- ngIRCd Users - lists.barton.de

ngIRCd 17
by Alexander Barton 07 Nov '10

07 Nov '10

Hello World! We proudly present … ngIRCd 17! :-) After three release candidates, here it is, the new release. The most prominent new features are that user mode „c“ (show connect/disconnect NOTICEs, required for e.g. BOPM) and „x“ (hostname cloaking) have been implemented, and support for user authentication using PAM has been added. New configuration options „NoPAM“, „NoZeroConf“, and „SyslogFacility“ were supplemented and allow even better customization of ngIRCd. In addition, the debug mode has been enhanced, the daemon can dump its internal state now, which eases bug fixing. And the power of IRC operators has been raised: now they can use MODE on any channel and always change channel modes. The NEWS file lists the following new features since our last release (r16): * doc: change path names in sample-ngircd.conf depending on sysconfdir * Generate ngIRCd version number from GIT tag. * Make sourcecode compatible with ansi2knr again. This allows to compile ngIRCd using a pre-ANSI K&R C compiler again. * New configuration option "NoZeroConf" to disable service registration at runtime even if ngIRCd is compiled with support for ZeroConf (e.g. using Howl, Avahi or on Mac OS X). * New configuration option "SyslogFacility" to define the syslog "facility" (the "target"), to which ngIRCd should send its log messages. Possible values are system dependant, but most probably "auth", "daemon", "user" and "local1" through "local7" are possible values; see syslog(3). Default is "local5" for historical reasons. * Dump the "internal server state" (configured servers, established connections and known clients) to the console or syslog when receiving the SIGUSR2 signal and debug mode is enabled. * Enable the daemon to disable and enable "debug mode" on runtime using signal SIGUSR1, when debug code is compiled in, not only on startup using the command line parameters. * Implement user mode "x": hostname cloaking (closes: #102). * Change MOTD file handling: ngIRCd now caches the contens of the MOTD file, so the daemon now requires a HUP signal or REHASH command to re-read the MOTD file when its content changed. * Allow IRC ops to change channel modes even without OperServerMode set. * Allow IRC operators to use MODE command on any channel (closes: #100). * New configuration option "NoPAM" to disable PAM. * Implement asynchronous user authentication using PAM, please see the file doc/PAM.txt for details. * Add some documentation for using BOPM with ngIRCd, see doc/Bopm.txt. * Implement user mode "c": receive connect/disconnect NOTICEs. Note that this new mode requires the user to be an IRC operator. * Show SSL status in WHOIS output, numeric 275. Please see the ChangeLog for a list of all changes — it is available in the archive and online: <http://ngircd.berlios.de/doc/ChangeLog>. The new release can be downloaded here: * <ftp://ftp.berlios.de/pub/ngircd/ngircd-17.tar.gz> * <ftp://ngircd.barton.de/pub/ngircd/ngircd-17.tar.gz> GnuPG signatures for release 17, a Mac OS X package, patches from release 16 and 17~rc3 as well as further archive formats are available as well and can be downloaded from here: * <ftp://ngircd.barton.de/pub/ngircd/> * <ftp://ftp.berlios.de/pub/ngircd/> You can have a look at the complete history and every single patch using the GIT web-frontend located at: * <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> I hope that there are now showstoppers … please report all bugs using our bug tracker (<http://ngircd.barton.de/bugzilla/index.cgi>) and/or this mailing list. Thanks for your support! Regards Alex P.S.: I updated my unofficial Debian packages already: <http://debian.barton.de>

1 0

ngIRCd 17~rc2
by Alexander Barton 25 Okt '10

25 Okt '10

Hi All! Not quite a week later, there is the 2nd release candidate for the upcoming ngIRCd release 17 available for downloading and testing. The changes since -rc1 are: * doc/Makefile.am: don¹t set docdir, automake handles it already * Only use "__attribute__ ((unused))" if GCC >=2.8 is used * Only try to set FD_CLOEXEC if this flag is defined * Added m68k/apple/aux3.0.1 (gcc 2.7.2) to doc/Platforms.txt * Don't use PARAMS() macro for function implementations * ./configure: check support for C prototypes again * ./configure: check if C compiler can compile ISO Standard C * Make sourcecode compatible with ansi2knr again * Generate ngIRCd version number from GIT tag * ZeroConf: include header files missing since commit a988bbc86a * New doc/HowToRelease.txt file describing the release process * Updated contrib/platformtest.sh (new version scheme) So the biggest change is that really old compilers (pre-ANSI) are supported again; this has been tested on ancient Apple UNIX (A/UX) 3.0. If you want to test ngIRCd yourself, you can use the sequenze tar xzf ngircd-17~rc2.tar.gz cd ngircd-17~rc2 contrib/platformtest.sh which will (hopefully) result in a nice summary after compiling the daemon and running the test suite. The release candidate is available for downloading and testing at the usual locations: <ftp://ftp.berlios.de/pub/ngircd/ngircd-17~rc2.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-17~rc2.tar.gz> The GnuPG signatures for 17~rc2 and patches from release 17~rc1 are available as well and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ftp.berlios.de/pub/ngircd/> You can have a look at the complete history and every single patch using the GIT web-frontend located at: <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> Please test this release candidate thoroughly and report all bugs and problems you encounter using the bug tracker (<http://ngircd.barton.de/bugzilla/index.cgi>) and/or this mailing list. I think I'll update my Debian packages with this release candidate as well as build a new Mac OS X package, so that this code gets even more testing. What do you think? Regards Alex

1 0

ngIRCd 17~rc1
by Alexander Barton 20 Okt '10

20 Okt '10

(My last mail didn’t reach its recipients; /me should use the correct mailing list address in the future – so here my next try…) Hello World! After about 6 months of development, we are in preparation for the next ngIRCd release which will be release 17. I uploaded the first release candidate for the upcoming release, please test! The following new features and changes have been coded, most notably ngIRCd now supports user authentication using PAM, supports hostname cloaking (user mode „x“), and supports BOPM, the open proxy monitor. The full NEWS file lists: * New configuration option "NoZeroConf" to disable service registration at runtime even if ngIRCd is compiled with support for ZeroConf (e.g. using Howl, Avahi or on Mac OS X). * New configuration option "SyslogFacility" to define the syslog "facility" (the "target"), to which ngIRCd should send its log messages. Possible values are system dependant, but most probably "auth", "daemon", "user" and "local1" through "local7" are possible values; see syslog(3). Default is "local5" for historical reasons. * Dump the "internal server state" (configured servers, established connections and known clients) to the console or syslog when receiving the SIGUSR2 signal and debug mode is enabled. * Enable the daemon to disable and enable "debug mode" on runtime using signal SIGUSR1, when debug code is compiled in, not only on startup using the command line parameters. * Implement user mode "x": hostname cloaking (closes: #102). * Change MOTD file handling: ngIRCd now caches the contens of the MOTD file, so the daemon now requires a HUP signal or REHASH command to re-read the MOTD file when its content changed. * Allow IRC ops to change channel modes even without OperServerMode set. * Allow IRC operators to use MODE command on any channel (closes: #100). * New configuration option "NoPAM" to disable PAM. * Implement asynchronous user authentication using PAM, please see the file doc/PAM.txt for details. * Implement user mode "c": receive connect/disconnect NOTICEs. Note that this new mode requires the user to be an IRC operator. * Show SSL status in WHOIS output, numeric 275. The release candidate is available for downloading and testing at the usual locations: <ftp://ftp.berlios.de/pub/ngircd/ngircd-17~rc1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-17~rc1.tar.gz> The GnuPG signatures for 17~rc1 and patches from release 16 are available as well and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ftp.berlios.de/pub/ngircd/> You can have a look at the complete history and every single patch using the GIT web-frontend located at: <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> Please test this release candidate thoroughly and report all bugs and problems you encounter using the bug tracker (<http://ngircd.barton.de/bugzilla/index.cgi>) and/or this mailing list. Are features missing, for release 17 or later? Please file „enhancement“ bugs as well! And patches are welcome, too ;-) Thanks a lot for your support! Regards Alex

1 0

nevermind....
by Rich Griese 15 Aug '10

15 Aug '10

Mivermind folks. I realized that my client was using a different /open name password than I had adjusted in the .conf what a moron I am. On the good news front. I have my server limited to my one channel, and don't have to worry about idiots using it. oh... now that I say that.... from a utilization of resources standpoint. How to you prevent people from joining the SERVER and not a channel. I mean, and this is a bit ridiculas, but two people could join the server, and not the one allowed channel, but could use the / msg command to talk to each other. I run this server on a mac mini, and am looking for every way to eliminate any user other than the ones that are using the server/channel for it's purpose. Any thoughts. Cheers! RichGriese(a)gmail.com

2 1

upgraded to release 16 and can limit to predefined, but....
by Rich Griese 15 Aug '10

15 Aug '10

I made progress. I upgraded to release 16 and now can limit channels to the one I want. problem though.... I am not longer getting ops. Can you tell me what commands need to go in the .conf file, AND what commands I execute to identify as a operator, and then op myself in my channel. Cheers! RichGriese(a)gmail.com

1 0

question about limiting channels
by Rich Griese 13 Aug '10

13 Aug '10

I have been running the ngircd for a while, now and it works fine. I have been using it as a user support system, and they way I have it confirmed is that I set up a server password, and was using cgi:irc. I could with that web client prevent the /join command from working. So in effect I did not have to worry about any connecting and using any channel than the one I had set up and monitored. The server required a password, the only one that knew the password was the cgi:irc client, and only using that client was forced to my one channel, and could not issue a /join command. I am not having troulbe with cgi:irc on a new server. My question. Can I set ngircd up so that users CANNOT create channels? I want to set up a mibbit widget that wil allow people to join the server, and my one user help channel, but I want to prevent anyone from doing a /join #yaya and creating another channel. I can't find anything in the docs for the .conf file, but found this mailing list and thought I would ask. I had been limiting channel access before by the one web client i allowed to access the system. Now I want to know if I can set the ngircd server itself to limit channels. Cheers! RichGriese(a)gmail.com

3 2

Patch to show SSL status in WHOIS
by Neale Pickett 23 Jun '10

23 Jun '10

I've been wanting this for years and finally took the 5 minutes to patch it in. I took the response code (275) from whatever's running OFTC's IRC network. diff -ur ngircd-16.orig/src/ngircd/irc-info.c ngircd-16/src/ngircd/irc-info.c --- ngircd-16.orig/src/ngircd/irc-info.c 2010-04-25 06:06:29.000000000 -0500 +++ ngircd-16/src/ngircd/irc-info.c 2010-06-11 17:23:18.000000000 -0500 @@ -993,6 +993,13 @@ if( ! IRC_WriteStrClient( from, RPL_WHOISOPERATOR_MSG, Client_ID( from ), Client_ID( c ))) return DISCONNECTED; } + + /* Connected using SSL? */ + if( Conn_UsesSSL(Client_Conn(c))) + { + if( ! IRC_WriteStrClient( from, RPL_WHOISSSL_MSG, Client_ID( from ), Client_ID( c ))) return DISCONNECTED; + } + /* Idle and signon time (local clients only!) */ if (Client_Conn(c) > NONE ) { if (! IRC_WriteStrClient(from, RPL_WHOISIDLE_MSG, diff -ur ngircd-16.orig/src/ngircd/messages.h ngircd-16/src/ngircd/messages.h --- ngircd-16.orig/src/ngircd/messages.h 2010-04-25 06:06:29.000000000 -0500 +++ ngircd-16/src/ngircd/messages.h 2010-06-11 17:20:34.000000000 -0500 @@ -47,6 +47,7 @@ #define RPL_LOCALUSERS_MSG "265 %s %lu %lu :Current local users: %lu, Max: %lu" #define RPL_NETUSERS_MSG "266 %s %lu %lu :Current global users: %lu, Max: %lu" #define RPL_STATSCONN_MSG "250 %s :Highest connection count: %lu (%lu connections received)" +#define RPL_WHOISSSL_MSG "275 %s %s :is connected via SSL (secure link)" #define RPL_AWAY_MSG "301 %s %s :%s" #define RPL_USERHOST_MSG "302 %s :"

2 2

ngIRCd Release 16
by Alexander Barton 04 Mai '10

04 Mai '10

Hi All! ngIRCd 16 has been released and is available from the usual locations, see below. Since our last stable release, release 15, the most notably changes are: - Enhace connection statistics counters: display total number of served connections on daemon shutdown and when a new client connects using the new numeric RPL_STATSCONN (250). - Various fixes to the build system and code cleanups. - Added missing modes to USERMODES #define - Show our name (IRCD=ngIRCd) in ISUPPORT (005) numeric - Quote received messages of ERROR commands in log output. - ngircd.conf manual page: document missing "Password" variable. - Implement WEBIRC command used by some Web-IRC frontends. The password required to secure this command must be configured using the new "WebircPassword" variable in the ngircd.conf file. - Remove limit on max number of configured irc operators. - Only link "nsl" library when really needed. - A new channel mode "secure connections only" (+z) has been implemented: Only clients using a SSL encrypted connection to the server are allowed to join such a channel. But please note three things: a) already joined clients are not checked when setting this mode, b) IRC operators are always allowed to join every channel, and c) remote clients using a server not supporting this mode are not checked either and therefore always allowed to join. Please see the ChangeLog file (or even better: the GIT history) for the complete list of all the changes committed. As usua, you can download ngIRCd release 15 from these locations: <ftp://ftp.berlios.de/pub/ngircd/ngircd-16.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-16.tar.gz> GnuPG signatures for release 16, a Mac OS X installer package, and patches from release 15 are available, too, and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ftp.berlios.de/pub/ngircd/> The relevant MD5 sums are: ngircd-15.tar.gz: c183a85eba6fe51255983848f099c8ae ngircd-14.1-15.patch.gz: 8a114618bfc23d3301762e0dbee704ac You can have a look at the complete history and every single patch using the GIT web-frontend located at: <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> Please let us know if you encounter any bugs or need more or better documentation (best is to file bugs using the bug tracker or to mail to this list). Thanks a lot for your support! Regards Alex

1 1

ngIRCd-Foo when leaving a channel?
by Götz Hoffart 29 Apr '10

29 Apr '10

Hi, since a few days I noticed a strange behaviour. See this log from irssi: […] 19:03:11 * po||ux macht dann mal Feierabend. 19:03:13 <@po||ux> bye! 19:06:25 < tooltime> cu 19:06:33 < SirLani> tschö 19:10:12 -!- po||ux [~alex@…] has left #… [Heim. Biergarten. Juhu!] 19:10:17 < po||ux> 7aw 19:10:32 < SirLani> eh 19:10:34 < SirLani> hä!? 19:10:58 < SirLani> po||ux hat den Chatraum verlassen und sagt danach noch was? 19:17:42 -!- sepia [~lucia@…] has joined #… 19:17:53 < sepia> Abend! 19:18:07 < tooltime> re 19:32:08 -!- You're now known as allerbest 19:32:20 < allerbest> re 19:32:32 < allerbest> SirLani: das hat mich neulich schon gewundert […] We had this a few days earlier, too—with exactly the same behaviour: po||ux leaves (timestamp 19:10:12) and a few seconds after that “he” writes “7aw” (timestamp 19:10:17) to the channel. Last time I thought of a random hiccup or cosmic ray or dying bits in non-ECC memory. But this time two people mentioned the behaviour _and_ it was the exact same string “po||ux” sent after leaving. I'm using: # ngircd --version ngircd 15-SYSLOG+ZLIB+SSL+TCPWRAP+IDENT+IRCPLUS+IPv6-i486/pc/linux-gnu Regards Götz

2 1

Web: Suggestion: clarify mailing list status
by usenet＠hoffart.de 25 Apr '10

25 Apr '10

Hi, I think it should be noted in the »Support« section of the website, that any feedback is appreciated even without subscribing to the mailing list. The text now doesn't state clearly that it is not necessary to subscribe to the mailing list. This can be hurdle for some users. Regards Götz -- http://www.knubbelmac.de/

2 1

ngIRCd 16~rc1
by Alexander Barton 25 Apr '10

25 Apr '10

Hi IRC’ers! We are in preparation of the next ngIRCd release, release 16, which will contain the following changes, enhancements, and new features: - Various fixes to the build system and code cleanups. - contrib/platformtest.sh: Only show latest commit. - Updatet doc/Platforms.txt, added new README-Interix.txt documenting how to tun ngIRCd on Microsoft Services for UNIX (MS SFU, MS SUA). - Updated links to the ngIRCd homepage (bug tracker, mailing list). - Added missing modes to USERMODES #define - Show our name (IRCD=ngIRCd) in ISUPPORT (005) numeric - Quote received messages of ERROR commands in log output. - ngircd.conf manual page: document missing "Password" variable. - Implement WEBIRC command used by some Web-IRC frontends. The password required to secure this command must be configured using the new "WebircPassword" variable in the ngircd.conf file. - Don't use port 6668 as example for both "Ports" and "SSLPorts". - Remove limit on max number of configured irc operators. - Only link "nsl" library when really needed. - A new channel mode "secure connections only" (+z) has been implemented: Only clients using a SSL encrypted connection to the server are allowed to join such a channel. But please note three things: a) already joined clients are not checked when setting this mode, b) IRC operators are always allowed to join every channel, and c) remote clients using a server not supporting this mode are not checked either and therefore always allowed to join. The first release candidate, version 16~rc1, is available for downloading and testing since a few days, and it would be really great if you could vive it a try and do some testing. Thanks! As usual the download locations are: <ftp://ftp.berlios.de/pub/ngircd/ngircd-16~rc1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-16~rc1.tar.gz> The GnuPG signatures for 16~rc1 and patches from release 15 are available as well and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ftp.berlios.de/pub/ngircd/> You can have a look at the complete history and every single patch using the GIT web-frontend located at: <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> Please test this release candidate and report all bugs you encounter using the bug tracker and/or this mailing list. Thanks a lot for your support! Regards Alex

1 1

ngIRCd Release 15
by Alexander Barton 12 Nov '09

12 Nov '09

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi! There have been no reports of problems in our release candidate for ngIRCd 15, so here it is! Since our last ngIRCd release 14.1 on April 9 quite a few fixes and enhancements have been developed and been integrated into our master branch. The changes are: - - Do not add default listening port (6667) if SSL ports were specified, so ngIRCd can be configured to only accept SSL encrypted connections now. - - Enable IRC operators to use the IRC command SQUIT (insted of the already implemented but non-standard DISCONNECT command). - - New configuration option "AllowRemoteOper" (disabled by default) that enables remote IRC operators to use the IRC commands SQUIT and CONNECT on the local server. - - Mac OS X: fix test for packagemaker(1) tool in Makefile and use gcc 4.0 for Mac OS X 10.4 compatibility in the Xcode project file. - - Fix --with-{openssl|gnutls} to accept path names. - - Fix LSB header of Debian init script. - - Updated doc/Platforms.txt and include new script contrib/platformtest.sh to ease generating platform reports. - - Fix connection information for already registered connections. - - Enforce upper limit on maximum number of handled commands. This implements a throttling scheme: an IRC client can send up to 3 commands or 256 bytes per second before a one second pause is enforced. - - Fix connection counter. - - Fix a few error handling glitches for SSL/TLS connections. - - Minor fixes to manual pages and documentation. You can download ngIRCd release 15 here: <ftp://ftp.berlios.de/pub/ngircd/ngircd-15.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-15.tar.gz> The GnuPG signatures for 15, a Mac OS X installer package, and patches from release 14.1 are available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ftp.berlios.de/pub/ngircd/> The relevant MD5 sums are: ngircd-15.tar.gz: c183a85eba6fe51255983848f099c8ae ngircd-14.1-15.patch.gz: 8a114618bfc23d3301762e0dbee704ac You can have a look at the complete history and every single patch using the GIT web-frontend located at: <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> Please let us know if you encounter any bugs or need more/better documentation (best is to file bugs using the bug tracker or to mail to this list). Thanks! Regards Alex -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) iQEcBAEBAgAGBQJK/A6rAAoJEIpFmtLqoVokCd8H/1LoJy2h5wMjnFo8vczMs5Tk O7Vula2j7+oMkDyg+s1jSVzElXWVR3ghThDCofYcpRlk70PLeeonwI/5q8IGCPcs uGeHgmhbCQUyLfCGffX6x/Gmrd0bo3PYnFQGoNHqmMdV2AL4KJtYz5b6VvbpWHS+ 6Y+gf6xp9f9dQC2J1250+FEKv6NtNv0ugIbMgZh9tdPnECqEFVGlDETvzR2YR7uM JDFq0eVzXqj6S9MFjOub7Boq0O7uEHYBuZ1suJ8rwRAgJVQrJoYvxm60YD5LIvus /9U7AfyPvXbEgr+Qfg2E1DlyXDcRr46lKSGuVgqfjATnmlhoIDtEQ14D+2rM8FE= =wATK -----END PGP SIGNATURE-----

1 0

ngIRCd 15~rc1
by Alexander Barton 25 Okt '09

25 Okt '09

Hi all! Since our last ngIRCd release 14.1 on April 9 quite a few fixes and enhancements have been developed and been integrated into our master branch. So here it is: the first release candidate for ngIRCd 15: The changes are: - Do not add default listening port (6667) if SSL ports were specified, so ngIRCd can be configured to only accept SSL encrypted connections now. - Enable IRC operators to use the IRC command SQUIT (insted of the already implemented but non-standard DISCONNECT command). - New configuration option "AllowRemoteOper" (disabled by default) that enables remote IRC operators to use the IRC commands SQUIT and CONNECT on the local server. - Mac OS X: fix test for packagemaker(1) tool in Makefile and use gcc 4.0 for Mac OS X 10.4 compatibility in the Xcode project file. - Fix --with-{openssl|gnutls} to accept path names. - Fix LSB header of Debian init script. - Updated doc/Platforms.txt and include new script contrib/platformtest.sh to ease generating platform reports. - Fix connection information for already registered connections. - Enforce upper limit on maximum number of handled commands. This implements a throttling scheme: an IRC client can send up to 3 commands or 256 bytes per second before a one second pause is enforced. - Fix connection counter. - Fix a few error handling glitches for SSL/TLS connections. - Minor fixes to manual pages and documentation. You can download ngIRCd 15~rc1 here: <ftp://ftp.berlios.de/pub/ngircd/ngircd-15~rc1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-15~rc1.tar.gz> The GnuPG signatures for 15~rc1 and patches from release 14.1 are available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ftp.berlios.de/pub/ngircd/> You can have a look at the complete history and every single patch using the GIT web-frontend located at: <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> Please test this release candidate and report all bugs you encounter using the bug tracker and/or this mailing list. Thanks! Regards Alex

1 0

Re: [ngIRCd-ML] how to install ngIRCd on Windows cygwin?
by a boy 21 Sep '09

21 Sep '09

On Fri, Sep 18, 2009 at 10:48 PM, Alexander Barton <alex(a)barton.de> wrote: > Am 17.09.2009 um 09:06 schrieb a boy: > > any direction? >> > > Installation on Cygwin should be just the same as on any other UNIX-like > platform and is explained in the INSTALL file: > > ./configure [<parameters> [...]] > ~~~~~~~~~~~~ ./configure is unix shell script and can not run on Cygwin and windows. so how to get the file "makefile"? > make > make install > > But I don't know if anybody has been testing ngIRCd on Cygwin recently, so > it probably could be that something is broken ... > If so, please report any errors you encounter! > > Regards > Alex > >

2 1

how to install ngIRCd on Windows cygwin?
by a boy 18 Sep '09

18 Sep '09

any direction?

2 1

ngircd 14.1 compiles fine on HP-UX 11i
by Götz Hoffart 22 Jul '09

22 Jul '09

Hi, just wanted you to know that ngircd 14.1 compiles fine on HP-UX 11i (PA-RISC) using gcc 4.2.3 and gmake: -------------- ngIRCd 14.1 has been configured with the following options: Target: hppa2.0w-hp-hpux11.11 Compiler: gcc Compiler flags: -g -O2 -pipe -W -Wall -Wpointer-arith -Wstrict- prototypes -fstack-protector -D_XOPEN_SOURCE_EXTENDED -DSYSCONFDIR='"$ (sysconfdir)"' Libraries: -lz -lnsl 'ngircd' binary: /usr/local/sbin Configuration file: /usr/local/etc Manual pages: /usr/local/man Documentation: /usr/local/share/doc/ngircd Syslog support: yes Enable debug code: no zlib compression: yes IRC sniffer: no Use TCP Wrappers: no Strict RFC mode: no Zeroconf support: no IRC+ protocol: yes IDENT support: no I/O backend: "poll()" IPv6 protocol: no SSL support: no real 0m22.769s user 0m9.110s sys 0m11.100s bash-3.2# time gmake -j2 real 0m13.663s user 0m16.200s sys 0m1.880s bash-3.2# ./ngircd --help ngircd 14.1-SYSLOG+ZLIB+IRCPLUS-hppa2.0w/hp/hpux11.11 bash-3.2# file ngircd ngircd: PA-RISC2.0 shared executable dynamically linked -not stripped -------------- (testing machine is a J6750 with two PA-RISC 8700+ CPUs with 875 MHz and enough RAM) Regards Götz

1 0

[Announce]: Initial certificate support for openssl backend
by Florian Westphal 18 Mai '09

18 Mai '09

Hi all, a patch (mostly untested 8->) that adds initial support for certificates to the openssl backend is located here: http://strlen.de/ngircd/0001-SSL-TLS-add-initial-certificate-support-to-ope… (signature: http://strlen.de/ngircd/0001-SSL-TLS-add-initial-certificate-support-to-ope…) The patch applies to ngircd release 14.1. For those that want to follow up on changes/patches, please consider tracking my repository at git://git.breakpoint.cc/fw/ngircd-tls.git , the certificate changes are in the "tls-master" branch. (gitweb: http://git.breakpoint.cc/cgi-bin/gitweb.cgi?p=fw/ngircd-tls.git;a=summary) By "initial" i mean that the basic functionality appears to work, but has not been tested much; noone has reviewed the code so far, there is no support for things like "accept this cerificate only from ip X", etc. Although I do not know when I can work on this again, I ask that everyone interested in seeing this feature in ngircd and with some spare time and a ngircd test installation to try out this patch and report any bugs/change requests, thanks a lot! The patch adds a few new config options: in GLOBAL section: SSLCAFile = /the/file/with/trusted/ca/certificates Filename pointing to the Trusted CA Certificates. Required for verifying peer certificates. SSLCRLFile = /the/file/with/revoked/certificates (WARNING: not tested AT ALL) SSLRequireClientCert = (yes|no) Do not accept incoming SSL connections from clients that do not have a valid certificate. in [SERVER] section: SSLVerify: (yes|no) Verify Server Peer Certificate. If this is an active connection (i.e. ngircd connects to the peer), the ssl handshake is aborted if the certificate of the remote server cannot be validated. If this is a passive connection (ngircd waits for the remote server to connect) and SSLRequireClientCert is false, the server link will only be established if the password matches and a valid certificate was received. (obvioulsy, because its not possible to know in advance if the incoming connection is from a server or an irc client, the ssl handshake always completes; the connection will be shut down again once the SERVER command is received and the connection was established without /invalid certificate.

