Re: [Virtio-fs] [PATCH V2] virtiofsd: Support remote posix locks

[Vivek Goyal <vgoyal@redhat.com>]

On Mon, Jun 10, 2019 at 08:00:26PM +0100, Dr. David Alan Gilbert wrote:

[..]
> >  struct lo_map_elem {
> >  	union {
> >  		struct lo_inode *inode;
> > @@ -86,6 +92,8 @@ struct lo_inode {
> >  	struct lo_key key;
> >  	uint64_t refcount; /* protected by lo->mutex */
> >  	fuse_ino_t fuse_ino;
> > +	pthread_mutex_t plock_mutex;
> > +	GHashTable *posix_locks; /* protected by lo_inode->plock_mutex */
> >  };
> >  
> >  struct lo_cred {
> > @@ -105,6 +113,7 @@ struct lo_data {
> >  	int norace;
> >  	int writeback;
> >  	int flock;
> > +	int posix_lock;
> >  	int xattr;
> >  	const char *source;
> >  	double timeout;
> > @@ -133,6 +142,10 @@ static const struct fuse_opt lo_opts[] =
> >  	  offsetof(struct lo_data, flock), 1 },
> >  	{ "no_flock",
> >  	  offsetof(struct lo_data, flock), 0 },
> > +	{ "posix_lock",
> > +	  offsetof(struct lo_data, posix_lock), 0 },
> 
> Shouldn't that be 1                             ^ ?
> 

Good catch. Havind said that, I am wondering why do we need this
"posix_lock" knob given we enable it by default. I guess only place it
could be useful if user wanted to force a certain behavior without knowing
what's the daemon default. Also defaults could change down the line.

I will post V3.

Vivek