[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->f_dentry->d_inode;
	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 {
		dout("mds returned error code %d", err);
	}
	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 {
		dout("mds error code %d", err);
	}
	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;

	dout("encoding %d flock and %d fcntl locks", num_flock_locks,
	     num_fcntl_locks);
	err = ceph_pagelist_append(pagelist, &num_fcntl_locks, sizeof(u32));
	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;
	}

	err = ceph_pagelist_append(pagelist, &num_flock_locks, sizeof(u32));
	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 GF	15	{
	16	struct inode *inode = file->f_dentry->d_inode;
	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 GF	35	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
40819f6f GF	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
	41	req->r_args.filelock_change.rule = lock_type;
	42	req->r_args.filelock_change.type = cmd;
637ae8d5	43	req->r_args.filelock_change.pid = cpu_to_le64((u64)fl->fl_pid);
40819f6f GF	44	/* This should be adjusted, but I'm not sure if
	45	namespaces actually get id numbers*/
	46	req->r_args.filelock_change.pid_namespace =
637ae8d5 HS	47	cpu_to_le64((u64)(unsigned long)fl->fl_nspid);
637ae8d5 HS	48	req->r_args.filelock_change.start = cpu_to_le64(fl->fl_start);
40819f6f GF	49	req->r_args.filelock_change.length = cpu_to_le64(length);
	50	req->r_args.filelock_change.wait = wait;
	51
	52	err = ceph_mdsc_do_request(mdsc, inode, req);
a5b10629 HS	53
	54	if ( operation == CEPH_MDS_OP_GETFILELOCK){
	55	fl->fl_pid = le64_to_cpu(req->r_reply_info.filelock_reply->pid);
	56	if (CEPH_LOCK_SHARED == req->r_reply_info.filelock_reply->type)
	57	fl->fl_type = F_RDLCK;
	58	else if (CEPH_LOCK_EXCL == req->r_reply_info.filelock_reply->type)
	59	fl->fl_type = F_WRLCK;
	60	else
	61	fl->fl_type = F_UNLCK;
	62
	63	fl->fl_start = le64_to_cpu(req->r_reply_info.filelock_reply->start);
	64	length = le64_to_cpu(req->r_reply_info.filelock_reply->start) +
	65	le64_to_cpu(req->r_reply_info.filelock_reply->length);
	66	if (length >= 1)
	67	fl->fl_end = length -1;
	68	else
	69	fl->fl_end = 0;
	70
	71	}
40819f6f GF	72	ceph_mdsc_put_request(req);
40819f6f GF	73	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
637ae8d5 HS	74	"length: %llu, wait: %d, type`: %d, err code %d", (int)lock_type,
	75	(int)operation, (u64)fl->fl_pid, fl->fl_start,
	76	length, wait, fl->fl_type, err);
40819f6f GF	77	return err;
	78	}
	79
	80	/**
	81	* Attempt to set an fcntl lock.
	82	* For now, this just goes away to the server. Later it may be more awesome.
	83	*/
	84	int ceph_lock(struct file file, int cmd, struct file_lock fl)
	85	{
40819f6f GF	86	u8 lock_cmd;
	87	int err;
	88	u8 wait = 0;
	89	u16 op = CEPH_MDS_OP_SETFILELOCK;
	90
	91	fl->fl_nspid = get_pid(task_tgid(current));
	92	dout("ceph_lock, fl_pid:%d", fl->fl_pid);
	93
	94	/* set wait bit as appropriate, then make command as Ceph expects it*/
	95	if (F_SETLKW == cmd)
	96	wait = 1;
	97	if (F_GETLK == cmd)
	98	op = CEPH_MDS_OP_GETFILELOCK;
	99
	100	if (F_RDLCK == fl->fl_type)
	101	lock_cmd = CEPH_LOCK_SHARED;
	102	else if (F_WRLCK == fl->fl_type)
	103	lock_cmd = CEPH_LOCK_EXCL;
	104	else
	105	lock_cmd = CEPH_LOCK_UNLOCK;
	106
637ae8d5	107	err = ceph_lock_message(CEPH_LOCK_FCNTL, op, file, lock_cmd, wait, fl);
40819f6f	108	if (!err) {
a5b10629 HS	109	if ( op != CEPH_MDS_OP_GETFILELOCK ){
	110	dout("mds locked, locking locally");
	111	err = posix_lock_file(file, fl, NULL);
	112	if (err && (CEPH_MDS_OP_SETFILELOCK == op)) {
	113	/* undo! This should only happen if the kernel detects
	114	* local deadlock. */
	115	ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
	116	CEPH_LOCK_UNLOCK, 0, fl);
	117	dout("got %d on posix_lock_file, undid lock", err);
	118	}
40819f6f	119	}
a5b10629	120
40819f6f GF	121	} else {
	122	dout("mds returned error code %d", err);
	123	}
	124	return err;
	125	}
	126
	127	int ceph_flock(struct file file, int cmd, struct file_lock fl)
	128	{
40819f6f GF	129	u8 lock_cmd;
	130	int err;
	131	u8 wait = 1;
	132
	133	fl->fl_nspid = get_pid(task_tgid(current));
	134	dout("ceph_flock, fl_pid:%d", fl->fl_pid);
	135
	136	/* set wait bit, then clear it out of cmd*/
	137	if (cmd & LOCK_NB)
	138	wait = 0;
	139	cmd = cmd & (LOCK_SH \| LOCK_EX \| LOCK_UN);
	140	/* set command sequence that Ceph wants to see:
	141	shared lock, exclusive lock, or unlock */
	142	if (LOCK_SH == cmd)
	143	lock_cmd = CEPH_LOCK_SHARED;
	144	else if (LOCK_EX == cmd)
	145	lock_cmd = CEPH_LOCK_EXCL;
	146	else
	147	lock_cmd = CEPH_LOCK_UNLOCK;
40819f6f GF	148
40819f6f GF	149	err = ceph_lock_message(CEPH_LOCK_FLOCK, CEPH_MDS_OP_SETFILELOCK,
637ae8d5	150	file, lock_cmd, wait, fl);
40819f6f GF	151	if (!err) {
	152	err = flock_lock_file_wait(file, fl);
	153	if (err) {
	154	ceph_lock_message(CEPH_LOCK_FLOCK,
	155	CEPH_MDS_OP_SETFILELOCK,
637ae8d5	156	file, CEPH_LOCK_UNLOCK, 0, fl);
40819f6f GF	157	dout("got %d on flock_lock_file_wait, undid lock", err);
	158	}
	159	} else {
	160	dout("mds error code %d", err);
	161	}
	162	return err;
	163	}
	164
	165	/**
	166	* Must be called with BKL already held. Fills in the passed
	167	* counter variables, so you can prepare pagelist metadata before calling
	168	* ceph_encode_locks.
