mirror of
git://sourceware.org/git/lvm2.git
synced 2024-12-22 17:35:59 +03:00
9e3e4d6994
- When defining configuration source, the code now uses separate CONFIG_PROFILE_COMMAND and CONFIG_PROFILE_METADATA markers (before, it was just CONFIG_PROFILE that did not make the difference between the two). This helps when checking the configuration if it contains correct set of options which are all in either command-profilable or metadata-profilable group without mixing these groups together - so it's a firm distinction. The "command profile" can't contain "metadata profile" and vice versa! This is strictly checked and if the settings are mixed, such profile is rejected and it's not used. So in the end, the CONFIG_PROFILE_COMMAND set of options and CONFIG_PROFILE_METADATA are mutually exclusive sets. - Marking configuration with one or the other marker will also determine the way these configuration sources are positioned in the configuration cascade which is now: CONFIG_STRING -> CONFIG_PROFILE_COMMAND -> CONFIG_PROFILE_METADATA -> CONFIG_FILE/CONFIG_MERGED_FILES - Marking configuration with one or the other marker will also make it possible to issue a command context refresh (will be probably a part of a future patch) if needed for settings in global profile set. For settings in metadata profile set this is impossible since we can't refresh cmd context in the middle of reading VG/LV metadata and for each VG/LV separately because each VG/LV can have a different metadata profile assinged and it's not possible to change these settings at this level. - When command profile is incorrect, it's rejected *and also* the command exits immediately - the profile *must* be correct for the command that was run with a profile to be executed. Before this patch, when the profile was found incorrect, there was just the warning message and the command continued without profile applied. But it's more correct to exit immediately in this case. - When metadata profile is incorrect, we reject it during command runtime (as we know the profile name from metadata and not early from command line as it is in case of command profiles) and we *do continue* with the command as we're in the middle of operation. Also, the metadata profile is applied directly and on the fly on find_config_tree_* fn call and even if the metadata profile is found incorrect, we still need to return the non-profiled value as found in the other configuration provided or default value. To exit immediately even in this case, we'd need to refactor existing find_config_tree_* fns so they can return error. Currently, these fns return only config values (which end up with default values in the end if the config is not found). - To check the profile validity before use to be sure it's correct, one can use : lvm dumpconfig --commandprofile/--metadataprofile ProfileName --validate (the --commandprofile/--metadataprofile for dumpconfig will come as part of the subsequent patch) - This patch also adds a reference to --commandprofile and --metadataprofile in the cmd help string (which was missing before for the --profile for some commands). We do not mention --profile now as people should use --commandprofile or --metadataprofile directly. However, the --profile is still supported for backward compatibility and it's translated as: --profile == --metadataprofile for lvcreate, vgcreate, lvchange and vgchange (as these commands are able to attach profile to metadata) --profile == --commandprofile for all the other commands (--metadataprofile is not allowed there as it makes no sense) - This patch also contains some cleanups to make the code handling the profiles more readable...
496 lines
13 KiB
C
496 lines
13 KiB
C
/*
|
|
* Copyright (C) 2003-2004 Sistina Software, Inc. All rights reserved.
|
|
* Copyright (C) 2004-2011 Red Hat, Inc. All rights reserved.
|
|
*
|
|
* This file is part of LVM2.
|
|
*
|
|
* This copyrighted material is made available to anyone wishing to use,
|
|
* modify, copy, or redistribute it subject to the terms and conditions
|
|
* of the GNU Lesser General Public License v.2.1.
