1
0
mirror of git://sourceware.org/git/lvm2.git synced 2024-10-28 20:25:52 +03:00
lvm2/lib/metadata/pv_map.c
Heinz Mauelshagen 302b6c99a7 raid_manip: v2 fix multi-segment misallocation on 'lvconvert --repair'
The previous patch felt short WRT disabling allocation on PVs holding other
legs of the RAID LV persistently; this patch introduces an internal,
transient PV flag PV_ALLOCATION_PROHIBITED to address this very problem.

General problem description for completeness:

An 'lvconvert --repair $RAID_LV" to replace a failed leg of a multi-segment
RAID10/4/5/6 logical volume can lead to allocation of (parts of) the replacement
image component pair on the physical volume of another image component
(e.g. image 0 allocated on the same PV as image 1 silently impeding resilience).

Patch fixes this severe resilince issue by prohibiting allocation on PVs
already holding other legs of the RAID set. It allows to allocate free space
on any operational PV already holding parts of the image component pair.
2015-01-16 13:44:16 +01:00

229 lines
5.0 KiB
C

/*
* Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
* Copyright (C) 2004-2006 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 Lesser General Public License v.2.1.
*
* You should have received a copy of the GNU Lesser 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 "lib.h"
#include "pv_map.h"
#include <assert.h>
/*
* Areas are maintained in size order, largest first.
*
* FIXME Cope with overlap.
*/
static void _insert_area(struct dm_list *head, struct pv_area *a, unsigned reduced)
{
struct pv_area *pva;
uint32_t count = reduced ? a->unreserved : a->count;
dm_list_iterate_items(pva, head)
if (count > pva->count)
break;
dm_list_add(&pva->list, &a->list);
a->map->pe_count += a->count;
}
static void _remove_area(struct pv_area *a)
{
dm_list_del(&a->list);
a->map->pe_count -= a->count;
}
static int _create_single_area(struct dm_pool *mem, struct pv_map *pvm,
uint32_t start, uint32_t length)
{
struct pv_area *pva;
if (!(pva = dm_pool_zalloc(mem, sizeof(*pva))))
return_0;
log_debug_alloc("Allowing allocation on %s start PE %" PRIu32 " length %"
PRIu32, pv_dev_name(pvm->pv), start, length);
pva->map = pvm;
pva->start = start;
pva->count = length;
pva->unreserved = pva->count;
_insert_area(&pvm->areas, pva, 0);
return 1;
}
static int _create_alloc_areas_for_pv(struct dm_pool *mem, struct pv_map *pvm,
uint32_t start, uint32_t count)
{
struct pv_segment *peg;
uint32_t pe, end, area_len;
/* Only select extents from start to end inclusive */
end = start + count - 1;
if (end > pvm->pv->pe_count - 1)
end = pvm->pv->pe_count - 1;
pe = start;
/* Walk through complete ordered list of device segments */
dm_list_iterate_items(peg, &pvm->pv->segments) {
/* pe holds the next extent we want to check */
/* Beyond the range we're interested in? */
if (pe > end)
break;
/* Skip if we haven't reached the first seg we want yet */
if (pe > peg->pe + peg->len - 1)
continue;
/* Free? */
if (peg->lvseg)
goto next;
/* How much of this peg do we need? */
area_len = (end >= peg->pe + peg->len - 1) ?
peg->len - (pe - peg->pe) : end - pe + 1;
if (!_create_single_area(mem, pvm, pe, area_len))
return_0;
next:
pe = peg->pe + peg->len;
}
return 1;
}
static int _create_all_areas_for_pv(struct dm_pool *mem, struct pv_map *pvm,
struct dm_list *pe_ranges)
{
struct pe_range *aa;
if (!pe_ranges) {
/* Use whole PV */
if (!_create_alloc_areas_for_pv(mem, pvm, UINT32_C(0),
pvm->pv->pe_count))
return_0;
return 1;
}
dm_list_iterate_items(aa, pe_ranges) {
if (!_create_alloc_areas_for_pv(mem, pvm, aa->start,
aa->count))
return_0;
}
return 1;
}
static int _create_maps(struct dm_pool *mem, struct dm_list *pvs, struct dm_list *pvms)
{
struct pv_map *pvm, *pvm2;
struct pv_list *pvl;
dm_list_iterate_items(pvl, pvs) {
if (!(pvl->pv->status & ALLOCATABLE_PV) ||
(pvl->pv->status & PV_ALLOCATION_PROHIBITED)) {
pvl->pv->status &= ~PV_ALLOCATION_PROHIBITED;
continue;
}
if (is_missing_pv(pvl->pv))
continue;
assert(pvl->pv->dev);
pvm = NULL;
dm_list_iterate_items(pvm2, pvms)
if (pvm2->pv->dev == pvl->pv->dev) {
pvm = pvm2;
break;
}
if (!pvm) {
if (!(pvm = dm_pool_zalloc(mem, sizeof(*pvm))))
return_0;
pvm->pv = pvl->pv;
dm_list_init(&pvm->areas);
dm_list_add(pvms, &pvm->list);
}
if (!_create_all_areas_for_pv(mem, pvm, pvl->pe_ranges))
return_0;
}
return 1;
}
/*
* Create list of PV areas available for this particular allocation
*/
struct dm_list *create_pv_maps(struct dm_pool *mem, struct volume_group *vg,
struct dm_list *allocatable_pvs)
{
struct dm_list *pvms;
if (!(pvms = dm_pool_zalloc(mem, sizeof(*pvms)))) {
log_error("create_pv_maps alloc failed");
return NULL;
}
dm_list_init(pvms);
if (!_create_maps(mem, allocatable_pvs, pvms)) {
log_error("Couldn't create physical volume maps in %s",
vg->name);
dm_pool_free(mem, pvms);
return NULL;
}
return pvms;
}
void consume_pv_area(struct pv_area *pva, uint32_t to_go)
{
_remove_area(pva);
assert(to_go <= pva->count);
if (to_go < pva->count) {
/* split the area */
pva->start += to_go;
pva->count -= to_go;
pva->unreserved = pva->count;
_insert_area(&pva->map->areas, pva, 0);
}
}
/*
* Remove an area from list and reinsert it based on its new size
* after a provisional allocation (or reverting one).
*/
void reinsert_changed_pv_area(struct pv_area *pva)
{
_remove_area(pva);
_insert_area(&pva->map->areas, pva, 1);
}
uint32_t pv_maps_size(struct dm_list *pvms)
{
struct pv_map *pvm;
uint32_t pe_count = 0;
dm_list_iterate_items(pvm, pvms)
pe_count += pvm->pe_count;
return pe_count;
}