mirror_zfs/lib/libzfs/libzfs_util.c
Serapheim Dimitropoulos d2734cce68 OpenZFS 9166 - zfs storage pool checkpoint
Details about the motivation of this feature and its usage can
be found in this blogpost:

    https://sdimitro.github.io/post/zpool-checkpoint/

A lightning talk of this feature can be found here:
https://www.youtube.com/watch?v=fPQA8K40jAM

Implementation details can be found in big block comment of
spa_checkpoint.c

Side-changes that are relevant to this commit but not explained
elsewhere:

* renames members of "struct metaslab trees to be shorter without
  losing meaning

* space_map_{alloc,truncate}() accept a block size as a
  parameter. The reason is that in the current state all space
  maps that we allocate through the DMU use a global tunable
  (space_map_blksz) which defauls to 4KB. This is ok for metaslab
  space maps in terms of bandwirdth since they are scattered all
  over the disk. But for other space maps this default is probably
  not what we want. Examples are device removal's vdev_obsolete_sm
  or vdev_chedkpoint_sm from this review. Both of these have a
  1:1 relationship with each vdev and could benefit from a bigger
  block size.

Porting notes:

* The part of dsl_scan_sync() which handles async destroys has
  been moved into the new dsl_process_async_destroys() function.

* Remove "VERIFY(!(flags & FWRITE))" in "kernel.c" so zhack can write
  to block device backed pools.

* ZTS:
  * Fix get_txg() in zpool_sync_001_pos due to "checkpoint_txg".

  * Don't use large dd block sizes on /dev/urandom under Linux in
    checkpoint_capacity.

  * Adopt Delphix-OS's setting of 4 (spa_asize_inflation =
    SPA_DVAS_PER_BP + 1) for the checkpoint_capacity test to speed
    its attempts to fill the pool

  * Create the base and nested pools with sync=disabled to speed up
    the "setup" phase.

  * Clear labels in test pool between checkpoint tests to avoid
    duplicate pool issues.

  * The import_rewind_device_replaced test has been marked as "known
    to fail" for the reasons listed in its DISCLAIMER.

  * New module parameters:

      zfs_spa_discard_memory_limit,
      zfs_remove_max_bytes_pause (not documented - debugging only)
      vdev_max_ms_count (formerly metaslabs_per_vdev)
      vdev_min_ms_count

Authored by: Serapheim Dimitropoulos <serapheim.dimitro@delphix.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: John Kennedy <john.kennedy@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: Brian Behlendorf <behlendorf1@llnl.gov>
Approved by: Richard Lowe <richlowe@richlowe.net>
Ported-by: Tim Chase <tim@chase2k.com>
Signed-off-by: Tim Chase <tim@chase2k.com>

OpenZFS-issue: https://illumos.org/issues/9166
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/7159fdb8
Closes #7570
2018-06-26 10:07:42 -07:00

