]> git.proxmox.com Git - mirror_zfs.git/blob - module/zfs/sa.c
Update to onnv_147
[mirror_zfs.git] / module / zfs / sa.c
1 /*
2 * CDDL HEADER START
3 *
4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
7 *
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
12 *
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 *
19 * CDDL HEADER END
20 */
21 /*
22 * Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
23 */
24
25 #include <sys/zfs_context.h>
26 #include <sys/types.h>
27 #include <sys/param.h>
28 #include <sys/systm.h>
29 #include <sys/sysmacros.h>
30 #include <sys/dmu.h>
31 #include <sys/dmu_impl.h>
32 #include <sys/dmu_objset.h>
33 #include <sys/dbuf.h>
34 #include <sys/dnode.h>
35 #include <sys/zap.h>
36 #include <sys/sa.h>
37 #include <sys/sunddi.h>
38 #include <sys/sa_impl.h>
39 #include <sys/dnode.h>
40 #include <sys/errno.h>
41 #include <sys/zfs_context.h>
42
43 /*
44 * ZFS System attributes:
45 *
46 * A generic mechanism to allow for arbitrary attributes
47 * to be stored in a dnode. The data will be stored in the bonus buffer of
48 * the dnode and if necessary a special "spill" block will be used to handle
49 * overflow situations. The spill block will be sized to fit the data
50 * from 512 - 128K. When a spill block is used the BP (blkptr_t) for the
51 * spill block is stored at the end of the current bonus buffer. Any
52 * attributes that would be in the way of the blkptr_t will be relocated
53 * into the spill block.
54 *
55 * Attribute registration:
56 *
57 * Stored persistently on a per dataset basis
58 * a mapping between attribute "string" names and their actual attribute
59 * numeric values, length, and byteswap function. The names are only used
60 * during registration. All attributes are known by their unique attribute
61 * id value. If an attribute can have a variable size then the value
62 * 0 will be used to indicate this.
63 *
64 * Attribute Layout:
65 *
66 * Attribute layouts are a way to compactly store multiple attributes, but
67 * without taking the overhead associated with managing each attribute
68 * individually. Since you will typically have the same set of attributes
69 * stored in the same order a single table will be used to represent that
70 * layout. The ZPL for example will usually have only about 10 different
71 * layouts (regular files, device files, symlinks,
72 * regular files + scanstamp, files/dir with extended attributes, and then
73 * you have the possibility of all of those minus ACL, because it would
74 * be kicked out into the spill block)
75 *
76 * Layouts are simply an array of the attributes and their
77 * ordering i.e. [0, 1, 4, 5, 2]
78 *
79 * Each distinct layout is given a unique layout number and that is whats
80 * stored in the header at the beginning of the SA data buffer.
81 *
82 * A layout only covers a single dbuf (bonus or spill). If a set of
83 * attributes is split up between the bonus buffer and a spill buffer then
84 * two different layouts will be used. This allows us to byteswap the
85 * spill without looking at the bonus buffer and keeps the on disk format of
86 * the bonus and spill buffer the same.
87 *
88 * Adding a single attribute will cause the entire set of attributes to
89 * be rewritten and could result in a new layout number being constructed
90 * as part of the rewrite if no such layout exists for the new set of
91 * attribues. The new attribute will be appended to the end of the already
92 * existing attributes.
93 *
94 * Both the attribute registration and attribute layout information are
95 * stored in normal ZAP attributes. Their should be a small number of
96 * known layouts and the set of attributes is assumed to typically be quite
97 * small.
98 *
99 * The registered attributes and layout "table" information is maintained
100 * in core and a special "sa_os_t" is attached to the objset_t.
101 *
102 * A special interface is provided to allow for quickly applying
103 * a large set of attributes at once. sa_replace_all_by_template() is
104 * used to set an array of attributes. This is used by the ZPL when
105 * creating a brand new file. The template that is passed into the function
106 * specifies the attribute, size for variable length attributes, location of
107 * data and special "data locator" function if the data isn't in a contiguous
108 * location.
109 *
110 * Byteswap implications:
111 * Since the SA attributes are not entirely self describing we can't do
112 * the normal byteswap processing. The special ZAP layout attribute and
113 * attribute registration attributes define the byteswap function and the
114 * size of the attributes, unless it is variable sized.
115 * The normal ZFS byteswapping infrastructure assumes you don't need
116 * to read any objects in order to do the necessary byteswapping. Whereas
117 * SA attributes can only be properly byteswapped if the dataset is opened
118 * and the layout/attribute ZAP attributes are available. Because of this
119 * the SA attributes will be byteswapped when they are first accessed by
120 * the SA code that will read the SA data.
121 */
122
123 typedef void (sa_iterfunc_t)(void *hdr, void *addr, sa_attr_type_t,
124 uint16_t length, int length_idx, boolean_t, void *userp);
125
126 static int sa_build_index(sa_handle_t *hdl, sa_buf_type_t buftype);
127 static void sa_idx_tab_hold(objset_t *os, sa_idx_tab_t *idx_tab);
128 static void *sa_find_idx_tab(objset_t *os, dmu_object_type_t bonustype,
129 void *data);
130 static void sa_idx_tab_rele(objset_t *os, void *arg);
131 static void sa_copy_data(sa_data_locator_t *func, void *start, void *target,
132 int buflen);
133 static int sa_modify_attrs(sa_handle_t *hdl, sa_attr_type_t newattr,
134 sa_data_op_t action, sa_data_locator_t *locator, void *datastart,
135 uint16_t buflen, dmu_tx_t *tx);
136
137 arc_byteswap_func_t *sa_bswap_table[] = {
138 byteswap_uint64_array,
139 byteswap_uint32_array,
140 byteswap_uint16_array,
141 byteswap_uint8_array,
142 zfs_acl_byteswap,
143 };
144
145 #define SA_COPY_DATA(f, s, t, l) \
146 { \
147 if (f == NULL) { \
148 if (l == 8) { \
149 *(uint64_t *)t = *(uint64_t *)s; \
150 } else if (l == 16) { \
151 *(uint64_t *)t = *(uint64_t *)s; \
152 *(uint64_t *)((uintptr_t)t + 8) = \
153 *(uint64_t *)((uintptr_t)s + 8); \
154 } else { \
155 bcopy(s, t, l); \
156 } \
157 } else \
158 sa_copy_data(f, s, t, l); \
159 }
160
161 /*
162 * This table is fixed and cannot be changed. Its purpose is to
163 * allow the SA code to work with both old/new ZPL file systems.
164 * It contains the list of legacy attributes. These attributes aren't
165 * stored in the "attribute" registry zap objects, since older ZPL file systems
166 * won't have the registry. Only objsets of type ZFS_TYPE_FILESYSTEM will
167 * use this static table.
168 */
169 sa_attr_reg_t sa_legacy_attrs[] = {
170 {"ZPL_ATIME", sizeof (uint64_t) * 2, SA_UINT64_ARRAY, 0},
171 {"ZPL_MTIME", sizeof (uint64_t) * 2, SA_UINT64_ARRAY, 1},
172 {"ZPL_CTIME", sizeof (uint64_t) * 2, SA_UINT64_ARRAY, 2},
173 {"ZPL_CRTIME", sizeof (uint64_t) * 2, SA_UINT64_ARRAY, 3},
174 {"ZPL_GEN", sizeof (uint64_t), SA_UINT64_ARRAY, 4},
175 {"ZPL_MODE", sizeof (uint64_t), SA_UINT64_ARRAY, 5},
176 {"ZPL_SIZE", sizeof (uint64_t), SA_UINT64_ARRAY, 6},
177 {"ZPL_PARENT", sizeof (uint64_t), SA_UINT64_ARRAY, 7},
178 {"ZPL_LINKS", sizeof (uint64_t), SA_UINT64_ARRAY, 8},
179 {"ZPL_XATTR", sizeof (uint64_t), SA_UINT64_ARRAY, 9},
180 {"ZPL_RDEV", sizeof (uint64_t), SA_UINT64_ARRAY, 10},
181 {"ZPL_FLAGS", sizeof (uint64_t), SA_UINT64_ARRAY, 11},
182 {"ZPL_UID", sizeof (uint64_t), SA_UINT64_ARRAY, 12},
183 {"ZPL_GID", sizeof (uint64_t), SA_UINT64_ARRAY, 13},
184 {"ZPL_PAD", sizeof (uint64_t) * 4, SA_UINT64_ARRAY, 14},
185 {"ZPL_ZNODE_ACL", 88, SA_UINT8_ARRAY, 15},
186 };
187
188 /*
189 * ZPL legacy layout
190 * This is only used for objects of type DMU_OT_ZNODE
191 */
192 sa_attr_type_t sa_legacy_zpl_layout[] = {
193 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
194 };
195
196 /*
197 * Special dummy layout used for buffers with no attributes.
