2 * Copyright (c) 2007-2008 The DragonFly Project. All rights reserved.
4 * This code is derived from software contributed to The DragonFly Project
5 * by Matthew Dillon <dillon@backplane.com>
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in
15 * the documentation and/or other materials provided with the
17 * 3. Neither the name of The DragonFly Project nor the names of its
18 * contributors may be used to endorse or promote products derived
19 * from this software without specific, prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
24 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
25 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
26 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
27 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
28 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
29 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
30 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
31 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * $DragonFly: src/sys/vfs/hammer/hammer_subs.c,v 1.28 2008/06/23 21:42:48 dillon Exp $
37 * HAMMER structural locking
41 #include <sys/dirent.h>
44 hammer_lock_ex_ident(struct hammer_lock *lock, const char *ident)
46 thread_t td = curthread;
48 KKASSERT(lock->refs > 0);
50 if (lock->locktd != td) {
51 while (lock->locktd != NULL || lock->lockcount) {
54 if (hammer_debug_locks) {
55 kprintf("hammer_lock_ex: held by %p\n",
58 ++hammer_contention_count;
59 tsleep(lock, 0, ident, 0);
60 if (hammer_debug_locks)
61 kprintf("hammer_lock_ex: try again\n");
66 KKASSERT(lock->lockcount >= 0);
72 * Try to obtain an exclusive lock
75 hammer_lock_ex_try(struct hammer_lock *lock)
77 thread_t td = curthread;
79 KKASSERT(lock->refs > 0);
81 if (lock->locktd != td) {
82 if (lock->locktd != NULL || lock->lockcount) {
88 KKASSERT(lock->lockcount >= 0);
95 * Obtain a shared lock
98 hammer_lock_sh(struct hammer_lock *lock)
100 KKASSERT(lock->refs > 0);
102 while (lock->locktd != NULL) {
103 if (lock->locktd == curthread) {
104 Debugger("hammer_lock_sh: lock_sh on exclusive");
110 tsleep(lock, 0, "hmrlck", 0);
112 KKASSERT(lock->lockcount <= 0);
118 * Obtain a shared lock at a lower priority then thread waiting for an
119 * exclusive lock. To avoid a deadlock this may only be done if no other
120 * shared locks are being held by the caller.
123 hammer_lock_sh_lowpri(struct hammer_lock *lock)
125 KKASSERT(lock->refs > 0);
127 while (lock->locktd != NULL || lock->exwanted) {
128 if (lock->locktd == curthread) {
129 Debugger("hammer_lock_sh: lock_sh on exclusive");
135 tsleep(lock, 0, "hmrlck", 0);
137 KKASSERT(lock->lockcount <= 0);
143 hammer_lock_sh_try(struct hammer_lock *lock)
145 KKASSERT(lock->refs > 0);
151 KKASSERT(lock->lockcount <= 0);
158 * Upgrade a shared lock to an exclusively held lock. This function will
159 * return EDEADLK If there is more then one shared holder.
161 * No error occurs and no action is taken if the lock is already exclusively
162 * held by the caller. If the lock is not held at all or held exclusively
163 * by someone else, this function will panic.
166 hammer_lock_upgrade(struct hammer_lock *lock)
171 if (lock->lockcount > 0) {
172 if (lock->locktd != curthread)
173 panic("hammer_lock_upgrade: illegal lock state");
175 } else if (lock->lockcount == -1) {
177 lock->locktd = curthread;
179 } else if (lock->lockcount != 0) {
182 panic("hammer_lock_upgrade: lock is not held");
191 * Downgrade an exclusively held lock to a shared lock.
