| 1 |
/*------------------------------------------------------------------------- |
| 2 |
* |
| 3 |
* xlogreader.c |
| 4 |
* Generic XLog reading facility |
| 5 |
* |
| 6 |
* Portions Copyright (c) 2013-2025, PostgreSQL Global Development Group |
| 7 |
* |
| 8 |
* IDENTIFICATION |
| 9 |
* src/backend/access/transam/xlogreader.c |
| 10 |
* |
| 11 |
* NOTES |
| 12 |
* See xlogreader.h for more notes on this facility. |
| 13 |
* |
| 14 |
* This file is compiled as both front-end and backend code, so it |
| 15 |
* may not use ereport, server-defined static variables, etc. |
| 16 |
*------------------------------------------------------------------------- |
| 17 |
*/ |
| 18 |
#include "postgres.h" |
| 19 |
|
| 20 |
#include <unistd.h> |
| 21 |
#ifdef USE_LZ4 |
| 22 |
#include <lz4.h> |
| 23 |
#endif |
| 24 |
#ifdef USE_ZSTD |
| 25 |
#include <zstd.h> |
| 26 |
#endif |
| 27 |
|
| 28 |
#include "access/transam.h" |
| 29 |
#include "access/xlog_internal.h" |
| 30 |
#include "access/xlogreader.h" |
| 31 |
#include "access/xlogrecord.h" |
| 32 |
#include "catalog/pg_control.h" |
| 33 |
#include "common/pg_lzcompress.h" |
| 34 |
#include "replication/origin.h" |
| 35 |
|
| 36 |
#ifndef FRONTEND |
| 37 |
#include "pgstat.h" |
| 38 |
#include "storage/bufmgr.h" |
| 39 |
#else |
| 40 |
#include "common/logging.h" |
| 41 |
#endif |
| 42 |
|
| 43 |
static void report_invalid_record(XLogReaderState *state, const char *fmt,...) |
| 44 |
pg_attribute_printf(2, 3); |
| 45 |
static void allocate_recordbuf(XLogReaderState *state, uint32 reclength); |
| 46 |
static int ReadPageInternal(XLogReaderState *state, XLogRecPtr pageptr, |
| 47 |
int reqLen); |
| 48 |
static void XLogReaderInvalReadState(XLogReaderState *state); |
| 49 |
static XLogPageReadResult XLogDecodeNextRecord(XLogReaderState *state, bool nonblocking); |
| 50 |
static bool ValidXLogRecordHeader(XLogReaderState *state, XLogRecPtr RecPtr, |
| 51 |
XLogRecPtr PrevRecPtr, XLogRecord *record, bool randAccess); |
| 52 |
static bool ValidXLogRecord(XLogReaderState *state, XLogRecord *record, |
| 53 |
XLogRecPtr recptr); |
| 54 |
static void ResetDecoder(XLogReaderState *state); |
| 55 |
static void WALOpenSegmentInit(WALOpenSegment *seg, WALSegmentContext *segcxt, |
| 56 |
int segsize, const char *waldir); |
| 57 |
|
| 58 |
/* size of the buffer allocated for error message. */ |
| 59 |
#define MAX_ERRORMSG_LEN 1000 |
| 60 |
|
| 61 |
/* |
| 62 |
* Default size; large enough that typical users of XLogReader won't often need |
| 63 |
* to use the 'oversized' memory allocation code path. |
| 64 |
*/ |
| 65 |
#define DEFAULT_DECODE_BUFFER_SIZE (64 * 1024) |
| 66 |
|
| 67 |
/* |
| 68 |
* Construct a string in state->errormsg_buf explaining what's wrong with |
| 69 |
* the current record being read. |
| 70 |
*/ |
| 71 |
static void |
| 72 |
report_invalid_record(XLogReaderState *state, const char *fmt,...) |
| 73 |
{ |
| 74 |
va_list args; |
| 75 |
|
| 76 |
fmt = _(fmt); |
| 77 |
|
| 78 |
va_start(args, fmt); |
| 79 |
vsnprintf(state->errormsg_buf, MAX_ERRORMSG_LEN, fmt, args); |
| 80 |
va_end(args); |
| 81 |
|
| 82 |
state->errormsg_deferred = true; |
| 83 |
} |
| 84 |
|
| 85 |
/* |
| 86 |
* Set the size of the decoding buffer. A pointer to a caller supplied memory |
| 87 |
* region may also be passed in, in which case non-oversized records will be |
| 88 |
* decoded there. |
| 89 |
*/ |
| 90 |
void |
| 91 |
XLogReaderSetDecodeBuffer(XLogReaderState *state, void *buffer, size_t size) |
| 92 |
{ |
| 93 |
Assert(state->decode_buffer == NULL); |
| 94 |
|
| 95 |
state->decode_buffer = buffer; |
| 96 |
state->decode_buffer_size = size; |
| 97 |
state->decode_buffer_tail = buffer; |
| 98 |
state->decode_buffer_head = buffer; |
| 99 |
} |
| 100 |
|
| 101 |
/* |
| 102 |
* Allocate and initialize a new XLogReader. |
| 103 |
* |
| 104 |
* Returns NULL if the xlogreader couldn't be allocated. |
| 105 |
*/ |
| 106 |
XLogReaderState * |
| 107 |
XLogReaderAllocate(int wal_segment_size, const char *waldir, |
| 108 |
XLogReaderRoutine *routine, void *private_data) |
| 109 |
{ |
| 110 |
XLogReaderState *state; |
| 111 |
|
| 112 |
state = (XLogReaderState *) |
| 113 |
palloc_extended(sizeof(XLogReaderState), |
| 114 |
MCXT_ALLOC_NO_OOM | MCXT_ALLOC_ZERO); |
| 115 |
if (!state) |
| 116 |
return NULL; |
| 117 |
|
| 118 |
/* initialize caller-provided support functions */ |
| 119 |
state->routine = *routine; |
| 120 |
|
| 121 |
/* |
| 122 |
* Permanently allocate readBuf. We do it this way, rather than just |
| 123 |
* making a static array, for two reasons: (1) no need to waste the |
| 124 |
* storage in most instantiations of the backend; (2) a static char array |
| 125 |
* isn't guaranteed to have any particular alignment, whereas |
| 126 |
* palloc_extended() will provide MAXALIGN'd storage. |
| 127 |
*/ |
| 128 |
state->readBuf = (char *) palloc_extended(XLOG_BLCKSZ, |
| 129 |
MCXT_ALLOC_NO_OOM); |
| 130 |
if (!state->readBuf) |
| 131 |
{ |
| 132 |
pfree(state); |
| 133 |
return NULL; |
| 134 |
} |
| 135 |
|
| 136 |
/* Initialize segment info. */ |
| 137 |
WALOpenSegmentInit(&state->seg, &state->segcxt, wal_segment_size, |
| 138 |
waldir); |
| 139 |
|
| 140 |
/* system_identifier initialized to zeroes above */ |
| 141 |
state->private_data = private_data; |
| 142 |
/* ReadRecPtr, EndRecPtr and readLen initialized to zeroes above */ |
| 143 |
state->errormsg_buf = palloc_extended(MAX_ERRORMSG_LEN + 1, |
| 144 |
MCXT_ALLOC_NO_OOM); |
| 145 |
if (!state->errormsg_buf) |
| 146 |
{ |
| 147 |
pfree(state->readBuf); |
| 148 |
pfree(state); |
| 149 |
return NULL; |
| 150 |
} |
| 151 |
state->errormsg_buf[0] = '\0'; |
| 152 |
|
| 153 |
/* |
| 154 |
* Allocate an initial readRecordBuf of minimal size, which can later be |
| 155 |
* enlarged if necessary. |
| 156 |
*/ |
| 157 |
allocate_recordbuf(state, 0); |
| 158 |
return state; |
| 159 |
} |
| 160 |
|
| 161 |
void |
| 162 |
XLogReaderFree(XLogReaderState *state) |
| 163 |
{ |
| 164 |
if (state->seg.ws_file != -1) |
| 165 |
state->routine.segment_close(state); |
| 166 |
|
| 167 |
if (state->decode_buffer && state->free_decode_buffer) |
| 168 |
pfree(state->decode_buffer); |
| 169 |
|
| 170 |
pfree(state->errormsg_buf); |
| 171 |
if (state->readRecordBuf) |
| 172 |
pfree(state->readRecordBuf); |
| 173 |
pfree(state->readBuf); |
| 174 |
pfree(state); |
| 175 |
} |
| 176 |
|
| 177 |
/* |
| 178 |
* Allocate readRecordBuf to fit a record of at least the given length. |
| 179 |
* |
| 180 |
* readRecordBufSize is set to the new buffer size. |
| 181 |
* |
| 182 |
* To avoid useless small increases, round its size to a multiple of |
| 183 |
* XLOG_BLCKSZ, and make sure it's at least 5*Max(BLCKSZ, XLOG_BLCKSZ) to start |
| 184 |
* with. (That is enough for all "normal" records, but very large commit or |
| 185 |
* abort records might need more space.) |
| 186 |
* |
| 187 |
* Note: This routine should *never* be called for xl_tot_len until the header |
| 188 |
* of the record has been fully validated. |
| 189 |
*/ |
| 190 |
static void |
| 191 |
allocate_recordbuf(XLogReaderState *state, uint32 reclength) |
| 192 |
{ |
| 193 |
uint32 newSize = reclength; |
| 194 |
|
| 195 |
newSize += XLOG_BLCKSZ - (newSize % XLOG_BLCKSZ); |
| 196 |
newSize = Max(newSize, 5 * Max(BLCKSZ, XLOG_BLCKSZ)); |
| 197 |
|
| 198 |
if (state->readRecordBuf) |
| 199 |
pfree(state->readRecordBuf); |
| 200 |
state->readRecordBuf = (char *) palloc(newSize); |
| 201 |
state->readRecordBufSize = newSize; |
| 202 |
} |
| 203 |
|
| 204 |
/* |
| 205 |
* Initialize the passed segment structs. |
| 206 |
*/ |
| 207 |
static void |
| 208 |
WALOpenSegmentInit(WALOpenSegment *seg, WALSegmentContext *segcxt, |
| 209 |
int segsize, const char *waldir) |
| 210 |
{ |
| 211 |
seg->ws_file = -1; |
| 212 |
seg->ws_segno = 0; |
| 213 |
seg->ws_tli = 0; |
| 214 |
|
| 215 |
segcxt->ws_segsize = segsize; |
| 216 |
if (waldir) |
| 217 |
snprintf(segcxt->ws_dir, MAXPGPATH, "%s", waldir); |
| 218 |
} |
| 219 |
|
| 220 |
/* |
| 221 |
* Begin reading WAL at 'RecPtr'. |
| 222 |
* |
| 223 |
* 'RecPtr' should point to the beginning of a valid WAL record. Pointing at |
| 224 |
* the beginning of a page is also OK, if there is a new record right after |
| 225 |
* the page header, i.e. not a continuation. |
| 226 |
* |
| 227 |
* This does not make any attempt to read the WAL yet, and hence cannot fail. |
| 228 |
* If the starting address is not correct, the first call to XLogReadRecord() |
| 229 |
* will error out. |
| 230 |
*/ |
| 231 |
void |
| 232 |
XLogBeginRead(XLogReaderState *state, XLogRecPtr RecPtr) |
| 233 |
{ |
| 234 |
Assert(!XLogRecPtrIsInvalid(RecPtr)); |
| 235 |
|
| 236 |
ResetDecoder(state); |
| 237 |
|
| 238 |
/* Begin at the passed-in record pointer. */ |
| 239 |
state->EndRecPtr = RecPtr; |
| 240 |
state->NextRecPtr = RecPtr; |
| 241 |
state->ReadRecPtr = InvalidXLogRecPtr; |
| 242 |
state->DecodeRecPtr = InvalidXLogRecPtr; |
| 243 |
} |
| 244 |
|
| 245 |
/* |
| 246 |
* Release the last record that was returned by XLogNextRecord(), if any, to |
| 247 |
* free up space. Returns the LSN past the end of the record. |
| 248 |
*/ |
| 249 |
XLogRecPtr |
| 250 |
XLogReleasePreviousRecord(XLogReaderState *state) |
| 251 |
{ |
| 252 |
DecodedXLogRecord *record; |
| 253 |
XLogRecPtr next_lsn; |
| 254 |
|
| 255 |
if (!state->record) |
| 256 |
return InvalidXLogRecPtr; |
| 257 |
|
| 258 |
/* |
| 259 |
* Remove it from the decoded record queue. It must be the oldest item |
| 260 |
* decoded, decode_queue_head. |
| 261 |
*/ |
| 262 |
record = state->record; |
| 263 |
next_lsn = record->next_lsn; |
| 264 |
Assert(record == state->decode_queue_head); |
| 265 |
state->record = NULL; |
| 266 |
state->decode_queue_head = record->next; |
| 267 |
|
| 268 |
/* It might also be the newest item decoded, decode_queue_tail. */ |
| 269 |
if (state->decode_queue_tail == record) |
| 270 |
state->decode_queue_tail = NULL; |
| 271 |
|
| 272 |
/* Release the space. */ |
| 273 |
if (unlikely(record->oversized)) |
| 274 |
{ |
| 275 |
/* It's not in the decode buffer, so free it to release space. */ |
