[net-next-2.6.git] / fs / ceph / export.c

#include <linux/ceph/ceph_debug.h>

#include <linux/exportfs.h>
#include <linux/slab.h>
#include <asm/unaligned.h>

#include "super.h"
#include "mds_client.h"

/*
 * NFS export support
 *
 * NFS re-export of a ceph mount is, at present, only semireliable.
 * The basic issue is that the Ceph architectures doesn't lend itself
 * well to generating filehandles that will remain valid forever.
 *
 * So, we do our best.  If you're lucky, your inode will be in the
 * client's cache.  If it's not, and you have a connectable fh, then
 * the MDS server may be able to find it for you.  Otherwise, you get
 * ESTALE.
 *
 * There are ways to this more reliable, but in the non-connectable fh
 * case, we won't every work perfectly, and in the connectable case,
 * some changes are needed on the MDS side to work better.
 */

/*
 * Basic fh
 */
struct ceph_nfs_fh {
	u64 ino;
} __attribute__ ((packed));

/*
 * Larger 'connectable' fh that includes parent ino and name hash.
 * Use this whenever possible, as it works more reliably.
 */
struct ceph_nfs_confh {
	u64 ino, parent_ino;
	u32 parent_name_hash;
} __attribute__ ((packed));

static int ceph_encode_fh(struct dentry *dentry, u32 *rawfh, int *max_len,
			  int connectable)
{
	int type;
	struct ceph_nfs_fh *fh = (void *)rawfh;
	struct ceph_nfs_confh *cfh = (void *)rawfh;
	struct dentry *parent = dentry->d_parent;
	struct inode *inode = dentry->d_inode;
	int connected_handle_length = sizeof(*cfh)/4;
	int handle_length = sizeof(*fh)/4;

	/* don't re-export snaps */
	if (ceph_snap(inode) != CEPH_NOSNAP)
		return -EINVAL;

	if (*max_len >= connected_handle_length) {
		dout("encode_fh %p connectable\n", dentry);
		cfh->ino = ceph_ino(dentry->d_inode);
		cfh->parent_ino = ceph_ino(parent->d_inode);
		cfh->parent_name_hash = parent->d_name.hash;
		*max_len = connected_handle_length;
		type = 2;
	} else if (*max_len >= handle_length) {
		if (connectable) {
			*max_len = connected_handle_length;
			return 255;
		}
		dout("encode_fh %p\n", dentry);
		fh->ino = ceph_ino(dentry->d_inode);
		*max_len = handle_length;
		type = 1;
	} else {
		*max_len = handle_length;
		return 255;
	}
	return type;
}

/*
 * convert regular fh to dentry
 *
 * FIXME: we should try harder by querying the mds for the ino.
 */
static struct dentry *__fh_to_dentry(struct super_block *sb,
				     struct ceph_nfs_fh *fh)
{
	struct inode *inode;
	struct dentry *dentry;
	struct ceph_vino vino;
	int err;

	dout("__fh_to_dentry %llx\n", fh->ino);
	vino.ino = fh->ino;
	vino.snap = CEPH_NOSNAP;
	inode = ceph_find_inode(sb, vino);
	if (!inode)
		return ERR_PTR(-ESTALE);

	dentry = d_obtain_alias(inode);
	if (IS_ERR(dentry)) {
		pr_err("fh_to_dentry %llx -- inode %p but ENOMEM\n",
		       fh->ino, inode);
		iput(inode);
		return dentry;
	}
	err = ceph_init_dentry(dentry);

	if (err < 0) {
		iput(inode);
		return ERR_PTR(err);
	}
	dout("__fh_to_dentry %llx %p dentry %p\n", fh->ino, inode, dentry);
	return dentry;
}

