Date: Tue, 14 May 2013 23:33:08 +0200 From: Henner Heck <Henner.Heck@web.de> To: Steven Hartland <killing@multiplay.co.uk> Cc: freebsd-fs@freebsd.org Subject: Re: Long delays during boot and zpool/zfs commands on 9.1-RELEASE (possibly because of unavailable pool?) Message-ID: <5192AD94.5020707@web.de> In-Reply-To: <A6A43A3F71994DC68F4F51124FA213E5@multiplay.co.uk> References: <968416157.282645.1368232366317.JavaMail.root@erie.cs.uoguelph.ca> <518EFE05.8010100@hub.org> <518F4130.6080201@hub.org> <518F4307.3060908@hub.org> <519285C9.8000306@web.de> <A6A43A3F71994DC68F4F51124FA213E5@multiplay.co.uk>
next in thread | previous in thread | raw e-mail | index | archive | help
This is a multi-part message in MIME format.
--------------040909000208010909070704
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Am 14.05.2013 21:22, schrieb Steven Hartland:
> ----- Original Message ----- From: "Henner Heck" <Henner.Heck@web.de>
>> Hello all,
>>
>> i set up a PC with FreeBSD 9.1-Release and 2 zfs pools.
>> One is a mirror from which FreeBSD is booting
>> (tough enough without getting "error 2"),
>> the other one a raidz2 for data.
>> The disks for the raidz2 are encrypted with geli and attached manually.
>>
>> I noticed that a "zpool status" or a "zfs list" before attaching
>> the encrypted disks waits for about one minute before showing output.
>> When they finally do, the output is as expected, the raidz2 pool is
>> shown as
>> UNAVAIL and its datasets are not listed.
>> When all the disks are attached with geli, the outputs are given
>> immediately.
>>
>> On boot there are 2 delays.
>> One of about 1 minute after the output
>> "Setting hostid: 0x........"
>> and one of 2 minutes after
>> "Mounting local file systems:.".
>> Both these outputs don't show up in dmesg, which ends with
>> "Trying to mount root from zfs:zroot/ROOT []..." shortly before.
>>
>> I suspect that the boot delays too are because of the encrypted pool.
>>
>> A different machine running FreeBSD 8.3-RELEASE has a delay of only
>> about 3 seconds on "zpool status" with an encrypted pool
>> and also the boot shows no annoying anomalies.
>>
>> Any idea how to get rid of these really long delays?
>
> Try the attached patch see if that helps?
>
> Regards
> Steve
>
> ================================================
> This e.mail is private and confidential between Multiplay (UK) Ltd.
> and the person or entity to whom it is addressed. In the event of
> misdirection, the recipient is prohibited from using, copying,
> printing or otherwise disseminating it or any information contained in
> it.
> In the event of misdirection, illegible or incomplete transmission
> please telephone +44 845 868 1337
> or return the E.mail to postmaster@multiplay.co.uk.
Hello Steven,
i tried to apply the patch, but patch didn't return and gave no output.
I checked my the zfs.c and it doesn't seem to fit this patch.
Please see attached file.
Regards,
Henner Heck
--------------040909000208010909070704
Content-Type: text/plain; charset=windows-1252;
name="zfs.c"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
filename="zfs.c"
/*-
* Copyright (c) 2007 Doug Rabson
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $FreeBSD: release/9.1.0/sys/boot/zfs/zfs.c 237771 2012-06-29 10:30:59Z avg $
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD: release/9.1.0/sys/boot/zfs/zfs.c 237771 2012-06-29 10:30:59Z avg $");
/*
* Stand-alone file reading package.
*/
#include <sys/param.h>
#include <sys/disklabel.h>
#include <sys/time.h>
#include <sys/queue.h>
#include <stddef.h>
#include <stdarg.h>
#include <string.h>
#include <stand.h>
#include <bootstrap.h>
#include "libzfs.h"
#include "zfsimpl.c"
static int zfs_open(const char *path, struct open_file *f);
static int zfs_write(struct open_file *f, void *buf, size_t size, size_t *resid);
static int zfs_close(struct open_file *f);
static int zfs_read(struct open_file *f, void *buf, size_t size, size_t *resid);
static off_t zfs_seek(struct open_file *f, off_t offset, int where);
static int zfs_stat(struct open_file *f, struct stat *sb);
static int zfs_readdir(struct open_file *f, struct dirent *d);
struct devsw zfs_dev;
struct fs_ops zfs_fsops = {
"zfs",
zfs_open,
zfs_close,
zfs_read,
zfs_write,
zfs_seek,
zfs_stat,
zfs_readdir
};
/*
* In-core open file.
