Leif Walsh
|
32698fa975
|
refs #5672 #5351 increase drd timeouts some more (ugh), make locktree stop reporting false positives
git-svn-id: file:///svn/toku/tokudb@51378 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:25 -04:00 |
|
John Esmet
|
2ce44e52ee
|
refs #5828 destroy the root node when the concurrent tree is destroyed.
git-svn-id: file:///svn/toku/tokudb@51345 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:24 -04:00 |
|
John Esmet
|
33a99affe4
|
fixes #5771 merge the single txnid optimization to main. single threaded write performance is up 20-50% in mysql and multithreaded performance is largely unchanged.
git-svn-id: file:///svn/toku/tokudb@51108 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:23 -04:00 |
|
John Esmet
|
c0f791cb4e
|
refs #5779 rebalance on the way down in treenode::insert(), add a test that makes sure the concurrent tree can survive 128k serial key insertions.
git-svn-id: file:///svn/toku/tokudb@50986 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:22 -04:00 |
|
John Esmet
|
2f8f9ff560
|
refs #5776 merge a fix for 5776 to main. will resolve as closed once 5771 can survive stress tests for 24 hours or so.
git-svn-id: file:///svn/toku/tokudb@50967 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:22 -04:00 |
|
Leif Walsh
|
032ed0c0ca
|
refs #5742 #5672 suppress more valgrind stuff
git-svn-id: file:///svn/toku/tokudb@50720 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:21 -04:00 |
|
John Esmet
|
a879883311
|
refs #5351 remove kruft
git-svn-id: file:///svn/toku/tokudb@50479 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:19 -04:00 |
|
John Esmet
|
689a93f5ff
|
fixes #5701 The assert half correct, half over-active. When the manager's mutex is held it is okay to transition an existing locktree's reference count from 0 to 1. If the lock isn't held, it's a race, since one thread could be destroying the zero-ref count locktree when you're incrementing the count. The failure occurred with the manager's mutx held (so the assert was overactive) and it's a layering-pain to assert that the manager's mutex is held otherwise, so just get rid of it.
git-svn-id: file:///svn/toku/tokudb@50297 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:18 -04:00 |
|
John Esmet
|
3ebf117c33
|
refs #5351 correct corner case that was causing +inf, +inf point to not be covered by unit test
git-svn-id: file:///svn/toku/tokudb@50109 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:17 -04:00 |
|
John Esmet
|
b15afefe89
|
refs #5351 fix an overactive assert. just ignore the bit for now because the lock request layer does not know that write locks do not realy exist.
git-svn-id: file:///svn/toku/tokudb@49962 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:16 -04:00 |
|
John Esmet
|
43d1d20ae6
|
fix DRD failures in the threaded stress framework and locktree
git-svn-id: file:///svn/toku/tokudb@49960 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:16 -04:00 |
|
John Esmet
|
3556f22396
|
refs #5351 put this in the race tools header, make sure it does not break a release build
git-svn-id: file:///svn/toku/tokudb@49925 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:16 -04:00 |
|
John Esmet
|
85594f4e49
|
refs #5351 fix various drd and helgrind related issues. tree rotations should no longer be problematic with race tools.
git-svn-id: file:///svn/toku/tokudb@49920 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:16 -04:00 |
|
John Esmet
|
d61566864e
|
refs #5351 set the mutex to zero first
git-svn-id: file:///svn/toku/tokudb@49918 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:16 -04:00 |
|
John Esmet
|
6ee5fff9f7
|
refs #5351 fix test
git-svn-id: file:///svn/toku/tokudb@49917 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:16 -04:00 |
|
John Esmet
|
f0a8035a35
|
refs #5351 remove printf
git-svn-id: file:///svn/toku/tokudb@49916 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:16 -04:00 |
|
John Esmet
|
d49dd4f15b
|
refs #5351 have the locktree link with portability and ft and make sure a printf uses a portable format specifier
git-svn-id: file:///svn/toku/tokudb@49888 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:16 -04:00 |
|
Leif Walsh
|
ffe2a5d87c
|
refs #5351 fix mutexattr stuff for darwin
git-svn-id: file:///svn/toku/tokudb@49885 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:16 -04:00 |
|
John Esmet
|
bcd522eac0
|
refs #5351 propset id on lock_request.h
git-svn-id: file:///svn/toku/tokudb@49866 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:15 -04:00 |
|
John Esmet
|
12803248af
|
refs #5351 do not use the static initializer in a dynamic context, because it confuses tools like helgrind which look for create/destroy pairs to validate mutex handling
git-svn-id: file:///svn/toku/tokudb@49865 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:15 -04:00 |
|
Leif Walsh
|
a67344b336
|
refs #5351 tighten up acquire_lock after cleanup
git-svn-id: file:///svn/toku/tokudb@49863 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:15 -04:00 |
|
Leif Walsh
|
c08bafb348
|
refs #5351 clang compilation cleanup after 5351 merge
git-svn-id: file:///svn/toku/tokudb@49862 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:15 -04:00 |
|
John Esmet
|
65cd284834
|
fixes #5351 fixes #5528 merge the new locktree to main. the locktree, locktree manager, and blocking lock request api are implemented at the toplevel under locktree/ and the ydb layer has been modified to use the new API. some kruft has been deleted from various parts of the source tree and these merge also includes a change where we kill 'includes.h'
git-svn-id: file:///svn/toku/tokudb@49851 c7de825b-a66e-492c-adef-691d508d4ae1
|
2013-04-17 00:01:15 -04:00 |
|