/*
 * convert connectable fh to dentry
 */
static struct dentry *__cfh_to_dentry(struct super_block *sb,
				      struct ceph_nfs_confh *cfh)
{
	struct ceph_mds_client *mdsc = ceph_sb_to_client(sb)->mdsc;
	struct inode *inode;
	struct dentry *dentry;
	struct ceph_vino vino;
	int err;

	dout("__cfh_to_dentry %llx (%llx/%x)\n",
	     cfh->ino, cfh->parent_ino, cfh->parent_name_hash);

	vino.ino = cfh->ino;
	vino.snap = CEPH_NOSNAP;
	inode = ceph_find_inode(sb, vino);
	if (!inode) {
		struct ceph_mds_request *req;

		req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_LOOKUPHASH,
					       USE_ANY_MDS);
		if (IS_ERR(req))
			return ERR_CAST(req);

		req->r_ino1 = vino;
		req->r_ino2.ino = cfh->parent_ino;
		req->r_ino2.snap = CEPH_NOSNAP;
		req->r_path2 = kmalloc(16, GFP_NOFS);
		snprintf(req->r_path2, 16, "%d", cfh->parent_name_hash);
		req->r_num_caps = 1;
		err = ceph_mdsc_do_request(mdsc, NULL, req);
		ceph_mdsc_put_request(req);
		inode = ceph_find_inode(sb, vino);
		if (!inode)
			return ERR_PTR(err ? err : -ESTALE);
	}

	dentry = d_obtain_alias(inode);
	if (IS_ERR(dentry)) {
		pr_err("cfh_to_dentry %llx -- inode %p but ENOMEM\n",
		       cfh->ino, inode);
		iput(inode);
		return dentry;
	}
	err = ceph_init_dentry(dentry);
	if (err < 0) {
		iput(inode);
		return ERR_PTR(err);
	}
	dout("__cfh_to_dentry %llx %p dentry %p\n", cfh->ino, inode, dentry);
	return dentry;
}

static struct dentry *ceph_fh_to_dentry(struct super_block *sb, struct fid *fid,
					int fh_len, int fh_type)
{
	if (fh_type == 1)
		return __fh_to_dentry(sb, (struct ceph_nfs_fh *)fid->raw);
	else
		return __cfh_to_dentry(sb, (struct ceph_nfs_confh *)fid->raw);
}

/*
 * get parent, if possible.
 *
 * FIXME: we could do better by querying the mds to discover the
 * parent.
 */
static struct dentry *ceph_fh_to_parent(struct super_block *sb,
					 struct fid *fid,
					int fh_len, int fh_type)
{
	struct ceph_nfs_confh *cfh = (void *)fid->raw;
	struct ceph_vino vino;
	struct inode *inode;
	struct dentry *dentry;
	int err;

	if (fh_type == 1)
		return ERR_PTR(-ESTALE);

	pr_debug("fh_to_parent %llx/%d\n", cfh->parent_ino,
		 cfh->parent_name_hash);

	vino.ino = cfh->ino;
	vino.snap = CEPH_NOSNAP;
	inode = ceph_find_inode(sb, vino);
	if (!inode)
		return ERR_PTR(-ESTALE);

	dentry = d_obtain_alias(inode);
	if (IS_ERR(dentry)) {
		pr_err("fh_to_parent %llx -- inode %p but ENOMEM\n",
		       cfh->ino, inode);
		iput(inode);
		return dentry;
	}
	err = ceph_init_dentry(dentry);
	if (err < 0) {
		iput(inode);
		return ERR_PTR(err);
	}
	dout("fh_to_parent %llx %p dentry %p\n", cfh->ino, inode, dentry);
	return dentry;
}

