mirror of
git://sourceware.org/git/lvm2.git
synced 2024-12-22 17:35:59 +03:00
5c9d70c955
Use set_lv() and dev_set() to wipe sections of devices. Add mirror_in_sync() flag to avoid unnecessary resync on activation.
1288 lines
29 KiB
C
1288 lines
29 KiB
C
/*
|
|
* Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
|
|
* Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
|
|
*
|
|
* 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 General Public License v.2.
|
|
*
|
|
* You should have received a copy of the GNU General Public License
|
|
* along with this program; if not, write to the Free Software Foundation,
|
|
* Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
*/
|
|
|
|
#include "tools.h"
|
|
#include "lv_alloc.h"
|
|
#include "xlate.h"
|
|
|
|
#include <sys/stat.h>
|
|
#include <sys/wait.h>
|
|
|
|
/* From linux/drivers/md/dm-log.c */
|
|
#define MIRROR_MAGIC 0x4D695272
|
|
#define MIRROR_DISK_VERSION 2
|
|
|
|
/* Command line args */
|
|
unsigned arg_count(struct cmd_context *cmd, int a)
|
|
{
|
|
return cmd->args[a].count;
|
|
}
|
|
|
|
const char *arg_value(struct cmd_context *cmd, int a)
|
|
{
|
|
return cmd->args[a].value;
|
|
}
|
|
|
|
const char *arg_str_value(struct cmd_context *cmd, int a, const char *def)
|
|
{
|
|
return arg_count(cmd, a) ? cmd->args[a].value : def;
|
|
}
|
|
|
|
int32_t arg_int_value(struct cmd_context *cmd, int a, const int32_t def)
|
|
{
|
|
return arg_count(cmd, a) ? cmd->args[a].i_value : def;
|
|
}
|
|
|
|
uint32_t arg_uint_value(struct cmd_context *cmd, int a, const uint32_t def)
|
|
{
|
|
return arg_count(cmd, a) ? cmd->args[a].ui_value : def;
|
|
}
|
|
|
|
int64_t arg_int64_value(struct cmd_context *cmd, int a, const int64_t def)
|
|
{
|
|
return arg_count(cmd, a) ? cmd->args[a].i64_value : def;
|
|
}
|
|
|
|
uint64_t arg_uint64_value(struct cmd_context *cmd, int a, const uint64_t def)
|
|
{
|
|
return arg_count(cmd, a) ? cmd->args[a].ui64_value : def;
|
|
}
|
|
|
|
const void *arg_ptr_value(struct cmd_context *cmd, int a, const void *def)
|
|
{
|
|
return arg_count(cmd, a) ? cmd->args[a].ptr : def;
|
|
}
|
|
|
|
sign_t arg_sign_value(struct cmd_context *cmd, int a, const sign_t def)
|
|
{
|
|
return arg_count(cmd, a) ? cmd->args[a].sign : def;
|
|
}
|
|
|
|
int arg_count_increment(struct cmd_context *cmd, int a)
|
|
{
|
|
return cmd->args[a].count++;
|
|
}
|
|
|
|
const char *command_name(struct cmd_context *cmd)
|
|
{
|
|
return cmd->command->name;
|
|
}
|
|
|
|
/*
|
|
* Metadata iteration functions
|
|
*/
|
|
int process_each_lv_in_vg(struct cmd_context *cmd, struct volume_group *vg,
|
|
struct list *arg_lvnames, struct list *tags,
|
|
void *handle,
|
|
int (*process_single) (struct cmd_context * cmd,
|
|
struct logical_volume * lv,
|
|
void *handle))
|
|
{
|
|
int ret_max = 0;
|
|
int ret = 0;
|
|
unsigned process_all = 0;
|
|
unsigned process_lv = 0;
|
|
unsigned tags_supplied = 0;
|
|
unsigned lvargs_supplied = 0;
|
|
unsigned lvargs_matched = 0;
|
|
|
|
struct lv_list *lvl;
|
|
|
|
if (vg->status & EXPORTED_VG) {
|
|
log_error("Volume group \"%s\" is exported", vg->name);
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
if (tags && !list_empty(tags))
|
|
tags_supplied = 1;
|
|
|
|
if (arg_lvnames && !list_empty(arg_lvnames))
|
|
lvargs_supplied = 1;
|
|
|
|
/* Process all LVs in this VG if no restrictions given */
|
|
if (!tags_supplied && !lvargs_supplied)
|
|
process_all = 1;
|
|
|
|
/* Or if VG tags match */
|
|
if (!process_lv && tags_supplied &&
|
|
str_list_match_list(tags, &vg->tags)) {
|
|
process_all = 1;
|
|
}
|
|
|
|
list_iterate_items(lvl, &vg->lvs) {
|
|
if (lvl->lv->status & SNAPSHOT)
|
|
continue;
|
|
|
|
/* Should we process this LV? */
|
|
if (process_all)
|
|
process_lv = 1;
|
|
else
|
|
process_lv = 0;
|
|
|
|
/* LV tag match? */
|
|
if (!process_lv && tags_supplied &&
|
|
str_list_match_list(tags, &lvl->lv->tags)) {
|
|
process_lv = 1;
|
|
}
|
|
|
|
/* LV name match? */
|
|
if (lvargs_supplied &&
|
|
str_list_match_item(arg_lvnames, lvl->lv->name)) {
