Re: [PATCH v2 02/10] nfsd: Avoid taking state_lock while holding inode lock in nfsd_break_one_deleg

[Jeff Layton <jeff.layton@primarydata.com>]

On Wed, 16 Jul 2014 21:34:41 -0400
"J. Bruce Fields" <bfields@fieldses.org> wrote:

> On Wed, Jul 16, 2014 at 10:31:57AM -0400, Jeff Layton wrote:
> > state_lock is a heavily contended global lock. We don't want to grab
> > that while simultaneously holding the inode->i_lock.
> > 
> > Add a new per-nfs4_file lock that we can use to protect the
> > per-nfs4_file delegation list. Hold that while walking the list in the
> > break_deleg callback and queue the workqueue job for each one.
> > 
> > The workqueue job can then take the state_lock and do the list
> > manipulations without the i_lock being held prior to starting the
> > rpc call.
> 
> Looks fine.
> 
> As a separate issue: does nfs4_state_shutdown net have the same problem
> as destroy_client?  It seems like there should be a cp->dl_time++ there
> too.
> 

I had looked at that before and convinced myself that it wasn't needed,
but now I can't recall why. I think you're correct though -- we ought
to bump it there too. It certainly won't hurt to do so and it might
prevent the same race there.

> Though I'm not sure what the point of that del_recall_lru loop is at
> all.  nfs4_state_destroy_net() destroys each client and its associated
> delegations.  Maybe that function should just be
> 
> 	nfs4_lock_state();
> 	nfs4_client_tracking_exit(net);
> 	nfs4_state_destroy_net(net);
> 	WARN_ON(!list_empty(&nn->del_recall_lru);
> 	nfs4_unlock_state();
> 
> --b.
> 

Yeah, it is odd. I end up relying on that for the "recall_delegations"
fault injector in a later patch in the series, so I'd probably prefer
to leave it as-is for now. We should be able to get rid of it
eventually though.