|
|
*
|
|
* You should have received a copy of the GNU Lesser General Public License
|
|
* along with this program; if not, write to the Free Software Foundation,
|
|
* Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
*/
|
|
|
|
#include "lib.h"
|
|
#include "memlock.h"
|
|
#include "defaults.h"
|
|
#include "config.h"
|
|
#include "toolcontext.h"
|
|
|
|
#include <limits.h>
|
|
#include <fcntl.h>
|
|
#include <unistd.h>
|
|
#include <sys/mman.h>
|
|
#include <sys/time.h>
|
|
#include <sys/resource.h>
|
|
|
|
#ifndef DEVMAPPER_SUPPORT
|
|
|
|
void memlock_inc_daemon(struct cmd_context *cmd)
|
|
{
|
|
return;
|
|
}
|
|
|
|
void memlock_dec_daemon(struct cmd_context *cmd)
|
|
{
|
|
return;
|
|
}
|
|
|
|
void critical_section_inc(struct cmd_context *cmd, const char *reason)
|
|
{
|
|
return;
|
|
}
|
|
|
|
void critical_section_dec(struct cmd_context *cmd, const char *reason)
|
|
{
|
|
return;
|
|
}
|
|
|
|
int critical_section(void)
|
|
{
|
|
return 0;
|
|
}
|
|
void memlock_init(struct cmd_context *cmd)
|
|
{
|
|
return;
|
|
}
|
|
|
|
void memlock_unlock(struct cmd_context *cmd)
|
|
{
|
|
return;
|
|
}
|
|
|
|
void memlock_reset(void)
|
|
{
|
|
return;
|
|
}
|
|
|
|
#else /* DEVMAPPER_SUPPORT */
|
|
|
|
static size_t _size_stack;
|
|
static size_t _size_malloc_tmp;
|
|
static size_t _size_malloc = 2000000;
|
|
|
|
static void *_malloc_mem = NULL;
|
|
static int _mem_locked = 0;
|
|
static int _critical_section = 0;
|
|
static int _memlock_count_daemon = 0;
|
|
static int _priority;
|
|
static int _default_priority;
|
|
|
|
/* list of maps, that are unconditionaly ignored */
|
|
static const char * const _ignore_maps[] = {
|
|
"[vdso]",
|
|
"[vsyscall]",
|
|
"[vectors]",
|
|
};
|
|
|
|
/* default blacklist for maps */
|
|
static const char * const _blacklist_maps[] = {
|
|
"locale/locale-archive",
|
|
"/LC_MESSAGES/",
|
|
"gconv/gconv-modules.cache",
|
|
"/libblkid.so.", /* not using lzma during mlock (selinux) */
|
|
"/liblzma.so.", /* not using lzma during mlock (selinux) */
|
|
"/libncurses.so.", /* not using ncurses during mlock */
|
|
"/libpcre.so.", /* not using pcre during mlock (selinux) */
|
|
"/libreadline.so.", /* not using readline during mlock */
|
|
"/libselinux.so.", /* not using selinux during mlock */
|
|
"/libsepol.so.", /* not using sepol during mlock */
|
|
"/libtinfo.so.", /* not using tinfo during mlock */
|
|
"/libuuid.so.", /* not using uuid during mlock (blkid) */
|
|
"/libdl-", /* not using dlopen,dlsym during mlock */
|
|
/* "/libdevmapper-event.so" */
|
|
};
|
|
|
|
typedef enum { LVM_MLOCK, LVM_MUNLOCK } lvmlock_t;
|
|
|
|
static unsigned _use_mlockall;
|
|
static int _maps_fd;
|
|
static size_t _maps_len = 8192; /* Initial buffer size for reading /proc/self/maps */
|
|
static char *_maps_buffer;
|
|
static char _procselfmaps[PATH_MAX] = "";
|
|
#define SELF_MAPS "/self/maps"
|
|
|
|
static size_t _mstats; /* statistic for maps locking */
|
|
|
|
static void _touch_memory(void *mem, size_t size)
|
|
{
|
|
size_t pagesize = lvm_getpagesize();
|
|
char *pos = mem;
|
|
char *end = pos + size - sizeof(long);
|
|
|
|
while (pos < end) {
|
|
*(long *) pos = 1;
|
|
pos += pagesize;
|
|
}
|
|
}
|
|
|
|
static void _allocate_memory(void)
|
|
{
|
|
void *stack_mem, *temp_malloc_mem;
|
|
struct rlimit limit;
|
|
|
|
/* Check if we could preallocate requested stack */
|
|
if ((getrlimit (RLIMIT_STACK, &limit) == 0) &&
|
|
((_size_stack * 2) < limit.rlim_cur) &&
|
|
((stack_mem = alloca(_size_stack))))
|
|
_touch_memory(stack_mem, _size_stack);
|
|
/* FIXME else warn user setting got ignored */
|
|
|
|
if ((temp_malloc_mem = malloc(_size_malloc_tmp)))
|
|
_touch_memory(temp_malloc_mem, _size_malloc_tmp);
|
|
|
|
if ((_malloc_mem = malloc(_size_malloc)))
|
|
_touch_memory(_malloc_mem, _size_malloc);
|