|
|
process_lv = 1;
|
|
lvargs_matched++;
|
|
}
|
|
|
|
if (!process_lv)
|
|
continue;
|
|
|
|
ret = process_single(cmd, lvl->lv, handle);
|
|
if (ret > ret_max)
|
|
ret_max = ret;
|
|
}
|
|
|
|
if (lvargs_supplied && lvargs_matched != list_size(arg_lvnames)) {
|
|
log_error("One or more specified logical volume(s) not found.");
|
|
if (ret_max < ECMD_FAILED)
|
|
ret_max = ECMD_FAILED;
|
|
}
|
|
|
|
return ret_max;
|
|
}
|
|
|
|
int process_each_lv(struct cmd_context *cmd, int argc, char **argv,
|
|
int lock_type, void *handle,
|
|
int (*process_single) (struct cmd_context * cmd,
|
|
struct logical_volume * lv,
|
|
void *handle))
|
|
{
|
|
int opt = 0;
|
|
int ret_max = 0;
|
|
int ret = 0;
|
|
int consistent;
|
|
|
|
struct list *tags_arg;
|
|
struct list *vgnames; /* VGs to process */
|
|
struct str_list *sll, *strl;
|
|
struct volume_group *vg;
|
|
struct list tags, lvnames;
|
|
struct list arg_lvnames; /* Cmdline vgname or vgname/lvname */
|
|
char *vglv;
|
|
size_t vglv_sz;
|
|
|
|
const char *vgname;
|
|
|
|
list_init(&tags);
|
|
list_init(&arg_lvnames);
|
|
|
|
if (argc) {
|
|
struct list arg_vgnames;
|
|
|
|
log_verbose("Using logical volume(s) on command line");
|
|
list_init(&arg_vgnames);
|
|
|
|
for (; opt < argc; opt++) {
|
|
const char *lv_name = argv[opt];
|
|
char *vgname_def;
|
|
int dev_dir_found = 0;
|
|
|
|
/* Do we have a tag or vgname or lvname? */
|
|
vgname = lv_name;
|
|
|
|
if (*vgname == '@') {
|
|
if (!validate_name(vgname + 1)) {
|
|
log_error("Skipping invalid tag %s",
|
|
vgname);
|
|
continue;
|
|
}
|
|
if (!str_list_add(cmd->mem, &tags,
|
|
dm_pool_strdup(cmd->mem,
|
|
vgname + 1))) {
|
|
log_error("strlist allocation failed");
|
|
return ECMD_FAILED;
|
|
}
|
|
continue;
|
|
}
|
|
|
|
/* FIXME Jumbled parsing */
|
|
if (*vgname == '/') {
|
|
while (*vgname == '/')
|
|
vgname++;
|
|
vgname--;
|
|
}
|
|
if (!strncmp(vgname, cmd->dev_dir,
|
|
strlen(cmd->dev_dir))) {
|
|
vgname += strlen(cmd->dev_dir);
|
|
dev_dir_found = 1;
|
|
while (*vgname == '/')
|
|
vgname++;
|
|
}
|
|
if (*vgname == '/') {
|
|
log_error("\"%s\": Invalid path for Logical "
|
|
"Volume", argv[opt]);
|
|
if (ret_max < ECMD_FAILED)
|
|
ret_max = ECMD_FAILED;
|
|
continue;
|
|
}
|
|
lv_name = vgname;
|
|
if (strchr(vgname, '/')) {
|
|
/* Must be an LV */
|
|
lv_name = strchr(vgname, '/');
|
|
while (*lv_name == '/')
|
|
lv_name++;
|
|
if (!(vgname = extract_vgname(cmd, vgname))) {
|
|
if (ret_max < ECMD_FAILED)
|
|
ret_max = ECMD_FAILED;
|
|
continue;
|
|
}
|
|
} else if (!dev_dir_found &&
|
|
(vgname_def = default_vgname(cmd))) {
|
|
vgname = vgname_def;
|
|
} else
|
|
lv_name = NULL;
|
|
|
|
if (!str_list_add(cmd->mem, &arg_vgnames,
|
|
dm_pool_strdup(cmd->mem, vgname))) {
|
|
log_error("strlist allocation failed");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
if (!lv_name) {
|
|
if (!str_list_add(cmd->mem, &arg_lvnames,
|
|
dm_pool_strdup(cmd->mem,
|
|
vgname))) {
|
|
log_error("strlist allocation failed");
|
|
return ECMD_FAILED;
|
|
}
|
|
} else {
|
|
vglv_sz = strlen(vgname) + strlen(lv_name) + 2;
|
|
if (!(vglv = dm_pool_alloc(cmd->mem, vglv_sz)) ||
|
|
lvm_snprintf(vglv, vglv_sz, "%s/%s", vgname,
|
|
lv_name) < 0) {
|
|
log_error("vg/lv string alloc failed");
|
|
return ECMD_FAILED;
|
|
}
|
|
if (!str_list_add(cmd->mem, &arg_lvnames, vglv)) {
|
|
log_error("strlist allocation failed");
|
|
return ECMD_FAILED;
|
|
}
|
|
}
|
|
}
|
|
vgnames = &arg_vgnames;
|
|
}
|
|
|
|
if (!argc || !list_empty(&tags)) {
|
|
log_verbose("Finding all logical volumes");
|
|
if (!(vgnames = get_vgs(cmd, 0)) || list_empty(vgnames)) {
|
|
log_error("No volume groups found");
|
|
return ret_max;
|
|
}
|
|
}
|
|
|
|
list_iterate_items(strl, vgnames) {
|
|
vgname = strl->str;
|
|
if (!vgname || !*vgname)
|
|
continue; /* FIXME Unnecessary? */
|
|
if (!lock_vol(cmd, vgname, lock_type)) {
|
|
log_error("Can't lock %s: skipping", vgname);
|
|
continue;
|
|
}
|
|
if (lock_type & LCK_WRITE)
|
|
consistent = 1;
|
|
else
|