1 0

ngIRCd Release 14.1
by Alexander Barton 12 Mai '09

12 Mai '09

Hi all! Florian Westphal found a serious bug in ngIRCd release 14 which affects all servers compiled with SSL-support linked in, whereas it is irrelevant whether SSL support is actually in use or not. This bug is remotely triggerable and causes the daemon to crash (DoS). So EVERYBODY using ngIRCd release 13 or 14 with SSL-support linked in SHOULD UPGRADE to nIRCcd release 14.1 as soon as possible! You can use the "ngircd --version" command to check the options your daemon provides: if it lists "SSL", you are affected! (e. g. "ngircd 14.1-SYSLOG+ZLIB+SSL+IRCPLUS+IPv6-i386/apple/darwin9.6" is affected) The full changelog lists (since release 14): - Security: fix remotely triggerable crash in SSL/TLS code. - BSD start script contrib/ngircd.sh has been renamed to ngircd-bsd.sh. - New start/stop script for RedHat-based distributions: contrib/ngircd-redhat.init, thanks to Naoya Nakazawa <naoya(a)sanow.net>. - Doxygen: update source code repository link to GIT. - Debian: build ngircd-full-dbg package. - Allow ping timeout quit messages to show the timeout value. - Fix error handling on compressed links. - Fix server list announcement. - Do not remove hostnames from info text. Direct download links for the source archive: <ftp://ftp.berlios.de/pub/ngircd/ngircd-14.1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-14.1.tar.gz> The ChangeLog can be found here: <http://ngircd.barton.de/doc/ChangeLog> <http://ngircd.berlios.de/doc/ChangeLog> GnuPG signatures and a patches from release 14 are available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ftp.berlios.de/pub/ngircd/> The relevant MD5 sums are: MD5 (ngircd-14.1.tar.gz) = eef90855414c35bfb6590d17e24ee06f MD5 (ngircd-14-14.1.patch.gz) = 896814187a7a350272ab5fb4119a381a You can habe a look at the complete history and every single patch using the GIT web-frontend located at: <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> Please let us know if you encounter any bugs or need more/better documentation (best is to file bugs using the bug tracker or to mail to this list). Thanks! Regards Alex

3 3

ngIRCd Release 14
by Alexander Barton 20 Apr '09

20 Apr '09

Hello! Since the last release candidate no serious problems have been found, and so here it is: ngIRCd Release 14. Prominent changes are the removal of the limit of pre-definded channels in ngircd.conf and the implemnetation of a "channel key file" which allows for individual pre-defined channel user keys (see "man ngircd.conf" and the variable "KeyFile" for details). The changelog lists: - Display IPv6 addresses as "[<addr>]" when accepting connections. - Updated Debian/Linux init script (see contrib/Debian/ngircd.init). - Allow creation of persistent modeless channels. - The INFO command reports the compile time now (if available). - Spell check and enhance ngIRCd manual pages. - Channel mode changes: break on syntax errors in MODE command. - Support individual channel keys for pre-defined channels: introduce new configuration variable "KeyFile" in [Channel] sections in ngircd.conf, here a file can be configured for each pre-defined channel which contains individual channel keys for different users. - Remove limit on maximum number of predefined channels in ngircd.conf. - Updated ngircd.spec file for building RPM packages. - Add new and missing files to Mac OS X Xcode project, and update project. - Reject masks with wildcard after last dot. - TLS/SSL: remove useless error message when ssl connection is closed. - Fix memory leak when a encrypted and compressed server link goes down. (closes bug #95, reported by Christoph) - Fix handling of channels containing dots. (closes ug #93, reported by Gonosz Csiga) Thanks a lot to all contributors! Direct download links for the source archive: <ftp://ftp.berlios.de/pub/ngircd/ngircd-14.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-14.tar.gz> The ChangeLog can be found here: <http://ngircd.barton.de/doc/ChangeLog> <http://ngircd.berlios.de/doc/ChangeLog> GnuPG signature and patches from release 13 as well as 14~rc1 are available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ftp.berlios.de/pub/ngircd/> The relevant MD5 sums are: MD5 (ngircd-14.tar.gz) = 0080f9bb9a466a8852a5aedf0b51f016 MD5 (ngircd-13-14.patch.gz) = 91edaf2e35379b8011e307671367e067 You can habe a look at the complete history and every single patch using the GIT web-frontend located at: <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> Please let us know if you encounter any bugs or need more/better documentation (best is to file bugs using the bug tracker or to mail to this list). Thanks! Regards Alex

1 0

ngIRCd 14~rc1
by Alexander Barton 09 Apr '09

09 Apr '09

Hello! Quite a few fixes and enhancements have landed in our GIT development branch, so the next release of ngIRCd is becoming due. So I'm announcing the first release candidate for ngIRCd 14 with the following changes: - Updated Debian/Linux init script (see contrib/Debian/ngircd.init). - Allow creation of persistent modeless channels. - The INFO command reports the compile time now (if available). - Spell check and enhance ngIRCd manual pages. - Channel mode changes: break on syntax errors in MODE command. - Support individual channel keys for pre-defined channels: introduce new configuration variable "KeyFile" in [Channel] sections in ngircd.conf, here a file can be configured for each pre-defined channel which contains individual channel keys for different users. - Remove limit on maximum number of predefined channels in ngircd.conf. - Updated ngircd.spec file for building RPM packages. - Add new and missing files to Mac OS X Xcode project, and update project. - Reject masks with wildcard after last dot. - TLS/SSL: remove useless error message when ssl connection is closed. - Fix memory leak when a encrypted and compressed server link goes down. (closes bug #95, reported by Christoph) - Fix handling of channels containing dots. (closes ug #93, reported by Gonosz Csiga) You can download the release candidate here: <ftp://ftp.berlios.de/pub/ngircd/ngircd-14~rc1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-14~rc1.tar.gz> The GnuPG signature and a patch from release 13 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ftp.berlios.de/pub/ngircd/> You can habe a look at the complete history and every single patch using the GIT web-frontend located at: <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> Please test this release candidate and report all bugs you encounter using the bug tracker and/or this mailing list. Thanks! Regards Alex

1 0

[PATCH] misspelling in ERR_NORECIPIENT reply
by Dana Dahlstrom 25 Mär '09

25 Mär '09

--- src/ngircd/messages.h | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/ngircd/messages.h b/src/ngircd/messages.h index 209485a..9e15b81 100644 --- a/src/ngircd/messages.h +++ b/src/ngircd/messages.h @@ -94,7 +94,7 @@ #define ERR_TOOMANYCHANNELS_MSG "405 %s %s :You have joined too many channels" #define ERR_WASNOSUCHNICK_MSG "406 %s %s :There was no such nickname" #define ERR_NOORIGIN_MSG "409 %s :No origin specified" -#define ERR_NORECIPIENT_MSG "411 %s :No receipient given (%s)" +#define ERR_NORECIPIENT_MSG "411 %s :No recipient given (%s)" #define ERR_NOTEXTTOSEND_MSG "412 %s :No text to send" #define ERR_WILDTOPLEVEL "414 %s :Wildcard in toplevel domain" #define ERR_UNKNOWNCOMMAND_MSG "421 %s %s :Unknown command" -- 1.4.4.4

2 1

What about BOPM?
by Kurt Kraut 14 Feb '09

14 Feb '09

Hi, Have anyone tested BOPM [0] with ngircd? BOPM is a great defense system for IRC networks. It scans for vulnerability each incoming IP address in the IRC network and bans if something is found AND it checks in blacklists [1] if the user IP address has been envolved in IRC attacks or SPAM of virus spread. So, anyone interested in making ngircd work with BOPM? Thanks in advance, Kurt Kraut (listas(a)kurtkraut.net) [0] http://wiki.blitzed.org/BOPM [1] like http://www.dronebl.org/

1 0

ngIRCd Release 13
by Alexander Barton 26 Dez '08

26 Dez '08

Hi all! The next stable release of ngIRCd has been published: ngIRCd 13. Starting with this release, the version numering scheme has been canged: The leading "0" is stripped off, so we have "<major>[.<minor>]" now. So ngIRCd 13 is the direct successor of ngIRCd 0.12.1. The most prominet changes are support for SSL encrypted client-server and server-server links using OpenSSL or GNUTLS (see doc/SSL.txt), support for IRC services (see doc/Services.txt), and server-local channels with prefix "&". And as usual, lots of minor enhancements and bug fixes are included as well. Since the release candidate 13~rc1 only minor changes have been committed, mostly to enhance the documentation and to fix the testsuite. Thanks a lot to all contributors! In addition, the ngIRCd website has been revised. Thanks go to Götz Hoffart (layout) and Lucia Hoffart (english translation): <http://ngircd.barton.de/> <http://ngircd.berlios.de/> Direct download links for the source archive: <ftp://ftp.berlios.de/pub/ngircd/ngircd-13.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-13.tar.gz> The ChangeLog can be found here: <http://ngircd.barton.de/doc/ChangeLog> <http://ngircd.berlios.de/doc/ChangeLog> GnuPG signature and patches from 0.12.1 as well as 13~rc1 are available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ftp.berlios.de/pub/ngircd/> The relevant MD5 sums are: MD5 (ngircd-13.tar.gz) = f26064f0232e050ddba6f312ad34ec94 MD5 (ngircd-0.12.1-13.patch.gz) = 48b34eb4c33b631f3ac90908ec7e5777 MD5 (ngircd-13~rc1-13.patch) = 839b294bf18b6b24b8c07cf8e37b4a1e You can habe a look at the complete history and every single patch using the GIT web-frontend located at: <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> Please let me know if you encounter any bugs or need more/better documentation (best is to file bugs using the bug tracker or to mail to this list). Thanks! Regards Alex

1 0

Fwd: [IRCServices] Services 5.1.14 released
by Alexander Barton 07 Dez '08

07 Dez '08

Hi! Andrew Church released the next version of IRC Services today, which includes fixes that obsolete all required patches/workarounds to make them work with ngIRCd 13~rc1 (and the final release 13). Great news I think :-) Regards Alex Anfang der weitergeleiteten E-Mail: > Von: achurch(a)achurch.org (Andrew Church) > Datum: 7. Dezember 2008 13:12:30 MEZ > An: services <ircservices(a)ircservices.za.net> > Betreff: [IRCServices] Services 5.1.14 released > Antwort an: IRC Services General Mailing List <ircservices(a)ircservices.za.net > > > X-Mailer: MMail v5.30 > > Services 5.1.14 has been released, and can be downloaded from: > > http://www.ircservices.za.net/download/ (Japan) > ftp://ftp.esper.net/ircservices/ (Western USA) > > 54047dbfe9b8aa76f41abab48958c3b5 ircservices-5.1.14.tar.gz > 597dfecdd99bcfba0cce9cf06ed7f738 ircservices-5.1.14.diff.gz > f096cbd8d8bfe12c038ecb8523ea751f ircservices-5.1.14-1.i386.rpm > bb62c92efa10543ee1ce2c8c3399ecd0 ircservices_5.1.14-1_i386.deb > > The mirrors should have it shortly. > > This release fixes the various bugs that have been reported on the > mailing list over the past few months. In particular, a bug which > could > prevent autokick bans from working properly has been fixed. > > Changes in version 5.1.14 > ------------------------- > 2008/12/07 Fixed potential infinite loop on detecting a fatal error. > Reported by Jille Timmermans <jille(a)quis.cx> > 2008/12/07 Fixed race condition in which Services might fail to send > its initial data to the server if the connection took > too long. Reported by Alexander Barton <alex(a)barton.de> > 2008/12/07 Fixed bug causing incorrect #defines to be written to > config.h in non-English locales. Reported by Alexander > Barton <alex(a)barton.de> > 2008/12/07 Fixed bug causing nickname enforcers to remain online > indefinitely if the associated nickname was dropped > before the enforcer was removed. Reported by Jille > Timmermans <jille(a)quis.cx> > 2008/12/07 Fixed bug causing channel ban lists to get desynchronized, > leading to failure to add bans when autokicking users. > Reported (and fix suggested) by Aragon Gouveia > <aragon(a)phat.za.net> > > --Andrew Church > achurch(a)achurch.org > http://achurch.org/

1 0

ngIRCd 13~rc1
by Alexander Barton 22 Nov '08

22 Nov '08

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi! The next release of ngIRCd is on the horizon, and here we go with the first release candidate. Starting with this release, the version numering scheme will be canged: the leading "0" is stripped off, so we have "<major>[.<minor>]". The next version will be named "Release 13". The most prominet changes are support for SSL encrypted client-server and server-server links using OpenSSL or GNUTLS, support for IRC services, and server-local channels with prefix "&". And as usual, lots of minor enhancements and bug fixes are included as well. Thanks a lot to all contributors! You can download this release candidate here: <ftp://ftp.berlios.de/pub/ngircd/ngircd-13~rc1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-13~rc1.tar.gz> The GnuPG signature and a patch from 0.12.1 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ftp.berlios.de/pub/ngircd/> The ChangeLog can be found here: <http://ngircd.barton.de/doc/ChangeLog> <http://ngircd.berlios.de/doc/ChangeLog> You can habe a look at the complete history and every single patch using the GIT web-frontend located at: <http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git> Please test this release candidate and report all bugs you encounter using the bug tracker and/or this mailing list. Thanks! And again: thanks a lot to all users, testers, contributors and co-authors! Regards Alex -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) iD8DBQFJJ1HuwkoPY342SFYRAm95AJ9arBix0Mlsu4eJH7Pg+hRjI/yvkACg37rh lxZFj4SJ7WnHJ+vZZZAVJcc= =DMRW -----END PGP SIGNATURE-----

1 0

Re: [ngIRCd-ML] ngIRCd and Services
by Alexander Barton 11 Nov '08

11 Nov '08

Hello Robert! Am 08.11.2008 um 19:08 schrieb Robert Pendell: > Ok. I've compiled and configured both the latest git code and > ircservices 5.1.13. I've setup ngircd per what you have stated > however > I still cannot get it to work. I used the rfc1483 module in it and it > never gets a connection. It tries once then dies however it does > indicate that it loaded successfully which at least means that there > isn't any syntax errors in the config. Attached are "sanitized" > versions of my config files. They are sanitized in that there are no > personal passwords there. I changed them for privacy. > > Here are the lines I am getting in my log file > > [Nov 08 13:05:10 2008] IRC Services 5.1.13 starting up > [Nov 08 13:05:31 2008] Read error from server: Connection reset by > peer This seems to be an other bug of IRCServices that I forgot to mention and had on all my boxes, as well: IRCServices don't flush the write buffer when the connection to the server is established and therefore never sends the USER and NICK commands to log in. Which in turn causes the server to disconnect IRCServices again because of no commands were received. Solution #1: Send IRCServices a HUP signal right after starting it up (kill -HUP <PID>). Solution #2: I fixed the bug in IRCServices, see my post to the IRCServices mailing list: <http://lists.ircservices.za.net/pipermail/ircservices/2008/005658.html> So you can a) apply this patch or b) wait for an IRCServices release with this patch. Regards Alex

1 0

ngIRCd and Services
by Robert Pendell 08 Nov '08

08 Nov '08

Currently I am running the latest development code from the git repository. What is the status of services support in the code? Are there any irc services that will work with ngircd yet? P.S. - I realize the "when it's done" think. Just need to know what kind of progress you are making. -- Robert Pendell shinji(a)elite-systems.org "A perfect world is one of chaos." Thawte Web of Trust Notary CAcert Assurer

3 3

Fwd: [IRCServices] Services 5.1.13 released
by Alexander Barton 07 Nov '08

07 Nov '08

Hello! Attached is the announcement of IRCServices 5.1.13, which fixes the bug that prevented ngIRCd as of the GIT master development version to use this IRC services package without patching. Rehards Alex > Von: achurch(a)achurch.org (Andrew Church) > Datum: 7. November 2008 18:41:28 MEZ > An: services <ircservices(a)ircservices.za.net> > Betreff: [IRCServices] Services 5.1.13 released > Antwort an: IRC Services General Mailing List <ircservices(a)ircservices.za.net > > > X-Mailer: MMail v5.30 > > Services 5.1.13 has been released, and can be downloaded from: > > http://www.ircservices.za.net/download/ (Japan) > ftp://ftp.esper.net/ircservices/ (Western USA) > > c124a94038f99a4bcd233834489ecc8b ircservices-5.1.13.tar.gz > 53523416d77964051ef842f684b90f78 ircservices-5.1.13.diff.gz > a113f2c7c38a11b271769ccfa205cc7a ircservices-5.1.13-1.i386.rpm > f2114c76a7b272e7e7c03ac2950c8e46 ircservices_5.1.13-1_i386.deb > > The mirrors should have it shortly. > > This release includes the rfc1459 protocol module fix recently > mentioned > on the mailing list. There are no other changes in this release. > > Changes in version 5.1.13 > ------------------------- > 2008/11/08 Fixed bug preventing remote users from being recognized by > the rfc1459 protocol module. Reported by Alexander > Barton <alex(a)barton.de> > > --Andrew Church > achurch(a)achurch.org > http://achurch.org/ > ------------------------------------------------------------------ > To unsubscribe or change your subscription options, visit: > http://lists.ircservices.za.net/mailman/listinfo/ircservices

1 0

ngIRCD-ML on GMANE
by Alexander Barton 16 Aug '08

16 Aug '08

Hi all! I just wanted to let you know that this mailing list is now include in the GMANE archive: <http://dir.gmane.org/gmane.network.irc.ngircd> Citing <http://gmane.org/>: "Gmane is a mailing list archive. Any public mailing list can be carried by Gmane, but the vast majority of the lists here deal with free software. For more information about the project, read the about page [<http://gmane.org/about.php>]. There are currently 10,359 mailing lists subscribed to Gmane, with a total of 60,457,704 messages." GMANE allows you to read this mailing list via different Web frontends, RSS feeds or via NNTP (Usenet). Please see our information page on GMANE mentioned above. Regards Alex PS.: The archive of this mailing list has been imported and is already available. Thanks to the folks at GMANE!

1 0

ngIRCd 0.12.1
by Alexander Barton 17 Jul '08

17 Jul '08

Hello World! ngIRCd 0.12.1 has been assembled a few days before and now officially released into the public. It can be downloaded from the usual locations: a) ftp://ngircd.barton.de/pub/ngircd/ ftp://ngircd.barton.de/pub/ngircd/ngircd-0.12.1.tar.gz b) ftp://ftp.berlios.de/pub/ngircd/ ftp://ftp.berlios.de/pub/ngircd/ngircd-0.12.1.tar.gz Patches from 0.12.0 as well as GnuGP signatures can be found there, too. Quoting the ChangeLog: - Allow mixed line terminations (CR+LF/CR/LF) in non-RFC-compliant mode - Don't allow stray \r or \n in command parameters - --configtest: return non-zero exit code if there are errors - Update ngIRCd manual pages - Add option aliases -V (for --version) and -h (for --help). - Fix 'no-ipv6' compile error. - Make Listen parameter a comma-seperated list of addresses. This also obsoletes ListenIPv4 and ListenIPv6 options. If Listen is unset, it is treated as Listen="::,0.0.0.0". Note: ListenIPv4 and ListenIPv6 options are still recognized, but ngircd will print a warning if they are used in the config file. Have fun! Regards Alex

1 0

Hash table backing store patch (and general optimization thoughts)
by Scott Perry 27 Mai '08

