node);
}
}
if (!rdev->rsite->vg_name)
continue;
if (!_add_dev_to_dtree(dm, dtree, rdev->lv, NULL))
return_0;
if (rdev->slog &&
!_add_dev_to_dtree(dm, dtree, rdev->slog, NULL))
return_0;
}
return 1;
}
struct thin_cb_data {
const struct logical_volume *pool_lv;
struct dev_manager *dm;
};
static int _thin_pool_callback(struct dm_tree_node *node,
dm_node_callback_t type, void *cb_data)
{
int ret, status;
const struct thin_cb_data *data = cb_data;
const char *dmdir = dm_dir();
const struct dm_config_node *cn;
const struct dm_config_value *cv;
const char *thin_check =
find_config_tree_str_allow_empty(data->pool_lv->vg->cmd,
"global/thin_check_executable",
THIN_CHECK_CMD);
const struct logical_volume *mlv = first_seg(data->pool_lv)->metadata_lv;
size_t len = strlen(dmdir) + 2 * (strlen(mlv->vg->name) + strlen(mlv->name)) + 3;
char meta_path[len];
int args = 0;
const char *argv[19]; /* Max supported 15 args */
char *split, *dm_name;
if (!thin_check[0])
return 1; /* Checking disabled */
if (!(dm_name = dm_build_dm_name(data->dm->mem, mlv->vg->name,
mlv->name, NULL)) ||
(dm_snprintf(meta_path, len, "%s/%s", dmdir, dm_name) < 0)) {
log_error("Failed to build thin metadata path.");
return 0;
}
if ((cn = find_config_tree_node(mlv->vg->cmd, "global/thin_check_options"))) {
for (cv = cn->v; cv && args < 16; cv = cv->next) {
if (cv->type != DM_CFG_STRING) {
log_error("Invalid string in config file: "
"global/thin_check_options");
return 0;
}
argv[++args] = cv->v.str;
}
} else {
/* Use default options (no support for options with spaces) */
if (!(split = dm_pool_strdup(data->dm->mem, DEFAULT_THIN_CHECK_OPTIONS))) {
log_error("Failed to duplicate thin check string.");
return 0;
}
args = dm_split_words(split, 16, 0, (char**) argv + 1);
}
if (args == 16) {
log_error("Too many options for thin check command.");
return 0;
}
argv[0] = thin_check;
argv[++args] = meta_path;
argv[++args] = NULL;
if (!(ret = exec_cmd(data->pool_lv->vg->cmd, (const char * const *)argv,
&status, 0))) {
switch (type) {
case DM_NODE_CALLBACK_PRELOADED:
log_err_once("Check of thin pool %s/%s failed (status:%d). "
"Manual repair required (thin_dump --repair %s)!",
data->pool_lv->vg->name, data->pool_lv->name,
status, meta_path);
break;
default:
log_warn("WARNING: Integrity check of metadata for thin pool "
"%s/%s failed.",
data->pool_lv->vg->name, data->pool_lv->name);
}
/*
* FIXME: What should we do here??
*
* Maybe mark the node, so it's not activating
* as thin_pool but as error/linear and let the
* dm tree resolve the issue.
*/
}
dm_pool_free(data->dm->mem, dm_name);
return ret;
}
static int _thin_pool_register_callback(struct dev_manager *dm,
struct dm_tree_node *node,
const struct logical_volume *lv)
{
struct thin_cb_data *data;
/* Skip metadata testing for unused pool. */
if (!first_seg(lv)->transaction_id)
return 1;
if (!(data = dm_pool_alloc(dm->mem, sizeof(*data)))) {
log_error("Failed to allocated path for callback.");
return 0;
}
data->dm = dm;
data->pool_lv = lv;
dm_tree_node_set_callback(node, _thin_pool_callback, data);
return 1;
}
/*
* Add LV and any known dependencies
*/
static int _add_lv_to_dtree(struct dev_manager *dm, struct dm_tree *dtree,
struct logical_volume *lv, int origin_only)
{
uint32_t s;
struct seg_list *sl;
struct lv_segment *seg = first_seg(lv);
struct dm_tree_node *thin_node;
const char *uuid;
if ((!origin_only || lv_is_thin_volume(lv)) &&
!_add_dev_to_dtree(dm, dtree, lv, NULL))
return_0;
/* FIXME Can we avoid doing this every time? */
2005-11-09 16:08:41 +03:00
if (!_add_dev_to_dtree(dm, dtree, lv, "real"))
return_0;
if (!origin_only && !_add_dev_to_dtree(dm, dtree, lv, "cow"))
return_0;
if ((lv->status & MIRRORED) && seg->log_lv &&
!_add_dev_to_dtree(dm, dtree, seg->log_lv, NULL))
return_0;
if (lv->status & RAID)
for (s = 0; s < seg->area_count; s++)
if (!_add_dev_to_dtree(dm, dtree,
seg_metalv(seg, s), NULL))
return_0;
/* Add any LVs referencing a PVMOVE LV unless told not to. */
if (dm->track_pvmove_deps && lv->status & PVMOVE)
dm_list_iterate_items(sl, &lv->segs_using_this_lv)
if (!_add_lv_to_dtree(dm, dtree, sl->seg->lv, origin_only))
return_0;
/* Adding LV head of replicator adds all other related devs */
if (lv_is_replicator_dev(lv) &&
!_add_partial_replicator_to_dtree(dm, dtree, lv))
return_0;
if (lv_is_thin_volume(lv)) {
2012-01-25 15:10:06 +04:00
#if 0
/* FIXME Implement dm_tree_node_skip_children optimisation */
if (origin_only) {
if (!(uuid = build_dm_uuid(dm->mem, lv->lvid.s, NULL)))
return_0;
if ((thin_node = dm_tree_find_node_by_uuid(dtree, uuid)))
2012-01-25 15:10:06 +04:00
dm_tree_node_skip_children(thin_node, 1);
}
2012-01-25 15:10:06 +04:00
#endif
/* Add thin pool LV layer */
lv = seg->pool_lv;
seg = first_seg(lv);
}
if (lv_is_thin_pool(lv)) {
if (!_add_lv_to_dtree(dm, dtree, seg->metadata_lv, 0))
return_0;
/* FIXME code from _create_partial_dtree() should be moved here */