|
consistent = 0;
|
|
if (!(vg = vg_read(cmd, vgname, NULL, &consistent)) || !consistent) {
|
|
unlock_vg(cmd, vgname);
|
|
if (!vg)
|
|
log_error("Volume group \"%s\" "
|
|
"not found", vgname);
|
|
else
|
|
log_error("Volume group \"%s\" "
|
|
"inconsistent", vgname);
|
|
if (!vg || !(vg = recover_vg(cmd, vgname, lock_type))) {
|
|
if (ret_max < ECMD_FAILED)
|
|
ret_max = ECMD_FAILED;
|
|
continue;
|
|
}
|
|
}
|
|
|
|
tags_arg = &tags;
|
|
list_init(&lvnames); /* LVs to be processed in this VG */
|
|
list_iterate_items(sll, &arg_lvnames) {
|
|
const char *vg_name = sll->str;
|
|
const char *lv_name = strchr(vg_name, '/');
|
|
|
|
if ((!lv_name && !strcmp(vg_name, vgname))) {
|
|
/* Process all LVs in this VG */
|
|
tags_arg = NULL;
|
|
list_init(&lvnames);
|
|
break;
|
|
} else if (!strncmp(vg_name, vgname, strlen(vgname)) &&
|
|
strlen(vgname) == lv_name - vg_name) {
|
|
if (!str_list_add(cmd->mem, &lvnames,
|
|
dm_pool_strdup(cmd->mem,
|
|
lv_name + 1))) {
|
|
log_error("strlist allocation failed");
|
|
return ECMD_FAILED;
|
|
}
|
|
}
|
|
}
|
|
|
|
ret = process_each_lv_in_vg(cmd, vg, &lvnames, tags_arg,
|
|
handle, process_single);
|
|
unlock_vg(cmd, vgname);
|
|
if (ret > ret_max)
|
|
ret_max = ret;
|
|
}
|
|
|
|
return ret_max;
|
|
}
|
|
|
|
int process_each_segment_in_pv(struct cmd_context *cmd,
|
|
struct volume_group *vg,
|
|
struct physical_volume *pv,
|
|
void *handle,
|
|
int (*process_single) (struct cmd_context * cmd,
|
|
struct volume_group * vg,
|
|
struct pv_segment * pvseg,
|
|
void *handle))
|
|
{
|
|
struct pv_segment *pvseg;
|
|
int ret_max = 0;
|
|
int ret;
|
|
|
|
list_iterate_items(pvseg, &pv->segments) {
|
|
ret = process_single(cmd, vg, pvseg, handle);
|
|
if (ret > ret_max)
|
|
ret_max = ret;
|
|
}
|
|
|
|
return ret_max;
|
|
}
|
|
|
|
int process_each_segment_in_lv(struct cmd_context *cmd,
|
|
struct logical_volume *lv,
|
|
void *handle,
|
|
int (*process_single) (struct cmd_context * cmd,
|
|
struct lv_segment * seg,
|
|
void *handle))
|
|
{
|
|
struct lv_segment *seg;
|
|
int ret_max = 0;
|
|
int ret;
|
|
|
|
list_iterate_items(seg, &lv->segments) {
|
|
ret = process_single(cmd, seg, handle);
|
|
if (ret > ret_max)
|
|
ret_max = ret;
|
|
}
|
|
|
|
return ret_max;
|
|
}
|
|
|
|
static int _process_one_vg(struct cmd_context *cmd, const char *vg_name,
|
|
const char *vgid,
|
|
struct list *tags, struct list *arg_vgnames,
|
|
int lock_type, int consistent, void *handle,
|
|
int ret_max,
|
|
int (*process_single) (struct cmd_context * cmd,
|
|
const char *vg_name,
|
|
struct volume_group * vg,
|
|
int consistent, void *handle))
|
|
{
|
|
struct volume_group *vg;
|
|
int ret = 0;
|
|
|
|
if (!lock_vol(cmd, vg_name, lock_type)) {
|
|
log_error("Can't lock %s: skipping", vg_name);
|
|
return ret_max;
|
|
}
|
|
|
|
log_verbose("Finding volume group \"%s\"", vg_name);
|
|
if (!(vg = vg_read(cmd, vg_name, vgid, &consistent))) {
|
|
log_error("Volume group \"%s\" not found", vg_name);
|
|
unlock_vg(cmd, vg_name);
|
|
return ret_max;
|
|
}
|
|
|
|
if (!list_empty(tags)) {
|
|
/* Only process if a tag matches or it's on arg_vgnames */
|
|
if (!str_list_match_item(arg_vgnames, vg_name) &&
|
|
!str_list_match_list(tags, &vg->tags)) {
|
|
unlock_vg(cmd, vg_name);
|
|
return ret_max;
|
|
}
|
|
}
|
|
|
|
if ((ret = process_single(cmd, vg_name, vg, consistent,
|
|
handle)) > ret_max) {
|
|
ret_max = ret;
|
|
}
|
|
|
|
unlock_vg(cmd, vg_name);
|
|
|
|
return ret_max;
|
|
}
|
|
|
|
int process_each_vg(struct cmd_context *cmd, int argc, char **argv,
|
|
int lock_type, int consistent, void *handle,
|
|
int (*process_single) (struct cmd_context * cmd,
|
|
const char *vg_name,
|
|
struct volume_group * vg,
|
|
int consistent, void *handle))
|
|
{
|
|
int opt = 0;
|
|
int ret_max = 0;
|
|
|
|
struct str_list *sl;
|
|
struct list *vgnames, *vgids;
|
|
struct list arg_vgnames, tags;
|
|
|
|
const char *vg_name, *vgid;
|
|
char *dev_dir = cmd->dev_dir;
|
|
|
|
list_init(&tags);
|
|
list_init(&arg_vgnames);
|
|
|
|
