[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)le32_to_cpu(ci->i_layout.fl_pg_preferred);
		if (copy_to_user(arg, &l, sizeof(l)))
			return -EFAULT;
	}

	return err;
}

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, i;

	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);
		nl.preferred_osd =
				(s32)le32_to_cpu(ci->i_layout.fl_pg_preferred);
	} 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;
	if (l.preferred_osd)
		nl.preferred_osd = l.preferred_osd;

	if ((nl.object_size & ~PAGE_MASK) ||
	    (nl.stripe_unit & ~PAGE_MASK) ||
	    ((unsigned)nl.object_size % (unsigned)nl.stripe_unit))
		return -EINVAL;

	/* make sure it's a valid data pool */
	if (l.data_pool > 0) {
		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;
	}

	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);
	req->r_args.setlayout.layout.fl_pg_preferred =
		cpu_to_le32(l.preferred_osd);

	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, i;
	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;

	if ((l.object_size & ~PAGE_MASK) ||
	    (l.stripe_unit & ~PAGE_MASK) ||
	    !l.stripe_unit ||
	    (l.object_size &&
	        (unsigned)l.object_size % (unsigned)l.stripe_unit))
		return -EINVAL;

	/* make sure it's a valid data pool */
	if (l.data_pool > 0) {
		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;
	}

	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);
	req->r_args.setlayout.layout.fl_pg_preferred =
			cpu_to_le32(l.preferred_osd);