if (!_add_lv_to_dtree(dm, dtree, seg_lv(seg, 0), 0))
return_0;
if (!_add_dev_to_dtree(dm, dtree, lv, _thin_layer))
return_0;
/* If the partial tree is used for deactivation, setup callback */
if (!(uuid = build_dm_uuid(dm->mem, lv->lvid.s, _thin_layer)))
return_0;
if ((thin_node = dm_tree_find_node_by_uuid(dtree, uuid)) &&
!_thin_pool_register_callback(dm, thin_node, lv))
return_0;
}
return 1;
}
static struct dm_tree *_create_partial_dtree(struct dev_manager *dm, struct logical_volume *lv, int origin_only)
{
2005-11-09 16:05:17 +03:00
struct dm_tree *dtree;
struct dm_list *snh;
struct lv_segment *seg;
uint32_t s;
2005-11-09 16:05:17 +03:00
if (!(dtree = dm_tree_create())) {
log_debug("Partial dtree creation failed for %s.", lv->name);
return NULL;
}
if (!_add_lv_to_dtree(dm, dtree, lv, (lv_is_origin(lv) || lv_is_thin_volume(lv)) ? origin_only : 0))
2008-01-30 16:19:47 +03:00
goto_bad;
/* Add any snapshots of this LV */
if (!origin_only && lv_is_origin(lv))
dm_list_iterate(snh, &lv->snapshot_segs)
if (!_add_lv_to_dtree(dm, dtree, dm_list_struct_base(snh, struct lv_segment, origin_list)->cow, 0))
goto_bad;
/* Add any LVs used by segments in this LV */
dm_list_iterate_items(seg, &lv->segments)
for (s = 0; s < seg->area_count; s++)
if (seg_type(seg, s) == AREA_LV && seg_lv(seg, s)) {
if (!_add_lv_to_dtree(dm, dtree, seg_lv(seg, s), 0))
2008-01-30 16:19:47 +03:00
goto_bad;
}
return dtree;
2008-01-30 16:19:47 +03:00
bad:
2005-11-09 16:05:17 +03:00
dm_tree_free(dtree);
return NULL;
}
static char *_add_error_device(struct dev_manager *dm, struct dm_tree *dtree,
struct lv_segment *seg, int s)
{
char *dlid, *name;
char errid[32];
struct dm_tree_node *node;
struct lv_segment *seg_i;
struct dm_info info;
int segno = -1, i = 0;
uint64_t size = (uint64_t) seg->len * seg->lv->vg->extent_size;
dm_list_iterate_items(seg_i, &seg->lv->segments) {
if (seg == seg_i)
segno = i;
++i;
}
if (segno < 0) {
log_error("_add_error_device called with bad segment");
2011-10-11 13:02:20 +04:00
return NULL;
}
sprintf(errid, "missing_%d_%d", segno, s);
if (!(dlid = build_dm_uuid(dm->mem, seg->lv->lvid.s, errid)))
return_NULL;
if (!(name = dm_build_dm_name(dm->mem, seg->lv->vg->name,
seg->lv->name, errid)))
return_NULL;
log_debug("Getting device info for %s [%s]", name, dlid);
if (!_info(dlid, 1, 0, &info, NULL)) {
log_error("Failed to get info for %s [%s].", name, dlid);
return 0;
}
if (!info.exists) {
/* Create new node */
if (!(node = dm_tree_add_new_dev(dtree, name, dlid, 0, 0, 0, 0, 0)))
return_NULL;
if (!dm_tree_node_add_error_target(node, size))
return_NULL;
} else {
/* Already exists */
if (!dm_tree_add_dev(dtree, info.major, info.minor)) {
log_error("Failed to add device (%" PRIu32 ":%" PRIu32") to dtree",
info.major, info.minor);
return_NULL;
}
}
return dlid;
}
static int _add_error_area(struct dev_manager *dm, struct dm_tree_node *node,
struct lv_segment *seg, int s)
{
char *dlid;
uint64_t extent_size = seg->lv->vg->extent_size;
if (!strcmp(dm->cmd->stripe_filler, "error")) {
/*
* FIXME, the tree pointer is first field of dm_tree_node, but
* we don't have the struct definition available.
*/
struct dm_tree **tree = (struct dm_tree **) node;
2011-10-11 13:03:33 +04:00
if (!(dlid = _add_error_device(dm, *tree, seg, s)))
return_0;
if (!dm_tree_node_add_target_area(node, NULL, dlid, extent_size * seg_le(seg, s)))
return_0;
2008-09-19 11:18:03 +04:00
} else
if (!dm_tree_node_add_target_area(node, dm->cmd->stripe_filler, NULL, UINT64_C(0)))
return_0;
2008-09-19 11:18:03 +04:00
return 1;
}
int add_areas_line(struct dev_manager *dm, struct lv_segment *seg,
struct dm_tree_node *node, uint32_t start_area,
uint32_t areas)
{
uint64_t extent_size = seg->lv->vg->extent_size;
uint32_t s;
2002-03-19 02:25:50 +03:00
char *dlid;
struct stat info;
const char *name;
unsigned num_error_areas = 0;
unsigned num_existing_areas = 0;
/* FIXME Avoid repeating identical stat in dm_tree_node_add_target_area */
for (s = start_area; s < areas; s++) {
if ((seg_type(seg, s) == AREA_PV &&
(!seg_pvseg(seg, s) || !seg_pv(seg, s) || !seg_dev(seg, s) ||
!(name = dev_name(seg_dev(seg, s))) || !*name ||
stat(name, &info) < 0 || !S_ISBLK(info.st_mode))) ||
(seg_type(seg, s) == AREA_LV && !seg_lv(seg, s))) {
if (!seg->lv->vg->cmd->partial_activation) {
log_error("Aborting. LV %s is now incomplete "
"and --partial was not specified.", seg->lv->name);
return 0;
}
if (!_add_error_area(dm, node, seg, s))
return_0;
num_error_areas++;
} else if (seg_type(seg, s) == AREA_PV) {
if (!dm_tree_node_add_target_area(node, dev_name(seg_dev(seg, s)), NULL,
(seg_pv(seg, s)->pe_start + (extent_size * seg_pe(seg, s)))))
return_0;
num_existing_areas++;
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
} else if (seg_is_raid(seg)) {
/*
* RAID can handle unassigned areas. It simple puts
* '- -' in for the metadata/data device pair. This
* is a valid way to indicate to the RAID target that
* the device is missing.
