Age Owner TLA Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * nbtree.h
4 : * header file for postgres btree access method implementation.
5 : *
6 : *
7 : * Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group
8 : * Portions Copyright (c) 1994, Regents of the University of California
9 : *
10 : * src/include/access/nbtree.h
11 : *
12 : *-------------------------------------------------------------------------
13 : */
14 : #ifndef NBTREE_H
15 : #define NBTREE_H
16 :
17 : #include "access/amapi.h"
18 : #include "access/itup.h"
19 : #include "access/sdir.h"
20 : #include "access/tableam.h"
21 : #include "access/xlogreader.h"
22 : #include "catalog/pg_am_d.h"
23 : #include "catalog/pg_index.h"
24 : #include "lib/stringinfo.h"
25 : #include "storage/bufmgr.h"
26 : #include "storage/shm_toc.h"
27 :
28 : /* There's room for a 16-bit vacuum cycle ID in BTPageOpaqueData */
29 : typedef uint16 BTCycleId;
30 :
31 : /*
32 : * BTPageOpaqueData -- At the end of every page, we store a pointer
33 : * to both siblings in the tree. This is used to do forward/backward
34 : * index scans. The next-page link is also critical for recovery when
35 : * a search has navigated to the wrong page due to concurrent page splits
36 : * or deletions; see src/backend/access/nbtree/README for more info.
37 : *
38 : * In addition, we store the page's btree level (counting upwards from
39 : * zero at a leaf page) as well as some flag bits indicating the page type
40 : * and status. If the page is deleted, a BTDeletedPageData struct is stored
41 : * in the page's tuple area, while a standard BTPageOpaqueData struct is
42 : * stored in the page special area.
43 : *
44 : * We also store a "vacuum cycle ID". When a page is split while VACUUM is
45 : * processing the index, a nonzero value associated with the VACUUM run is
46 : * stored into both halves of the split page. (If VACUUM is not running,
47 : * both pages receive zero cycleids.) This allows VACUUM to detect whether
48 : * a page was split since it started, with a small probability of false match
49 : * if the page was last split some exact multiple of MAX_BT_CYCLE_ID VACUUMs
50 : * ago. Also, during a split, the BTP_SPLIT_END flag is cleared in the left
51 : * (original) page, and set in the right page, but only if the next page
52 : * to its right has a different cycleid.
53 : *
54 : * NOTE: the BTP_LEAF flag bit is redundant since level==0 could be tested
55 : * instead.
56 : *
57 : * NOTE: the btpo_level field used to be a union type in order to allow
58 : * deleted pages to store a 32-bit safexid in the same field. We now store
59 : * 64-bit/full safexid values using BTDeletedPageData instead.
60 : */
61 :
62 : typedef struct BTPageOpaqueData
63 : {
64 : BlockNumber btpo_prev; /* left sibling, or P_NONE if leftmost */
65 : BlockNumber btpo_next; /* right sibling, or P_NONE if rightmost */
66 : uint32 btpo_level; /* tree level --- zero for leaf pages */
67 : uint16 btpo_flags; /* flag bits, see below */
68 : BTCycleId btpo_cycleid; /* vacuum cycle ID of latest split */
69 : } BTPageOpaqueData;
70 :
71 : typedef BTPageOpaqueData *BTPageOpaque;
72 :
73 : #define BTPageGetOpaque(page) ((BTPageOpaque) PageGetSpecialPointer(page))
74 :
75 : /* Bits defined in btpo_flags */
76 : #define BTP_LEAF (1 << 0) /* leaf page, i.e. not internal page */
77 : #define BTP_ROOT (1 << 1) /* root page (has no parent) */
78 : #define BTP_DELETED (1 << 2) /* page has been deleted from tree */
79 : #define BTP_META (1 << 3) /* meta-page */
80 : #define BTP_HALF_DEAD (1 << 4) /* empty, but still in tree */
81 : #define BTP_SPLIT_END (1 << 5) /* rightmost page of split group */
82 : #define BTP_HAS_GARBAGE (1 << 6) /* page has LP_DEAD tuples (deprecated) */
83 : #define BTP_INCOMPLETE_SPLIT (1 << 7) /* right sibling's downlink is missing */
84 : #define BTP_HAS_FULLXID (1 << 8) /* contains BTDeletedPageData */
85 :
86 : /*
87 : * The max allowed value of a cycle ID is a bit less than 64K. This is
88 : * for convenience of pg_filedump and similar utilities: we want to use
89 : * the last 2 bytes of special space as an index type indicator, and
90 : * restricting cycle ID lets btree use that space for vacuum cycle IDs
91 : * while still allowing index type to be identified.
92 : */
93 : #define MAX_BT_CYCLE_ID 0xFF7F
94 :
95 :
96 : /*
97 : * The Meta page is always the first page in the btree index.
98 : * Its primary purpose is to point to the location of the btree root page.
99 : * We also point to the "fast" root, which is the current effective root;
100 : * see README for discussion.
101 : */
102 :
103 : typedef struct BTMetaPageData
104 : {
105 : uint32 btm_magic; /* should contain BTREE_MAGIC */
106 : uint32 btm_version; /* nbtree version (always <= BTREE_VERSION) */
107 : BlockNumber btm_root; /* current root location */
108 : uint32 btm_level; /* tree level of the root page */
109 : BlockNumber btm_fastroot; /* current "fast" root location */
110 : uint32 btm_fastlevel; /* tree level of the "fast" root page */
111 : /* remaining fields only valid when btm_version >= BTREE_NOVAC_VERSION */
112 :
113 : /* number of deleted, non-recyclable pages during last cleanup */
114 : uint32 btm_last_cleanup_num_delpages;
115 : /* number of heap tuples during last cleanup (deprecated) */
116 : float8 btm_last_cleanup_num_heap_tuples;
117 :
118 : bool btm_allequalimage; /* are all columns "equalimage"? */
119 : } BTMetaPageData;
120 :
121 : #define BTPageGetMeta(p) \
122 : ((BTMetaPageData *) PageGetContents(p))
123 :
124 : /*
125 : * The current Btree version is 4. That's what you'll get when you create
126 : * a new index.
127 : *
128 : * Btree version 3 was used in PostgreSQL v11. It is mostly the same as
129 : * version 4, but heap TIDs were not part of the keyspace. Index tuples
130 : * with duplicate keys could be stored in any order. We continue to
131 : * support reading and writing Btree versions 2 and 3, so that they don't
132 : * need to be immediately re-indexed at pg_upgrade. In order to get the
133 : * new heapkeyspace semantics, however, a REINDEX is needed.
134 : *
135 : * Deduplication is safe to use when the btm_allequalimage field is set to
136 : * true. It's safe to read the btm_allequalimage field on version 3, but
137 : * only version 4 indexes make use of deduplication. Even version 4
138 : * indexes created on PostgreSQL v12 will need a REINDEX to make use of
139 : * deduplication, though, since there is no other way to set
140 : * btm_allequalimage to true (pg_upgrade hasn't been taught to set the
141 : * metapage field).
142 : *
143 : * Btree version 2 is mostly the same as version 3. There are two new
144 : * fields in the metapage that were introduced in version 3. A version 2
145 : * metapage will be automatically upgraded to version 3 on the first
146 : * insert to it. INCLUDE indexes cannot use version 2.
147 : */
148 : #define BTREE_METAPAGE 0 /* first page is meta */
149 : #define BTREE_MAGIC 0x053162 /* magic number in metapage */
150 : #define BTREE_VERSION 4 /* current version number */
151 : #define BTREE_MIN_VERSION 2 /* minimum supported version */
152 : #define BTREE_NOVAC_VERSION 3 /* version with all meta fields set */
153 :
154 : /*
155 : * Maximum size of a btree index entry, including its tuple header.
156 : *
157 : * We actually need to be able to fit three items on every page,
158 : * so restrict any one item to 1/3 the per-page available space.
159 : *
160 : * There are rare cases where _bt_truncate() will need to enlarge
161 : * a heap index tuple to make space for a tiebreaker heap TID
162 : * attribute, which we account for here.
