[linux.git] / fs / ceph / locks.c

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

#include <linux/file.h>
#include <linux/namei.h>
#include <linux/random.h>

#include "super.h"
#include "mds_client.h"
#include <linux/ceph/pagelist.h>

static u64 lock_secret;
static int ceph_lock_wait_for_completion(struct ceph_mds_client *mdsc,
                                         struct ceph_mds_request *req);

static inline u64 secure_addr(void *addr)
{
	u64 v = lock_secret ^ (u64)(unsigned long)addr;
	/*
	 * Set the most significant bit, so that MDS knows the 'owner'
	 * is sufficient to identify the owner of lock. (old code uses
	 * both 'owner' and 'pid')
	 */
	v |= (1ULL << 63);
	return v;
}

void __init ceph_flock_init(void)
{
	get_random_bytes(&lock_secret, sizeof(lock_secret));
}

/**
 * Implement fcntl and flock locking functions.
 */
static int ceph_lock_message(u8 lock_type, u16 operation, struct file *file,
			     int cmd, u8 wait, struct file_lock *fl)
{
	struct inode *inode = file_inode(file);
	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
	struct ceph_mds_request *req;
	int err;
	u64 length = 0;
	u64 owner;

	if (operation != CEPH_MDS_OP_SETFILELOCK || cmd == CEPH_LOCK_UNLOCK)
		wait = 0;

	req = ceph_mdsc_create_request(mdsc, operation, USE_AUTH_MDS);
	if (IS_ERR(req))
		return PTR_ERR(req);
	req->r_inode = inode;
	ihold(inode);
	req->r_num_caps = 1;

	/* mds requires start and length rather than start and end */
	if (LLONG_MAX == fl->fl_end)
		length = 0;
	else
		length = fl->fl_end - fl->fl_start + 1;

	owner = secure_addr(fl->fl_owner);

	dout("ceph_lock_message: rule: %d, op: %d, owner: %llx, pid: %llu, "
	     "start: %llu, length: %llu, wait: %d, type: %d", (int)lock_type,
	     (int)operation, owner, (u64)fl->fl_pid, fl->fl_start, length,
	     wait, fl->fl_type);

	req->r_args.filelock_change.rule = lock_type;
	req->r_args.filelock_change.type = cmd;
	req->r_args.filelock_change.owner = cpu_to_le64(owner);
	req->r_args.filelock_change.pid = cpu_to_le64((u64)fl->fl_pid);
	req->r_args.filelock_change.start = cpu_to_le64(fl->fl_start);
	req->r_args.filelock_change.length = cpu_to_le64(length);
	req->r_args.filelock_change.wait = wait;

	if (wait)
		req->r_wait_for_completion = ceph_lock_wait_for_completion;

	err = ceph_mdsc_do_request(mdsc, inode, req);

	if (operation == CEPH_MDS_OP_GETFILELOCK) {
		fl->fl_pid = -le64_to_cpu(req->r_reply_info.filelock_reply->pid);
		if (CEPH_LOCK_SHARED == req->r_reply_info.filelock_reply->type)
			fl->fl_type = F_RDLCK;
		else if (CEPH_LOCK_EXCL == req->r_reply_info.filelock_reply->type)
			fl->fl_type = F_WRLCK;
		else
			fl->fl_type = F_UNLCK;

		fl->fl_start = le64_to_cpu(req->r_reply_info.filelock_reply->start);
		length = le64_to_cpu(req->r_reply_info.filelock_reply->start) +
						 le64_to_cpu(req->r_reply_info.filelock_reply->length);
		if (length >= 1)
			fl->fl_end = length -1;
		else
			fl->fl_end = 0;

	}
	ceph_mdsc_put_request(req);
	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
	     "length: %llu, wait: %d, type: %d, err code %d", (int)lock_type,
	     (int)operation, (u64)fl->fl_pid, fl->fl_start,
	     length, wait, fl->fl_type, err);
	return err;
}

static int ceph_lock_wait_for_completion(struct ceph_mds_client *mdsc,
                                         struct ceph_mds_request *req)
{
	struct ceph_mds_request *intr_req;
	struct inode *inode = req->r_inode;
	int err, lock_type;

