xref: /freebsd/sys/contrib/openzfs/module/zfs/sa.c (revision 7fdf597e96a02165cfe22ff357b857d5fa15ed8a)
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 https://opensource.org/licenses/CDDL-1.0.
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 /*
23  * Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
24  * Copyright (c) 2013, 2017 by Delphix. All rights reserved.
25  * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
26  * Copyright 2023 RackTop Systems, Inc.
27  */
28 
29 #include <sys/zfs_context.h>
30 #include <sys/types.h>
31 #include <sys/param.h>
32 #include <sys/sysmacros.h>
33 #include <sys/dmu.h>
34 #include <sys/dmu_impl.h>
35 #include <sys/dmu_objset.h>
36 #include <sys/dmu_tx.h>
37 #include <sys/dbuf.h>
38 #include <sys/dnode.h>
39 #include <sys/zap.h>
40 #include <sys/sa.h>
41 #include <sys/sunddi.h>
42 #include <sys/sa_impl.h>
43 #include <sys/errno.h>
44 #include <sys/zfs_context.h>
45 
46 #ifdef _KERNEL
47 #include <sys/zfs_znode.h>
48 #endif
49 
50 /*
51  * ZFS System attributes:
52  *
53  * A generic mechanism to allow for arbitrary attributes
54  * to be stored in a dnode.  The data will be stored in the bonus buffer of
55  * the dnode and if necessary a special "spill" block will be used to handle
56  * overflow situations.  The spill block will be sized to fit the data
57  * from 512 - 128K.  When a spill block is used the BP (blkptr_t) for the
58  * spill block is stored at the end of the current bonus buffer.  Any
59  * attributes that would be in the way of the blkptr_t will be relocated
60  * into the spill block.
61  *
62  * Attribute registration:
63  *
64  * Stored persistently on a per dataset basis
65  * a mapping between attribute "string" names and their actual attribute
66  * numeric values, length, and byteswap function.  The names are only used
67  * during registration.  All  attributes are known by their unique attribute
68  * id value.  If an attribute can have a variable size then the value
69  * 0 will be used to indicate this.
70  *
71  * Attribute Layout:
72  *
73  * Attribute layouts are a way to compactly store multiple attributes, but
74  * without taking the overhead associated with managing each attribute
75  * individually.  Since you will typically have the same set of attributes
76  * stored in the same order a single table will be used to represent that
77  * layout.  The ZPL for example will usually have only about 10 different
78  * layouts (regular files, device files, symlinks,
79  * regular files + scanstamp, files/dir with extended attributes, and then
80  * you have the possibility of all of those minus ACL, because it would
81  * be kicked out into the spill block)
82  *
83  * Layouts are simply an array of the attributes and their
84  * ordering i.e. [0, 1, 4, 5, 2]
85  *
86  * Each distinct layout is given a unique layout number and that is what's
87  * stored in the header at the beginning of the SA data buffer.
88  *
89  * A layout only covers a single dbuf (bonus or spill).  If a set of
90  * attributes is split up between the bonus buffer and a spill buffer then
91  * two different layouts will be used.  This allows us to byteswap the
92  * spill without looking at the bonus buffer and keeps the on disk format of
93  * the bonus and spill buffer the same.
94  *
95  * Adding a single attribute will cause the entire set of attributes to
96  * be rewritten and could result in a new layout number being constructed
97  * as part of the rewrite if no such layout exists for the new set of
98  * attributes.  The new attribute will be appended to the end of the already
99  * existing attributes.
100  *
101  * Both the attribute registration and attribute layout information are
102  * stored in normal ZAP attributes.  Their should be a small number of
103  * known layouts and the set of attributes is assumed to typically be quite
104  * small.
105  *
106  * The registered attributes and layout "table" information is maintained
107  * in core and a special "sa_os_t" is attached to the objset_t.
108  *
109  * A special interface is provided to allow for quickly applying
110  * a large set of attributes at once.  sa_replace_all_by_template() is
111  * used to set an array of attributes.  This is used by the ZPL when
112  * creating a brand new file.  The template that is passed into the function
113  * specifies the attribute, size for variable length attributes, location of
114  * data and special "data locator" function if the data isn't in a contiguous
115  * location.
116  *
117  * Byteswap implications:
118  *
119  * Since the SA attributes are not entirely self describing we can't do
120  * the normal byteswap processing.  The special ZAP layout attribute and
121  * attribute registration attributes define the byteswap function and the
122  * size of the attributes, unless it is variable sized.
123  * The normal ZFS byteswapping infrastructure assumes you don't need
124  * to read any objects in order to do the necessary byteswapping.  Whereas
125  * SA attributes can only be properly byteswapped if the dataset is opened
126  * and the layout/attribute ZAP attributes are available.  Because of this
127  * the SA attributes will be byteswapped when they are first accessed by
128  * the SA code that will read the SA data.
129  */
130 
131 typedef void (sa_iterfunc_t)(void *hdr, void *addr, sa_attr_type_t,
132     uint16_t length, int length_idx, boolean_t, void *userp);
133 
134 static int sa_build_index(sa_handle_t *hdl, sa_buf_type_t buftype);
135 static void sa_idx_tab_hold(objset_t *os, sa_idx_tab_t *idx_tab);
136 static sa_idx_tab_t *sa_find_idx_tab(objset_t *os, dmu_object_type_t bonustype,
137     sa_hdr_phys_t *hdr);
138 static void sa_idx_tab_rele(objset_t *os, void *arg);
139 static void sa_copy_data(sa_data_locator_t *func, void *start, void *target,
140     int buflen);
141 static int sa_modify_attrs(sa_handle_t *hdl, sa_attr_type_t newattr,
142     sa_data_op_t action, sa_data_locator_t *locator, void *datastart,
143     uint16_t buflen, dmu_tx_t *tx);
144 
145 static arc_byteswap_func_t sa_bswap_table[] = {
146 	byteswap_uint64_array,
147 	byteswap_uint32_array,
148 	byteswap_uint16_array,
149 	byteswap_uint8_array,
150 	zfs_acl_byteswap,
151 };
152 
153 #ifdef HAVE_EFFICIENT_UNALIGNED_ACCESS
154 #define	SA_COPY_DATA(f, s, t, l)				\
155 do {								\
156 	if (f == NULL) {					\
157 		if (l == 8) {					\
158 			*(uint64_t *)t = *(uint64_t *)s;	\
159 		} else if (l == 16) {				\
160 			*(uint64_t *)t = *(uint64_t *)s;	\
161 			*(uint64_t *)((uintptr_t)t + 8) =	\
162 			    *(uint64_t *)((uintptr_t)s + 8);	\
163 		} else {					\
164 			memcpy(t, s, l);				\
165 		}						\
166 	} else {						\
167 		sa_copy_data(f, s, t, l);			\
168 	}							\
169 } while (0)
170 #else
171 #define	SA_COPY_DATA(f, s, t, l)	sa_copy_data(f, s, t, l)
172 #endif
173 
174 /*
175  * This table is fixed and cannot be changed.  Its purpose is to
176  * allow the SA code to work with both old/new ZPL file systems.
177  * It contains the list of legacy attributes.  These attributes aren't
178  * stored in the "attribute" registry zap objects, since older ZPL file systems
179  * won't have the registry.  Only objsets of type ZFS_TYPE_FILESYSTEM will
180  * use this static table.
181  */
182 static const sa_attr_reg_t sa_legacy_attrs[] = {
183 	{"ZPL_ATIME", sizeof (uint64_t) * 2, SA_UINT64_ARRAY, 0},
184 	{"ZPL_MTIME", sizeof (uint64_t) * 2, SA_UINT64_ARRAY, 1},
185 	{"ZPL_CTIME", sizeof (uint64_t) * 2, SA_UINT64_ARRAY, 2},
186 	{"ZPL_CRTIME", sizeof (uint64_t) * 2, SA_UINT64_ARRAY, 3},
187 	{"ZPL_GEN", sizeof (uint64_t), SA_UINT64_ARRAY, 4},
188 	{"ZPL_MODE", sizeof (uint64_t), SA_UINT64_ARRAY, 5},
189 	{"ZPL_SIZE", sizeof (uint64_t), SA_UINT64_ARRAY, 6},
190 	{"ZPL_PARENT", sizeof (uint64_t), SA_UINT64_ARRAY, 7},
191 	{"ZPL_LINKS", sizeof (uint64_t), SA_UINT64_ARRAY, 8},
192 	{"ZPL_XATTR", sizeof (uint64_t), SA_UINT64_ARRAY, 9},
193 	{"ZPL_RDEV", sizeof (uint64_t), SA_UINT64_ARRAY, 10},
194 	{"ZPL_FLAGS", sizeof (uint64_t), SA_UINT64_ARRAY, 11},
195 	{"ZPL_UID", sizeof (uint64_t), SA_UINT64_ARRAY, 12},
196 	{"ZPL_GID", sizeof (uint64_t), SA_UINT64_ARRAY, 13},
197 	{"ZPL_PAD", sizeof (uint64_t) * 4, SA_UINT64_ARRAY, 14},
198 	{"ZPL_ZNODE_ACL", 88, SA_UINT8_ARRAY, 15},
199 };
200 
201 /*
202  * This is only used for objects of type DMU_OT_ZNODE
203  */
204 static const sa_attr_type_t sa_legacy_zpl_layout[] = {
205     0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
206 };
207 
208 /*
209  * Special dummy layout used for buffers with no attributes.
