diff options
author | piaojun <piaojun@huawei.com> | 2016-08-02 14:02:19 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2016-08-02 17:31:41 -0400 |
commit | ee8f7fcbe638b07e8d1c3dc98e8be35e56306d05 (patch) | |
tree | 25b25d9131e4a07e700302c01a5f3924d9c8ae2b /fs/ocfs2/dlm/dlmthread.c | |
parent | 309e91911daede6adde0364f489e69909c3f6894 (diff) | |
download | linux-ee8f7fcbe638b07e8d1c3dc98e8be35e56306d05.tar.gz |
ocfs2/dlm: continue to purge recovery lockres when recovery master goes down
We found a dlm-blocked situation caused by continuous breakdown of
recovery masters described below. To solve this problem, we should
purge recovery lock once detecting recovery master goes down.
N3 N2 N1(reco master)
go down
pick up recovery lock and
begin recoverying for N2
go down
pick up recovery
lock failed, then
purge it:
dlm_purge_lockres
->DROPPING_REF is set
send deref to N1 failed,
recovery lock is not purged
find N1 go down, begin
recoverying for N1, but
blocked in dlm_do_recovery
as DROPPING_REF is set:
dlm_do_recovery
->dlm_pick_recovery_master
->dlmlock
->dlm_get_lock_resource
->__dlm_wait_on_lockres_flags(tmpres,
DLM_LOCK_RES_DROPPING_REF);
Fixes: 8c0343968163 ("ocfs2/dlm: clear DROPPING_REF flag when the master goes down")
Link: http://lkml.kernel.org/r/578453AF.8030404@huawei.com
Signed-off-by: Jun Piao <piaojun@huawei.com>
Reviewed-by: Joseph Qi <joseph.qi@huawei.com>
Reviewed-by: Jiufei Xue <xuejiufei@huawei.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Diffstat (limited to 'fs/ocfs2/dlm/dlmthread.c')
-rw-r--r-- | fs/ocfs2/dlm/dlmthread.c | 52 |
1 files changed, 48 insertions, 4 deletions
diff --git a/fs/ocfs2/dlm/dlmthread.c b/fs/ocfs2/dlm/dlmthread.c index ce397229acc04e..838a06d4066a6e 100644 --- a/fs/ocfs2/dlm/dlmthread.c +++ b/fs/ocfs2/dlm/dlmthread.c @@ -160,6 +160,52 @@ void dlm_lockres_calc_usage(struct dlm_ctxt *dlm, spin_unlock(&dlm->spinlock); } +/* + * Do the real purge work: + * unhash the lockres, and + * clear flag DLM_LOCK_RES_DROPPING_REF. + * It requires dlm and lockres spinlock to be taken. + */ +void __dlm_do_purge_lockres(struct dlm_ctxt *dlm, + struct dlm_lock_resource *res) +{ + assert_spin_locked(&dlm->spinlock); + assert_spin_locked(&res->spinlock); + + if (!list_empty(&res->purge)) { + mlog(0, "%s: Removing res %.*s from purgelist\n", + dlm->name, res->lockname.len, res->lockname.name); + list_del_init(&res->purge); + dlm_lockres_put(res); + dlm->purge_count--; + } + + if (!__dlm_lockres_unused(res)) { + mlog(ML_ERROR, "%s: res %.*s in use after deref\n", + dlm->name, res->lockname.len, res->lockname.name); + __dlm_print_one_lock_resource(res); + BUG(); + } + + __dlm_unhash_lockres(dlm, res); + + spin_lock(&dlm->track_lock); + if (!list_empty(&res->tracking)) + list_del_init(&res->tracking); + else { + mlog(ML_ERROR, "%s: Resource %.*s not on the Tracking list\n", + dlm->name, res->lockname.len, res->lockname.name); + __dlm_print_one_lock_resource(res); + } + spin_unlock(&dlm->track_lock); + + /* + * lockres is not in the hash now. drop the flag and wake up + * any processes waiting in dlm_get_lock_resource. + */ + res->state &= ~DLM_LOCK_RES_DROPPING_REF; +} + static void dlm_purge_lockres(struct dlm_ctxt *dlm, struct dlm_lock_resource *res) { @@ -176,10 +222,8 @@ static void dlm_purge_lockres(struct dlm_ctxt *dlm, if (!master) { if (res->state & DLM_LOCK_RES_DROPPING_REF) { - mlog(ML_NOTICE, "%s: res %.*s already in " - "DLM_LOCK_RES_DROPPING_REF state\n", - dlm->name, res->lockname.len, - res->lockname.name); + mlog(ML_NOTICE, "%s: res %.*s already in DLM_LOCK_RES_DROPPING_REF state\n", + dlm->name, res->lockname.len, res->lockname.name); spin_unlock(&res->spinlock); return; } |