2206 lines
53 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2018, Joyent, Inc. All rights reserved.
* Copyright (c) 2011, 2017 by Delphix. All rights reserved.
* Copyright 2016 Igor Kozhukhov <ikozhukhov@gmail.com>
* Copyright (c) 2017 Datto Inc.
*/
/*
* Internal utility routines for the ZFS library.
*/
#include <errno.h>
#include <fcntl.h>
#include <libintl.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <strings.h>
#include <unistd.h>
#include <ctype.h>
#include <math.h>
#include <sys/stat.h>
#include <sys/mnttab.h>
#include <sys/mntent.h>
#include <sys/types.h>
#include <sys/wait.h>
#include <libzfs.h>
#include <libzfs_core.h>
#include "libzfs_impl.h"
#include "zfs_prop.h"
#include "zfeature_common.h"
#include <zfs_fletcher.h>
int
libzfs_errno(libzfs_handle_t *hdl)
{
return (hdl->libzfs_error);
}
const char *
libzfs_error_init(int error)
{
switch (error) {
case ENXIO:
return (dgettext(TEXT_DOMAIN, "The ZFS modules are not "
"loaded.\nTry running '/sbin/modprobe zfs' as root "
"to load them.\n"));
case ENOENT:
return (dgettext(TEXT_DOMAIN, "/dev/zfs and /proc/self/mounts "
"are required.\nTry running 'udevadm trigger' and 'mount "
"-t proc proc /proc' as root.\n"));
case ENOEXEC:
return (dgettext(TEXT_DOMAIN, "The ZFS modules cannot be "
"auto-loaded.\nTry running '/sbin/modprobe zfs' as "
"root to manually load them.\n"));
case EACCES:
return (dgettext(TEXT_DOMAIN, "Permission denied the "
"ZFS utilities must be run as root.\n"));
default:
return (dgettext(TEXT_DOMAIN, "Failed to initialize the "
"libzfs library.\n"));
}
}
const char *
libzfs_error_action(libzfs_handle_t *hdl)
{
return (hdl->libzfs_action);
}
const char *
libzfs_error_description(libzfs_handle_t *hdl)
{
if (hdl->libzfs_desc[0] != '\0')
return (hdl->libzfs_desc);
switch (hdl->libzfs_error) {
case EZFS_NOMEM:
return (dgettext(TEXT_DOMAIN, "out of memory"));
case EZFS_BADPROP:
return (dgettext(TEXT_DOMAIN, "invalid property value"));
case EZFS_PROPREADONLY:
return (dgettext(TEXT_DOMAIN, "read-only property"));
case EZFS_PROPTYPE:
return (dgettext(TEXT_DOMAIN, "property doesn't apply to "
"datasets of this type"));
case EZFS_PROPNONINHERIT:
return (dgettext(TEXT_DOMAIN, "property cannot be inherited"));
case EZFS_PROPSPACE:
return (dgettext(TEXT_DOMAIN, "invalid quota or reservation"));
case EZFS_BADTYPE:
return (dgettext(TEXT_DOMAIN, "operation not applicable to "
"datasets of this type"));
case EZFS_BUSY:
return (dgettext(TEXT_DOMAIN, "pool or dataset is busy"));
case EZFS_EXISTS:
return (dgettext(TEXT_DOMAIN, "pool or dataset exists"));
case EZFS_NOENT:
return (dgettext(TEXT_DOMAIN, "no such pool or dataset"));
case EZFS_BADSTREAM:
return (dgettext(TEXT_DOMAIN, "invalid backup stream"));
case EZFS_DSREADONLY:
return (dgettext(TEXT_DOMAIN, "dataset is read-only"));
case EZFS_VOLTOOBIG:
return (dgettext(TEXT_DOMAIN, "volume size exceeds limit for "
"this system"));
case EZFS_INVALIDNAME:
return (dgettext(TEXT_DOMAIN, "invalid name"));
case EZFS_BADRESTORE:
return (dgettext(TEXT_DOMAIN, "unable to restore to "
"destination"));
case EZFS_BADBACKUP:
return (dgettext(TEXT_DOMAIN, "backup failed"));
case EZFS_BADTARGET:
return (dgettext(TEXT_DOMAIN, "invalid target vdev"));
case EZFS_NODEVICE:
return (dgettext(TEXT_DOMAIN, "no such device in pool"));
case EZFS_BADDEV:
return (dgettext(TEXT_DOMAIN, "invalid device"));
case EZFS_NOREPLICAS:
return (dgettext(TEXT_DOMAIN, "no valid replicas"));
case EZFS_RESILVERING:
return (dgettext(TEXT_DOMAIN, "currently resilvering"));
case EZFS_BADVERSION:
return (dgettext(TEXT_DOMAIN, "unsupported version or "
"feature"));
case EZFS_POOLUNAVAIL:
return (dgettext(TEXT_DOMAIN, "pool is unavailable"));
case EZFS_DEVOVERFLOW:
return (dgettext(TEXT_DOMAIN, "too many devices in one vdev"));
case EZFS_BADPATH:
return (dgettext(TEXT_DOMAIN, "must be an absolute path"));
case EZFS_CROSSTARGET:
return (dgettext(TEXT_DOMAIN, "operation crosses datasets or "
"pools"));
case EZFS_ZONED:
return (dgettext(TEXT_DOMAIN, "dataset in use by local zone"));
case EZFS_MOUNTFAILED:
return (dgettext(TEXT_DOMAIN, "mount failed"));
case EZFS_UMOUNTFAILED:
return (dgettext(TEXT_DOMAIN, "umount failed"));
case EZFS_UNSHARENFSFAILED:
return (dgettext(TEXT_DOMAIN, "unshare(1M) failed"));
case EZFS_SHARENFSFAILED:
return (dgettext(TEXT_DOMAIN, "share(1M) failed"));
case EZFS_UNSHARESMBFAILED:
return (dgettext(TEXT_DOMAIN, "smb remove share failed"));
case EZFS_SHARESMBFAILED:
return (dgettext(TEXT_DOMAIN, "smb add share failed"));
case EZFS_PERM:
return (dgettext(TEXT_DOMAIN, "permission denied"));
case EZFS_NOSPC:
return (dgettext(TEXT_DOMAIN, "out of space"));
case EZFS_FAULT:
return (dgettext(TEXT_DOMAIN, "bad address"));
case EZFS_IO:
return (dgettext(TEXT_DOMAIN, "I/O error"));
case EZFS_INTR:
return (dgettext(TEXT_DOMAIN, "signal received"));
case EZFS_ISSPARE:
return (dgettext(TEXT_DOMAIN, "device is reserved as a hot "
"spare"));
case EZFS_INVALCONFIG:
return (dgettext(TEXT_DOMAIN, "invalid vdev configuration"));
case EZFS_RECURSIVE:
return (dgettext(TEXT_DOMAIN, "recursive dataset dependency"));
case EZFS_NOHISTORY:
return (dgettext(TEXT_DOMAIN, "no history available"));
case EZFS_POOLPROPS:
return (dgettext(TEXT_DOMAIN, "failed to retrieve "
"pool properties"));
case EZFS_POOL_NOTSUP:
return (dgettext(TEXT_DOMAIN, "operation not supported "
"on this type of pool"));
case EZFS_POOL_INVALARG:
return (dgettext(TEXT_DOMAIN, "invalid argument for "
"this pool operation"));
case EZFS_NAMETOOLONG:
return (dgettext(TEXT_DOMAIN, "dataset name is too long"));
case EZFS_OPENFAILED:
return (dgettext(TEXT_DOMAIN, "open failed"));
case EZFS_NOCAP:
return (dgettext(TEXT_DOMAIN,
"disk capacity information could not be retrieved"));
case EZFS_LABELFAILED:
return (dgettext(TEXT_DOMAIN, "write of label failed"));
case EZFS_BADWHO:
return (dgettext(TEXT_DOMAIN, "invalid user/group"));
case EZFS_BADPERM:
return (dgettext(TEXT_DOMAIN, "invalid permission"));
case EZFS_BADPERMSET:
return (dgettext(TEXT_DOMAIN, "invalid permission set name"));
case EZFS_NODELEGATION:
return (dgettext(TEXT_DOMAIN, "delegated administration is "
"disabled on pool"));
case EZFS_BADCACHE:
return (dgettext(TEXT_DOMAIN, "invalid or missing cache file"));
case EZFS_ISL2CACHE:
return (dgettext(TEXT_DOMAIN, "device is in use as a cache"));
case EZFS_VDEVNOTSUP:
return (dgettext(TEXT_DOMAIN, "vdev specification is not "
"supported"));
case EZFS_NOTSUP:
return (dgettext(TEXT_DOMAIN, "operation not supported "
"on this dataset"));
case EZFS_ACTIVE_SPARE:
return (dgettext(TEXT_DOMAIN, "pool has active shared spare "
"device"));
case EZFS_UNPLAYED_LOGS:
return (dgettext(TEXT_DOMAIN, "log device has unplayed intent "
"logs"));
case EZFS_REFTAG_RELE:
return (dgettext(TEXT_DOMAIN, "no such tag on this dataset"));
case EZFS_REFTAG_HOLD:
return (dgettext(TEXT_DOMAIN, "tag already exists on this "
"dataset"));
case EZFS_TAGTOOLONG:
return (dgettext(TEXT_DOMAIN, "tag too long"));
case EZFS_PIPEFAILED:
return (dgettext(TEXT_DOMAIN, "pipe create failed"));
case EZFS_THREADCREATEFAILED:
return (dgettext(TEXT_DOMAIN, "thread create failed"));
case EZFS_POSTSPLIT_ONLINE:
