Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * syncscan.c
4 : * scan synchronization support
5 : *
6 : * When multiple backends run a sequential scan on the same table, we try
7 : * to keep them synchronized to reduce the overall I/O needed. The goal is
8 : * to read each page into shared buffer cache only once, and let all backends
9 : * that take part in the shared scan process the page before it falls out of
10 : * the cache.
11 : *
12 : * Since the "leader" in a pack of backends doing a seqscan will have to wait
13 : * for I/O, while the "followers" don't, there is a strong self-synchronizing
14 : * effect once we can get the backends examining approximately the same part
15 : * of the table at the same time. Hence all that is really needed is to get
16 : * a new backend beginning a seqscan to begin it close to where other backends
17 : * are reading. We can scan the table circularly, from block X up to the
18 : * end and then from block 0 to X-1, to ensure we visit all rows while still
19 : * participating in the common scan.
20 : *
21 : * To accomplish that, we keep track of the scan position of each table, and
22 : * start new scans close to where the previous scan(s) are. We don't try to
23 : * do any extra synchronization to keep the scans together afterwards; some
24 : * scans might progress much more slowly than others, for example if the
25 : * results need to be transferred to the client over a slow network, and we
26 : * don't want such queries to slow down others.
27 : *
28 : * There can realistically only be a few large sequential scans on different
29 : * tables in progress at any time. Therefore we just keep the scan positions
30 : * in a small LRU list which we scan every time we need to look up or update a
31 : * scan position. The whole mechanism is only applied for tables exceeding
32 : * a threshold size (but that is not the concern of this module).
33 : *
34 : * INTERFACE ROUTINES
35 : * ss_get_location - return current scan location of a relation
36 : * ss_report_location - update current scan location
37 : *
38 : *
39 : * Portions Copyright (c) 1996-2026, PostgreSQL Global Development Group
40 : * Portions Copyright (c) 1994, Regents of the University of California
41 : *
42 : * IDENTIFICATION
43 : * src/backend/access/common/syncscan.c
44 : *
45 : *-------------------------------------------------------------------------
46 : */
47 : #include "postgres.h"
48 :
49 : #include "access/syncscan.h"
50 : #include "miscadmin.h"
51 : #include "storage/lwlock.h"
52 : #include "storage/shmem.h"
53 : #include "storage/subsystems.h"
54 : #include "utils/rel.h"
55 :
56 :
57 : /* GUC variables */
58 : #ifdef TRACE_SYNCSCAN
59 : bool trace_syncscan = false;
60 : #endif
61 :
62 :
63 : /*
64 : * Size of the LRU list.
65 : *
66 : * Note: the code assumes that SYNC_SCAN_NELEM > 1.
67 : *
68 : * XXX: What's a good value? It should be large enough to hold the
69 : * maximum number of large tables scanned simultaneously. But a larger value
70 : * means more traversing of the LRU list when starting a new scan.
71 : */
72 : #define SYNC_SCAN_NELEM 20
73 :
74 : /*
75 : * Interval between reports of the location of the current scan, in pages.
76 : *
77 : * Note: This should be smaller than the ring size (see buffer/freelist.c)
78 : * we use for bulk reads. Otherwise a scan joining other scans might start
79 : * from a page that's no longer in the buffer cache. This is a bit fuzzy;
80 : * there's no guarantee that the new scan will read the page before it leaves
81 : * the buffer cache anyway, and on the other hand the page is most likely
82 : * still in the OS cache.
83 : */
84 : #define SYNC_SCAN_REPORT_INTERVAL (128 * 1024 / BLCKSZ)
85 :
86 :
87 : /*
88 : * The scan locations structure is essentially a doubly-linked LRU with head
89 : * and tail pointer, but designed to hold a fixed maximum number of elements in
90 : * fixed-size shared memory.