210  */
211 static const sa_attr_type_t sa_dummy_zpl_layout[] = { 0 };
212 
213 static const size_t sa_legacy_attr_count = ARRAY_SIZE(sa_legacy_attrs);
214 static kmem_cache_t *sa_cache = NULL;
215 
216 static int
217 sa_cache_constructor(void *buf, void *unused, int kmflag)
218 {
219 	(void) unused, (void) kmflag;
220 	sa_handle_t *hdl = buf;
221 
222 	mutex_init(&hdl->sa_lock, NULL, MUTEX_DEFAULT, NULL);
223 	return (0);
224 }
225 
226 static void
227 sa_cache_destructor(void *buf, void *unused)
228 {
229 	(void) unused;
230 	sa_handle_t *hdl = buf;
231 	mutex_destroy(&hdl->sa_lock);
232 }
233 
234 void
235 sa_cache_init(void)
236 {
237 	sa_cache = kmem_cache_create("sa_cache",
238 	    sizeof (sa_handle_t), 0, sa_cache_constructor,
239 	    sa_cache_destructor, NULL, NULL, NULL, KMC_RECLAIMABLE);
240 }
241 
242 void
243 sa_cache_fini(void)
244 {
245 	if (sa_cache)
246 		kmem_cache_destroy(sa_cache);
247 }
248 
249 static int
250 layout_num_compare(const void *arg1, const void *arg2)
251 {
252 	const sa_lot_t *node1 = (const sa_lot_t *)arg1;
253 	const sa_lot_t *node2 = (const sa_lot_t *)arg2;
254 
255 	return (TREE_CMP(node1->lot_num, node2->lot_num));
256 }
257 
258 static int
259 layout_hash_compare(const void *arg1, const void *arg2)
260 {
261 	const sa_lot_t *node1 = (const sa_lot_t *)arg1;
262 	const sa_lot_t *node2 = (const sa_lot_t *)arg2;
263 
264 	int cmp = TREE_CMP(node1->lot_hash, node2->lot_hash);
265 	if (likely(cmp))
266 		return (cmp);
267 
268 	return (TREE_CMP(node1->lot_instance, node2->lot_instance));
269 }
270 
271 static boolean_t
272 sa_layout_equal(sa_lot_t *tbf, sa_attr_type_t *attrs, int count)
273 {
274 	int i;
275 
276 	if (count != tbf->lot_attr_count)
277 		return (1);
278 
279 	for (i = 0; i != count; i++) {
280 		if (attrs[i] != tbf->lot_attrs[i])
281 			return (1);
282 	}
283 	return (0);
284 }
285 
286 #define	SA_ATTR_HASH(attr) (zfs_crc64_table[(-1ULL ^ attr) & 0xFF])
287 
288 static uint64_t
289 sa_layout_info_hash(const sa_attr_type_t *attrs, int attr_count)
290 {
291 	uint64_t crc = -1ULL;
292 
293 	for (int i = 0; i != attr_count; i++)
294 		crc ^= SA_ATTR_HASH(attrs[i]);
295 
296 	return (crc);
297 }
298 
299 static int
300 sa_get_spill(sa_handle_t *hdl)
301 {
302 	int rc;
303 	if (hdl->sa_spill == NULL) {
304 		if ((rc = dmu_spill_hold_existing(hdl->sa_bonus, NULL,
305 		    &hdl->sa_spill)) == 0)
306 			VERIFY(0 == sa_build_index(hdl, SA_SPILL));
307 	} else {
308 		rc = 0;
309 	}
310 
311 	return (rc);
312 }
313 
314 /*
315  * Main attribute lookup/update function
316  * returns 0 for success or non zero for failures
317  *
318  * Operates on bulk array, first failure will abort further processing
319  */
320 static int
321 sa_attr_op(sa_handle_t *hdl, sa_bulk_attr_t *bulk, int count,
322     sa_data_op_t data_op, dmu_tx_t *tx)
323 {
324 	sa_os_t *sa = hdl->sa_os->os_sa;
325 	int i;
326 	int error = 0;
327 	sa_buf_type_t buftypes;
328 
329 	buftypes = 0;
330 
331 	ASSERT(count > 0);
332 	for (i = 0; i != count; i++) {
333 		ASSERT(bulk[i].sa_attr <= hdl->sa_os->os_sa->sa_num_attrs);
334 
335 		bulk[i].sa_addr = NULL;
336 		/* First check the bonus buffer */
337 
338 		if (hdl->sa_bonus_tab && TOC_ATTR_PRESENT(
339 		    hdl->sa_bonus_tab->sa_idx_tab[bulk[i].sa_attr])) {
340 			SA_ATTR_INFO(sa, hdl->sa_bonus_tab,
341 			    SA_GET_HDR(hdl, SA_BONUS),
342 			    bulk[i].sa_attr, bulk[i], SA_BONUS, hdl);
343 			if (tx && !(buftypes & SA_BONUS)) {
344 				dmu_buf_will_dirty(hdl->sa_bonus, tx);
345 				buftypes |= SA_BONUS;
346 			}
347 		}
348 		if (bulk[i].sa_addr == NULL &&
349 		    ((error = sa_get_spill(hdl)) == 0)) {
350 			if (TOC_ATTR_PRESENT(
351 			    hdl->sa_spill_tab->sa_idx_tab[bulk[i].sa_attr])) {
352 				SA_ATTR_INFO(sa, hdl->sa_spill_tab,
353 				    SA_GET_HDR(hdl, SA_SPILL),
354 				    bulk[i].sa_attr, bulk[i], SA_SPILL, hdl);
355 				if (tx && !(buftypes & SA_SPILL) &&
356 				    bulk[i].sa_size == bulk[i].sa_length) {
357 					dmu_buf_will_dirty(hdl->sa_spill, tx);
358 					buftypes |= SA_SPILL;
359 				}
360 			}
361 		}
362 		if (error && error != ENOENT) {
363 			return ((error == ECKSUM) ? EIO : error);
364 		}
365 
366 		switch (data_op) {
367 		case SA_LOOKUP:
368 			if (bulk[i].sa_addr == NULL)
369 				return (SET_ERROR(ENOENT));
370 			if (bulk[i].sa_data) {
371 				SA_COPY_DATA(bulk[i].sa_data_func,
372 				    bulk[i].sa_addr, bulk[i].sa_data,
373 				    MIN(bulk[i].sa_size, bulk[i].sa_length));
374 			}
375 			continue;
376 
377 		case SA_UPDATE:
378 			/* existing rewrite of attr */
379 			if (bulk[i].sa_addr &&
380 			    bulk[i].sa_size == bulk[i].sa_length) {
381 				SA_COPY_DATA(bulk[i].sa_data_func,
382 				    bulk[i].sa_data, bulk[i].sa_addr,
383 				    bulk[i].sa_length);
384 				continue;
385 			} else if (bulk[i].sa_addr) { /* attr size change */
386 				error = sa_modify_attrs(hdl, bulk[i].sa_attr,
387 				    SA_REPLACE, bulk[i].sa_data_func,
388 				    bulk[i].sa_data, bulk[i].sa_length, tx);
389 			} else { /* adding new attribute */
390 				error = sa_modify_attrs(hdl, bulk[i].sa_attr,
391 				    SA_ADD, bulk[i].sa_data_func,
392 				    bulk[i].sa_data, bulk[i].sa_length, tx);
393 			}
394 			if (error)
395 				return (error);
396 			break;
397 		default:
398 			break;
399 		}
400 	}
401 	return (error);
402 }
403 
404 static sa_lot_t *
405 sa_add_layout_entry(objset_t *os, const sa_attr_type_t *attrs, int attr_count,
406     uint64_t lot_num, uint64_t hash, boolean_t zapadd, dmu_tx_t *tx)
407 {
408 	sa_os_t *sa = os->os_sa;
409 	sa_lot_t *tb, *findtb;
410 	int i;
411 	avl_index_t loc;
412 
413 	ASSERT(MUTEX_HELD(&sa->sa_lock));
414 	tb = kmem_zalloc(sizeof (sa_lot_t), KM_SLEEP);
415 	tb->lot_attr_count = attr_count;
416 	tb->lot_attrs = kmem_alloc(sizeof (sa_attr_type_t) * attr_count,
417 	    KM_SLEEP);
418 	memcpy(tb->lot_attrs, attrs, sizeof (sa_attr_type_t) * attr_count);
419 	tb->lot_num = lot_num;
420 	tb->lot_hash = hash;
421 	tb->lot_instance = 0;
422 
423 	if (zapadd) {
424 		char attr_name[8];
425 
426 		if (sa->sa_layout_attr_obj == 0) {
427 			sa->sa_layout_attr_obj = zap_create_link(os,
428 			    DMU_OT_SA_ATTR_LAYOUTS,
429 			    sa->sa_master_obj, SA_LAYOUTS, tx);
430 		}
431 
432 		(void) snprintf(attr_name, sizeof (attr_name),
433 		    "%d", (int)lot_num);
434 		VERIFY(0 == zap_update(os, os->os_sa->sa_layout_attr_obj,
435 		    attr_name, 2, attr_count, attrs, tx));
436 	}
437 
438 	list_create(&tb->lot_idx_tab, sizeof (sa_idx_tab_t),
439 	    offsetof(sa_idx_tab_t, sa_next));
440 
441 	for (i = 0; i != attr_count; i++) {
442 		if (sa->sa_attr_table[tb->lot_attrs[i]].sa_length == 0)
443 			tb->lot_var_sizes++;
444 	}
445 
446 	avl_add(&sa->sa_layout_num_tree, tb);
447 
448 	/* verify we don't have a hash collision */
449 	if ((findtb = avl_find(&sa->sa_layout_hash_tree, tb, &loc)) != NULL) {
450 		for (; findtb && findtb->lot_hash == hash;
451 		    findtb = AVL_NEXT(&sa->sa_layout_hash_tree, findtb)) {
452 			if (findtb->lot_instance != tb->lot_instance)
453 				break;
454 			tb->lot_instance++;
455 		}
456 	}
457 	avl_add(&sa->sa_layout_hash_tree, tb);
458 	return (tb);
459 }
460 
461 static void
462 sa_find_layout(objset_t *os, uint64_t hash, sa_attr_type_t *attrs,
463     int count, dmu_tx_t *tx, sa_lot_t **lot)
464 {
465 	sa_lot_t *tb, tbsearch;
466 	avl_index_t loc;
467 	sa_os_t *sa = os->os_sa;
468 	boolean_t found = B_FALSE;
469 
470 	mutex_enter(&sa->sa_lock);
471 	tbsearch.lot_hash = hash;
472 	tbsearch.lot_instance = 0;
473 	tb = avl_find(&sa->sa_layout_hash_tree, &tbsearch, &loc);
474 	if (tb) {
475 		for (; tb && tb->lot_hash == hash;
476 		    tb = AVL_NEXT(&sa->sa_layout_hash_tree, tb)) {
477 			if (sa_layout_equal(tb, attrs, count) == 0) {
478 				found = B_TRUE;
479 				break;
480 			}
481 		}
482 	}
483 	if (!found) {
484 		tb = sa_add_layout_entry(os, attrs, count,
485 		    avl_numnodes(&sa->sa_layout_num_tree), hash, B_TRUE, tx);
486 	}
487 	mutex_exit(&sa->sa_lock);
488 	*lot = tb;
489 }
490 
491 static int
492 sa_resize_spill(sa_handle_t *hdl, uint32_t size, dmu_tx_t *tx)
493 {
494 	int error;
495 	uint32_t blocksize;
496 
497 	if (size == 0) {
498 		blocksize = SPA_MINBLOCKSIZE;
499 	} else if (size > SPA_OLD_MAXBLOCKSIZE) {
500 		ASSERT(0);
501 		return (SET_ERROR(EFBIG));
502 	} else {
503 		blocksize = P2ROUNDUP_TYPED(size, SPA_MINBLOCKSIZE, uint32_t);
504 	}
505 
506 	error = dbuf_spill_set_blksz(hdl->sa_spill, blocksize, tx);
507 	ASSERT(error == 0);
508 	return (error);
509 }
510 
511 static void
512 sa_copy_data(sa_data_locator_t *func, void *datastart, void *target, int buflen)
513 {
514 	if (func == NULL) {
515 		memcpy(target, datastart, buflen);
516 	} else {
517 		boolean_t start;
518 		int bytes;
519 		void *dataptr;
520 		void *saptr = target;
521 		uint32_t length;
522 
523 		start = B_TRUE;
524 		bytes = 0;
525 		while (bytes < buflen) {
526 			func(&dataptr, &length, buflen, start, datastart);
527 			memcpy(saptr, dataptr, length);
528 			saptr = (void *)((caddr_t)saptr + length);
529 			bytes += length;
530 			start = B_FALSE;
531 		}
532 	}
533 }
534 
535 /*
536  * Determine several different values pertaining to system attribute
537  * buffers.
538  *
539  * Return the size of the sa_hdr_phys_t header for the buffer. Each
540  * variable length attribute except the first contributes two bytes to
541  * the header size, which is then rounded up to an 8-byte boundary.
542  *
543  * The following output parameters are also computed.
544  *
545  *  index - The index of the first attribute in attr_desc that will
546  *  spill over. Only valid if will_spill is set.
547  *
548  *  total - The total number of bytes of all system attributes described
549  *  in attr_desc.
550  *
551  *  will_spill - Set when spilling is necessary. It is only set when
552  *  the buftype is SA_BONUS.
553  */
554 static int
555 sa_find_sizes(sa_os_t *sa, sa_bulk_attr_t *attr_desc, int attr_count,
556     dmu_buf_t *db, sa_buf_type_t buftype, int full_space, int *index,
557     int *total, boolean_t *will_spill)
558 {
559 	int var_size_count = 0;
560 	int i;
561 	int hdrsize;
562 	int extra_hdrsize;
563 
564 	if (buftype == SA_BONUS && sa->sa_force_spill) {
565 		*total = 0;
566 		*index = 0;
567 		*will_spill = B_TRUE;
568 		return (0);
569 	}
570 
571 	*index = -1;
572 	*total = 0;
573 	*will_spill = B_FALSE;
574 
575 	extra_hdrsize = 0;
576 	hdrsize = (SA_BONUSTYPE_FROM_DB(db) == DMU_OT_ZNODE) ? 0 :
577 	    sizeof (sa_hdr_phys_t);
578 
579 	ASSERT(IS_P2ALIGNED(full_space, 8));
580 
581 	for (i = 0; i != attr_count; i++) {
582 		boolean_t is_var_sz, might_spill_here;
583 		int tmp_hdrsize;
584 
585 		*total = P2ROUNDUP(*total, 8);
586 		*total += attr_desc[i].sa_length;
587 		if (*will_spill)
588 			continue;
589 
590 		is_var_sz = (SA_REGISTERED_LEN(sa, attr_desc[i].sa_attr) == 0);
591 		if (is_var_sz)
592 			var_size_count++;
593 
594 		/*
595 		 * Calculate what the SA header size would be if this
596 		 * attribute doesn't spill.
