alexpeshkoff
e79f55708d
1. Cleanup - use Firebird::Semaphore instead of local events, removed code
...
related with local events from isc_sync.cpp.
2. Cleanup - removed unused parameter 'number of events' from ISC_event_wait()
and two unused parameters from ISC_event_init().
3. Added SYSV-semaphore based implementation of IPC to isc_sync.cpp.
This should fix CORE-2102: Firebird 2.5 does not build on MacOS (Darwin).
4. Fixed use of posic CS without (x)inetd (i.e. fb_inet_server -s).
5. Fixed (to be reviewed by Dmitry) recovery after some process death in lock manager.
2008-10-10 15:58:05 +00:00
hvlad
97bf925499
Use fb_utils::init_status instead of status[0] = 0
2008-09-10 11:41:36 +00:00
alexpeshkoff
13a15d4bfe
1. Final fix for CORE-1964 - finished vararg cleanup in jrd.
...
2. Generic cleanup of error reporting functions all around engine.
2008-08-27 12:20:47 +00:00
roman-simakov
59ae35d837
Style
2008-08-13 09:02:35 +00:00
hvlad
53f21b34c8
Fixed bug CORE-1970 : Lock conversion denied (215) error may occurs
2008-07-17 12:49:19 +00:00
alexpeshkoff
a2e416210b
Fixed CORE-1964: Use of 0 instead of isc_arg_end to terminate list of ERR_post's (and cousines) arguments is invalid.
...
Also cleaned up use of gds__log() in a few places.
2008-06-26 09:47:59 +00:00
asfernandes
4822c30c70
Misc
2008-06-06 22:40:15 +00:00
robocop
94224d3300
Totally misc.
2008-06-06 08:51:03 +00:00
robocop
b8ea332022
Misc.
2008-06-05 11:02:42 +00:00
hvlad
6bff6c6649
Replace fixed size array of shared latches (bdb_shared) by queue. It allows to have more than 20 simultaneous readers for hot pages and eliminate needs for scanning full array when searching for current thread's latch
2008-06-05 08:35:18 +00:00
dimitr
ad8ce57050
Enabled page latching code for [Super]Classic. This solves most of the synchronization issues between worker thread and ASTs.
2008-06-04 13:22:59 +00:00
dimitr
f84ad12e33
Misc.
2008-05-14 09:21:01 +00:00
robocop
89a57960db
Cleanup.
2008-05-11 12:27:51 +00:00
hvlad
422338e5e3
Lock att_mutex in Attachment's constructor
2008-05-08 07:45:19 +00:00
dimitr
0a9e8c8382
Implemented CORE-1751. The changes include:
...
1) Make memory counters aggregated.
2) Add attachment pool and allocate appropriate resources out of this pool.
3) Always release attachments explicitly (via destructor).
4) Always delete user requests prior to attachment deletion.
5) Introduce memory usage counters per every monitoring object.
6) Misc refactoring.
Some pieces are still incomplete (although everything basically works), but I'd like to get feedback and testing sooner rather than later.
2008-05-06 08:46:39 +00:00
hvlad
adc1ed0f03
Let cache_writer also lock its own attachment during execution
2008-05-04 21:25:56 +00:00
skidder
3d8af77572
Fix regression CORE-1876. Incremental backups with NBACKUP broken in 2.1
2008-05-01 01:40:22 +00:00
robocop
53bf05282d
Misc.
2008-04-26 14:35:12 +00:00
roman-simakov
dda2676ba5
1. Fixed unlocking header page while CCH_release. Thanks to Vlad.
...
2. Fixed writing page while nbackup merge state is stalled.
3. Fixed using dead dbb in AST handler while somebody is changing cached lock of the GlobalRWLock.
2008-04-16 09:16:45 +00:00
roman-simakov
a202daf6bb
Replaced nbackup dirty pages locking with RAII classes
2008-04-16 08:46:13 +00:00
dimitr
2f41d8e8ba
Corrections for the currently disabled code.
2008-04-02 12:42:09 +00:00
robocop
882955f5c3
const.
2008-03-29 13:01:49 +00:00
asfernandes
a930424d27
Fix the build
2008-03-19 15:55:11 +00:00
dimitr
26e92220d3
Cleanup/refactoring.
2008-03-19 11:46:47 +00:00
dimitr
ebad313cfd
1) Fixed CORE-1788.
...
2) Protected from other similar deadlocks.
3) Refactoring.
2008-03-13 14:20:32 +00:00
dimitr
afb970dba1
1) Cleanup.
...
2) Catch exceptions in AST routines.
2008-03-12 16:53:57 +00:00
robocop
2912e6a5be
Cleanup and misc.
2008-03-10 08:44:07 +00:00
robocop
10d74a7642
Style.
2008-02-13 12:57:04 +00:00
asfernandes
11b869588d
Misc
2008-02-08 01:25:41 +00:00
hvlad
078f55d9b0
Fixed bug CORE-1731 : Sometimes engine may "hang" using 100% CPU load and no IO activity for the few minutes
2008-02-06 17:06:43 +00:00
robocop
4c3381ef6b
Style, constness and macro cleanup.
