Viewing: mount_lustre.c
// SPDX-License-Identifier: GPL-2.0-only
/*
* Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
* Copyright (c) 2011, 2017, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
*
* lustre/utils/mount_lustre.c
*
* Author: Robert Read <rread@clusterfs.com>
* Author: Nathan Rutman <nathan@clusterfs.com>
*/
#ifndef _GNU_SOURCE
#define _GNU_SOURCE
#endif
#include <getopt.h>
#include <mntent.h>
#include <stdbool.h>
#include <stdlib.h>
#include <stdio.h>
#include <unistd.h>
#include <fcntl.h>
#include <errno.h>
#include <string.h>
#include <sys/mount.h>
#include <sys/param.h>
#include <linux/lustre/lustre_ver.h>
#include <ctype.h>
#include <limits.h>
#if defined(HAVE_LUSTRE_CRYPTO) && defined(HAVE_LIBKEYUTILS)
#include <keyutils.h>
#endif
#include <linux/lnet/nidstr.h>
#include <libcfs/util/string.h>
#include "obdctl.h"
#include "mount_utils.h"
#define MAX_RETRIES 99
#define PATH_FORMAT "/etc/lustre/mount.%s.params"
int verbose;
int version;
char *progname;
static void usage(FILE *out)
{
fprintf(out,
"\nThis mount helper should only be invoked via the mount (8) command,\ne.g. mount -t lustre dev dir\n\n");
fprintf(out, "usage: %s [-fhnvV] [-o <srvopt>] <device> <mountpt>\n",
progname);
fprintf(out, "usage: %s [-fhnvV] [-o <cliopt>] <mgstarget> <mountpt>\n",
progname);
fprintf(out,
"\t<device>: the local disk device when mounting a server\n"
"\t<mgstarget>: the server MGS and filesystem for a client:\n"
"\t\t<mgsnid>[:<altmgsnid>...]:/<filesystem>[/<subdir>]\n"
"\t\t\t<mgsnid>: MGS LNet Node Identifier (e.g. mgs01@o2ib)\n"
"\t\t\t<filesystem>: Lustre filesystem name (e.g. lustre1)\n"
"\t\t\t<subdir>: subdirectory of the filesystem to mount\n"
"\t<mountpt>: filesystem mountpoint (e.g. /mnt/lustre)\n"
"\t-f|--fake: fake mount (only update /etc/mtab)\n"
"\t-o force|--force: force mount even if already in /etc/mtab\n"
"\t-h|--help: print this usage message\n"
"\t-n|--nomtab: do not update /etc/mtab after mount\n"
"\t-v|--verbose: print verbose config settings\n"
"\t-V|--version: output build version of the utility and exit\n"
"\tdefault options are marked below with '*'\n"
"\t\t(no)flock: disable* or enable POSIX flock support\n"
"\t\t(no)user_xattr: disable or enable* user xattr namespace\n"
"\t<srvopt>: one or more comma separated server options:\n"
"\t\t(no)acl: disable or enable* POSIX ACL support completely\n"
"\t\tabort_recov: abort server recovery handling\n"
"\t\tnosvc: only start MGC/MGS without starting MDS/OSS\n"
"\t\tnomgs: only start target MDS/OSS, using existing MGS\n"
"\t\tnoclient: do NOT allow clients connections other than MGS local\n"
"\t\tnoscrub: do NOT auto start OI scrub unless requested\n"
"\t\tskip_lfsck: do NOT auto resume paused/crashed LFSCK\n"
"\t\tmd_stripe_cache_size=<num>: set MD RAID device stripe cache size\n"
"\t<cliopt>: one or more comma separated client options:\n"
"\t\texclude=<ostname>[:<ostname>]: list of inactive OSTs (e.g. lustre-OST0001)\n"
"\t\tlocalflock: enable POSIX flock only on local client\n"
"\tmgsname=HOSTNAME: MGS hostname for display in /proc/mounts\n"
"\t\tretry=<num>: number of times mount is retried by client\n"
#ifdef HAVE_GSS
"\t\tskpath=<file|directory>: path of keys to load into kernel keyring\n"
"\t\t\t(clients also auto-load <mountpoint>/.lgss if present)\n"
#endif
"\t\t(no)user_fid2path: disable* or enable user $MOUNT/.lustre/fid access\n"
"\t\t(no)checksum: disable or enable* data checksums\n"
"\t\t(no)lruresize: disable or enable* LDLM dynamic LRU size\n"
"\t\t(no)lazystatfs: disable or enable* statfs to work if OST is unavailable\n"
"\t\t32bitapi: return only 32-bit inode numbers to userspace\n"
"\t\t(no)verbose: disable or enable* messages at filesystem (un,re)mount\n"
#ifdef HAVE_LUSTRE_CRYPTO
#ifdef HAVE_LIBKEYUTILS
"\t\ttest_dummy_encryption: enable test dummy encryption mode\n"
#endif
"\t\tnoencrypt: disable client side encryption\n"
#endif
);
exit((out != stdout) ? EINVAL : 0);
}
/*****************************************************************************
*
* This part was cribbed from util-linux/mount/mount.c. There was no clear
* license information, but many other files in the package are identified as
* GNU GPL, so it's a pretty safe bet that was their intent.
