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

799 lines
19 KiB
C
Raw Normal View History

2005-04-20 00:52:35 +04:00
/*
* Copyright (C) 2003 Sistina Software, Inc. All rights reserved.
* Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
2005-04-20 00:52:35 +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.
2005-04-20 00:52:35 +04:00
*
* You should have received a copy of the GNU Lesser General Public License
2005-04-20 00:52:35 +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
*/
#include "lib.h"
#include "metadata.h"
#include "pv_alloc.h"
#include "toolcontext.h"
2007-08-31 00:30:41 +04:00
#include "locking.h"
#include "defaults.h"
2013-11-22 23:44:25 +04:00
#include "lvmcache.h"
#include "lvmetad.h"
#include "display.h"
#include "label.h"
#include "archiver.h"
2005-04-20 00:52:35 +04:00
static struct pv_segment *_alloc_pv_segment(struct dm_pool *mem,
2005-04-20 00:52:35 +04:00
struct physical_volume *pv,
uint32_t pe, uint32_t len,
struct lv_segment *lvseg,
uint32_t lv_area)
2005-04-20 00:52:35 +04:00
{
struct pv_segment *peg;
if (!(peg = dm_pool_zalloc(mem, sizeof(*peg)))) {
2005-04-20 00:52:35 +04:00
log_error("pv_segment allocation failed");
return NULL;
}
peg->pv = pv;
peg->pe = pe;
peg->len = len;
peg->lvseg = lvseg;
peg->lv_area = lv_area;
2005-04-20 00:52:35 +04:00
dm_list_init(&peg->list);
2005-04-20 00:52:35 +04:00
return peg;
}
int alloc_pv_segment_whole_pv(struct dm_pool *mem, struct physical_volume *pv)
2005-04-20 00:52:35 +04:00
{
struct pv_segment *peg;
if (!pv->pe_count)
return 1;
2005-04-20 00:52:35 +04:00
/* FIXME Cope with holes in PVs */
2008-01-30 16:19:47 +03:00
if (!(peg = _alloc_pv_segment(mem, pv, 0, pv->pe_count, NULL, 0)))
return_0;
2005-04-20 00:52:35 +04:00
dm_list_add(&pv->segments, &peg->list);
2005-04-20 00:52:35 +04:00
return 1;
2005-04-20 00:52:35 +04:00
}
int peg_dup(struct dm_pool *mem, struct dm_list *peg_new, struct dm_list *peg_old)
2005-04-20 00:52:35 +04:00
{
struct pv_segment *peg, *pego;
dm_list_init(peg_new);
2005-04-20 00:52:35 +04:00
dm_list_iterate_items(pego, peg_old) {
2005-04-20 00:52:35 +04:00
if (!(peg = _alloc_pv_segment(mem, pego->pv, pego->pe,
pego->len, pego->lvseg,
2008-01-30 16:19:47 +03:00
pego->lv_area)))
return_0;
dm_list_add(peg_new, &peg->list);
2005-04-20 00:52:35 +04:00
}
return 1;
}
/* Find segment at a given physical extent in a PV */
static struct pv_segment *find_peg_by_pe(const struct physical_volume *pv,
uint32_t pe)
{
struct pv_segment *pvseg;
/* search backwards to optimise mostly used last segment split */
dm_list_iterate_back_items(pvseg, &pv->segments)
if (pe >= pvseg->pe && pe < pvseg->pe + pvseg->len)
return pvseg;
return NULL;
}
/*
* Split peg at given extent.
* Second part is always not allocated to a LV and returned.
*/
static struct pv_segment *_pv_split_segment(struct dm_pool *mem,
struct physical_volume *pv,
struct pv_segment *peg,
uint32_t pe)
{
struct pv_segment *peg_new;
if (!(peg_new = _alloc_pv_segment(mem, peg->pv, pe,
peg->len + peg->pe - pe,
2008-01-30 16:19:47 +03:00
NULL, 0)))
return_NULL;
peg->len = peg->len - peg_new->len;
dm_list_add_h(&peg->list, &peg_new->list);
if (peg->lvseg) {
peg->pv->pe_alloc_count -= peg_new->len;
peg->lvseg->lv->vg->free_count += peg_new->len;
}
return peg_new;
}
/*
* Ensure there is a PV segment boundary at the given extent.
