2019-05-28 19:57:20 +03:00
|
|
|
// SPDX-License-Identifier: GPL-2.0-only
|
2006-01-18 12:30:29 +03:00
|
|
|
/******************************************************************************
|
|
|
|
*******************************************************************************
|
|
|
|
**
|
|
|
|
** Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
|
2011-11-02 23:30:58 +04:00
|
|
|
** Copyright (C) 2004-2011 Red Hat, Inc. All rights reserved.
|
2006-01-18 12:30:29 +03:00
|
|
|
**
|
|
|
|
**
|
|
|
|
*******************************************************************************
|
|
|
|
******************************************************************************/
|
|
|
|
|
|
|
|
#include "dlm_internal.h"
|
|
|
|
#include "lockspace.h"
|
|
|
|
#include "member.h"
|
|
|
|
#include "dir.h"
|
|
|
|
#include "ast.h"
|
|
|
|
#include "recover.h"
|
|
|
|
#include "lowcomms.h"
|
|
|
|
#include "lock.h"
|
|
|
|
#include "requestqueue.h"
|
|
|
|
#include "recoverd.h"
|
|
|
|
|
|
|
|
|
|
|
|
/* If the start for which we're re-enabling locking (seq) has been superseded
|
2007-09-28 00:53:38 +04:00
|
|
|
by a newer stop (ls_recover_seq), we need to leave locking disabled.
|
|
|
|
|
|
|
|
We suspend dlm_recv threads here to avoid the race where dlm_recv a) sees
|
|
|
|
locking stopped and b) adds a message to the requestqueue, but dlm_recoverd
|
|
|
|
enables locking and clears the requestqueue between a and b. */
|
2006-01-18 12:30:29 +03:00
|
|
|
|
|
|
|
static int enable_locking(struct dlm_ls *ls, uint64_t seq)
|
|
|
|
{
|
|
|
|
int error = -EINTR;
|
|
|
|
|
2007-09-28 00:53:38 +04:00
|
|
|
down_write(&ls->ls_recv_active);
|
|
|
|
|
2006-01-18 12:30:29 +03:00
|
|
|
spin_lock(&ls->ls_recover_lock);
|
|
|
|
if (ls->ls_recover_seq == seq) {
|
|
|
|
set_bit(LSFL_RUNNING, &ls->ls_flags);
|
2007-09-28 00:53:38 +04:00
|
|
|
/* unblocks processes waiting to enter the dlm */
|
2006-01-18 12:30:29 +03:00
|
|
|
up_write(&ls->ls_in_recovery);
|
2012-08-02 20:08:21 +04:00
|
|
|
clear_bit(LSFL_RECOVER_LOCK, &ls->ls_flags);
|
2006-01-18 12:30:29 +03:00
|
|
|
error = 0;
|
|
|
|
}
|
|
|
|
spin_unlock(&ls->ls_recover_lock);
|
2007-09-28 00:53:38 +04:00
|
|
|
|
|
|
|
up_write(&ls->ls_recv_active);
|
2006-01-18 12:30:29 +03:00
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
|
|
|
|
{
|
|
|
|
unsigned long start;
|
|
|
|
int error, neg = 0;
|
|
|
|
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover %llu", (unsigned long long)rv->seq);
|
2006-01-18 12:30:29 +03:00
|
|
|
|
2006-01-20 11:47:07 +03:00
|
|
|
mutex_lock(&ls->ls_recoverd_active);
|
2006-01-18 12:30:29 +03:00
|
|
|
|
2011-04-05 22:16:24 +04:00
|
|
|
dlm_callback_suspend(ls);
|
2006-01-18 12:30:29 +03:00
|
|
|
|
2012-05-10 19:18:07 +04:00
|
|
|
dlm_clear_toss(ls);
|
2006-01-18 12:30:29 +03:00
|
|
|
|
|
|
|
/*
|
2008-01-16 22:02:31 +03:00
|
|
|
* This list of root rsb's will be the basis of most of the recovery
|
|
|
|
* routines.