> > 
> > Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
> > Signed-off-by: Jeff Layton <jlayton@primarydata.com>
> > Reviewed-by: Christoph Hellwig <hch@lst.de>
> > ---
> >  fs/nfsd/nfs4callback.c | 25 ++++++++++++++++++----
> >  fs/nfsd/nfs4state.c    | 58
> > +++++++++++++++++++++++++++++++++-----------------
> > fs/nfsd/state.h        |  4 +++- 3 files changed, 62 insertions(+),
> > 25 deletions(-)
> > 
> > diff --git a/fs/nfsd/nfs4callback.c b/fs/nfsd/nfs4callback.c
> > index 30a71cb46001..a88a93e09d69 100644
> > --- a/fs/nfsd/nfs4callback.c
> > +++ b/fs/nfsd/nfs4callback.c
> > @@ -933,7 +933,7 @@ void nfsd4_shutdown_callback(struct nfs4_client
> > *clp) set_bit(NFSD4_CLIENT_CB_KILL, &clp->cl_flags);
> >  	/*
> >  	 * Note this won't actually result in a null callback;
> > -	 * instead, nfsd4_do_callback_rpc() will detect the killed
> > +	 * instead, nfsd4_run_cb_null() will detect the killed
> >  	 * client, destroy the rpc client, and stop:
> >  	 */
> >  	do_probe_callback(clp);
> > @@ -1011,10 +1011,9 @@ static void nfsd4_process_cb_update(struct
> > nfsd4_callback *cb) run_nfsd4_cb(cb);
> >  }
> >  
> > -void
> > -nfsd4_do_callback_rpc(struct work_struct *w)
> > +static void
> > +nfsd4_run_callback_rpc(struct nfsd4_callback *cb)
> >  {
> > -	struct nfsd4_callback *cb = container_of(w, struct
> > nfsd4_callback, cb_work); struct nfs4_client *clp = cb->cb_clp;
> >  	struct rpc_clnt *clnt;
> >  
> > @@ -1032,6 +1031,24 @@ nfsd4_do_callback_rpc(struct work_struct *w)
> >  			cb->cb_ops, cb);
> >  }
> >  
> > +void
> > +nfsd4_run_cb_null(struct work_struct *w)
> > +{
> > +	struct nfsd4_callback *cb = container_of(w, struct
> > nfsd4_callback,
> > +							cb_work);
> > +	nfsd4_run_callback_rpc(cb);
> > +}
> > +
> > +void
> > +nfsd4_run_cb_recall(struct work_struct *w)
> > +{
> > +	struct nfsd4_callback *cb = container_of(w, struct
> > nfsd4_callback,
> > +							cb_work);
> > +
> > +	nfsd4_prepare_cb_recall(cb->cb_op);
> > +	nfsd4_run_callback_rpc(cb);
> > +}
> > +
> >  void nfsd4_cb_recall(struct nfs4_delegation *dp)
> >  {
> >  	struct nfsd4_callback *cb = &dp->dl_recall;
> > diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> > index 56ea4f12803e..bdf8ac3393bd 100644
> > --- a/fs/nfsd/nfs4state.c
> > +++ b/fs/nfsd/nfs4state.c
> > @@ -254,6 +254,8 @@ static void nfsd4_free_file(struct nfs4_file *f)
> >  static inline void
> >  put_nfs4_file(struct nfs4_file *fi)
> >  {
> > +	might_lock(&state_lock);
> > +
> >  	if (atomic_dec_and_lock(&fi->fi_ref, &state_lock)) {
> >  		hlist_del(&fi->fi_hash);
> >  		spin_unlock(&state_lock);
> > @@ -554,6 +556,8 @@ static void block_delegations(struct knfsd_fh
> > *fh) u32 hash;
> >  	struct bloom_pair *bd = &blocked_delegations;
> >  
> > +	lockdep_assert_held(&state_lock);
> > +
> >  	hash = arch_fast_hash(&fh->fh_base, fh->fh_size, 0);
> >  
> >  	__set_bit(hash&255, bd->set[bd->new]);
> > @@ -592,7 +596,7 @@ alloc_init_deleg(struct nfs4_client *clp,
> > struct nfs4_ol_stateid *stp, struct sv fh_copy_shallow(&dp->dl_fh,
> > &current_fh->fh_handle); dp->dl_time = 0;
> >  	atomic_set(&dp->dl_count, 1);
> > -	INIT_WORK(&dp->dl_recall.cb_work, nfsd4_do_callback_rpc);
> > +	INIT_WORK(&dp->dl_recall.cb_work, nfsd4_run_cb_recall);
> >  	return dp;
> >  }
> >  
> > @@ -640,7 +644,9 @@ hash_delegation_locked(struct nfs4_delegation
> > *dp, struct nfs4_file *fp) lockdep_assert_held(&state_lock);
> >  
> >  	dp->dl_stid.sc_type = NFS4_DELEG_STID;
> > +	spin_lock(&fp->fi_lock);
> >  	list_add(&dp->dl_perfile, &fp->fi_delegations);
> > +	spin_unlock(&fp->fi_lock);
> >  	list_add(&dp->dl_perclnt,
> > &dp->dl_stid.sc_client->cl_delegations); }
> >  
> > @@ -648,14 +654,18 @@ hash_delegation_locked(struct nfs4_delegation
> > *dp, struct nfs4_file *fp) static void
> >  unhash_delegation(struct nfs4_delegation *dp)
> >  {
> > +	struct nfs4_file *fp = dp->dl_file;
> > +
> >  	spin_lock(&state_lock);
> >  	list_del_init(&dp->dl_perclnt);
> > -	list_del_init(&dp->dl_perfile);
> >  	list_del_init(&dp->dl_recall_lru);
> > +	spin_lock(&fp->fi_lock);
> > +	list_del_init(&dp->dl_perfile);
> > +	spin_unlock(&fp->fi_lock);
> >  	spin_unlock(&state_lock);
> > -	if (dp->dl_file) {
> > -		nfs4_put_deleg_lease(dp->dl_file);