597 		 */
598 		tmp_hdrsize = hdrsize + ((is_var_sz && var_size_count > 1) ?
599 		    sizeof (uint16_t) : 0);
600 
601 		/*
602 		 * Check whether this attribute spans into the space
603 		 * that would be used by the spill block pointer should
604 		 * a spill block be needed.
605 		 */
606 		might_spill_here =
607 		    buftype == SA_BONUS && *index == -1 &&
608 		    (*total + P2ROUNDUP(tmp_hdrsize, 8)) >
609 		    (full_space - sizeof (blkptr_t));
610 
611 		if (is_var_sz && var_size_count > 1) {
612 			if (buftype == SA_SPILL ||
613 			    tmp_hdrsize + *total < full_space) {
614 				/*
615 				 * Record the extra header size in case this
616 				 * increase needs to be reversed due to
617 				 * spill-over.
618 				 */
619 				hdrsize = tmp_hdrsize;
620 				if (*index != -1 || might_spill_here)
621 					extra_hdrsize += sizeof (uint16_t);
622 			} else {
623 				ASSERT(buftype == SA_BONUS);
624 				if (*index == -1)
625 					*index = i;
626 				*will_spill = B_TRUE;
627 				continue;
628 			}
629 		}
630 
631 		/*
632 		 * Store index of where spill *could* occur. Then
633 		 * continue to count the remaining attribute sizes. The
634 		 * sum is used later for sizing bonus and spill buffer.
635 		 */
636 		if (might_spill_here)
637 			*index = i;
638 
639 		if ((*total + P2ROUNDUP(hdrsize, 8)) > full_space &&
640 		    buftype == SA_BONUS)
641 			*will_spill = B_TRUE;
642 	}
643 
644 	if (*will_spill)
645 		hdrsize -= extra_hdrsize;
646 
647 	hdrsize = P2ROUNDUP(hdrsize, 8);
648 	return (hdrsize);
649 }
650 
651 #define	BUF_SPACE_NEEDED(total, header) (total + header)
652 
653 /*
654  * Find layout that corresponds to ordering of attributes
655  * If not found a new layout number is created and added to
656  * persistent layout tables.
657  */
658 static int
659 sa_build_layouts(sa_handle_t *hdl, sa_bulk_attr_t *attr_desc, int attr_count,
660     dmu_tx_t *tx)
661 {
662 	sa_os_t *sa = hdl->sa_os->os_sa;
663 	uint64_t hash;
664 	sa_buf_type_t buftype;
665 	sa_hdr_phys_t *sahdr;
666 	void *data_start;
667 	sa_attr_type_t *attrs, *attrs_start;
668 	int i, lot_count;
669 	int dnodesize;
670 	int spill_idx;
671 	int hdrsize;
672 	int spillhdrsize = 0;
673 	int used;
674 	dmu_object_type_t bonustype;
675 	sa_lot_t *lot;
676 	int len_idx;
677 	int spill_used;
678 	int bonuslen;
679 	boolean_t spilling;
680 
681 	dmu_buf_will_dirty(hdl->sa_bonus, tx);
682 	bonustype = SA_BONUSTYPE_FROM_DB(hdl->sa_bonus);
683 	dmu_object_dnsize_from_db(hdl->sa_bonus, &dnodesize);
684 	bonuslen = DN_BONUS_SIZE(dnodesize);
685 
686 	/* first determine bonus header size and sum of all attributes */
687 	hdrsize = sa_find_sizes(sa, attr_desc, attr_count, hdl->sa_bonus,
688 	    SA_BONUS, bonuslen, &spill_idx, &used, &spilling);
689 
690 	if (used > SPA_OLD_MAXBLOCKSIZE)
691 		return (SET_ERROR(EFBIG));
692 
693 	VERIFY0(dmu_set_bonus(hdl->sa_bonus, spilling ?
694 	    MIN(bonuslen - sizeof (blkptr_t), used + hdrsize) :
695 	    used + hdrsize, tx));
696 
697 	ASSERT((bonustype == DMU_OT_ZNODE && spilling == 0) ||
698 	    bonustype == DMU_OT_SA);
699 
700 	/* setup and size spill buffer when needed */
701 	if (spilling) {
702 		boolean_t dummy;
703 
704 		if (hdl->sa_spill == NULL) {
705 			VERIFY(dmu_spill_hold_by_bonus(hdl->sa_bonus, 0, NULL,
706 			    &hdl->sa_spill) == 0);
707 		}
708 		dmu_buf_will_dirty(hdl->sa_spill, tx);
709 
710 		spillhdrsize = sa_find_sizes(sa, &attr_desc[spill_idx],
711 		    attr_count - spill_idx, hdl->sa_spill, SA_SPILL,
712 		    hdl->sa_spill->db_size, &i, &spill_used, &dummy);
713 
714 		if (spill_used > SPA_OLD_MAXBLOCKSIZE)
715 			return (SET_ERROR(EFBIG));
716 
717 		if (BUF_SPACE_NEEDED(spill_used, spillhdrsize) >
718 		    hdl->sa_spill->db_size)
719 			VERIFY(0 == sa_resize_spill(hdl,
720 			    BUF_SPACE_NEEDED(spill_used, spillhdrsize), tx));
721 	}
722 
723 	/* setup starting pointers to lay down data */
724 	data_start = (void *)((uintptr_t)hdl->sa_bonus->db_data + hdrsize);
725 	sahdr = (sa_hdr_phys_t *)hdl->sa_bonus->db_data;
726 	buftype = SA_BONUS;
727 
728 	attrs_start = attrs = kmem_alloc(sizeof (sa_attr_type_t) * attr_count,
729 	    KM_SLEEP);
730 	lot_count = 0;
731 
732 	for (i = 0, len_idx = 0, hash = -1ULL; i != attr_count; i++) {
733 		uint16_t length;
734 
735 		ASSERT(IS_P2ALIGNED(data_start, 8));
736 		attrs[i] = attr_desc[i].sa_attr;
737 		length = SA_REGISTERED_LEN(sa, attrs[i]);
738 		if (length == 0)
739 			length = attr_desc[i].sa_length;
740 
741 		if (spilling && i == spill_idx) { /* switch to spill buffer */
742 			VERIFY(bonustype == DMU_OT_SA);
743 			if (buftype == SA_BONUS && !sa->sa_force_spill) {
744 				sa_find_layout(hdl->sa_os, hash, attrs_start,
745 				    lot_count, tx, &lot);
746 				SA_SET_HDR(sahdr, lot->lot_num, hdrsize);
747 			}
748 
749 			buftype = SA_SPILL;
750 			hash = -1ULL;
751 			len_idx = 0;
752 
753 			sahdr = (sa_hdr_phys_t *)hdl->sa_spill->db_data;
754 			sahdr->sa_magic = SA_MAGIC;
755 			data_start = (void *)((uintptr_t)sahdr +
756 			    spillhdrsize);
757 			attrs_start = &attrs[i];
758 			lot_count = 0;
759 		}
760 		hash ^= SA_ATTR_HASH(attrs[i]);
761 		attr_desc[i].sa_addr = data_start;
762 		attr_desc[i].sa_size = length;
763 		SA_COPY_DATA(attr_desc[i].sa_data_func, attr_desc[i].sa_data,
764 		    data_start, length);
765 		if (sa->sa_attr_table[attrs[i]].sa_length == 0) {
766 			sahdr->sa_lengths[len_idx++] = length;
767 		}
768 		data_start = (void *)P2ROUNDUP(((uintptr_t)data_start +
769 		    length), 8);
770 		lot_count++;
771 	}
772 
773 	sa_find_layout(hdl->sa_os, hash, attrs_start, lot_count, tx, &lot);
774 
775 	/*
776 	 * Verify that old znodes always have layout number 0.
777 	 * Must be DMU_OT_SA for arbitrary layouts
778 	 */
779 	VERIFY((bonustype == DMU_OT_ZNODE && lot->lot_num == 0) ||
780 	    (bonustype == DMU_OT_SA && lot->lot_num > 1));
781 
782 	if (bonustype == DMU_OT_SA) {
783 		SA_SET_HDR(sahdr, lot->lot_num,
784 		    buftype == SA_BONUS ? hdrsize : spillhdrsize);
785 	}
786 
787 	kmem_free(attrs, sizeof (sa_attr_type_t) * attr_count);
788 	if (hdl->sa_bonus_tab) {
789 		sa_idx_tab_rele(hdl->sa_os, hdl->sa_bonus_tab);
790 		hdl->sa_bonus_tab = NULL;
791 	}
792 	if (!sa->sa_force_spill)
793 		VERIFY(0 == sa_build_index(hdl, SA_BONUS));
794 	if (hdl->sa_spill) {
795 		sa_idx_tab_rele(hdl->sa_os, hdl->sa_spill_tab);
796 		if (!spilling) {
797 			/*
798 			 * remove spill block that is no longer needed.
799 			 */
800 			dmu_buf_rele(hdl->sa_spill, NULL);
801 			hdl->sa_spill = NULL;
802 			hdl->sa_spill_tab = NULL;
803 			VERIFY(0 == dmu_rm_spill(hdl->sa_os,
804 			    sa_handle_object(hdl), tx));
805 		} else {
806 			VERIFY(0 == sa_build_index(hdl, SA_SPILL));
807 		}
808 	}
809 
810 	return (0);
811 }
812 
813 static void
814 sa_free_attr_table(sa_os_t *sa)
815 {
816 	int i;
817 
818 	if (sa->sa_attr_table == NULL)
819 		return;
820 
821 	for (i = 0; i != sa->sa_num_attrs; i++) {
822 		if (sa->sa_attr_table[i].sa_name)
823 			kmem_free(sa->sa_attr_table[i].sa_name,
824 			    strlen(sa->sa_attr_table[i].sa_name) + 1);
825 	}
826 
827 	kmem_free(sa->sa_attr_table,
828 	    sizeof (sa_attr_table_t) * sa->sa_num_attrs);
829 
830 	sa->sa_attr_table = NULL;
831 }
832 
833 static int
834 sa_attr_table_setup(objset_t *os, const sa_attr_reg_t *reg_attrs, int count)
835 {
836 	sa_os_t *sa = os->os_sa;
837 	uint64_t sa_attr_count = 0;
838 	uint64_t sa_reg_count = 0;
839 	int error = 0;
840 	uint64_t attr_value;
841 	sa_attr_table_t *tb;
842 	zap_cursor_t zc;
843 	zap_attribute_t *za;
844 	int registered_count = 0;
845 	int i;
846 	dmu_objset_type_t ostype = dmu_objset_type(os);
847 
848 	sa->sa_user_table =
849 	    kmem_zalloc(count * sizeof (sa_attr_type_t), KM_SLEEP);
850 	sa->sa_user_table_sz = count * sizeof (sa_attr_type_t);
851 
852 	if (sa->sa_reg_attr_obj != 0) {
853 		error = zap_count(os, sa->sa_reg_attr_obj,
854 		    &sa_attr_count);
855 
856 		/*
857 		 * Make sure we retrieved a count and that it isn't zero
858 		 */
859 		if (error || (error == 0 && sa_attr_count == 0)) {
860 			if (error == 0)
861 				error = SET_ERROR(EINVAL);
862 			goto bail;
863 		}
864 		sa_reg_count = sa_attr_count;
865 	}
866 
867 	if (ostype == DMU_OST_ZFS && sa_attr_count == 0)
868 		sa_attr_count += sa_legacy_attr_count;
869 
870 	/* Allocate attribute numbers for attributes that aren't registered */
871 	for (i = 0; i != count; i++) {
872 		boolean_t found = B_FALSE;
873 		int j;
874 
875 		if (ostype == DMU_OST_ZFS) {
876 			for (j = 0; j != sa_legacy_attr_count; j++) {
877 				if (strcmp(reg_attrs[i].sa_name,
878 				    sa_legacy_attrs[j].sa_name) == 0) {
879 					sa->sa_user_table[i] =
880 					    sa_legacy_attrs[j].sa_attr;
881 					found = B_TRUE;
882 				}
883 			}
884 		}
885 		if (found)
886 			continue;
887 
888 		if (sa->sa_reg_attr_obj)
889 			error = zap_lookup(os, sa->sa_reg_attr_obj,
890 			    reg_attrs[i].sa_name, 8, 1, &attr_value);
891 		else
892 			error = SET_ERROR(ENOENT);
893 		switch (error) {
894 		case ENOENT:
895 			sa->sa_user_table[i] = (sa_attr_type_t)sa_attr_count;
896 			sa_attr_count++;
897 			break;
898 		case 0:
899 			sa->sa_user_table[i] = ATTR_NUM(attr_value);
900 			break;
901 		default:
902 			goto bail;
903 		}
904 	}
905 
906 	sa->sa_num_attrs = sa_attr_count;
907 	tb = sa->sa_attr_table =
908 	    kmem_zalloc(sizeof (sa_attr_table_t) * sa_attr_count, KM_SLEEP);
909 
910 	/*
911 	 * Attribute table is constructed from requested attribute list,
912 	 * previously foreign registered attributes, and also the legacy
913 	 * ZPL set of attributes.