|
2006-01-18 12:30:29 +03:00
|
|
|
*/
|
|
|
|
|
2008-01-16 22:02:31 +03:00
|
|
|
dlm_create_root_list(ls);
|
2006-01-18 12:30:29 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Add or remove nodes from the lockspace's ls_nodes list.
|
|
|
|
*/
|
|
|
|
|
|
|
|
error = dlm_recover_members(ls, rv, &neg);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_members error %d", error);
|
2006-01-18 12:30:29 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
2011-10-14 21:34:58 +04:00
|
|
|
|
2012-05-10 19:18:07 +04:00
|
|
|
dlm_recover_dir_nodeid(ls);
|
|
|
|
|
|
|
|
ls->ls_recover_dir_sent_res = 0;
|
|
|
|
ls->ls_recover_dir_sent_msg = 0;
|
dlm: fixes for nodir mode
The "nodir" mode (statically assign master nodes instead
of using the resource directory) has always been highly
experimental, and never seriously used. This commit
fixes a number of problems, making nodir much more usable.
- Major change to recovery: recover all locks and restart
all in-progress operations after recovery. In some
cases it's not possible to know which in-progess locks
to recover, so recover all. (Most require recovery
in nodir mode anyway since rehashing changes most
master nodes.)
- Change the way nodir mode is enabled, from a command
line mount arg passed through gfs2, into a sysfs
file managed by dlm_controld, consistent with the
other config settings.
- Allow recovering MSTCPY locks on an rsb that has not
yet been turned into a master copy.
- Ignore RCOM_LOCK and RCOM_LOCK_REPLY recovery messages
from a previous, aborted recovery cycle. Base this
on the local recovery status not being in the state
where any nodes should be sending LOCK messages for the
current recovery cycle.
- Hold rsb lock around dlm_purge_mstcpy_locks() because it
may run concurrently with dlm_recover_master_copy().
- Maintain highbast on process-copy lkb's (in addition to
the master as is usual), because the lkb can switch
back and forth between being a master and being a
process copy as the master node changes in recovery.
- When recovering MSTCPY locks, flag rsb's that have
non-empty convert or waiting queues for granting
at the end of recovery. (Rename flag from LOCKS_PURGED
to RECOVER_GRANT and similar for the recovery function,
because it's not only resources with purged locks
that need grant a grant attempt.)
- Replace a couple of unnecessary assertion panics with
error messages.
Signed-off-by: David Teigland <teigland@redhat.com>
2012-04-27 00:54:29 +04:00
|
|
|
ls->ls_recover_locks_in = 0;
|
|
|
|
|
2011-10-14 21:34:58 +04:00
|
|
|
dlm_set_recover_status(ls, DLM_RS_NODES);
|
|
|
|
|
|
|
|
error = dlm_recover_members_wait(ls);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_members_wait error %d", error);
|
2011-10-14 21:34:58 +04:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2006-01-18 12:30:29 +03:00
|
|
|
start = jiffies;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Rebuild our own share of the directory by collecting from all other
|
|
|
|
* nodes their master rsb names that hash to us.
|
|
|
|
*/
|
|
|
|
|
|
|
|
error = dlm_recover_directory(ls);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_directory error %d", error);
|
2006-01-18 12:30:29 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2011-10-14 21:34:58 +04:00
|
|
|
dlm_set_recover_status(ls, DLM_RS_DIR);
|
2006-01-18 12:30:29 +03:00
|
|
|
|
|
|
|
error = dlm_recover_directory_wait(ls);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_directory_wait error %d", error);
|
2006-01-18 12:30:29 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_directory %u out %u messages",
|
2012-05-10 19:18:07 +04:00
|
|
|
ls->ls_recover_dir_sent_res, ls->ls_recover_dir_sent_msg);
|
|
|
|
|
2006-01-18 12:30:29 +03:00
|
|
|
/*
|
|
|
|
* We may have outstanding operations that are waiting for a reply from
|
|
|
|
* a failed node. Mark these to be resent after recovery. Unlock and
|
|
|
|
* cancel ops can just be completed.
|
|
|
|
*/
|
|
|
|
|
|
|
|
dlm_recover_waiters_pre(ls);
|
|
|
|
|
|
|
|
error = dlm_recovery_stopped(ls);
|
|
|
|
if (error)
|
|
|
|
goto fail;
|
|
|
|
|
|
|
|
if (neg || dlm_no_directory(ls)) {
|
|
|
|
/*
|
|
|
|
* Clear lkb's for departed nodes.
|
|
|
|
*/
|
|
|
|
|
dlm: fixes for nodir mode
The "nodir" mode (statically assign master nodes instead
of using the resource directory) has always been highly
experimental, and never seriously used. This commit
fixes a number of problems, making nodir much more usable.
