Merge tag 'nfs-for-5.6-2' of git://git.linux-nfs.org/projects/anna/linux-nfs

Pull NFS client bugfixes from Anna Schumaker:
 "The only stable fix this time is the DMA scatter-gather list bug fixed
  by Chuck.

  The rest fix up races and refcounting issues that have been found
  during testing.

  Stable fix:
   - fix DMA scatter-gather list mapping imbalance

  The rest:
   - fix directory verifier races
   - fix races between open and dentry revalidation
   - fix revalidation of dentries with delegations
   - fix "cachethis" setting for writes
   - fix delegation and delegation cred pinning"

* tag 'nfs-for-5.6-2' of git://git.linux-nfs.org/projects/anna/linux-nfs:
  NFSv4: Ensure the delegation cred is pinned when we call delegreturn
  NFSv4: Ensure the delegation is pinned in nfs_do_return_delegation()
  NFSv4.1 make cachethis=no for writes
  xprtrdma: Fix DMA scatter-gather list mapping imbalance
  NFSv4: Fix revalidation of dentries with delegations
  NFSv4: Fix races between open and dentry revalidation
  NFS: Fix up directory verifier races

8 files changed, 188 insertions, 50 deletions

diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c
index 4a841071d8a7..1865322de142 100644
--- a/fs/nfs/delegation.c
+++ b/fs/nfs/delegation.c
@@ -42,13 +42,27 @@ static void nfs_mark_delegation_revoked(struct nfs_delegation *delegation)
 	if (!test_and_set_bit(NFS_DELEGATION_REVOKED, &delegation->flags)) {
 		delegation->stateid.type = NFS4_INVALID_STATEID_TYPE;
 		atomic_long_dec(&nfs_active_delegations);
+		if (!test_bit(NFS_DELEGATION_RETURNING, &delegation->flags))
+			nfs_clear_verifier_delegated(delegation->inode);
 	}
 }
 
+static struct nfs_delegation *nfs_get_delegation(struct nfs_delegation *delegation)
+{
+	refcount_inc(&delegation->refcount);
+	return delegation;
+}
+
+static void nfs_put_delegation(struct nfs_delegation *delegation)
+{
+	if (refcount_dec_and_test(&delegation->refcount))
+		__nfs_free_delegation(delegation);
+}
+
 static void nfs_free_delegation(struct nfs_delegation *delegation)
 {
 	nfs_mark_delegation_revoked(delegation);
-	__nfs_free_delegation(delegation);
+	nfs_put_delegation(delegation);
 }
 
 /**
@@ -241,13 +255,18 @@ void nfs_inode_reclaim_delegation(struct inode *inode, const struct cred *cred,
 
 static int nfs_do_return_delegation(struct inode *inode, struct nfs_delegation *delegation, int issync)
 {
+	const struct cred *cred;
 	int res = 0;
 
-	if (!test_bit(NFS_DELEGATION_REVOKED, &delegation->flags))
-		res = nfs4_proc_delegreturn(inode,
-				delegation->cred,
+	if (!test_bit(NFS_DELEGATION_REVOKED, &delegation->flags)) {
+		spin_lock(&delegation->lock);
+		cred = get_cred(delegation->cred);
+		spin_unlock(&delegation->lock);
+		res = nfs4_proc_delegreturn(inode, cred,
 				&delegation->stateid,
 				issync);
+		put_cred(cred);
+	}
 	return res;
 }
 
