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

#include <linux/ceph/ceph_debug.h>

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

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

/**
 * 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;

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

	/* 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;

	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
	     "length: %llu, wait: %d, type: %d", (int)lock_type,
	     (int)operation, (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.pid = cpu_to_le64((u64)fl->fl_pid);
	/* This should be adjusted, but I'm not sure if
	   namespaces actually get id numbers*/
	req->r_args.filelock_change.pid_namespace =
		cpu_to_le64((u64)(unsigned long)fl->fl_nspid);
	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;

	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;
}

/**
 * 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;

	fl->fl_nspid = get_pid(task_tgid(current));
	dout("ceph_lock, fl_pid:%d", fl->fl_pid);

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

	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);
			}
		}

	} else if (err == -ERESTARTSYS) {
		dout("undoing lock\n");
		ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
				  CEPH_LOCK_UNLOCK, 0, fl);
	}
	return err;
}

int ceph_flock(struct file *file, int cmd, struct file_lock *fl)
{
	u8 lock_cmd;
	int err;
	u8 wait = 1;

	fl->fl_nspid = get_pid(task_tgid(current));
	dout("ceph_flock, fl_pid:%d", fl->fl_pid);

	/* set wait bit, then clear it out of cmd*/
	if (cmd & LOCK_NB)
		wait = 0;
	cmd = cmd & (LOCK_SH | LOCK_EX | LOCK_UN);
	/* set command sequence that Ceph wants to see:
	   shared lock, exclusive lock, or unlock */
	if (LOCK_SH == cmd)
		lock_cmd = CEPH_LOCK_SHARED;
	else if (LOCK_EX == cmd)
		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 = flock_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 flock_lock_file_wait, undid lock", err);
		}
	} else if (err == -ERESTARTSYS) {
		dout("undoing lock\n");
		ceph_lock_message(CEPH_LOCK_FLOCK,
				  CEPH_MDS_OP_SETFILELOCK,
				  file, CEPH_LOCK_UNLOCK, 0, fl);
	}
	return err;
}

/**
 * Must be called with BKL already held. 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;

	*fcntl_count = 0;
	*flock_count = 0;

	for (lock = inode->i_flock; lock != NULL; lock = lock->fl_next) {
		if (lock->fl_flags & FL_POSIX)
			++(*fcntl_count);
		else if (lock->fl_flags & FL_FLOCK)
			++(*flock_count);
	}
	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 pagelist.
 * Format is: #fcntl locks, sequential fcntl locks, #flock locks,
 * sequential flock locks.
 * Must be called with lock_flocks() already held.
 * If we encounter more of a specific lock type than expected,
 * we return the value 1.
 */
int ceph_encode_locks(struct inode *inode, struct ceph_pagelist *pagelist,
		      int num_fcntl_locks, int num_flock_locks)
{
	struct file_lock *lock;
	struct ceph_filelock cephlock;
	int err = 0;
	int seen_fcntl = 0;
	int seen_flock = 0;
	__le32 nlocks;

	dout("encoding %d flock and %d fcntl locks", num_flock_locks,
	     num_fcntl_locks);
	nlocks = cpu_to_le32(num_fcntl_locks);
	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
	if (err)
		goto fail;
	for (lock = inode->i_flock; lock != NULL; lock = lock->fl_next) {
		if (lock->fl_flags & FL_POSIX) {
			++seen_fcntl;
			if (seen_fcntl > num_fcntl_locks) {
				err = -ENOSPC;
				goto fail;
			}
			err = lock_to_ceph_filelock(lock, &cephlock);
			if (err)
				goto fail;
			err = ceph_pagelist_append(pagelist, &cephlock,
					   sizeof(struct ceph_filelock));
		}
		if (err)
			goto fail;
	}

	nlocks = cpu_to_le32(num_flock_locks);
	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
	if (err)
		goto fail;
	for (lock = inode->i_flock; lock != NULL; lock = lock->fl_next) {
		if (lock->fl_flags & FL_FLOCK) {
			++seen_flock;
			if (seen_flock > num_flock_locks) {
				err = -ENOSPC;
				goto fail;
			}
			err = lock_to_ceph_filelock(lock, &cephlock);
			if (err)
				goto fail;
			err = ceph_pagelist_append(pagelist, &cephlock,
					   sizeof(struct ceph_filelock));
		}
		if (err)
			goto fail;
	}
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(lock->fl_pid);
	cephlock->pid_namespace =
	        cpu_to_le64((u64)(unsigned long)lock->fl_nspid);