if (argc) {
|
|
log_verbose("Using volume group(s) on command line");
|
|
|
|
for (; opt < argc; opt++) {
|
|
vg_name = argv[opt];
|
|
if (*vg_name == '@') {
|
|
if (!validate_name(vg_name + 1)) {
|
|
log_error("Skipping invalid tag %s",
|
|
vg_name);
|
|
continue;
|
|
}
|
|
if (!str_list_add(cmd->mem, &tags,
|
|
dm_pool_strdup(cmd->mem,
|
|
vg_name + 1))) {
|
|
log_error("strlist allocation failed");
|
|
return ECMD_FAILED;
|
|
}
|
|
continue;
|
|
}
|
|
|
|
if (*vg_name == '/') {
|
|
while (*vg_name == '/')
|
|
vg_name++;
|
|
vg_name--;
|
|
}
|
|
if (!strncmp(vg_name, dev_dir, strlen(dev_dir)))
|
|
vg_name += strlen(dev_dir);
|
|
if (strchr(vg_name, '/')) {
|
|
log_error("Invalid volume group name: %s",
|
|
vg_name);
|
|
continue;
|
|
}
|
|
if (!str_list_add(cmd->mem, &arg_vgnames,
|
|
dm_pool_strdup(cmd->mem, vg_name))) {
|
|
log_error("strlist allocation failed");
|
|
return ECMD_FAILED;
|
|
}
|
|
}
|
|
|
|
vgnames = &arg_vgnames;
|
|
}
|
|
|
|
if (!argc || !list_empty(&tags)) {
|
|
log_verbose("Finding all volume groups");
|
|
if (!(vgids = get_vgids(cmd, 0)) || list_empty(vgids)) {
|
|
log_error("No volume groups found");
|
|
return ret_max;
|
|
}
|
|
list_iterate_items(sl, vgids) {
|
|
vgid = sl->str;
|
|
if (!vgid || !(vg_name = vgname_from_vgid(cmd->mem, vgid)) ||
|
|
!*vg_name)
|
|
continue;
|
|
ret_max = _process_one_vg(cmd, vg_name, vgid, &tags,
|
|
&arg_vgnames,
|
|
lock_type, consistent, handle,
|
|
ret_max, process_single);
|
|
}
|
|
} else {
|
|
list_iterate_items(sl, vgnames) {
|
|
vg_name = sl->str;
|
|
if (!vg_name || !*vg_name)
|
|
continue; /* FIXME Unnecessary? */
|
|
ret_max = _process_one_vg(cmd, vg_name, NULL, &tags,
|
|
&arg_vgnames,
|
|
lock_type, consistent, handle,
|
|
ret_max, process_single);
|
|
}
|
|
}
|
|
|
|
return ret_max;
|
|
}
|
|
|
|
int process_each_pv_in_vg(struct cmd_context *cmd, struct volume_group *vg,
|
|
struct list *tags, void *handle,
|
|
int (*process_single) (struct cmd_context * cmd,
|
|
struct volume_group * vg,
|
|
struct physical_volume * pv,
|
|
void *handle))
|
|
{
|
|
int ret_max = 0;
|
|
int ret = 0;
|
|
struct pv_list *pvl;
|
|
|
|
list_iterate_items(pvl, &vg->pvs) {
|
|
if (tags && !list_empty(tags) &&
|
|
!str_list_match_list(tags, &pvl->pv->tags)) {
|
|
continue;
|
|
}
|
|
if ((ret = process_single(cmd, vg, pvl->pv, handle)) > ret_max)
|
|
ret_max = ret;
|
|
}
|
|
|
|
return ret_max;
|
|
}
|
|
|
|
static int _process_all_devs(struct cmd_context *cmd, void *handle,
|
|
int (*process_single) (struct cmd_context * cmd,
|
|
struct volume_group * vg,
|
|
struct physical_volume * pv,
|
|
void *handle))
|
|
{
|
|
struct physical_volume *pv;
|
|
struct physical_volume pv_dummy;
|
|
struct dev_iter *iter;
|
|
struct device *dev;
|
|
|
|
int ret_max = 0;
|
|
int ret = 0;
|
|
|
|
if (!(iter = dev_iter_create(cmd->filter, 1))) {
|
|
log_error("dev_iter creation failed");
|
|
return ECMD_FAILED;
|
|
}
|
|
|
|
while ((dev = dev_iter_get(iter))) {
|
|
if (!(pv = pv_read(cmd, dev_name(dev), NULL, NULL, 0))) {
|
|
memset(&pv_dummy, 0, sizeof(pv_dummy));
|
|
list_init(&pv_dummy.tags);
|
|
list_init(&pv_dummy.segments);
|
|
pv_dummy.dev = dev;
|
|
pv_dummy.fmt = NULL;
|
|
pv = &pv_dummy;
|
|
}
|
|
ret = process_single(cmd, NULL, pv, handle);
|
|
if (ret > ret_max)
|
|
ret_max = ret;
|
|
}
|
|
|
|
dev_iter_destroy(iter);
|
|
|
|
return ret_max;
|
|
}
|
|
|
|
int process_each_pv(struct cmd_context *cmd, int argc, char **argv,
|
|
struct volume_group *vg, void *handle,
|
|
int (*process_single) (struct cmd_context * cmd,
|
|
struct volume_group * vg,
|
|
struct physical_volume * pv,
|
|
void *handle))
|
|
{
|
|
int opt = 0;
|
|
int ret_max = 0;
|
|
int ret = 0;
|
|
|
|
struct pv_list *pvl;
|
|
struct physical_volume *pv;
|
|
struct list *pvslist, *vgnames;
|
|
struct list tags;
|
|
struct str_list *sll;
|
|
char *tagname;
|
|
int consistent = 1;
|
|
|
|
list_init(&tags);
|
|
|
|
if (argc) {
|
|
log_verbose("Using physical volume(s) on command line");
|
|
for (; opt < argc; opt++) {
|