- Major change to recovery: recover all locks and restart
all in-progress operations after recovery. In some
cases it's not possible to know which in-progess locks
to recover, so recover all. (Most require recovery
in nodir mode anyway since rehashing changes most
master nodes.)
- Change the way nodir mode is enabled, from a command
line mount arg passed through gfs2, into a sysfs
file managed by dlm_controld, consistent with the
other config settings.
- Allow recovering MSTCPY locks on an rsb that has not
yet been turned into a master copy.
- Ignore RCOM_LOCK and RCOM_LOCK_REPLY recovery messages
from a previous, aborted recovery cycle. Base this
on the local recovery status not being in the state
where any nodes should be sending LOCK messages for the
current recovery cycle.
- Hold rsb lock around dlm_purge_mstcpy_locks() because it
may run concurrently with dlm_recover_master_copy().
- Maintain highbast on process-copy lkb's (in addition to
the master as is usual), because the lkb can switch
back and forth between being a master and being a
process copy as the master node changes in recovery.
- When recovering MSTCPY locks, flag rsb's that have
non-empty convert or waiting queues for granting
at the end of recovery. (Rename flag from LOCKS_PURGED
to RECOVER_GRANT and similar for the recovery function,
because it's not only resources with purged locks
that need grant a grant attempt.)
- Replace a couple of unnecessary assertion panics with
error messages.
Signed-off-by: David Teigland <teigland@redhat.com>
2012-04-27 00:54:29 +04:00
|
|
|
dlm_recover_purge(ls);
|
2006-01-18 12:30:29 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Get new master nodeid's for rsb's that were mastered on
|
|
|
|
* departed nodes.
|
|
|
|
*/
|
|
|
|
|
|
|
|
error = dlm_recover_masters(ls);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_masters error %d", error);
|
2006-01-18 12:30:29 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Send our locks on remastered rsb's to the new masters.
|
|
|
|
*/
|
|
|
|
|
|
|
|
error = dlm_recover_locks(ls);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_locks error %d", error);
|
2006-01-18 12:30:29 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2011-10-14 21:34:58 +04:00
|
|
|
dlm_set_recover_status(ls, DLM_RS_LOCKS);
|
|
|
|
|
2006-01-18 12:30:29 +03:00
|
|
|
error = dlm_recover_locks_wait(ls);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_locks_wait error %d", error);
|
2006-01-18 12:30:29 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_locks %u in",
|
dlm: fixes for nodir mode
The "nodir" mode (statically assign master nodes instead
of using the resource directory) has always been highly
experimental, and never seriously used. This commit
fixes a number of problems, making nodir much more usable.
- Major change to recovery: recover all locks and restart
all in-progress operations after recovery. In some
cases it's not possible to know which in-progess locks
to recover, so recover all. (Most require recovery
in nodir mode anyway since rehashing changes most
master nodes.)
- Change the way nodir mode is enabled, from a command
line mount arg passed through gfs2, into a sysfs
file managed by dlm_controld, consistent with the
other config settings.
- Allow recovering MSTCPY locks on an rsb that has not
yet been turned into a master copy.
- Ignore RCOM_LOCK and RCOM_LOCK_REPLY recovery messages
from a previous, aborted recovery cycle. Base this
on the local recovery status not being in the state
where any nodes should be sending LOCK messages for the
current recovery cycle.
- Hold rsb lock around dlm_purge_mstcpy_locks() because it
may run concurrently with dlm_recover_master_copy().