| 276 |
pfree(record); |
| 277 |
} |
| 278 |
else |
| 279 |
{ |
| 280 |
/* It must be the head (oldest) record in the decode buffer. */ |
| 281 |
Assert(state->decode_buffer_head == (char *) record); |
| 282 |
|
| 283 |
/* |
| 284 |
* We need to update head to point to the next record that is in the |
| 285 |
* decode buffer, if any, being careful to skip oversized ones |
| 286 |
* (they're not in the decode buffer). |
| 287 |
*/ |
| 288 |
record = record->next; |
| 289 |
while (unlikely(record && record->oversized)) |
| 290 |
record = record->next; |
| 291 |
|
| 292 |
if (record) |
| 293 |
{ |
| 294 |
/* Adjust head to release space up to the next record. */ |
| 295 |
state->decode_buffer_head = (char *) record; |
| 296 |
} |
| 297 |
else |
| 298 |
{ |
| 299 |
/* |
| 300 |
* Otherwise we might as well just reset head and tail to the |
| 301 |
* start of the buffer space, because we're empty. This means |
| 302 |
* we'll keep overwriting the same piece of memory if we're not |
| 303 |
* doing any prefetching. |
| 304 |
*/ |
| 305 |
state->decode_buffer_head = state->decode_buffer; |
| 306 |
state->decode_buffer_tail = state->decode_buffer; |
| 307 |
} |
| 308 |
} |
| 309 |
|
| 310 |
return next_lsn; |
| 311 |
} |
| 312 |
|
| 313 |
/* |
| 314 |
* Attempt to read an XLOG record. |
| 315 |
* |
| 316 |
* XLogBeginRead() or XLogFindNextRecord() and then XLogReadAhead() must be |
| 317 |
* called before the first call to XLogNextRecord(). This functions returns |
| 318 |
* records and errors that were put into an internal queue by XLogReadAhead(). |
| 319 |
* |
| 320 |
* On success, a record is returned. |
| 321 |
* |
| 322 |
* The returned record (or *errormsg) points to an internal buffer that's |
| 323 |
* valid until the next call to XLogNextRecord. |
| 324 |
*/ |
| 325 |
DecodedXLogRecord * |
| 326 |
XLogNextRecord(XLogReaderState *state, char **errormsg) |
| 327 |
{ |
| 328 |
/* Release the last record returned by XLogNextRecord(). */ |
| 329 |
XLogReleasePreviousRecord(state); |
| 330 |
|
| 331 |
if (state->decode_queue_head == NULL) |
| 332 |
{ |
| 333 |
*errormsg = NULL; |
| 334 |
if (state->errormsg_deferred) |
| 335 |
{ |
| 336 |
if (state->errormsg_buf[0] != '\0') |
| 337 |
*errormsg = state->errormsg_buf; |
| 338 |
state->errormsg_deferred = false; |
| 339 |
} |
| 340 |
|
| 341 |
/* |
| 342 |
* state->EndRecPtr is expected to have been set by the last call to |
| 343 |
* XLogBeginRead() or XLogNextRecord(), and is the location of the |
| 344 |
* error. |
| 345 |
*/ |
| 346 |
Assert(!XLogRecPtrIsInvalid(state->EndRecPtr)); |
| 347 |
|
| 348 |
return NULL; |
| 349 |
} |
| 350 |
|
| 351 |
/* |
| 352 |
* Record this as the most recent record returned, so that we'll release |
| 353 |
* it next time. This also exposes it to the traditional |
| 354 |
* XLogRecXXX(xlogreader) macros, which work with the decoder rather than |
| 355 |
* the record for historical reasons. |
| 356 |
*/ |
| 357 |
state->record = state->decode_queue_head; |
| 358 |
|
| 359 |
/* |
| 360 |
* Update the pointers to the beginning and one-past-the-end of this |
| 361 |
* record, again for the benefit of historical code that expected the |
| 362 |
* decoder to track this rather than accessing these fields of the record |
| 363 |
* itself. |
| 364 |
*/ |
| 365 |
state->ReadRecPtr = state->record->lsn; |
| 366 |
state->EndRecPtr = state->record->next_lsn; |
| 367 |
|
| 368 |
*errormsg = NULL; |
| 369 |
|
| 370 |
return state->record; |
| 371 |
} |
| 372 |
|
| 373 |
/* |
| 374 |
* Attempt to read an XLOG record. |
| 375 |
* |
| 376 |
* XLogBeginRead() or XLogFindNextRecord() must be called before the first call |
| 377 |
* to XLogReadRecord(). |
| 378 |
* |
| 379 |
* If the page_read callback fails to read the requested data, NULL is |
| 380 |
* returned. The callback is expected to have reported the error; errormsg |
| 381 |
* is set to NULL. |
| 382 |
* |
| 383 |
* If the reading fails for some other reason, NULL is also returned, and |
| 384 |
* *errormsg is set to a string with details of the failure. |
| 385 |
* |
| 386 |
* The returned pointer (or *errormsg) points to an internal buffer that's |
| 387 |
* valid until the next call to XLogReadRecord. |
| 388 |
*/ |
| 389 |
XLogRecord * |
| 390 |
XLogReadRecord(XLogReaderState *state, char **errormsg) |
| 391 |
{ |
| 392 |
DecodedXLogRecord *decoded; |
| 393 |
|
| 394 |
/* |
| 395 |
* Release last returned record, if there is one. We need to do this so |
| 396 |
* that we can check for empty decode queue accurately. |
| 397 |
*/ |
| 398 |
XLogReleasePreviousRecord(state); |
| 399 |
|
| 400 |
/* |
| 401 |
* Call XLogReadAhead() in blocking mode to make sure there is something |
| 402 |
* in the queue, though we don't use the result. |
| 403 |
*/ |
| 404 |
if (!XLogReaderHasQueuedRecordOrError(state)) |
| 405 |
XLogReadAhead(state, false /* nonblocking */ ); |
| 406 |
|
| 407 |
/* Consume the head record or error. */ |
| 408 |
decoded = XLogNextRecord(state, errormsg); |
| 409 |
if (decoded) |
| 410 |
{ |
| 411 |
/* |
| 412 |
* This function returns a pointer to the record's header, not the |
| 413 |
* actual decoded record. The caller will access the decoded record |
| 414 |
* through the XLogRecGetXXX() macros, which reach the decoded |
| 415 |
* recorded as xlogreader->record. |
| 416 |
*/ |
| 417 |
Assert(state->record == decoded); |
| 418 |
return &decoded->header; |
| 419 |
} |
| 420 |
|
| 421 |
return NULL; |
| 422 |
} |
| 423 |
|
| 424 |
/* |
| 425 |
* Allocate space for a decoded record. The only member of the returned |
| 426 |
* object that is initialized is the 'oversized' flag, indicating that the |
| 427 |
* decoded record wouldn't fit in the decode buffer and must eventually be |
| 428 |
* freed explicitly. |
| 429 |
* |
| 430 |
* The caller is responsible for adjusting decode_buffer_tail with the real |
| 431 |
* size after successfully decoding a record into this space. This way, if |
| 432 |
* decoding fails, then there is nothing to undo unless the 'oversized' flag |
| 433 |
* was set and pfree() must be called. |
| 434 |
* |
| 435 |
* Return NULL if there is no space in the decode buffer and allow_oversized |
| 436 |
* is false, or if memory allocation fails for an oversized buffer. |
| 437 |
*/ |
| 438 |
static DecodedXLogRecord * |
| 439 |
XLogReadRecordAlloc(XLogReaderState *state, size_t xl_tot_len, bool allow_oversized) |
| 440 |
{ |
| 441 |
size_t required_space = DecodeXLogRecordRequiredSpace(xl_tot_len); |
| 442 |
DecodedXLogRecord *decoded = NULL; |
| 443 |
|
| 444 |
/* Allocate a circular decode buffer if we don't have one already. */ |
| 445 |
if (unlikely(state->decode_buffer == NULL)) |
| 446 |
{ |
| 447 |
if (state->decode_buffer_size == 0) |
| 448 |
state->decode_buffer_size = DEFAULT_DECODE_BUFFER_SIZE; |
| 449 |
state->decode_buffer = palloc(state->decode_buffer_size); |
| 450 |
state->decode_buffer_head = state->decode_buffer; |
| 451 |
state->decode_buffer_tail = state->decode_buffer; |
| 452 |
state->free_decode_buffer = true; |
| 453 |
} |
| 454 |
|
| 455 |
/* Try to allocate space in the circular decode buffer. */ |
| 456 |
if (state->decode_buffer_tail >= state->decode_buffer_head) |
| 457 |
{ |
| 458 |
/* Empty, or tail is to the right of head. */ |
| 459 |
if (required_space <= |
| 460 |
state->decode_buffer_size - |
| 461 |
(state->decode_buffer_tail - state->decode_buffer)) |
| 462 |
{ |
| 463 |
/*- |
| 464 |
* There is space between tail and end. |
| 465 |
* |
| 466 |
* +-----+--------------------+-----+ |
| 467 |
* | |////////////////////|here!| |
| 468 |
* +-----+--------------------+-----+ |
| 469 |
* ^ ^ |
| 470 |
* | | |
| 471 |
* h t |
| 472 |
*/ |
| 473 |
decoded = (DecodedXLogRecord *) state->decode_buffer_tail; |
| 474 |
decoded->oversized = false; |
| 475 |
return decoded; |
| 476 |
} |
| 477 |
else if (required_space < |
| 478 |
state->decode_buffer_head - state->decode_buffer) |
| 479 |
{ |
| 480 |
/*- |
| 481 |
* There is space between start and head. |
| 482 |
* |
| 483 |
* +-----+--------------------+-----+ |
| 484 |
* |here!|////////////////////| | |
| 485 |
* +-----+--------------------+-----+ |
| 486 |
* ^ ^ |
| 487 |
* | | |
| 488 |
* h t |
| 489 |
*/ |
| 490 |
decoded = (DecodedXLogRecord *) state->decode_buffer; |
| 491 |
decoded->oversized = false; |
| 492 |
return decoded; |
| 493 |
} |
| 494 |
} |
| 495 |
else |
| 496 |
{ |
| 497 |
/* Tail is to the left of head. */ |
| 498 |
if (required_space < |
| 499 |
state->decode_buffer_head - state->decode_buffer_tail) |
| 500 |
{ |
| 501 |
/*- |
| 502 |
* There is space between tail and head. |
| 503 |
* |
| 504 |
* +-----+--------------------+-----+ |
| 505 |
* |/////|here! |/////| |
| 506 |
* +-----+--------------------+-----+ |
| 507 |
* ^ ^ |
| 508 |
* | | |
| 509 |
* t h |
| 510 |
*/ |
| 511 |
decoded = (DecodedXLogRecord *) state->decode_buffer_tail; |
| 512 |
decoded->oversized = false; |
| 513 |
return decoded; |
| 514 |
} |
| 515 |
} |
| 516 |
|
| 517 |
/* Not enough space in the decode buffer. Are we allowed to allocate? */ |
| 518 |
if (allow_oversized) |
| 519 |
{ |
| 520 |
decoded = palloc(required_space); |
| 521 |
decoded->oversized = true; |
| 522 |
return decoded; |
| 523 |
} |
| 524 |
|
| 525 |
return NULL; |
| 526 |
} |
| 527 |
|
| 528 |
static XLogPageReadResult |
| 529 |
XLogDecodeNextRecord(XLogReaderState *state, bool nonblocking) |
| 530 |
{ |
| 531 |
XLogRecPtr RecPtr; |
| 532 |
XLogRecord *record; |
| 533 |
XLogRecPtr targetPagePtr; |
| 534 |
bool randAccess; |
| 535 |
uint32 len, |
| 536 |
total_len; |
| 537 |
uint32 targetRecOff; |
| 538 |
uint32 pageHeaderSize; |
| 539 |
bool assembled; |
| 540 |
bool gotheader; |
| 541 |
int readOff; |
| 542 |
DecodedXLogRecord *decoded; |
| 543 |
char *errormsg; /* not used */ |
| 544 |
|
| 545 |
/* |
| 546 |
* randAccess indicates whether to verify the previous-record pointer of |
| 547 |
* the record we're reading. We only do this if we're reading |
| 548 |
* sequentially, which is what we initially assume. |
| 549 |
*/ |
| 550 |