*
* If an image is marked as VISIBLE_LV and !LVM_WRITE,
* it means the device has temporarily been extracted
* from the array. It may come back at a future date,
* so the bitmap must track differences. Again, '- -'
* is used in the CTR table.
*/
if ((seg_type(seg, s) == AREA_UNASSIGNED) ||
((seg_lv(seg, s)->status & VISIBLE_LV) &&
!(seg_lv(seg, s)->status & LVM_WRITE))) {
/* One each for metadata area and data area */
if (!dm_tree_node_add_null_area(node, 0) ||
!dm_tree_node_add_null_area(node, 0))
return_0;
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
continue;
}
Add the ability to split an image from the mirror and track changes. ~> lvconvert --splitmirrors 1 --trackchanges vg/lv The '--trackchanges' option allows a user the ability to use an image of a RAID1 array for the purposes of temporary read-only access. The image can be merged back into the array at a later time and only the blocks that have changed in the array since the split will be resync'ed. This operation can be thought of as a partial split. The image is never completely extracted from the array, in that the array reserves the position the device occupied and tracks the differences between the array and the split image via a bitmap. The image itself is rendered read-only and the name (<LV>_rimage_*) cannot be changed. The user can complete the split (permanently splitting the image from the array) by re-issuing the 'lvconvert' command without the '--trackchanges' argument and specifying the '--name' argument. ~> lvconvert --splitmirrors 1 --name my_split vg/lv Merging the tracked image back into the array is done with the '--merge' option (included in a follow-on patch). ~> lvconvert --merge vg/lv_rimage_<n> The internal mechanics of this are relatively simple. The 'raid' device- mapper target allows for the specification of an empty slot in an array via '- -'. This is what will be used if a partial activation of an array is ever required. (It would also be possible to use 'error' targets in place of the '- -'.) If a RAID image is found to be both read-only and visible, then it is considered separate from the array and '- -' is used to hold it's position in the array. So, all that needs to be done to temporarily split an image from the array /and/ cause the kernel target's bitmap to track (aka "mark") changes made is to make the specified image visible and read-only. To merge the device back into the array, the image needs to be returned to the read/write state of the top-level LV and made invisible.
2011-08-18 23:38:26 +04:00
if (!(dlid = build_dm_uuid(dm->mem, seg_metalv(seg, s)->lvid.s, NULL)))
return_0;
if (!dm_tree_node_add_target_area(node, NULL, dlid, extent_size * seg_metale(seg, s)))
return_0;
if (!(dlid = build_dm_uuid(dm->mem, seg_lv(seg, s)->lvid.s, NULL)))
return_0;
if (!dm_tree_node_add_target_area(node, NULL, dlid, extent_size * seg_le(seg, s)))
return_0;
} else if (seg_type(seg, s) == AREA_LV) {
if (!(dlid = build_dm_uuid(dm->mem, seg_lv(seg, s)->lvid.s, NULL)))
return_0;
if (!dm_tree_node_add_target_area(node, NULL, dlid, extent_size * seg_le(seg, s)))
return_0;
} else {
log_error(INTERNAL_ERROR "Unassigned area found in LV %s.",
seg->lv->name);
return 0;
}
}
if (num_error_areas) {
/* Thins currently do not support partial activation */
if (lv_is_thin_type(seg->lv)) {
log_error("Cannot activate %s%s: pool incomplete.",
seg->lv->vg->name, seg->lv->name);
return 0;
}
}
return 1;
}
2005-11-09 16:08:41 +03:00
static int _add_origin_target_to_dtree(struct dev_manager *dm,
2005-11-09 16:05:17 +03:00
struct dm_tree_node *dnode,
struct logical_volume *lv)
{
const char *real_dlid;
if (!(real_dlid = build_dm_uuid(dm->mem, lv->lvid.s, "real")))
return_0;
2005-11-09 16:05:17 +03:00
if (!dm_tree_node_add_snapshot_origin_target(dnode, lv->size, real_dlid))
return_0;
return 1;
}
static int _add_snapshot_merge_target_to_dtree(struct dev_manager *dm,
struct dm_tree_node *dnode,
struct logical_volume *lv)
{
const char *origin_dlid, *cow_dlid, *merge_dlid;
struct lv_segment *merging_cow_seg = find_merging_cow(lv);
if (!(origin_dlid = build_dm_uuid(dm->mem, lv->lvid.s, "real")))
return_0;
if (!(cow_dlid = build_dm_uuid(dm->mem, merging_cow_seg->cow->lvid.s, "cow")))
return_0;
if (!(merge_dlid = build_dm_uuid(dm->mem, merging_cow_seg->cow->lvid.s, NULL)))
return_0;
if (!dm_tree_node_add_snapshot_merge_target(dnode, lv->size, origin_dlid,
cow_dlid, merge_dlid,
merging_cow_seg->chunk_size))
return_0;
return 1;
}
2005-11-09 16:08:41 +03:00
static int _add_snapshot_target_to_dtree(struct dev_manager *dm,
struct dm_tree_node *dnode,
struct logical_volume *lv,
struct lv_activate_opts *laopts)
{
const char *origin_dlid;
const char *cow_dlid;
struct lv_segment *snap_seg;
uint64_t size;
if (!(snap_seg = find_cow(lv))) {