	BUG_ON(req->r_op != CEPH_MDS_OP_SETFILELOCK);
	if (req->r_args.filelock_change.rule == CEPH_LOCK_FCNTL)
		lock_type = CEPH_LOCK_FCNTL_INTR;
	else if (req->r_args.filelock_change.rule == CEPH_LOCK_FLOCK)
		lock_type = CEPH_LOCK_FLOCK_INTR;
	else
		BUG_ON(1);
	BUG_ON(req->r_args.filelock_change.type == CEPH_LOCK_UNLOCK);

	err = wait_for_completion_interruptible(&req->r_completion);
	if (!err)
		return 0;

	dout("ceph_lock_wait_for_completion: request %llu was interrupted\n",
	     req->r_tid);

	mutex_lock(&mdsc->mutex);
	if (test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
		err = 0;
	} else {
		/*
		 * ensure we aren't running concurrently with
		 * ceph_fill_trace or ceph_readdir_prepopulate, which
		 * rely on locks (dir mutex) held by our caller.
		 */
		mutex_lock(&req->r_fill_mutex);
		req->r_err = err;
		set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
		mutex_unlock(&req->r_fill_mutex);

		if (!req->r_session) {
			// haven't sent the request
			err = 0;
		}
	}
	mutex_unlock(&mdsc->mutex);
	if (!err)
		return 0;

	intr_req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETFILELOCK,
					    USE_AUTH_MDS);
	if (IS_ERR(intr_req))
		return PTR_ERR(intr_req);

	intr_req->r_inode = inode;
	ihold(inode);
	intr_req->r_num_caps = 1;

	intr_req->r_args.filelock_change = req->r_args.filelock_change;
	intr_req->r_args.filelock_change.rule = lock_type;
	intr_req->r_args.filelock_change.type = CEPH_LOCK_UNLOCK;

	err = ceph_mdsc_do_request(mdsc, inode, intr_req);
	ceph_mdsc_put_request(intr_req);

	if (err && err != -ERESTARTSYS)
		return err;

	wait_for_completion_killable(&req->r_safe_completion);
	return 0;
}

/**
 * Attempt to set an fcntl lock.
 * For now, this just goes away to the server. Later it may be more awesome.
 */
int ceph_lock(struct file *file, int cmd, struct file_lock *fl)
{
	u8 lock_cmd;
	int err;
	u8 wait = 0;
	u16 op = CEPH_MDS_OP_SETFILELOCK;

	if (!(fl->fl_flags & FL_POSIX))
		return -ENOLCK;
	/* No mandatory locks */
	if (__mandatory_lock(file->f_mapping->host) && fl->fl_type != F_UNLCK)
		return -ENOLCK;

	dout("ceph_lock, fl_owner: %p", fl->fl_owner);

	/* set wait bit as appropriate, then make command as Ceph expects it*/
	if (IS_GETLK(cmd))
		op = CEPH_MDS_OP_GETFILELOCK;
	else if (IS_SETLKW(cmd))
		wait = 1;

	if (F_RDLCK == fl->fl_type)
		lock_cmd = CEPH_LOCK_SHARED;
	else if (F_WRLCK == fl->fl_type)
		lock_cmd = CEPH_LOCK_EXCL;
	else
		lock_cmd = CEPH_LOCK_UNLOCK;

	err = ceph_lock_message(CEPH_LOCK_FCNTL, op, file, lock_cmd, wait, fl);
	if (!err) {
		if (op != CEPH_MDS_OP_GETFILELOCK) {
			dout("mds locked, locking locally");
			err = posix_lock_file(file, fl, NULL);
			if (err && (CEPH_MDS_OP_SETFILELOCK == op)) {
				/* undo! This should only happen if
				 * the kernel detects local
				 * deadlock. */
				ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
						  CEPH_LOCK_UNLOCK, 0, fl);
				dout("got %d on posix_lock_file, undid lock",
				     err);
			}
		}
	}
	return err;
}