	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);
33d4909c SW	29	l.preferred_osd =
33d4909c SW	30	(s32)le32_to_cpu(ci->i_layout.fl_pg_preferred);
8f4e91de SW	31	if (copy_to_user(arg, &l, sizeof(l)))
	32	return -EFAULT;
	33	}
	34
	35	return err;
	36	}
	37
	38	static long ceph_ioctl_set_layout(struct file file, void __user arg)
	39	{
	40	struct inode *inode = file->f_dentry->d_inode;
5f21c96d	41	struct inode *parent_inode;
3d14c5d2	42	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
8f4e91de SW	43	struct ceph_mds_request *req;
8f4e91de SW	44	struct ceph_ioctl_layout l;
a35eca95 GF	45	struct ceph_inode_info *ci = ceph_inode(file->f_dentry->d_inode);
a35eca95 GF	46	struct ceph_ioctl_layout nl;
8f4e91de SW	47	int err, i;
8f4e91de SW	48
8f4e91de SW	49	if (copy_from_user(&l, arg, sizeof(l)))
	50	return -EFAULT;
	51
a35eca95 GF	52	/* validate changed params against current layout */
	53	err = ceph_do_getattr(file->f_dentry->d_inode, CEPH_STAT_CAP_LAYOUT);
	54	if (!err) {
	55	nl.stripe_unit = ceph_file_layout_su(ci->i_layout);
	56	nl.stripe_count = ceph_file_layout_stripe_count(ci->i_layout);
	57	nl.object_size = ceph_file_layout_object_size(ci->i_layout);
	58	nl.data_pool = le32_to_cpu(ci->i_layout.fl_pg_pool);
	59	nl.preferred_osd =
	60	(s32)le32_to_cpu(ci->i_layout.fl_pg_preferred);
	61	} else
	62	return err;
	63
	64	if (l.stripe_count)
	65	nl.stripe_count = l.stripe_count;
	66	if (l.stripe_unit)
	67	nl.stripe_unit = l.stripe_unit;
	68	if (l.object_size)
	69	nl.object_size = l.object_size;
	70	if (l.data_pool)
	71	nl.data_pool = l.data_pool;
	72	if (l.preferred_osd)
	73	nl.preferred_osd = l.preferred_osd;
	74
	75	if ((nl.object_size & ~PAGE_MASK) \|\|
	76	(nl.stripe_unit & ~PAGE_MASK) \|\|
	77	((unsigned)nl.object_size % (unsigned)nl.stripe_unit))
8f4e91de SW	78	return -EINVAL;
	79
	80	/* make sure it's a valid data pool */
	81	if (l.data_pool > 0) {
	82	mutex_lock(&mdsc->mutex);
	83	err = -EINVAL;
	84	for (i = 0; i < mdsc->mdsmap->m_num_data_pg_pools; i++)
	85	if (mdsc->mdsmap->m_data_pg_pools[i] == l.data_pool) {
	86	err = 0;
	87	break;
	88	}
	89	mutex_unlock(&mdsc->mutex);
	90	if (err)
	91	return err;
	92	}
	93
	94	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETLAYOUT,
	95	USE_AUTH_MDS);
	96	if (IS_ERR(req))
	97	return PTR_ERR(req);
70b666c3 SW	98	req->r_inode = inode;
70b666c3 SW	99	ihold(inode);
8f4e91de SW	100	req->r_inode_drop = CEPH_CAP_FILE_SHARED \| CEPH_CAP_FILE_EXCL;
	101
	102	req->r_args.setlayout.layout.fl_stripe_unit =
	103	cpu_to_le32(l.stripe_unit);
	104	req->r_args.setlayout.layout.fl_stripe_count =
	105	cpu_to_le32(l.stripe_count);
	106	req->r_args.setlayout.layout.fl_object_size =
	107	cpu_to_le32(l.object_size);
	108	req->r_args.setlayout.layout.fl_pg_pool = cpu_to_le32(l.data_pool);
33d4909c SW	109	req->r_args.setlayout.layout.fl_pg_preferred =
33d4909c SW	110	cpu_to_le32(l.preferred_osd);
8f4e91de	111
5f21c96d	112	parent_inode = ceph_get_dentry_parent_inode(file->f_dentry);
8f4e91de	113	err = ceph_mdsc_do_request(mdsc, parent_inode, req);
5f21c96d	114	iput(parent_inode);
8f4e91de SW	115	ceph_mdsc_put_request(req);
	116	return err;
	117	}
	118
571dba52 GF	119	/*
	120	* Set a layout policy on a directory inode. All items in the tree
	121	* rooted at this inode will inherit this layout on creation,
	122	* (It doesn't apply retroactively )
	123	* unless a subdirectory has its own layout policy.
	124	*/
	125	static long ceph_ioctl_set_layout_policy (struct file file, void __user arg)
	126	{
	127	struct inode *inode = file->f_dentry->d_inode;
	128	struct ceph_mds_request *req;
	129	struct ceph_ioctl_layout l;
	130	int err, i;
	131	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
	132
	133	/* copy and validate */
	134	if (copy_from_user(&l, arg, sizeof(l)))
	135	return -EFAULT;
	136
	137	if ((l.object_size & ~PAGE_MASK) \|\|
	138	(l.stripe_unit & ~PAGE_MASK) \|\|
	139	!l.stripe_unit \|\|
	140	(l.object_size &&
	141	(unsigned)l.object_size % (unsigned)l.stripe_unit))
	142	return -EINVAL;
	143
	144	/* make sure it's a valid data pool */
	145	if (l.data_pool > 0) {
	146	mutex_lock(&mdsc->mutex);
	147	err = -EINVAL;
	148	for (i = 0; i < mdsc->mdsmap->m_num_data_pg_pools; i++)
	149	if (mdsc->mdsmap->m_data_pg_pools[i] == l.data_pool) {
	150	err = 0;
	151	break;
	152	}
	153	mutex_unlock(&mdsc->mutex);
	154	if (err)
	155	return err;
	156	}
	157