*/
int pv_split_segment(struct dm_pool *mem,
struct physical_volume *pv, uint32_t pe,
struct pv_segment **pvseg_allocated)
{
struct pv_segment *pvseg, *pvseg_new = NULL;
if (pe == pv->pe_count)
goto out;
if (!(pvseg = find_peg_by_pe(pv, pe))) {
log_error("Segment with extent %" PRIu32 " in PV %s not found",
pe, pv_dev_name(pv));
return 0;
}
/* This is a peg start already */
if (pe == pvseg->pe) {
pvseg_new = pvseg;
goto out;
}
if (!(pvseg_new = _pv_split_segment(mem, pv, pvseg, pe)))
2008-01-30 16:19:47 +03:00
return_0;
out:
if (pvseg_allocated)
*pvseg_allocated = pvseg_new;
return 1;
}
2005-05-11 19:04:06 +04:00
static struct pv_segment null_pv_segment = {
.pv = NULL,
.pe = 0,
2005-05-11 19:04:06 +04:00
};
struct pv_segment *assign_peg_to_lvseg(struct physical_volume *pv,
uint32_t pe, uint32_t area_len,
struct lv_segment *seg,
uint32_t area_num)
{
struct pv_segment *peg = NULL;
2005-05-11 19:04:06 +04:00
/* Missing format1 PV */
if (!pv)
return &null_pv_segment;
if (!pv_split_segment(seg->lv->vg->vgmem, pv, pe, &peg) ||
!pv_split_segment(seg->lv->vg->vgmem, pv, pe + area_len, NULL))
2008-01-30 16:19:47 +03:00
return_NULL;
if (!peg) {
log_error("Missing PV segment on %s at %u.",
pv_dev_name(pv), pe);
return NULL;
}
peg->lvseg = seg;
peg->lv_area = area_num;
peg->pv->pe_alloc_count += area_len;
peg->lvseg->lv->vg->free_count -= area_len;
return peg;
}
int discard_pv_segment(struct pv_segment *peg, uint32_t discard_area_reduction)
{
uint64_t discard_offset_sectors;
uint64_t pe_start = peg->pv->pe_start;
char uuid[64] __attribute__((aligned(8)));
if (!peg->lvseg) {
log_error("discard_pv_segment with unallocated segment: "
"%s PE %" PRIu32, pv_dev_name(peg->pv), peg->pe);
return 0;
}
2005-06-01 20:51:55 +04:00
/*
* Only issue discards if enabled in lvm.conf and both
* the device and kernel (>= 2.6.35) supports discards.
*/
if (!find_config_tree_bool(peg->pv->fmt->cmd, devices_issue_discards_CFG, NULL))
return 1;
/* Missing PV? */
if (is_missing_pv(peg->pv) || !peg->pv->dev) {
if (!id_write_format(&peg->pv->id, uuid, sizeof(uuid)))
return_0;
log_verbose("Skipping discard on missing device with uuid %s.", uuid);
return 1;
}
if (!dev_discard_max_bytes(peg->pv->fmt->cmd->dev_types, peg->pv->dev) ||
!dev_discard_granularity(peg->pv->fmt->cmd->dev_types, peg->pv->dev))
return 1;
discard_offset_sectors = (peg->pe + peg->lvseg->area_len - discard_area_reduction) *
(uint64_t) peg->pv->vg->extent_size + pe_start;
if (!discard_offset_sectors) {
/*
* pe_start=0 and the PV's first extent contains the label.
* Must skip past the first extent.
