Re: [PATCH 2/3] NFSD: Increase the reference of lockowner when coping file_lock

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, 06 Aug 2014 21:35:40 +0800
Kinglong Mee <kinglongmee@xxxxxxxxx> wrote:

> Signed-off-by: Kinglong Mee <kinglongmee@xxxxxxxxx>
> ---
>  fs/nfsd/nfs4state.c | 33 +++++++++++++++++++++++++++++++--
>  1 file changed, 31 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> index 24168ae..07e4b5c 100644
> --- a/fs/nfsd/nfs4state.c
> +++ b/fs/nfsd/nfs4state.c
> @@ -4867,6 +4867,29 @@ nfs4_transform_lock_offset(struct file_lock *lock)
>  		lock->fl_end = OFFSET_MAX;
>  }
>  
> +static inline struct nfs4_lockowner *get_lockowner(struct nfs4_lockowner *lo)
> +{
> +	atomic_inc(&lo->lo_owner.so_count);
> +	return lo;
> +}
> +
> +static void nfsd4_fl_copy_lock(struct file_lock *dst, struct file_lock *src)
> +{
> +	struct nfs4_lockowner *lo = (struct nfs4_lockowner *) src->fl_owner;
> +	get_lockowner(lo);
> +}
> +
> +static void nfsd4_fl_release_lock(struct file_lock *fl)
> +{
> +	struct nfs4_lockowner *lo = (struct nfs4_lockowner *) fl->fl_owner;
> +	nfs4_put_stateowner(&lo->lo_owner);
> +}
> +
> +static const struct file_lock_operations nfsd4_fl_lock_ops = {
> +	.fl_copy_lock = nfsd4_fl_copy_lock,
> +	.fl_release_private = nfsd4_fl_release_lock,
> +};
> +
>  static inline void
>  nfs4_set_lock_denied(struct file_lock *fl, struct nfsd4_lock_denied *deny)
>  {
> @@ -5233,10 +5256,12 @@ nfsd4_lock(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
>  		status = nfserr_openmode;
>  		goto out;
>  	}
> -	file_lock->fl_owner = (fl_owner_t)lock_sop;
> +
> +	file_lock->fl_owner = (fl_owner_t)get_lockowner(lock_sop);
>  	file_lock->fl_pid = current->tgid;
>  	file_lock->fl_file = filp;
>  	file_lock->fl_flags = FL_POSIX | FL_NFSD;
> +	file_lock->fl_ops = &nfsd4_fl_lock_ops;
>  	file_lock->fl_start = lock->lk_offset;
>  	file_lock->fl_end = last_byte_offset(lock->lk_offset, lock->lk_length);
>  	nfs4_transform_lock_offset(file_lock);
> @@ -5399,6 +5424,7 @@ nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
>  	struct nfs4_ol_stateid *stp;
>  	struct file *filp = NULL;
>  	struct file_lock *file_lock = NULL;
> +	struct nfs4_lockowner *lock_sop = NULL;
>  	__be32 status;
>  	int err;
>  	struct nfsd_net *nn = net_generic(SVC_NET(rqstp), nfsd_net_id);
> @@ -5420,6 +5446,8 @@ nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
>  		status = nfserr_lock_range;
>  		goto put_stateid;
>  	}
> +
> +	lock_sop = lockowner(stp->st_stateowner);
>  	file_lock = locks_alloc_lock();
>  	if (!file_lock) {
>  		dprintk("NFSD: %s: unable to allocate lock!\n", __func__);
> @@ -5428,10 +5456,11 @@ nfsd4_locku(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
>  	}
>  	locks_init_lock(file_lock);
>  	file_lock->fl_type = F_UNLCK;
> -	file_lock->fl_owner = (fl_owner_t)lockowner(stp->st_stateowner);
> +	file_lock->fl_owner = (fl_owner_t)get_lockowner(lock_sop);
>  	file_lock->fl_pid = current->tgid;
>  	file_lock->fl_file = filp;
>  	file_lock->fl_flags = FL_POSIX | FL_NFSD;
> +	file_lock->fl_ops = &nfsd4_fl_lock_ops;
>  	file_lock->fl_start = locku->lu_offset;
>  
>  	file_lock->fl_end = last_byte_offset(locku->lu_offset,

(Sorry I didn't respond to these before -- I no longer work for Red
 Hat, so emails to jlayton@xxxxxxxxxx don't get to me).

This looks really wrong. In this case, knfsd is acting as a
(server-side) lock manager, not a filesystem:

        const struct file_lock_operations *fl_ops;      /* Callbacks for filesystems */
        const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */

So really, taking and putting of lockowner references is really the
purview of the lockmanager. If you had a filesystem that set fl_ops and
was exportable, then this will likely break it.

The fact that the file locking code uses struct file_lock for lock
requests and for tracking the locks themselves makes for a horribly
confusing interface. If I were designing this from scratch I would have
made those two distinct structures. Maybe we should do that anyway
sometime.

What you probably ought want to do is to declare some new fl_lmops for
taking and putting lockowner references and add hooks to call them at
the appropriate places.

-- 
Jeff Layton <jlayton@xxxxxxxxxxxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html




[Index of Archives]     [Linux Ext4 Filesystem]     [Union Filesystem]     [Filesystem Testing]     [Ceph Users]     [Ecryptfs]     [AutoFS]     [Kernel Newbies]     [Share Photos]     [Security]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux Cachefs]     [Reiser Filesystem]     [Linux RAID]     [Samba]     [Device Mapper]     [CEPH Development]
  Powered by Linux