[PATCH RFC 0/4] un-deprecate nfsdcld

[PATCH RFC 0/4] un-deprecate nfsdcld

[Scott Mayhew]

When nfsdcld was released, it was quickly deprecated in favor of the
nfsdcltrack usermodehelper, so as to not require another running daemon.
That prevents NFSv4 clients from reclaiming locks from nfsd's running in
containers, since neither nfsdcltrack nor the legacy client tracking
code work in containers.  These patches un-deprecate the use of nfsdcld
for NFSv4 client tracking.

These patches are intended to go alongside some nfs-utils patches that
introduce an enhancement that allows nfsd to "slurp" up the client
records during client tracking initialization and store them internally
in hash table.  This enables nfsd to check whether an NFSv4 client is 
allowed to reclaim without having to do an upcall to nfsdcld.  It also
allows nfsd to decide to end the v4 grace period early if the number of
RECLAIM_COMPLETE operations it has received from "known" clients is
equal to the number of entries in the hash table.  It also allows nfsd
to skip the v4 grace period altogether if it knows there are no clients
allowed to reclaim.

There is a fallback to allow nfsd to continue to work with older nfsdcld
daemons in the event that any are out in the wild (unlikely).
Everything should work fine except nfsd will not be able to exit the
grace period early or skip the grace period altogether.

Scott Mayhew (4):
  nfsd: fix a warning in __cld_pipe_upcall()
  nfsd: make nfs4_client_reclaim use an xdr_netobj instead of a fixed
    char array
  nfsd: un-deprecate nfsdcld
  nfsd: keep a tally of RECLAIM_COMPLETE operations when using nfsdcld

 fs/nfsd/netns.h               |   3 +
 fs/nfsd/nfs4recover.c         | 326 ++++++++++++++++++++++++++++++----
 fs/nfsd/nfs4state.c           |  82 +++++++--
 fs/nfsd/nfsctl.c              |   1 +
 fs/nfsd/state.h               |   8 +-
 include/uapi/linux/nfsd/cld.h |   1 +
 6 files changed, 372 insertions(+), 49 deletions(-)

-- 
2.17.1

[PATCH RFC 1/4] nfsd: fix a warning in __cld_pipe_upcall()

[Scott Mayhew]

__cld_pipe_upcall() emits a "do not call blocking ops when
!TASK_RUNNING" warning due to the dput() call in rpc_queue_upcall().
Fix it by using a completion instead of hand coding the wait.

Signed-off-by: Scott Mayhew <smayhew@redhat.com>
---
 fs/nfsd/nfs4recover.c | 17 ++++++-----------
 1 file changed, 6 insertions(+), 11 deletions(-)

diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
index 9c247fa1e959..5188f9f70c78 100644
--- a/fs/nfsd/nfs4recover.c
+++ b/fs/nfsd/nfs4recover.c
@@ -662,7 +662,7 @@ struct cld_net {
 struct cld_upcall {
 	struct list_head	 cu_list;
 	struct cld_net		*cu_net;
-	struct task_struct	*cu_task;
+	struct completion	 cu_done;
 	struct cld_msg		 cu_msg;
 };
 
@@ -671,23 +671,18 @@ __cld_pipe_upcall(struct rpc_pipe *pipe, struct cld_msg *cmsg)
 {
 	int ret;
 	struct rpc_pipe_msg msg;
+	struct cld_upcall *cup = container_of(cmsg, struct cld_upcall, cu_msg);
 
 	memset(&msg, 0, sizeof(msg));
 	msg.data = cmsg;
 	msg.len = sizeof(*cmsg);
 
-	/*
-	 * Set task state before we queue the upcall. That prevents
-	 * wake_up_process in the downcall from racing with schedule.
-	 */
-	set_current_state(TASK_UNINTERRUPTIBLE);
 	ret = rpc_queue_upcall(pipe, &msg);
 	if (ret < 0) {
-		set_current_state(TASK_RUNNING);
 		goto out;
 	}
 
-	schedule();
+	wait_for_completion(&cup->cu_done);
 
 	if (msg.errno < 0)
 		ret = msg.errno;
@@ -754,7 +749,7 @@ cld_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
 	if (copy_from_user(&cup->cu_msg, src, mlen) != 0)
 		return -EFAULT;
 
-	wake_up_process(cup->cu_task);
+	complete(&cup->cu_done);
 	return mlen;
 }
 
@@ -769,7 +764,7 @@ cld_pipe_destroy_msg(struct rpc_pipe_msg *msg)
 	if (msg->errno >= 0)
 		return;
 
-	wake_up_process(cup->cu_task);
+	complete(&cup->cu_done);
 }
 
 static const struct rpc_pipe_ops cld_upcall_ops = {
@@ -900,7 +895,7 @@ alloc_cld_upcall(struct cld_net *cn)
 			goto restart_search;
 		}
 	}
-	new->cu_task = current;
+	init_completion(&new->cu_done);
 	new->cu_msg.cm_vers = CLD_UPCALL_VERSION;
 	put_unaligned(cn->cn_xid++, &new->cu_msg.cm_xid);
 	new->cu_net = cn;
-- 
2.17.1

Re: [PATCH RFC 1/4] nfsd: fix a warning in __cld_pipe_upcall()

[J. Bruce Fields]

This looks fine, applying now for 4.21.--b.

On Tue, Nov 06, 2018 at 01:35:08PM -0500, Scott Mayhew wrote:
> __cld_pipe_upcall() emits a "do not call blocking ops when
> !TASK_RUNNING" warning due to the dput() call in rpc_queue_upcall().
> Fix it by using a completion instead of hand coding the wait.
> 
> Signed-off-by: Scott Mayhew <smayhew@redhat.com>
> ---
>  fs/nfsd/nfs4recover.c | 17 ++++++-----------
>  1 file changed, 6 insertions(+), 11 deletions(-)
> 
> diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
> index 9c247fa1e959..5188f9f70c78 100644
> --- a/fs/nfsd/nfs4recover.c
> +++ b/fs/nfsd/nfs4recover.c
> @@ -662,7 +662,7 @@ struct cld_net {
>  struct cld_upcall {
>  	struct list_head	 cu_list;
>  	struct cld_net		*cu_net;
> -	struct task_struct	*cu_task;
> +	struct completion	 cu_done;
>  	struct cld_msg		 cu_msg;
>  };
>  
> @@ -671,23 +671,18 @@ __cld_pipe_upcall(struct rpc_pipe *pipe, struct cld_msg *cmsg)
>  {
>  	int ret;
>  	struct rpc_pipe_msg msg;
> +	struct cld_upcall *cup = container_of(cmsg, struct cld_upcall, cu_msg);
>  
>  	memset(&msg, 0, sizeof(msg));
>  	msg.data = cmsg;
>  	msg.len = sizeof(*cmsg);
>  
> -	/*
> -	 * Set task state before we queue the upcall. That prevents
> -	 * wake_up_process in the downcall from racing with schedule.
> -	 */
> -	set_current_state(TASK_UNINTERRUPTIBLE);
>  	ret = rpc_queue_upcall(pipe, &msg);
>  	if (ret < 0) {
> -		set_current_state(TASK_RUNNING);
>  		goto out;
>  	}
>  
> -	schedule();
> +	wait_for_completion(&cup->cu_done);
>  
>  	if (msg.errno < 0)
>  		ret = msg.errno;
> @@ -754,7 +749,7 @@ cld_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
>  	if (copy_from_user(&cup->cu_msg, src, mlen) != 0)
>  		return -EFAULT;
>  
> -	wake_up_process(cup->cu_task);
> +	complete(&cup->cu_done);
>  	return mlen;
>  }
>  
> @@ -769,7 +764,7 @@ cld_pipe_destroy_msg(struct rpc_pipe_msg *msg)
>  	if (msg->errno >= 0)
>  		return;
>  
> -	wake_up_process(cup->cu_task);
> +	complete(&cup->cu_done);
>  }
>  
>  static const struct rpc_pipe_ops cld_upcall_ops = {
> @@ -900,7 +895,7 @@ alloc_cld_upcall(struct cld_net *cn)
>  			goto restart_search;
>  		}
>  	}
> -	new->cu_task = current;
> +	init_completion(&new->cu_done);
>  	new->cu_msg.cm_vers = CLD_UPCALL_VERSION;
>  	put_unaligned(cn->cn_xid++, &new->cu_msg.cm_xid);
>  	new->cu_net = cn;
> -- 
> 2.17.1

[PATCH RFC 2/4] nfsd: make nfs4_client_reclaim use an xdr_netobj instead of a fixed char array

[Scott Mayhew]

This will allow the reclaim_str_hashtbl to store either the recovery
directory names used by the legacy client tracking code or the full
client strings used by the nfsdcld client tracking code.

