* [Qemu-devel] problems with dual ethernet networking @ 2004-11-02 18:44 marco ghidinelli 2004-11-03 2:45 ` Jim C. Brown 0 siblings, 1 reply; 5+ messages in thread From: marco ghidinelli @ 2004-11-02 18:44 UTC (permalink / raw) To: qemu-devel i'm unable to get a running configuration with a dual ethernet card with qemu and vdeq. i try with this commands: vde_switch -daemon -sock /tmp/vde1.ctl chmod 777 /tmp/vde1.ctl vde_switch -daemon -sock /tmp/vde2.ctl chmod 777 /tmp/vde2.ctl vdeq qemu -sock /tmp/vde1.ctl,/tmp/vde2.ctl -nics 2 -hda vivian.img and i got this message: qemu: invalid fd for network interface 0 running: vdeq echo -sock /tmp/vde1.ctl,/tmp/vde2.ctl -nics 2 -hda vivian.img i got: -tun-fd 3,5 -nics 2 -hda images/vivian.img and looking in vl.c seems that qemu doesn't like very much the option -tun-fd 3,5 is this option implemented? where i am wrong? let me know. bye! -- BOFH excuse #10: hardware stress fractures ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [Qemu-devel] problems with dual ethernet networking 2004-11-02 18:44 [Qemu-devel] problems with dual ethernet networking marco ghidinelli @ 2004-11-03 2:45 ` Jim C. Brown 2004-11-03 9:35 ` Renzo Davoli 0 siblings, 1 reply; 5+ messages in thread From: Jim C. Brown @ 2004-11-03 2:45 UTC (permalink / raw) To: qemu-devel On Tue, Nov 02, 2004 at 07:44:43PM +0100, marco ghidinelli wrote: > i'm unable to get a running configuration with a dual ethernet card with > qemu and vdeq. > > i try with this commands: > > vde_switch -daemon -sock /tmp/vde1.ctl > chmod 777 /tmp/vde1.ctl > vde_switch -daemon -sock /tmp/vde2.ctl > chmod 777 /tmp/vde2.ctl > > vdeq qemu -sock /tmp/vde1.ctl,/tmp/vde2.ctl -nics 2 -hda vivian.img > > and i got this message: > > qemu: invalid fd for network interface 0 > > running: > > vdeq echo -sock /tmp/vde1.ctl,/tmp/vde2.ctl -nics 2 -hda vivian.img > > i got: > > -tun-fd 3,5 -nics 2 -hda images/vivian.img > > and looking in vl.c seems that qemu doesn't like very much the option > -tun-fd 3,5 > > is this option implemented? where i am wrong? No it is not implemented. vdeq is meant for only one sock at a time. You may be able to get qemu to work by using -tun-fd 3 -tun-fd 5 tho (not sure), but vdeq won't support that unless you change it. > let me know. > > bye! > > -- > BOFH excuse #10: > > hardware stress fractures > > > _______________________________________________ > Qemu-devel mailing list > Qemu-devel@nongnu.org > http://lists.nongnu.org/mailman/listinfo/qemu-devel > -- Infinite complexity begets infinite beauty. Infinite precision begets infinite perfection. ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [Qemu-devel] problems with dual ethernet networking 2004-11-03 2:45 ` Jim C. Brown @ 2004-11-03 9:35 ` Renzo Davoli 2004-11-03 12:40 ` [Qemu-devel] [patch] " Renzo Davoli 0 siblings, 1 reply; 5+ messages in thread From: Renzo Davoli @ 2004-11-03 9:35 UTC (permalink / raw) To: qemu-devel > > vdeq qemu -sock /tmp/vde1.ctl,/tmp/vde2.ctl -nics 2 -hda vivian.img > No it is not implemented. vdeq is meant for only one sock at a time. Actually, it was implemented in qemu and vdeq used the multiple -tun-fd option provided bu qemu. Now I'll try to see what has happened inside the qemu code. Stay tuned. renzo ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [Qemu-devel] [patch] problems with dual ethernet networking 2004-11-03 9:35 ` Renzo Davoli @ 2004-11-03 12:40 ` Renzo Davoli 2004-11-03 13:48 ` marco ghidinelli 0 siblings, 1 reply; 5+ messages in thread From: Renzo Davoli @ 2004-11-03 12:40 UTC (permalink / raw) To: qemu-devel [-- Attachment #1: Type: text/plain, Size: 1078 bytes --] On Wed, Nov 03, 2004 at 10:35:53AM +0100, Renzo Davoli wrote: > Actually, it was implemented in qemu and vdeq used the multiple > -tun-fd option provided bu qemu. > qemu has changed recently the option management, now it wants multiple -tun-fd options for several tun interfaces (previously it accepted several comma separated numbers). I have just committed on cvs an updated version of vde that