|
|
|
free(temp_malloc_mem);
|
|
}
|
|
|
|
static void _release_memory(void)
|
|
{
|
|
free(_malloc_mem);
|
|
}
|
|
|
|
/*
|
|
* mlock/munlock memory areas from /proc/self/maps
|
|
* format described in kernel/Documentation/filesystem/proc.txt
|
|
*/
|
|
static int _maps_line(const struct dm_config_node *cn, lvmlock_t lock,
|
|
const char *line, size_t *mstats)
|
|
{
|
|
const struct dm_config_value *cv;
|
|
long from, to;
|
|
int pos;
|
|
unsigned i;
|
|
char fr, fw, fx, fp;
|
|
size_t sz;
|
|
const char *lock_str = (lock == LVM_MLOCK) ? "mlock" : "munlock";
|
|
|
|
if (sscanf(line, "%lx-%lx %c%c%c%c%n",
|
|
&from, &to, &fr, &fw, &fx, &fp, &pos) != 6) {
|
|
log_error("Failed to parse maps line: %s", line);
|
|
return 0;
|
|
}
|
|
|
|
/* Select readable maps */
|
|
if (fr != 'r') {
|
|
log_debug_mem("%s area unreadable %s : Skipping.", lock_str, line);
|
|
return 1;
|
|
}
|
|
|
|
/* always ignored areas */
|
|
for (i = 0; i < DM_ARRAY_SIZE(_ignore_maps); ++i)
|
|
if (strstr(line + pos, _ignore_maps[i])) {
|
|
log_debug_mem("%s ignore filter '%s' matches '%s': Skipping.",
|
|
lock_str, _ignore_maps[i], line);
|
|
return 1;
|
|
}
|
|
|
|
sz = to - from;
|
|
if (!cn) {
|
|
/* If no blacklist configured, use an internal set */
|
|
for (i = 0; i < DM_ARRAY_SIZE(_blacklist_maps); ++i)
|
|
if (strstr(line + pos, _blacklist_maps[i])) {
|
|
log_debug_mem("%s default filter '%s' matches '%s': Skipping.",
|
|
lock_str, _blacklist_maps[i], line);
|
|
return 1;
|
|
}
|
|
} else {
|
|
for (cv = cn->v; cv; cv = cv->next) {
|
|
if ((cv->type != DM_CFG_STRING) || !cv->v.str[0])
|
|
continue;
|
|
if (strstr(line + pos, cv->v.str)) {
|
|
log_debug_mem("%s_filter '%s' matches '%s': Skipping.",
|
|
lock_str, cv->v.str, line);
|
|
return 1;
|
|
}
|
|
}
|
|
}
|
|
|
|
#ifdef VALGRIND_POOL
|
|
/*
|
|
* Valgrind is continually eating memory while executing code
|
|
* so we need to deactivate check of locked memory size
|
|
*/
|
|
sz -= sz; /* = 0, but avoids getting warning about dead assigment */
|
|
|
|
#endif
|
|
*mstats += sz;
|
|
log_debug_mem("%s %10ldKiB %12lx - %12lx %c%c%c%c%s", lock_str,
|
|
((long)sz + 1023) / 1024, from, to, fr, fw, fx, fp, line + pos);
|
|
|
|
if (lock == LVM_MLOCK) {
|
|
if (mlock((const void*)from, sz) < 0) {
|
|
log_sys_error("mlock", line);
|
|
return 0;
|
|
}
|
|
} else {
|
|
if (munlock((const void*)from, sz) < 0) {
|
|
log_sys_error("munlock", line);
|
|
return 0;
|
|
}
|
|
}
|
|
|
|
return 1;
|
|
}
|
|
|
|
static int _memlock_maps(struct cmd_context *cmd, lvmlock_t lock, size_t *mstats)
|
|
{
|
|
const struct dm_config_node *cn;
|
|
char *line, *line_end;
|
|
size_t len;
|
|
ssize_t n;
|
|
int ret = 1;
|
|
|
|
if (_use_mlockall) {
|
|
#ifdef MCL_CURRENT
|
|
if (lock == LVM_MLOCK) {
|
|
if (mlockall(MCL_CURRENT | MCL_FUTURE)) {
|
|
log_sys_error("mlockall", "");
|
|
return 0;
|
|
}
|
|
} else {
|
|
if (munlockall()) {
|
|
log_sys_error("munlockall", "");
|
|
return 0;
|
|
}
|
|
}
|
|
return 1;
|
|
#else
|
|
return 0;
|
|
#endif
|
|
}
|
|
|
|
/* Force libc.mo load */
|
|
if (lock == LVM_MLOCK)
|
|
(void)strerror(0);
|
|
/* Reset statistic counters */
|
|
*mstats = 0;
|
|
|
|
/* read mapping into a single memory chunk without reallocation
|
|
* in the middle of reading maps file */
|
|
for (len = 0;;) {
|
|
if (!_maps_buffer || len >= _maps_len) {
|
|
if (_maps_buffer)
|
|
_maps_len *= 2;
|
|
if (!(line = dm_realloc(_maps_buffer, _maps_len))) {
|
|
log_error("Allocation of maps buffer failed.");
|
|
return 0;
|
|
}
|
|
_maps_buffer = line;
|
|
}
|
|
if (lseek(_maps_fd, 0, SEEK_SET))
|
|
log_sys_error("lseek", _procselfmaps);
|
|
for (len = 0 ; len < _maps_len; len += n) {
|
|
if (!(n = read(_maps_fd, _maps_buffer + len, _maps_len - len)))
|
|
break; /* EOF */
|
|
if (n == -1) {
|
|
log_sys_error("read", _procselfmaps);