198 */
199
200 sa_attr_type_t sa_dummy_zpl_layout[] = { 0 };
201
202 static int sa_legacy_attr_count = 16;
203 static kmem_cache_t *sa_cache = NULL;
204
205 /*ARGSUSED*/
206 static int
207 sa_cache_constructor(void *buf, void *unused, int kmflag)
208 {
209 sa_handle_t *hdl = buf;
210
211 hdl->sa_bonus_tab = NULL;
212 hdl->sa_spill_tab = NULL;
213 hdl->sa_os = NULL;
214 hdl->sa_userp = NULL;
215 hdl->sa_bonus = NULL;
216 hdl->sa_spill = NULL;
217 mutex_init(&hdl->sa_lock, NULL, MUTEX_DEFAULT, NULL);
218 return (0);
219 }
220
221 /*ARGSUSED*/
222 static void
223 sa_cache_destructor(void *buf, void *unused)
224 {
225 sa_handle_t *hdl = buf;
226 mutex_destroy(&hdl->sa_lock);
227 }
228
229 void
230 sa_cache_init(void)
231 {
232 sa_cache = kmem_cache_create("sa_cache",
233 sizeof (sa_handle_t), 0, sa_cache_constructor,
234 sa_cache_destructor, NULL, NULL, NULL, 0);
235 }
236
237 void
238 sa_cache_fini(void)
239 {
240 if (sa_cache)
241 kmem_cache_destroy(sa_cache);
242 }
243
244 static int
245 layout_num_compare(const void *arg1, const void *arg2)
246 {
247 const sa_lot_t *node1 = arg1;
248 const sa_lot_t *node2 = arg2;
249
250 if (node1->lot_num > node2->lot_num)
251 return (1);
252 else if (node1->lot_num < node2->lot_num)
253 return (-1);
254 return (0);
255 }
256
257 static int
258 layout_hash_compare(const void *arg1, const void *arg2)
259 {
260 const sa_lot_t *node1 = arg1;
261 const sa_lot_t *node2 = arg2;
262
263 if (node1->lot_hash > node2->lot_hash)
264 return (1);
265 if (node1->lot_hash < node2->lot_hash)
266 return (-1);
267 if (node1->lot_instance > node2->lot_instance)
268 return (1);
269 if (node1->lot_instance < node2->lot_instance)
270 return (-1);
271 return (0);
272 }
273
274 boolean_t
275 sa_layout_equal(sa_lot_t *tbf, sa_attr_type_t *attrs, int count)
276 {
277 int i;
278
279 if (count != tbf->lot_attr_count)
280 return (1);
281
282 for (i = 0; i != count; i++) {
283 if (attrs[i] != tbf->lot_attrs[i])
284 return (1);
285 }
286 return (0);
287 }
288
289 #define SA_ATTR_HASH(attr) (zfs_crc64_table[(-1ULL ^ attr) & 0xFF])
290
291 static uint64_t
292 sa_layout_info_hash(sa_attr_type_t *attrs, int attr_count)
293 {
294 int i;
295 uint64_t crc = -1ULL;
296
297 for (i = 0; i != attr_count; i++)
298 crc ^= SA_ATTR_HASH(attrs[i]);
299
300 return (crc);
301 }
302
303 static int
304 sa_get_spill(sa_handle_t *hdl)
305 {
306 int rc;
307 if (hdl->sa_spill == NULL) {
308 if ((rc = dmu_spill_hold_existing(hdl->sa_bonus, NULL,
309 &hdl->sa_spill)) == 0)
310 VERIFY(0 == sa_build_index(hdl, SA_SPILL));
311 } else {
312 rc = 0;
313 }
314
315 return (rc);
316 }
317
318 /*
319 * Main attribute lookup/update function
320 * returns 0 for success or non zero for failures
321 *
322 * Operates on bulk array, first failure will abort further processing
323 */
324 int
325 sa_attr_op(sa_handle_t *hdl, sa_bulk_attr_t *bulk, int count,
326 sa_data_op_t data_op, dmu_tx_t *tx)
327 {
328 sa_os_t *sa = hdl->sa_os->os_sa;
329 int i;
330 int error = 0;
331 sa_buf_type_t buftypes;
332
333 buftypes = 0;
334
335 ASSERT(count > 0);
336 for (i = 0; i != count; i++) {
337 ASSERT(bulk[i].sa_attr <= hdl->sa_os->os_sa->sa_num_attrs);
338
339 bulk[i].sa_addr = NULL;
340 /* First check the bonus buffer */
341
342 if (hdl->sa_bonus_tab && TOC_ATTR_PRESENT(
343 hdl->sa_bonus_tab->sa_idx_tab[bulk[i].sa_attr])) {
344 SA_ATTR_INFO(sa, hdl->sa_bonus_tab,
345 SA_GET_HDR(hdl, SA_BONUS),
346 bulk[i].sa_attr, bulk[i], SA_BONUS, hdl);
347 if (tx && !(buftypes & SA_BONUS)) {
348 dmu_buf_will_dirty(hdl->sa_bonus, tx);
349 buftypes |= SA_BONUS;
350 }
351 }
352 if (bulk[i].sa_addr == NULL &&
353 ((error = sa_get_spill(hdl)) == 0)) {
354 if (TOC_ATTR_PRESENT(
355 hdl->sa_spill_tab->sa_idx_tab[bulk[i].sa_attr])) {
356 SA_ATTR_INFO(sa, hdl->sa_spill_tab,
357 SA_GET_HDR(hdl, SA_SPILL),
358 bulk[i].sa_attr, bulk[i], SA_SPILL, hdl);
359 if (tx && !(buftypes & SA_SPILL) &&
360 bulk[i].sa_size == bulk[i].sa_length) {
361 dmu_buf_will_dirty(hdl->sa_spill, tx);
362 buftypes |= SA_SPILL;
363 }
364 }
365 }
366 if (error && error != ENOENT) {
367 return ((error == ECKSUM) ? EIO : error);
368 }
369
370 switch (data_op) {
371 case SA_LOOKUP:
372 if (bulk[i].sa_addr == NULL)
373 return (ENOENT);
374 if (bulk[i].sa_data) {
375 SA_COPY_DATA(bulk[i].sa_data_func,
376 bulk[i].sa_addr, bulk[i].sa_data,
377 bulk[i].sa_size);
378 }
379 continue;
380
381 case SA_UPDATE:
382 /* existing rewrite of attr */
383 if (bulk[i].sa_addr &&
384 bulk[i].sa_size == bulk[i].sa_length) {
385 SA_COPY_DATA(bulk[i].sa_data_func,
386 bulk[i].sa_data, bulk[i].sa_addr,
387 bulk[i].sa_length);
388 continue;
389 } else if (bulk[i].sa_addr) { /* attr size change */
390 error = sa_modify_attrs(hdl, bulk[i].sa_attr,
391 SA_REPLACE, bulk[i].sa_data_func,
392 bulk[i].sa_data, bulk[i].sa_length, tx);
393 } else { /* adding new attribute */
394 error = sa_modify_attrs(hdl, bulk[i].sa_attr,
395 SA_ADD, bulk[i].sa_data_func,
396 bulk[i].sa_data, bulk[i].sa_length, tx);
397 }
398 if (error)
399 return (error);
400 break;
401 }
402 }
403 return (error);
404 }
405
406 static sa_lot_t *
407 sa_add_layout_entry(objset_t *os, sa_attr_type_t *attrs, int attr_count,
408 uint64_t lot_num, uint64_t hash, boolean_t zapadd, dmu_tx_t *tx)
409 {
410 sa_os_t *sa = os->os_sa;
411 sa_lot_t *tb, *findtb;
412 int i;
413 avl_index_t loc;
414
415 ASSERT(MUTEX_HELD(&sa->sa_lock));
416 tb = kmem_zalloc(sizeof (sa_lot_t), KM_SLEEP);
417 tb->lot_attr_count = attr_count;
418 tb->lot_attrs = kmem_alloc(sizeof (sa_attr_type_t) * attr_count,
419 KM_SLEEP);
420 bcopy(attrs, tb->lot_attrs, sizeof (sa_attr_type_t) * attr_count);
421 tb->lot_num = lot_num;
422 tb->lot_hash = hash;
423 tb->lot_instance = 0;
424
425 if (zapadd) {
426 char attr_name[8];
427
428 if (sa->sa_layout_attr_obj == 0) {
429 sa->sa_layout_attr_obj = zap_create(os,
430 DMU_OT_SA_ATTR_LAYOUTS, DMU_OT_NONE, 0, tx);
431 VERIFY(zap_add(os, sa->sa_master_obj, SA_LAYOUTS, 8, 1,
432 &sa->sa_layout_attr_obj, tx) == 0);
433 }
434
435 (void) snprintf(attr_name, sizeof (attr_name),
436 "%d", (int)lot_num);
437 VERIFY(0 == zap_update(os, os->os_sa->sa_layout_attr_obj,
438 attr_name, 2, attr_count, attrs, tx));
439 }
440
441 list_create(&tb->lot_idx_tab, sizeof (sa_idx_tab_t),
442 offsetof(sa_idx_tab_t, sa_next));
443
444 for (i = 0; i != attr_count; i++) {
445 if (sa->sa_attr_table[tb->lot_attrs[i]].sa_length == 0)
446 tb->lot_var_sizes++;
447 }
448
449 avl_add(&sa->sa_layout_num_tree, tb);
450
451 /* verify we don't have a hash collision */
452 if ((findtb = avl_find(&sa->sa_layout_hash_tree, tb, &loc)) != NULL) {
453 for (; findtb && findtb->lot_hash == hash;
454 findtb = AVL_NEXT(&sa->sa_layout_hash_tree, findtb)) {
455 if (findtb->lot_instance != tb->lot_instance)
456 break;
457 tb->lot_instance++;
458 }
459 }
460 avl_add(&sa->sa_layout_hash_tree, tb);
461 return (tb);
462 }
463
464 static void
465 sa_find_layout(objset_t *os, uint64_t hash, sa_attr_type_t *attrs,
466 int count, dmu_tx_t *tx, sa_lot_t **lot)
467 {
468 sa_lot_t *tb, tbsearch;
469 avl_index_t loc;
470 sa_os_t *sa = os->os_sa;
471 boolean_t found = B_FALSE;
472
473 mutex_enter(&sa->sa_lock);
474 tbsearch.lot_hash = hash;
475 tbsearch.lot_instance = 0;
476 tb = avl_find(&sa->sa_layout_hash_tree, &tbsearch, &loc);
477 if (tb) {
478 for (; tb && tb->lot_hash == hash;
479 tb = AVL_NEXT(&sa->sa_layout_hash_tree, tb)) {
