[deliverable/linux.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;
	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;

	spin_lock(&dentry->d_lock);
	parent = dget(dentry->d_parent);
	spin_unlock(&dentry->d_lock);

	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 = ceph_dentry_hash(parent->d_inode,
							 dentry);
		*max_len = connected_handle_length;
		type = 2;
	} else if (*max_len >= handle_length) {
		if (connectable) {
			*max_len = connected_handle_length;
			type = 255;
		} else {
			dout("encode_fh %p\n", dentry);
			fh->ino = ceph_ino(dentry->d_inode);
			*max_len = handle_length;
			type = 1;
		}
	} else {
		*max_len = handle_length;
		type = 255;
	}
	dput(parent);
	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 ceph_mds_client *mdsc = ceph_sb_to_client(sb)->mdsc;
	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) {
		struct ceph_mds_request *req;

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

		req->r_ino1 = vino;
		req->r_num_caps = 1;
		err = ceph_mdsc_do_request(mdsc, NULL, req);
		inode = req->r_target_inode;
		if (inode)
			ihold(inode);
		ceph_mdsc_put_request(req);
		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);
		inode = req->r_target_inode;
		if (inode)
			ihold(inode);
		ceph_mdsc_put_request(req);
		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;
a8e63b7d SW	48	struct ceph_nfs_confh cfh = (void )rawfh;
e5f86dc3	49	struct dentry *parent;
a8e63b7d	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
e5f86dc3 SW	58	spin_lock(&dentry->d_lock);
	59	parent = dget(dentry->d_parent);
	60	spin_unlock(&dentry->d_lock);
	61
92923dcb	62	if (*max_len >= connected_handle_length) {
a8e63b7d SW	63	dout("encode_fh %p connectable\n", dentry);
	64	cfh->ino = ceph_ino(dentry->d_inode);
	65	cfh->parent_ino = ceph_ino(parent->d_inode);
e5f86dc3 SW	66	cfh->parent_name_hash = ceph_dentry_hash(parent->d_inode,
e5f86dc3 SW	67	dentry);
92923dcb	68	*max_len = connected_handle_length;
a8e63b7d	69	type = 2;
92923dcb	70	} else if (*max_len >= handle_length) {
bba0cd0e AK	71	if (connectable) {
bba0cd0e AK	72	*max_len = connected_handle_length;
e5f86dc3 SW	73	type = 255;
	74	} else {
	75	dout("encode_fh %p\n", dentry);
	76	fh->ino = ceph_ino(dentry->d_inode);
	77	*max_len = handle_length;
	78	type = 1;
bba0cd0e	79	}
a8e63b7d	80	} else {
bba0cd0e	81	*max_len = handle_length;
e5f86dc3	82	type = 255;
a8e63b7d	83	}
e5f86dc3	84	dput(parent);
a8e63b7d SW	85	return type;
	86	}
	87
	88	/*
	89	* convert regular fh to dentry
	90	*
	91	* FIXME: we should try harder by querying the mds for the ino.
	92	*/
	93	static struct dentry __fh_to_dentry(struct super_block sb,
	94	struct ceph_nfs_fh *fh)
	95	{
3c454cf2	96	struct ceph_mds_client *mdsc = ceph_sb_to_client(sb)->mdsc;
a8e63b7d SW	97	struct inode *inode;
	98	struct dentry *dentry;
	99	struct ceph_vino vino;
	100	int err;
	101
	102	dout("__fh_to_dentry %llx\n", fh->ino);
	103	vino.ino = fh->ino;
	104	vino.snap = CEPH_NOSNAP;
	105	inode = ceph_find_inode(sb, vino);
3c454cf2 SW	106	if (!inode) {
	107	struct ceph_mds_request *req;
	108
	109	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_LOOKUPINO,
	110	USE_ANY_MDS);
	111	if (IS_ERR(req))
	112	return ERR_CAST(req);
	113
	114	req->r_ino1 = vino;
	115	req->r_num_caps = 1;
	116	err = ceph_mdsc_do_request(mdsc, NULL, req);
45e3d3ee SW	117	inode = req->r_target_inode;
45e3d3ee SW	118	if (inode)
70b666c3	119	ihold(inode);
3c454cf2	120	ceph_mdsc_put_request(req);
3c454cf2 SW	121	if (!inode)
	122	return ERR_PTR(-ESTALE);
	123	}
a8e63b7d SW	124
a8e63b7d SW	125	dentry = d_obtain_alias(inode);
0d509c94	126	if (IS_ERR(dentry)) {
a8e63b7d SW	127	pr_err("fh_to_dentry %llx -- inode %p but ENOMEM\n",