Signed-off-by: Scott Mayhew <smayhew@redhat.com>
---
 fs/nfsd/nfs4recover.c | 79 +++++++++++++++++++++++++++++++++++++------
 fs/nfsd/nfs4state.c   | 30 ++++++++--------
 fs/nfsd/state.h       |  8 ++---
 3 files changed, 87 insertions(+), 30 deletions(-)

diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
index 5188f9f70c78..b288b14273ff 100644
--- a/fs/nfsd/nfs4recover.c
+++ b/fs/nfsd/nfs4recover.c
@@ -178,6 +178,7 @@ nfsd4_create_clid_dir(struct nfs4_client *clp)
 	struct nfs4_client_reclaim *crp;
 	int status;
 	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
+	struct xdr_netobj name;
 
 	if (test_and_set_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
 		return;
@@ -222,9 +223,18 @@ nfsd4_create_clid_dir(struct nfs4_client *clp)
 	inode_unlock(d_inode(dir));
 	if (status == 0) {
 		if (nn->in_grace) {
-			crp = nfs4_client_to_reclaim(dname, nn);
-			if (crp)
-				crp->cr_clp = clp;
+			name.data = kmemdup(dname, HEXDIR_LEN, GFP_KERNEL);
+			if (name.data) {
+				name.len = HEXDIR_LEN;
+				crp = nfs4_client_to_reclaim(name, nn);
+				if (crp)
+					crp->cr_clp = clp;
+				else
+					kfree(name.data);
+			} else {
+				dprintk("%s: failed to allocate memory for name.data!\n",
+					__func__);
+			}
 		}
 		vfs_fsync(nn->rec_file, 0);
 	} else {
@@ -353,6 +363,7 @@ nfsd4_remove_clid_dir(struct nfs4_client *clp)
 	char dname[HEXDIR_LEN];
 	int status;
 	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
+	struct xdr_netobj name;
 
 	if (!nn->rec_file || !test_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
 		return;
@@ -376,9 +387,17 @@ nfsd4_remove_clid_dir(struct nfs4_client *clp)
 		vfs_fsync(nn->rec_file, 0);
 		if (nn->in_grace) {
 			/* remove reclaim record */
-			crp = nfsd4_find_reclaim_client(dname, nn);
-			if (crp)
-				nfs4_remove_reclaim_record(crp, nn);
+			name.data = kmemdup(dname, HEXDIR_LEN, GFP_KERNEL);
+			if (name.data) {
+				name.len = HEXDIR_LEN;
+				crp = nfsd4_find_reclaim_client(name, nn);
+				kfree(name.data);
+				if (crp)
+					nfs4_remove_reclaim_record(crp, nn);
+			} else {
+				dprintk("%s: failed to allocate memory for name.data!\n",
+					__func__);
+			}
 		}
 	}
 out_drop_write:
@@ -393,14 +412,31 @@ static int
 purge_old(struct dentry *parent, struct dentry *child, struct nfsd_net *nn)
 {
 	int status;
+	struct xdr_netobj name;
 
-	if (nfs4_has_reclaimed_state(child->d_name.name, nn))
+	if (child->d_name.len != HEXDIR_LEN - 1) {
+		printk("%s: illegal name %pd in recovery directory\n",
+				__func__, child);
+		/* Keep trying; maybe the others are OK: */
 		return 0;
+	}
+	name.data = kmemdup_nul(child->d_name.name, child->d_name.len, GFP_KERNEL);
+	if (!name.data) {
+		dprintk("%s: failed to allocate memory for name.data!\n",
+			__func__);
+		goto out;
+	}
+	name.len = HEXDIR_LEN;
+	if (nfs4_has_reclaimed_state(name, nn))
+		goto out_free;
 
 	status = vfs_rmdir(d_inode(parent), child);
 	if (status)
 		printk("failed to remove client recovery directory %pd\n",
 				child);
+out_free:
+	kfree(name.data);
+out:
 	/* Keep trying, success or failure: */
 	return 0;
 }
@@ -430,13 +466,24 @@ nfsd4_recdir_purge_old(struct nfsd_net *nn)
 static int
 load_recdir(struct dentry *parent, struct dentry *child, struct nfsd_net *nn)
 {
+	struct xdr_netobj name;
+
 	if (child->d_name.len != HEXDIR_LEN - 1) {
-		printk("nfsd4: illegal name %pd in recovery directory\n",
-				child);
+		printk("%s: illegal name %pd in recovery directory\n",
+				__func__, child);
 		/* Keep trying; maybe the others are OK: */
 		return 0;
 	}
-	nfs4_client_to_reclaim(child->d_name.name, nn);
+	name.data = kmemdup_nul(child->d_name.name, child->d_name.len, GFP_KERNEL);
+	if (!name.data) {
+		dprintk("%s: failed to allocate memory for name.data!\n",
+			__func__);
+		goto out;
+	}
+	name.len = HEXDIR_LEN;
+	if (!nfs4_client_to_reclaim(name, nn))
+		kfree(name.data);
+out:
 	return 0;
 }
 
@@ -616,6 +663,7 @@ nfsd4_check_legacy_client(struct nfs4_client *clp)
 	char dname[HEXDIR_LEN];
 	struct nfs4_client_reclaim *crp;
 	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
+	struct xdr_netobj name;
 
 	/* did we already find that this client is stable? */
 	if (test_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
@@ -628,13 +676,22 @@ nfsd4_check_legacy_client(struct nfs4_client *clp)
 	}
 
 	/* look for it in the reclaim hashtable otherwise */
-	crp = nfsd4_find_reclaim_client(dname, nn);
+	name.data = kmemdup(dname, HEXDIR_LEN, GFP_KERNEL);
+	if (!name.data) {
+		dprintk("%s: failed to allocate memory for name.data!\n",
+			__func__);
+		goto out_enoent;
+	}
+	name.len = HEXDIR_LEN;
+	crp = nfsd4_find_reclaim_client(name, nn);
+	kfree(name.data);
 	if (crp) {
 		set_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags);
 		crp->cr_clp = clp;
 		return 0;
 	}
 
+out_enoent:
 	return -ENOENT;
 }
 
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index f093fbe47133..777fbb0d2761 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -1058,9 +1058,9 @@ static unsigned int clientid_hashval(u32 id)
 	return id & CLIENT_HASH_MASK;
 }
 
-static unsigned int clientstr_hashval(const char *name)
+static unsigned int clientstr_hashval(struct xdr_netobj name)
 {
-	return opaque_hashval(name, 8) & CLIENT_HASH_MASK;
+	return opaque_hashval(name.data, 8) & CLIENT_HASH_MASK;
 }
 
 /*
@@ -2039,11 +2039,6 @@ compare_blob(const struct xdr_netobj *o1, const struct xdr_netobj *o2)
 	return memcmp(o1->data, o2->data, o1->len);
 }
 
-static int same_name(const char *n1, const char *n2)
-{
-	return 0 == memcmp(n1, n2, HEXDIR_LEN);
-}
-
 static int
 same_verf(nfs4_verifier *v1, nfs4_verifier *v2)
 {
@@ -6449,7 +6444,7 @@ alloc_reclaim(void)
 }
 
 bool
-nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
+nfs4_has_reclaimed_state(struct xdr_netobj name, struct nfsd_net *nn)
 {
 	struct nfs4_client_reclaim *crp;
 
@@ -6459,20 +6454,24 @@ nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
 
 /*
  * failure => all reset bets are off, nfserr_no_grace...
+ *
+ * The caller is responsible for freeing name.data if NULL is returned (it
+ * will be freed in nfs4_remove_reclaim_record in the normal case).
  */
 struct nfs4_client_reclaim *
-nfs4_client_to_reclaim(const char *name, struct nfsd_net *nn)
+nfs4_client_to_reclaim(struct xdr_netobj name, struct nfsd_net *nn)
 {
 	unsigned int strhashval;
 	struct nfs4_client_reclaim *crp;
 
-	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", HEXDIR_LEN, name);
+	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", name.len, name.data);
 	crp = alloc_reclaim();
 	if (crp) {
 		strhashval = clientstr_hashval(name);
 		INIT_LIST_HEAD(&crp->cr_strhash);
 		list_add(&crp->cr_strhash, &nn->reclaim_str_hashtbl[strhashval]);
-		memcpy(crp->cr_recdir, name, HEXDIR_LEN);
+		crp->cr_name.data = name.data;
+		crp->cr_name.len = name.len;
 		crp->cr_clp = NULL;
 		nn->reclaim_str_hashtbl_size++;
 	}
@@ -6483,6 +6482,7 @@ void
 nfs4_remove_reclaim_record(struct nfs4_client_reclaim *crp, struct nfsd_net *nn)
 {
 	list_del(&crp->cr_strhash);
+	kfree(crp->cr_name.data);
 	kfree(crp);
 	nn->reclaim_str_hashtbl_size--;
 }
@@ -6506,16 +6506,16 @@ nfs4_release_reclaim(struct nfsd_net *nn)
 /*
  * called from OPEN, CLAIM_PREVIOUS with a new clientid. */
 struct nfs4_client_reclaim *
-nfsd4_find_reclaim_client(const char *recdir, struct nfsd_net *nn)
+nfsd4_find_reclaim_client(struct xdr_netobj name, struct nfsd_net *nn)
 {
 	unsigned int strhashval;
 	struct nfs4_client_reclaim *crp = NULL;
 
-	dprintk("NFSD: nfs4_find_reclaim_client for recdir %s\n", recdir);
+	dprintk("NFSD: nfs4_find_reclaim_client for name %.*s\n", name.len, name.data);
 