	169	*/
	170	void ceph_count_locks(struct inode inode, int fcntl_count, int *flock_count)
	171	{
	172	struct file_lock *lock;
	173
	174	*fcntl_count = 0;
	175	*flock_count = 0;
	176
	177	for (lock = inode->i_flock; lock != NULL; lock = lock->fl_next) {
	178	if (lock->fl_flags & FL_POSIX)
	179	++(*fcntl_count);
	180	else if (lock->fl_flags & FL_FLOCK)
	181	++(*flock_count);
	182	}
	183	dout("counted %d flock locks and %d fcntl locks",
	184	flock_count, fcntl_count);
	185	}
	186
	187	/**
	188	* Encode the flock and fcntl locks for the given inode into the pagelist.
	189	* Format is: #fcntl locks, sequential fcntl locks, #flock locks,
	190	* sequential flock locks.
fca4451a GF	191	* Must be called with lock_flocks() already held.
	192	* If we encounter more of a specific lock type than expected,
	193	* we return the value 1.
40819f6f GF	194	*/
	195	int ceph_encode_locks(struct inode inode, struct ceph_pagelist pagelist,
	196	int num_fcntl_locks, int num_flock_locks)
	197	{
	198	struct file_lock *lock;
	199	struct ceph_filelock cephlock;
	200	int err = 0;
fca4451a GF	201	int seen_fcntl = 0;
fca4451a GF	202	int seen_flock = 0;
40819f6f GF	203
	204	dout("encoding %d flock and %d fcntl locks", num_flock_locks,
	205	num_fcntl_locks);
	206	err = ceph_pagelist_append(pagelist, &num_fcntl_locks, sizeof(u32));
	207	if (err)
	208	goto fail;
	209	for (lock = inode->i_flock; lock != NULL; lock = lock->fl_next) {
	210	if (lock->fl_flags & FL_POSIX) {
fca4451a GF	211	++seen_fcntl;
	212	if (seen_fcntl > num_fcntl_locks) {
	213	err = -ENOSPC;
	214	goto fail;
	215	}
40819f6f GF	216	err = lock_to_ceph_filelock(lock, &cephlock);
	217	if (err)
	218	goto fail;
	219	err = ceph_pagelist_append(pagelist, &cephlock,
	220	sizeof(struct ceph_filelock));
	221	}
	222	if (err)
	223	goto fail;
	224	}
	225
	226	err = ceph_pagelist_append(pagelist, &num_flock_locks, sizeof(u32));
	227	if (err)
	228	goto fail;
	229	for (lock = inode->i_flock; lock != NULL; lock = lock->fl_next) {
	230	if (lock->fl_flags & FL_FLOCK) {
fca4451a GF	231	++seen_flock;
	232	if (seen_flock > num_flock_locks) {
	233	err = -ENOSPC;
	234	goto fail;
	235	}
40819f6f GF	236	err = lock_to_ceph_filelock(lock, &cephlock);
	237	if (err)
	238	goto fail;
	239	err = ceph_pagelist_append(pagelist, &cephlock,
	240	sizeof(struct ceph_filelock));
	241	}
	242	if (err)
	243	goto fail;
	244	}
	245	fail:
	246	return err;
	247	}
	248
	249	/*
	250	* Given a pointer to a lock, convert it to a ceph filelock
	251	*/
	252	int lock_to_ceph_filelock(struct file_lock *lock,
	253	struct ceph_filelock *cephlock)
	254	{
	255	int err = 0;
	256
	257	cephlock->start = cpu_to_le64(lock->fl_start);
	258	cephlock->length = cpu_to_le64(lock->fl_end - lock->fl_start + 1);
	259	cephlock->client = cpu_to_le64(0);
	260	cephlock->pid = cpu_to_le64(lock->fl_pid);
ad8453ab AC	261	cephlock->pid_namespace =
ad8453ab AC	262	cpu_to_le64((u64)(unsigned long)lock->fl_nspid);
40819f6f GF	263
	264	switch (lock->fl_type) {
	265	case F_RDLCK:
	266	cephlock->type = CEPH_LOCK_SHARED;
	267	break;
	268	case F_WRLCK:
	269	cephlock->type = CEPH_LOCK_EXCL;
	270	break;
	271	case F_UNLCK:
	272	cephlock->type = CEPH_LOCK_UNLOCK;
	273	break;
	274	default:
	275	dout("Have unknown lock type %d", lock->fl_type);
	276	err = -EINVAL;
	277	}
	278
	279	return err;
	280	}