91 : */
92 : typedef struct ss_scan_location_t
93 : {
94 : RelFileLocator relfilelocator; /* identity of a relation */
95 : BlockNumber location; /* last-reported location in the relation */
96 : } ss_scan_location_t;
97 :
98 : typedef struct ss_lru_item_t
99 : {
100 : struct ss_lru_item_t *prev;
101 : struct ss_lru_item_t *next;
102 : ss_scan_location_t location;
103 : } ss_lru_item_t;
104 :
105 : typedef struct ss_scan_locations_t
106 : {
107 : ss_lru_item_t *head;
108 : ss_lru_item_t *tail;
109 : ss_lru_item_t items[FLEXIBLE_ARRAY_MEMBER]; /* SYNC_SCAN_NELEM items */
110 : } ss_scan_locations_t;
111 :
112 : #define SizeOfScanLocations(N) \
113 : (offsetof(ss_scan_locations_t, items) + (N) * sizeof(ss_lru_item_t))
114 :
115 : static void SyncScanShmemRequest(void *arg);
116 : static void SyncScanShmemInit(void *arg);
117 :
118 : const ShmemCallbacks SyncScanShmemCallbacks = {
119 : .request_fn = SyncScanShmemRequest,
120 : .init_fn = SyncScanShmemInit,
121 : };
122 :
123 : /* Pointer to struct in shared memory */
124 : static ss_scan_locations_t *scan_locations;
125 :
126 : /* prototypes for internal functions */
127 : static BlockNumber ss_search(RelFileLocator relfilelocator,
128 : BlockNumber location, bool set);
129 :
130 :
131 : /*
132 : * SyncScanShmemRequest --- register this module's shared memory
133 : */
134 : static void
135 1234 : SyncScanShmemRequest(void *arg)
136 : {
137 1234 : ShmemRequestStruct(.name = "Sync Scan Locations List",
138 : .size = SizeOfScanLocations(SYNC_SCAN_NELEM),
139 : .ptr = (void **) &scan_locations,
140 : );
141 1234 : }
142 :
143 : /*
144 : * SyncScanShmemInit --- initialize this module's shared memory
145 : */
146 : static void
147 1231 : SyncScanShmemInit(void *arg)
148 : {
149 : int i;
150 :
151 1231 : scan_locations->head = &scan_locations->items[0];
152 1231 : scan_locations->tail = &scan_locations->items[SYNC_SCAN_NELEM - 1];
153 :
154 25851 : for (i = 0; i < SYNC_SCAN_NELEM; i++)
155 : {
156 24620 : ss_lru_item_t *item = &scan_locations->items[i];
157 :
158 : /*
159 : * Initialize all slots with invalid values. As scans are started,
160 : * these invalid entries will fall off the LRU list and get replaced
161 : * with real entries.
162 : */
163 24620 : item->location.relfilelocator.spcOid = InvalidOid;
164 24620 : item->location.relfilelocator.dbOid = InvalidOid;
165 24620 : item->location.relfilelocator.relNumber = InvalidRelFileNumber;
166 24620 : item->location.location = InvalidBlockNumber;
167 :
168 24620 : item->prev = (i > 0) ?
169 24620 : (&scan_locations->items[i - 1]) : NULL;
170 24620 : item->next = (i < SYNC_SCAN_NELEM - 1) ?
171 24620 : (&scan_locations->items[i + 1]) : NULL;
172 : }
173 1231 : }
174 :
175 : /*
176 : * ss_search --- search the scan_locations structure for an entry with the
177 : * given relfilelocator.
178 : *
179 : * If "set" is true, the location is updated to the given location. If no
180 : * entry for the given relfilelocator is found, it will be created at the head
181 : * of the list with the given location, even if "set" is false.
182 : *
183 : * In any case, the location after possible update is returned.
184 : *
185 : * Caller is responsible for having acquired suitable lock on the shared
186 : * data structure.