-	strhashval = clientstr_hashval(recdir);
+	strhashval = clientstr_hashval(name);
 	list_for_each_entry(crp, &nn->reclaim_str_hashtbl[strhashval], cr_strhash) {
-		if (same_name(crp->cr_recdir, recdir)) {
+		if (compare_blob(&crp->cr_name, &name) == 0) {
 			return crp;
 		}
 	}
diff --git a/fs/nfsd/state.h b/fs/nfsd/state.h
index 6aacb325b6a0..95dd298c4620 100644
--- a/fs/nfsd/state.h
+++ b/fs/nfsd/state.h
@@ -367,7 +367,7 @@ struct nfs4_client {
 struct nfs4_client_reclaim {
 	struct list_head	cr_strhash;	/* hash by cr_name */
 	struct nfs4_client	*cr_clp;	/* pointer to associated clp */
-	char			cr_recdir[HEXDIR_LEN]; /* recover dir */
+	struct xdr_netobj	cr_name;	/* recovery dir name */
 };
 
 /* A reasonable value for REPLAY_ISIZE was estimated as follows:  
@@ -619,7 +619,7 @@ void nfs4_put_stid(struct nfs4_stid *s);
 void nfs4_inc_and_copy_stateid(stateid_t *dst, struct nfs4_stid *stid);
 void nfs4_remove_reclaim_record(struct nfs4_client_reclaim *, struct nfsd_net *);
 extern void nfs4_release_reclaim(struct nfsd_net *);
-extern struct nfs4_client_reclaim *nfsd4_find_reclaim_client(const char *recdir,
+extern struct nfs4_client_reclaim *nfsd4_find_reclaim_client(struct xdr_netobj name,
 							struct nfsd_net *nn);
 extern __be32 nfs4_check_open_reclaim(clientid_t *clid,
 		struct nfsd4_compound_state *cstate, struct nfsd_net *nn);
@@ -634,9 +634,9 @@ extern void nfsd4_destroy_callback_queue(void);
 extern void nfsd4_shutdown_callback(struct nfs4_client *);
 extern void nfsd4_shutdown_copy(struct nfs4_client *clp);
 extern void nfsd4_prepare_cb_recall(struct nfs4_delegation *dp);
-extern struct nfs4_client_reclaim *nfs4_client_to_reclaim(const char *name,
+extern struct nfs4_client_reclaim *nfs4_client_to_reclaim(struct xdr_netobj name,
 							struct nfsd_net *nn);
-extern bool nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn);
+extern bool nfs4_has_reclaimed_state(struct xdr_netobj name, struct nfsd_net *nn);
 
 struct nfs4_file *find_file(struct knfsd_fh *fh);
 void put_nfs4_file(struct nfs4_file *fi);
-- 
2.17.1

Re: [PATCH RFC 2/4] nfsd: make nfs4_client_reclaim use an xdr_netobj instead of a fixed char array

[J. Bruce Fields]

On Tue, Nov 06, 2018 at 01:35:09PM -0500, Scott Mayhew wrote:
> This will allow the reclaim_str_hashtbl to store either the recovery
> directory names used by the legacy client tracking code or the full
> client strings used by the nfsdcld client tracking code.
> 
> Signed-off-by: Scott Mayhew <smayhew@redhat.com>
> ---
>  fs/nfsd/nfs4recover.c | 79 +++++++++++++++++++++++++++++++++++++------
>  fs/nfsd/nfs4state.c   | 30 ++++++++--------
>  fs/nfsd/state.h       |  8 ++---
>  3 files changed, 87 insertions(+), 30 deletions(-)
> 
> diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
> index 5188f9f70c78..b288b14273ff 100644
> --- a/fs/nfsd/nfs4recover.c
> +++ b/fs/nfsd/nfs4recover.c
> @@ -178,6 +178,7 @@ nfsd4_create_clid_dir(struct nfs4_client *clp)
>  	struct nfs4_client_reclaim *crp;
>  	int status;
>  	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
> +	struct xdr_netobj name;
>  
>  	if (test_and_set_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
>  		return;
> @@ -222,9 +223,18 @@ nfsd4_create_clid_dir(struct nfs4_client *clp)
>  	inode_unlock(d_inode(dir));
>  	if (status == 0) {
>  		if (nn->in_grace) {
> -			crp = nfs4_client_to_reclaim(dname, nn);
> -			if (crp)
> -				crp->cr_clp = clp;
> +			name.data = kmemdup(dname, HEXDIR_LEN, GFP_KERNEL);
> +			if (name.data) {
> +				name.len = HEXDIR_LEN;
> +				crp = nfs4_client_to_reclaim(name, nn);
> +				if (crp)
> +					crp->cr_clp = clp;
> +				else
> +					kfree(name.data);
> +			} else {
> +				dprintk("%s: failed to allocate memory for name.data!\n",
> +					__func__);
> +			}

The usual pattern for error handling is

	try_something
	if failed
		goto error cleanup
	try_something_else
	if failed
		goto error cleanup
		...

and I find that a little easier to follow than this sort of:

	try_something
	if succeeded
		try something_else
		if succeeded
			...

Also, nfsd4_create_clid is getting a little long.

But, I don't know, I haven't tried, if the alternatives don't look like
clear improvements then I can live with it.

The basic idea of the patch looks fine to me.

--b.

