Commit Graph

405 Commits

Author SHA1 Message Date
Howard Chu
233632fa04 Use per-thread slap_counters to eliminate lock contention 2007-10-25 08:45:00 +00:00
Howard Chu
68ebee4726 Concurrency tweaks:
store conn->c_sd, don't use ber_sockbuf_ctrl to retrieve it.
  use per-thread free lists for operations, no mutexes needed.
2007-10-25 02:22:40 +00:00
Pierangelo Masarati
5acbf5997c re-sort funcs (and enums) in "logical" order 2007-08-17 09:49:17 +00:00
Pierangelo Masarati
c77e65c274 fix SASL ssf logging (ITS#5001) 2007-06-09 15:31:46 +00:00
Hallvard Furuseth
46d420565b thread pool data was accessed through the wrong type 2007-05-31 20:52:15 +00:00
Hallvard Furuseth
93fbb87545 Give OperationBuffer normal struct members instead of using
LBER_ALIGNED_BUFFER(OPERATION_BUFFER_SIZE), as discussed in ITS#4078.
Add a corresponding SyncOperationBuffer in overlays/syncprov.c.
2007-05-18 23:54:26 +00:00
Hallvard Furuseth
584ea90db2 Remove unused variables 2007-05-09 19:11:44 +00:00
Hallvard Furuseth
bbc719ca56 ITS#4948: '#if <undefined macro>' -> '#ifdef' warning cleanup 2007-05-09 18:17:10 +00:00
Howard Chu
63bef22335 Cleanup conditional arg in prev commit 2007-03-31 00:40:22 +00:00
Howard Chu
c6df30118f ITS#4893 fix prev commit 2007-03-25 12:18:06 +00:00
Howard Chu
7426ab07fc ITS#4893 define LDAP_PF_LOCAL_SENDMSG in <ac/socket.h> if a message must
be sent to transmit client credentials. Buffer the message data.
2007-03-25 04:40:22 +00:00
Pierangelo Masarati
639d678592 cleanup struct declarations 2007-03-20 01:07:10 +00:00
Howard Chu
06212e9de9 Plug leaks 2007-02-16 02:22:24 +00:00
Howard Chu
e28627f1e2 Added "slapd" rewrite map handler, connection_fake_init2 to use existing
tmpmemctx without reinitializing
2007-02-14 08:51:47 +00:00
Howard Chu
9b5dbf9e35 Remove extraneous debug statements 2007-01-25 13:53:08 +00:00
Howard Chu
4b4072eb44 Use thread keys to plug slapi leak in connection_fake_init() 2007-01-25 11:55:40 +00:00
Howard Chu
0126348461 connection_fake_init can leak SLAPI extensions. 2007-01-25 07:22:39 +00:00
Quanah Gibson-Mount
7b503e6290 logging cleanup 2007-01-12 22:55:39 +00:00
Kurt Zeilenga
da6d9eb046 happy new year 2007-01-02 20:00:42 +00:00
Howard Chu
ff93c6bef7 Don't log EAGAIN returns from ber_get_next() 2006-12-31 16:15:45 +00:00
Howard Chu
d2fc667fe7 Fix connection_get race condition on Windows 2006-12-26 23:40:00 +00:00
Howard Chu
df54b1a244 In connections_shutdown - check for Uninitialized struct first, since
attempting to lock an unint'd mutex may crash on some platforms. And
there is no chance for an unint'd conn to get init'd while shutdown
is occurring, so the unprotected check of struct state is safe.
2006-12-21 06:06:32 +00:00
Pierangelo Masarati
d29d83a80e protect assertion behind mutex (test should be atomic anyway...) 2006-12-14 19:22:00 +00:00
Howard Chu
4d9d4e2538 Fix connections_shutdown() 2006-12-14 02:12:39 +00:00
Pierangelo Masarati
9fdb9b6e02 give cleanup another chance to free the temporary callback (ITS#4771) 2006-12-07 19:44:04 +00:00
Pierangelo Masarati
0b29856b16 don't use conditionals in macro arguments (ITS#4769) 2006-12-06 11:39:07 +00:00
Howard Chu
47793aa3c9 Fix typo in prev commit 2006-11-09 06:47:47 +00:00
Howard Chu
6ac62a7573 Fix - failed Binds could leave a closing connection open. 2006-11-09 01:42:21 +00:00
Howard Chu
44475373db Fix prev commit again (ITS#4722) 2006-10-26 22:10:12 +00:00
Howard Chu
9a3138cb55 Fix typo in prev commit 2006-10-17 19:45:55 +00:00
Howard Chu
eb16636abe Tweak prev commit, add more notes (re ITS#4667) 2006-10-17 19:37:10 +00:00
Howard Chu
354e90a0d9 From rev 1.373, releasing the mutex means connection_close() may get called
from another thread. Drop the assert, just return instead.
2006-10-17 19:31:42 +00:00
Howard Chu
57c329a3af ITS#4606 errno is not per-thread on WIN32, always use WSAGet/SetLastError
(with notable exceptions: ignore tests for EINTR which winsock never sets)
2006-09-14 06:35:34 +00:00
Howard Chu
3e50687bf7 In connection_next, sockets may not be contiguous on Unix if syslog etc are
active. Just search to the end of the table, tracking the max isn't worth
the lock overhead.
2006-09-11 21:19:38 +00:00
Howard Chu
f82ee25ec1 ITS#4667 in connection_closing, release c_mutex to allow waiters
to clear out.
2006-09-08 23:31:53 +00:00
Howard Chu
da6a4bcd3e ITS#4667 fix assert in connection_next() for PENDING connections 2006-09-08 23:30:31 +00:00
Pierangelo Masarati
0d2e2772d5 move req2op selection into a helper 2006-09-04 07:17:34 +00:00
Pierangelo Masarati
faa13725e6 remove unnecessary assertion; add test for pending ops 2006-09-01 10:41:54 +00:00
Pierangelo Masarati
9e8725a3cc add an assertion to track ITS#4659; minor cleanup 2006-09-01 10:00:50 +00:00
Howard Chu
69fa2d9f93 Move ber_sockbuf_free() to slapd_remove() to avoid using AC_SOCKET_INVALID 2006-08-18 23:54:24 +00:00
Howard Chu
2e3e7ed14f Fix prev commit 2006-08-18 01:17:44 +00:00
Howard Chu
098153c5bb Move socket close into slapd_remove, don't close until after socket
is removed from daemon control.
2006-08-17 00:44:38 +00:00
Pierangelo Masarati
f26bef66c6 cleanup 2006-05-20 15:49:18 +00:00
Howard Chu
9de3d9c277 Fix idletimeout deadlock 2006-05-19 22:47:33 +00:00
Howard Chu
813b95941e Fixed version of rev 1.361 2006-05-14 03:37:17 +00:00
Howard Chu
1cf58aba1c Revert prev commit, spoke too soon, close race condition came back. 2006-05-13 12:35:12 +00:00
Howard Chu
d74a230229 Simplify connection mutexes 2006-05-13 11:50:00 +00:00
Luke Howard
87051a2a4a ITS#4525 2006-05-08 08:58:33 +00:00
Howard Chu
a8f8dfe747 Use ldap_pvt_thread_pool_tid() 2006-05-02 00:58:05 +00:00
Howard Chu
daec2febfb ITS#4476 revert connection_write processing back into main thread 2006-04-06 04:34:37 +00:00