*/
discard_offset_sectors = peg->pv->vg->extent_size;
discard_area_reduction--;
}
log_debug_alloc("Discarding %" PRIu32 " extents offset %" PRIu64 " sectors on %s.",
discard_area_reduction, discard_offset_sectors, dev_name(peg->pv->dev));
if (discard_area_reduction &&
!dev_discard_blocks(peg->pv->dev, discard_offset_sectors << SECTOR_SHIFT,
discard_area_reduction * (uint64_t) peg->pv->vg->extent_size * SECTOR_SIZE))
return_0;
return 1;
}
int release_pv_segment(struct pv_segment *peg, uint32_t area_reduction)
{
if (!peg->lvseg) {
log_error("release_pv_segment with unallocated segment: "
"%s PE %" PRIu32, pv_dev_name(peg->pv), peg->pe);
return 0;
}
if (peg->lvseg->area_len == area_reduction) {
peg->pv->pe_alloc_count -= area_reduction;
peg->lvseg->lv->vg->free_count += area_reduction;
peg->lvseg = NULL;
peg->lv_area = 0;
/* FIXME merge free space */
return 1;
}
if (!pv_split_segment(peg->lvseg->lv->vg->vgmem,
peg->pv, peg->pe + peg->lvseg->area_len -
area_reduction, NULL))
2008-01-30 16:19:47 +03:00
return_0;
return 1;
}
/*
* Only for use by lv_segment merging routines.
*/
void merge_pv_segments(struct pv_segment *peg1, struct pv_segment *peg2)
{
peg1->len += peg2->len;
dm_list_del(&peg2->list);
}
/*
* Calculate the overlap, in extents, between a struct pv_segment and
* a struct pe_range.
*/
static uint32_t _overlap_pe(const struct pv_segment *pvseg,
const struct pe_range *per)
{
uint32_t start;
uint32_t end;
start = max(pvseg->pe, per->start);
end = min(pvseg->pe + pvseg->len, per->start + per->count);
if (end < start)
return 0;
else
return end - start;
}
/*
* Returns: number of free PEs in a struct pv_list
*/
uint32_t pv_list_extents_free(const struct dm_list *pvh)
{
struct pv_list *pvl;
struct pe_range *per;
uint32_t extents = 0;
struct pv_segment *pvseg;
dm_list_iterate_items(pvl, pvh) {
dm_list_iterate_items(per, pvl->pe_ranges) {
dm_list_iterate_items(pvseg, &pvl->pv->segments) {
if (!pvseg_is_allocated(pvseg))
extents += _overlap_pe(pvseg, per);
}
}
}
return extents;
}
/*
* Check all pv_segments in VG for consistency
*/
int check_pv_segments(struct volume_group *vg)
{
struct physical_volume *pv;
struct pv_list *pvl;
struct pv_segment *peg;
unsigned s, segno;
2005-06-01 20:51:55 +04:00
uint32_t start_pe, alloced;
uint32_t pv_count = 0, free_count = 0, extent_count = 0;
int ret = 1;
dm_list_iterate_items(pvl, &vg->pvs) {
pv = pvl->pv;
segno = 0;
start_pe = 0;
2005-06-01 20:51:55 +04:00
alloced = 0;
pv_count++;
dm_list_iterate_items(peg, &pv->segments) {
s = peg->lv_area;
/* FIXME Remove this next line eventually */
log_debug_alloc("%s %u: %6u %6u: %s(%u:%u)",
pv_dev_name(pv), segno++, peg->pe, peg->len,
peg->lvseg ? peg->lvseg->lv->name : "NULL",
peg->lvseg ? peg->lvseg->le : 0, s);
/* FIXME Add details here on failure instead */
if (start_pe != peg->pe) {
2005-06-01 20:51:55 +04:00
log_error("Gap in pvsegs: %u, %u",
start_pe, peg->pe);
ret = 0;
}
if (peg->lvseg) {
2005-06-01 20:51:55 +04:00
if (seg_type(peg->lvseg, s) != AREA_PV) {
log_error("Wrong lvseg area type");
ret = 0;
}
2005-06-01 20:51:55 +04:00
if (seg_pvseg(peg->lvseg, s) != peg) {
log_error("Inconsistent pvseg pointers");
ret = 0;
}
if (peg->lvseg->area_len != peg->len) {
2005-06-01 20:51:55 +04:00
log_error("Inconsistent length: %u %u",
peg->len,
peg->lvseg->area_len);
ret = 0;
}
2005-06-01 20:51:55 +04:00
alloced += peg->len;
}
start_pe += peg->len;
}
2005-06-01 20:51:55 +04:00