>  		}
>  		vfs_fsync(nn->rec_file, 0);
>  	} else {
> @@ -353,6 +363,7 @@ nfsd4_remove_clid_dir(struct nfs4_client *clp)
>  	char dname[HEXDIR_LEN];
>  	int status;
>  	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
> +	struct xdr_netobj name;
>  
>  	if (!nn->rec_file || !test_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
>  		return;
> @@ -376,9 +387,17 @@ nfsd4_remove_clid_dir(struct nfs4_client *clp)
>  		vfs_fsync(nn->rec_file, 0);
>  		if (nn->in_grace) {
>  			/* remove reclaim record */
> -			crp = nfsd4_find_reclaim_client(dname, nn);
> -			if (crp)
> -				nfs4_remove_reclaim_record(crp, nn);
> +			name.data = kmemdup(dname, HEXDIR_LEN, GFP_KERNEL);
> +			if (name.data) {
> +				name.len = HEXDIR_LEN;
> +				crp = nfsd4_find_reclaim_client(name, nn);
> +				kfree(name.data);
> +				if (crp)
> +					nfs4_remove_reclaim_record(crp, nn);
> +			} else {
> +				dprintk("%s: failed to allocate memory for name.data!\n",
> +					__func__);
> +			}
>  		}
>  	}
>  out_drop_write:
> @@ -393,14 +412,31 @@ static int
>  purge_old(struct dentry *parent, struct dentry *child, struct nfsd_net *nn)
>  {
>  	int status;
> +	struct xdr_netobj name;
>  
> -	if (nfs4_has_reclaimed_state(child->d_name.name, nn))
> +	if (child->d_name.len != HEXDIR_LEN - 1) {
> +		printk("%s: illegal name %pd in recovery directory\n",
> +				__func__, child);
> +		/* Keep trying; maybe the others are OK: */
>  		return 0;
> +	}
> +	name.data = kmemdup_nul(child->d_name.name, child->d_name.len, GFP_KERNEL);
> +	if (!name.data) {
> +		dprintk("%s: failed to allocate memory for name.data!\n",
> +			__func__);
> +		goto out;
> +	}
> +	name.len = HEXDIR_LEN;
> +	if (nfs4_has_reclaimed_state(name, nn))
> +		goto out_free;
>  
>  	status = vfs_rmdir(d_inode(parent), child);
>  	if (status)
>  		printk("failed to remove client recovery directory %pd\n",
>  				child);
> +out_free:
> +	kfree(name.data);
> +out:
>  	/* Keep trying, success or failure: */
>  	return 0;
>  }
> @@ -430,13 +466,24 @@ nfsd4_recdir_purge_old(struct nfsd_net *nn)
>  static int
>  load_recdir(struct dentry *parent, struct dentry *child, struct nfsd_net *nn)
>  {
> +	struct xdr_netobj name;
> +
>  	if (child->d_name.len != HEXDIR_LEN - 1) {
> -		printk("nfsd4: illegal name %pd in recovery directory\n",
> -				child);
> +		printk("%s: illegal name %pd in recovery directory\n",
> +				__func__, child);
>  		/* Keep trying; maybe the others are OK: */
>  		return 0;
>  	}
> -	nfs4_client_to_reclaim(child->d_name.name, nn);
> +	name.data = kmemdup_nul(child->d_name.name, child->d_name.len, GFP_KERNEL);
> +	if (!name.data) {
> +		dprintk("%s: failed to allocate memory for name.data!\n",
> +			__func__);
> +		goto out;
> +	}
> +	name.len = HEXDIR_LEN;
> +	if (!nfs4_client_to_reclaim(name, nn))
> +		kfree(name.data);
> +out:
>  	return 0;
>  }
>  
> @@ -616,6 +663,7 @@ nfsd4_check_legacy_client(struct nfs4_client *clp)
>  	char dname[HEXDIR_LEN];
>  	struct nfs4_client_reclaim *crp;
>  	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
> +	struct xdr_netobj name;
>  
>  	/* did we already find that this client is stable? */
>  	if (test_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
> @@ -628,13 +676,22 @@ nfsd4_check_legacy_client(struct nfs4_client *clp)
>  	}
>  
>  	/* look for it in the reclaim hashtable otherwise */
> -	crp = nfsd4_find_reclaim_client(dname, nn);
> +	name.data = kmemdup(dname, HEXDIR_LEN, GFP_KERNEL);
> +	if (!name.data) {
> +		dprintk("%s: failed to allocate memory for name.data!\n",
> +			__func__);
> +		goto out_enoent;
> +	}
> +	name.len = HEXDIR_LEN;
> +	crp = nfsd4_find_reclaim_client(name, nn);
> +	kfree(name.data);
>  	if (crp) {
>  		set_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags);
>  		crp->cr_clp = clp;
>  		return 0;
>  	}
>  
> +out_enoent:
>  	return -ENOENT;
>  }
>  
> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> index f093fbe47133..777fbb0d2761 100644
> --- a/fs/nfsd/nfs4state.c
> +++ b/fs/nfsd/nfs4state.c
> @@ -1058,9 +1058,9 @@ static unsigned int clientid_hashval(u32 id)
>  	return id & CLIENT_HASH_MASK;
>  }
>  
> -static unsigned int clientstr_hashval(const char *name)
> +static unsigned int clientstr_hashval(struct xdr_netobj name)
>  {
> -	return opaque_hashval(name, 8) & CLIENT_HASH_MASK;
> +	return opaque_hashval(name.data, 8) & CLIENT_HASH_MASK;
>  }
>  
>  /*
> @@ -2039,11 +2039,6 @@ compare_blob(const struct xdr_netobj *o1, const struct xdr_netobj *o2)
>  	return memcmp(o1->data, o2->data, o1->len);
>  }
>  
> -static int same_name(const char *n1, const char *n2)
> -{
> -	return 0 == memcmp(n1, n2, HEXDIR_LEN);
> -}
> -
>  static int
>  same_verf(nfs4_verifier *v1, nfs4_verifier *v2)
>  {
> @@ -6449,7 +6444,7 @@ alloc_reclaim(void)
>  }
>  
>  bool
> -nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
> +nfs4_has_reclaimed_state(struct xdr_netobj name, struct nfsd_net *nn)
>  {
>  	struct nfs4_client_reclaim *crp;
>  
> @@ -6459,20 +6454,24 @@ nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn)
>  
>  /*
>   * failure => all reset bets are off, nfserr_no_grace...
> + *
> + * The caller is responsible for freeing name.data if NULL is returned (it
> + * will be freed in nfs4_remove_reclaim_record in the normal case).
>   */
>  struct nfs4_client_reclaim *
> -nfs4_client_to_reclaim(const char *name, struct nfsd_net *nn)
> +nfs4_client_to_reclaim(struct xdr_netobj name, struct nfsd_net *nn)
>  {
>  	unsigned int strhashval;
>  	struct nfs4_client_reclaim *crp;
>  
> -	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", HEXDIR_LEN, name);
> +	dprintk("NFSD nfs4_client_to_reclaim NAME: %.*s\n", name.len, name.data);
>  	crp = alloc_reclaim();
>  	if (crp) {
>  		strhashval = clientstr_hashval(name);
>  		INIT_LIST_HEAD(&crp->cr_strhash);
>  		list_add(&crp->cr_strhash, &nn->reclaim_str_hashtbl[strhashval]);
> -		memcpy(crp->cr_recdir, name, HEXDIR_LEN);
> +		crp->cr_name.data = name.data;
> +		crp->cr_name.len = name.len;
>  		crp->cr_clp = NULL;
>  		nn->reclaim_str_hashtbl_size++;
>  	}
> @@ -6483,6 +6482,7 @@ void
>  nfs4_remove_reclaim_record(struct nfs4_client_reclaim *crp, struct nfsd_net *nn)
>  {
>  	list_del(&crp->cr_strhash);
> +	kfree(crp->cr_name.data);
>  	kfree(crp);
>  	nn->reclaim_str_hashtbl_size--;
>  }
> @@ -6506,16 +6506,16 @@ nfs4_release_reclaim(struct nfsd_net *nn)
>  /*
>   * called from OPEN, CLAIM_PREVIOUS with a new clientid. */
>  struct nfs4_client_reclaim *
> -nfsd4_find_reclaim_client(const char *recdir, struct nfsd_net *nn)
> +nfsd4_find_reclaim_client(struct xdr_netobj name, struct nfsd_net *nn)
>  {
>  	unsigned int strhashval;
>  	struct nfs4_client_reclaim *crp = NULL;
>  
> -	dprintk("NFSD: nfs4_find_reclaim_client for recdir %s\n", recdir);
> +	dprintk("NFSD: nfs4_find_reclaim_client for name %.*s\n", name.len, name.data);
>  
> -	strhashval = clientstr_hashval(recdir);
> +	strhashval = clientstr_hashval(name);
>  	list_for_each_entry(crp, &nn->reclaim_str_hashtbl[strhashval], cr_strhash) {
> -		if (same_name(crp->cr_recdir, recdir)) {
> +		if (compare_blob(&crp->cr_name, &name) == 0) {
>  			return crp;
>  		}
>  	}
> diff --git a/fs/nfsd/state.h b/fs/nfsd/state.h
> index 6aacb325b6a0..95dd298c4620 100644
> --- a/fs/nfsd/state.h
> +++ b/fs/nfsd/state.h
> @@ -367,7 +367,7 @@ struct nfs4_client {
>  struct nfs4_client_reclaim {
>  	struct list_head	cr_strhash;	/* hash by cr_name */
>  	struct nfs4_client	*cr_clp;	/* pointer to associated clp */
> -	char			cr_recdir[HEXDIR_LEN]; /* recover dir */
> +	struct xdr_netobj	cr_name;	/* recovery dir name */
>  };
>  
>  /* A reasonable value for REPLAY_ISIZE was estimated as follows:  
> @@ -619,7 +619,7 @@ void nfs4_put_stid(struct nfs4_stid *s);
>  void nfs4_inc_and_copy_stateid(stateid_t *dst, struct nfs4_stid *stid);
>  void nfs4_remove_reclaim_record(struct nfs4_client_reclaim *, struct nfsd_net *);
>  extern void nfs4_release_reclaim(struct nfsd_net *);
> -extern struct nfs4_client_reclaim *nfsd4_find_reclaim_client(const char *recdir,
> +extern struct nfs4_client_reclaim *nfsd4_find_reclaim_client(struct xdr_netobj name,
>  							struct nfsd_net *nn);
>  extern __be32 nfs4_check_open_reclaim(clientid_t *clid,
>  		struct nfsd4_compound_state *cstate, struct nfsd_net *nn);
> @@ -634,9 +634,9 @@ extern void nfsd4_destroy_callback_queue(void);
>  extern void nfsd4_shutdown_callback(struct nfs4_client *);
>  extern void nfsd4_shutdown_copy(struct nfs4_client *clp);
>  extern void nfsd4_prepare_cb_recall(struct nfs4_delegation *dp);
> -extern struct nfs4_client_reclaim *nfs4_client_to_reclaim(const char *name,
> +extern struct nfs4_client_reclaim *nfs4_client_to_reclaim(struct xdr_netobj name,
>  							struct nfsd_net *nn);
> -extern bool nfs4_has_reclaimed_state(const char *name, struct nfsd_net *nn);
> +extern bool nfs4_has_reclaimed_state(struct xdr_netobj name, struct nfsd_net *nn);
>  
>  struct nfs4_file *find_file(struct knfsd_fh *fh);
>  void put_nfs4_file(struct nfs4_file *fi);
> -- 
> 2.17.1

[PATCH RFC 3/4] nfsd: un-deprecate nfsdcld

[Scott Mayhew]

When nfsdcld was released, it was quickly deprecated in favor of the
nfsdcltrack usermodehelper, so as to not require another running daemon.
That prevents NFSv4 clients from reclaiming locks from nfsd's running in
containers, since neither nfsdcltrack nor the legacy client tracking
code work in containers.

This commit un-deprecates the use of nfsdcld, with one twist: we will
populate the reclaim_str_hashtbl on startup.

During client tracking initialization, do an upcall ("GraceStart") to
nfsdcld to get a list of clients from the database.  nfsdcld will do
one downcall with a status of -EINPROGRESS for each client record in
the database, which in turn will cause an nfs4_client_reclaim to be
added to the reclaim_str_hashtbl.  When complete, nfsdcld will do a
final downcall with a status of 0.

This will save nfsd from having to do an upcall to the daemon during
nfs4_check_open_reclaim() processing.

Even though nfsdcld was quickly deprecated, there is a very small chance
of old nfsdcld daemons running in the wild.  These will respond to the
new "GraceStart" upcall with -EOPNOTSUPP, in which case we will log a
message and fall back to the original nfsdcld tracking ops (now called
nfsd4_cld_tracking_ops_v0).