*
****************************************************************************/
struct opt_map {
const char *opt; /* option name */
int inv; /* true if flag value should be inverted */
int mask; /* flag mask value */
};
static const struct opt_map opt_map[] = {
/* these flags are parsed by mount, not Lustre */
{ .opt = "async", .mask = MS_SYNCHRONOUS, .inv = 1 }, /* asynchronous I/O */
{ .opt = "atime", .mask = MS_NOATIME, .inv = 1 }, /* set access time */
{ .opt = "auto" }, /* allow auto mount */
{ .opt = "defaults" }, /* default options */
{ .opt = "dev", .mask = MS_NODEV, .inv = 1 }, /* interpret devs */
{ .opt = "exec", .mask = MS_NOEXEC, .inv = 1 }, /* allow execution */
{ .opt = "loop" },
{ .opt = "noatime", .mask = MS_NOATIME }, /* do not set atime */
{ .opt = "noauto" }, /* mount explicitly */
{ .opt = "nodev", .mask = MS_NODEV }, /* no interpret devs */
{ .opt = "noowner", .inv = 1 }, /* no special privs */
{ .opt = "nosuid", .mask = MS_NOSUID }, /* do not honor suid */
{ .opt = "nouser", .inv = 1 }, /* users cannot mount */
{ .opt = "nousers", .inv = 1 }, /* users cannot mount */
{ .opt = "_netdev" }, /* network only */
{ .opt = "noexec", .mask = MS_NOEXEC }, /* no execute */
{ .opt = "remount", .mask = MS_REMOUNT }, /* remount */
{ .opt = "ro", .mask = MS_RDONLY }, /* read-only */
{ .opt = "rw", .mask = MS_RDONLY, .inv = 1 }, /* read-write */
{ .opt = "suid", .mask = MS_NOSUID, .inv = 1 }, /* honor suid */
{ .opt = "sync", .mask = MS_SYNCHRONOUS }, /* synchronous I/O */
#ifdef MS_NODIRATIME
{ .opt = "diratime", /* set access time */
.mask = MS_NODIRATIME, .inv = 1 }, /* on read */
{ .opt = "nodiratime", /* do not set access */
.mask = MS_NODIRATIME }, /* time on read */
#endif
#ifdef MS_RELATIME
{ .opt = "norelatime", /* do not set rel */
.mask = MS_RELATIME, .inv = 1 }, /* access time */
{ .opt = "relatime", /* set relative */
.mask = MS_RELATIME }, /* access time */
#endif
#ifdef MS_STRICTATIME
{ .opt = "strictatime",
.mask = MS_STRICTATIME }, /* strict access time */
#endif
{ .opt = NULL } };
/****************************************************************************/
/*
* 1 = don't pass on to lustre
* 0 = pass on to lustre
*/
static int parse_one_option(const char *check, int *flagp)
{
const struct opt_map *opt;
for (opt = &opt_map[0]; opt->opt != NULL; opt++) {
if (strncmp(check, opt->opt, strlen(opt->opt)) == 0) {
if (opt->mask) {
if (opt->inv)
*flagp &= ~(opt->mask);
else
*flagp |= opt->mask;
}
return 1;
}
}
/*
* Assume any unknown options are valid and pass them on. The mount
* will fail if lmd_parse, ll_options or ldiskfs doesn't recognize it.
*/
return 0;
}
static size_t merge_strings(char *dst, const char *src, size_t size)
{
size_t dsize = strlen(dst);
size_t len = strlen(src);
size_t ret = dsize + len;
dst += dsize;
size -= dsize;
if (len >= size)
len = size - 1;
memcpy(dst, src, len);
dst[len] = '\0';
return ret;
}
static int append_option(char *options, size_t options_len,
const char *param, const char *value)
{
int rc;
if (options[0] != '\0') {
rc = merge_strings(options, ",", options_len);
if (rc >= options_len)
goto out_err;
}
rc = merge_strings(options, param, options_len);
if (rc >= options_len)
goto out_err;
if (value) {
rc = merge_strings(options, value, options_len);
if (rc >= options_len)
goto out_err;
}
return 0;
out_err:
/* this 'too long' message is checked in conf-sanity test_98 */
fprintf(stderr,
"error: add mount options %s%s too long, %zu over limit %zu\n",
param, value, strlen(options) + strlen(param) + strlen(value),
options_len);
return E2BIG;
}
/*
* Replace options with subset of Lustre-specific options, and
* fill in mount flags
*/
static int parse_options(struct mount_opts *mop, char *orig_options,
int *flagp, size_t options_len)
{
char *options, *opt, *nextopt, *arg, *val;
int rc = 0;
options = calloc(strlen(orig_options) + 1, 1);
if (!options)
return ENOMEM;
*flagp = 0;
nextopt = orig_options;
while ((opt = strsep(&nextopt, ","))) {
if (!*opt)
/* empty option */
continue;
/* Handle retries in a slightly different manner */
arg = opt;
val = strchr(opt, '=');
/*
* please note that some ldiskfs mount options are also in
* the form of param=value. We should pay attention not to
* remove those mount options, see bug 22097.