log_error("Couldn't find snapshot for '%s'.", lv->name);
2005-10-26 23:50:00 +04:00
return 0;
}
if (!(origin_dlid = build_dm_uuid(dm->mem, snap_seg->origin->lvid.s, "real")))
return_0;
if (!(cow_dlid = build_dm_uuid(dm->mem, snap_seg->cow->lvid.s, "cow")))
return_0;
size = (uint64_t) snap_seg->len * snap_seg->origin->vg->extent_size;
if (!laopts->no_merging && lv_is_merging_cow(lv)) {
/* cow is to be merged so load the error target */
if (!dm_tree_node_add_error_target(dnode, size))
return_0;
}
else if (!dm_tree_node_add_snapshot_target(dnode, size, origin_dlid,
cow_dlid, 1, snap_seg->chunk_size))
return_0;
return 1;
}
2005-11-09 16:08:41 +03:00
static int _add_target_to_dtree(struct dev_manager *dm,
struct dm_tree_node *dnode,
struct lv_segment *seg,
struct lv_activate_opts *laopts)
{
uint64_t extent_size = seg->lv->vg->extent_size;
if (!seg->segtype->ops->add_target_line) {
log_error(INTERNAL_ERROR "_emit_target cannot handle "
"segment type %s", seg->segtype->name);
return 0;
}
return seg->segtype->ops->add_target_line(dm, dm->mem, dm->cmd,
&dm->target_state, seg,
laopts, dnode,
extent_size * seg->len,
&dm-> pvmove_mirror_count);
2002-03-16 01:59:12 +03:00
}
2005-11-09 16:08:41 +03:00
static int _add_new_lv_to_dtree(struct dev_manager *dm, struct dm_tree *dtree,
struct logical_volume *lv,
struct lv_activate_opts *laopts,
const char *layer);
/* Add all replicators' LVs */
static int _add_replicator_dev_target_to_dtree(struct dev_manager *dm,
struct dm_tree *dtree,
struct lv_segment *seg,
struct lv_activate_opts *laopts)
{
struct replicator_device *rdev;
struct replicator_site *rsite;
/* For inactive replicator add linear mapping */
if (!lv_is_active_replicator_dev(seg->lv)) {
if (!_add_new_lv_to_dtree(dm, dtree, seg->lv->rdevice->lv, laopts, NULL))
return_0;
return 1;
}
/* Add rlog and replicator nodes */
if (!seg->replicator ||
!first_seg(seg->replicator)->rlog_lv ||
!_add_new_lv_to_dtree(dm, dtree,
first_seg(seg->replicator)->rlog_lv,
laopts, NULL) ||
!_add_new_lv_to_dtree(dm, dtree, seg->replicator, laopts, NULL))
return_0;
/* Activation of one replicator_dev node activates all other nodes */
dm_list_iterate_items(rsite, &seg->replicator->rsites) {
dm_list_iterate_items(rdev, &rsite->rdevices) {
if (rdev->lv &&
!_add_new_lv_to_dtree(dm, dtree, rdev->lv,
laopts, NULL))
return_0;
if (rdev->slog &&
!_add_new_lv_to_dtree(dm, dtree, rdev->slog,
laopts, NULL))
return_0;
}
}
/* Add remaining replicator-dev nodes in the second loop
* to avoid multiple retries for inserting all elements */
dm_list_iterate_items(rsite, &seg->replicator->rsites) {
if (rsite->state != REPLICATOR_STATE_ACTIVE)
continue;
dm_list_iterate_items(rdev, &rsite->rdevices) {
if (rdev->replicator_dev->lv == seg->lv)
continue;
if (!rdev->replicator_dev->lv ||
!_add_new_lv_to_dtree(dm, dtree,
rdev->replicator_dev->lv,
laopts, NULL))
return_0;
}
}
return 1;
}
2005-11-09 16:08:41 +03:00
static int _add_segment_to_dtree(struct dev_manager *dm,
struct dm_tree *dtree,
struct dm_tree_node *dnode,
struct lv_segment *seg,
struct lv_activate_opts *laopts,
const char *layer)
{
uint32_t s;
struct dm_list *snh;
struct lv_segment *seg_present;
const char *target_name;
struct lv_activate_opts lva;
/* Ensure required device-mapper targets are loaded */
seg_present = find_cow(seg->lv) ? : seg;
target_name = (seg_present->segtype->ops->target_name ?
seg_present->segtype->ops->target_name(seg_present, laopts) :
seg_present->segtype->name);
log_debug("Checking kernel supports %s segment type for %s%s%s",
target_name, seg->lv->name,
layer ? "-" : "", layer ? : "");
if (seg_present->segtype->ops->target_present &&
2009-02-28 23:04:24 +03:00
!seg_present->segtype->ops->target_present(seg_present->lv->vg->cmd,
seg_present, NULL)) {
log_error("Can't process LV %s: %s target support missing "
"from kernel?", seg->lv->name, target_name);
return 0;
}
/* Add mirror log */
if (seg->log_lv &&
!_add_new_lv_to_dtree(dm, dtree, seg->log_lv, laopts, NULL))
return_0;
if (seg_is_replicator_dev(seg)) {
if (!_add_replicator_dev_target_to_dtree(dm, dtree, seg, laopts))
return_0;
/* If this is a snapshot origin, add real LV */
/* If this is a snapshot origin + merging snapshot, add cow + real LV */
} else if (lv_is_origin(seg->lv) && !layer) {
if (!laopts->no_merging && lv_is_merging_origin(seg->lv)) {
if (!_add_new_lv_to_dtree(dm, dtree,
find_merging_cow(seg->lv)->cow, laopts, "cow"))
return_0;
/*
* Must also add "real" LV for use when
* snapshot-merge target is added
*/
}
if (!_add_new_lv_to_dtree(dm, dtree, seg->lv, laopts, "real"))
return_0;
} else if (lv_is_cow(seg->lv) && !layer) {
if (!_add_new_lv_to_dtree(dm, dtree, seg->lv, laopts, "cow"))
return_0;
} else if ((layer != _thin_layer) && seg_is_thin(seg)) {
lva = *laopts;
lva.real_pool = 1;
if (!_add_new_lv_to_dtree(dm, dtree, seg_is_thin_pool(seg) ?