int ceph_flock(struct file *file, int cmd, struct file_lock *fl)
{
	u8 lock_cmd;
	int err;
	u8 wait = 0;

	if (!(fl->fl_flags & FL_FLOCK))
		return -ENOLCK;
	/* No mandatory locks */
	if (fl->fl_type & LOCK_MAND)
		return -EOPNOTSUPP;

	dout("ceph_flock, fl_file: %p", fl->fl_file);

	if (IS_SETLKW(cmd))
		wait = 1;

	if (F_RDLCK == fl->fl_type)
		lock_cmd = CEPH_LOCK_SHARED;
	else if (F_WRLCK == fl->fl_type)
		lock_cmd = CEPH_LOCK_EXCL;
	else
		lock_cmd = CEPH_LOCK_UNLOCK;

	err = ceph_lock_message(CEPH_LOCK_FLOCK, CEPH_MDS_OP_SETFILELOCK,
				file, lock_cmd, wait, fl);
	if (!err) {
		err = locks_lock_file_wait(file, fl);
		if (err) {
			ceph_lock_message(CEPH_LOCK_FLOCK,
					  CEPH_MDS_OP_SETFILELOCK,
					  file, CEPH_LOCK_UNLOCK, 0, fl);
			dout("got %d on locks_lock_file_wait, undid lock", err);
		}
	}
	return err;
}

/*
 * Fills in the passed counter variables, so you can prepare pagelist metadata
 * before calling ceph_encode_locks.
 */
void ceph_count_locks(struct inode *inode, int *fcntl_count, int *flock_count)
{
	struct file_lock *lock;
	struct file_lock_context *ctx;

	*fcntl_count = 0;
	*flock_count = 0;

	ctx = inode->i_flctx;
	if (ctx) {
		spin_lock(&ctx->flc_lock);
		list_for_each_entry(lock, &ctx->flc_posix, fl_list)
			++(*fcntl_count);
		list_for_each_entry(lock, &ctx->flc_flock, fl_list)
			++(*flock_count);
		spin_unlock(&ctx->flc_lock);
	}
	dout("counted %d flock locks and %d fcntl locks",
	     *flock_count, *fcntl_count);
}

/**
 * Encode the flock and fcntl locks for the given inode into the ceph_filelock
 * array. Must be called with inode->i_lock already held.
 * If we encounter more of a specific lock type than expected, return -ENOSPC.
 */
int ceph_encode_locks_to_buffer(struct inode *inode,
				struct ceph_filelock *flocks,
				int num_fcntl_locks, int num_flock_locks)
{
	struct file_lock *lock;
	struct file_lock_context *ctx = inode->i_flctx;
	int err = 0;
	int seen_fcntl = 0;
	int seen_flock = 0;
	int l = 0;

	dout("encoding %d flock and %d fcntl locks", num_flock_locks,
	     num_fcntl_locks);

	if (!ctx)
		return 0;

	spin_lock(&ctx->flc_lock);
	list_for_each_entry(lock, &ctx->flc_posix, fl_list) {
		++seen_fcntl;
		if (seen_fcntl > num_fcntl_locks) {
			err = -ENOSPC;
			goto fail;
		}
		err = lock_to_ceph_filelock(lock, &flocks[l]);
		if (err)
			goto fail;
		++l;
	}
	list_for_each_entry(lock, &ctx->flc_flock, fl_list) {
		++seen_flock;
		if (seen_flock > num_flock_locks) {
			err = -ENOSPC;
			goto fail;
		}
		err = lock_to_ceph_filelock(lock, &flocks[l]);
		if (err)
			goto fail;
		++l;
	}
fail:
	spin_unlock(&ctx->flc_lock);
	return err;
}