if (start_pe != pv->pe_count) {
log_error("PV segment pe_count mismatch: %u != %u",
start_pe, pv->pe_count);
ret = 0;
}
if (alloced != pv->pe_alloc_count) {
log_error("PV segment pe_alloc_count mismatch: "
"%u != %u", alloced, pv->pe_alloc_count);
ret = 0;
}
extent_count += start_pe;
free_count += (start_pe - alloced);
}
if (pv_count != vg->pv_count) {
log_error("PV segment VG pv_count mismatch: %u != %u",
pv_count, vg->pv_count);
ret = 0;
}
if (free_count != vg->free_count) {
log_error("PV segment VG free_count mismatch: %u != %u",
free_count, vg->free_count);
ret = 0;
}
if (extent_count != vg->extent_count) {
log_error("PV segment VG extent_count mismatch: %u != %u",
extent_count, vg->extent_count);
ret = 0;
}
return ret;
}
static int _reduce_pv(struct physical_volume *pv, struct volume_group *vg,
uint32_t old_pe_count, uint32_t new_pe_count)
{
struct pv_segment *peg, *pegt;
if (new_pe_count < pv->pe_alloc_count) {
log_error("%s: cannot resize to %" PRIu32 " extents "
"as %" PRIu32 " are allocated.",
pv_dev_name(pv), new_pe_count,
pv->pe_alloc_count);
return 0;
}
/* Check PEs to be removed are not already allocated */
dm_list_iterate_items(peg, &pv->segments) {
if (peg->pe + peg->len <= new_pe_count)
continue;
if (peg->lvseg) {
log_error("%s: cannot resize to %" PRIu32 " extents as "
"later ones are allocated.",
pv_dev_name(pv), new_pe_count);
return 0;
}
}
if (!pv_split_segment(vg->vgmem, pv, new_pe_count, NULL))
2008-01-30 16:19:47 +03:00
return_0;
dm_list_iterate_items_safe(peg, pegt, &pv->segments) {
if (peg->pe + peg->len > new_pe_count)
dm_list_del(&peg->list);
}
pv->pe_count = new_pe_count;
vg->extent_count -= (old_pe_count - new_pe_count);
vg->free_count -= (old_pe_count - new_pe_count);
return 1;
}
static int _extend_pv(struct physical_volume *pv, struct volume_group *vg,
uint32_t old_pe_count, uint32_t new_pe_count)
{
struct pv_segment *peg;
if ((uint64_t) new_pe_count * pv->pe_size > pv->size ) {
log_error("%s: cannot resize to %" PRIu32 " extents as there "
"is only room for %" PRIu64 ".", pv_dev_name(pv),
new_pe_count, pv->size / pv->pe_size);
return 0;
}
2012-03-02 01:49:32 +04:00
if (!(peg = _alloc_pv_segment(pv->fmt->cmd->mem, pv,
old_pe_count,
new_pe_count - old_pe_count,
NULL, 0)))
return_0;
dm_list_add(&pv->segments, &peg->list);
pv->pe_count = new_pe_count;
vg->extent_count += (new_pe_count - old_pe_count);
vg->free_count += (new_pe_count - old_pe_count);
return 1;
}
/*
* Resize a PV in a VG, adding or removing segments as needed.
* New size must fit within pv->size.
*/
static int pv_resize(struct physical_volume *pv,
struct volume_group *vg,
uint64_t size)
{
uint32_t old_pe_count, new_pe_count = 0;
if (size < pv_min_size()) {
2011-02-21 16:09:27 +03:00
log_error("Size must exceed minimum of %" PRIu64 " sectors on PV %s.",
pv_min_size(), pv_dev_name(pv));
return 0;
}
if (size < pv_pe_start(pv)) {
log_error("Size must exceed physical extent start "
"of %" PRIu64 " sectors on PV %s.",
pv_pe_start(pv), pv_dev_name(pv));
}
old_pe_count = pv->pe_count;
if (!pv->fmt->ops->pv_resize(pv->fmt, pv, vg, size)) {
log_error("Format specific resize of PV %s failed.",
pv_dev_name(pv));
return 0;
}
/* pv->pe_count is 0 now! We need to recalculate! */
/* If there's a VG, calculate new PE count value. */
/* Don't do for orphan VG */
if (vg && !is_orphan_vg(vg->name)) {
/* FIXME: Maybe PE calculation should go into pv->fmt->resize?