	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
3d14c5d2	1	#include <linux/ceph/ceph_debug.h>
40819f6f GF	2
	3	#include <linux/file.h>
	4	#include <linux/namei.h>
	5
	6	#include "super.h"
	7	#include "mds_client.h"
3d14c5d2	8	#include <linux/ceph/pagelist.h>
40819f6f GF	9
	10	/**
	11	* Implement fcntl and flock locking functions.
	12	*/
	13	static int ceph_lock_message(u8 lock_type, u16 operation, struct file *file,
637ae8d5	14	int cmd, u8 wait, struct file_lock *fl)
40819f6f	15	{
496ad9aa	16	struct inode *inode = file_inode(file);
40819f6f	17	struct ceph_mds_client *mdsc =
3d14c5d2	18	ceph_sb_to_client(inode->i_sb)->mdsc;
40819f6f GF	19	struct ceph_mds_request *req;
40819f6f GF	20	int err;
637ae8d5	21	u64 length = 0;
40819f6f GF	22
	23	req = ceph_mdsc_create_request(mdsc, operation, USE_AUTH_MDS);
	24	if (IS_ERR(req))
	25	return PTR_ERR(req);
70b666c3 SW	26	req->r_inode = inode;
70b666c3 SW	27	ihold(inode);
40819f6f	28
637ae8d5 HS	29	/* mds requires start and length rather than start and end */
	30	if (LLONG_MAX == fl->fl_end)
	31	length = 0;
	32	else
	33	length = fl->fl_end - fl->fl_start + 1;
	34
40819f6f	35	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
0c1f91f2	36	"length: %llu, wait: %d, type: %d", (int)lock_type,
637ae8d5 HS	37	(int)operation, (u64)fl->fl_pid, fl->fl_start,
	38	length, wait, fl->fl_type);
	39
40819f6f GF	40	req->r_args.filelock_change.rule = lock_type;
40819f6f GF	41	req->r_args.filelock_change.type = cmd;
637ae8d5	42	req->r_args.filelock_change.pid = cpu_to_le64((u64)fl->fl_pid);
40819f6f GF	43	/* This should be adjusted, but I'm not sure if
	44	namespaces actually get id numbers*/
	45	req->r_args.filelock_change.pid_namespace =
637ae8d5 HS	46	cpu_to_le64((u64)(unsigned long)fl->fl_nspid);
637ae8d5 HS	47	req->r_args.filelock_change.start = cpu_to_le64(fl->fl_start);
40819f6f GF	48	req->r_args.filelock_change.length = cpu_to_le64(length);
	49	req->r_args.filelock_change.wait = wait;
	50
	51	err = ceph_mdsc_do_request(mdsc, inode, req);
a5b10629 HS	52
	53	if ( operation == CEPH_MDS_OP_GETFILELOCK){
	54	fl->fl_pid = le64_to_cpu(req->r_reply_info.filelock_reply->pid);
	55	if (CEPH_LOCK_SHARED == req->r_reply_info.filelock_reply->type)
	56	fl->fl_type = F_RDLCK;
	57	else if (CEPH_LOCK_EXCL == req->r_reply_info.filelock_reply->type)
	58	fl->fl_type = F_WRLCK;
	59	else
	60	fl->fl_type = F_UNLCK;
	61
	62	fl->fl_start = le64_to_cpu(req->r_reply_info.filelock_reply->start);
	63	length = le64_to_cpu(req->r_reply_info.filelock_reply->start) +
	64	le64_to_cpu(req->r_reply_info.filelock_reply->length);
	65	if (length >= 1)
	66	fl->fl_end = length -1;
	67	else
	68	fl->fl_end = 0;
	69
	70	}
40819f6f GF	71	ceph_mdsc_put_request(req);
40819f6f GF	72	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
0c1f91f2	73	"length: %llu, wait: %d, type: %d, err code %d", (int)lock_type,
637ae8d5 HS	74	(int)operation, (u64)fl->fl_pid, fl->fl_start,
637ae8d5 HS	75	length, wait, fl->fl_type, err);
40819f6f GF	76	return err;
	77	}
	78
	79	/**
	80	* Attempt to set an fcntl lock.
	81	* For now, this just goes away to the server. Later it may be more awesome.
	82	*/
	83	int ceph_lock(struct file file, int cmd, struct file_lock fl)