*/
struct file {
off_t f_seekp; /* seek pointer */
dnode_phys_t f_dnode;
uint64_t f_zap_type; /* zap type for readdir */
uint64_t f_num_leafs; /* number of fzap leaf blocks */
zap_leaf_phys_t *f_zap_leaf; /* zap leaf buffer */
};
/*
* Open a file.
*/
static int
zfs_open(const char *upath, struct open_file *f)
{
struct zfsmount *mount = (struct zfsmount *)f->f_devdata;
struct file *fp;
int rc;
if (f->f_dev != &zfs_dev)
return (EINVAL);
/* allocate file system specific data structure */
fp = malloc(sizeof(struct file));
bzero(fp, sizeof(struct file));
f->f_fsdata = (void *)fp;
rc = zfs_lookup(mount, upath, &fp->f_dnode);
fp->f_seekp = 0;
if (rc) {
f->f_fsdata = NULL;
free(fp);
}
return (rc);
}
static int
zfs_close(struct open_file *f)
{
struct file *fp = (struct file *)f->f_fsdata;
dnode_cache_obj = 0;
f->f_fsdata = (void *)0;
if (fp == (struct file *)0)
return (0);
free(fp);
return (0);
}
/*
* Copy a portion of a file into kernel memory.
* Cross block boundaries when necessary.
*/
static int
zfs_read(struct open_file *f, void *start, size_t size, size_t *resid /* out */)
{
const spa_t *spa = ((struct zfsmount *)f->f_devdata)->spa;
struct file *fp = (struct file *)f->f_fsdata;
struct stat sb;
size_t n;
int rc;
rc = zfs_stat(f, &sb);
if (rc)
return (rc);
n = size;
if (fp->f_seekp + n > sb.st_size)
n = sb.st_size - fp->f_seekp;
rc = dnode_read(spa, &fp->f_dnode, fp->f_seekp, start, n);
if (rc)
return (rc);
if (0) {
int i;
for (i = 0; i < n; i++)
putchar(((char*) start)[i]);
}
fp->f_seekp += n;
if (resid)
*resid = size - n;
return (0);
}
/*
* Don't be silly - the bootstrap has no business writing anything.
*/
static int
zfs_write(struct open_file *f, void *start, size_t size, size_t *resid /* out */)
{
return (EROFS);
}
static off_t
zfs_seek(struct open_file *f, off_t offset, int where)
{
struct file *fp = (struct file *)f->f_fsdata;
switch (where) {
case SEEK_SET:
fp->f_seekp = offset;
break;
case SEEK_CUR:
fp->f_seekp += offset;
break;
case SEEK_END:
{
struct stat sb;
int error;
error = zfs_stat(f, &sb);
if (error != 0) {
errno = error;
return (-1);
}
fp->f_seekp = sb.st_size - offset;
break;
}
default:
errno = EINVAL;
return (-1);
}
return (fp->f_seekp);
}
static int
zfs_stat(struct open_file *f, struct stat *sb)
{
const spa_t *spa = ((struct zfsmount *)f->f_devdata)->spa;
struct file *fp = (struct file *)f->f_fsdata;
return (zfs_dnode_stat(spa, &fp->f_dnode, sb));
}
static int
zfs_readdir(struct open_file *f, struct dirent *d)
{
const spa_t *spa = ((struct zfsmount *)f->f_devdata)->spa;
struct file *fp = (struct file *)f->f_fsdata;
mzap_ent_phys_t mze;
struct stat sb;
size_t bsize = fp->f_dnode.dn_datablkszsec << SPA_MINBLOCKSHIFT;
int rc;
rc = zfs_stat(f, &sb);
if (rc)
return (rc);
if (!S_ISDIR(sb.st_mode))
return (ENOTDIR);
/*
* If this is the first read, get the zap type.