randAccess = false; |
| 551 |
|
| 552 |
/* reset error state */ |
| 553 |
state->errormsg_buf[0] = '\0'; |
| 554 |
decoded = NULL; |
| 555 |
|
| 556 |
state->abortedRecPtr = InvalidXLogRecPtr; |
| 557 |
state->missingContrecPtr = InvalidXLogRecPtr; |
| 558 |
|
| 559 |
RecPtr = state->NextRecPtr; |
| 560 |
|
| 561 |
if (state->DecodeRecPtr != InvalidXLogRecPtr) |
| 562 |
{ |
| 563 |
/* read the record after the one we just read */ |
| 564 |
|
| 565 |
/* |
| 566 |
* NextRecPtr is pointing to end+1 of the previous WAL record. If |
| 567 |
* we're at a page boundary, no more records can fit on the current |
| 568 |
* page. We must skip over the page header, but we can't do that until |
| 569 |
* we've read in the page, since the header size is variable. |
| 570 |
*/ |
| 571 |
} |
| 572 |
else |
| 573 |
{ |
| 574 |
/* |
| 575 |
* Caller supplied a position to start at. |
| 576 |
* |
| 577 |
* In this case, NextRecPtr should already be pointing either to a |
| 578 |
* valid record starting position or alternatively to the beginning of |
| 579 |
* a page. See the header comments for XLogBeginRead. |
| 580 |
*/ |
| 581 |
Assert(RecPtr % XLOG_BLCKSZ == 0 || XRecOffIsValid(RecPtr)); |
| 582 |
randAccess = true; |
| 583 |
} |
| 584 |
|
| 585 |
restart: |
| 586 |
state->nonblocking = nonblocking; |
| 587 |
state->currRecPtr = RecPtr; |
| 588 |
assembled = false; |
| 589 |
|
| 590 |
targetPagePtr = RecPtr - (RecPtr % XLOG_BLCKSZ); |
| 591 |
targetRecOff = RecPtr % XLOG_BLCKSZ; |
| 592 |
|
| 593 |
/* |
| 594 |
* Read the page containing the record into state->readBuf. Request enough |
| 595 |
* byte to cover the whole record header, or at least the part of it that |
| 596 |
* fits on the same page. |
| 597 |
*/ |
| 598 |
readOff = ReadPageInternal(state, targetPagePtr, |
| 599 |
Min(targetRecOff + SizeOfXLogRecord, XLOG_BLCKSZ)); |
| 600 |
if (readOff == XLREAD_WOULDBLOCK) |
| 601 |
return XLREAD_WOULDBLOCK; |
| 602 |
else if (readOff < 0) |
| 603 |
goto err; |
| 604 |
|
| 605 |
/* |
| 606 |
* ReadPageInternal always returns at least the page header, so we can |
| 607 |
* examine it now. |
| 608 |
*/ |
| 609 |
pageHeaderSize = XLogPageHeaderSize((XLogPageHeader) state->readBuf); |
| 610 |
if (targetRecOff == 0) |
| 611 |
{ |
| 612 |
/* |
| 613 |
* At page start, so skip over page header. |
| 614 |
*/ |
| 615 |
RecPtr += pageHeaderSize; |
| 616 |
targetRecOff = pageHeaderSize; |
| 617 |
} |
| 618 |
else if (targetRecOff < pageHeaderSize) |
| 619 |
{ |
| 620 |
report_invalid_record(state, "invalid record offset at %X/%X: expected at least %u, got %u", |
| 621 |
LSN_FORMAT_ARGS(RecPtr), |
| 622 |
pageHeaderSize, targetRecOff); |
| 623 |
goto err; |
| 624 |
} |
| 625 |
|
| 626 |
if ((((XLogPageHeader) state->readBuf)->xlp_info & XLP_FIRST_IS_CONTRECORD) && |
| 627 |
targetRecOff == pageHeaderSize) |
| 628 |
{ |
| 629 |
report_invalid_record(state, "contrecord is requested by %X/%X", |
| 630 |
LSN_FORMAT_ARGS(RecPtr)); |
| 631 |
goto err; |
| 632 |
} |
| 633 |
|
| 634 |
/* ReadPageInternal has verified the page header */ |
| 635 |
Assert(pageHeaderSize <= readOff); |
| 636 |
|
| 637 |
/* |
| 638 |
* Read the record length. |
| 639 |
* |
| 640 |
* NB: Even though we use an XLogRecord pointer here, the whole record |
| 641 |
* header might not fit on this page. xl_tot_len is the first field of the |
| 642 |
* struct, so it must be on this page (the records are MAXALIGNed), but we |
| 643 |
* cannot access any other fields until we've verified that we got the |
| 644 |
* whole header. |
| 645 |
*/ |
| 646 |
record = (XLogRecord *) (state->readBuf + RecPtr % XLOG_BLCKSZ); |
| 647 |
total_len = record->xl_tot_len; |
| 648 |
|
| 649 |
/* |
| 650 |
* If the whole record header is on this page, validate it immediately. |
| 651 |
* Otherwise do just a basic sanity check on xl_tot_len, and validate the |
| 652 |
* rest of the header after reading it from the next page. The xl_tot_len |
| 653 |
* check is necessary here to ensure that we enter the "Need to reassemble |
| 654 |
* record" code path below; otherwise we might fail to apply |
| 655 |
* ValidXLogRecordHeader at all. |
| 656 |
*/ |
| 657 |
if (targetRecOff <= XLOG_BLCKSZ - SizeOfXLogRecord) |
| 658 |
{ |
| 659 |
if (!ValidXLogRecordHeader(state, RecPtr, state->DecodeRecPtr, record, |
| 660 |
randAccess)) |
| 661 |
goto err; |
| 662 |
gotheader = true; |
| 663 |
} |
| 664 |
else |
| 665 |
{ |
| 666 |
/* There may be no next page if it's too small. */ |
| 667 |
if (total_len < SizeOfXLogRecord) |
| 668 |
{ |
| 669 |
report_invalid_record(state, |
| 670 |
"invalid record length at %X/%X: expected at least %u, got %u", |
| 671 |
LSN_FORMAT_ARGS(RecPtr), |
| 672 |
(uint32) SizeOfXLogRecord, total_len); |
| 673 |
goto err; |
| 674 |
} |
| 675 |
/* We'll validate the header once we have the next page. */ |
| 676 |
gotheader = false; |
| 677 |
} |
| 678 |
|
| 679 |
/* |
| 680 |
* Try to find space to decode this record, if we can do so without |
| 681 |
* calling palloc. If we can't, we'll try again below after we've |
| 682 |
* validated that total_len isn't garbage bytes from a recycled WAL page. |
| 683 |
*/ |
| 684 |
decoded = XLogReadRecordAlloc(state, |
| 685 |
total_len, |
| 686 |
false /* allow_oversized */ ); |
| 687 |
if (decoded == NULL && nonblocking) |
| 688 |
{ |
| 689 |
/* |
| 690 |
* There is no space in the circular decode buffer, and the caller is |
| 691 |
* only reading ahead. The caller should consume existing records to |
| 692 |
* make space. |
| 693 |
*/ |
| 694 |
return XLREAD_WOULDBLOCK; |
| 695 |
} |
| 696 |
|
| 697 |
len = XLOG_BLCKSZ - RecPtr % XLOG_BLCKSZ; |
| 698 |
if (total_len > len) |
| 699 |
{ |
| 700 |
/* Need to reassemble record */ |
| 701 |
char *contdata; |
| 702 |
XLogPageHeader pageHeader; |
| 703 |
char *buffer; |
| 704 |
uint32 gotlen; |
| 705 |
|
| 706 |
assembled = true; |
| 707 |
|
| 708 |
/* |
| 709 |
* We always have space for a couple of pages, enough to validate a |
| 710 |
* boundary-spanning record header. |
| 711 |
*/ |
| 712 |
Assert(state->readRecordBufSize >= XLOG_BLCKSZ * 2); |
| 713 |
Assert(state->readRecordBufSize >= len); |
| 714 |
|
| 715 |
/* Copy the first fragment of the record from the first page. */ |
| 716 |
memcpy(state->readRecordBuf, |
| 717 |
state->readBuf + RecPtr % XLOG_BLCKSZ, len); |
| 718 |
buffer = state->readRecordBuf + len; |
| 719 |
gotlen = len; |
| 720 |
|
| 721 |
do |
| 722 |
{ |
| 723 |
/* Calculate pointer to beginning of next page */ |
| 724 |
targetPagePtr += XLOG_BLCKSZ; |
| 725 |
|
| 726 |
/* |
| 727 |
* Read the page header before processing the record data, so we |
| 728 |
* can handle the case where the previous record ended as being a |
| 729 |
* partial one. |
| 730 |
*/ |
| 731 |
readOff = ReadPageInternal(state, targetPagePtr, SizeOfXLogShortPHD); |
| 732 |
if (readOff == XLREAD_WOULDBLOCK) |
| 733 |
return XLREAD_WOULDBLOCK; |
| 734 |
else if (readOff < 0) |
| 735 |
goto err; |
| 736 |
|
| 737 |
Assert(SizeOfXLogShortPHD <= readOff); |
| 738 |
|
| 739 |
pageHeader = (XLogPageHeader) state->readBuf; |
| 740 |
|
| 741 |
/* |
| 742 |
* If we were expecting a continuation record and got an |
| 743 |
* "overwrite contrecord" flag, that means the continuation record |
| 744 |
* was overwritten with a different record. Restart the read by |
| 745 |
* assuming the address to read is the location where we found |
| 746 |
* this flag; but keep track of the LSN of the record we were |
| 747 |
* reading, for later verification. |
| 748 |
*/ |
| 749 |
if (pageHeader->xlp_info & XLP_FIRST_IS_OVERWRITE_CONTRECORD) |
| 750 |
{ |
| 751 |
state->overwrittenRecPtr = RecPtr; |
| 752 |
RecPtr = targetPagePtr; |
| 753 |
goto restart; |
| 754 |
} |
| 755 |
|
| 756 |
/* Check that the continuation on next page looks valid */ |
| 757 |
if (!(pageHeader->xlp_info & XLP_FIRST_IS_CONTRECORD)) |
| 758 |
{ |
| 759 |
report_invalid_record(state, |
| 760 |
"there is no contrecord flag at %X/%X", |
| 761 |
LSN_FORMAT_ARGS(RecPtr)); |
| 762 |
goto err; |
| 763 |
} |
| 764 |
|
| 765 |
/* |
| 766 |
* Cross-check that xlp_rem_len agrees with how much of the record |
| 767 |
* we expect there to be left. |
| 768 |
*/ |
| 769 |
if (pageHeader->xlp_rem_len == 0 || |
| 770 |
total_len != (pageHeader->xlp_rem_len + gotlen)) |
| 771 |
{ |
| 772 |
report_invalid_record(state, |
| 773 |
"invalid contrecord length %u (expected %lld) at %X/%X", |
| 774 |
pageHeader->xlp_rem_len, |
| 775 |
((long long) total_len) - gotlen, |
| 776 |
LSN_FORMAT_ARGS(RecPtr)); |
| 777 |
goto err; |
| 778 |
} |
| 779 |
|
| 780 |
/* Wait for the next page to become available */ |
| 781 |
readOff = ReadPageInternal(state, targetPagePtr, |
| 782 |
Min(total_len - gotlen + SizeOfXLogShortPHD, |
| 783 |
XLOG_BLCKSZ)); |
| 784 |
if (readOff == XLREAD_WOULDBLOCK) |
| 785 |
return XLREAD_WOULDBLOCK; |
| 786 |
else if (readOff < 0) |
| 787 |
goto err; |
| 788 |
|
| 789 |
/* Append the continuation from this page to the buffer */ |
| 790 |
pageHeaderSize = XLogPageHeaderSize(pageHeader); |
| 791 |
|
| 792 |
if (readOff < pageHeaderSize) |
| 793 |
readOff = ReadPageInternal(state, targetPagePtr, |
| 794 |
pageHeaderSize); |
| 795 |
|
| 796 |
Assert(pageHeaderSize <= readOff); |
| 797 |
|
| 798 |
contdata = (char *) state->readBuf + pageHeaderSize; |
| 799 |
len = XLOG_BLCKSZ - pageHeaderSize; |
| 800 |
if (pageHeader->xlp_rem_len < len) |
| 801 |
len = pageHeader->xlp_rem_len; |
| 802 |
|
| 803 |
if (readOff < pageHeaderSize + len) |
| 804 |
readOff = ReadPageInternal(state, targetPagePtr, |
| 805 |
pageHeaderSize + len); |
| 806 |
|
| 807 |
memcpy(buffer, contdata, len); |
| 808 |
buffer += len; |
| 809 |
gotlen += len; |
| 810 |
|
| 811 |
/* If we just reassembled the record header, validate it. */ |
| 812 |
if (!gotheader) |
| 813 |
{ |
| 814 |
record = (XLogRecord *) state->readRecordBuf; |
| 815 |
if (!ValidXLogRecordHeader(state, RecPtr, state->DecodeRecPtr, |
| 816 |
record, randAccess)) |
| 817 |
goto err; |
| 818 |
gotheader = true; |
| 819 |
} |
| 820 |
|
| 821 |
/* |
| 822 |
