From cvs at videolan.org Sun Apr 1 02:42:48 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sun, 1 Apr 2001 02:42:48 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010401004248.23F99233DD@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: Makefile logger.c main.c types.c types.h vlanserver.h Update of /cvs/videolan/network/VLANserver/config Modified Files: config.fl config.h Update of /cvs/videolan/network/VLANserver/interface Modified Files: interface.c interface.h telnet.c telnet.h Update of /cvs/videolan/network/VLANserver/server Modified Files: poll_switch.c reinit_switchs.c server.c server.h Update of /cvs/videolan/network/VLANserver/snmp Modified Files: callback.c if_snmp.c if_snmp.h snmp.c snmp.h snmp_switch.h Log Message: . Don't print logs with LOGDEBUG level when compiling without the -DDEBUG option . Wait the end of the snmp init before starting sending requests to it . New function inside telnetd: reload forces to flush the database . Support for matrixed switched config (need to enter the unit) default is 1 snmp changed in port-vlan and port-mac added a 3COM specific kludge -- nitrox From cvs at videolan.org Sun Apr 1 15:37:01 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sun, 1 Apr 2001 15:37:01 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010401133701.B8196233E4@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: logger.c logger.h types.c types.h Update of /cvs/videolan/network/VLANserver/interface Modified Files: telnet.c Log Message: . Few changes in telnet . Now we are able to log inside a log file -- nitrox From cvs at videolan.org Sun Apr 1 19:15:57 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sun, 1 Apr 2001 19:15:57 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010401171557.6E2D2233E4@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: logger.c Log Message: . Corrected a bug when logging inside a file. -- nitrox From cvs at videolan.org Mon Apr 2 17:13:20 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Mon, 2 Apr 2001 17:13:20 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010402151320.DA39C233A5@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: Makefile logger.c Update of /cvs/videolan/network/VLANserver/old-tools/VLANclient Modified Files: Makefile Update of /cvs/videolan/network/VLANserver/server Modified Files: poll_switch.c poll_switch.h reinit_switchs.c request_handler.c server.c server.h Update of /cvs/videolan/network/VLANserver/snmp Modified Files: if_snmp.c Log Message: . Changed the way switches are polled during initialization . Corrected a bug in logger.c . Cosmetic changes in snmp -- nitrox From cvs at videolan.org Wed Apr 4 01:13:45 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Wed, 4 Apr 2001 01:13:45 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010403231345.4970F233A5@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: Makefile Added Files: sleep.c sleep.h Update of /cvs/videolan/network/VLANserver/db Modified Files: db.c Update of /cvs/videolan/network/VLANserver/interface Modified Files: telnet.c Update of /cvs/videolan/network/VLANserver/server Modified Files: client.c connect.c poll_switch.c reinit_switchs.c request_handler.c server.c Log Message: * added a VS_sleep function in sleep.c, use it to replace all sleeps so that GDB still works * fatal bug correction in the database, the memory leak is back but will be removed soon * removed the DB_switchs_lock and DB_switchs_unlock function which weren't useful any more * cosmetic changes -- bbp From cvs at videolan.org Wed Apr 4 12:23:25 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Wed, 4 Apr 2001 12:23:25 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010404102325.624FB233A5@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver/db Modified Files: db.h Log Message: suppression of two useless an no longer existent functions prototypes in db.h -- bbp From cvs at videolan.org Thu Apr 5 01:54:18 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Thu, 5 Apr 2001 01:54:18 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010404235418.75798233A5@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver/db Modified Files: db.c Log Message: Removed a useless function in db.c, removed the memory leak. This is untested code because gunther uses a ~?@]^[~?]@[{^ umask and I can't read his cool vlanserver.conf :) -- bbp From cvs at videolan.org Thu Apr 5 08:42:52 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Thu, 5 Apr 2001 08:42:52 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010405064252.3BD5B233A5@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver/server Modified Files: client.c reinit_switchs.c request_handler.c server.c Update of /cvs/videolan/network/VLANserver/snmp Modified Files: if_snmp.c Log Message: - Modify the client and the server : the server communicate good with the client. It send a short message if the server did not succed to change the channel. - cosmetic modification in snmp : replace a LOGFATAL with a LOGERROR for the snmp traps -- gunther From nitrox at via.ecp.fr Thu Apr 5 13:18:27 2001 From: nitrox at via.ecp.fr (Damien Lucas) Date: Thu, 5 Apr 2001 13:18:27 +0200 Subject: Resultats de ma recherche de la (les ?) segfaults ... In-Reply-To: <20010405104415.A29676@via.ecp.fr>; from gunther@via.ecp.fr on Thu, Apr 05, 2001 at 10:44:15AM +0200 References: <20010405104415.A29676@via.ecp.fr> Message-ID: <20010405131827.B812@via.ecp.fr> On Thu, Apr 05, 2001, Laurent Rossier wrote: > Bon je ne l'ai pas trouve (evidemment dirait nitrox !) mais j'ai trouve > des trucs interressants qui pourraient vous aider (on ne sait jamais des > fois que je serve a qqchose ...) Voici les logs de gdb qd ca segfault > (en voici deux differents mais je pense que c'est la meme segfault non ? > ...) : T mortel ! > * Program received signal SIGSEGV, Segmentation fault. > * SNMP_is_inited (zwitch=0x7972746e) at snmp/snmp.c:458 > * 458 if(z->machine>0xffffffffffff) En fait, tu pourrais faire un 'print zwitch' dans gdb (? condition de compiler sans optim. Parce que pour moi, la seg fault vient de zwitch. A ce moment gdb te donne un dump hex de la memeoire ? ce point ou bien un 'cannot access memory' > * Program received signal SIGSEGV, Segmentation fault. > * 0x804d611 in SNMP_SW_callback (op=0, session=0x0, reqid=0, pdu=0x0, > * zwitch=0x0) at snmp/callback.c:257 > * 257 (*zwitch->walk_macs.machine)->machine=(VS_MachineId)mac; La c'est tr?s clair, le pointeur zwitch est le pointeur NULL (0x0) donc ca risue de