187 : */
188 : static BlockNumber
189 3905 : ss_search(RelFileLocator relfilelocator, BlockNumber location, bool set)
190 : {
191 : ss_lru_item_t *item;
192 :
193 3905 : item = scan_locations->head;
194 : for (;;)
195 665 : {
196 : bool match;
197 :
198 4570 : match = RelFileLocatorEquals(item->location.relfilelocator,
199 : relfilelocator);
200 :
201 4570 : if (match || item->next == NULL)
202 : {
203 : /*
204 : * If we reached the end of list and no match was found, take over
205 : * the last entry
206 : */
207 3905 : if (!match)
208 : {
209 35 : item->location.relfilelocator = relfilelocator;
210 35 : item->location.location = location;
211 : }
212 3870 : else if (set)
213 3825 : item->location.location = location;
214 :
215 : /* Move the entry to the front of the LRU list */
216 3905 : if (item != scan_locations->head)
217 : {
218 : /* unlink */
219 35 : if (item == scan_locations->tail)
220 35 : scan_locations->tail = item->prev;
221 35 : item->prev->next = item->next;
222 35 : if (item->next)
223 0 : item->next->prev = item->prev;
224 :
225 : /* link */
226 35 : item->prev = NULL;
227 35 : item->next = scan_locations->head;
228 35 : scan_locations->head->prev = item;
229 35 : scan_locations->head = item;
230 : }
231 :
232 3905 : return item->location.location;
233 : }
234 :
235 665 : item = item->next;
236 : }
237 :
238 : /* not reached */
239 : }
240 :
241 : /*
242 : * ss_get_location --- get the optimal starting location for scan
243 : *
244 : * Returns the last-reported location of a sequential scan on the
245 : * relation, or 0 if no valid location is found.
246 : *
247 : * We expect the caller has just done RelationGetNumberOfBlocks(), and
248 : * so that number is passed in rather than computing it again. The result
249 : * is guaranteed less than relnblocks (assuming that's > 0).
250 : */
251 : BlockNumber
252 80 : ss_get_location(Relation rel, BlockNumber relnblocks)
253 : {
254 : BlockNumber startloc;
255 :
256 80 : LWLockAcquire(SyncScanLock, LW_EXCLUSIVE);
257 80 : startloc = ss_search(rel->rd_locator, 0, false);
258 80 : LWLockRelease(SyncScanLock);
259 :
260 : /*
261 : * If the location is not a valid block number for this scan, start at 0.
262 : *
263 : * This can happen if for instance a VACUUM truncated the table since the
264 : * location was saved.
265 : */
266 80 : if (startloc >= relnblocks)
267 0 : startloc = 0;
268 :
269 : #ifdef TRACE_SYNCSCAN
270 : if (trace_syncscan)
271 : elog(LOG,
272 : "SYNC_SCAN: start \"%s\" (size %u) at %u",
273 : RelationGetRelationName(rel), relnblocks, startloc);
274 : #endif
275 :
276 80 : return startloc;
277 : }
278 :
279 : /*
280 : * ss_report_location --- update the current scan location
281 : *
282 : * Writes an entry into the shared Sync Scan state of the form
283 : * (relfilelocator, blocknumber), overwriting any existing entry for the
284 : * same relfilelocator.
285 : */
286 : void
287 60491 : ss_report_location(Relation rel, BlockNumber location)
288 : {
289 : #ifdef TRACE_SYNCSCAN
290 : if (trace_syncscan)
291 : {
292 : if ((location % 1024) == 0)
293 : elog(LOG,
294 : "SYNC_SCAN: scanning \"%s\" at %u",
295 : RelationGetRelationName(rel), location);
296 : }
297 : #endif
298 :
299 : /*
300 : * To reduce lock contention, only report scan progress every N pages. For
301 : * the same reason, don't block if the lock isn't immediately available.
302 : * Missing a few updates isn't critical, it just means that a new scan
303 : * that wants to join the pack will start a little bit behind the head of
304 : * the scan. Hopefully the pages are still in OS cache and the scan
305 : * catches up quickly.
306 : */
307 60491 : if ((location % SYNC_SCAN_REPORT_INTERVAL) == 0)
308 : {
309 3825 : if (LWLockConditionalAcquire(SyncScanLock, LW_EXCLUSIVE))
310 : {
311 3825 : (void) ss_search(rel->rd_locator, location, true);
312 3825 : LWLockRelease(SyncScanLock);
313 : }
314 : #ifdef TRACE_SYNCSCAN
315 : else if (trace_syncscan)
316 : elog(LOG,
317 : "SYNC_SCAN: missed update for \"%s\" at %u",
318 : RelationGetRelationName(rel), location);
319 : #endif
320 : }
321 60491 : }
|
