Merge commit 'v2.6.28-rc2' into tracing/urgent

6 files changed, 220 insertions, 126 deletions

diff --git a/fs/nfsd/export.c b/fs/nfsd/export.c
index 9dc036f18356..5839b229cd0e 100644
--- a/fs/nfsd/export.c
+++ b/fs/nfsd/export.c
@@ -99,7 +99,7 @@ static int expkey_parse(struct cache_detail *cd, char *mesg, int mlen)
 	int fsidtype;
 	char *ep;
 	struct svc_expkey key;
-	struct svc_expkey *ek;
+	struct svc_expkey *ek = NULL;
 
 	if (mesg[mlen-1] != '\n')
 		return -EINVAL;
@@ -107,7 +107,8 @@ static int expkey_parse(struct cache_detail *cd, char *mesg, int mlen)
 
 	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
 	err = -ENOMEM;
-	if (!buf) goto out;
+	if (!buf)
+		goto out;
 
 	err = -EINVAL;
 	if ((len=qword_get(&mesg, buf, PAGE_SIZE)) <= 0)
@@ -151,34 +152,32 @@ static int expkey_parse(struct cache_detail *cd, char *mesg, int mlen)
 
 	/* now we want a pathname, or empty meaning NEGATIVE  */
 	err = -EINVAL;
-	if ((len=qword_get(&mesg, buf, PAGE_SIZE)) < 0)
+	len = qword_get(&mesg, buf, PAGE_SIZE);
+	if (len < 0)
 		goto out;
 	dprintk("Path seems to be <%s>\n", buf);
 	err = 0;
 	if (len == 0) {
 		set_bit(CACHE_NEGATIVE, &key.h.flags);
 		ek = svc_expkey_update(&key, ek);
-		if (ek)
-			cache_put(&ek->h, &svc_expkey_cache);
-		else err = -ENOMEM;
+		if (!ek)
+			err = -ENOMEM;
 	} else {
-		struct nameidata nd;
-		err = path_lookup(buf, 0, &nd);
+		err = kern_path(buf, 0, &key.ek_path);
 		if (err)
 			goto out;
 
 		dprintk("Found the path %s\n", buf);
-		key.ek_path = nd.path;
 
 		ek = svc_expkey_update(&key, ek);
-		if (ek)
-			cache_put(&ek->h, &svc_expkey_cache);
-		else
+		if (!ek)
 			err = -ENOMEM;
-		path_put(&nd.path);
+		path_put(&key.ek_path);
 	}
 	cache_flush();
  out:
+	if (ek)
+		cache_put(&ek->h, &svc_expkey_cache);
 	if (dom)
 		auth_domain_put(dom);
 	kfree(buf);
@@ -500,35 +499,22 @@ static int svc_export_parse(struct cache_detail *cd, char *mesg, int mlen)
 	int len;
 	int err;
 	struct auth_domain *dom = NULL;
-	struct nameidata nd;
-	struct svc_export exp, *expp;
+	struct svc_export exp = {}, *expp;
 	int an_int;
 
-	nd.path.dentry = NULL;
-	exp.ex_pathname = NULL;
-
-	/* fs locations */
-	exp.ex_fslocs.locations = NULL;
-	exp.ex_fslocs.locations_count = 0;
-	exp.ex_fslocs.migrated = 0;
-
-	exp.ex_uuid = NULL;
-
-	/* secinfo */
-	exp.ex_nflavors = 0;
-
 	if (mesg[mlen-1] != '\n')
 		return -EINVAL;
 	mesg[mlen-1] = 0;
 
 	buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
-	err = -ENOMEM;
-	if (!buf) goto out;
+	if (!buf)
+		return -ENOMEM;
 
 	/* client */
-	len = qword_get(&mesg, buf, PAGE_SIZE);
 	err = -EINVAL;
-	if (len <= 0) goto out;
+	len = qword_get(&mesg, buf, PAGE_SIZE);
+	if (len <= 0)
+		goto out;
 
 	err = -ENOENT;
 	dom = auth_domain_find(buf);
@@ -537,25 +523,25 @@ static int svc_export_parse(struct cache_detail *cd, char *mesg, int mlen)
 