*/
if (fp->f_seekp == 0) {
rc = dnode_read(spa, &fp->f_dnode,
0, &fp->f_zap_type, sizeof(fp->f_zap_type));
if (rc)
return (rc);
if (fp->f_zap_type == ZBT_MICRO) {
fp->f_seekp = offsetof(mzap_phys_t, mz_chunk);
} else {
rc = dnode_read(spa, &fp->f_dnode,
offsetof(zap_phys_t, zap_num_leafs),
&fp->f_num_leafs,
sizeof(fp->f_num_leafs));
if (rc)
return (rc);
fp->f_seekp = bsize;
fp->f_zap_leaf = (zap_leaf_phys_t *)malloc(bsize);
rc = dnode_read(spa, &fp->f_dnode,
fp->f_seekp,
fp->f_zap_leaf,
bsize);
if (rc)
return (rc);
}
}
if (fp->f_zap_type == ZBT_MICRO) {
mzap_next:
if (fp->f_seekp >= bsize)
return (ENOENT);
rc = dnode_read(spa, &fp->f_dnode,
fp->f_seekp, &mze, sizeof(mze));
if (rc)
return (rc);
fp->f_seekp += sizeof(mze);
if (!mze.mze_name[0])
goto mzap_next;
d->d_fileno = ZFS_DIRENT_OBJ(mze.mze_value);
d->d_type = ZFS_DIRENT_TYPE(mze.mze_value);
strcpy(d->d_name, mze.mze_name);
d->d_namlen = strlen(d->d_name);
return (0);
} else {
zap_leaf_t zl;
zap_leaf_chunk_t *zc, *nc;
int chunk;
size_t namelen;
char *p;
uint64_t value;
/*
* Initialise this so we can use the ZAP size
* calculating macros.
*/
zl.l_bs = ilog2(bsize);
zl.l_phys = fp->f_zap_leaf;
/*
* Figure out which chunk we are currently looking at
* and consider seeking to the next leaf. We use the
* low bits of f_seekp as a simple chunk index.
*/
fzap_next:
chunk = fp->f_seekp & (bsize - 1);
if (chunk == ZAP_LEAF_NUMCHUNKS(&zl)) {
fp->f_seekp = (fp->f_seekp & ~(bsize - 1)) + bsize;
chunk = 0;
/*
* Check for EOF and read the new leaf.
*/
if (fp->f_seekp >= bsize * fp->f_num_leafs)
return (ENOENT);
rc = dnode_read(spa, &fp->f_dnode,
fp->f_seekp,
fp->f_zap_leaf,
bsize);
if (rc)
return (rc);
}
zc = &ZAP_LEAF_CHUNK(&zl, chunk);
fp->f_seekp++;
if (zc->l_entry.le_type != ZAP_CHUNK_ENTRY)
goto fzap_next;
namelen = zc->l_entry.le_name_length;
if (namelen > sizeof(d->d_name))
namelen = sizeof(d->d_name);
/*
* Paste the name back together.
*/
nc = &ZAP_LEAF_CHUNK(&zl, zc->l_entry.le_name_chunk);
p = d->d_name;
while (namelen > 0) {
int len;
len = namelen;
if (len > ZAP_LEAF_ARRAY_BYTES)
len = ZAP_LEAF_ARRAY_BYTES;
memcpy(p, nc->l_array.la_array, len);
p += len;
namelen -= len;
nc = &ZAP_LEAF_CHUNK(&zl, nc->l_array.la_next);
}
d->d_name[sizeof(d->d_name) - 1] = 0;
/*
* Assume the first eight bytes of the value are
* a uint64_t.
*/
value = fzap_leaf_value(&zl, zc);
d->d_fileno = ZFS_DIRENT_OBJ(value);
d->d_type = ZFS_DIRENT_TYPE(value);
d->d_namlen = strlen(d->d_name);
return (0);
}
}
static int
vdev_read(vdev_t *vdev, void *priv, off_t offset, void *buf, size_t size)
{
int fd;
fd = (uintptr_t) priv;
lseek(fd, offset, SEEK_SET);
if (read(fd, buf, size) == size) {
return 0;
} else {
return (EIO);
}
}
static int
zfs_dev_init(void)
{
zfs_init();
if (archsw.arch_zfs_probe == NULL)
return (ENXIO);
archsw.arch_zfs_probe();
return (0);
}
int
zfs_probe_dev(const char *devname, uint64_t *pool_guid)
{
spa_t *spa;
int fd;
int ret;
fd = open(devname, O_RDONLY);
if (fd == -1)
return (ENXIO);
ret = vdev_probe(vdev_read, (void *)(uintptr_t)fd, &spa);
if (ret != 0)
close(fd);
else if (pool_guid != NULL)
*pool_guid = spa->spa_guid;
return (0);
}
/*
* Print information about ZFS pools
*/
static void
zfs_dev_print(int verbose)
{
spa_t *spa;
char line[80];
if (verbose) {
spa_all_status();
return;
}
STAILQ_FOREACH(spa, &zfs_pools, spa_link) {
sprintf(line, " zfs:%s\n", spa->spa_name);
pager_output(line);
}
}
/*
* Attempt to open the pool described by (dev) for use by (f).