163 : */
164 : #define BTMaxItemSize(page) \
165 : (MAXALIGN_DOWN((PageGetPageSize(page) - \
166 : MAXALIGN(SizeOfPageHeaderData + 3*sizeof(ItemIdData)) - \
167 : MAXALIGN(sizeof(BTPageOpaqueData))) / 3) - \
168 : MAXALIGN(sizeof(ItemPointerData)))
169 : #define BTMaxItemSizeNoHeapTid(page) \
170 : MAXALIGN_DOWN((PageGetPageSize(page) - \
171 : MAXALIGN(SizeOfPageHeaderData + 3*sizeof(ItemIdData)) - \
172 : MAXALIGN(sizeof(BTPageOpaqueData))) / 3)
173 :
174 : /*
175 : * MaxTIDsPerBTreePage is an upper bound on the number of heap TIDs tuples
176 : * that may be stored on a btree leaf page. It is used to size the
177 : * per-page temporary buffers.
178 : *
179 : * Note: we don't bother considering per-tuple overheads here to keep
180 : * things simple (value is based on how many elements a single array of
181 : * heap TIDs must have to fill the space between the page header and
182 : * special area). The value is slightly higher (i.e. more conservative)
183 : * than necessary as a result, which is considered acceptable.
184 : */
185 : #define MaxTIDsPerBTreePage \
186 : (int) ((BLCKSZ - SizeOfPageHeaderData - sizeof(BTPageOpaqueData)) / \
187 : sizeof(ItemPointerData))
188 :
189 : /*
190 : * The leaf-page fillfactor defaults to 90% but is user-adjustable.
191 : * For pages above the leaf level, we use a fixed 70% fillfactor.
192 : * The fillfactor is applied during index build and when splitting
193 : * a rightmost page; when splitting non-rightmost pages we try to
194 : * divide the data equally. When splitting a page that's entirely
195 : * filled with a single value (duplicates), the effective leaf-page
196 : * fillfactor is 96%, regardless of whether the page is a rightmost
197 : * page.
198 : */
199 : #define BTREE_MIN_FILLFACTOR 10
200 : #define BTREE_DEFAULT_FILLFACTOR 90
201 : #define BTREE_NONLEAF_FILLFACTOR 70
202 : #define BTREE_SINGLEVAL_FILLFACTOR 96
203 :
204 : /*
205 : * In general, the btree code tries to localize its knowledge about
206 : * page layout to a couple of routines. However, we need a special
207 : * value to indicate "no page number" in those places where we expect
208 : * page numbers. We can use zero for this because we never need to
209 : * make a pointer to the metadata page.
210 : */
211 :
212 : #define P_NONE 0
213 :
214 : /*
215 : * Macros to test whether a page is leftmost or rightmost on its tree level,
216 : * as well as other state info kept in the opaque data.
217 : */
218 : #define P_LEFTMOST(opaque) ((opaque)->btpo_prev == P_NONE)
219 : #define P_RIGHTMOST(opaque) ((opaque)->btpo_next == P_NONE)
220 : #define P_ISLEAF(opaque) (((opaque)->btpo_flags & BTP_LEAF) != 0)
221 : #define P_ISROOT(opaque) (((opaque)->btpo_flags & BTP_ROOT) != 0)
222 : #define P_ISDELETED(opaque) (((opaque)->btpo_flags & BTP_DELETED) != 0)
223 : #define P_ISMETA(opaque) (((opaque)->btpo_flags & BTP_META) != 0)
224 : #define P_ISHALFDEAD(opaque) (((opaque)->btpo_flags & BTP_HALF_DEAD) != 0)
225 : #define P_IGNORE(opaque) (((opaque)->btpo_flags & (BTP_DELETED|BTP_HALF_DEAD)) != 0)
226 : #define P_HAS_GARBAGE(opaque) (((opaque)->btpo_flags & BTP_HAS_GARBAGE) != 0)
227 : #define P_INCOMPLETE_SPLIT(opaque) (((opaque)->btpo_flags & BTP_INCOMPLETE_SPLIT) != 0)
228 : #define P_HAS_FULLXID(opaque) (((opaque)->btpo_flags & BTP_HAS_FULLXID) != 0)
229 :
230 : /*
231 : * BTDeletedPageData is the page contents of a deleted page
232 : */
233 : typedef struct BTDeletedPageData
234 : {
235 : FullTransactionId safexid; /* See BTPageIsRecyclable() */
236 : } BTDeletedPageData;
237 :
774 pg 238 ECB : static inline void
774 pg 239 GIC 3420 : BTPageSetDeleted(Page page, FullTransactionId safexid)
240 : {
241 : BTPageOpaque opaque;
242 : PageHeader header;
243 : BTDeletedPageData *contents;
774 pg 244 ECB :
373 michael 245 CBC 3420 : opaque = BTPageGetOpaque(page);
774 pg 246 GIC 3420 : header = ((PageHeader) page);
774 pg 247 ECB :
774 pg 248 CBC 3420 : opaque->btpo_flags &= ~BTP_HALF_DEAD;
249 3420 : opaque->btpo_flags |= BTP_DELETED | BTP_HAS_FULLXID;
774 pg 250 GIC 3420 : header->pd_lower = MAXALIGN(SizeOfPageHeaderData) +
774 pg 251 ECB : sizeof(BTDeletedPageData);
774 pg 252 GIC 3420 : header->pd_upper = header->pd_special;
253 :
774 pg 254 ECB : /* Set safexid in deleted page */
774 pg 255 CBC 3420 : contents = ((BTDeletedPageData *) PageGetContents(page));
256 3420 : contents->safexid = safexid;
774 pg 257 GIC 3420 : }
258 :
774 pg 259 ECB : static inline FullTransactionId
774 pg 260 GIC 721 : BTPageGetDeleteXid(Page page)
261 : {
262 : BTPageOpaque opaque;
263 : BTDeletedPageData *contents;
264 :
774 pg 265 ECB : /* We only expect to be called with a deleted page */
774 pg 266 CBC 721 : Assert(!PageIsNew(page));
373 michael 267 721 : opaque = BTPageGetOpaque(page);
774 pg 268 GIC 721 : Assert(P_ISDELETED(opaque));
269 :
774 pg 270 ECB : /* pg_upgrade'd deleted page -- must be safe to delete now */
774 pg 271 GBC 721 : if (!P_HAS_FULLXID(opaque))
774 pg 272 UIC 0 : return FirstNormalFullTransactionId;
273 :
774 pg 274 ECB : /* Get safexid from deleted page */
774 pg 275 CBC 721 : contents = ((BTDeletedPageData *) PageGetContents(page));
774 pg 276 GIC 721 : return contents->safexid;
277 : }
278 :
279 : /*
280 : * Is an existing page recyclable?
281 : *
282 : * This exists to centralize the policy on which deleted pages are now safe to
283 : * re-use. However, _bt_pendingfsm_finalize() duplicates some of the same
284 : * logic because it doesn't work directly with pages -- keep the two in sync.
285 : *
286 : * Note: PageIsNew() pages are always safe to recycle, but we can't deal with
287 : * them here (caller is responsible for that case themselves). Caller might
288 : * well need special handling for new pages anyway.
289 : */
774 pg 290 ECB : static inline bool
6 pg 291 GNC 33197 : BTPageIsRecyclable(Page page, Relation heaprel)
292 : {
293 : BTPageOpaque opaque;
774 pg 294 ECB :
774 pg 295 GIC 33197 : Assert(!PageIsNew(page));
296 :
774 pg 297 ECB : /* Recycling okay iff page is deleted and safexid is old enough */
373 michael 298 CBC 33197 : opaque = BTPageGetOpaque(page);
774 pg 299 GIC 33197 : if (P_ISDELETED(opaque))
300 : {
301 : /*
302 : * The page was deleted, but when? If it was just deleted, a scan
303 : * might have seen the downlink to it, and will read the page later.
304 : * As long as that can happen, we must keep the deleted page around as
305 : * a tombstone.
306 : *
307 : * For that check if the deletion XID could still be visible to
308 : * anyone. If not, then no scan that's still in progress could have
309 : * seen its downlink, and we can recycle it.
310 : */
6 pg 311 GNC 493 : return GlobalVisCheckRemovableFullXid(heaprel, BTPageGetDeleteXid(page));
312 : }
313 :
774 pg 314 GIC 32704 : return false;
315 : }
316 :
317 : /*
318 : * BTVacState and BTPendingFSM are private nbtree.c state used during VACUUM.
319 : * They are exported for use by page deletion related code in nbtpage.c.