2008-02-03 10:41:44 +00:00
asfernandes
de8771570f
Misc
2008-01-29 01:36:41 +00:00
dimitr
132176a475
Major synchronization changes. Description of the commit will be posted to fb-devel.
2008-01-26 12:52:51 +00:00
dimitr
db7d3ecd4c
The merge continued.
2008-01-16 08:31:31 +00:00
roman-simakov
406a8cc3bf
small fix debug code under NBAK_DEBUG
2008-01-14 16:22:43 +00:00
asfernandes
41faeba3cc
Misc
2007-12-05 00:03:15 +00:00
alexpeshkoff
eea9070f48
Optimized RuntimeStaticstics (and related part of ThreadDb) for performance - some tests run 25faster
2007-12-03 15:46:39 +00:00
dimitr
1a64b5214c
Fixed CORE-1601.
2007-11-19 10:29:59 +00:00
dimitr
0e6c8c9abe
Moved the engine checkout/checkin logic from CCH to PIO. This allows to avoid hacks like usage of SCH_check_thread_enter().
2007-09-25 11:47:39 +00:00
hvlad
a576160b7a
Restore correct error message formating
2007-09-19 20:35:33 +00:00
dimitr
97a45d4704
Removed the asserts.
2007-09-05 06:44:12 +00:00
alexpeshkoff
6d66c28b13
Fixed solx86 build - not always offset_t and UINT64 are the same
2007-08-10 07:55:46 +00:00
alexpeshkoff
ea8ba3370d
Cleaned up mutexes - removed MUTX_T and related macros
2007-06-06 12:37:24 +00:00
mkubecek
9aa6368cc2
C++ specification does not allow unbraced linkage specification with storage class (and neither does GCC 4.1.2)
2007-05-20 12:34:36 +00:00
asfernandes
97843da8a3
Misc
2007-05-20 03:25:52 +00:00
robocop
5f3d9bbd54
Misc.
2007-05-18 06:15:26 +00:00
hvlad
18e0b65d96
Add missed THREAD_ENTER calls
2007-05-17 08:31:03 +00:00
hvlad
f40e794951
Implement improved dirty pages flush algoritm
2007-05-17 07:17:11 +00:00
robocop
b7446d9c64
Misc.
2007-05-16 06:53:47 +00:00
dimitr
533e28ce6c
Fixed CORE-1249 plus minor corrections for the single-user shutdown mode.
...
The change in CCH allows to report isc_shutdown instead of the LM error code in Classic (GFIX depends on this behaviour).
Also, refactored the SHUT logic a bit to assert on unexpected shutdown modes.
2007-05-08 08:31:48 +00:00
hvlad
a640aee5de
Few small fixes :
...
Removed not needed includes
Stop tracking precedence for temporary pages
THREAD_EXIT\THREAD_ENTER while page writes into nbackup delta file
Temporary pages was not written to disk in BACKUP state (nbak_state_stalled)
2007-05-04 22:10:11 +00:00
skidder
8539ceb992
Small NBAK performance optimization from Rom
...
an Simakov
2007-05-04 21:25:04 +00:00
dimitr
512bb7abca
Misc.
2007-05-01 06:48:20 +00:00
hvlad
8befd31986
Continue Nickolay's refactor : use clear_page_dirty_flag
2007-04-27 06:49:55 +00:00
asfernandes
c722690150
Misc
2007-04-27 02:28:57 +00:00
skidder
f231dc3d1e
Cleanup as requested by Dmitry and Vlad
...
- do not write pages of temporary tables to the NBAK delta files (engine still does a lot of unneeded locking and synchronization for temp pages, especially in CS builds, but this is another problem)
- refactor and clean up potential problems in the logic of dirty page -> clean page transition
2007-04-26 20:16:48 +00:00
hvlad
31af74fbf6
Fixed bug CORE-1228 : Database often became corrupted after an out of disk space condition
2007-04-25 20:45:03 +00:00
dimitr
9dd5a0efca
Minor cleanup.
2007-04-25 06:32:54 +00:00
asfernandes
676b447138
Misc
2007-04-25 02:26:33 +00:00
skidder
04b9ec818b
NBAK synchronization rework. This should fix CORE-1151 and other Nbackup stability issues
2007-04-24 14:05:46 +00:00
alexpeshkoff
d63491e3c5
Finished gcc-warnings cleanup
2007-04-11 16:05:40 +00:00
alexpeshkoff
6e735f7c53
Misc + gcc warning hunting
2007-04-09 16:15:29 +00:00
robocop
e71a03261c
This is the full SafeArg implementation for getting rid of the fixed parameter types in the messages db. Alas, the legacy messages (before FB2.1) can't be changed in number of arguments neither the types of arguments sent by callers of the MsgPrint family or the fb_msg_format functions can be changed because old clients know these legacy messages and would crash.
2007-03-28 04:20:36 +00:00
alexpeshkoff
8b2abbdb65
fixed correct logic, broken by my last commit
2007-02-24 11:28:51 +00:00
robocop
63a80b0db6
Misc.
