[linux-2.6-block.git] / fs / ceph / ioctl.c

#include <linux/ceph/ceph_debug.h>
#include <linux/in.h>

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


/*
 * ioctls
 */

/*
 * get and set the file layout
 */
static long ceph_ioctl_get_layout(struct file *file, void __user *arg)
{
	struct ceph_inode_info *ci = ceph_inode(file_inode(file));
	struct ceph_ioctl_layout l;
	int err;

	err = ceph_do_getattr(file_inode(file), CEPH_STAT_CAP_LAYOUT, false);
	if (!err) {
		l.stripe_unit = ceph_file_layout_su(ci->i_layout);
		l.stripe_count = ceph_file_layout_stripe_count(ci->i_layout);
		l.object_size = ceph_file_layout_object_size(ci->i_layout);
		l.data_pool = le32_to_cpu(ci->i_layout.fl_pg_pool);
		l.preferred_osd = (s32)-1;
		if (copy_to_user(arg, &l, sizeof(l)))
			return -EFAULT;
	}

	return err;
}

static long __validate_layout(struct ceph_mds_client *mdsc,
			      struct ceph_ioctl_layout *l)
{
	int i, err;

	/* validate striping parameters */
	if ((l->object_size & ~PAGE_MASK) ||
	    (l->stripe_unit & ~PAGE_MASK) ||
	    ((unsigned)l->stripe_unit != 0 &&
	     ((unsigned)l->object_size % (unsigned)l->stripe_unit)))
		return -EINVAL;

	/* make sure it's a valid data pool */
	mutex_lock(&mdsc->mutex);
	err = -EINVAL;
	for (i = 0; i < mdsc->mdsmap->m_num_data_pg_pools; i++)
		if (mdsc->mdsmap->m_data_pg_pools[i] == l->data_pool) {
			err = 0;
			break;
		}
	mutex_unlock(&mdsc->mutex);
	if (err)
		return err;

	return 0;
}

static long ceph_ioctl_set_layout(struct file *file, void __user *arg)
{
	struct inode *inode = file_inode(file);
	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
	struct ceph_mds_request *req;
	struct ceph_ioctl_layout l;
	struct ceph_inode_info *ci = ceph_inode(file_inode(file));
	struct ceph_ioctl_layout nl;
	int err;

	if (copy_from_user(&l, arg, sizeof(l)))
		return -EFAULT;

	/* validate changed params against current layout */
	err = ceph_do_getattr(file_inode(file), CEPH_STAT_CAP_LAYOUT, false);
	if (err)
		return err;

	memset(&nl, 0, sizeof(nl));
	if (l.stripe_count)
		nl.stripe_count = l.stripe_count;
	else
		nl.stripe_count = ceph_file_layout_stripe_count(ci->i_layout);
	if (l.stripe_unit)
		nl.stripe_unit = l.stripe_unit;
	else
		nl.stripe_unit = ceph_file_layout_su(ci->i_layout);
	if (l.object_size)
		nl.object_size = l.object_size;
	else
		nl.object_size = ceph_file_layout_object_size(ci->i_layout);
	if (l.data_pool)
		nl.data_pool = l.data_pool;
	else
		nl.data_pool = ceph_file_layout_pg_pool(ci->i_layout);

	/* this is obsolete, and always -1 */
	nl.preferred_osd = le64_to_cpu(-1);

	err = __validate_layout(mdsc, &nl);
	if (err)
		return err;

	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETLAYOUT,
				       USE_AUTH_MDS);
	if (IS_ERR(req))
		return PTR_ERR(req);
	req->r_inode = inode;
	ihold(inode);
	req->r_num_caps = 1;

	req->r_inode_drop = CEPH_CAP_FILE_SHARED | CEPH_CAP_FILE_EXCL;

	req->r_args.setlayout.layout.fl_stripe_unit =
		cpu_to_le32(l.stripe_unit);
	req->r_args.setlayout.layout.fl_stripe_count =
		cpu_to_le32(l.stripe_count);
	req->r_args.setlayout.layout.fl_object_size =
		cpu_to_le32(l.object_size);
	req->r_args.setlayout.layout.fl_pg_pool = cpu_to_le32(l.data_pool);

	err = ceph_mdsc_do_request(mdsc, NULL, req);
	ceph_mdsc_put_request(req);
	return err;
}