const struct export_operations ceph_export_ops = {
	.encode_fh = ceph_encode_fh,
	.fh_to_dentry = ceph_fh_to_dentry,
	.fh_to_parent = ceph_fh_to_parent,
};
Commit	Line	Data
3d14c5d2	1	#include <linux/ceph/ceph_debug.h>
a8e63b7d SW	2
a8e63b7d SW	3	#include <linux/exportfs.h>
5a0e3ad6	4	#include <linux/slab.h>
a8e63b7d SW	5	#include <asm/unaligned.h>
	6
	7	#include "super.h"
3d14c5d2	8	#include "mds_client.h"
a8e63b7d SW	9
	10	/*
	11	* NFS export support
	12	*
	13	* NFS re-export of a ceph mount is, at present, only semireliable.
	14	* The basic issue is that the Ceph architectures doesn't lend itself
	15	* well to generating filehandles that will remain valid forever.
	16	*
	17	* So, we do our best. If you're lucky, your inode will be in the
	18	* client's cache. If it's not, and you have a connectable fh, then
	19	* the MDS server may be able to find it for you. Otherwise, you get
	20	* ESTALE.
	21	*
	22	* There are ways to this more reliable, but in the non-connectable fh
	23	* case, we won't every work perfectly, and in the connectable case,
	24	* some changes are needed on the MDS side to work better.
	25	*/
	26
	27	/*
	28	* Basic fh
	29	*/
	30	struct ceph_nfs_fh {
	31	u64 ino;
	32	} __attribute__ ((packed));
	33
	34	/*
	35	* Larger 'connectable' fh that includes parent ino and name hash.
	36	* Use this whenever possible, as it works more reliably.
	37	*/
	38	struct ceph_nfs_confh {
	39	u64 ino, parent_ino;
	40	u32 parent_name_hash;
	41	} __attribute__ ((packed));
	42
	43	static int ceph_encode_fh(struct dentry dentry, u32 rawfh, int *max_len,
	44	int connectable)
	45	{
92923dcb	46	int type;
a8e63b7d SW	47	struct ceph_nfs_fh fh = (void )rawfh;
	48	struct ceph_nfs_confh cfh = (void )rawfh;
	49	struct dentry *parent = dentry->d_parent;
	50	struct inode *inode = dentry->d_inode;
92923dcb AK	51	int connected_handle_length = sizeof(*cfh)/4;
92923dcb AK	52	int handle_length = sizeof(*fh)/4;
a8e63b7d SW	53
	54	/* don't re-export snaps */
	55	if (ceph_snap(inode) != CEPH_NOSNAP)
	56	return -EINVAL;
	57
92923dcb	58	if (*max_len >= connected_handle_length) {
a8e63b7d SW	59	dout("encode_fh %p connectable\n", dentry);
	60	cfh->ino = ceph_ino(dentry->d_inode);
	61	cfh->parent_ino = ceph_ino(parent->d_inode);
	62	cfh->parent_name_hash = parent->d_name.hash;
92923dcb	63	*max_len = connected_handle_length;
a8e63b7d	64	type = 2;
92923dcb	65	} else if (*max_len >= handle_length) {
bba0cd0e AK	66	if (connectable) {
bba0cd0e AK	67	*max_len = connected_handle_length;
92923dcb	68	return 255;
bba0cd0e	69	}
a8e63b7d SW	70	dout("encode_fh %p\n", dentry);
a8e63b7d SW	71	fh->ino = ceph_ino(dentry->d_inode);
92923dcb	72	*max_len = handle_length;
a8e63b7d SW	73	type = 1;
a8e63b7d SW	74	} else {
bba0cd0e	75	*max_len = handle_length;
92923dcb	76	return 255;
a8e63b7d SW	77	}
	78	return type;
	79	}
	80
	81	/*
	82	* convert regular fh to dentry
	83	*
	84	* FIXME: we should try harder by querying the mds for the ino.
	85	*/
	86	static struct dentry __fh_to_dentry(struct super_block sb,
	87	struct ceph_nfs_fh *fh)
	88	{
	89	struct inode *inode;
	90	struct dentry *dentry;
	91	struct ceph_vino vino;
	92	int err;
	93
	94	dout("__fh_to_dentry %llx\n", fh->ino);
	95	vino.ino = fh->ino;
	96	vino.snap = CEPH_NOSNAP;
	97	inode = ceph_find_inode(sb, vino);
	98	if (!inode)
	99	return ERR_PTR(-ESTALE);
	100
	101	dentry = d_obtain_alias(inode);
0d509c94	102	if (IS_ERR(dentry)) {
a8e63b7d SW	103	pr_err("fh_to_dentry %llx -- inode %p but ENOMEM\n",
	104	fh->ino, inode);
	105	iput(inode);