*/
if (strcmp(opt, "force") == 0) {
/* XXX special check for 'force' option */
++mop->mo_force;
printf("force: %d\n", mop->mo_force);
} else if (val && strncmp(arg, "max_sectors_kb", 14) == 0) {
fprintf(stderr,
"%s: max_sectors_kb is ignored\n",
progname);
} else if (val &&
strncmp(arg, "md_stripe_cache_size", 20) == 0) {
mop->mo_md_stripe_cache_size = atoi(val + 1);
} else if (val && strncmp(arg, "mgsname", 7) == 0) {
rc = append_option(options, options_len, opt, NULL);
if (rc != 0)
goto out_options;
} else if (val && strncmp(arg, "mgssec", 6) == 0) {
rc = append_option(options, options_len, opt, NULL);
if (rc != 0)
goto out_options;
} else if (strncmp(arg, "nosvc", 5) == 0) {
mop->mo_nosvc = 1;
rc = append_option(options, options_len, opt, NULL);
if (rc != 0)
goto out_options;
} else if (val && strncmp(arg, "retry", 5) == 0) {
mop->mo_retry = atoi(val + 1);
if (mop->mo_retry > MAX_RETRIES)
mop->mo_retry = MAX_RETRIES;
else if (mop->mo_retry < 0)
mop->mo_retry = 0;
#ifdef HAVE_GSS
} else if (val && strncmp(opt, "skpath=", 7) == 0) {
if (strlen(val) + 1 >= sizeof(mop->mo_skpath)) {
fprintf(stderr,
"%s: shared key path too long\n",
progname);
free(options);
return EINVAL;
}
strncpy(mop->mo_skpath, val + 1,
sizeof(mop->mo_skpath) - 1);
#endif
#ifdef HAVE_LUSTRE_CRYPTO
} else if (strncmp(arg, "test_dummy_encryption", 21) == 0) {
#ifdef HAVE_LIBKEYUTILS
#ifdef HAVE_FSCRYPT_DUMMY_CONTEXT_ENABLED
/* Using dummy encryption mode requires inserting a
* special dummy key into the session keyring.
* Key type is "logon", key description is
* "fscrypt:4242424242424242", and key payload has to be
* in the form <mode><raw><size>, where:
* <mode> is "\x00\x00\x00\x00"
* <raw> is "$(printf ""\\\\x%02x"" {0..63})"
* <size> is "\x40\x00\x00\x00" for little endian,
* "\x00\x00\x00\x40" for big endian.
*/
char payload[72];
int *p = (int *)payload;
char *q = (char *)(p + 1);
int i = 0;
key_serial_t key;
*p = 0;
while (i < 0x40)
*(q++) = i++;
p = (int *)q;
*p = 0x40;
key = add_key("logon", "fscrypt:4242424242424242",
(const void *)payload, sizeof(payload),
KEY_SPEC_SESSION_KEYRING);
if (key == -1) {
fprintf(stderr,
"%s: test dummy encryption option ignored: could not insert dummy encryption key into session keyring\n",
progname);
} else {
/* pass this on as an option */
rc = append_option(options, options_len, opt,
NULL);
if (rc != 0)
goto out_options;
}
#else /* !HAVE_FSCRYPT_DUMMY_CONTEXT_ENABLED */
/* pass this on as an option */
rc = append_option(options, options_len, opt,
NULL);
if (rc != 0)
goto out_options;
#endif
#else /* HAVE_LIBKEYUTILS */
fprintf(stderr,
"%s: test dummy encryption option ignored: Lustre not built with libkeyutils support\n",
progname);
#endif
#endif
} else if (parse_one_option(opt, flagp) == 0) {
/* pass this on as an option */
rc = append_option(options, options_len, opt, NULL);
if (rc != 0)
goto out_options;
}
}
#ifdef MS_STRICTATIME
#if LUSTRE_VERSION_CODE > OBD_OCD_VERSION(3, 2, 53, 0)
/*
* LU-1783
* In the future when upstream fixes land in all supported kernels
* we should stop forcing MS_STRICTATIME in lustre mounts.
* We override the kernel level default of MS_RELATIME for now
* due to a kernel vfs level bug in atime updates that fails
* to reset timestamps from the future.
