1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
|
/******************************************************
MariaBackup: hot backup tool for InnoDB
(c) 2009-2013 Percona LLC and/or its affiliates.
Originally Created 3/3/2009 Yasufumi Kinoshita
Written by Alexey Kopytov, Aleksandr Kuzminsky, Stewart Smith, Vadim Tkachenko,
Yasufumi Kinoshita, Ignacio Nin and Baron Schwartz.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
*******************************************************/
/* Source file cursor implementation */
#include <my_base.h>
#include <fil0fil.h>
#include <fsp0fsp.h>
#include <srv0start.h>
#include <trx0sys.h>
#include "fil_cur.h"
#include "fil0crypt.h"
#include "fil0pagecompress.h"
#include "common.h"
#include "read_filt.h"
#include "xtrabackup.h"
#include "xb0xb.h"
/* Size of read buffer in pages (640 pages = 10M for 16K sized pages) */
#define XB_FIL_CUR_PAGES 640
/***********************************************************************
Extracts the relative path ("database/table.ibd") of a tablespace from a
specified possibly absolute path.
For user tablespaces both "./database/table.ibd" and
"/remote/dir/database/table.ibd" result in "database/table.ibd".
For system tablepsaces (i.e. When is_system is TRUE) both "/remote/dir/ibdata1"
and "./ibdata1" yield "ibdata1" in the output. */
const char *
xb_get_relative_path(
/*=================*/
const char* path, /*!< in: tablespace path (either
relative or absolute) */
ibool is_system) /*!< in: TRUE for system tablespaces,
i.e. when only the filename must be
returned. */
{
const char *next;
const char *cur;
const char *prev;
prev = NULL;
cur = path;
while ((next = strchr(cur, OS_PATH_SEPARATOR)) != NULL) {
prev = cur;
cur = next + 1;
}
if (is_system) {
return(cur);
} else {
return((prev == NULL) ? cur : prev);
}
}
/**********************************************************************//**
Closes a file. */
static
void
xb_fil_node_close_file(
/*===================*/
fil_node_t* node) /*!< in: file node */
{
ibool ret;
mutex_enter(&fil_system->mutex);
ut_ad(node);
ut_a(node->n_pending == 0);
ut_a(node->n_pending_flushes == 0);
ut_a(!node->being_extended);
if (!node->is_open()) {
mutex_exit(&fil_system->mutex);
return;
}
ret = os_file_close(node->handle);
ut_a(ret);
node->handle = OS_FILE_CLOSED;
ut_a(fil_system->n_open > 0);
fil_system->n_open--;
if (node->space->purpose == FIL_TYPE_TABLESPACE &&
fil_is_user_tablespace_id(node->space->id)) {
ut_a(UT_LIST_GET_LEN(fil_system->LRU) > 0);
/* The node is in the LRU list, remove it */
UT_LIST_REMOVE(fil_system->LRU, node);
}
mutex_exit(&fil_system->mutex);
}
/************************************************************************
Open a source file cursor and initialize the associated read filter.
@return XB_FIL_CUR_SUCCESS on success, XB_FIL_CUR_SKIP if the source file must
be skipped and XB_FIL_CUR_ERROR on error. */
xb_fil_cur_result_t
xb_fil_cur_open(
/*============*/
xb_fil_cur_t* cursor, /*!< out: source file cursor */
xb_read_filt_t* read_filter, /*!< in/out: the read filter */
fil_node_t* node, /*!< in: source tablespace node */
uint thread_n, /*!< thread number for diagnostics */
ulonglong max_file_size)
{
bool success;
int err;
/* Initialize these first so xb_fil_cur_close() handles them correctly
in case of error */
cursor->orig_buf = NULL;
cursor->node = NULL;
cursor->space_id = node->space->id;
strncpy(cursor->abs_path, node->name, sizeof(cursor->abs_path));
/* Get the relative path for the destination tablespace name, i.e. the
one that can be appended to the backup root directory. Non-system
tablespaces may have absolute paths for DATA DIRECTORY.