Signed-off-by: Scott Mayhew <smayhew@redhat.com>
---
 fs/nfsd/nfs4recover.c         | 225 ++++++++++++++++++++++++++++++++--
 include/uapi/linux/nfsd/cld.h |   1 +
 2 files changed, 218 insertions(+), 8 deletions(-)

diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
index b288b14273ff..10887fdce8c2 100644
--- a/fs/nfsd/nfs4recover.c
+++ b/fs/nfsd/nfs4recover.c
@@ -763,6 +763,34 @@ cld_pipe_upcall(struct rpc_pipe *pipe, struct cld_msg *cmsg)
 	return ret;
 }
 
+static ssize_t
+__cld_pipe_inprogress_downcall(const struct cld_msg __user *cmsg,
+		struct nfsd_net *nn)
+{
+	uint8_t cmd;
+	struct xdr_netobj name;
+	uint16_t namelen;
+
+	if (get_user(cmd, &cmsg->cm_cmd)) {
+		dprintk("%s: error when copying cmd from userspace", __func__);
+		return -EFAULT;
+	}
+	if (cmd == Cld_GraceStart) {
+		if (get_user(namelen, &cmsg->cm_u.cm_name.cn_len))
+			return -EFAULT;
+		name.data = memdup_user(&cmsg->cm_u.cm_name.cn_id, namelen);
+		if (IS_ERR_OR_NULL(name.data))
+			return -EFAULT;
+		name.len = namelen;
+		if (!nfs4_client_to_reclaim(name, nn)) {
+			kfree(name.data);
+			return -EFAULT;
+		}
+		return sizeof(*cmsg);
+	}
+	return -EFAULT;
+}
+
 static ssize_t
 cld_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
 {
@@ -772,6 +800,7 @@ cld_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
 	struct nfsd_net *nn = net_generic(file_inode(filp)->i_sb->s_fs_info,
 						nfsd_net_id);
 	struct cld_net *cn = nn->cld_net;
+	int16_t status;
 
 	if (mlen != sizeof(*cmsg)) {
 		dprintk("%s: got %zu bytes, expected %zu\n", __func__, mlen,
@@ -785,13 +814,24 @@ cld_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
 		return -EFAULT;
 	}
 
+	/*
+	 * copy the status so we know whether to remove the upcall from the
+	 * list (for -EINPROGRESS, we just want to make sure the xid is
+	 * valid, not remove the upcall from the list)
+	 */
+	if (get_user(status, &cmsg->cm_status)) {
+		dprintk("%s: error when copying status from userspace", __func__);
+		return -EFAULT;
+	}
+
 	/* walk the list and find corresponding xid */
 	cup = NULL;
 	spin_lock(&cn->cn_lock);
 	list_for_each_entry(tmp, &cn->cn_list, cu_list) {
 		if (get_unaligned(&tmp->cu_msg.cm_xid) == xid) {
 			cup = tmp;
-			list_del_init(&cup->cu_list);
+			if (status != -EINPROGRESS)
+				list_del_init(&cup->cu_list);
 			break;
 		}
 	}
@@ -803,6 +843,9 @@ cld_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
 		return -EINVAL;
 	}
 
+	if (status == -EINPROGRESS)
+		return __cld_pipe_inprogress_downcall(cmsg, nn);
+
 	if (copy_from_user(&cup->cu_msg, src, mlen) != 0)
 		return -EFAULT;
 
@@ -1049,9 +1092,14 @@ nfsd4_cld_remove(struct nfs4_client *clp)
 				"record from stable storage: %d\n", ret);
 }
 
-/* Check for presence of a record, and update its timestamp */
+/*
+ * For older nfsdcld's that do not allow us to "slurp" the clients
+ * from the tracking database during startup.
+ *
+ * Check for presence of a record, and update its timestamp
+ */
 static int
-nfsd4_cld_check(struct nfs4_client *clp)
+nfsd4_cld_check_v0(struct nfs4_client *clp)
 {
 	int ret;
 	struct cld_upcall *cup;
@@ -1084,8 +1132,61 @@ nfsd4_cld_check(struct nfs4_client *clp)
 	return ret;
 }
 
+/*
+ * For newer nfsdcld's that allow us to "slurp" the clients
+ * from the tracking database during startup.
+ *
+ * Check for presence of a record in the reclaim_str_hashtbl
+ */
+static int
+nfsd4_cld_check(struct nfs4_client *clp)
+{
+	struct nfs4_client_reclaim *crp;
+	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
+
+	/* did we already find that this client is stable? */
+	if (test_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
+		return 0;
+
+	/* look for it in the reclaim hashtable otherwise */
+	crp = nfsd4_find_reclaim_client(clp->cl_name, nn);
+	if (crp) {
+		crp->cr_clp = clp;
+		return 0;
+	}
+
+	return -ENOENT;
+}
+
+static int
+nfsd4_cld_grace_start(struct nfsd_net *nn)
+{
+	int ret;
+	struct cld_upcall *cup;
+	struct cld_net *cn = nn->cld_net;
+
+	cup = alloc_cld_upcall(cn);
+	if (!cup) {
+		ret = -ENOMEM;
+		goto out_err;
+	}
+
+	cup->cu_msg.cm_cmd = Cld_GraceStart;
+	ret = cld_pipe_upcall(cn->cn_pipe, &cup->cu_msg);
+	if (!ret)
+		ret = cup->cu_msg.cm_status;
+
+	free_cld_upcall(cup);
+out_err:
+	if (ret)
+		dprintk("%s: Unable to get clients from userspace: %d\n",
+			__func__, ret);
+	return ret;
+}
+
+/* For older nfsdcld's that need cm_gracetime */
 static void
-nfsd4_cld_grace_done(struct nfsd_net *nn)
+nfsd4_cld_grace_done_v0(struct nfsd_net *nn)
 {
 	int ret;
 	struct cld_upcall *cup;
@@ -1109,11 +1210,118 @@ nfsd4_cld_grace_done(struct nfsd_net *nn)
 		printk(KERN_ERR "NFSD: Unable to end grace period: %d\n", ret);
 }
 
-static const struct nfsd4_client_tracking_ops nfsd4_cld_tracking_ops = {
+/*
+ * For newer nfsdcld's that do not need cm_gracetime.  We also need to call
+ * nfs4_release_reclaim() to clear out the reclaim_str_hashtbl.
+ */
+static void
+nfsd4_cld_grace_done(struct nfsd_net *nn)
+{
+	int ret;
+	struct cld_upcall *cup;
+	struct cld_net *cn = nn->cld_net;
+
+	cup = alloc_cld_upcall(cn);
+	if (!cup) {
+		ret = -ENOMEM;
+		goto out_err;
+	}
+
+	cup->cu_msg.cm_cmd = Cld_GraceDone;
+	ret = cld_pipe_upcall(cn->cn_pipe, &cup->cu_msg);
+	if (!ret)
+		ret = cup->cu_msg.cm_status;
+
+	free_cld_upcall(cup);
+out_err:
+	nfs4_release_reclaim(nn);
+	if (ret)
+		printk(KERN_ERR "NFSD: Unable to end grace period: %d\n", ret);
+}
+
+static int
+nfs4_cld_state_init(struct net *net)
+{
+	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
+	int i;
+
+	nn->reclaim_str_hashtbl = kmalloc_array(CLIENT_HASH_SIZE,
+						sizeof(struct list_head),
+						GFP_KERNEL);
+	if (!nn->reclaim_str_hashtbl)
+		return -ENOMEM;
+
+	for (i = 0; i < CLIENT_HASH_SIZE; i++)
+		INIT_LIST_HEAD(&nn->reclaim_str_hashtbl[i]);
+	nn->reclaim_str_hashtbl_size = 0;
+
+	return 0;
+}
+
+static void
+nfs4_cld_state_shutdown(struct net *net)
+{
+	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
+
+	kfree(nn->reclaim_str_hashtbl);
+}
+
+static int
+nfsd4_cld_tracking_init(struct net *net)
+{
+	int status;
+	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
+
+	status = nfs4_cld_state_init(net);
+	if (status)
+		return status;
+
+	status = nfsd4_init_cld_pipe(net);
+	if (status)
+		goto err_shutdown;
+
+	status = nfsd4_cld_grace_start(nn);
+	if (status) {
+		if (status == -EOPNOTSUPP)
+			printk(KERN_WARNING "NFSD: Please upgrade nfsdcld.\n");
+		nfs4_release_reclaim(nn);
+		goto err_remove;
+	}
+	return 0;
+
+err_remove:
+	nfsd4_remove_cld_pipe(net);
+err_shutdown:
+	nfs4_cld_state_shutdown(net);
+	return status;
+}
+
+static void
+nfsd4_cld_tracking_exit(struct net *net)
+{
+	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
+
+	nfs4_release_reclaim(nn);
+	nfsd4_remove_cld_pipe(net);
+	nfs4_cld_state_shutdown(net);
+}
+
+/* For older nfsdcld's */
+static const struct nfsd4_client_tracking_ops nfsd4_cld_tracking_ops_v0 = {
 	.init		= nfsd4_init_cld_pipe,
 	.exit		= nfsd4_remove_cld_pipe,
 	.create		= nfsd4_cld_create,
 	.remove		= nfsd4_cld_remove,
+	.check		= nfsd4_cld_check_v0,
+	.grace_done	= nfsd4_cld_grace_done_v0,
+};
+
+/* For newer nfsdcld's */
+static const struct nfsd4_client_tracking_ops nfsd4_cld_tracking_ops = {
+	.init		= nfsd4_cld_tracking_init,
+	.exit		= nfsd4_cld_tracking_exit,
+	.create		= nfsd4_cld_create,
+	.remove		= nfsd4_cld_remove,
 	.check		= nfsd4_cld_check,
 	.grace_done	= nfsd4_cld_grace_done,
 };
@@ -1498,9 +1706,10 @@ nfsd4_client_tracking_init(struct net *net)
 