*/
#warn "remove MS_STRICTATIME override if kernel updates atime from the future"
#endif
/*
* set strictatime to default if NOATIME or RELATIME
* not given explicit
*/
if (!(*flagp & (MS_NOATIME | MS_RELATIME)))
*flagp |= MS_STRICTATIME;
#endif
strcpy(orig_options, options);
out_options:
free(options);
return rc;
}
#ifdef HAVE_SERVER_SUPPORT
/* Add mgsnids from ldd params */
static int add_mgsnids(struct mount_opts *mop, char *options,
const char *params, size_t options_len)
{
char *ptr = (char *)params;
char tmp, *sep;
int rc = 0;
while ((ptr = strstr(ptr, PARAM_MGSNODE)) != NULL) {
sep = strchr(ptr, ' ');
if (sep) {
tmp = *sep;
*sep = '\0';
}
rc = append_option(options, options_len, ptr, NULL);
if (rc != 0)
goto out;
mop->mo_have_mgsnid++;
if (sep) {
*sep = tmp;
ptr = sep;
} else {
break;
}
}
out:
return rc;
}
static int clear_update_ondisk(char *source, struct lustre_disk_data *ldd)
{
char wanted_mountopts[512] = "";
struct mkfs_opts mkop;
int ret;
int ret2;
memset(&mkop, 0, sizeof(mkop));
mkop.mo_ldd = *ldd;
mkop.mo_ldd.ldd_flags &= ~LDD_F_UPDATE;
mkop.mo_flags = MO_NOHOSTID_CHECK; /* Ignore missing hostid */
if (strlen(source) > sizeof(mkop.mo_device) - 1) {
fatal();
fprintf(stderr, "Device name too long: %s\n", source);
return -E2BIG;
}
strncpy(mkop.mo_device, source, sizeof(mkop.mo_device));
ret = osd_prepare_lustre(&mkop,
wanted_mountopts, sizeof(wanted_mountopts));
if (ret) {
fatal();
fprintf(stderr, "Can't prepare device %s: %s\n",
source, strerror(ret));
return ret;
}
/* Create the loopback file */
if (mkop.mo_flags & MO_IS_LOOP) {
ret = access(mkop.mo_device, F_OK);
if (ret) {
ret = errno;
fatal();
fprintf(stderr, "Can't access device %s: %s\n",
source, strerror(ret));
return ret;
}
ret = loop_setup(&mkop);
if (ret) {
fatal();
fprintf(stderr, "Loop device setup for %s failed: %s\n",
mkop.mo_device, strerror(ret));
return ret;
}
}
ret = osd_write_ldd(&mkop);
if (ret != 0) {
fatal();
fprintf(stderr, "failed to write local files: %s\n",
strerror(ret));
}
ret2 = loop_cleanup(&mkop);
if (ret == 0)
ret = ret2;
return ret;
}
static int parse_ldd(char *source, struct mount_opts *mop,
char *options, size_t options_len)
{
struct lustre_disk_data *ldd = &mop->mo_ldd;
char *cur, *start;
char *temp_options;
bool skip_mgsnode_param = false;
int rc = 0;
rc = osd_is_lustre(source, &ldd->ldd_mount_type);
if (rc == 0) {
fprintf(stderr,
"%s: %s has not been formatted with mkfs.lustre or the backend filesystem type is not supported by this tool\n",
progname, source);
return ENODEV;
}
rc = osd_read_ldd(source, ldd);
if (rc) {
fprintf(stderr,
"%s: %s failed to read permanent mount data: %s\n",
progname, source, rc >= 0 ? strerror(rc) : "");
return rc;
}
if ((IS_MDT(ldd) || IS_OST(ldd)) &&
(ldd->ldd_flags & LDD_F_NEED_INDEX)) {
fprintf(stderr,
"%s: %s has no index assigned (probably formatted with old mkfs)\n",
progname, source);
return EINVAL;
}
if (ldd->ldd_flags & LDD_F_UPDATE)
clear_update_ondisk(source, ldd);
/* Since we never rewrite ldd, ignore temp flags */
ldd->ldd_flags &= ~(LDD_F_VIRGIN | LDD_F_WRITECONF |
LDD_F_NO_LOCAL_LOGS);
/* This is to make sure default options go first */
temp_options = strdup(options);
if (!temp_options) {
fprintf(stderr, "%s: can't allocate memory for temp_options\n",
progname);
return ENOMEM;
}
strncpy(options, ldd->ldd_mount_opts, options_len);
rc = append_option(options, options_len, temp_options, NULL);
free(temp_options);
if (rc != 0)
return rc;
/* svname of the form lustre:OST1234 means never registered */
rc = strlen(ldd->ldd_svname);
if (strcmp(ldd->ldd_svname, "MGS") != 0) {
if (rc < 8) {
fprintf(stderr, "%s: invalid name '%s'\n",
progname, ldd->ldd_svname);
return EINVAL;
} else if (ldd->ldd_svname[rc - 8] == ':') {
ldd->ldd_svname[rc - 8] = '-';
ldd->ldd_flags |= LDD_F_VIRGIN;
} else if (ldd->ldd_svname[rc - 8] == '=') {
ldd->ldd_svname[rc - 8] = '-';
ldd->ldd_flags |= LDD_F_WRITECONF;
} else if (ldd->ldd_svname[rc - 8] == '+') {
ldd->ldd_svname[rc - 8] = '-';
ldd->ldd_flags |= LDD_F_NO_LOCAL_LOGS;
}
}
/* backend osd type */
rc = append_option(options, options_len, "osd=",
mt_type(ldd->ldd_mount_type));
if (rc != 0)
return rc;
if (!mop->mo_have_mgsnid) {
/*
* Only use disk data if mount -o mgsnode=nid wasn't
* specified
*/
if (ldd->ldd_flags & LDD_F_SV_TYPE_MGS) {
rc = append_option(options, options_len, "mgs", NULL);
if (rc != 0)
return rc;
mop->mo_have_mgsnid++;
} else {
if (add_mgsnids(mop, options, ldd->ldd_params,
options_len))
return E2BIG;
skip_mgsnode_param = true;
}
}
/* Better have an mgsnid by now */
if (!mop->mo_have_mgsnid) {
fprintf(stderr, "%s: missing option mgsnode=<nid>\n",
progname);
return EINVAL;
}
if (ldd->ldd_flags & LDD_F_VIRGIN) {
rc = append_option(options, options_len, "virgin", NULL);
if (rc != 0)
return rc;
}
if (ldd->ldd_flags & LDD_F_UPDATE) {
rc = append_option(options, options_len, "update", NULL);
if (rc != 0)
return rc;
}