320 : */
321 : typedef struct BTPendingFSM
322 : {
323 : BlockNumber target; /* Page deleted by current VACUUM */
324 : FullTransactionId safexid; /* Page's BTDeletedPageData.safexid */
325 : } BTPendingFSM;
326 :
327 : typedef struct BTVacState
328 : {
329 : IndexVacuumInfo *info;
330 : IndexBulkDeleteResult *stats;
331 : IndexBulkDeleteCallback callback;
332 : void *callback_state;
333 : BTCycleId cycleid;
334 : MemoryContext pagedelcontext;
335 :
336 : /*
337 : * _bt_pendingfsm_finalize() state
338 : */
339 : int bufsize; /* pendingpages space (in # elements) */
340 : int maxbufsize; /* max bufsize that respects work_mem */
341 : BTPendingFSM *pendingpages; /* One entry per newly deleted page */
342 : int npendingpages; /* current # valid pendingpages */
343 : } BTVacState;
344 :
345 : /*
346 : * Lehman and Yao's algorithm requires a ``high key'' on every non-rightmost
347 : * page. The high key is not a tuple that is used to visit the heap. It is
348 : * a pivot tuple (see "Notes on B-Tree tuple format" below for definition).
349 : * The high key on a page is required to be greater than or equal to any
350 : * other key that appears on the page. If we find ourselves trying to
351 : * insert a key that is strictly > high key, we know we need to move right
352 : * (this should only happen if the page was split since we examined the
353 : * parent page).
354 : *
355 : * Our insertion algorithm guarantees that we can use the initial least key
356 : * on our right sibling as the high key. Once a page is created, its high
357 : * key changes only if the page is split.
358 : *
359 : * On a non-rightmost page, the high key lives in item 1 and data items
360 : * start in item 2. Rightmost pages have no high key, so we store data
361 : * items beginning in item 1.
362 : */
363 :
364 : #define P_HIKEY ((OffsetNumber) 1)
365 : #define P_FIRSTKEY ((OffsetNumber) 2)
366 : #define P_FIRSTDATAKEY(opaque) (P_RIGHTMOST(opaque) ? P_HIKEY : P_FIRSTKEY)
367 :
368 : /*
369 : * Notes on B-Tree tuple format, and key and non-key attributes:
370 : *
371 : * INCLUDE B-Tree indexes have non-key attributes. These are extra
372 : * attributes that may be returned by index-only scans, but do not influence
373 : * the order of items in the index (formally, non-key attributes are not
374 : * considered to be part of the key space). Non-key attributes are only
375 : * present in leaf index tuples whose item pointers actually point to heap
376 : * tuples (non-pivot tuples). _bt_check_natts() enforces the rules
377 : * described here.
378 : *
379 : * Non-pivot tuple format (plain/non-posting variant):
380 : *
381 : * t_tid | t_info | key values | INCLUDE columns, if any
382 : *
383 : * t_tid points to the heap TID, which is a tiebreaker key column as of
384 : * BTREE_VERSION 4.
385 : *
386 : * Non-pivot tuples complement pivot tuples, which only have key columns.
387 : * The sole purpose of pivot tuples is to represent how the key space is
388 : * separated. In general, any B-Tree index that has more than one level
389 : * (i.e. any index that does not just consist of a metapage and a single
390 : * leaf root page) must have some number of pivot tuples, since pivot
391 : * tuples are used for traversing the tree. Suffix truncation can omit
392 : * trailing key columns when a new pivot is formed, which makes minus
393 : * infinity their logical value. Since BTREE_VERSION 4 indexes treat heap
394 : * TID as a trailing key column that ensures that all index tuples are
395 : * physically unique, it is necessary to represent heap TID as a trailing
396 : * key column in pivot tuples, though very often this can be truncated
397 : * away, just like any other key column. (Actually, the heap TID is
398 : * omitted rather than truncated, since its representation is different to
399 : * the non-pivot representation.)
400 : *
401 : * Pivot tuple format:
402 : *
403 : * t_tid | t_info | key values | [heap TID]
404 : *
405 : * We store the number of columns present inside pivot tuples by abusing
406 : * their t_tid offset field, since pivot tuples never need to store a real
407 : * offset (pivot tuples generally store a downlink in t_tid, though). The
408 : * offset field only stores the number of columns/attributes when the
409 : * INDEX_ALT_TID_MASK bit is set, which doesn't count the trailing heap
410 : * TID column sometimes stored in pivot tuples -- that's represented by
411 : * the presence of BT_PIVOT_HEAP_TID_ATTR. The INDEX_ALT_TID_MASK bit in
412 : * t_info is always set on BTREE_VERSION 4 pivot tuples, since
413 : * BTreeTupleIsPivot() must work reliably on heapkeyspace versions.
414 : *
415 : * In version 2 or version 3 (!heapkeyspace) indexes, INDEX_ALT_TID_MASK
416 : * might not be set in pivot tuples. BTreeTupleIsPivot() won't work
417 : * reliably as a result. The number of columns stored is implicitly the
418 : * same as the number of columns in the index, just like any non-pivot
419 : * tuple. (The number of columns stored should not vary, since suffix
420 : * truncation of key columns is unsafe within any !heapkeyspace index.)
421 : *
422 : * The 12 least significant bits from t_tid's offset number are used to
423 : * represent the number of key columns within a pivot tuple. This leaves 4
424 : * status bits (BT_STATUS_OFFSET_MASK bits), which are shared by all tuples
425 : * that have the INDEX_ALT_TID_MASK bit set (set in t_info) to store basic
426 : * tuple metadata. BTreeTupleIsPivot() and BTreeTupleIsPosting() use the
427 : * BT_STATUS_OFFSET_MASK bits.
428 : *
429 : * Sometimes non-pivot tuples also use a representation that repurposes
430 : * t_tid to store metadata rather than a TID. PostgreSQL v13 introduced a
431 : * new non-pivot tuple format to support deduplication: posting list
432 : * tuples. Deduplication merges together multiple equal non-pivot tuples
433 : * into a logically equivalent, space efficient representation. A posting
434 : * list is an array of ItemPointerData elements. Non-pivot tuples are
435 : * merged together to form posting list tuples lazily, at the point where
436 : * we'd otherwise have to split a leaf page.
437 : *
438 : * Posting tuple format (alternative non-pivot tuple representation):
439 : *
440 : * t_tid | t_info | key values | posting list (TID array)
441 : *
442 : * Posting list tuples are recognized as such by having the
443 : * INDEX_ALT_TID_MASK status bit set in t_info and the BT_IS_POSTING status
444 : * bit set in t_tid's offset number. These flags redefine the content of
445 : * the posting tuple's t_tid to store the location of the posting list
446 : * (instead of a block number), as well as the total number of heap TIDs
447 : * present in the tuple (instead of a real offset number).
448 : *
449 : * The 12 least significant bits from t_tid's offset number are used to
450 : * represent the number of heap TIDs present in the tuple, leaving 4 status
451 : * bits (the BT_STATUS_OFFSET_MASK bits). Like any non-pivot tuple, the
452 : * number of columns stored is always implicitly the total number in the
453 : * index (in practice there can never be non-key columns stored, since
454 : * deduplication is not supported with INCLUDE indexes).