seg->lv : seg->pool_lv, &lva, _thin_layer))
return_0;
} else {
if (seg_is_thin_pool(seg) &&
!_add_new_lv_to_dtree(dm, dtree, seg->metadata_lv, laopts, NULL))
return_0;
/* Add any LVs used by this segment */
for (s = 0; s < seg->area_count; s++) {
if ((seg_type(seg, s) == AREA_LV) &&
2009-12-03 13:01:30 +03:00
(!_add_new_lv_to_dtree(dm, dtree, seg_lv(seg, s),
laopts, NULL)))
return_0;
if (seg_is_raid(seg) &&
!_add_new_lv_to_dtree(dm, dtree, seg_metalv(seg, s),
laopts, NULL))
return_0;
}
}
/* Now we've added its dependencies, we can add the target itself */
if (lv_is_origin(seg->lv) && !layer) {
if (laopts->no_merging || !lv_is_merging_origin(seg->lv)) {
if (!_add_origin_target_to_dtree(dm, dnode, seg->lv))
return_0;
} else {
if (!_add_snapshot_merge_target_to_dtree(dm, dnode, seg->lv))
return_0;
}
} else if (lv_is_cow(seg->lv) && !layer) {
if (!_add_snapshot_target_to_dtree(dm, dnode, seg->lv, laopts))
return_0;
} else if (!_add_target_to_dtree(dm, dnode, seg, laopts))
return_0;
if (lv_is_origin(seg->lv) && !layer)
/* Add any snapshots of this LV */
dm_list_iterate(snh, &seg->lv->snapshot_segs)
if (!_add_new_lv_to_dtree(dm, dtree, dm_list_struct_base(snh, struct lv_segment, origin_list)->cow,
laopts, NULL))
return_0;
return 1;
}
This patch fixes issues with improper udev flags on sub-LVs. The current code does not always assign proper udev flags to sub-LVs (e.g. mirror images and log LVs). This shows up especially during a splitmirror operation in which an image is split off from a mirror to form a new LV. A mirror with a disk log is actually composed of 4 different LVs: the 2 mirror images, the log, and the top-level LV that "glues" them all together. When a 2-way mirror is split into two linear LVs, two of those LVs must be removed. The segments of the image which is not split off to form the new LV are transferred to the top-level LV. This is done so that the original LV can maintain its major/minor, UUID, and name. The sub-lv from which the segments were transferred gets an error segment as a transitory process before it is eventually removed. (Note that if the error target was not put in place, a resume_lv would result in two LVs pointing to the same segment! If the machine crashes before the eventual removal of the sub-LV, the result would be a residual LV with the same mapping as the original (now linear) LV.) So, the two LVs that need to be removed are now the log device and the sub-LV with the error segment. If udev_flags are not properly set, a resume will cause the error LV to come up and be scanned by udev. This causes I/O errors. Additionally, when udev scans sub-LVs (or former sub-LVs), it can cause races when we are trying to remove those LVs. This is especially bad during failure conditions. When the mirror is suspended, the top-level along with its sub-LVs are suspended. The changes (now 2 linear devices and the yet-to-be-removed log and error LV) are committed. When the resume takes place on the original LV, there are no longer links to the other sub-lvs through the LVM metadata. The links are implicitly handled by querying the kernel for a list of dependencies. This is done in the '_add_dev' function (which is recursively called for each dependency found) - called through the following chain: _add_dev dm_tree_add_dev_with_udev_flags <*** DM / LVM divide ***> _add_dev_to_dtree _add_lv_to_dtree _create_partial_dtree _tree_action dev_manager_activate _lv_activate_lv _lv_resume lv_resume_if_active When udev flags are calculated by '_get_udev_flags', it is done by referencing the 'logical_volume' structure. Those flags are then passed down into 'dm_tree_add_dev_with_udev_flags', which in turn passes them to '_add_dev'. Unfortunately, when '_add_dev' is finding the dependencies, it has no way to calculate their proper udev_flags. This is because it is below the DM/LVM divide - it doesn't have access to the logical_volume structure. In fact, '_add_dev' simply reuses the udev_flags given for the initial device! This virtually guarentees the udev_flags are wrong for all the dependencies unless they are reset by some other mechanism. The current code provides no such mechanism. Even if '_add_new_lv_to_dtree' were called on the sub-devices - which it isn't - entries already in the tree are simply passed over, failing to reset any udev_flags. The solution must retain its implicit nature of discovering dependencies and be able to go back over the dependencies found to properly set the udev_flags. My solution simply calls a new function before leaving '_add_new_lv_to_dtree' that iterates over the dtree nodes to properly reset the udev_flags of any children. It is important that this function occur after the '_add_dev' has done its job of querying the kernel for a list of dependencies. It is this list of children that we use to look up their respective LVs and properly calculate the udev_flags. This solution has worked for single machine, cluster, and cluster w/ exclusive activation.
2011-10-06 18:45:40 +04:00
static int _set_udev_flags_for_children(struct dev_manager *dm,
struct volume_group *vg,
struct dm_tree_node *dnode)
{
char *p;
const char *uuid;
void *handle = NULL;
struct dm_tree_node *child;
const struct dm_info *info;
struct lv_list *lvl;
while ((child = dm_tree_next_child(&handle, dnode, 0))) {
/* Ignore root node */
if (!(info = dm_tree_node_get_info(child)) || !info->exists)
continue;
if (!(uuid = dm_tree_node_get_uuid(child))) {
log_error(INTERNAL_ERROR
"Failed to get uuid for %" PRIu32 ":%" PRIu32,
info->major, info->minor);
continue;
}
/* Ignore non-LVM devices */
if (!(p = strstr(uuid, UUID_PREFIX)))
continue;
p += strlen(UUID_PREFIX);
/* Ignore LVs that belong to different VGs (due to stacking) */
if (strncmp(p, (char *)vg->id.uuid, ID_LEN))
continue;
/* Ignore LVM devices with 'layer' suffixes */
if (strrchr(p, '-'))
continue;
if (!(lvl = find_lv_in_vg_by_lvid(vg, (const union lvid *)p))) {
log_error(INTERNAL_ERROR
"%s (%" PRIu32 ":%" PRIu32 ") not found in VG",
dm_tree_node_get_name(child),
info->major, info->minor);
return 0;
}
dm_tree_node_set_udev_flags(child,
_get_udev_flags(dm, lvl->lv, NULL));
}
return 1;
}
2005-11-09 16:08:41 +03:00
static int _add_new_lv_to_dtree(struct dev_manager *dm, struct dm_tree *dtree,
struct logical_volume *lv, struct lv_activate_opts *laopts,
const char *layer)
2005-10-17 22:21:05 +04:00
{
struct lv_segment *seg;
struct lv_layer *lvlayer;
struct seg_list *sl;
2005-11-09 16:05:17 +03:00
struct dm_tree_node *dnode;
const struct dm_info *dinfo;
char *name, *dlid;
uint32_t max_stripe_size = UINT32_C(0);
uint32_t read_ahead = lv->read_ahead;
2007-11-29 18:04:12 +03:00
uint32_t read_ahead_flags = UINT32_C(0);
2005-10-17 22:21:05 +04:00
/* FIXME Seek a simpler way to lay out the snapshot-merge tree. */
if (lv_is_origin(lv) && lv_is_merging_origin(lv) && !layer) {
/*
* Clear merge attributes if merge isn't currently possible:
* either origin or merging snapshot are open
* - but use "snapshot-merge" if it is already in use
* - open_count is always retrieved (as of dm-ioctl 4.7.0)
* so just use the tree's existing nodes' info
*/
if (((dinfo = _cached_info(dm->mem, lv,
dtree)) && dinfo->open_count) ||
((dinfo = _cached_info(dm->mem, find_merging_cow(lv)->cow,
dtree)) && dinfo->open_count)) {
/* FIXME Is there anything simpler to check for instead? */
if (!lv_has_target_type(dm->mem, lv, NULL, "snapshot-merge"))
laopts->no_merging = 1;
}
}
if (!(name = dm_build_dm_name(dm->mem, lv->vg->name, lv->name, layer)))
return_0;
2005-10-17 22:21:05 +04:00
if (!(dlid = build_dm_uuid(dm->mem, lv->lvid.s, layer)))
return_0;
2005-10-17 22:21:05 +04:00
/* We've already processed this node if it already has a context ptr */
2005-11-09 16:05:17 +03:00
if ((dnode = dm_tree_find_node_by_uuid(dtree, dlid)) &&
dm_tree_node_get_context(dnode))
return 1;
2005-10-17 22:21:05 +04:00
if (!(lvlayer = dm_pool_alloc(dm->mem, sizeof(*lvlayer)))) {
2009-12-03 13:01:30 +03:00
log_error("_add_new_lv_to_dtree: pool alloc failed for %s %s.",
lv->name, layer);
2005-10-17 22:21:05 +04:00
return 0;
}
lvlayer->lv = lv;
/*
2005-11-09 16:08:41 +03:00
* Add LV to dtree.