return (dgettext(TEXT_DOMAIN, "disk was split from this pool "
"into a new one"));
case EZFS_SCRUB_PAUSED:
return (dgettext(TEXT_DOMAIN, "scrub is paused; "
"use 'zpool scrub' to resume"));
case EZFS_SCRUBBING:
return (dgettext(TEXT_DOMAIN, "currently scrubbing; "
"use 'zpool scrub -s' to cancel current scrub"));
case EZFS_NO_SCRUB:
return (dgettext(TEXT_DOMAIN, "there is no active scrub"));
case EZFS_DIFF:
return (dgettext(TEXT_DOMAIN, "unable to generate diffs"));
case EZFS_DIFFDATA:
return (dgettext(TEXT_DOMAIN, "invalid diff data"));
case EZFS_POOLREADONLY:
return (dgettext(TEXT_DOMAIN, "pool is read-only"));
case EZFS_NO_PENDING:
return (dgettext(TEXT_DOMAIN, "operation is not "
"in progress"));
case EZFS_CHECKPOINT_EXISTS:
return (dgettext(TEXT_DOMAIN, "checkpoint exists"));
case EZFS_DISCARDING_CHECKPOINT:
return (dgettext(TEXT_DOMAIN, "currently discarding "
"checkpoint"));
case EZFS_NO_CHECKPOINT:
return (dgettext(TEXT_DOMAIN, "checkpoint does not exist"));
case EZFS_DEVRM_IN_PROGRESS:
return (dgettext(TEXT_DOMAIN, "device removal in progress"));
case EZFS_VDEV_TOO_BIG:
return (dgettext(TEXT_DOMAIN, "device exceeds supported size"));
case EZFS_ACTIVE_POOL:
return (dgettext(TEXT_DOMAIN, "pool is imported on a "
"different host"));
case EZFS_CRYPTOFAILED:
return (dgettext(TEXT_DOMAIN, "encryption failure"));
case EZFS_UNKNOWN:
return (dgettext(TEXT_DOMAIN, "unknown error"));
default:
assert(hdl->libzfs_error == 0);
return (dgettext(TEXT_DOMAIN, "no error"));
}
}
/*PRINTFLIKE2*/
void
zfs_error_aux(libzfs_handle_t *hdl, const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
(void) vsnprintf(hdl->libzfs_desc, sizeof (hdl->libzfs_desc),
fmt, ap);
hdl->libzfs_desc_active = 1;
va_end(ap);
}
static void
zfs_verror(libzfs_handle_t *hdl, int error, const char *fmt, va_list ap)
{
(void) vsnprintf(hdl->libzfs_action, sizeof (hdl->libzfs_action),
fmt, ap);
hdl->libzfs_error = error;
if (hdl->libzfs_desc_active)
hdl->libzfs_desc_active = 0;
else
hdl->libzfs_desc[0] = '\0';
if (hdl->libzfs_printerr) {
if (error == EZFS_UNKNOWN) {
(void) fprintf(stderr, dgettext(TEXT_DOMAIN, "internal "
"error: %s\n"), libzfs_error_description(hdl));
abort();
}
(void) fprintf(stderr, "%s: %s\n", hdl->libzfs_action,
libzfs_error_description(hdl));
if (error == EZFS_NOMEM)
exit(1);
}
}
int
zfs_error(libzfs_handle_t *hdl, int error, const char *msg)
{
return (zfs_error_fmt(hdl, error, "%s", msg));
}
/*PRINTFLIKE3*/
int
zfs_error_fmt(libzfs_handle_t *hdl, int error, const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
zfs_verror(hdl, error, fmt, ap);
va_end(ap);
return (-1);
}
static int
zfs_common_error(libzfs_handle_t *hdl, int error, const char *fmt,
va_list ap)
{
switch (error) {
case EPERM:
case EACCES:
zfs_verror(hdl, EZFS_PERM, fmt, ap);
return (-1);
case ECANCELED:
zfs_verror(hdl, EZFS_NODELEGATION, fmt, ap);
return (-1);
case EIO:
zfs_verror(hdl, EZFS_IO, fmt, ap);
return (-1);
case EFAULT:
zfs_verror(hdl, EZFS_FAULT, fmt, ap);
return (-1);
case EINTR:
zfs_verror(hdl, EZFS_INTR, fmt, ap);
return (-1);
}
return (0);
}
int
zfs_standard_error(libzfs_handle_t *hdl, int error, const char *msg)
{
return (zfs_standard_error_fmt(hdl, error, "%s", msg));
}
/*PRINTFLIKE3*/
int
zfs_standard_error_fmt(libzfs_handle_t *hdl, int error, const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
if (zfs_common_error(hdl, error, fmt, ap) != 0) {
va_end(ap);
return (-1);
}
switch (error) {
case ENXIO:
case ENODEV:
case EPIPE:
zfs_verror(hdl, EZFS_IO, fmt, ap);
break;
case ENOENT:
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"dataset does not exist"));
zfs_verror(hdl, EZFS_NOENT, fmt, ap);
break;
case ENOSPC:
case EDQUOT:
zfs_verror(hdl, EZFS_NOSPC, fmt, ap);
break;
case EEXIST:
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"dataset already exists"));
zfs_verror(hdl, EZFS_EXISTS, fmt, ap);
break;
case EBUSY:
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"dataset is busy"));
zfs_verror(hdl, EZFS_BUSY, fmt, ap);
break;
case EROFS:
zfs_verror(hdl, EZFS_POOLREADONLY, fmt, ap);
break;
case ENAMETOOLONG:
zfs_verror(hdl, EZFS_NAMETOOLONG, fmt, ap);
break;
case ENOTSUP:
zfs_verror(hdl, EZFS_BADVERSION, fmt, ap);
break;
case EAGAIN:
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"pool I/O is currently suspended"));
zfs_verror(hdl, EZFS_POOLUNAVAIL, fmt, ap);
break;
case EREMOTEIO:
zfs_verror(hdl, EZFS_ACTIVE_POOL, fmt, ap);
break;
default:
zfs_error_aux(hdl, strerror(error));
zfs_verror(hdl, EZFS_UNKNOWN, fmt, ap);
break;
}
va_end(ap);
return (-1);
}
int
zpool_standard_error(libzfs_handle_t *hdl, int error, const char *msg)
{
return (zpool_standard_error_fmt(hdl, error, "%s", msg));
}
/*PRINTFLIKE3*/
int
zpool_standard_error_fmt(libzfs_handle_t *hdl, int error, const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
if (zfs_common_error(hdl, error, fmt, ap) != 0) {
va_end(ap);
return (-1);
}
switch (error) {
case ENODEV:
zfs_verror(hdl, EZFS_NODEVICE, fmt, ap);
break;
case ENOENT:
zfs_error_aux(hdl,
dgettext(TEXT_DOMAIN, "no such pool or dataset"));
zfs_verror(hdl, EZFS_NOENT, fmt, ap);
break;
case EEXIST:
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"pool already exists"));
zfs_verror(hdl, EZFS_EXISTS, fmt, ap);
break;
case EBUSY:
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN, "pool is busy"));
zfs_verror(hdl, EZFS_BUSY, fmt, ap);
break;
/* There is no pending operation to cancel */
case ENOTACTIVE:
zfs_verror(hdl, EZFS_NO_PENDING, fmt, ap);
break;
case ENXIO:
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"one or more devices is currently unavailable"));
zfs_verror(hdl, EZFS_BADDEV, fmt, ap);
break;
case ENAMETOOLONG:
zfs_verror(hdl, EZFS_DEVOVERFLOW, fmt, ap);
break;
case ENOTSUP:
zfs_verror(hdl, EZFS_POOL_NOTSUP, fmt, ap);
break;
case EINVAL:
zfs_verror(hdl, EZFS_POOL_INVALARG, fmt, ap);
break;
case ENOSPC:
case EDQUOT:
zfs_verror(hdl, EZFS_NOSPC, fmt, ap);
return (-1);
case EAGAIN:
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"pool I/O is currently suspended"));
zfs_verror(hdl, EZFS_POOLUNAVAIL, fmt, ap);
break;
case EROFS:
zfs_verror(hdl, EZFS_POOLREADONLY, fmt, ap);
break;
case EDOM:
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"block size out of range or does not match"));
zfs_verror(hdl, EZFS_BADPROP, fmt, ap);
break;
case EREMOTEIO:
zfs_verror(hdl, EZFS_ACTIVE_POOL, fmt, ap);
break;
case ZFS_ERR_CHECKPOINT_EXISTS:
zfs_verror(hdl, EZFS_CHECKPOINT_EXISTS, fmt, ap);
break;
case ZFS_ERR_DISCARDING_CHECKPOINT:
zfs_verror(hdl, EZFS_DISCARDING_CHECKPOINT, fmt, ap);
break;
case ZFS_ERR_NO_CHECKPOINT:
zfs_verror(hdl, EZFS_NO_CHECKPOINT, fmt, ap);
break;
case ZFS_ERR_DEVRM_IN_PROGRESS:
zfs_verror(hdl, EZFS_DEVRM_IN_PROGRESS, fmt, ap);
break;
case ZFS_ERR_VDEV_TOO_BIG:
zfs_verror(hdl, EZFS_VDEV_TOO_BIG, fmt, ap);
break;
default:
zfs_error_aux(hdl, strerror(error));
zfs_verror(hdl, EZFS_UNKNOWN, fmt, ap);
}
va_end(ap);
return (-1);
}
/*
* Display an out of memory error message and abort the current program.
*/
int
no_memory(libzfs_handle_t *hdl)
{
return (zfs_error(hdl, EZFS_NOMEM, "internal error"));
}
/*
* A safe form of malloc() which will die if the allocation fails.
*/
void *