if (ldd->ldd_flags & LDD_F_WRITECONF) {
rc = append_option(options, options_len, "writeconf", NULL);
if (rc != 0)
return rc;
}
if (ldd->ldd_flags & LDD_F_NO_LOCAL_LOGS) {
rc = append_option(options, options_len, "nolocallogs", NULL);
if (rc != 0)
return rc;
}
if (ldd->ldd_flags & LDD_F_NO_PRIMNODE) {
rc = append_option(options, options_len, "noprimnode", NULL);
if (rc != 0)
return rc;
}
/*
* prefix every lustre parameter with param= so that in-kernel
* mount can recognize them properly and send to MGS at registration
*/
start = ldd->ldd_params;
while (start && *start != '\0') {
while (*start == ' ')
start++;
if (*start == '\0')
break;
cur = start;
start = strchr(cur, ' ');
if (start) {
*start = '\0';
start++;
}
if (skip_mgsnode_param && !strncmp(cur, "mgsnode", 7))
continue;
rc = append_option(options, options_len, "param=", cur);
if (rc != 0)
return rc;
}
/* svname must be last option */
rc = append_option(options, options_len, "svname=", ldd->ldd_svname);
return rc;
}
#endif /* HAVE_SERVER_SUPPORT */
static void set_defaults(struct mount_opts *mop)
{
memset(mop, 0, sizeof(*mop));
mop->mo_usource = NULL;
mop->mo_source = NULL;
mop->mo_nomtab = 0;
mop->mo_fake = 0;
mop->mo_force = 0;
mop->mo_retry = 0;
mop->mo_have_mgsnid = 0;
mop->mo_md_stripe_cache_size = 16384;
mop->mo_orig_options = "";
mop->mo_nosvc = 0;
}
static int parse_opts(int argc, char *const argv[], struct mount_opts *mop,
char *options, size_t options_len)
{
static struct option long_opts[] = {
{ .val = 1, .name = "force", .has_arg = no_argument },
{ .val = 'f', .name = "fake", .has_arg = no_argument },
{ .val = 'h', .name = "help", .has_arg = no_argument },
{ .val = 'M', .name = "mgsname", .has_arg = required_argument },
{ .val = 'n', .name = "nomtab", .has_arg = no_argument },
{ .val = 'o', .name = "options", .has_arg = required_argument },
{ .val = 'v', .name = "verbose", .has_arg = no_argument },
{ .val = 'V', .name = "version", .has_arg = no_argument },
{ .name = NULL } };
char real_path[PATH_MAX] = {'\0'};
FILE *f;
char path[256], name[256];
size_t sz;
char *ptr;
int opt, rc;
while ((opt = getopt_long(argc, argv, "fhM:no:vV",
long_opts, NULL)) != EOF){
switch (opt) {
case 1:
++mop->mo_force;
printf("force: %d\n", mop->mo_force);
break;
case 'f':
++mop->mo_fake;
printf("fake: %d\n", mop->mo_fake);
break;
case 'h':
usage(stdout);
break;
case 'M':
rc = append_option(options, options_len, "mgsname=",
optarg);
if (rc != 0)
return rc;
break;
case 'n':
++mop->mo_nomtab;
printf("nomtab: %d\n", mop->mo_nomtab);
break;
case 'o':
mop->mo_orig_options = optarg;
break;
case 'v':
++verbose;
break;
case 'V':
++version;
fprintf(stdout, "%s %s (libmount)\n", progname,
LUSTRE_VERSION_STRING);
return 0;
default:
fprintf(stderr, "%s: unknown option '%c'\n",
progname, opt);
usage(stderr);
break;
}
}
if (optind + 2 > argc) {
fprintf(stderr, "%s: too few arguments\n", progname);
usage(stderr);
}
mop->mo_usource = argv[optind];
if (!mop->mo_usource)
usage(stderr);
#ifdef HAVE_SERVER_SUPPORT
/* osd-wbcfs lustre_tgt */
if (strcmp(mop->mo_usource, OSD_WBCFS_DEV) == 0) {
mop->mo_ldd.ldd_mount_type = LDD_MT_WBCFS;
mop->mo_source = strdup(mop->mo_usource);
if (!realpath(argv[optind + 1], mop->mo_target)) {
rc = errno;
fprintf(stderr, "warning: %s: cannot resolve: %s\n",
argv[optind], strerror(errno));
return rc;
}
return 0;
}
#endif
/**
* Try to get the real path to the device, in case it is a
* symbolic link for instance
*/
if (realpath(mop->mo_usource, real_path) != NULL) {
ptr = strrchr(real_path, '/');
if (ptr && strncmp(ptr, "/dm-", 4) == 0 &&
isdigit(*(ptr + 4))) {
snprintf(path, sizeof(path), "/sys/block/%s/dm/name",
ptr + 1);
if ((f = fopen(path, "r"))) {
/* read "<name>\n" from sysfs */
if (fgets(name, sizeof(name), f) &&
(sz = strlen(name)) > 1) {
name[sz - 1] = '\0';
snprintf(real_path, sizeof(real_path),
"/dev/mapper/%s", name);
}
fclose(f);
}
}
mop->mo_usource = strdup(real_path);
}
ptr = strstr(mop->mo_usource, ":/");
if (ptr) {
mop->mo_source = convert_hostnames(mop->mo_usource, true);
if (!mop->mo_source)
usage(stderr);
mop->mo_fsname = convert_fsname(mop->mo_usource);
} else {
mop->mo_source = strdup(mop->mo_usource);
}
if (realpath(argv[optind + 1], mop->mo_target) == NULL) {
rc = errno;
fprintf(stderr, "warning: %s: cannot resolve: %s\n",
argv[optind + 1], strerror(errno));
return rc;
}
return 0;
}
#ifdef HAVE_SERVER_SUPPORT
/*
* change label from <fsname>:<index> to
* <fsname>-<index> to indicate the device has
* been registered. only if the label is
* supposed to be changed and target service
* is supposed to start
*/
static void label_lustre(struct mount_opts *mop)
{
if (mop->mo_nosvc)
return;
if (mop->mo_ldd.ldd_flags & (LDD_F_VIRGIN | LDD_F_WRITECONF |
LDD_F_NO_LOCAL_LOGS)) {
(void)osd_label_lustre(mop);
} else {
struct lustre_disk_data ldd;
int rc;
/*
* device label could be changed after journal recovery,
* it should also be relabeled for mount has succeeded.