|
if (*argv[opt] == '@') {
|
|
tagname = argv[opt] + 1;
|
|
|
|
if (!validate_name(tagname)) {
|
|
log_error("Skipping invalid tag %s",
|
|
tagname);
|
|
if (ret_max < EINVALID_CMD_LINE)
|
|
ret_max = EINVALID_CMD_LINE;
|
|
continue;
|
|
}
|
|
if (!str_list_add(cmd->mem, &tags,
|
|
dm_pool_strdup(cmd->mem,
|
|
tagname))) {
|
|
log_error("strlist allocation failed");
|
|
return ECMD_FAILED;
|
|
}
|
|
continue;
|
|
}
|
|
if (vg) {
|
|
if (!(pvl = find_pv_in_vg(vg, argv[opt]))) {
|
|
log_error("Physical Volume \"%s\" not "
|
|
"found in Volume Group "
|
|
"\"%s\"", argv[opt],
|
|
vg->name);
|
|
ret_max = ECMD_FAILED;
|
|
continue;
|
|
}
|
|
pv = pvl->pv;
|
|
} else {
|
|
if (!(pv = pv_read(cmd, argv[opt], NULL,
|
|
NULL, 1))) {
|
|
log_error("Failed to read physical "
|
|
"volume \"%s\"", argv[opt]);
|
|
ret_max = ECMD_FAILED;
|
|
continue;
|
|
}
|
|
}
|
|
|
|
ret = process_single(cmd, vg, pv, handle);
|
|
if (ret > ret_max)
|
|
ret_max = ret;
|
|
}
|
|
if (!list_empty(&tags) && (vgnames = get_vgs(cmd, 0)) &&
|
|
!list_empty(vgnames)) {
|
|
list_iterate_items(sll, vgnames) {
|
|
if (!(vg = vg_read(cmd, sll->str, NULL, &consistent))) {
|
|
log_error("Volume group \"%s\" not found", sll->str);
|
|
ret_max = ECMD_FAILED;
|
|
continue;
|
|
}
|
|
if (!consistent)
|
|
continue;
|
|
ret = process_each_pv_in_vg(cmd, vg, &tags,
|
|
handle,
|
|
process_single);
|
|
if (ret > ret_max)
|
|
ret_max = ret;
|
|
}
|
|
}
|
|
} else {
|
|
if (vg) {
|
|
log_verbose("Using all physical volume(s) in "
|
|
"volume group");
|
|
ret = process_each_pv_in_vg(cmd, vg, NULL, handle,
|
|
process_single);
|
|
if (ret > ret_max)
|
|
ret_max = ret;
|
|
} else if (arg_count(cmd, all_ARG)) {
|
|
ret = _process_all_devs(cmd, handle, process_single);
|
|
if (ret > ret_max)
|
|
ret_max = ret;
|
|
} else {
|
|
log_verbose("Scanning for physical volume names");
|
|
if (!(pvslist = get_pvs(cmd)))
|
|
return ECMD_FAILED;
|
|
|
|
list_iterate_items(pvl, pvslist) {
|
|
ret = process_single(cmd, NULL, pvl->pv,
|
|
handle);
|
|
if (ret > ret_max)
|
|
ret_max = ret;
|
|
}
|
|
}
|
|
}
|
|
|
|
return ret_max;
|
|
}
|
|
|
|
/*
|
|
* Determine volume group name from a logical volume name
|
|
*/
|
|
const char *extract_vgname(struct cmd_context *cmd, const char *lv_name)
|
|
{
|
|
const char *vg_name = lv_name;
|
|
char *st;
|
|
char *dev_dir = cmd->dev_dir;
|
|
int dev_dir_provided = 0;
|
|
|
|
/* Path supplied? */
|
|
if (vg_name && strchr(vg_name, '/')) {
|
|
/* Strip dev_dir (optional) */
|
|
if (*vg_name == '/') {
|
|
while (*vg_name == '/')
|
|
vg_name++;
|
|
vg_name--;
|
|
}
|
|
if (!strncmp(vg_name, dev_dir, strlen(dev_dir))) {
|
|
vg_name += strlen(dev_dir);
|
|
dev_dir_provided = 1;
|
|
while (*vg_name == '/')
|
|
vg_name++;
|
|
}
|
|
if (*vg_name == '/') {
|
|
log_error("\"%s\": Invalid path for Logical "
|
|
"Volume", lv_name);
|
|
return 0;
|
|
}
|
|
|
|
/* Require exactly one set of consecutive slashes */
|
|
if ((st = strchr(vg_name, '/')))
|
|
while (*st == '/')
|
|
st++;
|
|
|
|
if (!strchr(vg_name, '/') || strchr(st, '/')) {
|
|
log_error("\"%s\": Invalid path for Logical Volume",
|
|
lv_name);
|
|
return 0;
|
|
}
|
|
|
|
vg_name = dm_pool_strdup(cmd->mem, vg_name);
|
|
if (!vg_name) {
|
|
log_error("Allocation of vg_name failed");
|
|
return 0;
|
|
}
|
|
|
|
*strchr(vg_name, '/') = '\0';
|
|
return vg_name;
|
|
}
|
|
|
|
if (!(vg_name = default_vgname(cmd))) {
|
|
if (lv_name)
|
|
log_error("Path required for Logical Volume \"%s\"",
|
|
lv_name);
|
|
return 0;
|
|
}
|
|
|
|
return vg_name;
|
|
}
|
|
|
|
/*
|
|
* Extract default volume group name from environment
|
|
*/
|
|
char *default_vgname(struct cmd_context *cmd)
|
|
{
|
|
char *vg_path;
|
|
char *dev_dir = cmd->dev_dir;
|
|
|
|
/* Take default VG from environment? */
|
|
vg_path = getenv("LVM_VG_NAME");
|
|
if (!vg_path)
|
|
return 0;
|
|
|
|
/* Strip dev_dir (optional) */
|
|
if (*vg_path == '/') {
|
|
while (*vg_path == '/')
|
|
vg_path++;
|
|
vg_path--;
|
|
}
|
|
if (!strncmp(vg_path, dev_dir, strlen(dev_dir)))
