Percona XtraBackup moved to https://jira.percona.com/projects/PXB

Merge lp:~akopytov/percona-xtrabackup/rebase-2.1-on-newer-server-versions into lp:percona-xtrabackup/2.1

  1. rebase-2.1-on-newer-server-versions
  2. Merge into 2.1
Proposed by Alexey Kopytov
Status: Merged
Approved by: Laurynas Biveinis
Approved revision: no longer in the source branch.
Merged at revision: 630
Proposed branch: lp:~akopytov/percona-xtrabackup/rebase-2.1-on-newer-server-versions
Merge into: lp:percona-xtrabackup/2.1
Diff against target: 4471 lines (+762/-724)
12 files modified
Makefile (+9/-39)
patches/innodb51.patch (+87/-88)
patches/innodb55.patch (+99/-92)
patches/innodb56.patch (+70/-67)
patches/xtradb51.patch (+114/-119)
patches/xtradb55.patch (+238/-228)
src/Makefile (+1/-1)
src/compact.cc (+1/-1)
src/innodb_int.cc (+58/-2)
src/innodb_int.h (+52/-2)
src/xtrabackup.cc (+8/-12)
utils/build.sh (+25/-73)
To merge this branch: bzr merge lp:~akopytov/percona-xtrabackup/rebase-2.1-on-newer-server-versions
Reviewer Review Type Date Requested Status
Laurynas Biveinis (community) Approve
Review via email: mp+176200@code.launchpad.net

Description of the change

Rebased XtraBackup on MySQL versions 5.1.70, 5.5.30, 5.6.11 and Percona
Server versions 5.1.70-rel14.8 and 5.5.31-rel30.3.

To post a comment you must log in.
Revision history for this message
Alexey Kopytov (akopytov) wrote :

http://jenkins.percona.com/view/XtraBackup/job/percona-xtrabackup-2.1-param/382/

innodb56/xtradb56 failures are bug #1203669.

galera55 failures are bug #1201686 and bug #1203668 (currently being looked at by Raghu).

Revision history for this message
Laurynas Biveinis (laurynas-biveinis) wrote :

Indentation or tabs vs spaces in diff lines 4377 to the end?

review: Needs Fixing
Revision history for this message
Alexey Kopytov (akopytov) wrote :

I have standardized on spaces for indentation for XB shell scripts, because that's what the majority of code uses. I fix that in existing code whenever I have to touch a line. I deliberately kept tabs on the lines I didn't have to touch to simplify the patch.

Revision history for this message
Laurynas Biveinis (laurynas-biveinis) wrote :