|
|
return 0;
|
|
}
|
|
}
|
|
if (len < _maps_len) { /* fits in buffer */
|
|
_maps_buffer[len] = '\0';
|
|
break;
|
|
}
|
|
}
|
|
|
|
line = _maps_buffer;
|
|
cn = find_config_tree_node(cmd, activation_mlock_filter_CFG, NULL);
|
|
|
|
while ((line_end = strchr(line, '\n'))) {
|
|
*line_end = '\0'; /* remove \n */
|
|
if (!_maps_line(cn, lock, line, mstats))
|
|
ret = 0;
|
|
line = line_end + 1;
|
|
}
|
|
|
|
log_debug_mem("%socked %ld bytes",
|
|
(lock == LVM_MLOCK) ? "L" : "Unl", (long)*mstats);
|
|
|
|
return ret;
|
|
}
|
|
|
|
/* Stop memory getting swapped out */
|
|
static void _lock_mem(struct cmd_context *cmd)
|
|
{
|
|
_allocate_memory();
|
|
|
|
/*
|
|
* For daemon we need to use mlockall()
|
|
* so even future adition of thread which may not even use lvm lib
|
|
* will not block memory locked thread
|
|
* Note: assuming _memlock_count_daemon is updated before _memlock_count
|
|
*/
|
|
_use_mlockall = _memlock_count_daemon ? 1 :
|
|
find_config_tree_bool(cmd, activation_use_mlockall_CFG, NULL);
|
|
|
|
if (!_use_mlockall) {
|
|
if (!*_procselfmaps &&
|
|
dm_snprintf(_procselfmaps, sizeof(_procselfmaps),
|
|
"%s" SELF_MAPS, cmd->proc_dir) < 0) {
|
|
log_error("proc_dir too long");
|
|
return;
|
|
}
|
|
|
|
if (!(_maps_fd = open(_procselfmaps, O_RDONLY))) {
|
|
log_sys_error("open", _procselfmaps);
|
|
return;
|
|
}
|
|
}
|
|
|
|
log_very_verbose("Locking memory");
|
|
if (!_memlock_maps(cmd, LVM_MLOCK, &_mstats))
|
|
stack;
|
|
|
|
errno = 0;
|
|
if (((_priority = getpriority(PRIO_PROCESS, 0)) == -1) && errno)
|
|
log_sys_error("getpriority", "");
|
|
else
|
|
if (setpriority(PRIO_PROCESS, 0, _default_priority))
|
|
log_error("setpriority %d failed: %s",
|
|
_default_priority, strerror(errno));
|
|
}
|
|
|
|
static void _unlock_mem(struct cmd_context *cmd)
|
|
{
|
|
size_t unlock_mstats;
|
|
|
|
log_very_verbose("Unlocking memory");
|
|
|
|
if (!_memlock_maps(cmd, LVM_MUNLOCK, &unlock_mstats))
|
|
stack;
|
|
|
|
if (!_use_mlockall) {
|
|
if (close(_maps_fd))
|
|
log_sys_error("close", _procselfmaps);
|
|
dm_free(_maps_buffer);
|
|
_maps_buffer = NULL;
|
|
if (_mstats < unlock_mstats) {
|
|
if ((_mstats + lvm_getpagesize()) < unlock_mstats)
|
|
log_error(INTERNAL_ERROR
|
|
"Reserved memory (%ld) not enough: used %ld. Increase activation/reserved_memory?",
|
|
(long)_mstats, (long)unlock_mstats);
|
|
else
|
|
/* FIXME Believed due to incorrect use of yes_no_prompt while locks held */
|
|
log_debug_mem("Suppressed internal error: Maps lock %ld < unlock %ld, a one-page difference.",
|
|
(long)_mstats, (long)unlock_mstats);
|
|
}
|
|
}
|
|
|
|
if (setpriority(PRIO_PROCESS, 0, _priority))
|
|
log_error("setpriority %u failed: %s", _priority,
|
|
strerror(errno));
|
|
_release_memory();
|
|
}
|
|
|
|
static void _lock_mem_if_needed(struct cmd_context *cmd)
|
|
{
|
|
log_debug_mem("Lock: Memlock counters: locked:%d critical:%d daemon:%d suspended:%d",
|
|
_mem_locked, _critical_section, _memlock_count_daemon, dm_get_suspended_counter());
|
|
if (!_mem_locked &&
|
|
((_critical_section + _memlock_count_daemon) == 1)) {
|
|
_mem_locked = 1;
|
|
_lock_mem(cmd);
|
|
}
|
|
}
|
|
|
|
static void _unlock_mem_if_possible(struct cmd_context *cmd)
|
|
{
|
|
log_debug_mem("Unlock: Memlock counters: locked:%d critical:%d daemon:%d suspended:%d",
|
|
_mem_locked, _critical_section, _memlock_count_daemon, dm_get_suspended_counter());
|
|
if (_mem_locked &&
|
|
!_critical_section &&
|
|
!_memlock_count_daemon) {
|
|
_unlock_mem(cmd);
|
|
_mem_locked = 0;
|
|
}
|
|
}
|
|
|
|
void critical_section_inc(struct cmd_context *cmd, const char *reason)
|
|
{
|
|
/*
|
|
* Profiles are loaded on-demand so make sure that before
|
|
* entering the critical section all needed profiles are
|
|
* loaded to avoid the disk access later.