should be able to work with multiple interfaces. the -nics option should not be needed. Sourceforge sometimes is late in updating the public access cvs, thus I am sending here enclosed the patch for vdeq.c. I have had no time to test it personally, please let me know if it works correctly. BTW: if you are connecting also several qemu machines on the same vde network remember to change the MAC addresses! The default configuration MAC is 52:54:00:12:34:56. (for eth0, 52:54:00:12:34:57 for eth1 etc.). Several machines on the same network using the standard configuration end up having the same MAC: networking does not work. i.e. it is a net-does-not-work. renzo [-- Attachment #2: vdeq.diff --] [-- Type: text/plain, Size: 1556 bytes --] --- vde-1.5.7/qemu/vdeq.c 2004-11-03 12:57:35.185373176 +0100 +++ vde-1.5.8/qemu/vdeq.c 2004-11-03 13:11:43.333435072 +0100 @@ -78,12 +78,10 @@ unsigned char bufin[BUFSIZE]; -//struct pollfd pollv[]={{STDIN_FILENO,POLLIN|POLLHUP,0},{0,POLLIN|POLLHUP,0}}; struct pollfd *pollv; char *filename; char *vdeqname; -char *numfd; #define NUMW 10 static int countnics(const char *s) @@ -169,19 +167,12 @@ perror("malloc nics"); exit(1); } - if ((numfd= (char *) malloc(nb_nics * NUMW * sizeof (char)))<0) { - perror("malloc numfd"); - exit(1); - } for (i=0; i<nb_nics; i++) { - char curfd[NUMW]; if (socketpair(AF_UNIX, SOCK_DGRAM, 0, sp[i]) < 0){ perror("socketpair"); exit(1); } - snprintf(curfd,NUMW,"%d%c",sp[i][0],(i<nb_nics-1)?',':0); - strcat(numfd,curfd); } if ((sockname= (char **) malloc(sizeof(char *) * nb_nics))<0) { @@ -207,17 +198,21 @@ printf("as %s\n",argsock); for (i=0; i<nb_nics; i++) printf("%d -> %s\n",i,sockname[i]); */ - newargc=argc+3-args; + newargc=argc+1+(2*nb_nics)-args; if ((newargv=(char **) malloc ((newargc+1)* sizeof(char *))) <0) { perror("malloc"); exit(1); } newargv[0]=filename; - newargv[1]="-tun-fd"; - newargv[2]=numfd; + for (i=0; i<nb_nics; i++) { + char numfd[10]; + sprintf(numfd,"%d",sp[i][0]); + newargv[2*i+1]="-tun-fd"; + newargv[2*i+2]=strdup(numfd); + } - for (i=3;args<argc;i++,args++) newargv[i]=argv[args]; + for (i=(2*nb_nics)+1;args<argc;i++,args++) newargv[i]=argv[args]; newargv[i]=0; ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [Qemu-devel] [patch] problems with dual ethernet networking 2004-11-03 12:40 ` [Qemu-devel] [patch] " Renzo Davoli @ 2004-11-03 13:48 ` marco ghidinelli 0 siblings, 0 replies; 5+ messages in thread From: marco ghidinelli @ 2004-11-03 13:48 UTC (permalink / raw) To: qemu-devel On Wed, Nov 03, 2004 at 01:40:25PM +0100, Renzo Davoli wrote: > On Wed, Nov 03, 2004 at 10:35:53AM +0100, Renzo Davoli wrote: > > Actually, it was implemented in qemu and vdeq used the multiple > > -tun-fd option provided bu qemu. > > > qemu has changed recently the option management, now it wants multiple > -tun-fd options for several tun interfaces (previously it accepted several > comma separated numbers). > > I have just committed on cvs an updated version of vde that should be able > to work with multiple interfaces. > the -nics option should not be needed. > > Sourceforge sometimes is late in updating the public access cvs, thus I > am sending here enclosed the patch for vdeq.c. > > I have had no time to test it personally, please let me know if it works > correctly. it works perfectly. great. > BTW: if you are connecting also several qemu machines on the same vde > network remember to change the MAC addresses! > The default configuration MAC is 52:54:00:12:34:56. (for eth0, > 52:54:00:12:34:57 for eth1 etc.). yes, of course! :) > renzo thankyou very much! :) ^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2004-11-03 13:56 UTC | newest] Thread overview: 5+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2004-11-02 18:44 [Qemu-devel] problems with dual ethernet networking marco ghidinelli 2004-11-03 2:45 ` Jim C. Brown 2004-11-03 9:35 ` Renzo Davoli 2004-11-03 12:40 ` [Qemu-devel] [patch] " Renzo Davoli 2004-11-03 13:48 ` marco ghidinelli
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).