> > -		put_nfs4_file(dp->dl_file);
> > +	if (fp) {
> > +		nfs4_put_deleg_lease(fp);
> > +		put_nfs4_file(fp);
> >  		dp->dl_file = NULL;
> >  	}
> >  }
> > @@ -1677,7 +1687,7 @@ static struct nfs4_client
> > *create_client(struct xdr_netobj name,
> > spin_unlock(&nn->client_lock); return NULL;
> >  	}
> > -	INIT_WORK(&clp->cl_cb_null.cb_work, nfsd4_do_callback_rpc);
> > +	INIT_WORK(&clp->cl_cb_null.cb_work, nfsd4_run_cb_null);
> >  	clp->cl_time = get_seconds();
> >  	clear_bit(0, &clp->cl_cb_slot_busy);
> >  	copy_verf(clp, verf);
> > @@ -3079,30 +3089,38 @@ nfs4_share_conflict(struct svc_fh
> > *current_fh, unsigned int deny_type) return ret;
> >  }
> >  
> > -static void nfsd_break_one_deleg(struct nfs4_delegation *dp)
> > +void nfsd4_prepare_cb_recall(struct nfs4_delegation *dp)
> >  {
> >  	struct nfs4_client *clp = dp->dl_stid.sc_client;
> >  	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
> >  
> > -	lockdep_assert_held(&state_lock);
> > -	/* We're assuming the state code never drops its reference
> > -	 * without first removing the lease.  Since we're in this
> > lease
> > -	 * callback (and since the lease code is serialized by the
> > kernel
> > -	 * lock) we know the server hasn't removed the lease yet,
> > we know
> > -	 * it's safe to take a reference: */
> > -	atomic_inc(&dp->dl_count);
> > -
> > +	/*
> > +	 * We can't do this in nfsd_break_deleg_cb because it is
> > +	 * already holding inode->i_lock
> > +	 */
> > +	spin_lock(&state_lock);
> > +	block_delegations(&dp->dl_fh);
> >  	/*
> >  	 * If the dl_time != 0, then we know that it has already
> > been
> >  	 * queued for a lease break. Don't queue it again.
> >  	 */
> >  	if (dp->dl_time == 0) {
> > -		list_add_tail(&dp->dl_recall_lru,
> > &nn->del_recall_lru); dp->dl_time = get_seconds();
> > +		list_add_tail(&dp->dl_recall_lru,
> > &nn->del_recall_lru); }
> > +	spin_unlock(&state_lock);
> > +}
> >  
> > -	block_delegations(&dp->dl_fh);
> > -
> > +static void nfsd_break_one_deleg(struct nfs4_delegation *dp)
> > +{
> > +	/*
> > +	 * We're assuming the state code never drops its reference
> > +	 * without first removing the lease.  Since we're in this
> > lease
> > +	 * callback (and since the lease code is serialized by the
> > kernel
> > +	 * lock) we know the server hasn't removed the lease yet,
> > we know
> > +	 * it's safe to take a reference.
> > +	 */
> > +	atomic_inc(&dp->dl_count);
> >  	nfsd4_cb_recall(dp);
> >  }
> >  
> > @@ -3127,11 +3145,11 @@ static void nfsd_break_deleg_cb(struct
> > file_lock *fl) */
> >  	fl->fl_break_time = 0;
> >  
> > -	spin_lock(&state_lock);
> >  	fp->fi_had_conflict = true;
> > +	spin_lock(&fp->fi_lock);
> >  	list_for_each_entry(dp, &fp->fi_delegations, dl_perfile)
> >  		nfsd_break_one_deleg(dp);
> > -	spin_unlock(&state_lock);
> > +	spin_unlock(&fp->fi_lock);
> >  }
> >  
> >  static
> > diff --git a/fs/nfsd/state.h b/fs/nfsd/state.h
> > index 20857142773f..81b7522e3f67 100644
> > --- a/fs/nfsd/state.h
> > +++ b/fs/nfsd/state.h
> > @@ -436,7 +436,8 @@ extern struct nfs4_client_reclaim
> > *nfsd4_find_reclaim_client(const char *recdir, extern __be32
> > nfs4_check_open_reclaim(clientid_t *clid, struct
> > nfsd4_compound_state *cstate, struct nfsd_net *nn); extern int
> > set_callback_cred(void); -void nfsd4_do_callback_rpc(struct
> > work_struct *w); +void nfsd4_run_cb_null(struct work_struct *w);
> > +void nfsd4_run_cb_recall(struct work_struct *w);
> >  extern void nfsd4_probe_callback(struct nfs4_client *clp);
> >  extern void nfsd4_probe_callback_sync(struct nfs4_client *clp);
> >  extern void nfsd4_change_callback(struct nfs4_client *clp, struct
> > nfs4_cb_conn *); @@ -444,6 +445,7 @@ extern void
> > nfsd4_cb_recall(struct nfs4_delegation *dp); extern int
> > nfsd4_create_callback_queue(void); extern void
> > nfsd4_destroy_callback_queue(void); extern void
> > nfsd4_shutdown_callback(struct nfs4_client *); +extern void
> > nfsd4_prepare_cb_recall(struct nfs4_delegation *dp); extern void
> > nfs4_put_delegation(struct nfs4_delegation *dp); extern struct
> > nfs4_client_reclaim *nfs4_client_to_reclaim(const char *name,
> > struct nfsd_net *nn); -- 
> > 1.9.3
> > 


-- 
Jeff Layton <jlayton@primarydata.com>