nfs: add export operations

This support for opening files on NFS by file handle, both through the
open_by_handle syscall, and for re-exporting NFS (for example using a
different version).  The support is very basic for now, as each open by
handle will have to do an NFSv4 open operation on the wire.  In the
future this will hopefully be mitigated by an open file cache, as well
as various optimizations in NFS for this specific case.

Signed-off-by: Peng Tao <tao.peng@primarydata.com>
[hch: incorporated various changes, resplit the patches, new changelog]
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>

4 files changed, 182 insertions, 1 deletions

diff --git a/fs/nfs/Makefile b/fs/nfs/Makefile
index 98f4e57..1fb1189 100644
--- a/fs/nfs/Makefile
+++ b/fs/nfs/Makefile
@@ -7,7 +7,7 @@ obj-$(CONFIG_NFS_FS) += nfs.o
 CFLAGS_nfstrace.o += -I$(src)
 nfs-y 			:= client.o dir.o file.o getroot.o inode.o super.o \
 			   io.o direct.o pagelist.o read.o symlink.o unlink.o \
-			   write.o namespace.o mount_clnt.o nfstrace.o
+			   write.o namespace.o mount_clnt.o nfstrace.o export.o
 nfs-$(CONFIG_ROOT_NFS)	+= nfsroot.o
 nfs-$(CONFIG_SYSCTL)	+= sysctl.o
 nfs-$(CONFIG_NFS_FSCACHE) += fscache.o fscache-index.o
diff --git a/fs/nfs/export.c b/fs/nfs/export.c
new file mode 100644
index 0000000..249cb96
--- /dev/null
+++ b/fs/nfs/export.c
@@ -0,0 +1,177 @@
+/*
+ * Copyright (c) 2015, Primary Data, Inc. All rights reserved.
+ *
+ * Tao Peng <bergwolf@primarydata.com>
+ */
+#include <linux/dcache.h>
+#include <linux/exportfs.h>
+#include <linux/nfs.h>
+#include <linux/nfs_fs.h>
+
+#include "internal.h"
+#include "nfstrace.h"
+
+#define NFSDBG_FACILITY		NFSDBG_VFS
+
+enum {
+	FILEID_HIGH_OFF = 0,	/* inode fileid high */
+	FILEID_LOW_OFF,		/* inode fileid low */
+	FILE_I_TYPE_OFF,	/* inode type */
+	EMBED_FH_OFF		/* embeded server fh */
+};
+
+
+static struct nfs_fh *nfs_exp_embedfh(__u32 *p)
+{
+	return (struct nfs_fh *)(p + EMBED_FH_OFF);
+}
+
+/*
+ * Let's break subtree checking for now... otherwise we'll have to embed parent fh
+ * but there might not be enough space.
+ */
+static int
+nfs_encode_fh(struct inode *inode, __u32 *p, int *max_len, struct inode *parent)
+{
+	struct nfs_fh *server_fh = NFS_FH(inode);
+	struct nfs_fh *clnt_fh = nfs_exp_embedfh(p);
+	size_t fh_size = offsetof(struct nfs_fh, data) + server_fh->size;
+	int len = EMBED_FH_OFF + XDR_QUADLEN(fh_size);
+
+	dprintk("%s: max fh len %d inode %p parent %p",
+		__func__, *max_len, inode, parent);
+
+	if (*max_len < len || IS_AUTOMOUNT(inode)) {
+		dprintk("%s: fh len %d too small, required %d\n",
+			__func__, *max_len, len);
+		*max_len = len;
+		return FILEID_INVALID;
+	}
+	if (IS_AUTOMOUNT(inode)) {
+		*max_len = FILEID_INVALID;
+		goto out;
+	}
+
+	p[FILEID_HIGH_OFF] = NFS_FILEID(inode) >> 32;
+	p[FILEID_LOW_OFF] = NFS_FILEID(inode);
+	p[FILE_I_TYPE_OFF] = inode->i_mode & S_IFMT;
+	p[len - 1] = 0; /* Padding */
+	nfs_copy_fh(clnt_fh, server_fh);
+	*max_len = len;
+out:
+	dprintk("%s: result fh fileid %llu mode %u size %d\n",
+		__func__, NFS_FILEID(inode), inode->i_mode, *max_len);
+	return *max_len;
+}
+
+static struct dentry *
+nfs_fh_to_dentry(struct super_block *sb, struct fid *fid,
+		 int fh_len, int fh_type)
+{
+	struct nfs4_label *label = NULL;
+	struct nfs_fattr *fattr = NULL;
+	struct nfs_fh *server_fh = nfs_exp_embedfh(fid->raw);
+	size_t fh_size = offsetof(struct nfs_fh, data) + server_fh->size;