 	/* Finally, try to use nfsdcld */
 	nn->client_tracking_ops = &nfsd4_cld_tracking_ops;
-	printk(KERN_WARNING "NFSD: the nfsdcld client tracking upcall will be "
-			"removed in 3.10. Please transition to using "
-			"nfsdcltrack.\n");
+	status = nn->client_tracking_ops->init(net);
+	if (!status)
+		return status;
+	nn->client_tracking_ops = &nfsd4_cld_tracking_ops_v0;
 do_init:
 	status = nn->client_tracking_ops->init(net);
 	if (status) {
diff --git a/include/uapi/linux/nfsd/cld.h b/include/uapi/linux/nfsd/cld.h
index f8f5cccad749..b1e9de4f07d5 100644
--- a/include/uapi/linux/nfsd/cld.h
+++ b/include/uapi/linux/nfsd/cld.h
@@ -36,6 +36,7 @@ enum cld_command {
 	Cld_Remove,		/* remove record of this cm_id */
 	Cld_Check,		/* is this cm_id allowed? */
 	Cld_GraceDone,		/* grace period is complete */
+	Cld_GraceStart,
 };
 
 /* representation of long-form NFSv4 client ID */
-- 
2.17.1

Re: [PATCH RFC 3/4] nfsd: un-deprecate nfsdcld

[J. Bruce Fields]

On Tue, Nov 06, 2018 at 01:35:10PM -0500, Scott Mayhew wrote:
> When nfsdcld was released, it was quickly deprecated in favor of the
> nfsdcltrack usermodehelper, so as to not require another running daemon.
> That prevents NFSv4 clients from reclaiming locks from nfsd's running in
> containers, since neither nfsdcltrack nor the legacy client tracking
> code work in containers.
> 
> This commit un-deprecates the use of nfsdcld, with one twist: we will
> populate the reclaim_str_hashtbl on startup.
> 
> During client tracking initialization, do an upcall ("GraceStart") to
> nfsdcld to get a list of clients from the database.  nfsdcld will do
> one downcall with a status of -EINPROGRESS for each client record in
> the database, which in turn will cause an nfs4_client_reclaim to be
> added to the reclaim_str_hashtbl.  When complete, nfsdcld will do a
> final downcall with a status of 0.

Kind of a  funny convention, but OK, sounds like it should work.

> This will save nfsd from having to do an upcall to the daemon during
> nfs4_check_open_reclaim() processing.
> 
> Even though nfsdcld was quickly deprecated, there is a very small chance
> of old nfsdcld daemons running in the wild.  These will respond to the
> new "GraceStart" upcall with -EOPNOTSUPP, in which case we will log a
> message and fall back to the original nfsdcld tracking ops (now called
> nfsd4_cld_tracking_ops_v0).

Sounds good.

The code looks OK to me.

--b.

> 
> Signed-off-by: Scott Mayhew <smayhew@redhat.com>
> ---
>  fs/nfsd/nfs4recover.c         | 225 ++++++++++++++++++++++++++++++++--
>  include/uapi/linux/nfsd/cld.h |   1 +
>  2 files changed, 218 insertions(+), 8 deletions(-)
> 
> diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
> index b288b14273ff..10887fdce8c2 100644
> --- a/fs/nfsd/nfs4recover.c
> +++ b/fs/nfsd/nfs4recover.c
> @@ -763,6 +763,34 @@ cld_pipe_upcall(struct rpc_pipe *pipe, struct cld_msg *cmsg)
>  	return ret;
>  }
>  
> +static ssize_t
> +__cld_pipe_inprogress_downcall(const struct cld_msg __user *cmsg,
> +		struct nfsd_net *nn)
> +{
> +	uint8_t cmd;
> +	struct xdr_netobj name;
> +	uint16_t namelen;
> +
> +	if (get_user(cmd, &cmsg->cm_cmd)) {
> +		dprintk("%s: error when copying cmd from userspace", __func__);
> +		return -EFAULT;
> +	}
> +	if (cmd == Cld_GraceStart) {
> +		if (get_user(namelen, &cmsg->cm_u.cm_name.cn_len))
> +			return -EFAULT;
> +		name.data = memdup_user(&cmsg->cm_u.cm_name.cn_id, namelen);
> +		if (IS_ERR_OR_NULL(name.data))
> +			return -EFAULT;
> +		name.len = namelen;
> +		if (!nfs4_client_to_reclaim(name, nn)) {
> +			kfree(name.data);
> +			return -EFAULT;
> +		}
> +		return sizeof(*cmsg);
> +	}
> +	return -EFAULT;
> +}
> +
>  static ssize_t
>  cld_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
>  {
> @@ -772,6 +800,7 @@ cld_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
>  	struct nfsd_net *nn = net_generic(file_inode(filp)->i_sb->s_fs_info,
>  						nfsd_net_id);
>  	struct cld_net *cn = nn->cld_net;
> +	int16_t status;
>  
>  	if (mlen != sizeof(*cmsg)) {
>  		dprintk("%s: got %zu bytes, expected %zu\n", __func__, mlen,
> @@ -785,13 +814,24 @@ cld_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
>  		return -EFAULT;
>  	}
>  
> +	/*
> +	 * copy the status so we know whether to remove the upcall from the
> +	 * list (for -EINPROGRESS, we just want to make sure the xid is
> +	 * valid, not remove the upcall from the list)
> +	 */
> +	if (get_user(status, &cmsg->cm_status)) {
> +		dprintk("%s: error when copying status from userspace", __func__);
> +		return -EFAULT;
> +	}
> +
>  	/* walk the list and find corresponding xid */
>  	cup = NULL;
>  	spin_lock(&cn->cn_lock);
>  	list_for_each_entry(tmp, &cn->cn_list, cu_list) {
>  		if (get_unaligned(&tmp->cu_msg.cm_xid) == xid) {
>  			cup = tmp;
> -			list_del_init(&cup->cu_list);
> +			if (status != -EINPROGRESS)
> +				list_del_init(&cup->cu_list);
>  			break;
>  		}
>  	}
> @@ -803,6 +843,9 @@ cld_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
>  		return -EINVAL;
>  	}
>  
> +	if (status == -EINPROGRESS)
> +		return __cld_pipe_inprogress_downcall(cmsg, nn);
> +
>  	if (copy_from_user(&cup->cu_msg, src, mlen) != 0)
>  		return -EFAULT;
>  
> @@ -1049,9 +1092,14 @@ nfsd4_cld_remove(struct nfs4_client *clp)
>  				"record from stable storage: %d\n", ret);
>  }
>  
> -/* Check for presence of a record, and update its timestamp */
> +/*
> + * For older nfsdcld's that do not allow us to "slurp" the clients
> + * from the tracking database during startup.
> + *
> + * Check for presence of a record, and update its timestamp
> + */
>  static int
> -nfsd4_cld_check(struct nfs4_client *clp)
> +nfsd4_cld_check_v0(struct nfs4_client *clp)
>  {
>  	int ret;
>  	struct cld_upcall *cup;
> @@ -1084,8 +1132,61 @@ nfsd4_cld_check(struct nfs4_client *clp)
>  	return ret;
>  }
>  
> +/*
> + * For newer nfsdcld's that allow us to "slurp" the clients
> + * from the tracking database during startup.
> + *
> + * Check for presence of a record in the reclaim_str_hashtbl
> + */
> +static int
> +nfsd4_cld_check(struct nfs4_client *clp)
> +{
> +	struct nfs4_client_reclaim *crp;
> +	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
> +
> +	/* did we already find that this client is stable? */
> +	if (test_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
> +		return 0;
> +
> +	/* look for it in the reclaim hashtable otherwise */
> +	crp = nfsd4_find_reclaim_client(clp->cl_name, nn);
> +	if (crp) {
> +		crp->cr_clp = clp;
> +		return 0;
> +	}
> +
> +	return -ENOENT;
> +}
> +
> +static int
> +nfsd4_cld_grace_start(struct nfsd_net *nn)
> +{
> +	int ret;
> +	struct cld_upcall *cup;
> +	struct cld_net *cn = nn->cld_net;
> +
> +	cup = alloc_cld_upcall(cn);
> +	if (!cup) {
> +		ret = -ENOMEM;
> +		goto out_err;
> +	}
> +
> +	cup->cu_msg.cm_cmd = Cld_GraceStart;
> +	ret = cld_pipe_upcall(cn->cn_pipe, &cup->cu_msg);
> +	if (!ret)
> +		ret = cup->cu_msg.cm_status;
> +
> +	free_cld_upcall(cup);
> +out_err:
> +	if (ret)
> +		dprintk("%s: Unable to get clients from userspace: %d\n",
> +			__func__, ret);
> +	return ret;
> +}
> +
> +/* For older nfsdcld's that need cm_gracetime */
>  static void
> -nfsd4_cld_grace_done(struct nfsd_net *nn)
> +nfsd4_cld_grace_done_v0(struct nfsd_net *nn)
>  {
>  	int ret;
>  	struct cld_upcall *cup;
> @@ -1109,11 +1210,118 @@ nfsd4_cld_grace_done(struct nfsd_net *nn)
>  		printk(KERN_ERR "NFSD: Unable to end grace period: %d\n", ret);
>  }
>  
> -static const struct nfsd4_client_tracking_ops nfsd4_cld_tracking_ops = {
> +/*
> + * For newer nfsdcld's that do not need cm_gracetime.  We also need to call
> + * nfs4_release_reclaim() to clear out the reclaim_str_hashtbl.
> + */
> +static void
> +nfsd4_cld_grace_done(struct nfsd_net *nn)
> +{
> +	int ret;
> +	struct cld_upcall *cup;
> +	struct cld_net *cn = nn->cld_net;
> +
> +	cup = alloc_cld_upcall(cn);
> +	if (!cup) {
> +		ret = -ENOMEM;
> +		goto out_err;
> +	}
> +
> +	cup->cu_msg.cm_cmd = Cld_GraceDone;
> +	ret = cld_pipe_upcall(cn->cn_pipe, &cup->cu_msg);
> +	if (!ret)
> +		ret = cup->cu_msg.cm_status;
> +
> +	free_cld_upcall(cup);
> +out_err:
> +	nfs4_release_reclaim(nn);
> +	if (ret)
> +		printk(KERN_ERR "NFSD: Unable to end grace period: %d\n", ret);
> +}
> +
> +static int
> +nfs4_cld_state_init(struct net *net)
> +{
> +	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
> +	int i;
> +
> +	nn->reclaim_str_hashtbl = kmalloc_array(CLIENT_HASH_SIZE,
> +						sizeof(struct list_head),
> +						GFP_KERNEL);
> +	if (!nn->reclaim_str_hashtbl)
> +		return -ENOMEM;
> +
> +	for (i = 0; i < CLIENT_HASH_SIZE; i++)
> +		INIT_LIST_HEAD(&nn->reclaim_str_hashtbl[i]);
> +	nn->reclaim_str_hashtbl_size = 0;
> +
> +	return 0;
> +}
> +
> +static void
> +nfs4_cld_state_shutdown(struct net *net)
> +{
> +	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
> +
> +	kfree(nn->reclaim_str_hashtbl);
> +}
> +
> +static int
> +nfsd4_cld_tracking_init(struct net *net)
> +{
> +	int status;
> +	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
> +
> +	status = nfs4_cld_state_init(net);
> +	if (status)
> +		return status;
> +
> +	status = nfsd4_init_cld_pipe(net);
> +	if (status)
> +		goto err_shutdown;
> +
> +	status = nfsd4_cld_grace_start(nn);
> +	if (status) {
> +		if (status == -EOPNOTSUPP)
> +			printk(KERN_WARNING "NFSD: Please upgrade nfsdcld.\n");
> +		nfs4_release_reclaim(nn);
> +		goto err_remove;
> +	}
> +	return 0;
> +
> +err_remove:
> +	nfsd4_remove_cld_pipe(net);
> +err_shutdown:
> +	nfs4_cld_state_shutdown(net);
> +	return status;
> +}
> +
> +static void
> +nfsd4_cld_tracking_exit(struct net *net)
> +{
> +	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
> +
> +	nfs4_release_reclaim(nn);
> +	nfsd4_remove_cld_pipe(net);
> +	nfs4_cld_state_shutdown(net);
> +}
> +
> +/* For older nfsdcld's */
> +static const struct nfsd4_client_tracking_ops nfsd4_cld_tracking_ops_v0 = {
>  	.init		= nfsd4_init_cld_pipe,
>  	.exit		= nfsd4_remove_cld_pipe,
>  	.create		= nfsd4_cld_create,
>  	.remove		= nfsd4_cld_remove,
> +	.check		= nfsd4_cld_check_v0,
> +	.grace_done	= nfsd4_cld_grace_done_v0,
> +};
> +
> +/* For newer nfsdcld's */
> +static const struct nfsd4_client_tracking_ops nfsd4_cld_tracking_ops = {
> +	.init		= nfsd4_cld_tracking_init,
> +	.exit		= nfsd4_cld_tracking_exit,
> +	.create		= nfsd4_cld_create,
> +	.remove		= nfsd4_cld_remove,
>  	.check		= nfsd4_cld_check,
>  	.grace_done	= nfsd4_cld_grace_done,
>  };
> @@ -1498,9 +1706,10 @@ nfsd4_client_tracking_init(struct net *net)
>  
>  	/* Finally, try to use nfsdcld */
>  	nn->client_tracking_ops = &nfsd4_cld_tracking_ops;
> -	printk(KERN_WARNING "NFSD: the nfsdcld client tracking upcall will be "
> -			"removed in 3.10. Please transition to using "
> -			"nfsdcltrack.\n");
> +	status = nn->client_tracking_ops->init(net);
> +	if (!status)
> +		return status;
> +	nn->client_tracking_ops = &nfsd4_cld_tracking_ops_v0;
>  do_init:
>  	status = nn->client_tracking_ops->init(net);
>  	if (status) {
> diff --git a/include/uapi/linux/nfsd/cld.h b/include/uapi/linux/nfsd/cld.h
> index f8f5cccad749..b1e9de4f07d5 100644
> --- a/include/uapi/linux/nfsd/cld.h
> +++ b/include/uapi/linux/nfsd/cld.h
> @@ -36,6 +36,7 @@ enum cld_command {
>  	Cld_Remove,		/* remove record of this cm_id */
>  	Cld_Check,		/* is this cm_id allowed? */
>  	Cld_GraceDone,		/* grace period is complete */
> +	Cld_GraceStart,
>  };
>  
>  /* representation of long-form NFSv4 client ID */
> -- 
> 2.17.1

