| Message ID | 5518C165.10106@huawei.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Alex, On 03/30/2015 11:22 AM, alex chen wrote: > If ocfs2 lockres has not been initialized before calling ocfs2_dlm_lock, > the lock won't be dropped and then will lead umount hung. The case is > described below: > > ocfs2_mknod > ocfs2_mknod_locked > __ocfs2_mknod_locked > ocfs2_journal_access_di > Failed because of -ENOMEM or other reasons, the inode lockres > has not been initialized yet. If failed here, is OCFS2_I(inode)->ip_inode_lockres initialized? If not how can you break __ocfs2_cluster_lock with the following condition? if (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) Thanks, Junxiao. > > iput(inode) > ocfs2_evict_inode > ocfs2_delete_inode > ocfs2_inode_lock > ocfs2_inode_lock_full_nested > __ocfs2_cluster_lock > Succeeds and allocates a new dlm lockres. > ocfs2_clear_inode > ocfs2_open_unlock > ocfs2_drop_inode_locks > ocfs2_drop_lock > Since lockres has not been initialized, the lock > can't be dropped and the lockres can't be > migrated, thus umount will hang forever. > > Signed-off-by: Alex Chen <alex.chen@huawei.com> > Reviewed-by: Joseph Qi <joseph.qi@huawei.com> > Reviewed-by: joyce.xue <xuejiufei@huawei.com> > > --- > fs/ocfs2/dlmglue.c | 5 +++++ > 1 file changed, 5 insertions(+) > > diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c > index 11849a4..8b23aa2 100644 > --- a/fs/ocfs2/dlmglue.c > +++ b/fs/ocfs2/dlmglue.c > @@ -1391,6 +1391,11 @@ static int __ocfs2_cluster_lock(struct ocfs2_super *osb, > int noqueue_attempted = 0; > int dlm_locked = 0; > > + if (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) { > + mlog_errno(-EINVAL); > + return -EINVAL; > + } > + > ocfs2_init_mask_waiter(&mw); > > if (lockres->l_ops->flags & LOCK_TYPE_USES_LVB) >
Hi Junxiao, On 2015/4/16 15:28, Junxiao Bi wrote: > Hi Alex, > > On 03/30/2015 11:22 AM, alex chen wrote: >> If ocfs2 lockres has not been initialized before calling ocfs2_dlm_lock, >> the lock won't be dropped and then will lead umount hung. The case is >> described below: >> >> ocfs2_mknod >> ocfs2_mknod_locked >> __ocfs2_mknod_locked >> ocfs2_journal_access_di >> Failed because of -ENOMEM or other reasons, the inode lockres >> has not been initialized yet. > > If failed here, is OCFS2_I(inode)->ip_inode_lockres initialized? If not The OCFS2_I(inode)->ip_inode_lockres is initialized as follows: __ocfs2_mknod_locked ocfs2_populate_inode ocfs2_inode_lock_res_init ocfs2_lock_res_init_common So if ocfs2_journal_access_di is failed, the ip_inode_lockres will not be initialized. In this situation, we should not allocate a new dlm lockres through calling ocfs2_dlm_lock() in __ocfs2_cluster_lock(), otherwise it will lead umount hung. So we need bread __ocfs2_cluster_lock() if the inode lockres is not be initialized, that is the condition (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) is TRUE. > how can you break __ocfs2_cluster_lock with the following condition? > > if (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) > > Thanks, > Junxiao. > >> >> iput(inode) >> ocfs2_evict_inode >> ocfs2_delete_inode >> ocfs2_inode_lock >> ocfs2_inode_lock_full_nested >> __ocfs2_cluster_lock >> Succeeds and allocates a new dlm lockres. >> ocfs2_clear_inode >> ocfs2_open_unlock >> ocfs2_drop_inode_locks >> ocfs2_drop_lock >> Since lockres has not been initialized, the lock >> can't be dropped and the lockres can't be >> migrated, thus umount will hang forever. >> >> Signed-off-by: Alex Chen <alex.chen@huawei.com> >> Reviewed-by: Joseph Qi <joseph.qi@huawei.com> >> Reviewed-by: joyce.xue <xuejiufei@huawei.com> >> >> --- >> fs/ocfs2/dlmglue.c | 5 +++++ >> 1 file changed, 5 insertions(+) >> >> diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c >> index 11849a4..8b23aa2 100644 >> --- a/fs/ocfs2/dlmglue.c >> +++ b/fs/ocfs2/dlmglue.c >> @@ -1391,6 +1391,11 @@ static int __ocfs2_cluster_lock(struct ocfs2_super *osb, >> int noqueue_attempted = 0; >> int dlm_locked = 0; >> >> + if (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) { >> + mlog_errno(-EINVAL); >> + return -EINVAL; >> + } >> + >> ocfs2_init_mask_waiter(&mw); >> >> if (lockres->l_ops->flags & LOCK_TYPE_USES_LVB) >> > > > . >