zfs_alloc(libzfs_handle_t *hdl, size_t size)
{
void *data;
if ((data = calloc(1, size)) == NULL)
(void) no_memory(hdl);
return (data);
}
/*
* A safe form of asprintf() which will die if the allocation fails.
*/
/*PRINTFLIKE2*/
char *
zfs_asprintf(libzfs_handle_t *hdl, const char *fmt, ...)
{
va_list ap;
char *ret;
int err;
va_start(ap, fmt);
err = vasprintf(&ret, fmt, ap);
va_end(ap);
if (err < 0)
(void) no_memory(hdl);
return (ret);
}
/*
* A safe form of realloc(), which also zeroes newly allocated space.
*/
void *
zfs_realloc(libzfs_handle_t *hdl, void *ptr, size_t oldsize, size_t newsize)
{
void *ret;
if ((ret = realloc(ptr, newsize)) == NULL) {
(void) no_memory(hdl);
return (NULL);
}
bzero((char *)ret + oldsize, (newsize - oldsize));
return (ret);
}
/*
* A safe form of strdup() which will die if the allocation fails.
*/
char *
zfs_strdup(libzfs_handle_t *hdl, const char *str)
{
char *ret;
if ((ret = strdup(str)) == NULL)
(void) no_memory(hdl);
return (ret);
}
/*
* Convert a number to an appropriately human-readable output.
*/
void
zfs_nicenum_format(uint64_t num, char *buf, size_t buflen,
enum zfs_nicenum_format format)
{
uint64_t n = num;
int index = 0;
const char *u;
const char *units[3][7] = {
[ZFS_NICENUM_1024] = {"", "K", "M", "G", "T", "P", "E"},
[ZFS_NICENUM_BYTES] = {"B", "K", "M", "G", "T", "P", "E"},
[ZFS_NICENUM_TIME] = {"ns", "us", "ms", "s", "?", "?", "?"}
};
const int units_len[] = {[ZFS_NICENUM_1024] = 6,
[ZFS_NICENUM_BYTES] = 6,
[ZFS_NICENUM_TIME] = 4};
const int k_unit[] = { [ZFS_NICENUM_1024] = 1024,
[ZFS_NICENUM_BYTES] = 1024,
[ZFS_NICENUM_TIME] = 1000};
double val;
if (format == ZFS_NICENUM_RAW) {
snprintf(buf, buflen, "%llu", (u_longlong_t)num);
return;
} else if (format == ZFS_NICENUM_RAWTIME && num > 0) {
snprintf(buf, buflen, "%llu", (u_longlong_t)num);
return;
} else if (format == ZFS_NICENUM_RAWTIME && num == 0) {
snprintf(buf, buflen, "%s", "-");
return;
}
while (n >= k_unit[format] && index < units_len[format]) {
n /= k_unit[format];
index++;
}
u = units[format][index];
/* Don't print zero latencies since they're invalid */
if ((format == ZFS_NICENUM_TIME) && (num == 0)) {
(void) snprintf(buf, buflen, "-");
} else if ((index == 0) || ((num %
(uint64_t)powl(k_unit[format], index)) == 0)) {
/*
* If this is an even multiple of the base, always display
* without any decimal precision.
*/
(void) snprintf(buf, buflen, "%llu%s", (u_longlong_t)n, u);
} else {
/*
* We want to choose a precision that reflects the best choice
* for fitting in 5 characters. This can get rather tricky when
* we have numbers that are very close to an order of magnitude.
* For example, when displaying 10239 (which is really 9.999K),
* we want only a single place of precision for 10.0K. We could
* develop some complex heuristics for this, but it's much
* easier just to try each combination in turn.
*/
int i;
for (i = 2; i >= 0; i--) {
val = (double)num /
(uint64_t)powl(k_unit[format], index);
/*
* Don't print floating point values for time. Note,
* we use floor() instead of round() here, since
* round can result in undesirable results. For
* example, if "num" is in the range of
* 999500-999999, it will print out "1000us". This
* doesn't happen if we use floor().
*/
if (format == ZFS_NICENUM_TIME) {
if (snprintf(buf, buflen, "%d%s",
(unsigned int) floor(val), u) <= 5)
break;
} else {
if (snprintf(buf, buflen, "%.*f%s", i,
val, u) <= 5)
break;
}
}
}
}
/*
* Convert a number to an appropriately human-readable output.
*/
void
zfs_nicenum(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_1024);
}
/*
* Convert a time to an appropriately human-readable output.
* @num: Time in nanoseconds
*/
void
zfs_nicetime(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_TIME);
}
/*
* Print out a raw number with correct column spacing
*/
void
zfs_niceraw(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_RAW);
}
/*
* Convert a number of bytes to an appropriately human-readable output.
*/
void
zfs_nicebytes(uint64_t num, char *buf, size_t buflen)
{
zfs_nicenum_format(num, buf, buflen, ZFS_NICENUM_BYTES);
}
void
libzfs_print_on_error(libzfs_handle_t *hdl, boolean_t printerr)
{
hdl->libzfs_printerr = printerr;
}
static int
libzfs_module_loaded(const char *module)
{
const char path_prefix[] = "/sys/module/";
char path[256];
memcpy(path, path_prefix, sizeof (path_prefix) - 1);
strcpy(path + sizeof (path_prefix) - 1, module);
return (access(path, F_OK) == 0);
}
/*
* Read lines from an open file descriptor and store them in an array of
* strings until EOF. lines[] will be allocated and populated with all the
* lines read. All newlines are replaced with NULL terminators for
* convenience. lines[] must be freed after use with libzfs_free_str_array().
*
* Returns the number of lines read.
*/
static int
libzfs_read_stdout_from_fd(int fd, char **lines[])
{
FILE *fp;
int lines_cnt = 0;
size_t len = 0;
char *line = NULL;
char **tmp_lines = NULL, **tmp;
char *nl = NULL;
int rc;
fp = fdopen(fd, "r");
if (fp == NULL)
return (0);
while (1) {
rc = getline(&line, &len, fp);
if (rc == -1)
break;
tmp = realloc(tmp_lines, sizeof (*tmp_lines) * (lines_cnt + 1));
if (tmp == NULL) {
/* Return the lines we were able to process */
break;
}
tmp_lines = tmp;
/* Terminate newlines */
if ((nl = strchr(line, '\n')) != NULL)
*nl = '\0';
tmp_lines[lines_cnt] = line;
lines_cnt++;
line = NULL;
}
fclose(fp);
*lines = tmp_lines;
return (lines_cnt);
}
static int
libzfs_run_process_impl(const char *path, char *argv[], char *env[], int flags,
char **lines[], int *lines_cnt)
{
pid_t pid;
int error, devnull_fd;
int link[2];
/*
* Setup a pipe between our child and parent process if we're
* reading stdout.
*/
if ((lines != NULL) && pipe(link) == -1)
return (-ESTRPIPE);
pid = vfork();
if (pid == 0) {
/* Child process */
devnull_fd = open("/dev/null", O_WRONLY);
if (devnull_fd < 0)
_exit(-1);
if (!(flags & STDOUT_VERBOSE) && (lines == NULL))
(void) dup2(devnull_fd, STDOUT_FILENO);
else if (lines != NULL) {
/* Save the output to lines[] */
dup2(link[1], STDOUT_FILENO);
close(link[0]);
close(link[1]);
}
if (!(flags & STDERR_VERBOSE))
(void) dup2(devnull_fd, STDERR_FILENO);
close(devnull_fd);
if (flags & NO_DEFAULT_PATH) {
if (env == NULL)
execv(path, argv);
else
execve(path, argv, env);
} else {
if (env == NULL)
execvp(path, argv);
else
execvpe(path, argv, env);
}
_exit(-1);
} else if (pid > 0) {
/* Parent process */
int status;
while ((error = waitpid(pid, &status, 0)) == -1 &&
errno == EINTR) { }
if (error < 0 || !WIFEXITED(status))
return (-1);
if (lines != NULL) {
close(link[1]);
*lines_cnt = libzfs_read_stdout_from_fd(link[0], lines);
}
return (WEXITSTATUS(status));
}
return (-1);
}
int
libzfs_run_process(const char *path, char *argv[], int flags)
{
return (libzfs_run_process_impl(path, argv, NULL, flags, NULL, NULL));
}
/*
* Run a command and store its stdout lines in an array of strings (lines[]).
* lines[] is allocated and populated for you, and the number of lines is set in
* lines_cnt. lines[] must be freed after use with libzfs_free_str_array().
* All newlines (\n) in lines[] are terminated for convenience.
*/
int
libzfs_run_process_get_stdout(const char *path, char *argv[], char *env[],