OK

review: Approve

Preview Diff

[H/L] Next/Prev Comment, [J/K] Next/Prev File, [N/P] Next/Prev Hunk
1=== modified file 'Makefile'
2--- Makefile 2013-06-18 08:07:48 +0000
3+++ Makefile 2013-07-22 14:01:36 +0000
4@@ -4,61 +4,31 @@
5
6 # Here we specify what server versions we build against
7
8-MYSQL_51_SOURCE=mysql-5.1.59.tar.gz
9-MYSQL_55_SOURCE=mysql-5.5.17.tar.gz
10-MYSQL_56_SOURCE=mysql-5.6.10.tar.gz
11-PS_51_VERSION=5.1.59-13.0
12-PS_51_SOURCE=Percona-Server-XtraBackup-$(PS_51_VERSION).tar.gz
13-PS_55_VERSION=5.5.16-22.0
14-PS_55_SOURCE=Percona-Server-XtraBackup-$(PS_55_VERSION).tar.gz
15+MYSQL_51_SOURCE=mysql-5.1.70.tar.gz
16+MYSQL_55_SOURCE=mysql-5.5.31.tar.gz
17+MYSQL_56_SOURCE=mysql-5.6.11.tar.gz
18+PS_51_SOURCE=Percona-Server-5.1.70-rel14.8.tar.gz
19+PS_55_SOURCE=Percona-Server-5.5.31-rel30.3.tar.gz
20 BZR_REVNO=$(shell bzr revno 2>/dev/null || cat REVNO)
21 XTRABACKUP_VERSION=$(shell sed -e 's/XTRABACKUP_VERSION=//' < VERSION)
22
23 # targets for fetching server source tarballs
24
25-SERVER_SOURCE_TARBALLS=$(MYSQL_51_SOURCE) $(MYSQL_55_SOURCE) $(MYSQL_56_SOURCE)
26-PS_SERVER=$(PS_51_SOURCE) $(PS_55_SOURCE)
27-
28-# PS server source tarballs are inadequate
29-# for these older versions, so we have to make them ourselves
30+SERVER_SOURCE_TARBALLS=$(MYSQL_51_SOURCE) $(MYSQL_55_SOURCE) \
31+ $(MYSQL_56_SOURCE) $(PS_51_SOURCE) $(PS_55_SOURCE)
32
33 .PHONY: ps51source ps55source
34
35 ps51source: $(PS_51_SOURCE)
36
37-
38-$(PS_51_SOURCE):
39- rm -rf percona-server-5.1-xtrabackup
40- bzr branch -r tag:Percona-Server-$(PS_51_VERSION) lp:percona-server/5.1 percona-server-5.1-xtrabackup
41- cd percona-server-5.1-xtrabackup && bzr export percona-server-5.1-xtrabackup.tar.gz
42- mv percona-server-5.1-xtrabackup/percona-server-5.1-xtrabackup.tar.gz .
43- rm -rf percona-server-5.1-xtrabackup
44- tar xfz percona-server-5.1-xtrabackup.tar.gz
45- cd percona-server-5.1-xtrabackup && make mysql-$(shell echo $(PS_51_VERSION) | sed -e 's/-.*//').tar.gz
46- rm percona-server-5.1-xtrabackup.tar.gz
47- tar cfz $(PS_51_SOURCE) percona-server-5.1-xtrabackup
48- rm -rf percona-server-5.1-xtrabackup
49-
50 ps55source: $(PS_55_SOURCE)
51
52-$(PS_55_SOURCE):
53- rm -rf percona-server-5.5-xtrabackup
54- bzr branch -r tag:Percona-Server-$(PS_55_VERSION) lp:percona-server/5.5 percona-server-5.5-xtrabackup
55- cd percona-server-5.5-xtrabackup && bzr export percona-server-5.5-xtrabackup.tar.gz
56- mv percona-server-5.5-xtrabackup/percona-server-5.5-xtrabackup.tar.gz .
57- rm -rf percona-server-5.5-xtrabackup
58- tar xfz percona-server-5.5-xtrabackup.tar.gz
59- cd percona-server-5.5-xtrabackup && make mysql-$(shell echo $(PS_55_VERSION) | sed -e 's/-.*//').tar.gz
60- rm percona-server-5.5-xtrabackup.tar.gz
61- tar cfz $(PS_55_SOURCE) percona-server-5.5-xtrabackup
62- rm -rf percona-server-5.5-xtrabackup
63-
64 # source dist targets
65
66-dist: $(SERVER_SOURCE_TARBALLS) $(PS_SERVER)
67+dist: $(SERVER_SOURCE_TARBALLS)
68 bzr export percona-xtrabackup-$(XTRABACKUP_VERSION).tar.gz
69 tar xfz percona-xtrabackup-$(XTRABACKUP_VERSION).tar.gz
70- cp $(SERVER_SOURCE_TARBALLS) $(PS_SERVER) percona-xtrabackup-$(XTRABACKUP_VERSION)/
71+ cp $(SERVER_SOURCE_TARBALLS) percona-xtrabackup-$(XTRABACKUP_VERSION)/
72 echo $(BZR_REVNO) > percona-xtrabackup-$(XTRABACKUP_VERSION)/REVNO
73 rm percona-xtrabackup-$(XTRABACKUP_VERSION).tar.gz
74 tar cfz percona-xtrabackup-$(XTRABACKUP_VERSION)-$(BZR_REVNO).tar.gz percona-xtrabackup-$(XTRABACKUP_VERSION)
75
76=== modified file 'patches/innodb51.patch'
77--- patches/innodb51.patch 2013-05-07 13:32:47 +0000
78+++ patches/innodb51.patch 2013-07-22 14:01:36 +0000
79@@ -9,7 +9,7 @@
80 buf_block_t*
81 btr_root_block_get(
82 /*===============*/
83-@@ -1122,7 +1122,7 @@
84+@@ -1169,7 +1169,7 @@
85 /************************************************************//**
86 Returns the child page of a node pointer and x-latches it.
87 @return child page, x-latched */
88@@ -18,16 +18,6 @@
89 buf_block_t*
90 btr_node_ptr_get_child(
91 /*===================*/
92---- a/storage/innodb_plugin/btr/btr0cur.c
93-+++ b/storage/innodb_plugin/btr/btr0cur.c
94-@@ -3308,6 +3308,7 @@
95- byte* field;
96-
97- field = rec_get_nth_field(rec, offsets, i, &rec_len);
98-+ (void)field;
99-
100- if (rec_len != UNIV_SQL_NULL) {
101- n_not_null[i]++;
102 --- a/storage/innodb_plugin/buf/buf0buf.c
103 +++ b/storage/innodb_plugin/buf/buf0buf.c
104 @@ -51,6 +51,7 @@
105@@ -38,7 +28,7 @@
106
107 /*
108 IMPLEMENTATION OF THE BUFFER POOL
109-@@ -358,7 +359,7 @@
110+@@ -350,7 +351,7 @@
111 return(TRUE);
112 }
113
114@@ -47,7 +37,7 @@
115 if (recv_lsn_checks_on) {
116 ib_uint64_t current_lsn;
117
118-@@ -670,6 +671,7 @@
119+@@ -662,6 +663,7 @@
120 block->page.in_flush_list = FALSE;
121 block->page.in_free_list = FALSE;
122 block->page.in_LRU_list = FALSE;
123@@ -55,7 +45,7 @@
124 block->in_unzip_LRU_list = FALSE;
125 #endif /* UNIV_DEBUG */
126 #if defined UNIV_AHI_DEBUG || defined UNIV_DEBUG
127-@@ -1690,7 +1692,9 @@
128+@@ -1588,7 +1590,9 @@
129 ut_ad(zip_size == fil_space_get_zip_size(space));
130 ut_ad(ut_is_2pow(zip_size));
131 #ifndef UNIV_LOG_DEBUG
132@@ -66,7 +56,7 @@
133 #endif
134 buf_pool->stat.n_page_gets++;
135 loop:
136-@@ -2802,6 +2806,12 @@
137+@@ -2704,6 +2708,12 @@
138 frame = ((buf_block_t*) bpage)->frame;
139 }
140
141@@ -79,7 +69,7 @@
142 /* If this page is not uninitialized and not in the
143 doublewrite buffer, then the page number and space id
144 should be the same as in block. */
145-@@ -2894,7 +2904,8 @@
146+@@ -2796,7 +2806,8 @@
147 recv_recover_page(TRUE, (buf_block_t*) bpage);
148 }
149
150@@ -199,7 +189,7 @@
151
152 /*
153 IMPLEMENTATION OF THE TABLESPACE MEMORY CACHE
154-@@ -287,7 +289,7 @@
155+@@ -288,7 +290,7 @@
156
157 /** The tablespace memory cache. This variable is NULL before the module is
158 initialized. */
159@@ -208,7 +198,7 @@
160
161
162 /********************************************************************//**
163-@@ -298,7 +300,7 @@
164+@@ -299,7 +301,7 @@
165 off the LRU list if it is in the LRU list. The caller must hold the fil_sys
166 mutex. */
167 static
168@@ -217,7 +207,7 @@
169 fil_node_prepare_for_io(
170 /*====================*/
171 fil_node_t* node, /*!< in: file node */
172-@@ -623,7 +625,7 @@
173+@@ -624,7 +626,7 @@
174 Opens a the file of a node of a tablespace. The caller must own the fil_system
175 mutex. */
176 static
177@@ -226,7 +216,7 @@
178 fil_node_open_file(
179 /*===============*/
180 fil_node_t* node, /*!< in: file node */
181-@@ -657,7 +659,17 @@
182+@@ -658,7 +660,17 @@
183 node->name, OS_FILE_OPEN, OS_FILE_READ_ONLY, &success);
184 if (!success) {
185 /* The following call prints an error message */
186@@ -245,7 +235,7 @@
187
188 ut_print_timestamp(stderr);
189
190-@@ -718,12 +730,15 @@
191+@@ -719,12 +731,15 @@
192
193 if (UNIV_UNLIKELY(space_id != space->id)) {
194 fprintf(stderr,
195@@ -264,7 +254,7 @@
196 }
197
198 if (UNIV_UNLIKELY(space_id == ULINT_UNDEFINED
199-@@ -747,8 +762,8 @@
200+@@ -748,8 +763,8 @@
201 }
202
203 if (size_bytes >= 1024 * 1024) {
204@@ -275,7 +265,7 @@
205 }
206
207 if (!(flags & DICT_TF_ZSSIZE_MASK)) {
208-@@ -793,6 +808,8 @@
209+@@ -794,6 +809,8 @@
210 /* Put the node to the LRU list */
211 UT_LIST_ADD_FIRST(LRU, system->LRU, node);
212 }
213@@ -284,7 +274,7 @@
214 }
215
216 /**********************************************************************//**
217-@@ -1418,7 +1435,12 @@
218+@@ -1438,7 +1455,12 @@
219 the file yet; the following calls will open it and update the
220 size fields */
221
222@@ -298,7 +288,7 @@
223 fil_node_complete_io(node, fil_system, OS_FILE_READ);
224 }
225
226-@@ -1470,7 +1492,12 @@
227+@@ -1490,7 +1512,12 @@
228 the file yet; the following calls will open it and update the
229 size fields */
230
231@@ -312,7 +302,7 @@
232 fil_node_complete_io(node, fil_system, OS_FILE_READ);
233 }
234
235-@@ -1910,7 +1937,7 @@
236+@@ -1929,7 +1956,7 @@
237 mem_free(path);
238 }
239
240@@ -321,7 +311,7 @@
241 /********************************************************//**
242 Writes a log record about an .ibd file create/rename/delete. */
243 static
244-@@ -2134,7 +2161,7 @@
245+@@ -2153,7 +2180,7 @@
246 if (fil_create_new_single_table_tablespace(
247 space_id, name, FALSE, flags,
248 FIL_IBD_FILE_INITIAL_SIZE) != DB_SUCCESS) {
249@@ -330,7 +320,7 @@
250 }
251 }
252
253-@@ -2297,7 +2324,7 @@
254+@@ -2317,7 +2344,7 @@
255 }
256
257 if (success) {
258@@ -339,16 +329,16 @@
259 /* Write a log record about the deletion of the .ibd
260 file, so that ibbackup can replay it in the
261 --apply-log phase. We use a dummy mtr and the familiar
262-@@ -2598,7 +2625,7 @@
263+@@ -2618,7 +2645,7 @@
264
265 mutex_exit(&fil_system->mutex);
266
267 -#ifndef UNIV_HOTBACKUP
268 +#ifdef UNDEFINED
269- if (success) {
270+ if (success && !recv_recovery_on) {
271 mtr_t mtr;
272
273-@@ -2788,7 +2815,7 @@
274+@@ -2808,7 +2835,7 @@
275
276 fil_node_create(path, size, space_id, FALSE);
277
278@@ -357,7 +347,7 @@
279 {
280 mtr_t mtr;
281
282-@@ -3041,19 +3068,97 @@
283+@@ -3061,19 +3088,97 @@
284 "InnoDB: open the tablespace file ", stderr);
285 ut_print_filename(stderr, filepath);
286 fputs("!\n"
287@@ -465,7 +455,7 @@
288 return(FALSE);
289 }
290
291-@@ -3284,7 +3389,7 @@
292+@@ -3304,7 +3409,7 @@
293 cannot be ok. */
294
295 size = (((ib_int64_t)size_high) << 32) + (ib_int64_t)size_low;
296@@ -474,7 +464,7 @@
297 if (size < FIL_IBD_FILE_INITIAL_SIZE * UNIV_PAGE_SIZE) {
298 fprintf(stderr,
299 "InnoDB: Error: the size of single-table tablespace"
300-@@ -3414,7 +3519,51 @@
301+@@ -3434,7 +3539,51 @@
302
303 fil_node_create(filepath, 0, space_id, FALSE);
304 func_exit:
305@@ -527,7 +517,7 @@
306 ut_free(buf2);
307 mem_free(filepath);
308 }
309-@@ -3425,7 +3574,7 @@
310+@@ -3445,7 +3594,7 @@
311 idea is to read as much good data as we can and jump over bad data.
312 @return 0 if ok, -1 if error even after the retries, 1 if at the end
313 of the directory */
314@@ -536,7 +526,7 @@
315 int
316 fil_file_readdir_next_file(
317 /*=======================*/
318-@@ -3469,7 +3618,7 @@
319+@@ -3489,7 +3638,7 @@
320 @return DB_SUCCESS or error number */
321 UNIV_INTERN
322 ulint
323@@ -545,7 +535,7 @@
324 /*===================================*/
325 {
326 int ret;
327-@@ -3525,7 +3674,9 @@
328+@@ -3545,7 +3694,9 @@
329 dbinfo.name);
330 srv_normalize_path_for_win(dbpath);
331
332@@ -556,7 +546,7 @@
333
334 if (dbdir != NULL) {
335 /* printf("Opened dir %s\n", dbinfo.name); */
336-@@ -3551,8 +3702,11 @@
337+@@ -3571,8 +3722,11 @@
338 ".ibd")) {
339 /* The name ends in .ibd; try opening
340 the file */
341@@ -569,7 +559,7 @@
342 }
343 next_file_item:
344 ret = fil_file_readdir_next_file(&err,
345-@@ -3724,15 +3878,97 @@
346+@@ -3744,15 +3898,97 @@
347 "InnoDB: in InnoDB data dictionary"
348 " has tablespace id %lu,\n"
349 "InnoDB: but tablespace with that id"
350@@ -675,7 +665,7 @@
351 } else {
352 ut_print_timestamp(stderr);
353 fputs(" InnoDB: Error: table ", stderr);
354-@@ -4121,7 +4357,7 @@
355+@@ -4141,7 +4377,7 @@
356 off the LRU list if it is in the LRU list. The caller must hold the fil_sys
357 mutex. */
358 static
359@@ -684,7 +674,7 @@
360 fil_node_prepare_for_io(
361 /*====================*/
362 fil_node_t* node, /*!< in: file node */
363-@@ -4141,10 +4377,13 @@
364+@@ -4161,10 +4397,13 @@
365 }
366
367 if (node->open == FALSE) {
368@@ -699,7 +689,7 @@
369 }
370
371 if (node->n_pending == 0 && space->purpose == FIL_TABLESPACE
372-@@ -4157,6 +4396,8 @@
373+@@ -4177,6 +4416,8 @@
374 }
375
376 node->n_pending++;
377@@ -708,7 +698,7 @@
378 }
379
380 /********************************************************************//**
381-@@ -4292,7 +4533,9 @@
382+@@ -4312,7 +4553,9 @@
383 ut_ad(recv_no_ibuf_operations || (type == OS_FILE_WRITE)
384 || !ibuf_bitmap_page(zip_size, block_offset)
385 || sync || is_log);
386@@ -719,7 +709,7 @@
387 || ibuf_page(space_id, zip_size, block_offset, NULL));
388 # endif /* UNIV_LOG_DEBUG */
389 if (sync) {
390-@@ -4341,6 +4584,16 @@
391+@@ -4361,6 +4604,16 @@
392
393 ut_ad((mode != OS_AIO_IBUF) || (space->purpose == FIL_TABLESPACE));
394
395@@ -736,6 +726,26 @@
396 node = UT_LIST_GET_FIRST(space->chain);
397
398 for (;;) {
399+@@ -4869,6 +5122,12 @@
400+ const char* tmp_name) /*!< in: temp table name used while
401+ swapping */
402+ {
403++ (void)old_space_id;
404++ (void)old_name;
405++ (void)new_space_id;
406++ (void)new_name;
407++ (void)tmp_name;
408++#ifdef UNDEFINED
409+ mtr_t mtr;
410+ mtr_start(&mtr);
411+ fil_op_write_log(MLOG_FILE_RENAME, old_space_id,
412+@@ -4876,5 +5135,6 @@
413+ fil_op_write_log(MLOG_FILE_RENAME, new_space_id,
414+ 0, 0, new_name, old_name, &mtr);
415+ mtr_commit(&mtr);
416++#endif
417+ }
418+
419 --- a/storage/innodb_plugin/ibuf/ibuf0ibuf.c
420 +++ b/storage/innodb_plugin/ibuf/ibuf0ibuf.c
421 @@ -1061,6 +1061,9 @@
422@@ -748,7 +758,7 @@
423 if (ibuf_fixed_addr_page(space, zip_size, page_no)) {
424
425 return(TRUE);
426-@@ -2277,6 +2280,9 @@
427+@@ -2275,6 +2278,9 @@
428 ulint n_bytes;
429 ulint n_pag2;
430
431@@ -760,7 +770,7 @@
432
433 --- a/storage/innodb_plugin/include/fil0fil.h
434 +++ b/storage/innodb_plugin/include/fil0fil.h
435-@@ -504,7 +504,7 @@
436+@@ -505,7 +505,7 @@
437 @return DB_SUCCESS or error number */
438 UNIV_INTERN
439 ulint
440@@ -771,7 +781,7 @@
441 Returns TRUE if a single-table tablespace does not exist in the memory cache,
442 --- a/storage/innodb_plugin/include/mem0mem.ic
443 +++ b/storage/innodb_plugin/include/mem0mem.ic
444-@@ -367,7 +367,7 @@
445+@@ -363,7 +363,7 @@
446 buf = (byte*)buf + MEM_FIELD_HEADER_SIZE;
447
448 /* Check that the field lengths agree */
449@@ -782,7 +792,7 @@
450 return(buf);
451 --- a/storage/innodb_plugin/include/mtr0mtr.ic
452 +++ b/storage/innodb_plugin/include/mtr0mtr.ic
453-@@ -160,7 +160,7 @@
454+@@ -161,7 +161,7 @@
455 while (offset > 0) {
456 offset -= sizeof(mtr_memo_slot_t);
457
458@@ -989,7 +999,7 @@
459 ibool
460 log_block_checksum_is_ok_or_old_format(
461 /*===================================*/
462-@@ -1486,6 +1487,7 @@
463+@@ -1492,6 +1493,7 @@
464 buf_block_get_page_no(block));
465
466 if ((recv_addr == NULL)
467@@ -997,7 +1007,7 @@
468 || (recv_addr->state == RECV_BEING_PROCESSED)
469 || (recv_addr->state == RECV_PROCESSED)) {
470
471-@@ -1575,9 +1577,16 @@
472+@@ -1581,9 +1583,16 @@
473 if (page_zip) {
474 memset(FIL_PAGE_LSN + page_zip->data, 0, 8);
475 }
476@@ -1015,7 +1025,7 @@
477
478 ib_uint64_t end_lsn;
479
480-@@ -1762,6 +1771,18 @@
481+@@ -1769,6 +1778,18 @@
482 ulint zip_size = fil_space_get_zip_size(space);
483 ulint page_no = recv_addr->page_no;
484
485@@ -1034,7 +1044,7 @@
486 if (recv_addr->state == RECV_NOT_PROCESSED) {
487 if (!has_printed) {
488 ut_print_timestamp(stderr);
489-@@ -1795,7 +1816,7 @@
490+@@ -1802,7 +1823,7 @@
491
492 mutex_enter(&(recv_sys->mutex));
493 }
494@@ -1043,7 +1053,7 @@
495 recv_addr = HASH_GET_NEXT(addr_hash, recv_addr);
496 }
497
498-@@ -2297,7 +2318,7 @@
499+@@ -2304,7 +2325,7 @@
500 || type == MLOG_FILE_RENAME
501 || type == MLOG_FILE_DELETE) {
502 ut_a(space);
503@@ -1052,7 +1062,7 @@
504 if (recv_replay_file_ops) {
505
506 /* In ibbackup --apply-log, replay an .ibd file
507-@@ -2320,7 +2341,7 @@
508+@@ -2327,7 +2348,7 @@
509 ut_error;
510 }
511 }
512@@ -1061,7 +1071,7 @@
513 /* In normal mysqld crash recovery we do not try to
514 replay file operations */
515 #ifdef UNIV_LOG_LSN_DEBUG
516-@@ -2737,8 +2758,11 @@
517+@@ -2744,8 +2765,11 @@
518
519 fprintf(stderr,
520 "InnoDB: Doing recovery: scanned up to"
521@@ -1075,7 +1085,7 @@
522 }
523 }
524
525-@@ -2841,7 +2865,7 @@
526+@@ -2848,7 +2872,7 @@
527 "InnoDB: Reading tablespace information"
528 " from the .ibd files...\n");
529
530@@ -1084,7 +1094,7 @@
531
532 /* If we are using the doublewrite method, we will
533 check if there are half-written pages in data files,
534-@@ -2850,12 +2874,14 @@
535+@@ -2857,12 +2881,14 @@
536
537 if (srv_force_recovery < SRV_FORCE_NO_LOG_REDO) {
538
539@@ -1100,7 +1110,7 @@
540 }
541 }
542
543-@@ -3005,6 +3031,7 @@
544+@@ -3012,6 +3038,7 @@
545 recv_sys->recovered_lsn = checkpoint_lsn;
546
547 srv_start_lsn = checkpoint_lsn;
548@@ -1108,7 +1118,7 @@
549 }
550
551 contiguous_lsn = ut_uint64_align_down(recv_sys->scanned_lsn,
552-@@ -3286,6 +3313,7 @@
553+@@ -3293,6 +3320,7 @@
554 that the data dictionary tables will be free of any locks.
555 The data dictionary latch should guarantee that there is at
556 most one data dictionary transaction active at a time. */
557@@ -1127,7 +1137,7 @@
558 #if defined(UNIV_HOTBACKUP) || defined(__WIN__) || defined(__NETWARE__)
559 /* InnoDB Hot Backup does not lock the data files.
560 * On Windows, mandatory locking is used.
561-@@ -1316,8 +1316,9 @@
562+@@ -1324,8 +1324,9 @@
563 attributes = 0;
564 ut_error;
565 }
566@@ -1138,7 +1148,7 @@
567 GENERIC_READ | GENERIC_WRITE, /* read and write
568 access */
569 share_mode, /* File can be read also by other
570-@@ -1376,7 +1377,11 @@
571+@@ -1391,7 +1392,11 @@
572 if (create_mode == OS_FILE_OPEN || create_mode == OS_FILE_OPEN_RAW
573 || create_mode == OS_FILE_OPEN_RETRY) {
574 mode_str = "OPEN";
575@@ -1151,7 +1161,7 @@
576 } else if (create_mode == OS_FILE_CREATE) {
577 mode_str = "CREATE";
578 create_flag = O_RDWR | O_CREAT | O_EXCL;
579-@@ -2545,6 +2550,9 @@
580+@@ -2560,6 +2565,9 @@
581 ut_a((offset & 0xFFFFFFFFUL) == offset);
582 ut_a((n & 0xFFFFFFFFUL) == n);
583
584@@ -1161,7 +1171,7 @@
585 os_n_file_writes++;
586
587 ut_ad(file);
588-@@ -2669,6 +2677,9 @@
589+@@ -2684,6 +2692,9 @@
590 #else
591 ssize_t ret;
592
593@@ -1193,7 +1203,7 @@
594
595 --- a/storage/innodb_plugin/srv/srv0srv.c
596 +++ b/storage/innodb_plugin/srv/srv0srv.c
597-@@ -374,6 +374,10 @@
598+@@ -383,6 +383,10 @@
599
600 UNIV_INTERN ulong srv_replication_delay = 0;
601
602@@ -1204,7 +1214,7 @@
603 /*-------------------------------------------*/
604 UNIV_INTERN ulong srv_n_spin_wait_rounds = 30;
605 UNIV_INTERN ulong srv_n_free_tickets_to_enter = 500;
606-@@ -1002,7 +1006,7 @@
607+@@ -1011,7 +1015,7 @@
608 }
609
610 /* Initialize some INFORMATION SCHEMA internal structures */
611@@ -1213,7 +1223,7 @@
612 }
613
614 /*********************************************************************//**
615-@@ -1013,6 +1017,7 @@
616+@@ -1022,6 +1026,7 @@
617 /*==========*/
618 {
619 os_fast_mutex_free(&srv_conc_mutex);
620@@ -1221,7 +1231,7 @@
621 mem_free(srv_conc_slots);
622 srv_conc_slots = NULL;
623
624-@@ -1026,6 +1031,7 @@
625+@@ -1035,6 +1040,7 @@
626 srv_mysql_table = NULL;
627
628 trx_i_s_cache_free(trx_i_s_cache);
629@@ -1334,9 +1344,9 @@
630 srv_was_started = TRUE;
631 --- a/storage/innodb_plugin/trx/trx0purge.c