- Maintain highbast on process-copy lkb's (in addition to
the master as is usual), because the lkb can switch
back and forth between being a master and being a
process copy as the master node changes in recovery.
- When recovering MSTCPY locks, flag rsb's that have
non-empty convert or waiting queues for granting
at the end of recovery. (Rename flag from LOCKS_PURGED
to RECOVER_GRANT and similar for the recovery function,
because it's not only resources with purged locks
that need grant a grant attempt.)
- Replace a couple of unnecessary assertion panics with
error messages.
Signed-off-by: David Teigland <teigland@redhat.com>
2012-04-27 00:54:29 +04:00
|
|
|
ls->ls_recover_locks_in);
|
|
|
|
|
2006-01-18 12:30:29 +03:00
|
|
|
/*
|
|
|
|
* Finalize state in master rsb's now that all locks can be
|
|
|
|
* checked. This includes conversion resolution and lvb
|
|
|
|
* settings.
|
|
|
|
*/
|
|
|
|
|
|
|
|
dlm_recover_rsbs(ls);
|
2006-10-31 20:56:01 +03:00
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* Other lockspace members may be going through the "neg" steps
|
|
|
|
* while also adding us to the lockspace, in which case they'll
|
2006-11-01 18:31:48 +03:00
|
|
|
* be doing the recover_locks (RS_LOCKS) barrier.
|
2006-10-31 20:56:01 +03:00
|
|
|
*/
|
|
|
|
dlm_set_recover_status(ls, DLM_RS_LOCKS);
|
2006-11-01 18:31:48 +03:00
|
|
|
|
|
|
|
error = dlm_recover_locks_wait(ls);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_locks_wait error %d", error);
|
2006-11-01 18:31:48 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
2006-01-18 12:30:29 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
dlm_release_root_list(ls);
|
|
|
|
|
2006-11-27 20:31:22 +03:00
|
|
|
/*
|
|
|
|
* Purge directory-related requests that are saved in requestqueue.
|
|
|
|
* All dir requests from before recovery are invalid now due to the dir
|
|
|
|
* rebuild and will be resent by the requesting nodes.
|
|
|
|
*/
|
|
|
|
|
|
|
|
dlm_purge_requestqueue(ls);
|
|
|
|
|
2006-01-18 12:30:29 +03:00
|
|
|
dlm_set_recover_status(ls, DLM_RS_DONE);
|
2011-10-14 21:34:58 +04:00
|
|
|
|
2006-01-18 12:30:29 +03:00
|
|
|
error = dlm_recover_done_wait(ls);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_done_wait error %d", error);
|
2006-01-18 12:30:29 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
|
|
|
dlm_clear_members_gone(ls);
|
|
|
|
|
2007-05-18 17:59:31 +04:00
|
|
|
dlm_adjust_timeouts(ls);
|
|
|
|
|
2011-04-05 22:16:24 +04:00
|
|
|
dlm_callback_resume(ls);
|
|
|
|
|
2006-01-18 12:30:29 +03:00
|
|
|
error = enable_locking(ls, rv->seq);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "enable_locking error %d", error);
|
2006-01-18 12:30:29 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
|
|
|
error = dlm_process_requestqueue(ls);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_process_requestqueue error %d", error);
|
2006-01-18 12:30:29 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
|
|
|
error = dlm_recover_waiters_post(ls);
|
|
|
|
if (error) {
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover_waiters_post error %d", error);
|
2006-01-18 12:30:29 +03:00
|
|
|
goto fail;
|
|
|
|
}
|
|
|
|
|
dlm: fixes for nodir mode
The "nodir" mode (statically assign master nodes instead
of using the resource directory) has always been highly
experimental, and never seriously used. This commit
fixes a number of problems, making nodir much more usable.
- Major change to recovery: recover all locks and restart
all in-progress operations after recovery. In some
cases it's not possible to know which in-progess locks
to recover, so recover all. (Most require recovery
in nodir mode anyway since rehashing changes most
master nodes.)
- Change the way nodir mode is enabled, from a command
line mount arg passed through gfs2, into a sysfs
file managed by dlm_controld, consistent with the
other config settings.