segfaulter. C'est cette erreur qui est la plus embettante. si pdu vaut NULL, ca veut dire que entre le moment ou le pacquet est arrvi? et le snmp_read dans if_snmp.c on doit pourrir la m?moire. > Il se chie dessus dans SNMP_is_inited qd il fait le test avec la mac car > la mac doit etre fausse - ne doit pas etre au bon format - et donc a la > ligne 458 qd il la compare avec 0xffffffffffff et ben ca segfault ... > Sinon le pb viendrait aussi de la mac mais en amont dans la fonction > Callback_Macs() ou il remplit la table des macs qui serait > completement fausse. > M'enfin c'est ce que je pense ... Je pense que c'est la struct zwitch toute enti?re qui est en question. > Puis en affichant les macs qu'il voit j'ai trouve une ligne super > cheloud : > > 05/04 10:22:09 [SNMP : debug ] mac in callback 102afc3dc > 05/04 10:22:09 [SNMP : debug ] mac in callback 102dc0465 > 05/04 10:22:09 [SNMP : debug ] mac in callback b4b7 > 05/04 10:22:09 [SNMP : debug ] mac in callback 50044cb079 > 05/04 10:22:09 [SNMP : debug ] mac in callback 50044cacdc A v?rifer: Ou bien ce sont les switches qui sont capables de nbous envoyer des conneries. Ou bien ces l'interpr?tation du pacquet qui chie Ou bien on a pourri la m?moire entre les deux. (Tu as affich? quoi ? le local unisgned int macs ou la variable qu'on chope genre zwitch->.... ? ) EN tout cas avec ce que tu asd trouv? on va pouvoir avanc? plus vite. -- nitrox From gunther at via.ecp.fr Thu Apr 5 15:23:28 2001 From: gunther at via.ecp.fr (Laurent Rossier) Date: Thu, 5 Apr 2001 15:23:28 +0200 Subject: Resultats de ma recherche de la (les ?) segfaults ... In-Reply-To: <20010405131827.B812@via.ecp.fr>; from nitrox@via.ecp.fr on Thu, Apr 05, 2001 at 01:18:27PM +0200 References: <20010405104415.A29676@via.ecp.fr> <20010405131827.B812@via.ecp.fr> Message-ID: <20010405152328.A4559@via.ecp.fr> On Thu, Apr 05, 2001, Damien Lucas wrote: > A v?rifer: > Ou bien ce sont les switches qui sont capables de nbous envoyer des conneries. Je ne pense pas que ca vienne des switchs car arparp fait des requetes snmp il me semble t il donne les bonnes macs. > Ou bien ces l'interpr?tation du pacquet qui chie > Ou bien on a pourri la m?moire entre les deux. > (Tu as affich? quoi ? le local unisgned int macs > ou la variable qu'on chope genre zwitch->.... ? ) Oue j'ai affiche mac qui est un unsigned long long int et qui apres sert a faire : (*zwitch->walk_macs.machine)->machine=(VS_MachineId)mac; -- Gunther From cvs at videolan.org Sat Apr 14 01:20:14 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sat, 14 Apr 2001 01:20:14 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010413232014.A8A7B233E4@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: Makefile Update of /cvs/videolan/network/VLANserver/interface Modified Files: interface.c interface.h telnet.c Update of /cvs/videolan/network/VLANserver/server Modified Files: request_handler.c Log Message: . added lool's code (function to info about a mac by the interface) lool: please have a look, I changed few things. . added a test to knnow if mac is alone on the port or not This test is actually only in interface part I will soon include it inside server's part to bridge ALL macs . changed a type definition in server/request_handler long long int instead of VS_MachineId (but this should not be reason of the segfault) -- nitrox From nitrox at via.ecp.fr Sun Apr 15 23:50:30 2001 From: nitrox at via.ecp.fr (Damien Lucas) Date: Sun, 15 Apr 2001 23:50:30 +0200 Subject: [VLANSERVER] Developping Guidelines Message-ID: <20010415235030.A17741@via.ecp.fr> For those who weren't subscribed when it was posted. -- nitrox ----- Forwarded message from Damien Lucas ----- From: Damien Lucas Date: Fri, 15 Dec 2000 02:58:53 +0100 To: videolan_devel at via.ecp.fr Subject: [VLANSERVER] Developping Guidelines [out of date guidelines] 3- CODING CONVENTIONS --------------------- . 2 spaces for each indentation . variables should be declared like this: char * string; struct SNMP_switch * zwitch; . no p_ or i_ or ps_ ... in variables' name . begin declarations by bigger datas: in this order: long long, long, int, short, pointer, char . don't cast calls to "malloc". . no space between the name of the function and the ) : function(); . one space after each , : function(arg1, arg2,, arg3); . never 2 blank lines. . comments inside code with /* comments */ and as few as we can. . comments before beginning of each function: /************************************************** * FUNCTION * ************************************************** * Synopsis * **************************************************/ 80 char lines . KLUDGE parts should be identified with: { /* KLUDGE */ ... } . No variable definitions in the middle of a function. I wish I were clear and did not forget anything. -- 'trox ----- End forwarded message ----- From nitrox at via.ecp.fr Mon Apr 16 02:11:07 2001 From: nitrox at via.ecp.fr (Damien Lucas) Date: Mon, 16 Apr 2001 02:11:07 +0200 Subject: Questions Message-ID: <20010416021106.A29567@via.ecp.fr> I was working on bridging all the macs known on a port (ie when you got a hub, you 'd like all you're computers to be bridged when you ask for vlanchanging). There is no problem to know all the macs (see telnet/info) BUT we need the IPs. 4 solutions: 1- The client sends the IP list 2- We make a reverse ARP 3- We ask to the VLANbridge 4- We ask to Webbase These are the only 4 solutions I've seen. Do you have another one ? The problems are: 1- What about the security (I can ask for bridging anyone ...) ? 2- What if the macs are not in the same VLAN as VLANserver ? What do you think of that ? -- nitrox From benny at via.ecp.fr Mon Apr 16 03:02:46 2001 From: benny at via.ecp.fr (Benoit Steiner) Date: Sun, 15 Apr 2001 18:02:46 -0700 Subject: Questions References: <20010416021106.A29567@via.ecp.fr> Message-ID: <3ADA44B6.9B0CB0C4@via.ecp.fr> 1/ Very good to test the robustness of the vlanserver since a user will sooner or later misconfigure its vlc ;) And also very good to train the diffs to diagnose quickly a problem ;) 2/ Why not ? Seems robust and reliable to me. 3/ Means that the vlanserver depends on the vlanbridge to be able to work. Until now, there has been no such strong dependency. 