char **lines[], int *lines_cnt)
{
return (libzfs_run_process_impl(path, argv, env, 0, lines, lines_cnt));
}
/*
* Same as libzfs_run_process_get_stdout(), but run without $PATH set. This
* means that *path needs to be the full path to the executable.
*/
int
libzfs_run_process_get_stdout_nopath(const char *path, char *argv[],
char *env[], char **lines[], int *lines_cnt)
{
return (libzfs_run_process_impl(path, argv, env, NO_DEFAULT_PATH,
lines, lines_cnt));
}
/*
* Free an array of strings. Free both the strings contained in the array and
* the array itself.
*/
void
libzfs_free_str_array(char **strs, int count)
{
while (--count >= 0)
free(strs[count]);
free(strs);
}
/*
* Returns 1 if environment variable is set to "YES", "yes", "ON", "on", or
* a non-zero number.
*
* Returns 0 otherwise.
*/
int
libzfs_envvar_is_set(char *envvar)
{
char *env = getenv(envvar);
if (env && (strtoul(env, NULL, 0) > 0 ||
(!strncasecmp(env, "YES", 3) && strnlen(env, 4) == 3) ||
(!strncasecmp(env, "ON", 2) && strnlen(env, 3) == 2)))
return (1);
return (0);
}
/*
* Verify the required ZFS_DEV device is available and optionally attempt
* to load the ZFS modules. Under normal circumstances the modules
* should already have been loaded by some external mechanism.
*
* Environment variables:
* - ZFS_MODULE_LOADING="YES|yes|ON|on" - Attempt to load modules.
* - ZFS_MODULE_TIMEOUT="<seconds>" - Seconds to wait for ZFS_DEV
*/
static int
libzfs_load_module(const char *module)
{
char *argv[4] = {"/sbin/modprobe", "-q", (char *)module, (char *)0};
char *load_str, *timeout_str;
long timeout = 10; /* seconds */
long busy_timeout = 10; /* milliseconds */
int load = 0, fd;
hrtime_t start;
/* Optionally request module loading */
if (!libzfs_module_loaded(module)) {
load_str = getenv("ZFS_MODULE_LOADING");
if (load_str) {
if (!strncasecmp(load_str, "YES", strlen("YES")) ||
!strncasecmp(load_str, "ON", strlen("ON")))
load = 1;
else
load = 0;
}
if (load) {
if (libzfs_run_process("/sbin/modprobe", argv, 0))
return (ENOEXEC);
if (!libzfs_module_loaded(module))
return (ENXIO);
}
}
/*
* Device creation by udev is asynchronous and waiting may be
* required. Busy wait for 10ms and then fall back to polling every
* 10ms for the allowed timeout (default 10s, max 10m). This is
* done to optimize for the common case where the device is
* immediately available and to avoid penalizing the possible
* case where udev is slow or unable to create the device.
*/
timeout_str = getenv("ZFS_MODULE_TIMEOUT");
if (timeout_str) {
timeout = strtol(timeout_str, NULL, 0);
timeout = MAX(MIN(timeout, (10 * 60)), 0); /* 0 <= N <= 600 */
}
start = gethrtime();
do {
fd = open(ZFS_DEV, O_RDWR);
if (fd >= 0) {
(void) close(fd);
return (0);
} else if (errno != ENOENT) {
return (errno);
} else if (NSEC2MSEC(gethrtime() - start) < busy_timeout) {
sched_yield();
} else {
usleep(10 * MILLISEC);
}
} while (NSEC2MSEC(gethrtime() - start) < (timeout * MILLISEC));
return (ENOENT);
}
libzfs_handle_t *
libzfs_init(void)
{
libzfs_handle_t *hdl;
int error;
error = libzfs_load_module(ZFS_DRIVER);
if (error) {
errno = error;
return (NULL);
}
if ((hdl = calloc(1, sizeof (libzfs_handle_t))) == NULL) {
return (NULL);
}
if ((hdl->libzfs_fd = open(ZFS_DEV, O_RDWR)) < 0) {
free(hdl);
return (NULL);
}
#ifdef HAVE_SETMNTENT
if ((hdl->libzfs_mnttab = setmntent(MNTTAB, "r")) == NULL) {
#else
if ((hdl->libzfs_mnttab = fopen(MNTTAB, "r")) == NULL) {
#endif
(void) close(hdl->libzfs_fd);
free(hdl);
return (NULL);
}
hdl->libzfs_sharetab = fopen(ZFS_SHARETAB, "r");
if (libzfs_core_init() != 0) {
(void) close(hdl->libzfs_fd);
(void) fclose(hdl->libzfs_mnttab);
if (hdl->libzfs_sharetab)
(void) fclose(hdl->libzfs_sharetab);
free(hdl);
return (NULL);
}
zfs_prop_init();
zpool_prop_init();
zpool_feature_init();
libzfs_mnttab_init(hdl);
fletcher_4_init();
if (getenv("ZFS_PROP_DEBUG") != NULL) {
hdl->libzfs_prop_debug = B_TRUE;
}
return (hdl);
}
void
libzfs_fini(libzfs_handle_t *hdl)
{
(void) close(hdl->libzfs_fd);
if (hdl->libzfs_mnttab)
#ifdef HAVE_SETMNTENT
(void) endmntent(hdl->libzfs_mnttab);
#else
(void) fclose(hdl->libzfs_mnttab);
#endif
if (hdl->libzfs_sharetab)
(void) fclose(hdl->libzfs_sharetab);
zfs_uninit_libshare(hdl);
zpool_free_handles(hdl);
namespace_clear(hdl);
libzfs_mnttab_fini(hdl);
libzfs_core_fini();
fletcher_4_fini();
free(hdl);
}
libzfs_handle_t *
zpool_get_handle(zpool_handle_t *zhp)
{
return (zhp->zpool_hdl);
}
libzfs_handle_t *
zfs_get_handle(zfs_handle_t *zhp)
{
return (zhp->zfs_hdl);
}
zpool_handle_t *
zfs_get_pool_handle(const zfs_handle_t *zhp)
{
return (zhp->zpool_hdl);
}
/*
* Given a name, determine whether or not it's a valid path
* (starts with '/' or "./"). If so, walk the mnttab trying
* to match the device number. If not, treat the path as an
* fs/vol/snap/bkmark name.
*/
zfs_handle_t *
zfs_path_to_zhandle(libzfs_handle_t *hdl, char *path, zfs_type_t argtype)
{
struct stat64 statbuf;
struct extmnttab entry;
int ret;
if (path[0] != '/' && strncmp(path, "./", strlen("./")) != 0) {
/*
* It's not a valid path, assume it's a name of type 'argtype'.
*/
return (zfs_open(hdl, path, argtype));
}
if (stat64(path, &statbuf) != 0) {
(void) fprintf(stderr, "%s: %s\n", path, strerror(errno));
return (NULL);
}
/* Reopen MNTTAB to prevent reading stale data from open file */
if (freopen(MNTTAB, "r", hdl->libzfs_mnttab) == NULL)
return (NULL);
while ((ret = getextmntent(hdl->libzfs_mnttab, &entry, 0)) == 0) {
if (makedevice(entry.mnt_major, entry.mnt_minor) ==
statbuf.st_dev) {
break;
}
}
if (ret != 0) {
return (NULL);
}
if (strcmp(entry.mnt_fstype, MNTTYPE_ZFS) != 0) {
(void) fprintf(stderr, gettext("'%s': not a ZFS filesystem\n"),
path);
return (NULL);
}
return (zfs_open(hdl, entry.mnt_special, ZFS_TYPE_FILESYSTEM));
}
/*
* Append partition suffix to an otherwise fully qualified device path.
* This is used to generate the name the full path as its stored in
* ZPOOL_CONFIG_PATH for whole disk devices. On success the new length
* of 'path' will be returned on error a negative value is returned.
*/
int
zfs_append_partition(char *path, size_t max_len)
{
int len = strlen(path);
if ((strncmp(path, UDISK_ROOT, strlen(UDISK_ROOT)) == 0) ||
(strncmp(path, ZVOL_ROOT, strlen(ZVOL_ROOT)) == 0)) {
if (len + 6 >= max_len)
return (-1);
(void) strcat(path, "-part1");
len += 6;
} else {
if (len + 2 >= max_len)
return (-1);
if (isdigit(path[len-1])) {
(void) strcat(path, "p1");
len += 2;
} else {
(void) strcat(path, "1");
len += 1;
}
}
return (len);
}
/*
* Given a shorthand device name check if a file by that name exists in any
* of the 'zpool_default_import_path' or ZPOOL_IMPORT_PATH directories. If
* one is found, store its fully qualified path in the 'path' buffer passed
* by the caller and return 0, otherwise return an error.
*/
int
zfs_resolve_shortname(const char *name, char *path, size_t len)
{
int i, error = -1;
char *dir, *env, *envdup;
env = getenv("ZPOOL_IMPORT_PATH");
errno = ENOENT;
if (env) {
envdup = strdup(env);
dir = strtok(envdup, ":");
while (dir && error) {
(void) snprintf(path, len, "%s/%s", dir, name);
error = access(path, F_OK);
dir = strtok(NULL, ":");
}
free(envdup);
} else {