914 	 */
915 
916 	if (sa->sa_reg_attr_obj) {
917 		za = zap_attribute_alloc();
918 		for (zap_cursor_init(&zc, os, sa->sa_reg_attr_obj);
919 		    (error = zap_cursor_retrieve(&zc, za)) == 0;
920 		    zap_cursor_advance(&zc)) {
921 			uint64_t value;
922 			value  = za->za_first_integer;
923 
924 			registered_count++;
925 			tb[ATTR_NUM(value)].sa_attr = ATTR_NUM(value);
926 			tb[ATTR_NUM(value)].sa_length = ATTR_LENGTH(value);
927 			tb[ATTR_NUM(value)].sa_byteswap = ATTR_BSWAP(value);
928 			tb[ATTR_NUM(value)].sa_registered = B_TRUE;
929 
930 			if (tb[ATTR_NUM(value)].sa_name) {
931 				continue;
932 			}
933 			tb[ATTR_NUM(value)].sa_name =
934 			    kmem_zalloc(strlen(za->za_name) +1, KM_SLEEP);
935 			(void) strlcpy(tb[ATTR_NUM(value)].sa_name, za->za_name,
936 			    strlen(za->za_name) +1);
937 		}
938 		zap_cursor_fini(&zc);
939 		zap_attribute_free(za);
940 		/*
941 		 * Make sure we processed the correct number of registered
942 		 * attributes
943 		 */
944 		if (registered_count != sa_reg_count) {
945 			ASSERT(error != 0);
946 			goto bail;
947 		}
948 
949 	}
950 
951 	if (ostype == DMU_OST_ZFS) {
952 		for (i = 0; i != sa_legacy_attr_count; i++) {
953 			if (tb[i].sa_name)
954 				continue;
955 			tb[i].sa_attr = sa_legacy_attrs[i].sa_attr;
956 			tb[i].sa_length = sa_legacy_attrs[i].sa_length;
957 			tb[i].sa_byteswap = sa_legacy_attrs[i].sa_byteswap;
958 			tb[i].sa_registered = B_FALSE;
959 			tb[i].sa_name =
960 			    kmem_zalloc(strlen(sa_legacy_attrs[i].sa_name) +1,
961 			    KM_SLEEP);
962 			(void) strlcpy(tb[i].sa_name,
963 			    sa_legacy_attrs[i].sa_name,
964 			    strlen(sa_legacy_attrs[i].sa_name) + 1);
965 		}
966 	}
967 
968 	for (i = 0; i != count; i++) {
969 		sa_attr_type_t attr_id;
970 
971 		attr_id = sa->sa_user_table[i];
972 		if (tb[attr_id].sa_name)
973 			continue;
974 
975 		tb[attr_id].sa_length = reg_attrs[i].sa_length;
976 		tb[attr_id].sa_byteswap = reg_attrs[i].sa_byteswap;
977 		tb[attr_id].sa_attr = attr_id;
978 		tb[attr_id].sa_name =
979 		    kmem_zalloc(strlen(reg_attrs[i].sa_name) + 1, KM_SLEEP);
980 		(void) strlcpy(tb[attr_id].sa_name, reg_attrs[i].sa_name,
981 		    strlen(reg_attrs[i].sa_name) + 1);
982 	}
983 
984 	sa->sa_need_attr_registration =
985 	    (sa_attr_count != registered_count);
986 
987 	return (0);
988 bail:
989 	kmem_free(sa->sa_user_table, count * sizeof (sa_attr_type_t));
990 	sa->sa_user_table = NULL;
991 	sa_free_attr_table(sa);
992 	ASSERT(error != 0);
993 	return (error);
994 }
995 
996 int
997 sa_setup(objset_t *os, uint64_t sa_obj, const sa_attr_reg_t *reg_attrs,
998     int count, sa_attr_type_t **user_table)
999 {
1000 	zap_cursor_t zc;
1001 	zap_attribute_t *za;
1002 	sa_os_t *sa;
1003 	dmu_objset_type_t ostype = dmu_objset_type(os);
1004 	sa_attr_type_t *tb;
1005 	int error;
1006 
1007 	mutex_enter(&os->os_user_ptr_lock);
1008 	if (os->os_sa) {
1009 		mutex_enter(&os->os_sa->sa_lock);
1010 		mutex_exit(&os->os_user_ptr_lock);
1011 		tb = os->os_sa->sa_user_table;
1012 		mutex_exit(&os->os_sa->sa_lock);
1013 		*user_table = tb;
1014 		return (0);
1015 	}
1016 
1017 	sa = kmem_zalloc(sizeof (sa_os_t), KM_SLEEP);
1018 	mutex_init(&sa->sa_lock, NULL, MUTEX_NOLOCKDEP, NULL);
1019 	sa->sa_master_obj = sa_obj;
1020 
1021 	os->os_sa = sa;
1022 	mutex_enter(&sa->sa_lock);
1023 	mutex_exit(&os->os_user_ptr_lock);
1024 	avl_create(&sa->sa_layout_num_tree, layout_num_compare,
1025 	    sizeof (sa_lot_t), offsetof(sa_lot_t, lot_num_node));
1026 	avl_create(&sa->sa_layout_hash_tree, layout_hash_compare,
1027 	    sizeof (sa_lot_t), offsetof(sa_lot_t, lot_hash_node));
1028 
1029 	if (sa_obj) {
1030 		error = zap_lookup(os, sa_obj, SA_LAYOUTS,
1031 		    8, 1, &sa->sa_layout_attr_obj);
1032 		if (error != 0 && error != ENOENT)
1033 			goto fail;
1034 		error = zap_lookup(os, sa_obj, SA_REGISTRY,
1035 		    8, 1, &sa->sa_reg_attr_obj);
1036 		if (error != 0 && error != ENOENT)
1037 			goto fail;
1038 	}
1039 
1040 	if ((error = sa_attr_table_setup(os, reg_attrs, count)) != 0)
1041 		goto fail;
1042 
1043 	if (sa->sa_layout_attr_obj != 0) {
1044 		uint64_t layout_count;
1045 
1046 		error = zap_count(os, sa->sa_layout_attr_obj,
1047 		    &layout_count);
1048 
1049 		/*
1050 		 * Layout number count should be > 0
1051 		 */
1052 		if (error || (error == 0 && layout_count == 0)) {
1053 			if (error == 0)
1054 				error = SET_ERROR(EINVAL);
1055 			goto fail;
1056 		}
1057 
1058 		za = zap_attribute_alloc();
1059 		for (zap_cursor_init(&zc, os, sa->sa_layout_attr_obj);
1060 		    (error = zap_cursor_retrieve(&zc, za)) == 0;
1061 		    zap_cursor_advance(&zc)) {
1062 			sa_attr_type_t *lot_attrs;
1063 			uint64_t lot_num;
1064 
1065 			lot_attrs = kmem_zalloc(sizeof (sa_attr_type_t) *
1066 			    za->za_num_integers, KM_SLEEP);
1067 
1068 			if ((error = (zap_lookup(os, sa->sa_layout_attr_obj,
1069 			    za->za_name, 2, za->za_num_integers,
1070 			    lot_attrs))) != 0) {
1071 				kmem_free(lot_attrs, sizeof (sa_attr_type_t) *
1072 				    za->za_num_integers);
1073 				break;
1074 			}
1075 			VERIFY0(ddi_strtoull(za->za_name, NULL, 10,
1076 			    (unsigned long long *)&lot_num));
1077 
1078 			(void) sa_add_layout_entry(os, lot_attrs,
1079 			    za->za_num_integers, lot_num,
1080 			    sa_layout_info_hash(lot_attrs,
1081 			    za->za_num_integers), B_FALSE, NULL);
1082 			kmem_free(lot_attrs, sizeof (sa_attr_type_t) *
1083 			    za->za_num_integers);
1084 		}
1085 		zap_cursor_fini(&zc);
1086 		zap_attribute_free(za);
1087 
1088 		/*
1089 		 * Make sure layout count matches number of entries added
1090 		 * to AVL tree
1091 		 */
1092 		if (avl_numnodes(&sa->sa_layout_num_tree) != layout_count) {
1093 			ASSERT(error != 0);
1094 			goto fail;
1095 		}
1096 	}
1097 
1098 	/* Add special layout number for old ZNODES */
1099 	if (ostype == DMU_OST_ZFS) {
1100 		(void) sa_add_layout_entry(os, sa_legacy_zpl_layout,
1101 		    sa_legacy_attr_count, 0,
1102 		    sa_layout_info_hash(sa_legacy_zpl_layout,
1103 		    sa_legacy_attr_count), B_FALSE, NULL);
1104 
1105 		(void) sa_add_layout_entry(os, sa_dummy_zpl_layout, 0, 1,
1106 		    0, B_FALSE, NULL);
1107 	}
1108 	*user_table = os->os_sa->sa_user_table;
1109 	mutex_exit(&sa->sa_lock);
1110 	return (0);
1111 fail:
1112 	os->os_sa = NULL;
1113 	sa_free_attr_table(sa);
1114 	if (sa->sa_user_table)
1115 		kmem_free(sa->sa_user_table, sa->sa_user_table_sz);
1116 	mutex_exit(&sa->sa_lock);
1117 	avl_destroy(&sa->sa_layout_hash_tree);
1118 	avl_destroy(&sa->sa_layout_num_tree);
1119 	mutex_destroy(&sa->sa_lock);
1120 	kmem_free(sa, sizeof (sa_os_t));
1121 	return ((error == ECKSUM) ? EIO : error);
1122 }
1123 
1124 void
1125 sa_tear_down(objset_t *os)
1126 {
1127 	sa_os_t *sa = os->os_sa;
1128 	sa_lot_t *layout;
1129 	void *cookie;
1130 
1131 	kmem_free(sa->sa_user_table, sa->sa_user_table_sz);
1132 
1133 	/* Free up attr table */
1134 
1135 	sa_free_attr_table(sa);
1136 
1137 	cookie = NULL;
1138 	while ((layout =
1139 	    avl_destroy_nodes(&sa->sa_layout_hash_tree, &cookie))) {
1140 		sa_idx_tab_t *tab;
1141 		while ((tab = list_head(&layout->lot_idx_tab))) {
1142 			ASSERT(zfs_refcount_count(&tab->sa_refcount));
1143 			sa_idx_tab_rele(os, tab);
1144 		}
1145 	}
1146 
1147 	cookie = NULL;
1148 	while ((layout = avl_destroy_nodes(&sa->sa_layout_num_tree, &cookie))) {
1149 		kmem_free(layout->lot_attrs,
1150 		    sizeof (sa_attr_type_t) * layout->lot_attr_count);
1151 		kmem_free(layout, sizeof (sa_lot_t));
1152 	}
1153 
1154 	avl_destroy(&sa->sa_layout_hash_tree);
1155 	avl_destroy(&sa->sa_layout_num_tree);
1156 	mutex_destroy(&sa->sa_lock);
1157 
1158 	kmem_free(sa, sizeof (sa_os_t));
1159 	os->os_sa = NULL;
1160 }
1161 
1162 static void
1163 sa_build_idx_tab(void *hdr, void *attr_addr, sa_attr_type_t attr,
1164     uint16_t length, int length_idx, boolean_t var_length, void *userp)
1165 {
1166 	sa_idx_tab_t *idx_tab = userp;
1167 
1168 	if (var_length) {
1169 		ASSERT(idx_tab->sa_variable_lengths);
1170 		idx_tab->sa_variable_lengths[length_idx] = length;
1171 	}
1172 	TOC_ATTR_ENCODE(idx_tab->sa_idx_tab[attr], length_idx,
1173 	    (uint32_t)((uintptr_t)attr_addr - (uintptr_t)hdr));
1174 }
1175 
1176 static void
1177 sa_attr_iter(objset_t *os, sa_hdr_phys_t *hdr, dmu_object_type_t type,
1178     sa_iterfunc_t func, sa_lot_t *tab, void *userp)
1179 {
1180 	void *data_start;
1181 	sa_lot_t *tb = tab;
1182 	sa_lot_t search;
1183 	avl_index_t loc;
1184 	sa_os_t *sa = os->os_sa;
1185 	int i;
1186 	uint16_t *length_start = NULL;
1187 	uint8_t length_idx = 0;
1188 
1189 	if (tab == NULL) {
1190 		search.lot_num = SA_LAYOUT_NUM(hdr, type);
1191 		tb = avl_find(&sa->sa_layout_num_tree, &search, &loc);
1192 		ASSERT(tb);
1193 	}
1194 
1195 	if (IS_SA_BONUSTYPE(type)) {
1196 		data_start = (void *)P2ROUNDUP(((uintptr_t)hdr +
1197 		    offsetof(sa_hdr_phys_t, sa_lengths) +
1198 		    (sizeof (uint16_t) * tb->lot_var_sizes)), 8);
1199 		length_start = hdr->sa_lengths;
1200 	} else {
1201 		data_start = hdr;
1202 	}
1203 