455 : */
456 : #define INDEX_ALT_TID_MASK INDEX_AM_RESERVED_BIT
457 :
458 : /* Item pointer offset bit masks */
459 : #define BT_OFFSET_MASK 0x0FFF
460 : #define BT_STATUS_OFFSET_MASK 0xF000
461 : /* BT_STATUS_OFFSET_MASK status bits */
462 : #define BT_PIVOT_HEAP_TID_ATTR 0x1000
463 : #define BT_IS_POSTING 0x2000
464 :
465 : /*
466 : * Mask allocated for number of keys in index tuple must be able to fit
467 : * maximum possible number of index attributes
468 : */
469 : StaticAssertDecl(BT_OFFSET_MASK >= INDEX_MAX_KEYS,
470 : "BT_OFFSET_MASK can't fit INDEX_MAX_KEYS");
471 :
472 : /*
473 : * Note: BTreeTupleIsPivot() can have false negatives (but not false
474 : * positives) when used with !heapkeyspace indexes
475 : */
476 : static inline bool
1138 477 710208502 : BTreeTupleIsPivot(IndexTuple itup)
478 : {
1138 pg 479 CBC 710208502 : if ((itup->t_info & INDEX_ALT_TID_MASK) == 0)
1138 pg 480 GIC 484022165 : return false;
1138 pg 481 ECB : /* absence of BT_IS_POSTING in offset number indicates pivot tuple */
1138 pg 482 CBC 226186337 : if ((ItemPointerGetOffsetNumberNoCheck(&itup->t_tid) & BT_IS_POSTING) != 0)
1138 pg 483 GIC 11241898 : return false;
1138 pg 484 ECB :
1138 pg 485 CBC 214944439 : return true;
486 : }
1138 pg 487 ECB :
488 : static inline bool
1138 pg 489 GIC 574156255 : BTreeTupleIsPosting(IndexTuple itup)
490 : {
1138 pg 491 CBC 574156255 : if ((itup->t_info & INDEX_ALT_TID_MASK) == 0)
1138 pg 492 GIC 363628961 : return false;
1138 pg 493 ECB : /* presence of BT_IS_POSTING in offset number indicates posting tuple */
1138 pg 494 CBC 210527294 : if ((ItemPointerGetOffsetNumberNoCheck(&itup->t_tid) & BT_IS_POSTING) == 0)
1138 pg 495 GIC 155203594 : return false;
1138 pg 496 ECB :
1138 pg 497 CBC 55323700 : return true;
498 : }
1138 pg 499 ECB :
500 : static inline void
1082 pg 501 GIC 378731 : BTreeTupleSetPosting(IndexTuple itup, uint16 nhtids, int postingoffset)
502 : {
1082 pg 503 CBC 378731 : Assert(nhtids > 1);
1082 pg 504 GIC 378731 : Assert((nhtids & BT_STATUS_OFFSET_MASK) == 0);
1133 pg 505 CBC 378731 : Assert((size_t) postingoffset == MAXALIGN(postingoffset));
1138 506 378731 : Assert(postingoffset < INDEX_SIZE_MASK);
1082 507 378731 : Assert(!BTreeTupleIsPivot(itup));
1138 pg 508 ECB :
1138 pg 509 CBC 378731 : itup->t_info |= INDEX_ALT_TID_MASK;
1138 pg 510 GIC 378731 : ItemPointerSetOffsetNumber(&itup->t_tid, (nhtids | BT_IS_POSTING));
1138 pg 511 CBC 378731 : ItemPointerSetBlockNumber(&itup->t_tid, postingoffset);
512 378731 : }
1138 pg 513 ECB :
514 : static inline uint16
1138 pg 515 GIC 18767101 : BTreeTupleGetNPosting(IndexTuple posting)
516 : {
1138 pg 517 ECB : OffsetNumber existing;
518 :
1138 pg 519 GIC 18767101 : Assert(BTreeTupleIsPosting(posting));
520 :
1138 pg 521 CBC 18767101 : existing = ItemPointerGetOffsetNumberNoCheck(&posting->t_tid);
1138 pg 522 GIC 18767101 : return (existing & BT_OFFSET_MASK);
1138 pg 523 ECB : }
1816 teodor 524 :
525 : static inline uint32
1138 pg 526 GIC 21080154 : BTreeTupleGetPostingOffset(IndexTuple posting)
527 : {
1138 pg 528 CBC 21080154 : Assert(BTreeTupleIsPosting(posting));
529 :
530 21080154 : return ItemPointerGetBlockNumberNoCheck(&posting->t_tid);
531 : }
1138 pg 532 ECB :
533 : static inline ItemPointer
1138 pg 534 GIC 18895118 : BTreeTupleGetPosting(IndexTuple posting)
535 : {
1138 pg 536 CBC 37790236 : return (ItemPointer) ((char *) posting +
1138 pg 537 GIC 18895118 : BTreeTupleGetPostingOffset(posting));
1138 pg 538 ECB : }
539 :
540 : static inline ItemPointer
1138 pg 541 GIC 15471349 : BTreeTupleGetPostingN(IndexTuple posting, int n)
542 : {
1138 pg 543 CBC 15471349 : return BTreeTupleGetPosting(posting) + n;
544 : }
1828 teodor 545 ECB :
546 : /*
547 : * Get/set downlink block number in pivot tuple.
548 : *
549 : * Note: Cannot assert that tuple is a pivot tuple. If we did so then
550 : * !heapkeyspace indexes would exhibit false positive assertion failures.
551 : */
552 : static inline BlockNumber
1138 pg 553 GIC 12322514 : BTreeTupleGetDownLink(IndexTuple pivot)
554 : {
1138 pg 555 CBC 12322514 : return ItemPointerGetBlockNumberNoCheck(&pivot->t_tid);
556 : }
1138 pg 557 ECB :
558 : static inline void
1138 pg 559 GIC 72893 : BTreeTupleSetDownLink(IndexTuple pivot, BlockNumber blkno)
560 : {
1138 pg 561 CBC 72893 : ItemPointerSetBlockNumber(&pivot->t_tid, blkno);
1138 pg 562 GIC 72893 : }
1812 teodor 563 ECB :
1816 564 : /*
565 : * Get number of attributes within tuple.
566 : *
567 : * Note that this does not include an implicit tiebreaker heap TID
568 : * attribute, if any. Note also that the number of key attributes must be
569 : * explicitly represented in all heapkeyspace pivot tuples.
570 : *
571 : * Note: This is defined as a macro rather than an inline function to
572 : * avoid including rel.h.
573 : */
574 : #define BTreeTupleGetNAtts(itup, rel) \
575 : ( \
576 : (BTreeTupleIsPivot(itup)) ? \
577 : ( \
578 : ItemPointerGetOffsetNumberNoCheck(&(itup)->t_tid) & BT_OFFSET_MASK \
579 : ) \
580 : : \
581 : IndexRelationGetNumberOfAttributes(rel) \
582 : )
583 :
584 : /*
585 : * Set number of key attributes in tuple.
586 : *
587 : * The heap TID tiebreaker attribute bit may also be set here, indicating that
588 : * a heap TID value will be stored at the end of the tuple (i.e. using the
589 : * special pivot tuple representation).
590 : */
591 : static inline void
1097 pg 592 GIC 96252 : BTreeTupleSetNAtts(IndexTuple itup, uint16 nkeyatts, bool heaptid)
593 : {
1097 pg 594 CBC 96252 : Assert(nkeyatts <= INDEX_MAX_KEYS);
1082 pg 595 GIC 96252 : Assert((nkeyatts & BT_STATUS_OFFSET_MASK) == 0);
1097 pg 596 CBC 96252 : Assert(!heaptid || nkeyatts > 0);
597 96252 : Assert(!BTreeTupleIsPivot(itup) || nkeyatts == 0);
1138 pg 598 ECB :
1138 pg 599 CBC 96252 : itup->t_info |= INDEX_ALT_TID_MASK;
600 :
1097 601 96252 : if (heaptid)
1097 pg 602 GIC 516 : nkeyatts |= BT_PIVOT_HEAP_TID_ATTR;
1138 pg 603 ECB :
1097 604 : /* BT_IS_POSTING bit is deliberately unset here */
1097 pg 605 GIC 96252 : ItemPointerSetOffsetNumber(&itup->t_tid, nkeyatts);
606 96252 : Assert(BTreeTupleIsPivot(itup));
1138 pg 607 CBC 96252 : }
1138 pg 608 ECB :
609 : /*
610 : * Get/set leaf page's "top parent" link from its high key. Used during page
611 : * deletion.
612 : *
613 : * Note: Cannot assert that tuple is a pivot tuple. If we did so then
614 : * !heapkeyspace indexes would exhibit false positive assertion failures.
615 : */
616 : static inline BlockNumber
1138 pg 617 GIC 2755 : BTreeTupleGetTopParent(IndexTuple leafhikey)
618 : {
1138 pg 619 CBC 2755 : return ItemPointerGetBlockNumberNoCheck(&leafhikey->t_tid);
620 : }
1138 pg 621 ECB :
622 : static inline void
1138 pg 623 GIC 3420 : BTreeTupleSetTopParent(IndexTuple leafhikey, BlockNumber blkno)
624 : {
1138 pg 625 CBC 3420 : ItemPointerSetBlockNumber(&leafhikey->t_tid, blkno);
1097 pg 626 GIC 3420 : BTreeTupleSetNAtts(leafhikey, 0, false);
1138 pg 627 CBC 3420 : }
1138 pg 628 ECB :
629 : /*
630 : * Get tiebreaker heap TID attribute, if any.
631 : *
632 : * This returns the first/lowest heap TID in the case of a posting list tuple.
