summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* | | | doc: spec lookup table typesRainer Gerhards2013-01-241-10/+24
| | | |
* | | | doc: some spec changes for lookup tablesRainer Gerhards2013-01-241-42/+24
| | | |
* | | | Merge branch 'v7-stable'Rainer Gerhards2013-01-242-2/+41
|\| | |
| * | | added omusrmsg changes to v7 compatibility docRainer Gerhards2013-01-241-1/+32
| | | | | | | | | | | | | | | | Thanks to Tomas Heinrich for pointing out that this was missing.
| * | | - added $!all-json to property replacerRadu Gheorghe2013-01-241-1/+9
| | | | | | | | | | | | | | | | - specified that $!<name> can be used with CEE-enhanced syslog (once parsed via mmjsonparse)
* | | | bump version number for experimental releaseRainer Gerhards2013-01-221-1/+1
| | | |
* | | | optimze: reduce memory operations during dns resolution/hostname settingRainer Gerhards2013-01-228-124/+145
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | previously, hostname and ip strings were shuffled to the msg object, which created a property out of them. Now the cache holds the property, and it is resused (almost) everywhere, what saves a lot of memory operations. The only exception is imtcp session setup, where different handling of the hostname is done, which we need to sort out (but that's another story).
* | | | some fixes to commit 440aaff114607e82aefee56336846611a2e35c9cRainer Gerhards2013-01-222-12/+14
| | | | | | | | | | | | | | | | extended testbench run brought up some issues...
* | | | optimize: move lowercasing of hostnames to cacheRainer Gerhards2013-01-246-54/+85
| | | | | | | | | | | | | | | | so we save this on each name lookup!
* | | | refactor cvthname() for cleaner codeRainer Gerhards2013-01-241-33/+27
| | | |
* | | | fix: gtls stream driver must support new interfaceRainer Gerhards2013-01-241-5/+3
| | | |
* | | | testbench: remove warnings from nettesterRainer Gerhards2013-01-241-1/+1
| | | | | | | | | | | | | | | | | | | | the no-longer-existing -c option was still used when calling rsyslogd
* | | | undo accidental enable of debug system in testbenchRainer Gerhards2013-01-241-2/+2
| | | |
* | | | Merge branch 'master' into master-dns-propsRainer Gerhards2013-01-241-6/+10
|\ \ \ \
| * | | | regression fix: programname was not properly handledRainer Gerhards2013-01-241-6/+10
| | | | | | | | | | | | | | | | | | | | Regression from recent refactoring - never released.
* | | | | optimize: have dns cache pre-create rsyslog prop_t'sRainer Gerhards2013-01-2418-99/+107
|/ / / /
* | | | yet another version number bump (experimental version)Rainer Gerhards2013-01-181-1/+1
| | | |
* | | | fix memory leak (regression from batch optimization work - unreleased)Rainer Gerhards2013-01-181-0/+1
| | | |
* | | | refactor / simplifyRainer Gerhards2013-01-181-5/+3
| | | |
* | | | bump version number for experimental buildRainer Gerhards2013-01-181-1/+1
| | | |
* | | | optimize: reduce nbr of strcpy() in FROMHOST processingRainer Gerhards2013-01-187-33/+45
| | | |
* | | | dnscache: switch to use hashtable instead of linked listRainer Gerhards2013-01-182-72/+58
| | | |
* | | | design: draft lookup table functionalityRainer Gerhards2013-01-174-3/+218
| | | |
* | | | push version for private experimental releaseRainer Gerhards2013-01-171-1/+1
| | | |
* | | | fix compile problem with new stats countersRainer Gerhards2013-01-171-10/+10
| | | | | | | | | | | | | | | | mutexes on systems without atomic instructions were incorrect
* | | | omfile: finalize dynafile cache statsRainer Gerhards2013-01-172-13/+30
| | | |
* | | | push version numberRainer Gerhards2013-01-171-1/+1
| | | |
* | | | omfile: add stats counters to track dynafile performanceRainer Gerhards2013-01-162-1/+39
| | | |
* | | | push version number for private test releaseRainer Gerhards2013-01-151-1/+1
| | | |
* | | | optimize: another round of removing isdigit()Rainer Gerhards2013-01-151-2/+2
| | | |
* | | | optimize memory layout for much better cache hitsRainer Gerhards2013-01-155-36/+44
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Moave element status out of batch_obj_t because we get a *much* better cache hit ratio this way. Note that this is really a HUGE saving, even if it doesn't look so (both profiler data as well as practical tests indicate that!).