@@ -273,9 +292,13 @@ nfs_start_delegation_return_locked(struct nfs_inode *nfsi)
 	if (delegation == NULL)
 		goto out;
 	spin_lock(&delegation->lock);
-	if (!test_and_set_bit(NFS_DELEGATION_RETURNING, &delegation->flags))
-		ret = delegation;
+	if (!test_and_set_bit(NFS_DELEGATION_RETURNING, &delegation->flags)) {
+		/* Refcount matched in nfs_end_delegation_return() */
+		ret = nfs_get_delegation(delegation);
+	}
 	spin_unlock(&delegation->lock);
+	if (ret)
+		nfs_clear_verifier_delegated(&nfsi->vfs_inode);
 out:
 	return ret;
 }
@@ -393,6 +416,7 @@ int nfs_inode_set_delegation(struct inode *inode, const struct cred *cred,
 	if (delegation == NULL)
 		return -ENOMEM;
 	nfs4_stateid_copy(&delegation->stateid, stateid);
+	refcount_set(&delegation->refcount, 1);
 	delegation->type = type;
 	delegation->pagemod_limit = pagemod_limit;
 	delegation->change_attr = inode_peek_iversion_raw(inode);
@@ -492,6 +516,8 @@ static int nfs_end_delegation_return(struct inode *inode, struct nfs_delegation
 
 	err = nfs_do_return_delegation(inode, delegation, issync);
 out:
+	/* Refcount matched in nfs_start_delegation_return_locked() */
+	nfs_put_delegation(delegation);
 	return err;
 }
 
@@ -686,9 +712,12 @@ void nfs4_inode_return_delegation_on_close(struct inode *inode)
 		    list_empty(&NFS_I(inode)->open_files) &&
 		    !test_and_set_bit(NFS_DELEGATION_RETURNING, &delegation->flags)) {
 			clear_bit(NFS_DELEGATION_RETURN_IF_CLOSED, &delegation->flags);
-			ret = delegation;
+			/* Refcount matched in nfs_end_delegation_return() */
+			ret = nfs_get_delegation(delegation);
 		}
 		spin_unlock(&delegation->lock);
+		if (ret)
+			nfs_clear_verifier_delegated(inode);
 	}
 out:
 	rcu_read_unlock();
@@ -1088,10 +1117,11 @@ restart:
 			delegation = nfs_start_delegation_return_locked(NFS_I(inode));
 			rcu_read_unlock();
 			if (delegation != NULL) {
-				delegation = nfs_detach_delegation(NFS_I(inode),
-					delegation, server);
-				if (delegation != NULL)
+				if (nfs_detach_delegation(NFS_I(inode), delegation,
+							server) != NULL)
 					nfs_free_delegation(delegation);
+				/* Match nfs_start_delegation_return_locked */
+				nfs_put_delegation(delegation);
 			}
 			iput(inode);
 			nfs_sb_deactive(server->super);
diff --git a/fs/nfs/delegation.h b/fs/nfs/delegation.h
index 31b84604d383..9b00a0b7f832 100644
--- a/fs/nfs/delegation.h
+++ b/fs/nfs/delegation.h
@@ -22,6 +22,7 @@ struct nfs_delegation {
 	unsigned long pagemod_limit;
 	__u64 change_attr;
 	unsigned long flags;
+	refcount_t refcount;
 	spinlock_t lock;
 	struct rcu_head rcu;
 };
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 1320288ff9ec..193d6fb363b7 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -155,6 +155,7 @@ typedef struct {
 	loff_t		current_index;
 	decode_dirent_t	decode;
 
+	unsigned long	dir_verifier;
 	unsigned long	timestamp;
 	unsigned long	gencount;
 	unsigned int	cache_entry_index;
@@ -353,6 +354,7 @@ int nfs_readdir_xdr_filler(struct page **pages, nfs_readdir_descriptor_t *desc,
  again:
 	timestamp = jiffies;
 	gencount = nfs_inc_attr_generation_counter();
+	desc->dir_verifier = nfs_save_change_attribute(inode);
 	error = NFS_PROTO(inode)->readdir(file_dentry(file), cred, entry->cookie, pages,
 					  NFS_SERVER(inode)->dtsize, desc->plus);
 	if (error < 0) {
@@ -455,13 +457,13 @@ void nfs_force_use_readdirplus(struct inode *dir)
 }
 
 static
-void nfs_prime_dcache(struct dentry *parent, struct nfs_entry *entry)
+void nfs_prime_dcache(struct dentry *parent, struct nfs_entry *entry,
+		unsigned long dir_verifier)
 {
 	struct qstr filename = QSTR_INIT(entry->name, entry->len);
 	DECLARE_WAIT_QUEUE_HEAD_ONSTACK(wq);
 	struct dentry *dentry;
 	struct dentry *alias;
-	struct inode *dir = d_inode(parent);
 	struct inode *inode;
 	int status;
 