2007-02-23 05:55:37 +00:00
asfernandes
d87a30bf0c
Misc
2007-02-23 01:42:10 +00:00
alexpeshkoff
dfef55e991
Fix for CORE-968 - Assertion failure in posix/isc_ipc.cpp
2007-02-22 16:47:39 +00:00
dimitr
15c3681d8a
Fixed CORE-1135.
2007-02-22 14:12:04 +00:00
robocop
409ff041dd
Misc and replaced "IB" by "FB" where it made sense to me.
2007-02-11 09:04:54 +00:00
dimitr
fd7a702e7d
Optimization. Removed redundant NBAK calls (they are pretty expensive).
2007-02-08 14:01:53 +00:00
asfernandes
90337587ff
Misc
2007-02-08 00:03:36 +00:00
dimitr
ecc261de7c
Cleanup.
2007-02-06 08:03:43 +00:00
dimitr
f042dc9f8f
Fixed CORE-1121. This time the fix itself is separated from the cleanup to make the reviewers happy.
...
The changes are identical to my previous commit into B2_0_Release.
2007-02-06 07:47:59 +00:00
alexpeshkoff
a13e291d5f
front-ported fox for CORE-1098
2007-01-18 15:50:18 +00:00
dimitr
5af8f3c3a9
Fixed CORE-1039.
2006-12-04 21:26:43 +00:00
dimitr
988782a96a
1) A new class to handle aggregated multi-level performance counters.
...
2) Cleanup and minor fixing.
3) A few monitoring changes.
2006-10-07 10:53:01 +00:00
robocop
7040d668fd
Misc.
2006-05-25 08:40:23 +00:00
dimitr
211645e3dc
Getting rid of the old allocator routines.
2006-05-23 10:17:00 +00:00
hvlad
701f2cbda4
Global temporary tables implementation
2006-05-21 22:07:35 +00:00
robocop
e77a46b011
Misc.
2006-05-20 03:55:54 +00:00
asfernandes
292ad66e96
Misc
2006-05-20 01:53:51 +00:00
alexpeshkoff
b056a118b5
1. Avoid use of STL
...
2. Reworked exceptions to make status_exception always hold correct status vector
3. Avoid use of RTTI
2006-05-19 15:17:02 +00:00
robocop
c5f2284b15
Style.
2006-04-06 08:18:53 +00:00
dimitr
233aef595a
gds_alloc() no longer throws exceptions. Let's handle it properly.
2006-04-05 04:36:23 +00:00
robocop
1f84bac700
Mark unused data members, discover more code associated to conditional compilation, clean some places, replace custom loops by memcpy and memset when possible, etc.
2006-02-23 05:08:26 +00:00
robocop
97122ee389
Replace jrd_time.h by our timestamp.h, but I left timestamp.h's #includes commented where it *seemed* not necessary.
2006-02-03 08:39:36 +00:00
robocop
25c08aea67
The files rng.cpp, rng.h, rng_proto.h, bookmark.cpp and bookmark.h belong exclusively to PC_ENGINE, that's obsolete functionality.
...
Eighth step: dbb_refresh_ranges belongs to the obsolete functionality and since it's never set but tested in cch.cpp with two AND, the whole condition will be always false, hence remove the code.
2006-02-02 07:57:36 +00:00
dimitr
a5a967c648
Little cleanup.
2006-01-05 10:10:37 +00:00
hvlad
ddc489f6f0
Restored accidentally committed wrong file
2005-12-26 21:17:16 +00:00
hvlad
52086cd806
Avoid frequent (unneeded and inefficient) allocation \ deallocation of stack entryes
2005-12-26 21:13:03 +00:00
hvlad
9d6b361f69
Eliminate unnecessary JRD_get_thread_data call in nbak.
...
Checked against B2_0_BV_Stable3 branch
2005-11-21 23:33:20 +00:00
skidder
699c914cbf
little debug logging improvements
2005-07-19 23:27:38 +00:00
hvlad
3e5ce4257d
Temporary fix to prevent 100% cpu load in case of write failure
2005-02-17 07:29:50 +00:00
robocop
8d4b89723f
Welcome to paranoid programming.
2005-01-25 06:33:07 +00:00
dimitr
4ef8631580
Cleanup.
2005-01-24 07:58:34 +00:00
robocop
af6a16107a
Misc.
2004-11-30 06:18:39 +00:00
robocop
d6535351a2
Misc: style.
2004-11-24 09:22:07 +00:00
robocop
628b088d2c
Treat string literals and string constants with more respect.
2004-11-10 04:26:45 +00:00
alexpeshkoff
6e89d3e6d8
avoid warnings in gcc
2004-11-07 14:30:38 +00:00
dimitr
421f259a4e
1) Locks related cleanup
...
2) Got rid of TRA_nowait + transaction waits related cleanup
3) Return isc_lock_timeout when it's necessary
TODO: Enhance SET TRANSACTION
2004-10-07 09:15:32 +00:00