/**
 * Copy the encoded flock and fcntl locks into the pagelist.
 * Format is: #fcntl locks, sequential fcntl locks, #flock locks,
 * sequential flock locks.
 * Returns zero on success.
 */
int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
			   struct ceph_pagelist *pagelist,
			   int num_fcntl_locks, int num_flock_locks)
{
	int err = 0;
	__le32 nlocks;

	nlocks = cpu_to_le32(num_fcntl_locks);
	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
	if (err)
		goto out_fail;

	err = ceph_pagelist_append(pagelist, flocks,
				   num_fcntl_locks * sizeof(*flocks));
	if (err)
		goto out_fail;

	nlocks = cpu_to_le32(num_flock_locks);
	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
	if (err)
		goto out_fail;

	err = ceph_pagelist_append(pagelist,
				   &flocks[num_fcntl_locks],
				   num_flock_locks * sizeof(*flocks));
out_fail:
	return err;
}

/*
 * Given a pointer to a lock, convert it to a ceph filelock
 */
int lock_to_ceph_filelock(struct file_lock *lock,
			  struct ceph_filelock *cephlock)
{
	int err = 0;
	cephlock->start = cpu_to_le64(lock->fl_start);
	cephlock->length = cpu_to_le64(lock->fl_end - lock->fl_start + 1);
	cephlock->client = cpu_to_le64(0);
	cephlock->pid = cpu_to_le64((u64)lock->fl_pid);
	cephlock->owner = cpu_to_le64(secure_addr(lock->fl_owner));

	switch (lock->fl_type) {
	case F_RDLCK:
		cephlock->type = CEPH_LOCK_SHARED;
		break;
	case F_WRLCK:
		cephlock->type = CEPH_LOCK_EXCL;
		break;
	case F_UNLCK:
		cephlock->type = CEPH_LOCK_UNLOCK;
		break;
	default:
		dout("Have unknown lock type %d", lock->fl_type);
		err = -EINVAL;
	}