@@ -500,7 +502,7 @@ again:
 		if (nfs_same_file(dentry, entry)) {
 			if (!entry->fh->size)
 				goto out;
-			nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
+			nfs_set_verifier(dentry, dir_verifier);
 			status = nfs_refresh_inode(d_inode(dentry), entry->fattr);
 			if (!status)
 				nfs_setsecurity(d_inode(dentry), entry->fattr, entry->label);
@@ -526,7 +528,7 @@ again:
 		dput(dentry);
 		dentry = alias;
 	}
-	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
+	nfs_set_verifier(dentry, dir_verifier);
 out:
 	dput(dentry);
 }
@@ -564,7 +566,8 @@ int nfs_readdir_page_filler(nfs_readdir_descriptor_t *desc, struct nfs_entry *en
 		count++;
 
 		if (desc->plus)
-			nfs_prime_dcache(file_dentry(desc->file), entry);
+			nfs_prime_dcache(file_dentry(desc->file), entry,
+					desc->dir_verifier);
 
 		status = nfs_readdir_add_to_array(entry, page);
 		if (status != 0)
@@ -983,14 +986,113 @@ static int nfs_fsync_dir(struct file *filp, loff_t start, loff_t end,
  * full lookup on all child dentries of 'dir' whenever a change occurs
  * on the server that might have invalidated our dcache.
  *
+ * Note that we reserve bit '0' as a tag to let us know when a dentry
+ * was revalidated while holding a delegation on its inode.
+ *
  * The caller should be holding dir->i_lock
  */
 void nfs_force_lookup_revalidate(struct inode *dir)
 {
-	NFS_I(dir)->cache_change_attribute++;
+	NFS_I(dir)->cache_change_attribute += 2;
 }
 EXPORT_SYMBOL_GPL(nfs_force_lookup_revalidate);
 
+/**
+ * nfs_verify_change_attribute - Detects NFS remote directory changes
+ * @dir: pointer to parent directory inode
+ * @verf: previously saved change attribute
+ *
+ * Return "false" if the verifiers doesn't match the change attribute.
+ * This would usually indicate that the directory contents have changed on
+ * the server, and that any dentries need revalidating.
+ */
+static bool nfs_verify_change_attribute(struct inode *dir, unsigned long verf)
+{
+	return (verf & ~1UL) == nfs_save_change_attribute(dir);
+}
+
+static void nfs_set_verifier_delegated(unsigned long *verf)
+{
+	*verf |= 1UL;
+}
+
+#if IS_ENABLED(CONFIG_NFS_V4)
+static void nfs_unset_verifier_delegated(unsigned long *verf)
+{
+	*verf &= ~1UL;
+}
+#endif /* IS_ENABLED(CONFIG_NFS_V4) */
+
+static bool nfs_test_verifier_delegated(unsigned long verf)
+{
+	return verf & 1;
+}
+
+static bool nfs_verifier_is_delegated(struct dentry *dentry)
+{
+	return nfs_test_verifier_delegated(dentry->d_time);
+}
+
+static void nfs_set_verifier_locked(struct dentry *dentry, unsigned long verf)
+{
+	struct inode *inode = d_inode(dentry);
+
+	if (!nfs_verifier_is_delegated(dentry) &&
+	    !nfs_verify_change_attribute(d_inode(dentry->d_parent), verf))
+		goto out;
+	if (inode && NFS_PROTO(inode)->have_delegation(inode, FMODE_READ))
+		nfs_set_verifier_delegated(&verf);
+out:
+	dentry->d_time = verf;
+}
+
+/**
+ * nfs_set_verifier - save a parent directory verifier in the dentry
+ * @dentry: pointer to dentry
+ * @verf: verifier to save
+ *
+ * Saves the parent directory verifier in @dentry. If the inode has
+ * a delegation, we also tag the dentry as having been revalidated
+ * while holding a delegation so that we know we don't have to
+ * look it up again after a directory change.
+ */
+void nfs_set_verifier(struct dentry *dentry, unsigned long verf)
+{
+
+	spin_lock(&dentry->d_lock);
+	nfs_set_verifier_locked(dentry, verf);
+	spin_unlock(&dentry->d_lock);
+}
+EXPORT_SYMBOL_GPL(nfs_set_verifier);
+
+#if IS_ENABLED(CONFIG_NFS_V4)
+/**
+ * nfs_clear_verifier_delegated - clear the dir verifier delegation tag
+ * @inode: pointer to inode
+ *
+ * Iterates through the dentries in the inode alias list and clears
+ * the tag used to indicate that the dentry has been revalidated
+ * while holding a delegation.
+ * This function is intended for use when the delegation is being
+ * returned or revoked.
+ */
+void nfs_clear_verifier_delegated(struct inode *inode)
+{
+	struct dentry *alias;
+
+	if (!inode)
+		return;
+	spin_lock(&inode->i_lock);
+	hlist_for_each_entry(alias, &inode->i_dentry, d_u.d_alias) {
+		spin_lock(&alias->d_lock);
+		nfs_unset_verifier_delegated(&alias->d_time);
+		spin_unlock(&alias->d_lock);
+	}
+	spin_unlock(&inode->i_lock);
+}
+EXPORT_SYMBOL_GPL(nfs_clear_verifier_delegated);
+#endif /* IS_ENABLED(CONFIG_NFS_V4) */
+
 /*
  * A check for whether or not the parent directory has changed.
  * In the case it has, we assume that the dentries are untrustworthy
@@ -1159,6 +1261,7 @@ nfs_lookup_revalidate_dentry(struct inode *dir, struct dentry *dentry,
 	struct nfs_fh *fhandle;
 	struct nfs_fattr *fattr;
 	struct nfs4_label *label;
+	unsigned long dir_verifier;
 	int ret;
 