for (i = 0; i < DEFAULT_IMPORT_PATH_SIZE && error < 0; i++) {
(void) snprintf(path, len, "%s/%s",
zpool_default_import_path[i], name);
error = access(path, F_OK);
}
}
return (error ? ENOENT : 0);
}
/*
* Given a shorthand device name look for a match against 'cmp_name'. This
* is done by checking all prefix expansions using either the default
* 'zpool_default_import_paths' or the ZPOOL_IMPORT_PATH environment
* variable. Proper partition suffixes will be appended if this is a
* whole disk. When a match is found 0 is returned otherwise ENOENT.
*/
static int
zfs_strcmp_shortname(char *name, char *cmp_name, int wholedisk)
{
int path_len, cmp_len, i = 0, error = ENOENT;
char *dir, *env, *envdup = NULL;
char path_name[MAXPATHLEN];
cmp_len = strlen(cmp_name);
env = getenv("ZPOOL_IMPORT_PATH");
if (env) {
envdup = strdup(env);
dir = strtok(envdup, ":");
} else {
dir = zpool_default_import_path[i];
}
while (dir) {
/* Trim trailing directory slashes from ZPOOL_IMPORT_PATH */
while (dir[strlen(dir)-1] == '/')
dir[strlen(dir)-1] = '\0';
path_len = snprintf(path_name, MAXPATHLEN, "%s/%s", dir, name);
if (wholedisk)
path_len = zfs_append_partition(path_name, MAXPATHLEN);
if ((path_len == cmp_len) && strcmp(path_name, cmp_name) == 0) {
error = 0;
break;
}
if (env) {
dir = strtok(NULL, ":");
} else if (++i < DEFAULT_IMPORT_PATH_SIZE) {
dir = zpool_default_import_path[i];
} else {
dir = NULL;
}
}
if (env)
free(envdup);
return (error);
}
/*
* Given either a shorthand or fully qualified path name look for a match
* against 'cmp'. The passed name will be expanded as needed for comparison
* purposes and redundant slashes stripped to ensure an accurate match.
*/
int
zfs_strcmp_pathname(char *name, char *cmp, int wholedisk)
{
int path_len, cmp_len;
char path_name[MAXPATHLEN];
char cmp_name[MAXPATHLEN];
char *dir, *dup;
/* Strip redundant slashes if one exists due to ZPOOL_IMPORT_PATH */
memset(cmp_name, 0, MAXPATHLEN);
dup = strdup(cmp);
dir = strtok(dup, "/");
while (dir) {
strlcat(cmp_name, "/", sizeof (cmp_name));
strlcat(cmp_name, dir, sizeof (cmp_name));
dir = strtok(NULL, "/");
}
free(dup);
if (name[0] != '/')
return (zfs_strcmp_shortname(name, cmp_name, wholedisk));
(void) strlcpy(path_name, name, MAXPATHLEN);
path_len = strlen(path_name);
cmp_len = strlen(cmp_name);
if (wholedisk) {
path_len = zfs_append_partition(path_name, MAXPATHLEN);
if (path_len == -1)
return (ENOMEM);
}
if ((path_len != cmp_len) || strcmp(path_name, cmp_name))
return (ENOENT);
return (0);
}
/*
* Given a full path to a device determine if that device appears in the
* import search path. If it does return the first match and store the
* index in the passed 'order' variable, otherwise return an error.
*/
int
zfs_path_order(char *name, int *order)
{
int i = 0, error = ENOENT;
char *dir, *env, *envdup;
env = getenv("ZPOOL_IMPORT_PATH");
if (env) {
envdup = strdup(env);
dir = strtok(envdup, ":");
while (dir) {
if (strncmp(name, dir, strlen(dir)) == 0) {
*order = i;
error = 0;
break;
}
dir = strtok(NULL, ":");
i++;
}
free(envdup);
} else {
for (i = 0; i < DEFAULT_IMPORT_PATH_SIZE; i++) {
if (strncmp(name, zpool_default_import_path[i],
strlen(zpool_default_import_path[i])) == 0) {
*order = i;
error = 0;
break;
}
}
}
return (error);
}
/*
* Initialize the zc_nvlist_dst member to prepare for receiving an nvlist from
* an ioctl().
*/
int
zcmd_alloc_dst_nvlist(libzfs_handle_t *hdl, zfs_cmd_t *zc, size_t len)
{
if (len == 0)
len = 16 * 1024;
zc->zc_nvlist_dst_size = len;
zc->zc_nvlist_dst =
(uint64_t)(uintptr_t)zfs_alloc(hdl, zc->zc_nvlist_dst_size);
if (zc->zc_nvlist_dst == 0)
return (-1);
return (0);
}
/*
* Called when an ioctl() which returns an nvlist fails with ENOMEM. This will
* expand the nvlist to the size specified in 'zc_nvlist_dst_size', which was
* filled in by the kernel to indicate the actual required size.
*/
int
zcmd_expand_dst_nvlist(libzfs_handle_t *hdl, zfs_cmd_t *zc)
{
free((void *)(uintptr_t)zc->zc_nvlist_dst);
zc->zc_nvlist_dst =
(uint64_t)(uintptr_t)zfs_alloc(hdl, zc->zc_nvlist_dst_size);
if (zc->zc_nvlist_dst == 0)
return (-1);
return (0);
}
/*
* Called to free the src and dst nvlists stored in the command structure.
*/
void
zcmd_free_nvlists(zfs_cmd_t *zc)
{
free((void *)(uintptr_t)zc->zc_nvlist_conf);
free((void *)(uintptr_t)zc->zc_nvlist_src);
free((void *)(uintptr_t)zc->zc_nvlist_dst);
zc->zc_nvlist_conf = 0;
zc->zc_nvlist_src = 0;
zc->zc_nvlist_dst = 0;
}
static int
zcmd_write_nvlist_com(libzfs_handle_t *hdl, uint64_t *outnv, uint64_t *outlen,
nvlist_t *nvl)
{
char *packed;
size_t len;
verify(nvlist_size(nvl, &len, NV_ENCODE_NATIVE) == 0);
if ((packed = zfs_alloc(hdl, len)) == NULL)
return (-1);
verify(nvlist_pack(nvl, &packed, &len, NV_ENCODE_NATIVE, 0) == 0);
*outnv = (uint64_t)(uintptr_t)packed;
*outlen = len;
return (0);
}
int
zcmd_write_conf_nvlist(libzfs_handle_t *hdl, zfs_cmd_t *zc, nvlist_t *nvl)
{
return (zcmd_write_nvlist_com(hdl, &zc->zc_nvlist_conf,
&zc->zc_nvlist_conf_size, nvl));
}
int
zcmd_write_src_nvlist(libzfs_handle_t *hdl, zfs_cmd_t *zc, nvlist_t *nvl)
{
return (zcmd_write_nvlist_com(hdl, &zc->zc_nvlist_src,
&zc->zc_nvlist_src_size, nvl));
}
/*
* Unpacks an nvlist from the ZFS ioctl command structure.
*/
int
zcmd_read_dst_nvlist(libzfs_handle_t *hdl, zfs_cmd_t *zc, nvlist_t **nvlp)
{
if (nvlist_unpack((void *)(uintptr_t)zc->zc_nvlist_dst,
zc->zc_nvlist_dst_size, nvlp, 0) != 0)
return (no_memory(hdl));
return (0);
}
int
zfs_ioctl(libzfs_handle_t *hdl, int request, zfs_cmd_t *zc)
{
return (ioctl(hdl->libzfs_fd, request, zc));
}
/*
* ================================================================
* API shared by zfs and zpool property management
* ================================================================
*/
static void
zprop_print_headers(zprop_get_cbdata_t *cbp, zfs_type_t type)
{
zprop_list_t *pl = cbp->cb_proplist;
int i;
char *title;
size_t len;
cbp->cb_first = B_FALSE;
if (cbp->cb_scripted)
return;
/*
* Start with the length of the column headers.
*/
cbp->cb_colwidths[GET_COL_NAME] = strlen(dgettext(TEXT_DOMAIN, "NAME"));
cbp->cb_colwidths[GET_COL_PROPERTY] = strlen(dgettext(TEXT_DOMAIN,
"PROPERTY"));
cbp->cb_colwidths[GET_COL_VALUE] = strlen(dgettext(TEXT_DOMAIN,
"VALUE"));
cbp->cb_colwidths[GET_COL_RECVD] = strlen(dgettext(TEXT_DOMAIN,
"RECEIVED"));
cbp->cb_colwidths[GET_COL_SOURCE] = strlen(dgettext(TEXT_DOMAIN,
"SOURCE"));
/* first property is always NAME */
assert(cbp->cb_proplist->pl_prop ==
((type == ZFS_TYPE_POOL) ? ZPOOL_PROP_NAME : ZFS_PROP_NAME));
/*
* Go through and calculate the widths for each column. For the
* 'source' column, we kludge it up by taking the worst-case scenario of
* inheriting from the longest name. This is acceptable because in the
* majority of cases 'SOURCE' is the last column displayed, and we don't
* use the width anyway. Note that the 'VALUE' column can be oversized,
* if the name of the property is much longer than any values we find.
*/
for (pl = cbp->cb_proplist; pl != NULL; pl = pl->pl_next) {
/*
* 'PROPERTY' column
*/
if (pl->pl_prop != ZPROP_INVAL) {
const char *propname = (type == ZFS_TYPE_POOL) ?
zpool_prop_to_name(pl->pl_prop) :
zfs_prop_to_name(pl->pl_prop);