*/
static int
zfs_dev_open(struct open_file *f, ...)
{
va_list args;
struct zfs_devdesc *dev;
struct zfsmount *mount;
spa_t *spa;
int rv;
va_start(args, f);
dev = va_arg(args, struct zfs_devdesc *);
va_end(args);
spa = spa_find_by_guid(dev->pool_guid);
if (!spa)
return (ENXIO);
rv = zfs_spa_init(spa);
if (rv != 0)
return (rv);
mount = malloc(sizeof(*mount));
rv = zfs_mount(spa, dev->root_guid, mount);
if (rv != 0) {
free(mount);
return (rv);
}
if (mount->objset.os_type != DMU_OST_ZFS) {
printf("Unexpected object set type %ju\n",
(uintmax_t)mount->objset.os_type);
free(mount);
return (EIO);
}
f->f_devdata = mount;
free(dev);
return (0);
}
static int
zfs_dev_close(struct open_file *f)
{
free(f->f_devdata);
f->f_devdata = NULL;
return (0);
}
static int
zfs_dev_strategy(void *devdata, int rw, daddr_t dblk, size_t size, char *buf, size_t *rsize)
{
return (ENOSYS);
}
struct devsw zfs_dev = {
.dv_name = "zfs",
.dv_type = DEVT_ZFS,
.dv_init = zfs_dev_init,
.dv_strategy = zfs_dev_strategy,
.dv_open = zfs_dev_open,
.dv_close = zfs_dev_close,
.dv_ioctl = noioctl,
.dv_print = zfs_dev_print,
.dv_cleanup = NULL
};
int
zfs_parsedev(struct zfs_devdesc *dev, const char *devspec, const char **path)
{
static char rootname[ZFS_MAXNAMELEN];
static char poolname[ZFS_MAXNAMELEN];
spa_t *spa;
const char *end;
const char *np;
const char *sep;
int rv;
np = devspec;
if (*np != ':')
return (EINVAL);
np++;
end = strchr(np, ':');
if (end == NULL)
return (EINVAL);
sep = strchr(np, '/');
if (sep == NULL || sep >= end)
sep = end;
memcpy(poolname, np, sep - np);
poolname[sep - np] = '\0';
if (sep < end) {
sep++;
memcpy(rootname, sep, end - sep);
rootname[end - sep] = '\0';
}
else
rootname[0] = '\0';
spa = spa_find_by_name(poolname);
if (!spa)
return (ENXIO);
rv = zfs_spa_init(spa);
if (rv != 0)
return (rv);
dev->pool_guid = spa->spa_guid;
if (rootname[0] != '\0') {
rv = zfs_lookup_dataset(spa, rootname, &dev->root_guid);
if (rv != 0)
return (rv);
} else
dev->root_guid = 0;
if (path != NULL)
*path = (*end == '\0') ? end : end + 1;
dev->d_dev = &zfs_dev;
dev->d_type = zfs_dev.dv_type;
return (0);
}
char *
zfs_fmtdev(void *vdev)
{
static char rootname[ZFS_MAXNAMELEN];
static char buf[2 * ZFS_MAXNAMELEN + 8];
struct zfs_devdesc *dev = (struct zfs_devdesc *)vdev;
spa_t *spa;
buf[0] = '\0';
if (dev->d_type != DEVT_ZFS)
return (buf);
spa = spa_find_by_guid(dev->pool_guid);
if (spa == NULL) {
printf("ZFS: can't find pool by guid\n");
return (buf);
}
if (zfs_spa_init(spa) != 0) {
printf("ZFS: can't init pool\n");
return (buf);
}
if (dev->root_guid == 0 && zfs_get_root(spa, &dev->root_guid)) {
printf("ZFS: can't find root filesystem\n");
return (buf);
}
if (zfs_rlookup(spa, dev->root_guid, rootname)) {
printf("ZFS: can't find filesystem by guid\n");
return (buf);
}
if (rootname[0] == '\0')
sprintf(buf, "%s:%s:", dev->d_dev->dv_name, spa->spa_name);
else
sprintf(buf, "%s:%s/%s:", dev->d_dev->dv_name, spa->spa_name,
rootname);
return (buf);
}
--------------040909000208010909070704--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5192AD94.5020707>