 	/* path */
 	err = -EINVAL;
-	if ((len=qword_get(&mesg, buf, PAGE_SIZE)) <= 0)
-		goto out;
-	err = path_lookup(buf, 0, &nd);
-	if (err) goto out_no_path;
+	if ((len = qword_get(&mesg, buf, PAGE_SIZE)) <= 0)
+		goto out1;
+
+	err = kern_path(buf, 0, &exp.ex_path);
+	if (err)
+		goto out1;
 
-	exp.h.flags = 0;
 	exp.ex_client = dom;
-	exp.ex_path.mnt = nd.path.mnt;
-	exp.ex_path.dentry = nd.path.dentry;
-	exp.ex_pathname = kstrdup(buf, GFP_KERNEL);
+
 	err = -ENOMEM;
+	exp.ex_pathname = kstrdup(buf, GFP_KERNEL);
 	if (!exp.ex_pathname)
-		goto out;
+		goto out2;
 
 	/* expiry */
 	err = -EINVAL;
 	exp.h.expiry_time = get_expiry(&mesg);
 	if (exp.h.expiry_time == 0)
-		goto out;
+		goto out3;
 
 	/* flags */
 	err = get_int(&mesg, &an_int);
@@ -563,22 +549,26 @@ static int svc_export_parse(struct cache_detail *cd, char *mesg, int mlen)
 		err = 0;
 		set_bit(CACHE_NEGATIVE, &exp.h.flags);
 	} else {
-		if (err || an_int < 0) goto out;	
+		if (err || an_int < 0)
+			goto out3;
 		exp.ex_flags= an_int;
 	
 		/* anon uid */
 		err = get_int(&mesg, &an_int);
-		if (err) goto out;
+		if (err)
+			goto out3;
 		exp.ex_anon_uid= an_int;
 
 		/* anon gid */
 		err = get_int(&mesg, &an_int);
-		if (err) goto out;
+		if (err)
+			goto out3;
 		exp.ex_anon_gid= an_int;
 
 		/* fsid */
 		err = get_int(&mesg, &an_int);
-		if (err) goto out;
+		if (err)
+			goto out3;
 		exp.ex_fsid = an_int;
 
 		while ((len = qword_get(&mesg, buf, PAGE_SIZE)) > 0) {
@@ -604,12 +594,13 @@ static int svc_export_parse(struct cache_detail *cd, char *mesg, int mlen)
 				 */
 				break;
 			if (err)
-				goto out;
+				goto out4;
 		}
 
-		err = check_export(nd.path.dentry->d_inode, exp.ex_flags,
+		err = check_export(exp.ex_path.dentry->d_inode, exp.ex_flags,
 				   exp.ex_uuid);
-		if (err) goto out;
+		if (err)
+			goto out4;
 	}
 
 	expp = svc_export_lookup(&exp);
@@ -622,15 +613,16 @@ static int svc_export_parse(struct cache_detail *cd, char *mesg, int mlen)
 		err = -ENOMEM;
 	else
 		exp_put(expp);
- out:
+out4:
 	nfsd4_fslocs_free(&exp.ex_fslocs);
 	kfree(exp.ex_uuid);
+out3:
 	kfree(exp.ex_pathname);
-	if (nd.path.dentry)
-		path_put(&nd.path);
- out_no_path:
-	if (dom)
-		auth_domain_put(dom);
+out2:
+	path_put(&exp.ex_path);
+out1:
+	auth_domain_put(dom);
+out:
 	kfree(buf);
 	return err;
 }
@@ -998,7 +990,7 @@ exp_export(struct nfsctl_export *nxp)
 	struct svc_export	*exp = NULL;
 	struct svc_export	new;
 	struct svc_expkey	*fsid_key = NULL;
-	struct nameidata nd;
+	struct path path;
 	int		err;
 
 	/* Consistency check */
@@ -1021,12 +1013,12 @@ exp_export(struct nfsctl_export *nxp)
 
 
 	/* Look up the dentry */
-	err = path_lookup(nxp->ex_path, 0, &nd);
+	err = kern_path(nxp->ex_path, 0, &path);
 	if (err)
 		goto out_put_clp;
 	err = -EINVAL;
 