len = strlen(propname);
if (len > cbp->cb_colwidths[GET_COL_PROPERTY])
cbp->cb_colwidths[GET_COL_PROPERTY] = len;
} else {
len = strlen(pl->pl_user_prop);
if (len > cbp->cb_colwidths[GET_COL_PROPERTY])
cbp->cb_colwidths[GET_COL_PROPERTY] = len;
}
/*
* 'VALUE' column. The first property is always the 'name'
* property that was tacked on either by /sbin/zfs's
* zfs_do_get() or when calling zprop_expand_list(), so we
* ignore its width. If the user specified the name property
* to display, then it will be later in the list in any case.
*/
if (pl != cbp->cb_proplist &&
pl->pl_width > cbp->cb_colwidths[GET_COL_VALUE])
cbp->cb_colwidths[GET_COL_VALUE] = pl->pl_width;
/* 'RECEIVED' column. */
if (pl != cbp->cb_proplist &&
pl->pl_recvd_width > cbp->cb_colwidths[GET_COL_RECVD])
cbp->cb_colwidths[GET_COL_RECVD] = pl->pl_recvd_width;
/*
* 'NAME' and 'SOURCE' columns
*/
if (pl->pl_prop == (type == ZFS_TYPE_POOL ? ZPOOL_PROP_NAME :
ZFS_PROP_NAME) &&
pl->pl_width > cbp->cb_colwidths[GET_COL_NAME]) {
cbp->cb_colwidths[GET_COL_NAME] = pl->pl_width;
cbp->cb_colwidths[GET_COL_SOURCE] = pl->pl_width +
strlen(dgettext(TEXT_DOMAIN, "inherited from"));
}
}
/*
* Now go through and print the headers.
*/
for (i = 0; i < ZFS_GET_NCOLS; i++) {
switch (cbp->cb_columns[i]) {
case GET_COL_NAME:
title = dgettext(TEXT_DOMAIN, "NAME");
break;
case GET_COL_PROPERTY:
title = dgettext(TEXT_DOMAIN, "PROPERTY");
break;
case GET_COL_VALUE:
title = dgettext(TEXT_DOMAIN, "VALUE");
break;
case GET_COL_RECVD:
title = dgettext(TEXT_DOMAIN, "RECEIVED");
break;
case GET_COL_SOURCE:
title = dgettext(TEXT_DOMAIN, "SOURCE");
break;
default:
title = NULL;
}
if (title != NULL) {
if (i == (ZFS_GET_NCOLS - 1) ||
cbp->cb_columns[i + 1] == GET_COL_NONE)
(void) printf("%s", title);
else
(void) printf("%-*s ",
cbp->cb_colwidths[cbp->cb_columns[i]],
title);
}
}
(void) printf("\n");
}
/*
* Display a single line of output, according to the settings in the callback
* structure.
*/
void
zprop_print_one_property(const char *name, zprop_get_cbdata_t *cbp,
const char *propname, const char *value, zprop_source_t sourcetype,
const char *source, const char *recvd_value)
{
int i;
const char *str = NULL;
char buf[128];
/*
* Ignore those source types that the user has chosen to ignore.
*/
if ((sourcetype & cbp->cb_sources) == 0)
return;
if (cbp->cb_first)
zprop_print_headers(cbp, cbp->cb_type);
for (i = 0; i < ZFS_GET_NCOLS; i++) {
switch (cbp->cb_columns[i]) {
case GET_COL_NAME:
str = name;
break;
case GET_COL_PROPERTY:
str = propname;
break;
case GET_COL_VALUE:
str = value;
break;
case GET_COL_SOURCE:
switch (sourcetype) {
case ZPROP_SRC_NONE:
str = "-";
break;
case ZPROP_SRC_DEFAULT:
str = "default";
break;
case ZPROP_SRC_LOCAL:
str = "local";
break;
case ZPROP_SRC_TEMPORARY:
str = "temporary";
break;
case ZPROP_SRC_INHERITED:
(void) snprintf(buf, sizeof (buf),
"inherited from %s", source);
str = buf;
break;
case ZPROP_SRC_RECEIVED:
str = "received";
break;
default:
str = NULL;
assert(!"unhandled zprop_source_t");
}
break;
case GET_COL_RECVD:
str = (recvd_value == NULL ? "-" : recvd_value);
break;
default:
continue;
}
if (i == (ZFS_GET_NCOLS - 1) ||
cbp->cb_columns[i + 1] == GET_COL_NONE)
(void) printf("%s", str);
else if (cbp->cb_scripted)
(void) printf("%s\t", str);
else
(void) printf("%-*s ",
cbp->cb_colwidths[cbp->cb_columns[i]],
str);
}
(void) printf("\n");
}
/*
* Given a numeric suffix, convert the value into a number of bits that the
* resulting value must be shifted.
*/
static int
str2shift(libzfs_handle_t *hdl, const char *buf)
{
const char *ends = "BKMGTPEZ";
int i;
if (buf[0] == '\0')
return (0);
for (i = 0; i < strlen(ends); i++) {
if (toupper(buf[0]) == ends[i])
break;
}
if (i == strlen(ends)) {
if (hdl)
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"invalid numeric suffix '%s'"), buf);
return (-1);
}
/*
* Allow 'G' = 'GB' = 'GiB', case-insensitively.
* However, 'BB' and 'BiB' are disallowed.
*/
if (buf[1] == '\0' ||
(toupper(buf[0]) != 'B' &&
((toupper(buf[1]) == 'B' && buf[2] == '\0') ||
(toupper(buf[1]) == 'I' && toupper(buf[2]) == 'B' &&
buf[3] == '\0'))))
return (10 * i);
if (hdl)
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"invalid numeric suffix '%s'"), buf);
return (-1);
}
/*
* Convert a string of the form '100G' into a real number. Used when setting
* properties or creating a volume. 'buf' is used to place an extended error
* message for the caller to use.
*/
int
zfs_nicestrtonum(libzfs_handle_t *hdl, const char *value, uint64_t *num)
{
char *end;
int shift;
*num = 0;
/* Check to see if this looks like a number. */
if ((value[0] < '0' || value[0] > '9') && value[0] != '.') {
if (hdl)
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"bad numeric value '%s'"), value);
return (-1);
}
/* Rely on strtoull() to process the numeric portion. */
errno = 0;
*num = strtoull(value, &end, 10);
/*
* Check for ERANGE, which indicates that the value is too large to fit
* in a 64-bit value.
*/
if (errno == ERANGE) {
if (hdl)
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"numeric value is too large"));
return (-1);
}
/*
* If we have a decimal value, then do the computation with floating
* point arithmetic. Otherwise, use standard arithmetic.
*/
if (*end == '.') {
double fval = strtod(value, &end);
if ((shift = str2shift(hdl, end)) == -1)
return (-1);
fval *= pow(2, shift);
if (fval > UINT64_MAX) {
if (hdl)
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"numeric value is too large"));
return (-1);
}
*num = (uint64_t)fval;
} else {
if ((shift = str2shift(hdl, end)) == -1)
return (-1);
/* Check for overflow */
if (shift >= 64 || (*num << shift) >> shift != *num) {
if (hdl)
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"numeric value is too large"));
return (-1);
}
*num <<= shift;
}
return (0);
}
/*
* Given a propname=value nvpair to set, parse any numeric properties
* (index, boolean, etc) if they are specified as strings and add the
* resulting nvpair to the returned nvlist.
*
* At the DSL layer, all properties are either 64-bit numbers or strings.
* We want the user to be able to ignore this fact and specify properties
* as native values (numbers, for example) or as strings (to simplify
* command line utilities). This also handles converting index types
* (compression, checksum, etc) from strings to their on-disk index.
*/
int
zprop_parse_value(libzfs_handle_t *hdl, nvpair_t *elem, int prop,
zfs_type_t type, nvlist_t *ret, char **svalp, uint64_t *ivalp,
const char *errbuf)
{
data_type_t datatype = nvpair_type(elem);
zprop_type_t proptype;
const char *propname;
char *value;
boolean_t isnone = B_FALSE;
boolean_t isauto = B_FALSE;
int err = 0;
if (type == ZFS_TYPE_POOL) {
proptype = zpool_prop_get_type(prop);
propname = zpool_prop_to_name(prop);
} else {
proptype = zfs_prop_get_type(prop);
propname = zfs_prop_to_name(prop);
}
/*
* Convert any properties to the internal DSL value types.
*/
*svalp = NULL;
*ivalp = 0;
switch (proptype) {
case PROP_TYPE_STRING:
if (datatype != DATA_TYPE_STRING) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"'%s' must be a string"), nvpair_name(elem));
goto error;
}
err = nvpair_value_string(elem, svalp);