|
|
vg_path += strlen(dev_dir);
|
|
|
|
if (strchr(vg_path, '/')) {
|
|
log_error("Environment Volume Group in LVM_VG_NAME invalid: "
|
|
"\"%s\"", vg_path);
|
|
return 0;
|
|
}
|
|
|
|
return dm_pool_strdup(cmd->mem, vg_path);
|
|
}
|
|
|
|
/*
|
|
* Process physical extent range specifiers
|
|
*/
|
|
static int _add_pe_range(struct dm_pool *mem, struct list *pe_ranges,
|
|
uint32_t start, uint32_t count)
|
|
{
|
|
struct pe_range *per;
|
|
|
|
log_debug("Adding PE range: start PE %" PRIu32 " length %" PRIu32,
|
|
start, count);
|
|
|
|
/* Ensure no overlap with existing areas */
|
|
list_iterate_items(per, pe_ranges) {
|
|
if (((start < per->start) && (start + count - 1 >= per->start))
|
|
|| ((start >= per->start) &&
|
|
(per->start + per->count - 1) >= start)) {
|
|
log_error("Overlapping PE ranges detected (%" PRIu32
|
|
"-%" PRIu32 ", %" PRIu32 "-%" PRIu32 ")",
|
|
start, start + count - 1, per->start,
|
|
per->start + per->count - 1);
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
if (!(per = dm_pool_alloc(mem, sizeof(*per)))) {
|
|
log_error("Allocation of list failed");
|
|
return 0;
|
|
}
|
|
|
|
per->start = start;
|
|
per->count = count;
|
|
list_add(pe_ranges, &per->list);
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _parse_pes(struct dm_pool *mem, char *c, struct list *pe_ranges,
|
|
uint32_t size)
|
|
{
|
|
char *endptr;
|
|
uint32_t start, end;
|
|
|
|
/* Default to whole PV */
|
|
if (!c) {
|
|
if (!_add_pe_range(mem, pe_ranges, UINT32_C(0), size)) {
|
|
stack;
|
|
return 0;
|
|
}
|
|
return 1;
|
|
}
|
|
|
|
while (*c) {
|
|
if (*c != ':')
|
|
goto error;
|
|
|
|
c++;
|
|
|
|
/* Disallow :: and :\0 */
|
|
if (*c == ':' || !*c)
|
|
goto error;
|
|
|
|
/* Default to whole range */
|
|
start = UINT32_C(0);
|
|
end = size - 1;
|
|
|
|
/* Start extent given? */
|
|
if (isdigit(*c)) {
|
|
start = (uint32_t) strtoul(c, &endptr, 10);
|
|
if (endptr == c)
|
|
goto error;
|
|
c = endptr;
|
|
/* Just one number given? */
|
|
if (!*c || *c == ':')
|
|
end = start;
|
|
}
|
|
/* Range? */
|
|
if (*c == '-') {
|
|
c++;
|
|
if (isdigit(*c)) {
|
|
end = (uint32_t) strtoul(c, &endptr, 10);
|
|
if (endptr == c)
|
|
goto error;
|
|
c = endptr;
|
|
}
|
|
}
|
|
if (*c && *c != ':')
|
|
goto error;
|
|
|
|
if ((start > end) || (end > size - 1)) {
|
|
log_error("PE range error: start extent %" PRIu32 " to "
|
|
"end extent %" PRIu32, start, end);
|
|
return 0;
|
|
}
|
|
|
|
if (!_add_pe_range(mem, pe_ranges, start, end - start + 1)) {
|
|
stack;
|
|
return 0;
|
|
}
|
|
|
|
}
|
|
|
|
return 1;
|
|
|
|
error:
|
|
log_error("Physical extent parsing error at %s", c);
|
|
return 0;
|
|
}
|
|
|
|
static void _create_pv_entry(struct dm_pool *mem, struct pv_list *pvl,
|
|
char *colon, int allocatable_only, struct list *r)
|
|
{
|
|
const char *pvname;
|
|
struct pv_list *new_pvl;
|
|
struct list *pe_ranges;
|
|
|
|
pvname = dev_name(pvl->pv->dev);
|
|
if (allocatable_only && !(pvl->pv->status & ALLOCATABLE_PV)) {
|
|
log_error("Physical volume %s not allocatable", pvname);
|
|
return;
|
|
}
|
|
|
|
if (allocatable_only &&
|
|
(pvl->pv->pe_count == pvl->pv->pe_alloc_count)) {
|
|
log_err("No free extents on physical volume \"%s\"", pvname);
|
|
return;
|
|
}
|
|
|
|
if (!(new_pvl = dm_pool_alloc(mem, sizeof(*new_pvl)))) {
|
|
log_err("Unable to allocate physical volume list.");
|
|
return;
|
|
}
|
|
|
|
memcpy(new_pvl, pvl, sizeof(*new_pvl));
|
|
|
|
if (!(pe_ranges = dm_pool_alloc(mem, sizeof(*pe_ranges)))) {
|
|
log_error("Allocation of pe_ranges list failed");
|
|
return;
|
|
}
|
|
list_init(pe_ranges);
|
|
|
|
/* Determine selected physical extents */
|
|
if (!_parse_pes(mem, colon, pe_ranges, pvl->pv->pe_count)) {
|
|
stack;
|
|
return;
|
|
}
|
|
new_pvl->pe_ranges = pe_ranges;
|
|
|
|
list_add(r, &new_pvl->list);
|
|
}
|
|
|
|
struct list *create_pv_list(struct dm_pool *mem, struct volume_group *vg, int argc,
|
|
char **argv, int allocatable_only)
|
|
{
|
|
struct list *r;
|
|
struct pv_list *pvl;
|