27 Mai '08

This message is for discussion, the patches attached are not intended to be committed quite yet. I'd like to spend at least one more night of heavy break-in testing before finally signing off on my modified patch or the original. Florian sent me a patch (attached: hashtable.diff) last week or so. Applying it changes the data structure backing store for users and channels from using linked lists to using a hash table. I wrote a test harness and ran 1000 clients against a local ngircd using the hashtable patch to test for memory leaks and logical errors. The test clients ran in a tight loop, and as often as possible they would change their nicks, join and/or part channels, send messages to channels (existent and not) for 18 hours. While using my own local client to watch the carnage, the ircd stayed very responsive and the test harness did not experience any errors with data constistency when comparing each client's local channel list with the results of WHOIS. Other than memory issues already present (getpwd() in libc leaks a small amount of memory, but it is run once at startup), there were no memory leaks that I could find. I'm going to retest the ircd on a different system tonight (one where I have better tools for profiling and memory) to ensure that there are no memory issues. I've attached my own modified version of the patch (hashtable.mine.diff) for subsequent review. There is only one notable change, in Client_Destroy(), where I've restructured nearly the entire function. Comments on the patch are inline below (please note that the quoted text is *not* the complete patch. the quoted text is still in valid diff syntax, but I removed large portions for brevity). If you just want the final thoughts, jump to the end of this message. > Index: ngircd/src/ngircd/channel.c > =================================================================== > --- ngircd.orig/src/ngircd/channel.c 2008-05-25 10:52:59.000000000 > -0700 > +++ ngircd/src/ngircd/channel.c 2008-05-25 13:05:55.000000000 -0700 this is a huge win. every single channel search goes from being linear in time to almost constant time. > @@ -226,26 +217,14 @@ Channel_Part(CLIENT * Client, CLIENT * O > assert(Name != NULL); > assert(Reason != NULL); > > - /* Check that specified channel exists */ > - chan = Channel_Search(Name); > - if (!chan) { > - IRC_WriteStrClient(Client, ERR_NOSUCHCHANNEL_MSG, > - Client_ID(Client), Name); > - return false; > - } > - > - /* Check that the client is in the channel */ > - if (!Get_Cl2Chan(chan, Client)) { > - IRC_WriteStrClient(Client, ERR_NOTONCHANNEL_MSG, > - Client_ID(Client), Name); > + chan = Channel_Search( Name ); > + if(( ! chan ) || ( ! Cl2Chan_Get( chan, Client ))) > + { > + IRC_WriteStrClient( Client, ERR_NOSUCHCHANNEL_MSG, > Client_ID( Client ), Name ); > return false; > } > > - /* Part client from channel */ > - if (!Remove_Client(REMOVE_PART, chan, Client, Origin, Reason, true)) > - return false; > - else > - return true; > + return Remove_Client( REMOVE_PART, chan, Client, Origin, Reason, > true); > } /* Channel_Part */ > > > @@ -896,46 +711,43 @@ Remove_Client( int Type, CHANNEL *Chan, retaining the variable c to represent the channel structure instead of using the variable Chan in this change would make this portion of the diff a lot less intimidating and invasive. However, I did not remove this portion of the diff in my version because it cleans up the method considerably, making variable naming throughout the function much more consistent. > * so only inform other clients in same channel. */ > assert( InformServer == false ); > LogDebug("User \"%s\" left channel \"%s\" (%s).", > - Client_Mask( Client ), c->name, Reason ); > + Client_Mask( Client ), Chan->name, Reason ); > break; > case REMOVE_KICK: > /* User was KICKed: inform other servers and all users in channel > */ > if( InformServer ) > IRC_WriteStrServersPrefix( Client_NextHop( Origin ), > - Origin, "KICK %s %s :%s", c->name, Client_ID( Client ), Reason); > - IRC_WriteStrChannelPrefix(Client, c, Origin, false, "KICK %s %s : > %s", > - c->name, Client_ID( Client ), Reason ); > + Origin, "KICK %s %s :%s", Chan->name, Client_ID( Client ), > Reason); > + IRC_WriteStrChannelPrefix(Client, Chan, Origin, false, "KICK %s > %s :%s", > + Chan->name, Client_ID( Client ), Reason ); > if ((Client_Conn(Client) > NONE) && > (Client_Type(Client) == CLIENT_USER)) > { > IRC_WriteStrClientPrefix(Client, Origin, "KICK %s %s :%s", > - c->name, Client_ID( Client ), Reason); > + Chan->name, Client_ID( Client ), Reason); > } > LogDebug("User \"%s\" has been kicked off \"%s\" by \"%s\": %s.", > - Client_Mask( Client ), c->name, Client_ID(Origin), Reason); > + Client_Mask( Client ), Chan->name, Client_ID(Origin), Reason); > break; > default: /* PART */ > if (InformServer) > - IRC_WriteStrServersPrefix(Origin, Client, "PART %s :%s", c- > >name, Reason); > + IRC_WriteStrServersPrefix(Origin, Client, "PART %s :%s", Chan- > >name, Reason); > > - IRC_WriteStrChannelPrefix(Origin, c, Client, false, "PART %s :%s", > - c->name, Reason); > + IRC_WriteStrChannelPrefix(Origin, Chan, Client, false, "PART %s : > %s", > + Chan->name, Reason); > > if ((Client_Conn(Origin) > NONE) && > (Client_Type(Origin) == CLIENT_USER)) > { > - IRC_WriteStrClientPrefix( Origin, Client, "PART %s :%s", c- > >name, Reason); > + IRC_WriteStrClientPrefix( Origin, Client, "PART %s :%s", Chan- > >name, Reason); > LogDebug("User \"%s\" left channel \"%s\" (%s).", > - Client_Mask(Client), c->name, Reason); > + Client_Mask(Client), Chan->name, Reason); > } > } > > - /* Wenn Channel nun leer und nicht pre-defined: loeschen */ > - if( ! strchr( Channel_Modes( Chan ), 'P' )) > - { > - if( ! Get_First_Cl2Chan( NULL, Chan )) Delete_Channel( Chan ); > - } > - > + /* if channel empty and not predefined -> remove */ > + if ((hashtable_getfillcount(&Chan->members) == 0) && ! > strchr(Channel_Modes( Chan ), 'P')) > + hashtable_delete(&My_Channels, Chan->name); > return true; > } /* Remove_Client */ > > Index: ngircd/src/ngircd/cl2chan.h > =================================================================== > --- /dev/null 1970-01-01 00:00:00.000000000 +0000 > +++ ngircd/src/ngircd/cl2chan.h 2008-05-25 13:05:55.000000000 -0700 separating out the cl2chan functions is a really good refactoring move as far as code cleanup goes. > Index: ngircd/src/ngircd/client.c > =================================================================== > --- ngircd.orig/src/ngircd/client.c 2008-05-25 10:52:59.000000000 > -0700 > +++ ngircd/src/ngircd/client.c 2008-05-25 13:05:55.000000000 -0700 again this looks more like it could have been it's own patch instead of being included here. that said, it works properly; since it's already here we may as well merge it in. > @@ -141,8 +146,17 @@ Client_ThisServer( void ) > GLOBAL CLIENT * > Client_NewLocal(CONN_ID Idx, char *Hostname, int Type, bool Idented) > { > - return Init_New_Client(Idx, This_Server, NULL, Type, NULL, NULL, > - Hostname, NULL, 0, 0, NULL, Idented); > + /* # is illegal in a nick name, so #+CONN_ID is unique. > + * We need this as a placeholder until the clients sets > + the real nick. */ > + char fake_nick[CLIENT_NICK_LEN]; > + > + snprintf(fake_nick, sizeof fake_nick, "#%x#", Idx); > +#ifdef DEBUG > + assert(Client_Search(fake_nick) == NULL); > +#endif > + return Init_New_Client(Idx, This_Server, NULL, Type, fake_nick, > NULL, > + Hostname, NULL, 0, 0, NULL, Idented); > } /* Client_NewLocal */ > > > @@ -241,93 +269,110 @@ Client_Destroy( CLIENT *Client, char *Lo > strlcat(msg, Client->id, sizeof (msg)); > } > > - last = NULL; > - c = My_Clients; > - while( c ) > +#ifdef DEBUG > + strlcpy(nickbuf, Client->id, sizeof nickbuf); > + if (hashtable_get(&My_Clients, nickbuf) == NULL) > + LogDebug("Client_Destroy() called for unknown nickname %s!", > nickbuf); > +#endif > + hashtable_iterator_reset(&My_Clients); > + while(( c = hashtable_iterator_getnext(&My_Clients))) > { > - if(( Client->type == CLIENT_SERVER ) && ( c->introducer == > Client ) && ( c != Client )) > - { > - /* der Client, der geloescht wird ist ein Server. Der Client, > den wir gerade > - * pruefen, ist ein Child von diesem und muss daher auch > entfernt werden */ > - Client_Destroy( c, NULL, msg, false ); > - last = NULL; > - c = My_Clients; > + LogDebug("Testing client nick: %s, want %s", c->id, Client->id); > + I think there's a serious optimization to be made here: If Client is not a server, then no iteration is necessary, so maybe separating this function into two functional portions, one for normal clients and one for servers, would be a good idea, since this function is needlessly inefficient if the Client is a server. This is an especially large bottleneck with a multiserver network, since destroying a server would have taken O(n^2) as written in this patch. This portion of the code was changed in my patch. > + if (c != Client) { > + if ((Client->type == CLIENT_SERVER) && (c->introducer == > Client)) { > + /* The client that is about to be removed is a server, > + * the client we are checking right now is a child of that > + * server and thus has to be removed, too. */ > + Client_Destroy( c, NULL, msg, false ); > + hashtable_iterator_reset(&My_Clients); > + } > continue; > } > - if( c == Client ) > - { > - /* Wir haben den Client gefunden: entfernen */ > - if( last ) last->next = c->next; > - else My_Clients = (CLIENT *)c->next; > - > - if( c->type == CLIENT_USER ) > - { > - if( c->conn_id != NONE ) > - { > - /* Ein lokaler User */ > - Log( LOG_NOTICE, "User \"%s\" unregistered (connection %d): > %s", Client_Mask( c ), c->conn_id, txt ); > - > - if( SendQuit ) > - { > - /* Alle andere Server informieren! */ > - if( FwdMsg ) IRC_WriteStrServersPrefix( NULL, c, "QUIT :%s", > FwdMsg ); > - else IRC_WriteStrServersPrefix( NULL, c, "QUIT :" ); > + /* c == Client */ …<snip>… > + LogDebug("Remove Nick \"%s\" from hashtable (Destroy)", c->id); > + hashtable_delete(&My_Clients, c->id); > + break; > } > + hashtable_iterator_reset(&My_Clients); > + > +#ifdef DEBUG > + assert(hashtable_get(&My_Clients, nickbuf) == NULL); > +#endif > } /* Client_Destroy */ > > > @@ -575,13 +635,14 @@ Client_GetFromToken( CLIENT *Client, int this looks a lot less efficient than it could be. luckily, it's called only once during a connection's lifetime, but if a connection has not been fully registered and logged in, than why not use the token (or a variation of it) as the hashtable's key? > assert( Client != NULL ); > assert( Token > 0 ); > > - c = My_Clients; > - while( c ) > - { > - if(( c->type == CLIENT_SERVER ) && ( c->introducer == Client ) && > ( c->token == Token )) return c; > - c = (CLIENT *)c->next; > + hashtable_iterator_reset(&My_Clients); > + while((c = hashtable_iterator_getnext(&My_Clients))) { > + if(( c->type == CLIENT_SERVER ) && > + ( c->introducer == Client ) && > + ( c->token == Token )) break; > } > - return NULL; > + hashtable_iterator_reset(&My_Clients); > + return c; > } /* Client_GetFromToken */ > > > @@ -884,12 +939,11 @@ Client_MyServerCount( void ) this comment applies to all the Client_*Count() functions: performance in these functions is very unfortunate, changing from a hash table to a linked list does not make this any faster, but if performance in these sections of the code is important, then the counts should probably be cached. for extra credit, the cached values could be updated in some way that does not include a O(n) search of the hash table. > CLIENT *c; > unsigned long cnt = 0; > > - c = My_Clients; > - while( c ) > - { > + hashtable_iterator_reset(&My_Clients); > + while ((c = hashtable_iterator_getnext(&My_Clients))) > if(( c->type == CLIENT_SERVER ) && ( c->hops == 1 )) cnt++; > - c = (CLIENT *)c->next; > - } > + > + hashtable_iterator_reset(&My_Clients); > return cnt; > } /* Client_MyServerCount */ > > @@ -1062,18 +1118,14 @@ Generate_MyToken( CLIENT *Client ) this function could also stand to use some help from caching. it seems raucously inefficient. > CLIENT *c; > int token; > > - c = My_Clients; > token = 2; > - while( c ) > - { > - if( c->mytoken == token ) > - { > - /* Das Token wurde bereits vergeben */ > + hashtable_iterator_reset(&My_Clients); > + while ((c = hashtable_iterator_getnext(&My_Clients))) { > + if( c->mytoken == token ) { > + /* Token already in use */ > token++; > - c = My_Clients; > - continue; > + hashtable_iterator_reset(&My_Clients); > } > - else c = (CLIENT *)c->next; > } > Client->mytoken = token; > Log( LOG_DEBUG, "Assigned token %d to server \"%s\".", token, > Client->id ); > Index: ngircd/src/ngircd/hash.c > =================================================================== > --- ngircd.orig/src/ngircd/hash.c 2008-05-25 10:52:59.000000000 -0700 > +++ ngircd/src/ngircd/hash.c 2008-05-25 13:05:55.000000000 -0700 > @@ -19,16 +20,33 @@ static char UNUSED id[] = "$Id: hash.c,v > #include "imp.h" > #include <assert.h> > #include <string.h> > +#include <errno.h> > +#include <stdlib.h> > > #include "defines.h" > +#include "log.h" > #include "tool.h" > > #include "exp.h" > #include "hash.h" > allowing some way for the builder to set this at configure-time (instead of adding -DNITIAL_HASHSIZE in their CFLAGS) would probably be much appreciated, since network size varies enormously. if nothing else, the initial size should err on the side of being too large, as hash table performance does not suffer considerably from being too empty. > +#define INITIAL_HASHSIZE 23 > + > +static struct bucket *hashtable_bucket_get(hashtable *h, const void > *key); > +static struct bucket *hashtable_bucket_steal(hashtable *h, const > void *key); > > static UINT32 jenkins_hash PARAMS(( register UINT8 *k, register > UINT32 length, register UINT32 initval )); > > +/* Enable Extra Debug Output in Hash Insertion/Removal/Iteration > code. > + * This creates a _LOT_ of debug messages. */ > +/* #define DEBUG_HASH */ > + > +#ifdef DEBUG > +static void hashtable_stats(const hashtable * const h); > +#else > +static inline void hashtable_stats(const UNUSED hashtable * const > h) { } > +#endif > + > > GLOBAL UINT32 > Hash( const char *String ) > @@ -122,4 +140,493 @@ jenkins_hash( register UINT8 *k, registe > } /* jenkins_hash */ > > > -/* -eof- */ > +static size_t get_larger_prime(size_t old) > +{ > + size_t i; since I've never seen an IRC network with more than 6664223 users, this is probably fine. having artificial limitations allows for a much faster get_larger_primes() than would otherwise be possible. I'd suggest using consecutive primes that are approximately twice the value of the previous in the list (extra credit: as close to a power of 2 without exceeding). > + static const size_t primes[] = { 223, 523, 823, 1123, > + 1523, 1823, 2423, 4523, 8123, > + 20323, 28723, 47123, 76423, > + 96223, 133723, 220123, > + 386723, 647723, 999623, > + 1923323, 4735823, 6664223 }; > + > + for (i=0; i < sizeof primes/sizeof primes[0];i++) { > + if (old < primes[i]) > + return primes[i]; > + } > +#ifdef DEBUG > + Log(LOG_DEBUG, "Could not find larger prime than %u", old); > +#endif > + return old*2 + 1; > +} …<snip>… > +static struct bucket *hashtable_iterator_getnextlist(hashtable *h, > size_t startpos) > +{ > + struct bucket *b = NULL; > + size_t i = startpos; > + > + assert(h->hash_size > 0); > + while (i < h->hash_size) { > + b = h->buckets[i]; > + if (b) break; > + i++; > + } > + > + h->iterator.nextres = b; > + h->iterator.curhash = i; > +#ifdef DEBUG_HASH > + LogDebug("hashtable_iterator_getnextlist: found %sresult, start at > pos %u, now at %u", > + b?"":"no ", startpos,i); > +#endif > + return b; > +} > + > + while we're here, it's a good time to mention: iterating over a hash table should be done as little as possible. cscope reports that there are 16 places in the code that call hashtable_iterator_getnext(), which is probably at least 15 too many. > +void *hashtable_iterator_getnext(hashtable *h) > +{ > + void *retval; > + struct bucket *nextres; > + > + assert(h != NULL); > + nextres = h->iterator.nextres; > +#ifdef DEBUG_HASH > + if (!nextres) LogDebug("hashtable_iterator_getnext(): initial h- > >iterator.nextres == NULL"); > +#endif > + if (h->iterator.curhash == 0 && !nextres) > + nextres = hashtable_iterator_getnextlist(h, 0); > + > + if (!nextres || h->iterator.curhash >= h->hash_size) > + return NULL; > + > + retval = nextres->elemptr; > + h->iterator.nextres = nextres->next; > + > + if (!h->iterator.nextres) /* no more buckets in this list, find > next list */ > + (void)hashtable_iterator_getnextlist(h, h->iterator.curhash+1); > +#ifdef DEBUG_HASH > + LogDebug("hashtable_iterator_getnext(): %s, next result: %s", > + retval ? "ok": "(NULL)", h->iterator.nextres ? "yes" :"(NULL)"); > + if (h->iterator.nextres) > + assert(hashtable_checkref(h, h->iterator.nextres->keyptr) == 1); > +#endif > + return retval; > +} > + > + as per my recent change using strtok_r() in IRC_JOIN(), it's probably worth looking into allowing a caller to provide/use their own iterator context variables. as much as iterating a hash table is awful, sometimes it's necessary. this is more a note for the future, as adding support for iterator contexts provided by the caller would be an unagreeably large change to add to an already large patch. > +void hashtable_iterator_reset(hashtable *h) > +{ > + assert(h != NULL); > + h->iterator.curhash = 0; > + h->iterator.nextres = NULL; > +} > + …<snip>… it's good that this function is provided, but it should be called as little as possible via allowing the size of the table to be set easily at compile-time or (even more flexible) in the config file. > +static bool rehash(hashtable *h, size_t newsize) > +{ > + hashtable newhash; > + size_t i; > +#ifdef DEBUG > + unsigned int collisions = 0; > +#endif > + > + hashtable_stats(h); > + > +bool hashtable_add(hashtable *h, const void *key, void *item) > +{ > + size_t pos; > + struct bucket *b; > + > + assert(h->hash_size > 0); > + assert(key != NULL); > + assert(item != NULL); > + > + if ((h->hash_fillcount * 100 / h->hash_size) > 90) > + rehash(h, get_larger_prime(h->hash_size)); > + > + errno = 0; > + if (hashtable_bucket_get(h, key)) { > + errno = EEXIST; /* dupe */ > + return false; > + } > + > + b = bucket_alloc(); > + if (!b) return false; > + There's nothing in the hash table's implementation that seems to require a prime table size (usually only required by doublehashing, which does not use a list of buckets on collisions). A minor optimization could include enforcing a power-of-2 table size, since compilers for some platforms compile modulus to a multiply and divide, whereas modulus by a power of 2 is strength-reduced by all at-least- semi intelligent compilers as a simple and instruction. Using powers of 2 for the table size would also make the get_larger_prime function faster and easier, if not misleadingly named. > + pos = h->hashfunc(key) % h->hash_size; > + b->keyptr = key; > + b->elemptr = item; > + b->next = h->buckets[pos]; > + h->buckets[pos] = b; > + > + h->hash_fillcount++; > +#ifdef DEBUG_HASH > + LogDebug("hashtable_insert(): pos %u", pos); > + assert(hashtable_checkref(h, key) == 1); > +#endif > + return true; > +} > + > + …<snip>… > Index: ngircd/src/ngircd/irc-info.c > =================================================================== > --- ngircd.orig/src/ngircd/irc-info.c 2008-05-25 10:52:59.000000000 > -0700 > +++ ngircd/src/ngircd/irc-info.c 2008-05-25 13:05:55.000000000 -0700 there should probably be a faster way to iterate through servers than this. iterating through all open connections feels a lot like using a shotgun in a crowd to kill a pickpocket. for a future optimization, maybe servers and users could warrant separate tables? > @@ -225,7 +226,7 @@ IRC_LINKS( CLIENT *Client, REQUEST *Req > { > if( ! IRC_WriteStrClient( target, RPL_LINKS_MSG, > Client_ID( target ), Client_ID( c ), > Client_ID( Client_TopServer( c ) ? Client_TopServer( c ) : > Client_ThisServer( )), Client_Hops( c ), Client_Info( c ))) return > DISCONNECTED; > } > - c = Client_Next( c ); > + c = Client_Next(); > } > > IRC_SetPenalty( target, 1 ); this is slow, but so is calling /names on any network. for future scaling, it might be worth looking into supporting the ability to return a "server too busy" reply if the server's input socket has waiting data being blocked by running this command? > @@ -382,7 +383,7 @@ IRC_NAMES( CLIENT *Client, REQUEST *Req > if( ! IRC_Send_NAMES( from, chan )) return DISCONNECTED; > > /* naechster Channel */ > - chan = Channel_Next( chan ); > + chan = Channel_Next(); > } > > /* Nun noch alle Clients ausgeben, die in keinem Channel sind */ > Index: ngircd/src/ngircd/irc-login.c > =================================================================== > --- ngircd.orig/src/ngircd/irc-login.c 2008-05-25 10:54:27.000000000 > -0700 > +++ ngircd/src/ngircd/irc-login.c 2008-05-25 13:05:55.000000000 -0700 this change is related to the change in Client_SetID() (more precisely, Client_NewLocal() and Init_New_Client() ) (where a unique identifier is used for clients that have not logged in, but are connected). It probably should have been part of a different change, but it makes a lot of use of the new hash table backend, so extracting it out is not worth the trouble. > @@ -244,7 +244,9 @@ IRC_NICK( CLIENT *Client, REQUEST *Req ) > Client_Conn( Client )); > > /* Register new nickname of this client */ > - Client_SetID( target, Req->argv[0] ); > + if (!Client_SetID( target, Req->argv[0] )) > + return IRC_WriteStrClient( Client, ERR_NICKNAMEINUSE_MSG, > + Client_ID( Client ), Req->argv[0] ); > > /* If we received a valid USER command already then > * register the new client! */ > Index: ngircd/src/ngircd/irc-oper.c > =================================================================== > --- ngircd.orig/src/ngircd/irc-oper.c 2008-05-25 10:52:59.000000000 > -0700 > +++ ngircd/src/ngircd/irc-oper.c 2008-05-25 13:05:55.000000000 -0700 > @@ -292,7 +292,7 @@ IRC_WALLOPS( CLIENT *Client, REQUEST *Re again with the iterating. this is really inefficient and something to look at in the future. > if (!from) > return IRC_WriteStrClient(Client, ERR_NOSUCHNICK_MSG, > Client_ID(Client), Req->prefix); > > - for (to=Client_First(); to != NULL; to=Client_Next(to)) { > + for (to=Client_First(); to != NULL; to=Client_Next()) { > if (Client_Conn(to) < 0) /* no local connection or WALLOPS origin */ > continue; > > Index: ngircd/src/ngircd/irc-write.c > =================================================================== > --- ngircd.orig/src/ngircd/irc-write.c 2008-05-25 10:52:59.000000000 > -0700 > +++ ngircd/src/ngircd/irc-write.c 2008-05-25 13:05:55.000000000 -0700 > @@ -310,7 +311,7 @@ va_dcl again, sending messages to many other servers should not require iterating through the entirety of the client list. > /* Ziel-Server gefunden. Nun noch pruefen, ob Flags stimmen */ > if(( Flag == '\0' ) || ( strchr( Client_Flags( c ), Flag ) != > NULL )) IRC_WriteStrClientPrefix( c, Prefix, "%s", buffer ); > } > - c = Client_Next( c ); > + c = Client_Next(); > } > } /* IRC_WriteStrServersPrefixFlag */ > > Index: ngircd/src/ngircd/irc.c > =================================================================== > --- ngircd.orig/src/ngircd/irc.c 2008-05-25 10:52:59.000000000 -0700 > +++ ngircd/src/ngircd/irc.c 2008-05-25 13:05:55.000000000 -0700 > @@ -279,7 +279,7 @@ IRC_TRACE( CLIENT *Client, REQUEST *Req again, iterating for servers should not require iterating through the entirety of the client list. > if( ! IRC_WriteStrClient( from, RPL_TRACEOPERATOR_MSG, > Client_ID( from ), Client_ID( c ))) return DISCONNECTED; > } > } > - c = Client_Next( c ); > + c = Client_Next(); > } > > IRC_SetPenalty( Client, 3 ); Final thoughts: Most of the lines in this diff are actually due to function renaming or variable renaming. Some of this was necessary (the arguments to Client_Next() and Channel_Next() changed, for instance), but for the most part the function name changes can be seen as code cleanup. Some portions of it are valid bugfixes that could have been split out, but in a few cases the bugfixes interact closely with the new code, so it's very justifiable to include them. As I've thoroughly reviewed and tested this patch, I feel that despite it's size, it is not unnecessarily large or risky. Users and servers (and in the future, services) should probably be stored in separate hash tables. Iterating over a hash table of the servers will be notably faster than iterating over a list of connections and filtering out all clients that are not servers. Almost anywhere that Client_Next() or Channel_Next() is called is a possible place to optimize. In a lot of cases Client_Next() was being used, but most of the set was not being used (when trying to iterate over all servers or opers, for instance). This would be a considerable speedup, especially since iterating over the hash table takes longer than iterating over a linked list by approximately the m/ n fill ratio. As ngircd grows and gets used on larger networks, it might be worth finding expensive operations that cannot be sped up more (like /names, or /list) and return a "server too busy" message if processing the command would result in too many waiting inputs going unprocessed for too long. Multithreading would fix this, but most of the code (especially the hash table in this patch and the linked list currently in use) is not thread safe. /* comments welcome */ ./scott

3 8

[patch] use strtok_r instead of strchr in IRC_JOIN
by Scott Perry 26 Mai '08

26 Mai '08

This patch does significant cleanup on the join code by using strtok_r instead of mangling strchr to parse channel names and keys in parallel when a JOIN command contains a list of channels and keys. since strtok_r is not universally portable, Florian was kind enough to provide a working /src/portab/strtok_r.c which I've included into the configure path. I'd like to more widely move the codebase from using strtok to using strtok_r where possible. A few reasons for this: 1) strtok_r and strtok are the same speed and functionality. there are no compelling reasons to pick strtok() over strtok_r() outside of portability. Since strtok_r is fairly universal (all modern operating systems support it in their standard libraries), and a solid implementation exists in /src/portab, portability is not a compelling concern. 2) strtok_r will alleviate concerns about program state in the future. This is important because functions may need to tokenize strings while a function higher in the stack has not completed tokenization on a different string. This would lead to incorrect behaviour. By using strtok_r, this is no longer a concern when adding features or fixing bugs. There are other reasons, but most of them relate to thread-safety. While ngircd could benefit from having worker threads processing server input (since reading from a socket is a low cost operation compared to actually processing the input), this isn't a very good argument for ngircd at the moment. There are more useful ways to make ngircd faster at the moment that don't include reorganizing vast swaths of code. More on that later. Anyway, back to the original point: this patch cleans up IRC_JOIN considerably and provides strtok_r() for future use through /src/portab. The attached patch applies with no warnings to the latest git head. ./scott

3 2

Reloading certificate upon SIGHUP?
by Christoph Biedl 16 Mai '08

16 Mai '08

Hello, I am using ngircd 0.11 with gnutls support enabled. For a certain reason (guess why) I had to change the certificate but appearently after a reload ngircd still presents the old certificate. Did I miss something (or did I something wrong)? Else I'd like to add a feature request to reload the SSL credentials upon SIGHUP. Christoph

3 4

[PATCH] Add tests for MOTD, VERSION, TIME, USERHOST
by Brandon Beresini 15 Mai '08

15 Mai '08

The misc-test.e file is also somewhat reorganized and commented here. Note the old $Id$ tag from CVS remains and won't be updated. (Thoughts?) --- src/testsuite/misc-test.e | 127 +++++++++++++++++++++++++++++++++++++++++++- 1 files changed, 124 insertions(+), 3 deletions(-) diff --git a/src/testsuite/misc-test.e b/src/testsuite/misc-test.e index 4b88379..f9142df 100644 --- a/src/testsuite/misc-test.e +++ b/src/testsuite/misc-test.e @@ -13,26 +13,147 @@ expect { "376" } +# RFC 2812 Section 3.4.1 + +send "motd\r" +expect { + timeout { exit 1 } + "375" +} +expect { + timeout { exit 1 } + "372" +} +expect { + timeout { exit 1 } + "376" +} + +send "motd ngircd.test.server\r" +expect { + timeout { exit 1 } + "375" +} +expect { + timeout { exit 1 } + "372" +} +expect { + timeout { exit 1 } + "376" +} + +send "motd doesnotexist\r" +expect { + timeout { exit 1 } + "402" +# note this is not specified in RFC 2812, but probably should be +} + +# RFC 2812 Section 3.4.3 + +send "version\r" +expect { + timeout { exit 1 } + "351" +} + +send "version ngircd.test.server\r" +expect { + timeout { exit 1 } + "351" +} + +send "version doesnotexist\r" +expect { + timeout { exit 1 } + "402" +} + +# RFC 2812 Section 3.4.6 + +send "time\r" +expect { + timeout { exit 1 } + "391" +} + +send "time ngircd.test.server\r" +expect { + timeout { exit 1 } + "391" +} + +send "time doesnotexist\r" +expect { + timeout { exit 1 } + "402" +} + +# RFC 2812 Section 3.4.10 + +send "info\r" +expect { + timeout { exit 1 } + "371" +} +expect { + timeout { exit 1 } + "374" +} + +# RFC 2812 Section 4.5 + send "summon\r" expect { timeout { exit 1 } "445" } +# RFC 2812 Section 4.6 + send "users\r" expect { timeout { exit 1 } "446" } -send "info\r" +# RFC 2812 Section 4.8 + +send "userhost\r" expect { timeout { exit 1 } - "371" + "461" } + +send "userhost nick\r" expect { timeout { exit 1 } - "374" + -re ":ngircd.test.server 302 nick :?nick=+.*@(localhost|127.0.0.1)" +} + +send "userhost doesnotexist\r" +expect { + timeout { exit 1 } + ":ngircd.test.server 302 nick :\r" +} + +send "userhost nick doesnotexist nick doesnotexist\r" +expect { + timeout { exit 1 } + -re ":ngircd.test.server 302 nick :nick=+.*@(localhost|127.0.0.1) nick=+.*(a)(localhost|127.0.0.1)" +} + +send "away :testing\r" +expect { + timeout { exit 1 } + "306 nick" +} + +send "userhost nick nick nick nick nick nick\r" +expect { + timeout { exit 1 } + -re ":ngircd.test.server 302 nick :nick=-.*@(localhost|127.0.0.1) nick=-.*(a)(localhost|127.0.0.1) nick=-.*(a)(localhost|127.0.0.1) nick=-.*(a)(localhost|127.0.0.1) nick=-.*(a)(localhost|127.0.0.1)\r" } send "quit\r" -- 1.4.4.4

3 3

[PATCH] Fix handling of localhost in Cygwin on Vista
by Ali Shemiran 15 Mai '08

15 Mai '08

--- src/ngircd/client.c | 6 ++++++ 1 files changed, 6 insertions(+), 0 deletions(-) diff --git a/src/ngircd/client.c b/src/ngircd/client.c index 1b72adc..4479792 100644 --- a/src/ngircd/client.c +++ b/src/ngircd/client.c @@ -636,6 +636,12 @@ GLOBAL char * Client_Hostname( CLIENT *Client ) { assert( Client != NULL ); + +#ifdef __CYGWIN__ + if (strcmp(Client->host, "127.0.0.1") == 0) + return "localhost"; +#endif + return Client->host; } /* Client_Hostname */ -- 1.4.4.4

2 1

ngIRCd 0.12.0
by Alexander Barton 13 Mai '08

13 Mai '08

Hi! ngIRCd 0.12.0 has been released and can be downloaded from the following locations, as usual: a) ftp://ngircd.barton.de/pub/ngircd/ ftp://ngircd.barton.de/pub/ngircd/ngircd-0.12.0.tar.gz b) ftp://ftp.berlios.de/pub/ngircd/ ftp://ftp.berlios.de/pub/ngircd/ngircd-0.12.0.tar.gz Patches from 0.11.1 (about 220K uncompressed) and 0.12.0-pre2 as well as GnuPG signatures are available, too. The most prominent enhancement is support for IPv6, and the complete ChangeLog since 0.11.1 is: - Fix Bug: 85: "WHO #SecretChannel" that user is not a member of now returns proper RPL_ENDOFWHO_MSG instead of nothing. (Ali Shemiran) - Fix complie on FreeBSD 5.4 and AIX. - If bind() fails, also print ip address and not just the port number. ngIRCd 0.12.0-pre2 (2008-04-29) - IPv6: Add config options to disabe ipv4/ipv6 support. - Don't include doc/CVS.txt in distribution archive, use doc/GIT.txt now! - Documentation: get rid of some more references to CVS, switch to GIT. - Get rid of cvs-version.* and CVSDATE definition. - Report ERR_NOTONCHANNEL when trying to part a channel one is not member of. - Testsuite: remove erroneous ConfUID setting in config file. ngIRCd 0.12.0-pre1 (2008-04-20) - Include Mac OS X Xcode project in distribution archives. - Do not exit on SIGHUP or /REHASH if the config file cannot opened. - Add IPv6 support. - Install a LaunchDaemon script to start/stop ngIRCd on Mac OS X. - Implemented IRC commands INFO, SUMMON (dummy), and USERS (dummy) and enhanced test suite to check these commands. (Dana Dahlstrom) - RPL_WHOREPLY messages generated by IRC_WHO didn't include flags (*,@,+). (Dana Dahlstrom) - IRC_WHO now supports search patterns and will test this against user nickname/servername/hostname, etc. as required by RFC 2812, Section 3.6.1. (reported by Dana Dahlstrom) - Add test cases for "WHO" command. (Dana Dahlstrom) - Implement RFC 2812 handling of "0" argument to 'JOIN': must be treated as if the user had sent PART commands for all channels the user is a member of. (Dana Dahlstrom) - Allow NOTICEs to be sent to a channel. (Fabian Schlager) Nine people already committed to the post-0.12.0 GIT repository, so there are good chances that the next release of ngIRCd won't be too far in the future ;-) You can use this mailing liste, the ngIRCd bug tracker (http://arthur.barton.de/cgi-bin/bugzilla/index.cgi ) or the IRC channel #ngircd on irc.barton.de:6667 (irc://irc.barton.de/ngircd ) to get in touch with us. Regards Alex

