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

// SPDX-License-Identifier: GPL-2.0
#include <linux/ceph/ceph_debug.h>
#include <linux/in.h>

#include "super.h"
#include "mds_client.h"
#include "ioctl.h"
#include <linux/ceph/striper.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 = ci->i_layout.stripe_unit;
		l.stripe_count = ci->i_layout.stripe_count;
		l.object_size = ci->i_layout.object_size;
		l.data_pool = ci->i_layout.pool_id;
		l.preferred_osd = -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 = ci->i_layout.stripe_count;
	if (l.stripe_unit)
		nl.stripe_unit = l.stripe_unit;
	else
		nl.stripe_unit = ci->i_layout.stripe_unit;
	if (l.object_size)
		nl.object_size = l.object_size;
	else
		nl.object_size = ci->i_layout.object_size;
	if (l.data_pool)
		nl.data_pool = l.data_pool;
	else
		nl.data_pool = ci->i_layout.pool_id;

	/* this is obsolete, and always -1 */
	nl.preferred_osd = -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;
	CEPH_DEFINE_OID_ONSTACK(oid);
	u32 xlen;
	u64 tmp;
	struct ceph_pg pgid;
	int r;

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

	down_read(&osdc->lock);
	ceph_calc_file_object_mapping(&ci->i_layout, dl.file_offset, 1,
				      &dl.object_no, &dl.object_offset, &xlen);
	dl.file_offset -= dl.object_offset;
	dl.object_size = ci->i_layout.object_size;
	dl.block_size = ci->i_layout.stripe_unit;

	/* 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 = ci->i_layout.pool_id;
	oloc.pool_ns = ceph_try_get_string(ci->i_layout.pool_ns);
	ceph_oid_printf(&oid, "%s", dl.object_name);

	r = ceph_object_locator_to_pg(osdc->osdmap, &oid, &oloc, &pgid);

	ceph_oloc_destroy(&oloc);
	if (r < 0) {
		up_read(&osdc->lock);
		return r;
	}

	dl.osd = ceph_pg_to_acting_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->lock);

	/* 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);
	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;