	128	fh->ino, inode);
	129	iput(inode);
0d509c94	130	return dentry;
a8e63b7d SW	131	}
a8e63b7d SW	132	err = ceph_init_dentry(dentry);
a8e63b7d SW	133	if (err < 0) {
	134	iput(inode);
	135	return ERR_PTR(err);
	136	}
	137	dout("__fh_to_dentry %llx %p dentry %p\n", fh->ino, inode, dentry);
	138	return dentry;
	139	}
	140
	141	/*
	142	* convert connectable fh to dentry
	143	*/
	144	static struct dentry __cfh_to_dentry(struct super_block sb,
	145	struct ceph_nfs_confh *cfh)
	146	{
3d14c5d2	147	struct ceph_mds_client *mdsc = ceph_sb_to_client(sb)->mdsc;
a8e63b7d SW	148	struct inode *inode;
	149	struct dentry *dentry;
	150	struct ceph_vino vino;
	151	int err;
	152
	153	dout("__cfh_to_dentry %llx (%llx/%x)\n",
	154	cfh->ino, cfh->parent_ino, cfh->parent_name_hash);
	155
	156	vino.ino = cfh->ino;
	157	vino.snap = CEPH_NOSNAP;
	158	inode = ceph_find_inode(sb, vino);
	159	if (!inode) {
	160	struct ceph_mds_request *req;
	161
	162	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_LOOKUPHASH,
	163	USE_ANY_MDS);
	164	if (IS_ERR(req))
7e34bc52	165	return ERR_CAST(req);
a8e63b7d SW	166
	167	req->r_ino1 = vino;
	168	req->r_ino2.ino = cfh->parent_ino;
	169	req->r_ino2.snap = CEPH_NOSNAP;
	170	req->r_path2 = kmalloc(16, GFP_NOFS);
	171	snprintf(req->r_path2, 16, "%d", cfh->parent_name_hash);
	172	req->r_num_caps = 1;
	173	err = ceph_mdsc_do_request(mdsc, NULL, req);
45e3d3ee SW	174	inode = req->r_target_inode;
45e3d3ee SW	175	if (inode)
70b666c3	176	ihold(inode);
a8e63b7d	177	ceph_mdsc_put_request(req);
a8e63b7d SW	178	if (!inode)
	179	return ERR_PTR(err ? err : -ESTALE);
	180	}
	181
	182	dentry = d_obtain_alias(inode);
0d509c94	183	if (IS_ERR(dentry)) {
a8e63b7d SW	184	pr_err("cfh_to_dentry %llx -- inode %p but ENOMEM\n",
	185	cfh->ino, inode);
	186	iput(inode);
0d509c94	187	return dentry;
a8e63b7d SW	188	}
	189	err = ceph_init_dentry(dentry);
	190	if (err < 0) {
	191	iput(inode);
	192	return ERR_PTR(err);
	193	}
	194	dout("__cfh_to_dentry %llx %p dentry %p\n", cfh->ino, inode, dentry);
	195	return dentry;
	196	}
	197
	198	static struct dentry ceph_fh_to_dentry(struct super_block sb, struct fid *fid,
	199	int fh_len, int fh_type)
	200	{
	201	if (fh_type == 1)
	202	return __fh_to_dentry(sb, (struct ceph_nfs_fh *)fid->raw);
	203	else
	204	return __cfh_to_dentry(sb, (struct ceph_nfs_confh *)fid->raw);
	205	}
	206
	207	/*
	208	* get parent, if possible.
	209	*
	210	* FIXME: we could do better by querying the mds to discover the
	211	* parent.
	212	*/
	213	static struct dentry ceph_fh_to_parent(struct super_block sb,
	214	struct fid *fid,
	215	int fh_len, int fh_type)
	216	{
	217	struct ceph_nfs_confh cfh = (void )fid->raw;
	218	struct ceph_vino vino;
	219	struct inode *inode;
	220	struct dentry *dentry;
	221	int err;
	222
	223	if (fh_type == 1)
	224	return ERR_PTR(-ESTALE);
	225
	226	pr_debug("fh_to_parent %llx/%d\n", cfh->parent_ino,
	227	cfh->parent_name_hash);
	228
	229	vino.ino = cfh->ino;
	230	vino.snap = CEPH_NOSNAP;
	231	inode = ceph_find_inode(sb, vino);
	232	if (!inode)
	233	return ERR_PTR(-ESTALE);
	234
	235	dentry = d_obtain_alias(inode);
0d509c94	236	if (IS_ERR(dentry)) {
a8e63b7d SW	237	pr_err("fh_to_parent %llx -- inode %p but ENOMEM\n",
	238	cfh->ino, inode);
	239	iput(inode);
0d509c94	240	return dentry;
a8e63b7d SW	241	}
	242	err = ceph_init_dentry(dentry);
	243	if (err < 0) {
	244	iput(inode);
	245	return ERR_PTR(err);
	246	}
	247	dout("fh_to_parent %llx %p dentry %p\n", cfh->ino, inode, dentry);
	248	return dentry;
	249	}
	250
	251	const struct export_operations ceph_export_ops = {
	252	.encode_fh = ceph_encode_fh,
	253	.fh_to_dentry = ceph_fh_to_dentry,
	254	.fh_to_parent = ceph_fh_to_parent,
	255	};