|
struct list tags, arg_pvnames;
|
|
const char *pvname = NULL;
|
|
char *colon, *tagname;
|
|
int i;
|
|
|
|
/* Build up list of PVs */
|
|
if (!(r = dm_pool_alloc(mem, sizeof(*r)))) {
|
|
log_error("Allocation of list failed");
|
|
return NULL;
|
|
}
|
|
list_init(r);
|
|
|
|
list_init(&tags);
|
|
list_init(&arg_pvnames);
|
|
|
|
for (i = 0; i < argc; i++) {
|
|
if (*argv[i] == '@') {
|
|
tagname = argv[i] + 1;
|
|
if (!validate_name(tagname)) {
|
|
log_error("Skipping invalid tag %s", tagname);
|
|
continue;
|
|
}
|
|
list_iterate_items(pvl, &vg->pvs) {
|
|
if (str_list_match_item(&pvl->pv->tags,
|
|
tagname)) {
|
|
_create_pv_entry(mem, pvl, NULL,
|
|
allocatable_only, r);
|
|
}
|
|
}
|
|
continue;
|
|
}
|
|
|
|
pvname = argv[i];
|
|
|
|
if ((colon = strchr(pvname, ':'))) {
|
|
if (!(pvname = dm_pool_strndup(mem, pvname,
|
|
(unsigned) (colon -
|
|
pvname)))) {
|
|
log_error("Failed to clone PV name");
|
|
return NULL;
|
|
}
|
|
}
|
|
|
|
if (!(pvl = find_pv_in_vg(vg, pvname))) {
|
|
log_err("Physical Volume \"%s\" not found in "
|
|
"Volume Group \"%s\"", pvname, vg->name);
|
|
return NULL;
|
|
}
|
|
_create_pv_entry(mem, pvl, colon, allocatable_only, r);
|
|
}
|
|
|
|
if (list_empty(r))
|
|
log_error("No specified PVs have space available");
|
|
|
|
return list_empty(r) ? NULL : r;
|
|
}
|
|
|
|
struct list *clone_pv_list(struct dm_pool *mem, struct list *pvsl)
|
|
{
|
|
struct list *r;
|
|
struct pv_list *pvl, *new_pvl;
|
|
|
|
/* Build up list of PVs */
|
|
if (!(r = dm_pool_alloc(mem, sizeof(*r)))) {
|
|
log_error("Allocation of list failed");
|
|
return NULL;
|
|
}
|
|
list_init(r);
|
|
|
|
list_iterate_items(pvl, pvsl) {
|
|
if (!(new_pvl = dm_pool_zalloc(mem, sizeof(*new_pvl)))) {
|
|
log_error("Unable to allocate physical volume list.");
|
|
return NULL;
|
|
}
|
|
|
|
memcpy(new_pvl, pvl, sizeof(*new_pvl));
|
|
list_add(r, &new_pvl->list);
|
|
}
|
|
|
|
return r;
|
|
}
|
|
|
|
/*
|
|
* Attempt metadata recovery
|
|
*/
|
|
struct volume_group *recover_vg(struct cmd_context *cmd, const char *vgname,
|
|
int lock_type)
|
|
{
|
|
int consistent = 1;
|
|
|
|
lock_type &= ~LCK_TYPE_MASK;
|
|
lock_type |= LCK_WRITE;
|
|
|
|
if (!lock_vol(cmd, vgname, lock_type)) {
|
|
log_error("Can't lock %s for metadata recovery: skipping",
|
|
vgname);
|
|
return NULL;
|
|
}
|
|
|
|
return vg_read(cmd, vgname, NULL, &consistent);
|
|
}
|
|
|
|
int apply_lvname_restrictions(const char *name)
|
|
{
|
|
if (!strncmp(name, "snapshot", 8)) {
|
|
log_error("Names starting \"snapshot\" are reserved. "
|
|
"Please choose a different LV name.");
|
|
return 0;
|
|
}
|
|
|
|
if (!strncmp(name, "pvmove", 6)) {
|
|
log_error("Names starting \"pvmove\" are reserved. "
|
|
"Please choose a different LV name.");
|
|
return 0;
|
|
}
|
|
|
|
if (strstr(name, "_mlog")) {
|
|
log_error("Names including \"_mlog\" are reserved. "
|
|
"Please choose a different LV name.");
|
|
return 0;
|
|
}
|
|
|
|
if (strstr(name, "_mimage")) {
|
|
log_error("Names including \"_mimage\" are reserved. "
|
|
"Please choose a different LV name.");
|
|
return 0;
|
|
}
|
|
|
|
return 1;
|
|
}
|
|
|
|
int validate_vg_name(struct cmd_context *cmd, const char *vg_name)
|
|
{
|
|
char vg_path[PATH_MAX];
|
|
|
|
if (!validate_name(vg_name))
|
|
return 0;
|
|
|
|
snprintf(vg_path, PATH_MAX, "%s%s", cmd->dev_dir, vg_name);
|
|
if (path_exists(vg_path)) {
|
|
log_error("%s: already exists in filesystem", vg_path);
|
|
return 0;
|
|
}
|
|
|
|
return 1;
|
|
}
|
|
|
|
int generate_log_name_format(struct volume_group *vg __attribute((unused)),
|
|
const char *lv_name, char *buffer, size_t size)
|
|
{
|
|
if (lvm_snprintf(buffer, size, "%s_mlog", lv_name) < 0) {
|
|
stack;
|
|
return 0;
|
|
}
|
|
|
|
/* FIXME I think we can cope without this. Cf. _add_lv_to_dtree()
|
|
if (find_lv_in_vg(vg, buffer) &&
|
|
lvm_snprintf(buffer, size, "%s_mlog_%%d",
|
|
lv_name) < 0) {
|
|
stack;
|
|
return 0;
|
|
}
|
|
*******/
|
|
|
|
return 1;
|
|
}
|
|
|
|
/*
|
|
* Initialize the LV with 'value'.