	return err;
}
Commit	Line	Data
b2441318	1	// SPDX-License-Identifier: GPL-2.0
3d14c5d2	2	#include <linux/ceph/ceph_debug.h>
40819f6f GF	3
	4	#include <linux/file.h>
	5	#include <linux/namei.h>
eb13e832	6	#include <linux/random.h>
40819f6f GF	7
	8	#include "super.h"
	9	#include "mds_client.h"
3d14c5d2	10	#include <linux/ceph/pagelist.h>
40819f6f	11
eb13e832	12	static u64 lock_secret;
9280be24 YZ	13	static int ceph_lock_wait_for_completion(struct ceph_mds_client *mdsc,
9280be24 YZ	14	struct ceph_mds_request *req);
eb13e832 YZ	15
	16	static inline u64 secure_addr(void *addr)
	17	{
	18	u64 v = lock_secret ^ (u64)(unsigned long)addr;
	19	/*
	20	* Set the most significant bit, so that MDS knows the 'owner'
	21	* is sufficient to identify the owner of lock. (old code uses
	22	* both 'owner' and 'pid')
	23	*/
	24	v \|= (1ULL << 63);
	25	return v;
	26	}
	27
	28	void __init ceph_flock_init(void)
	29	{
	30	get_random_bytes(&lock_secret, sizeof(lock_secret));
	31	}
	32
40819f6f GF	33	/**
	34	* Implement fcntl and flock locking functions.
	35	*/
	36	static int ceph_lock_message(u8 lock_type, u16 operation, struct file *file,
637ae8d5	37	int cmd, u8 wait, struct file_lock *fl)
40819f6f	38	{
496ad9aa	39	struct inode *inode = file_inode(file);
eb13e832	40	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
40819f6f GF	41	struct ceph_mds_request *req;
40819f6f GF	42	int err;
637ae8d5	43	u64 length = 0;
eb13e832	44	u64 owner;
40819f6f	45
9280be24 YZ	46	if (operation != CEPH_MDS_OP_SETFILELOCK \|\| cmd == CEPH_LOCK_UNLOCK)
	47	wait = 0;
	48
40819f6f GF	49	req = ceph_mdsc_create_request(mdsc, operation, USE_AUTH_MDS);
	50	if (IS_ERR(req))
	51	return PTR_ERR(req);
70b666c3 SW	52	req->r_inode = inode;
70b666c3 SW	53	ihold(inode);
3bd58143	54	req->r_num_caps = 1;
40819f6f	55
637ae8d5 HS	56	/* mds requires start and length rather than start and end */
	57	if (LLONG_MAX == fl->fl_end)
	58	length = 0;
	59	else
	60	length = fl->fl_end - fl->fl_start + 1;
	61
130d1f95	62	owner = secure_addr(fl->fl_owner);
eb13e832 YZ	63
	64	dout("ceph_lock_message: rule: %d, op: %d, owner: %llx, pid: %llu, "
	65	"start: %llu, length: %llu, wait: %d, type: %d", (int)lock_type,
	66	(int)operation, owner, (u64)fl->fl_pid, fl->fl_start, length,
	67	wait, fl->fl_type);
637ae8d5	68
40819f6f GF	69	req->r_args.filelock_change.rule = lock_type;
40819f6f GF	70	req->r_args.filelock_change.type = cmd;
eb13e832	71	req->r_args.filelock_change.owner = cpu_to_le64(owner);
637ae8d5	72	req->r_args.filelock_change.pid = cpu_to_le64((u64)fl->fl_pid);
637ae8d5	73	req->r_args.filelock_change.start = cpu_to_le64(fl->fl_start);
40819f6f GF	74	req->r_args.filelock_change.length = cpu_to_le64(length);
	75	req->r_args.filelock_change.wait = wait;
	76
9280be24 YZ	77	if (wait)
	78	req->r_wait_for_completion = ceph_lock_wait_for_completion;
	79
40819f6f	80	err = ceph_mdsc_do_request(mdsc, inode, req);
a5b10629	81
eb13e832	82	if (operation == CEPH_MDS_OP_GETFILELOCK) {
9d5b86ac	83	fl->fl_pid = -le64_to_cpu(req->r_reply_info.filelock_reply->pid);
a5b10629 HS	84	if (CEPH_LOCK_SHARED == req->r_reply_info.filelock_reply->type)
	85	fl->fl_type = F_RDLCK;
	86	else if (CEPH_LOCK_EXCL == req->r_reply_info.filelock_reply->type)
	87	fl->fl_type = F_WRLCK;
	88	else
	89	fl->fl_type = F_UNLCK;
	90
	91	fl->fl_start = le64_to_cpu(req->r_reply_info.filelock_reply->start);