1 0

[PATCH] Added Cygwin uid kludge from Florian. (verified working)
by Ali Shemiran 08 Mai '08

08 Mai '08

--- src/ngircd/ngircd.c | 13 +++++++++++++ 1 files changed, 13 insertions(+), 0 deletions(-) diff --git a/src/ngircd/ngircd.c b/src/ngircd/ngircd.c index 626b8b2..d839618 100644 --- a/src/ngircd/ngircd.c +++ b/src/ngircd/ngircd.c @@ -671,6 +671,19 @@ NGIRCd_getNobodyID(uid_t *uid, gid_t *gid ) { struct passwd *pwd; +#ifdef __CYGWIN__ + /* Cygwin kludge. + * It can return EINVAL instead of EPERM + * so, if we are already unprivileged, + * use id of current user. + */ + if (geteuid() && getuid()) { + *uid = getuid(); + *gid = getgid(); + return true; + } +#endif + pwd = getpwnam("nobody"); if (!pwd) return false; -- 1.5.5.1

1 0

[PATCH] Anglified and improved comments in channel.c and channel.h
by Eric Grunow 08 Mai '08

08 Mai '08

--- src/ngircd/channel.c | 45 +++++++++++++++++++++++++++++++-------------- src/ngircd/channel.h | 2 +- 2 files changed, 32 insertions(+), 15 deletions(-) diff --git a/src/ngircd/channel.c b/src/ngircd/channel.c index c923404..fde51a3 100644 --- a/src/ngircd/channel.c +++ b/src/ngircd/channel.c @@ -89,7 +89,7 @@ Channel_GetListInvites(CHANNEL *c) GLOBAL void Channel_InitPredefined( void ) { - /* Vordefinierte persistente Channels erzeugen */ + /* Generate predefined persistent channels */ CHANNEL *chan; char *c; @@ -98,9 +98,10 @@ Channel_InitPredefined( void ) for( i = 0; i < Conf_Channel_Count; i++ ) { /* Ist ein Name konfiguriert? */ + /* Check for Name configuration */ if( ! Conf_Channel[i].name[0] ) continue; - /* Gueltiger Channel-Name? */ + /* Check for invalid channel name */ if( ! Channel_IsValidName( Conf_Channel[i].name )) { Log( LOG_ERR, "Can't create pre-defined channel: invalid name: \"%s\"!", Conf_Channel[i].name ); @@ -108,7 +109,7 @@ Channel_InitPredefined( void ) continue; } - /* Gibt es den Channel bereits? */ + /* Check if the channel name is already in use */ chan = Channel_Search( Conf_Channel[i].name ); if( chan ) { @@ -149,7 +150,7 @@ Channel_Exit( void ) CHANNEL *c, *c_next; CL2CHAN *cl2chan, *cl2chan_next; - /* Channel-Strukturen freigeben */ + /* free struct Channel */ c = My_Channels; while( c ) { @@ -159,7 +160,7 @@ Channel_Exit( void ) c = c_next; } - /* Channel-Zuordnungstabelle freigeben */ + /* Free Channel allocation table */ cl2chan = My_Cl2Chan; while( c ) { @@ -169,7 +170,14 @@ Channel_Exit( void ) } } /* Channel_Exit */ - +/** + * Join Channel + * This function lets a client join a channel. First, the function + * checks that the specified channel name is valid and that the client + * isn't already a member. If the specified channel doesn't exist, + * a new channel is created. Client is added to channel by function + * Add_Client(). + */ GLOBAL bool Channel_Join( CLIENT *Client, char *Name ) { @@ -178,6 +186,7 @@ Channel_Join( CLIENT *Client, char *Name ) assert( Client != NULL ); assert( Name != NULL ); + /* Check that the channel name is valid */ if( ! Channel_IsValidName( Name )) { IRC_WriteStrClient( Client, ERR_NOSUCHCHANNEL_MSG, Client_ID( Client ), Name ); return false; @@ -185,25 +194,25 @@ Channel_Join( CLIENT *Client, char *Name ) chan = Channel_Search( Name ); if( chan ) { - /* Ist der Client bereits Mitglied? */ + /* Check if the client is already in the channel */ if( Get_Cl2Chan( chan, Client )) return false; } else { - /* Gibt es noch nicht? Dann neu anlegen: */ + /* If the specified channel doesn't exist, the channel is created */ chan = Channel_Create( Name ); if (!chan) return false; } - /* User dem Channel hinzufuegen */ + /* Add user to Channel */ if( ! Add_Client( chan, Client )) return false; else return true; } /* Channel_Join */ /** - * Remove client from channel. - * This function lets a client lead a channel. First, the function checks + * Part client from channel. + * This function lets a client part from a channel. First, the function checks * if the channel exists and the client is a member of it and sends out * appropriate error messages if not. The real work is done by the function * Remove_Client(). @@ -217,25 +226,30 @@ Channel_Part(CLIENT * Client, CLIENT * Origin, const char *Name, const char *Rea assert(Name != NULL); assert(Reason != NULL); + /* Check that specified channel exists */ chan = Channel_Search(Name); if (!chan) { IRC_WriteStrClient(Client, ERR_NOSUCHCHANNEL_MSG, Client_ID(Client), Name); return false; } + + /* Check that the client is in the channel */ if (!Get_Cl2Chan(chan, Client)) { IRC_WriteStrClient(Client, ERR_NOTONCHANNEL_MSG, Client_ID(Client), Name); return false; } + + /* Part client from channel */ if (!Remove_Client(REMOVE_PART, chan, Client, Origin, Reason, true)) return false; else return true; } /* Channel_Part */ - +/* Kick user from Channel */ GLOBAL void Channel_Kick( CLIENT *Client, CLIENT *Origin, const char *Name, const char *Reason ) { @@ -246,6 +260,7 @@ Channel_Kick( CLIENT *Client, CLIENT *Origin, const char *Name, const char *Reas assert( Name != NULL ); assert( Reason != NULL ); + /* Check that channel exists */ chan = Channel_Search( Name ); if( ! chan ) { @@ -253,26 +268,28 @@ Channel_Kick( CLIENT *Client, CLIENT *Origin, const char *Name, const char *Reas return; } + /* Check that user is on the specified channel */ if( ! Channel_IsMemberOf( chan, Origin )) { IRC_WriteStrClient( Origin, ERR_NOTONCHANNEL_MSG, Client_ID( Origin ), Name ); return; } - /* Is User Channel-Operator? */ + /* Check if user has operator status */ if( ! strchr( Channel_UserModes( chan, Origin ), 'o' )) { IRC_WriteStrClient( Origin, ERR_CHANOPRIVSNEEDED_MSG, Client_ID( Origin ), Name); return; } - /* Ist the kickED User member of channel? */ + /* Check that the client to be kicked is on the specified channel */ if( ! Channel_IsMemberOf( chan, Client )) { IRC_WriteStrClient( Origin, ERR_USERNOTINCHANNEL_MSG, Client_ID( Origin ), Client_ID( Client ), Name ); return; } + /* Kick Client from channel */ Remove_Client( REMOVE_KICK, chan, Client, Origin, Reason, true); } /* Channel_Kick */ diff --git a/src/ngircd/channel.h b/src/ngircd/channel.h index 1483b4a..1f78bf8 100644 --- a/src/ngircd/channel.h +++ b/src/ngircd/channel.h @@ -46,7 +46,7 @@ typedef struct _CLIENT2CHAN struct _CLIENT2CHAN *next; CLIENT *client; CHANNEL *channel; - char modes[CHANNEL_MODE_LEN]; /* User-Modes in dem Channel */ + char modes[CHANNEL_MODE_LEN]; /* User-Modes in Channel */ } CL2CHAN; #else -- 1.5.5.1

3 2

ngIRCd 0.12.0-pre2 & source code repository switched to GIT
by Alexander Barton 04 Mai '08

04 Mai '08

Hello all, We are steadily moving towards ngIRCd 0.12.0 and I released ngIRCd 0.12.0-pre2 for testing a few days ago (0.12.0-pre1 has been released a little bit too quickly, missed some "needed" features, and therefore hasn't been announced loudly). Most notably, ngIRCd will support IPv6, became more RFC-compliant, has enhanced support for Mac OS X, and the source code repository has been switched from CVS to GIT (see below). The ChangeLog for the upcoming 0.12.0 release reads as following: ngIRCd 0.12.0-pre2 (2008-04-29) - IPv6: Add config options to disabe ipv4/ipv6 support. - Don't include doc/CVS.txt in distribution archive, use doc/GIT.txt now! - Documentation: get rid of some more references to CVS, switch to GIT. - Get rid of cvs-version.* and CVSDATE definition. - Report ERR_NOTONCHANNEL when trying to part a channel one is not member of. - Testsuite: remove erroneous ConfUID setting in config file. ngIRCd 0.12.0-pre1 (2008-04-20) - Include Mac OS X Xcode project in distribution archives. - Do not exit on SIGHUP or /REHASH if the config file cannot opened. - Add IPv6 support. - Install a LaunchDaemon script to start/stop ngIRCd on Mac OS X. - Implemented IRC commands INFO, SUMMON (dummy), and USERS (dummy) and enhanced test suite to check these commands. (Dana Dahlstrom) - RPL_WHOREPLY messages generated by IRC_WHO didn't include flags (*,@,+). (Dana Dahlstrom) - IRC_WHO now supports search patterns and will test this against user nickname/servername/hostname, etc. as required by RFC 2812, Section 3.6.1. (reported by Dana Dahlstrom) - Add test cases for "WHO" command. (Dana Dahlstrom) - Implement RFC 2812 handling of "0" argument to 'JOIN': must be treated as if the user had sent PART commands for all channels the user is a member of. (Dana Dahlstrom) - Allow NOTICEs to be sent to a channel. (Fabian Schlager) The pre-release (archives, patches, GnuPG signatures) can be downloaded from the usual places: - ftp://ngircd.barton.de/pub/ngircd/ ftp://ngircd.barton.de/pub/ngircd/ngircd-0.12.0-pre2.tar.gz - ftp://ftp.berlios.de/pub/ngircd/ ftp://ftp.berlios.de/pub/ngircd/ngircd-0.12.0-pre2.tar.gz Please download this release and throw it at as many machines as you can reach ;-) And please let us know about all problems you encounter -- thanks! GIT Source Code Repository During the last weeks we switched the source code repsoitory from CVS to GIT, because Florian and I think that GIT eases distributed development a lot -- and the first results do look very promising, have a look at the great work done by Dana and his team. Thanks! Actually the old CVS repository is still up and running, but no new code will be committed to it. Eventually I'll shut it down, but are a little bit unsure about when and how ... The new official GIT repository of ngIRCd can be reached as following: - git://ngircd.barton.de/ngircd.git So you can use the command "git clone git://ngircd.barton.de/ ngircd.git" to download a local copy of this repository. Please have a look at doc/GIT.txt and the excellent GIT documentation located at http://git.or.cz/ for more details. A "GITweb" interface (as replacement for CVSweb) is available, too: - http://ngircd.barton.de/ngircd.git [which will redirect to:] - http://ngircd.barton.de/cgi-bin/gitweb.cgi?p=ngircd.git So please test ngIRCd, report bugs, code patches, and let us know how ngIRCd could be enhanced. Thanks! You can use this mailing liste, the ngIRCd bug tracker (http://arthur.barton.de/cgi-bin/bugzilla/index.cgi ) or the IRC channel #ngircd on irc.barton.de:6667 (irc://irc.barton.de/ngircd ) to get in touch with us. Regards Alex

1 0

tests for JOIN and various replies
by Dana Dahlstrom 04 Mai '08

04 Mai '08

Brandon Beresini sent me a patch yesterday adding tests for JOIN under various circumstances, which I believe he worked on with Bryan Caldwell and Ali Shemiran. I made a few modifications; the result is below. We couldn't seem to come up with a good way to elicit a proper ERR_BANNEDFROMCHAN reply because, while the configuration file can specify channel modes, it apparently can't pass the required parameter to the +b mode in order to effect a ban. Any proposals? Also in the process of investigating this we discovered some rather unusual behavior. A user who has already joined a channel and tries to join the channel again will normally receive no reply---this is how other IRC daemons behave as well. However, if a ban is in effect for that user, the user receives an ERR_BANNEDFROMCHAN reply even though the user is already in the channel---other servers don't do this: join #chan :nick!~user@csegrad.ucsd.edu JOIN :#chan :irc.bit.netz 353 nick = #chan :@nick :irc.bit.netz 366 nick #chan :End of NAMES list join #chan mode #chan +b nick :nick!~user@csegrad.ucsd.edu MODE #chan +b nick!*@* join #chan :irc.bit.netz 474 nick #chan :Cannot join channel (+b) This might qualify as a bug. Thoughts? Dana diff --git a/src/testsuite/Makefile.am b/src/testsuite/Makefile.am index bc8df84..b382a01 100644 --- a/src/testsuite/Makefile.am +++ b/src/testsuite/Makefile.am @@ -20,8 +20,8 @@ EXTRA_DIST = \ README functions.inc getpid.sh \ start-server.sh stop-server.sh tests.sh stress-server.sh \ test-loop.sh wait-tests.sh \ - channel-test.e connect-test.e check-idle.e misc-test.e mode-test.e \ - opless-channel-test.e \ + channel-test.e connect-test.e check-idle.e join-test.e \ + misc-test.e mode-test.e opless-channel-test.e \ who-test.e stress-A.e stress-B.e \ ngircd-test.conf @@ -48,6 +48,10 @@ channel-test: tests.sh rm -f channel-test ln -s $(srcdir)/tests.sh channel-test +join-test: tests.sh + rm -f join-test + ln -s $(srcdir)/tests.sh join-test + opless-channel-test: tests.sh rm -f opless-channel-test ln -s $(srcdir)/tests.sh opless-channel-test @@ -67,10 +71,11 @@ mode-test: tests.sh TESTS = start-server.sh \ connect-test \ channel-test \ + join-test \ misc-test \ mode-test \ - who-test \ opless-channel-test \ + who-test \ stress-server.sh \ stop-server.sh diff --git a/src/testsuite/join-test.e b/src/testsuite/join-test.e new file mode 100644 index 0000000..41a5168 --- /dev/null +++ b/src/testsuite/join-test.e @@ -0,0 +1,68 @@ +spawn telnet localhost 6789 +expect { + timeout { exit 1 } + "Connected" +} + +send "nick nick\r" +send "user user . . :User\r" +expect { + timeout { exit 1 } + "376" +} + +send "JOIN\r" +expect { + timeout { exit 1} + "461" +} + +send "JOIN #InviteChannel\r" +expect { + timeout { exit 1 } + "473" +} + +send "JOIN #FullKeyed\r" +expect { + timeout { exit 1 } + "475" +} + +send "JOIN #FullKeyed WrongKey\r" +expect { + timeout { exit 1 } + "475" +} + +send "JOIN #FullKeyed Secret\r" +expect { + timeout { exit 1 } + "471" +} + +send "JOIN #TopicChannel\r" +expect { + timeout { exit 1 } + "@* JOIN :#TopicChannel" +} +expect { + timeout { exit 1 } + "332" +} + +send "JOIN 0\r" +send "JOIN #1,#2,#3,#4\r" +send "JOIN #5\r" +expect { + timeout { exit 1 } + "405" +} + +send "quit\r" +expect { + timeout { exit 1 } + "Connection closed" +} + +# -eof- diff --git a/src/testsuite/ngircd-test.conf b/src/testsuite/ngircd-test.conf index 0bec96d..eeace76 100644 --- a/src/testsuite/ngircd-test.conf +++ b/src/testsuite/ngircd-test.conf @@ -8,9 +8,25 @@ AdminEMail = admin(a)irc.server MaxConnectionsIP = 0 OperCanUseMode = yes + MaxJoins = 4 [Operator] Name = TestOp Password = 123 +[CHANNEL] + Name = #InviteChannel + Modes = i + +[CHANNEL] + Name = #FullKeyed + Modes = lk + MaxUsers = 0 + Key = Secret + +[CHANNEL] + Name = #TopicChannel + Modes = t + Topic = the topic + # -eof-

2 1

patch for KICK to handle comma-delimited lists of channels and nicks
by Dana Dahlstrom 03 Mai '08

03 Mai '08

Bryan Caldwell and Brandon Beresini showed me this patch yesterday: http://arthur.barton.de/cgi-bin/bugzilla/attachment.cgi?id=12&action=view I've reviewed and tested it, and we're now running it on cse70.ucsd.edu. It appears to resolve Bug 75: http://arthur.barton.de/cgi-bin/bugzilla/show_bug.cgi?id=75 Dana

2 1

ngIRCd on Wikipedia
by Dana Dahlstrom 01 Mai '08

01 Mai '08

I just added rows for ngIRCd in the tables here: http://en.wikipedia.org/wiki/Comparison_of_Internet_Relay_Chat_daemons Take a look, and correct any mistakes I may have made! :) Dana

1 0

Proposed test suite changes
by bcaldwel＠ucsd.edu 26 Apr '08

26 Apr '08

Hi everyone, Brandon, Ali, and I were using the test suite and noticed that we could add some useful features. As we were playing with the code, we would run the test and some would fail. We felt it took more work than necessary to figure out which test failed. As a result we made some changes to the testsuite. changes: -We added puts before each test to act as a label so we could tell which tests log messages belonged to. =This helped us quickly determine what was causing test to fail, because we could easily see the erroneous server communication. =Test labels must begin with ">--". This allows them to easily and quickly be seen in the log files and allows the second change to function -Added the -s option when running the individual tests =This option will show all the test that were run and if a test fails gives the user the option of viewing all the log messages associated with the failed test. While these changes add slightly more work to writing tests (adding a label) we have found them very useful when the tests don't pass. We added -s as an option instead of a default because there are some situations where the added information isn't needed and to make the change transparent to those who don't want to use it. -Ali, Brandon, Bryan --------------------------------------------- diff --git a/src/testsuite/channel-test.e b/src/testsuite/channel-test.e index 16c67ad..88388c1 100644 --- a/src/testsuite/channel-test.e +++ b/src/testsuite/channel-test.e @@ -1,11 +1,13 @@ # $Id: channel-test.e,v 1.4 2008/02/05 13:31:51 fw Exp $ +puts ">-- Connecting to test server ... " spawn telnet localhost 6789 expect { timeout { exit 1 } "Connected" } +puts ">-- setting nick ... " send "nick nick\r" send "user user . . :User\r" expect { @@ -13,22 +15,26 @@ expect { "376" } +puts ">-- join #channel test ... " send "join #channel\r" expect { timeout { exit 1 } "@* JOIN :#channel" } +puts ">-- names list test ... " expect { timeout { exit 1 } "366" } +puts ">-- set topic test ... " send "topic #channel :Test-Topic\r" expect { timeout { exit 1 } "@* TOPIC #channel :Test-Topic" } +puts ">-- who #channel test ... " send "who #channel\r" expect { timeout { exit 1 } @@ -38,37 +44,44 @@ expect { timeout { exit 1 } "* nick H@ :0 User" } +puts ">-- end of who test ... " expect { timeout { exit 1 } "315 nick #channel" } +puts ">-- names #channel test ... " send "names #channel\r" expect { timeout { exit 1 } "353 nick = #channel :@nick" } +puts ">-- end of names test ... " expect { timeout { exit 1 } "366 nick #channel" } +puts ">-- list test ... " send "list\r" expect { timeout { exit 1 } "322 nick #channel 1 :Test-Topic" } +puts ">-- end of list test ... " expect { timeout { exit 1 } "323 nick :End of LIST" } +puts ">-- part test ... " send "part #channel\r" expect { timeout { exit 1 } "@* PART #channel :nick" } +puts ">-- re-join test ... " send "join #channel\r" expect { timeout { exit 1 } @@ -79,6 +92,7 @@ expect { "366" } +puts ">-- join #channel2 test ... " send "join #channel2\r" expect { timeout { exit 1 } @@ -89,6 +103,7 @@ expect { "366" } +puts ">-- join 0 test ... " send "join 0\r" expect { timeout { exit 1 } @@ -99,6 +114,7 @@ expect { "@* PART #channel :nick" } +puts ">-- quitting ... " send "quit\r" expect { timeout { exit 1 } diff --git a/src/testsuite/check-idle.e b/src/testsuite/check-idle.e index e994c14..04191f6 100644 --- a/src/testsuite/check-idle.e +++ b/src/testsuite/check-idle.e @@ -1,11 +1,13 @@ # $Id: check-idle.e,v 1.2 2004/09/04 14:22:13 alex Exp $ +puts ">-- Connecting to test server ... " spawn telnet localhost 6789 expect { timeout { exit 1 } "Connected" } +puts ">-- Setting nick nick ... " send "nick IdleTest\r" send "user idle . . :Idle-Test\r" expect { @@ -14,6 +16,7 @@ expect { "376" } +puts ">-- lusers test ... " send "lusers\r" expect { timeout { exit 1 } @@ -21,6 +24,7 @@ expect { "251 IdleTest :There are" { set r 99 } } +puts ">-- quitting ... " send "quit\r" expect { timeout { exit 1 } diff --git a/src/testsuite/connect-test.e b/src/testsuite/connect-test.e index 49ce70c..9fc28e1 100644 --- a/src/testsuite/connect-test.e +++ b/src/testsuite/connect-test.e @@ -1,17 +1,20 @@ # $Id: connect-test.e,v 1.1 2002/09/09 10:16:24 alex Exp $ +puts ">-- Connecting to test server ... " spawn telnet localhost 6789 expect { timeout { exit 1 } "Connected" } +puts ">-- oper test ... " send "oper\r" expect { timeout { exit 1 } "451" } +puts ">-- quitting ... " send "quit\r" expect { timeout { exit 1 } diff --git a/src/testsuite/misc-test.e b/src/testsuite/misc-test.e index 4b88379..5b404c3 100644 --- a/src/testsuite/misc-test.e +++ b/src/testsuite/misc-test.e @@ -1,11 +1,13 @@ # $Id: misc-test.e,v 1.2 2008/02/17 13:51:00 alex Exp $ +puts ">-- Connecting to test server ... " spawn telnet localhost 6789 expect { timeout { exit 1 } "Connected" } +puts ">-- setting nick ... " send "nick nick\r" send "user user . . :User\r" expect { @@ -13,28 +15,33 @@ expect { "376" } +puts ">-- summon is disabled test ... " send "summon\r" expect { timeout { exit 1 } "445" } +puts ">-- users is disabled test ... " send "users\r" expect { timeout { exit 1 } "446" } +puts ">-- info test ... " send "info\r" expect { timeout { exit 1 } "371" } +puts ">-- end of info test ... " expect { timeout { exit 1 } "374" } +puts ">-- quitting ... " send "quit\r" expect { timeout { exit 1 } diff --git a/src/testsuite/mode-test.e b/src/testsuite/mode-test.e index 260cd03..946833b 100644 --- a/src/testsuite/mode-test.e +++ b/src/testsuite/mode-test.e @@ -1,11 +1,13 @@ # $Id: mode-test.e,v 1.7 2008/02/16 11:27:49 fw Exp $ +puts ">-- Connecting to test server ... " spawn telnet localhost 6789 expect { timeout { exit 1 } "Connected" } +puts ">-- setting nick ... " send "nick nick\r" send "user user . . :User\r" expect { @@ -13,24 +15,28 @@ expect { "376" } +puts ">-- set mode test ... " send "mode nick +i\r" expect { timeout { exit 1 } "@* MODE nick :+i" } +puts ">-- get mode test ... " send "mode nick\r" expect { timeout { exit 1 } "221 nick +i" } +puts ">-- unset mode test ... " send "mode nick -i\r" expect { timeout { exit 1 } "@* MODE nick :-i" } +puts ">-- oper test ... " send "oper TestOp 123\r" expect { timeout { exit 1 } @@ -41,12 +47,14 @@ expect { "381 nick" } +puts ">-- is op test ... " send "mode nick\r" expect { timeout { exit 1 } "221 nick +o" } +puts ">-- joining #channel ... " send "join #channel\r" expect { timeout { exit 1 } @@ -57,48 +65,56 @@ expect { "366" } +puts ">-- set #channel mode test ... " send "mode #channel +tn\r" expect { timeout { exit 1 } "@* MODE #channel +tn" } +puts ">-- get #channel mode test ... " send "mode #channel\r" expect { timeout { exit 1 } "324 nick #channel +tn" } +puts ">-- voice nick test ... " send "mode #channel +v nick\r" expect { timeout { exit 1 } "@* MODE #channel +v nick\r" } +puts ">-- invite only with nick mask test ... " send "mode #channel +I nick1\r" expect { timeout { exit 1 } "@* MODE #channel +I nick1!*@*" } +puts ">-- ban nick2 with domain mask test ... " send "mode #channel +b nick2@domain\r" expect { timeout { exit 1 } "@* MODE #channel +b nick2!*@domain" } +puts ">-- invite only with user mask test ... " send "mode #channel +I nick3!user\r" expect { timeout { exit 1 } "@* MODE #channel +I nick3!user@*" } +puts ">-- unset #channel mode test ... " send "mode #channel -vo nick nick\r" expect { timeout { exit 1 } "@* MODE #channel -vo nick nick\r" } +puts ">-- quitting ... " send "quit\r" expect { timeout { exit 1 } diff --git a/src/testsuite/stress-A.e b/src/testsuite/stress-A.e index b22a2f3..18fe86d 100644 --- a/src/testsuite/stress-A.e +++ b/src/testsuite/stress-A.e @@ -2,6 +2,7 @@ set timeout 30 +puts ">-- Connecting to test server ... " spawn telnet localhost 6789 expect { timeout { exit 1 } diff --git a/src/testsuite/stress-B.e b/src/testsuite/stress-B.e index e260c6d..f0c6539 100644 --- a/src/testsuite/stress-B.e +++ b/src/testsuite/stress-B.e @@ -1,5 +1,6 @@ # $Id: stress-B.e,v 1.3 2005/12/30 22:12:28 alex Exp $ +puts ">-- setting user ... " send "user user . . :User\r" expect { timeout { exit 1 } @@ -8,6 +9,7 @@ expect { sleep 2 +puts ">-- becoming op ... " send "oper TestOp 123\r" expect { timeout { exit 1 } @@ -20,6 +22,7 @@ expect { sleep 2 +puts ">-- joining #channel ... " send "join #channel\r" expect { timeout { exit 1 } @@ -30,12 +33,14 @@ expect { " 366 * #channel :" } +puts ">-- getting #channel mode ... " send "mode #channel\r" expect { timeout { exit 1 } " 324 test* #channel" } +puts ">-- joining #channel2 ... " send "join #channel2\r" expect { timeout { exit 1 } @@ -46,6 +51,7 @@ expect { " 366 * #channel2 :" } +puts ">-- listing names ... " send "names\r" expect { timeout { exit 1 } @@ -54,12 +60,14 @@ expect { sleep 3 +puts ">-- parting #channel2 ... " send "part #channel2\r" expect { timeout { exit 1 } " PART #channel2 " } +puts ">-- parting #channel ... " send "part #channel\r" expect { timeout { exit 1 } @@ -68,6 +76,7 @@ expect { sleep 1 +puts ">-- quitting ... " send "quit\r" expect { timeout { exit 1 } diff --git a/src/testsuite/tests.sh b/src/testsuite/tests.sh index 2bd9ae4..0eaba50 100755 --- a/src/testsuite/tests.sh +++ b/src/testsuite/tests.sh @@ -26,9 +26,35 @@ if [ $? -ne 0 ]; then echo " ${name}: \"telnet\" not found."; exit 77 fi -echo_n " running ${test} ..." +if [ "$1" == "-s" ]; then + echo " running ${test} ..." +else + echo_n " running ${test} ..." +fi expect ${srcdir}/${test}.e > logs/${test}.log 2>&1; r=$? -[ $r -eq 0 ] && echo " ok." || echo " failure!" +if [ "$1" == "-s" ]; then + cat logs/${test}.log | grep ">\--" +fi + +if [ $r -eq 0 ]; then + echo " ok." +else + if [ "$1" == "-s" ]; then + echo "--> failed." + echo_n "Display log for last test? (y/[n]): " + read response + if [ "$response" == "y" ]; then + begLastTest=`grep -n ">\--" logs/${test}.log | tail -1` + begLastTest=${begLastTest%:*} + length=`wc -l logs/${test}.log` + length=${length% *} + fromEnd=`expr $length - $begLastTest + 1` + tail -n $fromEnd logs/${test}.log + fi + else + echo " failed." + fi +fi exit $r diff --git a/src/testsuite/who-test.e b/src/testsuite/who-test.e index c54a190..1573cb2 100644 --- a/src/testsuite/who-test.e +++ b/src/testsuite/who-test.e @@ -1,9 +1,12 @@ + +puts ">-- Connecting to test server ... " spawn telnet localhost 6789 expect { timeout { exit 1 } "Connected" } +puts ">-- setting nick ... " send "nick nick\r" send "user user . . :Real Name\r" expect { @@ -11,72 +14,84 @@ expect { "376" } +puts ">-- who test ... " send "who\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick \* ~user localhost ngircd.test.server nick H :0 Real Name" } +puts ">-- joining #channel ... " send "join #channel\r" expect { timeout { exit 1 } "@* JOIN :#channel" } +puts ">-- who 0 test ... " send "who 0\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick #channel ~user localhost ngircd.test.server nick H@ :0 Real Name" } +puts ">-- set away test ... " send "away :testing\r" expect { timeout { exit 1 } "306 nick" } +puts ">-- away who test ... " send "who *\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick #channel ~user localhost ngircd.test.server nick G@ :0 Real Name" } +puts ">-- voicing nick on #channel ... " send "mode #channel +v nick\r" expect { timeout { exit 1 } "@* MODE #channel +v nick\r" } +puts ">-- who server by ip test ... " send "who localhost\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick #channel ~user localhost ngircd.test.server nick G@ :0 Real Name" } +puts ">-- un-op nick on #channel ... " send "mode #channel -o nick\r" expect { timeout { exit 1 } "@* MODE #channel -o nick\r" } +puts ">-- who server by name test ... " send "who ngircd.test.server\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick #channel ~user localhost ngircd.test.server nick G+ :0 Real Name" } +puts ">-- parting #channel ... " send "part #channel\r" expect { timeout { exit 1 } "@* PART #channel :nick" } +puts ">-- who by real name test ... " send "who Real?Name\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick \* ~user localhost ngircd.test.server nick G :0 Real Name" } +puts ">-- becoming op ... " send "oper TestOp 123\r" expect { timeout { exit 1 } @@ -87,72 +102,84 @@ expect { "381 nick" } +puts ">-- who 0 with mask test ... " send "who 0 o\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick \* ~user localhost ngircd.test.server nick G* :0 Real Name" } +puts ">-- unaway test ... " send "away\r" expect { timeout { exit 1 } "305 nick" } +puts ">-- who with mask test ... " send "who *cal*ho??\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick \* ~user localhost ngircd.test.server nick H* :0 Real Name" } +puts ">-- joining #opers ... " send "join #opers\r" expect { timeout { exit 1 } "@* JOIN :#opers" } +puts ">-- who channel test ... " send "who #opers\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick #opers ~user localhost ngircd.test.server nick H*@ :0 Real Name" } +puts ">-- un-op nick on #opers ... " send "mode #opers -o nick\r" expect { timeout { exit 1 } "@* MODE #opers -o nick\r" } +puts ">-- who server mask test ... " send "who *.server\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick #opers ~user localhost ngircd.test.server nick H* :0 Real Name" } +puts ">-- voicing nick in #opers ... " send "mode #opers +v nick\r" expect { timeout { exit 1 } "@* MODE #opers +v nick\r" } +puts ">-- who real name mask test ... " send "who Real*me\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick #opers ~user localhost ngircd.test.server nick H*+ :0 Real Name" } +puts ">-- making #opers secret ... " send "mode #opers +s\r" expect { timeout { exit 1 } "@* MODE #opers +s\r" } +puts ">-- who with mask test ... " send "who n?c?\r" expect { timeout { exit 1 } ":ngircd.test.server 352 nick \* ~user localhost ngircd.test.server nick H* :0 Real Name" } +puts ">-- quitting ... " send "quit\r" expect { timeout { exit 1 }