* We might need a bigger buffer. We have validated the record |
| 823 |
* header, in the case that it split over a page boundary. We've |
| 824 |
* also cross-checked total_len against xlp_rem_len on the second |
| 825 |
* page, and verified xlp_pageaddr on both. |
| 826 |
*/ |
| 827 |
if (total_len > state->readRecordBufSize) |
| 828 |
{ |
| 829 |
char save_copy[XLOG_BLCKSZ * 2]; |
| 830 |
|
| 831 |
/* |
| 832 |
* Save and restore the data we already had. It can't be more |
| 833 |
* than two pages. |
| 834 |
*/ |
| 835 |
Assert(gotlen <= lengthof(save_copy)); |
| 836 |
Assert(gotlen <= state->readRecordBufSize); |
| 837 |
memcpy(save_copy, state->readRecordBuf, gotlen); |
| 838 |
allocate_recordbuf(state, total_len); |
| 839 |
memcpy(state->readRecordBuf, save_copy, gotlen); |
| 840 |
buffer = state->readRecordBuf + gotlen; |
| 841 |
} |
| 842 |
} while (gotlen < total_len); |
| 843 |
Assert(gotheader); |
| 844 |
|
| 845 |
record = (XLogRecord *) state->readRecordBuf; |
| 846 |
if (!ValidXLogRecord(state, record, RecPtr)) |
| 847 |
goto err; |
| 848 |
|
| 849 |
pageHeaderSize = XLogPageHeaderSize((XLogPageHeader) state->readBuf); |
| 850 |
state->DecodeRecPtr = RecPtr; |
| 851 |
state->NextRecPtr = targetPagePtr + pageHeaderSize |
| 852 |
+ MAXALIGN(pageHeader->xlp_rem_len); |
| 853 |
} |
| 854 |
else |
| 855 |
{ |
| 856 |
/* Wait for the record data to become available */ |
| 857 |
readOff = ReadPageInternal(state, targetPagePtr, |
| 858 |
Min(targetRecOff + total_len, XLOG_BLCKSZ)); |
| 859 |
if (readOff == XLREAD_WOULDBLOCK) |
| 860 |
return XLREAD_WOULDBLOCK; |
| 861 |
else if (readOff < 0) |
| 862 |
goto err; |
| 863 |
|
| 864 |
/* Record does not cross a page boundary */ |
| 865 |
if (!ValidXLogRecord(state, record, RecPtr)) |
| 866 |
goto err; |
| 867 |
|
| 868 |
state->NextRecPtr = RecPtr + MAXALIGN(total_len); |
| 869 |
|
| 870 |
state->DecodeRecPtr = RecPtr; |
| 871 |
} |
| 872 |
|
| 873 |
/* |
| 874 |
* Special processing if it's an XLOG SWITCH record |
| 875 |
*/ |
| 876 |
if (record->xl_rmid == RM_XLOG_ID && |
| 877 |
(record->xl_info & ~XLR_INFO_MASK) == XLOG_SWITCH) |
| 878 |
{ |
| 879 |
/* Pretend it extends to end of segment */ |
| 880 |
state->NextRecPtr += state->segcxt.ws_segsize - 1; |
| 881 |
state->NextRecPtr -= XLogSegmentOffset(state->NextRecPtr, state->segcxt.ws_segsize); |
| 882 |
} |
| 883 |
|
| 884 |
/* |
| 885 |
* If we got here without a DecodedXLogRecord, it means we needed to |
| 886 |
* validate total_len before trusting it, but by now we've done that. |
| 887 |
*/ |
| 888 |
if (decoded == NULL) |
| 889 |
{ |
| 890 |
Assert(!nonblocking); |
| 891 |
decoded = XLogReadRecordAlloc(state, |
| 892 |
total_len, |
| 893 |
true /* allow_oversized */ ); |
| 894 |
/* allocation should always happen under allow_oversized */ |
| 895 |
Assert(decoded != NULL); |
| 896 |
} |
| 897 |
|
| 898 |
if (DecodeXLogRecord(state, decoded, record, RecPtr, &errormsg)) |
| 899 |
{ |
| 900 |
/* Record the location of the next record. */ |
| 901 |
decoded->next_lsn = state->NextRecPtr; |
| 902 |
|
| 903 |
/* |
| 904 |
* If it's in the decode buffer, mark the decode buffer space as |
| 905 |
* occupied. |
| 906 |
*/ |
| 907 |
if (!decoded->oversized) |
| 908 |
{ |
| 909 |
/* The new decode buffer head must be MAXALIGNed. */ |
| 910 |
Assert(decoded->size == MAXALIGN(decoded->size)); |
| 911 |
if ((char *) decoded == state->decode_buffer) |
| 912 |
state->decode_buffer_tail = state->decode_buffer + decoded->size; |
| 913 |
else |
| 914 |
state->decode_buffer_tail += decoded->size; |
| 915 |
} |
| 916 |
|
| 917 |
/* Insert it into the queue of decoded records. */ |
| 918 |
Assert(state->decode_queue_tail != decoded); |
| 919 |
if (state->decode_queue_tail) |
| 920 |
state->decode_queue_tail->next = decoded; |
| 921 |
state->decode_queue_tail = decoded; |
| 922 |
if (!state->decode_queue_head) |
| 923 |
state->decode_queue_head = decoded; |
| 924 |
return XLREAD_SUCCESS; |
| 925 |
} |
| 926 |
|
| 927 |
err: |
| 928 |
if (assembled) |
| 929 |
{ |
| 930 |
/* |
| 931 |
* We get here when a record that spans multiple pages needs to be |
| 932 |
* assembled, but something went wrong -- perhaps a contrecord piece |
| 933 |
* was lost. If caller is WAL replay, it will know where the aborted |
| 934 |
* record was and where to direct followup WAL to be written, marking |
| 935 |
* the next piece with XLP_FIRST_IS_OVERWRITE_CONTRECORD, which will |
| 936 |
* in turn signal downstream WAL consumers that the broken WAL record |
| 937 |
* is to be ignored. |
| 938 |
*/ |
| 939 |
state->abortedRecPtr = RecPtr; |
| 940 |
state->missingContrecPtr = targetPagePtr; |
| 941 |
|
| 942 |
/* |
| 943 |
* If we got here without reporting an error, make sure an error is |
| 944 |
* queued so that XLogPrefetcherReadRecord() doesn't bring us back a |
| 945 |
* second time and clobber the above state. |
| 946 |
*/ |
| 947 |
state->errormsg_deferred = true; |
| 948 |
} |
| 949 |
|
| 950 |
if (decoded && decoded->oversized) |
| 951 |
pfree(decoded); |
| 952 |
|
| 953 |
/* |
| 954 |
* Invalidate the read state. We might read from a different source after |
| 955 |
* failure. |
| 956 |
*/ |
| 957 |
XLogReaderInvalReadState(state); |
| 958 |
|
| 959 |
/* |
| 960 |
* If an error was written to errormsg_buf, it'll be returned to the |
| 961 |
* caller of XLogReadRecord() after all successfully decoded records from |
| 962 |
* the read queue. |
| 963 |
*/ |
| 964 |
|
| 965 |
return XLREAD_FAIL; |
| 966 |
} |
| 967 |
|
| 968 |
/* |
| 969 |
* Try to decode the next available record, and return it. The record will |
| 970 |
* also be returned to XLogNextRecord(), which must be called to 'consume' |
| 971 |
* each record. |
| 972 |
* |
| 973 |
* If nonblocking is true, may return NULL due to lack of data or WAL decoding |
| 974 |
* space. |
| 975 |
*/ |
| 976 |
DecodedXLogRecord * |
| 977 |
XLogReadAhead(XLogReaderState *state, bool nonblocking) |
| 978 |
{ |
| 979 |
XLogPageReadResult result; |
| 980 |
|
| 981 |
if (state->errormsg_deferred) |
| 982 |
return NULL; |
| 983 |
|
| 984 |
result = XLogDecodeNextRecord(state, nonblocking); |
| 985 |
if (result == XLREAD_SUCCESS) |
| 986 |
{ |
| 987 |
Assert(state->decode_queue_tail != NULL); |
| 988 |
return state->decode_queue_tail; |
| 989 |
} |
| 990 |
|
| 991 |
return NULL; |
| 992 |
} |
| 993 |
|
| 994 |
/* |
| 995 |
* Read a single xlog page including at least [pageptr, reqLen] of valid data |
| 996 |
* via the page_read() callback. |
| 997 |
* |
| 998 |
* Returns XLREAD_FAIL if the required page cannot be read for some |
| 999 |
* reason; errormsg_buf is set in that case (unless the error occurs in the |
| 1000 |
* page_read callback). |
| 1001 |
* |
| 1002 |
* Returns XLREAD_WOULDBLOCK if the requested data can't be read without |
| 1003 |
* waiting. This can be returned only if the installed page_read callback |
| 1004 |
* respects the state->nonblocking flag, and cannot read the requested data |
| 1005 |
* immediately. |
| 1006 |
* |
| 1007 |
* We fetch the page from a reader-local cache if we know we have the required |
| 1008 |
* data and if there hasn't been any error since caching the data. |
| 1009 |
*/ |
| 1010 |
static int |
| 1011 |
ReadPageInternal(XLogReaderState *state, XLogRecPtr pageptr, int reqLen) |
| 1012 |
{ |
| 1013 |
int readLen; |
| 1014 |
uint32 targetPageOff; |
| 1015 |
XLogSegNo targetSegNo; |
| 1016 |
XLogPageHeader hdr; |
| 1017 |
|
| 1018 |
Assert((pageptr % XLOG_BLCKSZ) == 0); |
| 1019 |
|
| 1020 |
XLByteToSeg(pageptr, targetSegNo, state->segcxt.ws_segsize); |
| 1021 |
targetPageOff = XLogSegmentOffset(pageptr, state->segcxt.ws_segsize); |
| 1022 |
|
| 1023 |
/* check whether we have all the requested data already */ |
| 1024 |
if (targetSegNo == state->seg.ws_segno && |
| 1025 |
targetPageOff == state->segoff && reqLen <= state->readLen) |
| 1026 |
return state->readLen; |
| 1027 |
|
| 1028 |
/* |
| 1029 |
* Invalidate contents of internal buffer before read attempt. Just set |
| 1030 |
* the length to 0, rather than a full XLogReaderInvalReadState(), so we |
| 1031 |
* don't forget the segment we last successfully read. |
| 1032 |
*/ |
| 1033 |
state->readLen = 0; |
| 1034 |
|
| 1035 |
/* |
| 1036 |
* Data is not in our buffer. |
| 1037 |
* |
| 1038 |
* Every time we actually read the segment, even if we looked at parts of |
| 1039 |
* it before, we need to do verification as the page_read callback might |
| 1040 |
* now be rereading data from a different source. |
| 1041 |
* |
| 1042 |
* Whenever switching to a new WAL segment, we read the first page of the |
| 1043 |
* file and validate its header, even if that's not where the target |
| 1044 |
* record is. This is so that we can check the additional identification |
| 1045 |
* info that is present in the first page's "long" header. |
| 1046 |
*/ |
| 1047 |
if (targetSegNo != state->seg.ws_segno && targetPageOff != 0) |
| 1048 |
{ |
| 1049 |
XLogRecPtr targetSegmentPtr = pageptr - targetPageOff; |
| 1050 |
|
| 1051 |
readLen = state->routine.page_read(state, targetSegmentPtr, XLOG_BLCKSZ, |
| 1052 |
state->currRecPtr, |
| 1053 |
state->readBuf); |
| 1054 |
if (readLen == XLREAD_WOULDBLOCK) |
| 1055 |
return XLREAD_WOULDBLOCK; |
| 1056 |
else if (readLen < 0) |
| 1057 |
goto err; |
| 1058 |
|
| 1059 |
/* we can be sure to have enough WAL available, we scrolled back */ |
| 1060 |
Assert(readLen == XLOG_BLCKSZ); |
| 1061 |
|
| 1062 |
if (!XLogReaderValidatePageHeader(state, targetSegmentPtr, |
| 1063 |
state->readBuf)) |
| 1064 |
goto err; |
| 1065 |
} |
| 1066 |
|
| 1067 |
/* |
| 1068 |
* First, read the requested data length, but at least a short page header |
| 1069 |
* so that we can validate it. |
| 1070 |
*/ |
| 1071 |
readLen = state->routine.page_read(state, pageptr, Max(reqLen, SizeOfXLogShortPHD), |
| 1072 |
state->currRecPtr, |
| 1073 |
state->readBuf); |
| 1074 |
if (readLen == XLREAD_WOULDBLOCK) |
| 1075 |
return XLREAD_WOULDBLOCK; |
| 1076 |
else if (readLen < 0) |
| 1077 |
goto err; |
| 1078 |
|
| 1079 |
Assert(readLen <= XLOG_BLCKSZ); |
| 1080 |
|