* If we're working with precommitted metadata, clear any
* existing inactive table left behind.
* Major/minor settings only apply to the visible layer.
*/
/* FIXME Move the clear from here until later, so we can leave
* identical inactive tables untouched. (For pvmove.)
*/
if (!(dnode = dm_tree_add_new_dev_with_udev_flags(dtree, name, dlid,
layer ? UINT32_C(0) : (uint32_t) lv->major,
layer ? UINT32_C(0) : (uint32_t) lv->minor,
read_only_lv(lv, laopts),
((lv->vg->status & PRECOMMITTED) | laopts->revert) ? 1 : 0,
lvlayer,
_get_udev_flags(dm, lv, layer))))
return_0;
/* Store existing name so we can do rename later */
2005-11-09 16:05:17 +03:00
lvlayer->old_name = dm_tree_node_get_name(dnode);
/* Create table */
dm->pvmove_mirror_count = 0u;
dm_list_iterate_items(seg, &lv->segments) {
if (!_add_segment_to_dtree(dm, dtree, dnode, seg, laopts, layer))
return_0;
/* These aren't real segments in the LVM2 metadata */
if (lv_is_origin(lv) && !layer)
break;
if (!laopts->no_merging && lv_is_cow(lv) && !layer)
break;
if (max_stripe_size < seg->stripe_size * seg->area_count)
max_stripe_size = seg->stripe_size * seg->area_count;
}
2005-10-17 22:21:05 +04:00
if (read_ahead == DM_READ_AHEAD_AUTO) {
/* we need RA at least twice a whole stripe - see the comment in md/raid0.c */
read_ahead = max_stripe_size * 2;
if (!read_ahead)
lv_calculate_readahead(lv, &read_ahead);
2007-11-29 18:04:12 +03:00
read_ahead_flags = DM_READ_AHEAD_MINIMUM_FLAG;
}
2007-11-29 18:04:12 +03:00
dm_tree_node_set_read_ahead(dnode, read_ahead, read_ahead_flags);
/* Setup thin pool callback */
if (layer && lv_is_thin_pool(lv) &&
!_thin_pool_register_callback(dm, dnode, lv))
return_0;
/* Add any LVs referencing a PVMOVE LV unless told not to */
if (dm->track_pvmove_deps && (lv->status & PVMOVE))
dm_list_iterate_items(sl, &lv->segs_using_this_lv)
if (!_add_new_lv_to_dtree(dm, dtree, sl->seg->lv, laopts, NULL))
return_0;
This patch fixes issues with improper udev flags on sub-LVs. The current code does not always assign proper udev flags to sub-LVs (e.g. mirror images and log LVs). This shows up especially during a splitmirror operation in which an image is split off from a mirror to form a new LV. A mirror with a disk log is actually composed of 4 different LVs: the 2 mirror images, the log, and the top-level LV that "glues" them all together. When a 2-way mirror is split into two linear LVs, two of those LVs must be removed. The segments of the image which is not split off to form the new LV are transferred to the top-level LV. This is done so that the original LV can maintain its major/minor, UUID, and name. The sub-lv from which the segments were transferred gets an error segment as a transitory process before it is eventually removed. (Note that if the error target was not put in place, a resume_lv would result in two LVs pointing to the same segment! If the machine crashes before the eventual removal of the sub-LV, the result would be a residual LV with the same mapping as the original (now linear) LV.) So, the two LVs that need to be removed are now the log device and the sub-LV with the error segment. If udev_flags are not properly set, a resume will cause the error LV to come up and be scanned by udev. This causes I/O errors. Additionally, when udev scans sub-LVs (or former sub-LVs), it can cause races when we are trying to remove those LVs. This is especially bad during failure conditions. When the mirror is suspended, the top-level along with its sub-LVs are suspended. The changes (now 2 linear devices and the yet-to-be-removed log and error LV) are committed. When the resume takes place on the original LV, there are no longer links to the other sub-lvs through the LVM metadata. The links are implicitly handled by querying the kernel for a list of dependencies. This is done in the '_add_dev' function (which is recursively called for each dependency found) - called through the following chain: _add_dev dm_tree_add_dev_with_udev_flags <*** DM / LVM divide ***> _add_dev_to_dtree _add_lv_to_dtree _create_partial_dtree _tree_action dev_manager_activate _lv_activate_lv _lv_resume lv_resume_if_active When udev flags are calculated by '_get_udev_flags', it is done by referencing the 'logical_volume' structure. Those flags are then passed down into 'dm_tree_add_dev_with_udev_flags', which in turn passes them to '_add_dev'. Unfortunately, when '_add_dev' is finding the dependencies, it has no way to calculate their proper udev_flags. This is because it is below the DM/LVM divide - it doesn't have access to the logical_volume structure. In fact, '_add_dev' simply reuses the udev_flags given for the initial device! This virtually guarentees the udev_flags are wrong for all the dependencies unless they are reset by some other mechanism. The current code provides no such mechanism. Even if '_add_new_lv_to_dtree' were called on the sub-devices - which it isn't - entries already in the tree are simply passed over, failing to reset any udev_flags. The solution must retain its implicit nature of discovering dependencies and be able to go back over the dependencies found to properly set the udev_flags. My solution simply calls a new function before leaving '_add_new_lv_to_dtree' that iterates over the dtree nodes to properly reset the udev_flags of any children. It is important that this function occur after the '_add_dev' has done its job of querying the kernel for a list of dependencies. It is this list of children that we use to look up their respective LVs and properly calculate the udev_flags. This solution has worked for single machine, cluster, and cluster w/ exclusive activation.