/*
 * Set a layout policy on a directory inode. All items in the tree
 * rooted at this inode will inherit this layout on creation,
 * (It doesn't apply retroactively )
 * unless a subdirectory has its own layout policy.
 */
static long ceph_ioctl_set_layout_policy (struct file *file, void __user *arg)
{
	struct inode *inode = file_inode(file);
	struct ceph_mds_request *req;
	struct ceph_ioctl_layout l;
	int err;
	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;

	/* copy and validate */
	if (copy_from_user(&l, arg, sizeof(l)))
		return -EFAULT;

	err = __validate_layout(mdsc, &l);
	if (err)
		return err;

	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETDIRLAYOUT,
				       USE_AUTH_MDS);

	if (IS_ERR(req))
		return PTR_ERR(req);
	req->r_inode = inode;
	ihold(inode);
	req->r_num_caps = 1;

	req->r_args.setlayout.layout.fl_stripe_unit =
			cpu_to_le32(l.stripe_unit);
	req->r_args.setlayout.layout.fl_stripe_count =
			cpu_to_le32(l.stripe_count);
	req->r_args.setlayout.layout.fl_object_size =
			cpu_to_le32(l.object_size);
	req->r_args.setlayout.layout.fl_pg_pool =
			cpu_to_le32(l.data_pool);

	err = ceph_mdsc_do_request(mdsc, inode, req);
	ceph_mdsc_put_request(req);
	return err;
}

/*
 * Return object name, size/offset information, and location (OSD
 * number, network address) for a given file offset.
 */
static long ceph_ioctl_get_dataloc(struct file *file, void __user *arg)
{
	struct ceph_ioctl_dataloc dl;
	struct inode *inode = file_inode(file);
	struct ceph_inode_info *ci = ceph_inode(inode);
	struct ceph_osd_client *osdc =
		&ceph_sb_to_client(inode->i_sb)->client->osdc;
	struct ceph_object_locator oloc;
	struct ceph_object_id oid;
	u64 len = 1, olen;
	u64 tmp;
	struct ceph_pg pgid;
	int r;

	/* copy and validate */
	if (copy_from_user(&dl, arg, sizeof(dl)))
		return -EFAULT;

	down_read(&osdc->map_sem);
	r = ceph_calc_file_object_mapping(&ci->i_layout, dl.file_offset, len,
					  &dl.object_no, &dl.object_offset,
					  &olen);
	if (r < 0) {
		up_read(&osdc->map_sem);
		return -EIO;
	}
	dl.file_offset -= dl.object_offset;
	dl.object_size = ceph_file_layout_object_size(ci->i_layout);
	dl.block_size = ceph_file_layout_su(ci->i_layout);

	/* block_offset = object_offset % block_size */
	tmp = dl.object_offset;
	dl.block_offset = do_div(tmp, dl.block_size);

	snprintf(dl.object_name, sizeof(dl.object_name), "%llx.%08llx",
		 ceph_ino(inode), dl.object_no);

	oloc.pool = ceph_file_layout_pg_pool(ci->i_layout);
	ceph_oid_set_name(&oid, dl.object_name);

	r = ceph_oloc_oid_to_pg(osdc->osdmap, &oloc, &oid, &pgid);
	if (r < 0) {
		up_read(&osdc->map_sem);
		return r;
	}

	dl.osd = ceph_calc_pg_primary(osdc->osdmap, pgid);
	if (dl.osd >= 0) {
		struct ceph_entity_addr *a =
			ceph_osd_addr(osdc->osdmap, dl.osd);
		if (a)
			memcpy(&dl.osd_addr, &a->in_addr, sizeof(dl.osd_addr));
	} else {
		memset(&dl.osd_addr, 0, sizeof(dl.osd_addr));
	}
	up_read(&osdc->map_sem);