| 1081 |
/* Do we have enough data to check the header length? */ |
| 1082 |
if (readLen <= SizeOfXLogShortPHD) |
| 1083 |
goto err; |
| 1084 |
|
| 1085 |
Assert(readLen >= reqLen); |
| 1086 |
|
| 1087 |
hdr = (XLogPageHeader) state->readBuf; |
| 1088 |
|
| 1089 |
/* still not enough */ |
| 1090 |
if (readLen < XLogPageHeaderSize(hdr)) |
| 1091 |
{ |
| 1092 |
readLen = state->routine.page_read(state, pageptr, XLogPageHeaderSize(hdr), |
| 1093 |
state->currRecPtr, |
| 1094 |
state->readBuf); |
| 1095 |
if (readLen == XLREAD_WOULDBLOCK) |
| 1096 |
return XLREAD_WOULDBLOCK; |
| 1097 |
else if (readLen < 0) |
| 1098 |
goto err; |
| 1099 |
} |
| 1100 |
|
| 1101 |
/* |
| 1102 |
* Now that we know we have the full header, validate it. |
| 1103 |
*/ |
| 1104 |
if (!XLogReaderValidatePageHeader(state, pageptr, (char *) hdr)) |
| 1105 |
goto err; |
| 1106 |
|
| 1107 |
/* update read state information */ |
| 1108 |
state->seg.ws_segno = targetSegNo; |
| 1109 |
state->segoff = targetPageOff; |
| 1110 |
state->readLen = readLen; |
| 1111 |
|
| 1112 |
return readLen; |
| 1113 |
|
| 1114 |
err: |
| 1115 |
XLogReaderInvalReadState(state); |
| 1116 |
|
| 1117 |
return XLREAD_FAIL; |
| 1118 |
} |
| 1119 |
|
| 1120 |
/* |
| 1121 |
* Invalidate the xlogreader's read state to force a re-read. |
| 1122 |
*/ |
| 1123 |
static void |
| 1124 |
XLogReaderInvalReadState(XLogReaderState *state) |
| 1125 |
{ |
| 1126 |
state->seg.ws_segno = 0; |
| 1127 |
state->segoff = 0; |
| 1128 |
state->readLen = 0; |
| 1129 |
} |
| 1130 |
|
| 1131 |
/* |
| 1132 |
* Validate an XLOG record header. |
| 1133 |
* |
| 1134 |
* This is just a convenience subroutine to avoid duplicated code in |
| 1135 |
* XLogReadRecord. It's not intended for use from anywhere else. |
| 1136 |
*/ |
| 1137 |
static bool |
| 1138 |
ValidXLogRecordHeader(XLogReaderState *state, XLogRecPtr RecPtr, |
| 1139 |
XLogRecPtr PrevRecPtr, XLogRecord *record, |
| 1140 |
bool randAccess) |
| 1141 |
{ |
| 1142 |
if (record->xl_tot_len < SizeOfXLogRecord) |
| 1143 |
{ |
| 1144 |
report_invalid_record(state, |
| 1145 |
"invalid record length at %X/%X: expected at least %u, got %u", |
| 1146 |
LSN_FORMAT_ARGS(RecPtr), |
| 1147 |
(uint32) SizeOfXLogRecord, record->xl_tot_len); |
| 1148 |
return false; |
| 1149 |
} |
| 1150 |
if (!RmgrIdIsValid(record->xl_rmid)) |
| 1151 |
{ |
| 1152 |
report_invalid_record(state, |
| 1153 |
"invalid resource manager ID %u at %X/%X", |
| 1154 |
record->xl_rmid, LSN_FORMAT_ARGS(RecPtr)); |
| 1155 |
return false; |
| 1156 |
} |
| 1157 |
if (randAccess) |
| 1158 |
{ |
| 1159 |
/* |
| 1160 |
* We can't exactly verify the prev-link, but surely it should be less |
| 1161 |
* than the record's own address. |
| 1162 |
*/ |
| 1163 |
if (!(record->xl_prev < RecPtr)) |
| 1164 |
{ |
| 1165 |
report_invalid_record(state, |
| 1166 |
"record with incorrect prev-link %X/%X at %X/%X", |
| 1167 |
LSN_FORMAT_ARGS(record->xl_prev), |
| 1168 |
LSN_FORMAT_ARGS(RecPtr)); |
| 1169 |
return false; |
| 1170 |
} |
| 1171 |
} |
| 1172 |
else |
| 1173 |
{ |
| 1174 |
/* |
| 1175 |
* Record's prev-link should exactly match our previous location. This |
| 1176 |
* check guards against torn WAL pages where a stale but valid-looking |
| 1177 |
* WAL record starts on a sector boundary. |
| 1178 |
*/ |
| 1179 |
if (record->xl_prev != PrevRecPtr) |
| 1180 |
{ |
| 1181 |
report_invalid_record(state, |
| 1182 |
"record with incorrect prev-link %X/%X at %X/%X", |
| 1183 |
LSN_FORMAT_ARGS(record->xl_prev), |
| 1184 |
LSN_FORMAT_ARGS(RecPtr)); |
| 1185 |
return false; |
| 1186 |
} |
| 1187 |
} |
| 1188 |
|
| 1189 |
return true; |
| 1190 |
} |
| 1191 |
|
| 1192 |
|
| 1193 |
/* |
| 1194 |
* CRC-check an XLOG record. We do not believe the contents of an XLOG |
| 1195 |
* record (other than to the minimal extent of computing the amount of |
| 1196 |
* data to read in) until we've checked the CRCs. |
| 1197 |
* |
| 1198 |
* We assume all of the record (that is, xl_tot_len bytes) has been read |
| 1199 |
* into memory at *record. Also, ValidXLogRecordHeader() has accepted the |
| 1200 |
* record's header, which means in particular that xl_tot_len is at least |
| 1201 |
* SizeOfXLogRecord. |
| 1202 |
*/ |
| 1203 |
static bool |
| 1204 |
ValidXLogRecord(XLogReaderState *state, XLogRecord *record, XLogRecPtr recptr) |
| 1205 |
{ |
| 1206 |
pg_crc32c crc; |
| 1207 |
|
| 1208 |
Assert(record->xl_tot_len >= SizeOfXLogRecord); |
| 1209 |
|
| 1210 |
/* Calculate the CRC */ |
| 1211 |
INIT_CRC32C(crc); |
| 1212 |
COMP_CRC32C(crc, ((char *) record) + SizeOfXLogRecord, record->xl_tot_len - SizeOfXLogRecord); |
| 1213 |
/* include the record header last */ |
| 1214 |
COMP_CRC32C(crc, (char *) record, offsetof(XLogRecord, xl_crc)); |
| 1215 |
FIN_CRC32C(crc); |
| 1216 |
|
| 1217 |
if (!EQ_CRC32C(record->xl_crc, crc)) |
| 1218 |
{ |
| 1219 |
report_invalid_record(state, |
| 1220 |
"incorrect resource manager data checksum in record at %X/%X", |
| 1221 |
LSN_FORMAT_ARGS(recptr)); |
| 1222 |
return false; |
| 1223 |
} |
| 1224 |
|
| 1225 |
return true; |
| 1226 |
} |
| 1227 |
|
| 1228 |
/* |
| 1229 |
* Validate a page header. |
| 1230 |
* |
| 1231 |
* Check if 'phdr' is valid as the header of the XLog page at position |
| 1232 |
* 'recptr'. |
| 1233 |
*/ |
| 1234 |
bool |
| 1235 |
XLogReaderValidatePageHeader(XLogReaderState *state, XLogRecPtr recptr, |
| 1236 |
char *phdr) |
| 1237 |
{ |
| 1238 |
XLogSegNo segno; |
| 1239 |
int32 offset; |
| 1240 |
XLogPageHeader hdr = (XLogPageHeader) phdr; |
| 1241 |
|
| 1242 |
Assert((recptr % XLOG_BLCKSZ) == 0); |
| 1243 |
|
| 1244 |
XLByteToSeg(recptr, segno, state->segcxt.ws_segsize); |
| 1245 |
offset = XLogSegmentOffset(recptr, state->segcxt.ws_segsize); |
| 1246 |
|
| 1247 |
if (hdr->xlp_magic != XLOG_PAGE_MAGIC) |
| 1248 |
{ |
| 1249 |
char fname[MAXFNAMELEN]; |
| 1250 |
|
| 1251 |
XLogFileName(fname, state->seg.ws_tli, segno, state->segcxt.ws_segsize); |
| 1252 |
|
| 1253 |
report_invalid_record(state, |
| 1254 |
"invalid magic number %04X in WAL segment %s, LSN %X/%X, offset %u", |
| 1255 |
hdr->xlp_magic, |
| 1256 |
fname, |
| 1257 |
LSN_FORMAT_ARGS(recptr), |
| 1258 |
offset); |
| 1259 |
return false; |
| 1260 |
} |
| 1261 |
|
| 1262 |
if ((hdr->xlp_info & ~XLP_ALL_FLAGS) != 0) |
| 1263 |
{ |
| 1264 |
char fname[MAXFNAMELEN]; |
| 1265 |
|
| 1266 |
XLogFileName(fname, state->seg.ws_tli, segno, state->segcxt.ws_segsize); |
| 1267 |
|
| 1268 |
report_invalid_record(state, |
| 1269 |
"invalid info bits %04X in WAL segment %s, LSN %X/%X, offset %u", |
| 1270 |
hdr->xlp_info, |
| 1271 |
fname, |
| 1272 |
LSN_FORMAT_ARGS(recptr), |
| 1273 |
offset); |
| 1274 |
return false; |
| 1275 |
} |
| 1276 |
|
| 1277 |
if (hdr->xlp_info & XLP_LONG_HEADER) |
| 1278 |
{ |
| 1279 |
XLogLongPageHeader longhdr = (XLogLongPageHeader) hdr; |
| 1280 |
|
| 1281 |
if (state->system_identifier && |
| 1282 |
longhdr->xlp_sysid != state->system_identifier) |
| 1283 |
{ |
| 1284 |
report_invalid_record(state, |
| 1285 |
"WAL file is from different database system: WAL file database system identifier is %" PRIu64 ", pg_control database system identifier is %" PRIu64, |
| 1286 |
longhdr->xlp_sysid, |
| 1287 |
state->system_identifier); |
| 1288 |
return false; |
| 1289 |
} |
| 1290 |
else if (longhdr->xlp_seg_size != state->segcxt.ws_segsize) |
| 1291 |
{ |
| 1292 |
report_invalid_record(state, |
| 1293 |
"WAL file is from different database system: incorrect segment size in page header"); |
| 1294 |
return false; |
| 1295 |
} |
| 1296 |
else if (longhdr->xlp_xlog_blcksz != XLOG_BLCKSZ) |
| 1297 |
{ |
| 1298 |
report_invalid_record(state, |
| 1299 |
"WAL file is from different database system: incorrect XLOG_BLCKSZ in page header"); |
| 1300 |
return false; |
| 1301 |
} |
| 1302 |
} |
| 1303 |
else if (offset == 0) |
| 1304 |
{ |
| 1305 |
char fname[MAXFNAMELEN]; |
| 1306 |
|
| 1307 |
XLogFileName(fname, state->seg.ws_tli, segno, state->segcxt.ws_segsize); |
| 1308 |
|
| 1309 |
/* hmm, first page of file doesn't have a long header? */ |
| 1310 |
report_invalid_record(state, |
| 1311 |
"invalid info bits %04X in WAL segment %s, LSN %X/%X, offset %u", |
| 1312 |
hdr->xlp_info, |
| 1313 |
fname, |
| 1314 |
LSN_FORMAT_ARGS(recptr), |
| 1315 |
offset); |
| 1316 |
return false; |
| 1317 |
} |
| 1318 |
|
| 1319 |
/* |
| 1320 |
* Check that the address on the page agrees with what we expected. This |
| 1321 |
* check typically fails when an old WAL segment is recycled, and hasn't |
| 1322 |
* yet been overwritten with new data yet. |
| 1323 |
*/ |
| 1324 |
if (hdr->xlp_pageaddr != recptr) |
| 1325 |
{ |
| 1326 |
char fname[MAXFNAMELEN]; |
| 1327 |
|
| 1328 |
XLogFileName(fname, state->seg.ws_tli, segno, state->segcxt.ws_segsize); |
| 1329 |
|
| 1330 |
report_invalid_record(state, |
| 1331 |
"unexpected pageaddr %X/%X in WAL segment %s, LSN %X/%X, offset %u", |
| 1332 |
LSN_FORMAT_ARGS(hdr->xlp_pageaddr), |
| 1333 |
fname, |
| 1334 |
LSN_FORMAT_ARGS(recptr), |
| 1335 |
offset); |
| 1336 |
return false; |
| 1337 |
} |
| 1338 |
|
| 1339 |
/* |
| 1340 |
* Since child timelines are always assigned a TLI greater than their |
| 1341 |
* immediate parent's TLI, we should never see TLI go backwards across |
| 1342 |
* successive pages of a consistent WAL sequence. |
| 1343 |
* |
| 1344 |
* Sometimes we re-read a segment that's already been (partially) read. So |
| 1345 |
* we only verify TLIs for pages that are later than the last remembered |
| 1346 |
* LSN. |
| 1347 |
*/ |
| 1348 |
if (recptr > state->latestPagePtr) |
| 1349 |
{ |
| 1350 |
if (hdr->xlp_tli < state->latestPageTLI) |
| 1351 |
{ |
| 1352 |
char fname[MAXFNAMELEN]; |
| 1353 |
|
| 1354 |
XLogFileName(fname, state->seg.ws_tli, segno, state->segcxt.ws_segsize); |
| 1355 |
|
| 1356 |
report_invalid_record(state, |
| 1357 |
"out-of-sequence timeline ID %u (after %u) in WAL segment %s, LSN %X/%X, offset %u", |
| 1358 |
hdr->xlp_tli, |
| 1359 |
state->latestPageTLI, |
| 1360 |
fname, |
| 1361 |
LSN_FORMAT_ARGS(recptr), |
| 1362 |
offset); |
| 1363 |
return false; |
| 1364 |
} |
| 1365 |
} |
| 1366 |
state->latestPagePtr = recptr; |
| 1367 |
state->latestPageTLI = hdr->xlp_tli; |
| 1368 |
|
| 1369 |
return true; |
| 1370 |
} |
| 1371 |
|
| 1372 |
/* |
| 1373 |
* Forget about an error produced by XLogReaderValidatePageHeader(). |
| 1374 |
*/ |
| 1375 |
void |
| 1376 |
XLogReaderResetError(XLogReaderState *state) |
| 1377 |
{ |
| 1378 |
state->errormsg_buf[0] = '\0'; |
| 1379 |
state->errormsg_deferred = false; |
| 1380 |
} |
| 1381 |
|
| 1382 |
/* |
| 1383 |
* Find the first record with an lsn >= RecPtr. |
| 1384 |
* |
| 1385 |
* This is different from XLogBeginRead() in that RecPtr doesn't need to point |
| 1386 |
* to a valid record boundary. Useful for checking whether RecPtr is a valid |
| 1387 |
* xlog address for reading, and to find the first valid address after some |
| 1388 |
* address when dumping records for debugging purposes. |
| 1389 |
* |
| 1390 |
* This positions the reader, like XLogBeginRead(), so that the next call to |
| 1391 |
* XLogReadRecord() will read the next valid record. |
| 1392 |
*/ |
| 1393 |
XLogRecPtr |
| 1394 |
XLogFindNextRecord(XLogReaderState *state, XLogRecPtr RecPtr) |
| 1395 |
{ |
| 1396 |
XLogRecPtr tmpRecPtr; |
| 1397 |
XLogRecPtr found = InvalidXLogRecPtr; |
| 1398 |
XLogPageHeader header; |
| 1399 |
char *errormsg; |
| 1400 |
|
| 1401 |
Assert(!XLogRecPtrIsInvalid(RecPtr)); |
| 1402 |
|
| 1403 |
/* Make sure ReadPageInternal() can't return XLREAD_WOULDBLOCK. */ |
| 1404 |
state->nonblocking = false; |
| 1405 |
|
| 1406 |
/* |
| 1407 |
* skip over potential continuation data, keeping in mind that it may span |
| 1408 |
* multiple pages |
| 1409 |
*/ |
| 1410 |
tmpRecPtr = RecPtr; |
| 1411 |
while (true) |
| 1412 |
{ |
| 1413 |
XLogRecPtr targetPagePtr; |
| 1414 |
int targetRecOff; |
| 1415 |
uint32 pageHeaderSize; |
| 1416 |
int readLen; |
| 1417 |
|
| 1418 |
/* |
| 1419 |
* Compute targetRecOff. It should typically be equal or greater than |
| 1420 |
* short page-header since a valid record can't start anywhere before |
| 1421 |
* that, except when caller has explicitly specified the offset that |
| 1422 |
* falls somewhere there or when we are skipping multi-page |
| 1423 |
* continuation record. It doesn't matter though because |
| 1424 |
* ReadPageInternal() is prepared to handle that and will read at |
| 1425 |
* least short page-header worth of data |
| 1426 |
*/ |
| 1427 |
targetRecOff = tmpRecPtr % XLOG_BLCKSZ; |
| 1428 |
|
| 1429 |
/* scroll back to page boundary */ |
| 1430 |
targetPagePtr = tmpRecPtr - targetRecOff; |
| 1431 |
|
| 1432 |
/* Read the page containing the record */ |
| 1433 |
readLen = ReadPageInternal(state, targetPagePtr, targetRecOff); |
| 1434 |
if (readLen < 0) |
| 1435 |
goto err; |
| 1436 |
|
| 1437 |
header = (XLogPageHeader) state->readBuf; |
| 1438 |
|
| 1439 |
pageHeaderSize = XLogPageHeaderSize(header); |
| 1440 |
|
| 1441 |
/* make sure we have enough data for the page header */ |
| 1442 |
readLen = ReadPageInternal(state, targetPagePtr, pageHeaderSize); |
| 1443 |
if (readLen < 0) |
| 1444 |
goto err; |
| 1445 |
|
| 1446 |
/* skip over potential continuation data */ |
| 1447 |
if (header->xlp_info & XLP_FIRST_IS_CONTRECORD) |
| 1448 |
{ |
| 1449 |
/* |
| 1450 |
* If the length of the remaining continuation data is more than |
| 1451 |
* what can fit in this page, the continuation record crosses over |
| 1452 |
* this page. Read the next page and try again. xlp_rem_len in the |
| 1453 |
* next page header will contain the remaining length of the |
| 1454 |
* continuation data |
| 1455 |
* |
| 1456 |
* Note that record headers are MAXALIGN'ed |
| 1457 |
*/ |
| 1458 |
if (MAXALIGN(header->xlp_rem_len) >= (XLOG_BLCKSZ - pageHeaderSize)) |
| 1459 |
tmpRecPtr = targetPagePtr + XLOG_BLCKSZ; |
| 1460 |
else |
| 1461 |
{ |
| 1462 |
/* |
| 1463 |
* The previous continuation record ends in this page. Set |
| 1464 |
* tmpRecPtr to point to the first valid record |
| 1465 |
*/ |
| 1466 |
tmpRecPtr = targetPagePtr + pageHeaderSize |
| 1467 |
+ MAXALIGN(header->xlp_rem_len); |
| 1468 |
break; |
| 1469 |
} |
| 1470 |
} |
| 1471 |
else |
| 1472 |
{ |
| 1473 |
tmpRecPtr = targetPagePtr + pageHeaderSize; |
| 1474 |
break; |
| 1475 |
} |
| 1476 |
} |
| 1477 |
|
| 1478 |
/* |
| 1479 |
* we know now that tmpRecPtr is an address pointing to a valid XLogRecord |
| 1480 |
* because either we're at the first record after the beginning of a page |
| 1481 |
* or we just jumped over the remaining data of a continuation. |
| 1482 |
*/ |
| 1483 |
XLogBeginRead(state, tmpRecPtr); |
| 1484 |
while (XLogReadRecord(state, &errormsg) != NULL) |
| 1485 |
{ |
| 1486 |
/* past the record we've found, break out */ |
| 1487 |
if (RecPtr <= state->ReadRecPtr) |
| 1488 |
{ |
| 1489 |
/* Rewind the reader to the beginning of the last record. */ |
| 1490 |
found = state->ReadRecPtr; |
| 1491 |
XLogBeginRead(state, found); |
| 1492 |
return found; |
| 1493 |
} |
| 1494 |
} |
| 1495 |
|
| 1496 |
err: |
| 1497 |
XLogReaderInvalReadState(state); |
| 1498 |
|
| 1499 |
return InvalidXLogRecPtr; |
| 1500 |
} |
| 1501 |
|
| 1502 |
/* |
| 1503 |
* Helper function to ease writing of XLogReaderRoutine->page_read callbacks. |
| 1504 |
* If this function is used, caller must supply a segment_open callback in |
| 1505 |
* 'state', as that is used here. |
| 1506 |
* |
| 1507 |
* Read 'count' bytes into 'buf', starting at location 'startptr', from WAL |
| 1508 |
* fetched from timeline 'tli'. |
| 1509 |
* |
| 1510 |
* Returns true if succeeded, false if an error occurs, in which case |
| 1511 |
* 'errinfo' receives error details. |
| 1512 |
*/ |
| 1513 |
bool |
| 1514 |
WALRead(XLogReaderState *state, |
| 1515 |
char *buf, XLogRecPtr startptr, Size count, TimeLineID tli, |
| 1516 |
WALReadError *errinfo) |
| 1517 |
{ |
| 1518 |
char *p; |
| 1519 |
XLogRecPtr recptr; |
| 1520 |
Size nbytes; |
| 1521 |
#ifndef FRONTEND |
| 1522 |
instr_time io_start; |
| 1523 |
#endif |
| 1524 |
|
| 1525 |
p = buf; |
| 1526 |
recptr = startptr; |
| 1527 |
nbytes = count; |
| 1528 |
|
| 1529 |
while (nbytes > 0) |
| 1530 |
{ |
| 1531 |
uint32 startoff; |
| 1532 |
int segbytes; |
| 1533 |
int readbytes; |
| 1534 |
|
| 1535 |
startoff = XLogSegmentOffset(recptr, state->segcxt.ws_segsize); |
| 1536 |
|
| 1537 |
/* |
| 1538 |
* If the data we want is not in a segment we have open, close what we |
| 1539 |
* have (if anything) and open the next one, using the caller's |
| 1540 |
* provided segment_open callback. |
| 1541 |
*/ |
| 1542 |
if (state->seg.ws_file < 0 || |
| 1543 |
!XLByteInSeg(recptr, state->seg.ws_segno, state->segcxt.ws_segsize) || |
| 1544 |
tli != state->seg.ws_tli) |
| 1545 |
{ |
| 1546 |
XLogSegNo nextSegNo; |
| 1547 |
|
| 1548 |
if (state->seg.ws_file >= 0) |
| 1549 |
state->routine.segment_close(state); |
| 1550 |
|
| 1551 |
XLByteToSeg(recptr, nextSegNo, state->segcxt.ws_segsize); |
| 1552 |
state->routine.segment_open(state, nextSegNo, &tli); |
| 1553 |
|
| 1554 |
/* This shouldn't happen -- indicates a bug in segment_open */ |
| 1555 |
Assert(state->seg.ws_file >= 0); |
| 1556 |
|
| 1557 |
/* Update the current segment info. */ |
| 1558 |
state->seg.ws_tli = tli; |
| 1559 |
state->seg.ws_segno = nextSegNo; |
| 1560 |
} |
| 1561 |
|
| 1562 |
/* How many bytes are within this segment? */ |
| 1563 |
if (nbytes > (state->segcxt.ws_segsize - startoff)) |
| 1564 |
segbytes = state->segcxt.ws_segsize - startoff; |
| 1565 |
else |
| 1566 |
segbytes = nbytes; |
| 1567 |
|
| 1568 |
#ifndef FRONTEND |
| 1569 |
/* Measure I/O timing when reading segment */ |
| 1570 |
io_start = pgstat_prepare_io_time(track_wal_io_timing); |
| 1571 |
|
| 1572 |
pgstat_report_wait_start(WAIT_EVENT_WAL_READ); |
| 1573 |
#endif |
| 1574 |
|
| 1575 |
/* Reset errno first; eases reporting non-errno-affecting errors */ |
| 1576 |
errno = 0; |
| 1577 |
readbytes = pg_pread(state->seg.ws_file, p, segbytes, (off_t) startoff); |
| 1578 |
|
| 1579 |
#ifndef FRONTEND |
| 1580 |
pgstat_report_wait_end(); |
| 1581 |
#endif |
| 1582 |
|
| 1583 |
if (readbytes <= 0) |
| 1584 |
{ |
| 1585 |
errinfo->wre_errno = errno; |
| 1586 |
errinfo->wre_req = segbytes; |
| 1587 |
errinfo->wre_read = readbytes; |
| 1588 |
errinfo->wre_off = startoff; |
| 1589 |
errinfo->wre_seg = state->seg; |
| 1590 |
return false; |
| 1591 |
} |
| 1592 |
|
| 1593 |
#ifndef FRONTEND |
| 1594 |
pgstat_count_io_op_time(IOOBJECT_WAL, IOCONTEXT_NORMAL, IOOP_READ, |
| 1595 |
io_start, 1, readbytes); |
| 1596 |
#endif |
| 1597 |
|
| 1598 |
/* Update state for read */ |
| 1599 |
recptr += readbytes; |
| 1600 |
nbytes -= readbytes; |
| 1601 |
p += readbytes; |
| 1602 |
} |
| 1603 |
|
| 1604 |
return true; |
| 1605 |
} |
| 1606 |
|
| 1607 |
/* ---------------------------------------- |
| 1608 |
* Functions for decoding the data and block references in a record. |
| 1609 |
* ---------------------------------------- |
| 1610 |
*/ |
| 1611 |
|
| 1612 |
/* |
| 1613 |
* Private function to reset the state, forgetting all decoded records, if we |
| 1614 |
* are asked to move to a new read position. |
| 1615 |
*/ |
| 1616 |
static void |
| 1617 |
ResetDecoder(XLogReaderState *state) |
| 1618 |
{ |
| 1619 |
DecodedXLogRecord *r; |
| 1620 |
|
| 1621 |
/* Reset the decoded record queue, freeing any oversized records. */ |
| 1622 |
while ((r = state->decode_queue_head) != NULL) |
| 1623 |
{ |
| 1624 |
state->decode_queue_head = r->next; |
| 1625 |
if (r->oversized) |
| 1626 |
pfree(r); |
| 1627 |
} |
| 1628 |
state->decode_queue_tail = NULL; |
| 1629 |
state->decode_queue_head = NULL; |
| 1630 |
state->record = NULL; |
| 1631 |
|
| 1632 |
/* Reset the decode buffer to empty. */ |
| 1633 |
state->decode_buffer_tail = state->decode_buffer; |
| 1634 |
state->decode_buffer_head = state->decode_buffer; |
| 1635 |
|
| 1636 |
/* Clear error state. */ |
| 1637 |
state->errormsg_buf[0] = '\0'; |
| 1638 |
state->errormsg_deferred = false; |
| 1639 |
} |
| 1640 |
|
| 1641 |
/* |
| 1642 |
* Compute the maximum possible amount of padding that could be required to |
| 1643 |
* decode a record, given xl_tot_len from the record's header. This is the |