*/
memset(&ldd, 0, sizeof(ldd));
ldd.ldd_mount_type = mop->mo_ldd.ldd_mount_type;
rc = osd_label_read(mop->mo_source, &ldd);
if (rc == 0) {
rc = strlen(ldd.ldd_svname);
if (rc >= 8 && ldd.ldd_svname[rc - 8] != '-')
(void)osd_label_lustre(mop);
}
}
}
#endif /* HAVE_SERVER_SUPPORT */
/* no-op version for mount, since it only needs temporary parameters */
int jt_lcfg_setparam_perm(int argc, char **argv, struct param_opts *popt)
{
return 0;
}
const char *jt_cmdname(const char *func)
{
return func;
}
struct sp_workq { int unused; };
int spwq_init(struct sp_workq *wq, struct param_opts *popt)
{ return 0; }
int spwq_destroy(struct sp_workq *wq)
{ return 0; }
int spwq_expand(struct sp_workq *wq, size_t num_items)
{ return 0; }
int spwq_add_item(struct sp_workq *wq, char *path,
char *param_name, char *value)
{ return 0; }
int sp_run_threads(struct sp_workq *wq)
{ return 0; }
int parse_param_file(char *path)
{
int rc = 0;
FILE *file = fopen(path, "r");
if (file) {
char *param = NULL;
size_t len = 0;
while (getline(¶m, &len, file) != -1) {
char *tmp;
/* skip any comments on lines */
tmp = strchr(param, '#');
if (tmp) {
if (tmp == param)
continue;
*tmp = '\0';
}
/* remove trailing newline/whitespace. embedded OK */
tmp = strchr(param, '\n');
if (tmp)
*tmp = '\0';
if (!*param)
continue;
rc = jt_lcfg_setparam(2, (char*[3])
{ "mount.params", param, NULL });
}
free(param);
}
return rc;
}
/**
* check_vm_dirty_ratio() - Check vm.dirty_ratio
*
* Check if vm.dirty_ratio is set to a value that may cause performance issues.
* Lustre's writeback support does not work well with vm_dirty_ratio set to 0.
* Values below 10 can cause inefficient write behavior under load.
*/
static void check_vm_dirty_ratio(void)
{
int dirty_ratio = 0;
FILE *fp = fopen("/proc/sys/vm/dirty_ratio", "r");
if (fp) {
if (fscanf(fp, "%d", &dirty_ratio) == 1 && dirty_ratio < 10) {
fprintf(stderr,
"Warning: vm.dirty_ratio is set to %d, which may hurt performance\n",
dirty_ratio);
fprintf(stderr, "Consider increasing it to at least 10 using:\n");
fprintf(stderr, " sysctl -w vm.dirty_ratio=10\n");
}
fclose(fp);
return;
}
return;
}
int set_client_params(char *fsname)
{
char path[PATH_MAX];
int rc, rc1;
snprintf(path, sizeof(path), PATH_FORMAT, "client");
rc = parse_param_file(path);
if (fsname) {
snprintf(path, sizeof(path), PATH_FORMAT, fsname);
rc1 = parse_param_file(path);
if (rc1 && !rc)
rc = rc1;
}
return 0;
}
int main(int argc, char *const argv[])
{
struct mount_opts mop;
#ifdef HAVE_GSS
unsigned int key = -1;
#endif
char *options;
int i, flags;
int rc;
bool client;
size_t maxopt_len;
size_t g_pagesize;
char *fstype = "lustre";
progname = strrchr(argv[0], '/');
progname = progname ? progname + 1 : argv[0];
set_defaults(&mop);
g_pagesize = sysconf(_SC_PAGESIZE);
if (g_pagesize == -1) {
rc = errno;
printf("error: failed to get page size: %s\n", strerror(rc));
return rc;
}
maxopt_len = MIN(g_pagesize, 64 * 1024);
options = malloc(maxopt_len);
if (!options) {
fprintf(stderr, "can't allocate memory for options\n");
rc = ENOMEM;
goto out_mo_source;
}
options[0] = '\0';
rc = parse_opts(argc, argv, &mop, options, maxopt_len);
if (rc || version)
goto out_options;
if (verbose) {
for (i = 0; i < argc; i++)
printf("arg[%d] = %s\n", i, argv[i]);
printf("source = %s (%s), target = %s\n", mop.mo_usource,
mop.mo_source, mop.mo_target);
printf("options(%zu/%zu) = %s\n", strlen(mop.mo_orig_options),
maxopt_len, mop.mo_orig_options);
}
if (strlen(mop.mo_orig_options) >= maxopt_len) {
/* this 'too long' message is checked in conf-sanity test_98 */
fprintf(stderr,
"error: mount options too long, %zu over limit %zu\n",
strlen(mop.mo_orig_options), maxopt_len);
rc = E2BIG;
goto out_options;
}
strcpy(options, mop.mo_orig_options);
rc = parse_options(&mop, options, &flags, maxopt_len);
if (rc) {
fprintf(stderr, "%s: can't parse options: %s\n",
progname, options);
goto out_options;
}
/* Prefer filesystem type given on mount command line
* so it appears correctly in the /proc/mounts output.