632 +++ b/storage/innodb_plugin/trx/trx0purge.c
633-@@ -1087,6 +1087,9 @@
634+@@ -1093,6 +1093,9 @@
635 /* que_thr_t* thr2; */
636- ulint old_pages_handled;
637+ ulonglong old_pages_handled;
638
639 + if (srv_fake_write)
640 + return(0);
641@@ -1360,7 +1370,7 @@
642
643 --- a/storage/innodb_plugin/trx/trx0sys.c
644 +++ b/storage/innodb_plugin/trx/trx0sys.c
645-@@ -1586,12 +1586,14 @@
646+@@ -1583,12 +1583,14 @@
647 mem_free(trx_doublewrite);
648 trx_doublewrite = NULL;
649
650@@ -1375,7 +1385,7 @@
651
652 /* There can't be any active transactions. */
653 rseg = UT_LIST_GET_FIRST(trx_sys->rseg_list);
654-@@ -1617,10 +1619,12 @@
655+@@ -1614,10 +1616,12 @@
656 UT_LIST_REMOVE(view_list, trx_sys->view_list, prev_view);
657 }
658
659@@ -1425,9 +1435,9 @@
660
661 /*
662 WAIT ARRAY
663-@@ -927,6 +928,13 @@
664- ibool fatal = FALSE;
665- double longest_diff = 0;
666+@@ -932,6 +933,13 @@
667+ return(FALSE);
668+ }
669
670 + if (srv_rebuild_indexes) {
671 +
672@@ -1436,12 +1446,12 @@
673 + return(FALSE);
674 + }
675 +
676+ sync_array_enter(sync_primary_wait_array);
677+
678 for (i = 0; i < sync_primary_wait_array->n_cells; i++) {
679-
680- double diff;
681 --- a/storage/innodb_plugin/include/buf0buf.h
682 +++ b/storage/innodb_plugin/include/buf0buf.h
683-@@ -1191,6 +1191,8 @@
684+@@ -1270,6 +1270,8 @@
685 frees a page in buffer pool */
686 # endif /* UNIV_DEBUG_FILE_ACCESSES || UNIV_DEBUG */
687 #endif /* !UNIV_HOTBACKUP */
688@@ -1493,14 +1503,3 @@
689
690 UNIV_MEM_FREE(buf, n);
691 }
692---- a/configure.in
693-+++ b/configure.in
694-@@ -22,7 +22,7 @@
695- AM_INIT_AUTOMAKE([1.9 tar-ustar])
696- AC_PROG_LIBTOOL
697-
698--AM_CONFIG_HEADER([include/config.h])
699-+AC_CONFIG_HEADERS([include/config.h])
700-
701- # Request support for automake silent-rules if available.
702- # Default to verbose output. One can use the configure-time
703
704=== modified file 'patches/innodb55.patch'
705--- patches/innodb55.patch 2013-05-07 13:32:47 +0000
706+++ patches/innodb55.patch 2013-07-22 14:01:36 +0000
707@@ -1,6 +1,6 @@
708 --- a/storage/innobase/btr/btr0btr.c
709 +++ b/storage/innobase/btr/btr0btr.c
710-@@ -674,7 +674,7 @@
711+@@ -698,7 +698,7 @@
712 /**************************************************************//**
713 Gets the root node of a tree and x-latches it.
714 @return root page, x-latched */
715@@ -9,7 +9,7 @@
716 buf_block_t*
717 btr_root_block_get(
718 /*===============*/
719-@@ -1123,7 +1123,7 @@
720+@@ -1183,7 +1183,7 @@
721 /************************************************************//**
722 Returns the child page of a node pointer and x-latches it.
723 @return child page, x-latched */
724@@ -28,7 +28,7 @@
725
726 /*
727 IMPLEMENTATION OF THE BUFFER POOL
728-@@ -518,7 +519,7 @@
729+@@ -538,7 +539,7 @@
730 return(TRUE);
731 }
732
733@@ -37,7 +37,7 @@
734 if (recv_lsn_checks_on) {
735 ib_uint64_t current_lsn;
736
737-@@ -881,6 +882,7 @@
738+@@ -918,6 +919,7 @@
739 block->page.in_flush_list = FALSE;
740 block->page.in_free_list = FALSE;
741 block->page.in_LRU_list = FALSE;
742@@ -45,7 +45,7 @@
743 block->in_unzip_LRU_list = FALSE;
744 #endif /* UNIV_DEBUG */
745 #if defined UNIV_AHI_DEBUG || defined UNIV_DEBUG
746-@@ -2299,7 +2301,8 @@
747+@@ -2256,7 +2258,8 @@
748 ut_ad(zip_size == fil_space_get_zip_size(space));
749 ut_ad(ut_is_2pow(zip_size));
750 #ifndef UNIV_LOG_DEBUG
751@@ -55,7 +55,7 @@
752 || ibuf_page_low(space, zip_size, offset,
753 FALSE, file, line, NULL));
754 #endif
755-@@ -3567,6 +3570,13 @@
756+@@ -3535,6 +3538,13 @@
757 frame = ((buf_block_t*) bpage)->frame;
758 }
759
760@@ -69,7 +69,7 @@
761 /* If this page is not uninitialized and not in the
762 doublewrite buffer, then the page number and space id
763 should be the same as in block. */
764-@@ -3668,7 +3678,8 @@
765+@@ -3637,7 +3647,8 @@
766 recv_recover_page(TRUE, (buf_block_t*) bpage);
767 }
768
769@@ -81,7 +81,7 @@
770 bpage->offset, buf_page_get_zip_size(bpage),
771 --- a/storage/innobase/buf/buf0rea.c
772 +++ b/storage/innobase/buf/buf0rea.c
773-@@ -124,6 +124,45 @@
774+@@ -162,6 +162,45 @@
775 bpage = buf_page_init_for_read(err, mode, space, zip_size, unzip,
776 tablespace_version, offset);
777 if (bpage == NULL) {
778@@ -127,7 +127,7 @@
779
780 return(0);
781 }
782-@@ -768,6 +807,50 @@
783+@@ -820,6 +859,50 @@
784 /* It is a single table tablespace and the .ibd file is
785 missing: do nothing */
786
787@@ -189,7 +189,7 @@
788 #ifndef UNIV_HOTBACKUP
789 # include "buf0lru.h"
790 # include "ibuf0ibuf.h"
791-@@ -297,7 +299,7 @@
792+@@ -300,7 +302,7 @@
793
794 /** The tablespace memory cache. This variable is NULL before the module is
795 initialized. */
796@@ -198,7 +198,7 @@
797
798 #ifdef UNIV_DEBUG
799 /** Try fil_validate() every this many times */
800-@@ -336,7 +338,7 @@
801+@@ -339,7 +341,7 @@
802 off the LRU list if it is in the LRU list. The caller must hold the fil_sys
803 mutex. */
804 static
805@@ -207,7 +207,7 @@
806 fil_node_prepare_for_io(
807 /*====================*/
808 fil_node_t* node, /*!< in: file node */
809-@@ -661,7 +663,7 @@
810+@@ -664,7 +666,7 @@
811 Opens a the file of a node of a tablespace. The caller must own the fil_system
812 mutex. */
813 static
814@@ -216,7 +216,7 @@
815 fil_node_open_file(
816 /*===============*/
817 fil_node_t* node, /*!< in: file node */
818-@@ -696,7 +698,17 @@
819+@@ -699,7 +701,17 @@
820 OS_FILE_READ_ONLY, &success);
821 if (!success) {
822 /* The following call prints an error message */
823@@ -235,7 +235,7 @@
824
825 ut_print_timestamp(stderr);
826
827-@@ -757,12 +769,15 @@
828+@@ -760,12 +772,15 @@
829
830 if (UNIV_UNLIKELY(space_id != space->id)) {
831 fprintf(stderr,
832@@ -254,7 +254,7 @@
833 }
834
835 if (UNIV_UNLIKELY(space_id == ULINT_UNDEFINED
836-@@ -786,8 +801,8 @@
837+@@ -789,8 +804,8 @@
838 }
839
840 if (size_bytes >= 1024 * 1024) {
841@@ -265,7 +265,7 @@
842 }
843
844 if (!(flags & DICT_TF_ZSSIZE_MASK)) {
845-@@ -838,6 +853,8 @@
846+@@ -841,6 +856,8 @@
847 /* Put the node to the LRU list */
848 UT_LIST_ADD_FIRST(LRU, system->LRU, node);
849 }
850@@ -274,7 +274,7 @@
851 }
852
853 /**********************************************************************//**
854-@@ -1464,7 +1481,12 @@
855+@@ -1504,7 +1521,12 @@
856 the file yet; the following calls will open it and update the
857 size fields */
858
859@@ -288,7 +288,7 @@
860 fil_node_complete_io(node, fil_system, OS_FILE_READ);
861 }
862
863-@@ -1516,7 +1538,12 @@
864+@@ -1573,7 +1595,12 @@
865 the file yet; the following calls will open it and update the
866 size fields */
867
868@@ -302,7 +302,7 @@
869 fil_node_complete_io(node, fil_system, OS_FILE_READ);
870 }
871
872-@@ -1957,7 +1984,7 @@
873+@@ -2021,7 +2048,7 @@
874 mem_free(path);
875 }
876
877@@ -311,7 +311,7 @@
878 /********************************************************//**
879 Writes a log record about an .ibd file create/rename/delete. */
880 static
881-@@ -2181,7 +2208,7 @@
882+@@ -2245,7 +2272,7 @@
883 if (fil_create_new_single_table_tablespace(
884 space_id, name, FALSE, flags,
885 FIL_IBD_FILE_INITIAL_SIZE) != DB_SUCCESS) {
886@@ -320,7 +320,7 @@
887 }
888 }
889
890-@@ -2344,7 +2371,7 @@
891+@@ -2434,7 +2461,7 @@
892 }
893
894 if (success) {
895@@ -329,16 +329,16 @@
896 /* Write a log record about the deletion of the .ibd
897 file, so that ibbackup can replay it in the
898 --apply-log phase. We use a dummy mtr and the familiar
899-@@ -2645,7 +2672,7 @@
900+@@ -2735,7 +2762,7 @@
901
902 mutex_exit(&fil_system->mutex);
903
904 -#ifndef UNIV_HOTBACKUP
905 +#ifdef UNDEFINED
906- if (success) {
907+ if (success && !recv_recovery_on) {
908 mtr_t mtr;
909
910-@@ -2836,7 +2863,7 @@
911+@@ -2926,7 +2953,7 @@
912
913 fil_node_create(path, size, space_id, FALSE);
914
915@@ -347,7 +347,7 @@
916 {
917 mtr_t mtr;
918
919-@@ -3091,19 +3118,97 @@
920+@@ -3181,19 +3208,97 @@
921 "InnoDB: open the tablespace file ", stderr);
922 ut_print_filename(stderr, filepath);
923 fputs("!\n"
924@@ -455,7 +455,7 @@
925 return(FALSE);
926 }
927
928-@@ -3335,7 +3440,7 @@
929+@@ -3425,7 +3530,7 @@
930 cannot be ok. */
931
932 size = (((ib_int64_t)size_high) << 32) + (ib_int64_t)size_low;
933@@ -464,7 +464,7 @@
934 if (size < FIL_IBD_FILE_INITIAL_SIZE * UNIV_PAGE_SIZE) {
935 fprintf(stderr,
936 "InnoDB: Error: the size of single-table tablespace"
937-@@ -3465,7 +3570,51 @@
938+@@ -3555,7 +3660,51 @@
939
940 fil_node_create(filepath, 0, space_id, FALSE);
941 func_exit:
942@@ -517,7 +517,7 @@
943 ut_free(buf2);
944 mem_free(filepath);
945 }
946-@@ -3476,7 +3625,7 @@
947+@@ -3566,7 +3715,7 @@
948 idea is to read as much good data as we can and jump over bad data.
949 @return 0 if ok, -1 if error even after the retries, 1 if at the end
950 of the directory */
951@@ -526,7 +526,7 @@
952 int
953 fil_file_readdir_next_file(
954 /*=======================*/
955-@@ -3520,7 +3669,7 @@
956+@@ -3610,7 +3759,7 @@
957 @return DB_SUCCESS or error number */
958 UNIV_INTERN
959 ulint
960@@ -535,7 +535,7 @@
961 /*===================================*/
962 {
963 int ret;
964-@@ -3576,7 +3725,9 @@
965+@@ -3666,7 +3815,9 @@
966 dbinfo.name);
967 srv_normalize_path_for_win(dbpath);
968
969@@ -546,7 +546,7 @@
970
971 if (dbdir != NULL) {
972 /* printf("Opened dir %s\n", dbinfo.name); */
973-@@ -3602,8 +3753,11 @@
974+@@ -3692,8 +3843,11 @@
975 ".ibd")) {
976 /* The name ends in .ibd; try opening
977 the file */
978@@ -559,7 +559,7 @@
979 }
980 next_file_item:
981 ret = fil_file_readdir_next_file(&err,
982-@@ -3775,15 +3929,97 @@
983+@@ -3865,15 +4019,97 @@
984 "InnoDB: in InnoDB data dictionary"
985 " has tablespace id %lu,\n"
986 "InnoDB: but tablespace with that id"
987@@ -665,7 +665,7 @@
988 } else {
989 ut_print_timestamp(stderr);
990 fputs(" InnoDB: Error: table ", stderr);
991-@@ -4172,7 +4408,7 @@
992+@@ -4262,7 +4498,7 @@
993 off the LRU list if it is in the LRU list. The caller must hold the fil_sys
994 mutex. */
995 static
996@@ -674,7 +674,7 @@
997 fil_node_prepare_for_io(
998 /*====================*/
999 fil_node_t* node, /*!< in: file node */
1000-@@ -4192,10 +4428,13 @@
1001+@@ -4282,10 +4518,13 @@
1002 }
1003
1004 if (node->open == FALSE) {
1005@@ -689,7 +689,7 @@
1006 }
1007
1008 if (node->n_pending == 0 && space->purpose == FIL_TABLESPACE
1009-@@ -4208,6 +4447,8 @@
1010+@@ -4298,6 +4537,8 @@
1011 }
1012
1013 node->n_pending++;
1014@@ -698,7 +698,7 @@
1015 }
1016
1017 /********************************************************************//**
1018-@@ -4390,6 +4631,16 @@
1019+@@ -4482,6 +4723,16 @@
1020
1021 ut_ad((mode != OS_AIO_IBUF) || (space->purpose == FIL_TABLESPACE));
1022
1023@@ -715,9 +715,29 @@
1024 node = UT_LIST_GET_FIRST(space->chain);
1025
1026 for (;;) {
1027+@@ -4997,6 +5248,12 @@
1028+ const char* tmp_name) /*!< in: temp table name used while
1029+ swapping */
1030+ {
1031++ (void)old_space_id;
1032++ (void)old_name;
1033++ (void)new_space_id;
1034++ (void)new_name;
1035++ (void)tmp_name;
1036++#ifdef UNDEFINED
1037+ mtr_t mtr;
1038+ mtr_start(&mtr);
1039+ fil_op_write_log(MLOG_FILE_RENAME, old_space_id,
1040+@@ -5004,5 +5261,6 @@
1041+ fil_op_write_log(MLOG_FILE_RENAME, new_space_id,
1042+ 0, 0, new_name, old_name, &mtr);
1043+ mtr_commit(&mtr);
1044++#endif
1045+ }
1046+
1047 --- a/storage/innobase/ibuf/ibuf0ibuf.c
1048 +++ b/storage/innobase/ibuf/ibuf0ibuf.c
1049-@@ -1148,6 +1148,9 @@
1050+@@ -1157,6 +1157,9 @@
1051 ut_ad(!recv_no_ibuf_operations);
1052 ut_ad(x_latch || mtr == NULL);
1053
1054@@ -727,7 +747,7 @@
1055 if (ibuf_fixed_addr_page(space, zip_size, page_no)) {
1056
1057 return(TRUE);
1058-@@ -2715,6 +2718,9 @@
1059+@@ -2729,6 +2732,9 @@
1060 ulint n_bytes;
1061 ulint n_pag2;
1062
1063@@ -739,7 +759,7 @@
1064
1065 --- a/storage/innobase/include/fil0fil.h
1066 +++ b/storage/innobase/include/fil0fil.h
1067-@@ -505,7 +505,7 @@
1068+@@ -513,7 +513,7 @@
1069 @return DB_SUCCESS or error number */
1070 UNIV_INTERN
1071 ulint
1072@@ -803,16 +823,16 @@
1073
1074 if (log_sys->n_pending_checkpoint_writes > 0) {
1075 /* A checkpoint write is running */
1076-@@ -3114,6 +3116,7 @@
1077+@@ -3141,6 +3143,7 @@
1078 shutdown, because the InnoDB layer may have committed or
1079 prepared transactions and we don't want to lose them. */
1080
1081 + if (!srv_apply_log_only) {
1082 server_busy = trx_n_mysql_transactions > 0
1083 || UT_LIST_GET_LEN(trx_sys->trx_list) > trx_n_prepared;
1084- mutex_exit(&kernel_mutex);
1085-@@ -3121,6 +3124,10 @@
1086- if (server_busy || srv_is_any_background_thread_active()) {
1087+
1088+@@ -3193,6 +3196,10 @@
1089+
1090 goto loop;
1091 }
1092 + }
1093@@ -902,7 +922,7 @@
1094 ibool
1095 log_block_checksum_is_ok_or_old_format(
1096 /*===================================*/
1097-@@ -1496,6 +1497,7 @@
1098+@@ -1502,6 +1503,7 @@
1099 buf_block_get_page_no(block));
1100
1101 if ((recv_addr == NULL)
1102@@ -910,7 +930,7 @@
1103 || (recv_addr->state == RECV_BEING_PROCESSED)
1104 || (recv_addr->state == RECV_PROCESSED)) {
1105
1106-@@ -1585,9 +1587,16 @@
1107+@@ -1591,9 +1593,16 @@
1108 if (page_zip) {
1109 memset(FIL_PAGE_LSN + page_zip->data, 0, 8);
1110 }
1111@@ -928,7 +948,7 @@
1112
1113 ib_uint64_t end_lsn;
1114
1115-@@ -1774,6 +1783,18 @@
1116+@@ -1780,6 +1789,18 @@
1117 ulint zip_size = fil_space_get_zip_size(space);
1118 ulint page_no = recv_addr->page_no;
1119
1120@@ -947,7 +967,7 @@
1121 if (recv_addr->state == RECV_NOT_PROCESSED) {
1122 if (!has_printed) {
1123 ut_print_timestamp(stderr);
1124-@@ -1807,7 +1828,7 @@
1125+@@ -1813,7 +1834,7 @@
1126
1127 mutex_enter(&(recv_sys->mutex));
1128 }
1129@@ -956,7 +976,7 @@
1130 recv_addr = HASH_GET_NEXT(addr_hash, recv_addr);
1131 }
1132
1133-@@ -2308,7 +2329,7 @@
1134+@@ -2314,7 +2335,7 @@
1135 || type == MLOG_FILE_RENAME
1136 || type == MLOG_FILE_DELETE) {
1137 ut_a(space);
1138@@ -965,7 +985,7 @@
1139 if (recv_replay_file_ops) {
1140
1141 /* In ibbackup --apply-log, replay an .ibd file
1142-@@ -2331,7 +2352,7 @@
1143+@@ -2337,7 +2358,7 @@
1144 ut_error;
1145 }
1146 }
1147@@ -974,7 +994,7 @@
1148 /* In normal mysqld crash recovery we do not try to
1149 replay file operations */
1150 #ifdef UNIV_LOG_LSN_DEBUG
1151-@@ -2748,8 +2769,11 @@
1152+@@ -2754,8 +2775,11 @@
1153
1154 fprintf(stderr,
1155 "InnoDB: Doing recovery: scanned up to"
1156@@ -988,7 +1008,7 @@
1157 }
1158 }
1159
1160-@@ -2854,7 +2878,7 @@
1161+@@ -2860,7 +2884,7 @@
1162 "InnoDB: Reading tablespace information"
1163 " from the .ibd files...\n");
1164
1165@@ -997,7 +1017,7 @@
1166
1167 /* If we are using the doublewrite method, we will
1168 check if there are half-written pages in data files,
1169-@@ -2863,12 +2887,14 @@
1170+@@ -2869,12 +2893,14 @@
1171
1172 if (srv_force_recovery < SRV_FORCE_NO_LOG_REDO) {
1173
1174@@ -1013,7 +1033,7 @@
1175 }
1176 }
1177
1178-@@ -3018,6 +3044,7 @@
1179+@@ -3024,6 +3050,7 @@
1180 recv_sys->recovered_lsn = checkpoint_lsn;
1181
1182 srv_start_lsn = checkpoint_lsn;
1183@@ -1021,7 +1041,7 @@
1184 }
1185
1186 contiguous_lsn = ut_uint64_align_down(recv_sys->scanned_lsn,
1187-@@ -3299,6 +3326,7 @@
1188+@@ -3305,6 +3332,7 @@
1189 that the data dictionary tables will be free of any locks.
1190 The data dictionary latch should guarantee that there is at
1191 most one data dictionary transaction active at a time. */
1192@@ -1031,7 +1051,7 @@
1193
1194 --- a/storage/innobase/os/os0file.c
1195 +++ b/storage/innobase/os/os0file.c
1196-@@ -656,7 +656,7 @@
1197+@@ -658,7 +658,7 @@
1198 }
1199
1200 #undef USE_FILE_LOCK
1201@@ -1040,7 +1060,7 @@
1202 #if defined(UNIV_HOTBACKUP) || defined(__WIN__)
1203 /* InnoDB Hot Backup does not lock the data files.
1204 * On Windows, mandatory locking is used.
1205-@@ -1456,8 +1456,9 @@
1206+@@ -1465,8 +1465,9 @@
1207 attributes = 0;
1208 ut_error;
1209 }
1210@@ -1051,7 +1071,7 @@
1211 GENERIC_READ | GENERIC_WRITE, /* read and write
1212 access */
1213 share_mode, /* File can be read also by other
1214-@@ -1516,7 +1517,11 @@
1215+@@ -1531,7 +1532,11 @@
1216 if (create_mode == OS_FILE_OPEN || create_mode == OS_FILE_OPEN_RAW
1217 || create_mode == OS_FILE_OPEN_RETRY) {
1218 mode_str = "OPEN";
1219@@ -1064,7 +1084,7 @@
1220 } else if (create_mode == OS_FILE_CREATE) {
1221 mode_str = "CREATE";
1222 create_flag = O_RDWR | O_CREAT | O_EXCL;
1223-@@ -2694,6 +2699,9 @@
1224+@@ -2709,6 +2714,9 @@
1225 ut_a((offset & 0xFFFFFFFFUL) == offset);
1226 ut_a((n & 0xFFFFFFFFUL) == n);
1227
1228@@ -1074,7 +1094,7 @@
1229 os_n_file_writes++;
1230
1231 ut_ad(file);
1232-@@ -2818,6 +2826,9 @@
1233+@@ -2833,6 +2841,9 @@
1234 #else
1235 ssize_t ret;
1236
1237@@ -1105,7 +1125,7 @@
1238 }
1239 --- a/storage/innobase/srv/srv0srv.c
1240 +++ b/storage/innobase/srv/srv0srv.c
1241-@@ -399,6 +399,10 @@
1242+@@ -405,6 +405,10 @@
1243
1244 UNIV_INTERN ulong srv_replication_delay = 0;
1245
1246@@ -1116,7 +1136,7 @@
1247 /*-------------------------------------------*/
1248 UNIV_INTERN ulong srv_n_spin_wait_rounds = 30;
1249 UNIV_INTERN ulong srv_n_free_tickets_to_enter = 500;
1250-@@ -1078,7 +1082,7 @@
1251+@@ -1084,7 +1088,7 @@
1252 }
1253
1254 /* Initialize some INFORMATION SCHEMA internal structures */
1255@@ -1125,7 +1145,7 @@
1256 }
1257
1258 /*********************************************************************//**
1259-@@ -1089,6 +1093,7 @@
1260+@@ -1095,6 +1099,7 @@
1261 /*==========*/
1262 {
1263 os_fast_mutex_free(&srv_conc_mutex);
1264@@ -1133,7 +1153,7 @@
1265 mem_free(srv_conc_slots);
1266 srv_conc_slots = NULL;
1267
1268-@@ -1102,6 +1107,7 @@
1269+@@ -1108,6 +1113,7 @@
1270 srv_mysql_table = NULL;
1271
1272 trx_i_s_cache_free(trx_i_s_cache);
1273@@ -1141,7 +1161,7 @@
1274 }
1275
1276 /*********************************************************************//**
1277-@@ -1717,7 +1723,7 @@
1278+@@ -1727,7 +1733,7 @@
1279 }
1280
1281 /* Record the lock wait time for this thread */
1282@@ -1187,7 +1207,7 @@
1283 ulint
1284 open_or_create_data_files(
1285 /*======================*/
1286-@@ -1419,7 +1422,7 @@
1287+@@ -1448,7 +1451,7 @@
1288 }
1289 #endif /* UNIV_LOG_ARCHIVE */
1290
1291@@ -1196,7 +1216,7 @@
1292 ut_print_timestamp(stderr);
1293 fprintf(stderr,
1294 " InnoDB: Error: combined size of log files"
1295-@@ -1714,6 +1717,10 @@
1296+@@ -1743,6 +1746,10 @@
1297 are initialized in trx_sys_init_at_db_start(). */
1298
1299 recv_recovery_from_checkpoint_finish();
1300@@ -1207,7 +1227,7 @@
1301 if (srv_force_recovery < SRV_FORCE_NO_IBUF_MERGE) {
1302 /* The following call is necessary for the insert
1303 buffer to work with multiple tablespaces. We must
1304-@@ -1873,6 +1880,17 @@
1305+@@ -1902,6 +1909,17 @@
1306 if (!srv_auto_extend_last_data_file
1307 && sum_of_data_file_sizes != tablespace_size_in_header) {
1308
1309@@ -1225,7 +1245,7 @@
1310 ut_print_timestamp(stderr);
