[deliverable/linux.git] / fs / ceph / ioctl.c

#include <linux/in.h>

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