2011-10-06 18:45:40 +04:00
if (!_set_udev_flags_for_children(dm, lv->vg, dnode))
return_0;
return 1;
2005-10-17 22:21:05 +04:00
}
/* FIXME: symlinks should be created/destroyed at the same time
* as the kernel devices but we can't do that from within libdevmapper
* at present so we must walk the tree twice instead. */
/*
* Create LV symlinks for children of supplied root node.
*/
2005-11-09 16:05:17 +03:00
static int _create_lv_symlinks(struct dev_manager *dm, struct dm_tree_node *root)
2005-10-17 22:21:05 +04:00
{
void *handle = NULL;
2005-11-09 16:05:17 +03:00
struct dm_tree_node *child;
struct lv_layer *lvlayer;
char *old_vgname, *old_lvname, *old_layer;
char *new_vgname, *new_lvname, *new_layer;
const char *name;
int r = 1;
2005-10-17 22:21:05 +04:00
/* Nothing to do if udev fallback is disabled. */
if (!dm->cmd->current_settings.udev_fallback) {
fs_set_create();
return 1;
}
2005-11-09 16:05:17 +03:00
while ((child = dm_tree_next_child(&handle, root, 0))) {
2009-12-03 12:59:54 +03:00
if (!(lvlayer = dm_tree_node_get_context(child)))
continue;
2005-10-17 22:21:05 +04:00
/* Detect rename */
2005-11-09 16:05:17 +03:00
name = dm_tree_node_get_name(child);
2005-10-17 22:21:05 +04:00
if (name && lvlayer->old_name && *lvlayer->old_name && strcmp(name, lvlayer->old_name)) {
if (!dm_split_lvm_name(dm->mem, lvlayer->old_name, &old_vgname, &old_lvname, &old_layer)) {
2008-01-30 17:00:02 +03:00
log_error("_create_lv_symlinks: Couldn't split up old device name %s", lvlayer->old_name);
return 0;
}
if (!dm_split_lvm_name(dm->mem, name, &new_vgname, &new_lvname, &new_layer)) {
log_error("_create_lv_symlinks: Couldn't split up new device name %s", name);
return 0;
}
if (!fs_rename_lv(lvlayer->lv, name, old_vgname, old_lvname))
r = 0;
continue;
}
if (lv_is_visible(lvlayer->lv)) {
if (!_dev_manager_lv_mknodes(lvlayer->lv))
r = 0;
continue;
}
if (!_dev_manager_lv_rmnodes(lvlayer->lv))
r = 0;
2005-10-17 22:21:05 +04:00
}
return r;
}
/*
* Remove LV symlinks for children of supplied root node.
*/
static int _remove_lv_symlinks(struct dev_manager *dm, struct dm_tree_node *root)
{
void *handle = NULL;
struct dm_tree_node *child;
char *vgname, *lvname, *layer;
int r = 1;
/* Nothing to do if udev fallback is disabled. */
if (!dm->cmd->current_settings.udev_fallback)
return 1;
while ((child = dm_tree_next_child(&handle, root, 0))) {
2008-01-30 17:00:02 +03:00
if (!dm_split_lvm_name(dm->mem, dm_tree_node_get_name(child), &vgname, &lvname, &layer)) {
r = 0;
continue;
}
if (!*vgname)
continue;
/* only top level layer has symlinks */
if (*layer)
continue;
fs_del_lv_byname(dm->cmd->dev_dir, vgname, lvname,
dm->cmd->current_settings.udev_rules);
}
return r;
}
static int _clean_tree(struct dev_manager *dm, struct dm_tree_node *root, char *non_toplevel_tree_dlid)
{
void *handle = NULL;
2005-11-09 16:05:17 +03:00
struct dm_tree_node *child;
char *vgname, *lvname, *layer;
const char *name, *uuid;
2005-11-09 16:05:17 +03:00
while ((child = dm_tree_next_child(&handle, root, 0))) {
if (!(name = dm_tree_node_get_name(child)))
continue;
2005-11-09 16:05:17 +03:00
if (!(uuid = dm_tree_node_get_uuid(child)))
continue;
2008-01-30 17:00:02 +03:00
if (!dm_split_lvm_name(dm->mem, name, &vgname, &lvname, &layer)) {
log_error("_clean_tree: Couldn't split up device name %s.", name);
return 0;
}
/* Not meant to be top level? */
if (!*layer)
continue;
/* If operation was performed on a partial tree, don't remove it */
if (non_toplevel_tree_dlid && !strcmp(non_toplevel_tree_dlid, uuid))
continue;
if (!dm_tree_deactivate_children(root, uuid, strlen(uuid)))
2009-08-03 22:01:45 +04:00
return_0;
}
return 1;
2005-10-17 22:21:05 +04:00
}
static int _tree_action(struct dev_manager *dm, struct logical_volume *lv,
struct lv_activate_opts *laopts, action_t action)
2005-10-17 22:21:05 +04:00
{
const size_t DLID_SIZE = ID_LEN + sizeof(UUID_PREFIX) - 1;
2005-11-09 16:05:17 +03:00
struct dm_tree *dtree;
struct dm_tree_node *root;
char *dlid;
2005-10-17 22:21:05 +04:00
int r = 0;
laopts->is_activate = (action == ACTIVATE);
if (!(dtree = _create_partial_dtree(dm, lv, laopts->origin_only)))
return_0;
2005-10-17 22:21:05 +04:00
2005-11-09 16:05:17 +03:00
if (!(root = dm_tree_find_node(dtree, 0, 0))) {
log_error("Lost dependency tree root node");
goto out_no_root;
2005-10-17 22:21:05 +04:00
}
/* Restore fs cookie */
dm_tree_set_cookie(root, fs_get_cookie());