We want to make "local" copies for the backup. */
strncpy(cursor->rel_path,
xb_get_relative_path(cursor->abs_path, cursor->is_system()),
sizeof(cursor->rel_path));
/* In the backup mode we should already have a tablespace handle created
by fil_ibd_load() unless it is a system
tablespace. Otherwise we open the file here. */
if (!node->is_open()) {
ut_ad(cursor->is_system()
|| srv_operation == SRV_OPERATION_RESTORE_DELTA
|| xb_close_files);
node->handle = os_file_create_simple_no_error_handling(
0, node->name,
OS_FILE_OPEN,
OS_FILE_READ_ALLOW_DELETE, true, &success);
if (!success) {
/* The following call prints an error message */
os_file_get_last_error(TRUE);
msg("[%02u] mariabackup: error: cannot open "
"tablespace %s\n",
thread_n, cursor->abs_path);
return(XB_FIL_CUR_SKIP);
}
mutex_enter(&fil_system->mutex);
fil_system->n_open++;
if (node->space->purpose == FIL_TYPE_TABLESPACE &&
fil_is_user_tablespace_id(node->space->id)) {
/* Put the node to the LRU list */
UT_LIST_ADD_FIRST(fil_system->LRU, node);
}
mutex_exit(&fil_system->mutex);
}
ut_ad(node->is_open());
cursor->node = node;
cursor->file = node->handle;
#ifdef _WIN32
HANDLE hDup;
DuplicateHandle(GetCurrentProcess(),cursor->file.m_file,
GetCurrentProcess(), &hDup, 0, FALSE, DUPLICATE_SAME_ACCESS);
int filenr = _open_osfhandle((intptr_t)hDup, 0);
if (filenr < 0) {
err = EINVAL;
}
else {
err = _fstat64(filenr, &cursor->statinfo);
close(filenr);
}
#else
err = fstat(cursor->file.m_file, &cursor->statinfo);
#endif
if (max_file_size < (ulonglong)cursor->statinfo.st_size) {
cursor->statinfo.st_size = (ulonglong)max_file_size;
}
if (err) {
msg("[%02u] mariabackup: error: cannot fstat %s\n",
thread_n, cursor->abs_path);
xb_fil_cur_close(cursor);
return(XB_FIL_CUR_SKIP);
}
if (srv_file_flush_method == SRV_O_DIRECT
|| srv_file_flush_method == SRV_O_DIRECT_NO_FSYNC) {
os_file_set_nocache(cursor->file, node->name, "OPEN");
}
posix_fadvise(cursor->file, 0, 0, POSIX_FADV_SEQUENTIAL);
const page_size_t page_size(node->space->flags);
cursor->page_size = page_size;
/* Allocate read buffer */
cursor->buf_size = XB_FIL_CUR_PAGES * page_size.physical();
cursor->orig_buf = static_cast<byte *>
(malloc(cursor->buf_size + UNIV_PAGE_SIZE));
cursor->buf = static_cast<byte *>
(ut_align(cursor->orig_buf, UNIV_PAGE_SIZE));
cursor->buf_read = 0;
cursor->buf_npages = 0;
cursor->buf_offset = 0;
cursor->buf_page_no = 0;
cursor->thread_n = thread_n;
if (!node->space->crypt_data
&& os_file_read(IORequestRead,
node->handle, cursor->buf, 0,
page_size.physical())) {
mutex_enter(&fil_system->mutex);
if (!node->space->crypt_data) {
node->space->crypt_data
= fil_space_read_crypt_data(page_size,
cursor->buf);
}
mutex_exit(&fil_system->mutex);
}
cursor->space_size = (ulint)(cursor->statinfo.st_size
/ page_size.physical());
cursor->read_filter = read_filter;
cursor->read_filter->init(&cursor->read_filter_ctxt, cursor,
node->space->id);
return(XB_FIL_CUR_SUCCESS);
}
static bool page_is_corrupted(byte *page, ulint page_no, xb_fil_cur_t *cursor,
fil_space_t *space)
{
byte tmp_frame[UNIV_PAGE_SIZE_MAX];
byte tmp_page[UNIV_PAGE_SIZE_MAX];
const ulint page_size = cursor->page_size.physical();
ulint page_type = mach_read_from_2(page + FIL_PAGE_TYPE);
/* We ignore the doublewrite buffer pages.*/
if (cursor->space_id == TRX_SYS_SPACE
&& page_no >= FSP_EXTENT_SIZE
&& page_no < FSP_EXTENT_SIZE * 3) {
return false;
}
/* Validate page number. */
if (mach_read_from_4(page + FIL_PAGE_OFFSET) != page_no
&& cursor->space_id != TRX_SYS_SPACE) {
/* On pages that are not all zero, the
page number must match.
There may be a mismatch on tablespace ID,
because files may be renamed during backup.
We disable the page number check
on the system tablespace, because it may consist
of multiple files, and here we count the pages
from the start of each file.)
The first 38 and last 8 bytes are never encrypted. */
const ulint* p = reinterpret_cast<ulint*>(page);
const ulint* const end = reinterpret_cast<ulint*>(
page + page_size);
do {
if (*p++) {
return true;
}
} while (p != end);
/* Whole zero page is valid. */
return false;
}
/* Validate encrypted pages. The first page is never encrypted.