2 1

Making ngIRCd work with services
by egrunow＠ucsd.edu 21 Apr '08

21 Apr '08

Hi everyone, Scott Perry and I are going to be working to make ngIRCd work with services. To begin with, we'll try to make it work with atheme and IRC Services, hopefully this will make us (more?) compatible with other services packages as well. As an aside, Scott and I are enjoying learning German from the comments in the code. Eric Grunow

4 4

student interns at UC San Diego
by Dana Dahlstrom 15 Apr '08

15 Apr '08

Hello all, Let me introduce to you 5 outstanding students who will be working with us on improvements to ngIRCd from now until June as part of a team project I'm coordinating. They are: Ali Shemiran Brandon Beresini Bryan Caldwell Eric Grunow Scott Perry You'll hear from them on this list and encounter them on #ngircd (if you haven't already). Please give them a warm welcome and plenty of suggestions and feedback. There are barely 2 months left in our short academic term, so they'll need to get working right away, and any guidance you can offer is greatly appreciated. Here's to continuing progress! Dana

4 6

ngircd with freebsd and pf
by Marc Freimann 13 Apr '08

13 Apr '08

Hello, I installed ngircd at freebsd 7.0. There are problems and questions about ngircd and irc in general. The first problem: ---------------------------- I can´t reach ngircd from the internet when I use pf at freebsd. All other services working fine without problems. With command "tcpdump -i pflog0" I received the following errormessage: 10:46:11.913489 IP pD96FFF32.dip.t-dialin.net.7238 > test.ircd: tcp 24 [bad hdr length 0 - too short, < 20] I tested the connection with 3 different irc clients at 2 different internet connection with the same result. Are the realy differents between "normal" packets (for example http or ssh and irc packets about size? Is there a solution for this problem ? ----------------------------- The secound problem: ----------------------------- When I start ngircd by restart freebsd (rc.conf: ngircd_enable=YES) I always received the errormessage: Can´t change working directory to "/nonexistent": No such file or directory". At this situation, the ChrootDir parameter is not active in ngircd.conf. What can I do ? When I activate Chrootdir /var/empty (defaultsetting) the irc server don´t work. (fatal error) What can I do ? ----------------------------- The first question: ----------------------------- What is the best nickserv for ngircd ? Is there a howto about implement this nickserv to ngircd ----------------------------- Best Regards Marc -- Psst! Geheimtipp: Online Games kostenlos spielen bei den GMX Free Games! http://games.entertainment.gmx.net/de/entertainment/games/free

3 2

Re: ngIRCd-ML Digest, Vol 46, Issue 3
by Armin Plöger 15 Mär '08

15 Mär '08

Hallo Florian Ich nutze ngircd 0.10.0-2etch1 und habe den Befehl das der Chater keine channels betreten dürfen nocheinmal getestet und es klappt leider nicht. Wo genau muss der befehl rein ? unter global denke ich doch mal oder ? Was du mir sagtest mit dem Chroot das Verstehe ich leider noch nicht so wirklich was du mir damit sagen möchtest. Ich bin darauf leider nicht schlaugeworden wie ich das ausschalten nach einem rehas vermeiden kann. Jedoch ist mir noch ein Problem aufgefallen (ich weiss langsam nerve ich) Wie kann ich es zulassen das im usernamen auch sonderzeichen wie []{}-_ genutzt werden können ? Gruss Armin

2 1

Re: Diverse Einstellungen in ngircd
by Armin Plöger 15 Mär '08

15 Mär '08

Hallo nochmal nachdem mir Florian ja schon gute Infos gegeben hat habe ich aber dennoch ein paar Sachen. der befehle PredefChannelsOnly = yes funktioniert leider nicht. ich habe den befehle schon an mehreren stellen der config Datei eingeschrieben jedoch ohne erfolge. ich kann immer noch in beliebige räume joinen. also selbst erstellte channels sind immer noch möglich. wie kann ich den [Channel] Block aus einer externen Datei auslesen ? da ich als Service einen bot nutze der die Datei dann in regelmäßigen Abständen schreiben soll. dann ist die nächste Sache wie kann ich ngirc dazu bewegen das er die Konfigurationsdatei ohne Neustart einliest ? als admin mit dem befehle /rehash wird ngircd beendet und ich muss ihn über die Konsole neu starten Gruß Armin

2 1

Diverse Einstellungen in ngircd
by Armin Plöger 13 Mär '08

13 Mär '08

Hallo liebe Mailingliste, ich habe mir jetzt den ngircd installiert, es ist für einen der nicht viel ahnung von irc servern hat ein wirklich super IRC server das richitge für mich also. Soweit bin ich mit der Funktion des Servers sehr zufrieden. Finde jedoch diverse Einstellungen nicht :-( kann mir da von euch evt jemand Tipps geben wie ich das regeln kann ? Ich möchte ein kleines Privates IRC Netz aufmachen für meine Homepage also soll als ein zigster Client cgi:irc sein. 1. Wie kann ich im Server IP Adressen unterbinden ? Es sollen NUR Logins von 2 ip Adressen gewährt werden. Vom localhost und von meinem Bouncer (auf einer anderen Shell) 2. wie kann ich es realisieren das der Server KEINE User räume nimmt ? Es sollen also nur vordefinierte räume genutzt werden. 3. Kann ich es irgendwie hin bekommen das ich die räume (da ich diese dynamisch erstellen möchte) aus einer mysql Datenbank raus lese ? Also das die erlaubten registrierten räume möchte ich über die Webseite anlegen und der Benutzer kann diese räume dann mit cgi:irc betreten. Wie schaffe ich das ? Gibt es dafür Services wie chanserv der das kann ? Wo finde ich eine Konfigurationsanleitung dann dafür ? Vielen dank für hoffentlich kommende antworten Gruß Armin

2 1

ngircd.conf connecting to other servers
by Jim Wirtz 05 Mär '08

05 Mär '08

I really like this sever but I'm having a bit of a problem setting up my ngircd.conf. Yes this should be a simple thing but it not working like I expected. I have 3 servers setup and would like them to connect to each other. main server main.com 2nd server second.com 3rd server third.com on main.com ------------ [SERVER] Name = main.com Host = Port = 0 MyPassword = PeerPassword = Group = -1 Passive = no [SERVER] Name = second.com Host = Port = 0 MyPassword = PeerPassword = pleaseLetMeIn Group = -1 Passive = no [SERVER] Name = third.com Host = Port = 0 MyPassword = PeerPassword = pleaseLetMeIn Group = -1 Passive = no -------------------------------------------- on 2nd server ------------ [SERVER] Name = second.com Host = main.com Port = 6667 MyPassword = pleaseLetMeIn PeerPassword = Group = -1 Passive = no --------------------------------- on 3rd server ------------ [SERVER] Name = third.com Host = main.com Port = 6667 MyPassword = pleaseLetMeIn PeerPassword = Group = -1 Passive = no ------------------------------------------------- What am I missing??? Jim

2 1

ngIRCd 0.11.1
by Alexander Barton 26 Feb '08

26 Feb '08

Hi All! ngIRCd 0.11.1, a bugfix release of the 0.11.x branch, has been released and can be downloaded from the usual locations: <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.11.1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.11.1.tar.gz> GnuPG signatures and a patch from 0.11.0 are available and can be downloaded from here: <ftp://ngircd.berlios.de/pub/ngircd/> <ftp://ngircd.barton.de/pub/ngircd/> This release has been tagged as "rel-0-11-1" in CVS. Changes since 0.11.0 are: - Fix sending of JOIN commands between servers when remote server appended mode flags. (Rolf Eike Beer) [from HEAD] - Send "G" instead of "H" flag in WHO replies. (reported by Dana Dahlstrom) - Under some circumstances ngIRCd issued channel MODE message with a trailing space. (Dana Dahlstrom) [from HEAD] You can view the complete NEWS and ChangeLog files online here: <http://ngircd.barton.de/doc/NEWS> <http://ngircd.barton.de/doc/ChangeLog> Have fun! Regards Alex

1 0

messages > 512 characterss
by Dana Dahlstrom 24 Feb '08

24 Feb '08

Right now ngIRCd terminates any connection that sends in a message longer than 512 chars including line termination (see conn.c:1232). This strikes me as a bit brutal. Other possibilities include: truncating the message, ignoring the message, or replying with some error code. RFC 2812 doesn't directly specify what should happen when messages exceed the maximum length, but in at least once place it implies they should be truncated. In the description of ISON, it states, "The only limit on the number of nicks that may be checked is that the combined length MUST NOT be too large as to cause the server to chop it off so it fits in 512 characters." Thoughts? Dana

3 3

patch for SQUIT, INFO, SUMMON, USERS
by Dana Dahlstrom 17 Feb '08

17 Feb '08

RFC 2812 states that the SQUIT command is available to (IRC) operators, but when any client (IRCop or not) issues an SQUIT command, ngIRCd replies with ERR_NOTREGISTERED, which is not permitted by the RFC. Also, quoting from RFC 2812: The INFO command is REQUIRED to return information describing the server: its version, when it was compiled, the patchlevel, when it was started, and any other miscellaneous information which may be considered to be relevant. Implementing the SUMMON command is not required, but: If summon is not enabled in a server, it MUST return the ERR_SUMMONDISABLED numeric. Likewise for the USERS command: If disabled, the correct numeric MUST be returned to indicate this. The patch below adds a test (called misc-test for lack of a better name) for these commands and modifies: * testsuite/Makefile.am to run misc-test; * parse.c and irc-server.c to make IRC_SQUIT handle client requests; and * parse.c and irc-info.{h,c} to reply to INFO, SUMMON, and USERS. It is perhaps worth noting the commands SERVICE, SERVLIST, and SQUERY, which remain unimplemented, are described in RFC 2812 section 3, which also states that "All commands described in this section MUST be implemented by any server for this protocol." Dana --- /dev/null 1 Jan 1970 00:00:00 -0000 +++ src/testsuite/misc-test.e 17 Feb 2008 06:47:42 -0000 @@ -0,0 +1,50 @@ +# $Id$ + +spawn telnet localhost 6789 +expect { + timeout { exit 1 } + "Connected" +} + +send "nick nick\r" +send "user user . . :User\r" +expect { + timeout { exit 1 } + "376" +} + +send "summon\r" +expect { + timeout { exit 1 } + "445" +} + +send "users\r" +expect { + timeout { exit 1 } + "446" +} + +send "info\r" +expect { + timeout { exit 1 } + "371" +} +expect { + timeout { exit 1 } + "374" +} + +send "squit\r" +expect { + timeout { exit 1 } + "481" +} + +send "quit\r" +expect { + timeout { exit 1 } + "ERROR" +} + +# -eof- --- src/testsuite/Makefile.am 17 Feb 2008 00:00:13 -0000 1.17 +++ src/testsuite/Makefile.am 17 Feb 2008 06:47:42 -0000 @@ -21,7 +21,7 @@ EXTRA_DIST = \ start-server.sh stop-server.sh tests.sh stress-server.sh \ test-loop.sh wait-tests.sh \ connect-test.e channel-test.e mode-test.e \ - who-test.e + who-test.e misc-test.e stress-A.e stress-B.e check-idle.e \ ngircd-test.conf @@ -52,6 +52,10 @@ who-test: tests.sh rm -f who-test ln -s $(srcdir)/tests.sh who-test +misc-test: tests.sh + rm -f misc-test + ln -s $(srcdir)/tests.sh misc-test + mode-test: tests.sh rm -f mode-test ln -s $(srcdir)/tests.sh mode-test @@ -60,6 +64,7 @@ TESTS = start-server.sh \ connect-test \ channel-test \ who-test \ + misc-test \ mode-test \ stress-server.sh \ stop-server.sh --- src/ngircd/parse.c 5 Feb 2008 13:07:14 -0000 1.71 +++ src/ngircd/parse.c 17 Feb 2008 06:47:42 -0000 @@ -67,6 +67,7 @@ static COMMAND My_Commands[] = { "DISCONNECT", IRC_DISCONNECT, CLIENT_USER, 0, 0, 0 }, { "ERROR", IRC_ERROR, 0xFFFF, 0, 0, 0 }, { "HELP", IRC_HELP, CLIENT_USER, 0, 0, 0 }, + { "INFO", IRC_INFO, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, { "INVITE", IRC_INVITE, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, { "ISON", IRC_ISON, CLIENT_USER, 0, 0, 0 }, { "JOIN", IRC_JOIN, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, @@ -91,13 +92,15 @@ static COMMAND My_Commands[] = { "REHASH", IRC_REHASH, CLIENT_USER, 0, 0, 0 }, { "RESTART", IRC_RESTART, CLIENT_USER, 0, 0, 0 }, { "SERVER", IRC_SERVER, 0xFFFF, 0, 0, 0 }, - { "SQUIT", IRC_SQUIT, CLIENT_SERVER, 0, 0, 0 }, + { "SQUIT", IRC_SQUIT, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, { "STATS", IRC_STATS, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, + { "SUMMON", IRC_SUMMON, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, { "TIME", IRC_TIME, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, { "TOPIC", IRC_TOPIC, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, { "TRACE", IRC_TRACE, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, { "USER", IRC_USER, 0xFFFF, 0, 0, 0 }, { "USERHOST", IRC_USERHOST, CLIENT_USER, 0, 0, 0 }, + { "USERS", IRC_USERS, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, { "VERSION", IRC_VERSION, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, { "WALLOPS", IRC_WALLOPS, CLIENT_USER|CLIENT_SERVER, 0, 0, 0 }, { "WHO", IRC_WHO, CLIENT_USER, 0, 0, 0 }, --- src/ngircd/irc-info.h 11 Feb 2008 11:06:31 -0000 1.5 +++ src/ngircd/irc-info.h 17 Feb 2008 06:47:42 -0000 @@ -19,14 +19,17 @@ GLOBAL bool IRC_ADMIN PARAMS(( CLIENT *Client, REQUEST *Req )); +GLOBAL bool IRC_INFO PARAMS(( CLIENT *Client, REQUEST *Req )); GLOBAL bool IRC_ISON PARAMS(( CLIENT *Client, REQUEST *Req )); GLOBAL bool IRC_LINKS PARAMS(( CLIENT *Client, REQUEST *Req )); GLOBAL bool IRC_LUSERS PARAMS(( CLIENT *Client, REQUEST *Req )); GLOBAL bool IRC_MOTD PARAMS(( CLIENT *Client, REQUEST *Req )); GLOBAL bool IRC_NAMES PARAMS(( CLIENT *Client, REQUEST *Req )); GLOBAL bool IRC_STATS PARAMS(( CLIENT *Client, REQUEST *Req )); +GLOBAL bool IRC_SUMMON PARAMS(( CLIENT *Client, REQUEST *Req )); GLOBAL bool IRC_TIME PARAMS(( CLIENT *Client, REQUEST *Req )); GLOBAL bool IRC_USERHOST PARAMS(( CLIENT *Client, REQUEST *Req )); +GLOBAL bool IRC_USERS PARAMS(( CLIENT *Client, REQUEST *Req )); GLOBAL bool IRC_VERSION PARAMS(( CLIENT *Client, REQUEST *Req )); GLOBAL bool IRC_WHO PARAMS(( CLIENT *Client, REQUEST *Req )); GLOBAL bool IRC_WHOIS PARAMS(( CLIENT *Client, REQUEST *Req )); --- src/ngircd/messages.h 11 Dec 2007 11:29:44 -0000 1.74 +++ src/ngircd/messages.h 17 Feb 2008 06:47:42 -0000 @@ -77,6 +77,8 @@ #define RPL_BANLIST_MSG "367 %s %s %s" #define RPL_ENDOFBANLIST_MSG "368 %s %s :End of channel ban list" #define RPL_ENDOFWHOWAS_MSG "369 %s %s :End of WHOWAS list" +#define RPL_INFO_MSG "371 %s :%s" +#define RPL_ENDOFINFO_MSG "374 %s :End of INFO list" #define RPL_MOTD_MSG "372 %s :- %s" #define RPL_MOTDSTART_MSG "375 %s :- %s message of the day" #define RPL_ENDOFMOTD_MSG "376 %s :End of MOTD command" @@ -100,6 +102,8 @@ #define ERR_USERNOTINCHANNEL_MSG "441 %s %s %s :They aren't on that channel" #define ERR_NOTONCHANNEL_MSG "442 %s %s :You are not on that channel" #define ERR_USERONCHANNEL_MSG "443 %s %s %s :is already on channel" +#define ERR_SUMMONDISABLED_MSG "445 %s %s :Command disabled" +#define ERR_USERSDISABLED_MSG "446 %s %s :Command disabled" #define ERR_NOTREGISTERED_MSG "451 %s :Connection not registered" #define ERR_NOTREGISTEREDSERVER_MSG "451 %s :Connection not registered as server link" #define ERR_NEEDMOREPARAMS_MSG "461 %s %s :Syntax error" --- src/ngircd/irc-server.c 21 Nov 2007 12:16:36 -0000 1.46 +++ src/ngircd/irc-server.c 17 Feb 2008 06:47:42 -0000 @@ -285,6 +285,9 @@ IRC_SQUIT( CLIENT *Client, REQUEST *Req assert( Client != NULL ); assert( Req != NULL ); + if (! (Client_OperByMe(Client) || Client_Type(Client) == CLIENT_SERVER)) + return IRC_WriteStrClient( Client, ERR_NOPRIVILEGES_MSG, Client_ID( Client ) ); + /* Falsche Anzahl Parameter? */ if( Req->argc != 2 ) return IRC_WriteStrClient( Client, ERR_NEEDMOREPARAMS_MSG, Client_ID( Client ), Req->command ); --- src/ngircd/irc-info.c 17 Feb 2008 00:00:12 -0000 1.43 +++ src/ngircd/irc-info.c 17 Feb 2008 06:47:42 -0000 @@ -86,6 +86,71 @@ IRC_ADMIN(CLIENT *Client, REQUEST *Req ) GLOBAL bool +IRC_INFO( CLIENT *Client, REQUEST *Req ) +{ + CLIENT *target, *prefix; + + assert( Client != NULL ); + assert( Req != NULL ); + + /* Wrong number of parameters? */ + if(( Req->argc > 1 )) return IRC_WriteStrClient( Client, ERR_NEEDMOREPARAMS_MSG, Client_ID( Client ), Req->command ); + + /* Look for a target. */ + if( Req->argc == 1 ) target = Client_Search( Req->argv[0] ); + else target = Client_ThisServer( ); + + /* Determine prefix. */ + if( Client_Type( Client ) == CLIENT_SERVER ) prefix = Client_Search( Req->prefix ); + else prefix = Client; + if( ! prefix ) return IRC_WriteStrClient( Client, ERR_NOSUCHNICK_MSG, Client_ID( Client ), Req->prefix ); + + /* Pass on to another server? */ + if( target != Client_ThisServer( )) + { + if(( ! target ) || ( Client_Type( target ) != CLIENT_SERVER )) return IRC_WriteStrClient( prefix, ERR_NOSUCHSERVER_MSG, Client_ID( prefix ), Req->argv[0] ); + + /* forward */ + IRC_WriteStrClientPrefix( target, prefix, "INFO %s", Req->argv[0] ); + return CONNECTED; + } + + /* Answer with version info. */ + { + char msg[510]; + + strlcpy(msg, PACKAGE_NAME, sizeof(msg)); + strlcat(msg, "-", sizeof(msg)); + strlcat(msg, PACKAGE_VERSION, sizeof(msg)); + if( ! IRC_WriteStrClient( Client, RPL_INFO_MSG, Client_ID( prefix ), msg )) return DISCONNECTED; + +#ifdef CVSDATE + strlcpy(msg, "last revision ", sizeof(msg)); + strlcat(msg, CVSDATE, sizeof(msg)); + if( ! IRC_WriteStrClient( Client, RPL_INFO_MSG, Client_ID( prefix ), msg )) return DISCONNECTED; +#endif + + strlcpy(msg, "compiled ", sizeof(msg)); + strlcat(msg, __DATE__, sizeof(msg)); + if( ! IRC_WriteStrClient( Client, RPL_INFO_MSG, Client_ID( prefix ), msg )) return DISCONNECTED; + + strlcpy(msg, "with options ", sizeof(msg)); + strlcat(msg, NGIRCd_VersionAddition, sizeof(msg)); + if( ! IRC_WriteStrClient( Client, RPL_INFO_MSG, Client_ID( prefix ), msg )) return DISCONNECTED; + + strlcpy(msg, "started ", sizeof(msg)); + strlcat(msg, NGIRCd_StartStr, sizeof(msg)); + if( ! IRC_WriteStrClient( Client, RPL_INFO_MSG, Client_ID( prefix ), msg )) return DISCONNECTED; + + if( ! IRC_WriteStrClient( Client, RPL_ENDOFINFO_MSG, Client_ID( prefix ) )) return DISCONNECTED; + } + + IRC_SetPenalty( Client, 1 ); + return CONNECTED; +} /* IRC_INFO */ + + +GLOBAL bool IRC_ISON( CLIENT *Client, REQUEST *Req ) { char rpl[COMMAND_LEN]; @@ -470,6 +535,13 @@ IRC_STATS( CLIENT *Client, REQUEST *Req GLOBAL bool +IRC_SUMMON( CLIENT *Client, REQUEST *Req ) +{ + return IRC_WriteStrClient( Client, ERR_SUMMONDISABLED_MSG, Client_ID( Client ), Req->command ); +} /* IRC_SUMMON */ + + +GLOBAL bool IRC_TIME( CLIENT *Client, REQUEST *Req ) { CLIENT *from, *target; @@ -547,6 +619,13 @@ IRC_USERHOST( CLIENT *Client, REQUEST *R GLOBAL bool +IRC_USERS( CLIENT *Client, REQUEST *Req ) +{ + return IRC_WriteStrClient( Client, ERR_USERSDISABLED_MSG, Client_ID( Client ), Req->command ); +} /* IRC_USERS */ + + +GLOBAL bool IRC_VERSION( CLIENT *Client, REQUEST *Req ) { CLIENT *target, *prefix;

2 1

patch for missing RPL_WHOREPLY flags
by Dana Dahlstrom 17 Feb '08

17 Feb '08

RPL_WHOREPLY messages generated by IRC_WHO don't include flags (*,@,+) that should appear according to this description: http://www.mishscript.de/reference/rawhelp3.htm#raw352 Other IRC servers do include the flags. The following patch modifies who-away-test.e to expose missing flags, modifies ngircd-test.conf to accommodate who-away-test.e, and modifies irc-info.c to correct these problems. Two additional notes: (1) The modifications to irc-info.c on lines 767-772 create even more code duplicated in both IRC_WHO and IRC_Send_WHO. These functions could probably benefit from some refactoring. (2) The file who-away-test.e, especially as modified here, would probably be more accurately called simply who-test.e. Dana --- ngircd-CVSHEAD/src/testsuite/who-away-test.e 2008-02-11 03:06:32.000000000 -0800 +++ ngircd-WHO-fixes/src/testsuite/who-away-test.e 2008-02-16 11:55:58.000000000 -0800 @@ -16,19 +16,19 @@ expect { send "who\r" expect { timeout { exit 1 } - ":ngircd.test.server 352 nick * ~user localhost ngircd.test.server nick H :0 Real Name" + ":ngircd.test.server 352 nick \* ~user localhost ngircd.test.server nick H :0 Real Name" } -send "who 0\r" +send "join #channel\r" expect { timeout { exit 1 } - ":ngircd.test.server 352 nick * ~user localhost ngircd.test.server nick H :0 Real Name" + "@* JOIN :#channel" } -send "who *\r" +send "who 0\r" expect { timeout { exit 1 } - ":ngircd.test.server 352 nick * ~user localhost ngircd.test.server nick H :0 Real Name" + ":ngircd.test.server 352 nick #channel ~user localhost ngircd.test.server nick H@ :0 Real Name" } send "away :testing\r" @@ -37,28 +37,62 @@ expect { "306 nick" } +send "who *\r" +expect { + timeout { exit 1 } + ":ngircd.test.server 352 nick #channel ~user localhost ngircd.test.server nick G@ :0 Real Name" +} + +send "mode #channel +v nick\r" +expect { + timeout { exit 1 } + "@* MODE #channel +v nick\r" +} + send "who localhost\r" expect { timeout { exit 1 } - ":ngircd.test.server 352 nick * ~user localhost ngircd.test.server nick G :0 Real Name" + ":ngircd.test.server 352 nick #channel ~user localhost ngircd.test.server nick G@ :0 Real Name" +} + +send "mode #channel -o nick\r" +expect { + timeout { exit 1 } + "@* MODE #channel -o nick\r" } send "who ngircd.test.server\r" expect { timeout { exit 1 } - ":ngircd.test.server 352 nick * ~user localhost ngircd.test.server nick G :0 Real Name" + ":ngircd.test.server 352 nick #channel ~user localhost ngircd.test.server nick G+ :0 Real Name" +} + +send "part #channel\r" +expect { + timeout { exit 1 } + "@* PART #channel :nick" } send "who Real?Name\r" expect { timeout { exit 1 } - ":ngircd.test.server 352 nick * ~user localhost ngircd.test.server nick G :0 Real Name" + ":ngircd.test.server 352 nick \* ~user localhost ngircd.test.server nick G :0 Real Name" } -send "who nick\r" +send "oper TestOp 123\r" +expect { + timeout { exit 1 } + "MODE nick :+o" +} expect { timeout { exit 1 } - ":ngircd.test.server 352 nick * ~user localhost ngircd.test.server nick G :0 Real Name" + "381 nick" +} + +send "who 0 o\r" +expect { + timeout { exit 1 } + ":ngircd.test.server 352 nick \* ~user localhost ngircd.test.server nick G* :0 Real Name" } send "away\r" @@ -70,25 +104,55 @@ expect { send "who *cal*ho??\r" expect { timeout { exit 1 } - ":ngircd.test.server 352 nick * ~user localhost ngircd.test.server nick H :0 Real Name" + ":ngircd.test.server 352 nick \* ~user localhost ngircd.test.server nick H* :0 Real Name" +} + +send "join #opers\r" +expect { + timeout { exit 1 } + "@* JOIN :#opers" +} + +send "who #opers\r" +expect { + timeout { exit 1 } + ":ngircd.test.server 352 nick #opers ~user localhost ngircd.test.server nick H*@ :0 Real Name" +} + +send "mode #opers -o nick\r" +expect { + timeout { exit 1 } + "@* MODE #opers -o nick\r" } send "who *.server\r" expect { timeout { exit 1 } - ":ngircd.test.server 352 nick * ~user localhost ngircd.test.server nick H :0 Real Name" + ":ngircd.test.server 352 nick #opers ~user localhost ngircd.test.server nick H* :0 Real Name" +} + +send "mode #opers +v nick\r" +expect { + timeout { exit 1 } + "@* MODE #opers +v nick\r" } send "who Real*me\r" expect { timeout { exit 1 } - ":ngircd.test.server 352 nick * ~user localhost ngircd.test.server nick H :0 Real Name" + ":ngircd.test.server 352 nick #opers ~user localhost ngircd.test.server nick H*+ :0 Real Name" +} + +send "mode #opers +s\r" +expect { + timeout { exit 1 } + "@* MODE #opers +s\r" } send "who n?c?\r" expect { timeout { exit 1 } - ":ngircd.test.server 352 nick * ~user localhost ngircd.test.server nick H :0 Real Name" + ":ngircd.test.server 352 nick \* ~user localhost ngircd.test.server nick H* :0 Real Name" } send "quit\r" --- ngircd-CVSHEAD/src/testsuite/ngircd-test.conf 2007-11-18 07:07:16.000000000 -0800 +++ ngircd-WHO-fixes/src/testsuite/ngircd-test.conf 2008-02-16 11:53:36.000000000 -0800 @@ -9,6 +9,7 @@ MaxConnectionsIP = 0 ServerUID = 1 ServerGID = 1 + OperCanUseMode = yes [Operator] Name = TestOp --- ngircd-CVSHEAD/src/ngircd/irc-info.c 2008-02-11 15:15:21.000000000 -0800 +++ ngircd-WHO-fixes/src/ngircd/irc-info.c 2008-02-16 11:05:16.000000000 -0800 @@ -671,7 +671,7 @@ GLOBAL bool IRC_WHO( CLIENT *Client, REQUEST *Req ) { bool only_ops, have_arg, client_match; - const char *channelname, *client_modes; + const char *channelname, *client_modes, *chan_user_modes; char pattern[COMMAND_LEN]; char flags[4]; CL2CHAN *cl2chan; @@ -748,7 +748,7 @@ IRC_WHO( CLIENT *Client, REQUEST *Req ) else strcpy(flags, "H"); - if (only_ops) /* this client is an operator */ + if (strchr(client_modes, 'o')) /* this client is an operator */ strlcat(flags, "*", sizeof(flags)); /* Search suitable channel */ @@ -763,7 +763,14 @@ IRC_WHO( CLIENT *Client, REQUEST *Req ) } cl2chan = Channel_NextChannelOf(c, cl2chan); } - if (!cl2chan) + if (cl2chan) { + chan = Channel_GetChannel(cl2chan); + chan_user_modes = Channel_UserModes(chan, c); + if (strchr(chan_user_modes, 'o')) + strlcat(flags, "@", sizeof(flags)); + else if (strchr(chan_user_modes, 'v')) + strlcat(flags, "+", sizeof(flags)); + } else channelname = "*"; if (!write_whoreply(Client, c, channelname, flags))

2 1

IRC_WHO always returns "H" flag
by Dana Dahlstrom 17 Feb '08

17 Feb '08

Do you think the IRC_WHO function (in irc-info.c, at line 651) should always return the "H" flag in RPL_WHOREPLY messages? According to RFC 2812 this flag can be either "H" or "G". Unfortunately the RFC doesn't seem to suggest what this means, but I've found some pages that indicate it means "here" or "gone" (depending on the user's away status): http://www.mishscript.de/reference/rawhelp3.htm#raw352 http://www.mircscripts.org/forums.php?cid=3&id=159227 http://gametz.com/forum/Site_Feedback/topic/234776.html#M2688608 Dana

2 6

patch for channel-mode bug
by Dana Dahlstrom 16 Feb '08

16 Feb '08

Under some circumstances ngIRCd currently issues a channel MODE message with a trailing space after the last parameter, which isn't permitted by the grammar in RFC 2812 section 2.3.1: http://tools.ietf.org/html/rfc2812#section-2.3.1 The following patch modifies mode-test.e to expose this, and modifies irc-mode.c to correct it. Dana diff -pru ngircd-CVSHEAD/src/ngircd/irc-mode.c ngircd-mode-fix/src/ngircd/irc-mode.c --- ngircd-CVSHEAD/src/ngircd/irc-mode.c 2007-10-14 05:08:57.000000000 -0700 +++ ngircd-mode-fix/src/ngircd/irc-mode.c 2008-02-15 19:30:14.000000000 -0800 @@ -317,7 +317,7 @@ Channel_Mode( CLIENT *Client, REQUEST *R /* Prepare reply string */ if( set ) strcpy( the_modes, "+" ); else strcpy( the_modes, "-" ); - strcpy( the_args, " " ); + the_args[0] = '\0'; x[1] = '\0'; ok = CONNECTED; @@ -528,8 +528,8 @@ Channel_Mode( CLIENT *Client, REQUEST *R /* Channel-User-Mode */ if( Channel_UserModeAdd( Channel, client, x[0] )) { - strlcat( the_args, Client_ID( client ), sizeof( the_args )); strlcat( the_args, " ", sizeof( the_args )); + strlcat( the_args, Client_ID( client ), sizeof( the_args )); strlcat( the_modes, x, sizeof( the_modes )); Log( LOG_DEBUG, "User \"%s\": Mode change on %s, now \"%s\"", Client_Mask( client ), Channel_Name( Channel ), Channel_UserModes( Channel, client )); } @@ -552,8 +552,8 @@ Channel_Mode( CLIENT *Client, REQUEST *R /* Channel-User-Mode */ if( Channel_UserModeDel( Channel, client, x[0] )) { - strlcat( the_args, Client_ID( client ), sizeof( the_args )); strlcat( the_args, " ", sizeof( the_args )); + strlcat( the_args, Client_ID( client ), sizeof( the_args )); strlcat( the_modes, x, sizeof( the_modes )); Log( LOG_DEBUG, "User \"%s\": Mode change on %s, now \"%s\"", Client_Mask( client ), Channel_Name( Channel ), Channel_UserModes( Channel, client )); } @@ -572,8 +572,7 @@ Channel_Mode( CLIENT *Client, REQUEST *R /* Are there additional arguments to add? */ if( argadd[0] ) { - len = strlen( the_args ) - 1; - if( the_args[len] != ' ' ) strlcat( the_args, " ", sizeof( the_args )); + strlcat( the_args, " ", sizeof( the_args )); strlcat( the_args, argadd, sizeof( the_args )); } } @@ -586,9 +585,6 @@ chan_exit: len = strlen( the_modes ) - 1; if(( the_modes[len] == '+' ) || ( the_modes[len] == '-' )) the_modes[len] = '\0'; - /* Clean up argument string if there are none */ - if( ! the_args[1] ) the_args[0] = '\0'; - if( Client_Type( Client ) == CLIENT_SERVER ) { /* Forward mode changes to channel users and other servers */ diff -pru ngircd-CVSHEAD/src/testsuite/mode-test.e ngircd-mode-fix/src/testsuite/mode-test.e --- ngircd-CVSHEAD/src/testsuite/mode-test.e 2004-03-10 12:40:06.000000000 -0800 +++ ngircd-mode-fix/src/testsuite/mode-test.e 2008-02-15 19:33:48.000000000 -0800 @@ -72,7 +72,7 @@ expect { send "mode #channel +v nick\r" expect { timeout { exit 1 } - "@* MODE #channel +v nick" + "@* MODE #channel +v nick\r" } send "mode #channel +I nick1\r" @@ -96,7 +96,7 @@ expect { send "mode #channel -vo nick nick\r" expect { timeout { exit 1 } - "@* MODE #channel -vo nick nick" + "@* MODE #channel -vo nick nick\r" } send "quit\r"

2 1

patches from Debian package (0.10.3-2)
by Dana Dahlstrom 16 Feb '08

16 Feb '08

I see many patches from the older versions of the Debian package have been incorporated into ngIRCd releases, but there are a few differences from the most recent package that aren't incorporated yet. I don't know autoconf and automake very well, so I can't tell if these are important, but they're included inline below, extracted from here: http://ftp.de.debian.org/debian/pool/main/n/ngircd/ngircd_0.10.3-2.diff.gz --- ngircd-0.10.3.orig/config.guess +++ ngircd-0.10.3/config.guess @@ -4,7 +4,7 @@ # 2000, 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, # Inc. -timestamp='2007-03-06' +timestamp='2007-07-22' # This file is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by @@ -330,7 +330,7 @@ sun4*:SunOS:5.*:* | tadpole*:SunOS:5.*:*) echo sparc-sun-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'` exit ;; - i86pc:SunOS:5.*:*) + i86pc:SunOS:5.*:* | i86xen:SunOS:5.*:*) echo i386-pc-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'` exit ;; sun4*:SunOS:6*:*) @@ -793,7 +793,7 @@ exit ;; *:Interix*:[3456]*) case ${UNAME_MACHINE} in - x86) + x86) echo i586-pc-interix${UNAME_RELEASE} exit ;; EM64T | authenticamd) --- ngircd-0.10.3.orig/config.sub +++ ngircd-0.10.3/config.sub @@ -4,7 +4,7 @@ # 2000, 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, # Inc. -timestamp='2007-01-18' +timestamp='2007-06-28' # This file is (in principle) common to ALL GNU software. # The presence of a machine in this file suggests that SOME GNU # software @@ -475,8 +475,8 @@ basic_machine=craynv-cray os=-unicosmp ;; - cr16c) - basic_machine=cr16c-unknown + cr16) + basic_machine=cr16-unknown os=-elf ;; crds | unos) @@ -683,6 +683,10 @@ basic_machine=i386-pc os=-mingw32 ;; + mingw32ce) + basic_machine=arm-unknown + os=-mingw32ce + ;; miniframe) basic_machine=m68000-convergent ;;

2 2

patch for "JOIN 0" command (in RFC 2812)
by Dana Dahlstrom 05 Feb '08

05 Feb '08

Folks, The students in my software-engineering class are writing IRC clients in Java, and I'm running ngIRCd as a sandbox for them to play in. We noticed ngIRCd doesn't obey the "JOIN 0" command specified in RFC 2812: JOIN 0 ; Leave all currently joined channels. http://tools.ietf.org/html/rfc2812#section-3.2.1 I believe the following patch addresses this. Cheers! -- Dana Dahlstrom Lecturer, Computer Science and Engineering University of California, San Diego (Remember, reality is only an approximation of theory.) Index: src/ngircd/irc-channel.c =================================================================== RCS file: /srv/cvs/ngircd/ngircd/src/ngircd/irc-channel.c,v retrieving revision 1.41 diff -u -p -r1.41 irc-channel.c --- src/ngircd/irc-channel.c 7 Jan 2008 11:42:00 -0000 1.41 +++ src/ngircd/irc-channel.c 5 Feb 2008 05:53:11 -0000 @@ -46,6 +46,7 @@ IRC_JOIN( CLIENT *Client, REQUEST *Req ) char *channame, *channame_ptr, *key, *key_ptr, *flags, *topic, modes[8]; bool is_new_chan, is_invited, is_banned; CLIENT *target; + CL2CHAN *cl2chan; CHANNEL *chan; assert( Client != NULL ); @@ -61,6 +62,23 @@ IRC_JOIN( CLIENT *Client, REQUEST *Req ) else target = Client; if( ! target ) return IRC_WriteStrClient( Client, ERR_NOSUCHNICK_MSG, Client_ID( Client ), Req->prefix ); + /* Is argument "0"? */ + if (Req->argc == 1 && !strncmp("0", Req->argv[0], 2)) { + + /* Part from all channels. */ + cl2chan = Channel_FirstChannelOf( target ); + while (cl2chan) + { + chan = Channel_GetChannel( cl2chan ); + assert( chan != NULL ); + Channel_Part( target, Client, Channel_Name( chan ), Client_ID( target )); + + /* next */ + cl2chan = Channel_FirstChannelOf( target ); + } + return CONNECTED; + } + /* Are channel keys given? */ if (Req->argc > 1) { key = Req->argv[1]; Index: src/testsuite/channel-test.e =================================================================== RCS file: /srv/cvs/ngircd/ngircd/src/testsuite/channel-test.e,v retrieving revision 1.3 diff -u -p -r1.3 channel-test.e --- src/testsuite/channel-test.e 27 Dec 2003 13:01:12 -0000 1.3 +++ src/testsuite/channel-test.e 5 Feb 2008 05:53:11 -0000 @@ -69,6 +69,36 @@ expect { "@* PART #channel :nick" } +send "join #channel\r" +expect { + timeout { exit 1 } + "@* JOIN :#channel" +} +expect { + timeout { exit 1 } + "366" +} + +send "join #channel2\r" +expect { + timeout { exit 1 } + "@* JOIN :#channel2" +} +expect { + timeout { exit 1 } + "366" +} + +send "join 0\r" +expect { + timeout { exit 1 } + "@* PART #channel2 :nick" +} +expect { + timeout { exit 1 } + "@* PART #channel :nick" +} + send "quit\r" expect { timeout { exit 1 }

3 5

Patch: Specify restricted nicknames
by Fabian Schlager 23 Jan '08

23 Jan '08

Hi! I wrote a small patch that allows you to specify a file in the config containing forbidden nicks. This may be useful to disallow users changing their nicks to NickServ or ChanServ or any other nick. regards, Fabian

4 3

ngIRCd 0.11.0
by Alexander Barton 16 Jan '08

16 Jan '08

Hello! After releasing ngIRCd 0.10.4, which is "only" a bug-fix release, we are going to release our new "major release" 0.11.0 containing new features: the most prominent is the introduction of the "MaxNickLength" configuration option, which enables administrators to define the maxmimum nick name length dynamically. As usual, ngIRCd 0.11.0 can be downloaded from: <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.11.0.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.11.0.tar.gz> GnuPG signatures and a patch from 0.10.4 are available and can be downloaded from here: <ftp://ngircd.berlios.de/pub/ngircd/> <ftp://ngircd.barton.de/pub/ngircd/> The 0.11.x branch is tagged as "branch-0-11-x" in CVS, and this release is tagged as "rel-0-11-0". The NEWS file of ngIRCd 0.11.0 lists the following enhancements: - Add support for /STAT u (server uptime) command. - New [Server] configuration Option "Bind" allows to specify the source ip adress to use when connecting to remote server. - New configuration option "MaxNickLength" to specify the allowed length of user nick names. Note: must be unique in an IRC network! - Numeric 317: implemented "signon time" (displayed in WHOIS result). - Added new server configuration option "Passive" for "Server" blocks to disable automatic outgoing connections (similar to -p option to ngircd, but only for the specified server). (Tassilo Schweyer) - Added support for the WALLOPS command. Usage is restricted to IRC operators. You can view the complete NEWS and ChangeLog files online here: <http://ngircd.barton.de/doc/NEWS> <http://ngircd.barton.de/doc/ChangeLog> There have been no changes since ngIRCd 0.11.0-pre2. Have fun! Regards Alex

1 0

Bug 70 NOTICE doesn't take channels as targets
by Fabian Schlager 16 Jan '08

16 Jan '08

Hello, crawling through the sources and the bug tracker, I tried to solve BUG 70. As this is my first patch (and I'm not that good in programming C) I hope I did not make too many mistakes. The patch was tested on a local ngircd instance (MacOS X 10.4.11) and is attached to this mail. I haven't found any specific conditions for a user to be able to NOTICE a channel in the RFC's, so I assumed that they are the same as used for PRIVMSG, is this correct? Regards, Fabian

2 1

ngIRCd 0.10.4 – Security
by Alexander Barton 09 Jan '08

09 Jan '08

Hello All! ngIRCd 0.10.4 has been released, fixing a security bug which could cause the server to access invalid memory and to crash while handling specially crafted PART commands. All installations should be upgraded! <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.10.4.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.10.4.tar.gz> The GnuPG signature and a patch from 0.10.3 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> <ftp://ngircd.berlios.de/pub/ngircd/> All the above and more information, bug tracker, and CVS web frontend can be found via the homepage of ngIRCd which is located here: <http://ngircd.barton.de/> <http://ngircd.berlios.de/> This release is tagged as "rel-0-10-4" in CVS. The fix for the above mentioned security bug is the only change since release 0.10.3. In addition, ngIRCd 0.11.0-pre2 has been released which fixes this bug as well. It can be obtained via the usual FTP servers listed above. Regards Alex

3 5

ngIRCd 0.11.0-pre1
by Alexander Barton 03 Jan '08

03 Jan '08

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hello and happy new year! The next release of ngIRCd is overdue, but here is the first pointer: ngIRCd 0.11.0-pre1. So dont wait, go grab it, test it, and report bugs! ;-) <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.11.0-pre1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.11.0-pre1.tar.gz> The GnuPG signature and a patch from 0.10.3 is available and can be downloaded from here: <ftp://ngircd.berlios.de/pub/ngircd/> <ftp://ngircd.barton.de/pub/ngircd/> Changes since ngIRCd 0.10.3 are: - - Use dotted-decimal IP address if hostname is >= 64. - - Add support for /STAT u (server uptime) command. - - New [Server] configuration Option "Bind" allows to specify the source ip adress to use when connecting to remote server. - - New configuration option "MaxNickLength" to specify the allowed maximum length of user nick names. Note: must be unique in an IRC network! - - Enhanced the IRC+ protocol to support an enhanced "server handshake" and enable server to recognice numeric 005 (ISUPPORT) and 376 (ENDOFMOTD). See doc/Protocol.txt for details. - - Re-added doc/SSL.txt to distribution -- got lost somewhere!? - - Fixes the wrong logging output when nested servers are introduced to the network as well as the wrong output of the LINKS command. - - Update Mac OS X Xcode project file for Xcode 3. - - Adjust test suite to be usable on HP/UX 11.11 :-) - - Fix code to compile using K&R C compiler and ansi2kr again. - - New config option NoDNS: Disables DNS lookups when clients connect. - - Fixed propagation of channel mode 'P' on server links. - - Numeric 317: implemented "signon time" (displayed in WHOIS result). - - Fixed code that prevented GCC 2.95 to compile ngIRCd. - - Adjust path names in manual pages according to "./configure" settings. - - Added new server configuration option "Passive" for "Server" blocks to disable automatic outgoing connections (similar to -p option to ngircd, but only for the specified server). (Tassilo Schweyer) - - Don't connect to a server if a connection to another server within the same group is already in progress. - - Added support for the WALLOPS command. Usage is restricted to IRC operators. Please test this release and let us know which problems you encounter and which changes should be made for the final 0.11.0 release. Thanks! Regards Alex -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (Darwin) iD8DBQFHfBvSwkoPY342SFYRAlqwAKDEVz6S3eNxpHuyxnaMeLOn5Y5s8QCg9Sbi tQPWo6cUG0C1IitawWxlvKQ= =1WjZ -----END PGP SIGNATURE-----

1 0

Re: [ngIRCd-ML] Authentified users
by Florian Westphal 28 Dez '07

28 Dez '07

Damien Gerard <milipili(a)shikami.org> wrote: > However how can I specify an operator for a given channel ? You can't. However, an irc operator can use the /mode command (including /mode +o) if you enable the "OperCanUseMode" option in ngircd.conf. Regards, Florian -- Florian Westphal <fw(a)strlen.de> // http://www.strlen.de 1024D/F260502D 2005-10-20 Key fingerprint = 1C81 1AD5 EA8F 3047 7555 E8EE 5E2F DA6C F260 502D

1 0

Authentified users
by Damien Gerard 27 Dez '07

27 Dez '07

Hi ! I am trying ngircd because I love its simple configuration file :) However I have a little problem with "Operator". I define the following : [Operator] # Password of the IRC operator Password = test # Optional Mask from which /OPER will be accepted Mask = Milipili!*@* When I try to do the following : /oper Milipili test I have "bad password" In fact at the begining I would like to autoconnect a irssi client using its -w option. Where am I wrong ? -- Damien Gerard milipili(a)shikami.org People who used magic without knowing what they were doing usually came to a sticky end. All over the entire room, sometimes. -- (Terry Pratchett, Moving Pictures)

2 1

Re: [ngIRCd-ML] len nick
by Adam McGreggor 14 Okt '07

14 Okt '07

On Sun, Oct 14, 2007 at 02:06:53PM +0200, Mancausoft wrote: > I use the ngircd from debian package and the nick len is 9 char > I would support 20/30 char for nick :( I raised this back in May, q.v. http://arthur.ath.cx/pipermail/ngircd-ml/2007-May/000284.html > Where is the option for change this limit? Is hardcoded?? If is > harcoded, why? Yeah, hardcoded in one of the files, I can't remember which, but a quick grep should throw light on the matter; I couldn't be bothered re-compiling, and currently tollerate it as it as (on the principle that imo, the daemon's still a lot better than hybrid/dancer.) a

1 0

len nick
by Mancausoft 14 Okt '07

14 Okt '07

I use the ngircd from debian package and the nick len is 9 char I would support 20/30 char for nick :( Where is the option for change this limit? Is hardcoded?? If is harcoded, why? Mancausoft

2 1

ngIRCd 0.10.3 – Security
by Alexander Barton 01 Aug '07

01 Aug '07

Hello World! ngIRCd 0.10.3 has been released, fixing a security bug which could cause the server to crash while handling specially crafted JOIN commands. All installations should be upgraded! <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.10.3.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.10.3.tar.gz> The GnuPG signature and a patch from 0.10.2 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> All the above and more information, bug tracker, and CVS web frontend can be found via the homepage of ngIRCd which is located here: <http://ngircd.barton.de/> <http://ngircd.berlios.de/> This release is tagged as "rel-0-10-3" in CVS. The fix for the above mentioned security bug is the only change since release 0.10.2. Regards Alex

1 1

Away message
by Daniel Falk 20 Jul '07

20 Jul '07

Hi, I can't figure out how to leave an away message that's longer than one word. The server complains, saying "away Syntax Error" What do I need to do? Thanks.

2 1

ngIRCd 0.10.2
by Alexander Barton 08 Jun '07

08 Jun '07

Hello All! This morning ngIRCd 0.10.2 has been released. You can download it from via the homepage and the usual places: <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.10.1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.10.1.tar.gz> The GnuPG signature and a patch from 0.10.1 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> All the above and more information, bug tracker, and CVS web frontend can be found via the homepage of ngIRCd which can be found here: <http://ngircd.barton.de/> <http://ngircd.berlios.de/> This release is tagged as "rel-0-10-2" in CVS. There have been no changes since 0.10.2-pre2, the complete ChangeLog starting from 0.10.1 is attached: ngIRCd 0.10.2 ngIRCd 0.10.2-pre2 (2007-05-19) - Server links are allowed to use larger write buffers now (up to 50 KB). ngIRCd 0.10.2-pre1 (2007-05-05) - Fix compressed server links (broken since 0.10.0). - Predefined Channel configuration now allows specification of channel key (mode k) and maximum user count (mode l). - When using epoll() IO interface, compile in the select() interface as well and fall back to it when epoll() isn't available on runtime. - New configure option "--without-select" to disable select() IO API (even when using epoll(), see above). - Added support for IO APIs "poll()" and "/dev/poll". - Reorganized internal handling of invite and ban lists. Regards Alex

1 0

ngIRCd 0.10.2-pre2
by Alexander Barton 20 Mai '07

20 Mai '07

Hello World! This is the announcement for the second test version before ngIRCd 0.10.2 will be released. There is one change since -pre1, that allows server links to use bigger write buffers which should solve problems which occures when "big servers" tried to peer. Please test! I hope that this is the last pre-release and 0.10.2 is next! Download: <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.10.2-pre2.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.10.2-pre2.tar.gz> The GnuPG signature and a patch from 0.10.2-pre1 is available and can be downloaded from here: <ftp://ftp.berlios.de/pub/ngircd/> <ftp://ngircd.barton.de/pub/ngircd/> This pre-release is tagged as "rel-0-10-2-pre2" in CVS (the branch for the 0.10.x releases is named "branch-0-10-x"). Changes since ngIRCd 0.10.2-pre1 are: - Server links are allowed to use larger write buffers now (up to 50 KB). Changes since ngIRCd 0.10.1 are: - Fix compressed server links (broken since 0.10.0). - Predefined Channel configuration now allows specification of channel key (mode k) and maximum user count (mode l). - When using epoll() IO interface, compile in the select() interface as well and fall back to it when epoll() isn't available on runtime. - New configure option "--without-select" to disable select() IO API (even when using epoll(), see above). - Added support for IO APIs "poll()" and "/dev/poll". - Reorganized internal handling of invite and ban lists. Please test this release and let us know which problems you encounter and which changes should be made for the final 0.10.2 release. Thanks! Regards Alex

1 0

ngIRCd 0.10.2-pre1
by Alexander Barton 06 Mai '07

06 Mai '07

Hi All! We are going to test the next release of ngIRCd, version 0.10.2. Here it is: <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.10.2-pre1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.10.2-pre1.tar.gz> The GnuPG signature and a patch from 0.9.2 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> This pre-release is tagged as "rel-0-10-2-pre1" in CVS (the branch for the 0.10.x releases is named "branch-0-10-x"). Changes since ngIRCd 0.10.1 are: - Fix compressed server links (broken since 0.10.0). - Predefined Channel configuration now allows specification of channel key (mode k) and maximum user count (mode l). - When using epoll() IO interface, compile in the select() interface as well and fall back to it when epoll() isn't available on runtime. - New configure option "--without-select" to disable select() IO API (even when using epoll(), see above). - Added support for IO APIs "poll()" and "/dev/poll". - Reorganized internal handling of invite and ban lists. Please test this release and let us know which problems you encounter and which changes should be made for the final 0.10.2 release. Thanks! Regards Alex

1 0

setting nick maxlength?
by Adam McGreggor 01 Mai '07

01 Mai '07

Hiya, I'm back running a ngirc instance, and have encountered an annoyance. I want to be able to set the maximum allowed nick-length; it would only appear to be set-able at compilation time, not through the config file. Is that the case, or is there a setting in the config file to so do? fwiw, i'm using rhe Debain ported^Wpackaged version. TIA a

3 2

[Announce] TLS/SSL Support
by Florian Westphal 17 Feb '07

17 Feb '07

Hi. I have added SSL/TLS support to ngircd. What works: - incoming ssl/tls client and server connections - outgoing server connections (server links) What does not work yet: - (X509) Certificate support. The patch is here (created against ngircd CVSHEAD): http://www.strlen.de/ngircd/ngircd-ssl-2007-01-28.diff.bz2 Alternatively, you can grab ngircd-ssl from svn: svn co svn://svn.strlen.de/fw/ngircd-ssl/branches/ngircd-ssl-0.1.X You need to pass either --with-openssl or --with-gnutls to the configure script. A sample config file is included in the docs directory. Please test and let me know if it works on your platform. Thanks, Florian

2 4

how to login as administrator or operator
by arnadelo＠edu.xunta.es 10 Feb '07

10 Feb '07

We have just installed ngircd on ubuntu at school. We have created 3 channels, and we can join this channels and chat. What we need now is to login as administrators or operators. The program we use to chat is xchat. Thans

2 1

ngIRCd authentication hooks
by Glen Stewart 25 Jan '07

25 Jan '07

I have a semi-closed system of 1500 users that I would like to offer IRC services to - but only if they authenticate against a custom MySQL table entry. I'd like to call a perl script that will do the lookup of a given identity & password pair, and return a 0 exit code for success, and a non-zero code for failure to authenticate. So my question to the ngIRCd community & author is whether there are plans, or interest in an authentication hook being added. And, whether this kind of approach would be incompatible with standards-compliant IRC clients. Thanks! -- Associate.com - THE Place to Associate! http://freegroups.net Is Your E-mail Group Here? See http://freegroups.net/faq/your-group-here.html

3 2

ngIRCd 0.10.1
by Alexander Barton 17 Dez '06

17 Dez '06

Hello "World"! Since the release of ngIRCd 0.10.0 there have been a few fixes and enhancements to the CVS branch of ngIRCd 0.10.x, so here we go: ngIRCd 0.10.1 has been released and can be downloaded from the usual locations: <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.10.1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.10.1.tar.gz> And soon it should be available via HTTP from this site: <http://prdownload.berlios.de/ngircd/ngircd-0.10.1.tar.gz> The GnuPG signature and a patch from 0.9.0 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> All the above and more information, bug tracker, and CVS web frontend can be found via the homepage of ngIRCd which can be found here: <http://ngircd.barton.de/> <http://ngircd.berlios.de/> This release is tagged as "rel-0-10-1" in CVS. Changes since ngIRCd 0.10.0 are: - Fixed validation of server names containing digits. - Update the "info text" of the local server after re-reading configuration. - Changed Numerics 265 and 266 to follow ircd 2.11.x "standards". - Allow PASS syntax defined in RFC 1459 for server links, too. - Enhanced ISUPPORT message (005 numeric). - New configuration option "PredefChannelsOnly": if set, clients can only join predefined channels. - Code cleanups: use "LogDebug(...)" instead of "Log(LOG_DEBUG, ...)", use "strcspn()", unsigned vs. signed, use "const", fix whitespaces, ... Regards Alex

1 0

[PATCH] Matche
by Dieter Rauschenberger 05 Okt '06

05 Okt '06

Hello All! I changed the Matche() function in match.c to work with REGCOMP(3) (regular expressions). This adds several possibilities in the conf file. Now you can write something like: [Operator] Mask = d-?ra[0-9x]?!~?d-?ra@.*\.d-ra\.net The default Mask had to be changed slightly: Mask = .*!ident(a)somewhere\.example\.com Another advantage: The source code is much more shorter! Keep it as simple as possible - but not simpler! The patch: Index: ngircd/doc/sample-ngircd.conf =================================================================== RCS file: /srv/cvs/ngircd/ngircd/doc/sample-ngircd.conf,v retrieving revision 1.37 diff -u -p -r1.37 sample-ngircd.conf --- ngircd/doc/sample-ngircd.conf 9 Apr 2006 12:27:23 -0000 1.37 +++ ngircd/doc/sample-ngircd.conf 5 Oct 2006 16:14:45 -0000 @@ -115,7 +115,7 @@ ;Password = ThePwd # Optional Mask from which /OPER will be accepted - ;Mask = *!ident(a)somewhere.example.com + ;Mask = .*!ident(a)somewhere\.example\.com [Operator] # More [Operator] sections, if you like ... Index: ngircd/src/ngircd/match.c =================================================================== RCS file: /srv/cvs/ngircd/ngircd/src/ngircd/match.c,v retrieving revision 1.4 diff -u -p -r1.4 match.c --- ngircd/src/ngircd/match.c 31 Jul 2005 20:13:08 -0000 1.4 +++ ngircd/src/ngircd/match.c 5 Oct 2006 16:14:46 -0000 @@ -19,6 +19,8 @@ static char UNUSED id[] = "$Id: match.c, #include "imp.h" #include <assert.h> #include <string.h> +#include <sys/types.h> +#include <regex.h> #include "exp.h" #include "match.h" @@ -33,9 +35,9 @@ static char UNUSED id[] = "$Id: match.c, static int Matche PARAMS(( char *p, char *t )); -static int Matche_After_Star PARAMS(( char *p, char *t )); +#define MATCH_NONE 7 /* no match in regexec */ #define MATCH_PATTERN 6 /* bad pattern */ #define MATCH_LITERAL 5 /* match failure on literal match */ #define MATCH_RANGE 4 /* match failure on [..] construct */ @@ -52,202 +54,27 @@ Match( char *Pattern, char *String ) else return false; } /* Match */ - +/* + * p = pattern + * t = input text + * + * */ static int Matche( char *p, char *t ) { - register char range_start, range_end; - bool invert; - bool member_match; - bool loop; + int erg, status; + regex_t preg; - for( ; *p; p++, t++ ) + status = regcomp(&preg, p, REG_EXTENDED|REG_NOSUB); + if ( status != 0 ) + erg = MATCH_PATTERN; + else { - /* if this is the end of the text then this is the end of the match */ - if( ! *t ) - { - return ( *p == '*' && *++p == '\0' ) ? MATCH_VALID : MATCH_ABORT; - } - - /* determine and react to pattern type */ - switch( *p ) - { - case '?': /* single any character match */ - break; - - case '*': /* multiple any character match */ - return Matche_After_Star( p, t ); - - case '[': /* [..] construct, single member/exclusion character match */ - /* move to beginning of range */ - p++; - - /* check if this is a member match or exclusion match */ - invert = false; - if( *p == '!' || *p == '^' ) - { - invert = true; - p++; - } - - /* if closing bracket here or at range start then we have a malformed pattern */ - if ( *p == ']' ) return MATCH_PATTERN; - - member_match = false; - loop = true; - - while( loop ) - { - /* if end of construct then loop is done */ - if( *p == ']' ) - { - loop = false; - continue; - } - - /* matching a '!', '^', '-', '\' or a ']' */ - if( *p == '\\' ) range_start = range_end = *++p; - else range_start = range_end = *p; - - /* if end of pattern then bad pattern (Missing ']') */ - if( ! *p ) return MATCH_PATTERN; - - /* check for range bar */ - if( *++p == '-' ) - { - /* get the range end */ - range_end = *++p; - - /* if end of pattern or construct then bad pattern */ - if( range_end == '\0' || range_end == ']' ) return MATCH_PATTERN; - - /* special character range end */ - if( range_end == '\\' ) - { - range_end = *++p; - - /* if end of text then we have a bad pattern */ - if ( ! range_end ) return MATCH_PATTERN; - } - - /* move just beyond this range */ - p++; - } - - /* if the text character is in range then match found. make sure the range - * letters have the proper relationship to one another before comparison */ - if( range_start < range_end ) - { - if( *t >= range_start && *t <= range_end ) - { - member_match = true; - loop = false; - } - } - else - { - if( *t >= range_end && *t <= range_start ) - { - member_match = true; - loop = false; - } - } - } - - /* if there was a match in an exclusion set then no match */ - /* if there was no match in a member set then no match */ - if(( invert && member_match ) || ! ( invert || member_match )) return MATCH_RANGE; - - /* if this is not an exclusion then skip the rest of the [...] - * construct that already matched. */ - if( member_match ) - { - while( *p != ']' ) - { - /* bad pattern (Missing ']') */ - if( ! *p ) return MATCH_PATTERN; - - /* skip exact match */ - if( *p == '\\' ) - { - p++; - - /* if end of text then we have a bad pattern */ - if( ! *p ) return MATCH_PATTERN; - } - - /* move to next pattern char */ - p++; - } - } - break; - case '\\': /* next character is quoted and must match exactly */ - /* move pattern pointer to quoted char and fall through */ - p++; - - /* if end of text then we have a bad pattern */ - if( ! *p ) return MATCH_PATTERN; - - /* must match this character exactly */ - default: - if( *p != *t ) return MATCH_LITERAL; - } + status = regexec(&preg, t, (size_t)0, NULL, 0); + erg = status ? MATCH_NONE : MATCH_VALID; } - /* if end of text not reached then the pattern fails */ - - if( *t ) return MATCH_END; - else return MATCH_VALID; + regfree(&preg); + return erg; } /* Matche */ - -static int -Matche_After_Star( char *p, char *t ) -{ - register int nextp, match = 0; - - /* pass over existing ? and * in pattern */ - while( *p == '?' || *p == '*' ) - { - /* take one char for each ? and + */ - if (*p == '?') - { - /* if end of text then no match */ - if( ! *t++ ) return MATCH_ABORT; - } - - /* move to next char in pattern */ - p++; - } - - /* if end of pattern we have matched regardless of text left */ - if( ! *p ) return MATCH_VALID; - - /* get the next character to match which must be a literal or '[' */ - nextp = *p; - if( nextp == '\\' ) - { - nextp = p[1]; - - /* if end of text then we have a bad pattern */ - if( ! nextp ) return MATCH_PATTERN; - } - - /* Continue until we run out of text or definite result seen */ - do - { - /* a precondition for matching is that the next character - * in the pattern match the next character in the text or that - * the next pattern char is the beginning of a range. Increment - * text pointer as we go here */ - if( nextp == *t || nextp == '[' ) match = Matche( p, t ); - - /* if the end of text is reached then no match */ - if( ! *t++ ) match = MATCH_ABORT; - } while( match != MATCH_VALID && match != MATCH_ABORT && match != MATCH_PATTERN ); - - /* return result */ - return match; -} /* Matche_After_Star */ - - /* -eof- */ Regards Dieter

2 2

ngIRCd 0.10.0
by Alexander Barton 01 Okt '06

01 Okt '06

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hello All! All reported problems of the prereleases have been fixed, here we go: ngIRCd 0.10.0 has been released! It can be downloaded from the usual places: <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.10.0.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.10.0.tar.gz> And soon it should be available via HTTP from this location: <http://prdownload.berlios.de/ngircd/ngircd-0.10.0.tar.gz> The GnuPG signature and a patch from 0.9.2 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> The 0.10.x branch has been tagged as "branch-0-10-x" in CVS, and this release is tagged as "rel-0-10-0". Changes since ngIRCd 0.9.2 are: - Fixed file handle leak when daemon is not able to send MOTD to a client. - Fixed build problems with GCC option -fstack-protector. - Minor documentation updates. - Validate "ServerName" (see RFC 2812, section 2.3.1). - Enhanced DIE to accept a single parameter ("comment text") which is sent to all locally connected clients before the server goes down. - The ngIRCd handles time shifts backwards more gracefully now (the timeout handling doesn't disconnect clients by mistake any more). - Internal: Restructured connection handling (the connection ID is equal to the file descriptor of the connection). - Internal: Simplified resolver code. - JOIN now supports more than one channel key at a time. - Implemented numeric "333": Time and user name who set a channel topic. - Enhanced the handler for PING and PONG commands: fix forwarding and enable back-passing of a client supplied additional argument of PING. - Changed handling of timeouts for unregistered connections: don't reset the counter if data is received and disconnect clients earlier. - Removed unnecessary #define of "LOCAL", now use plain C "static" instead. - Channel topics are no longer limited to 127 characters: now the only limit is the maximum length of an IRC command, i. e. 512 bytes (in practice, this limits the topic to about 490 characters due to protocol overhead). - Reverse DNS lookup code now checks the result by doing an additional lookup to prevent spoofing. - Added new IO layer which (optionally) supports epoll() and kqueue() in addition to the select() interface. Regards Alex -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFFH/zWwkoPY342SFYRAvt/AJsELm3WIc27Lr4OYzvtWPoBOpd/ZgCgjgdw Cj6wPn7cMeBaQiMQ88S/3ho= =Y3OQ -----END PGP SIGNATURE-----

1 0

ngIRCd 0.10.0-pre1
by Alexander Barton 30 Aug '06

30 Aug '06

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi All! After a quite long time, the next release of ngIRCd, version 0.10.0, is pending. Before releasing this new version, we will publish at least one pre-release which this mail is now announcing: ngIRCd 0.10.0-pre1! ngIRCd 0.10.0 can be downloaded from the following locations: <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.10.0-pre1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.10.0-pre1.tar.gz> The GnuPG signature and a patch from 0.9.2 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> This pre-release is tagged as "rel-0-10-0-pre1" in CVS; the branch for the 0.10.x releases is named "branch-0-10-x". Changes since ngIRCd 0.9.2 are: - Validate "ServerName" (see RFC 2812, section 2.3.1). - Enhanced DIE to accept a single parameter ("comment text") which is sent to all locally connected clients before the server goes down. - The ngIRCd handles time shifts backwards more gracefully now (the timeout handling doesn't disconnect clients by mistake any more). - Internal: Restructured connection handling (the connection ID is equal to the file descriptor of the connection). - Internal: Simplified resolver code. - JOIN now supports more than one channel key at a time. - Implemented numeric "333": Time and user name who set a channel topic. - Enhanced the handler for PING and PONG commands: fix forwarding and enable back-passing of a client supplied additional argument of PING. - Changed handling of timeouts for unregistered connections: don't reset the counter if data is received and disconnect clients earlier. - Removed unnecessary #define of "LOCAL", now use plain C "static" instead. - Channel topics are no longer limited to 127 characters: now the only limit is the maximum length of an IRC command, i. e. 512 bytes (in practice, this limits the topic to about 490 characters due to protocol overhead). - Reverse DNS lookup code now checks the result by doing an additional lookup to prevent spoofing. - Added new IO layer which (optionally) supports epoll() and kqueue () in addition to the select() interface. Please test this release and let us know which problems you encounter and which changes should be made for the final 0.10.0 release. Thanks! Regards Alex -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (Darwin) iD8DBQFE01CLwkoPY342SFYRAgkRAJ9nUDzKIe580W5CUnUFXQCUKlZv8wCgqlP8 eY+Y5NaVQnbjFOJNEVvrKEs= =d8Zm -----END PGP SIGNATURE-----

3 3

Newbie Questions about ngIRCd
by Stephan Fabel 02 Mär '06

02 Mär '06

Hi list, I'm new to the topic of IRC (as an admin), and am in the following situation: I would like to set up an IRC server, with the following features: - four permanent groups, two of those should be password protected - assign a password for connection to the server - use SSL as a connection possibility - user administration: make certain users (identified by a password) channel operators on "their" channels I know about stunnel, but would prefer a "native" solution. If this is not possible at the moment, stunnel would work. I already assigned a password to the server for connection, and so far this works as well. What doesn't seem to work (or I am too blind to find it in the docs) is how to make a permanent channel password protected and how to "administer" the operators insofar as when they log in, they automatically get +o set on their channel. If a user doesn't provide a password (to become operator), he should not be denied access, but simply not be able to get op status or open new channels. If possible, I would like to avoid a channel bot. Is there a way for ngIRCd to have these features? Thanks, Stephan

1 0

Channel Operators
by Michael Belanger 01 Feb '06

01 Feb '06

Greetings, I have been running ngIRC for a few days and am impressed with how quickly it compiled and just worked nearly out of the box! I have currently two static channels defined in the config file for ngirc. However, I seem to be missing a critical piece of information to setup channel operators. It appears that the operator is normally the channel creator -- when the channel is created on the fly.. For these static channels, I would think that it is predefined in the configuration somewhere. The example conf does not show such an option. Can anyone guide me to the solution? Much thanks! -M -- Michael Belanger CICLOPS, Space Science Institute Join us on IRC: irc.ciclops.org channel #ciclops

2 2

Compile without NickServ support?
by AA 29 Jan '06

29 Jan '06

Hello, I would like to not have to mess with NickServer. I will be running this IRC server on a private LAN with no outside access and do not want to have to mess with registration. Would I just be able to add --disable-nickserv when running the ./configure script? Thanks much, Aaron

2 1

How to register nicks and channels?
by Martin Wegner 16 Jan '06

16 Jan '06

Hello. Is it possible for users to register nicknames or channels in ngircd? If yes, how? NickServ and ChanServ I know from other IRC servers do not exist, right? I have created an oper in ngircd.conf: [Operator] Name = martin Password = xxxxxxxxxx I have my username set to "martin". How can I identify myself with a password to be the operator? How can I define OPs for a persistent channel? Thanks in advance, martin -- Get my public GPG key from pgp.mit.edu or wwwkeys.pgp.net Key ID: 0x44085D12 -- Homepage: http://mroot.net/ Powered by Gentoo Linux (http://gentoo.org/)

3 2

ngIRCd in Debian!
by Alexander Barton 06 Dez '05

06 Dez '05

Hi all! I just want to let you know that ngIRCd has been added to the "unstable" and "testing" distribution of Debian Linux: <http://packages.debian.org/testing/net/ngircd> Thanks to Mario Iseli! Regards Alex PS.: Now I have to test this package myself :-)

1 0

ngIRCd 0.9.2
by Alexander Barton 15 Okt '05

15 Okt '05

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi All! There have been a few fixes for the 0.9.x branch of ngIRCd, so here comes the next release: ngIRCd 0.9.2! ngIRCd 0.9.2 can be downloaded from the following places: <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.9.2.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.9.2.tar.gz> And soon it should be available via HTTP from this site: <http://download.berlios.de/ngircd/ngircd-0.9.2.tar.gz> The GnuPG signature and a patch from 0.9.0 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> This release is tagged as "rel-0-9-2" in CVS. Changes since ngIRCd 0.9.1 are: - Fixed a bug that could cause the damon to crash when outgoing server connections can't be established. - Fixed a bug that caused the daemon to leak file descriptors when no resolver subprocesses could be created. - Fixed server NOTICEs to users with "s" mode ("server messages"). - Fixed a format string bug in "connection statistics" messages to clients. Regards Alex - -- Alexander Barton, Freiburg, Germany alex(a)barton.de, http://www.barton.de/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFDUQXXwkoPY342SFYRAqo/AKDWRzZ9H9XmP7OwSU5RwLw2uVN8dwCfVYBT pVSa282J2+iDRvpvg1gZrLI= =y66f -----END PGP SIGNATURE-----

1 0

Kicking idle people who are not on a channel
by Jens Kleine 13 Aug '05

13 Aug '05

Hi, another quite unusual hack but maybe usefull for some of you: Adding these two lines to conn.c after line 1313: if(Channel_FirstChannelOf(c) == NULL && Conn_GetIdle(Client_Conn(c)) > 60) Conn_Close( i, NULL, "Not on a channel timeout", true ); will auto-close connections to people who hang around on your server but don't chat on any of the channels. (I just can't stand them) It gives them 60 seconds to join a channel or switch between them. Yours, Jens. -- Many an ancient lord's last words had been, 'You can't kill me because I've got magic aaargh.' (Terry Pratchett - Interesting Times)

2 3

rc.d start-up
by Adam McGreggor 13 Aug '05

13 Aug '05

Helo, I've recently installed ngirc on a FreeBSD install, using the ports collection. There's one thing that appears to be missing, and that's a shell script to start the daemon when the server's rebooted. I see in http://arthur.barton.de/pipermail/ngircd-ml/2005-January/000218.html that a ngircd.sh has been written, but can't see this - should it be there? I can't find much usefulness with a web-search, nor from /docs TIA, Adam

3 3

ngIRCd 0.9.1
by Alexander Barton 04 Aug '05

04 Aug '05

Hello All! Due to a severe bug in the handler for the IRC KILL command in ngIRCd 0.9.0, we are releasing version 0.9.1. ngIRCd 0.9.1 can be downloaded from the following places: <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.9.1.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.9.1.tar.gz> And soon it should be available via HTTP from this site: <http://download.berlios.de/ngircd/ngircd-0.9.1.tar.gz> The GnuPG signature and a patch from 0.9.0 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> This release is tagged as "rel-0-9-1" in CVS. Changes since ngIRCd 0.9.0 are: - The KILL command killed much more than desired (including server links!) when the target user is connected to a remote server. Bug introduced in ngIRCd 0.9.0 ... Reported by <qssl(a)fastmail.fm>, Thanks! - Changed some constants to be "signed" (instead of unsigned) to solve problems with old (pre-ANSI) compilers. Regards Alex -- Alexander Barton, Freiburg, Germany http://www.barton.de/, alex(a)barton.de