- Allow recovering MSTCPY locks on an rsb that has not
yet been turned into a master copy.
- Ignore RCOM_LOCK and RCOM_LOCK_REPLY recovery messages
from a previous, aborted recovery cycle. Base this
on the local recovery status not being in the state
where any nodes should be sending LOCK messages for the
current recovery cycle.
- Hold rsb lock around dlm_purge_mstcpy_locks() because it
may run concurrently with dlm_recover_master_copy().
- Maintain highbast on process-copy lkb's (in addition to
the master as is usual), because the lkb can switch
back and forth between being a master and being a
process copy as the master node changes in recovery.
- When recovering MSTCPY locks, flag rsb's that have
non-empty convert or waiting queues for granting
at the end of recovery. (Rename flag from LOCKS_PURGED
to RECOVER_GRANT and similar for the recovery function,
because it's not only resources with purged locks
that need grant a grant attempt.)
- Replace a couple of unnecessary assertion panics with
error messages.
Signed-off-by: David Teigland <teigland@redhat.com>
2012-04-27 00:54:29 +04:00
|
|
|
dlm_recover_grant(ls);
|
2006-01-18 12:30:29 +03:00
|
|
|
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover %llu generation %u done: %u ms",
|
2011-11-02 23:30:58 +04:00
|
|
|
(unsigned long long)rv->seq, ls->ls_generation,
|
2006-01-18 12:30:29 +03:00
|
|
|
jiffies_to_msecs(jiffies - start));
|
2006-01-20 11:47:07 +03:00
|
|
|
mutex_unlock(&ls->ls_recoverd_active);
|
2006-01-18 12:30:29 +03:00
|
|
|
|
2011-11-02 23:30:58 +04:00
|
|
|
dlm_lsop_recover_done(ls);
|
2006-01-18 12:30:29 +03:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
fail:
|
|
|
|
dlm_release_root_list(ls);
|
2014-02-14 21:54:44 +04:00
|
|
|
log_rinfo(ls, "dlm_recover %llu error %d",
|
2006-11-29 17:33:48 +03:00
|
|
|
(unsigned long long)rv->seq, error);
|
2006-01-20 11:47:07 +03:00
|
|
|
mutex_unlock(&ls->ls_recoverd_active);
|
2006-01-18 12:30:29 +03:00
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
2006-10-31 20:56:08 +03:00
|
|
|
/* The dlm_ls_start() that created the rv we take here may already have been
|
|
|
|
stopped via dlm_ls_stop(); in that case we need to leave the RECOVERY_STOP
|
|
|
|
flag set. */
|
|
|
|
|
2006-01-18 12:30:29 +03:00
|
|
|
static void do_ls_recovery(struct dlm_ls *ls)
|
|
|
|
{
|
|
|
|
struct dlm_recover *rv = NULL;
|
|
|
|
|
|
|
|
spin_lock(&ls->ls_recover_lock);
|
|
|
|
rv = ls->ls_recover_args;
|
|
|
|
ls->ls_recover_args = NULL;
|
2006-10-31 20:56:08 +03:00
|
|
|
if (rv && ls->ls_recover_seq == rv->seq)
|
2012-08-02 20:08:21 +04:00
|
|
|
clear_bit(LSFL_RECOVER_STOP, &ls->ls_flags);
|
2006-01-18 12:30:29 +03:00
|
|
|
spin_unlock(&ls->ls_recover_lock);
|
|
|
|
|
|
|
|
if (rv) {
|
|
|
|
ls_recover(ls, rv);
|
2011-11-02 23:30:58 +04:00
|
|
|
kfree(rv->nodes);
|
2006-01-18 12:30:29 +03:00
|
|
|
kfree(rv);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int dlm_recoverd(void *arg)
|
|
|
|
{
|
|
|
|
struct dlm_ls *ls;
|
|
|
|
|
|
|
|
ls = dlm_find_lockspace_local(arg);
|
2006-08-24 23:47:20 +04:00
|
|
|
if (!ls) {
|
|
|
|
log_print("dlm_recoverd: no lockspace %p", arg);
|
|
|
|
return -1;
|
|
|
|
}
|
2006-01-18 12:30:29 +03:00
|
|
|
|
2012-08-02 20:08:21 +04:00
|
|
|
down_write(&ls->ls_in_recovery);
|
|
|
|
set_bit(LSFL_RECOVER_LOCK, &ls->ls_flags);
|
|
|
|
wake_up(&ls->ls_recover_lock_wait);
|
|
|
|
|
2017-09-12 11:56:30 +03:00
|
|
|
while (1) {
|
|
|
|
/*
|
|
|
|
* We call kthread_should_stop() after set_current_state().