1204 	for (i = 0; i != tb->lot_attr_count; i++) {
1205 		int attr_length, reg_length;
1206 		uint8_t idx_len;
1207 
1208 		reg_length = sa->sa_attr_table[tb->lot_attrs[i]].sa_length;
1209 		IMPLY(reg_length == 0, IS_SA_BONUSTYPE(type));
1210 		if (reg_length) {
1211 			attr_length = reg_length;
1212 			idx_len = 0;
1213 		} else {
1214 			attr_length = length_start[length_idx];
1215 			idx_len = length_idx++;
1216 		}
1217 
1218 		func(hdr, data_start, tb->lot_attrs[i], attr_length,
1219 		    idx_len, reg_length == 0 ? B_TRUE : B_FALSE, userp);
1220 
1221 		data_start = (void *)P2ROUNDUP(((uintptr_t)data_start +
1222 		    attr_length), 8);
1223 	}
1224 }
1225 
1226 static void
1227 sa_byteswap_cb(void *hdr, void *attr_addr, sa_attr_type_t attr,
1228     uint16_t length, int length_idx, boolean_t variable_length, void *userp)
1229 {
1230 	(void) hdr, (void) length_idx, (void) variable_length;
1231 	sa_handle_t *hdl = userp;
1232 	sa_os_t *sa = hdl->sa_os->os_sa;
1233 
1234 	sa_bswap_table[sa->sa_attr_table[attr].sa_byteswap](attr_addr, length);
1235 }
1236 
1237 static void
1238 sa_byteswap(sa_handle_t *hdl, sa_buf_type_t buftype)
1239 {
1240 	sa_hdr_phys_t *sa_hdr_phys = SA_GET_HDR(hdl, buftype);
1241 	dmu_buf_impl_t *db;
1242 	int num_lengths = 1;
1243 	int i;
1244 	sa_os_t *sa __maybe_unused = hdl->sa_os->os_sa;
1245 
1246 	ASSERT(MUTEX_HELD(&sa->sa_lock));
1247 	if (sa_hdr_phys->sa_magic == SA_MAGIC)
1248 		return;
1249 
1250 	db = SA_GET_DB(hdl, buftype);
1251 
1252 	if (buftype == SA_SPILL) {
1253 		arc_release(db->db_buf, NULL);
1254 		arc_buf_thaw(db->db_buf);
1255 	}
1256 
1257 	sa_hdr_phys->sa_magic = BSWAP_32(sa_hdr_phys->sa_magic);
1258 	sa_hdr_phys->sa_layout_info = BSWAP_16(sa_hdr_phys->sa_layout_info);
1259 
1260 	/*
1261 	 * Determine number of variable lengths in header
1262 	 * The standard 8 byte header has one for free and a
1263 	 * 16 byte header would have 4 + 1;
1264 	 */
1265 	if (SA_HDR_SIZE(sa_hdr_phys) > 8)
1266 		num_lengths += (SA_HDR_SIZE(sa_hdr_phys) - 8) >> 1;
1267 	for (i = 0; i != num_lengths; i++)
1268 		sa_hdr_phys->sa_lengths[i] =
1269 		    BSWAP_16(sa_hdr_phys->sa_lengths[i]);
1270 
1271 	sa_attr_iter(hdl->sa_os, sa_hdr_phys, DMU_OT_SA,
1272 	    sa_byteswap_cb, NULL, hdl);
1273 
1274 	if (buftype == SA_SPILL)
1275 		arc_buf_freeze(((dmu_buf_impl_t *)hdl->sa_spill)->db_buf);
1276 }
1277 
1278 static int
1279 sa_build_index(sa_handle_t *hdl, sa_buf_type_t buftype)
1280 {
1281 	sa_hdr_phys_t *sa_hdr_phys;
1282 	dmu_buf_impl_t *db = SA_GET_DB(hdl, buftype);
1283 	dmu_object_type_t bonustype = SA_BONUSTYPE_FROM_DB(db);
1284 	sa_os_t *sa = hdl->sa_os->os_sa;
1285 	sa_idx_tab_t *idx_tab;
1286 
1287 	sa_hdr_phys = SA_GET_HDR(hdl, buftype);
1288 
1289 	mutex_enter(&sa->sa_lock);
1290 
1291 	/* Do we need to byteswap? */
1292 
1293 	/* only check if not old znode */
1294 	if (IS_SA_BONUSTYPE(bonustype) && sa_hdr_phys->sa_magic != SA_MAGIC &&
1295 	    sa_hdr_phys->sa_magic != 0) {
1296 		if (BSWAP_32(sa_hdr_phys->sa_magic) != SA_MAGIC) {
1297 			mutex_exit(&sa->sa_lock);
1298 			zfs_dbgmsg("Buffer Header: %x != SA_MAGIC:%x "
1299 			    "object=%#llx\n", sa_hdr_phys->sa_magic, SA_MAGIC,
1300 			    (u_longlong_t)db->db.db_object);
1301 			return (SET_ERROR(EIO));
1302 		}
1303 		sa_byteswap(hdl, buftype);
1304 	}
1305 
1306 	idx_tab = sa_find_idx_tab(hdl->sa_os, bonustype, sa_hdr_phys);
1307 
1308 	if (buftype == SA_BONUS)
1309 		hdl->sa_bonus_tab = idx_tab;
1310 	else
1311 		hdl->sa_spill_tab = idx_tab;
1312 
1313 	mutex_exit(&sa->sa_lock);
1314 	return (0);
1315 }
1316 
1317 static void
1318 sa_evict_sync(void *dbu)
1319 {
1320 	(void) dbu;
1321 	panic("evicting sa dbuf\n");
1322 }
1323 
1324 static void
1325 sa_idx_tab_rele(objset_t *os, void *arg)
1326 {
1327 	sa_os_t *sa = os->os_sa;
1328 	sa_idx_tab_t *idx_tab = arg;
1329 
1330 	if (idx_tab == NULL)
1331 		return;
1332 
1333 	mutex_enter(&sa->sa_lock);
1334 	if (zfs_refcount_remove(&idx_tab->sa_refcount, NULL) == 0) {
1335 		list_remove(&idx_tab->sa_layout->lot_idx_tab, idx_tab);
1336 		if (idx_tab->sa_variable_lengths)
1337 			kmem_free(idx_tab->sa_variable_lengths,
1338 			    sizeof (uint16_t) *
1339 			    idx_tab->sa_layout->lot_var_sizes);
1340 		zfs_refcount_destroy(&idx_tab->sa_refcount);
1341 		kmem_free(idx_tab->sa_idx_tab,
1342 		    sizeof (uint32_t) * sa->sa_num_attrs);
1343 		kmem_free(idx_tab, sizeof (sa_idx_tab_t));
1344 	}
1345 	mutex_exit(&sa->sa_lock);
1346 }
1347 
1348 static void
1349 sa_idx_tab_hold(objset_t *os, sa_idx_tab_t *idx_tab)
1350 {
1351 	sa_os_t *sa __maybe_unused = os->os_sa;
1352 
1353 	ASSERT(MUTEX_HELD(&sa->sa_lock));
1354 	(void) zfs_refcount_add(&idx_tab->sa_refcount, NULL);
1355 }
1356 
1357 void
1358 sa_spill_rele(sa_handle_t *hdl)
1359 {
1360 	mutex_enter(&hdl->sa_lock);
1361 	if (hdl->sa_spill) {
1362 		sa_idx_tab_rele(hdl->sa_os, hdl->sa_spill_tab);
1363 		dmu_buf_rele(hdl->sa_spill, NULL);
1364 		hdl->sa_spill = NULL;
1365 		hdl->sa_spill_tab = NULL;
1366 	}
1367 	mutex_exit(&hdl->sa_lock);
1368 }
1369 
1370 void
1371 sa_handle_destroy(sa_handle_t *hdl)
1372 {
1373 	dmu_buf_t *db = hdl->sa_bonus;
1374 
1375 	mutex_enter(&hdl->sa_lock);
1376 	(void) dmu_buf_remove_user(db, &hdl->sa_dbu);
1377 
1378 	if (hdl->sa_bonus_tab)
1379 		sa_idx_tab_rele(hdl->sa_os, hdl->sa_bonus_tab);
1380 
1381 	if (hdl->sa_spill_tab)
1382 		sa_idx_tab_rele(hdl->sa_os, hdl->sa_spill_tab);
1383 
1384 	dmu_buf_rele(hdl->sa_bonus, NULL);
1385 
1386 	if (hdl->sa_spill)
1387 		dmu_buf_rele(hdl->sa_spill, NULL);
1388 	mutex_exit(&hdl->sa_lock);
1389 
1390 	kmem_cache_free(sa_cache, hdl);
1391 }
1392 
1393 int
1394 sa_handle_get_from_db(objset_t *os, dmu_buf_t *db, void *userp,
1395     sa_handle_type_t hdl_type, sa_handle_t **handlepp)
1396 {
1397 	int error = 0;
1398 	sa_handle_t *handle = NULL;
1399 #ifdef ZFS_DEBUG
1400 	dmu_object_info_t doi;
1401 
1402 	dmu_object_info_from_db(db, &doi);
1403 	ASSERT(doi.doi_bonus_type == DMU_OT_SA ||
1404 	    doi.doi_bonus_type == DMU_OT_ZNODE);
1405 #endif
1406 	/* find handle, if it exists */
1407 	/* if one doesn't exist then create a new one, and initialize it */
1408 
1409 	if (hdl_type == SA_HDL_SHARED)
1410 		handle = dmu_buf_get_user(db);
1411 
1412 	if (handle == NULL) {
1413 		sa_handle_t *winner = NULL;
1414 
1415 		handle = kmem_cache_alloc(sa_cache, KM_SLEEP);
1416 		handle->sa_dbu.dbu_evict_func_sync = NULL;
1417 		handle->sa_dbu.dbu_evict_func_async = NULL;
1418 		handle->sa_userp = userp;
1419 		handle->sa_bonus = db;
1420 		handle->sa_os = os;
1421 		handle->sa_spill = NULL;
1422 		handle->sa_bonus_tab = NULL;
1423 		handle->sa_spill_tab = NULL;
1424 
1425 		error = sa_build_index(handle, SA_BONUS);
1426 
1427 		if (hdl_type == SA_HDL_SHARED) {
1428 			dmu_buf_init_user(&handle->sa_dbu, sa_evict_sync, NULL,
1429 			    NULL);
1430 			winner = dmu_buf_set_user_ie(db, &handle->sa_dbu);
1431 		}
1432 
1433 		if (winner != NULL) {
1434 			kmem_cache_free(sa_cache, handle);
1435 			handle = winner;
1436 		}
1437 	}
1438 	*handlepp = handle;
1439 
1440 	return (error);
1441 }
1442 
1443 int
1444 sa_handle_get(objset_t *objset, uint64_t objid, void *userp,
1445     sa_handle_type_t hdl_type, sa_handle_t **handlepp)
1446 {
1447 	dmu_buf_t *db;
1448 	int error;
1449 
1450 	if ((error = dmu_bonus_hold(objset, objid, NULL, &db)))
1451 		return (error);
1452 
1453 	return (sa_handle_get_from_db(objset, db, userp, hdl_type,
1454 	    handlepp));
1455 }
1456 
1457 int
1458 sa_buf_hold(objset_t *objset, uint64_t obj_num, const void *tag, dmu_buf_t **db)
1459 {
1460 	return (dmu_bonus_hold(objset, obj_num, tag, db));
1461 }
1462 
1463 void
1464 sa_buf_rele(dmu_buf_t *db, const void *tag)
1465 {
1466 	dmu_buf_rele(db, tag);
1467 }
1468 
1469 static int
1470 sa_lookup_impl(sa_handle_t *hdl, sa_bulk_attr_t *bulk, int count)
1471 {
1472 	ASSERT(hdl);
1473 	ASSERT(MUTEX_HELD(&hdl->sa_lock));
1474 	return (sa_attr_op(hdl, bulk, count, SA_LOOKUP, NULL));
1475 }
1476 
1477 static int
1478 sa_lookup_locked(sa_handle_t *hdl, sa_attr_type_t attr, void *buf,
1479     uint32_t buflen)
1480 {
1481 	int error;
1482 	sa_bulk_attr_t bulk;
1483 
1484 	VERIFY3U(buflen, <=, SA_ATTR_MAX_LEN);
1485 
1486 	bulk.sa_attr = attr;
1487 	bulk.sa_data = buf;
1488 	bulk.sa_length = buflen;
1489 	bulk.sa_data_func = NULL;
1490 
1491 	ASSERT(hdl);
1492 	error = sa_lookup_impl(hdl, &bulk, 1);
1493 	return (error);
1494 }
1495 
1496 int
1497 sa_lookup(sa_handle_t *hdl, sa_attr_type_t attr, void *buf, uint32_t buflen)
1498 {
1499 	int error;
1500 
1501 	mutex_enter(&hdl->sa_lock);
1502 	error = sa_lookup_locked(hdl, attr, buf, buflen);
1503 	mutex_exit(&hdl->sa_lock);
1504 
1505 	return (error);
1506 }
1507 
1508 /*
1509  * Return size of an attribute
1510  */
1511 
1512 static int
1513 sa_size_locked(sa_handle_t *hdl, sa_attr_type_t attr, int *size)
1514 {
1515 	sa_bulk_attr_t bulk;
1516 	int error;
1517 
1518 	bulk.sa_data = NULL;
1519 	bulk.sa_attr = attr;
1520 	bulk.sa_data_func = NULL;
1521 
1522 	ASSERT(hdl);
1523 	ASSERT(MUTEX_HELD(&hdl->sa_lock));