1 0

ngIRCd 0.9.0
by Alexander Barton 25 Jul '05

25 Jul '05

Hello World! We got no bug and/or regression reports for ngIRCd 0.9.0-pre1, so here it is: ngIRCd 0.9.0! There have been no changes to the source code of the daemon since the pre-release and only minor changes (read "enhancements") to the build system have been applied. Therefore 0.9.0 should perform just as well as 0.9.0-pre1 -- Have fun! ngIRCd 0.9.0 can be downloaded from the following places: <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.9.0.tar.gz> <ftp://ngircd.barton.de/pub/ngircd/ngircd-0.9.0.tar.gz> And soon it should be available via HTTP from this site: <http://download.berlios.de/ngircd/ngircd-0.9.0.tar.gz> The GnuPG signature and a patch from 0.8.3 is available and can be downloaded from here: <ftp://ngircd.barton.de/pub/ngircd/> The 0.9.x branch has been tagged as "branch-0-9-x" in CVS, and this release is tagged as "rel-0-9-0". Changes since ngIRCd 0.8.3 are: - Fixed maximum length of user names, now allow up to 9 characters. - Cut off oversized IRC messages that should be sent to the network instead of shuttding down the (wrong) connection. - Don't generate error messages for unknown commands received before the client is registered with the server (like the original ircd). - Never run with root privileges but always switch the user ID. - Make "netsplit" messages RFC compliant. - Fix handling of QUIT Messages: send only one message, even if the client is member of multiple channels. - Don't exit server if closing of a socket fails; instead ignore it and pray that this will be "the right thing" ... - Implemented the IRC function "WHOWAS". - Don't enable assert() calls when not ./configure'd with --enable-debug. - Fixed ./configure test for TCP Wrappers: now it runs on Apple Mac OS X as well. - Enhanced configure script: now you can pass an (optional) search path to all --with-XXX parameters, e. g. "--with-ident=/opt/ident". - Removed typedefs for the native C datatypes. Use stdbool.h / inttypes.h if available. - New configuration option "OperServerMode" to enable a workaround needed when running an network with ircd2 servers and "OperCanUseMode" enabled to prevent the ircd2 daemon to drop mode changes of IRC operators. Patch by Florian Westphal, <westphal(a)foo.fh-furtwangen.de>. - Implemented support for "secret channels" (channel mode "s"). - New configuration option "Mask" for [Operator] sections to limit OPER commands to users with a specific IRC mask. Patch from Florian Westphal. - Write "error file" (/tmp/ngircd-XXX.err) only if compiled with debug code ("--enable-debug") and running as daemon process. - Don't create version information string each time a client connects but instead on server startup. By Florian Westphal. - New configuration variable "PidFile", section "[Global]": if defined, the server writes its process ID (PID) to this file. Default: off. Idea of Florian Westphal, westphal(a)foo.fh-furtwangen.de>. - Code cleanups from Florian, <westphal(a)foo.fh-furtwangen.de>. - Raised the maximum length of passwords to 20 characters. - Fixed a memory leak when resizing the connection pool and realloc() failed. Now we don't fall back to malloc(), which should be sane anyway. Patch from Florian Westphal, <westphal(a)foo.fh-furtwangen.de>. - Added support for the Howl (http://www.porchdogsoft.com/products/ howl/) Rendezvous API, in addition to the API of Apple (Mac OS X). The available API will be autodetected when you call "./configure --with-rendezvous". - Made ngIRCd compile on HP/UX 10.20 with native HP pre-ANSI C compiler and most probably other older C compilers on other systems. - When the daemon should switch to another user ID (ServerID is defined in the configuration file) and is not running in a chroot environment, it changes its working directory to the home directory of this user. This should enable the system to write proper core files when not running with root privileges ... Regards Alex -- Alexander Barton, Freiburg, Germany http://www.barton.de/, alex(a)barton.de

1 0

ngIRCd 0.9.0-pre1
by Alexander Barton 09 Jul '05

09 Jul '05

Hi all! A new pre-release version of ngIRCd, version 0.9.0-pre1, is available for testing and can be downloaded from: <ftp://Arthur.Ath.CX/pub/Users/alex/ngircd/ngircd-0.9.0-pre1.tar.gz> The GnuPG signature and a patch from 0.7.7 is available and can be downloaded from here: <ftp://Arthur.Ath.CX/pub/Users/alex/ngircd/> Please test & report any bug you find to this list. Thanks! :-)) The 0.9.x branch has been tagged as "branch-0-9-x" in CVS, this pre- release is tagged as "rel-0-9-0-pre1". Changes since ngIRCd 0.8.3 are: - Fixed maximum length of user names, now allow up to 9 characters. - Cut off oversized IRC messages that should be sent to the network instead of shuttding down the (wrong) connection. - Don't generate error messages for unknown commands received before the client is registered with the server (like the original ircd). - Never run with root privileges but always switch the user ID. - Make "netsplit" messages RFC compliant. - Fix handling of QUIT Messages: send only one message, even if the client is member of multiple channels. - Don't exit server if closing of a socket fails; instead ignore it and pray that this will be "the right thing" ... - Implemented the IRC function "WHOWAS". - Don't enable assert() calls when not ./configure'd with --enable- debug. - Fixed ./configure test for TCP Wrappers: now it runs on Mac OS X as well. - Enhanced configure script: now you can pass an (optional) search path to all --with-XXX parameters, e. g. "--with-ident=/opt/ident". - Removed typedefs for the native C datatypes. Use stdbool.h / inttypes.h if available. - New configuration option "OperServerMode" to enable a workaround needed when running an network with ircd2 servers and "OperCanUseMode" enabled to prevent the ircd2 daemon to drop mode changes of IRC operators. Patch by Florian Westphal, <westphal(a)foo.fh-furtwangen.de>. - Implemented support for "secret channels" (channel mode "s"). - New configuration option "Mask" for [Operator] sections to limit OPER commands to users with a specific IRC mask. Patch from Florian Westphal. - Write "error file" (/tmp/ngircd-XXX.err) only if compiled with debug code ("--enable-debug") and running as daemon process. - Don't create version information string each time a client connects but instead on server startup. By Florian Westphal. - New configuration variable "PidFile", section "[Global]": if defined, the server writes its process ID (PID) to this file. Default: off. Idea of Florian Westphal, <westphal(a)foo.fh-furtwangen.de>. - Code cleanups from Florian Westphal, <westphal(a)foo.fh- furtwangen.de>. - Raised the maximum length of passwords to 20 characters. - Fixed a memory leak when resizing the connection pool and realloc() failed. Now we don't fall back to malloc(), which should be sane anyway. Patch from Florian Westphal, <westphal(a)foo.fh-furtwangen.de>. - Added support for the Howl (http://www.porchdogsoft.com/products/ howl/) Rendezvous API, in addition to the API of Apple (Mac OS X). The available API will be autodetected when you call "./configure --with- rendezvous". - Made ngIRCd compile on HP/UX 10.20 with native HP pre-ANSI C compiler and most probably other older C compilers on other systems. - When the daemon should switch to another user ID (ServerID is defined in the configuration file) and is not running in a chroot environment, it changes its working directory to the home directory of this user. This should enable the system to write proper core files when not running with root privileges ... Regards Alex -- Alexander Barton, Freiburg, Germany http://www.barton.de/, alex(a)barton.de

1 0

How to track development?
by Alexander Barton 10 Apr '05

10 Apr '05

Hi all! Here are some hints for those interested in tracking the ongoing development of ngIRCd: - You can visit <http://cia.navi.cx/stats/project/ngIRCd> to review all the latest changes committed to the CVS ("Concurrent Versions Systems", the software used to manage the source files of ngIRCd). - You can subscribe to the RSS feed located at <http://cia.navi.cx/stats/project/ngIRCd/.rss> to be notified of all CVS commits (as displayed at the page above). - And you can subscript to the "ngIRCd-CVS" mailing list to receive an email each time changes become committed to the CVS. The homepage of this list is located at <http://arthur.ath.cx/cgi-bin/mailman/listinfo/ngircd-cvs-ml>. You'll "only" get the changelog of the affected files this way, not the complete patch. - Use the "ViewCVS" interface located at <http://arthur.ath.cx/cgi-bin/viewcvs.cgi/ngircd/> to review the current CVS repository, including the possibility to review all old versions of all files, to generate specific patches and to download a current archive ("tarball"). All "regular" releases will be announced via this mailing list, on the homepage <http://arthur.ath.cx/~alex/ngircd/>, and on freshmeat.net where you can subscribe to be notified on new releases via email: <http://freshmeat.net/projects/ngircd/>. Regards Alex -- Alexander Barton, Freiburg, Germany alex(a)barton.de, http://www.barton.de/

1 0

Re: [ngIRCd-ML] IRC Services
by Stefan Johnson 21 Mär '05

21 Mär '05

--- Alexander Barton <alex(a)barton.de> wrote: > Hi Stefan! > > > I got to looking around and > > discovered this little gem of a server. I > downloaded > > it, compiled it (no problems what-so-ever) and > > installed it without a single hitch. I even went > in > > and found the correct line in the defines.h file > to > > modify the nick length quickly because the code is > > clean and self documenting. KUDOS. > > Thanks :-) No problem. I don't speak or read German, but your code is self-documenting and I can understand what's going on without bothering with the comments :) > > > It's my understanding that services work by > connecting > > to the IRC servers they service as "just another > IRC > > server." The servers don't know anything about > the > > services server other than it's another IRC server > on > > the network. > > This is one possibility. And this one should/could > already work with > the ngIRCd. I agree. I haven't found a service package to try this with as of yet. Hybserv2 acts as another server on the network as I described. Hybrid7 doesn't have support for the service related commands by default, and may not have them at all. > > The alternative is that the service registers itself > as "service" using > the SERVICE command as defined by RFC 2812, section > 3.1.6. This variant > isn't implemented by the ngIRCd at the moment. I've never actually seen a service implemented this way, although I do remember seeing this section in the RFC's while I was learning the protocol to write my own IRC bot. I just looked in to this and it looks like it wouldn't take too much to add this kind of functionality to ngIRCd. I looked at the client.c file, and I know another wrapper needs to be added to Client_New() called Client_NewRemoteService(). I'll dig deeper and see if I can't implement this. I'd rather find an IRC server that has this working already to test how the protocol is suppose to work, but *something* is better than *nothing* :) and I'm sure people will correct anything I do wrong if I get this working and submit the appropriate patches. By the way, I compiled this on a Tru64 5.1b box as well without problems. Can't remember if it's in the list of reported "working" or not, but if it's not, now it can be added. Thanks again, Stefan Johnson __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/

1 0

IRC Services
by Stefan Johnson 19 Mär '05

19 Mär '05

Hi, I'm new to the list. I run hybrid7 with hybserv2 services on my linux server at home. I recently acquired an HP C3000 workstation (PA-RISC) with HP-UX 11i installed. I installed the GNU C compiler suite, and tried to install hybrid7 to no avail. It seems the only time hybrid7 has been successfully compiled on HP-UX is via HP's ansi-c compiler ($1500.00 or so price tag for that.) I got to looking around and discovered this little gem of a server. I downloaded it, compiled it (no problems what-so-ever) and installed it without a single hitch. I even went in and found the correct line in the defines.h file to modify the nick length quickly because the code is clean and self documenting. KUDOS. I was browsing the mailing list archive and discovered that I'm not the only one interested in IRC services. The last discussion I remember reading on it is from January of this year. It seems noone was quite sure how the services actually work. I'd like to introduce 2 concepts of how services are suppose to work (from my experience... guesswork, but the end result would be the same) plus I'd like to give an idea for "faking" services using something as simple as an eggdrop irc bot. It's my understanding that services work by connecting to the IRC servers they service as "just another IRC server." The servers don't know anything about the services server other than it's another IRC server on the network. The difference is this... The services "server" "cheats" by sending nick change events for users, disconnecting users, and such depending on what commands are given by a user who has authenticated with those services. In order to write the services, someone might consider starting with ngircd and cutting out the "user can connect" portions, and modifying it to keep an authentication database for users, channels, services ops, and memos (if a memo service is desired.) I'd be happy to help whomever is wanting to write the services, but my time is somewhat limited, so if I end up writing them alone, it will take quite a while to complete. The pseudo-services idea I have is a little more "dangerous" and requires a thorough understanding of how to set up and manage an eggdrop or other irc bot. The irc bot should be running on a server that only trusted users have access to. The bot should be set up to NOT allow "bot op" connections, telnet connections from "any ip," or otherwise present any way for arbitrary users to connect to it and give it commands. The bot should be stripped of non-services scripts. The last part (and this is WHY the tight security) is that the bot should be able to (and must) have OPER privileges on the irc server it connects to. Why? The setting that allows IRC operators to make themselves CHANNEL operators makes this a god-send for "quick and dirty" services-via-bot setups. Scripts can be written to handel "chanserve" functionality with little effort. Modifications to ngircd would need to be made to allow IRC Operators the ability to change other users' nicknames on the fly for NickServe functionality to exist, but this may (or may not) be easily accomplished. Anyway, these are the ideas I've had, and hopefully they'll help someone out. Comments, flames, and other verbage are welcome :) Stefan __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/

2 1

Fwd: ngIRCd <= v0.8.2 Format String Vulnerability
by Alexander Barton 04 Feb '05

04 Feb '05

For your information. Regards Alex > Von: CoKi <coki(a)nosystem.com.ar> > Datum: 3. Februar 2005 3:09:09 Uhr MEZ > An: bugtraq(a)securityfocus.com > Betreff: ngIRCd <= v0.8.2 Format String Vulnerability > X-Mailer: MIME-tools 5.411 (Entity 5.404) > > > > ------------------------------------------------- > No System Group - Advisory #11 - 03/02/05 > ------------------------------------------------- > Program: ngIRCd > Homepage: http://arthur.ath.cx/~alex/ngircd/ > Operating System: Linux and Unix-Compatible > Vulnerable Versions: ngIRCd v0.8.2 and prior > Risk: High > Impact: Remote Format String Vulnerability > ------------------------------------------------- > > > - DESCRIPTION > ------------------------------------------------- > ngIRCd is a portable IRC daemon written from scratch. > It is easy to configure, supports server links (even > with original ircds) and runs on hosts with changing > IP addresses (such as dial-in networks). Currently > supported platforms are AIX, A/UX, Darwin/Mac OS X, > FreeBSD, HP-UX, IRIX, Linux, NetBSD, SunOS/Solaris, > and Windows with Cygwin. > > More informations at: http://arthur.ath.cx/~alex/ngircd/ > > > - DETAILS > ------------------------------------------------- > Exist a format string bug in the Log_Resolver() function > of log.c when parses erroneous arguments to the syslog() > function. This may give a remote shell with privileges of > ngIRCd, it's 'root' by default. > > ----------- log.c ------------ > 261: vsnprintf( msg, MAX_LOG_MSG_LEN, Format, ap ); > 262: va_end( ap ); > 263: > 264: /* Output */ > 265: if( NGIRCd_NoDaemon ) > 266: { > 267: /* Output to console */ > 268: fprintf( stdout, "[%d:%d] %s\n", (INT)getpid( ), > Level, msg ); > 269: fflush( stdout ); > 270: } > 271: #ifdef SYSLOG > 272: else syslog( Level, msg ); > 273: #endif > ----------- log.c ------------ > > Here we can look that exist a format string vulnerability > in 272 line of log.c > > > - EXPLOIT > ------------------------------------------------- > To obtain a successful exploitation, we need that > ngIRCd has been compiled with IDENT, logging to > SYSLOG and DEBUG enabled. > > We now will look a part of the code of resolve.c: > > ----------- resolve.c ------------ > 96: /* For sub-process */ > 97: pid = fork( ); > 98: if( pid > 0 ) > 99: { > 100: /* Main process */ > 101: Log( LOG_DEBUG, "Resolver for %s created (PID > %d).", inet_ntoa( Addr->sin_addr ), pid ); > 102: FD_SET( s->pipe[0], &Resolver_FDs ); > 103: if( s->pipe[0] > Conn_MaxFD ) Conn_MaxFD = > s->pipe[0]; > 104: s->pid = pid; > 105: s->stage = 0; > 106: s->bufpos = 0; > 107: return s; > 108: } > > [...] > > 230: #ifdef IDENTAUTH > 231: /* Do "IDENT" (aka "AUTH") lookup and write result to > parent */ > 232: Log_Resolver( LOG_DEBUG, "Doing IDENT lookup on socket %d > ...", Sock ); > 233: res = ident_id( Sock, 10 ); > 234: Log_Resolver( LOG_DEBUG, "Ok, IDENT lookup on socket %d > done: \"%s\"", Sock, res ? res : "" ); > 235: > 236: /* Write IDENT result into pipe to parent */ > 237: len = strlen( res ? res : "" ); > 238: if( res != NULL ) res[len] = '\n'; > 239: len++; > 240: if( (size_t)write( w_fd, res ? res : "\n", len ) != > (size_t)len ) > 241: { > 242: Log_Resolver( LOG_CRIT, "Resolver: Can't write to > parent (IDENT): %s!", strerror( errno )); > 243: close( w_fd ); > 244: } > 245: free( res ); > 246: #endif > ----------- resolve.c ------------ > > At 97 lines of resolve.c we can see that is > created a new process with fork() function, > this will allow us to obtain the exact address > of RET of automatic way without ircd dies. > We can see too that this process do IDENT lookup > when a host is connected to the ircd daemon, > and using the vulnerable function for logging > with syslog at 242 lines of this code. > > I have written a code exploit that use this > method for obtain a root shell in the target. > > http://www.nosystem.com.ar/exploits/ngircd_fsexp.c > > root@servidor:/home/coki/audit# ./ngircd_fsexp -h victim -t 1 -o 10000 > > ngIRCd <= 0.8.2 remote format string exploit > by CoKi <coki(a)nosystem.com.ar> > > [*] host : victim > [*] system : Slackware Linux 10.0 > [*] ircd version : ngircd-0.8.2.tar.gz > [*] syslog GOT address : 0x08068094 > [*] verifying host : 10.0.0.2 > > [*] trying RET address : 0x0806d710 (offset 10000) > [*] building evil buffer : done! > [*] running fake ident server : 0.0.0.0:113 > > [*] connecting to ircd... : 10.0.0.2:6667 connected > [*] waiting for answer... : 10.0.0.1:43260 connected > [*] sending evil ident... : done! > [*] checking for shell... : done! > > [!] you have a shell :) > > Linux victim 2.4.26 #29 Mon Jun 14 19:22:30 PDT 2004 i686 unknown > unknown GNU/Linux > uid=0(root) gid=0(root) > groups=0(root),1(bin),2(daemon),3(sys),4(adm),6(disk),10(wheel),11(flop > py) > > > - SOLUTIONS > ------------------------------------------------- > The patch is included here: > > --- src/ngircd/log.c 2004-06-26 06:06:27.000000000 -0300 > +++ src/ngircd/log.c.patch 2005-02-02 12:53:33.000000000 -0300 > @@ -269,7 +269,7 @@ > fflush( stdout ); > } > #ifdef SYSLOG > - else syslog( Level, msg ); > + else syslog( Level, "%s", msg ); > #endif > } /* Log_Resolver */ > > > - REFERENCES > ------------------------------------------------- > http://www.nosystem.com.ar/advisories/advisory-11.txt > > > - CREDITS > ------------------------------------------------- > Discovered by CoKi <coki(a)nosystem.com.ar> > > No System Group - http://www.nosystem.com.ar -- Alexander Barton, Freiburg, Germany http://www.barton.de/, alex(a)barton.de

1 0

ngIRCd 0.8.2
by Alexander Barton 03 Feb '05

03 Feb '05

Hi All! ngIRCd 0.8.2, the "hey there are contributors" release has just hit the ground :-) Many thanks go to Florian Westphal for his work! Everybody running an older version of ngIRCd should upgrade to this version because some remotely exploitable security related bugs (that could cause the daemon to crash) have been fixed! Note: if you are using the CVS HEAD branch then you should run an CVS update and use the current code base. Changes since version 0.8.1 are: - Added doc/SSL.txt to distribution. - Fixed a buffer overflow that could cause the daemon to crash. Bug found by Florian Westphal, <westphal(a)foo.fh-furtwangen.de>. - Fixed a possible buffer underrun when reading the MOTD file. Thanks to Florian Westphal, <westphal(a)foo.fh-furtwangen.de>. - Fixed detection of IRC lines which are too long to send. Detected by Florian Westphal, <westphal(a)foo.fh-furtwangen.de>. - Fixed return values of our own implementation of strlcpy(). The code has been taken from rsync and they fixed it, but we didn't until today :-/ It has only been used when the system didn't implement strlcpy by itself, not on "modern" systems. Florian Westphal, <westphal(a)foo.fh-furtwangen.de>. You can download ngIRCd 0.8.2 (~271 KB) from: - <http://download.berlios.de/ngircd/ngircd-0.8.2.tar.gz> - <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.8.2.tar.gz> - <ftp://Arthur.Ath.CX/pub/Users/alex/ngircd/ngircd-0.8.2.tar.gz> And the patch from 0.8.1 to 0.8.2 (~3 KB) as well as GnuPG signatures can be found here: - <ftp://ftp.berlios.de/pub/ngircd/> - <ftp://Arthur.Ath.CX/pub/Users/alex/ngircd/> This release has been tagged as "rel-0-8-2" in the CVS. Regards Alex -- Alexander Barton, Freiburg, Germany http://www.barton.de/, alex(a)barton.de

2 7

ngIRCd 0.8.3
by Alexander Barton 03 Feb '05

03 Feb '05

Hi All! An other exploitable bug in ngIRCd has been found, so here is jet an other update: ngIRCd 0.8.3 has been released! The bug is only exploitable when the daemon is compiled to do IDENT lookups, which is not the default configuration. If you don't use IDENT lookups with your ngIRCd, you are safe. All other should update as soon as possible as this bugs allows remote attackers to execute arbitrary code with the priviledges of the ngIRCd. The only change since version 0.8.2 is: - Fixed a bug that could case a root exploit when the daemon is compiled to do IDENT lookups and is logging to syslog. Bug discovered by CoKi, <coki(a)nosystem.com.ar>, thanks a lot! (http://www.nosystem.com.ar/advisories/advisory-11.txt) You can download ngIRCd 0.8.3 (~271 KB) from: - <ftp://ftp.berlios.de/pub/ngircd/ngircd-0.8.3.tar.gz> - <ftp://Arthur.Ath.CX/pub/Users/alex/ngircd/ngircd-0.8.3.tar.gz> - <http://download.berlios.de/ngircd/ngircd-0.8.3.tar.gz> [soon ...] And the patch from 0.8.2 to 0.8.3 (~3 KB) as well as GnuPG signatures can be found here: - <ftp://ftp.berlios.de/pub/ngircd/> - <ftp://Arthur.Ath.CX/pub/Users/alex/ngircd/> This release has been tagged as "rel-0-8-3" in the CVS. Regards Alex -- Alexander Barton, Freiburg, Germany alex(a)barton.de, http://www.barton.de/

1 0

Fwd: [ GLSA 200501-40 ] ngIRCd: Buffer overflow
by Alexander Barton 29 Jan '05

29 Jan '05

For your information. But now the bug is published, so all servers should really be updated as fast as possible. Regards Alex > Von: Thierry Carrez <koon(a)gentoo.org> > Datum: 28. Januar 2005 23:07:30 Uhr MEZ > An: gentoo-announce(a)lists.gentoo.org > Cc: bugtraq(a)securityfocus.com, full-disclosure(a)lists.netsys.com, > security-alerts(a)linuxsecurity.com > Betreff: [ GLSA 200501-40 ] ngIRCd: Buffer overflow > User-Agent: Mozilla Thunderbird 1.0 (X11/20041208) > > - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - > Gentoo Linux Security Advisory GLSA 200501-40 > - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - > http://security.gentoo.org/ > - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - > > Severity: High > Title: ngIRCd: Buffer overflow > Date: January 28, 2005 > Bugs: #79705 > ID: 200501-40 > > - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - > > Synopsis > ======== > > ngIRCd is vulnerable to a buffer overflow that can be used to crash the > daemon and possibly execute arbitrary code. > > Background > ========== > > ngIRCd is a free open source daemon for Internet Relay Chat (IRC). > > Affected packages > ================= > > ------------------------------------------------------------------- > Package / Vulnerable / Unaffected > ------------------------------------------------------------------- > 1 net-irc/ngircd < 0.8.2 >= 0.8.2 > > Description > =========== > > Florian Westphal discovered a buffer overflow caused by an integer > underflow in the Lists_MakeMask() function of lists.c. > > Impact > ====== > > A remote attacker can exploit this buffer overflow to crash the ngIRCd > daemon and possibly execute arbitrary code with the rights of the > ngIRCd daemon process. > > Workaround > ========== > > There is no known workaround at this time. > > Resolution > ========== > > All ngIRCd users should upgrade to the latest version: > > # emerge --sync > # emerge --ask --oneshot --verbose ">=net-irc/ngIRCd-0.8.2" > > References > ========== > > [ 1 ] ngIRCd Release Annoucement > > http://arthur.ath.cx/pipermail/ngircd-ml/2005-January/000228.html > > Availability > ============ > > This GLSA and any updates to it are available for viewing at > the Gentoo Security Website: > > http://security.gentoo.org/glsa/glsa-200501-40.xml > > Concerns? > ========= > > Security is a primary focus of Gentoo Linux and ensuring the > confidentiality and security of our users machines is of utmost > importance to us. Any security concerns should be addressed to > security(a)gentoo.org or alternatively, you may file a bug at > http://bugs.gentoo.org. > > License > ======= > > Copyright 2005 Gentoo Foundation, Inc; referenced text > belongs to its owner(s). > > The contents of this document are licensed under the > Creative Commons - Attribution / Share Alike license. > > http://creativecommons.org/licenses/by-sa/2.0 -- Alexander Barton, Freiburg, Germany http://www.barton.de/, alex(a)barton.de

1 0

Custom Features?
by Josh Coombs 22 Jan '05

22 Jan '05

Got a couple oddball feature requests I'd like to bounce off you... 1 - External Auth Mechanism Currently, ngircd supports a single global server password option. I'd like to be able to do specific user/pass pairs on connect. Either setting them up in the conf, an external flat file, or the option to call an external script, pass it the user/pass pair, and allow the connect if it returns without error (allowing radius/ldap/etc based auth without having to support it directly in ngricd) would be slick. 2 - Toggle to restrict use of /nick and channel creation to those with +o. By itself this doesn't do much, but in combination with the above, this allows much more control of the ircd's use, making it appropriate to use in an internal LAN/buisiness environment. One last question, have the services provisions been fleshed out enough to warrant trying third party service apps, or are they still too rough? Joshua Coombs

2 2

[Patch] +s (secret) channel mode
by Florian Westphal 22 Jan '05

22 Jan '05

Hi, this simple patch adds support for secret '+s' channels to ngircd. A "secret" channel does not show up in /LIST output (unless user is on that channel) Florian

2 1

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

ngIRCd Users