|
|
*/
|
|
(void) load_pending_profiles(cmd);
|
|
|
|
if (!_critical_section) {
|
|
_critical_section = 1;
|
|
log_debug_mem("Entering critical section (%s).", reason);
|
|
}
|
|
|
|
_lock_mem_if_needed(cmd);
|
|
}
|
|
|
|
void critical_section_dec(struct cmd_context *cmd, const char *reason)
|
|
{
|
|
if (_critical_section && !dm_get_suspended_counter()) {
|
|
_critical_section = 0;
|
|
log_debug_mem("Leaving critical section (%s).", reason);
|
|
}
|
|
}
|
|
|
|
int critical_section(void)
|
|
{
|
|
return _critical_section;
|
|
}
|
|
|
|
/*
|
|
* The memlock_*_daemon functions will force the mlockall() call that we need
|
|
* to stay in memory, but they will have no effect on device scans (unlike
|
|
* normal critical_section_inc/dec). Memory is kept locked as long as either
|
|
* of critical_section or memlock_daemon is in effect.
|
|
*/
|
|
|
|
void memlock_inc_daemon(struct cmd_context *cmd)
|
|
{
|
|
++_memlock_count_daemon;
|
|
if (_memlock_count_daemon == 1 && _critical_section > 0)
|
|
log_error(INTERNAL_ERROR "_memlock_inc_daemon used in critical section.");
|
|
log_debug_mem("memlock_count_daemon inc to %d", _memlock_count_daemon);
|
|
_lock_mem_if_needed(cmd);
|
|
}
|
|
|
|
void memlock_dec_daemon(struct cmd_context *cmd)
|
|
{
|
|
if (!_memlock_count_daemon)
|
|
log_error(INTERNAL_ERROR "_memlock_count_daemon has dropped below 0.");
|
|
--_memlock_count_daemon;
|
|
log_debug_mem("memlock_count_daemon dec to %d", _memlock_count_daemon);
|
|
if (!_memlock_count_daemon && _critical_section && _mem_locked) {
|
|
log_error("Unlocking daemon memory in critical section.");
|
|
_unlock_mem(cmd);
|
|
_mem_locked = 0;
|
|
}
|
|
_unlock_mem_if_possible(cmd);
|
|
}
|
|
|
|
void memlock_init(struct cmd_context *cmd)
|
|
{
|
|
/* When threaded, caller already limited stack size so just use the default. */
|
|
_size_stack = 1024ULL * (cmd->threaded ? DEFAULT_RESERVED_STACK :
|
|
find_config_tree_int(cmd, activation_reserved_stack_CFG, NULL));
|
|
_size_malloc_tmp = find_config_tree_int(cmd, activation_reserved_memory_CFG, NULL) * 1024ULL;
|
|
_default_priority = find_config_tree_int(cmd, activation_process_priority_CFG, NULL);
|
|
}
|
|
|
|
void memlock_reset(void)
|
|
{
|
|
log_debug_mem("memlock reset.");
|
|
_mem_locked = 0;
|
|
_critical_section = 0;
|
|
_memlock_count_daemon = 0;
|
|
}
|
|
|
|
void memlock_unlock(struct cmd_context *cmd)
|
|
{
|
|
_unlock_mem_if_possible(cmd);
|
|
}
|
|
|
|
#endif
|