	/* send result back to user */
	if (copy_to_user(arg, &dl, sizeof(dl)))
		return -EFAULT;

	return 0;
}

static long ceph_ioctl_lazyio(struct file *file)
{
	struct ceph_file_info *fi = file->private_data;
	struct inode *inode = file_inode(file);
	struct ceph_inode_info *ci = ceph_inode(inode);

	if ((fi->fmode & CEPH_FILE_MODE_LAZY) == 0) {
		spin_lock(&ci->i_ceph_lock);
		ci->i_nr_by_mode[fi->fmode]--;
		fi->fmode |= CEPH_FILE_MODE_LAZY;
		ci->i_nr_by_mode[fi->fmode]++;
		spin_unlock(&ci->i_ceph_lock);
		dout("ioctl_layzio: file %p marked lazy\n", file);

		ceph_check_caps(ci, 0, NULL);
	} else {
		dout("ioctl_layzio: file %p already lazy\n", file);
	}
	return 0;
}

static long ceph_ioctl_syncio(struct file *file)
{
	struct ceph_file_info *fi = file->private_data;

	fi->flags |= CEPH_F_SYNC;
	return 0;
}

long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
	dout("ioctl file %p cmd %u arg %lu\n", file, cmd, arg);
	switch (cmd) {
	case CEPH_IOC_GET_LAYOUT:
		return ceph_ioctl_get_layout(file, (void __user *)arg);

	case CEPH_IOC_SET_LAYOUT:
		return ceph_ioctl_set_layout(file, (void __user *)arg);

	case CEPH_IOC_SET_LAYOUT_POLICY:
		return ceph_ioctl_set_layout_policy(file, (void __user *)arg);

	case CEPH_IOC_GET_DATALOC:
		return ceph_ioctl_get_dataloc(file, (void __user *)arg);

	case CEPH_IOC_LAZYIO:
		return ceph_ioctl_lazyio(file);