4/ Even worse than 1, because the database is full of mistakes. About question 1, who cares anyway :) Seriously a user should only make requests for its IP and then the server determines if there are other machine to move. About question 2, the only solution that works is to let the vlanbridge do the reverse ARP. But I don't like it. It's up to you. The biggest problem here is that when a computer boots the vlanbridge assumes it's on vlan2. So what happens if this computer is connected to a hub that is on vlan4 ? There are a lot of problems like this. We should consider that a port on a switch is used by a group of machines on the same vlan, not a single one. But then we have no easy way to automaticly and dynamicly populate such groups. As a result, supporting hubs is a real pain. We should really require everybody to buy a switch: They're no more that expensive ;) Benny Damien Lucas wrote: > > I was working on bridging all the macs known on a port > (ie when you got a hub, you 'd like all you're computers to be bridged > when you ask for vlanchanging). > > There is no problem to know all the macs (see telnet/info) > BUT we need the IPs. > > 4 solutions: > > 1- The client sends the IP list > 2- We make a reverse ARP > 3- We ask to the VLANbridge > 4- We ask to Webbase > > These are the only 4 solutions I've seen. Do you have another one ? > > The problems are: > 1- What about the security (I can ask for bridging anyone ...) ? > 2- What if the macs are not in the same VLAN as VLANserver ? > > What do you think of that ? > > -- > nitrox From nitrox at via.ecp.fr Mon Apr 16 03:45:46 2001 From: nitrox at via.ecp.fr (Damien Lucas) Date: Mon, 16 Apr 2001 03:45:46 +0200 Subject: Questions In-Reply-To: <3ADA44B6.9B0CB0C4@via.ecp.fr>; from benny@via.ecp.fr on Sun, Apr 15, 2001 at 06:02:46PM -0700 References: <20010416021106.A29567@via.ecp.fr> <3ADA44B6.9B0CB0C4@via.ecp.fr> Message-ID: <20010416034546.B29567@via.ecp.fr> On Sun, Apr 15, 2001, Benoit Steiner wrote: > About question 2, the only solution that works is to let the vlanbridge > do the reverse ARP. But I don't like it. It's up to you. What about having an ATM card in the VLANserver so that VLANserver would be able to do the reverse ARP on its own ? > The biggest problem here is that when a computer boots the vlanbridge > assumes it's on vlan2. If the computer is alone we got a SNMP trap link-up. Then we should ask for the switch the mac and vlan and then ask the VLANbridge to bridge. BUT if the link was active when the computer boots (ie there's a hub) I don't think there is a method to be informed, is there ? The solution is to flush the database and poll all the switch regularly but now, polling the switches takes up to 4 minutes. If we have to do it every 30min, it makes 15% downtime ... (we can't send snmp request while polling the switches !). -- nitrox From cvs at videolan.org Mon Apr 16 04:12:19 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Mon, 16 Apr 2001 04:12:19 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010416021219.9B29E23400@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: TODO logger.c logger.h types.c Update of /cvs/videolan/network/VLANserver/config Modified Files: config.fl config.h Update of /cvs/videolan/network/VLANserver/doc Modified Files: vlanserver.conf.sample Update of /cvs/videolan/network/VLANserver/interface Modified Files: interface.c telnet.c Update of /cvs/videolan/network/VLANserver/server Modified Files: client.c request_handler.c Log Message: 1. Added return values of sscanf in order to prevent from bad messages and added few comments (request_handler.c) 2. In section doc, wrote about the way of using . unit (cf support of matrixed switches) . logfile, logmethod (see below) 3. Changed the LogMethod: . logmethod can be NONE, FILE_ONLY, SCREEN_ONLY, BOTH . we are now able to change the logmethod in runtime . At the beginning we use SCREEN_ONLY . After parsing the config file, we take cfg->logmethod into account . Todo: do the same thing with loglevels -- nitrox From bbp at via.ecp.fr Tue Apr 17 11:33:20 2001 From: bbp at via.ecp.fr (Brieuc Jeunhomme) Date: Tue, 17 Apr 2001 11:33:20 +0200 Subject: =?iso-8859-1?Q?probl=E8me_de_m=E9moire_lors_de_la_r=E9initialisation?= Message-ID: <20010417113320.B16807@via.ecp.fr> Gunther avait rep?r? un probl?me de m?moire (les malloc ne fonctionnaient plus) lors de la r?initialisation de la base, alors que d'apr?s top, le programme ne consommait pas ?norm?ment de m?moire. Hypoth?se sur la cause : 1) on faisait un setrlimit pour limiter la consommation de m?moire a pas beaucoup 2) comme les timeout SNMP sont par d?faut de -1 seconde (ie 4 milliards et des patates soit environ 30-40 ans), ben toutes les requ?tes auxquelles les switchs ne repondaient pas bouffaient de la memoire En combinant ces deux points, on obtient peut-?tre l'explication du pourquoi du comment du probl?me. Voil? voil?. -- BBP From massiot at via.ecp.fr Tue Apr 17 11:38:37 2001 From: massiot at via.ecp.fr (Christophe Massiot) Date: Tue, 17 Apr 2001 11:38:37 +0200 Subject: =?iso-8859-1?Q?probl=E8me_de_m=E9moire_lors_de_la_r=E9initialisation?= In-Reply-To: <20010417113320.B16807@via.ecp.fr>; from bbp@via.ecp.fr on Tue, Apr 17, 2001 at 11:33:20AM +0200 References: <20010417113320.B16807@via.ecp.fr> Message-ID: <20010417113837.A16035@via.ecp.fr> On Tue, Apr 17, 2001, Brieuc Jeunhomme wrote: > Hypoth?se sur la cause : > 1) on faisait un setrlimit pour limiter la consommation de > m?moire a pas beaucoup > 2) comme les timeout SNMP sont par d?faut de -1 seconde (ie 4 > milliards et des patates soit environ 30-40 ans), ben toutes > les requ?tes auxquelles les switchs ne repondaient pas > bouffaient de la memoire 3) vous avez ?cras? les drapeaux malloc du tas (corruption du tas) en ?crivant ? un endroit non-allou? et vous ?tes profond?ment dans la merde. Remarquez ?a pourrait ?tre pire, vous pourriez d?molir la pile aussi. -- Christophe Massiot. From sam at zoy.org Tue Apr 17 12:09:26 2001 From: sam at zoy.org (Samuel Hocevar) Date: Tue, 17 Apr 2001 12:09:26 +0200 Subject: =?iso-8859-1?Q?probl=E8me_de_m=E9moire_lors_de_la_r=E9initialisation?= In-Reply-To: <20010417113320.B16807@via.ecp.fr>; from bbp@via.ecp.fr on Tue, Apr 17, 2001 at 11:33:20AM +0200 References: <20010417113320.B16807@via.ecp.fr> Message-ID: <20010417120925.M27642@zoy.org> On Tue, Apr 17, 2001, Brieuc Jeunhomme wrote: > Gunther avait rep?r? un probl?me de m?moire (les malloc ne > fonctionnaient plus) Sous Linux malloc fonctionne toujours, m?me avec des setrlimit partout. S'il renvoie NULL c'est que le tas est corrompu. -- Sam. From nitrox at via.ecp.fr Tue Apr 17 13:39:04 2001 From: nitrox at via.ecp.fr (Damien Lucas) Date: Tue, 17 Apr 2001 13:39:04 +0200 Subject: =?iso-8859-1?Q?probl=E8me_de_m=E9moire_lors_de_la_r=E9initialisation?