if (!(dlid = build_dm_uuid(dm->mem, lv->lvid.s, (lv_is_origin(lv) && laopts->origin_only) ? "real" : NULL)))
goto_out;
2005-10-17 22:21:05 +04:00
2005-10-19 17:59:18 +04:00
/* Only process nodes with uuid of "LVM-" plus VG id. */
switch(action) {
case CLEAN:
/* Deactivate any unused non-toplevel nodes */
if (!_clean_tree(dm, root, laopts->origin_only ? dlid : NULL))
goto_out;
break;
case DEACTIVATE:
if (retry_deactivation())
dm_tree_retry_remove(root);
/* Deactivate LV and all devices it references that nothing else has open. */
if (!dm_tree_deactivate_children(root, dlid, DLID_SIZE))
2009-08-03 22:01:45 +04:00
goto_out;
if (!_remove_lv_symlinks(dm, root))
log_warn("Failed to remove all device symlinks associated with %s.", lv->name);
break;
case SUSPEND:
dm_tree_skip_lockfs(root);
if (!dm->flush_required && !seg_is_raid(first_seg(lv)) &&
(lv->status & MIRRORED) && !(lv->status & PVMOVE))
dm_tree_use_no_flush_suspend(root);
/* Fall through */
case SUSPEND_WITH_LOCKFS:
if (!dm_tree_suspend_children(root, dlid, DLID_SIZE))
goto_out;
break;
case PRELOAD:
case ACTIVATE:
/* Add all required new devices to tree */
if (!_add_new_lv_to_dtree(dm, dtree, lv, laopts, (lv_is_origin(lv) && laopts->origin_only) ? "real" : NULL))
goto_out;
/* Preload any devices required before any suspensions */
if (!dm_tree_preload_children(root, dlid, DLID_SIZE))
2009-08-03 22:01:45 +04:00
goto_out;
if (dm_tree_node_size_changed(root))
dm->flush_required = 1;
if (action == ACTIVATE) {
if (!dm_tree_activate_children(root, dlid, DLID_SIZE))
2009-08-03 22:01:45 +04:00
goto_out;
if (!_create_lv_symlinks(dm, root))
log_warn("Failed to create symlinks for %s.", lv->name);
}
break;
default:
log_error("_tree_action: Action %u not supported.", action);
2005-10-17 22:21:05 +04:00
goto out;
2009-12-03 12:58:30 +03:00
}
2005-10-17 22:21:05 +04:00
r = 1;
out:
/* Save fs cookie for udev settle, do not wait here */
fs_set_cookie(dm_tree_get_cookie(root));
out_no_root:
2005-11-09 16:05:17 +03:00
dm_tree_free(dtree);
2005-10-17 22:21:05 +04:00
return r;
}
/* origin_only may only be set if we are resuming (not activating) an origin LV */
int dev_manager_activate(struct dev_manager *dm, struct logical_volume *lv,
struct lv_activate_opts *laopts)
{
if (!_tree_action(dm, lv, laopts, ACTIVATE))
return_0;
if (!_tree_action(dm, lv, laopts, CLEAN))
return_0;
return 1;
}
/* origin_only may only be set if we are resuming (not activating) an origin LV */
int dev_manager_preload(struct dev_manager *dm, struct logical_volume *lv,
struct lv_activate_opts *laopts, int *flush_required)
{
if (!_tree_action(dm, lv, laopts, PRELOAD))
return_0;
*flush_required = dm->flush_required;
return 1;
}
2005-10-26 18:13:52 +04:00
int dev_manager_deactivate(struct dev_manager *dm, struct logical_volume *lv)
{
struct lv_activate_opts laopts = { 0 };
if (!_tree_action(dm, lv, &laopts, DEACTIVATE))
return_0;
return 1;
}
int dev_manager_suspend(struct dev_manager *dm, struct logical_volume *lv,
struct lv_activate_opts *laopts, int lockfs, int flush_required)
2005-10-26 18:13:52 +04:00
{
dm->flush_required = flush_required;
if (!_tree_action(dm, lv, laopts, lockfs ? SUSPEND_WITH_LOCKFS : SUSPEND))
return_0;
return 1;
2005-10-26 18:13:52 +04:00
}
/*
* Does device use VG somewhere in its construction?
* Returns 1 if uncertain.
*/
int dev_manager_device_uses_vg(struct device *dev,
struct volume_group *vg)
{
2005-11-09 16:05:17 +03:00
struct dm_tree *dtree;
struct dm_tree_node *root;
char dlid[sizeof(UUID_PREFIX) + sizeof(struct id) - 1] __attribute__((aligned(8)));
int r = 1;
2005-11-09 16:05:17 +03:00
if (!(dtree = dm_tree_create())) {
2005-11-09 16:08:41 +03:00
log_error("partial dtree creation failed");
return r;
}
if (!dm_tree_add_dev(dtree, (uint32_t) MAJOR(dev->dev), (uint32_t) MINOR(dev->dev))) {
2005-11-09 16:08:41 +03:00
log_error("Failed to add device %s (%" PRIu32 ":%" PRIu32") to dtree",
dev_name(dev), (uint32_t) MAJOR(dev->dev), (uint32_t) MINOR(dev->dev));
goto out;
}
memcpy(dlid, UUID_PREFIX, sizeof(UUID_PREFIX) - 1);
memcpy(dlid + sizeof(UUID_PREFIX) - 1, &vg->id.uuid[0], sizeof(vg->id));
2005-11-09 16:05:17 +03:00
if (!(root = dm_tree_find_node(dtree, 0, 0))) {
log_error("Lost dependency tree root node");
goto out;
}
2005-11-09 16:05:17 +03:00
if (dm_tree_children_use_uuid(root, dlid, sizeof(UUID_PREFIX) + sizeof(vg->id) - 1))
goto_out;
r = 0;
out:
2005-11-09 16:05:17 +03:00
dm_tree_free(dtree);
return r;
}