	158	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETDIRLAYOUT,
	159	USE_AUTH_MDS);
	160
	161	if (IS_ERR(req))
	162	return PTR_ERR(req);
70b666c3 SW	163	req->r_inode = inode;
70b666c3 SW	164	ihold(inode);
571dba52 GF	165
	166	req->r_args.setlayout.layout.fl_stripe_unit =
	167	cpu_to_le32(l.stripe_unit);
	168	req->r_args.setlayout.layout.fl_stripe_count =
	169	cpu_to_le32(l.stripe_count);
	170	req->r_args.setlayout.layout.fl_object_size =
	171	cpu_to_le32(l.object_size);
	172	req->r_args.setlayout.layout.fl_pg_pool =
	173	cpu_to_le32(l.data_pool);
	174	req->r_args.setlayout.layout.fl_pg_preferred =
	175	cpu_to_le32(l.preferred_osd);
	176
	177	err = ceph_mdsc_do_request(mdsc, inode, req);
	178	ceph_mdsc_put_request(req);
	179	return err;
	180	}
	181
8f4e91de SW	182	/*
	183	* Return object name, size/offset information, and location (OSD
	184	* number, network address) for a given file offset.
	185	*/
	186	static long ceph_ioctl_get_dataloc(struct file file, void __user arg)
	187	{
	188	struct ceph_ioctl_dataloc dl;
	189	struct inode *inode = file->f_dentry->d_inode;
	190	struct ceph_inode_info *ci = ceph_inode(inode);
3d14c5d2 YS	191	struct ceph_osd_client *osdc =
3d14c5d2 YS	192	&ceph_sb_to_client(inode->i_sb)->client->osdc;
8f4e91de SW	193	u64 len = 1, olen;
	194	u64 tmp;
	195	struct ceph_object_layout ol;
51042122	196	struct ceph_pg pgid;
8f4e91de SW	197
	198	/* copy and validate */
	199	if (copy_from_user(&dl, arg, sizeof(dl)))
	200	return -EFAULT;
	201
	202	down_read(&osdc->map_sem);
	203	ceph_calc_file_object_mapping(&ci->i_layout, dl.file_offset, &len,
	204	&dl.object_no, &dl.object_offset, &olen);
	205	dl.file_offset -= dl.object_offset;
	206	dl.object_size = ceph_file_layout_object_size(ci->i_layout);
	207	dl.block_size = ceph_file_layout_su(ci->i_layout);
	208
	209	/* block_offset = object_offset % block_size */
	210	tmp = dl.object_offset;
	211	dl.block_offset = do_div(tmp, dl.block_size);
	212
	213	snprintf(dl.object_name, sizeof(dl.object_name), "%llx.%08llx",
	214	ceph_ino(inode), dl.object_no);
	215	ceph_calc_object_layout(&ol, dl.object_name, &ci->i_layout,
	216	osdc->osdmap);
	217
51042122	218	pgid = ol.ol_pgid;
8f4e91de SW	219	dl.osd = ceph_calc_pg_primary(osdc->osdmap, pgid);
	220	if (dl.osd >= 0) {
	221	struct ceph_entity_addr *a =
	222	ceph_osd_addr(osdc->osdmap, dl.osd);
	223	if (a)
	224	memcpy(&dl.osd_addr, &a->in_addr, sizeof(dl.osd_addr));
	225	} else {
	226	memset(&dl.osd_addr, 0, sizeof(dl.osd_addr));
	227	}
	228	up_read(&osdc->map_sem);
	229
	230	/* send result back to user */
	231	if (copy_to_user(arg, &dl, sizeof(dl)))
	232	return -EFAULT;
	233
	234	return 0;
	235	}
	236
8c6e9229 SW	237	static long ceph_ioctl_lazyio(struct file *file)
	238	{
	239	struct ceph_file_info *fi = file->private_data;
	240	struct inode *inode = file->f_dentry->d_inode;
	241	struct ceph_inode_info *ci = ceph_inode(inode);
	242
	243	if ((fi->fmode & CEPH_FILE_MODE_LAZY) == 0) {
be655596	244	spin_lock(&ci->i_ceph_lock);
8c6e9229 SW	245	ci->i_nr_by_mode[fi->fmode]--;
	246	fi->fmode \|= CEPH_FILE_MODE_LAZY;
	247	ci->i_nr_by_mode[fi->fmode]++;
be655596	248	spin_unlock(&ci->i_ceph_lock);
8c6e9229 SW	249	dout("ioctl_layzio: file %p marked lazy\n", file);
	250
	251	ceph_check_caps(ci, 0, NULL);
	252	} else {
	253	dout("ioctl_layzio: file %p already lazy\n", file);
	254	}
	255	return 0;
	256	}
	257
4918b6d1 SW	258	static long ceph_ioctl_syncio(struct file *file)
	259	{
	260	struct ceph_file_info *fi = file->private_data;
	261
	262	fi->flags \|= CEPH_F_SYNC;
	263	return 0;
	264	}
	265
8f4e91de SW	266	long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
	267	{
	268	dout("ioctl file %p cmd %u arg %lu\n", file, cmd, arg);
	269	switch (cmd) {
	270	case CEPH_IOC_GET_LAYOUT:
	271	return ceph_ioctl_get_layout(file, (void __user *)arg);
	272
	273	case CEPH_IOC_SET_LAYOUT:
	274	return ceph_ioctl_set_layout(file, (void __user *)arg);
	275
571dba52 GF	276	case CEPH_IOC_SET_LAYOUT_POLICY:
	277	return ceph_ioctl_set_layout_policy(file, (void __user *)arg);
	278
8f4e91de SW	279	case CEPH_IOC_GET_DATALOC:
8f4e91de SW	280	return ceph_ioctl_get_dataloc(file, (void __user *)arg);
8c6e9229 SW	281
	282	case CEPH_IOC_LAZYIO:
	283	return ceph_ioctl_lazyio(file);
4918b6d1 SW	284
	285	case CEPH_IOC_SYNCIO:
	286	return ceph_ioctl_syncio(file);
8f4e91de	287	}
571dba52	288
8f4e91de SW	289	return -ENOTTY;
8f4e91de SW	290	}