= In-Reply-To: <20010417120925.M27642@zoy.org>; from sam@zoy.org on Tue, Apr 17, 2001 at 12:09:26PM +0200 References: <20010417113320.B16807@via.ecp.fr> <20010417120925.M27642@zoy.org> Message-ID: <20010417133904.A26088@via.ecp.fr> On Tue, Apr 17, 2001, Samuel Hocevar wrote: > S'il renvoie NULL c'est que le tas est corrompu. Question, quand on utilise MALLOC_CHECK, qu'est-ce que ca ait fexactement ? -- nitrox From massiot at via.ecp.fr Tue Apr 17 13:43:35 2001 From: massiot at via.ecp.fr (Christophe Massiot) Date: Tue, 17 Apr 2001 13:43:35 +0200 Subject: =?iso-8859-1?Q?probl=E8me_de_m=E9moire_lors_de_la_r=E9initialisation?= In-Reply-To: <20010417133904.A26088@via.ecp.fr>; from nitrox@via.ecp.fr on Tue, Apr 17, 2001 at 01:39:04PM +0200 References: <20010417113320.B16807@via.ecp.fr> <20010417120925.M27642@zoy.org> <20010417133904.A26088@via.ecp.fr> Message-ID: <20010417134335.A29704@via.ecp.fr> On Tue, Apr 17, 2001, Damien Lucas wrote: > > S'il renvoie NULL c'est que le tas est corrompu. > > Question, quand on utilise MALLOC_CHECK, > qu'est-ce que ca ait fexactement ? C'est MALLOC_CHECK_ et man malloc. Ce que tu cherches est man efence. -- Christophe Massiot. From sam at zoy.org Tue Apr 17 13:54:00 2001 From: sam at zoy.org (Samuel Hocevar) Date: Tue, 17 Apr 2001 13:54:00 +0200 Subject: =?iso-8859-1?Q?probl=E8me_de_m=E9moire_lors_de_la_r=E9initialisation?= In-Reply-To: <20010417133904.A26088@via.ecp.fr>; from nitrox@via.ecp.fr on Tue, Apr 17, 2001 at 01:39:04PM +0200 References: <20010417113320.B16807@via.ecp.fr> <20010417120925.M27642@zoy.org> <20010417133904.A26088@via.ecp.fr> Message-ID: <20010417135400.Q27642@zoy.org> On Tue, Apr 17, 2001, Damien Lucas wrote: > Question, quand on utilise MALLOC_CHECK, > qu'est-ce que ca ait fexactement ? Je ne sais pas fexactement ce que ?a ait, mais je sais qu'au moins il v?rifie l'int?grit? de la zone de donn?es qu'on essaye d'allouer ou de d?sallouer, en regardant les octets qui sont un peu avant. Si ?a r?le alors qu'on est s?r de d?sallouer un truc allou? pr?c?demment comme il faut, alors c'est qu'on a d?bord? dans un buffer. J'essaye de aire un exemple. ?a c'est le tas avant un malloc : ------------------------------------------------------------- | x | ------------------------------------------------------------- 'x' c'est un pointeur qui donne des infos sur ce qu'il y a apr?s, en gros le prochain truc libre, le pr?c?dent, et des trucs du genre. Si maintenant on ait un malloc, on va se retrouver avec ?a : ------------------------------------------------------------- | x | BLORP | y | ------------------------------------------------------------- Avec BLORP la zone qu'on vient d'allouer, et 'y' une nouvelle zone d'information comme 'x', qui dit o? est le morceau pr?c?dent, le suivant, etc. On reait un malloc : ------------------------------------------------------------- | x | BLORP | y | BROOZLE | z | ------------------------------------------------------------- Et si maintenant on ait un free(BROOZLE), il va lire dans 'y' (qu'il trouve ? partir de l'adresse de BROOZLE) et updater tout ?a : ------------------------------------------------------------- | x | BLORP | y | ------------------------------------------------------------- Maintenant, imaginons qu'un programmeur un peu porc, qui aurait par exemple index? son tableau BLORP de 1 ? n au lieu de 0 ? n-1, d?passe un peu de BLORP. ?a va ?crire des cochonneries dans 'y', et si on ait un free(BROOZLE), il va lire n'importe quoi dans 'y', et il y a 99% de chances que ?a segfaulte tout de suite, ou au prochain malloc, ou que malloc commence ? avoir un comportement irrationnel. Si on a MALLOC_CHECK_ activ?, il va aire des v?rifications suppl?mentaires sur 'y', et probablement dire "cette zone est corrompue, on ne va plus y toucher", histoire que le programme continue ? tourner. Mais d?s que MALLOC_CHECK_ ait le moindre warning c'est un bug. -- Sam. From bbp at via.ecp.fr Fri Apr 20 01:27:37 2001 From: bbp at via.ecp.fr (Brieuc Jeunhomme) Date: Fri, 20 Apr 2001 01:27:37 +0200 Subject: =?iso-8859-1?Q?probl=E8me_de_m=E9moire_lors_de_la_r=E9initialisation?= In-Reply-To: <20010417120925.M27642@zoy.org>; from sam@zoy.org on Tue, Apr 17, 2001 at 12:09:26PM +0200 References: <20010417113320.B16807@via.ecp.fr> <20010417120925.M27642@zoy.org> Message-ID: <20010420012737.A8401@via.ecp.fr> > Sous Linux malloc fonctionne toujours, m?me avec des setrlimit > partout. S'il renvoie NULL c'est que le tas est corrompu. Bon, j'ai quand m?me essay?. J'ai fait tourner les deux programmes suivants : bbp at obelix /tmp 0% cat a.c #include #include #include #include int main () { struct rlimit a; a.rlim_cur=1000; a.rlim_max=1000; setrlimit(RLIMIT_DATA,&a); setrlimit(RLIMIT_STACK,&a); setrlimit(RLIMIT_MEMLOCK,&a); setrlimit(RLIMIT_AS,&a); printf("%p\n",malloc(2000000)); return 0; } bbp at obelix /tmp 0% ./a (nil) bbp at obelix /tmp 0% cat b.c #include int main() { printf("%p\n",malloc(2000000)); return 0; } bbp at obelix /tmp 0% ./b 0x40132008 Donc, ben malloc renvoie parfois NULL et mon hypoth?se tient la route, m?me si il reste effectivement possible que le probl?me vienne d'une corruption du tas. -- BBP From cvs at videolan.org Fri Apr 20 06:21:09 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Fri, 20 Apr 2001 06:21:09 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010420042109.2EDD523400@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: Makefile TODO types.c Update of /cvs/videolan/network/VLANserver/config Modified Files: config.fl Update of /cvs/videolan/network/VLANserver/server Modified Files: connect.c connect.h poll_switch.c poll_switch.h request_handler.c server.c server.h Update of /cvs/videolan/network/VLANserver/snmp Modified Files: callback.c if_snmp.c if_snmp.h snmp.c snmp.h snmp_switch.h Log Message: . Added an option in .fl in order to avoid the gcc warning . Changed DEBUG option in DEBUG for logs and MEM_DEBUG for memory purpose . Removed *vs argument (vs is now a global variable) in each function . Updated vlanserver.conf.sample . snmp time out is yet implemented . so switches are not polled the same way: no second init, just reasking 5 times aborting after . Added the good #includes in order not to have to put the snmp prototypes in our code . Removed a segfault in snmp when using traps All this seems to work well. BUT there is always the current segfault. I think now that it relies on double use of libsnmp I mean call to snmp_create_pdu inside one of our function which is itself called by snmp_read. Seems not to be allowed. I'm gonna try not to resend anything in callback.c but pushing requests to main loop in order to execute snmp functions outside snmp_read. -- nitrox From nitrox at via.ecp.fr Fri Apr 20 06:32:07 2001 From: nitrox at via.ecp.fr (Damien Lucas) Date: Fri, 20 Apr 2001 06:32:07 +0200 Subject: =?iso-8859-1?Q?probl=E8me_de_m=E9moire_lors_de_la_r=E9initialisation?