[linux-2.6-block.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;

	/* 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)
		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);

	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
e49bf4c5 SW	42	/* validate striping parameters */
	43	if ((l->object_size & ~PAGE_MASK) \|\|
	44	(l->stripe_unit & ~PAGE_MASK) \|\|
	45	((unsigned)l->object_size % (unsigned)l->stripe_unit))
	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)
	64	{
	65	struct inode *inode = file->f_dentry->d_inode;
5f21c96d	66	struct inode *parent_inode;
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;
a35eca95 GF	70	struct ceph_inode_info *ci = ceph_inode(file->f_dentry->d_inode);
a35eca95 GF	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 GF	77	/* validate changed params against current layout */
a35eca95 GF	78	err = ceph_do_getattr(file->f_dentry->d_inode, 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
5f21c96d	123	parent_inode = ceph_get_dentry_parent_inode(file->f_dentry);
8f4e91de	124	err = ceph_mdsc_do_request(mdsc, parent_inode, req);
5f21c96d	125	iput(parent_inode);
8f4e91de SW	126	ceph_mdsc_put_request(req);
	127	return err;
	128	}
	129
571dba52 GF	130	/*
	131	* Set a layout policy on a directory inode. All items in the tree
	132	* rooted at this inode will inherit this layout on creation,
	133	* (It doesn't apply retroactively )
	134	* unless a subdirectory has its own layout policy.
	135	*/
	136	static long ceph_ioctl_set_layout_policy (struct file file, void __user arg)
	137	{
	138	struct inode *inode = file->f_dentry->d_inode;
	139	struct ceph_mds_request *req;
	140	struct ceph_ioctl_layout l;
e49bf4c5	141	int err;
571dba52 GF	142	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
	143
	144	/* copy and validate */
	145	if (copy_from_user(&l, arg, sizeof(l)))
	146	return -EFAULT;
	147
e49bf4c5 SW	148	err = __validate_layout(mdsc, &l);
	149	if (err)
	150	return err;
571dba52 GF	151
	152	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETDIRLAYOUT,
	153	USE_AUTH_MDS);
	154
	155	if (IS_ERR(req))
	156	return PTR_ERR(req);
70b666c3 SW	157	req->r_inode = inode;
70b666c3 SW	158	ihold(inode);
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;
	181	struct inode *inode = file->f_dentry->d_inode;
	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;
8f4e91de SW	185	u64 len = 1, olen;
	186	u64 tmp;
	187	struct ceph_object_layout ol;
51042122	188	struct ceph_pg pgid;
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);
	195	ceph_calc_file_object_mapping(&ci->i_layout, dl.file_offset, &len,
	196	&dl.object_no, &dl.object_offset, &olen);
	197	dl.file_offset -= dl.object_offset;
	198	dl.object_size = ceph_file_layout_object_size(ci->i_layout);
	199	dl.block_size = ceph_file_layout_su(ci->i_layout);
	200
	201	/* block_offset = object_offset % block_size */
	202	tmp = dl.object_offset;
	203	dl.block_offset = do_div(tmp, dl.block_size);
	204
	205	snprintf(dl.object_name, sizeof(dl.object_name), "%llx.%08llx",
	206	ceph_ino(inode), dl.object_no);
	207	ceph_calc_object_layout(&ol, dl.object_name, &ci->i_layout,
	208	osdc->osdmap);
	209
51042122	210	pgid = ol.ol_pgid;
8f4e91de SW	211	dl.osd = ceph_calc_pg_primary(osdc->osdmap, pgid);
	212	if (dl.osd >= 0) {
	213	struct ceph_entity_addr *a =
	214	ceph_osd_addr(osdc->osdmap, dl.osd);
	215	if (a)
	216	memcpy(&dl.osd_addr, &a->in_addr, sizeof(dl.osd_addr));
	217	} else {
	218	memset(&dl.osd_addr, 0, sizeof(dl.osd_addr));
	219	}
	220	up_read(&osdc->map_sem);
	221
	222	/* send result back to user */
	223	if (copy_to_user(arg, &dl, sizeof(dl)))
	224	return -EFAULT;
	225
	226	return 0;
	227	}
	228
8c6e9229 SW	229	static long ceph_ioctl_lazyio(struct file *file)
	230	{
	231	struct ceph_file_info *fi = file->private_data;
	232	struct inode *inode = file->f_dentry->d_inode;
	233	struct ceph_inode_info *ci = ceph_inode(inode);
	234
	235	if ((fi->fmode & CEPH_FILE_MODE_LAZY) == 0) {
be655596	236	spin_lock(&ci->i_ceph_lock);
8c6e9229 SW	237	ci->i_nr_by_mode[fi->fmode]--;
	238	fi->fmode \|= CEPH_FILE_MODE_LAZY;
	239	ci->i_nr_by_mode[fi->fmode]++;
be655596	240	spin_unlock(&ci->i_ceph_lock);
8c6e9229 SW	241	dout("ioctl_layzio: file %p marked lazy\n", file);
	242
	243	ceph_check_caps(ci, 0, NULL);
	244	} else {
	245	dout("ioctl_layzio: file %p already lazy\n", file);
	246	}
	247	return 0;
	248	}
	249
4918b6d1 SW	250	static long ceph_ioctl_syncio(struct file *file)
	251	{
	252	struct ceph_file_info *fi = file->private_data;
	253
	254	fi->flags \|= CEPH_F_SYNC;
	255	return 0;
	256	}
	257
8f4e91de SW	258	long ceph_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
	259	{
	260	dout("ioctl file %p cmd %u arg %lu\n", file, cmd, arg);
	261	switch (cmd) {
	262	case CEPH_IOC_GET_LAYOUT:
	263	return ceph_ioctl_get_layout(file, (void __user *)arg);
	264
	265	case CEPH_IOC_SET_LAYOUT:
	266	return ceph_ioctl_set_layout(file, (void __user *)arg);
	267
571dba52 GF	268	case CEPH_IOC_SET_LAYOUT_POLICY:
	269	return ceph_ioctl_set_layout_policy(file, (void __user *)arg);
	270
8f4e91de SW	271	case CEPH_IOC_GET_DATALOC:
8f4e91de SW	272	return ceph_ioctl_get_dataloc(file, (void __user *)arg);
8c6e9229 SW	273
	274	case CEPH_IOC_LAZYIO:
	275	return ceph_ioctl_lazyio(file);
4918b6d1 SW	276
	277	case CEPH_IOC_SYNCIO:
	278	return ceph_ioctl_syncio(file);
8f4e91de	279	}
571dba52	280
8f4e91de SW	281	return -ENOTTY;
8f4e91de SW	282	}