[PATCH RFC 4/4] nfsd: keep a tally of RECLAIM_COMPLETE operations when using nfsdcld

[Scott Mayhew]

When using nfsdcld for NFSv4 client tracking, track the number of
RECLAIM_COMPLETE operations we receive from "known" clients to help in
deciding if we can lift the grace period early (or whether we need to
start a v4 grace period at all).

Signed-off-by: Scott Mayhew <smayhew@redhat.com>
---
 fs/nfsd/netns.h       |  3 +++
 fs/nfsd/nfs4recover.c |  5 +++++
 fs/nfsd/nfs4state.c   | 52 +++++++++++++++++++++++++++++++++++++++++++
 fs/nfsd/nfsctl.c      |  1 +
 4 files changed, 61 insertions(+)

diff --git a/fs/nfsd/netns.h b/fs/nfsd/netns.h
index 32cb8c027483..b42aaa22fba2 100644
--- a/fs/nfsd/netns.h
+++ b/fs/nfsd/netns.h
@@ -104,6 +104,9 @@ struct nfsd_net {
 	time_t nfsd4_grace;
 	bool somebody_reclaimed;
 
+	bool track_reclaim_completes;
+	atomic_t nr_reclaim_complete;
+
 	bool nfsd_net_up;
 	bool lockd_up;
 
diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
index 10887fdce8c2..171c013d63ba 100644
--- a/fs/nfsd/nfs4recover.c
+++ b/fs/nfsd/nfs4recover.c
@@ -1235,6 +1235,7 @@ nfsd4_cld_grace_done(struct nfsd_net *nn)
 	free_cld_upcall(cup);
 out_err:
 	nfs4_release_reclaim(nn);
+	atomic_set(&nn->nr_reclaim_complete, 0);
 	if (ret)
 		printk(KERN_ERR "NFSD: Unable to end grace period: %d\n", ret);
 }
@@ -1254,6 +1255,8 @@ nfs4_cld_state_init(struct net *net)
 	for (i = 0; i < CLIENT_HASH_SIZE; i++)
 		INIT_LIST_HEAD(&nn->reclaim_str_hashtbl[i]);
 	nn->reclaim_str_hashtbl_size = 0;
+	nn->track_reclaim_completes = true;
+	atomic_set(&nn->nr_reclaim_complete, 0);
 
 	return 0;
 }
@@ -1263,6 +1266,7 @@ nfs4_cld_state_shutdown(struct net *net)
 {
 	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
 
+	nn->track_reclaim_completes = false;
 	kfree(nn->reclaim_str_hashtbl);
 }
 
@@ -1302,6 +1306,7 @@ nfsd4_cld_tracking_exit(struct net *net)
 	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
 
 	nfs4_release_reclaim(nn);
+	atomic_set(&nn->nr_reclaim_complete, 0);
 	nfsd4_remove_cld_pipe(net);
 	nfs4_cld_state_shutdown(net);
 }
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 777fbb0d2761..5dfc3cd51d08 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -77,6 +77,7 @@ static u64 current_sessionid = 1;
 /* forward declarations */
 static bool check_for_locks(struct nfs4_file *fp, struct nfs4_lockowner *lowner);
 static void nfs4_free_ol_stateid(struct nfs4_stid *stid);
+void nfsd4_end_grace(struct nfsd_net *nn);
 
 /* Locking: */
 
@@ -1988,10 +1989,41 @@ destroy_client(struct nfs4_client *clp)
 	__destroy_client(clp);
 }
 