480 if (sa_layout_equal(tb, attrs, count) == 0) {
481 found = B_TRUE;
482 break;
483 }
484 }
485 }
486 if (!found) {
487 tb = sa_add_layout_entry(os, attrs, count,
488 avl_numnodes(&sa->sa_layout_num_tree), hash, B_TRUE, tx);
489 }
490 mutex_exit(&sa->sa_lock);
491 *lot = tb;
492 }
493
494 static int
495 sa_resize_spill(sa_handle_t *hdl, uint32_t size, dmu_tx_t *tx)
496 {
497 int error;
498 uint32_t blocksize;
499
500 if (size == 0) {
501 blocksize = SPA_MINBLOCKSIZE;
502 } else if (size > SPA_MAXBLOCKSIZE) {
503 ASSERT(0);
504 return (EFBIG);
505 } else {
506 blocksize = P2ROUNDUP_TYPED(size, SPA_MINBLOCKSIZE, uint32_t);
507 }
508
509 error = dbuf_spill_set_blksz(hdl->sa_spill, blocksize, tx);
510 ASSERT(error == 0);
511 return (error);
512 }
513
514 static void
515 sa_copy_data(sa_data_locator_t *func, void *datastart, void *target, int buflen)
516 {
517 if (func == NULL) {
518 bcopy(datastart, target, buflen);
519 } else {
520 boolean_t start;
521 int bytes;
522 void *dataptr;
523 void *saptr = target;
524 uint32_t length;
525
526 start = B_TRUE;
527 bytes = 0;
528 while (bytes < buflen) {
529 func(&dataptr, &length, buflen, start, datastart);
530 bcopy(dataptr, saptr, length);
531 saptr = (void *)((caddr_t)saptr + length);
532 bytes += length;
533 start = B_FALSE;
534 }
535 }
536 }
537
538 /*
539 * Determine several different sizes
540 * first the sa header size
541 * the number of bytes to be stored
542 * if spill would occur the index in the attribute array is returned
543 *
544 * the boolean will_spill will be set when spilling is necessary. It
545 * is only set when the buftype is SA_BONUS
546 */
547 static int
548 sa_find_sizes(sa_os_t *sa, sa_bulk_attr_t *attr_desc, int attr_count,
549 dmu_buf_t *db, sa_buf_type_t buftype, int *index, int *total,
550 boolean_t *will_spill)
551 {
552 int var_size = 0;
553 int i;
554 int full_space;
555 int hdrsize;
556 boolean_t done = B_FALSE;
557
558 if (buftype == SA_BONUS && sa->sa_force_spill) {
559 *total = 0;
560 *index = 0;
561 *will_spill = B_TRUE;
562 return (0);
563 }
564
565 *index = -1;
566 *total = 0;
567
568 if (buftype == SA_BONUS)
569 *will_spill = B_FALSE;
570
571 hdrsize = (SA_BONUSTYPE_FROM_DB(db) == DMU_OT_ZNODE) ? 0 :
572 sizeof (sa_hdr_phys_t);
573
574 full_space = (buftype == SA_BONUS) ? DN_MAX_BONUSLEN : db->db_size;
575
576 for (i = 0; i != attr_count; i++) {
577 boolean_t is_var_sz;
578
579 *total += attr_desc[i].sa_length;
580 if (done)
581 goto next;
582
583 is_var_sz = (SA_REGISTERED_LEN(sa, attr_desc[i].sa_attr) == 0);
584 if (is_var_sz) {
585 var_size++;
586 }
587
588 if (is_var_sz && var_size > 1) {
589 if (P2ROUNDUP(hdrsize + sizeof (uint16_t), 8) +
590 *total < full_space) {
591 hdrsize += sizeof (uint16_t);
592 } else {
593 done = B_TRUE;
594 *index = i;
595 if (buftype == SA_BONUS)
596 *will_spill = B_TRUE;
597 continue;
598 }
599 }
600
601 /*
602 * find index of where spill *could* occur.
603 * Then continue to count of remainder attribute
604 * space. The sum is used later for sizing bonus
605 * and spill buffer.
606 */
607 if (buftype == SA_BONUS && *index == -1 &&
608 P2ROUNDUP(*total + hdrsize, 8) >
609 (full_space - sizeof (blkptr_t))) {
610 *index = i;
611 done = B_TRUE;
612 }
613
614 next:
615 if (P2ROUNDUP(*total + hdrsize, 8) > full_space &&
616 buftype == SA_BONUS)
617 *will_spill = B_TRUE;
618 }
619
620 hdrsize = P2ROUNDUP(hdrsize, 8);
621 return (hdrsize);
622 }
623
624 #define BUF_SPACE_NEEDED(total, header) (total + header)
625
626 /*
627 * Find layout that corresponds to ordering of attributes
628 * If not found a new layout number is created and added to
629 * persistent layout tables.
630 */
631 static int
632 sa_build_layouts(sa_handle_t *hdl, sa_bulk_attr_t *attr_desc, int attr_count,
633 dmu_tx_t *tx)
634 {
635 sa_os_t *sa = hdl->sa_os->os_sa;
636 uint64_t hash;
637 sa_buf_type_t buftype;
638 sa_hdr_phys_t *sahdr;
639 void *data_start;
640 int buf_space;
641 sa_attr_type_t *attrs, *attrs_start;
642 int i, lot_count;
643 int hdrsize, spillhdrsize;
644 int used;
645 dmu_object_type_t bonustype;
646 sa_lot_t *lot;
647 int len_idx;
648 int spill_used;
649 boolean_t spilling;
650
651 dmu_buf_will_dirty(hdl->sa_bonus, tx);
652 bonustype = SA_BONUSTYPE_FROM_DB(hdl->sa_bonus);
653
654 /* first determine bonus header size and sum of all attributes */
655 hdrsize = sa_find_sizes(sa, attr_desc, attr_count, hdl->sa_bonus,
656 SA_BONUS, &i, &used, &spilling);
657
658 if (used > SPA_MAXBLOCKSIZE)
659 return (EFBIG);
660
661 VERIFY(0 == dmu_set_bonus(hdl->sa_bonus, spilling ?
662 MIN(DN_MAX_BONUSLEN - sizeof (blkptr_t), used + hdrsize) :
663 used + hdrsize, tx));
664
665 ASSERT((bonustype == DMU_OT_ZNODE && spilling == 0) ||
666 bonustype == DMU_OT_SA);
667
668 /* setup and size spill buffer when needed */
669 if (spilling) {
670 boolean_t dummy;
671
672 if (hdl->sa_spill == NULL) {
673 VERIFY(dmu_spill_hold_by_bonus(hdl->sa_bonus, NULL,
674 &hdl->sa_spill) == 0);
675 }
676 dmu_buf_will_dirty(hdl->sa_spill, tx);
677
678 spillhdrsize = sa_find_sizes(sa, &attr_desc[i],
679 attr_count - i, hdl->sa_spill, SA_SPILL, &i,
680 &spill_used, &dummy);
681
682 if (spill_used > SPA_MAXBLOCKSIZE)
683 return (EFBIG);
684
685 buf_space = hdl->sa_spill->db_size - spillhdrsize;
686 if (BUF_SPACE_NEEDED(spill_used, spillhdrsize) >
687 hdl->sa_spill->db_size)
688 VERIFY(0 == sa_resize_spill(hdl,
689 BUF_SPACE_NEEDED(spill_used, spillhdrsize), tx));
690 }
691
692 /* setup starting pointers to lay down data */
693 data_start = (void *)((uintptr_t)hdl->sa_bonus->db_data + hdrsize);
694 sahdr = (sa_hdr_phys_t *)hdl->sa_bonus->db_data;
695 buftype = SA_BONUS;
696
697 if (spilling)
698 buf_space = (sa->sa_force_spill) ?
699 0 : SA_BLKPTR_SPACE - hdrsize;
700 else
701 buf_space = hdl->sa_bonus->db_size - hdrsize;
702
703 attrs_start = attrs = kmem_alloc(sizeof (sa_attr_type_t) * attr_count,
704 KM_SLEEP);
705 lot_count = 0;
706
707 for (i = 0, len_idx = 0, hash = -1ULL; i != attr_count; i++) {
708 uint16_t length;
709
710 attrs[i] = attr_desc[i].sa_attr;
711 length = SA_REGISTERED_LEN(sa, attrs[i]);
712 if (length == 0)
713 length = attr_desc[i].sa_length;
714
715 if (buf_space < length) { /* switch to spill buffer */
716 VERIFY(bonustype == DMU_OT_SA);
717 if (buftype == SA_BONUS && !sa->sa_force_spill) {
718 sa_find_layout(hdl->sa_os, hash, attrs_start,
719 lot_count, tx, &lot);
720 SA_SET_HDR(sahdr, lot->lot_num, hdrsize);
721 }
722
723 buftype = SA_SPILL;
724 hash = -1ULL;
725 len_idx = 0;
726
727 sahdr = (sa_hdr_phys_t *)hdl->sa_spill->db_data;
728 sahdr->sa_magic = SA_MAGIC;
729 data_start = (void *)((uintptr_t)sahdr +
730 spillhdrsize);
731 attrs_start = &attrs[i];
732 buf_space = hdl->sa_spill->db_size - spillhdrsize;
733 lot_count = 0;
734 }
735 hash ^= SA_ATTR_HASH(attrs[i]);
736 attr_desc[i].sa_addr = data_start;
737 attr_desc[i].sa_size = length;
738 SA_COPY_DATA(attr_desc[i].sa_data_func, attr_desc[i].sa_data,
739 data_start, length);
740 if (sa->sa_attr_table[attrs[i]].sa_length == 0) {
741 sahdr->sa_lengths[len_idx++] = length;
742 }
743 data_start = (void *)P2ROUNDUP(((uintptr_t)data_start +
744 length), 8);
745 buf_space -= P2ROUNDUP(length, 8);
746 lot_count++;
747 }
748
749 sa_find_layout(hdl->sa_os, hash, attrs_start, lot_count, tx, &lot);
750
751 /*
752 * Verify that old znodes always have layout number 0.
753 * Must be DMU_OT_SA for arbitrary layouts
754 */
755 VERIFY((bonustype == DMU_OT_ZNODE && lot->lot_num == 0) ||
756 (bonustype == DMU_OT_SA && lot->lot_num > 1));
757
758 if (bonustype == DMU_OT_SA) {
759 SA_SET_HDR(sahdr, lot->lot_num,
760 buftype == SA_BONUS ? hdrsize : spillhdrsize);
761 }
762
763 kmem_free(attrs, sizeof (sa_attr_type_t) * attr_count);
764 if (hdl->sa_bonus_tab) {
765 sa_idx_tab_rele(hdl->sa_os, hdl->sa_bonus_tab);
766 hdl->sa_bonus_tab = NULL;
767 }
768 if (!sa->sa_force_spill)
769 VERIFY(0 == sa_build_index(hdl, SA_BONUS));
770 if (hdl->sa_spill) {
771 sa_idx_tab_rele(hdl->sa_os, hdl->sa_spill_tab);
772 if (!spilling) {
773 /*
774 * remove spill block that is no longer needed.
775 */
776 dmu_buf_rele(hdl->sa_spill, NULL);
777 hdl->sa_spill = NULL;
778 hdl->sa_spill_tab = NULL;
779 VERIFY(0 == dmu_rm_spill(hdl->sa_os,
780 sa_handle_object(hdl), tx));
781 } else {
782 VERIFY(0 == sa_build_index(hdl, SA_SPILL));
783 }
784 }
785
786 return (0);
787 }
788
789 static void
790 sa_free_attr_table(sa_os_t *sa)
791 {
792 int i;
793
794 if (sa->sa_attr_table == NULL)
795 return;
796
797 for (i = 0; i != sa->sa_num_attrs; i++) {
798 if (sa->sa_attr_table[i].sa_name)
799 kmem_free(sa->sa_attr_table[i].sa_name,
800 strlen(sa->sa_attr_table[i].sa_name) + 1);
801 }
802
803 kmem_free(sa->sa_attr_table,
804 sizeof (sa_attr_table_t) * sa->sa_num_attrs);
805
806 sa->sa_attr_table = NULL;
807 }
808
809 static int
810 sa_attr_table_setup(objset_t *os, sa_attr_reg_t *reg_attrs, int count)
811 {
812 sa_os_t *sa = os->os_sa;
813 uint64_t sa_attr_count = 0;
814 uint64_t sa_reg_count;
815 int error = 0;
816 uint64_t attr_value;
817 sa_attr_table_t *tb;
818 zap_cursor_t zc;
819 zap_attribute_t za;
820 int registered_count = 0;
821 int i;
822 dmu_objset_type_t ostype = dmu_objset_type(os);
823
824 sa->sa_user_table =
825 kmem_zalloc(count * sizeof (sa_attr_type_t), KM_SLEEP);
826 sa->sa_user_table_sz = count * sizeof (sa_attr_type_t);
827
828 if (sa->sa_reg_attr_obj != 0) {
829 error = zap_count(os, sa->sa_reg_attr_obj,
830 &sa_attr_count);
831
832 /*
833 * Make sure we retrieved a count and that it isn't zero
834 */
835 if (error || (error == 0 && sa_attr_count == 0)) {
836 if (error == 0)
837 error = EINVAL;
838 goto bail;
839 }
840 sa_reg_count = sa_attr_count;
841 }
842
843 if (ostype == DMU_OST_ZFS && sa_attr_count == 0)
844 sa_attr_count += sa_legacy_attr_count;
845
846 /* Allocate attribute numbers for attributes that aren't registered */
847 for (i = 0; i != count; i++) {
848 boolean_t found = B_FALSE;
849 int j;
850
851 if (ostype == DMU_OST_ZFS) {
852 for (j = 0; j != sa_legacy_attr_count; j++) {
853 if (strcmp(reg_attrs[i].sa_name,
854 sa_legacy_attrs[j].sa_name) == 0) {
855 sa->sa_user_table[i] =
856 sa_legacy_attrs[j].sa_attr;
857 found = B_TRUE;
858 }
859 }
860 }
861 if (found)
862 continue;
863
864 if (sa->sa_reg_attr_obj)
865 error = zap_lookup(os, sa->sa_reg_attr_obj,
866 reg_attrs[i].sa_name, 8, 1, &attr_value);
867 else
868 error = ENOENT;
869 switch (error) {
870 case ENOENT:
871 sa->sa_user_table[i] = (sa_attr_type_t)sa_attr_count;
872 sa_attr_count++;
873 break;
874 case 0:
875 sa->sa_user_table[i] = ATTR_NUM(attr_value);
876 break;
877 default:
878 goto bail;
879 }
880 }
881
882 sa->sa_num_attrs = sa_attr_count;
883 tb = sa->sa_attr_table =
884 kmem_zalloc(sizeof (sa_attr_table_t) * sa_attr_count, KM_SLEEP);
885
886 /*
887 * Attribute table is constructed from requested attribute list,
888 * previously foreign registered attributes, and also the legacy
889 * ZPL set of attributes.
890 */
891
892 if (sa->sa_reg_attr_obj) {
893 for (zap_cursor_init(&zc, os, sa->sa_reg_attr_obj);
894 (error = zap_cursor_retrieve(&zc, &za)) == 0;
895 zap_cursor_advance(&zc)) {
896 uint64_t value;
897 value = za.za_first_integer;
898
899 registered_count++;
900 tb[ATTR_NUM(value)].sa_attr = ATTR_NUM(value);
901 tb[ATTR_NUM(value)].sa_length = ATTR_LENGTH(value);
902 tb[ATTR_NUM(value)].sa_byteswap = ATTR_BSWAP(value);
903 tb[ATTR_NUM(value)].sa_registered = B_TRUE;
904
905 if (tb[ATTR_NUM(value)].sa_name) {
906 continue;
907 }
908 tb[ATTR_NUM(value)].sa_name =
909 kmem_zalloc(strlen(za.za_name) +1, KM_SLEEP);
910 (void) strlcpy(tb[ATTR_NUM(value)].sa_name, za.za_name,
911 strlen(za.za_name) +1);
912 }
913 zap_cursor_fini(&zc);
914 /*
915 * Make sure we processed the correct number of registered
916 * attributes
917 */
918 if (registered_count != sa_reg_count) {
919 ASSERT(error != 0);
920 goto bail;
921 }
922
923 }
924
925 if (ostype == DMU_OST_ZFS) {
926 for (i = 0; i != sa_legacy_attr_count; i++) {
927 if (tb[i].sa_name)
928 continue;
929 tb[i].sa_attr = sa_legacy_attrs[i].sa_attr;
930 tb[i].sa_length = sa_legacy_attrs[i].sa_length;
931 tb[i].sa_byteswap = sa_legacy_attrs[i].sa_byteswap;
932 tb[i].sa_registered = B_FALSE;
933 tb[i].sa_name =
934 kmem_zalloc(strlen(sa_legacy_attrs[i].sa_name) +1,
935 KM_SLEEP);
936 (void) strlcpy(tb[i].sa_name,
937 sa_legacy_attrs[i].sa_name,
938 strlen(sa_legacy_attrs[i].sa_name) + 1);
939 }
940 }
941
942 for (i = 0; i != count; i++) {
943 sa_attr_type_t attr_id;
944
945 attr_id = sa->sa_user_table[i];
946 if (tb[attr_id].sa_name)
947 continue;
948
949 tb[attr_id].sa_length = reg_attrs[i].sa_length;
950 tb[attr_id].sa_byteswap = reg_attrs[i].sa_byteswap;
951 tb[attr_id].sa_attr = attr_id;
952 tb[attr_id].sa_name =
953 kmem_zalloc(strlen(reg_attrs[i].sa_name) + 1, KM_SLEEP);
954 (void) strlcpy(tb[attr_id].sa_name, reg_attrs[i].sa_name,
955 strlen(reg_attrs[i].sa_name) + 1);
956 }
957
958 sa->sa_need_attr_registration =
959 (sa_attr_count != registered_count);
960
961 return (0);
962 bail:
963 kmem_free(sa->sa_user_table, count * sizeof (sa_attr_type_t));
964 sa->sa_user_table = NULL;
965 sa_free_attr_table(sa);
966 return ((error != 0) ? error : EINVAL);
967 }
968
969 int
970 sa_setup(objset_t *os, uint64_t sa_obj, sa_attr_reg_t *reg_attrs, int count,
971 sa_attr_type_t **user_table)
972 {
973 zap_cursor_t zc;
974 zap_attribute_t za;
975 sa_os_t *sa;
976 dmu_objset_type_t ostype = dmu_objset_type(os);
977 sa_attr_type_t *tb;
978 int error;
979
980 mutex_enter(&os->os_lock);
981 if (os->os_sa) {
982 mutex_enter(&os->os_sa->sa_lock);
983 mutex_exit(&os->os_lock);
984 tb = os->os_sa->sa_user_table;
985 mutex_exit(&os->os_sa->sa_lock);
986 *user_table = tb;
987 return (0);
988 }
989
990 sa = kmem_zalloc(sizeof (sa_os_t), KM_SLEEP);
991 mutex_init(&sa->sa_lock, NULL, MUTEX_DEFAULT, NULL);
992 sa->sa_master_obj = sa_obj;
993
994 os->os_sa = sa;
995 mutex_enter(&sa->sa_lock);
996 mutex_exit(&os->os_lock);
997 avl_create(&sa->sa_layout_num_tree, layout_num_compare,
998 sizeof (sa_lot_t), offsetof(sa_lot_t, lot_num_node));
999 avl_create(&sa->sa_layout_hash_tree, layout_hash_compare,
1000 sizeof (sa_lot_t), offsetof(sa_lot_t, lot_hash_node));
1001
1002 if (sa_obj) {
1003 error = zap_lookup(os, sa_obj, SA_LAYOUTS,
1004 8, 1, &sa->sa_layout_attr_obj);
1005 if (error != 0 && error != ENOENT)
1006 goto fail;
1007 error = zap_lookup(os, sa_obj, SA_REGISTRY,
1008 8, 1, &sa->sa_reg_attr_obj);
1009 if (error != 0 && error != ENOENT)
1010 goto fail;
1011 }
1012
1013 if ((error = sa_attr_table_setup(os, reg_attrs, count)) != 0)
1014 goto fail;
1015
1016 if (sa->sa_layout_attr_obj != 0) {
1017 uint64_t layout_count;
1018
1019 error = zap_count(os, sa->sa_layout_attr_obj,
1020 &layout_count);
1021
1022 /*
1023 * Layout number count should be > 0
1024 */
1025 if (error || (error == 0 && layout_count == 0)) {
1026 if (error == 0)
1027 error = EINVAL;
1028 goto fail;
1029 }
1030
1031 for (zap_cursor_init(&zc, os, sa->sa_layout_attr_obj);
1032 (error = zap_cursor_retrieve(&zc, &za)) == 0;
1033 zap_cursor_advance(&zc)) {
1034 sa_attr_type_t *lot_attrs;
1035 uint64_t lot_num;
1036
1037 lot_attrs = kmem_zalloc(sizeof (sa_attr_type_t) *
1038 za.za_num_integers, KM_SLEEP);
1039
1040 if ((error = (zap_lookup(os, sa->sa_layout_attr_obj,
1041 za.za_name, 2, za.za_num_integers,
1042 lot_attrs))) != 0) {
1043 kmem_free(lot_attrs, sizeof (sa_attr_type_t) *
1044 za.za_num_integers);
1045 break;
1046 }
1047 VERIFY(ddi_strtoull(za.za_name, NULL, 10,
1048 (unsigned long long *)&lot_num) == 0);
1049
1050 (void) sa_add_layout_entry(os, lot_attrs,
1051 za.za_num_integers, lot_num,
1052 sa_layout_info_hash(lot_attrs,
1053 za.za_num_integers), B_FALSE, NULL);
1054 kmem_free(lot_attrs, sizeof (sa_attr_type_t) *
1055 za.za_num_integers);
1056 }
1057 zap_cursor_fini(&zc);
1058
1059 /*
1060 * Make sure layout count matches number of entries added
1061 * to AVL tree
1062 */
1063 if (avl_numnodes(&sa->sa_layout_num_tree) != layout_count) {
1064 ASSERT(error != 0);
1065 goto fail;
1066 }
1067 }
1068
1069 /* Add special layout number for old ZNODES */
1070 if (ostype == DMU_OST_ZFS) {
1071 (void) sa_add_layout_entry(os, sa_legacy_zpl_layout,
1072 sa_legacy_attr_count, 0,
1073 sa_layout_info_hash(sa_legacy_zpl_layout,
1074 sa_legacy_attr_count), B_FALSE, NULL);
1075
1076 (void) sa_add_layout_entry(os, sa_dummy_zpl_layout, 0, 1,
1077 0, B_FALSE, NULL);
1078 }
1079 *user_table = os->os_sa->sa_user_table;
1080 mutex_exit(&sa->sa_lock);
1081 return (0);
1082 fail:
1083 os->os_sa = NULL;
1084 sa_free_attr_table(sa);
1085 if (sa->sa_user_table)
1086 kmem_free(sa->sa_user_table, sa->sa_user_table_sz);
1087 mutex_exit(&sa->sa_lock);
1088 kmem_free(sa, sizeof (sa_os_t));
1089 return ((error == ECKSUM) ? EIO : error);
1090 }
1091
1092 void
1093 sa_tear_down(objset_t *os)
1094 {
1095 sa_os_t *sa = os->os_sa;
1096 sa_lot_t *layout;
1097 void *cookie;
1098
1099 kmem_free(sa->sa_user_table, sa->sa_user_table_sz);
1100
1101 /* Free up attr table */
1102
1103 sa_free_attr_table(sa);
1104
1105 cookie = NULL;
1106 while (layout = avl_destroy_nodes(&sa->sa_layout_hash_tree, &cookie)) {
1107 sa_idx_tab_t *tab;
1108 while (tab = list_head(&layout->lot_idx_tab)) {
1109 ASSERT(refcount_count(&tab->sa_refcount));
1110 sa_idx_tab_rele(os, tab);
1111 }
1112 }
1113
1114 cookie = NULL;
1115 while (layout = avl_destroy_nodes(&sa->sa_layout_num_tree, &cookie)) {
1116 kmem_free(layout->lot_attrs,
1117 sizeof (sa_attr_type_t) * layout->lot_attr_count);
1118 kmem_free(layout, sizeof (sa_lot_t));
1119 }
1120
1121 avl_destroy(&sa->sa_layout_hash_tree);
1122 avl_destroy(&sa->sa_layout_num_tree);
1123
1124 kmem_free(sa, sizeof (sa_os_t));
1125 os->os_sa = NULL;
1126 }
1127
1128 void
1129 sa_build_idx_tab(void *hdr, void *attr_addr, sa_attr_type_t attr,
1130 uint16_t length, int length_idx, boolean_t var_length, void *userp)
1131 {
1132 sa_idx_tab_t *idx_tab = userp;
1133
1134 if (var_length) {
1135 ASSERT(idx_tab->sa_variable_lengths);
1136 idx_tab->sa_variable_lengths[length_idx] = length;
1137 }
1138 TOC_ATTR_ENCODE(idx_tab->sa_idx_tab[attr], length_idx,
1139 (uint32_t)((uintptr_t)attr_addr - (uintptr_t)hdr));
1140 }
1141
1142 static void
1143 sa_attr_iter(objset_t *os, sa_hdr_phys_t *hdr, dmu_object_type_t type,
1144 sa_iterfunc_t func, sa_lot_t *tab, void *userp)
1145 {
1146 void *data_start;
1147 sa_lot_t *tb = tab;
1148 sa_lot_t search;
1149 avl_index_t loc;
1150 sa_os_t *sa = os->os_sa;
1151 int i;
1152 uint16_t *length_start = NULL;
1153 uint8_t length_idx = 0;
1154
1155 if (tab == NULL) {
1156 search.lot_num = SA_LAYOUT_NUM(hdr, type);
1157 tb = avl_find(&sa->sa_layout_num_tree, &search, &loc);
1158 ASSERT(tb);
1159 }
1160
1161 if (IS_SA_BONUSTYPE(type)) {
1162 data_start = (void *)P2ROUNDUP(((uintptr_t)hdr +
1163 offsetof(sa_hdr_phys_t, sa_lengths) +
1164 (sizeof (uint16_t) * tb->lot_var_sizes)), 8);
1165 length_start = hdr->sa_lengths;
1166 } else {
1167 data_start = hdr;
1168 }
1169
1170 for (i = 0; i != tb->lot_attr_count; i++) {
1171 int attr_length, reg_length;
1172 uint8_t idx_len;
1173
1174 reg_length = sa->sa_attr_table[tb->lot_attrs[i]].sa_length;
1175 if (reg_length) {
1176 attr_length = reg_length;
1177 idx_len = 0;
1178 } else {
1179 attr_length = length_start[length_idx];
1180 idx_len = length_idx++;
1181 }
1182
1183 func(hdr, data_start, tb->lot_attrs[i], attr_length,
1184 idx_len, reg_length == 0 ? B_TRUE : B_FALSE, userp);
1185
1186 data_start = (void *)P2ROUNDUP(((uintptr_t)data_start +
1187 attr_length), 8);
1188 }
1189 }
1190
1191 /*ARGSUSED*/
1192 void
1193 sa_byteswap_cb(void *hdr, void *attr_addr, sa_attr_type_t attr,
1194 uint16_t length, int length_idx, boolean_t variable_length, void *userp)
1195 {
1196 sa_handle_t *hdl = userp;
1197 sa_os_t *sa = hdl->sa_os->os_sa;
1198
1199 sa_bswap_table[sa->sa_attr_table[attr].sa_byteswap](attr_addr, length);
1200 }
1201
1202 void
1203 sa_byteswap(sa_handle_t *hdl, sa_buf_type_t buftype)
1204 {
1205 sa_hdr_phys_t *sa_hdr_phys = SA_GET_HDR(hdl, buftype);
1206 dmu_buf_impl_t *db;
1207 sa_os_t *sa = hdl->sa_os->os_sa;
1208 int num_lengths = 1;
1209 int i;
1210
1211 ASSERT(MUTEX_HELD(&sa->sa_lock));
1212 if (sa_hdr_phys->sa_magic == SA_MAGIC)
1213 return;
1214
1215 db = SA_GET_DB(hdl, buftype);
1216
1217 if (buftype == SA_SPILL) {
1218 arc_release(db->db_buf, NULL);
1219 arc_buf_thaw(db->db_buf);
1220 }
1221
1222 sa_hdr_phys->sa_magic = BSWAP_32(sa_hdr_phys->sa_magic);
1223 sa_hdr_phys->sa_layout_info = BSWAP_16(sa_hdr_phys->sa_layout_info);
1224
1225 /*
1226 * Determine number of variable lenghts in header
1227 * The standard 8 byte header has one for free and a
1228 * 16 byte header would have 4 + 1;
1229 */
1230 if (SA_HDR_SIZE(sa_hdr_phys) > 8)
1231 num_lengths += (SA_HDR_SIZE(sa_hdr_phys) - 8) >> 1;
1232 for (i = 0; i != num_lengths; i++)
1233 sa_hdr_phys->sa_lengths[i] =
1234 BSWAP_16(sa_hdr_phys->sa_lengths[i]);
1235
1236 sa_attr_iter(hdl->sa_os, sa_hdr_phys, DMU_OT_SA,
1237 sa_byteswap_cb, NULL, hdl);
1238
1239 if (buftype == SA_SPILL)
1240 arc_buf_freeze(((dmu_buf_impl_t *)hdl->sa_spill)->db_buf);
1241 }
1242
1243 static int
1244 sa_build_index(sa_handle_t *hdl, sa_buf_type_t buftype)
1245 {
1246 sa_hdr_phys_t *sa_hdr_phys;
1247 dmu_buf_impl_t *db = SA_GET_DB(hdl, buftype);
1248 dmu_object_type_t bonustype = SA_BONUSTYPE_FROM_DB(db);
1249 sa_os_t *sa = hdl->sa_os->os_sa;
1250 sa_idx_tab_t *idx_tab;
1251
1252 sa_hdr_phys = SA_GET_HDR(hdl, buftype);
1253
1254 mutex_enter(&sa->sa_lock);
1255
1256 /* Do we need to byteswap? */
1257
1258 /* only check if not old znode */
1259 if (IS_SA_BONUSTYPE(bonustype) && sa_hdr_phys->sa_magic != SA_MAGIC &&
1260 sa_hdr_phys->sa_magic != 0) {
1261 VERIFY(BSWAP_32(sa_hdr_phys->sa_magic) == SA_MAGIC);
1262 sa_byteswap(hdl, buftype);
1263 }
1264
1265 idx_tab = sa_find_idx_tab(hdl->sa_os, bonustype, sa_hdr_phys);
1266
1267 if (buftype == SA_BONUS)
1268 hdl->sa_bonus_tab = idx_tab;
1269 else
1270 hdl->sa_spill_tab = idx_tab;
1271
1272 mutex_exit(&sa->sa_lock);
1273 return (0);
1274 }
1275
1276 /*ARGSUSED*/
1277 void
1278 sa_evict(dmu_buf_t *db, void *sap)
1279 {
1280 panic("evicting sa dbuf %p\n", (void *)db);
1281 }
1282
1283 static void
1284 sa_idx_tab_rele(objset_t *os, void *arg)
1285 {
1286 sa_os_t *sa = os->os_sa;
1287 sa_idx_tab_t *idx_tab = arg;
1288
1289 if (idx_tab == NULL)
1290 return;
1291
1292 mutex_enter(&sa->sa_lock);
1293 if (refcount_remove(&idx_tab->sa_refcount, NULL) == 0) {
1294 list_remove(&idx_tab->sa_layout->lot_idx_tab, idx_tab);
1295 if (idx_tab->sa_variable_lengths)
1296 kmem_free(idx_tab->sa_variable_lengths,
1297 sizeof (uint16_t) *
1298 idx_tab->sa_layout->lot_var_sizes);
1299 refcount_destroy(&idx_tab->sa_refcount);
1300 kmem_free(idx_tab->sa_idx_tab,
1301 sizeof (uint32_t) * sa->sa_num_attrs);
1302 kmem_free(idx_tab, sizeof (sa_idx_tab_t));
1303 }
1304 mutex_exit(&sa->sa_lock);
1305 }
1306
1307 static void
1308 sa_idx_tab_hold(objset_t *os, sa_idx_tab_t *idx_tab)
1309 {
1310 sa_os_t *sa = os->os_sa;
1311
1312 ASSERT(MUTEX_HELD(&sa->sa_lock));
1313 (void) refcount_add(&idx_tab->sa_refcount, NULL);
1314 }
1315
1316 void
1317 sa_handle_destroy(sa_handle_t *hdl)
1318 {
1319 mutex_enter(&hdl->sa_lock);
1320 (void) dmu_buf_update_user((dmu_buf_t *)hdl->sa_bonus, hdl,
1321 NULL, NULL, NULL);
1322
1323 if (hdl->sa_bonus_tab) {
1324 sa_idx_tab_rele(hdl->sa_os, hdl->sa_bonus_tab);
1325 hdl->sa_bonus_tab = NULL;
1326 }
1327 if (hdl->sa_spill_tab) {
1328 sa_idx_tab_rele(hdl->sa_os, hdl->sa_spill_tab);
1329 hdl->sa_spill_tab = NULL;
1330 }
1331
1332 dmu_buf_rele(hdl->sa_bonus, NULL);
1333
1334 if (hdl->sa_spill)
1335 dmu_buf_rele((dmu_buf_t *)hdl->sa_spill, NULL);
1336 mutex_exit(&hdl->sa_lock);
1337
1338 kmem_cache_free(sa_cache, hdl);
1339 }
1340
1341 int
1342 sa_handle_get_from_db(objset_t *os, dmu_buf_t *db, void *userp,
1343 sa_handle_type_t hdl_type, sa_handle_t **handlepp)
1344 {
1345 int error = 0;
1346 dmu_object_info_t doi;
1347 sa_handle_t *handle;
1348
1349 #ifdef ZFS_DEBUG
1350 dmu_object_info_from_db(db, &doi);
1351 ASSERT(doi.doi_bonus_type == DMU_OT_SA ||
1352 doi.doi_bonus_type == DMU_OT_ZNODE);
1353 #endif
1354 /* find handle, if it exists */
1355 /* if one doesn't exist then create a new one, and initialize it */
1356
1357 handle = (hdl_type == SA_HDL_SHARED) ? dmu_buf_get_user(db) : NULL;
1358 if (handle == NULL) {
1359 sa_handle_t *newhandle;
1360 handle = kmem_cache_alloc(sa_cache, KM_SLEEP);
1361 handle->sa_userp = userp;
1362 handle->sa_bonus = db;
1363 handle->sa_os = os;
1364 handle->sa_spill = NULL;
1365
1366 error = sa_build_index(handle, SA_BONUS);
1367 newhandle = (hdl_type == SA_HDL_SHARED) ?
1368 dmu_buf_set_user_ie(db, handle,
1369 NULL, sa_evict) : NULL;
1370
1371 if (newhandle != NULL) {
1372 kmem_cache_free(sa_cache, handle);
1373 handle = newhandle;
1374 }
1375 }
1376 *handlepp = handle;
1377
1378 return (error);
1379 }
1380
1381 int
1382 sa_handle_get(objset_t *objset, uint64_t objid, void *userp,
1383 sa_handle_type_t hdl_type, sa_handle_t **handlepp)
1384 {
1385 dmu_buf_t *db;
1386 int error;
1387
1388 if (error = dmu_bonus_hold(objset, objid, NULL, &db))
1389 return (error);
1390
1391 return (sa_handle_get_from_db(objset, db, userp, hdl_type,
1392 handlepp));
1393 }
1394
1395 int
1396 sa_buf_hold(objset_t *objset, uint64_t obj_num, void *tag, dmu_buf_t **db)
1397 {
1398 return (dmu_bonus_hold(objset, obj_num, tag, db));
1399 }
1400
1401 void
1402 sa_buf_rele(dmu_buf_t *db, void *tag)
1403 {
1404 dmu_buf_rele(db, tag);
1405 }
1406
1407 int
1408 sa_lookup_impl(sa_handle_t *hdl, sa_bulk_attr_t *bulk, int count)
1409 {
1410 ASSERT(hdl);
1411 ASSERT(MUTEX_HELD(&hdl->sa_lock));
1412 return (sa_attr_op(hdl, bulk, count, SA_LOOKUP, NULL));
1413 }
1414
1415 int
1416 sa_lookup(sa_handle_t *hdl, sa_attr_type_t attr, void *buf, uint32_t buflen)
1417 {
1418 int error;
1419 sa_bulk_attr_t bulk;
1420
1421 bulk.sa_attr = attr;
1422 bulk.sa_data = buf;
1423 bulk.sa_length = buflen;
1424 bulk.sa_data_func = NULL;
1425
1426 ASSERT(hdl);
1427 mutex_enter(&hdl->sa_lock);
1428 error = sa_lookup_impl(hdl, &bulk, 1);
1429 mutex_exit(&hdl->sa_lock);
1430 return (error);
1431 }
1432
1433 #ifdef _KERNEL
1434 int
1435 sa_lookup_uio(sa_handle_t *hdl, sa_attr_type_t attr, uio_t *uio)
1436 {
1437 int error;
1438 sa_bulk_attr_t bulk;
1439
1440 bulk.sa_data = NULL;
1441 bulk.sa_attr = attr;
1442 bulk.sa_data_func = NULL;
1443
1444 ASSERT(hdl);
1445
1446 mutex_enter(&hdl->sa_lock);
1447 if ((error = sa_attr_op(hdl, &bulk, 1, SA_LOOKUP, NULL)) == 0) {
1448 error = uiomove((void *)bulk.sa_addr, MIN(bulk.sa_size,
1449 uio->uio_resid), UIO_READ, uio);
1450 }
1451 mutex_exit(&hdl->sa_lock);
1452 return (error);
1453
1454 }
1455 #endif
1456
1457 void *
1458 sa_find_idx_tab(objset_t *os, dmu_object_type_t bonustype, void *data)
1459 {
1460 sa_idx_tab_t *idx_tab;
1461 sa_hdr_phys_t *hdr = (sa_hdr_phys_t *)data;
1462 sa_os_t *sa = os->os_sa;
1463 sa_lot_t *tb, search;
1464 avl_index_t loc;
1465
1466 /*
1467 * Deterimine layout number. If SA node and header == 0 then
1468 * force the index table to the dummy "1" empty layout.
1469 *
1470 * The layout number would only be zero for a newly created file
1471 * that has not added any attributes yet, or with crypto enabled which
1472 * doesn't write any attributes to the bonus buffer.
1473 */
1474
1475 search.lot_num = SA_LAYOUT_NUM(hdr, bonustype);
1476
1477 tb = avl_find(&sa->sa_layout_num_tree, &search, &loc);
1478
1479 /* Verify header size is consistent with layout information */
1480 ASSERT(tb);
1481 ASSERT(IS_SA_BONUSTYPE(bonustype) &&
1482 SA_HDR_SIZE_MATCH_LAYOUT(hdr, tb) || !IS_SA_BONUSTYPE(bonustype) ||
1483 (IS_SA_BONUSTYPE(bonustype) && hdr->sa_layout_info == 0));
1484
1485 /*
1486 * See if any of the already existing TOC entries can be reused?
1487 */
1488
1489 for (idx_tab = list_head(&tb->lot_idx_tab); idx_tab;
1490 idx_tab = list_next(&tb->lot_idx_tab, idx_tab)) {
1491 boolean_t valid_idx = B_TRUE;
1492 int i;
1493
1494 if (tb->lot_var_sizes != 0 &&
1495 idx_tab->sa_variable_lengths != NULL) {
1496 for (i = 0; i != tb->lot_var_sizes; i++) {
1497 if (hdr->sa_lengths[i] !=
1498 idx_tab->sa_variable_lengths[i]) {
1499 valid_idx = B_FALSE;
1500 break;
1501 }
1502 }
1503 }
1504 if (valid_idx) {
1505 sa_idx_tab_hold(os, idx_tab);
1506 return (idx_tab);
1507 }
1508 }
1509
1510 /* No such luck, create a new entry */
1511 idx_tab = kmem_zalloc(sizeof (sa_idx_tab_t), KM_SLEEP);
1512 idx_tab->sa_idx_tab =
1513 kmem_zalloc(sizeof (uint32_t) * sa->sa_num_attrs, KM_SLEEP);
1514 idx_tab->sa_layout = tb;
1515 refcount_create(&idx_tab->sa_refcount);
1516 if (tb->lot_var_sizes)
1517 idx_tab->sa_variable_lengths = kmem_alloc(sizeof (uint16_t) *
1518 tb->lot_var_sizes, KM_SLEEP);
1519
1520 sa_attr_iter(os, hdr, bonustype, sa_build_idx_tab,
1521 tb, idx_tab);
1522 sa_idx_tab_hold(os, idx_tab); /* one hold for consumer */
1523 sa_idx_tab_hold(os, idx_tab); /* one for layout */
1524 list_insert_tail(&tb->lot_idx_tab, idx_tab);
1525 return (idx_tab);
1526 }
1527
1528 void
1529 sa_default_locator(void **dataptr, uint32_t *len, uint32_t total_len,
1530 boolean_t start, void *userdata)
1531 {
1532 ASSERT(start);
1533
1534 *dataptr = userdata;
1535 *len = total_len;
1536 }
1537
1538 static void
1539 sa_attr_register_sync(sa_handle_t *hdl, dmu_tx_t *tx)
1540 {
1541 uint64_t attr_value = 0;
1542 sa_os_t *sa = hdl->sa_os->os_sa;
1543 sa_attr_table_t *tb = sa->sa_attr_table;
1544 int i;
1545
1546 mutex_enter(&sa->sa_lock);
1547
1548 if (!sa->sa_need_attr_registration || sa->sa_master_obj == NULL) {
1549 mutex_exit(&sa->sa_lock);
1550 return;
1551 }
1552
1553 if (sa->sa_reg_attr_obj == NULL) {
1554 sa->sa_reg_attr_obj = zap_create(hdl->sa_os,
1555 DMU_OT_SA_ATTR_REGISTRATION, DMU_OT_NONE, 0, tx);
1556 VERIFY(zap_add(hdl->sa_os, sa->sa_master_obj,
1557 SA_REGISTRY, 8, 1, &sa->sa_reg_attr_obj, tx) == 0);
1558 }
1559 for (i = 0; i != sa->sa_num_attrs; i++) {
1560 if (sa->sa_attr_table[i].sa_registered)
1561 continue;
1562 ATTR_ENCODE(attr_value, tb[i].sa_attr, tb[i].sa_length,
1563 tb[i].sa_byteswap);
1564 VERIFY(0 == zap_update(hdl->sa_os, sa->sa_reg_attr_obj,
1565 tb[i].sa_name, 8, 1, &attr_value, tx));
1566 tb[i].sa_registered = B_TRUE;
1567 }
1568 sa->sa_need_attr_registration = B_FALSE;
1569 mutex_exit(&sa->sa_lock);
1570 }
1571
1572 /*
1573 * Replace all attributes with attributes specified in template.
1574 * If dnode had a spill buffer then those attributes will be
1575 * also be replaced, possibly with just an empty spill block
1576 *
1577 * This interface is intended to only be used for bulk adding of
1578 * attributes for a new file. It will also be used by the ZPL
1579 * when converting and old formatted znode to native SA support.
1580 */
1581 int
1582 sa_replace_all_by_template_locked(sa_handle_t *hdl, sa_bulk_attr_t *attr_desc,
1583 int attr_count, dmu_tx_t *tx)
1584 {
1585 sa_os_t *sa = hdl->sa_os->os_sa;
1586
1587 if (sa->sa_need_attr_registration)
1588 sa_attr_register_sync(hdl, tx);
1589 return (sa_build_layouts(hdl, attr_desc, attr_count, tx));
1590 }
1591
1592 int
1593 sa_replace_all_by_template(sa_handle_t *hdl, sa_bulk_attr_t *attr_desc,
1594 int attr_count, dmu_tx_t *tx)
1595 {
1596 int error;
1597
1598 mutex_enter(&hdl->sa_lock);
1599 error = sa_replace_all_by_template_locked(hdl, attr_desc,
1600 attr_count, tx);
1601 mutex_exit(&hdl->sa_lock);
1602 return (error);
1603 }
1604
1605 /*
1606 * add/remove/replace a single attribute and then rewrite the entire set
1607 * of attributes.
1608 */
1609 static int
1610 sa_modify_attrs(sa_handle_t *hdl, sa_attr_type_t newattr,
1611 sa_data_op_t action, sa_data_locator_t *locator, void *datastart,
1612 uint16_t buflen, dmu_tx_t *tx)
1613 {
1614 sa_os_t *sa = hdl->sa_os->os_sa;
1615 dmu_buf_impl_t *db = (dmu_buf_impl_t *)hdl->sa_bonus;
1616 dnode_t *dn;
1617 sa_bulk_attr_t *attr_desc;
1618 void *old_data[2];
1619 int bonus_attr_count = 0;
1620 int bonus_data_size, spill_data_size;
1621 int spill_attr_count = 0;
1622 int error;
1623 uint16_t length;
1624 int i, j, k, length_idx;
1625 sa_hdr_phys_t *hdr;
1626 sa_idx_tab_t *idx_tab;
1627 int attr_count;
1628 int count;
1629
1630 ASSERT(MUTEX_HELD(&hdl->sa_lock));
1631
1632 /* First make of copy of the old data */
1633
1634 DB_DNODE_ENTER(db);
1635 dn = DB_DNODE(db);
1636 if (dn->dn_bonuslen != 0) {
1637 bonus_data_size = hdl->sa_bonus->db_size;
1638 old_data[0] = kmem_alloc(bonus_data_size, KM_SLEEP);
1639 bcopy(hdl->sa_bonus->db_data, old_data[0],
1640 hdl->sa_bonus->db_size);
1641 bonus_attr_count = hdl->sa_bonus_tab->sa_layout->lot_attr_count;
1642 } else {
1643 old_data[0] = NULL;
1644 }
1645 DB_DNODE_EXIT(db);
1646
1647 /* Bring spill buffer online if it isn't currently */
1648
1649 if ((error = sa_get_spill(hdl)) == 0) {
1650 spill_data_size = hdl->sa_spill->db_size;
1651 old_data[1] = kmem_alloc(spill_data_size, KM_SLEEP);
1652 bcopy(hdl->sa_spill->db_data, old_data[1],
1653 hdl->sa_spill->db_size);
1654 spill_attr_count =
1655 hdl->sa_spill_tab->sa_layout->lot_attr_count;
1656 } else if (error && error != ENOENT) {
1657 if (old_data[0])
1658 kmem_free(old_data[0], bonus_data_size);
1659 return (error);
1660 } else {
1661 old_data[1] = NULL;
1662 }
1663
1664 /* build descriptor of all attributes */
1665
1666 attr_count = bonus_attr_count + spill_attr_count;
1667 if (action == SA_ADD)
1668 attr_count++;
1669 else if (action == SA_REMOVE)
1670 attr_count--;
1671
1672 attr_desc = kmem_zalloc(sizeof (sa_bulk_attr_t) * attr_count, KM_SLEEP);
1673
1674 /*
1675 * loop through bonus and spill buffer if it exists, and
1676 * build up new attr_descriptor to reset the attributes
1677 */
1678 k = j = 0;
1679 count = bonus_attr_count;
1680 hdr = SA_GET_HDR(hdl, SA_BONUS);
1681 idx_tab = SA_IDX_TAB_GET(hdl, SA_BONUS);
1682 for (; k != 2; k++) {
1683 /* iterate over each attribute in layout */