if (err != 0) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"'%s' is invalid"), nvpair_name(elem));
goto error;
}
if (strlen(*svalp) >= ZFS_MAXPROPLEN) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"'%s' is too long"), nvpair_name(elem));
goto error;
}
break;
case PROP_TYPE_NUMBER:
if (datatype == DATA_TYPE_STRING) {
(void) nvpair_value_string(elem, &value);
if (strcmp(value, "none") == 0) {
isnone = B_TRUE;
} else if (strcmp(value, "auto") == 0) {
isauto = B_TRUE;
} else if (zfs_nicestrtonum(hdl, value, ivalp) != 0) {
goto error;
}
} else if (datatype == DATA_TYPE_UINT64) {
(void) nvpair_value_uint64(elem, ivalp);
} else {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"'%s' must be a number"), nvpair_name(elem));
goto error;
}
/*
* Quota special: force 'none' and don't allow 0.
*/
if ((type & ZFS_TYPE_DATASET) && *ivalp == 0 && !isnone &&
(prop == ZFS_PROP_QUOTA || prop == ZFS_PROP_REFQUOTA)) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"use 'none' to disable quota/refquota"));
goto error;
}
/*
* Special handling for "*_limit=none". In this case it's not
* 0 but UINT64_MAX.
*/
if ((type & ZFS_TYPE_DATASET) && isnone &&
(prop == ZFS_PROP_FILESYSTEM_LIMIT ||
prop == ZFS_PROP_SNAPSHOT_LIMIT)) {
*ivalp = UINT64_MAX;
}
/*
* Special handling for setting 'refreservation' to 'auto'. Use
* UINT64_MAX to tell the caller to use zfs_fix_auto_resv().
* 'auto' is only allowed on volumes.
*/
if (isauto) {
switch (prop) {
case ZFS_PROP_REFRESERVATION:
if ((type & ZFS_TYPE_VOLUME) == 0) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"'%s=auto' only allowed on "
"volumes"), nvpair_name(elem));
goto error;
}
*ivalp = UINT64_MAX;
break;
default:
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"'auto' is invalid value for '%s'"),
nvpair_name(elem));
goto error;
}
}
break;
case PROP_TYPE_INDEX:
if (datatype != DATA_TYPE_STRING) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"'%s' must be a string"), nvpair_name(elem));
goto error;
}
(void) nvpair_value_string(elem, &value);
if (zprop_string_to_index(prop, value, ivalp, type) != 0) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"'%s' must be one of '%s'"), propname,
zprop_values(prop, type));
goto error;
}
break;
default:
abort();
}
/*
* Add the result to our return set of properties.
*/
if (*svalp != NULL) {
if (nvlist_add_string(ret, propname, *svalp) != 0) {
(void) no_memory(hdl);
return (-1);
}
} else {
if (nvlist_add_uint64(ret, propname, *ivalp) != 0) {
(void) no_memory(hdl);
return (-1);
}
}
return (0);
error:
(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
return (-1);
}
static int
addlist(libzfs_handle_t *hdl, char *propname, zprop_list_t **listp,
zfs_type_t type)
{
int prop;
zprop_list_t *entry;
prop = zprop_name_to_prop(propname, type);
if (prop != ZPROP_INVAL && !zprop_valid_for_type(prop, type, B_FALSE))
prop = ZPROP_INVAL;
/*
* When no property table entry can be found, return failure if
* this is a pool property or if this isn't a user-defined
* dataset property,
*/
if (prop == ZPROP_INVAL && ((type == ZFS_TYPE_POOL &&
!zpool_prop_feature(propname) &&
!zpool_prop_unsupported(propname)) ||
(type == ZFS_TYPE_DATASET && !zfs_prop_user(propname) &&
!zfs_prop_userquota(propname) && !zfs_prop_written(propname)))) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"invalid property '%s'"), propname);
return (zfs_error(hdl, EZFS_BADPROP,
dgettext(TEXT_DOMAIN, "bad property list")));
}
if ((entry = zfs_alloc(hdl, sizeof (zprop_list_t))) == NULL)
return (-1);
entry->pl_prop = prop;
if (prop == ZPROP_INVAL) {
if ((entry->pl_user_prop = zfs_strdup(hdl, propname)) ==
NULL) {
free(entry);
return (-1);
}
entry->pl_width = strlen(propname);
} else {
entry->pl_width = zprop_width(prop, &entry->pl_fixed,
type);
}
*listp = entry;
return (0);
}
/*
* Given a comma-separated list of properties, construct a property list
* containing both user-defined and native properties. This function will
* return a NULL list if 'all' is specified, which can later be expanded
* by zprop_expand_list().
*/
int
zprop_get_list(libzfs_handle_t *hdl, char *props, zprop_list_t **listp,
zfs_type_t type)
{
*listp = NULL;
/*
* If 'all' is specified, return a NULL list.
*/
if (strcmp(props, "all") == 0)
return (0);
/*
* If no props were specified, return an error.
*/
if (props[0] == '\0') {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"no properties specified"));
return (zfs_error(hdl, EZFS_BADPROP, dgettext(TEXT_DOMAIN,
"bad property list")));
}
/*
* It would be nice to use getsubopt() here, but the inclusion of column
* aliases makes this more effort than it's worth.
*/
while (*props != '\0') {
size_t len;
char *p;
char c;
if ((p = strchr(props, ',')) == NULL) {
len = strlen(props);
p = props + len;
} else {
len = p - props;
}
/*
* Check for empty options.
*/
if (len == 0) {
zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
"empty property name"));
return (zfs_error(hdl, EZFS_BADPROP,
dgettext(TEXT_DOMAIN, "bad property list")));
}
/*
* Check all regular property names.
*/
c = props[len];
props[len] = '\0';
if (strcmp(props, "space") == 0) {
static char *spaceprops[] = {
"name", "avail", "used", "usedbysnapshots",
"usedbydataset", "usedbyrefreservation",
"usedbychildren", NULL
};
int i;
for (i = 0; spaceprops[i]; i++) {
if (addlist(hdl, spaceprops[i], listp, type))
return (-1);
listp = &(*listp)->pl_next;
}
} else {
if (addlist(hdl, props, listp, type))
return (-1);
listp = &(*listp)->pl_next;
}
props = p;
if (c == ',')
props++;
}
return (0);
}
void
zprop_free_list(zprop_list_t *pl)
{
zprop_list_t *next;
while (pl != NULL) {
next = pl->pl_next;
free(pl->pl_user_prop);
free(pl);
pl = next;
}
}
typedef struct expand_data {
zprop_list_t **last;
libzfs_handle_t *hdl;
zfs_type_t type;
} expand_data_t;
int
zprop_expand_list_cb(int prop, void *cb)
{
zprop_list_t *entry;
expand_data_t *edp = cb;
if ((entry = zfs_alloc(edp->hdl, sizeof (zprop_list_t))) == NULL)
return (ZPROP_INVAL);
entry->pl_prop = prop;
entry->pl_width = zprop_width(prop, &entry->pl_fixed, edp->type);
entry->pl_all = B_TRUE;
*(edp->last) = entry;
edp->last = &entry->pl_next;
return (ZPROP_CONT);
}
int
zprop_expand_list(libzfs_handle_t *hdl, zprop_list_t **plp, zfs_type_t type)
{
zprop_list_t *entry;
zprop_list_t **last;
expand_data_t exp;
if (*plp == NULL) {
/*
* If this is the very first time we've been called for an 'all'
* specification, expand the list to include all native
* properties.
*/
last = plp;
exp.last = last;
exp.hdl = hdl;
exp.type = type;
if (zprop_iter_common(zprop_expand_list_cb, &exp, B_FALSE,
B_FALSE, type) == ZPROP_INVAL)
return (-1);
/*
* Add 'name' to the beginning of the list, which is handled
* specially.
*/
if ((entry = zfs_alloc(hdl, sizeof (zprop_list_t))) == NULL)
return (-1);
entry->pl_prop = (type == ZFS_TYPE_POOL) ? ZPOOL_PROP_NAME :
ZFS_PROP_NAME;
entry->pl_width = zprop_width(entry->pl_prop,
&entry->pl_fixed, type);
entry->pl_all = B_TRUE;
entry->pl_next = *plp;
*plp = entry;
}
return (0);
}
int
zprop_iter(zprop_func func, void *cb, boolean_t show_all, boolean_t ordered,
zfs_type_t type)
{
return (zprop_iter_common(func, cb, show_all, ordered, type));
}