+static void inc_reclaim_complete(struct nfs4_client *clp)
+{
+	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
+
+	if (!nn->track_reclaim_completes)
+		return;
+	if (!test_bit(NFSD4_CLIENT_RECLAIM_COMPLETE, &clp->cl_flags))
+		return;
+	if (!nfsd4_find_reclaim_client(clp->cl_name, nn))
+		return;
+	if (atomic_inc_return(&nn->nr_reclaim_complete) ==
+			nn->reclaim_str_hashtbl_size) {
+		printk(KERN_INFO "NFSD: all clients done reclaiming, ending NFSv4 grace period (net %x)\n",
+				clp->net->ns.inum);
+		nfsd4_end_grace(nn);
+	}
+}
+
+static void dec_reclaim_complete(struct nfs4_client *clp)
+{
+	struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
+
+	if (!nn->track_reclaim_completes)
+		return;
+	if (!test_bit(NFSD4_CLIENT_RECLAIM_COMPLETE, &clp->cl_flags))
+		return;
+	if (nfsd4_find_reclaim_client(clp->cl_name, nn))
+		atomic_dec(&nn->nr_reclaim_complete);
+}
+
 static void expire_client(struct nfs4_client *clp)
 {
 	unhash_client(clp);
 	nfsd4_client_record_remove(clp);
+	dec_reclaim_complete(clp);
 	__destroy_client(clp);
 }
 
@@ -3339,6 +3371,7 @@ nfsd4_reclaim_complete(struct svc_rqst *rqstp,
 
 	status = nfs_ok;
 	nfsd4_client_record_create(cstate->session->se_client);
+	inc_reclaim_complete(cstate->session->se_client);
 out:
 	return status;
 }
@@ -4734,6 +4767,10 @@ static bool clients_still_reclaiming(struct nfsd_net *nn)
 	unsigned long double_grace_period_end = nn->boot_time +
 						2 * nn->nfsd4_lease;
 
+	if (nn->track_reclaim_completes &&
+			atomic_read(&nn->nr_reclaim_complete) ==
+			nn->reclaim_str_hashtbl_size)
+		return false;
 	if (!nn->somebody_reclaimed)
 		return false;
 	nn->somebody_reclaimed = false;
@@ -7253,10 +7290,25 @@ nfs4_state_start_net(struct net *net)
 		return ret;
 	locks_start_grace(net, &nn->nfsd4_manager);
 	nfsd4_client_tracking_init(net);
+	if (nn->track_reclaim_completes && nn->reclaim_str_hashtbl_size == 0)
+		goto skip_grace;
 	printk(KERN_INFO "NFSD: starting %ld-second grace period (net %x)\n",
 	       nn->nfsd4_grace, net->ns.inum);
 	queue_delayed_work(laundry_wq, &nn->laundromat_work, nn->nfsd4_grace * HZ);
 	return 0;
+
+skip_grace:
+	printk(KERN_INFO "NFSD: no clients to reclaim, skipping NFSv4 grace period (net %x)\n",
+			net->ns.inum);
+	queue_delayed_work(laundry_wq, &nn->laundromat_work, nn->nfsd4_lease * HZ);
+	/*
+	 * we could call nfsd4_end_grace() here, but it has a dprintk()
+	 * that would be confusing if debug logging is enabled
+	 */
+	nn->grace_ended = true;
+	nfsd4_record_grace_done(nn);
+	locks_end_grace(&nn->nfsd4_manager);
+	return 0;
 }
 
 /* initialization to perform when the nfsd service is started: */
diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
index 6384c9b94898..950ac6683be9 100644
--- a/fs/nfsd/nfsctl.c
+++ b/fs/nfsd/nfsctl.c
@@ -1240,6 +1240,7 @@ static __net_init int nfsd_init_net(struct net *net)
 	nn->nfsd4_lease = 45;	/* default lease time */
 	nn->nfsd4_grace = 45;
 	nn->somebody_reclaimed = false;
+	nn->track_reclaim_completes = false;
 	nn->clverifier_counter = prandom_u32();
 	nn->clientid_counter = prandom_u32();
 	nn->s2s_cp_cl_id = nn->clientid_counter++;
-- 
2.17.1

Re: [PATCH RFC 0/4] un-deprecate nfsdcld

[J. Bruce Fields]

Thanks for doing this, we've really been wanting it.  I should be able
to get to it in the next few days....

Out of curiosity, what are you using for testing?

--b.

On Tue, Nov 06, 2018 at 01:35:07PM -0500, Scott Mayhew wrote:
> When nfsdcld was released, it was quickly deprecated in favor of the
> nfsdcltrack usermodehelper, so as to not require another running daemon.
> That prevents NFSv4 clients from reclaiming locks from nfsd's running in
> containers, since neither nfsdcltrack nor the legacy client tracking
> code work in containers.  These patches un-deprecate the use of nfsdcld
> for NFSv4 client tracking.
> 
> These patches are intended to go alongside some nfs-utils patches that
> introduce an enhancement that allows nfsd to "slurp" up the client
> records during client tracking initialization and store them internally
> in hash table.  This enables nfsd to check whether an NFSv4 client is 
> allowed to reclaim without having to do an upcall to nfsdcld.  It also
> allows nfsd to decide to end the v4 grace period early if the number of
> RECLAIM_COMPLETE operations it has received from "known" clients is
> equal to the number of entries in the hash table.  It also allows nfsd
> to skip the v4 grace period altogether if it knows there are no clients
> allowed to reclaim.
> 
> There is a fallback to allow nfsd to continue to work with older nfsdcld
> daemons in the event that any are out in the wild (unlikely).
> Everything should work fine except nfsd will not be able to exit the
> grace period early or skip the grace period altogether.
> 
> Scott Mayhew (4):
>   nfsd: fix a warning in __cld_pipe_upcall()
>   nfsd: make nfs4_client_reclaim use an xdr_netobj instead of a fixed
>     char array
>   nfsd: un-deprecate nfsdcld
>   nfsd: keep a tally of RECLAIM_COMPLETE operations when using nfsdcld
> 
>  fs/nfsd/netns.h               |   3 +
>  fs/nfsd/nfs4recover.c         | 326 ++++++++++++++++++++++++++++++----
>  fs/nfsd/nfs4state.c           |  82 +++++++--
>  fs/nfsd/nfsctl.c              |   1 +
>  fs/nfsd/state.h               |   8 +-
>  include/uapi/linux/nfsd/cld.h |   1 +
>  6 files changed, 372 insertions(+), 49 deletions(-)
> 
> -- 
> 2.17.1

Re: [PATCH RFC 0/4] un-deprecate nfsdcld

[Scott Mayhew]

On Wed, 07 Nov 2018, J. Bruce Fields wrote:

> Thanks for doing this, we've really been wanting it.  I should be able
> to get to it in the next few days....
> 
> Out of curiosity, what are you using for testing?

For initially testing the GraceStart upcall, I'd just loaded a bunch of
bogus clientid's into the database and wrote a crash macro to dump out
the reclaim_str_hashtbl to make sure everything got there on startup.

For testing nfsd in a container, right now I'm using plain docker.  I
have one container running nfsd.  I made another image that just runs
cthon against an ip that you pass in via an environment variable.  I
start up a bunch of containers using the cthon image and then
periodically restart the nfsd container.  I'd like to try this stuff
out on OpenShift at some point too.

I'm open to other ideas for testing if you have any.

-Scott
> 
> --b.
> 
> On Tue, Nov 06, 2018 at 01:35:07PM -0500, Scott Mayhew wrote:
> > When nfsdcld was released, it was quickly deprecated in favor of the
> > nfsdcltrack usermodehelper, so as to not require another running daemon.
> > That prevents NFSv4 clients from reclaiming locks from nfsd's running in
> > containers, since neither nfsdcltrack nor the legacy client tracking
> > code work in containers.  These patches un-deprecate the use of nfsdcld
> > for NFSv4 client tracking.
> > 
> > These patches are intended to go alongside some nfs-utils patches that
> > introduce an enhancement that allows nfsd to "slurp" up the client
> > records during client tracking initialization and store them internally
> > in hash table.  This enables nfsd to check whether an NFSv4 client is 
> > allowed to reclaim without having to do an upcall to nfsdcld.  It also
> > allows nfsd to decide to end the v4 grace period early if the number of
> > RECLAIM_COMPLETE operations it has received from "known" clients is
> > equal to the number of entries in the hash table.  It also allows nfsd
> > to skip the v4 grace period altogether if it knows there are no clients
> > allowed to reclaim.
> > 
> > There is a fallback to allow nfsd to continue to work with older nfsdcld
> > daemons in the event that any are out in the wild (unlikely).
> > Everything should work fine except nfsd will not be able to exit the
> > grace period early or skip the grace period altogether.
> > 
> > Scott Mayhew (4):
> >   nfsd: fix a warning in __cld_pipe_upcall()
> >   nfsd: make nfs4_client_reclaim use an xdr_netobj instead of a fixed
> >     char array
> >   nfsd: un-deprecate nfsdcld
> >   nfsd: keep a tally of RECLAIM_COMPLETE operations when using nfsdcld
> > 
> >  fs/nfsd/netns.h               |   3 +
> >  fs/nfsd/nfs4recover.c         | 326 ++++++++++++++++++++++++++++++----
> >  fs/nfsd/nfs4state.c           |  82 +++++++--
> >  fs/nfsd/nfsctl.c              |   1 +
> >  fs/nfsd/state.h               |   8 +-
> >  include/uapi/linux/nfsd/cld.h |   1 +
> >  6 files changed, 372 insertions(+), 49 deletions(-)
> > 
> > -- 
> > 2.17.1