 	ret = -ENOMEM;
@@ -1168,6 +1271,7 @@ nfs_lookup_revalidate_dentry(struct inode *dir, struct dentry *dentry,
 	if (fhandle == NULL || fattr == NULL || IS_ERR(label))
 		goto out;
 
+	dir_verifier = nfs_save_change_attribute(dir);
 	ret = NFS_PROTO(dir)->lookup(dir, dentry, fhandle, fattr, label);
 	if (ret < 0) {
 		switch (ret) {
@@ -1188,7 +1292,7 @@ nfs_lookup_revalidate_dentry(struct inode *dir, struct dentry *dentry,
 		goto out;
 
 	nfs_setsecurity(inode, fattr, label);
-	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
+	nfs_set_verifier(dentry, dir_verifier);
 
 	/* set a readdirplus hint that we had a cache miss */
 	nfs_force_use_readdirplus(dir);
@@ -1230,7 +1334,7 @@ nfs_do_lookup_revalidate(struct inode *dir, struct dentry *dentry,
 		goto out_bad;
 	}
 
-	if (NFS_PROTO(dir)->have_delegation(inode, FMODE_READ))
+	if (nfs_verifier_is_delegated(dentry))
 		return nfs_lookup_revalidate_delegated(dir, dentry, inode);
 
 	/* Force a full look up iff the parent directory has changed */
@@ -1415,6 +1519,7 @@ struct dentry *nfs_lookup(struct inode *dir, struct dentry * dentry, unsigned in
 	struct nfs_fh *fhandle = NULL;
 	struct nfs_fattr *fattr = NULL;
 	struct nfs4_label *label = NULL;
+	unsigned long dir_verifier;
 	int error;
 
 	dfprintk(VFS, "NFS: lookup(%pd2)\n", dentry);
@@ -1440,6 +1545,7 @@ struct dentry *nfs_lookup(struct inode *dir, struct dentry * dentry, unsigned in
 	if (IS_ERR(label))
 		goto out;
 
+	dir_verifier = nfs_save_change_attribute(dir);
 	trace_nfs_lookup_enter(dir, dentry, flags);
 	error = NFS_PROTO(dir)->lookup(dir, dentry, fhandle, fattr, label);
 	if (error == -ENOENT)
@@ -1463,7 +1569,7 @@ no_entry:
 			goto out_label;
 		dentry = res;
 	}
-	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
+	nfs_set_verifier(dentry, dir_verifier);
 out_label:
 	trace_nfs_lookup_exit(dir, dentry, flags, error);
 	nfs4_label_free(label);
@@ -1668,7 +1774,7 @@ nfs4_do_lookup_revalidate(struct inode *dir, struct dentry *dentry,
 	if (inode == NULL)
 		goto full_reval;
 
-	if (NFS_PROTO(dir)->have_delegation(inode, FMODE_READ))
+	if (nfs_verifier_is_delegated(dentry))
 		return nfs_lookup_revalidate_delegated(dir, dentry, inode);
 
 	/* NFS only supports OPEN on regular files */
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 1309e6f47f3d..11bf15800ac9 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -2114,6 +2114,7 @@ static void init_once(void *foo)
 	init_rwsem(&nfsi->rmdir_sem);
 	mutex_init(&nfsi->commit_mutex);
 	nfs4_init_once(nfsi);
+	nfsi->cache_change_attribute = 0;
 }
 
 static int __init nfs_init_inodecache(void)
diff --git a/fs/nfs/nfs4file.c b/fs/nfs/nfs4file.c
index be4eb720d5b6..1297919e0fce 100644
--- a/fs/nfs/nfs4file.c
+++ b/fs/nfs/nfs4file.c
@@ -87,7 +87,6 @@ nfs4_file_open(struct inode *inode, struct file *filp)
 	if (inode != d_inode(dentry))
 		goto out_drop;
 
-	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
 	nfs_file_set_open_context(filp, ctx);
 	nfs_fscache_open_file(inode, filp);
 	err = 0;
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 95d07a3dc5d1..69b7ab7a5815 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -2974,10 +2974,13 @@ static int _nfs4_open_and_get_state(struct nfs4_opendata *opendata,
 	struct dentry *dentry;
 	struct nfs4_state *state;
 	fmode_t acc_mode = _nfs4_ctx_to_accessmode(ctx);
+	struct inode *dir = d_inode(opendata->dir);
+	unsigned long dir_verifier;
 	unsigned int seq;
 	int ret;
 
 	seq = raw_seqcount_begin(&sp->so_reclaim_seqcount);
+	dir_verifier = nfs_save_change_attribute(dir);
 
 	ret = _nfs4_proc_open(opendata, ctx);
 	if (ret != 0)
@@ -3005,8 +3008,19 @@ static int _nfs4_open_and_get_state(struct nfs4_opendata *opendata,
 			dput(ctx->dentry);
 			ctx->dentry = dentry = alias;
 		}
-		nfs_set_verifier(dentry,
-				nfs_save_change_attribute(d_inode(opendata->dir)));
+	}
+
+	switch(opendata->o_arg.claim) {
+	default:
+		break;
+	case NFS4_OPEN_CLAIM_NULL:
+	case NFS4_OPEN_CLAIM_DELEGATE_CUR:
+	case NFS4_OPEN_CLAIM_DELEGATE_PREV:
+		if (!opendata->rpc_done)
+			break;
+		if (opendata->o_res.delegation_type != 0)
+			dir_verifier = nfs_save_change_attribute(dir);
+		nfs_set_verifier(dentry, dir_verifier);
 	}
 
 	/* Parse layoutget results before we check for access */
@@ -5322,7 +5336,7 @@ static void nfs4_proc_write_setup(struct nfs_pgio_header *hdr,
 	hdr->timestamp   = jiffies;
 
 	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE];
-	nfs4_init_sequence(&hdr->args.seq_args, &hdr->res.seq_res, 1, 0);
+	nfs4_init_sequence(&hdr->args.seq_args, &hdr->res.seq_res, 0, 0);
 	nfs4_state_protect_write(server->nfs_client, clnt, msg, hdr);
 }
 