194 hammer_lock_downgrade(struct hammer_lock *lock)
196 KKASSERT(lock->lockcount == 1 && lock->locktd == curthread);
198 lock->lockcount = -1;
205 /* XXX memory barrier */
209 hammer_unlock(struct hammer_lock *lock)
212 KKASSERT(lock->lockcount != 0);
213 if (lock->lockcount < 0) {
214 if (++lock->lockcount == 0 && lock->wanted) {
219 KKASSERT(lock->locktd == curthread);
220 if (--lock->lockcount == 0) {
233 hammer_ref(struct hammer_lock *lock)
235 KKASSERT(lock->refs >= 0);
242 hammer_unref(struct hammer_lock *lock)
244 KKASSERT(lock->refs > 0);
251 * The sync_lock must be held when doing any modifying operations on
252 * meta-data. The flusher holds the lock exclusively while the reblocker
253 * and pruner use a shared lock.
255 * Modifying operations can run in parallel until the flusher needs to
256 * sync the disk media.
259 hammer_sync_lock_ex(hammer_transaction_t trans)
261 ++trans->sync_lock_refs;
262 hammer_lock_ex(&trans->hmp->sync_lock);
266 hammer_sync_lock_sh(hammer_transaction_t trans)
268 ++trans->sync_lock_refs;
269 hammer_lock_sh(&trans->hmp->sync_lock);
273 hammer_sync_lock_sh_try(hammer_transaction_t trans)
277 ++trans->sync_lock_refs;
278 if ((error = hammer_lock_sh_try(&trans->hmp->sync_lock)) != 0)
279 --trans->sync_lock_refs;
284 hammer_sync_unlock(hammer_transaction_t trans)
286 --trans->sync_lock_refs;
287 hammer_unlock(&trans->hmp->sync_lock);
294 hammer_to_unix_xid(uuid_t *uuid)
296 return(*(u_int32_t *)&uuid->node[2]);
300 hammer_guid_to_uuid(uuid_t *uuid, u_int32_t guid)
302 bzero(uuid, sizeof(*uuid));
303 *(u_int32_t *)&uuid->node[2] = guid;
307 hammer_time_to_timespec(u_int64_t xtime, struct timespec *ts)
309 ts->tv_sec = (unsigned long)(xtime / 1000000);
310 ts->tv_nsec = (unsigned int)(xtime % 1000000) * 1000L;
314 hammer_timespec_to_time(struct timespec *ts)
318 xtime = (unsigned)(ts->tv_nsec / 1000) +
319 (unsigned long)ts->tv_sec * 1000000ULL;
325 * Convert a HAMMER filesystem object type to a vnode type
328 hammer_get_vnode_type(u_int8_t obj_type)
331 case HAMMER_OBJTYPE_DIRECTORY:
333 case HAMMER_OBJTYPE_REGFILE:
335 case HAMMER_OBJTYPE_DBFILE:
337 case HAMMER_OBJTYPE_FIFO:
339 case HAMMER_OBJTYPE_CDEV:
341 case HAMMER_OBJTYPE_BDEV:
343 case HAMMER_OBJTYPE_SOFTLINK:
352 hammer_get_dtype(u_int8_t obj_type)
355 case HAMMER_OBJTYPE_DIRECTORY:
357 case HAMMER_OBJTYPE_REGFILE:
359 case HAMMER_OBJTYPE_DBFILE:
361 case HAMMER_OBJTYPE_FIFO:
363 case HAMMER_OBJTYPE_CDEV:
365 case HAMMER_OBJTYPE_BDEV:
367 case HAMMER_OBJTYPE_SOFTLINK:
376 hammer_get_obj_type(enum vtype vtype)
380 return(HAMMER_OBJTYPE_DIRECTORY);
382 return(HAMMER_OBJTYPE_REGFILE);
384 return(HAMMER_OBJTYPE_DBFILE);
386 return(HAMMER_OBJTYPE_FIFO);
388 return(HAMMER_OBJTYPE_CDEV);
390 return(HAMMER_OBJTYPE_BDEV);
392 return(HAMMER_OBJTYPE_SOFTLINK);
394 return(HAMMER_OBJTYPE_UNKNOWN);
400 hammer_nohistory(hammer_inode_t ip)
402 if (ip->hmp->hflags & HMNT_NOHISTORY)
404 if (ip->ino_data.uflags & (SF_NOHISTORY|UF_NOHISTORY))
410 * Return a namekey hash. The 64 bit namekey hash consists of a 32 bit
411 * crc in the MSB and 0 in the LSB. The caller will use the low bits to
412 * generate a unique key and will scan all entries with the same upper
413 * 32 bits when issuing a lookup.
