147 Commits

Author	SHA1	Message	Date
Howard Chu	57c5bbf7e4	Just use a readlock in cache loader	2007-02-08 09:27:35 +00:00
Howard Chu	ea8a2894c4	Free main-thread keys in db_close, can't do it later because the env ... will be gone.	2007-01-25 12:00:40 +00:00
Howard Chu	5d4c8f86bf	Fix prev commit	2007-01-17 10:02:53 +00:00
Howard Chu	ecb06964c5	Cleanup prev commit	2007-01-17 02:46:41 +00:00
Howard Chu	9898983d58	Minor tweaks, put dup EntryInfo's on free list instead of destroying them	2007-01-17 02:43:05 +00:00
Howard Chu	b941856404	Set upper bound on EntryInfos, fixed at 4x Entry cachesize. ... Probably should be tunable. Probably should add hit/miss counters to monitor to assist in tuning...	2007-01-12 07:35:34 +00:00
Howard Chu	813cca89a6	ITS#4797	2007-01-08 19:10:23 +00:00
Howard Chu	2082b1f9f0	Avoid calling purge too often	2007-01-05 06:49:25 +00:00
Howard Chu	001ce8e493	Partially revert prev commit, fix lru_purge loop	2007-01-04 11:01:59 +00:00
Howard Chu	e48f72c1b5	Unify use of BDB lockers	2007-01-04 08:36:59 +00:00
Kurt Zeilenga	da6d9eb046	happy new year	2007-01-02 20:00:42 +00:00
Howard Chu	822888eafc	Bypass a few redundant locks	2007-01-01 11:44:39 +00:00
Howard Chu	6074f48fa7	Tweak cache_lru_print	2007-01-01 11:17:13 +00:00
Howard Chu	a5546a22a8	Fix prev commit, hdb_cache_find_parent	2007-01-01 07:09:22 +00:00
Howard Chu	cd35516bdc	Fix typo in prev commit	2007-01-01 02:16:25 +00:00
Howard Chu	37dfbd3e8b	Fix prev commit	2007-01-01 01:07:57 +00:00
Howard Chu	d7fa1e764d	partial revert - keep acquiring BDB lock in lru_purge.	2007-01-01 01:00:19 +00:00
Howard Chu	b8309b4cf8	Cleanup db_lock/unlock parameters	2006-12-31 23:37:06 +00:00
Howard Chu	a720011c8a	Ditch LRU cache replacement in favor of 2nd-chance/clock. ... Much better concurrency.	2006-12-31 16:16:48 +00:00
Howard Chu	e844a6956d	In bdb_cache_add use a readlock outside the txn to protect the entry ... until send_ldap_result completes.	2006-12-30 12:05:55 +00:00
Howard Chu	eb47d50da7	Tweak bei_state so cache_lru_add doesn't ever try to free just-added ... entries. This allows us to use the frontend's entry directly instead of having to entry_dup it before adding to the cache.	2006-12-01 14:48:38 +00:00
Howard Chu	bd95f8ab1c	ITS#4738 provisional fix, still needs confirmation	2006-11-13 00:24:59 +00:00
Quanah Gibson-Mount	88ce76f24e	fix e_private value, goes along with memory leak fix to accesslog.c	2006-08-26 23:20:00 +00:00
Howard Chu	2a2115d70c	Fix prev commit, update kid counters too	2006-07-19 00:06:57 +00:00
Howard Chu	0340a1f1bf	ITS#4616 fix cache_modrdn bei_state management	2006-07-18 10:22:48 +00:00
Howard Chu	bab03f6d5c	Additional purge changes for rev 1.120, don't keep the current thread ... busy unnecessarily	2006-07-12 01:55:21 +00:00
Howard Chu	70512edd5a	ITS#4572 don't unlock non-existent locks	2006-06-01 02:03:10 +00:00
Howard Chu	c76eb501ab	Revert cache-purge task, go back to purging inline	2006-04-05 07:17:11 +00:00
Howard Chu	c002351982	cleanup	2006-03-31 22:24:39 +00:00
Howard Chu	9eca89acc8	Improve prev commit	2006-02-09 07:02:40 +00:00
Howard Chu	6344d36df0	Same as ITS#4385, must wake listener after submitting job to runqueue	2006-02-09 06:53:49 +00:00
Pierangelo Masarati	697df7cc06	silence warning	2006-01-07 11:06:32 +00:00
Howard Chu	a614a1c16e	Cleanup task declarations	2006-01-06 04:06:18 +00:00
Howard Chu	e82ce6444c	ITS#4308 fix cache flush task interval	2006-01-06 03:51:56 +00:00
Kurt Zeilenga	acbb5cf689	Happy new year!	2006-01-03 23:11:52 +00:00
Pierangelo Masarati	9c7aca5b47	fix indentation	2006-01-02 08:18:29 +00:00
Howard Chu	553b2884ab	cache LRU concurrency improvements, reorder DB accesses for consistency	2005-12-09 00:20:18 +00:00
Howard Chu	3e062a3f35	ITS#4254 fix lock ordering	2005-12-08 23:45:47 +00:00
Pierangelo Masarati	af23be3675	silence warnings...	2005-11-14 21:47:17 +00:00
Pierangelo Masarati	474dfbc8fd	don't trust strchr/strrchr with bervals	2005-11-13 22:26:53 +00:00
Howard Chu	a14fd7cc3b	Cleanup unused def	2005-10-18 21:36:06 +00:00
Howard Chu	ed5e8db572	Eliminate long-lived read transaction, we can just set a cursor's locker ... ID explicitly to our long-lived locker. Should have thought of this before...	2005-10-18 21:35:07 +00:00
Howard Chu	8f7da062f5	ITS#4088 force cursors to use same locker	2005-10-18 20:52:04 +00:00
Howard Chu	c44210a8bf	ITS#4088 proposed fix	2005-10-16 07:37:36 +00:00
Howard Chu	9944c9c461	ITS#4015 silence pointer/integer size warnings	2005-09-08 23:47:19 +00:00
Howard Chu	aa897eae90	ITS#4010 fixes complaint, larger problem remains	2005-09-08 15:28:36 +00:00
Hallvard Furuseth	3226cdf2a3	bdb_locker_id, bdb_locker_id_free: BerkeleyDB lock ids are u_int32_t, not int.	2005-07-27 12:19:16 +00:00
Hallvard Furuseth	81ecb0b153	assert expects int. (int)<nonnull ptr/long> can be 0. Use assert(arg!=0/NULL).	2005-07-18 06:22:33 +00:00
Hallvard Furuseth	d6cc947561	Silence "unused <something>" warnings	2005-07-14 19:52:08 +00:00
Howard Chu	fbbfac6771	ITS#3857 fix bottom-up modrdn cascades, use a global modrdn counter.	2005-07-12 13:17:56 +00:00