1311 fprintf(stderr,
1312 " InnoDB: Error: tablespace size"
1313-@@ -1952,6 +1970,7 @@
1314+@@ -1981,6 +1999,7 @@
1315
1316 return(DB_ERROR);
1317 }
1318@@ -1233,7 +1253,7 @@
1319 }
1320
1321 /* Check that os_fast_mutexes work as expected */
1322-@@ -1976,6 +1995,10 @@
1323+@@ -2005,6 +2024,10 @@
1324
1325 os_fast_mutex_free(&srv_os_test_mutex);
1326
1327@@ -1244,7 +1264,7 @@
1328 if (srv_print_verbose_log) {
1329 ut_print_timestamp(stderr);
1330 fprintf(stderr,
1331-@@ -2064,6 +2087,7 @@
1332+@@ -2093,6 +2116,7 @@
1333 ibuf_update_max_tablespace_id();
1334 }
1335
1336@@ -1268,7 +1288,7 @@
1337
1338 --- a/storage/innobase/trx/trx0sys.c
1339 +++ b/storage/innobase/trx/trx0sys.c
1340-@@ -1655,12 +1655,14 @@
1341+@@ -1656,12 +1656,14 @@
1342 mem_free(trx_doublewrite);
1343 trx_doublewrite = NULL;
1344
1345@@ -1283,7 +1303,7 @@
1346
1347 /* There can't be any active transactions. */
1348 rseg = UT_LIST_GET_FIRST(trx_sys->rseg_list);
1349-@@ -1686,10 +1688,12 @@
1350+@@ -1687,10 +1689,12 @@
1351 UT_LIST_REMOVE(view_list, trx_sys->view_list, prev_view);
1352 }
1353
1354@@ -1325,7 +1345,7 @@
1355 "InnoDB: Since"
1356 --- a/storage/innobase/trx/trx0purge.c
1357 +++ b/storage/innobase/trx/trx0purge.c
1358-@@ -1129,6 +1129,9 @@
1359+@@ -1141,6 +1141,9 @@
1360 que_thr_t* thr;
1361 ulint old_pages_handled;
1362
1363@@ -1335,19 +1355,6 @@
1364 ut_a(purge_sys->trx->n_active_thrs == 0);
1365
1366 rw_lock_x_lock(&purge_sys->latch);
1367---- a/configure.cmake
1368-+++ b/configure.cmake
1369-@@ -149,7 +149,9 @@
1370- SET(CMAKE_REQUIRED_LIBRARIES
1371- ${LIBM} ${LIBNSL} ${LIBBIND} ${LIBCRYPT} ${LIBSOCKET} ${LIBDL} ${CMAKE_THREAD_LIBS_INIT} ${LIBRT})
1372-
1373-- LIST(REMOVE_DUPLICATES CMAKE_REQUIRED_LIBRARIES)
1374-+ IF(CMAKE_REQUIRED_LIBRARIES)
1375-+ LIST(REMOVE_DUPLICATES CMAKE_REQUIRED_LIBRARIES)
1376-+ ENDIF()
1377- LINK_LIBRARIES(${CMAKE_THREAD_LIBS_INIT})
1378-
1379- OPTION(WITH_LIBWRAP "Compile with tcp wrappers support" OFF)
1380 --- a/storage/innobase/sync/sync0arr.c
1381 +++ b/storage/innobase/sync/sync0arr.c
1382 @@ -41,6 +41,7 @@
1383@@ -1358,9 +1365,9 @@
1384
1385 /*
1386 WAIT ARRAY
1387-@@ -928,6 +929,13 @@
1388- ibool fatal = FALSE;
1389- double longest_diff = 0;
1390+@@ -929,6 +930,13 @@
1391+ return(FALSE);
1392+ }
1393
1394 + if (srv_rebuild_indexes) {
1395 +
1396@@ -1374,7 +1381,7 @@
1397 extremely slowly. UNIV_DEBUG_VALGRIND does not necessary mean that
1398 --- a/storage/innobase/include/buf0buf.h
1399 +++ b/storage/innobase/include/buf0buf.h
1400-@@ -1441,6 +1441,8 @@
1401+@@ -1531,6 +1531,8 @@
1402 frees a page in buffer pool */
1403 # endif /* UNIV_DEBUG_FILE_ACCESSES || UNIV_DEBUG */
1404 #endif /* !UNIV_HOTBACKUP */
1405@@ -1385,7 +1392,7 @@
1406 /** The buffer control block structure */
1407 --- a/storage/innobase/mem/mem0dbg.c
1408 +++ b/storage/innobase/mem/mem0dbg.c
1409-@@ -280,18 +280,10 @@
1410+@@ -282,18 +282,10 @@
1411 byte* buf, /*!< in: pointer to buffer */
1412 ulint n) /*!< in: length of buffer */
1413 {
1414@@ -1406,7 +1413,7 @@
1415
1416 UNIV_MEM_INVALID(buf, n);
1417 }
1418-@@ -306,17 +298,10 @@
1419+@@ -308,17 +300,10 @@
1420 byte* buf, /*!< in: pointer to buffer */
1421 ulint n) /*!< in: length of buffer */
1422 {
1423
1424=== modified file 'patches/innodb56.patch'
1425--- patches/innodb56.patch 2013-07-10 12:29:40 +0000
1426+++ patches/innodb56.patch 2013-07-22 14:01:36 +0000
1427@@ -52,7 +52,7 @@
1428 block->in_unzip_LRU_list = FALSE;
1429 #endif /* UNIV_DEBUG */
1430 #if defined UNIV_AHI_DEBUG || defined UNIV_DEBUG
1431-@@ -3951,6 +3959,13 @@
1432+@@ -3963,6 +3971,13 @@
1433 frame = ((buf_block_t*) bpage)->frame;
1434 }
1435
1436@@ -66,7 +66,7 @@
1437 /* If this page is not uninitialized and not in the
1438 doublewrite buffer, then the page number and space id
1439 should be the same as in block. */
1440-@@ -4068,7 +4083,7 @@
1441+@@ -4080,7 +4095,7 @@
1442 recv_recover_page(TRUE, (buf_block_t*) bpage);
1443 }
1444
1445@@ -177,7 +177,7 @@
1446 }
1447
1448 /**********************************************************************//**
1449-@@ -1491,7 +1510,12 @@
1450+@@ -1506,7 +1525,12 @@
1451 the file yet; the following calls will open it and update the
1452 size fields */
1453
1454@@ -191,7 +191,7 @@
1455 fil_node_complete_io(node, fil_system, OS_FILE_READ);
1456 }
1457
1458-@@ -2095,7 +2119,7 @@
1459+@@ -2109,7 +2133,7 @@
1460 mem_free(path);
1461 }
1462
1463@@ -200,7 +200,7 @@
1464 /********************************************************//**
1465 Writes a log record about an .ibd file create/rename/delete. */
1466 static
1467-@@ -2329,7 +2353,7 @@
1468+@@ -2343,7 +2367,7 @@
1469 space_id, name, path, flags,
1470 DICT_TF2_USE_TABLESPACE,
1471 FIL_IBD_FILE_INITIAL_SIZE) != DB_SUCCESS) {
1472@@ -209,7 +209,7 @@
1473 }
1474 }
1475
1476-@@ -2687,7 +2711,7 @@
1477+@@ -2702,7 +2726,7 @@
1478 }
1479
1480 if (err == DB_SUCCESS) {
1481@@ -218,7 +218,7 @@
1482 /* Write a log record about the deletion of the .ibd
1483 file, so that ibbackup can replay it in the
1484 --apply-log phase. We use a dummy mtr and the familiar
1485-@@ -3042,7 +3066,7 @@
1486+@@ -3057,7 +3081,7 @@
1487
1488 mutex_exit(&fil_system->mutex);
1489
1490@@ -227,7 +227,7 @@
1491 if (success && !recv_recovery_on) {
1492 mtr_t mtr;
1493
1494-@@ -3426,7 +3450,7 @@
1495+@@ -3441,7 +3465,7 @@
1496 goto error_exit_1;
1497 }
1498
1499@@ -236,7 +236,7 @@
1500 {
1501 mtr_t mtr;
1502 ulint mlog_file_flag = 0;
1503-@@ -3504,6 +3528,97 @@
1504+@@ -3519,6 +3543,97 @@
1505 #endif /* UNIV_LOG_ARCHIVE */
1506 };
1507
1508@@ -334,7 +334,7 @@
1509 /********************************************************************//**
1510 Tries to open a single-table tablespace and optionally checks that the
1511 space id in it is correct. If this does not succeed, print an error message
1512-@@ -3569,6 +3684,9 @@
1513+@@ -3584,6 +3699,9 @@
1514 in the default location. If it is remote, it should not be here. */
1515 def.filepath = fil_make_ibd_name(tablename, false);
1516
1517@@ -344,7 +344,7 @@
1518 /* The path_in was read from SYS_DATAFILES. */
1519 if (path_in) {
1520 if (strcmp(def.filepath, path_in)) {
1521-@@ -3616,6 +3734,7 @@
1522+@@ -3631,6 +3749,7 @@
1523 tablespaces_found++;
1524 }
1525 }
1526@@ -352,7 +352,7 @@
1527
1528 /* Always look for a file at the default location. */
1529 ut_a(def.filepath);
1530-@@ -3712,11 +3831,15 @@
1531+@@ -3727,11 +3846,15 @@
1532 /* The following call prints an error message */
1533 os_file_get_last_error(true);
1534
1535@@ -369,7 +369,7 @@
1536
1537 err = DB_CORRUPTION;
1538
1539-@@ -3974,6 +4097,9 @@
1540+@@ -3989,6 +4112,9 @@
1541 #ifdef UNIV_HOTBACKUP
1542 fil_space_t* space;
1543 #endif
1544@@ -379,7 +379,7 @@
1545
1546 memset(&def, 0, sizeof(def));
1547 memset(&remote, 0, sizeof(remote));
1548-@@ -4018,6 +4144,9 @@
1549+@@ -4033,6 +4159,9 @@
1550 # endif /* !UNIV_HOTBACKUP */
1551 #endif
1552
1553@@ -389,7 +389,7 @@
1554 /* Check for a link file which locates a remote tablespace. */
1555 remote.success = fil_open_linked_file(
1556 tablename, &remote.filepath, &remote.file);
1557-@@ -4028,8 +4157,20 @@
1558+@@ -4043,8 +4172,20 @@
1559 if (!remote.success) {
1560 os_file_close(remote.file);
1561 mem_free(remote.filepath);
1562@@ -410,7 +410,7 @@
1563
1564
1565 /* Try to open the tablespace in the datadir. */
1566-@@ -4042,6 +4183,17 @@
1567+@@ -4057,6 +4198,17 @@
1568 fil_validate_single_table_tablespace(tablename, &def);
1569 if (!def.success) {
1570 os_file_close(def.file);
1571@@ -428,7 +428,7 @@
1572 }
1573 }
1574
1575-@@ -4114,7 +4266,7 @@
1576+@@ -4129,7 +4281,7 @@
1577 /* At this point, only one tablespace is open */
1578 ut_a(def.success == !remote.success);
1579
1580@@ -437,7 +437,7 @@
1581
1582 /* Get and test the file size. */
1583 size = os_file_get_size(fsp->file);
1584-@@ -4133,9 +4285,9 @@
1585+@@ -4148,9 +4300,9 @@
1586
1587 /* Every .ibd file is created >= 4 pages in size. Smaller files
1588 cannot be ok. */
1589@@ -449,7 +449,7 @@
1590 ib_logf(IB_LOG_LEVEL_ERROR,
1591 "The size of single-table tablespace file %s "
1592 "is only " UINT64PF ", should be at least %lu!",
1593-@@ -4216,7 +4368,7 @@
1594+@@ -4231,7 +4383,7 @@
1595 }
1596 mutex_exit(&fil_system->mutex);
1597 #endif /* UNIV_HOTBACKUP */
1598@@ -458,7 +458,7 @@
1599 tablename, fsp->id, fsp->flags, FIL_TABLESPACE);
1600
1601 if (!file_space_create_success) {
1602-@@ -4243,13 +4395,55 @@
1603+@@ -4258,13 +4410,55 @@
1604 }
1605
1606 func_exit:
1607@@ -518,7 +518,7 @@
1608 mem_free(tablename);
1609 if (remote.success) {
1610 mem_free(remote.filepath);
1611-@@ -4263,7 +4457,7 @@
1612+@@ -4278,7 +4472,7 @@
1613 idea is to read as much good data as we can and jump over bad data.
1614 @return 0 if ok, -1 if error even after the retries, 1 if at the end
1615 of the directory */
1616@@ -527,7 +527,7 @@
1617 int
1618 fil_file_readdir_next_file(
1619 /*=======================*/
1620-@@ -4303,7 +4497,7 @@
1621+@@ -4318,7 +4512,7 @@
1622 @return DB_SUCCESS or error number */
1623 UNIV_INTERN
1624 dberr_t
1625@@ -536,7 +536,7 @@
1626 /*===================================*/
1627 {
1628 int ret;
1629-@@ -4359,7 +4553,9 @@
1630+@@ -4374,7 +4568,9 @@
1631 "%s/%s", fil_path_to_mysql_datadir, dbinfo.name);
1632 srv_normalize_path_for_win(dbpath);
1633
1634@@ -547,7 +547,7 @@
1635
1636 if (dbdir != NULL) {
1637
1638-@@ -4380,9 +4576,15 @@
1639+@@ -4395,9 +4591,15 @@
1640 && (0 == strcmp(fileinfo.name
1641 + strlen(fileinfo.name) - 4,
1642 ".ibd")
1643@@ -565,7 +565,7 @@
1644 /* The name ends in .ibd or .isl;
1645 try opening the file */
1646 fil_load_single_table_tablespace(
1647-@@ -4538,6 +4740,7 @@
1648+@@ -4553,6 +4755,7 @@
1649 {
1650 fil_space_t* fnamespace;
1651 fil_space_t* space;
1652@@ -573,7 +573,7 @@
1653
1654 ut_ad(fil_system);
1655
1656-@@ -4615,6 +4818,10 @@
1657+@@ -4630,6 +4833,10 @@
1658 if (fnamespace == NULL) {
1659 if (print_error_if_does_not_exist) {
1660 fil_report_missing_tablespace(name, id);
1661@@ -584,7 +584,7 @@
1662 }
1663 } else {
1664 ut_print_timestamp(stderr);
1665-@@ -4638,6 +4845,10 @@
1666+@@ -4653,6 +4860,10 @@
1667
1668 mutex_exit(&fil_system->mutex);
1669
1670@@ -595,7 +595,7 @@
1671 return(FALSE);
1672 }
1673
1674-@@ -4728,6 +4939,7 @@
1675+@@ -4743,6 +4954,7 @@
1676 ulint page_size;
1677 ulint pages_added;
1678 ibool success;
1679@@ -603,7 +603,7 @@
1680
1681 ut_ad(!srv_read_only_mode);
1682
1683-@@ -4772,13 +4984,17 @@
1684+@@ -4787,13 +4999,17 @@
1685 goto retry;
1686 }
1687
1688@@ -622,7 +622,7 @@
1689 start_page_no = space->size;
1690 file_start_page_no = space->size - node->size;
1691
1692-@@ -5024,7 +5240,7 @@
1693+@@ -5039,7 +5255,7 @@
1694 off the LRU list if it is in the LRU list. The caller must hold the fil_sys
1695 mutex. */
1696 static
1697@@ -631,7 +631,7 @@
1698 fil_node_prepare_for_io(
1699 /*====================*/
1700 fil_node_t* node, /*!< in: file node */
1701-@@ -5044,9 +5260,12 @@
1702+@@ -5059,9 +5275,12 @@
1703 }
1704
1705 if (node->open == FALSE) {
1706@@ -645,7 +645,7 @@
1707 }
1708
1709 if (node->n_pending == 0 && fil_space_belongs_in_lru(space)) {
1710-@@ -5058,6 +5277,8 @@
1711+@@ -5073,6 +5292,8 @@
1712 }
1713
1714 node->n_pending++;
1715@@ -654,7 +654,7 @@
1716 }
1717
1718 /********************************************************************//**
1719-@@ -5259,6 +5480,16 @@
1720+@@ -5274,6 +5495,16 @@
1721
1722 ut_ad(mode != OS_AIO_IBUF || space->purpose == FIL_TABLESPACE);
1723
1724@@ -671,7 +671,7 @@
1725 node = UT_LIST_GET_FIRST(space->chain);
1726
1727 for (;;) {
1728-@@ -5290,7 +5521,11 @@
1729+@@ -5305,7 +5536,11 @@
1730 }
1731
1732 /* Open file if closed */
1733@@ -684,7 +684,7 @@
1734
1735 /* Check that at least the start offset is within the bounds of a
1736 single-table tablespace, including rollback tablespaces. */
1737-@@ -6164,6 +6399,7 @@
1738+@@ -6179,6 +6414,7 @@
1739 return(err);
1740 }
1741
1742@@ -692,11 +692,10 @@
1743 /****************************************************************//**
1744 Generate redo logs for swapping two .ibd files */
1745 UNIV_INTERN
1746-@@ -6187,4 +6423,4 @@
1747- 0, 0, new_name, old_name, &mtr);
1748- mtr_commit(&mtr);
1749+@@ -6205,3 +6441,4 @@
1750+ 0, 0, new_name, old_name, mtr);
1751+ }
1752 }
1753--
1754 +#endif
1755 --- a/storage/innobase/handler/ha_innodb.cc
1756 +++ b/storage/innobase/handler/ha_innodb.cc
1757@@ -709,7 +708,7 @@
1758
1759 if (thd && thd_sql_command(thd) == SQLCOM_DROP_TABLE) {
1760
1761-@@ -2423,43 +2423,19 @@
1762+@@ -2459,43 +2459,19 @@
1763 ulint buflen, /*!< in: length of buf, in bytes */
1764 const char* id, /*!< in: identifier to convert */
1765 ulint idlen, /*!< in: length of id, in bytes */
1766@@ -758,7 +757,7 @@
1767 if (UNIV_UNLIKELY(idlen > buflen)) {
1768 idlen = buflen;
1769 }
1770-@@ -16642,45 +16618,24 @@
1771+@@ -16774,45 +16750,24 @@
1772 void
1773 ib_logf(
1774 /*====*/
1775@@ -1022,20 +1021,6 @@
1776 #if defined(UNIV_HOTBACKUP) || defined(__WIN__)
1777 /* InnoDB Hot Backup does not lock the data files.
1778 * On Windows, mandatory locking is used.
1779---- a/storage/innobase/row/row0merge.cc
1780-+++ b/storage/innobase/row/row0merge.cc
1781-@@ -3227,9 +3227,11 @@
1782- goto err_exit;
1783- }
1784-
1785-+#if 0
1786- /* Generate the redo logs for file operations */
1787- fil_mtr_rename_log(old_table->space, old_name,
1788- new_table->space, new_table->name, tmp_name);
1789-+#endif
1790-
1791- /* What if the redo logs are flushed to disk here? This is
1792- tested with following crash point */
1793 --- a/storage/innobase/srv/srv0srv.cc
1794 +++ b/storage/innobase/srv/srv0srv.cc
1795 @@ -349,6 +349,10 @@
1796@@ -1049,7 +1034,7 @@
1797 /*-------------------------------------------*/
1798 UNIV_INTERN ulong srv_n_spin_wait_rounds = 30;
1799 UNIV_INTERN ulong srv_spin_wait_delay = 6;
1800-@@ -1808,7 +1812,8 @@
1801+@@ -1814,7 +1818,8 @@
1802 if (ret == SRV_NONE
1803 && srv_shutdown_state != SRV_SHUTDOWN_NONE
1804 && trx_purge_state() != PURGE_STATE_DISABLED
1805@@ -1173,7 +1158,7 @@
1806 if (srv_force_recovery < SRV_FORCE_NO_IBUF_MERGE) {
1807 /* The following call is necessary for the insert
1808 buffer to work with multiple tablespaces. We must
1809-@@ -2439,7 +2452,9 @@
1810+@@ -2439,7 +2453,9 @@
1811 value. Important to note that we can do it ONLY after
1812 we have finished the recovery process so that the
1813 image of TRX_SYS_PAGE_NO is not stale. */
1814@@ -1184,7 +1169,7 @@
1815 }
1816
1817 if (!create_new_db && sum_of_new_sizes > 0) {
1818-@@ -2647,6 +2662,7 @@
1819+@@ -2647,6 +2663,7 @@
1820 && srv_auto_extend_last_data_file
1821 && sum_of_data_file_sizes < tablespace_size_in_header) {
1822
1823@@ -1192,7 +1177,7 @@
1824 ut_print_timestamp(stderr);
1825 fprintf(stderr,
1826 " InnoDB: Error: tablespace size stored in header"
1827-@@ -2683,6 +2698,7 @@
1828+@@ -2683,6 +2700,7 @@
1829
1830 return(DB_ERROR);
1831 }
1832@@ -1200,7 +1185,7 @@
1833 }
1834
1835 /* Check that os_fast_mutexes work as expected */
1836-@@ -2707,6 +2723,10 @@
1837+@@ -2707,6 +2725,10 @@
1838
1839 os_fast_mutex_free(&srv_os_test_mutex);
1840
1841@@ -1211,7 +1196,7 @@
1842 if (srv_print_verbose_log) {
1843 ib_logf(IB_LOG_LEVEL_INFO,
1844 "%s started; log sequence number " LSN_PF "",
1845-@@ -2739,6 +2759,7 @@
1846+@@ -2739,6 +2761,7 @@
1847 fts_optimize_init();
1848 }
1849
1850@@ -1219,7 +1204,7 @@
1851 srv_was_started = TRUE;
1852
1853 return(DB_SUCCESS);
1854-@@ -2794,7 +2815,7 @@
1855+@@ -2794,7 +2817,7 @@
1856 return(DB_SUCCESS);
1857 }
1858
1859@@ -1285,7 +1270,7 @@
1860 total_trx = UT_LIST_GET_LEN(trx_sys->rw_trx_list)
1861 --- a/storage/innobase/trx/trx0trx.cc
1862 +++ b/storage/innobase/trx/trx0trx.cc
1863-@@ -2053,7 +2053,8 @@
1864+@@ -2084,7 +2084,8 @@
1865 scenario where some undo generated by a transaction,
1866 has XA stuff, and other undo, generated by the same
1867 transaction, doesn't. */
1868@@ -1320,7 +1305,7 @@
1869
1870 /*
1871 WAIT ARRAY
1872-@@ -903,6 +904,13 @@
1873+@@ -899,6 +900,13 @@
1874 ibool fatal = FALSE;
1875 double longest_diff = 0;
1876
1877@@ -1336,7 +1321,7 @@
1878 return(FALSE);
1879 --- a/storage/innobase/include/buf0buf.h
1880 +++ b/storage/innobase/include/buf0buf.h
1881-@@ -1583,6 +1583,8 @@
1882+@@ -1580,6 +1580,8 @@
1883 or buf_block_t::mutex. */
1884 # endif /* UNIV_DEBUG_FILE_ACCESSES || UNIV_DEBUG */
1885 #endif /* !UNIV_HOTBACKUP */
1886@@ -1390,7 +1375,7 @@
1887 }
1888 --- a/storage/innobase/include/srv0start.h
1889 +++ b/storage/innobase/include/srv0start.h
1890-@@ -77,6 +77,23 @@
1891+@@ -68,6 +68,23 @@
1892 srv_add_path_separator_if_needed(
1893 /*=============================*/
1894 char* str); /*!< in: null-terminated character string */
1895@@ -1416,7 +1401,7 @@
1896 Starts Innobase and creates a new database if database files
1897 --- a/configure.cmake
1898 +++ b/configure.cmake
1899-@@ -965,14 +965,6 @@
1900+@@ -979,14 +979,6 @@
1901 ENDIF()
1902
1903 CHECK_C_SOURCE_COMPILES("
1904@@ -1448,3 +1433,21 @@
1905 return tmp && tmp->init ? &tmp->dbug : 0;
1906 }
1907 #endif /* DBUG_OFF */
1908+--- a/storage/innobase/handler/handler0alter.cc
1909++++ b/storage/innobase/handler/handler0alter.cc
1910+@@ -5434,6 +5434,7 @@
1911+ = static_cast<ha_innobase_inplace_ctx*>(*pctx);
1912+
1913+ DBUG_ASSERT(ctx->need_rebuild());
1914++#if 0
1915+ /* Generate the redo log for the file
1916+ operations that will be performed in
1917+ commit_cache_rebuild(). */
1918+@@ -5442,6 +5443,7 @@
1919+ ctx->new_table->space,
1920+ ctx->new_table->name,
1921+ ctx->tmp_name, &mtr);
1922++#endif
1923+ DBUG_INJECT_CRASH("ib_commit_inplace_crash",
1924+ crash_inject_count++);
1925+ }
1926
1927=== modified file 'patches/xtradb51.patch'
1928--- patches/xtradb51.patch 2013-05-07 13:32:47 +0000
1929+++ patches/xtradb51.patch 2013-07-22 14:01:36 +0000
1930@@ -1,6 +1,6 @@
1931 --- a/storage/innodb_plugin/btr/btr0btr.c
1932 +++ b/storage/innodb_plugin/btr/btr0btr.c
1933-@@ -674,7 +674,7 @@
1934+@@ -680,7 +680,7 @@
1935 /**************************************************************//**
1936 Gets the root node of a tree and x-latches it.
1937 @return root page, x-latched */
1938@@ -9,7 +9,7 @@
1939 buf_block_t*
1940 btr_root_block_get(
1941 /*===============*/
1942-@@ -1134,7 +1134,7 @@
1943+@@ -1198,7 +1198,7 @@
1944 /************************************************************//**
1945 Returns the child page of a node pointer and x-latches it.
1946 @return child page, x-latched */
1947@@ -28,7 +28,7 @@
1948
1949 /* prototypes for new functions added to ha_innodb.cc */
1950 trx_t* innobase_get_trx();
1951-@@ -412,7 +413,7 @@
1952+@@ -414,7 +415,7 @@
1953 return(TRUE);
1954 }
1955
1956@@ -37,7 +37,7 @@
1957 if (recv_lsn_checks_on) {