415 * We strip bit 63 in order to provide a positive key, this way a seek
416 * offset of 0 will represent the base of the directory.
418 * This function can never return 0. We use the MSB-0 space to synthesize
419 * artificial directory entries such as "." and "..".
422 hammer_directory_namekey(const void *name, int len)
426 key = (int64_t)(crc32(name, len) & 0x7FFFFFFF) << 32;
428 key |= 0x100000000LL;
433 hammer_str_to_tid(const char *str)
437 tid = strtouq(str, NULL, 0); /* full TID */
442 hammer_crc_set_blockmap(hammer_blockmap_t blockmap)
444 blockmap->entry_crc = crc32(blockmap, HAMMER_BLOCKMAP_CRCSIZE);
448 hammer_crc_set_volume(hammer_volume_ondisk_t ondisk)
450 ondisk->vol_crc = crc32(ondisk, HAMMER_VOL_CRCSIZE1) ^
451 crc32(&ondisk->vol_crc + 1, HAMMER_VOL_CRCSIZE2);
455 hammer_crc_test_blockmap(hammer_blockmap_t blockmap)
459 crc = crc32(blockmap, HAMMER_BLOCKMAP_CRCSIZE);
460 return (blockmap->entry_crc == crc);
464 hammer_crc_test_volume(hammer_volume_ondisk_t ondisk)
468 crc = crc32(ondisk, HAMMER_VOL_CRCSIZE1) ^
469 crc32(&ondisk->vol_crc + 1, HAMMER_VOL_CRCSIZE2);
470 return (ondisk->vol_crc == crc);
474 hammer_crc_test_btree(hammer_node_ondisk_t ondisk)
478 crc = crc32(&ondisk->crc + 1, HAMMER_BTREE_CRCSIZE);
479 return (ondisk->crc == crc);
483 * Test or set the leaf->data_crc field. Deal with any special cases given
484 * a generic B-Tree leaf element and its data.
486 * NOTE: Inode-data: the atime and mtime fields are not CRCd, allowing them
487 * to be updated in-place.
490 hammer_crc_test_leaf(void *data, hammer_btree_leaf_elm_t leaf)
494 if (leaf->data_len == 0) {
497 switch(leaf->base.rec_type) {
498 case HAMMER_RECTYPE_INODE:
499 if (leaf->data_len != sizeof(struct hammer_inode_data))
501 crc = crc32(data, HAMMER_INODE_CRCSIZE);
504 crc = crc32(data, leaf->data_len);
508 return (leaf->data_crc == crc);
512 hammer_crc_set_leaf(void *data, hammer_btree_leaf_elm_t leaf)
514 if (leaf->data_len == 0) {
517 switch(leaf->base.rec_type) {
518 case HAMMER_RECTYPE_INODE:
519 KKASSERT(leaf->data_len ==
520 sizeof(struct hammer_inode_data));
521 leaf->data_crc = crc32(data, HAMMER_INODE_CRCSIZE);
524 leaf->data_crc = crc32(data, leaf->data_len);
531 hkprintf(const char *ctl, ...)
535 if (hammer_debug_debug) {
543 * Return the block size at the specified file offset.
546 hammer_blocksize(int64_t file_offset)
548 if (file_offset < HAMMER_XDEMARC)
549 return(HAMMER_BUFSIZE);
551 return(HAMMER_XBUFSIZE);
555 * Return the demarkation point between the two offsets where
556 * the block size changes.
559 hammer_blockdemarc(int64_t file_offset1, int64_t file_offset2)
561 if (file_offset1 < HAMMER_XDEMARC) {
562 if (file_offset2 <= HAMMER_XDEMARC)
563 return(file_offset2);
564 return(HAMMER_XDEMARC);
566 panic("hammer_blockdemarc: illegal range %lld %lld\n",
567 file_offset1, file_offset2);