(like it is for pv->fmt->setup) */
if (!(new_pe_count = pv_size(pv) / vg->extent_size)) {
log_error("Size must leave space for at least one physical "
"extent of %" PRIu32 " sectors on PV %s.",
pv_pe_size(pv), pv_dev_name(pv));
return 0;
}
if (new_pe_count == old_pe_count) {
pv->pe_count = old_pe_count;
log_verbose("No change to size of physical volume %s.",
pv_dev_name(pv));
return 1;
}
log_verbose("Resizing physical volume %s from %" PRIu32
" to %" PRIu32 " extents.",
pv_dev_name(pv), pv->pe_count, new_pe_count);
if (new_pe_count > old_pe_count)
return _extend_pv(pv, vg, old_pe_count, new_pe_count);
else
return _reduce_pv(pv, vg, old_pe_count, new_pe_count);
}
return 1;
}
int pv_resize_single(struct cmd_context *cmd,
struct volume_group *vg,
struct physical_volume *pv,
const uint64_t new_size)
{
struct pv_list *pvl;
uint64_t size = 0;
int r = 0;
const char *pv_name = pv_dev_name(pv);
const char *vg_name = pv->vg_name;
struct volume_group *old_vg = vg;
int vg_needs_pv_write = 0;
vg = vg_read_for_update(cmd, vg_name, NULL, 0);
if (vg_read_error(vg)) {
release_vg(vg);
log_error("Unable to read volume group \"%s\".",
vg_name);
return 0;
}
if (!(pvl = find_pv_in_vg(vg, pv_name))) {
log_error("Unable to find \"%s\" in volume group \"%s\"",
pv_name, vg->name);
goto out;
}
pv = pvl->pv;
if (!archive(vg))
goto out;
if (!(pv->fmt->features & FMT_RESIZE_PV)) {
log_error("Physical volume %s format does not support resizing.",
pv_name);
goto out;
}
/* Get new size */
if (!dev_get_size(pv_dev(pv), &size)) {
log_error("%s: Couldn't get size.", pv_name);
goto out;
}
if (new_size) {
if (new_size > size)
log_warn("WARNING: %s: Overriding real size. "
"You could lose data.", pv_name);
log_verbose("%s: Pretending size is %" PRIu64 " not %" PRIu64
" sectors.", pv_name, new_size, pv_size(pv));
size = new_size;
}
log_verbose("Resizing volume \"%s\" to %" PRIu64 " sectors.",
pv_name, pv_size(pv));
if (!pv_resize(pv, vg, size))
goto_out;
log_verbose("Updating physical volume \"%s\"", pv_name);
/* Write PV label only if this an orphan PV or it has 2nd mda. */
if ((is_orphan_vg(vg_name) ||
(vg_needs_pv_write = (fid_get_mda_indexed(vg->fid,
(const char *) &pv->id, ID_LEN, 1) != NULL))) &&
!pv_write(cmd, pv, 1)) {
log_error("Failed to store physical volume \"%s\"",
pv_name);
goto out;
}
if (!is_orphan_vg(vg_name)) {
if (!vg_write(vg) || !vg_commit(vg)) {
log_error("Failed to store physical volume \"%s\" in "
"volume group \"%s\"", pv_name, vg_name);
goto out;
}
backup(vg);
}
log_print_unless_silent("Physical volume \"%s\" changed", pv_name);
r = 1;
out:
if (!r && vg_needs_pv_write)
log_error("Use pvcreate and vgcfgrestore "
"to repair from archived metadata.");
unlock_vg(cmd, vg_name);
if (!old_vg)
release_vg(vg);
return r;
}
const char _really_wipe[] =
"Really WIPE LABELS from physical volume \"%s\" of volume group \"%s\" [y/n]? ";
/*
* Decide whether it is "safe" to wipe the labels on this device.
