Commit graph

86430 commits

Author SHA1 Message Date
Rich Prohaska
507da246c8 DB-718 test case for broken error handling in tokudb_fractal_tree_info info schema plugin 2014-09-03 08:13:49 -04:00
Rich Prohaska
b7feee730c Merge branch 'master' of github.com:Tokutek/tokudb-engine 2014-09-03 07:50:03 -04:00
Rich Prohaska
f7c43e4b9c DB-718 handle errors in tokudb info schema plugins 2014-09-03 07:49:52 -04:00
Jan Lindström
454037b794 MDEV-6682 innodb.innodb_simulate_comp_failures_small is too slow
if it's run on a real disk
  
  Made test smaller.
2014-09-03 07:37:13 +03:00
Rich Prohaska
ddc9e7488a DB-717 add iterations limit to tokustat 2014-09-02 16:23:32 -04:00
Rich Prohaska
d4fb687de3 Merge branch 'master' of github.com:Tokutek/tokudb-engine 2014-09-02 15:44:49 -04:00
Rich Prohaska
561edb14e8 DB-717 add iterations limit to tokustat 2014-09-02 15:44:40 -04:00
Rich Prohaska
6ee0e1c22e DB-716 use jemalloc 3.6.0 in tokudb builds 2014-09-02 13:01:56 -04:00
Rich Prohaska
9d28e3a6cf FT-548 fix dirty upgrade build on osx problem 2014-09-02 12:29:55 -04:00
Rich Prohaska
2f04fc27c8 FT-548 fix dirty upgrade build on osx problem 2014-09-02 11:53:49 -04:00
Rich Prohaska
ed9df111a5 FT-548 support upgrade after dirty shutdown of versions 25 through 27 2014-09-02 09:55:24 -04:00
Sergei Golubchik
6389fd3c79 MDEV-6673 I_S.SESSION_VARIABLES shows global values
only look at lex->option_type if it's a SHOW command, not a SELECT
2014-08-31 19:55:11 +02:00
Rich Prohaska
6827d3bd51 FT-312 compile big-shutdown on osx 2014-08-30 06:35:59 -04:00
Sergey Vojtovich
c01c819209 Backport from 10.0:
MDEV-6483 - Deadlock around rw_lock_debug_mutex on PPC64

This problem affects only debug builds on PPC64.

There are at least two race conditions around
rw_lock_debug_mutex_enter and rw_lock_debug_mutex_exit:

- rw_lock_debug_waiters was loaded/stored without setting
  appropriate locks/memory barriers.
- there is a gap between calls to os_event_reset() and
  os_event_wait() and in such case we're supposed to pass
  return value of the former to the latter.

Fixed by replacing self-cooked spinlocks with system mutexes.
These days system mutexes offer much better performance. OTOH
performance is not that critical for debug builds.
2014-08-29 16:14:11 +04:00
Sergey Vojtovich
40497577ff Backport from 10.0:
MDEV-6450 - MariaDB crash on Power8 when built with advance tool
            chain

InnoDB mutex_exit() function calls __sync_test_and_set() to release
the lock. According to manual this function is supposed to create
"acquire" memory barrier whereas in fact we need "release" memory
barrier at mutex_exit().

The problem isn't repeatable with gcc because it creates
"acquire-release" memory barrier for __sync_test_and_set().
ATC creates just "acquire" barrier.