	84	{
40819f6f GF	85	u8 lock_cmd;
	86	int err;
	87	u8 wait = 0;
	88	u16 op = CEPH_MDS_OP_SETFILELOCK;
	89
	90	fl->fl_nspid = get_pid(task_tgid(current));
	91	dout("ceph_lock, fl_pid:%d", fl->fl_pid);
	92
	93	/* set wait bit as appropriate, then make command as Ceph expects it*/
	94	if (F_SETLKW == cmd)
	95	wait = 1;
	96	if (F_GETLK == cmd)
	97	op = CEPH_MDS_OP_GETFILELOCK;
	98
	99	if (F_RDLCK == fl->fl_type)
	100	lock_cmd = CEPH_LOCK_SHARED;
	101	else if (F_WRLCK == fl->fl_type)
	102	lock_cmd = CEPH_LOCK_EXCL;
	103	else
	104	lock_cmd = CEPH_LOCK_UNLOCK;
	105
637ae8d5	106	err = ceph_lock_message(CEPH_LOCK_FCNTL, op, file, lock_cmd, wait, fl);
40819f6f	107	if (!err) {
a5b10629 HS	108	if ( op != CEPH_MDS_OP_GETFILELOCK ){
	109	dout("mds locked, locking locally");
	110	err = posix_lock_file(file, fl, NULL);
	111	if (err && (CEPH_MDS_OP_SETFILELOCK == op)) {
0c1f91f2 SW	112	/* undo! This should only happen if
	113	* the kernel detects local
	114	* deadlock. */
a5b10629 HS	115	ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
a5b10629 HS	116	CEPH_LOCK_UNLOCK, 0, fl);
0c1f91f2 SW	117	dout("got %d on posix_lock_file, undid lock",
0c1f91f2 SW	118	err);
a5b10629	119	}
40819f6f	120	}
a5b10629	121
0c1f91f2 SW	122	} else if (err == -ERESTARTSYS) {
	123	dout("undoing lock\n");
	124	ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
	125	CEPH_LOCK_UNLOCK, 0, fl);
40819f6f GF	126	}
	127	return err;
	128	}
	129
	130	int ceph_flock(struct file file, int cmd, struct file_lock fl)
	131	{
40819f6f GF	132	u8 lock_cmd;
	133	int err;
	134	u8 wait = 1;
	135
	136	fl->fl_nspid = get_pid(task_tgid(current));
	137	dout("ceph_flock, fl_pid:%d", fl->fl_pid);
	138
	139	/* set wait bit, then clear it out of cmd*/
	140	if (cmd & LOCK_NB)
	141	wait = 0;
	142	cmd = cmd & (LOCK_SH \| LOCK_EX \| LOCK_UN);
	143	/* set command sequence that Ceph wants to see:
	144	shared lock, exclusive lock, or unlock */
	145	if (LOCK_SH == cmd)
	146	lock_cmd = CEPH_LOCK_SHARED;
	147	else if (LOCK_EX == cmd)
	148	lock_cmd = CEPH_LOCK_EXCL;
	149	else
	150	lock_cmd = CEPH_LOCK_UNLOCK;
40819f6f GF	151
40819f6f GF	152	err = ceph_lock_message(CEPH_LOCK_FLOCK, CEPH_MDS_OP_SETFILELOCK,
637ae8d5	153	file, lock_cmd, wait, fl);
40819f6f GF	154	if (!err) {
	155	err = flock_lock_file_wait(file, fl);
	156	if (err) {
	157	ceph_lock_message(CEPH_LOCK_FLOCK,
	158	CEPH_MDS_OP_SETFILELOCK,
637ae8d5	159	file, CEPH_LOCK_UNLOCK, 0, fl);
40819f6f GF	160	dout("got %d on flock_lock_file_wait, undid lock", err);
40819f6f GF	161	}
0c1f91f2 SW	162	} else if (err == -ERESTARTSYS) {
	163	dout("undoing lock\n");
	164	ceph_lock_message(CEPH_LOCK_FLOCK,
	165	CEPH_MDS_OP_SETFILELOCK,
	166	file, CEPH_LOCK_UNLOCK, 0, fl);
40819f6f GF	167	}
	168	return err;
	169	}
	170
	171	/**
	172	* Must be called with BKL already held. Fills in the passed
	173	* counter variables, so you can prepare pagelist metadata before calling
	174	* ceph_encode_locks.
	175	*/
	176	void ceph_count_locks(struct inode inode, int fcntl_count, int *flock_count)
	177	{
	178	struct file_lock *lock;
	179
	180	*fcntl_count = 0;
	181	*flock_count = 0;
	182
	183	for (lock = inode->i_flock; lock != NULL; lock = lock->fl_next) {