	case CEPH_IOC_SYNCIO:
		return ceph_ioctl_syncio(file);
	}

	return -ENOTTY;
}
Commit	Line	Data
96c57ade	1	#include <linux/ceph/ceph_debug.h>
8f4e91de SW	2	#include <linux/in.h>
8f4e91de SW	3
8f4e91de	4	#include "super.h"
3d14c5d2	5	#include "mds_client.h"
3d14c5d2	6	#include "ioctl.h"
8f4e91de SW	7
	8
	9	/*
	10	* ioctls
	11	*/
	12
	13	/*
	14	* get and set the file layout
	15	*/
	16	static long ceph_ioctl_get_layout(struct file file, void __user arg)
	17	{
496ad9aa	18	struct ceph_inode_info *ci = ceph_inode(file_inode(file));
8f4e91de SW	19	struct ceph_ioctl_layout l;
	20	int err;
	21
508b32d8	22	err = ceph_do_getattr(file_inode(file), CEPH_STAT_CAP_LAYOUT, false);
8f4e91de SW	23	if (!err) {
	24	l.stripe_unit = ceph_file_layout_su(ci->i_layout);
	25	l.stripe_count = ceph_file_layout_stripe_count(ci->i_layout);
	26	l.object_size = ceph_file_layout_object_size(ci->i_layout);
	27	l.data_pool = le32_to_cpu(ci->i_layout.fl_pg_pool);
3469ac1a	28	l.preferred_osd = (s32)-1;
8f4e91de SW	29	if (copy_to_user(arg, &l, sizeof(l)))
	30	return -EFAULT;
	31	}
	32
	33	return err;
	34	}
	35
e49bf4c5 SW	36	static long __validate_layout(struct ceph_mds_client *mdsc,
	37	struct ceph_ioctl_layout *l)
	38	{
	39	int i, err;
	40
e49bf4c5 SW	41	/* validate striping parameters */
	42	if ((l->object_size & ~PAGE_MASK) \|\|
	43	(l->stripe_unit & ~PAGE_MASK) \|\|
0bc62284	44	((unsigned)l->stripe_unit != 0 &&
45f2e081	45	((unsigned)l->object_size % (unsigned)l->stripe_unit)))
e49bf4c5 SW	46	return -EINVAL;
	47
	48	/* make sure it's a valid data pool */
	49	mutex_lock(&mdsc->mutex);
	50	err = -EINVAL;
	51	for (i = 0; i < mdsc->mdsmap->m_num_data_pg_pools; i++)
	52	if (mdsc->mdsmap->m_data_pg_pools[i] == l->data_pool) {
	53	err = 0;
	54	break;
	55	}
	56	mutex_unlock(&mdsc->mutex);
	57	if (err)
	58	return err;
	59
	60	return 0;
	61	}
	62
8f4e91de SW	63	static long ceph_ioctl_set_layout(struct file file, void __user arg)
8f4e91de SW	64	{
496ad9aa	65	struct inode *inode = file_inode(file);
3d14c5d2	66	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
8f4e91de SW	67	struct ceph_mds_request *req;
8f4e91de SW	68	struct ceph_ioctl_layout l;
496ad9aa	69	struct ceph_inode_info *ci = ceph_inode(file_inode(file));
a35eca95	70	struct ceph_ioctl_layout nl;
e49bf4c5	71	int err;
8f4e91de	72
8f4e91de SW	73	if (copy_from_user(&l, arg, sizeof(l)))
	74	return -EFAULT;
	75
a35eca95	76	/* validate changed params against current layout */
508b32d8	77	err = ceph_do_getattr(file_inode(file), CEPH_STAT_CAP_LAYOUT, false);
702aeb1f	78	if (err)
a35eca95 GF	79	return err;
a35eca95 GF	80
702aeb1f	81	memset(&nl, 0, sizeof(nl));
a35eca95 GF	82	if (l.stripe_count)
a35eca95 GF	83	nl.stripe_count = l.stripe_count;
702aeb1f SW	84	else
702aeb1f SW	85	nl.stripe_count = ceph_file_layout_stripe_count(ci->i_layout);
a35eca95 GF	86	if (l.stripe_unit)
a35eca95 GF	87	nl.stripe_unit = l.stripe_unit;
702aeb1f SW	88	else
702aeb1f SW	89	nl.stripe_unit = ceph_file_layout_su(ci->i_layout);
a35eca95 GF	90	if (l.object_size)
a35eca95 GF	91	nl.object_size = l.object_size;
702aeb1f SW	92	else
702aeb1f SW	93	nl.object_size = ceph_file_layout_object_size(ci->i_layout);
a35eca95 GF	94	if (l.data_pool)
a35eca95 GF	95	nl.data_pool = l.data_pool;
702aeb1f SW	96	else
	97	nl.data_pool = ceph_file_layout_pg_pool(ci->i_layout);