|
|
*/
|
|
int set_lv(struct cmd_context *cmd, struct logical_volume *lv, int value)
|
|
{
|
|
struct device *dev;
|
|
char *name;
|
|
|
|
/*
|
|
* FIXME:
|
|
* <clausen> also, more than 4k
|
|
* <clausen> say, reiserfs puts it's superblock 32k in, IIRC
|
|
* <ejt_> k, I'll drop a fixme to that effect
|
|
* (I know the device is at least 4k, but not 32k)
|
|
*/
|
|
if (!(name = dm_pool_alloc(cmd->mem, PATH_MAX))) {
|
|
log_error("Name allocation failed - device not cleared");
|
|
return 0;
|
|
}
|
|
|
|
if (lvm_snprintf(name, PATH_MAX, "%s%s/%s", cmd->dev_dir,
|
|
lv->vg->name, lv->name) < 0) {
|
|
log_error("Name too long - device not cleared (%s)", lv->name);
|
|
return 0;
|
|
}
|
|
|
|
log_verbose("Clearing start of logical volume \"%s\"", lv->name);
|
|
|
|
if (!(dev = dev_cache_get(name, NULL))) {
|
|
log_error("%s: not found: device not cleared", name);
|
|
return 0;
|
|
}
|
|
|
|
if (!dev_open_quiet(dev))
|
|
return 0;
|
|
|
|
dev_set(dev, UINT64_C(0), (size_t) 4096, value);
|
|
dev_close_immediate(dev);
|
|
|
|
return 1;
|
|
}
|
|
|
|
|
|
/*
|
|
* This function writes a new header to the mirror log header to the lv
|
|
*
|
|
* Returns: 1 on success, 0 on failure
|
|
*/
|
|
static int _write_log_header(struct cmd_context *cmd, struct logical_volume *lv)
|
|
{
|
|
struct device *dev;
|
|
char *name;
|
|
struct { /* The mirror log header */
|
|
uint32_t magic;
|
|
uint32_t version;
|
|
uint64_t nr_regions;
|
|
} log_header;
|
|
|
|
log_header.magic = xlate32(MIRROR_MAGIC);
|
|
log_header.version = xlate32(MIRROR_DISK_VERSION);
|
|
log_header.nr_regions = xlate64((uint64_t)-1);
|
|
|
|
if (!(name = dm_pool_alloc(cmd->mem, PATH_MAX))) {
|
|
log_error("Name allocation failed - log header not written (%s)",
|
|
lv->name);
|
|
return 0;
|
|
}
|
|
|
|
if (lvm_snprintf(name, PATH_MAX, "%s%s/%s", cmd->dev_dir,
|
|
lv->vg->name, lv->name) < 0) {
|
|
log_error("Name too long - log header not written (%s)", lv->name);
|
|
return 0;
|
|
}
|
|
|
|
log_verbose("Writing log header to device, %s", lv->name);
|
|
|
|
if (!(dev = dev_cache_get(name, NULL))) {
|
|
log_error("%s: not found: log header not written", name);
|
|
return 0;
|
|
}
|
|
|
|
if (!dev_open_quiet(dev))
|
|
return 0;
|
|
|
|
if (!dev_write(dev, UINT64_C(0), sizeof(log_header), &log_header)) {
|
|
log_error("Failed to write log header to %s", name);
|
|
dev_close_immediate(dev);
|
|
return 0;
|
|
}
|
|
|
|
dev_close_immediate(dev);
|
|
|
|
return 1;
|
|
}
|
|
|
|
struct logical_volume *create_mirror_log(struct cmd_context *cmd,
|
|
struct volume_group *vg,
|
|
struct alloc_handle *ah,
|
|
alloc_policy_t alloc,
|
|
const char *lv_name,
|
|
int in_sync)
|
|
{
|
|
struct logical_volume *log_lv;
|
|
char *log_name;
|
|
size_t len;
|
|
|
|
len = strlen(lv_name) + 32;
|
|
if (!(log_name = alloca(len)) ||
|
|
!(generate_log_name_format(vg, lv_name, log_name, len))) {
|
|
log_error("log_name allocation failed. "
|
|
"Remove new LV and retry.");
|
|
return NULL;
|
|
}
|
|
|
|
if (!(log_lv = lv_create_empty(vg->fid, log_name, NULL,
|
|
VISIBLE_LV | LVM_READ | LVM_WRITE,
|
|
alloc, 0, vg))) {
|
|
stack;
|
|
return NULL;
|
|
}
|
|
|
|
if (!lv_add_log_segment(ah, log_lv)) {
|
|
stack;
|
|
goto error;
|
|
}
|
|
|
|
/* store mirror log on disk(s) */
|
|
if (!vg_write(vg)) {
|
|
stack;
|
|
goto error;
|
|
}
|
|
|
|
backup(vg);
|
|
|
|
if (!vg_commit(vg)) {
|
|
stack;
|
|
goto error;
|
|
}
|
|
|
|
if (!activate_lv(cmd, log_lv)) {
|
|
log_error("Aborting. Failed to activate mirror log. "
|
|
"Remove new LVs and retry.");
|
|
goto error;
|
|
}
|
|
|
|
if (activation() && !set_lv(cmd, log_lv, in_sync)) {
|
|
log_error("Aborting. Failed to wipe mirror log. "
|
|
"Remove new LV and retry.");
|
|
goto error;
|
|
}
|
|
|
|
if (!_write_log_header(cmd, log_lv)) {
|
|
log_error("Aborting. Failed to write mirror log header. "
|
|
"Remove new LV and retry.");
|
|
goto error;
|
|
}
|
|
|
|
if (!deactivate_lv(cmd, log_lv)) {
|
|
log_error("Aborting. Failed to deactivate mirror log. "
|
|
"Remove new LV and retry.");
|
|
goto error;
|
|
}
|
|
|
|
log_lv->status &= ~VISIBLE_LV;
|
|
|
|
return log_lv;
|
|
error:
|
|
/* FIXME Attempt to clean up. */
|
|
return NULL;
|
|
}
|