1958 ib_uint64_t current_lsn;
1959
1960-@@ -744,6 +745,7 @@
1961+@@ -746,6 +747,7 @@
1962 block->page.zip_list.prev = NULL;
1963 block->page.zip_list.next = NULL;
1964 block->page.in_LRU_list = FALSE;
1965@@ -45,7 +45,7 @@
1966 block->in_unzip_LRU_list = FALSE;
1967 #if defined UNIV_AHI_DEBUG || defined UNIV_DEBUG
1968 block->n_pointers = 0;
1969-@@ -1888,7 +1890,9 @@
1970+@@ -1787,7 +1789,9 @@
1971 ut_ad(zip_size == fil_space_get_zip_size(space));
1972 ut_ad(ut_is_2pow(zip_size));
1973 #ifndef UNIV_LOG_DEBUG
1974@@ -56,7 +56,7 @@
1975 #endif
1976 if (innobase_get_slow_log()) {
1977 trx = innobase_get_trx();
1978-@@ -3190,6 +3194,13 @@
1979+@@ -3096,6 +3100,13 @@
1980 frame = ((buf_block_t*) bpage)->frame;
1981 }
1982
1983@@ -70,7 +70,7 @@
1984 /* If this page is not uninitialized and not in the
1985 doublewrite buffer, then the page number and space id
1986 should be the same as in block. */
1987-@@ -3302,7 +3313,8 @@
1988+@@ -3208,7 +3219,8 @@
1989 recv_recover_page(TRUE, (buf_block_t*) bpage);
1990 }
1991
1992@@ -82,7 +82,7 @@
1993 (srv_pass_corrupt_table && bpage->is_corrupt) ? NULL :
1994 --- a/storage/innodb_plugin/fil/fil0fil.c
1995 +++ b/storage/innodb_plugin/fil/fil0fil.c
1996-@@ -297,7 +297,7 @@
1997+@@ -298,7 +298,7 @@
1998
1999 /** The tablespace memory cache. This variable is NULL before the module is
2000 initialized. */
2001@@ -91,7 +91,7 @@
2002
2003
2004 /********************************************************************//**
2005-@@ -308,7 +308,7 @@
2006+@@ -309,7 +309,7 @@
2007 off the LRU list if it is in the LRU list. The caller must hold the fil_sys
2008 mutex. */
2009 static
2010@@ -100,7 +100,7 @@
2011 fil_node_prepare_for_io(
2012 /*====================*/
2013 fil_node_t* node, /*!< in: file node */
2014-@@ -633,7 +633,7 @@
2015+@@ -634,7 +634,7 @@
2016 Opens a the file of a node of a tablespace. The caller must own the fil_system
2017 mutex. */
2018 static
2019@@ -109,7 +109,7 @@
2020 fil_node_open_file(
2021 /*===============*/
2022 fil_node_t* node, /*!< in: file node */
2023-@@ -667,7 +667,17 @@
2024+@@ -668,7 +668,17 @@
2025 node->name, OS_FILE_OPEN, OS_FILE_READ_ONLY, &success);
2026 if (!success) {
2027 /* The following call prints an error message */
2028@@ -128,7 +128,7 @@
2029
2030 ut_print_timestamp(stderr);
2031
2032-@@ -728,12 +738,14 @@
2033+@@ -729,12 +739,14 @@
2034
2035 if (UNIV_UNLIKELY(space_id != space->id)) {
2036 fprintf(stderr,
2037@@ -147,7 +147,7 @@
2038 }
2039
2040 if (UNIV_UNLIKELY(space_id == ULINT_UNDEFINED
2041-@@ -757,8 +769,8 @@
2042+@@ -758,8 +770,8 @@
2043 }
2044
2045 if (size_bytes >= 1024 * 1024) {
2046@@ -158,7 +158,7 @@
2047 }
2048
2049 if (!(flags & DICT_TF_ZSSIZE_MASK)) {
2050-@@ -803,6 +815,8 @@
2051+@@ -804,6 +816,8 @@
2052 /* Put the node to the LRU list */
2053 UT_LIST_ADD_FIRST(LRU, system->LRU, node);
2054 }
2055@@ -167,7 +167,7 @@
2056 }
2057
2058 /**********************************************************************//**
2059-@@ -1431,7 +1445,12 @@
2060+@@ -1451,7 +1465,12 @@
2061 the file yet; the following calls will open it and update the
2062 size fields */
2063
2064@@ -181,7 +181,7 @@
2065 fil_node_complete_io(node, fil_system, OS_FILE_READ);
2066 }
2067
2068-@@ -1483,7 +1502,12 @@
2069+@@ -1503,7 +1522,12 @@
2070 the file yet; the following calls will open it and update the
2071 size fields */
2072
2073@@ -195,7 +195,7 @@
2074 fil_node_complete_io(node, fil_system, OS_FILE_READ);
2075 }
2076
2077-@@ -1933,7 +1957,7 @@
2078+@@ -1952,7 +1976,7 @@
2079 mem_free(path);
2080 }
2081
2082@@ -204,7 +204,7 @@
2083 /********************************************************//**
2084 Writes a log record about an .ibd file create/rename/delete. */
2085 static
2086-@@ -2157,7 +2181,7 @@
2087+@@ -2176,7 +2200,7 @@
2088 if (fil_create_new_single_table_tablespace(
2089 space_id, name, FALSE, flags,
2090 FIL_IBD_FILE_INITIAL_SIZE) != DB_SUCCESS) {
2091@@ -213,7 +213,7 @@
2092 }
2093 }
2094
2095-@@ -2324,7 +2348,7 @@
2096+@@ -2344,7 +2368,7 @@
2097 }
2098
2099 if (success) {
2100@@ -222,16 +222,16 @@
2101 /* Write a log record about the deletion of the .ibd
2102 file, so that ibbackup can replay it in the
2103 --apply-log phase. We use a dummy mtr and the familiar
2104-@@ -2625,7 +2649,7 @@
2105+@@ -2645,7 +2669,7 @@
2106
2107 mutex_exit(&fil_system->mutex);
2108
2109 -#ifndef UNIV_HOTBACKUP
2110 +#ifdef UNDEFINED
2111- if (success) {
2112+ if (success && !recv_recovery_on) {
2113 mtr_t mtr;
2114
2115-@@ -2815,7 +2839,7 @@
2116+@@ -2835,7 +2859,7 @@
2117
2118 fil_node_create(path, size, space_id, FALSE);
2119
2120@@ -240,7 +240,7 @@
2121 {
2122 mtr_t mtr;
2123
2124-@@ -3159,19 +3183,97 @@
2125+@@ -3182,19 +3206,97 @@
2126 "InnoDB: open the tablespace file ", stderr);
2127 ut_print_filename(stderr, filepath);
2128 fputs("!\n"
2129@@ -348,7 +348,7 @@
2130 return(FALSE);
2131 }
2132
2133-@@ -4106,7 +4208,7 @@
2134+@@ -4138,7 +4240,7 @@
2135 cannot be ok. */
2136
2137 size = (((ib_int64_t)size_high) << 32) + (ib_int64_t)size_low;
2138@@ -357,7 +357,7 @@
2139 if (size < FIL_IBD_FILE_INITIAL_SIZE * (lint)UNIV_PAGE_SIZE) {
2140 fprintf(stderr,
2141 "InnoDB: Error: the size of single-table tablespace"
2142-@@ -4236,7 +4338,51 @@
2143+@@ -4268,7 +4370,51 @@
2144
2145 fil_node_create(filepath, 0, space_id, FALSE);
2146 func_exit:
2147@@ -410,7 +410,7 @@
2148 ut_free(buf2);
2149 mem_free(filepath);
2150 }
2151-@@ -4247,7 +4393,7 @@
2152+@@ -4279,7 +4425,7 @@
2153 idea is to read as much good data as we can and jump over bad data.
2154 @return 0 if ok, -1 if error even after the retries, 1 if at the end
2155 of the directory */
2156@@ -419,7 +419,7 @@
2157 int
2158 fil_file_readdir_next_file(
2159 /*=======================*/
2160-@@ -4291,7 +4437,7 @@
2161+@@ -4323,7 +4469,7 @@
2162 @return DB_SUCCESS or error number */
2163 UNIV_INTERN
2164 ulint
2165@@ -428,7 +428,7 @@
2166 /*===================================*/
2167 {
2168 int ret;
2169-@@ -4347,7 +4493,9 @@
2170+@@ -4379,7 +4525,9 @@
2171 dbinfo.name);
2172 srv_normalize_path_for_win(dbpath);
2173
2174@@ -439,7 +439,7 @@
2175
2176 if (dbdir != NULL) {
2177 /* printf("Opened dir %s\n", dbinfo.name); */
2178-@@ -4373,8 +4521,11 @@
2179+@@ -4405,8 +4553,11 @@
2180 ".ibd")) {
2181 /* The name ends in .ibd; try opening
2182 the file */
2183@@ -452,7 +452,7 @@
2184 }
2185 next_file_item:
2186 ret = fil_file_readdir_next_file(&err,
2187-@@ -4546,15 +4697,97 @@
2188+@@ -4578,15 +4729,97 @@
2189 "InnoDB: in InnoDB data dictionary"
2190 " has tablespace id %lu,\n"
2191 "InnoDB: but tablespace with that id"
2192@@ -558,7 +558,7 @@
2193 } else {
2194 ut_print_timestamp(stderr);
2195 fputs(" InnoDB: Error: table ", stderr);
2196-@@ -4953,7 +5186,7 @@
2197+@@ -4985,7 +5218,7 @@
2198 off the LRU list if it is in the LRU list. The caller must hold the fil_sys
2199 mutex. */
2200 static
2201@@ -567,7 +567,7 @@
2202 fil_node_prepare_for_io(
2203 /*====================*/
2204 fil_node_t* node, /*!< in: file node */
2205-@@ -4973,10 +5206,13 @@
2206+@@ -5005,10 +5238,13 @@
2207 }
2208
2209 if (node->open == FALSE) {
2210@@ -582,7 +582,7 @@
2211 }
2212
2213 if (node->n_pending == 0 && space->purpose == FIL_TABLESPACE
2214-@@ -4989,6 +5225,8 @@
2215+@@ -5021,6 +5257,8 @@
2216 }
2217
2218 node->n_pending++;
2219@@ -591,7 +591,7 @@
2220 }
2221
2222 /********************************************************************//**
2223-@@ -5125,7 +5363,9 @@
2224+@@ -5157,7 +5395,9 @@
2225 ut_ad(recv_no_ibuf_operations || (type == OS_FILE_WRITE)
2226 || !ibuf_bitmap_page(zip_size, block_offset)
2227 || sync || is_log);
2228@@ -602,7 +602,7 @@
2229 || ibuf_page(space_id, zip_size, block_offset, NULL));
2230 # endif /* UNIV_LOG_DEBUG */
2231 if (sync) {
2232-@@ -5190,6 +5430,16 @@
2233+@@ -5222,6 +5462,16 @@
2234
2235 ut_ad((mode != OS_AIO_IBUF) || (space->purpose == FIL_TABLESPACE));
2236
2237@@ -619,9 +619,24 @@
2238 node = UT_LIST_GET_FIRST(space->chain);
2239
2240 for (;;) {
2241+@@ -5924,6 +6174,7 @@
2242+ const char* tmp_name) /*!< in: temp table name used while
2243+ swapping */
2244+ {
2245++#ifdef UNDEFINED
2246+ mtr_t mtr;
2247+ mtr_start(&mtr);
2248+ fil_op_write_log(MLOG_FILE_RENAME, old_space_id,
2249+@@ -5931,5 +6182,6 @@
2250+ fil_op_write_log(MLOG_FILE_RENAME, new_space_id,
2251+ 0, 0, new_name, old_name, &mtr);
2252+ mtr_commit(&mtr);
2253++#endif
2254+ }
2255+
2256 --- a/storage/innodb_plugin/handler/ha_innodb.cc
2257 +++ b/storage/innodb_plugin/handler/ha_innodb.cc
2258-@@ -365,12 +365,6 @@
2259+@@ -388,12 +388,6 @@
2260 "Timeout in seconds an InnoDB transaction may wait for a lock before being rolled back. Values above 100000000 disable the timeout.",
2261 NULL, NULL, 50, 1, 1024 * 1024 * 1024, 0);
2262
2263@@ -634,7 +649,7 @@
2264 static MYSQL_THDVAR_BOOL(fake_changes, PLUGIN_VAR_OPCMDARG,
2265 "In the transaction after enabled, UPDATE, INSERT and DELETE only move the cursor to the records "
2266 "and do nothing other operations (no changes, no ibuf, no undo, no transaction log) in the transaction. "
2267-@@ -760,17 +754,6 @@
2268+@@ -789,17 +783,6 @@
2269 }
2270
2271 /******************************************************************//**
2272@@ -652,7 +667,7 @@
2273 Returns true if expand_fast_index_creation is enabled for the current
2274 session.
2275 @return the value of the server's expand_fast_index_creation variable */
2276-@@ -11966,7 +11949,6 @@
2277+@@ -12551,7 +12534,6 @@
2278 MYSQL_SYSVAR(flush_neighbor_pages),
2279 MYSQL_SYSVAR(read_ahead),
2280 MYSQL_SYSVAR(adaptive_checkpoint),
2281@@ -672,7 +687,7 @@
2282 if (ibuf_fixed_addr_page(space, zip_size, page_no)) {
2283
2284 return(TRUE);
2285-@@ -2280,6 +2283,9 @@
2286+@@ -2278,6 +2281,9 @@
2287 ulint n_bytes;
2288 ulint n_pag2;
2289
2290@@ -684,7 +699,7 @@
2291
2292 --- a/storage/innodb_plugin/include/fil0fil.h
2293 +++ b/storage/innodb_plugin/include/fil0fil.h
2294-@@ -505,7 +505,7 @@
2295+@@ -509,7 +509,7 @@
2296 @return DB_SUCCESS or error number */
2297 UNIV_INTERN
2298 ulint
2299@@ -712,7 +727,7 @@
2300 @return the value of the server's innodb_expand_fast_index_creation variable */
2301 --- a/storage/innodb_plugin/include/log0recv.h
2302 +++ b/storage/innodb_plugin/include/log0recv.h
2303-@@ -444,7 +444,7 @@
2304+@@ -481,7 +481,7 @@
2305 /* defined in os0file.c */
2306 #define OS_AIO_MERGE_N_CONSECUTIVE 64
2307 /* defined in log0recv.c */
2308@@ -723,7 +738,7 @@
2309
2310 --- a/storage/innodb_plugin/include/mem0mem.ic
2311 +++ b/storage/innodb_plugin/include/mem0mem.ic
2312-@@ -367,7 +367,7 @@
2313+@@ -363,7 +363,7 @@
2314 buf = (byte*)buf + MEM_FIELD_HEADER_SIZE;
2315
2316 /* Check that the field lengths agree */
2317@@ -734,7 +749,7 @@
2318 return(buf);
2319 --- a/storage/innodb_plugin/include/srv0srv.h
2320 +++ b/storage/innodb_plugin/include/srv0srv.h
2321-@@ -226,6 +226,11 @@
2322+@@ -243,6 +243,11 @@
2323
2324 extern ulong srv_replication_delay;
2325
2326@@ -816,7 +831,7 @@
2327
2328 --- a/storage/innodb_plugin/log/log0log.c
2329 +++ b/storage/innodb_plugin/log/log0log.c
2330-@@ -1370,7 +1370,7 @@
2331+@@ -1449,7 +1449,7 @@
2332 #endif /* UNIV_DEBUG */
2333 ulint unlock;
2334
2335@@ -825,7 +840,7 @@
2336 /* Recovery is running and no operations on the log files are
2337 allowed yet (the variable name .._no_ibuf_.. is misleading) */
2338
2339-@@ -2032,7 +2032,7 @@
2340+@@ -2117,7 +2117,7 @@
2341 return(TRUE);
2342 }
2343
2344@@ -834,7 +849,7 @@
2345
2346 if (log_sys->n_pending_checkpoint_writes > 0) {
2347 /* A checkpoint write is running */
2348-@@ -3127,6 +3127,7 @@
2349+@@ -3232,6 +3232,7 @@
2350 shutdown, because the InnoDB layer may have committed or
2351 prepared transactions and we don't want to lose them. */
2352
2353@@ -842,7 +857,7 @@
2354 if (trx_n_mysql_transactions > 0
2355 || UT_LIST_GET_LEN(trx_sys->trx_list) > trx_n_prepared) {
2356
2357-@@ -3134,6 +3135,7 @@
2358+@@ -3239,6 +3240,7 @@
2359
2360 goto loop;
2361 }
2362@@ -885,7 +900,7 @@
2363
2364 /** The recovery system */
2365 UNIV_INTERN recv_sys_t* recv_sys = NULL;
2366-@@ -621,7 +622,7 @@
2367+@@ -622,7 +623,7 @@
2368 /***********************************************************************//**
2369 Checks the consistency of the checkpoint info
2370 @return TRUE if ok */
2371@@ -894,7 +909,7 @@
2372 ibool
2373 recv_check_cp_is_consistent(
2374 /*========================*/
2375-@@ -651,7 +652,7 @@
2376+@@ -652,7 +653,7 @@
2377 /********************************************************//**
2378 Looks for the maximum consistent checkpoint from the log groups.
2379 @return error code or DB_SUCCESS */
2380@@ -903,16 +918,7 @@
2381 ulint
2382 recv_find_max_checkpoint(
2383 /*=====================*/
2384-@@ -840,7 +841,7 @@
2385- InnoDB-3.23.52 where the checksum field contains the log block number.
2386- @return TRUE if ok, or if the log block may be in the format of InnoDB
2387- version predating 3.23.52 */
2388--static
2389-+//static
2390- ibool
2391- log_block_checksum_is_ok_or_old_format(
2392- /*===================================*/
2393-@@ -1629,9 +1630,16 @@
2394+@@ -1636,9 +1637,16 @@
2395 if (page_zip) {
2396 memset(FIL_PAGE_LSN + page_zip->data, 0, 8);
2397 }
2398@@ -930,7 +936,7 @@
2399
2400 ib_uint64_t end_lsn;
2401
2402-@@ -1830,6 +1838,18 @@
2403+@@ -1838,6 +1846,18 @@
2404 ulint zip_size = fil_space_get_zip_size(space);
2405 ulint page_no = recv_addr->page_no;
2406
2407@@ -949,7 +955,7 @@
2408 if (recv_addr->state == RECV_NOT_PROCESSED) {
2409 if (!has_printed) {
2410 ut_print_timestamp(stderr);
2411-@@ -1863,7 +1883,7 @@
2412+@@ -1871,7 +1891,7 @@
2413
2414 mutex_enter(&(recv_sys->mutex));
2415 }
2416@@ -958,7 +964,7 @@
2417 recv_addr = HASH_GET_NEXT(addr_hash, recv_addr);
2418 }
2419
2420-@@ -2369,7 +2389,7 @@
2421+@@ -2377,7 +2397,7 @@
2422 || type == MLOG_FILE_RENAME
2423 || type == MLOG_FILE_DELETE) {
2424 ut_a(space);
2425@@ -967,7 +973,7 @@
2426 if (recv_replay_file_ops) {
2427
2428 /* In ibbackup --apply-log, replay an .ibd file
2429-@@ -2392,7 +2412,7 @@
2430+@@ -2400,7 +2420,7 @@
2431 ut_error;
2432 }
2433 }
2434@@ -976,7 +982,7 @@
2435 /* In normal mysqld crash recovery we do not try to
2436 replay file operations */
2437 #ifdef UNIV_LOG_LSN_DEBUG
2438-@@ -2809,8 +2829,11 @@
2439+@@ -2817,8 +2837,11 @@
2440
2441 fprintf(stderr,
2442 "InnoDB: Doing recovery: scanned up to"
2443@@ -990,7 +996,7 @@
2444 }
2445 }
2446
2447-@@ -2913,7 +2936,7 @@
2448+@@ -2921,7 +2944,7 @@
2449 "InnoDB: Reading tablespace information"
2450 " from the .ibd files...\n");
2451
2452@@ -999,7 +1005,7 @@
2453
2454 /* If we are using the doublewrite method, we will
2455 check if there are half-written pages in data files,
2456-@@ -2922,12 +2945,14 @@
2457+@@ -2930,12 +2953,14 @@
2458
2459 if (srv_force_recovery < SRV_FORCE_NO_LOG_REDO) {
2460
2461@@ -1015,7 +1021,7 @@
2462 }
2463 }
2464
2465-@@ -3095,6 +3120,7 @@
2466+@@ -3103,6 +3128,7 @@
2467 recv_sys->recovered_lsn = checkpoint_lsn;
2468
2469 srv_start_lsn = checkpoint_lsn;
2470@@ -1023,7 +1029,7 @@
2471 }
2472
2473 contiguous_lsn = ut_uint64_align_down(recv_sys->scanned_lsn,
2474-@@ -3453,6 +3479,7 @@
2475+@@ -3461,6 +3487,7 @@
2476 that the data dictionary tables will be free of any locks.
2477 The data dictionary latch should guarantee that there is at
2478 most one data dictionary transaction active at a time. */
2479@@ -1042,7 +1048,7 @@
2480 #if defined(UNIV_HOTBACKUP) || defined(__WIN__) || defined(__NETWARE__)
2481 /* InnoDB Hot Backup does not lock the data files.
2482 * On Windows, mandatory locking is used.
2483-@@ -1357,8 +1357,9 @@
2484+@@ -1365,8 +1365,9 @@
2485 attributes = 0;
2486 ut_error;
2487 }
2488@@ -1053,7 +1059,7 @@
2489 GENERIC_READ | GENERIC_WRITE, /* read and write
2490 access */
2491 share_mode, /* File can be read also by other
2492-@@ -1417,7 +1418,11 @@
2493+@@ -1432,7 +1433,11 @@
2494 if (create_mode == OS_FILE_OPEN || create_mode == OS_FILE_OPEN_RAW
2495 || create_mode == OS_FILE_OPEN_RETRY) {
2496 mode_str = "OPEN";
2497@@ -1066,7 +1072,7 @@
2498 } else if (create_mode == OS_FILE_CREATE) {
2499 mode_str = "CREATE";
2500 create_flag = O_RDWR | O_CREAT | O_EXCL;
2501-@@ -2682,6 +2687,9 @@
2502+@@ -2716,6 +2721,9 @@
2503 ut_a((offset & 0xFFFFFFFFUL) == offset);
2504 ut_a((n & 0xFFFFFFFFUL) == n);
2505
2506@@ -1076,7 +1082,7 @@
2507 os_n_file_writes++;
2508
2509 ut_ad(file);
2510-@@ -2806,6 +2814,9 @@
2511+@@ -2840,6 +2848,9 @@
2512 #else
2513 ssize_t ret;
2514
2515@@ -1108,9 +1114,9 @@
2516
2517 --- a/storage/innodb_plugin/srv/srv0srv.c
2518 +++ b/storage/innodb_plugin/srv/srv0srv.c
2519-@@ -86,11 +86,6 @@
2520- #include "trx0i_s.h"
2521- #include "os0sync.h" /* for HAVE_ATOMIC_BUILTINS */
2522+@@ -96,11 +96,6 @@
2523+ # include "my_sys.h" /* DEBUG_SYNC_C */
2524+ #endif
2525
2526 -/* prototypes of new functions added to ha_innodb.cc for kill_idle_transaction */
2527 -ibool innobase_thd_is_idle(const void* thd);
2528@@ -1120,7 +1126,7 @@
2529 /* prototypes for new functions added to ha_innodb.cc */
2530 ibool innobase_get_slow_log();
2531
2532-@@ -415,6 +410,11 @@
2533+@@ -438,6 +433,11 @@
2534
2535 UNIV_INTERN ulong srv_replication_delay = 0;
2536
2537@@ -1132,7 +1138,7 @@
2538 UNIV_INTERN long long srv_ibuf_max_size = 0;
2539 UNIV_INTERN ulint srv_ibuf_active_contract = 0; /* 0:disable 1:enable */
2540 UNIV_INTERN ulint srv_ibuf_accel_rate = 100;
2541-@@ -1065,7 +1065,7 @@
2542+@@ -1095,7 +1095,7 @@
2543 }
2544
2545 /* Initialize some INFORMATION SCHEMA internal structures */
2546@@ -1141,7 +1147,7 @@
2547 }
2548
2549 /*********************************************************************//**
2550-@@ -1076,6 +1076,7 @@
2551+@@ -1106,6 +1106,7 @@
2552 /*==========*/
2553 {
2554 os_fast_mutex_free(&srv_conc_mutex);
2555@@ -1149,7 +1155,7 @@
2556 mem_free(srv_conc_slots);
2557 srv_conc_slots = NULL;
2558
2559-@@ -1089,6 +1090,7 @@
2560+@@ -1119,6 +1120,7 @@
2561 srv_mysql_table = NULL;
2562
2563 trx_i_s_cache_free(trx_i_s_cache);
2564@@ -1157,7 +1163,7 @@
2565 }
2566
2567 /*********************************************************************//**
2568-@@ -2570,36 +2572,6 @@
2569+@@ -2634,36 +2636,6 @@
2570 old_sema = sema;
2571 }
2572
2573@@ -1196,7 +1202,7 @@
2574
2575 --- a/storage/innodb_plugin/srv/srv0start.c
2576 +++ b/storage/innodb_plugin/srv/srv0start.c
2577-@@ -62,6 +62,7 @@
2578+@@ -63,6 +63,7 @@
2579 #include "ibuf0ibuf.h"
2580 #include "srv0start.h"
2581 #include "srv0srv.h"
2582@@ -1204,7 +1210,7 @@
2583 #ifndef UNIV_HOTBACKUP
2584 # include "os0proc.h"
2585 # include "sync0sync.h"
2586-@@ -95,6 +96,8 @@
2587+@@ -96,6 +97,8 @@
2588 /** Log sequence number at shutdown */
2589 UNIV_INTERN ib_uint64_t srv_shutdown_lsn;
2590