* | | | Merge branch 'master' into master-optimizedRainer Gerhards2013-01-152-1/+3
|\ \ \ \
| * | | | bugfix: imklog issued wrong facility in error messagesRainer Gerhards2013-01-152-1/+3
| | | | | | | | | | | | | | | | | | | | ...what could lead to problems in other parts of the code
* | | | | bugfix: regression from textpri() optimizationRainer Gerhards2013-01-151-1/+2
| | | | |
* | | | | optimize: do not iterate over batch items if unneeded ;)Rainer Gerhards2013-01-151-5/+8
| | | | |
* | | | | optimize: use fixed size (8 bits) instead of enumRainer Gerhards2013-01-151-7/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | looks like GCC, even if optimizing, uses 32 bits - at least this is suggested by the profiler results (both in terms of runtime and cache misses).
* | | | | optimize: re-use already computed value!Rainer Gerhards2013-01-141-2/+2
| | | | |
* | | | | optimize: use built-in str comparison, as this is optimized in assemblyRainer Gerhards2013-01-141-7/+1
| | | | | | | | | | | | | | | | | | | | | | | | | do only when possible. However, the profiler only shows as *very* minimal effect.
* | | | | optimize: save inspection of already-inspected dataRainer Gerhards2013-01-141-2/+8
| | | | | | | | | | | | | | | | | | | | this is just a small improvement, but let's get the benefit ;)
* | | | | optimize: slight improvement due to isdigit() replacementRainer Gerhards2013-01-141-2/+3
| | | | |
* | | | | optimize: minor cleanup (long scheduled...)Rainer Gerhards2013-01-141-9/+0
| | | | | | | | | | | | | | | | | | | | | | | | | I did yet another safety-check that there are no problems with removing that de-init.
* | | | | optimize: iscntrl() seems to be surprisingly slowRainer Gerhards2013-01-141-4/+3
| | | | | | | | | | | | | | | | | | | | at least so tells the profiler...
* | | | | optimize: obtaining programnameRainer Gerhards2013-01-143-64/+35
| | | | | | | | | | | | | | | | | | | | included some additional refactoring for cleaner code
* | | | | optimzize: msg/getNow() - remove snprintfRainer Gerhards2013-01-141-8/+26
| | | | | | | | | | | | | | | | | | | | instead, we use a lookup table for the values.
* | | | | optimize: permit cancel during object destructionRainer Gerhards2013-01-141-4/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 1) usually, no cancellation happens 2) even if so, there is no cancellation point inside the destructors, so disabeling cancellation was mood in the first place...
* | | | | optimize msg.c/textpriRainer Gerhards2013-01-142-11/+14
|/ / / / | | | | | | | | | | | | remove snprintf() in favor for quicker code
* | | | omfile: improved async writing, finally enabled full async writeRainer Gerhards2013-01-143-12/+25
| | | | | | | | | | | | | | | | also fixed a couple of smaller issues along that way
* | | | Merge branch 'v7-stable-libnet-frag'Rainer Gerhards2013-01-1410-6/+48
|\ \ \ \ | | | | | | | | | | | | | | | | | | | | Conflicts: ChangeLog
| * | | | Merge branch 'v7-stable' into v7-stable-libnet-fragRainer Gerhards2013-01-143-2/+11
| |\| | |
| | * | | Merge branch 'v5-stable' into v7-stableRainer Gerhards2013-01-143-1/+7
| | |\ \ \ | | | | |/ | | | |/|
generated by cgit v1.2.3 (git 2.25.1) at 2025-07-06 22:39:43 +0000