On 04/21/2015 10:54 AM, alex chen wrote: > Hi Junxiao, > > On 2015/4/16 15:28, Junxiao Bi wrote: >> Hi Alex, >> >> On 03/30/2015 11:22 AM, alex chen wrote: >>> If ocfs2 lockres has not been initialized before calling ocfs2_dlm_lock, >>> the lock won't be dropped and then will lead umount hung. The case is >>> described below: >>> >>> ocfs2_mknod >>> ocfs2_mknod_locked >>> __ocfs2_mknod_locked >>> ocfs2_journal_access_di >>> Failed because of -ENOMEM or other reasons, the inode lockres >>> has not been initialized yet. >> >> If failed here, is OCFS2_I(inode)->ip_inode_lockres initialized? If not > > The OCFS2_I(inode)->ip_inode_lockres is initialized as follows: > __ocfs2_mknod_locked > ocfs2_populate_inode > ocfs2_inode_lock_res_init > ocfs2_lock_res_init_common > So if ocfs2_journal_access_di is failed, the ip_inode_lockres will not be > initialized. > In this situation, we should not allocate a new dlm lockres through calling > ocfs2_dlm_lock() in __ocfs2_cluster_lock(), otherwise it will lead umount > hung. So we need bread __ocfs2_cluster_lock() if the inode lockres is not > be initialized, that is the condition > (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) is TRUE. Looks good. Just didn't remind that inode_info is init once when it is allocated from "ocfs2_inode_cache". lockres->l_flags is init there. Thanks, Junxiao. > >> how can you break __ocfs2_cluster_lock with the following condition? >> >> if (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) >> >> Thanks, >> Junxiao. >> >>> >>> iput(inode) >>> ocfs2_evict_inode >>> ocfs2_delete_inode >>> ocfs2_inode_lock >>> ocfs2_inode_lock_full_nested >>> __ocfs2_cluster_lock >>> Succeeds and allocates a new dlm lockres. >>> ocfs2_clear_inode >>> ocfs2_open_unlock >>> ocfs2_drop_inode_locks >>> ocfs2_drop_lock >>> Since lockres has not been initialized, the lock >>> can't be dropped and the lockres can't be >>> migrated, thus umount will hang forever. >>> >>> Signed-off-by: Alex Chen <alex.chen@huawei.com> >>> Reviewed-by: Joseph Qi <joseph.qi@huawei.com> >>> Reviewed-by: joyce.xue <xuejiufei@huawei.com> >>> >>> --- >>> fs/ocfs2/dlmglue.c | 5 +++++ >>> 1 file changed, 5 insertions(+) >>> >>> diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c >>> index 11849a4..8b23aa2 100644 >>> --- a/fs/ocfs2/dlmglue.c >>> +++ b/fs/ocfs2/dlmglue.c >>> @@ -1391,6 +1391,11 @@ static int __ocfs2_cluster_lock(struct ocfs2_super *osb, >>> int noqueue_attempted = 0; >>> int dlm_locked = 0; >>> >>> + if (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) { >>> + mlog_errno(-EINVAL); >>> + return -EINVAL; >>> + } >>> + >>> ocfs2_init_mask_waiter(&mw); >>> >>> if (lockres->l_ops->flags & LOCK_TYPE_USES_LVB) >>> >> >> >> . >> >
On 2015/4/22 9:46, Junxiao Bi wrote: > On 04/21/2015 10:54 AM, alex chen wrote: >> Hi Junxiao, >> >> On 2015/4/16 15:28, Junxiao Bi wrote: >>> Hi Alex, >>> >>> On 03/30/2015 11:22 AM, alex chen wrote: >>>> If ocfs2 lockres has not been initialized before calling ocfs2_dlm_lock, >>>> the lock won't be dropped and then will lead umount hung. The case is >>>> described below: >>>> >>>> ocfs2_mknod >>>> ocfs2_mknod_locked >>>> __ocfs2_mknod_locked >>>> ocfs2_journal_access_di >>>> Failed because of -ENOMEM or other reasons, the inode lockres >>>> has not been initialized yet. >>> >>> If failed