633 : */
634 : static inline ItemPointer
1138 pg 635 GIC 72430312 : BTreeTupleGetHeapTID(IndexTuple itup)
636 : {
1138 pg 637 CBC 72430312 : if (BTreeTupleIsPivot(itup))
638 : {
1138 pg 639 ECB : /* Pivot tuple heap TID representation? */
1138 pg 640 GIC 50192192 : if ((ItemPointerGetOffsetNumberNoCheck(&itup->t_tid) &
641 : BT_PIVOT_HEAP_TID_ATTR) != 0)
1138 pg 642 CBC 243318 : return (ItemPointer) ((char *) itup + IndexTupleSize(itup) -
643 : sizeof(ItemPointerData));
1138 pg 644 ECB :
645 : /* Heap TID attribute was truncated */
1138 pg 646 GIC 49948874 : return NULL;
647 : }
1138 pg 648 CBC 22238120 : else if (BTreeTupleIsPosting(itup))
1138 pg 649 GIC 989727 : return BTreeTupleGetPosting(itup);
1138 pg 650 ECB :
1138 pg 651 CBC 21248393 : return &itup->t_tid;
652 : }
1138 pg 653 ECB :
654 : /*
655 : * Get maximum heap TID attribute, which could be the only TID in the case of
656 : * a non-pivot tuple that does not have a posting list tuple.
657 : *
658 : * Works with non-pivot tuples only.
659 : */
660 : static inline ItemPointer
1138 pg 661 GIC 170967 : BTreeTupleGetMaxHeapTID(IndexTuple itup)
662 : {
1138 pg 663 CBC 170967 : Assert(!BTreeTupleIsPivot(itup));
664 :
665 170967 : if (BTreeTupleIsPosting(itup))
666 : {
667 170471 : uint16 nposting = BTreeTupleGetNPosting(itup);
668 :
669 170471 : return BTreeTupleGetPostingN(itup, nposting - 1);
670 : }
1138 pg 671 ECB :
1138 pg 672 GIC 496 : return &itup->t_tid;
673 : }
1481 pg 674 ECB :
675 : /*
676 : * Operator strategy numbers for B-tree have been moved to access/stratnum.h,
677 : * because many places need to use them in ScanKeyInit() calls.
678 : *
679 : * The strategy numbers are chosen so that we can commute them by
680 : * subtraction, thus:
681 : */
682 : #define BTCommuteStrategyNumber(strat) (BTMaxStrategyNumber + 1 - (strat))
683 :
684 : /*
685 : * When a new operator class is declared, we require that the user
686 : * supply us with an amproc procedure (BTORDER_PROC) for determining
687 : * whether, for two keys a and b, a < b, a = b, or a > b. This routine
688 : * must return < 0, 0, > 0, respectively, in these three cases.
689 : *
690 : * To facilitate accelerated sorting, an operator class may choose to
691 : * offer a second procedure (BTSORTSUPPORT_PROC). For full details, see
692 : * src/include/utils/sortsupport.h.
693 : *
694 : * To support window frames defined by "RANGE offset PRECEDING/FOLLOWING",
695 : * an operator class may choose to offer a third amproc procedure
696 : * (BTINRANGE_PROC), independently of whether it offers sortsupport.
697 : * For full details, see doc/src/sgml/btree.sgml.
698 : *
699 : * To facilitate B-Tree deduplication, an operator class may choose to
700 : * offer a forth amproc procedure (BTEQUALIMAGE_PROC). For full details,
701 : * see doc/src/sgml/btree.sgml.
702 : */
703 :
704 : #define BTORDER_PROC 1
705 : #define BTSORTSUPPORT_PROC 2
706 : #define BTINRANGE_PROC 3
707 : #define BTEQUALIMAGE_PROC 4
708 : #define BTOPTIONS_PROC 5
709 : #define BTNProcs 5
710 :
711 : /*
712 : * We need to be able to tell the difference between read and write
713 : * requests for pages, in order to do locking correctly.
714 : */
715 :
716 : #define BT_READ BUFFER_LOCK_SHARE
717 : #define BT_WRITE BUFFER_LOCK_EXCLUSIVE
718 :
719 : /*
720 : * BTStackData -- As we descend a tree, we push the location of pivot
721 : * tuples whose downlink we are about to follow onto a private stack. If
722 : * we split a leaf, we use this stack to walk back up the tree and insert
723 : * data into its parent page at the correct location. We also have to
724 : * recursively insert into the grandparent page if and when the parent page
725 : * splits. Our private stack can become stale due to concurrent page
726 : * splits and page deletions, but it should never give us an irredeemably
727 : * bad picture.
728 : */
729 : typedef struct BTStackData
730 : {
731 : BlockNumber bts_blkno;
732 : OffsetNumber bts_offset;
733 : struct BTStackData *bts_parent;
734 : } BTStackData;
735 :
736 : typedef BTStackData *BTStack;
737 :
738 : /*
739 : * BTScanInsertData is the btree-private state needed to find an initial
740 : * position for an indexscan, or to insert new tuples -- an "insertion
741 : * scankey" (not to be confused with a search scankey). It's used to descend
742 : * a B-Tree using _bt_search.
743 : *
744 : * heapkeyspace indicates if we expect all keys in the index to be physically
745 : * unique because heap TID is used as a tiebreaker attribute, and if index may
746 : * have truncated key attributes in pivot tuples. This is actually a property
747 : * of the index relation itself (not an indexscan). heapkeyspace indexes are
748 : * indexes whose version is >= version 4. It's convenient to keep this close
749 : * by, rather than accessing the metapage repeatedly.
750 : *
751 : * allequalimage is set to indicate that deduplication is safe for the index.
752 : * This is also a property of the index relation rather than an indexscan.
753 : *
754 : * anynullkeys indicates if any of the keys had NULL value when scankey was
755 : * built from index tuple (note that already-truncated tuple key attributes
756 : * set NULL as a placeholder key value, which also affects value of
757 : * anynullkeys). This is a convenience for unique index non-pivot tuple
758 : * insertion, which usually temporarily unsets scantid, but shouldn't iff
759 : * anynullkeys is true. Value generally matches non-pivot tuple's HasNulls
760 : * bit, but may not when inserting into an INCLUDE index (tuple header value
761 : * is affected by the NULL-ness of both key and non-key attributes).
762 : *
763 : * When nextkey is false (the usual case), _bt_search and _bt_binsrch will
764 : * locate the first item >= scankey. When nextkey is true, they will locate
765 : * the first item > scan key.
766 : *
767 : * pivotsearch is set to true by callers that want to re-find a leaf page
768 : * using a scankey built from a leaf page's high key. Most callers set this
769 : * to false.
770 : *
771 : * scantid is the heap TID that is used as a final tiebreaker attribute. It
772 : * is set to NULL when index scan doesn't need to find a position for a
773 : * specific physical tuple. Must be set when inserting new tuples into
774 : * heapkeyspace indexes, since every tuple in the tree unambiguously belongs
775 : * in one exact position (it's never set with !heapkeyspace indexes, though).
776 : * Despite the representational difference, nbtree search code considers
777 : * scantid to be just another insertion scankey attribute.
778 : *
779 : * scankeys is an array of scan key entries for attributes that are compared
780 : * before scantid (user-visible attributes). keysz is the size of the array.
781 : * During insertion, there must be a scan key for every attribute, but when
782 : * starting a regular index scan some can be omitted. The array is used as a
783 : * flexible array member, though it's sized in a way that makes it possible to
784 : * use stack allocations. See nbtree/README for full details.
785 : */
786 : typedef struct BTScanInsertData
787 : {
788 : bool heapkeyspace;
789 : bool allequalimage;
790 : bool anynullkeys;
791 : bool nextkey;
792 : bool pivotsearch;
793 : ItemPointer scantid; /* tiebreaker for scankeys */
794 : int keysz; /* Size of scankeys array */
795 : ScanKeyData scankeys[INDEX_MAX_KEYS]; /* Must appear last */
796 : } BTScanInsertData;
797 :
798 : typedef BTScanInsertData *BTScanInsert;
799 :
800 : /*
801 : * BTInsertStateData is a working area used during insertion.
802 : *
803 : * This is filled in after descending the tree to the first leaf page the new
804 : * tuple might belong on. Tracks the current position while performing
805 : * uniqueness check, before we have determined which exact page to insert
806 : * to.