*/
if (strstr(argv[0], "mount.lustre_tgt"))
fstype = "lustre_tgt";
if (!mop.mo_force) {
rc = check_mtab_entry(mop.mo_usource, mop.mo_source,
mop.mo_target, fstype);
if (rc && !(flags & MS_REMOUNT)) {
fprintf(stderr,
"%s: according to %s %s is already mounted on %s\n",
progname, MOUNTED, mop.mo_usource,
mop.mo_target);
rc = EEXIST;
goto out_options;
}
if (!rc && (flags & MS_REMOUNT)) {
fprintf(stderr,
"%s: according to %s %s is not already mounted on %s\n",
progname, MOUNTED, mop.mo_usource,
mop.mo_target);
rc = ENOENT;
goto out_options;
}
}
if (flags & MS_REMOUNT)
mop.mo_nomtab++;
rc = access(mop.mo_target, F_OK);
if (rc) {
rc = errno;
fprintf(stderr, "%s: %s inaccessible: %s\n", progname,
mop.mo_target, strerror(errno));
goto out_options;
}
client = (strstr(mop.mo_usource, ":/") != NULL);
if (client) {
/* Check vm.dirty_ratio for client mounts */
check_vm_dirty_ratio();
} else {
#ifdef HAVE_SERVER_SUPPORT
rc = osd_init();
if (rc)
goto out_options;
rc = parse_ldd(mop.mo_source, &mop, options, maxopt_len);
if (rc)
goto out_osd;
#else
rc = EINVAL;
fprintf(stderr, "%s: cannot mount %s: no server support\n",
progname, mop.mo_usource);
goto out_options;
#endif
}
#ifdef HAVE_GSS
/* For client mounts, auto check for .lgss file in mount point */
if (client && mop.mo_skpath[0] == '\0') {
char lgss_path[PATH_MAX];
struct stat lgss_stat;
int ret;
ret = snprintf(lgss_path, sizeof(lgss_path), "%s/.lgss",
mop.mo_target);
if (ret > 0 && ret < sizeof(lgss_path) &&
stat(lgss_path, &lgss_stat) == 0 &&
S_ISREG(lgss_stat.st_mode)) {
if (verbose)
printf("Found SSK key file %s, loading\n",
lgss_path);
/* Set the skpath to the .lgss file */
strscpy(mop.mo_skpath, lgss_path,
sizeof(mop.mo_skpath));
}
}
if (mop.mo_skpath[0] != '\0') {
/* Treat shared key failures as fatal */
rc = load_shared_keys(&mop, client);
key = rc;
if (rc < 0) {
fprintf(stderr, "%s: Error loading shared keys: %s\n",
progname, strerror(-rc));
goto out_unload;
}
/* rc > 0 means a valid key serial was returned, so forward it
* to the kernel
*/
if (rc) {
char keyid[32];
snprintf(keyid, sizeof(keyid), "%d", key);
rc = append_option(options, maxopt_len, "skid=", keyid);
if (rc != 0)
goto out_unload;
}
}
#endif /* HAVE_GSS */
/*
* Auto-generate mgsname from device string if not already specified,
* and convert_hostname() has replaced the hostname with an IP address.
*/
if (client && !strstr(options, "mgsname=") &&
isalpha(mop.mo_usource[0])) {
char *end = strpbrk(mop.mo_usource, ",:");
if (end && strncmp(mop.mo_usource, mop.mo_source,
end - mop.mo_usource) != 0) {
char sep = *end;
*end = '\0'; /* temporarily NUL terminate mgsname */
rc = append_option(options, maxopt_len, "mgsname=",
mop.mo_usource);
if (verbose)
printf("add option mgsname=%s: rc = %d\n",
mop.mo_usource, rc);
*end = sep;
rc = 0;
}
}
/*
* Since Linux 2.4, the target device doesn't get passed to any of our
* functions. So we'll stick it on the end of the options.