	98
	99	/* this is obsolete, and always -1 */
	100	nl.preferred_osd = le64_to_cpu(-1);
a35eca95	101
e49bf4c5 SW	102	err = __validate_layout(mdsc, &nl);
	103	if (err)
	104	return err;
8f4e91de SW	105
	106	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETLAYOUT,
	107	USE_AUTH_MDS);
	108	if (IS_ERR(req))
	109	return PTR_ERR(req);
70b666c3 SW	110	req->r_inode = inode;
70b666c3 SW	111	ihold(inode);
3bd58143 YZ	112	req->r_num_caps = 1;
3bd58143 YZ	113
8f4e91de SW	114	req->r_inode_drop = CEPH_CAP_FILE_SHARED \| CEPH_CAP_FILE_EXCL;
	115
	116	req->r_args.setlayout.layout.fl_stripe_unit =
	117	cpu_to_le32(l.stripe_unit);
	118	req->r_args.setlayout.layout.fl_stripe_count =
	119	cpu_to_le32(l.stripe_count);
	120	req->r_args.setlayout.layout.fl_object_size =
	121	cpu_to_le32(l.object_size);
	122	req->r_args.setlayout.layout.fl_pg_pool = cpu_to_le32(l.data_pool);
8f4e91de	123
752c8bdc	124	err = ceph_mdsc_do_request(mdsc, NULL, req);
8f4e91de SW	125	ceph_mdsc_put_request(req);
	126	return err;
	127	}
	128
571dba52 GF	129	/*
	130	* Set a layout policy on a directory inode. All items in the tree
	131	* rooted at this inode will inherit this layout on creation,
	132	* (It doesn't apply retroactively )
	133	* unless a subdirectory has its own layout policy.
	134	*/
	135	static long ceph_ioctl_set_layout_policy (struct file file, void __user arg)
	136	{
496ad9aa	137	struct inode *inode = file_inode(file);
571dba52 GF	138	struct ceph_mds_request *req;
571dba52 GF	139	struct ceph_ioctl_layout l;
e49bf4c5	140	int err;
571dba52 GF	141	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
	142
	143	/* copy and validate */
	144	if (copy_from_user(&l, arg, sizeof(l)))
	145	return -EFAULT;
	146
e49bf4c5 SW	147	err = __validate_layout(mdsc, &l);
	148	if (err)
	149	return err;
571dba52 GF	150
	151	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETDIRLAYOUT,
	152	USE_AUTH_MDS);
	153
	154	if (IS_ERR(req))
	155	return PTR_ERR(req);
70b666c3 SW	156	req->r_inode = inode;
70b666c3 SW	157	ihold(inode);
3bd58143	158	req->r_num_caps = 1;
571dba52 GF	159
	160	req->r_args.setlayout.layout.fl_stripe_unit =
	161	cpu_to_le32(l.stripe_unit);
	162	req->r_args.setlayout.layout.fl_stripe_count =
	163	cpu_to_le32(l.stripe_count);
	164	req->r_args.setlayout.layout.fl_object_size =
	165	cpu_to_le32(l.object_size);
	166	req->r_args.setlayout.layout.fl_pg_pool =
	167	cpu_to_le32(l.data_pool);
571dba52 GF	168
	169	err = ceph_mdsc_do_request(mdsc, inode, req);
	170	ceph_mdsc_put_request(req);
	171	return err;
	172	}
	173
8f4e91de SW	174	/*
	175	* Return object name, size/offset information, and location (OSD
	176	* number, network address) for a given file offset.
	177	*/
	178	static long ceph_ioctl_get_dataloc(struct file file, void __user arg)
	179	{
	180	struct ceph_ioctl_dataloc dl;
496ad9aa	181	struct inode *inode = file_inode(file);
8f4e91de	182	struct ceph_inode_info *ci = ceph_inode(inode);
3d14c5d2 YS	183	struct ceph_osd_client *osdc =
3d14c5d2 YS	184	&ceph_sb_to_client(inode->i_sb)->client->osdc;
7c13cb64 ID	185	struct ceph_object_locator oloc;
7c13cb64 ID	186	struct ceph_object_id oid;
8f4e91de SW	187	u64 len = 1, olen;
8f4e91de SW	188	u64 tmp;
51042122	189	struct ceph_pg pgid;
457712a0	190	int r;
8f4e91de SW	191
	192	/* copy and validate */