Fixed by creating proper barrier at mutex_exit() by using
__sync_lock_release() instead of __sync_test_and_set().
2014-08-29 16:02:46 +04:00
Leif Walsh
31a941185b MX-1217 fix TokuMergeLibs to handle empty libs 2014-08-28 16:25:14 -04:00
Rich Prohaska
f7ee3d4a26 DB-712 fix tokudb locks info schema test results due to new schema 2014-08-28 15:11:55 -04:00
Rich Prohaska
967fcd6177 Merge branch 'db703' of github.com:Tokutek/tokudb-engine into db703 2014-08-28 06:19:59 -04:00
Rich Prohaska
08eb88eec4 DB-712 split locks and lock_waits dname into schema, table, and dictionary 2014-08-28 06:19:32 -04:00
Rich Prohaska
8c5cd26053 DB-703 DB-704 split dname into schema, table, and dictionary 2014-08-28 06:19:32 -04:00
Rich Prohaska
b585b4f71b DB-713 separate some long running tokudb tests so that valgrind runs without --big-test can exclude them 2014-08-27 18:00:57 -04:00
Rich Prohaska
5c4c580ef1 DB-713 separate some long running tokudb tests so that valgrind runs without --big-test can exclude them 2014-08-27 18:00:19 -04:00
Rich Prohaska
f704ecb91e DB-712 split locks and lock_waits dname into schema, table, and dictionary 2014-08-27 14:49:11 -04:00
Rich Prohaska
aed2c264b7 DB-703 DB-704 split dname into schema, table, and dictionary 2014-08-27 14:07:16 -04:00
Rich Prohaska
66504277bf DB-702 print upgrade failed clean shutdown required error message 2014-08-27 10:31:45 -04:00
John Esmet
8db687e6c7 FT-590 Calculate a node's weight using a 64 bit integer to prevent
overflow
2014-08-26 14:28:16 -04:00
Murthy Narkedimilli
b142bfd087 Renaming the enterprise packages to commercial 2014-08-26 14:01:38 +02:00
Sergei Golubchik
dd25e7f0ad MDEV-6601 Assertion `!thd->in_active_multi_stmt_transa ction() || thd->in_multi_stmt_transaction_mode()' failed on executing a stored procedure with commit
Don't restore the whole of thd->server_status after a routine invocation,
only restore SERVER_STATUS_CURSOR_EXISTS and SERVER_STATUS_LAST_ROW_SENT,
as --ps --embedded needs.
In particular, don't restore SERVER_STATUS_IN_TRANS.
2014-08-25 16:58:19 +02:00
Rich Prohaska
a0150feed6 FT-312 fix centos compile 2014-08-24 12:56:35 -04:00
Rich Prohaska
88a7ade688 FT-312 speedup shutdown by parallelizing compression 2014-08-24 08:45:19 -04:00
Rich Prohaska
b5dd1e7f5f FT-586 FT-563 change ft-verify to work with promotion 2014-08-23 10:58:26 -04:00
Harin Vadodaria
b9f2b1c135 Bug#19370676 : YASSL PRE-AUTH BUFFER OVERFLOW WHEN CLIENT
LIES ABOUT SUITE_LEN_
               and
Bug#19355577 : YASSL PRE-AUTH BUFFER OVERFLOW WHEN CLIENT
               LIES ABOUT COMP_LEN_

Description : Updating yaSSL to version 2.3.4.
2014-08-23 08:59:03 +05:30
Rich Prohaska
61195094c7 FT-584 use trylock inside of the lock tree manager get_status function so that it is non-blocking 2014-08-22 07:57:47 -04:00
Rich Prohaska
ff5a25cca3 FT-584 use trylock inside of the lock tree manager get_status function so that it is non-blocking 2014-08-21 19:21:21 -04:00
John Esmet
4cb144d9fb Add common cmake build directory names to .gitignore 2014-08-21 16:29:50 -04:00
John Esmet
f176b29b5e FT-396 Rename various tokudb-containing functions and comments to
instead refer to tokuft
2014-08-21 15:35:07 -04:00
John Esmet
7316bacdeb FT-396 Rename TokuDB to TokuFT in each license header 2014-08-21 15:35:06 -04:00
John Esmet
d569f7a4c0 FT-513 Remove some windows cruft 2014-08-21 15:35:06 -04:00
John Esmet
a57387dfa7 FT-512 Remove the majority of the remaining BDB artifacts 2014-08-21 15:35:06 -04:00
John Esmet
feb5b70c26 FT-582 Remove remaining artifacts from the ICC days 2014-08-21 15:35:06 -04:00
John Esmet
1a34a1315b FT-581 Remove dead rdtsc code 2014-08-21 15:35:06 -04:00
John Esmet
addd9c3c59 FT-580 Remove unused scripts 2014-08-21 15:35:06 -04:00
John Esmet
674454b0f7 FT-583 Remove sub_block_map.h, dead code from old format verisons 2014-08-21 15:35:06 -04:00
John Esmet
9b0047636b FT-585 Move serialize and compression size calculations around so we can malloc
one large buffer to serialize a node instead of many smaller ones, which
should hopefully put less pressure on jemalloc during checkpoints etc.
2014-08-21 15:34:22 -04:00
John Esmet
effa06ec47 Clean up toku_ft_handle_close 2014-08-21 15:34:18 -04:00
John Esmet
5a18a1b046 FT-440 Use a scoped malloc during garbage collection and for checkpoint
cachefiles, also clean up some formatting.
2014-08-21 15:34:14 -04:00
Tor Didriksen
ab727cec04 Bug#18928848 II. MALLOC OF UNINITIALIZED MEMORY SIZE
Several string functions have optimizations for constant
sub-expressions which lead to setting max_length == 0.

For subqueries, where we need a temporary table to holde the result,
we need to ensure that we use a VARCHAR(0) column rather than a
CHAR(0) column when such expressions take part in grouping.
With CHAR(0) end_update() may write garbage into the next field.
2014-08-21 16:42:04 +02:00
Rich Prohaska
1826cca4c5 Tokutek/mariadb-5.5#71 run part_index_scan on mariadb 2014-08-20 11:22:26 -04:00
Rich Prohaska
56ed8eb5cf #261 run multiple queries in part_index_scan 2014-08-20 11:21:35 -04:00
Balasubramanian Kandasamy
c2fc7d39b2 Add my.cnf.d to regular rpm for EL7 build 2014-08-20 09:46:38 +02:00