= In-Reply-To: <20010417113320.B16807@via.ecp.fr>; from bbp@via.ecp.fr on Tue, Apr 17, 2001 at 11:33:20AM +0200 References: <20010417113320.B16807@via.ecp.fr> Message-ID: <20010420063207.A16653@via.ecp.fr> On Tue, Apr 17, 2001, Brieuc Jeunhomme wrote: > Gunther avait rep?r? un probl?me de m?moire (les malloc ne > fonctionnaient plus) lors de la r?initialisation de la base, alors que Etait-ce tous les mallocs ou juste ceux du snmp ? Parce que si c'?tait seulement ceux du snmp, j'ai id?e de ce que ca pourrait ?tre. En passant, qqn sait-il si on peut avoir les symboles de debug dans les .so ? -- nitrox From sam at zoy.org Fri Apr 20 10:56:57 2001 From: sam at zoy.org (Samuel Hocevar) Date: Fri, 20 Apr 2001 10:56:57 +0200 Subject: =?iso-8859-1?Q?probl=E8me_de_m=E9moire_lors_de_la_r=E9initialisation?= In-Reply-To: <20010420063207.A16653@via.ecp.fr>; from nitrox@via.ecp.fr on Fri, Apr 20, 2001 at 06:32:07AM +0200 References: <20010417113320.B16807@via.ecp.fr> <20010420063207.A16653@via.ecp.fr> Message-ID: <20010420105657.A11129@zoy.org> On Fri, Apr 20, 2001, Damien Lucas wrote: > En passant, qqn sait-il si on peut avoir les symboles de debug dans les .so ? Si c'est des packages Debian, il y a parfois libfoo-dbg ? installer ? la place de libfoo et qui contient les symboles de debug. Sinon, faut se le recompiler en -g ? la main. -- Sam. From sam at zoy.org Fri Apr 20 11:02:17 2001 From: sam at zoy.org (Samuel Hocevar) Date: Fri, 20 Apr 2001 11:02:17 +0200 Subject: =?iso-8859-1?Q?probl=E8me_de_m=E9moire_lors_de_la_r=E9initialisation?= In-Reply-To: <20010420012737.A8401@via.ecp.fr>; from bbp@via.ecp.fr on Fri, Apr 20, 2001 at 01:27:37AM +0200 References: <20010417113320.B16807@via.ecp.fr> <20010417120925.M27642@zoy.org> <20010420012737.A8401@via.ecp.fr> Message-ID: <20010420110217.B11129@zoy.org> On Fri, Apr 20, 2001, Brieuc Jeunhomme wrote: > setrlimit(RLIMIT_AS,&a); Ah ouais, un point mongolo?de pour moi, j'avais test? qu'avec RLIMIT_DATA. Grmpfltz. -- Sam. From bbp at via.ecp.fr Fri Apr 20 12:42:42 2001 From: bbp at via.ecp.fr (Brieuc Jeunhomme) Date: Fri, 20 Apr 2001 12:42:42 +0200 Subject: =?iso-8859-1?Q?probl=E8me_de_m=E9moire_lors_de_la_r=E9initialisation?= In-Reply-To: <20010420063207.A16653@via.ecp.fr>; from nitrox@via.ecp.fr on Fri, Apr 20, 2001 at 06:32:07AM +0200 References: <20010417113320.B16807@via.ecp.fr> <20010420063207.A16653@via.ecp.fr> Message-ID: <20010420124242.A4651@via.ecp.fr> > Etait-ce tous les mallocs ou juste ceux du snmp ? Tous. -- BBP From cvs at videolan.org Fri Apr 20 20:21:50 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Fri, 20 Apr 2001 20:21:50 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010420182150.C814A23400@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: Makefile Update of /cvs/videolan/network/VLANserver/snmp Modified Files: callback.c if_snmp.c if_snmp.h snmp.c snmp.h snmp_switch.h Log Message: . Few changes inside snmp . Biggest one is that snmp_send is no more called inside the callback It's much better like that, But it doesn't seems to be the reason of the segfault. -- nitrox From cvs at videolan.org Sat Apr 21 02:14:53 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sat, 21 Apr 2001 02:14:53 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010421001453.0356E23400@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver/interface Modified Files: telnet.c Update of /cvs/videolan/network/VLANserver/server Modified Files: client.c server.c Update of /cvs/videolan/network/VLANserver/snmp Modified Files: callback.c if_snmp.c if_snmp.h snmp.c snmp_switch.h Log Message: . added a portid[port,unit] table definition . added functions to init this table Snmp_PORT & PORT_Snmp are not fully removed because of conceptual problems of initialisation... -- tooney From cvs at videolan.org Sat Apr 21 21:33:33 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sat, 21 Apr 2001 21:33:33 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010421193333.7066923400@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: Makefile main.c types.c Added Files: mem.c mem.h Update of /cvs/videolan/network/VLANserver/snmp Modified Files: callback.c snmp.c Log Message: . Segfault is dead. One word about what it was: A switch has zwitch->nbports ports (defined in the config) If you put 24 for a 26 ports switch, you won't take care of the 2 lasts anywhere but when walking on the macs, you'll got macs on 25 and 26 and we did fill zwitch->machine[25] which was not allocated. Doing that we were corrupting the memory in the snmplib code . Add tools for debugging in prevision of next segfault ;-) set MEM_DBG=1 in Makefile to compile with. It activated the Memory limitation And it logs all calls of malloc and free even inside dynamic libs Ready to test ! -- nitrox From cvs at videolan.org Sun Apr 22 04:33:23 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sun, 22 Apr 2001 04:33:23 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010422023323.6F21123400@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: Makefile common.h logger.c logger.h main.c mem.h sleep.c types.c types.h vlanserver.h Update of /cvs/videolan/network/VLANserver/config Modified Files: config.fl Update of /cvs/videolan/network/VLANserver/db Modified Files: db.c db.h Update of /cvs/videolan/network/VLANserver/interface Modified Files: interface.c interface.h telnet.c telnet.h Update of /cvs/videolan/network/VLANserver/server Modified Files: client.c connect.c poll_switch.c reinit_switchs.c request_handler.c send.c server.c Update of /cvs/videolan/network/VLANserver/snmp Modified Files: callback.c if_snmp.c if_snmp.h snmp.c snmp_switch.h Log Message: . Checked all #include's in order to remove all #include inside .h files and to call only the usefull .h files . Few chages in Makefile . Added #ifndef __XXXX_H__ inside .h files #define __XXXX_H__ [..] #endif . Changed the vlanserver struct, now vs->cfg, vs->snmp, vs->db are pointers -- nitrox From cvs at videolan.org Sun Apr 22 13:55:04 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sun, 22 Apr 2001 13:55:04 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010422115504.2814A23409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: main.c Log Message: Hum, let's say I was a bit tired for my last commit . Added few malloc ... -- nitrox From cvs at videolan.org Sun Apr 22 16:35:36 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sun, 22 Apr 2001 16:35:36 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010422143536.0E47823409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver/interface Modified Files: interface.c Update of /cvs/videolan/network/VLANserver/server Modified Files: poll_switch.c server.c Log Message: . Debugging with network test. . Seems to work well . Added a vlanid->vlan conversion after end of init Part of it is a kludge. I would change that soon! -- nitrox From cvs at videolan.org Mon Apr 23 00:33:58 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Mon, 23 Apr 2001 00:33:58 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010422223358.388A123409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver/server Modified Files: server.c Log Message: . Removed a polioterie in my last commit commented lines in server/server.c must be commented. . I found somes bugs: some mac addresses are not correctly learned I don't know if it is a problem with the swicthes, the snmp or the db. -- nitrox From cvs at videolan.org Mon Apr 23 01:58:22 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Mon, 23 Apr 2001 01:58:22 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010422235822.1371D23409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver/server Modified Files: server.c Log Message: I'd better not commit in the following hours .... . Corrected a polioterie in the previous commit -- nitrox From cvs at videolan.org Mon Apr 23 02:59:57 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Mon, 23 Apr 2001 02:59:57 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010423005957.AE87C23409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: Makefile Update of /cvs/videolan/network/VLANserver/interface Modified Files: interface.h telnet.c Log Message: This is the first version of telnet authentification. It works with passwords generated by crypt stored in the file "./passwd". The file format is at this time the following : user_name crypted_password user_name crypted_password ... Afterwards, it should reflect the format of /etc/passwd to take care of user permissions. -- lool From cvs at videolan.org Mon Apr 23 10:31:49 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Mon, 23 Apr 2001 10:31:49 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010423083149.C491023409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: types.h Update of /cvs/videolan/network/VLANserver/config Modified Files: config.fl Update of /cvs/videolan/network/VLANserver/db Modified Files: db.c Update of /cvs/videolan/network/VLANserver/doc Modified Files: vlanserver.conf.sample Update of /cvs/videolan/network/VLANserver/server Modified Files: client.c Log Message: -Add a defecports in the config. It can't be define in the switch type. Hope there are no segfaults ... -- gunther From bbp at via.ecp.fr Mon Apr 23 11:54:00 2001 From: bbp at via.ecp.fr (Brieuc Jeunhomme) Date: Mon, 23 Apr 2001 11:54:00 +0200 Subject: CVS Commit: network In-Reply-To: <20010423083149.C491023409@brahma.via.ecp.fr>; from cvs@videolan.org on Mon, Apr 23, 2001 at 10:31:49AM +0200 References: <20010423083149.C491023409@brahma.via.ecp.fr> Message-ID: <20010423115400.A20296@via.ecp.fr> > -Add a defecports in the config. It can't be define in the switch type. a WHAT ? -- BBP From gunther at via.ecp.fr Mon Apr 23 18:07:56 2001 From: gunther at via.ecp.fr (Laurent Rossier) Date: Mon, 23 Apr 2001 18:07:56 +0200 Subject: CVS Commit: network In-Reply-To: <20010423115400.A20296@via.ecp.fr>; from bbp@via.ecp.fr on Mon, Apr 23, 2001 at 11:54:00AM +0200 References: <20010423083149.C491023409@brahma.via.ecp.fr> <20010423115400.A20296@via.ecp.fr> Message-ID: <20010423180756.A23298@via.ecp.fr> On Mon, Apr 23, 2001, Brieuc Jeunhomme wrote: > a WHAT ? a port which enables you to defec ... Seriously this is for a port which doesn't work any more. -- Gunther From cvs at videolan.org Wed Apr 25 00:28:56 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Wed, 25 Apr 2001 00:28:56 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010424222856.72A9723409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: TODO logger.c types.h Update of /cvs/videolan/network/VLANserver/config Modified Files: config.fl config.h Update of /cvs/videolan/network/VLANserver/doc Modified Files: vlanserver.conf.sample Update of /cvs/videolan/network/VLANserver/server Modified Files: client.c server.c Log Message: -Add the dns resolution : we can give the dns of the vlanbridge in the vlanserver.conf and the dns of the server in the client -Remove this job from the TODO list and a new job -Clean the logger.c and the types.h : modify the ERR_CODE -- gunther From gunther at via.ecp.fr Wed Apr 25 00:30:43 2001 From: gunther at via.ecp.fr (Laurent Rossier) Date: Wed, 25 Apr 2001 00:30:43 +0200 Subject: CVS Commit: network In-Reply-To: <20010424222856.72A9723409@brahma.via.ecp.fr>; from cvs@videolan.org on Wed, Apr 25, 2001 at 12:28:56AM +0200 References: <20010424222856.72A9723409@brahma.via.ecp.fr> Message-ID: <20010425003043.A9137@via.ecp.fr> Log Message: -Add the dns resolution : we can give the dns of the vlanbridge in the vlanserver.conf and the dns of the server in the client -Remove this job from the TODO list and a new job -Clean the logger.c and the types.h : modify the ERR_CODE Voila c mieux comme ca desole pour la gene occasionn?e ;p -- Gunther From cvs at videolan.org Wed Apr 25 13:11:18 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Wed, 25 Apr 2001 13:11:18 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010425111118.E984E23409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: TODO Update of /cvs/videolan/network/VLANserver/config Modified Files: config.fl Log Message: -Write a regular expression for the dns in the config -Update the TODO list -- gunther From nitrox at via.ecp.fr Sat Apr 28 01:29:13 2001 From: nitrox at via.ecp.fr (Damien Lucas) Date: Sat, 28 Apr 2001 01:29:13 +0200 Subject: =?iso-8859-1?Q?CR_R=E9union_=2F_Orientations?