807 : *
808 : * (This should be private to nbtinsert.c, but it's also used by
809 : * _bt_binsrch_insert)
810 : */
811 : typedef struct BTInsertStateData
812 : {
813 : IndexTuple itup; /* Item we're inserting */
814 : Size itemsz; /* Size of itup -- should be MAXALIGN()'d */
815 : BTScanInsert itup_key; /* Insertion scankey */
816 :
817 : /* Buffer containing leaf page we're likely to insert itup on */
818 : Buffer buf;
819 :
820 : /*
821 : * Cache of bounds within the current buffer. Only used for insertions
822 : * where _bt_check_unique is called. See _bt_binsrch_insert and
823 : * _bt_findinsertloc for details.
824 : */
825 : bool bounds_valid;
826 : OffsetNumber low;
827 : OffsetNumber stricthigh;
828 :
829 : /*
830 : * if _bt_binsrch_insert found the location inside existing posting list,
831 : * save the position inside the list. -1 sentinel value indicates overlap
832 : * with an existing posting list tuple that has its LP_DEAD bit set.
833 : */
834 : int postingoff;
835 : } BTInsertStateData;
836 :
837 : typedef BTInsertStateData *BTInsertState;
838 :
839 : /*
840 : * State used to representing an individual pending tuple during
841 : * deduplication.
842 : */
843 : typedef struct BTDedupInterval
844 : {
845 : OffsetNumber baseoff;
846 : uint16 nitems;
847 : } BTDedupInterval;
848 :
849 : /*
850 : * BTDedupStateData is a working area used during deduplication.
851 : *
852 : * The status info fields track the state of a whole-page deduplication pass.
853 : * State about the current pending posting list is also tracked.
854 : *
855 : * A pending posting list is comprised of a contiguous group of equal items
856 : * from the page, starting from page offset number 'baseoff'. This is the
857 : * offset number of the "base" tuple for new posting list. 'nitems' is the
858 : * current total number of existing items from the page that will be merged to
859 : * make a new posting list tuple, including the base tuple item. (Existing
860 : * items may themselves be posting list tuples, or regular non-pivot tuples.)
861 : *
862 : * The total size of the existing tuples to be freed when pending posting list
863 : * is processed gets tracked by 'phystupsize'. This information allows
864 : * deduplication to calculate the space saving for each new posting list
865 : * tuple, and for the entire pass over the page as a whole.
866 : */
867 : typedef struct BTDedupStateData
868 : {
869 : /* Deduplication status info for entire pass over page */
870 : bool deduplicate; /* Still deduplicating page? */
871 : int nmaxitems; /* Number of max-sized tuples so far */
872 : Size maxpostingsize; /* Limit on size of final tuple */
873 :
874 : /* Metadata about base tuple of current pending posting list */
875 : IndexTuple base; /* Use to form new posting list */
876 : OffsetNumber baseoff; /* page offset of base */
877 : Size basetupsize; /* base size without original posting list */
878 :
879 : /* Other metadata about pending posting list */
880 : ItemPointer htids; /* Heap TIDs in pending posting list */
881 : int nhtids; /* Number of heap TIDs in htids array */
882 : int nitems; /* Number of existing tuples/line pointers */
883 : Size phystupsize; /* Includes line pointer overhead */
884 :
885 : /*
886 : * Array of tuples to go on new version of the page. Contains one entry
887 : * for each group of consecutive items. Note that existing tuples that
888 : * will not become posting list tuples do not appear in the array (they
889 : * are implicitly unchanged by deduplication pass).
890 : */
891 : int nintervals; /* current number of intervals in array */
892 : BTDedupInterval intervals[MaxIndexTuplesPerPage];
893 : } BTDedupStateData;
894 :
895 : typedef BTDedupStateData *BTDedupState;
896 :
897 : /*
898 : * BTVacuumPostingData is state that represents how to VACUUM (or delete) a
899 : * posting list tuple when some (though not all) of its TIDs are to be
900 : * deleted.
901 : *
902 : * Convention is that itup field is the original posting list tuple on input,
903 : * and palloc()'d final tuple used to overwrite existing tuple on output.
904 : */
905 : typedef struct BTVacuumPostingData
906 : {
907 : /* Tuple that will be/was updated */
908 : IndexTuple itup;
909 : OffsetNumber updatedoffset;
910 :
911 : /* State needed to describe final itup in WAL */
912 : uint16 ndeletedtids;
913 : uint16 deletetids[FLEXIBLE_ARRAY_MEMBER];
914 : } BTVacuumPostingData;
915 :
916 : typedef BTVacuumPostingData *BTVacuumPosting;
917 :
918 : /*
919 : * BTScanOpaqueData is the btree-private state needed for an indexscan.
920 : * This consists of preprocessed scan keys (see _bt_preprocess_keys() for
921 : * details of the preprocessing), information about the current location
922 : * of the scan, and information about the marked location, if any. (We use
923 : * BTScanPosData to represent the data needed for each of current and marked
924 : * locations.) In addition we can remember some known-killed index entries
925 : * that must be marked before we can move off the current page.
926 : *
927 : * Index scans work a page at a time: we pin and read-lock the page, identify
928 : * all the matching items on the page and save them in BTScanPosData, then
929 : * release the read-lock while returning the items to the caller for
930 : * processing. This approach minimizes lock/unlock traffic. Note that we
931 : * keep the pin on the index page until the caller is done with all the items
932 : * (this is needed for VACUUM synchronization, see nbtree/README). When we
933 : * are ready to step to the next page, if the caller has told us any of the
934 : * items were killed, we re-lock the page to mark them killed, then unlock.
935 : * Finally we drop the pin and step to the next page in the appropriate
936 : * direction.
937 : *
938 : * If we are doing an index-only scan, we save the entire IndexTuple for each
939 : * matched item, otherwise only its heap TID and offset. The IndexTuples go
940 : * into a separate workspace array; each BTScanPosItem stores its tuple's
941 : * offset within that array. Posting list tuples store a "base" tuple once,
942 : * allowing the same key to be returned for each TID in the posting list
943 : * tuple.
944 : */
945 :
946 : typedef struct BTScanPosItem /* what we remember about each match */
947 : {
948 : ItemPointerData heapTid; /* TID of referenced heap item */
949 : OffsetNumber indexOffset; /* index item's location within page */
950 : LocationIndex tupleOffset; /* IndexTuple's offset in workspace, if any */
951 : } BTScanPosItem;
952 :
953 : typedef struct BTScanPosData
954 : {
955 : Buffer buf; /* if valid, the buffer is pinned */
956 :
957 : XLogRecPtr lsn; /* pos in the WAL stream when page was read */
958 : BlockNumber currPage; /* page referenced by items array */
959 : BlockNumber nextPage; /* page's right link when we scanned it */
960 :
961 : /*
962 : * moreLeft and moreRight track whether we think there may be matching
963 : * index entries to the left and right of the current page, respectively.
964 : * We can clear the appropriate one of these flags when _bt_checkkeys()
965 : * returns continuescan = false.
966 : */
967 : bool moreLeft;
968 : bool moreRight;
969 :
970 : /*
971 : * If we are doing an index-only scan, nextTupleOffset is the first free
972 : * location in the associated tuple storage workspace.
973 : */
974 : int nextTupleOffset;
975 :
976 : /*
977 : * The items array is always ordered in index order (ie, increasing
978 : * indexoffset). When scanning backwards it is convenient to fill the
979 : * array back-to-front, so we start at the last slot and fill downwards.
980 : * Hence we need both a first-valid-entry and a last-valid-entry counter.
981 : * itemIndex is a cursor showing which entry was last returned to caller.