	92	length = le64_to_cpu(req->r_reply_info.filelock_reply->start) +
	93	le64_to_cpu(req->r_reply_info.filelock_reply->length);
	94	if (length >= 1)
	95	fl->fl_end = length -1;
	96	else
	97	fl->fl_end = 0;
	98
	99	}
40819f6f GF	100	ceph_mdsc_put_request(req);
40819f6f GF	101	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
0c1f91f2	102	"length: %llu, wait: %d, type: %d, err code %d", (int)lock_type,
637ae8d5 HS	103	(int)operation, (u64)fl->fl_pid, fl->fl_start,
637ae8d5 HS	104	length, wait, fl->fl_type, err);
40819f6f GF	105	return err;
	106	}
	107
9280be24 YZ	108	static int ceph_lock_wait_for_completion(struct ceph_mds_client *mdsc,
	109	struct ceph_mds_request *req)
	110	{
	111	struct ceph_mds_request *intr_req;
	112	struct inode *inode = req->r_inode;
	113	int err, lock_type;
	114
	115	BUG_ON(req->r_op != CEPH_MDS_OP_SETFILELOCK);
	116	if (req->r_args.filelock_change.rule == CEPH_LOCK_FCNTL)
	117	lock_type = CEPH_LOCK_FCNTL_INTR;
	118	else if (req->r_args.filelock_change.rule == CEPH_LOCK_FLOCK)
	119	lock_type = CEPH_LOCK_FLOCK_INTR;
	120	else
	121	BUG_ON(1);
	122	BUG_ON(req->r_args.filelock_change.type == CEPH_LOCK_UNLOCK);
	123
	124	err = wait_for_completion_interruptible(&req->r_completion);
	125	if (!err)
	126	return 0;
	127
	128	dout("ceph_lock_wait_for_completion: request %llu was interrupted\n",
	129	req->r_tid);
	130
92e57e62 YZ	131	mutex_lock(&mdsc->mutex);
	132	if (test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
	133	err = 0;
	134	} else {
	135	/*
	136	* ensure we aren't running concurrently with
	137	* ceph_fill_trace or ceph_readdir_prepopulate, which
	138	* rely on locks (dir mutex) held by our caller.
	139	*/
	140	mutex_lock(&req->r_fill_mutex);
	141	req->r_err = err;
	142	set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
	143	mutex_unlock(&req->r_fill_mutex);
	144
	145	if (!req->r_session) {
	146	// haven't sent the request
	147	err = 0;
	148	}
	149	}
	150	mutex_unlock(&mdsc->mutex);
	151	if (!err)
	152	return 0;
	153
9280be24 YZ	154	intr_req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_SETFILELOCK,
	155	USE_AUTH_MDS);
	156	if (IS_ERR(intr_req))
	157	return PTR_ERR(intr_req);
	158
	159	intr_req->r_inode = inode;
	160	ihold(inode);
	161	intr_req->r_num_caps = 1;
	162
	163	intr_req->r_args.filelock_change = req->r_args.filelock_change;
	164	intr_req->r_args.filelock_change.rule = lock_type;
	165	intr_req->r_args.filelock_change.type = CEPH_LOCK_UNLOCK;
	166
	167	err = ceph_mdsc_do_request(mdsc, inode, intr_req);
	168	ceph_mdsc_put_request(intr_req);
	169
	170	if (err && err != -ERESTARTSYS)
	171	return err;
	172
92e57e62	173	wait_for_completion_killable(&req->r_safe_completion);
9280be24 YZ	174	return 0;
	175	}
	176
40819f6f GF	177	/**
	178	* Attempt to set an fcntl lock.
	179	* For now, this just goes away to the server. Later it may be more awesome.
	180	*/
	181	int ceph_lock(struct file file, int cmd, struct file_lock fl)
	182	{
40819f6f GF	183	u8 lock_cmd;
	184	int err;
	185	u8 wait = 0;
	186	u16 op = CEPH_MDS_OP_SETFILELOCK;
	187
eb70c0ce YZ	188	if (!(fl->fl_flags & FL_POSIX))
	189	return -ENOLCK;
	190	/* No mandatory locks */
	191	if (__mandatory_lock(file->f_mapping->host) && fl->fl_type != F_UNLCK)
	192	return -ENOLCK;
	193
eb13e832	194	dout("ceph_lock, fl_owner: %p", fl->fl_owner);
40819f6f GF	195
40819f6f GF	196	/* set wait bit as appropriate, then make command as Ceph expects it*/
0e8e95d6	197	if (IS_GETLK(cmd))
40819f6f	198	op = CEPH_MDS_OP_GETFILELOCK;
0e8e95d6 YZ	199	else if (IS_SETLKW(cmd))
0e8e95d6 YZ	200	wait = 1;
40819f6f GF	201
	202	if (F_RDLCK == fl->fl_type)
	203	lock_cmd = CEPH_LOCK_SHARED;
	204	else if (F_WRLCK == fl->fl_type)
	205	lock_cmd = CEPH_LOCK_EXCL;
	206	else
	207	lock_cmd = CEPH_LOCK_UNLOCK;
	208
637ae8d5	209	err = ceph_lock_message(CEPH_LOCK_FCNTL, op, file, lock_cmd, wait, fl);
40819f6f	210	if (!err) {
eb13e832	211	if (op != CEPH_MDS_OP_GETFILELOCK) {
a5b10629 HS	212	dout("mds locked, locking locally");
	213	err = posix_lock_file(file, fl, NULL);
	214	if (err && (CEPH_MDS_OP_SETFILELOCK == op)) {
0c1f91f2 SW	215	/* undo! This should only happen if
	216	* the kernel detects local
	217	* deadlock. */
a5b10629 HS	218	ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
a5b10629 HS	219	CEPH_LOCK_UNLOCK, 0, fl);
0c1f91f2 SW	220	dout("got %d on posix_lock_file, undid lock",
0c1f91f2 SW	221	err);
a5b10629	222	}
40819f6f	223	}
40819f6f GF	224	}
	225	return err;
	226	}
	227
	228	int ceph_flock(struct file file, int cmd, struct file_lock fl)
	229	{
40819f6f GF	230	u8 lock_cmd;
40819f6f GF	231	int err;
0e8e95d6	232	u8 wait = 0;
40819f6f	233
eb70c0ce YZ	234	if (!(fl->fl_flags & FL_FLOCK))
	235	return -ENOLCK;
	236	/* No mandatory locks */
db4a63aa YZ	237	if (fl->fl_type & LOCK_MAND)
db4a63aa YZ	238	return -EOPNOTSUPP;
eb70c0ce	239
eb13e832	240	dout("ceph_flock, fl_file: %p", fl->fl_file);
40819f6f	241
0e8e95d6 YZ	242	if (IS_SETLKW(cmd))
	243	wait = 1;
	244
	245	if (F_RDLCK == fl->fl_type)
40819f6f	246	lock_cmd = CEPH_LOCK_SHARED;
0e8e95d6	247	else if (F_WRLCK == fl->fl_type)
40819f6f GF	248	lock_cmd = CEPH_LOCK_EXCL;
	249	else
	250	lock_cmd = CEPH_LOCK_UNLOCK;
40819f6f GF	251
40819f6f GF	252	err = ceph_lock_message(CEPH_LOCK_FLOCK, CEPH_MDS_OP_SETFILELOCK,
637ae8d5	253	file, lock_cmd, wait, fl);
40819f6f	254	if (!err) {
4f656367	255	err = locks_lock_file_wait(file, fl);
40819f6f GF	256	if (err) {
	257	ceph_lock_message(CEPH_LOCK_FLOCK,
	258	CEPH_MDS_OP_SETFILELOCK,
637ae8d5	259	file, CEPH_LOCK_UNLOCK, 0, fl);
4f656367	260	dout("got %d on locks_lock_file_wait, undid lock", err);
40819f6f	261	}
40819f6f GF	262	}
	263	return err;
	264	}
	265
5263e31e JL	266	/*
	267	* Fills in the passed counter variables, so you can prepare pagelist metadata
	268	* before calling ceph_encode_locks.
40819f6f GF	269	*/
	270	void ceph_count_locks(struct inode inode, int fcntl_count, int *flock_count)
	271	{
e084c1bd	272	struct file_lock *lock;
5263e31e	273	struct file_lock_context *ctx;
40819f6f GF	274
	275	*fcntl_count = 0;
	276	*flock_count = 0;
	277
5263e31e JL	278	ctx = inode->i_flctx;
5263e31e JL	279	if (ctx) {
e084c1bd JL	280	spin_lock(&ctx->flc_lock);
	281	list_for_each_entry(lock, &ctx->flc_posix, fl_list)
	282	++(*fcntl_count);
	283	list_for_each_entry(lock, &ctx->flc_flock, fl_list)
	284	++(*flock_count);
	285	spin_unlock(&ctx->flc_lock);
40819f6f GF	286	}
	287	dout("counted %d flock locks and %d fcntl locks",
	288	flock_count, fcntl_count);
	289	}
	290
	291	/**
39be95e9	292	* Encode the flock and fcntl locks for the given inode into the ceph_filelock
1c8c601a	293	* array. Must be called with inode->i_lock already held.