1524 	if ((error = sa_attr_op(hdl, &bulk, 1, SA_LOOKUP, NULL)) != 0) {
1525 		return (error);
1526 	}
1527 	*size = bulk.sa_size;
1528 
1529 	return (0);
1530 }
1531 
1532 int
1533 sa_size(sa_handle_t *hdl, sa_attr_type_t attr, int *size)
1534 {
1535 	int error;
1536 
1537 	mutex_enter(&hdl->sa_lock);
1538 	error = sa_size_locked(hdl, attr, size);
1539 	mutex_exit(&hdl->sa_lock);
1540 
1541 	return (error);
1542 }
1543 
1544 #ifdef _KERNEL
1545 int
1546 sa_lookup_uio(sa_handle_t *hdl, sa_attr_type_t attr, zfs_uio_t *uio)
1547 {
1548 	int error;
1549 	sa_bulk_attr_t bulk;
1550 
1551 	bulk.sa_data = NULL;
1552 	bulk.sa_attr = attr;
1553 	bulk.sa_data_func = NULL;
1554 
1555 	ASSERT(hdl);
1556 
1557 	mutex_enter(&hdl->sa_lock);
1558 	if ((error = sa_attr_op(hdl, &bulk, 1, SA_LOOKUP, NULL)) == 0) {
1559 		error = zfs_uiomove((void *)bulk.sa_addr, MIN(bulk.sa_size,
1560 		    zfs_uio_resid(uio)), UIO_READ, uio);
1561 	}
1562 	mutex_exit(&hdl->sa_lock);
1563 	return (error);
1564 }
1565 
1566 /*
1567  * For the existed object that is upgraded from old system, its ondisk layout
1568  * has no slot for the project ID attribute. But quota accounting logic needs
1569  * to access related slots by offset directly. So we need to adjust these old
1570  * objects' layout to make the project ID to some unified and fixed offset.
1571  */
1572 int
1573 sa_add_projid(sa_handle_t *hdl, dmu_tx_t *tx, uint64_t projid)
1574 {
1575 	znode_t *zp = sa_get_userdata(hdl);
1576 	dmu_buf_t *db = sa_get_db(hdl);
1577 	zfsvfs_t *zfsvfs = ZTOZSB(zp);
1578 	int count = 0, err = 0;
1579 	sa_bulk_attr_t *bulk, *attrs;
1580 	zfs_acl_locator_cb_t locate = { 0 };
1581 	uint64_t uid, gid, mode, rdev, xattr = 0, parent, gen, links;
1582 	uint64_t crtime[2], mtime[2], ctime[2], atime[2];
1583 	zfs_acl_phys_t znode_acl = { 0 };
1584 	char scanstamp[AV_SCANSTAMP_SZ];
1585 	char *dxattr_obj = NULL;
1586 	int dxattr_size = 0;
1587 
1588 	if (zp->z_acl_cached == NULL) {
1589 		zfs_acl_t *aclp;
1590 
1591 		mutex_enter(&zp->z_acl_lock);
1592 		err = zfs_acl_node_read(zp, B_FALSE, &aclp, B_FALSE);
1593 		mutex_exit(&zp->z_acl_lock);
1594 		if (err != 0 && err != ENOENT)
1595 			return (err);
1596 	}
1597 
1598 	bulk = kmem_zalloc(sizeof (sa_bulk_attr_t) * ZPL_END, KM_SLEEP);
1599 	attrs = kmem_zalloc(sizeof (sa_bulk_attr_t) * ZPL_END, KM_SLEEP);
1600 	mutex_enter(&hdl->sa_lock);
1601 	mutex_enter(&zp->z_lock);
1602 
1603 	err = sa_lookup_locked(hdl, SA_ZPL_PROJID(zfsvfs), &projid,
1604 	    sizeof (uint64_t));
1605 	if (unlikely(err == 0))
1606 		/* Someone has added project ID attr by race. */
1607 		err = EEXIST;
1608 	if (err != ENOENT)
1609 		goto out;
1610 
1611 	/* First do a bulk query of the attributes that aren't cached */
1612 	if (zp->z_is_sa) {
1613 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL,
1614 		    &mode, 8);
1615 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zfsvfs), NULL,
1616 		    &gen, 8);
1617 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
1618 		    &uid, 8);
1619 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs), NULL,
1620 		    &gid, 8);
1621 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zfsvfs), NULL,
1622 		    &parent, 8);
1623 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL,
1624 		    &atime, 16);
1625 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL,
1626 		    &mtime, 16);
1627 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
1628 		    &ctime, 16);
1629 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CRTIME(zfsvfs), NULL,
1630 		    &crtime, 16);
1631 		if (Z_ISBLK(ZTOTYPE(zp)) || Z_ISCHR(ZTOTYPE(zp)))
1632 			SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_RDEV(zfsvfs), NULL,
1633 			    &rdev, 8);
1634 	} else {
1635 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL,
1636 		    &atime, 16);
1637 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL,
1638 		    &mtime, 16);
1639 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
1640 		    &ctime, 16);
1641 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CRTIME(zfsvfs), NULL,
1642 		    &crtime, 16);
1643 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zfsvfs), NULL,
1644 		    &gen, 8);
1645 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL,
1646 		    &mode, 8);
1647 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zfsvfs), NULL,
1648 		    &parent, 8);
1649 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_XATTR(zfsvfs), NULL,
1650 		    &xattr, 8);
1651 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_RDEV(zfsvfs), NULL,
1652 		    &rdev, 8);
1653 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
1654 		    &uid, 8);
1655 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs), NULL,
1656 		    &gid, 8);
1657 		SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ZNODE_ACL(zfsvfs), NULL,
1658 		    &znode_acl, 88);
1659 	}
1660 	err = sa_bulk_lookup_locked(hdl, bulk, count);
1661 	if (err != 0)
1662 		goto out;
1663 
1664 	err = sa_lookup_locked(hdl, SA_ZPL_XATTR(zfsvfs), &xattr, 8);
1665 	if (err != 0 && err != ENOENT)
1666 		goto out;
1667 
1668 	err = sa_size_locked(hdl, SA_ZPL_DXATTR(zfsvfs), &dxattr_size);
1669 	if (err != 0 && err != ENOENT)
1670 		goto out;
1671 	if (dxattr_size != 0) {
1672 		dxattr_obj = vmem_alloc(dxattr_size, KM_SLEEP);
1673 		err = sa_lookup_locked(hdl, SA_ZPL_DXATTR(zfsvfs), dxattr_obj,
1674 		    dxattr_size);
1675 		if (err != 0 && err != ENOENT)
1676 			goto out;
1677 	}
1678 
1679 	zp->z_projid = projid;
1680 	zp->z_pflags |= ZFS_PROJID;
1681 	links = ZTONLNK(zp);
1682 	count = 0;
1683 	err = 0;
1684 
1685 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_MODE(zfsvfs), NULL, &mode, 8);
1686 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_SIZE(zfsvfs), NULL,
1687 	    &zp->z_size, 8);
1688 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_GEN(zfsvfs), NULL, &gen, 8);
1689 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_UID(zfsvfs), NULL, &uid, 8);
1690 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_GID(zfsvfs), NULL, &gid, 8);
1691 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_PARENT(zfsvfs), NULL, &parent, 8);
1692 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_FLAGS(zfsvfs), NULL,
1693 	    &zp->z_pflags, 8);
1694 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_ATIME(zfsvfs), NULL, &atime, 16);
1695 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
1696 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);
1697 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_CRTIME(zfsvfs), NULL,
1698 	    &crtime, 16);
1699 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_LINKS(zfsvfs), NULL, &links, 8);
1700 	SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_PROJID(zfsvfs), NULL, &projid, 8);
1701 
1702 	if (Z_ISBLK(ZTOTYPE(zp)) || Z_ISCHR(ZTOTYPE(zp)))
1703 		SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_RDEV(zfsvfs), NULL,
1704 		    &rdev, 8);
1705 
1706 	if (zp->z_acl_cached != NULL) {
1707 		SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_DACL_COUNT(zfsvfs), NULL,
1708 		    &zp->z_acl_cached->z_acl_count, 8);
1709 		if (zp->z_acl_cached->z_version < ZFS_ACL_VERSION_FUID)
1710 			zfs_acl_xform(zp, zp->z_acl_cached, CRED());
1711 		locate.cb_aclp = zp->z_acl_cached;
1712 		SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_DACL_ACES(zfsvfs),
1713 		    zfs_acl_data_locator, &locate,
1714 		    zp->z_acl_cached->z_acl_bytes);
1715 	}
1716 
1717 	if (xattr)
1718 		SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_XATTR(zfsvfs), NULL,
1719 		    &xattr, 8);
1720 
1721 	if (zp->z_pflags & ZFS_BONUS_SCANSTAMP) {
1722 		memcpy(scanstamp,
1723 		    (caddr_t)db->db_data + ZFS_OLD_ZNODE_PHYS_SIZE,
1724 		    AV_SCANSTAMP_SZ);
1725 		SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_SCANSTAMP(zfsvfs), NULL,
1726 		    scanstamp, AV_SCANSTAMP_SZ);
1727 		zp->z_pflags &= ~ZFS_BONUS_SCANSTAMP;
1728 	}
1729 
1730 	if (dxattr_obj) {
1731 		SA_ADD_BULK_ATTR(attrs, count, SA_ZPL_DXATTR(zfsvfs),
1732 		    NULL, dxattr_obj, dxattr_size);
1733 	}
1734 
1735 	VERIFY(dmu_set_bonustype(db, DMU_OT_SA, tx) == 0);
1736 	VERIFY(sa_replace_all_by_template_locked(hdl, attrs, count, tx) == 0);
1737 	if (znode_acl.z_acl_extern_obj) {
1738 		VERIFY(0 == dmu_object_free(zfsvfs->z_os,
1739 		    znode_acl.z_acl_extern_obj, tx));
1740 	}
1741 
1742 	zp->z_is_sa = B_TRUE;
1743 
1744 out:
1745 	mutex_exit(&zp->z_lock);
1746 	mutex_exit(&hdl->sa_lock);
1747 	kmem_free(attrs, sizeof (sa_bulk_attr_t) * ZPL_END);
1748 	kmem_free(bulk, sizeof (sa_bulk_attr_t) * ZPL_END);
1749 	if (dxattr_obj)
1750 		vmem_free(dxattr_obj, dxattr_size);
1751 	return (err);
1752 }
1753 #endif
1754 
1755 static sa_idx_tab_t *
1756 sa_find_idx_tab(objset_t *os, dmu_object_type_t bonustype, sa_hdr_phys_t *hdr)
1757 {
1758 	sa_idx_tab_t *idx_tab;
1759 	sa_os_t *sa = os->os_sa;
1760 	sa_lot_t *tb, search;
1761 	avl_index_t loc;
1762 
1763 	/*
1764 	 * Deterimine layout number.  If SA node and header == 0 then
1765 	 * force the index table to the dummy "1" empty layout.