2591@@ -1213,7 +1219,7 @@
2592 #ifdef HAVE_DARWIN_THREADS
2593 # include <sys/utsname.h>
2594 /** TRUE if the F_FULLFSYNC option is available */
2595-@@ -545,7 +548,7 @@
2596+@@ -540,7 +543,7 @@
2597 /*********************************************************************//**
2598 Creates or opens the log files and closes them.
2599 @return DB_SUCCESS or error code */
2600@@ -1222,7 +1228,7 @@
2601 ulint
2602 open_or_create_log_file(
2603 /*====================*/
2604-@@ -703,7 +706,7 @@
2605+@@ -698,7 +701,7 @@
2606 /*********************************************************************//**
2607 Creates or opens database data files and closes them.
2608 @return DB_SUCCESS or error code */
2609@@ -1231,7 +1237,7 @@
2610 ulint
2611 open_or_create_data_files(
2612 /*======================*/
2613-@@ -1782,6 +1785,10 @@
2614+@@ -1804,6 +1807,10 @@
2615 are initialized in trx_sys_init_at_db_start(). */
2616
2617 recv_recovery_from_checkpoint_finish();
2618@@ -1242,7 +1248,7 @@
2619 if (srv_force_recovery < SRV_FORCE_NO_IBUF_MERGE) {
2620 /* The following call is necessary for the insert
2621 buffer to work with multiple tablespaces. We must
2622-@@ -1962,7 +1969,18 @@
2623+@@ -1990,7 +1997,18 @@
2624
2625 if (srv_auto_extend_last_data_file
2626 && sum_of_data_file_sizes < tablespace_size_in_header) {
2627@@ -1261,7 +1267,7 @@
2628 fprintf(stderr,
2629 "InnoDB: Error: tablespace size stored in header"
2630 " is %lu pages, but\n"
2631-@@ -1987,6 +2005,7 @@
2632+@@ -2015,6 +2033,7 @@
2633
2634 return(DB_ERROR);
2635 }
2636@@ -1269,7 +1275,7 @@
2637 }
2638
2639 /* Check that os_fast_mutexes work as expected */
2640-@@ -2008,6 +2027,10 @@
2641+@@ -2036,6 +2055,10 @@
2642
2643 os_fast_mutex_free(&srv_os_test_mutex);
2644
2645@@ -1280,7 +1286,7 @@
2646 if (!srv_file_per_table_original_value
2647 && srv_pass_corrupt_table) {
2648 fprintf(stderr, "InnoDB: Warning:"
2649-@@ -2089,6 +2112,7 @@
2650+@@ -2117,6 +2140,7 @@
2651 ibuf_update_max_tablespace_id();
2652 }
2653
2654@@ -1290,9 +1296,9 @@
2655 srv_was_started = TRUE;
2656 --- a/storage/innodb_plugin/trx/trx0purge.c
2657 +++ b/storage/innodb_plugin/trx/trx0purge.c
2658-@@ -1113,6 +1113,9 @@
2659+@@ -1119,6 +1119,9 @@
2660 /* que_thr_t* thr2; */
2661- ulint old_pages_handled;
2662+ ulonglong old_pages_handled;
2663
2664 + if (srv_fake_write)
2665 + return(0);
2666@@ -1316,7 +1322,7 @@
2667
2668 --- a/storage/innodb_plugin/trx/trx0sys.c
2669 +++ b/storage/innodb_plugin/trx/trx0sys.c
2670-@@ -1910,12 +1910,14 @@
2671+@@ -1944,12 +1944,14 @@
2672 mem_free(trx_doublewrite);
2673 trx_doublewrite = NULL;
2674
2675@@ -1331,7 +1337,7 @@
2676
2677 /* There can't be any active transactions. */
2678 rseg = UT_LIST_GET_FIRST(trx_sys->rseg_list);
2679-@@ -1941,10 +1943,12 @@
2680+@@ -1975,10 +1977,12 @@
2681 UT_LIST_REMOVE(view_list, trx_sys->view_list, prev_view);
2682 }
2683
2684@@ -1389,7 +1395,7 @@
2685 mutex_enter(&kernel_mutex);
2686
2687 ret = trx_start_low(trx, rseg_id);
2688-@@ -990,7 +984,6 @@
2689+@@ -1000,7 +994,6 @@
2690 trx->read_view = NULL;
2691
2692 if (lsn) {
2693@@ -1397,7 +1403,7 @@
2694
2695 mutex_exit(&kernel_mutex);
2696
2697-@@ -999,11 +992,6 @@
2698+@@ -1009,11 +1002,6 @@
2699 trx_undo_insert_cleanup(trx);
2700 }
2701
2702@@ -1409,7 +1415,7 @@
2703
2704 /* NOTE that we could possibly make a group commit more
2705 efficient here: call os_thread_yield here to allow also other
2706-@@ -1036,9 +1024,9 @@
2707+@@ -1046,9 +1034,9 @@
2708 if (trx->flush_log_later) {
2709 /* Do nothing yet */
2710 trx->must_flush_log_later = TRUE;
2711@@ -1421,7 +1427,7 @@
2712 if (srv_unix_file_flush_method == SRV_UNIX_NOSYNC) {
2713 /* Write the log but do not flush it to disk */
2714
2715-@@ -1050,7 +1038,7 @@
2716+@@ -1060,7 +1048,7 @@
2717
2718 log_write_up_to(lsn, LOG_WAIT_ONE_GROUP, TRUE);
2719 }
2720@@ -1430,7 +1436,7 @@
2721
2722 /* Write the log but do not flush it to disk */
2723
2724-@@ -1730,23 +1718,16 @@
2725+@@ -1742,23 +1730,16 @@
2726 trx_t* trx) /*!< in: trx handle */
2727 {
2728 ib_uint64_t lsn = trx->commit_lsn;
2729@@ -1456,7 +1462,7 @@
2730 if (srv_unix_file_flush_method == SRV_UNIX_NOSYNC) {
2731 /* Write the log but do not flush it to disk */
2732
2733-@@ -1757,7 +1738,7 @@
2734+@@ -1769,7 +1750,7 @@
2735
2736 log_write_up_to(lsn, LOG_WAIT_ONE_GROUP, TRUE);
2737 }
2738@@ -1465,7 +1471,7 @@
2739
2740 /* Write the log but do not flush it to disk */
2741
2742-@@ -2018,8 +1999,6 @@
2743+@@ -2047,8 +2028,6 @@
2744 /*--------------------------------------*/
2745
2746 if (lsn) {
2747@@ -1474,7 +1480,7 @@
2748 /* Depending on the my.cnf options, we may now write the log
2749 buffer to the log files, making the prepared state of the
2750 transaction durable if the OS does not crash. We may also
2751-@@ -2039,15 +2018,9 @@
2752+@@ -2068,15 +2047,9 @@
2753
2754 mutex_exit(&kernel_mutex);
2755
2756@@ -1492,7 +1498,7 @@
2757 if (srv_unix_file_flush_method == SRV_UNIX_NOSYNC) {
2758 /* Write the log but do not flush it to disk */
2759
2760-@@ -2059,7 +2032,7 @@
2761+@@ -2088,7 +2061,7 @@
2762
2763 log_write_up_to(lsn, LOG_WAIT_ONE_GROUP, TRUE);
2764 }
2765@@ -1511,9 +1517,9 @@
2766
2767 /*
2768 WAIT ARRAY
2769-@@ -926,6 +927,13 @@
2770- ibool fatal = FALSE;
2771- double longest_diff = 0;
2772+@@ -931,6 +932,13 @@
2773+ return(FALSE);
2774+ }
2775
2776 + if (srv_rebuild_indexes) {
2777 +
2778@@ -1522,12 +1528,12 @@
2779 + return(FALSE);
2780 + }
2781 +
2782+ sync_array_enter(sync_primary_wait_array);
2783+
2784 for (i = 0; i < sync_primary_wait_array->n_cells; i++) {
2785-
2786- double diff;
2787 --- a/storage/innodb_plugin/include/buf0buf.h
2788 +++ b/storage/innodb_plugin/include/buf0buf.h
2789-@@ -1211,6 +1211,8 @@
2790+@@ -1295,6 +1295,8 @@
2791 frees a page in buffer pool */
2792 # endif /* UNIV_DEBUG_FILE_ACCESSES || UNIV_DEBUG */
2793 #endif /* !UNIV_HOTBACKUP */
2794@@ -1579,14 +1585,3 @@
2795
2796 UNIV_MEM_FREE(buf, n);
2797 }
2798---- a/configure.in
2799-+++ b/configure.in
2800-@@ -22,7 +22,7 @@
2801- AM_INIT_AUTOMAKE([1.9 tar-ustar])
2802- AC_PROG_LIBTOOL
2803-
2804--AM_CONFIG_HEADER([include/config.h])
2805-+AC_CONFIG_HEADERS([include/config.h])
2806-
2807- # Request support for automake silent-rules if available.
2808- # Default to verbose output. One can use the configure-time
2809
2810=== modified file 'patches/xtradb55.patch'
2811--- patches/xtradb55.patch 2013-05-07 13:32:47 +0000
2812+++ patches/xtradb55.patch 2013-07-22 14:01:36 +0000
2813@@ -1,6 +1,7 @@
2814+=== modified file 'Percona-Server/storage/innobase/btr/btr0btr.c'
2815 --- a/storage/innobase/btr/btr0btr.c
2816 +++ b/storage/innobase/btr/btr0btr.c
2817-@@ -674,7 +674,7 @@
2818+@@ -704,7 +704,7 @@
2819 /**************************************************************//**
2820 Gets the root node of a tree and x-latches it.
2821 @return root page, x-latched */
2822@@ -9,7 +10,7 @@
2823 buf_block_t*
2824 btr_root_block_get(
2825 /*===============*/
2826-@@ -1134,7 +1134,7 @@
2827+@@ -1209,7 +1209,7 @@
2828 /************************************************************//**
2829 Returns the child page of a node pointer and x-latches it.
2830 @return child page, x-latched */
2831@@ -28,7 +29,7 @@
2832
2833 /* prototypes for new functions added to ha_innodb.cc */
2834 trx_t* innobase_get_trx();
2835-@@ -581,7 +582,7 @@
2836+@@ -599,7 +600,7 @@
2837 return(TRUE);
2838 }
2839
2840@@ -37,7 +38,7 @@
2841 if (recv_lsn_checks_on) {
2842 ib_uint64_t current_lsn;
2843
2844-@@ -965,6 +966,7 @@
2845+@@ -1000,6 +1001,7 @@
2846 block->page.zip_list.prev = NULL;
2847 block->page.zip_list.next = NULL;
2848 block->page.in_LRU_list = FALSE;
2849@@ -45,7 +46,7 @@
2850 block->in_unzip_LRU_list = FALSE;
2851 #if defined UNIV_AHI_DEBUG || defined UNIV_DEBUG
2852 block->n_pointers = 0;
2853-@@ -2542,7 +2544,8 @@
2854+@@ -2471,7 +2473,8 @@
2855 ut_ad(zip_size == fil_space_get_zip_size(space));
2856 ut_ad(ut_is_2pow(zip_size));
2857 #ifndef UNIV_LOG_DEBUG
2858@@ -55,7 +56,7 @@
2859 || ibuf_page_low(space, zip_size, offset,
2860 FALSE, file, line, NULL));
2861 #endif
2862-@@ -3952,6 +3955,13 @@
2863+@@ -3902,6 +3905,13 @@
2864 frame = ((buf_block_t*) bpage)->frame;
2865 }
2866
2867@@ -69,19 +70,19 @@
2868 /* If this page is not uninitialized and not in the
2869 doublewrite buffer, then the page number and space id
2870 should be the same as in block. */
2871-@@ -4064,7 +4074,8 @@
2872+@@ -4025,7 +4035,8 @@
2873 recv_recover_page(TRUE, (buf_block_t*) bpage);
2874 }
2875
2876 - if (uncompressed && !recv_no_ibuf_operations) {
2877 + if (uncompressed && !recv_no_ibuf_operations &&
2878 + !srv_fake_write && !bpage->is_compacted) {
2879- ibuf_merge_or_delete_for_page(
2880- /* Delete possible entries, if bpage is_corrupt */
2881- (srv_pass_corrupt_table && bpage->is_corrupt) ? NULL :
2882+
2883+ buf_block_t* block;
2884+ ibool update_ibuf_bitmap;
2885 --- a/storage/innobase/fil/fil0fil.c
2886 +++ b/storage/innobase/fil/fil0fil.c
2887-@@ -307,7 +307,7 @@
2888+@@ -310,7 +310,7 @@
2889
2890 /** The tablespace memory cache. This variable is NULL before the module is
2891 initialized. */
2892@@ -90,7 +91,7 @@
2893
2894 #ifdef UNIV_DEBUG
2895 /** Try fil_validate() every this many times */
2896-@@ -346,7 +346,7 @@
2897+@@ -349,7 +349,7 @@
2898 off the LRU list if it is in the LRU list. The caller must hold the fil_sys
2899 mutex. */
2900 static
2901@@ -99,7 +100,7 @@
2902 fil_node_prepare_for_io(
2903 /*====================*/
2904 fil_node_t* node, /*!< in: file node */
2905-@@ -671,7 +671,7 @@
2906+@@ -674,7 +674,7 @@
2907 Opens a the file of a node of a tablespace. The caller must own the fil_system
2908 mutex. */
2909 static
2910@@ -108,7 +109,7 @@
2911 fil_node_open_file(
2912 /*===============*/
2913 fil_node_t* node, /*!< in: file node */
2914-@@ -706,7 +706,17 @@
2915+@@ -709,7 +709,17 @@
2916 OS_FILE_READ_ONLY, &success);
2917 if (!success) {
2918 /* The following call prints an error message */
2919@@ -127,7 +128,7 @@
2920
2921 ut_print_timestamp(stderr);
2922
2923-@@ -767,12 +777,15 @@
2924+@@ -770,12 +780,15 @@
2925
2926 if (UNIV_UNLIKELY(space_id != space->id)) {
2927 fprintf(stderr,
2928@@ -146,7 +147,7 @@
2929 }
2930
2931 if (UNIV_UNLIKELY(space_id == ULINT_UNDEFINED
2932-@@ -796,8 +809,8 @@
2933+@@ -799,8 +812,8 @@
2934 }
2935
2936 if (size_bytes >= 1024 * 1024) {
2937@@ -157,7 +158,7 @@
2938 }
2939
2940 if (!(flags & DICT_TF_ZSSIZE_MASK)) {
2941-@@ -848,6 +861,8 @@
2942+@@ -851,6 +864,8 @@
2943 /* Put the node to the LRU list */
2944 UT_LIST_ADD_FIRST(LRU, system->LRU, node);
2945 }
2946@@ -166,7 +167,7 @@
2947 }
2948
2949 /**********************************************************************//**
2950-@@ -1477,7 +1492,12 @@
2951+@@ -1517,7 +1532,12 @@
2952 the file yet; the following calls will open it and update the
2953 size fields */
2954
2955@@ -180,7 +181,7 @@
2956 fil_node_complete_io(node, fil_system, OS_FILE_READ);
2957 }
2958
2959-@@ -1529,7 +1549,12 @@
2960+@@ -1586,7 +1606,12 @@
2961 the file yet; the following calls will open it and update the
2962 size fields */
2963
2964@@ -194,7 +195,7 @@
2965 fil_node_complete_io(node, fil_system, OS_FILE_READ);
2966 }
2967
2968-@@ -1981,7 +2006,7 @@
2969+@@ -2045,7 +2070,7 @@
2970 mem_free(path);
2971 }
2972
2973@@ -203,7 +204,7 @@
2974 /********************************************************//**
2975 Writes a log record about an .ibd file create/rename/delete. */
2976 static
2977-@@ -2205,7 +2230,7 @@
2978+@@ -2269,7 +2294,7 @@
2979 if (fil_create_new_single_table_tablespace(
2980 space_id, name, FALSE, flags,
2981 FIL_IBD_FILE_INITIAL_SIZE) != DB_SUCCESS) {
2982@@ -212,7 +213,7 @@
2983 }
2984 }
2985
2986-@@ -2372,7 +2397,7 @@
2987+@@ -2460,7 +2485,7 @@
2988 }
2989
2990 if (success) {
2991@@ -221,16 +222,16 @@
2992 /* Write a log record about the deletion of the .ibd
2993 file, so that ibbackup can replay it in the
2994 --apply-log phase. We use a dummy mtr and the familiar
2995-@@ -2673,7 +2698,7 @@
2996+@@ -2761,7 +2786,7 @@
2997
2998 mutex_exit(&fil_system->mutex);
2999
3000 -#ifndef UNIV_HOTBACKUP
3001 +#ifdef UNDEFINED
3002- if (success) {
3003+ if (success && !recv_recovery_on) {
3004 mtr_t mtr;
3005
3006-@@ -2864,7 +2889,7 @@
3007+@@ -2952,7 +2977,7 @@
3008
3009 fil_node_create(path, size, space_id, FALSE);
3010
3011@@ -239,7 +240,7 @@
3012 {
3013 mtr_t mtr;
3014
3015-@@ -3210,19 +3235,97 @@
3016+@@ -3301,19 +3326,97 @@
3017 "InnoDB: open the tablespace file ", stderr);
3018 ut_print_filename(stderr, filepath);
3019 fputs("!\n"
3020@@ -347,7 +348,7 @@
3021 return(FALSE);
3022 }
3023
3024-@@ -4158,7 +4261,7 @@
3025+@@ -4258,7 +4361,7 @@
3026 cannot be ok. */
3027
3028 size = (((ib_int64_t)size_high) << 32) + (ib_int64_t)size_low;
3029@@ -356,7 +357,7 @@
3030 if (size < FIL_IBD_FILE_INITIAL_SIZE * (lint)UNIV_PAGE_SIZE) {
3031 fprintf(stderr,
3032 "InnoDB: Error: the size of single-table tablespace"
3033-@@ -4288,7 +4391,51 @@
3034+@@ -4388,7 +4491,51 @@
3035
3036 fil_node_create(filepath, 0, space_id, FALSE);
3037 func_exit:
3038@@ -409,7 +410,7 @@
3039 ut_free(buf2);
3040 mem_free(filepath);
3041 }
3042-@@ -4299,7 +4446,7 @@
3043+@@ -4399,7 +4546,7 @@
3044 idea is to read as much good data as we can and jump over bad data.
3045 @return 0 if ok, -1 if error even after the retries, 1 if at the end
3046 of the directory */
3047@@ -418,7 +419,7 @@
3048 int
3049 fil_file_readdir_next_file(
3050 /*=======================*/
3051-@@ -4343,7 +4490,7 @@
3052+@@ -4443,7 +4590,7 @@
3053 @return DB_SUCCESS or error number */
3054 UNIV_INTERN
3055 ulint
3056@@ -427,7 +428,7 @@
3057 /*===================================*/
3058 {
3059 int ret;
3060-@@ -4399,7 +4546,9 @@
3061+@@ -4499,7 +4646,9 @@
3062 dbinfo.name);
3063 srv_normalize_path_for_win(dbpath);
3064
3065@@ -438,7 +439,7 @@
3066
3067 if (dbdir != NULL) {
3068 /* printf("Opened dir %s\n", dbinfo.name); */
3069-@@ -4425,8 +4574,11 @@
3070+@@ -4525,8 +4674,11 @@
3071 ".ibd")) {
3072 /* The name ends in .ibd; try opening
3073 the file */
3074@@ -451,7 +452,7 @@
3075 }
3076 next_file_item:
3077 ret = fil_file_readdir_next_file(&err,
3078-@@ -4598,15 +4750,97 @@
3079+@@ -4698,15 +4850,97 @@
3080 "InnoDB: in InnoDB data dictionary"
3081 " has tablespace id %lu,\n"
3082 "InnoDB: but tablespace with that id"
3083@@ -557,7 +558,7 @@
3084 } else {
3085 ut_print_timestamp(stderr);
3086 fputs(" InnoDB: Error: table ", stderr);
3087-@@ -5005,7 +5239,7 @@
3088+@@ -5129,7 +5363,7 @@
3089 off the LRU list if it is in the LRU list. The caller must hold the fil_sys
3090 mutex. */
3091 static
3092@@ -566,7 +567,7 @@
3093 fil_node_prepare_for_io(
3094 /*====================*/
3095 fil_node_t* node, /*!< in: file node */
3096-@@ -5025,10 +5259,13 @@
3097+@@ -5149,10 +5383,13 @@
3098 }
3099
3100 if (node->open == FALSE) {
3101@@ -581,7 +582,7 @@
3102 }
3103
3104 if (node->n_pending == 0 && space->purpose == FIL_TABLESPACE
3105-@@ -5041,6 +5278,8 @@
3106+@@ -5165,6 +5402,8 @@
3107 }
3108
3109 node->n_pending++;
3110@@ -590,7 +591,7 @@
3111 }
3112
3113 /********************************************************************//**
3114-@@ -5240,6 +5479,16 @@
3115+@@ -5350,6 +5589,16 @@
3116
3117 ut_ad((mode != OS_AIO_IBUF) || (space->purpose == FIL_TABLESPACE));
3118
3119@@ -607,9 +608,28 @@
3120 node = UT_LIST_GET_FIRST(space->chain);
3121
3122 for (;;) {
3123+@@ -6028,6 +6277,12 @@
3124+ const char* tmp_name) /*!< in: temp table name used while
3125+ swapping */
3126+ {
3127++ (void)old_space_id;
3128++ (void)old_name;
3129++ (void)new_space_id;
3130++ (void)new_name;
3131++ (void)tmp_name;
3132++#ifdef UNDEFINED
3133+ mtr_t mtr;
3134+ mtr_start(&mtr);
3135+ fil_op_write_log(MLOG_FILE_RENAME, old_space_id,
3136+@@ -6035,4 +6290,5 @@
3137+ fil_op_write_log(MLOG_FILE_RENAME, new_space_id,
3138+ 0, 0, new_name, old_name, &mtr);
3139+ mtr_commit(&mtr);
3140++#endif
3141+ }
3142 --- a/storage/innobase/handler/ha_innodb.cc
3143 +++ b/storage/innobase/handler/ha_innodb.cc
3144-@@ -482,12 +482,6 @@
3145+@@ -514,12 +514,6 @@
3146 "Timeout in seconds an InnoDB transaction may wait for a lock before being rolled back. Values above 100000000 disable the timeout.",
3147 NULL, NULL, 50, 1, 1024 * 1024 * 1024, 0);
3148
3149@@ -622,7 +642,7 @@
3150 static MYSQL_THDVAR_BOOL(fake_changes, PLUGIN_VAR_OPCMDARG,
3151 "In the transaction after enabled, UPDATE, INSERT and DELETE only move the cursor to the records "
3152 "and do nothing other operations (no changes, no ibuf, no undo, no transaction log) in the transaction. "
3153-@@ -989,17 +983,6 @@
3154+@@ -1038,17 +1032,6 @@
3155 }
3156
3157 /******************************************************************//**
3158@@ -640,7 +660,7 @@
3159 Returns true if expand_fast_index_creation is enabled for the current
3160 session.
3161 @return the value of the server's expand_fast_index_creation variable */
3162-@@ -11806,17 +11789,12 @@
3163+@@ -12637,17 +12620,12 @@
3164 innodb_file_format_max_update, "Antelope");
3165
3166 /* Changed to the THDVAR */
3167@@ -664,7 +684,7 @@
3168
3169 static MYSQL_SYSVAR_STR(flush_method, innobase_file_flush_method,
3170 PLUGIN_VAR_RQCMDARG | PLUGIN_VAR_READONLY,
3171-@@ -12294,7 +12272,6 @@
3172+@@ -13234,7 +13212,6 @@
3173 MYSQL_SYSVAR(file_format_check),
3174 MYSQL_SYSVAR(file_format_max),
3175 MYSQL_SYSVAR(flush_log_at_trx_commit),
3176@@ -674,7 +694,7 @@
3177 MYSQL_SYSVAR(large_prefix),
3178 --- a/storage/innobase/ibuf/ibuf0ibuf.c
3179 +++ b/storage/innobase/ibuf/ibuf0ibuf.c
3180-@@ -1190,6 +1190,9 @@
3181+@@ -1199,6 +1199,9 @@
3182 ut_ad(!recv_no_ibuf_operations);
3183 ut_ad(x_latch || mtr == NULL);
3184
3185@@ -684,7 +704,7 @@
3186 if (ibuf_fixed_addr_page(space, zip_size, page_no)) {
3187
3188 return(TRUE);
3189-@@ -2757,6 +2760,9 @@
3190+@@ -2771,6 +2774,9 @@
3191 ulint n_bytes;
3192 ulint n_pag2;
3193
3194@@ -694,9 +714,20 @@
3195 while (sum_pages < n_pages) {
3196 n_bytes = ibuf_contract_ext(&n_pag2, sync);
3197
3198+--- a/storage/innobase/include/buf0buf.h
3199++++ b/storage/innobase/include/buf0buf.h
3200+@@ -1594,6 +1594,8 @@
3201+ frees a page in buffer pool */
3202+ # endif /* UNIV_DEBUG_FILE_ACCESSES || UNIV_DEBUG */
3203+ #endif /* !UNIV_HOTBACKUP */
3204++ ibool is_compacted; /*!< TRUE if the page was skipped in
3205++ compact backups */
3206+ };
3207+
3208+ /** The buffer control block structure */
3209 --- a/storage/innobase/include/fil0fil.h
3210 +++ b/storage/innobase/include/fil0fil.h
3211-@@ -506,7 +506,7 @@
3212+@@ -518,7 +518,7 @@
3213 @return DB_SUCCESS or error number */
3214 UNIV_INTERN
3215 ulint
3216@@ -723,7 +754,7 @@
3217 Get the current setting of the lower_case_table_names global parameter from