	184	if (lock->fl_flags & FL_POSIX)
	185	++(*fcntl_count);
	186	else if (lock->fl_flags & FL_FLOCK)
	187	++(*flock_count);
	188	}
	189	dout("counted %d flock locks and %d fcntl locks",
	190	flock_count, fcntl_count);
	191	}
	192
	193	/**
	194	* Encode the flock and fcntl locks for the given inode into the pagelist.
	195	* Format is: #fcntl locks, sequential fcntl locks, #flock locks,
	196	* sequential flock locks.
fca4451a GF	197	* Must be called with lock_flocks() already held.
	198	* If we encounter more of a specific lock type than expected,
	199	* we return the value 1.
40819f6f GF	200	*/
	201	int ceph_encode_locks(struct inode inode, struct ceph_pagelist pagelist,
	202	int num_fcntl_locks, int num_flock_locks)
	203	{
	204	struct file_lock *lock;
	205	struct ceph_filelock cephlock;
	206	int err = 0;
fca4451a GF	207	int seen_fcntl = 0;
fca4451a GF	208	int seen_flock = 0;
c420276a	209	__le32 nlocks;
40819f6f GF	210
	211	dout("encoding %d flock and %d fcntl locks", num_flock_locks,
	212	num_fcntl_locks);
c420276a JS	213	nlocks = cpu_to_le32(num_fcntl_locks);
c420276a JS	214	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
40819f6f GF	215	if (err)
	216	goto fail;
	217	for (lock = inode->i_flock; lock != NULL; lock = lock->fl_next) {
	218	if (lock->fl_flags & FL_POSIX) {
fca4451a GF	219	++seen_fcntl;
	220	if (seen_fcntl > num_fcntl_locks) {
	221	err = -ENOSPC;
	222	goto fail;
	223	}
40819f6f GF	224	err = lock_to_ceph_filelock(lock, &cephlock);
	225	if (err)
	226	goto fail;
	227	err = ceph_pagelist_append(pagelist, &cephlock,
	228	sizeof(struct ceph_filelock));
	229	}
	230	if (err)
	231	goto fail;
	232	}
	233
c420276a JS	234	nlocks = cpu_to_le32(num_flock_locks);
c420276a JS	235	err = ceph_pagelist_append(pagelist, &nlocks, sizeof(nlocks));
40819f6f GF	236	if (err)
	237	goto fail;
	238	for (lock = inode->i_flock; lock != NULL; lock = lock->fl_next) {
	239	if (lock->fl_flags & FL_FLOCK) {
fca4451a GF	240	++seen_flock;
	241	if (seen_flock > num_flock_locks) {
	242	err = -ENOSPC;
	243	goto fail;
	244	}
40819f6f GF	245	err = lock_to_ceph_filelock(lock, &cephlock);
	246	if (err)
	247	goto fail;
	248	err = ceph_pagelist_append(pagelist, &cephlock,
	249	sizeof(struct ceph_filelock));
	250	}
	251	if (err)
	252	goto fail;
	253	}
	254	fail:
	255	return err;
	256	}
	257
	258	/*
	259	* Given a pointer to a lock, convert it to a ceph filelock
	260	*/
	261	int lock_to_ceph_filelock(struct file_lock *lock,
	262	struct ceph_filelock *cephlock)
	263	{
	264	int err = 0;
	265
	266	cephlock->start = cpu_to_le64(lock->fl_start);
	267	cephlock->length = cpu_to_le64(lock->fl_end - lock->fl_start + 1);
	268	cephlock->client = cpu_to_le64(0);
	269	cephlock->pid = cpu_to_le64(lock->fl_pid);
ad8453ab AC	270	cephlock->pid_namespace =
ad8453ab AC	271	cpu_to_le64((u64)(unsigned long)lock->fl_nspid);
40819f6f GF	272
	273	switch (lock->fl_type) {
	274	case F_RDLCK:
	275	cephlock->type = CEPH_LOCK_SHARED;
	276	break;
	277	case F_WRLCK:
	278	cephlock->type = CEPH_LOCK_EXCL;
	279	break;
	280	case F_UNLCK:
	281	cephlock->type = CEPH_LOCK_UNLOCK;
	282	break;
	283	default:
	284	dout("Have unknown lock type %d", lock->fl_type);
	285	err = -EINVAL;
	286	}
	287
	288	return err;
	289	}