1766 	 *
1767 	 * The layout number would only be zero for a newly created file
1768 	 * that has not added any attributes yet, or with crypto enabled which
1769 	 * doesn't write any attributes to the bonus buffer.
1770 	 */
1771 
1772 	search.lot_num = SA_LAYOUT_NUM(hdr, bonustype);
1773 
1774 	tb = avl_find(&sa->sa_layout_num_tree, &search, &loc);
1775 
1776 	/* Verify header size is consistent with layout information */
1777 	ASSERT(tb);
1778 	ASSERT((IS_SA_BONUSTYPE(bonustype) &&
1779 	    SA_HDR_SIZE_MATCH_LAYOUT(hdr, tb)) || !IS_SA_BONUSTYPE(bonustype) ||
1780 	    (IS_SA_BONUSTYPE(bonustype) && hdr->sa_layout_info == 0));
1781 
1782 	/*
1783 	 * See if any of the already existing TOC entries can be reused?
1784 	 */
1785 
1786 	for (idx_tab = list_head(&tb->lot_idx_tab); idx_tab;
1787 	    idx_tab = list_next(&tb->lot_idx_tab, idx_tab)) {
1788 		boolean_t valid_idx = B_TRUE;
1789 		int i;
1790 
1791 		if (tb->lot_var_sizes != 0 &&
1792 		    idx_tab->sa_variable_lengths != NULL) {
1793 			for (i = 0; i != tb->lot_var_sizes; i++) {
1794 				if (hdr->sa_lengths[i] !=
1795 				    idx_tab->sa_variable_lengths[i]) {
1796 					valid_idx = B_FALSE;
1797 					break;
1798 				}
1799 			}
1800 		}
1801 		if (valid_idx) {
1802 			sa_idx_tab_hold(os, idx_tab);
1803 			return (idx_tab);
1804 		}
1805 	}
1806 
1807 	/* No such luck, create a new entry */
1808 	idx_tab = kmem_zalloc(sizeof (sa_idx_tab_t), KM_SLEEP);
1809 	idx_tab->sa_idx_tab =
1810 	    kmem_zalloc(sizeof (uint32_t) * sa->sa_num_attrs, KM_SLEEP);
1811 	idx_tab->sa_layout = tb;
1812 	zfs_refcount_create(&idx_tab->sa_refcount);
1813 	if (tb->lot_var_sizes)
1814 		idx_tab->sa_variable_lengths = kmem_alloc(sizeof (uint16_t) *
1815 		    tb->lot_var_sizes, KM_SLEEP);
1816 
1817 	sa_attr_iter(os, hdr, bonustype, sa_build_idx_tab,
1818 	    tb, idx_tab);
1819 	sa_idx_tab_hold(os, idx_tab);   /* one hold for consumer */
1820 	sa_idx_tab_hold(os, idx_tab);	/* one for layout */
1821 	list_insert_tail(&tb->lot_idx_tab, idx_tab);
1822 	return (idx_tab);
1823 }
1824 
1825 void
1826 sa_default_locator(void **dataptr, uint32_t *len, uint32_t total_len,
1827     boolean_t start, void *userdata)
1828 {
1829 	ASSERT(start);
1830 
1831 	*dataptr = userdata;
1832 	*len = total_len;
1833 }
1834 
1835 static void
1836 sa_attr_register_sync(sa_handle_t *hdl, dmu_tx_t *tx)
1837 {
1838 	uint64_t attr_value = 0;
1839 	sa_os_t *sa = hdl->sa_os->os_sa;
1840 	sa_attr_table_t *tb = sa->sa_attr_table;
1841 	int i;
1842 
1843 	mutex_enter(&sa->sa_lock);
1844 
1845 	if (!sa->sa_need_attr_registration || sa->sa_master_obj == 0) {
1846 		mutex_exit(&sa->sa_lock);
1847 		return;
1848 	}
1849 
1850 	if (sa->sa_reg_attr_obj == 0) {
1851 		sa->sa_reg_attr_obj = zap_create_link(hdl->sa_os,
1852 		    DMU_OT_SA_ATTR_REGISTRATION,
1853 		    sa->sa_master_obj, SA_REGISTRY, tx);
1854 	}
1855 	for (i = 0; i != sa->sa_num_attrs; i++) {
1856 		if (sa->sa_attr_table[i].sa_registered)
1857 			continue;
1858 		ATTR_ENCODE(attr_value, tb[i].sa_attr, tb[i].sa_length,
1859 		    tb[i].sa_byteswap);
1860 		VERIFY(0 == zap_update(hdl->sa_os, sa->sa_reg_attr_obj,
1861 		    tb[i].sa_name, 8, 1, &attr_value, tx));
1862 		tb[i].sa_registered = B_TRUE;
1863 	}
1864 	sa->sa_need_attr_registration = B_FALSE;
1865 	mutex_exit(&sa->sa_lock);
1866 }
1867 
1868 /*
1869  * Replace all attributes with attributes specified in template.
1870  * If dnode had a spill buffer then those attributes will be
1871  * also be replaced, possibly with just an empty spill block
1872  *
1873  * This interface is intended to only be used for bulk adding of
1874  * attributes for a new file.  It will also be used by the ZPL
1875  * when converting and old formatted znode to native SA support.
1876  */
1877 int
1878 sa_replace_all_by_template_locked(sa_handle_t *hdl, sa_bulk_attr_t *attr_desc,
1879     int attr_count, dmu_tx_t *tx)
1880 {
1881 	sa_os_t *sa = hdl->sa_os->os_sa;
1882 
1883 	if (sa->sa_need_attr_registration)
1884 		sa_attr_register_sync(hdl, tx);
1885 	return (sa_build_layouts(hdl, attr_desc, attr_count, tx));
1886 }
1887 
1888 int
1889 sa_replace_all_by_template(sa_handle_t *hdl, sa_bulk_attr_t *attr_desc,
1890     int attr_count, dmu_tx_t *tx)
1891 {
1892 	int error;
1893 
1894 	mutex_enter(&hdl->sa_lock);
1895 	error = sa_replace_all_by_template_locked(hdl, attr_desc,
1896 	    attr_count, tx);
1897 	mutex_exit(&hdl->sa_lock);
1898 	return (error);
1899 }
1900 
1901 /*
1902  * Add/remove a single attribute or replace a variable-sized attribute value
1903  * with a value of a different size, and then rewrite the entire set
1904  * of attributes.
1905  * Same-length attribute value replacement (including fixed-length attributes)
1906  * is handled more efficiently by the upper layers.
1907  */
1908 static int
1909 sa_modify_attrs(sa_handle_t *hdl, sa_attr_type_t newattr,
1910     sa_data_op_t action, sa_data_locator_t *locator, void *datastart,
1911     uint16_t buflen, dmu_tx_t *tx)
1912 {
1913 	sa_os_t *sa = hdl->sa_os->os_sa;
1914 	dmu_buf_impl_t *db = (dmu_buf_impl_t *)hdl->sa_bonus;
1915 	sa_bulk_attr_t *attr_desc;
1916 	void *old_data[2];
1917 	int bonus_attr_count = 0;
1918 	int bonus_data_size = 0;
1919 	int spill_data_size = 0;
1920 	int spill_attr_count = 0;
1921 	int error;
1922 	uint16_t length, reg_length;
1923 	int i, j, k, length_idx;
1924 	sa_hdr_phys_t *hdr;
1925 	sa_idx_tab_t *idx_tab;
1926 	int attr_count;
1927 	int count;
1928 
1929 	ASSERT(MUTEX_HELD(&hdl->sa_lock));
1930 
1931 	/* First make of copy of the old data */
1932 
1933 	DB_DNODE_ENTER(db);
1934 	if (DB_DNODE(db)->dn_bonuslen != 0) {
1935 		bonus_data_size = hdl->sa_bonus->db_size;
1936 		old_data[0] = kmem_alloc(bonus_data_size, KM_SLEEP);
1937 		memcpy(old_data[0], hdl->sa_bonus->db_data,
1938 		    hdl->sa_bonus->db_size);
1939 		bonus_attr_count = hdl->sa_bonus_tab->sa_layout->lot_attr_count;
1940 	} else {
1941 		old_data[0] = NULL;
1942 	}
1943 	DB_DNODE_EXIT(db);
1944 
1945 	/* Bring spill buffer online if it isn't currently */
1946 
1947 	if ((error = sa_get_spill(hdl)) == 0) {
1948 		spill_data_size = hdl->sa_spill->db_size;
1949 		old_data[1] = vmem_alloc(spill_data_size, KM_SLEEP);
1950 		memcpy(old_data[1], hdl->sa_spill->db_data,
1951 		    hdl->sa_spill->db_size);
1952 		spill_attr_count =
1953 		    hdl->sa_spill_tab->sa_layout->lot_attr_count;
1954 	} else if (error && error != ENOENT) {
1955 		if (old_data[0])
1956 			kmem_free(old_data[0], bonus_data_size);
1957 		return (error);
1958 	} else {
1959 		old_data[1] = NULL;
1960 	}
1961 
1962 	/* build descriptor of all attributes */
1963 
1964 	attr_count = bonus_attr_count + spill_attr_count;
1965 	if (action == SA_ADD)
1966 		attr_count++;
1967 	else if (action == SA_REMOVE)
1968 		attr_count--;
1969 
1970 	attr_desc = kmem_zalloc(sizeof (sa_bulk_attr_t) * attr_count, KM_SLEEP);
1971 
1972 	/*
1973 	 * loop through bonus and spill buffer if it exists, and
1974 	 * build up new attr_descriptor to reset the attributes
1975 	 */
1976 	k = j = 0;
1977 	count = bonus_attr_count;
1978 	hdr = SA_GET_HDR(hdl, SA_BONUS);
1979 	idx_tab = SA_IDX_TAB_GET(hdl, SA_BONUS);
1980 	for (; ; k++) {
1981 		/*
1982 		 * Iterate over each attribute in layout.  Fetch the
1983 		 * size of variable-length attributes needing rewrite
1984 		 * from sa_lengths[].