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

		ceph_check_caps(ci, 0);
	} 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
b2441318	1	// SPDX-License-Identifier: GPL-2.0
96c57ade	2	#include <linux/ceph/ceph_debug.h>
8f4e91de SW	3	#include <linux/in.h>
8f4e91de SW	4
8f4e91de	5	#include "super.h"
3d14c5d2	6	#include "mds_client.h"
3d14c5d2	7	#include "ioctl.h"
08c1ac50	8	#include <linux/ceph/striper.h>
8f4e91de SW	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
508b32d8	23	err = ceph_do_getattr(file_inode(file), CEPH_STAT_CAP_LAYOUT, false);
8f4e91de	24	if (!err) {
7627151e YZ	25	l.stripe_unit = ci->i_layout.stripe_unit;
	26	l.stripe_count = ci->i_layout.stripe_count;
	27	l.object_size = ci->i_layout.object_size;
	28	l.data_pool = ci->i_layout.pool_id;
24c149ad	29	l.preferred_osd = -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) \|\|
0bc62284	45	((unsigned)l->stripe_unit != 0 &&
45f2e081	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 */
508b32d8	78	err = ceph_do_getattr(file_inode(file), CEPH_STAT_CAP_LAYOUT, false);
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	85	else
7627151e	86	nl.stripe_count = ci->i_layout.stripe_count;
a35eca95 GF	87	if (l.stripe_unit)
a35eca95 GF	88	nl.stripe_unit = l.stripe_unit;
702aeb1f	89	else
7627151e	90	nl.stripe_unit = ci->i_layout.stripe_unit;
a35eca95 GF	91	if (l.object_size)
a35eca95 GF	92	nl.object_size = l.object_size;
702aeb1f	93	else
7627151e	94	nl.object_size = ci->i_layout.object_size;
a35eca95 GF	95	if (l.data_pool)
a35eca95 GF	96	nl.data_pool = l.data_pool;
702aeb1f	97	else
7627151e	98	nl.data_pool = ci->i_layout.pool_id;
702aeb1f SW	99
702aeb1f SW	100	/* this is obsolete, and always -1 */
24c149ad	101	nl.preferred_osd = -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);
3bd58143 YZ	113	req->r_num_caps = 1;
3bd58143 YZ	114
8f4e91de SW	115	req->r_inode_drop = CEPH_CAP_FILE_SHARED \| CEPH_CAP_FILE_EXCL;
	116
	117	req->r_args.setlayout.layout.fl_stripe_unit =
	118	cpu_to_le32(l.stripe_unit);
	119	req->r_args.setlayout.layout.fl_stripe_count =
	120	cpu_to_le32(l.stripe_count);
	121	req->r_args.setlayout.layout.fl_object_size =
	122	cpu_to_le32(l.object_size);
	123	req->r_args.setlayout.layout.fl_pg_pool = cpu_to_le32(l.data_pool);
8f4e91de	124
752c8bdc	125	err = ceph_mdsc_do_request(mdsc, NULL, req);
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	{
496ad9aa	138	struct inode *inode = file_inode(file);
571dba52 GF	139	struct ceph_mds_request *req;
571dba52 GF	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);
3bd58143	159	req->r_num_caps = 1;
571dba52 GF	160
	161	req->r_args.setlayout.layout.fl_stripe_unit =
	162	cpu_to_le32(l.stripe_unit);
	163	req->r_args.setlayout.layout.fl_stripe_count =
	164	cpu_to_le32(l.stripe_count);
	165	req->r_args.setlayout.layout.fl_object_size =
	166	cpu_to_le32(l.object_size);
	167	req->r_args.setlayout.layout.fl_pg_pool =
	168	cpu_to_le32(l.data_pool);
571dba52 GF	169
	170	err = ceph_mdsc_do_request(mdsc, inode, req);
	171	ceph_mdsc_put_request(req);
	172	return err;
	173	}
	174
8f4e91de SW	175	/*
	176	* Return object name, size/offset information, and location (OSD
	177	* number, network address) for a given file offset.
	178	*/
	179	static long ceph_ioctl_get_dataloc(struct file file, void __user arg)
	180	{
	181	struct ceph_ioctl_dataloc dl;
496ad9aa	182	struct inode *inode = file_inode(file);
8f4e91de	183	struct ceph_inode_info *ci = ceph_inode(inode);
3d14c5d2 YS	184	struct ceph_osd_client *osdc =
3d14c5d2 YS	185	&ceph_sb_to_client(inode->i_sb)->client->osdc;
7c13cb64	186	struct ceph_object_locator oloc;
281dbe5d	187	CEPH_DEFINE_OID_ONSTACK(oid);
dccbf080	188	u32 xlen;
8f4e91de	189	u64 tmp;
51042122	190	struct ceph_pg pgid;
457712a0	191	int r;
8f4e91de SW	192
	193	/* copy and validate */
	194	if (copy_from_user(&dl, arg, sizeof(dl)))
	195	return -EFAULT;
	196
5aea3dcd	197	down_read(&osdc->lock);
dccbf080 ID	198	ceph_calc_file_object_mapping(&ci->i_layout, dl.file_offset, 1,
dccbf080 ID	199	&dl.object_no, &dl.object_offset, &xlen);
8f4e91de	200	dl.file_offset -= dl.object_offset;
7627151e YZ	201	dl.object_size = ci->i_layout.object_size;
7627151e YZ	202	dl.block_size = ci->i_layout.stripe_unit;
8f4e91de SW	203
	204	/* block_offset = object_offset % block_size */
	205	tmp = dl.object_offset;
	206	dl.block_offset = do_div(tmp, dl.block_size);
	207
	208	snprintf(dl.object_name, sizeof(dl.object_name), "%llx.%08llx",
	209	ceph_ino(inode), dl.object_no);
41766f87	210
7627151e	211	oloc.pool = ci->i_layout.pool_id;
779fe0fb	212	oloc.pool_ns = ceph_try_get_string(ci->i_layout.pool_ns);
d30291b9	213	ceph_oid_printf(&oid, "%s", dl.object_name);
7c13cb64	214
d9591f5e	215	r = ceph_object_locator_to_pg(osdc->osdmap, &oid, &oloc, &pgid);
779fe0fb YZ	216
779fe0fb YZ	217	ceph_oloc_destroy(&oloc);
2fbcbff1	218	if (r < 0) {
5aea3dcd	219	up_read(&osdc->lock);
2fbcbff1	220	return r;
2fbcbff1	221	}
8f4e91de	222
f81f1633	223	dl.osd = ceph_pg_to_acting_primary(osdc->osdmap, &pgid);
8f4e91de SW	224	if (dl.osd >= 0) {
	225	struct ceph_entity_addr *a =
	226	ceph_osd_addr(osdc->osdmap, dl.osd);
	227	if (a)
	228	memcpy(&dl.osd_addr, &a->in_addr, sizeof(dl.osd_addr));
	229	} else {
	230	memset(&dl.osd_addr, 0, sizeof(dl.osd_addr));
	231	}
5aea3dcd	232	up_read(&osdc->lock);
8f4e91de SW	233
	234	/* send result back to user */
	235	if (copy_to_user(arg, &dl, sizeof(dl)))
	236	return -EFAULT;
	237
	238	return 0;
	239	}
	240
8c6e9229 SW	241	static long ceph_ioctl_lazyio(struct file *file)
	242	{
	243	struct ceph_file_info *fi = file->private_data;
496ad9aa	244	struct inode *inode = file_inode(file);
8c6e9229	245	struct ceph_inode_info *ci = ceph_inode(inode);
719a2514	246	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
8c6e9229 SW	247
8c6e9229 SW	248	if ((fi->fmode & CEPH_FILE_MODE_LAZY) == 0) {
be655596	249	spin_lock(&ci->i_ceph_lock);
8c6e9229	250	fi->fmode \|= CEPH_FILE_MODE_LAZY;
774a6a11	251	ci->i_nr_by_mode[ffs(CEPH_FILE_MODE_LAZY)]++;
719a2514	252	__ceph_touch_fmode(ci, mdsc, fi->fmode);
be655596	253	spin_unlock(&ci->i_ceph_lock);
8c6e9229 SW	254	dout("ioctl_layzio: file %p marked lazy\n", file);
8c6e9229 SW	255
e4b731cc	256	ceph_check_caps(ci, 0);
8c6e9229 SW	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	}