982 : */
983 : int firstItem; /* first valid index in items[] */
984 : int lastItem; /* last valid index in items[] */
985 : int itemIndex; /* current index in items[] */
986 :
987 : BTScanPosItem items[MaxTIDsPerBTreePage]; /* MUST BE LAST */
988 : } BTScanPosData;
989 :
990 : typedef BTScanPosData *BTScanPos;
991 :
992 : #define BTScanPosIsPinned(scanpos) \
993 : ( \
994 : AssertMacro(BlockNumberIsValid((scanpos).currPage) || \
995 : !BufferIsValid((scanpos).buf)), \
996 : BufferIsValid((scanpos).buf) \
997 : )
998 : #define BTScanPosUnpin(scanpos) \
999 : do { \
1000 : ReleaseBuffer((scanpos).buf); \
1001 : (scanpos).buf = InvalidBuffer; \
1002 : } while (0)
1003 : #define BTScanPosUnpinIfPinned(scanpos) \
1004 : do { \
1005 : if (BTScanPosIsPinned(scanpos)) \
1006 : BTScanPosUnpin(scanpos); \
1007 : } while (0)
1008 :
1009 : #define BTScanPosIsValid(scanpos) \
1010 : ( \
1011 : AssertMacro(BlockNumberIsValid((scanpos).currPage) || \
1012 : !BufferIsValid((scanpos).buf)), \
1013 : BlockNumberIsValid((scanpos).currPage) \
1014 : )
1015 : #define BTScanPosInvalidate(scanpos) \
1016 : do { \
1017 : (scanpos).currPage = InvalidBlockNumber; \
1018 : (scanpos).nextPage = InvalidBlockNumber; \
1019 : (scanpos).buf = InvalidBuffer; \
1020 : (scanpos).lsn = InvalidXLogRecPtr; \
1021 : (scanpos).nextTupleOffset = 0; \
1022 : } while (0)
1023 :
1024 : /* We need one of these for each equality-type SK_SEARCHARRAY scan key */
1025 : typedef struct BTArrayKeyInfo
1026 : {
1027 : int scan_key; /* index of associated key in arrayKeyData */
1028 : int cur_elem; /* index of current element in elem_values */
1029 : int mark_elem; /* index of marked element in elem_values */
1030 : int num_elems; /* number of elems in current array value */
1031 : Datum *elem_values; /* array of num_elems Datums */
1032 : } BTArrayKeyInfo;
1033 :
1034 : typedef struct BTScanOpaqueData
1035 : {
1036 : /* these fields are set by _bt_preprocess_keys(): */
1037 : bool qual_ok; /* false if qual can never be satisfied */
1038 : int numberOfKeys; /* number of preprocessed scan keys */
1039 : ScanKey keyData; /* array of preprocessed scan keys */
1040 :
1041 : /* workspace for SK_SEARCHARRAY support */
1042 : ScanKey arrayKeyData; /* modified copy of scan->keyData */
1043 : int numArrayKeys; /* number of equality-type array keys (-1 if
1044 : * there are any unsatisfiable array keys) */
1045 : int arrayKeyCount; /* count indicating number of array scan keys
1046 : * processed */
1047 : BTArrayKeyInfo *arrayKeys; /* info about each equality-type array key */
1048 : MemoryContext arrayContext; /* scan-lifespan context for array data */
1049 :
1050 : /* info about killed items if any (killedItems is NULL if never used) */
1051 : int *killedItems; /* currPos.items indexes of killed items */
1052 : int numKilled; /* number of currently stored items */
1053 :
1054 : /*
1055 : * If we are doing an index-only scan, these are the tuple storage
1056 : * workspaces for the currPos and markPos respectively. Each is of size
1057 : * BLCKSZ, so it can hold as much as a full page's worth of tuples.
1058 : */
1059 : char *currTuples; /* tuple storage for currPos */
1060 : char *markTuples; /* tuple storage for markPos */
1061 :
1062 : /*
1063 : * If the marked position is on the same page as current position, we
1064 : * don't use markPos, but just keep the marked itemIndex in markItemIndex
1065 : * (all the rest of currPos is valid for the mark position). Hence, to
1066 : * determine if there is a mark, first look at markItemIndex, then at
1067 : * markPos.
1068 : */
1069 : int markItemIndex; /* itemIndex, or -1 if not valid */
1070 :
1071 : /* keep these last in struct for efficiency */
1072 : BTScanPosData currPos; /* current position data */
1073 : BTScanPosData markPos; /* marked position, if any */
1074 : } BTScanOpaqueData;
1075 :
1076 : typedef BTScanOpaqueData *BTScanOpaque;
1077 :
1078 : /*
1079 : * We use some private sk_flags bits in preprocessed scan keys. We're allowed
1080 : * to use bits 16-31 (see skey.h). The uppermost bits are copied from the
1081 : * index's indoption[] array entry for the index attribute.
1082 : */
1083 : #define SK_BT_REQFWD 0x00010000 /* required to continue forward scan */
1084 : #define SK_BT_REQBKWD 0x00020000 /* required to continue backward scan */
1085 : #define SK_BT_INDOPTION_SHIFT 24 /* must clear the above bits */
1086 : #define SK_BT_DESC (INDOPTION_DESC << SK_BT_INDOPTION_SHIFT)
1087 : #define SK_BT_NULLS_FIRST (INDOPTION_NULLS_FIRST << SK_BT_INDOPTION_SHIFT)
1088 :
1089 : typedef struct BTOptions
1090 : {
1091 : int32 varlena_header_; /* varlena header (do not touch directly!) */
1092 : int fillfactor; /* page fill factor in percent (0..100) */
1093 : float8 vacuum_cleanup_index_scale_factor; /* deprecated */
1094 : bool deduplicate_items; /* Try to deduplicate items? */
1095 : } BTOptions;
1096 :
1097 : #define BTGetFillFactor(relation) \
1098 : (AssertMacro(relation->rd_rel->relkind == RELKIND_INDEX && \
1099 : relation->rd_rel->relam == BTREE_AM_OID), \
1100 : (relation)->rd_options ? \
1101 : ((BTOptions *) (relation)->rd_options)->fillfactor : \
1102 : BTREE_DEFAULT_FILLFACTOR)
1103 : #define BTGetTargetPageFreeSpace(relation) \
1104 : (BLCKSZ * (100 - BTGetFillFactor(relation)) / 100)
1105 : #define BTGetDeduplicateItems(relation) \
1106 : (AssertMacro(relation->rd_rel->relkind == RELKIND_INDEX && \
1107 : relation->rd_rel->relam == BTREE_AM_OID), \
1108 : ((relation)->rd_options ? \
1109 : ((BTOptions *) (relation)->rd_options)->deduplicate_items : true))
1110 :
1111 : /*
1112 : * Constant definition for progress reporting. Phase numbers must match
1113 : * btbuildphasename.
1114 : */
1115 : /* PROGRESS_CREATEIDX_SUBPHASE_INITIALIZE is 1 (see progress.h) */
1116 : #define PROGRESS_BTREE_PHASE_INDEXBUILD_TABLESCAN 2
1117 : #define PROGRESS_BTREE_PHASE_PERFORMSORT_1 3
1118 : #define PROGRESS_BTREE_PHASE_PERFORMSORT_2 4
1119 : #define PROGRESS_BTREE_PHASE_LEAF_LOAD 5
1120 :
1121 : /*
1122 : * external entry points for btree, in nbtree.c
1123 : */
1124 : extern void btbuildempty(Relation index);
1125 : extern bool btinsert(Relation rel, Datum *values, bool *isnull,
1126 : ItemPointer ht_ctid, Relation heapRel,
1127 : IndexUniqueCheck checkUnique,
1128 : bool indexUnchanged,
1129 : struct IndexInfo *indexInfo);
1130 : extern IndexScanDesc btbeginscan(Relation rel, int nkeys, int norderbys);
1131 : extern Size btestimateparallelscan(void);
1132 : extern void btinitparallelscan(void *target);
1133 : extern bool btgettuple(IndexScanDesc scan, ScanDirection dir);
1134 : extern int64 btgetbitmap(IndexScanDesc scan, TIDBitmap *tbm);
1135 : extern void btrescan(IndexScanDesc scan, ScanKey scankey, int nscankeys,
1136 : ScanKey orderbys, int norderbys);
1137 : extern void btparallelrescan(IndexScanDesc scan);
1138 : extern void btendscan(IndexScanDesc scan);
1139 : extern void btmarkpos(IndexScanDesc scan);
1140 : extern void btrestrpos(IndexScanDesc scan);
1141 : extern IndexBulkDeleteResult *btbulkdelete(IndexVacuumInfo *info,
1142 : IndexBulkDeleteResult *stats,
1143 : IndexBulkDeleteCallback callback,
1144 : void *callback_state);
1145 : extern IndexBulkDeleteResult *btvacuumcleanup(IndexVacuumInfo *info,
1146 : IndexBulkDeleteResult *stats);