here, is OCFS2_I(inode)->ip_inode_lockres initialized? If not >> >> The OCFS2_I(inode)->ip_inode_lockres is initialized as follows: >> __ocfs2_mknod_locked >> ocfs2_populate_inode >> ocfs2_inode_lock_res_init >> ocfs2_lock_res_init_common >> So if ocfs2_journal_access_di is failed, the ip_inode_lockres will not be >> initialized. >> In this situation, we should not allocate a new dlm lockres through calling >> ocfs2_dlm_lock() in __ocfs2_cluster_lock(), otherwise it will lead umount >> hung. So we need bread __ocfs2_cluster_lock() if the inode lockres is not >> be initialized, that is the condition >> (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) is TRUE. > > Looks good. Just didn't remind that inode_info is init once when it is > allocated from "ocfs2_inode_cache". lockres->l_flags is init there. > > Thanks, > Junxiao. > Thank you for your remind. If the subject is "check if lockres->l_flags be initialized to OCFS2_LOCK_INITIALIZED before calling ocfs2_dlm_lock" will be better. >> >>> how can you break __ocfs2_cluster_lock with the following condition? >>> >>> if (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) >>> >>> Thanks, >>> Junxiao. >>> >>>> >>>> iput(inode) >>>> ocfs2_evict_inode >>>> ocfs2_delete_inode >>>> ocfs2_inode_lock >>>> ocfs2_inode_lock_full_nested >>>> __ocfs2_cluster_lock >>>> Succeeds and allocates a new dlm lockres. >>>> ocfs2_clear_inode >>>> ocfs2_open_unlock >>>> ocfs2_drop_inode_locks >>>> ocfs2_drop_lock >>>> Since lockres has not been initialized, the lock >>>> can't be dropped and the lockres can't be >>>> migrated, thus umount will hang forever. >>>> >>>> Signed-off-by: Alex Chen <alex.chen@huawei.com> >>>> Reviewed-by: Joseph Qi <joseph.qi@huawei.com> >>>> Reviewed-by: joyce.xue <xuejiufei@huawei.com> >>>> >>>> --- >>>> fs/ocfs2/dlmglue.c | 5 +++++ >>>> 1 file changed, 5 insertions(+) >>>> >>>> diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c >>>> index 11849a4..8b23aa2 100644 >>>> --- a/fs/ocfs2/dlmglue.c >>>> +++ b/fs/ocfs2/dlmglue.c >>>> @@ -1391,6 +1391,11 @@ static int __ocfs2_cluster_lock(struct ocfs2_super *osb, >>>> int noqueue_attempted = 0; >>>> int dlm_locked = 0; >>>> >>>> + if (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) { >>>> + mlog_errno(-EINVAL); >>>> + return -EINVAL; >>>> + } >>>> + >>>> ocfs2_init_mask_waiter(&mw); >>>> >>>> if (lockres->l_ops->flags & LOCK_TYPE_USES_LVB) >>>> >>> >>> >>> . >>> >> > > > . >
diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c index 11849a4..8b23aa2 100644 --- a/fs/ocfs2/dlmglue.c +++ b/fs/ocfs2/dlmglue.c @@ -1391,6 +1391,11 @@ static int __ocfs2_cluster_lock(struct ocfs2_super *osb, int noqueue_attempted = 0; int dlm_locked = 0; + if (!(lockres->l_flags & OCFS2_LOCK_INITIALIZED)) { + mlog_errno(-EINVAL); + return -EINVAL; + } + ocfs2_init_mask_waiter(&mw); if (lockres->l_ops->flags & LOCK_TYPE_USES_LVB)
If ocfs2 lockres has not been initialized before calling ocfs2_dlm_lock, the lock won't be dropped and then will lead umount hung. The case is described below: ocfs2_mknod ocfs2_mknod_locked __ocfs2_mknod_locked ocfs2_journal_access_di Failed because of -ENOMEM or other reasons, the inode lockres has not been initialized yet. iput(inode) ocfs2_evict_inode ocfs2_delete_inode ocfs2_inode_lock ocfs2_inode_lock_full_nested __ocfs2_cluster_lock Succeeds and allocates a new dlm lockres. ocfs2_clear_inode ocfs2_open_unlock ocfs2_drop_inode_locks ocfs2_drop_lock Since lockres has not been initialized, the lock can't be dropped and the lockres can't be migrated, thus umount will hang forever. Signed-off-by: Alex Chen <alex.chen@huawei.com> Reviewed-by: Joseph Qi <joseph.qi@huawei.com> Reviewed-by: joyce.xue <xuejiufei@huawei.com> --- fs/ocfs2/dlmglue.c | 5 +++++ 1 file changed, 5 insertions(+)