3218 --- a/storage/innobase/include/log0recv.h
3219 +++ b/storage/innobase/include/log0recv.h
3220-@@ -444,7 +444,7 @@
3221+@@ -481,7 +481,7 @@
3222 /* defined in os0file.c */
3223 #define OS_AIO_MERGE_N_CONSECUTIVE 64
3224 /* defined in log0recv.c */
3225@@ -734,7 +765,7 @@
3226
3227 --- a/storage/innobase/include/srv0srv.h
3228 +++ b/storage/innobase/include/srv0srv.h
3229-@@ -145,8 +145,7 @@
3230+@@ -159,8 +159,7 @@
3231 extern ulint srv_n_log_files;
3232 extern ulint srv_log_file_size;
3233 extern ulint srv_log_buffer_size;
3234@@ -743,19 +774,19 @@
3235 +extern ulong srv_flush_log_at_trx_commit;
3236 extern char srv_adaptive_flushing;
3237
3238-
3239-@@ -247,6 +246,11 @@
3240- extern ulint srv_dict_size_limit;
3241-
3242- extern ulint srv_lazy_drop_table;
3243-+
3244+ /* If this flag is TRUE, then we will load the indexes' (and tables') metadata
3245+@@ -271,6 +270,11 @@
3246+ extern ulint srv_expand_import;
3247+ extern ulint srv_pass_corrupt_table;
3248+
3249 +extern ibool srv_read_only;
3250 +extern ibool srv_fake_write;
3251 +extern ibool srv_apply_log_only;
3252 +extern ibool srv_backup_mode;
3253- /*-------------------------------------------*/
3254-
3255- extern ulint srv_n_rows_inserted;
3256++
3257+ /* Helper macro to support srv_pass_corrupt_table checks. If 'cond' is FALSE,
3258+ execute 'code' if srv_pass_corrupt_table is non-zero, or trigger a fatal error
3259+ otherwise. The break statement in 'code' will obviously not work as expected. */
3260 --- a/storage/innobase/include/srv0start.h
3261 +++ b/storage/innobase/include/srv0start.h
3262 @@ -91,6 +91,8 @@
3263@@ -769,7 +800,7 @@
3264 extern ibool srv_have_fullfsync;
3265 --- a/storage/innobase/log/log0log.c
3266 +++ b/storage/innobase/log/log0log.c
3267-@@ -1154,7 +1154,7 @@
3268+@@ -1237,7 +1237,7 @@
3269 if (srv_unix_file_flush_method != SRV_UNIX_O_DSYNC
3270 && srv_unix_file_flush_method != SRV_UNIX_ALL_O_DIRECT
3271 && srv_unix_file_flush_method != SRV_UNIX_NOSYNC
3272@@ -778,7 +809,7 @@
3273
3274 fil_flush(group->space_id, FALSE);
3275 }
3276-@@ -1389,7 +1389,7 @@
3277+@@ -1472,7 +1472,7 @@
3278 #endif /* UNIV_DEBUG */
3279 ulint unlock;
3280
3281@@ -787,7 +818,7 @@
3282 /* Recovery is running and no operations on the log files are
3283 allowed yet (the variable name .._no_ibuf_.. is misleading) */
3284
3285-@@ -2051,7 +2051,7 @@
3286+@@ -2140,7 +2140,7 @@
3287 return(TRUE);
3288 }
3289
3290@@ -796,16 +827,16 @@
3291
3292 if (log_sys->n_pending_checkpoint_writes > 0) {
3293 /* A checkpoint write is running */
3294-@@ -3151,6 +3151,7 @@
3295+@@ -3277,6 +3277,7 @@
3296 shutdown, because the InnoDB layer may have committed or
3297 prepared transactions and we don't want to lose them. */
3298
3299 + if (!srv_apply_log_only) {
3300 server_busy = trx_n_mysql_transactions > 0
3301 || UT_LIST_GET_LEN(trx_sys->trx_list) > trx_n_prepared;
3302- mutex_exit(&kernel_mutex);
3303-@@ -3158,6 +3159,9 @@
3304- if (server_busy || srv_is_any_background_thread_active()) {
3305+
3306+@@ -3329,6 +3330,9 @@
3307+
3308 goto loop;
3309 }
3310 + } else {
3311@@ -867,16 +898,7 @@
3312 ulint
3313 recv_find_max_checkpoint(
3314 /*=====================*/
3315-@@ -850,7 +851,7 @@
3316- InnoDB-3.23.52 where the checksum field contains the log block number.
3317- @return TRUE if ok, or if the log block may be in the format of InnoDB
3318- version predating 3.23.52 */
3319--static
3320-+//static
3321- ibool
3322- log_block_checksum_is_ok_or_old_format(
3323- /*===================================*/
3324-@@ -1639,9 +1640,16 @@
3325+@@ -1645,9 +1646,16 @@
3326 if (page_zip) {
3327 memset(FIL_PAGE_LSN + page_zip->data, 0, 8);
3328 }
3329@@ -894,7 +916,7 @@
3330
3331 ib_uint64_t end_lsn;
3332
3333-@@ -1842,6 +1850,18 @@
3334+@@ -1848,6 +1856,18 @@
3335 ulint zip_size = fil_space_get_zip_size(space);
3336 ulint page_no = recv_addr->page_no;
3337
3338@@ -913,7 +935,7 @@
3339 if (recv_addr->state == RECV_NOT_PROCESSED) {
3340 if (!has_printed) {
3341 ut_print_timestamp(stderr);
3342-@@ -1875,7 +1895,7 @@
3343+@@ -1881,7 +1901,7 @@
3344
3345 mutex_enter(&(recv_sys->mutex));
3346 }
3347@@ -922,7 +944,7 @@
3348 recv_addr = HASH_GET_NEXT(addr_hash, recv_addr);
3349 }
3350
3351-@@ -2380,7 +2400,7 @@
3352+@@ -2386,7 +2406,7 @@
3353 || type == MLOG_FILE_RENAME
3354 || type == MLOG_FILE_DELETE) {
3355 ut_a(space);
3356@@ -931,7 +953,7 @@
3357 if (recv_replay_file_ops) {
3358
3359 /* In ibbackup --apply-log, replay an .ibd file
3360-@@ -2403,7 +2423,7 @@
3361+@@ -2409,7 +2429,7 @@
3362 ut_error;
3363 }
3364 }
3365@@ -940,7 +962,7 @@
3366 /* In normal mysqld crash recovery we do not try to
3367 replay file operations */
3368 #ifdef UNIV_LOG_LSN_DEBUG
3369-@@ -2820,8 +2840,11 @@
3370+@@ -2826,8 +2846,11 @@
3371
3372 fprintf(stderr,
3373 "InnoDB: Doing recovery: scanned up to"
3374@@ -954,7 +976,7 @@
3375 }
3376 }
3377
3378-@@ -2926,7 +2949,7 @@
3379+@@ -2932,7 +2955,7 @@
3380 "InnoDB: Reading tablespace information"
3381 " from the .ibd files...\n");
3382
3383@@ -963,7 +985,7 @@
3384
3385 /* If we are using the doublewrite method, we will
3386 check if there are half-written pages in data files,
3387-@@ -2935,12 +2958,14 @@
3388+@@ -2941,12 +2964,14 @@
3389
3390 if (srv_force_recovery < SRV_FORCE_NO_LOG_REDO) {
3391
3392@@ -979,7 +1001,7 @@
3393 }
3394 }
3395
3396-@@ -3109,6 +3134,7 @@
3397+@@ -3115,6 +3140,7 @@
3398 recv_sys->recovered_lsn = checkpoint_lsn;
3399
3400 srv_start_lsn = checkpoint_lsn;
3401@@ -987,7 +1009,7 @@
3402 }
3403
3404 contiguous_lsn = ut_uint64_align_down(recv_sys->scanned_lsn,
3405-@@ -3474,6 +3500,7 @@
3406+@@ -3480,6 +3506,7 @@
3407 that the data dictionary tables will be free of any locks.
3408 The data dictionary latch should guarantee that there is at
3409 most one data dictionary transaction active at a time. */
3410@@ -995,9 +1017,52 @@
3411 trx_rollback_or_clean_recovered(FALSE);
3412 }
3413
3414+--- a/storage/innobase/mem/mem0dbg.c
3415++++ b/storage/innobase/mem/mem0dbg.c
3416+@@ -282,18 +282,10 @@
3417+ byte* buf, /*!< in: pointer to buffer */
3418+ ulint n) /*!< in: length of buffer */
3419+ {
3420+- byte* ptr;
3421+-
3422+ UNIV_MEM_ASSERT_W(buf, n);
3423+
3424+- for (ptr = buf; ptr < buf + n; ptr++) {
3425+-
3426+- if (ut_rnd_gen_ibool()) {
3427+- *ptr = 0xBA;
3428+- } else {
3429+- *ptr = 0xBE;
3430+- }
3431+- }
3432++ /* Fix https://bugs.launchpad.net/percona-xtrabackup/+bug/1158154 */
3433++ memset(buf, 0xBA, n);
3434+
3435+ UNIV_MEM_INVALID(buf, n);
3436+ }
3437+@@ -308,17 +300,10 @@
3438+ byte* buf, /*!< in: pointer to buffer */
3439+ ulint n) /*!< in: length of buffer */
3440+ {
3441+- byte* ptr;
3442+-
3443+ UNIV_MEM_ASSERT_W(buf, n);
3444+
3445+- for (ptr = buf; ptr < buf + n; ptr++) {
3446+- if (ut_rnd_gen_ibool()) {
3447+- *ptr = 0xDE;
3448+- } else {
3449+- *ptr = 0xAD;
3450+- }
3451+- }
3452++ /* Fix https://bugs.launchpad.net/percona-xtrabackup/+bug/1158154 */
3453++ memset(buf, 0xDE, n);
3454+
3455+ UNIV_MEM_FREE(buf, n);
3456+ }
3457 --- a/storage/innobase/os/os0file.c
3458 +++ b/storage/innobase/os/os0file.c
3459-@@ -660,7 +660,7 @@
3460+@@ -670,7 +670,7 @@
3461 }
3462
3463 #undef USE_FILE_LOCK
3464@@ -1006,7 +1071,7 @@
3465 #if defined(UNIV_HOTBACKUP) || defined(__WIN__)
3466 /* InnoDB Hot Backup does not lock the data files.
3467 * On Windows, mandatory locking is used.
3468-@@ -1428,7 +1428,7 @@
3469+@@ -1474,7 +1474,7 @@
3470 #endif
3471 #ifdef UNIV_NON_BUFFERED_IO
3472 # ifndef UNIV_HOTBACKUP
3473@@ -1015,7 +1080,7 @@
3474 /* Do not use unbuffered i/o to log files because
3475 value 2 denotes that we do not flush the log at every
3476 commit, but only once per second */
3477-@@ -1444,7 +1444,7 @@
3478+@@ -1490,7 +1490,7 @@
3479 attributes = 0;
3480 #ifdef UNIV_NON_BUFFERED_IO
3481 # ifndef UNIV_HOTBACKUP
3482@@ -1024,7 +1089,7 @@
3483 /* Do not use unbuffered i/o to log files because
3484 value 2 denotes that we do not flush the log at every
3485 commit, but only once per second */
3486-@@ -1460,8 +1460,9 @@
3487+@@ -1506,8 +1506,9 @@
3488 attributes = 0;
3489 ut_error;
3490 }
3491@@ -1035,7 +1100,7 @@
3492 GENERIC_READ | GENERIC_WRITE, /* read and write
3493 access */
3494 share_mode, /* File can be read also by other
3495-@@ -1520,7 +1521,11 @@
3496+@@ -1572,7 +1573,11 @@
3497 if (create_mode == OS_FILE_OPEN || create_mode == OS_FILE_OPEN_RAW
3498 || create_mode == OS_FILE_OPEN_RETRY) {
3499 mode_str = "OPEN";
3500@@ -1048,7 +1113,7 @@
3501 } else if (create_mode == OS_FILE_CREATE) {
3502 mode_str = "CREATE";
3503 create_flag = O_RDWR | O_CREAT | O_EXCL;
3504-@@ -2791,6 +2796,9 @@
3505+@@ -2888,6 +2893,9 @@
3506 ut_a((offset & 0xFFFFFFFFUL) == offset);
3507 ut_a((n & 0xFFFFFFFFUL) == n);
3508
3509@@ -1058,7 +1123,7 @@
3510 os_n_file_writes++;
3511
3512 ut_ad(file);
3513-@@ -2915,6 +2923,9 @@
3514+@@ -3012,6 +3020,9 @@
3515 #else
3516 ssize_t ret;
3517
3518@@ -1089,19 +1154,20 @@
3519 }
3520 --- a/storage/innobase/srv/srv0srv.c
3521 +++ b/storage/innobase/srv/srv0srv.c
3522-@@ -87,11 +87,6 @@
3523+@@ -90,12 +90,6 @@
3524 #include "mysql/plugin.h"
3525 #include "mysql/service_thd_wait.h"
3526
3527 -/* prototypes of new functions added to ha_innodb.cc for kill_idle_transaction */
3528 -ibool innobase_thd_is_idle(const void* thd);
3529 -ib_int64_t innobase_thd_get_start_time(const void* thd);
3530--void innobase_thd_kill(void* thd);
3531+-void innobase_thd_kill(ulong thd_id);
3532+-ulong innobase_thd_get_thread_id(const void* thd);
3533 -
3534 /* prototypes for new functions added to ha_innodb.cc */
3535 ibool innobase_get_slow_log();
3536
3537-@@ -198,8 +193,7 @@
3538+@@ -212,8 +206,7 @@
3539 UNIV_INTERN ulint srv_log_file_size = ULINT_MAX;
3540 /* size in database pages */
3541 UNIV_INTERN ulint srv_log_buffer_size = ULINT_MAX;
3542@@ -1111,10 +1177,10 @@
3543
3544 /* Try to flush dirty pages so as to avoid IO bursts at
3545 the checkpoints. */
3546-@@ -457,6 +451,11 @@
3547+@@ -435,6 +428,11 @@
3548+ UNIV_INTERN ulint srv_pass_corrupt_table = 0; /* 0:disable 1:enable */
3549+
3550 UNIV_INTERN ulint srv_dict_size_limit = 0;
3551-
3552- UNIV_INTERN ulint srv_lazy_drop_table = 0;
3553 +
3554 +UNIV_INTERN ibool srv_read_only = FALSE;
3555 +UNIV_INTERN ibool srv_fake_write = FALSE;
3556@@ -1123,7 +1189,7 @@
3557 /*-------------------------------------------*/
3558 UNIV_INTERN ulong srv_n_spin_wait_rounds = 30;
3559 UNIV_INTERN ulong srv_n_free_tickets_to_enter = 500;
3560-@@ -1137,7 +1136,7 @@
3561+@@ -1185,7 +1183,7 @@
3562 }
3563
3564 /* Initialize some INFORMATION SCHEMA internal structures */
3565@@ -1132,7 +1198,7 @@
3566 }
3567
3568 /*********************************************************************//**
3569-@@ -1148,6 +1147,7 @@
3570+@@ -1196,6 +1194,7 @@
3571 /*==========*/
3572 {
3573 os_fast_mutex_free(&srv_conc_mutex);
3574@@ -1140,7 +1206,7 @@
3575 mem_free(srv_conc_slots);
3576 srv_conc_slots = NULL;
3577
3578-@@ -1161,6 +1161,7 @@
3579+@@ -1209,6 +1208,7 @@
3580 srv_mysql_table = NULL;
3581
3582 trx_i_s_cache_free(trx_i_s_cache);
3583@@ -1148,7 +1214,7 @@
3584 }
3585
3586 /*********************************************************************//**
3587-@@ -1886,7 +1887,7 @@
3588+@@ -1938,7 +1938,7 @@
3589 }
3590
3591 /* Record the lock wait time for this thread */
3592@@ -1157,7 +1223,7 @@
3593 }
3594
3595 if (trx->was_chosen_as_deadlock_victim) {
3596-@@ -2842,36 +2843,6 @@
3597+@@ -2946,36 +2946,6 @@
3598 old_sema = sema;
3599 }
3600
3601@@ -1169,12 +1235,12 @@
3602 - mutex_enter(&kernel_mutex);
3603 - trx = UT_LIST_GET_FIRST(trx_sys->mysql_trx_list);
3604 - while (trx) {
3605-- if (trx->conc_state == TRX_ACTIVE
3606+- if (trx->state == TRX_ACTIVE
3607 - && trx->mysql_thd
3608 - && innobase_thd_is_idle(trx->mysql_thd)) {
3609-- ib_int64_t start_time; /* as stmt ID */
3610+- ib_int64_t start_time = innobase_thd_get_start_time(trx->mysql_thd);
3611+- ulong thd_id = innobase_thd_get_thread_id(trx->mysql_thd);
3612 -
3613-- start_time = innobase_thd_get_start_time(trx->mysql_thd);
3614 - if (trx->last_stmt_start != start_time) {
3615 - trx->idle_start = now;
3616 - trx->last_stmt_start = start_time;
3617@@ -1182,7 +1248,7 @@
3618 - > srv_kill_idle_transaction) {
3619 - /* kill the session */
3620 - mutex_exit(&kernel_mutex);
3621-- innobase_thd_kill(trx->mysql_thd);
3622+- innobase_thd_kill(thd_id);
3623 - goto rescan_idle;
3624 - }
3625 - }
3626@@ -1196,7 +1262,7 @@
3627
3628 --- a/storage/innobase/srv/srv0start.c
3629 +++ b/storage/innobase/srv/srv0start.c
3630-@@ -62,6 +62,7 @@
3631+@@ -63,6 +63,7 @@
3632 #include "ibuf0ibuf.h"
3633 #include "srv0start.h"
3634 #include "srv0srv.h"
3635@@ -1204,7 +1270,7 @@
3636 #ifndef UNIV_HOTBACKUP
3637 # include "os0proc.h"
3638 # include "sync0sync.h"
3639-@@ -94,6 +95,8 @@
3640+@@ -95,6 +96,8 @@
3641 /** Log sequence number at shutdown */
3642 UNIV_INTERN ib_uint64_t srv_shutdown_lsn;
3643
3644@@ -1213,7 +1279,7 @@
3645 #ifdef HAVE_DARWIN_THREADS
3646 # include <sys/utsname.h>
3647 /** TRUE if the F_FULLFSYNC option is available */
3648-@@ -549,7 +552,7 @@
3649+@@ -551,7 +554,7 @@
3650 /*********************************************************************//**
3651 Creates or opens the log files and closes them.
3652 @return DB_SUCCESS or error code */
3653@@ -1222,7 +1288,7 @@
3654 ulint
3655 open_or_create_log_file(
3656 /*====================*/
3657-@@ -709,7 +712,7 @@
3658+@@ -711,7 +714,7 @@
3659 /*********************************************************************//**
3660 Creates or opens database data files and closes them.
3661 @return DB_SUCCESS or error code */
3662@@ -1231,7 +1297,7 @@
3663 ulint
3664 open_or_create_data_files(
3665 /*======================*/
3666-@@ -1875,6 +1878,10 @@
3667+@@ -1934,6 +1937,10 @@
3668 are initialized in trx_sys_init_at_db_start(). */
3669
3670 recv_recovery_from_checkpoint_finish();
3671@@ -1242,7 +1308,7 @@
3672 if (srv_force_recovery < SRV_FORCE_NO_IBUF_MERGE) {
3673 /* The following call is necessary for the insert
3674 buffer to work with multiple tablespaces. We must
3675-@@ -2054,6 +2061,17 @@
3676+@@ -2119,6 +2126,17 @@
3677 if (!srv_auto_extend_last_data_file
3678 && sum_of_data_file_sizes != tablespace_size_in_header) {
3679
3680@@ -1260,7 +1326,7 @@
3681 ut_print_timestamp(stderr);
3682 fprintf(stderr,
3683 " InnoDB: Error: tablespace size"
3684-@@ -2133,6 +2151,7 @@
3685+@@ -2198,6 +2216,7 @@
3686
3687 return(DB_ERROR);
3688 }
3689@@ -1268,7 +1334,7 @@
3690 }
3691
3692 /* Check that os_fast_mutexes work as expected */
3693-@@ -2157,6 +2176,10 @@
3694+@@ -2222,6 +2241,10 @@
3695
3696 os_fast_mutex_free(&srv_os_test_mutex);
3697
3698@@ -1279,7 +1345,7 @@
3699 if (!srv_file_per_table_original_value
3700 && srv_pass_corrupt_table) {
3701 fprintf(stderr, "InnoDB: Warning:"
3702-@@ -2252,6 +2275,7 @@
3703+@@ -2317,6 +2340,7 @@
3704 ibuf_update_max_tablespace_id();
3705 }
3706
3707@@ -1287,9 +1353,33 @@
3708 srv_file_per_table = srv_file_per_table_original_value;
3709
3710 srv_was_started = TRUE;
3711+--- a/storage/innobase/sync/sync0arr.c
3712++++ b/storage/innobase/sync/sync0arr.c
3713+@@ -41,6 +41,7 @@
3714+ #include "os0file.h"
3715+ #include "srv0srv.h"
3716+ #include "ha_prototypes.h"
3717++#include "xb0xb.h"
3718+
3719+ /*
3720+ WAIT ARRAY
3721+@@ -927,6 +928,13 @@
3722+ return(FALSE);
3723+ }
3724+
3725++ if (srv_rebuild_indexes) {
3726++
3727++ /* Avoid long semaphore warnings when rebuilding indexes */
3728++
3729++ return(FALSE);
3730++ }
3731++
3732+ #ifdef UNIV_DEBUG_VALGRIND
3733+ /* Increase the timeouts if running under valgrind because it executes
3734+ extremely slowly. UNIV_DEBUG_VALGRIND does not necessary mean that
3735 --- a/storage/innobase/trx/trx0purge.c
3736 +++ b/storage/innobase/trx/trx0purge.c
3737-@@ -1129,6 +1129,9 @@
3738+@@ -1149,6 +1149,9 @@
3739 que_thr_t* thr;
3740 ulint old_pages_handled;
3741
3742@@ -1315,7 +1405,7 @@
3743
3744 --- a/storage/innobase/trx/trx0sys.c
3745 +++ b/storage/innobase/trx/trx0sys.c
3746-@@ -1952,12 +1952,14 @@
3747+@@ -2000,12 +2000,14 @@
3748 mem_free(trx_doublewrite);
3749 trx_doublewrite = NULL;
3750
3751@@ -1330,7 +1420,7 @@
3752
3753 /* There can't be any active transactions. */
3754 rseg = UT_LIST_GET_FIRST(trx_sys->rseg_list);
3755-@@ -1983,10 +1985,12 @@
3756+@@ -2031,10 +2033,12 @@
3757 UT_LIST_REMOVE(view_list, trx_sys->view_list, prev_view);
3758 }
3759
3760@@ -1341,36 +1431,36 @@
3761 ut_a(UT_LIST_GET_LEN(trx_sys->mysql_trx_list) == 0);
3762 + }
3763
3764- mem_free(trx_sys);
3765-
3766+ ut_ad(trx_sys->descr_n_used == 0);
3767+ ut_free(trx_sys->descriptors);
3768 --- a/storage/innobase/trx/trx0trx.c
3769 +++ b/storage/innobase/trx/trx0trx.c
3770-@@ -563,9 +563,8 @@
3771- (ullint) trx->id);
3772+@@ -692,8 +692,8 @@
3773
3774 if (srv_force_recovery == 0) {
3775--
3776-- trx->conc_state = TRX_PREPARED;
3777+
3778+- trx->state = TRX_PREPARED;
3779 - trx_n_prepared++;
3780-+ /* xtrabackup should rollback it */
3781-+ trx->conc_state = TRX_ACTIVE;
3782++ /* xtrabackup should rollback it */
3783++ trx->state = TRX_ACTIVE;
3784 } else {
3785 fprintf(stderr,
3786 "InnoDB: Since"
3787-@@ -639,10 +638,9 @@
3788- (ullint) trx->id);
3789+@@ -772,9 +772,11 @@
3790
3791 if (srv_force_recovery == 0) {
3792--
3793-+ /* xtrabackup should rollback it */
3794- trx->conc_state
3795+
3796++ /* xtrabackup should
3797++ rollback it */
3798+ trx->state
3799 - = TRX_PREPARED;
3800 - trx_n_prepared++;
3801 + = TRX_ACTIVE;
3802++ trx_reserve_descriptor(trx);
3803 } else {
3804 fprintf(stderr,
3805 "InnoDB: Since"
3806-@@ -1051,7 +1049,6 @@
3807+@@ -1211,7 +1213,6 @@
3808 trx->read_view = NULL;
3809
3810 if (lsn) {
3811@@ -1378,7 +1468,7 @@
3812
3813 mutex_exit(&kernel_mutex);
3814
3815-@@ -1060,12 +1057,6 @@
3816+@@ -1220,12 +1221,6 @@
3817 trx_undo_insert_cleanup(trx);
3818 }
3819
3820@@ -1391,7 +1481,7 @@
3821 /* NOTE that we could possibly make a group commit more
3822 efficient here: call os_thread_yield here to allow also other
3823 trxs to come to commit! */
3824-@@ -1097,9 +1088,9 @@
3825+@@ -1257,9 +1252,9 @@
3826 if (trx->flush_log_later) {
3827 /* Do nothing yet */
3828 trx->must_flush_log_later = TRUE;
3829@@ -1403,7 +1493,7 @@
3830 if (srv_unix_file_flush_method == SRV_UNIX_NOSYNC) {
3831 /* Write the log but do not flush it to disk */
3832
3833-@@ -1111,7 +1102,7 @@
3834+@@ -1271,7 +1266,7 @@
3835
3836 log_write_up_to(lsn, LOG_WAIT_ONE_GROUP, TRUE);
3837 }
3838@@ -1412,7 +1502,7 @@
3839
3840 /* Write the log but do not flush it to disk */
3841
3842-@@ -1791,23 +1782,16 @@
3843+@@ -1956,23 +1951,16 @@
3844 trx_t* trx) /*!< in: trx handle */
3845 {
3846 ib_uint64_t lsn = trx->commit_lsn;
3847@@ -1438,7 +1528,7 @@