-	exp = exp_get_by_name(clp, nd.path.mnt, nd.path.dentry, NULL);
+	exp = exp_get_by_name(clp, path.mnt, path.dentry, NULL);
 
 	memset(&new, 0, sizeof(new));
 
@@ -1034,8 +1026,8 @@ exp_export(struct nfsctl_export *nxp)
 	if ((nxp->ex_flags & NFSEXP_FSID) &&
 	    (!IS_ERR(fsid_key = exp_get_fsid_key(clp, nxp->ex_dev))) &&
 	    fsid_key->ek_path.mnt &&
-	    (fsid_key->ek_path.mnt != nd.path.mnt ||
-	     fsid_key->ek_path.dentry != nd.path.dentry))
+	    (fsid_key->ek_path.mnt != path.mnt ||
+	     fsid_key->ek_path.dentry != path.dentry))
 		goto finish;
 
 	if (!IS_ERR(exp)) {
@@ -1051,7 +1043,7 @@ exp_export(struct nfsctl_export *nxp)
 		goto finish;
 	}
 
-	err = check_export(nd.path.dentry->d_inode, nxp->ex_flags, NULL);
+	err = check_export(path.dentry->d_inode, nxp->ex_flags, NULL);
 	if (err) goto finish;
 
 	err = -ENOMEM;
@@ -1064,7 +1056,7 @@ exp_export(struct nfsctl_export *nxp)
 	if (!new.ex_pathname)
 		goto finish;
 	new.ex_client = clp;
-	new.ex_path = nd.path;
+	new.ex_path = path;
 	new.ex_flags = nxp->ex_flags;
 	new.ex_anon_uid = nxp->ex_anon_uid;
 	new.ex_anon_gid = nxp->ex_anon_gid;
@@ -1090,7 +1082,7 @@ finish:
 		exp_put(exp);
 	if (fsid_key && !IS_ERR(fsid_key))
 		cache_put(&fsid_key->h, &svc_expkey_cache);
-	path_put(&nd.path);
+	path_put(&path);
 out_put_clp:
 	auth_domain_put(clp);
 out_unlock:
@@ -1121,7 +1113,7 @@ exp_unexport(struct nfsctl_export *nxp)
 {
 	struct auth_domain *dom;
 	svc_export *exp;
-	struct nameidata nd;
+	struct path path;
 	int		err;
 
 	/* Consistency check */
@@ -1138,13 +1130,13 @@ exp_unexport(struct nfsctl_export *nxp)
 		goto out_unlock;
 	}
 
-	err = path_lookup(nxp->ex_path, 0, &nd);
+	err = kern_path(nxp->ex_path, 0, &path);
 	if (err)
 		goto out_domain;
 
 	err = -EINVAL;
-	exp = exp_get_by_name(dom, nd.path.mnt, nd.path.dentry, NULL);
-	path_put(&nd.path);
+	exp = exp_get_by_name(dom, path.mnt, path.dentry, NULL);
+	path_put(&path);
 	if (IS_ERR(exp))
 		goto out_domain;
 
@@ -1166,26 +1158,26 @@ out_unlock:
  * since its harder to fool a kernel module than a user space program.
  */
 int
-exp_rootfh(svc_client *clp, char *path, struct knfsd_fh *f, int maxsize)
+exp_rootfh(svc_client *clp, char *name, struct knfsd_fh *f, int maxsize)
 {
 	struct svc_export	*exp;
-	struct nameidata	nd;
+	struct path		path;
 	struct inode		*inode;
 	struct svc_fh		fh;
 	int			err;
 
 	err = -EPERM;
 	/* NB: we probably ought to check that it's NUL-terminated */
-	if (path_lookup(path, 0, &nd)) {
-		printk("nfsd: exp_rootfh path not found %s", path);
+	if (kern_path(name, 0, &path)) {
+		printk("nfsd: exp_rootfh path not found %s", name);
 		return err;
 	}
-	inode = nd.path.dentry->d_inode;
+	inode = path.dentry->d_inode;
 
 	dprintk("nfsd: exp_rootfh(%s [%p] %s:%s/%ld)\n",
-		 path, nd.path.dentry, clp->name,
+		 name, path.dentry, clp->name,
 		 inode->i_sb->s_id, inode->i_ino);
-	exp = exp_parent(clp, nd.path.mnt, nd.path.dentry, NULL);
+	exp = exp_parent(clp, path.mnt, path.dentry, NULL);
 	if (IS_ERR(exp)) {
 		err = PTR_ERR(exp);
 		goto out;
@@ -1195,7 +1187,7 @@ exp_rootfh(svc_client *clp, char *path, struct knfsd_fh *f, int maxsize)
 	 * fh must be initialized before calling fh_compose
 	 */
 	fh_init(&fh, maxsize);
-	if (fh_compose(&fh, exp, nd.path.dentry, NULL))
+	if (fh_compose(&fh, exp, path.dentry, NULL))
 		err = -EINVAL;
 	else
 		err = 0;
@@ -1203,7 +1195,7 @@ exp_rootfh(svc_client *clp, char *path, struct knfsd_fh *f, int maxsize)
 	fh_put(&fh);
 	exp_put(exp);
 out:
-	path_put(&nd.path);
+	path_put(&path);
 	return err;
 }
 
diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
index 145b3c877a27..bb93946ace22 100644
--- a/fs/nfsd/nfs4recover.c
+++ b/fs/nfsd/nfs4recover.c
@@ -51,7 +51,7 @@
 #define NFSDDBG_FACILITY                NFSDDBG_PROC
 
 /* Globals */
-static struct nameidata rec_dir;
+static struct path rec_dir;
 static int rec_dir_init = 0;
 
 static void
@@ -121,9 +121,9 @@ out_no_tfm:
 static void
 nfsd4_sync_rec_dir(void)
 {
-	mutex_lock(&rec_dir.path.dentry->d_inode->i_mutex);
-	nfsd_sync_dir(rec_dir.path.dentry);
-	mutex_unlock(&rec_dir.path.dentry->d_inode->i_mutex);
+	mutex_lock(&rec_dir.dentry->d_inode->i_mutex);
+	nfsd_sync_dir(rec_dir.dentry);
+	mutex_unlock(&rec_dir.dentry->d_inode->i_mutex);
 }
 
 int
@@ -143,9 +143,9 @@ nfsd4_create_clid_dir(struct nfs4_client *clp)
 	nfs4_save_user(&uid, &gid);
 
 	/* lock the parent */
-	mutex_lock(&rec_dir.path.dentry->d_inode->i_mutex);
+	mutex_lock(&rec_dir.dentry->d_inode->i_mutex);
 
-	dentry = lookup_one_len(dname, rec_dir.path.dentry, HEXDIR_LEN-1);
+	dentry = lookup_one_len(dname, rec_dir.dentry, HEXDIR_LEN-1);
 	if (IS_ERR(dentry)) {
 		status = PTR_ERR(dentry);
 		goto out_unlock;
@@ -155,15 +155,15 @@ nfsd4_create_clid_dir(struct nfs4_client *clp)
 		dprintk("NFSD: nfsd4_create_clid_dir: DIRECTORY EXISTS\n");
 		goto out_put;
 	}
-	status = mnt_want_write(rec_dir.path.mnt);
+	status = mnt_want_write(rec_dir.mnt);
 	if (status)
 		goto out_put;
-	status = vfs_mkdir(rec_dir.path.dentry->d_inode, dentry, S_IRWXU);
-	mnt_drop_write(rec_dir.path.mnt);
+	status = vfs_mkdir(rec_dir.dentry->d_inode, dentry, S_IRWXU);
+	mnt_drop_write(rec_dir.mnt);
 out_put:
 	dput(dentry);
 out_unlock:
-	mutex_unlock(&rec_dir.path.dentry->d_inode->i_mutex);
+	mutex_unlock(&rec_dir.dentry->d_inode->i_mutex);
 	if (status == 0) {
 		clp->cl_firststate = 1;
 		nfsd4_sync_rec_dir();
@@ -226,7 +226,7 @@ nfsd4_list_rec_dir(struct dentry *dir, recdir_func *f)
 
 	nfs4_save_user(&uid, &gid);
 
-	filp = dentry_open(dget(dir), mntget(rec_dir.path.mnt), O_RDONLY);
+	filp = dentry_open(dget(dir), mntget(rec_dir.mnt), O_RDONLY);
 	status = PTR_ERR(filp);
 	if (IS_ERR(filp))
 		goto out;
@@ -291,9 +291,9 @@ nfsd4_unlink_clid_dir(char *name, int namlen)
 
 	dprintk("NFSD: nfsd4_unlink_clid_dir. name %.*s\n", namlen, name);
 
-	mutex_lock(&rec_dir.path.dentry->d_inode->i_mutex);
-	dentry = lookup_one_len(name, rec_dir.path.dentry, namlen);
-	mutex_unlock(&rec_dir.path.dentry->d_inode->i_mutex);
+	mutex_lock(&rec_dir.dentry->d_inode->i_mutex);
+	dentry = lookup_one_len(name, rec_dir.dentry, namlen);
+	mutex_unlock(&rec_dir.dentry->d_inode->i_mutex);
 	if (IS_ERR(dentry)) {
 		status = PTR_ERR(dentry);
 		return status;
@@ -302,7 +302,7 @@ nfsd4_unlink_clid_dir(char *name, int namlen)
 	if (!dentry->d_inode)
 		goto out;
 
-	status = nfsd4_clear_clid_dir(rec_dir.path.dentry, dentry);
+	status = nfsd4_clear_clid_dir(rec_dir.dentry, dentry);
 out:
 	dput(dentry);
 	return status;
@@ -318,7 +318,7 @@ nfsd4_remove_clid_dir(struct nfs4_client *clp)
 	if (!rec_dir_init || !clp->cl_firststate)
 		return;
 
-	status = mnt_want_write(rec_dir.path.mnt);
+	status = mnt_want_write(rec_dir.mnt);
 	if (status)
 		goto out;
 	clp->cl_firststate = 0;
@@ -327,7 +327,7 @@ nfsd4_remove_clid_dir(struct nfs4_client *clp)
 	nfs4_reset_user(uid, gid);
 	if (status == 0)
 		nfsd4_sync_rec_dir();
-	mnt_drop_write(rec_dir.path.mnt);
+	mnt_drop_write(rec_dir.mnt);
 out:
 	if (status)
 		printk("NFSD: Failed to remove expired client state directory"
@@ -357,17 +357,17 @@ nfsd4_recdir_purge_old(void) {
 
 	if (!rec_dir_init)
 		return;
-	status = mnt_want_write(rec_dir.path.mnt);
+	status = mnt_want_write(rec_dir.mnt);
 	if (status)
 		goto out;
-	status = nfsd4_list_rec_dir(rec_dir.path.dentry, purge_old);
+	status = nfsd4_list_rec_dir(rec_dir.dentry, purge_old);
 	if (status == 0)
 		nfsd4_sync_rec_dir();
-	mnt_drop_write(rec_dir.path.mnt);
+	mnt_drop_write(rec_dir.mnt);
 out:
 	if (status)
 		printk("nfsd4: failed to purge old clients from recovery"
-			" directory %s\n", rec_dir.path.dentry->d_name.name);
+			" directory %s\n", rec_dir.dentry->d_name.name);
 }
 
 static int
@@ -387,10 +387,10 @@ int
 nfsd4_recdir_load(void) {
 	int status;
 
-	status = nfsd4_list_rec_dir(rec_dir.path.dentry, load_recdir);
+	status = nfsd4_list_rec_dir(rec_dir.dentry, load_recdir);
 	if (status)
 		printk("nfsd4: failed loading clients from recovery"
-			" directory %s\n", rec_dir.path.dentry->d_name.name);
+			" directory %s\n", rec_dir.dentry->d_name.name);
 	return status;
 }
 
@@ -412,7 +412,7 @@ nfsd4_init_recdir(char *rec_dirname)
 
 	nfs4_save_user(&uid, &gid);
 
-	status = path_lookup(rec_dirname, LOOKUP_FOLLOW | LOOKUP_DIRECTORY,
+	status = kern_path(rec_dirname, LOOKUP_FOLLOW | LOOKUP_DIRECTORY,
 			&rec_dir);
 	if (status)
 		printk("NFSD: unable to find recovery directory %s\n",
@@ -429,5 +429,5 @@ nfsd4_shutdown_recdir(void)
 	if (!rec_dir_init)
 		return;
 	rec_dir_init = 0;
-	path_put(&rec_dir.path);
+	path_put(&rec_dir);
 }
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 0cc7ff5d5ab5..b0bebc552a11 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -3284,17 +3284,17 @@ int
 nfs4_reset_recoverydir(char *recdir)
 {
 	int status;
-	struct nameidata nd;
+	struct path path;
 
-	status = path_lookup(recdir, LOOKUP_FOLLOW, &nd);
+	status = kern_path(recdir, LOOKUP_FOLLOW, &path);
 	if (status)
 		return status;
 	status = -ENOTDIR;
-	if (S_ISDIR(nd.path.dentry->d_inode->i_mode)) {
+	if (S_ISDIR(path.dentry->d_inode->i_mode)) {
 		nfs4_set_recdir(recdir);
 		status = 0;
 	}
-	path_put(&nd.path);
+	path_put(&path);
 	return status;
 }
 
diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
index 97543df58242..e3f9783fdcf7 100644
--- a/fs/nfsd/nfsctl.c
+++ b/fs/nfsd/nfsctl.c
@@ -341,7 +341,7 @@ static ssize_t failover_unlock_ip(struct file *file, char *buf, size_t size)
 
 static ssize_t failover_unlock_fs(struct file *file, char *buf, size_t size)
 {
-	struct nameidata nd;
+	struct path path;
 	char *fo_path;
 	int error;
 
@@ -356,13 +356,13 @@ static ssize_t failover_unlock_fs(struct file *file, char *buf, size_t size)
 	if (qword_get(&buf, fo_path, size) < 0)
 		return -EINVAL;
 
-	error = path_lookup(fo_path, 0, &nd);
+	error = kern_path(fo_path, 0, &path);
 	if (error)
 		return error;
 
-	error = nlmsvc_unlock_all_by_sb(nd.path.mnt->mnt_sb);
+	error = nlmsvc_unlock_all_by_sb(path.mnt->mnt_sb);
 
-	path_put(&nd.path);
+	path_put(&path);
 	return error;
 }
 
diff --git a/fs/nfsd/nfssvc.c b/fs/nfsd/nfssvc.c
index 59eeb46f82c5..07e4f5d7baa8 100644
--- a/fs/nfsd/nfssvc.c
+++ b/fs/nfsd/nfssvc.c
@@ -249,6 +249,10 @@ static int nfsd_init_socks(int port)
 	if (error < 0)
 		return error;
 
+	error = lockd_up();
+	if (error < 0)
+		return error;
+
 	error = svc_create_xprt(nfsd_serv, "tcp", port,
 					SVC_SOCK_DEFAULTS);
 	if (error < 0)
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index aa1d0d6489a1..0bc56f6d9276 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -410,6 +410,7 @@ out_nfserr:
 static ssize_t nfsd_getxattr(struct dentry *dentry, char *key, void **buf)
 {
 	ssize_t buflen;
+	ssize_t ret;
 
 	buflen = vfs_getxattr(dentry, key, NULL, 0);
 	if (buflen <= 0)
@@ -419,7 +420,10 @@ static ssize_t nfsd_getxattr(struct dentry *dentry, char *key, void **buf)
 	if (!*buf)
 		return -ENOMEM;
 
-	return vfs_getxattr(dentry, key, *buf, buflen);
+	ret = vfs_getxattr(dentry, key, *buf, buflen);
+	if (ret < 0)
+		kfree(*buf);
+	return ret;
 }
 #endif
 
@@ -1814,6 +1818,115 @@ out:
 }
 
 /*
+ * We do this buffering because we must not call back into the file
+ * system's ->lookup() method from the filldir callback. That may well
+ * deadlock a number of file systems.
+ *
+ * This is based heavily on the implementation of same in XFS.
+ */
+struct buffered_dirent {
+	u64		ino;
+	loff_t		offset;
+	int		namlen;
+	unsigned int	d_type;
+	char		name[];
+};
+
+struct readdir_data {
+	char		*dirent;
+	size_t		used;
+	int		full;
+};
+
+static int nfsd_buffered_filldir(void *__buf, const char *name, int namlen,
+				 loff_t offset, u64 ino, unsigned int d_type)
+{
+	struct readdir_data *buf = __buf;
+	struct buffered_dirent *de = (void *)(buf->dirent + buf->used);
+	unsigned int reclen;
+
+	reclen = ALIGN(sizeof(struct buffered_dirent) + namlen, sizeof(u64));
+	if (buf->used + reclen > PAGE_SIZE) {
+		buf->full = 1;
+		return -EINVAL;
+	}
+
+	de->namlen = namlen;
+	de->offset = offset;
+	de->ino = ino;
+	de->d_type = d_type;
+	memcpy(de->name, name, namlen);
+	buf->used += reclen;
+
+	return 0;
+}
+
+static int nfsd_buffered_readdir(struct file *file, filldir_t func,
+				 struct readdir_cd *cdp, loff_t *offsetp)
+{
+	struct readdir_data buf;
+	struct buffered_dirent *de;
+	int host_err;
+	int size;
+	loff_t offset;
+
+	buf.dirent = (void *)__get_free_page(GFP_KERNEL);
+	if (!buf.dirent)
+		return -ENOMEM;
+
+	offset = *offsetp;
+	cdp->err = nfserr_eof; /* will be cleared on successful read */
+
+	while (1) {
+		unsigned int reclen;
+
+		buf.used = 0;
+		buf.full = 0;
+
+		host_err = vfs_readdir(file, nfsd_buffered_filldir, &buf);
+		if (buf.full)
+			host_err = 0;
+
+		if (host_err < 0)
+			break;
+
+		size = buf.used;
+
+		if (!size)
+			break;
+
+		de = (struct buffered_dirent *)buf.dirent;
+		while (size > 0) {
+			offset = de->offset;
+
+			if (func(cdp, de->name, de->namlen, de->offset,
+				 de->ino, de->d_type))
+				goto done;
+
+			if (cdp->err != nfs_ok)
+				goto done;
+
+			reclen = ALIGN(sizeof(*de) + de->namlen,
+				       sizeof(u64));
+			size -= reclen;
+			de = (struct buffered_dirent *)((char *)de + reclen);
+		}
+		offset = vfs_llseek(file, 0, SEEK_CUR);
+		if (!buf.full)
+			break;
+	}
+
+ done:
+	free_page((unsigned long)(buf.dirent));
+
+	if (host_err)
+		return nfserrno(host_err);
+
+	*offsetp = offset;
+	return cdp->err;
+}
+
+/*
  * Read entries from a directory.
  * The  NFSv3/4 verifier we ignore for now.
  */
@@ -1822,7 +1935,6 @@ nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp,
 	     struct readdir_cd *cdp, filldir_t func)
 {
 	__be32		err;
-	int 		host_err;
 	struct file	*file;
 	loff_t		offset = *offsetp;
 
@@ -1836,21 +1948,7 @@ nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp,
 		goto out_close;
 	}
 
-	/*
-	 * Read the directory entries. This silly loop is necessary because
-	 * readdir() is not guaranteed to fill up the entire buffer, but
-	 * may choose to do less.
-	 */
-
-	do {
-		cdp->err = nfserr_eof; /* will be cleared on successful read */
-		host_err = vfs_readdir(file, func, cdp);
-	} while (host_err >=0 && cdp->err == nfs_ok);
-	if (host_err)
-		err = nfserrno(host_err);
-	else
-		err = cdp->err;
-	*offsetp = vfs_llseek(file, 0, 1);
+	err = nfsd_buffered_readdir(file, func, cdp, offsetp);
 
 	if (err == nfserr_eof || err == nfserr_toosmall)
 		err = nfs_ok; /* can still be found in ->err */