In the system tablespace, the first page would be written with
FIL_PAGE_FILE_FLUSH_LSN at shutdown, and if the LSN exceeds
4,294,967,295, the mach_read_from_4() below would wrongly
interpret the page as encrypted. We prevent that by checking
page_no first. */
if (page_no
&& mach_read_from_4(page + FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION)
&& space->crypt_data
&& space->crypt_data->type != CRYPT_SCHEME_UNENCRYPTED) {
if (!fil_space_verify_crypt_checksum(page, cursor->page_size))
return true;
/* Compressed encrypted need to be decrypted
and decompressed for verification. */
if (page_type != FIL_PAGE_PAGE_COMPRESSED_ENCRYPTED
&& !opt_extended_validation)
return false;
memcpy(tmp_page, page, page_size);
bool decrypted = false;
if (!fil_space_decrypt(space, tmp_frame, tmp_page, &decrypted)) {
return true;
}
if (page_type != FIL_PAGE_PAGE_COMPRESSED_ENCRYPTED) {
return buf_page_is_corrupted(true, tmp_page,
cursor->page_size, space);
}
}
if (page_type == FIL_PAGE_PAGE_COMPRESSED) {
memcpy(tmp_page, page, page_size);
}
if (page_type == FIL_PAGE_PAGE_COMPRESSED
|| page_type == FIL_PAGE_PAGE_COMPRESSED_ENCRYPTED) {
ulint decomp = fil_page_decompress(tmp_frame, tmp_page);
page_type = mach_read_from_2(tmp_page + FIL_PAGE_TYPE);
return (!decomp
|| (decomp != srv_page_size
&& cursor->page_size.is_compressed())
|| page_type == FIL_PAGE_PAGE_COMPRESSED
|| page_type == FIL_PAGE_PAGE_COMPRESSED_ENCRYPTED
|| buf_page_is_corrupted(true, tmp_page,
cursor->page_size, space));
}
return buf_page_is_corrupted(true, page, cursor->page_size, space);
}
/************************************************************************
Reads and verifies the next block of pages from the source
file. Positions the cursor after the last read non-corrupted page.
@return XB_FIL_CUR_SUCCESS if some have been read successfully, XB_FIL_CUR_EOF
if there are no more pages to read and XB_FIL_CUR_ERROR on error. */
xb_fil_cur_result_t
xb_fil_cur_read(
/*============*/
xb_fil_cur_t* cursor) /*!< in/out: source file cursor */
{
byte* page;
ulint i;
ulint npages;
ulint retry_count;
xb_fil_cur_result_t ret;
ib_int64_t offset;
ib_int64_t to_read;
const ulint page_size = cursor->page_size.physical();
xb_ad(!cursor->is_system() || page_size == UNIV_PAGE_SIZE);
cursor->read_filter->get_next_batch(&cursor->read_filter_ctxt,
&offset, &to_read);
if (to_read == 0LL) {
return(XB_FIL_CUR_EOF);
}
if (to_read > (ib_int64_t) cursor->buf_size) {
to_read = (ib_int64_t) cursor->buf_size;
}
xb_a(to_read > 0 && to_read <= 0xFFFFFFFFLL);
if ((to_read & ~(page_size - 1))
&& offset + to_read == cursor->statinfo.st_size) {
if (to_read < (ib_int64_t) page_size) {
msg("[%02u] mariabackup: Warning: junk at the end of "
"%s:\n", cursor->thread_n, cursor->abs_path);
msg("[%02u] mariabackup: Warning: offset = %llu, "
"to_read = %llu\n",
cursor->thread_n,
(unsigned long long) offset,
(unsigned long long) to_read);
return(XB_FIL_CUR_EOF);
}
to_read = (ib_int64_t) (((ulint) to_read) &
~(page_size - 1));
}
xb_a((to_read & (page_size - 1)) == 0);
npages = (ulint) (to_read / page_size);
retry_count = 10;
ret = XB_FIL_CUR_SUCCESS;
fil_space_t *space = fil_space_acquire_for_io(cursor->space_id);
if (!space) {
return XB_FIL_CUR_ERROR;
}
read_retry:
xtrabackup_io_throttling();
cursor->buf_read = 0;
cursor->buf_npages = 0;
cursor->buf_offset = offset;
cursor->buf_page_no = (ulint)(offset / page_size);
if (!os_file_read(IORequestRead, cursor->file, cursor->buf, offset,
(ulint) to_read)) {
ret = XB_FIL_CUR_ERROR;
goto func_exit;
}
/* check pages for corruption and re-read if necessary. i.e. in case of
partially written pages */
for (page = cursor->buf, i = 0; i < npages;
page += page_size, i++) {
ulint page_no = cursor->buf_page_no + i;
if (page_is_corrupted(page, page_no, cursor, space)){
retry_count--;
if (retry_count == 0) {
msg("[%02u] mariabackup: "
"Error: failed to read page after "
"10 retries. File %s seems to be "
"corrupted.\n", cursor->thread_n,
cursor->abs_path);
ret = XB_FIL_CUR_ERROR;
break;
}
msg("[%02u] mariabackup: "
"Database page corruption detected at page "
ULINTPF ", retrying...\n", cursor->thread_n,
page_no);
os_thread_sleep(100000);
goto read_retry;
}
cursor->buf_read += page_size;
cursor->buf_npages++;
}
posix_fadvise(cursor->file, offset, to_read, POSIX_FADV_DONTNEED);
func_exit:
fil_space_release_for_io(space);
return(ret);
}
/************************************************************************
Close the source file cursor opened with xb_fil_cur_open() and its
associated read filter. */
void
xb_fil_cur_close(
/*=============*/
xb_fil_cur_t *cursor) /*!< in/out: source file cursor */
{
if (cursor->read_filter) {
cursor->read_filter->deinit(&cursor->read_filter_ctxt);
}
free(cursor->orig_buf);
if (cursor->node != NULL) {
xb_fil_node_close_file(cursor->node);
cursor->file = OS_FILE_CLOSED;
}
}
|