| 1644 |
* amount of output buffer space that we need to decode a record, though we |
| 1645 |
* might not finish up using it all. |
| 1646 |
* |
| 1647 |
* This computation is pessimistic and assumes the maximum possible number of |
| 1648 |
* blocks, due to lack of better information. |
| 1649 |
*/ |
| 1650 |
size_t |
| 1651 |
DecodeXLogRecordRequiredSpace(size_t xl_tot_len) |
| 1652 |
{ |
| 1653 |
size_t size = 0; |
| 1654 |
|
| 1655 |
/* Account for the fixed size part of the decoded record struct. */ |
| 1656 |
size += offsetof(DecodedXLogRecord, blocks[0]); |
| 1657 |
/* Account for the flexible blocks array of maximum possible size. */ |
| 1658 |
size += sizeof(DecodedBkpBlock) * (XLR_MAX_BLOCK_ID + 1); |
| 1659 |
/* Account for all the raw main and block data. */ |
| 1660 |
size += xl_tot_len; |
| 1661 |
/* We might insert padding before main_data. */ |
| 1662 |
size += (MAXIMUM_ALIGNOF - 1); |
| 1663 |
/* We might insert padding before each block's data. */ |
| 1664 |
size += (MAXIMUM_ALIGNOF - 1) * (XLR_MAX_BLOCK_ID + 1); |
| 1665 |
/* We might insert padding at the end. */ |
| 1666 |
size += (MAXIMUM_ALIGNOF - 1); |
| 1667 |
|
| 1668 |
return size; |
| 1669 |
} |
| 1670 |
|
| 1671 |
/* |
| 1672 |
* Decode a record. "decoded" must point to a MAXALIGNed memory area that has |
| 1673 |
* space for at least DecodeXLogRecordRequiredSpace(record) bytes. On |
| 1674 |
* success, decoded->size contains the actual space occupied by the decoded |
| 1675 |
* record, which may turn out to be less. |
| 1676 |
* |
| 1677 |
* Only decoded->oversized member must be initialized already, and will not be |
| 1678 |
* modified. Other members will be initialized as required. |
| 1679 |
* |
| 1680 |
* On error, a human-readable error message is returned in *errormsg, and |
| 1681 |
* the return value is false. |
| 1682 |
*/ |
| 1683 |
bool |
| 1684 |
DecodeXLogRecord(XLogReaderState *state, |
| 1685 |
DecodedXLogRecord *decoded, |
| 1686 |
XLogRecord *record, |
| 1687 |
XLogRecPtr lsn, |
| 1688 |
char **errormsg) |
| 1689 |
{ |
| 1690 |
/* |
| 1691 |
* read next _size bytes from record buffer, but check for overrun first. |
| 1692 |
*/ |
| 1693 |
#define COPY_HEADER_FIELD(_dst, _size) \ |
| 1694 |
do { \ |
| 1695 |
if (remaining < _size) \ |
| 1696 |
goto shortdata_err; \ |
| 1697 |
memcpy(_dst, ptr, _size); \ |
| 1698 |
ptr += _size; \ |
| 1699 |
remaining -= _size; \ |
| 1700 |
} while(0) |
| 1701 |
|
| 1702 |
char *ptr; |
| 1703 |
char *out; |
| 1704 |
uint32 remaining; |
| 1705 |
uint32 datatotal; |
| 1706 |
RelFileLocator *rlocator = NULL; |
| 1707 |
uint8 block_id; |
| 1708 |
|
| 1709 |
decoded->header = *record; |
| 1710 |
decoded->lsn = lsn; |
| 1711 |
decoded->next = NULL; |
| 1712 |
decoded->record_origin = InvalidRepOriginId; |
| 1713 |
decoded->toplevel_xid = InvalidTransactionId; |
| 1714 |
decoded->main_data = NULL; |
| 1715 |
decoded->main_data_len = 0; |
| 1716 |
decoded->max_block_id = -1; |
| 1717 |
ptr = (char *) record; |
| 1718 |
ptr += SizeOfXLogRecord; |
| 1719 |
remaining = record->xl_tot_len - SizeOfXLogRecord; |
| 1720 |
|
| 1721 |
/* Decode the headers */ |
| 1722 |
datatotal = 0; |
| 1723 |
while (remaining > datatotal) |
| 1724 |
{ |
| 1725 |
COPY_HEADER_FIELD(&block_id, sizeof(uint8)); |
| 1726 |
|
| 1727 |
if (block_id == XLR_BLOCK_ID_DATA_SHORT) |
| 1728 |
{ |
| 1729 |
/* XLogRecordDataHeaderShort */ |
| 1730 |
uint8 main_data_len; |
| 1731 |
|
| 1732 |
COPY_HEADER_FIELD(&main_data_len, sizeof(uint8)); |
| 1733 |
|
| 1734 |
decoded->main_data_len = main_data_len; |
| 1735 |
datatotal += main_data_len; |
| 1736 |
break; /* by convention, the main data fragment is |
| 1737 |
* always last */ |
| 1738 |
} |
| 1739 |
else if (block_id == XLR_BLOCK_ID_DATA_LONG) |
| 1740 |
{ |
| 1741 |
/* XLogRecordDataHeaderLong */ |
| 1742 |
uint32 main_data_len; |
| 1743 |
|
| 1744 |
COPY_HEADER_FIELD(&main_data_len, sizeof(uint32)); |
| 1745 |
decoded->main_data_len = main_data_len; |
| 1746 |
datatotal += main_data_len; |
| 1747 |
break; /* by convention, the main data fragment is |
| 1748 |
* always last */ |
| 1749 |
} |
| 1750 |
else if (block_id == XLR_BLOCK_ID_ORIGIN) |
| 1751 |
{ |
| 1752 |
COPY_HEADER_FIELD(&decoded->record_origin, sizeof(RepOriginId)); |
| 1753 |
} |
| 1754 |
else if (block_id == XLR_BLOCK_ID_TOPLEVEL_XID) |
| 1755 |
{ |
| 1756 |
COPY_HEADER_FIELD(&decoded->toplevel_xid, sizeof(TransactionId)); |
| 1757 |
} |
| 1758 |
else if (block_id <= XLR_MAX_BLOCK_ID) |
| 1759 |
{ |
| 1760 |
/* XLogRecordBlockHeader */ |
| 1761 |
DecodedBkpBlock *blk; |
| 1762 |
uint8 fork_flags; |
| 1763 |
|
| 1764 |
/* mark any intervening block IDs as not in use */ |
| 1765 |
for (int i = decoded->max_block_id + 1; i < block_id; ++i) |
| 1766 |
decoded->blocks[i].in_use = false; |
| 1767 |
|
| 1768 |
if (block_id <= decoded->max_block_id) |
| 1769 |
{ |
| 1770 |
report_invalid_record(state, |
| 1771 |
"out-of-order block_id %u at %X/%X", |
| 1772 |
block_id, |
| 1773 |
LSN_FORMAT_ARGS(state->ReadRecPtr)); |
| 1774 |
goto err; |
| 1775 |
} |
| 1776 |
decoded->max_block_id = block_id; |
| 1777 |
|
| 1778 |
blk = &decoded->blocks[block_id]; |
| 1779 |
blk->in_use = true; |
| 1780 |
blk->apply_image = false; |
| 1781 |
|
| 1782 |
COPY_HEADER_FIELD(&fork_flags, sizeof(uint8)); |
| 1783 |
blk->forknum = fork_flags & BKPBLOCK_FORK_MASK; |
| 1784 |
blk->flags = fork_flags; |
| 1785 |
blk->has_image = ((fork_flags & BKPBLOCK_HAS_IMAGE) != 0); |
| 1786 |
blk->has_data = ((fork_flags & BKPBLOCK_HAS_DATA) != 0); |
| 1787 |
|
| 1788 |
blk->prefetch_buffer = InvalidBuffer; |
| 1789 |
|
| 1790 |
COPY_HEADER_FIELD(&blk->data_len, sizeof(uint16)); |
| 1791 |
/* cross-check that the HAS_DATA flag is set iff data_length > 0 */ |
| 1792 |
if (blk->has_data && blk->data_len == 0) |
| 1793 |
{ |
| 1794 |
report_invalid_record(state, |
| 1795 |
"BKPBLOCK_HAS_DATA set, but no data included at %X/%X", |
| 1796 |
LSN_FORMAT_ARGS(state->ReadRecPtr)); |
| 1797 |
goto err; |
| 1798 |
} |
| 1799 |
if (!blk->has_data && blk->data_len != 0) |
| 1800 |
{ |
| 1801 |
report_invalid_record(state, |
| 1802 |
"BKPBLOCK_HAS_DATA not set, but data length is %u at %X/%X", |
| 1803 |
(unsigned int) blk->data_len, |
| 1804 |
LSN_FORMAT_ARGS(state->ReadRecPtr)); |
| 1805 |
goto err; |
| 1806 |
} |
| 1807 |
datatotal += blk->data_len; |
| 1808 |
|
| 1809 |
if (blk->has_image) |
| 1810 |
{ |
| 1811 |
COPY_HEADER_FIELD(&blk->bimg_len, sizeof(uint16)); |
| 1812 |
COPY_HEADER_FIELD(&blk->hole_offset, sizeof(uint16)); |
| 1813 |
COPY_HEADER_FIELD(&blk->bimg_info, sizeof(uint8)); |
| 1814 |
|
| 1815 |
blk->apply_image = ((blk->bimg_info & BKPIMAGE_APPLY) != 0); |
| 1816 |
|
| 1817 |
if (BKPIMAGE_COMPRESSED(blk->bimg_info)) |
| 1818 |
{ |
| 1819 |
if (blk->bimg_info & BKPIMAGE_HAS_HOLE) |
| 1820 |
COPY_HEADER_FIELD(&blk->hole_length, sizeof(uint16)); |
| 1821 |
else |
| 1822 |
blk->hole_length = 0; |
| 1823 |
} |
| 1824 |
else |
| 1825 |
blk->hole_length = BLCKSZ - blk->bimg_len; |
| 1826 |
datatotal += blk->bimg_len; |
| 1827 |
|
| 1828 |
/* |
| 1829 |
* cross-check that hole_offset > 0, hole_length > 0 and |
| 1830 |
* bimg_len < BLCKSZ if the HAS_HOLE flag is set. |
| 1831 |
*/ |
| 1832 |
if ((blk->bimg_info & BKPIMAGE_HAS_HOLE) && |
| 1833 |
(blk->hole_offset == 0 || |
| 1834 |
blk->hole_length == 0 || |
| 1835 |
blk->bimg_len == BLCKSZ)) |
| 1836 |
{ |
| 1837 |
report_invalid_record(state, |
| 1838 |
"BKPIMAGE_HAS_HOLE set, but hole offset %u length %u block image length %u at %X/%X", |
| 1839 |
(unsigned int) blk->hole_offset, |
| 1840 |
(unsigned int) blk->hole_length, |
| 1841 |
(unsigned int) blk->bimg_len, |
| 1842 |
LSN_FORMAT_ARGS(state->ReadRecPtr)); |
| 1843 |
goto err; |
| 1844 |
} |
| 1845 |
|
| 1846 |
/* |
| 1847 |
* cross-check that hole_offset == 0 and hole_length == 0 if |
| 1848 |
* the HAS_HOLE flag is not set. |
| 1849 |
*/ |
| 1850 |
if (!(blk->bimg_info & BKPIMAGE_HAS_HOLE) && |
| 1851 |
(blk->hole_offset != 0 || blk->hole_length != 0)) |
| 1852 |
{ |
| 1853 |
report_invalid_record(state, |
| 1854 |
"BKPIMAGE_HAS_HOLE not set, but hole offset %u length %u at %X/%X", |
| 1855 |
(unsigned int) blk->hole_offset, |
| 1856 |
(unsigned int) blk->hole_length, |
| 1857 |
LSN_FORMAT_ARGS(state->ReadRecPtr)); |
| 1858 |
goto err; |
| 1859 |
} |
| 1860 |
|
| 1861 |
/* |
| 1862 |
* Cross-check that bimg_len < BLCKSZ if it is compressed. |
| 1863 |
*/ |
| 1864 |
if (BKPIMAGE_COMPRESSED(blk->bimg_info) && |
| 1865 |
blk->bimg_len == BLCKSZ) |
| 1866 |
{ |
| 1867 |
report_invalid_record(state, |
| 1868 |
"BKPIMAGE_COMPRESSED set, but block image length %u at %X/%X", |
| 1869 |
(unsigned int) blk->bimg_len, |
| 1870 |
LSN_FORMAT_ARGS(state->ReadRecPtr)); |
| 1871 |
goto err; |
| 1872 |
} |
| 1873 |
|
| 1874 |
/* |
| 1875 |
* cross-check that bimg_len = BLCKSZ if neither HAS_HOLE is |
| 1876 |
* set nor COMPRESSED(). |
| 1877 |
*/ |
| 1878 |
if (!(blk->bimg_info & BKPIMAGE_HAS_HOLE) && |
| 1879 |
!BKPIMAGE_COMPRESSED(blk->bimg_info) && |
| 1880 |
blk->bimg_len != BLCKSZ) |
| 1881 |
{ |
| 1882 |
report_invalid_record(state, |
| 1883 |
"neither BKPIMAGE_HAS_HOLE nor BKPIMAGE_COMPRESSED set, but block image length is %u at %X/%X", |
| 1884 |
(unsigned int) blk->data_len, |
| 1885 |
LSN_FORMAT_ARGS(state->ReadRecPtr)); |
| 1886 |
goto err; |
| 1887 |
} |
| 1888 |
} |
| 1889 |
if (!(fork_flags & BKPBLOCK_SAME_REL)) |
| 1890 |
{ |
| 1891 |
COPY_HEADER_FIELD(&blk->rlocator, sizeof(RelFileLocator)); |
| 1892 |
rlocator = &blk->rlocator; |
| 1893 |
} |
| 1894 |
else |
| 1895 |
{ |
| 1896 |
if (rlocator == NULL) |
| 1897 |
{ |
| 1898 |
report_invalid_record(state, |
| 1899 |
"BKPBLOCK_SAME_REL set but no previous rel at %X/%X", |
| 1900 |
LSN_FORMAT_ARGS(state->ReadRecPtr)); |
| 1901 |
goto err; |
| 1902 |
} |
| 1903 |
|
| 1904 |
blk->rlocator = *rlocator; |
| 1905 |
} |
| 1906 |
COPY_HEADER_FIELD(&blk->blkno, sizeof(BlockNumber)); |
| 1907 |
} |
| 1908 |
else |
| 1909 |
{ |
| 1910 |
report_invalid_record(state, |
| 1911 |
"invalid block_id %u at %X/%X", |
| 1912 |