* 0 indicates we may not.
*/
static int pvremove_check(struct cmd_context *cmd, const char *name,
unsigned force_count, unsigned prompt)
{
struct device *dev;
struct label *label;
struct pv_list *pvl;
struct dm_list *pvslist;
struct physical_volume *pv = NULL;
int r = 0;
/* FIXME Check partition type is LVM unless --force is given */
if (!(dev = dev_cache_get(name, cmd->filter))) {
log_error("Device %s not found", name);
return 0;
}
/* Is there a pv here already? */
/* If not, this is an error unless you used -f. */
if (!label_read(dev, &label, 0)) {
if (force_count)
return 1;
log_error("No PV label found on %s.", name);
return 0;
}
lvmcache_seed_infos_from_lvmetad(cmd);
if (!(pvslist = get_pvs(cmd)))
return_0;
dm_list_iterate_items(pvl, pvslist)
if (pvl->pv->dev == dev)
pv = pvl->pv;
if (!pv) {
log_error(INTERNAL_ERROR "Physical Volume %s has a label,"
" but is neither in a VG nor orphan.", name);
goto out; /* better safe than sorry */
}
if (is_orphan(pv)) {
r = 1;
goto out;
}
/* we must have -ff to overwrite a non orphan */
if (force_count < 2) {
log_error("PV %s belongs to Volume Group %s so please use vgreduce first.", name, pv_vg_name(pv));
log_error("(If you are certain you need pvremove, then confirm by using --force twice.)");
goto out;
}
/* prompt */
if (!prompt &&
yes_no_prompt("Really WIPE LABELS from physical volume \"%s\" "
"of volume group \"%s\" [y/n]? ",
name, pv_vg_name(pv)) == 'n') {
log_error("%s: physical volume label not removed", name);
goto out;
}
if (force_count) {
log_warn("WARNING: Wiping physical volume label from "
"%s%s%s%s", name,
!is_orphan(pv) ? " of volume group \"" : "",
pv_vg_name(pv),
!is_orphan(pv) ? "\"" : "");
}
r = 1;
out:
if (pvslist)
dm_list_iterate_items(pvl, pvslist)
free_pv_fid(pvl->pv);
return r;
}
int pvremove_single(struct cmd_context *cmd, const char *pv_name,
void *handle __attribute__((unused)), unsigned force_count,
unsigned prompt)
{
struct device *dev;
int r = 0;
if (!lock_vol(cmd, VG_ORPHANS, LCK_VG_WRITE, NULL)) {
log_error("Can't get lock for orphan PVs");
return 0;
}
if (!pvremove_check(cmd, pv_name, force_count, prompt))
goto out;
if (!(dev = dev_cache_get(pv_name, cmd->filter))) {
log_error("%s: Couldn't find device. Check your filters?",
pv_name);
goto out;
}
if (!dev_test_excl(dev)) {
/* FIXME Detect whether device-mapper is still using the device */
log_error("Can't open %s exclusively - not removing. "
"Mounted filesystem?", dev_name(dev));
goto out;
}
/* Wipe existing label(s) */
if (!label_remove(dev)) {
log_error("Failed to wipe existing label(s) on %s", pv_name);
goto out;
}
if (!lvmetad_pv_gone_by_dev(dev, NULL))
goto_out;
log_print_unless_silent("Labels on physical volume \"%s\" successfully wiped",
pv_name);
r = 1;
out:
unlock_vg(cmd, VG_ORPHANS);
return r;
}
int pvcreate_single(struct cmd_context *cmd, const char *pv_name,
struct pvcreate_params *pp)
{
int r = 0;
if (!lock_vol(cmd, VG_ORPHANS, LCK_VG_WRITE, NULL)) {
log_error("Can't get lock for orphan PVs");
return 0;
}
if (!(pvcreate_vol(cmd, pv_name, pp, 1)))
goto_out;
r = 1;
out:
unlock_vg(cmd, VG_ORPHANS);
return r;
}