+	const struct nfs_rpc_ops *rpc_ops;
+	struct dentry *dentry;
+	struct inode *inode;
+	int len = EMBED_FH_OFF + XDR_QUADLEN(fh_size);
+	u32 *p = fid->raw;
+	int ret;
+
+	/* NULL translates to ESTALE */
+	if (fh_len < len || fh_type != len)
+		return NULL;
+
+	fattr = nfs_alloc_fattr();
+	if (fattr == NULL) {
+		dentry = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+
+	fattr->fileid = ((u64)p[FILEID_HIGH_OFF] << 32) + p[FILEID_LOW_OFF];
+	fattr->mode = p[FILE_I_TYPE_OFF];
+	fattr->valid |= NFS_ATTR_FATTR_FILEID | NFS_ATTR_FATTR_TYPE;
+
+	dprintk("%s: fileid %llu mode %d\n", __func__, fattr->fileid, fattr->mode);
+
+	inode = nfs_ilookup(sb, fattr, server_fh);
+	if (inode)
+		goto out_found;
+
+	label = nfs4_label_alloc(NFS_SB(sb), GFP_KERNEL);
+	if (IS_ERR(label)) {
+		dentry = ERR_CAST(label);
+		goto out_free_fattr;
+	}
+
+	rpc_ops = NFS_SB(sb)->nfs_client->rpc_ops;
+	ret = rpc_ops->getattr(NFS_SB(sb), server_fh, fattr, label);
+	if (ret) {
+		dprintk("%s: getattr failed %d\n", __func__, ret);
+		dentry = ERR_PTR(ret);
+		goto out_free_label;
+	}
+
+	inode = nfs_fhget(sb, server_fh, fattr, label);
+
+out_found:
+	dentry = d_obtain_alias(inode);
+
+out_free_label:
+	nfs4_label_free(label);
+out_free_fattr:
+	nfs_free_fattr(fattr);
+out:
+	return dentry;
+}
+
+static struct dentry *
+nfs_get_parent(struct dentry *dentry)
+{
+	int ret;
+	struct inode *inode = d_inode(dentry), *pinode;
+	struct super_block *sb = inode->i_sb;
+	struct nfs_server *server = NFS_SB(sb);
+	struct nfs_fattr *fattr = NULL;
+	struct nfs4_label *label = NULL;
+	struct dentry *parent;
+	struct nfs_rpc_ops const *ops = server->nfs_client->rpc_ops;
+	struct nfs_fh fh;
+
+	if (!ops->lookupp)
+		return ERR_PTR(-EACCES);
+
+	fattr = nfs_alloc_fattr();
+	if (fattr == NULL) {
+		parent = ERR_PTR(-ENOMEM);
+		goto out;
+	}
+
+	label = nfs4_label_alloc(server, GFP_KERNEL);
+	if (IS_ERR(label)) {
+		parent = ERR_CAST(label);
+		goto out_free_fattr;
+	}
+
+	ret = ops->lookupp(inode, &fh, fattr, label);
+	if (ret) {
+		parent = ERR_PTR(ret);
+		goto out_free_label;
+	}
+
+	pinode = nfs_fhget(sb, &fh, fattr, label);
+	parent = d_obtain_alias(pinode);
+out_free_label:
+	nfs4_label_free(label);
+out_free_fattr:
+	nfs_free_fattr(fattr);
+out:
+	return parent;
+}
+
+const struct export_operations nfs_export_ops = {
+	.encode_fh = nfs_encode_fh,
+	.fh_to_dentry = nfs_fh_to_dentry,
+	.get_parent = nfs_get_parent,
+};
diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
index c5054ed..2ebd574 100644
--- a/fs/nfs/internal.h
+++ b/fs/nfs/internal.h
@@ -10,6 +10,8 @@
 
 #define NFS_MS_MASK (MS_RDONLY|MS_NOSUID|MS_NODEV|MS_NOEXEC|MS_SYNCHRONOUS)
 
+extern const struct export_operations nfs_export_ops;
+
 struct nfs_string;
 
 /* Maximum number of readahead requests
diff --git a/fs/nfs/super.c b/fs/nfs/super.c
index b417639..b527164 100644
--- a/fs/nfs/super.c
+++ b/fs/nfs/super.c
@@ -2339,6 +2339,7 @@ void nfs_fill_super(struct super_block *sb, struct nfs_mount_info *mount_info)
 		 */
 		sb->s_flags |= MS_POSIXACL;
 		sb->s_time_gran = 1;
+		sb->s_export_op = &nfs_export_ops;
 	}
 
  	nfs_initialise_sb(sb);
@@ -2360,6 +2361,7 @@ static void nfs_clone_super(struct super_block *sb,
 	sb->s_xattr = old_sb->s_xattr;
 	sb->s_op = old_sb->s_op;
 	sb->s_time_gran = 1;
+	sb->s_export_op = old_sb->s_export_op;
 
 	if (server->nfs_client->rpc_ops->version != 2) {
 		/* The VFS shouldn't apply the umask to mode bits. We will do