block_id, LSN_FORMAT_ARGS(state->ReadRecPtr)); |
| 1913 |
goto err; |
| 1914 |
} |
| 1915 |
} |
| 1916 |
|
| 1917 |
if (remaining != datatotal) |
| 1918 |
goto shortdata_err; |
| 1919 |
|
| 1920 |
/* |
| 1921 |
* Ok, we've parsed the fragment headers, and verified that the total |
| 1922 |
* length of the payload in the fragments is equal to the amount of data |
| 1923 |
* left. Copy the data of each fragment to contiguous space after the |
| 1924 |
* blocks array, inserting alignment padding before the data fragments so |
| 1925 |
* they can be cast to struct pointers by REDO routines. |
| 1926 |
*/ |
| 1927 |
out = ((char *) decoded) + |
| 1928 |
offsetof(DecodedXLogRecord, blocks) + |
| 1929 |
sizeof(decoded->blocks[0]) * (decoded->max_block_id + 1); |
| 1930 |
|
| 1931 |
/* block data first */ |
| 1932 |
for (block_id = 0; block_id <= decoded->max_block_id; block_id++) |
| 1933 |
{ |
| 1934 |
DecodedBkpBlock *blk = &decoded->blocks[block_id]; |
| 1935 |
|
| 1936 |
if (!blk->in_use) |
| 1937 |
continue; |
| 1938 |
|
| 1939 |
Assert(blk->has_image || !blk->apply_image); |
| 1940 |
|
| 1941 |
if (blk->has_image) |
| 1942 |
{ |
| 1943 |
/* no need to align image */ |
| 1944 |
blk->bkp_image = out; |
| 1945 |
memcpy(out, ptr, blk->bimg_len); |
| 1946 |
ptr += blk->bimg_len; |
| 1947 |
out += blk->bimg_len; |
| 1948 |
} |
| 1949 |
if (blk->has_data) |
| 1950 |
{ |
| 1951 |
out = (char *) MAXALIGN(out); |
| 1952 |
blk->data = out; |
| 1953 |
memcpy(blk->data, ptr, blk->data_len); |
| 1954 |
ptr += blk->data_len; |
| 1955 |
out += blk->data_len; |
| 1956 |
} |
| 1957 |
} |
| 1958 |
|
| 1959 |
/* and finally, the main data */ |
| 1960 |
if (decoded->main_data_len > 0) |
| 1961 |
{ |
| 1962 |
out = (char *) MAXALIGN(out); |
| 1963 |
decoded->main_data = out; |
| 1964 |
memcpy(decoded->main_data, ptr, decoded->main_data_len); |
| 1965 |
ptr += decoded->main_data_len; |
| 1966 |
out += decoded->main_data_len; |
| 1967 |
} |
| 1968 |
|
| 1969 |
/* Report the actual size we used. */ |
| 1970 |
decoded->size = MAXALIGN(out - (char *) decoded); |
| 1971 |
Assert(DecodeXLogRecordRequiredSpace(record->xl_tot_len) >= |
| 1972 |
decoded->size); |
| 1973 |
|
| 1974 |
return true; |
| 1975 |
|
| 1976 |
shortdata_err: |
| 1977 |
report_invalid_record(state, |
| 1978 |
"record with invalid length at %X/%X", |
| 1979 |
LSN_FORMAT_ARGS(state->ReadRecPtr)); |
| 1980 |
err: |
| 1981 |
*errormsg = state->errormsg_buf; |
| 1982 |
|
| 1983 |
return false; |
| 1984 |
} |
| 1985 |
|
| 1986 |
/* |
| 1987 |
* Returns information about the block that a block reference refers to. |
| 1988 |
* |
| 1989 |
* This is like XLogRecGetBlockTagExtended, except that the block reference |
| 1990 |
* must exist and there's no access to prefetch_buffer. |
| 1991 |
*/ |
| 1992 |
void |
| 1993 |
XLogRecGetBlockTag(XLogReaderState *record, uint8 block_id, |
| 1994 |
RelFileLocator *rlocator, ForkNumber *forknum, |
| 1995 |
BlockNumber *blknum) |
| 1996 |
{ |
| 1997 |
if (!XLogRecGetBlockTagExtended(record, block_id, rlocator, forknum, |
| 1998 |
blknum, NULL)) |
| 1999 |
{ |
| 2000 |
#ifndef FRONTEND |
| 2001 |
elog(ERROR, "could not locate backup block with ID %d in WAL record", |
| 2002 |
block_id); |
| 2003 |
#else |
| 2004 |
pg_fatal("could not locate backup block with ID %d in WAL record", |
| 2005 |
block_id); |
| 2006 |
#endif |
| 2007 |
} |
| 2008 |
} |
| 2009 |
|
| 2010 |
/* |
| 2011 |
* Returns information about the block that a block reference refers to, |
| 2012 |
* optionally including the buffer that the block may already be in. |
| 2013 |
* |
| 2014 |
* If the WAL record contains a block reference with the given ID, *rlocator, |
| 2015 |
* *forknum, *blknum and *prefetch_buffer are filled in (if not NULL), and |
| 2016 |
* returns true. Otherwise returns false. |
| 2017 |
*/ |
| 2018 |
bool |
| 2019 |
XLogRecGetBlockTagExtended(XLogReaderState *record, uint8 block_id, |
| 2020 |
RelFileLocator *rlocator, ForkNumber *forknum, |
| 2021 |
BlockNumber *blknum, |
| 2022 |
Buffer *prefetch_buffer) |
| 2023 |
{ |
| 2024 |
DecodedBkpBlock *bkpb; |
| 2025 |
|
| 2026 |
if (!XLogRecHasBlockRef(record, block_id)) |
| 2027 |
return false; |
| 2028 |
|
| 2029 |
bkpb = &record->record->blocks[block_id]; |
| 2030 |
if (rlocator) |
| 2031 |
*rlocator = bkpb->rlocator; |
| 2032 |
if (forknum) |
| 2033 |
*forknum = bkpb->forknum; |
| 2034 |
if (blknum) |
| 2035 |
*blknum = bkpb->blkno; |
| 2036 |
if (prefetch_buffer) |
| 2037 |
*prefetch_buffer = bkpb->prefetch_buffer; |
| 2038 |
return true; |
| 2039 |
} |
| 2040 |
|
| 2041 |
/* |
| 2042 |
* Returns the data associated with a block reference, or NULL if there is |
| 2043 |
* no data (e.g. because a full-page image was taken instead). The returned |
| 2044 |
* pointer points to a MAXALIGNed buffer. |
| 2045 |
*/ |
| 2046 |
char * |
| 2047 |
XLogRecGetBlockData(XLogReaderState *record, uint8 block_id, Size *len) |
| 2048 |
{ |
| 2049 |
DecodedBkpBlock *bkpb; |
| 2050 |
|
| 2051 |
if (block_id > record->record->max_block_id || |
| 2052 |
!record->record->blocks[block_id].in_use) |
| 2053 |
return NULL; |
| 2054 |
|
| 2055 |
bkpb = &record->record->blocks[block_id]; |
| 2056 |
|
| 2057 |
if (!bkpb->has_data) |
| 2058 |
{ |
| 2059 |
if (len) |
| 2060 |
*len = 0; |
| 2061 |
return NULL; |
| 2062 |
} |
| 2063 |
else |
| 2064 |
{ |
| 2065 |
if (len) |
| 2066 |
*len = bkpb->data_len; |
| 2067 |
return bkpb->data; |
| 2068 |
} |
| 2069 |
} |
| 2070 |
|
| 2071 |
/* |
| 2072 |
* Restore a full-page image from a backup block attached to an XLOG record. |
| 2073 |
* |
| 2074 |
* Returns true if a full-page image is restored, and false on failure with |
| 2075 |
* an error to be consumed by the caller. |
| 2076 |
*/ |
| 2077 |
bool |
| 2078 |
RestoreBlockImage(XLogReaderState *record, uint8 block_id, char *page) |
| 2079 |
{ |
| 2080 |
DecodedBkpBlock *bkpb; |
| 2081 |
char *ptr; |
| 2082 |
PGAlignedBlock tmp; |
| 2083 |
|
| 2084 |
if (block_id > record->record->max_block_id || |
| 2085 |
!record->record->blocks[block_id].in_use) |
| 2086 |
{ |
| 2087 |
report_invalid_record(record, |
| 2088 |
"could not restore image at %X/%X with invalid block %d specified", |
| 2089 |
LSN_FORMAT_ARGS(record->ReadRecPtr), |
| 2090 |
block_id); |
| 2091 |
return false; |
| 2092 |
} |
| 2093 |
if (!record->record->blocks[block_id].has_image) |
| 2094 |
{ |
| 2095 |
report_invalid_record(record, "could not restore image at %X/%X with invalid state, block %d", |
| 2096 |
LSN_FORMAT_ARGS(record->ReadRecPtr), |
| 2097 |
block_id); |
| 2098 |
return false; |
| 2099 |
} |
| 2100 |
|
| 2101 |
bkpb = &record->record->blocks[block_id]; |
| 2102 |
ptr = bkpb->bkp_image; |
| 2103 |
|
| 2104 |
if (BKPIMAGE_COMPRESSED(bkpb->bimg_info)) |
| 2105 |
{ |
| 2106 |
/* If a backup block image is compressed, decompress it */ |
| 2107 |
bool decomp_success = true; |
| 2108 |
|
| 2109 |
if ((bkpb->bimg_info & BKPIMAGE_COMPRESS_PGLZ) != 0) |
| 2110 |
{ |
| 2111 |
if (pglz_decompress(ptr, bkpb->bimg_len, tmp.data, |
| 2112 |
BLCKSZ - bkpb->hole_length, true) < 0) |
| 2113 |
decomp_success = false; |
| 2114 |
} |
| 2115 |
else if ((bkpb->bimg_info & BKPIMAGE_COMPRESS_LZ4) != 0) |
| 2116 |
{ |
| 2117 |
#ifdef USE_LZ4 |
| 2118 |
if (LZ4_decompress_safe(ptr, tmp.data, |
| 2119 |
bkpb->bimg_len, BLCKSZ - bkpb->hole_length) <= 0) |
| 2120 |
decomp_success = false; |
| 2121 |
#else |
| 2122 |
report_invalid_record(record, "could not restore image at %X/%X compressed with %s not supported by build, block %d", |
| 2123 |
LSN_FORMAT_ARGS(record->ReadRecPtr), |
| 2124 |
"LZ4", |
| 2125 |
block_id); |
| 2126 |
return false; |
| 2127 |
#endif |
| 2128 |
} |
| 2129 |
else if ((bkpb->bimg_info & BKPIMAGE_COMPRESS_ZSTD) != 0) |
| 2130 |
{ |
| 2131 |
#ifdef USE_ZSTD |
| 2132 |
size_t decomp_result = ZSTD_decompress(tmp.data, |
| 2133 |
BLCKSZ - bkpb->hole_length, |
| 2134 |
ptr, bkpb->bimg_len); |
| 2135 |
|
| 2136 |
if (ZSTD_isError(decomp_result)) |
| 2137 |
decomp_success = false; |
| 2138 |
#else |
| 2139 |
report_invalid_record(record, "could not restore image at %X/%X compressed with %s not supported by build, block %d", |
| 2140 |
LSN_FORMAT_ARGS(record->ReadRecPtr), |
| 2141 |
"zstd", |
| 2142 |
block_id); |
| 2143 |
return false; |
| 2144 |
#endif |
| 2145 |
} |
| 2146 |
else |
| 2147 |
{ |
| 2148 |
report_invalid_record(record, "could not restore image at %X/%X compressed with unknown method, block %d", |
| 2149 |
LSN_FORMAT_ARGS(record->ReadRecPtr), |
| 2150 |
block_id); |
| 2151 |
return false; |
| 2152 |
} |
| 2153 |
|
| 2154 |
if (!decomp_success) |
| 2155 |
{ |
| 2156 |
report_invalid_record(record, "could not decompress image at %X/%X, block %d", |
| 2157 |
LSN_FORMAT_ARGS(record->ReadRecPtr), |
| 2158 |
block_id); |
| 2159 |
return false; |
| 2160 |
} |
| 2161 |
|
| 2162 |
ptr = tmp.data; |
| 2163 |
} |
| 2164 |
|
| 2165 |
/* generate page, taking into account hole if necessary */ |
| 2166 |
if (bkpb->hole_length == 0) |
| 2167 |
{ |
| 2168 |
memcpy(page, ptr, BLCKSZ); |
| 2169 |
} |
| 2170 |
else |
| 2171 |
{ |
| 2172 |
memcpy(page, ptr, bkpb->hole_offset); |
| 2173 |
/* must zero-fill the hole */ |
| 2174 |
MemSet(page + bkpb->hole_offset, 0, bkpb->hole_length); |
| 2175 |
memcpy(page + (bkpb->hole_offset + bkpb->hole_length), |
| 2176 |
ptr + bkpb->hole_offset, |
| 2177 |
BLCKSZ - (bkpb->hole_offset + bkpb->hole_length)); |
| 2178 |
} |
| 2179 |
|
| 2180 |
return true; |
| 2181 |
} |
| 2182 |
|
| 2183 |
#ifndef FRONTEND |
| 2184 |
|
| 2185 |
/* |
| 2186 |
* Extract the FullTransactionId from a WAL record. |
| 2187 |
*/ |
| 2188 |
FullTransactionId |
| 2189 |
XLogRecGetFullXid(XLogReaderState *record) |
| 2190 |
{ |
| 2191 |
/* |
| 2192 |
* This function is only safe during replay, because it depends on the |
| 2193 |
* replay state. See AdvanceNextFullTransactionIdPastXid() for more. |
| 2194 |
*/ |
| 2195 |
Assert(AmStartupProcess() || !IsUnderPostmaster); |
| 2196 |
|
| 2197 |
return FullTransactionIdFromAllowableAt(TransamVariables->nextXid, |
| 2198 |
XLogRecGetXid(record)); |
| 2199 |
} |
| 2200 |
|
| 2201 |
#endif |