*/
rc = append_option(options, maxopt_len, "device=", mop.mo_source);
if (rc != 0)
goto out_unload;
if (verbose)
printf("mounting device %s at %s, flags=%#x options=%s\n",
mop.mo_source, mop.mo_target, flags, options);
#ifdef HAVE_SERVER_SUPPORT
if (!client && osd_tune_lustre(mop.mo_source, &mop)) {
if (verbose)
fprintf(stderr,
"%s: unable to set tunables for %s (may cause reduced IO performance)\n",
argv[0], mop.mo_source);
}
#endif
if (!mop.mo_fake) {
/*
* flags and target get to lustre_get_sb(), but not
* lustre_fill_super(). Lustre ignores the flags, but mount
* does not.
*/
for (i = 0, rc = -EAGAIN; i <= mop.mo_retry && rc != 0; i++) {
rc = mount(mop.mo_source, mop.mo_target, fstype,
flags, (void *)options);
if (rc != 0) {
if (verbose) {
fprintf(stderr,
"%s: mount -t %s %s at %s failed: %s retries left: %d\n",
basename(progname), fstype,
mop.mo_usource, mop.mo_target,
strerror(errno),
mop.mo_retry - i);
}
#if LUSTRE_VERSION_CODE > OBD_OCD_VERSION(2, 18, 53, 0)
/* Pre-2.13 Lustre without 'lustre_tgt' type?
* Try with 'lustre' instead. Eventually this
* can be removed (e.g. 2.18 or whenever).
*/
if (errno == ENODEV &&
strcmp(fstype, "lustre_tgt") == 0) {
fstype = "lustre";
i--;
continue;
}
#endif
if (mop.mo_retry) {
int limit = i / 2 > 5 ? i / 2 : 5;
sleep(1 << limit);
} else {
rc = errno;
}
#ifdef HAVE_SERVER_SUPPORT
} else {
if (!client)
label_lustre(&mop);
#endif
}
}
}
if (rc) {
char *cli;
rc = errno;
cli = strrchr(mop.mo_usource, ':');
if (cli && (strlen(cli) > 2))
cli += 2;
else
cli = NULL;
fprintf(stderr, "%s: mount %s at %s failed: %s\n", progname,
mop.mo_usource, mop.mo_target, strerror(errno));
if (errno == EBUSY || errno == EINPROGRESS)
fprintf(stderr,
"Is the backend filesystem mounted?\n Check /proc/mounts\n");
else if (errno == ENODEV)
fprintf(stderr,
"Are the lustre modules loaded?\n Check /etc/modprobe.conf and /proc/filesystems\n");
else if (errno == ENOTBLK)
fprintf(stderr, "Do you need -o loop?\n");
else if (errno == ENOMEDIUM)
fprintf(stderr,
"This filesystem needs at least 1 OST\n");
else if (errno == ENOENT) {
fprintf(stderr, "Is the MGS specification correct?\n");
fprintf(stderr, "Is the filesystem name correct?\n");
fprintf(stderr,
"If upgrading, is the copied client log valid? (see upgrade docs)\n");
}
else if (errno == EALREADY)
fprintf(stderr,
"The target service is already running. (%s)\n",
mop.mo_usource);
else if (errno == ENXIO)
fprintf(stderr,
"The target service failed to start (bad config log?) (%s). See /var/log/messages.\n",
mop.mo_usource);
else if (errno == EIO)
fprintf(stderr, "Is the MGS running?\n");
else if (errno == EADDRINUSE)
fprintf(stderr,
"The target service's index is already in use. (%s)\n",
mop.mo_usource);
else if (errno == EINVAL) {
fprintf(stderr, "This may have multiple causes.\n");
if (cli)
fprintf(stderr,
"Is '%s' the correct filesystem name?\n",
cli);
fprintf(stderr, "Are the mount options correct?\n");
fprintf(stderr, "Check the syslog for more info.\n");
}
/* May as well try to clean up loop devs */
if (strncmp(mop.mo_usource, "/dev/loop", 9) == 0) {
char cmd[256];
int ret;
sprintf(cmd, "/sbin/losetup -d %s", mop.mo_usource);
if ((ret = system(cmd)) < 0)
rc = errno;
else if (ret > 0)
rc = WEXITSTATUS(ret);
}
} else {
/*
* Deal with utab just for client. Note that we ignore
* the return value here since it is not worth to fail
* mount by prevent some rare cases
* Client specific parameters are stored in either
* '/etc/lustre/mount.params' or '/etc/lustre/FSNAME.params'
* and are set here.
*/
if (client) {
update_utab_entry(&mop);
rc = set_client_params(mop.mo_fsname);
}
if (!mop.mo_nomtab) {
rc = update_mtab_entry(mop.mo_usource, mop.mo_target,
"lustre", mop.mo_orig_options,
0, 0, 0);
}
}
out_unload:
#ifdef HAVE_GSS
/* unload temporary key, most useful if kernel does not handle it */
unload_shared_key(key);
#endif
#ifdef HAVE_SERVER_SUPPORT
out_osd:
if (!client)
osd_fini();
#endif
out_options:
free(options);
out_mo_source:
/* mo_usource should be freed, but we can rely on the kernel */
free(mop.mo_source);
free(mop.mo_fsname);
return rc;
}