39be95e9	294	* If we encounter more of a specific lock type than expected, return -ENOSPC.
40819f6f	295	*/
39be95e9 JS	296	int ceph_encode_locks_to_buffer(struct inode *inode,
	297	struct ceph_filelock *flocks,
	298	int num_fcntl_locks, int num_flock_locks)
40819f6f GF	299	{
40819f6f GF	300	struct file_lock *lock;
bd61e0a9	301	struct file_lock_context *ctx = inode->i_flctx;
40819f6f	302	int err = 0;
fca4451a GF	303	int seen_fcntl = 0;
fca4451a GF	304	int seen_flock = 0;
39be95e9	305	int l = 0;
40819f6f GF	306
	307	dout("encoding %d flock and %d fcntl locks", num_flock_locks,
	308	num_fcntl_locks);
39be95e9	309
bd61e0a9 JL	310	if (!ctx)
	311	return 0;
	312
6109c850	313	spin_lock(&ctx->flc_lock);
f6762cb2	314	list_for_each_entry(lock, &ctx->flc_posix, fl_list) {
bd61e0a9 JL	315	++seen_fcntl;
	316	if (seen_fcntl > num_fcntl_locks) {
	317	err = -ENOSPC;
	318	goto fail;
40819f6f	319	}
bd61e0a9 JL	320	err = lock_to_ceph_filelock(lock, &flocks[l]);
	321	if (err)
	322	goto fail;
	323	++l;
40819f6f	324	}
bd61e0a9 JL	325	list_for_each_entry(lock, &ctx->flc_flock, fl_list) {
	326	++seen_flock;
	327	if (seen_flock > num_flock_locks) {
	328	err = -ENOSPC;
	329	goto fail;
40819f6f	330	}
bd61e0a9 JL	331	err = lock_to_ceph_filelock(lock, &flocks[l]);
	332	if (err)
	333	goto fail;
	334	++l;
40819f6f GF	335	}
40819f6f GF	336	fail:
6109c850	337	spin_unlock(&ctx->flc_lock);
40819f6f GF	338	return err;
	339	}
	340
39be95e9 JS	341	/**
	342	* Copy the encoded flock and fcntl locks into the pagelist.
	343	* Format is: #fcntl locks, sequential fcntl locks, #flock locks,
	344	* sequential flock locks.
	345	* Returns zero on success.
	346	*/
	347	int ceph_locks_to_pagelist(struct ceph_filelock *flocks,
	348	struct ceph_pagelist *pagelist,
	349	int num_fcntl_locks, int num_flock_locks)
	350	{
	351	int err = 0;
	352	__le32 nlocks;
	353
	354	nlocks = cpu_to_le32(num_fcntl_locks);
	355	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
	356	if (err)
	357	goto out_fail;
	358
	359	err = ceph_pagelist_append(pagelist, flocks,
	360	num_fcntl_locks * sizeof(*flocks));
	361	if (err)
	362	goto out_fail;
	363
	364	nlocks = cpu_to_le32(num_flock_locks);
	365	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
	366	if (err)
	367	goto out_fail;
	368
	369	err = ceph_pagelist_append(pagelist,
	370	&flocks[num_fcntl_locks],
	371	num_flock_locks * sizeof(*flocks));
	372	out_fail:
	373	return err;
	374	}
	375
40819f6f GF	376	/*
	377	* Given a pointer to a lock, convert it to a ceph filelock
	378	*/
	379	int lock_to_ceph_filelock(struct file_lock *lock,
	380	struct ceph_filelock *cephlock)
	381	{
	382	int err = 0;
40819f6f GF	383	cephlock->start = cpu_to_le64(lock->fl_start);
	384	cephlock->length = cpu_to_le64(lock->fl_end - lock->fl_start + 1);
	385	cephlock->client = cpu_to_le64(0);
eb13e832	386	cephlock->pid = cpu_to_le64((u64)lock->fl_pid);
130d1f95	387	cephlock->owner = cpu_to_le64(secure_addr(lock->fl_owner));
40819f6f GF	388
	389	switch (lock->fl_type) {
	390	case F_RDLCK:
	391	cephlock->type = CEPH_LOCK_SHARED;
	392	break;
	393	case F_WRLCK:
	394	cephlock->type = CEPH_LOCK_EXCL;
	395	break;
	396	case F_UNLCK:
	397	cephlock->type = CEPH_LOCK_UNLOCK;
	398	break;
	399	default:
	400	dout("Have unknown lock type %d", lock->fl_type);
	401	err = -EINVAL;
	402	}
	403
	404	return err;
	405	}