mirror_zfs/lib/libzfs/os/linux/libzfs_pool_os.c
George Wilson 8e82ffba7b
pool may become suspended during device expansion
When expanding a device zfs needs to rescan the partition table to
get the correct size. This can only happen when we're in the kernel
and requires the device to be closed. As part of the rescan, udev is
notified and the device links are removed and recreated. This leave a
window where the vdev code may try to reopen the device before udev
has recreated the link. If that happens, then the pool may end up in
a suspended state.

To correct this, we leverage the BLKPG_RESIZE_PARTITION ioctl which
allows the partition information to be modified even while it's in use.
This ioctl also does not remove the device link associated with the zfs
data partition so it eliminates the race condition that can occur in
the kernel.

Reviewed-by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: George Wilson <gwilson@delphix.com>
Closes #10897
2020-09-17 20:03:10 -07:00

343 lines
9.2 KiB
C

/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright 2015 Nexenta Systems, Inc. All rights reserved.
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2011, 2018 by Delphix. All rights reserved.
* Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>
* Copyright (c) 2018 Datto Inc.
* Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
* Copyright (c) 2017, Intel Corporation.
* Copyright (c) 2018, loli10K <ezomori.nozomu@gmail.com>
*/
#include <errno.h>
#include <libintl.h>
#include <stdio.h>
#include <stdlib.h>
#include <strings.h>
#include <unistd.h>
#include <libgen.h>
#include <zone.h>
#include <sys/stat.h>
#include <sys/efi_partition.h>
#include <sys/systeminfo.h>
#include <sys/vtoc.h>
#include <sys/zfs_ioctl.h>
#include <sys/vdev_disk.h>
#include <dlfcn.h>
#include <libzutil.h>
#include "zfs_namecheck.h"
#include "zfs_prop.h"
#include "libzfs_impl.h"
#include "zfs_comutil.h"
#include "zfeature_common.h"
/*
* If the device has being dynamically expanded then we need to relabel
* the disk to use the new unallocated space.
*/
int
zpool_relabel_disk(libzfs_handle_t *hdl, const char *path, const char *msg)
{
int fd, error;
if ((fd = open(path, O_RDWR|O_DIRECT)) < 0) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot "
"relabel '%s': unable to open device: %d"), path, errno);
return (zfs_error(hdl, EZFS_OPENFAILED, msg));
}
/*
* It's possible that we might encounter an error if the device
* does not have any unallocated space left. If so, we simply
* ignore that error and continue on.
*/
error = efi_use_whole_disk(fd);
/* Flush the buffers to disk and invalidate the page cache. */
(void) fsync(fd);
(void) ioctl(fd, BLKFLSBUF);
(void) close(fd);
if (error && error != VT_ENOSPC) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot "
"relabel '%s': unable to read disk capacity"), path);
return (zfs_error(hdl, EZFS_NOCAP, msg));
}
return (0);
}
/*
* Read the EFI label from the config, if a label does not exist then
* pass back the error to the caller. If the caller has passed a non-NULL
* diskaddr argument then we set it to the starting address of the EFI
* partition.
*/
static int
read_efi_label(nvlist_t *config, diskaddr_t *sb)
{
char *path;
int fd;
char diskname[MAXPATHLEN];
int err = -1;
if (nvlist_lookup_string(config, ZPOOL_CONFIG_PATH, &path) != 0)
return (err);
(void) snprintf(diskname, sizeof (diskname), "%s%s", DISK_ROOT,
strrchr(path, '/'));
if ((fd = open(diskname, O_RDONLY|O_DIRECT)) >= 0) {
struct dk_gpt *vtoc;
if ((err = efi_alloc_and_read(fd, &vtoc)) >= 0) {
if (sb != NULL)
*sb = vtoc->efi_parts[0].p_start;
efi_free(vtoc);
}
(void) close(fd);
}
return (err);
}
/*
* determine where a partition starts on a disk in the current
* configuration
*/
static diskaddr_t
find_start_block(nvlist_t *config)
{
nvlist_t **child;
uint_t c, children;
diskaddr_t sb = MAXOFFSET_T;
uint64_t wholedisk;
if (nvlist_lookup_nvlist_array(config,
ZPOOL_CONFIG_CHILDREN, &child, &children) != 0) {
if (nvlist_lookup_uint64(config,
ZPOOL_CONFIG_WHOLE_DISK,
&wholedisk) != 0 || !wholedisk) {
return (MAXOFFSET_T);
}
if (read_efi_label(config, &sb) < 0)
sb = MAXOFFSET_T;
return (sb);
}
for (c = 0; c < children; c++) {
sb = find_start_block(child[c]);
if (sb != MAXOFFSET_T) {
return (sb);
}
}
return (MAXOFFSET_T);
}
static int
zpool_label_disk_check(char *path)
{
struct dk_gpt *vtoc;
int fd, err;
if ((fd = open(path, O_RDONLY|O_DIRECT)) < 0)
return (errno);
if ((err = efi_alloc_and_read(fd, &vtoc)) != 0) {
(void) close(fd);
return (err);
}
if (vtoc->efi_flags & EFI_GPT_PRIMARY_CORRUPT) {
efi_free(vtoc);
(void) close(fd);