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index a5f8f03ecd59..5d5b91e54f73 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -337,35 +337,17 @@ static inline int nfs_server_capable(struct inode *inode, int cap)
 	return NFS_SERVER(inode)->caps & cap;
 }
 
-static inline void nfs_set_verifier(struct dentry * dentry, unsigned long verf)
-{
-	dentry->d_time = verf;
-}
-
 /**
  * nfs_save_change_attribute - Returns the inode attribute change cookie
  * @dir - pointer to parent directory inode
- * The "change attribute" is updated every time we finish an operation
- * that will result in a metadata change on the server.
+ * The "cache change attribute" is updated when we need to revalidate
+ * our dentry cache after a directory was seen to change on the server.
  */
 static inline unsigned long nfs_save_change_attribute(struct inode *dir)
 {
 	return NFS_I(dir)->cache_change_attribute;
 }
 
-/**
- * nfs_verify_change_attribute - Detects NFS remote directory changes
- * @dir - pointer to parent directory inode
- * @chattr - previously saved change attribute
- * Return "false" if the verifiers doesn't match the change attribute.
- * This would usually indicate that the directory contents have changed on
- * the server, and that any dentries need revalidating.
- */
-static inline int nfs_verify_change_attribute(struct inode *dir, unsigned long chattr)
-{
-	return chattr == NFS_I(dir)->cache_change_attribute;
-}
-
 /*
  * linux/fs/nfs/inode.c
  */
@@ -495,6 +477,10 @@ extern const struct file_operations nfs_dir_operations;
 extern const struct dentry_operations nfs_dentry_operations;
 
 extern void nfs_force_lookup_revalidate(struct inode *dir);
+extern void nfs_set_verifier(struct dentry * dentry, unsigned long verf);
+#if IS_ENABLED(CONFIG_NFS_V4)
+extern void nfs_clear_verifier_delegated(struct inode *inode);
+#endif /* IS_ENABLED(CONFIG_NFS_V4) */
 extern struct dentry *nfs_add_or_obtain(struct dentry *dentry,
 			struct nfs_fh *fh, struct nfs_fattr *fattr,
 			struct nfs4_label *label);
diff --git a/net/sunrpc/xprtrdma/frwr_ops.c b/net/sunrpc/xprtrdma/frwr_ops.c
index 095be887753e..125297c9aa3e 100644
--- a/net/sunrpc/xprtrdma/frwr_ops.c
+++ b/net/sunrpc/xprtrdma/frwr_ops.c
@@ -288,8 +288,8 @@ struct rpcrdma_mr_seg *frwr_map(struct rpcrdma_xprt *r_xprt,
 {
 	struct rpcrdma_ia *ia = &r_xprt->rx_ia;
 	struct ib_reg_wr *reg_wr;
+	int i, n, dma_nents;
 	struct ib_mr *ibmr;
-	int i, n;
 	u8 key;
 
 	if (nsegs > ia->ri_max_frwr_depth)
@@ -313,15 +313,16 @@ struct rpcrdma_mr_seg *frwr_map(struct rpcrdma_xprt *r_xprt,
 			break;
 	}
 	mr->mr_dir = rpcrdma_data_dir(writing);
+	mr->mr_nents = i;
 
-	mr->mr_nents =
-		ib_dma_map_sg(ia->ri_id->device, mr->mr_sg, i, mr->mr_dir);
-	if (!mr->mr_nents)
+	dma_nents = ib_dma_map_sg(ia->ri_id->device, mr->mr_sg, mr->mr_nents,
+				  mr->mr_dir);
+	if (!dma_nents)
 		goto out_dmamap_err;
 
 	ibmr = mr->frwr.fr_mr;
-	n = ib_map_mr_sg(ibmr, mr->mr_sg, mr->mr_nents, NULL, PAGE_SIZE);
-	if (unlikely(n != mr->mr_nents))
+	n = ib_map_mr_sg(ibmr, mr->mr_sg, dma_nents, NULL, PAGE_SIZE);
+	if (n != dma_nents)
 		goto out_mapmr_err;
 
 	ibmr->iova &= 0x00000000ffffffff;