0d509c94	106	return dentry;
a8e63b7d SW	107	}
	108	err = ceph_init_dentry(dentry);
	109
	110	if (err < 0) {
	111	iput(inode);
	112	return ERR_PTR(err);
	113	}
	114	dout("__fh_to_dentry %llx %p dentry %p\n", fh->ino, inode, dentry);
	115	return dentry;
	116	}
	117
	118	/*
	119	* convert connectable fh to dentry
	120	*/
	121	static struct dentry __cfh_to_dentry(struct super_block sb,
	122	struct ceph_nfs_confh *cfh)
	123	{
3d14c5d2	124	struct ceph_mds_client *mdsc = ceph_sb_to_client(sb)->mdsc;
a8e63b7d SW	125	struct inode *inode;
	126	struct dentry *dentry;
	127	struct ceph_vino vino;
	128	int err;
	129
	130	dout("__cfh_to_dentry %llx (%llx/%x)\n",
	131	cfh->ino, cfh->parent_ino, cfh->parent_name_hash);
	132
	133	vino.ino = cfh->ino;
	134	vino.snap = CEPH_NOSNAP;
	135	inode = ceph_find_inode(sb, vino);
	136	if (!inode) {
	137	struct ceph_mds_request *req;
	138
	139	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_LOOKUPHASH,
	140	USE_ANY_MDS);
	141	if (IS_ERR(req))
7e34bc52	142	return ERR_CAST(req);
a8e63b7d SW	143
	144	req->r_ino1 = vino;
	145	req->r_ino2.ino = cfh->parent_ino;
	146	req->r_ino2.snap = CEPH_NOSNAP;
	147	req->r_path2 = kmalloc(16, GFP_NOFS);
	148	snprintf(req->r_path2, 16, "%d", cfh->parent_name_hash);
	149	req->r_num_caps = 1;
	150	err = ceph_mdsc_do_request(mdsc, NULL, req);
	151	ceph_mdsc_put_request(req);
	152	inode = ceph_find_inode(sb, vino);
	153	if (!inode)
	154	return ERR_PTR(err ? err : -ESTALE);
	155	}
	156
	157	dentry = d_obtain_alias(inode);
0d509c94	158	if (IS_ERR(dentry)) {
a8e63b7d SW	159	pr_err("cfh_to_dentry %llx -- inode %p but ENOMEM\n",
	160	cfh->ino, inode);
	161	iput(inode);
0d509c94	162	return dentry;
a8e63b7d SW	163	}
	164	err = ceph_init_dentry(dentry);
	165	if (err < 0) {
	166	iput(inode);
	167	return ERR_PTR(err);
	168	}
	169	dout("__cfh_to_dentry %llx %p dentry %p\n", cfh->ino, inode, dentry);
	170	return dentry;
	171	}
	172
	173	static struct dentry ceph_fh_to_dentry(struct super_block sb, struct fid *fid,
	174	int fh_len, int fh_type)
	175	{
	176	if (fh_type == 1)
	177	return __fh_to_dentry(sb, (struct ceph_nfs_fh *)fid->raw);
	178	else
	179	return __cfh_to_dentry(sb, (struct ceph_nfs_confh *)fid->raw);
	180	}
	181
	182	/*
	183	* get parent, if possible.
	184	*
	185	* FIXME: we could do better by querying the mds to discover the
	186	* parent.
	187	*/
	188	static struct dentry ceph_fh_to_parent(struct super_block sb,
	189	struct fid *fid,
	190	int fh_len, int fh_type)
	191	{
	192	struct ceph_nfs_confh cfh = (void )fid->raw;
	193	struct ceph_vino vino;
	194	struct inode *inode;
	195	struct dentry *dentry;
	196	int err;
	197
	198	if (fh_type == 1)
	199	return ERR_PTR(-ESTALE);
	200
	201	pr_debug("fh_to_parent %llx/%d\n", cfh->parent_ino,
	202	cfh->parent_name_hash);
	203
	204	vino.ino = cfh->ino;
	205	vino.snap = CEPH_NOSNAP;
	206	inode = ceph_find_inode(sb, vino);
	207	if (!inode)
	208	return ERR_PTR(-ESTALE);
	209
	210	dentry = d_obtain_alias(inode);
0d509c94	211	if (IS_ERR(dentry)) {
a8e63b7d SW	212	pr_err("fh_to_parent %llx -- inode %p but ENOMEM\n",
	213	cfh->ino, inode);
	214	iput(inode);
0d509c94	215	return dentry;
a8e63b7d SW	216	}
	217	err = ceph_init_dentry(dentry);
	218	if (err < 0) {
	219	iput(inode);
	220	return ERR_PTR(err);
	221	}
	222	dout("fh_to_parent %llx %p dentry %p\n", cfh->ino, inode, dentry);
	223	return dentry;
	224	}
	225
	226	const struct export_operations ceph_export_ops = {
	227	.encode_fh = ceph_encode_fh,
	228	.fh_to_dentry = ceph_fh_to_dentry,
	229	.fh_to_parent = ceph_fh_to_parent,
	230	};