1985 		 */
1986 		for (i = 0, length_idx = 0; i != count; i++) {
1987 			sa_attr_type_t attr;
1988 
1989 			attr = idx_tab->sa_layout->lot_attrs[i];
1990 			reg_length = SA_REGISTERED_LEN(sa, attr);
1991 			if (reg_length == 0) {
1992 				length = hdr->sa_lengths[length_idx];
1993 				length_idx++;
1994 			} else {
1995 				length = reg_length;
1996 			}
1997 			if (attr == newattr) {
1998 				/*
1999 				 * There is nothing to do for SA_REMOVE,
2000 				 * so it is just skipped.
2001 				 */
2002 				if (action == SA_REMOVE)
2003 					continue;
2004 
2005 				/*
2006 				 * Duplicate attributes are not allowed, so the
2007 				 * action can not be SA_ADD here.
2008 				 */
2009 				ASSERT3S(action, ==, SA_REPLACE);
2010 
2011 				/*
2012 				 * Only a variable-sized attribute can be
2013 				 * replaced here, and its size must be changing.
2014 				 */
2015 				ASSERT3U(reg_length, ==, 0);
2016 				ASSERT3U(length, !=, buflen);
2017 				SA_ADD_BULK_ATTR(attr_desc, j, attr,
2018 				    locator, datastart, buflen);
2019 			} else {
2020 				SA_ADD_BULK_ATTR(attr_desc, j, attr,
2021 				    NULL, (void *)
2022 				    (TOC_OFF(idx_tab->sa_idx_tab[attr]) +
2023 				    (uintptr_t)old_data[k]), length);
2024 			}
2025 		}
2026 		if (k == 0 && hdl->sa_spill) {
2027 			hdr = SA_GET_HDR(hdl, SA_SPILL);
2028 			idx_tab = SA_IDX_TAB_GET(hdl, SA_SPILL);
2029 			count = spill_attr_count;
2030 		} else {
2031 			break;
2032 		}
2033 	}
2034 	if (action == SA_ADD) {
2035 		reg_length = SA_REGISTERED_LEN(sa, newattr);
2036 		IMPLY(reg_length != 0, reg_length == buflen);
2037 		SA_ADD_BULK_ATTR(attr_desc, j, newattr, locator,
2038 		    datastart, buflen);
2039 	}
2040 	ASSERT3U(j, ==, attr_count);
2041 
2042 	error = sa_build_layouts(hdl, attr_desc, attr_count, tx);
2043 
2044 	if (old_data[0])
2045 		kmem_free(old_data[0], bonus_data_size);
2046 	if (old_data[1])
2047 		vmem_free(old_data[1], spill_data_size);
2048 	kmem_free(attr_desc, sizeof (sa_bulk_attr_t) * attr_count);
2049 
2050 	return (error);
2051 }
2052 
2053 static int
2054 sa_bulk_update_impl(sa_handle_t *hdl, sa_bulk_attr_t *bulk, int count,
2055     dmu_tx_t *tx)
2056 {
2057 	int error;
2058 	sa_os_t *sa = hdl->sa_os->os_sa;
2059 	dmu_object_type_t bonustype;
2060 	dmu_buf_t *saved_spill;
2061 
2062 	ASSERT(hdl);
2063 	ASSERT(MUTEX_HELD(&hdl->sa_lock));
2064 
2065 	bonustype = SA_BONUSTYPE_FROM_DB(SA_GET_DB(hdl, SA_BONUS));
2066 	saved_spill = hdl->sa_spill;
2067 
2068 	/* sync out registration table if necessary */
2069 	if (sa->sa_need_attr_registration)
2070 		sa_attr_register_sync(hdl, tx);
2071 
2072 	error = sa_attr_op(hdl, bulk, count, SA_UPDATE, tx);
2073 	if (error == 0 && !IS_SA_BONUSTYPE(bonustype) && sa->sa_update_cb)
2074 		sa->sa_update_cb(hdl, tx);
2075 
2076 	/*
2077 	 * If saved_spill is NULL and current sa_spill is not NULL that
2078 	 * means we increased the refcount of the spill buffer through
2079 	 * sa_get_spill() or dmu_spill_hold_by_dnode().  Therefore we
2080 	 * must release the hold before calling dmu_tx_commit() to avoid
2081 	 * making a copy of this buffer in dbuf_sync_leaf() due to the
2082 	 * reference count now being greater than 1.
2083 	 */
2084 	if (!saved_spill && hdl->sa_spill) {
2085 		if (hdl->sa_spill_tab) {
2086 			sa_idx_tab_rele(hdl->sa_os, hdl->sa_spill_tab);
2087 			hdl->sa_spill_tab = NULL;
2088 		}
2089 
2090 		dmu_buf_rele(hdl->sa_spill, NULL);
2091 		hdl->sa_spill = NULL;
2092 	}
2093 
2094 	return (error);
2095 }
2096 
2097 /*
2098  * update or add new attribute
2099  */
2100 int
2101 sa_update(sa_handle_t *hdl, sa_attr_type_t type,
2102     void *buf, uint32_t buflen, dmu_tx_t *tx)
2103 {
2104 	int error;
2105 	sa_bulk_attr_t bulk;
2106 
2107 	VERIFY3U(buflen, <=, SA_ATTR_MAX_LEN);
2108 
2109 	bulk.sa_attr = type;
2110 	bulk.sa_data_func = NULL;
2111 	bulk.sa_length = buflen;
2112 	bulk.sa_data = buf;
2113 
2114 	mutex_enter(&hdl->sa_lock);
2115 	error = sa_bulk_update_impl(hdl, &bulk, 1, tx);
2116 	mutex_exit(&hdl->sa_lock);
2117 	return (error);
2118 }
2119 
2120 int
2121 sa_bulk_lookup_locked(sa_handle_t *hdl, sa_bulk_attr_t *attrs, int count)
2122 {
2123 	ASSERT(hdl);
2124 	ASSERT(MUTEX_HELD(&hdl->sa_lock));
2125 	return (sa_lookup_impl(hdl, attrs, count));
2126 }
2127 
2128 int
2129 sa_bulk_lookup(sa_handle_t *hdl, sa_bulk_attr_t *attrs, int count)
2130 {
2131 	int error;
2132 
2133 	ASSERT(hdl);
2134 	mutex_enter(&hdl->sa_lock);
2135 	error = sa_bulk_lookup_locked(hdl, attrs, count);
2136 	mutex_exit(&hdl->sa_lock);
2137 	return (error);
2138 }
2139 
2140 int
2141 sa_bulk_update(sa_handle_t *hdl, sa_bulk_attr_t *attrs, int count, dmu_tx_t *tx)
2142 {
2143 	int error;
2144 
2145 	ASSERT(hdl);
2146 	mutex_enter(&hdl->sa_lock);
2147 	error = sa_bulk_update_impl(hdl, attrs, count, tx);
2148 	mutex_exit(&hdl->sa_lock);
2149 	return (error);
2150 }
2151 
2152 int
2153 sa_remove(sa_handle_t *hdl, sa_attr_type_t attr, dmu_tx_t *tx)
2154 {
2155 	int error;
2156 
2157 	mutex_enter(&hdl->sa_lock);
2158 	error = sa_modify_attrs(hdl, attr, SA_REMOVE, NULL,
2159 	    NULL, 0, tx);
2160 	mutex_exit(&hdl->sa_lock);
2161 	return (error);
2162 }
2163 
2164 void
2165 sa_object_info(sa_handle_t *hdl, dmu_object_info_t *doi)
2166 {
2167 	dmu_object_info_from_db(hdl->sa_bonus, doi);
2168 }
2169 
2170 void
2171 sa_object_size(sa_handle_t *hdl, uint32_t *blksize, u_longlong_t *nblocks)
2172 {
2173 	dmu_object_size_from_db(hdl->sa_bonus,
2174 	    blksize, nblocks);
2175 }
2176 
2177 void
2178 sa_set_userp(sa_handle_t *hdl, void *ptr)
2179 {
2180 	hdl->sa_userp = ptr;
2181 }
2182 
2183 dmu_buf_t *
2184 sa_get_db(sa_handle_t *hdl)
2185 {
2186 	return (hdl->sa_bonus);
2187 }
2188 
2189 void *
2190 sa_get_userdata(sa_handle_t *hdl)
2191 {
2192 	return (hdl->sa_userp);
2193 }
2194 
2195 void
2196 sa_register_update_callback_locked(objset_t *os, sa_update_cb_t *func)
2197 {
2198 	ASSERT(MUTEX_HELD(&os->os_sa->sa_lock));
2199 	os->os_sa->sa_update_cb = func;
2200 }
2201 
2202 void
2203 sa_register_update_callback(objset_t *os, sa_update_cb_t *func)
2204 {
2205 
2206 	mutex_enter(&os->os_sa->sa_lock);
2207 	sa_register_update_callback_locked(os, func);
2208 	mutex_exit(&os->os_sa->sa_lock);
2209 }
2210 
2211 uint64_t
2212 sa_handle_object(sa_handle_t *hdl)
2213 {
2214 	return (hdl->sa_bonus->db_object);
2215 }
2216 
2217 boolean_t
2218 sa_enabled(objset_t *os)
2219 {
2220 	return (os->os_sa == NULL);
2221 }
2222 
2223 int
2224 sa_set_sa_object(objset_t *os, uint64_t sa_object)
2225 {
2226 	sa_os_t *sa = os->os_sa;
2227 
2228 	if (sa->sa_master_obj)
2229 		return (1);
2230 
2231 	sa->sa_master_obj = sa_object;
2232 
2233 	return (0);
2234 }
2235 
2236 int
2237 sa_hdrsize(void *arg)
2238 {
2239 	sa_hdr_phys_t *hdr = arg;
2240 
2241 	return (SA_HDR_SIZE(hdr));
2242 }
2243 
2244 void
2245 sa_handle_lock(sa_handle_t *hdl)
2246 {
2247 	ASSERT(hdl);
2248 	mutex_enter(&hdl->sa_lock);
2249 }
2250 
2251 void
2252 sa_handle_unlock(sa_handle_t *hdl)
2253 {
2254 	ASSERT(hdl);
2255 	mutex_exit(&hdl->sa_lock);
2256 }
2257 
2258 #ifdef _KERNEL
2259 EXPORT_SYMBOL(sa_handle_get);
2260 EXPORT_SYMBOL(sa_handle_get_from_db);
2261 EXPORT_SYMBOL(sa_handle_destroy);
2262 EXPORT_SYMBOL(sa_buf_hold);
2263 EXPORT_SYMBOL(sa_buf_rele);
2264 EXPORT_SYMBOL(sa_spill_rele);
2265 EXPORT_SYMBOL(sa_lookup);
2266 EXPORT_SYMBOL(sa_update);
2267 EXPORT_SYMBOL(sa_remove);
2268 EXPORT_SYMBOL(sa_bulk_lookup);
2269 EXPORT_SYMBOL(sa_bulk_lookup_locked);
2270 EXPORT_SYMBOL(sa_bulk_update);
2271 EXPORT_SYMBOL(sa_size);
2272 EXPORT_SYMBOL(sa_object_info);
2273 EXPORT_SYMBOL(sa_object_size);
2274 EXPORT_SYMBOL(sa_get_userdata);
2275 EXPORT_SYMBOL(sa_set_userp);
2276 EXPORT_SYMBOL(sa_get_db);
2277 EXPORT_SYMBOL(sa_handle_object);
2278 EXPORT_SYMBOL(sa_register_update_callback);
2279 EXPORT_SYMBOL(sa_setup);
2280 EXPORT_SYMBOL(sa_replace_all_by_template);
2281 EXPORT_SYMBOL(sa_replace_all_by_template_locked);
2282 EXPORT_SYMBOL(sa_enabled);
2283 EXPORT_SYMBOL(sa_cache_init);
2284 EXPORT_SYMBOL(sa_cache_fini);
2285 EXPORT_SYMBOL(sa_set_sa_object);
2286 EXPORT_SYMBOL(sa_hdrsize);
2287 EXPORT_SYMBOL(sa_handle_lock);
2288 EXPORT_SYMBOL(sa_handle_unlock);
2289 EXPORT_SYMBOL(sa_lookup_uio);
2290 EXPORT_SYMBOL(sa_add_projid);
2291 #endif /* _KERNEL */
2292