1684 for (i = 0, length_idx = 0; i != count; i++) {
1685 sa_attr_type_t attr;
1686
1687 attr = idx_tab->sa_layout->lot_attrs[i];
1688 if (attr == newattr) {
1689 if (action == SA_REMOVE) {
1690 j++;
1691 continue;
1692 }
1693 ASSERT(SA_REGISTERED_LEN(sa, attr) == 0);
1694 ASSERT(action == SA_REPLACE);
1695 SA_ADD_BULK_ATTR(attr_desc, j, attr,
1696 locator, datastart, buflen);
1697 } else {
1698 length = SA_REGISTERED_LEN(sa, attr);
1699 if (length == 0) {
1700 length = hdr->sa_lengths[length_idx++];
1701 }
1702
1703 SA_ADD_BULK_ATTR(attr_desc, j, attr,
1704 NULL, (void *)
1705 (TOC_OFF(idx_tab->sa_idx_tab[attr]) +
1706 (uintptr_t)old_data[k]), length);
1707 }
1708 }
1709 if (k == 0 && hdl->sa_spill) {
1710 hdr = SA_GET_HDR(hdl, SA_SPILL);
1711 idx_tab = SA_IDX_TAB_GET(hdl, SA_SPILL);
1712 count = spill_attr_count;
1713 } else {
1714 break;
1715 }
1716 }
1717 if (action == SA_ADD) {
1718 length = SA_REGISTERED_LEN(sa, newattr);
1719 if (length == 0) {
1720 length = buflen;
1721 }
1722 SA_ADD_BULK_ATTR(attr_desc, j, newattr, locator,
1723 datastart, buflen);
1724 }
1725
1726 error = sa_build_layouts(hdl, attr_desc, attr_count, tx);
1727
1728 if (old_data[0])
1729 kmem_free(old_data[0], bonus_data_size);
1730 if (old_data[1])
1731 kmem_free(old_data[1], spill_data_size);
1732 kmem_free(attr_desc, sizeof (sa_bulk_attr_t) * attr_count);
1733
1734 return (error);
1735 }
1736
1737 static int
1738 sa_bulk_update_impl(sa_handle_t *hdl, sa_bulk_attr_t *bulk, int count,
1739 dmu_tx_t *tx)
1740 {
1741 int error;
1742 sa_os_t *sa = hdl->sa_os->os_sa;
1743 dmu_object_type_t bonustype;
1744
1745 bonustype = SA_BONUSTYPE_FROM_DB(SA_GET_DB(hdl, SA_BONUS));
1746
1747 ASSERT(hdl);
1748 ASSERT(MUTEX_HELD(&hdl->sa_lock));
1749
1750 /* sync out registration table if necessary */
1751 if (sa->sa_need_attr_registration)
1752 sa_attr_register_sync(hdl, tx);
1753
1754 error = sa_attr_op(hdl, bulk, count, SA_UPDATE, tx);
1755 if (error == 0 && !IS_SA_BONUSTYPE(bonustype) && sa->sa_update_cb)
1756 sa->sa_update_cb(hdl, tx);
1757
1758 return (error);
1759 }
1760
1761 /*
1762 * update or add new attribute
1763 */
1764 int
1765 sa_update(sa_handle_t *hdl, sa_attr_type_t type,
1766 void *buf, uint32_t buflen, dmu_tx_t *tx)
1767 {
1768 int error;
1769 sa_bulk_attr_t bulk;
1770
1771 bulk.sa_attr = type;
1772 bulk.sa_data_func = NULL;
1773 bulk.sa_length = buflen;
1774 bulk.sa_data = buf;
1775
1776 mutex_enter(&hdl->sa_lock);
1777 error = sa_bulk_update_impl(hdl, &bulk, 1, tx);
1778 mutex_exit(&hdl->sa_lock);
1779 return (error);
1780 }
1781
1782 int
1783 sa_update_from_cb(sa_handle_t *hdl, sa_attr_type_t attr,
1784 uint32_t buflen, sa_data_locator_t *locator, void *userdata, dmu_tx_t *tx)
1785 {
1786 int error;
1787 sa_bulk_attr_t bulk;
1788
1789 bulk.sa_attr = attr;
1790 bulk.sa_data = userdata;
1791 bulk.sa_data_func = locator;
1792 bulk.sa_length = buflen;
1793
1794 mutex_enter(&hdl->sa_lock);
1795 error = sa_bulk_update_impl(hdl, &bulk, 1, tx);
1796 mutex_exit(&hdl->sa_lock);
1797 return (error);
1798 }
1799
1800 /*
1801 * Return size of an attribute
1802 */
1803
1804 int
1805 sa_size(sa_handle_t *hdl, sa_attr_type_t attr, int *size)
1806 {
1807 sa_bulk_attr_t bulk;
1808 int error;
1809
1810 bulk.sa_data = NULL;
1811 bulk.sa_attr = attr;
1812 bulk.sa_data_func = NULL;
1813
1814 ASSERT(hdl);
1815 mutex_enter(&hdl->sa_lock);
1816 if ((error = sa_attr_op(hdl, &bulk, 1, SA_LOOKUP, NULL)) != 0) {
1817 mutex_exit(&hdl->sa_lock);
1818 return (error);
1819 }
1820 *size = bulk.sa_size;
1821
1822 mutex_exit(&hdl->sa_lock);
1823 return (0);
1824 }
1825
1826 int
1827 sa_bulk_lookup_locked(sa_handle_t *hdl, sa_bulk_attr_t *attrs, int count)
1828 {
1829 ASSERT(hdl);
1830 ASSERT(MUTEX_HELD(&hdl->sa_lock));
1831 return (sa_lookup_impl(hdl, attrs, count));
1832 }
1833
1834 int
1835 sa_bulk_lookup(sa_handle_t *hdl, sa_bulk_attr_t *attrs, int count)
1836 {
1837 int error;
1838
1839 ASSERT(hdl);
1840 mutex_enter(&hdl->sa_lock);
1841 error = sa_bulk_lookup_locked(hdl, attrs, count);
1842 mutex_exit(&hdl->sa_lock);
1843 return (error);
1844 }
1845
1846 int
1847 sa_bulk_update(sa_handle_t *hdl, sa_bulk_attr_t *attrs, int count, dmu_tx_t *tx)
1848 {
1849 int error;
1850
1851 ASSERT(hdl);
1852 mutex_enter(&hdl->sa_lock);
1853 error = sa_bulk_update_impl(hdl, attrs, count, tx);
1854 mutex_exit(&hdl->sa_lock);
1855 return (error);
1856 }
1857
1858 int
1859 sa_remove(sa_handle_t *hdl, sa_attr_type_t attr, dmu_tx_t *tx)
1860 {
1861 int error;
1862
1863 mutex_enter(&hdl->sa_lock);
1864 error = sa_modify_attrs(hdl, attr, SA_REMOVE, NULL,
1865 NULL, 0, tx);
1866 mutex_exit(&hdl->sa_lock);
1867 return (error);
1868 }
1869
1870 void
1871 sa_object_info(sa_handle_t *hdl, dmu_object_info_t *doi)
1872 {
1873 dmu_object_info_from_db((dmu_buf_t *)hdl->sa_bonus, doi);
1874 }
1875
1876 void
1877 sa_object_size(sa_handle_t *hdl, uint32_t *blksize, u_longlong_t *nblocks)
1878 {
1879 dmu_object_size_from_db((dmu_buf_t *)hdl->sa_bonus,
1880 blksize, nblocks);
1881 }
1882
1883 void
1884 sa_update_user(sa_handle_t *newhdl, sa_handle_t *oldhdl)
1885 {
1886 (void) dmu_buf_update_user((dmu_buf_t *)newhdl->sa_bonus,
1887 oldhdl, newhdl, NULL, sa_evict);
1888 oldhdl->sa_bonus = NULL;
1889 }
1890
1891 void
1892 sa_set_userp(sa_handle_t *hdl, void *ptr)
1893 {
1894 hdl->sa_userp = ptr;
1895 }
1896
1897 dmu_buf_t *
1898 sa_get_db(sa_handle_t *hdl)
1899 {
1900 return ((dmu_buf_t *)hdl->sa_bonus);
1901 }
1902
1903 void *
1904 sa_get_userdata(sa_handle_t *hdl)
1905 {
1906 return (hdl->sa_userp);
1907 }
1908
1909 void
1910 sa_register_update_callback_locked(objset_t *os, sa_update_cb_t *func)
1911 {
1912 ASSERT(MUTEX_HELD(&os->os_sa->sa_lock));
1913 os->os_sa->sa_update_cb = func;
1914 }
1915
1916 void
1917 sa_register_update_callback(objset_t *os, sa_update_cb_t *func)
1918 {
1919
1920 mutex_enter(&os->os_sa->sa_lock);
1921 sa_register_update_callback_locked(os, func);
1922 mutex_exit(&os->os_sa->sa_lock);
1923 }
1924
1925 uint64_t
1926 sa_handle_object(sa_handle_t *hdl)
1927 {
1928 return (hdl->sa_bonus->db_object);
1929 }
1930
1931 boolean_t
1932 sa_enabled(objset_t *os)
1933 {
1934 return (os->os_sa == NULL);
1935 }
1936
1937 int
1938 sa_set_sa_object(objset_t *os, uint64_t sa_object)
1939 {
1940 sa_os_t *sa = os->os_sa;
1941
1942 if (sa->sa_master_obj)
1943 return (1);
1944
1945 sa->sa_master_obj = sa_object;
1946
1947 return (0);
1948 }
1949
1950 int
1951 sa_hdrsize(void *arg)
1952 {
1953 sa_hdr_phys_t *hdr = arg;
1954
1955 return (SA_HDR_SIZE(hdr));
1956 }
1957
1958 void
1959 sa_handle_lock(sa_handle_t *hdl)
1960 {
1961 ASSERT(hdl);
1962 mutex_enter(&hdl->sa_lock);
1963 }
1964
1965 void
1966 sa_handle_unlock(sa_handle_t *hdl)
1967 {
1968 ASSERT(hdl);
1969 mutex_exit(&hdl->sa_lock);
1970 }