|
|
|
|
* This is because it works correctly if kthread_stop() is
|
|
|
|
* called just before set_current_state().
|
|
|
|
*/
|
2006-01-18 12:30:29 +03:00
|
|
|
set_current_state(TASK_INTERRUPTIBLE);
|
2017-09-12 11:56:30 +03:00
|
|
|
if (kthread_should_stop()) {
|
|
|
|
set_current_state(TASK_RUNNING);
|
|
|
|
break;
|
|
|
|
}
|
2012-08-02 20:08:21 +04:00
|
|
|
if (!test_bit(LSFL_RECOVER_WORK, &ls->ls_flags) &&
|
2017-09-25 10:47:50 +03:00
|
|
|
!test_bit(LSFL_RECOVER_DOWN, &ls->ls_flags)) {
|
|
|
|
if (kthread_should_stop())
|
|
|
|
break;
|
2006-01-18 12:30:29 +03:00
|
|
|
schedule();
|
2017-09-25 10:47:50 +03:00
|
|
|
}
|
2006-01-18 12:30:29 +03:00
|
|
|
set_current_state(TASK_RUNNING);
|
|
|
|
|
2012-08-02 20:08:21 +04:00
|
|
|
if (test_and_clear_bit(LSFL_RECOVER_DOWN, &ls->ls_flags)) {
|
|
|
|
down_write(&ls->ls_in_recovery);
|
|
|
|
set_bit(LSFL_RECOVER_LOCK, &ls->ls_flags);
|
|
|
|
wake_up(&ls->ls_recover_lock_wait);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (test_and_clear_bit(LSFL_RECOVER_WORK, &ls->ls_flags))
|
2006-01-18 12:30:29 +03:00
|
|
|
do_ls_recovery(ls);
|
|
|
|
}
|
|
|
|
|
2012-08-02 20:08:21 +04:00
|
|
|
if (test_bit(LSFL_RECOVER_LOCK, &ls->ls_flags))
|
|
|
|
up_write(&ls->ls_in_recovery);
|
|
|
|
|
2006-01-18 12:30:29 +03:00
|
|
|
dlm_put_lockspace(ls);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
int dlm_recoverd_start(struct dlm_ls *ls)
|
|
|
|
{
|
|
|
|
struct task_struct *p;
|
|
|
|
int error = 0;
|
|
|
|
|
|
|
|
p = kthread_run(dlm_recoverd, ls, "dlm_recoverd");
|
|
|
|
if (IS_ERR(p))
|
|
|
|
error = PTR_ERR(p);
|
|
|
|
else
|
|
|
|
ls->ls_recoverd_task = p;
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
void dlm_recoverd_stop(struct dlm_ls *ls)
|
|
|
|
{
|
|
|
|
kthread_stop(ls->ls_recoverd_task);
|
|
|
|
}
|
|
|
|
|
|
|
|
void dlm_recoverd_suspend(struct dlm_ls *ls)
|
|
|
|
{
|
2006-08-09 02:06:07 +04:00
|
|
|
wake_up(&ls->ls_wait_general);
|
2006-01-20 11:47:07 +03:00
|
|
|
mutex_lock(&ls->ls_recoverd_active);
|
2006-01-18 12:30:29 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
void dlm_recoverd_resume(struct dlm_ls *ls)
|
|
|
|
{
|
2006-01-20 11:47:07 +03:00
|
|
|
mutex_unlock(&ls->ls_recoverd_active);
|
2006-01-18 12:30:29 +03:00
|
|
|
}
|
|
|
|
|