1147 : extern bool btcanreturn(Relation index, int attno);
1148 :
1149 : /*
1150 : * prototypes for internal functions in nbtree.c
1151 : */
1152 : extern bool _bt_parallel_seize(IndexScanDesc scan, BlockNumber *pageno);
1153 : extern void _bt_parallel_release(IndexScanDesc scan, BlockNumber scan_page);
1154 : extern void _bt_parallel_done(IndexScanDesc scan);
1155 : extern void _bt_parallel_advance_array_keys(IndexScanDesc scan);
1156 :
1157 : /*
1158 : * prototypes for functions in nbtdedup.c
1159 : */
1160 : extern void _bt_dedup_pass(Relation rel, Buffer buf, Relation heapRel,
1161 : IndexTuple newitem, Size newitemsz,
1162 : bool bottomupdedup);
1163 : extern bool _bt_bottomupdel_pass(Relation rel, Buffer buf, Relation heapRel,
1164 : Size newitemsz);
1165 : extern void _bt_dedup_start_pending(BTDedupState state, IndexTuple base,
1166 : OffsetNumber baseoff);
1167 : extern bool _bt_dedup_save_htid(BTDedupState state, IndexTuple itup);
1168 : extern Size _bt_dedup_finish_pending(Page newpage, BTDedupState state);
1169 : extern IndexTuple _bt_form_posting(IndexTuple base, ItemPointer htids,
1170 : int nhtids);
1171 : extern void _bt_update_posting(BTVacuumPosting vacposting);
1172 : extern IndexTuple _bt_swap_posting(IndexTuple newitem, IndexTuple oposting,
1173 : int postingoff);
1174 :
1175 : /*
1176 : * prototypes for functions in nbtinsert.c
1177 : */
1178 : extern bool _bt_doinsert(Relation rel, IndexTuple itup,
1179 : IndexUniqueCheck checkUnique, bool indexUnchanged,
1180 : Relation heapRel);
1181 : extern void _bt_finish_split(Relation rel, Relation heaprel, Buffer lbuf,
1182 : BTStack stack);
1183 : extern Buffer _bt_getstackbuf(Relation rel, Relation heaprel, BTStack stack,
1184 : BlockNumber child);
1185 :
1186 : /*
1187 : * prototypes for functions in nbtsplitloc.c
1188 : */
1189 : extern OffsetNumber _bt_findsplitloc(Relation rel, Page origpage,
1190 : OffsetNumber newitemoff, Size newitemsz, IndexTuple newitem,
1191 : bool *newitemonleft);
1192 :
1193 : /*
1194 : * prototypes for functions in nbtpage.c
1195 : */
1196 : extern void _bt_initmetapage(Page page, BlockNumber rootbknum, uint32 level,
1197 : bool allequalimage);
1198 : extern bool _bt_vacuum_needs_cleanup(Relation rel, Relation heaprel);
1199 : extern void _bt_set_cleanup_info(Relation rel, Relation heaprel,
1200 : BlockNumber num_delpages);
1201 : extern void _bt_upgrademetapage(Page page);
1202 : extern Buffer _bt_getroot(Relation rel, Relation heaprel, int access);
1203 : extern Buffer _bt_gettrueroot(Relation rel, Relation heaprel);
1204 : extern int _bt_getrootheight(Relation rel, Relation heaprel);
1205 : extern void _bt_metaversion(Relation rel, Relation heaprel, bool *heapkeyspace,
1206 : bool *allequalimage);
1207 : extern void _bt_checkpage(Relation rel, Buffer buf);
1208 : extern Buffer _bt_getbuf(Relation rel, Relation heaprel, BlockNumber blkno,
1209 : int access);
1210 : extern Buffer _bt_relandgetbuf(Relation rel, Buffer obuf,
1211 : BlockNumber blkno, int access);
1212 : extern void _bt_relbuf(Relation rel, Buffer buf);
1213 : extern void _bt_lockbuf(Relation rel, Buffer buf, int access);
1214 : extern void _bt_unlockbuf(Relation rel, Buffer buf);
1215 : extern bool _bt_conditionallockbuf(Relation rel, Buffer buf);
1216 : extern void _bt_upgradelockbufcleanup(Relation rel, Buffer buf);
1217 : extern void _bt_pageinit(Page page, Size size);
1218 : extern void _bt_delitems_vacuum(Relation rel, Buffer buf,
1219 : OffsetNumber *deletable, int ndeletable,
1220 : BTVacuumPosting *updatable, int nupdatable);
1221 : extern void _bt_delitems_delete_check(Relation rel, Buffer buf,
1222 : Relation heapRel,
1223 : TM_IndexDeleteOp *delstate);
1224 : extern void _bt_pagedel(Relation rel, Buffer leafbuf, BTVacState *vstate);
1225 : extern void _bt_pendingfsm_init(Relation rel, BTVacState *vstate,
1226 : bool cleanuponly);
1227 : extern void _bt_pendingfsm_finalize(Relation rel, BTVacState *vstate);
1228 :
1229 : /*
1230 : * prototypes for functions in nbtsearch.c
1231 : */
1232 : extern BTStack _bt_search(Relation rel, Relation heaprel, BTScanInsert key,
1233 : Buffer *bufP, int access, Snapshot snapshot);
1234 : extern Buffer _bt_moveright(Relation rel, Relation heaprel, BTScanInsert key,
1235 : Buffer buf, bool forupdate, BTStack stack,
1236 : int access, Snapshot snapshot);
1237 : extern OffsetNumber _bt_binsrch_insert(Relation rel, BTInsertState insertstate);
1238 : extern int32 _bt_compare(Relation rel, BTScanInsert key, Page page, OffsetNumber offnum);
1239 : extern bool _bt_first(IndexScanDesc scan, ScanDirection dir);
1240 : extern bool _bt_next(IndexScanDesc scan, ScanDirection dir);
1241 : extern Buffer _bt_get_endpoint(Relation rel, Relation heaprel, uint32 level,
1242 : bool rightmost, Snapshot snapshot);
1243 :
1244 : /*
1245 : * prototypes for functions in nbtutils.c
1246 : */
1247 : extern BTScanInsert _bt_mkscankey(Relation rel, Relation heaprel, IndexTuple itup);
1248 : extern void _bt_freestack(BTStack stack);
1249 : extern void _bt_preprocess_array_keys(IndexScanDesc scan);
1250 : extern void _bt_start_array_keys(IndexScanDesc scan, ScanDirection dir);
1251 : extern bool _bt_advance_array_keys(IndexScanDesc scan, ScanDirection dir);
1252 : extern void _bt_mark_array_keys(IndexScanDesc scan);
1253 : extern void _bt_restore_array_keys(IndexScanDesc scan);
1254 : extern void _bt_preprocess_keys(IndexScanDesc scan);
1255 : extern bool _bt_checkkeys(IndexScanDesc scan, IndexTuple tuple,
1256 : int tupnatts, ScanDirection dir, bool *continuescan);
1257 : extern void _bt_killitems(IndexScanDesc scan);
1258 : extern BTCycleId _bt_vacuum_cycleid(Relation rel);
1259 : extern BTCycleId _bt_start_vacuum(Relation rel);
1260 : extern void _bt_end_vacuum(Relation rel);
1261 : extern void _bt_end_vacuum_callback(int code, Datum arg);
1262 : extern Size BTreeShmemSize(void);
1263 : extern void BTreeShmemInit(void);
1264 : extern bytea *btoptions(Datum reloptions, bool validate);
1265 : extern bool btproperty(Oid index_oid, int attno,
1266 : IndexAMProperty prop, const char *propname,
1267 : bool *res, bool *isnull);
1268 : extern char *btbuildphasename(int64 phasenum);
1269 : extern IndexTuple _bt_truncate(Relation rel, IndexTuple lastleft,
1270 : IndexTuple firstright, BTScanInsert itup_key);
1271 : extern int _bt_keep_natts_fast(Relation rel, IndexTuple lastleft,
1272 : IndexTuple firstright);
1273 : extern bool _bt_check_natts(Relation rel, bool heapkeyspace, Page page,
1274 : OffsetNumber offnum);
1275 : extern void _bt_check_third_page(Relation rel, Relation heap,
1276 : bool needheaptidspace, Page page, IndexTuple newtup);
1277 : extern bool _bt_allequalimage(Relation rel, bool debugmessage);
1278 :
1279 : /*
1280 : * prototypes for functions in nbtvalidate.c
1281 : */
1282 : extern bool btvalidate(Oid opclassoid);
1283 : extern void btadjustmembers(Oid opfamilyoid,
1284 : Oid opclassoid,
1285 : List *operators,
1286 : List *functions);
1287 :
1288 : /*
1289 : * prototypes for functions in nbtsort.c
1290 : */
1291 : extern IndexBuildResult *btbuild(Relation heap, Relation index,
1292 : struct IndexInfo *indexInfo);
1293 : extern void _bt_parallel_build_main(dsm_segment *seg, shm_toc *toc);
1294 :
1295 : #endif /* NBTREE_H */
|