return (EIDRM);
}
efi_free(vtoc);
(void) close(fd);
return (0);
}
/*
* Generate a unique partition name for the ZFS member. Partitions must
* have unique names to ensure udev will be able to create symlinks under
* /dev/disk/by-partlabel/ for all pool members. The partition names are
* of the form <pool>-<unique-id>.
*/
static void
zpool_label_name(char *label_name, int label_size)
{
uint64_t id = 0;
int fd;
fd = open("/dev/urandom", O_RDONLY);
if (fd >= 0) {
if (read(fd, &id, sizeof (id)) != sizeof (id))
id = 0;
close(fd);
}
if (id == 0)
id = (((uint64_t)rand()) << 32) | (uint64_t)rand();
snprintf(label_name, label_size, "zfs-%016llx", (u_longlong_t)id);
}
/*
* Label an individual disk. The name provided is the short name,
* stripped of any leading /dev path.
*/
int
zpool_label_disk(libzfs_handle_t *hdl, zpool_handle_t *zhp, const char *name)
{
char path[MAXPATHLEN];
struct dk_gpt *vtoc;
int rval, fd;
size_t resv = EFI_MIN_RESV_SIZE;
uint64_t slice_size;
diskaddr_t start_block;
char errbuf[1024];
/* prepare an error message just in case */
(void) snprintf(errbuf, sizeof (errbuf),
dgettext(TEXT_DOMAIN, "cannot label '%s'"), name);
if (zhp) {
nvlist_t *nvroot;
verify(nvlist_lookup_nvlist(zhp->zpool_config,
ZPOOL_CONFIG_VDEV_TREE, &nvroot) == 0);
if (zhp->zpool_start_block == 0)
start_block = find_start_block(nvroot);
else
start_block = zhp->zpool_start_block;
zhp->zpool_start_block = start_block;
} else {
/* new pool */
start_block = NEW_START_BLOCK;
}
(void) snprintf(path, sizeof (path), "%s/%s", DISK_ROOT, name);
if ((fd = open(path, O_RDWR|O_DIRECT|O_EXCL)) < 0) {
/*
* This shouldn't happen. We've long since verified that this
* is a valid device.
*/
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot "
"label '%s': unable to open device: %d"), path, errno);
return (zfs_error(hdl, EZFS_OPENFAILED, errbuf));
}
if (efi_alloc_and_init(fd, EFI_NUMPAR, &vtoc) != 0) {
/*
* The only way this can fail is if we run out of memory, or we
* were unable to read the disk's capacity
*/
if (errno == ENOMEM)
(void) no_memory(hdl);
(void) close(fd);
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "cannot "
"label '%s': unable to read disk capacity"), path);
return (zfs_error(hdl, EZFS_NOCAP, errbuf));
}
slice_size = vtoc->efi_last_u_lba + 1;
slice_size -= EFI_MIN_RESV_SIZE;
if (start_block == MAXOFFSET_T)
start_block = NEW_START_BLOCK;
slice_size -= start_block;
slice_size = P2ALIGN(slice_size, PARTITION_END_ALIGNMENT);
vtoc->efi_parts[0].p_start = start_block;
vtoc->efi_parts[0].p_size = slice_size;
/*
* Why we use V_USR: V_BACKUP confuses users, and is considered
* disposable by some EFI utilities (since EFI doesn't have a backup
* slice). V_UNASSIGNED is supposed to be used only for zero size
* partitions, and efi_write() will fail if we use it. V_ROOT, V_BOOT,
* etc. were all pretty specific. V_USR is as close to reality as we
* can get, in the absence of V_OTHER.
*/
vtoc->efi_parts[0].p_tag = V_USR;
zpool_label_name(vtoc->efi_parts[0].p_name, EFI_PART_NAME_LEN);
vtoc->efi_parts[8].p_start = slice_size + start_block;
vtoc->efi_parts[8].p_size = resv;
vtoc->efi_parts[8].p_tag = V_RESERVED;
rval = efi_write(fd, vtoc);
/* Flush the buffers to disk and invalidate the page cache. */
(void) fsync(fd);
(void) ioctl(fd, BLKFLSBUF);
if (rval == 0)
rval = efi_rescan(fd);
/*
* Some block drivers (like pcata) may not support EFI GPT labels.
* Print out a helpful error message directing the user to manually
* label the disk and give a specific slice.
*/
if (rval != 0) {
(void) close(fd);
efi_free(vtoc);
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "try using "
"parted(8) and then provide a specific slice: %d"), rval);
return (zfs_error(hdl, EZFS_LABELFAILED, errbuf));
}
(void) close(fd);
efi_free(vtoc);
(void) snprintf(path, sizeof (path), "%s/%s", DISK_ROOT, name);
(void) zfs_append_partition(path, MAXPATHLEN);
/* Wait to udev to signal use the device has settled. */
rval = zpool_label_disk_wait(path, DISK_LABEL_WAIT);
if (rval) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "failed to "
"detect device partitions on '%s': %d"), path, rval);
return (zfs_error(hdl, EZFS_LABELFAILED, errbuf));
}
/* We can't be to paranoid. Read the label back and verify it. */
(void) snprintf(path, sizeof (path), "%s/%s", DISK_ROOT, name);
rval = zpool_label_disk_check(path);
if (rval) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "freshly written "
"EFI label on '%s' is damaged. Ensure\nthis device "
"is not in use, and is functioning properly: %d"),
path, rval);
return (zfs_error(hdl, EZFS_LABELFAILED, errbuf));
}
return (0);
}