	193	if (copy_from_user(&dl, arg, sizeof(dl)))
	194	return -EFAULT;
	195
	196	down_read(&osdc->map_sem);
e8afad65	197	r = ceph_calc_file_object_mapping(&ci->i_layout, dl.file_offset, len,
457712a0 SW	198	&dl.object_no, &dl.object_offset,
457712a0 SW	199	&olen);
494ddd11	200	if (r < 0) {
494ddd11	201	up_read(&osdc->map_sem);
457712a0	202	return -EIO;
494ddd11	203	}
8f4e91de SW	204	dl.file_offset -= dl.object_offset;
	205	dl.object_size = ceph_file_layout_object_size(ci->i_layout);
	206	dl.block_size = ceph_file_layout_su(ci->i_layout);
	207
	208	/* block_offset = object_offset % block_size */
	209	tmp = dl.object_offset;
	210	dl.block_offset = do_div(tmp, dl.block_size);
	211
	212	snprintf(dl.object_name, sizeof(dl.object_name), "%llx.%08llx",
	213	ceph_ino(inode), dl.object_no);
41766f87	214
7c13cb64 ID	215	oloc.pool = ceph_file_layout_pg_pool(ci->i_layout);
	216	ceph_oid_set_name(&oid, dl.object_name);
	217
	218	r = ceph_oloc_oid_to_pg(osdc->osdmap, &oloc, &oid, &pgid);
2fbcbff1	219	if (r < 0) {
	220	up_read(&osdc->map_sem);
	221	return r;
	222	}
8f4e91de	223
8f4e91de SW	224	dl.osd = ceph_calc_pg_primary(osdc->osdmap, pgid);
	225	if (dl.osd >= 0) {
	226	struct ceph_entity_addr *a =
	227	ceph_osd_addr(osdc->osdmap, dl.osd);
	228	if (a)
	229	memcpy(&dl.osd_addr, &a->in_addr, sizeof(dl.osd_addr));
	230	} else {
	231	memset(&dl.osd_addr, 0, sizeof(dl.osd_addr));
	232	}
	233	up_read(&osdc->map_sem);
	234
	235	/* send result back to user */
	236	if (copy_to_user(arg, &dl, sizeof(dl)))
	237	return -EFAULT;
	238
	239	return 0;
	240	}
	241
8c6e9229 SW	242	static long ceph_ioctl_lazyio(struct file *file)
	243	{
	244	struct ceph_file_info *fi = file->private_data;
496ad9aa	245	struct inode *inode = file_inode(file);
8c6e9229 SW	246	struct ceph_inode_info *ci = ceph_inode(inode);
	247
	248	if ((fi->fmode & CEPH_FILE_MODE_LAZY) == 0) {
be655596	249	spin_lock(&ci->i_ceph_lock);
8c6e9229 SW	250	ci->i_nr_by_mode[fi->fmode]--;
	251	fi->fmode \|= CEPH_FILE_MODE_LAZY;
	252	ci->i_nr_by_mode[fi->fmode]++;
be655596	253	spin_unlock(&ci->i_ceph_lock);
8c6e9229 SW	254	dout("ioctl_layzio: file %p marked lazy\n", file);
	255
	256	ceph_check_caps(ci, 0, NULL);
	257	} else {
	258	dout("ioctl_layzio: file %p already lazy\n", file);
	259	}
	260	return 0;
	261	}
	262
4918b6d1 SW	263	static long ceph_ioctl_syncio(struct file *file)
	264	{
	265	struct ceph_file_info *fi = file->private_data;
	266
	267	fi->flags \|= CEPH_F_SYNC;
	268	return 0;
	269	}
	270
8f4e91de SW	271	long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
	272	{
	273	dout("ioctl file %p cmd %u arg %lu\n", file, cmd, arg);
	274	switch (cmd) {
	275	case CEPH_IOC_GET_LAYOUT:
	276	return ceph_ioctl_get_layout(file, (void __user *)arg);
	277
	278	case CEPH_IOC_SET_LAYOUT:
	279	return ceph_ioctl_set_layout(file, (void __user *)arg);
	280
571dba52 GF	281	case CEPH_IOC_SET_LAYOUT_POLICY:
	282	return ceph_ioctl_set_layout_policy(file, (void __user *)arg);
	283
8f4e91de SW	284	case CEPH_IOC_GET_DATALOC:
8f4e91de SW	285	return ceph_ioctl_get_dataloc(file, (void __user *)arg);
8c6e9229 SW	286
	287	case CEPH_IOC_LAZYIO:
	288	return ceph_ioctl_lazyio(file);
4918b6d1 SW	289
	290	case CEPH_IOC_SYNCIO:
	291	return ceph_ioctl_syncio(file);
8f4e91de	292	}
571dba52	293
8f4e91de SW	294	return -ENOTTY;
8f4e91de SW	295	}