3848 if (srv_unix_file_flush_method == SRV_UNIX_NOSYNC) {
3849 /* Write the log but do not flush it to disk */
3850
3851-@@ -1818,7 +1802,7 @@
3852+@@ -1983,7 +1971,7 @@
3853
3854 log_write_up_to(lsn, LOG_WAIT_ONE_GROUP, TRUE);
3855 }
3856@@ -1447,7 +1537,7 @@
3857
3858 /* Write the log but do not flush it to disk */
3859
3860-@@ -2066,7 +2050,6 @@
3861+@@ -2252,7 +2240,6 @@
3862 /*--------------------------------------*/
3863
3864 if (lsn) {
3865@@ -1455,7 +1545,7 @@
3866
3867 /* Depending on the my.cnf options, we may now write the log
3868 buffer to the log files, making the prepared state of the
3869-@@ -2087,15 +2070,9 @@
3870+@@ -2273,15 +2260,9 @@
3871
3872 mutex_exit(&kernel_mutex);
3873
3874@@ -1473,7 +1563,7 @@
3875 if (srv_unix_file_flush_method == SRV_UNIX_NOSYNC) {
3876 /* Write the log but do not flush it to disk */
3877
3878-@@ -2107,7 +2084,7 @@
3879+@@ -2293,7 +2274,7 @@
3880
3881 log_write_up_to(lsn, LOG_WAIT_ONE_GROUP, TRUE);
3882 }
3883@@ -1482,94 +1572,14 @@
3884
3885 /* Write the log but do not flush it to disk */
3886
3887---- a/configure.cmake
3888-+++ b/configure.cmake
3889-@@ -149,7 +149,9 @@
3890- SET(CMAKE_REQUIRED_LIBRARIES
3891- ${LIBM} ${LIBNSL} ${LIBBIND} ${LIBCRYPT} ${LIBSOCKET} ${LIBDL} ${CMAKE_THREAD_LIBS_INIT} ${LIBRT})
3892-
3893-- LIST(REMOVE_DUPLICATES CMAKE_REQUIRED_LIBRARIES)
3894-+ IF(CMAKE_REQUIRED_LIBRARIES)
3895-+ LIST(REMOVE_DUPLICATES CMAKE_REQUIRED_LIBRARIES)
3896-+ ENDIF()
3897- LINK_LIBRARIES(${CMAKE_THREAD_LIBS_INIT})
3898-
3899- OPTION(WITH_LIBWRAP "Compile with tcp wrappers support" OFF)
3900---- a/storage/innobase/sync/sync0arr.c
3901-+++ b/storage/innobase/sync/sync0arr.c
3902-@@ -41,6 +41,7 @@
3903- #include "os0file.h"
3904- #include "srv0srv.h"
3905- #include "ha_prototypes.h"
3906-+#include "xb0xb.h"
3907-
3908- /*
3909- WAIT ARRAY
3910-@@ -926,6 +927,13 @@
3911- ibool fatal = FALSE;
3912- double longest_diff = 0;
3913-
3914-+ if (srv_rebuild_indexes) {
3915-+
3916-+ /* Avoid long semaphore warnings when rebuilding indexes */
3917-+
3918-+ return(FALSE);
3919-+ }
3920-+
3921- #ifdef UNIV_DEBUG_VALGRIND
3922- /* Increase the timeouts if running under valgrind because it executes
3923- extremely slowly. UNIV_DEBUG_VALGRIND does not necessary mean that
3924---- a/storage/innobase/include/buf0buf.h
3925-+++ b/storage/innobase/include/buf0buf.h
3926-@@ -1509,6 +1509,8 @@
3927- frees a page in buffer pool */
3928- # endif /* UNIV_DEBUG_FILE_ACCESSES || UNIV_DEBUG */
3929- #endif /* !UNIV_HOTBACKUP */
3930-+ ibool is_compacted; /*!< TRUE if the page was skipped in
3931-+ compact backups */
3932- };
3933-
3934- /** The buffer control block structure */
3935---- a/storage/innobase/mem/mem0dbg.c
3936-+++ b/storage/innobase/mem/mem0dbg.c
3937-@@ -280,18 +280,10 @@
3938- byte* buf, /*!< in: pointer to buffer */
3939- ulint n) /*!< in: length of buffer */
3940- {
3941-- byte* ptr;
3942--
3943- UNIV_MEM_ASSERT_W(buf, n);
3944-
3945-- for (ptr = buf; ptr < buf + n; ptr++) {
3946--
3947-- if (ut_rnd_gen_ibool()) {
3948-- *ptr = 0xBA;
3949-- } else {
3950-- *ptr = 0xBE;
3951-- }
3952-- }
3953-+ /* Fix https://bugs.launchpad.net/percona-xtrabackup/+bug/1158154 */
3954-+ memset(buf, 0xBA, n);
3955-
3956- UNIV_MEM_INVALID(buf, n);
3957- }
3958-@@ -306,17 +298,10 @@
3959- byte* buf, /*!< in: pointer to buffer */
3960- ulint n) /*!< in: length of buffer */
3961- {
3962-- byte* ptr;
3963--
3964- UNIV_MEM_ASSERT_W(buf, n);
3965-
3966-- for (ptr = buf; ptr < buf + n; ptr++) {
3967-- if (ut_rnd_gen_ibool()) {
3968-- *ptr = 0xDE;
3969-- } else {
3970-- *ptr = 0xAD;
3971-- }
3972-- }
3973-+ /* Fix https://bugs.launchpad.net/percona-xtrabackup/+bug/1158154 */
3974-+ memset(buf, 0xDE, n);
3975-
3976- UNIV_MEM_FREE(buf, n);
3977- }
3978+--- a/storage/innobase/row/row0merge.c
3979++++ b/storage/innobase/row/row0merge.c
3980+@@ -2774,7 +2774,7 @@
3981+ ut_ad(indexes);
3982+ ut_ad(n_indexes);
3983+
3984+- merge_sort_block_size = thd_merge_sort_block_size(trx->mysql_thd);
3985++ merge_sort_block_size = 1UL << 20;
3986+
3987+ trx_start_if_not_started(trx);
3988+
3989
3990=== modified file 'src/Makefile'
3991--- src/Makefile 2013-05-18 03:04:06 +0000
3992+++ src/Makefile 2013-07-22 14:01:36 +0000
3993@@ -183,7 +183,7 @@
3994 usr0sess.o ut0byte.o ut0dbg.o \
3995 ut0list.o ut0mem.o ut0rnd.o \
3996 ut0ut.o ut0vec.o ut0wqueue.o \
3997- ut0rbt.o)
3998+ ut0rbt.o log0online.o)
3999 xtradb: MYSQLOBJS = $(addprefix $(MYSQL_ROOT_DIR)/, mysys/libmysys.a \
4000 strings/libmystrings.a zlib/.libs/libzlt.a dbug/libdbug.a)
4001 xtradb: DEFS += -DXTRADB_BASED
4002
4003=== modified file 'src/compact.cc'
4004--- src/compact.cc 2013-05-03 05:17:03 +0000
4005+++ src/compact.cc 2013-07-22 14:01:36 +0000
4006@@ -815,7 +815,7 @@
4007 index->to_be_dropped = 1;
4008
4009 /* Patch up any FK referencing this index with NULL */
4010- dict_foreign_replace_index(table, index, trx);
4011+ dict_foreign_replace_index(table, NULL, index);
4012
4013 dict_index_remove_from_cache(table, index);
4014
4015
4016=== modified file 'src/innodb_int.cc'
4017--- src/innodb_int.cc 2013-04-27 12:08:11 +0000
4018+++ src/innodb_int.cc 2013-07-22 14:01:36 +0000
4019@@ -47,6 +47,25 @@
4020
4021 long innobase_mirrored_log_groups = 1;
4022
4023+/******************************************************//**
4024+Reads a specified log segment to a buffer. */
4025+UNIV_INTERN
4026+void
4027+xb_log_group_read_log_seg(
4028+/*===================*/
4029+ ulint type, /*!< in: LOG_ARCHIVE or LOG_RECOVER */
4030+ byte* buf, /*!< in: buffer where to read */
4031+ log_group_t* group, /*!< in: log group */
4032+ ib_uint64_t start_lsn, /*!< in: read area start */
4033+ ib_uint64_t end_lsn) /*!< in: read area end */
4034+{
4035+#if defined(XTRADB_BASED) && !defined(XTRADB55)
4036+ log_group_read_log_seg(type, buf, group, start_lsn, end_lsn, FALSE);
4037+#else
4038+ log_group_read_log_seg(type, buf, group, start_lsn, end_lsn);
4039+#endif
4040+}
4041+
4042 /****************************************************************//**
4043 A simple function to open or create a file.
4044 @return own: handle to the file, not defined if error, error number
4045@@ -244,7 +263,7 @@
4046 if (!ret) {
4047 msg("xtrabackup: cannot set size for file %s\n", path);
4048 os_file_close(*file);
4049- os_file_delete(path);
4050+ xb_file_delete(path);
4051 return ret;
4052 }
4053
4054@@ -289,7 +308,7 @@
4055 msg("xtrabackup: could not write the first page to %s\n",
4056 path);
4057 os_file_close(*file);
4058- os_file_delete(path);
4059+ xb_file_delete(path);
4060 return ret;
4061 }
4062
4063@@ -334,6 +353,43 @@
4064 /* do nothing */
4065 }
4066
4067+my_bool
4068+innobase_check_identifier_length(
4069+/*=============================*/
4070+ const char* id __attribute__((unused)))
4071+{
4072+ msg("xtrabackup: innobase_check_identifier_length() called.\n");
4073+ ut_a(0);
4074+ return(false);
4075+}
4076+
4077+uint
4078+innobase_convert_to_filename_charset(
4079+/*=================================*/
4080+ char* to __attribute__((unused)),
4081+ const char* from __attribute__((unused)),
4082+ ulint len __attribute__((unused)))
4083+{
4084+ msg("xtrabackup: innobase_convert_to_filename_charset() called.\n");
4085+ ut_a(0);
4086+
4087+ return(0);
4088+}
4089+
4090+uint
4091+innobase_convert_to_system_charset(
4092+/*=================================*/
4093+ char* to __attribute__((unused)),
4094+ const char* from __attribute__((unused)),
4095+ ulint len __attribute__((unused)),
4096+ uint* errors __attribute__((unused)))
4097+{
4098+ msg("xtrabackup: innobase_convert_to__charset() called.\n");
4099+ ut_a(0);
4100+
4101+ return(0);
4102+}
4103+
4104 #if MYSQL_VERSION_ID >= 50500
4105
4106 /**********************************************************************//**
4107
4108=== modified file 'src/innodb_int.h'
4109--- src/innodb_int.h 2013-05-04 05:26:19 +0000
4110+++ src/innodb_int.h 2013-07-22 14:01:36 +0000
4111@@ -144,6 +144,8 @@
4112 fil_rename_tablespace(old_name_in, id, new_name, new_path)
4113 # define xb_btr_root_block_get(index, mode, mtr) \
4114 btr_root_block_get(index, mode, mtr)
4115+# define xb_file_delete(path) \
4116+ os_file_delete_func(path)
4117 typedef ib_mutex_t mutex_t;
4118 # define INNODB_LOG_DIR srv_log_group_home_dir
4119 # define DEFAULT_LOG_FILE_SIZE 48*1024*1024
4120@@ -154,6 +156,7 @@
4121 # define xb_os_event_create(name) os_event_create()
4122 #else /* MYSQL_VERSION_ID >= 50600 */
4123 /* MySQL and Percona Server 5.1 and 5.5 */
4124+# define xb_file_delete(path) os_file_delete(path)
4125 # define xb_buf_page_is_corrupted(page, zip_size) \
4126 buf_page_is_corrupted(page, zip_size)
4127 # define xb_os_event_create(name) os_event_create(name)
4128@@ -360,6 +363,7 @@
4129 .ibd file of tablespace and want to
4130 stop temporarily posting of new i/o
4131 requests on the file */
4132+#if MYSQL_VERSION_ID < 50500
4133 ibool stop_ibuf_merges;
4134 /*!< we set this TRUE when we start
4135 deleting a single-table tablespace */
4136@@ -370,6 +374,20 @@
4137 or flush requests can be placed on this space,
4138 though there may be such requests still being
4139 processed on this space */
4140+#else
4141+ ibool stop_new_ops;
4142+ /*!< we set this TRUE when we start
4143+ deleting a single-table tablespace.
4144+ When this is set following new ops
4145+ are not allowed:
4146+ * read IO request
4147+ * ibuf merge
4148+ * file flush
4149+ Note that we can still possibly have
4150+ new write operations because we don't
4151+ check this flag when doing flush
4152+ batches. */
4153+#endif
4154 ulint purpose;/*!< FIL_TABLESPACE, FIL_LOG, or
4155 FIL_ARCH_LOG */
4156 UT_LIST_BASE_NODE_T(fil_node_t) chain;
4157@@ -746,14 +764,14 @@
4158 lsn_t* max_flushed_lsn,/* out: */
4159 ulint* sum_of_new_sizes);/* out: sum of sizes of the new files added
4160 */
4161-
4162+#ifndef XTRADB_BASED
4163 ibool
4164 log_block_checksum_is_ok_or_old_format(
4165 /*===================================*/
4166 /* out: TRUE if ok, or if the log block may be in the
4167 format of InnoDB version < 3.23.52 */
4168 byte* block); /* in: pointer to a log block */
4169-
4170+#endif /* XTRADB_BASED */
4171 #endif /* MYSQL_VERSION_ID >= 50600 */
4172
4173 buf_block_t*
4174@@ -801,6 +819,18 @@
4175 } /* extern "C" */
4176 #endif
4177
4178+/******************************************************//**
4179+Reads a specified log segment to a buffer. */
4180+UNIV_INTERN
4181+void
4182+xb_log_group_read_log_seg(
4183+/*===================*/
4184+ ulint type, /*!< in: LOG_ARCHIVE or LOG_RECOVER */
4185+ byte* buf, /*!< in: buffer where to read */
4186+ log_group_t* group, /*!< in: log group */
4187+ ib_uint64_t start_lsn, /*!< in: read area start */
4188+ ib_uint64_t end_lsn); /*!< in: read area end */
4189+
4190 /****************************************************************//**
4191 A simple function to open or create a file.
4192 @return own: handle to the file, not defined if error, error number
4193@@ -967,6 +997,26 @@
4194
4195 #if MYSQL_VERSION_ID < 50600
4196
4197+my_bool
4198+innobase_check_identifier_length(
4199+/*=============================*/
4200+ const char* id);
4201+
4202+uint
4203+innobase_convert_to_filename_charset(
4204+/*=================================*/
4205+ char* to,
4206+ const char* from,
4207+ ulint len);
4208+
4209+uint
4210+innobase_convert_to_system_charset(
4211+/*=================================*/
4212+ char* to,
4213+ const char* from,
4214+ ulint len,
4215+ uint* errors);
4216+
4217 void
4218 innobase_invalidate_query_cache(
4219 trx_t* trx,
4220
4221=== modified file 'src/xtrabackup.cc'
4222--- src/xtrabackup.cc 2013-07-10 12:29:40 +0000
4223+++ src/xtrabackup.cc 2013-07-22 14:01:36 +0000
4224@@ -1367,6 +1367,11 @@
4225 }
4226 #endif
4227
4228+#ifdef XTRADB_BASED
4229+ /* Workaround for LP #1203308. */
4230+ srv_track_changed_pages = FALSE;
4231+#endif
4232+
4233 return(FALSE);
4234
4235 error:
4236@@ -2056,8 +2061,8 @@
4237
4238 mutex_enter(&log_sys->mutex);
4239
4240- log_group_read_log_seg(LOG_RECOVER, log_sys->buf,
4241- group, start_lsn, end_lsn);
4242+ xb_log_group_read_log_seg(LOG_RECOVER, log_sys->buf,
4243+ group, start_lsn, end_lsn);
4244
4245
4246 /* reference recv_scan_log_recs() */
4247@@ -3487,11 +3492,7 @@
4248 for (;;) {
4249 mtr_start(&local_mtr);
4250
4251-#if (MYSQL_VERSION_ID < 50517)
4252- local_block = btr_block_get(space_id, zip_size, page_no, RW_S_LATCH, &local_mtr);
4253-#else
4254 local_block = btr_block_get(space_id, zip_size, page_no, RW_S_LATCH, index, &local_mtr);
4255-#endif
4256 local_page = buf_block_get_frame(local_block);
4257 blob_header = local_page + offset;
4258 #define BTR_BLOB_HDR_PART_LEN 0
4259@@ -3532,13 +3533,8 @@
4260 mtr_commit(&mtr);
4261 if (right_page_no != FIL_NULL) {
4262 mtr_start(&mtr);
4263-#if (MYSQL_VERSION_ID < 50517)
4264- block = btr_block_get(space, zip_size, right_page_no,
4265- RW_X_LATCH, &mtr);
4266-#else
4267 block = btr_block_get(space, zip_size, right_page_no,
4268 RW_X_LATCH, index, &mtr);
4269-#endif
4270 page = buf_block_get_frame(block);
4271 goto loop;
4272 }
4273@@ -4565,7 +4561,7 @@
4274 if (!table) {
4275 snprintf(name, FN_REFLEN, "%s/%s/%s", data_home_dir,
4276 db_name, file_name);
4277- return os_file_delete(name);
4278+ return xb_file_delete(name);
4279 }
4280
4281 return(TRUE);
4282
4283=== modified file 'utils/build.sh'
4284--- utils/build.sh 2013-06-05 08:39:54 +0000
4285+++ utils/build.sh 2013-07-22 14:01:36 +0000
4286@@ -2,9 +2,9 @@
4287
4288 set -e
4289
4290-MYSQL_51_VERSION=5.1.59
4291-MYSQL_55_VERSION=5.5.17
4292-MYSQL_56_VERSION=5.6.10
4293+MYSQL_51_VERSION=5.1.70
4294+MYSQL_55_VERSION=5.5.31
4295+MYSQL_56_VERSION=5.6.11
4296 PS_51_VERSION=5.1.59-13.0
4297 PS_55_VERSION=5.5.16-22.0
4298
4299@@ -125,10 +125,11 @@
4300 ################################################################################
4301 function unpack_and_patch()
4302 {
4303+ local dirname=`basename "$1" ".tar.gz"`
4304 tar xzf $top_dir/$1
4305- cd `basename "$1" ".tar.gz"`
4306+ cd $dirname
4307 patch -p1 < $top_dir/patches/$2
4308- cd ..
4309+ cd -
4310 }
4311
4312 ################################################################################
4313@@ -215,8 +216,8 @@
4314 {
4315 local type=$1
4316 local mysql_version_short=${mysql_version:0:3}
4317- server_dir=$top_dir/mysql-$mysql_version_short
4318- server_tarball=mysql-$mysql_version.tar.gz
4319+ local dirname=`basename "$server_tarball" ".tar.gz"`
4320+
4321 innodb_dir=$server_dir/storage/$innodb_name
4322
4323 echo "Downloading sources"
4324@@ -226,7 +227,7 @@
4325
4326 echo "Preparing sources"
4327 unpack_and_patch $server_tarball $server_patch
4328- mv $top_dir/mysql-$mysql_version $server_dir
4329+ mv $top_dir/$dirname $server_dir
4330
4331 build_server $type
4332
4333@@ -241,10 +242,12 @@
4334
4335 case "$type" in
4336 "innodb51")
4337+ server_dir=$top_dir/mysql-5.1
4338 mysql_version=$MYSQL_51_VERSION
4339 server_patch=innodb51.patch
4340 innodb_name=innodb_plugin
4341 xtrabackup_target=plugin
4342+ server_tarball=mysql-5.1.70.tar.gz
4343 configure_cmd="./configure --enable-local-infile \
4344 --enable-thread-safe-client \
4345 --with-plugins=innodb_plugin \
4346@@ -255,10 +258,12 @@
4347 build_all $type
4348 ;;
4349 "innodb55")
4350+ server_dir=$top_dir/mysql-5.5
4351 mysql_version=$MYSQL_55_VERSION
4352 server_patch=innodb55.patch
4353 innodb_name=innobase
4354 xtrabackup_target=5.5
4355+ server_tarball=mysql-5.5.31.tar.gz
4356 # We need to build with partitioning due to MySQL bug #58632
4357 configure_cmd="cmake . \
4358 -DENABLED_LOCAL_INFILE=ON \
4359@@ -272,10 +277,12 @@
4360 ;;
4361
4362 "innodb56" )
4363+ server_dir=$top_dir/mysql-5.6
4364 mysql_version=$MYSQL_56_VERSION
4365 server_patch=innodb56.patch
4366 innodb_name=innobase
4367 xtrabackup_target=5.6
4368+ server_tarball=mysql-5.6.11.tar.gz
4369 configure_cmd="cmake . \
4370 -DWITH_INNOBASE_STORAGE_ENGINE=ON \
4371 -DWITH_ZLIB=bundled \
4372@@ -288,8 +295,10 @@
4373 "xtradb51" )
4374 server_dir=$top_dir/Percona-Server
4375 branch_dir=percona-server-5.1-xtrabackup
4376- innodb_dir=$server_dir/storage/innodb_plugin
4377- xtrabackup_target=xtradb
4378+ innodb_name=innodb_plugin
4379+ xtrabackup_target=xtradb
4380+ server_patch=xtradb51.patch
4381+ server_tarball=Percona-Server-5.1.70-rel14.8.tar.gz
4382 configure_cmd="./configure --enable-local-infile \
4383 --enable-thread-safe-client \
4384 --with-plugins=innodb_plugin \
4385@@ -301,42 +310,15 @@
4386 configure_cmd="LIBS=-lrt $configure_cmd"
4387 fi
4388
4389-
4390- echo "Downloading sources"
4391-
4392- # Get Percona Server
4393- if [ -d $branch_dir ]
4394- then
4395- rm -rf $branch_dir
4396- fi
4397-
4398- if [ ! -f Percona-Server-XtraBackup-$PS_51_VERSION.tar.gz ]
4399- then
4400- make ps51source
4401- fi
4402-
4403- rm -rf $branch_dir
4404- tar xfz Percona-Server-XtraBackup-$PS_51_VERSION.tar.gz
4405- cd $branch_dir
4406- $MAKE_CMD main
4407- cd $top_dir
4408- rm -rf $server_dir
4409- ln -s $branch_dir/Percona-Server $server_dir
4410-
4411- # Patch Percona Server
4412- cd $server_dir
4413- patch -p1 < $top_dir/patches/xtradb51.patch
4414-
4415- build_server $type
4416-
4417- build_xtrabackup
4418-
4419+ build_all $type
4420 ;;
4421 "xtradb55" )
4422 server_dir=$top_dir/Percona-Server-5.5
4423 branch_dir=percona-server-5.5-xtrabackup
4424- innodb_dir=$server_dir/storage/innobase
4425- xtrabackup_target=xtradb55
4426+ innodb_name=innobase
4427+ xtrabackup_target=xtradb55
4428+ server_patch=xtradb55.patch
4429+ server_tarball=Percona-Server-5.5.31-rel30.3.tar.gz
4430 # We need to build with partitioning due to MySQL bug #58632
4431 configure_cmd="cmake . \
4432 -DENABLED_LOCAL_INFILE=ON \
4433@@ -350,37 +332,7 @@
4434 configure_cmd="LIBS=-lrt $configure_cmd"
4435 fi
4436
4437-
4438- echo "Downloading sources"
4439-
4440- # Get Percona Server
4441- if [ -d $branch_dir ]
4442- then
4443- rm -rf $branch_dir
4444- fi
4445-
4446- if [ ! -f Percona-Server-XtraBackup-$PS_55_VERSION.tar.gz ]
4447- then
4448- make ps55source
4449- fi
4450-
4451- rm -rf $branch_dir
4452- tar xfz Percona-Server-XtraBackup-$PS_55_VERSION.tar.gz
4453- cd $branch_dir
4454-
4455- $MAKE_CMD PERCONA_SERVER=Percona-Server-5.5 main
4456- cd $top_dir
4457- rm -rf $server_dir
4458- ln -s $branch_dir/Percona-Server $server_dir
4459-
4460- # Patch Percona Server
4461- cd $server_dir
4462- patch -p1 < $top_dir/patches/xtradb55.patch
4463-
4464- build_server $type
4465-
4466- build_xtrabackup
4467-
4468+ build_all $type
4469 ;;
4470 *)
4471 usage

Subscribers

People subscribed via source and target branches

to all changes: