[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->f_dentry->d_inode);
	struct ceph_ioctl_layout l;
	int err;

	err = ceph_do_getattr(file->f_dentry->d_inode, 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;

	/* preferred_osd is no longer supported */
	if (l->preferred_osd != -1)
		return -EINVAL;

	/* validate striping parameters */
	if ((l->object_size & ~PAGE_MASK) ||
	    (l->stripe_unit & ~PAGE_MASK) ||
	    ((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->f_dentry->d_inode;
	struct inode *parent_inode;
	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->f_dentry->d_inode);
	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->f_dentry->d_inode, CEPH_STAT_CAP_LAYOUT);
	if (!err) {
		nl.stripe_unit = ceph_file_layout_su(ci->i_layout);
		nl.stripe_count = ceph_file_layout_stripe_count(ci->i_layout);
		nl.object_size = ceph_file_layout_object_size(ci->i_layout);
		nl.data_pool = le32_to_cpu(ci->i_layout.fl_pg_pool);
	} else
		return err;

	if (l.stripe_count)
		nl.stripe_count = l.stripe_count;
	if (l.stripe_unit)
		nl.stripe_unit = l.stripe_unit;
	if (l.object_size)
		nl.object_size = l.object_size;
	if (l.data_pool)
		nl.data_pool = l.data_pool;

	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);

	parent_inode = ceph_get_dentry_parent_inode(file->f_dentry);
	err = ceph_mdsc_do_request(mdsc, parent_inode, req);
	iput(parent_inode);
	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->f_dentry->d_inode;
	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->f_dentry->d_inode;
	struct ceph_inode_info *ci = ceph_inode(inode);
	struct ceph_osd_client *osdc =
		&ceph_sb_to_client(inode->i_sb)->client->osdc;
	u64 len = 1, olen;
	u64 tmp;
	struct ceph_object_layout ol;
	struct ceph_pg pgid;

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

	down_read(&osdc->map_sem);
	ceph_calc_file_object_mapping(&ci->i_layout, dl.file_offset, &len,
				      &dl.object_no, &dl.object_offset, &olen);
	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);
	ceph_calc_object_layout(&ol, dl.object_name, &ci->i_layout,
				osdc->osdmap);

	pgid = ol.ol_pgid;
	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->f_dentry->d_inode;
	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	{
	19	struct ceph_inode_info *ci = ceph_inode(file->f_dentry->d_inode);
	20	struct ceph_ioctl_layout l;
	21	int err;
	22
	23	err = ceph_do_getattr(file->f_dentry->d_inode, CEPH_STAT_CAP_LAYOUT);
	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
	42	/* preferred_osd is no longer supported */
	43	if (l->preferred_osd != -1)
	44	return -EINVAL;
	45
	46	/* validate striping parameters */
	47	if ((l->object_size & ~PAGE_MASK) \|\|
	48	(l->stripe_unit & ~PAGE_MASK) \|\|
	49	((unsigned)l->object_size % (unsigned)l->stripe_unit))
	50	return -EINVAL;
	51
	52	/* make sure it's a valid data pool */
	53	mutex_lock(&mdsc->mutex);
	54	err = -EINVAL;
	55	for (i = 0; i < mdsc->mdsmap->m_num_data_pg_pools; i++)
	56	if (mdsc->mdsmap->m_data_pg_pools[i] == l->data_pool) {
	57	err = 0;
	58	break;
	59	}
	60	mutex_unlock(&mdsc->mutex);
	61	if (err)
	62	return err;
	63
	64	return 0;
	65	}
	66
8f4e91de SW	67	static long ceph_ioctl_set_layout(struct file file, void __user arg)
	68	{
	69	struct inode *inode = file->f_dentry->d_inode;
5f21c96d	70	struct inode *parent_inode;
3d14c5d2	71	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
8f4e91de SW	72	struct ceph_mds_request *req;
8f4e91de SW	73	struct ceph_ioctl_layout l;
a35eca95 GF	74	struct ceph_inode_info *ci = ceph_inode(file->f_dentry->d_inode);
a35eca95 GF	75	struct ceph_ioctl_layout nl;
e49bf4c5	76	int err;
8f4e91de	77
8f4e91de SW	78	if (copy_from_user(&l, arg, sizeof(l)))
	79	return -EFAULT;
	80
a35eca95 GF	81	/* validate changed params against current layout */
	82	err = ceph_do_getattr(file->f_dentry->d_inode, CEPH_STAT_CAP_LAYOUT);
	83	if (!err) {
	84	nl.stripe_unit = ceph_file_layout_su(ci->i_layout);
	85	nl.stripe_count = ceph_file_layout_stripe_count(ci->i_layout);
	86	nl.object_size = ceph_file_layout_object_size(ci->i_layout);
	87	nl.data_pool = le32_to_cpu(ci->i_layout.fl_pg_pool);
a35eca95 GF	88	} else
	89	return err;
	90
	91	if (l.stripe_count)
	92	nl.stripe_count = l.stripe_count;
	93	if (l.stripe_unit)
	94	nl.stripe_unit = l.stripe_unit;
	95	if (l.object_size)
	96	nl.object_size = l.object_size;
	97	if (l.data_pool)
	98	nl.data_pool = l.data_pool;
a35eca95	99
e49bf4c5 SW	100	err = __validate_layout(mdsc, &nl);
	101	if (err)
	102	return err;
8f4e91de SW	103
	104	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETLAYOUT,
	105	USE_AUTH_MDS);
	106	if (IS_ERR(req))
	107	return PTR_ERR(req);
70b666c3 SW	108	req->r_inode = inode;
70b666c3 SW	109	ihold(inode);
8f4e91de SW	110	req->r_inode_drop = CEPH_CAP_FILE_SHARED \| CEPH_CAP_FILE_EXCL;
	111
	112	req->r_args.setlayout.layout.fl_stripe_unit =
	113	cpu_to_le32(l.stripe_unit);
	114	req->r_args.setlayout.layout.fl_stripe_count =
	115	cpu_to_le32(l.stripe_count);
	116	req->r_args.setlayout.layout.fl_object_size =
	117	cpu_to_le32(l.object_size);
	118	req->r_args.setlayout.layout.fl_pg_pool = cpu_to_le32(l.data_pool);
8f4e91de	119
5f21c96d	120	parent_inode = ceph_get_dentry_parent_inode(file->f_dentry);
8f4e91de	121	err = ceph_mdsc_do_request(mdsc, parent_inode, req);
5f21c96d	122	iput(parent_inode);
8f4e91de SW	123	ceph_mdsc_put_request(req);
	124	return err;
	125	}
	126
571dba52 GF	127	/*
	128	* Set a layout policy on a directory inode. All items in the tree
	129	* rooted at this inode will inherit this layout on creation,
	130	* (It doesn't apply retroactively )
	131	* unless a subdirectory has its own layout policy.
	132	*/
	133	static long ceph_ioctl_set_layout_policy (struct file file, void __user arg)
	134	{
	135	struct inode *inode = file->f_dentry->d_inode;
	136	struct ceph_mds_request *req;
	137	struct ceph_ioctl_layout l;
e49bf4c5	138	int err;
571dba52 GF	139	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
	140
	141	/* copy and validate */
	142	if (copy_from_user(&l, arg, sizeof(l)))
	143	return -EFAULT;
	144
e49bf4c5 SW	145	err = __validate_layout(mdsc, &l);
	146	if (err)
	147	return err;
571dba52 GF	148
	149	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETDIRLAYOUT,
	150	USE_AUTH_MDS);
	151
	152	if (IS_ERR(req))
	153	return PTR_ERR(req);
70b666c3 SW	154	req->r_inode = inode;
70b666c3 SW	155	ihold(inode);
571dba52 GF	156
	157	req->r_args.setlayout.layout.fl_stripe_unit =
	158	cpu_to_le32(l.stripe_unit);
	159	req->r_args.setlayout.layout.fl_stripe_count =
	160	cpu_to_le32(l.stripe_count);
	161	req->r_args.setlayout.layout.fl_object_size =
	162	cpu_to_le32(l.object_size);
	163	req->r_args.setlayout.layout.fl_pg_pool =
	164	cpu_to_le32(l.data_pool);
571dba52 GF	165
	166	err = ceph_mdsc_do_request(mdsc, inode, req);
	167	ceph_mdsc_put_request(req);
	168	return err;
	169	}
	170
8f4e91de SW	171	/*
	172	* Return object name, size/offset information, and location (OSD
	173	* number, network address) for a given file offset.
	174	*/
	175	static long ceph_ioctl_get_dataloc(struct file file, void __user arg)
	176	{
	177	struct ceph_ioctl_dataloc dl;
	178	struct inode *inode = file->f_dentry->d_inode;
	179	struct ceph_inode_info *ci = ceph_inode(inode);
3d14c5d2 YS	180	struct ceph_osd_client *osdc =
3d14c5d2 YS	181	&ceph_sb_to_client(inode->i_sb)->client->osdc;
8f4e91de SW	182	u64 len = 1, olen;
	183	u64 tmp;
	184	struct ceph_object_layout ol;
51042122	185	struct ceph_pg pgid;
8f4e91de SW	186
	187	/* copy and validate */
	188	if (copy_from_user(&dl, arg, sizeof(dl)))
	189	return -EFAULT;
	190
	191	down_read(&osdc->map_sem);
	192	ceph_calc_file_object_mapping(&ci->i_layout, dl.file_offset, &len,
	193	&dl.object_no, &dl.object_offset, &olen);
	194	dl.file_offset -= dl.object_offset;
	195	dl.object_size = ceph_file_layout_object_size(ci->i_layout);
	196	dl.block_size = ceph_file_layout_su(ci->i_layout);
	197
	198	/* block_offset = object_offset % block_size */
	199	tmp = dl.object_offset;
	200	dl.block_offset = do_div(tmp, dl.block_size);
	201
	202	snprintf(dl.object_name, sizeof(dl.object_name), "%llx.%08llx",
	203	ceph_ino(inode), dl.object_no);
	204	ceph_calc_object_layout(&ol, dl.object_name, &ci->i_layout,
	205	osdc->osdmap);
	206
51042122	207	pgid = ol.ol_pgid;
8f4e91de SW	208	dl.osd = ceph_calc_pg_primary(osdc->osdmap, pgid);
	209	if (dl.osd >= 0) {
	210	struct ceph_entity_addr *a =
	211	ceph_osd_addr(osdc->osdmap, dl.osd);
	212	if (a)
	213	memcpy(&dl.osd_addr, &a->in_addr, sizeof(dl.osd_addr));
	214	} else {
	215	memset(&dl.osd_addr, 0, sizeof(dl.osd_addr));
	216	}
	217	up_read(&osdc->map_sem);
	218
	219	/* send result back to user */
	220	if (copy_to_user(arg, &dl, sizeof(dl)))
	221	return -EFAULT;
	222
	223	return 0;
	224	}
	225
8c6e9229 SW	226	static long ceph_ioctl_lazyio(struct file *file)
	227	{
	228	struct ceph_file_info *fi = file->private_data;
	229	struct inode *inode = file->f_dentry->d_inode;
	230	struct ceph_inode_info *ci = ceph_inode(inode);
	231
	232	if ((fi->fmode & CEPH_FILE_MODE_LAZY) == 0) {
be655596	233	spin_lock(&ci->i_ceph_lock);
8c6e9229 SW	234	ci->i_nr_by_mode[fi->fmode]--;
	235	fi->fmode \|= CEPH_FILE_MODE_LAZY;
	236	ci->i_nr_by_mode[fi->fmode]++;
be655596	237	spin_unlock(&ci->i_ceph_lock);
8c6e9229 SW	238	dout("ioctl_layzio: file %p marked lazy\n", file);
	239
	240	ceph_check_caps(ci, 0, NULL);
	241	} else {
	242	dout("ioctl_layzio: file %p already lazy\n", file);
	243	}
	244	return 0;
	245	}
	246
4918b6d1 SW	247	static long ceph_ioctl_syncio(struct file *file)
	248	{
	249	struct ceph_file_info *fi = file->private_data;
	250
	251	fi->flags \|= CEPH_F_SYNC;
	252	return 0;
	253	}
	254
8f4e91de SW	255	long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
	256	{
	257	dout("ioctl file %p cmd %u arg %lu\n", file, cmd, arg);
	258	switch (cmd) {
	259	case CEPH_IOC_GET_LAYOUT:
	260	return ceph_ioctl_get_layout(file, (void __user *)arg);
	261
	262	case CEPH_IOC_SET_LAYOUT:
	263	return ceph_ioctl_set_layout(file, (void __user *)arg);
	264
571dba52 GF	265	case CEPH_IOC_SET_LAYOUT_POLICY:
	266	return ceph_ioctl_set_layout_policy(file, (void __user *)arg);
	267
8f4e91de SW	268	case CEPH_IOC_GET_DATALOC:
8f4e91de SW	269	return ceph_ioctl_get_dataloc(file, (void __user *)arg);
8c6e9229 SW	270
	271	case CEPH_IOC_LAZYIO:
	272	return ceph_ioctl_lazyio(file);
4918b6d1 SW	273
	274	case CEPH_IOC_SYNCIO:
	275	return ceph_ioctl_syncio(file);
8f4e91de	276	}
571dba52	277
8f4e91de SW	278	return -ENOTTY;
8f4e91de SW	279	}