locks.c 8.07 KB
Newer Older
1
#include <linux/ceph/ceph_debug.h>
2 3 4 5 6 7

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

#include "super.h"
#include "mds_client.h"
8
#include <linux/ceph/pagelist.h>
9 10 11 12 13

/**
 * Implement fcntl and flock locking functions.
 */
static int ceph_lock_message(u8 lock_type, u16 operation, struct file *file,
14
			     int cmd, u8 wait, struct file_lock *fl)
15
{
Al Viro's avatar
Al Viro committed
16
	struct inode *inode = file_inode(file);
17
	struct ceph_mds_client *mdsc =
18
		ceph_sb_to_client(inode->i_sb)->mdsc;
19 20
	struct ceph_mds_request *req;
	int err;
21
	u64 length = 0;
22 23 24 25

	req = ceph_mdsc_create_request(mdsc, operation, USE_AUTH_MDS);
	if (IS_ERR(req))
		return PTR_ERR(req);
26 27
	req->r_inode = inode;
	ihold(inode);
28

29 30 31 32 33 34
	/* 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;

35
	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
36
	     "length: %llu, wait: %d, type: %d", (int)lock_type,
37 38 39
	     (int)operation, (u64)fl->fl_pid, fl->fl_start,
	     length, wait, fl->fl_type);

40 41
	req->r_args.filelock_change.rule = lock_type;
	req->r_args.filelock_change.type = cmd;
42
	req->r_args.filelock_change.pid = cpu_to_le64((u64)fl->fl_pid);
43 44 45
	/* This should be adjusted, but I'm not sure if
	   namespaces actually get id numbers*/
	req->r_args.filelock_change.pid_namespace =
46 47
		cpu_to_le64((u64)(unsigned long)fl->fl_nspid);
	req->r_args.filelock_change.start = cpu_to_le64(fl->fl_start);
48 49 50 51
	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);
52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70

	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;

	}
71 72
	ceph_mdsc_put_request(req);
	dout("ceph_lock_message: rule: %d, op: %d, pid: %llu, start: %llu, "
73
	     "length: %llu, wait: %d, type: %d, err code %d", (int)lock_type,
74 75
	     (int)operation, (u64)fl->fl_pid, fl->fl_start,
	     length, wait, fl->fl_type, err);
76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105
	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;

106
	err = ceph_lock_message(CEPH_LOCK_FCNTL, op, file, lock_cmd, wait, fl);
107
	if (!err) {
108 109 110 111
		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)) {
112 113 114
				/* undo! This should only happen if
				 * the kernel detects local
				 * deadlock. */
115 116
				ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
						  CEPH_LOCK_UNLOCK, 0, fl);
117 118
				dout("got %d on posix_lock_file, undid lock",
				     err);
119
			}
120
		}
121

122 123 124 125
	} else if (err == -ERESTARTSYS) {
		dout("undoing lock\n");
		ceph_lock_message(CEPH_LOCK_FCNTL, op, file,
				  CEPH_LOCK_UNLOCK, 0, fl);
126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152
	}
	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,
153
				file, lock_cmd, wait, fl);
154 155 156 157 158
	if (!err) {
		err = flock_lock_file_wait(file, fl);
		if (err) {
			ceph_lock_message(CEPH_LOCK_FLOCK,
					  CEPH_MDS_OP_SETFILELOCK,
159
					  file, CEPH_LOCK_UNLOCK, 0, fl);
160 161
			dout("got %d on flock_lock_file_wait, undid lock", err);
		}
162 163 164 165 166
	} else if (err == -ERESTARTSYS) {
		dout("undoing lock\n");
		ceph_lock_message(CEPH_LOCK_FLOCK,
				  CEPH_MDS_OP_SETFILELOCK,
				  file, CEPH_LOCK_UNLOCK, 0, fl);
167 168 169 170 171
	}
	return err;
}

/**
172
 * Must be called with lock_flocks() already held. Fills in the passed
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193
 * 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);
}

/**
194
 * Encode the flock and fcntl locks for the given inode into the ceph_filelock
195
 * array. Must be called with inode->i_lock already held.
196
 * If we encounter more of a specific lock type than expected, return -ENOSPC.
197
 */
198 199 200
int ceph_encode_locks_to_buffer(struct inode *inode,
				struct ceph_filelock *flocks,
				int num_fcntl_locks, int num_flock_locks)
201 202 203
{
	struct file_lock *lock;
	int err = 0;
204 205
	int seen_fcntl = 0;
	int seen_flock = 0;
206
	int l = 0;
207 208 209

	dout("encoding %d flock and %d fcntl locks", num_flock_locks,
	     num_fcntl_locks);
210

211 212
	for (lock = inode->i_flock; lock != NULL; lock = lock->fl_next) {
		if (lock->fl_flags & FL_POSIX) {
213 214 215 216 217
			++seen_fcntl;
			if (seen_fcntl > num_fcntl_locks) {
				err = -ENOSPC;
				goto fail;
			}
218
			err = lock_to_ceph_filelock(lock, &flocks[l]);
219 220
			if (err)
				goto fail;
221
			++l;
222 223 224 225
		}
	}
	for (lock = inode->i_flock; lock != NULL; lock = lock->fl_next) {
		if (lock->fl_flags & FL_FLOCK) {
226 227 228 229 230
			++seen_flock;
			if (seen_flock > num_flock_locks) {
				err = -ENOSPC;
				goto fail;
			}
231
			err = lock_to_ceph_filelock(lock, &flocks[l]);
232 233
			if (err)
				goto fail;
234
			++l;
235 236 237 238 239 240
		}
	}
fail:
	return err;
}

241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275
/**
 * 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;
}

276 277 278 279 280 281 282 283 284 285 286 287
/*
 * 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);
Alan Cox's avatar
Alan Cox committed
288 289
	cephlock->pid_namespace =
	        cpu_to_le64((u64)(unsigned long)lock->fl_nspid);
290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307

	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;
}