= Message-ID: <20010428012913.B13103@via.ecp.fr> A l'odj: 1- Les GPAs 2- Le VLANserver actuel 3- La release 4- Le nouveau CHANNELserver (vlcs) 5- Quelques "d?tails" 1- Les GPAs ----------- Tonney: a d?j? cod? une fonction dans snmp Lool: a d?j? cod? une focntion dans interface/telnet Marcari: a d?j? cod? l'int?gration du changement de chaine au vlc Babal: codera peut-?tre plus tard sur le vlcs 2- Le VLANserver actuel ----------------------- Seul bug connu ? l'heure actuel, certaines MACs ne sont pas trouv?es alors que la machine associ?e est conue du switch. 3- La Release ------------- Objectif: Ce week-end D'ici l?: . Eviter si possible le message de warning dans la libsnmp (nitrox) . Proprifier le code (. en-t?te avant les fonctions, . wrapper les lignes . Rajouter un tag CVS dans le fichier: $Id) (nitrox) . Ecrire une FAQ (bbp) . Anonce WEB/ML (gunther) 4- Le nouveau CHANNELserver --------------------------- 4-1 Description Le noyau central SERVER: son r?le est de faire communiquer les modules. il transmet les requ?tes ? l'ensemble des modules qui en ont besoin Le module INTF finalement pas s?par? avec un INTF2 il comportera entre autre les modules VIDEOLANP, TELNETD, ... c'est la partie visible de l'ext?rieur Le module DB il contient la base de donn?es. Le module NETWORK Divis? en L2 et L3 puis (ETH,TR) et (ATM,IP) ? nouveau redivis? Le module INTFS D?signe la communication avec les serveurs video (Le module STAT) Il pourrait stater toutes les demandes de changement de chaine par exemple 4-2 Fonctionnement Chacun des modules cit?s ci-dessus doit fournir une liste d'action. Par exemple pour INTF: Changer de chaine Obtenir des infos Demander la reload de la config ... 4-3 Exemple ci dessous: la description de la demande de changement de chaine: VIDEOLANP_ch_channel() { ... INTF_ch_channel(); ... } INTF_ch_channel() { ... SERVER_ch_channel(); ... } SERVER_ch_channel() { STATS_ch_channel(); NETWORK_ch_channel(); } NETWORK_ch_channel(); { ... SERVER_get_info(); /* sur la machine qui demande */ ... if(method==L2) L2_ch_channel(); else if(method==L3) L3_ch_channel(); endif } SERVER_get_info() { DB_get_info(); } 4-4 R?partition du travail Il s'agit de r?fl?chir ? la liste des "fonctions effectives" que l'ont souhaite voir dans chaque module et la liste des objets importants du module (vu de l'ext?rieur ie pas les objets internes) intfs: babal mardi apr?s la rdb config: nitrox samedi 16h30 DB: bbp samedi 14h intf; gunther lool marc mardi 14h network: nitrox tooney dimanche 16h 5- Quelques d?tails ------------------- Il restait quelques questions: 5-1 DNS/IP Il a ?t? d?cid? qu'il serait bon de pouvoir sp?cifier dans le fichier de config si les DNS devaient ?tre r?solu ? chaque fois ou au lancement ce qui demande de stocker un cache DNS d'une mani?re ou d'une autre Il a ?t? ?voque l'id?e de mettre pour chaque entit? un champ DNS et un champ IP 5-2 libsnmp La quesion ?tait de savoir si compte-tenu de toutes merdes qu'elles contient on ?crirait notre propre lib. Je (nitrox) pense qu'avant de voir ? ?crire notre propre lib, il faudrait conna?tre ? fond celle-ci afin de l'utiliser correctement. Dans un premier temps, nous reposerons donc sur la libsnmp (net-snmp 4.2) 5-3 Le Vlanbridge Deux questions: 1- Faut-il le releaser? 2- Faut-il relancer son d?veloppement? A la premi?re question il a ?t? r?pondu non dans la mesure o? personne ne connait suffisamment le code pour le maintenir. A la desi?me question, il a ?t? r?pondu non dans la mesure o? rien n'urge dans ce qu'il reste ? y coder. Faites p?ter les remarques ! -- nitrox From cvs at videolan.org Sat Apr 28 03:01:27 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sat, 28 Apr 2001 03:01:27 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010428010127.1C38A23409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: TODO logger.c main.c types.c vlanserver.h Update of /cvs/videolan/network/VLANserver/config Modified Files: config.fl Update of /cvs/videolan/network/VLANserver/db Modified Files: db.c db.h Update of /cvs/videolan/network/VLANserver/interface Modified Files: interface.c telnet.c Update of /cvs/videolan/network/VLANserver/server Modified Files: client.c connect.c poll_switch.c reinit_switchs.c request_handler.c send.c server.c server.h Update of /cvs/videolan/network/VLANserver/snmp Modified Files: callback.c if_snmp.c if_snmp.h snmp.c snmp_switch.h Log Message: Code cleaning in order of the release: . added header before each function . wrapped all the lines to 79 with \ . added some comments -- nitrox From cvs at videolan.org Sat Apr 28 03:29:24 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sat, 28 Apr 2001 03:29:24 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010428012924.D0A9223409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: .cvsignore Makefile Update of /cvs/videolan/network/VLANserver/config Added Files: .cvsignore Update of /cvs/videolan/network/VLANserver/db Added Files: .cvsignore Update of /cvs/videolan/network/VLANserver/doc Added Files: Makefile.options.sample Update of /cvs/videolan/network/VLANserver/interface Added Files: .cvsignore Update of /cvs/videolan/network/VLANserver/server Added Files: .cvsignore Update of /cvs/videolan/network/VLANserver/snmp Added Files: .cvsignore Log Message: . Added .cvsignore in subdirectories . Changed Makefile. You should NOT change Makegfile But you have to create a Makefile.options (you can find an example in doc/ ) This changed should avoid to recommit Makefile each time somebody changed DEBUG to 1 -- nitrox From cvs at videolan.org Sat Apr 28 03:30:17 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sat, 28 Apr 2001 03:30:17 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010428013017.B72C423409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: .cvsignore Log Message: Oups, ... -- nitrox From cvs at videolan.org Sat Apr 28 06:00:39 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sat, 28 Apr 2001 06:00:39 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010428040039.8112A23409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver/doc Modified Files: vlanserver.conf.sample Log Message: better explanations in the vlanserver.conf.sample file -- bbp From bbp at via.ecp.fr Sat Apr 28 06:02:14 2001 From: bbp at via.ecp.fr (Brieuc Jeunhomme) Date: Sat, 28 Apr 2001 06:02:14 +0200 Subject: CVS Commit: network In-Reply-To: <20010428040039.8112A23409@brahma.via.ecp.fr>; from cvs@videolan.org on Sat, Apr 28, 2001 at 06:00:39AM +0200 References: <20010428040039.8112A23409@brahma.via.ecp.fr> Message-ID: <20010428060214.A31606@via.ecp.fr> > better explanations in the vlanserver.conf.sample file If we create a documentation sample file, this is not to see things like this : # the aob ports aobports 2 12 42 -- BBP From cvs at videolan.org Sat Apr 28 22:10:01 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sat, 28 Apr 2001 22:10:01 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010428201001.843B32340A@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver/doc Added Files: FAQ Log Message: First version of the FAQ for the release. -- bbp From bbp at via.ecp.fr Sat Apr 28 22:11:21 2001 From: bbp at via.ecp.fr (Brieuc Jeunhomme) Date: Sat, 28 Apr 2001 22:11:21 +0200 Subject: CVS Commit: network In-Reply-To: <20010428201001.843B32340A@brahma.via.ecp.fr>; from cvs@videolan.org on Sat, Apr 28, 2001 at 10:10:01PM +0200 References: <20010428201001.843B32340A@brahma.via.ecp.fr> Message-ID: <20010428221121.A19241@via.ecp.fr> > First version of the FAQ for the release. Probably needs corrections and ADD-ONs. Please wrap lines to 80 chars and don't forget the history at the top of the file. -- BBP From nitrox at via.ecp.fr Sun Apr 29 00:21:42 2001 From: nitrox at via.ecp.fr (Damien Lucas) Date: Sun, 29 Apr 2001 00:21:42 +0200 Subject: =?iso-8859-1?Q?CR_R=E9union_=2F_Orientations?= In-Reply-To: <20010428012913.B13103@via.ecp.fr>; from nitrox@via.ecp.fr on Sat, Apr 28, 2001 at 01:29:13AM +0200 References: <20010428012913.B13103@via.ecp.fr> Message-ID: <20010429002142.A31363@via.ecp.fr> On Sat, Apr 28, 2001, Damien Lucas wrote: > network: nitrox tooney dimanche 16h Finalement, dimanche 14h -- 'trox From cvs at videolan.org Sun Apr 29 00:46:37 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sun, 29 Apr 2001 00:46:37 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010428224637.78EDA23409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver/db Modified Files: db.c Update of /cvs/videolan/network/VLANserver/doc Modified Files: FAQ Update of /cvs/videolan/network/VLANserver/snmp Modified Files: callback.c Log Message: . Changed ML address in FAQ . Removed all occurences of 24 as number of ports (The number of ports of a switch is zwitch->nb_ports not 24 and the unit of a switch is switch->unit not 1) ... -- nitrox From cvs at videolan.org Sun Apr 29 02:48:41 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sun, 29 Apr 2001 02:48:41 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010429004841.40DDD23409@brahma.via.ecp.fr> -- nitrox From cvs at videolan.org Sun Apr 29 05:41:49 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Sun, 29 Apr 2001 05:41:49 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010429034149.C219023409@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: Makefile TODO common.h config.h logger.c logger.h main.c types.c types.h vlanserver.h Added Files: AUTHORS COPYING ChangeLog Removed Files: mem.c mem.h sleep.c sleep.h Update of /cvs/videolan/network/VLANserver/config Modified Files: config.fl config.h Update of /cvs/videolan/network/VLANserver/db Modified Files: db.c Update of /cvs/videolan/network/VLANserver/debug Added Files: mem.c mem.h sleep.c sleep.h Update of /cvs/videolan/network/VLANserver/interface Modified Files: interface.c interface.h telnet.c telnet.h Update of /cvs/videolan/network/VLANserver/old-tools/VLANclient Modified Files: sh_vlan.c Update of /cvs/videolan/network/VLANserver/old-tools/mini-VLANserver Modified Files: vlanc.c vlans-h2.sh vlans.c vlans.sh Update of /cvs/videolan/network/VLANserver/old-tools/vlb-emu Modified Files: vlb-emu.py Update of /cvs/videolan/network/VLANserver/server Modified Files: client.c client.h connect.c connect.h poll_switch.c poll_switch.h reinit_switchs.c request_handler.c request_handler.h send.c send.h server.c server.h Update of /cvs/videolan/network/VLANserver/snmp Modified Files: callback.c callback.h if_snmp.c if_snmp.h snmp.c snmp.h snmp_switch.h Log Message: . Added COPYING file (GPL license) . Added AUTHORS file . Added ChangeLog file (currently empty) . Use of VS_sleep is now just in case of debugging . Compilation of VS_sleep just in case of debugging . Moved mem.c/mem.h/sleep.c/sleep.h in a debug directory . Added header in each file with -a CVS tag $Id$ -the name of the file and a summary of it does -a copyrigth information -a word to say this file is GPL -the name of the author PLEASE, check AUTHORS and your files (I wasn't sure of the author of each) -- nitrox From cvs at videolan.org Mon Apr 30 04:29:03 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Mon, 30 Apr 2001 04:29:03 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010430022903.CE9D92340B@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver/interface Modified Files: telnet.c Log Message: This just : - adds a log message (LOG_DEBUG) for sucessful logins and - corrects a little bug in interface/telnet.c:CatchWord() : now you may only enter alphanums chars, and this corrects the drawing in the command line. -- lool From cvs at videolan.org Mon Apr 30 06:40:25 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Mon, 30 Apr 2001 06:40:25 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010430044025.CA5972340B@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: Makefile Log Message: . Fixed a bug in Makefile -- nitrox From cvs at videolan.org Mon Apr 30 16:58:54 2001 From: cvs at videolan.org (cvs at videolan.org) Date: Mon, 30 Apr 2001 16:58:54 +0200 (CEST) Subject: CVS Commit: network Message-ID: <20010430145854.C4BBE2340B@brahma.via.ecp.fr> Update of /cvs/videolan/network/VLANserver Modified Files: types.h Update of /cvs/videolan/network/VLANserver/config Modified Files: config.fl Update of /cvs/videolan/network/VLANserver/db Modified Files: db.c Update of /cvs/videolan/network/VLANserver/doc Modified Files: vlanserver.conf.sample Update of /cvs/videolan/network/VLANserver/server Modified Files: server.c Update of /cvs/videolan/network/VLANserver/snmp Modified Files: callback.c Log Message: -changed the defecportswith unusedports. -changed the way the flags and the protections of the ports are initialized. -added a test in the snmp : if the port has an ATM or an UNUSED flag we don't initialize the portids. -- gunther From gunther at via.ecp.fr Mon Apr 30 17:15:41 2001 From: gunther at via.ecp.fr (Laurent Rossier) Date: Mon, 30 Apr 2001 17:15:41 +0200 Subject: CVS Commit: network In-Reply-To: <20010430145854.C4BBE2340B@brahma.via.ecp.fr>; from cvs@videolan.org on Mon, Apr 30, 2001 at 04:58:54PM +0200 References: <20010430145854.C4BBE2340B@brahma.via.ecp.fr> Message-ID: <20010430171541.A523@via.ecp.fr> > -changed the way the flags and the protections of the ports are initialized. Because it has to be initialized before the VS_switch_init function... -- Gunther