Version 0.0.72

Wed Jul 1 19:10:00 2026 UTC

I've just uploaded fedanet-0.0.72.tbz2, in which a minor fix is made to the code. Thanks go to a guy with nickname posyx for reporting where to look. Hopefully, this fixes both issues with non-intel platforms (primarily ARM), and with NetBSD. Looks like it was because "default" char is unsigned there, while in one weird part of the code it was assumed to be signed.

I'm not sure if this will really work on ARMs as well as on NetBSD (as I have neither of them right now), so if you have any of them, please update and report back your experience, both positive or negative.


anonymous

From vjkarbo (unverified) Fri Jul 3 22:18:30 2026 UTC

pencil

sucessfull building on DragonFly

Thanks for including the build code, the build was successful. How did you solve the problem of running non-root on bsd systems? I run #su -m fedanet -c `/home/user1/.fedanet/fedaserv -v -E -f -c /home/user1/.fedanet/serv.conf' this way and get the same error as in version *.71.: open: tun0: Permission denied; failed to set up the tun iface. (the system has a fedanet pseudo-user and rights to the fedanet:fedanet directory and files). I also experimented with /etc/devfs.conf and /etc/devfs.rules but `uucp dialer on /dev/tun interface` is clearly a different privilege level.

anonymous

From EVI (unverified) Sat Jul 4 20:45:53 2026 UTC in reply to this comment

pencil

Re: sucessfull building on DragonFly

In OpenBSD I did this way:

# ifconfig tun0 inet6  group tun
# chown dummy /dev/tun0
# su dummy
$ ~/fedanet-0.0.70/src/fedaserv -f -E -v -v

NetBSD ignores file permissions for /dev/tun.

I checked DragonFlyBSD man 4 tun and didn't find quirks like in NetBSD. So maybe you should do it this this way:

# chown fedanet /dev/tun0
anonymous

From dragonflybsd cd-live (unverified) Sun Jul 5 13:47:34 2026 UTC in reply to this comment

pencil

Re: Re: sucessfull building on DragonFly

> chown user /dev/tunN That doesn't help.

Maybe just run it as root, and then call the setuid() system call at the right place.

Or write the program that does this as root, then calls the setuid() system call. After that, use `exec fedaserv ... -t fd`.

admin.jpg's userpic

From Andrey Stolyarov profile Sun Jul 5 14:59:25 2026 UTC in reply to this comment

pencil

Re: Re: Re: sucessfull building on DragonFly

> Maybe just run it as root, and then call the setuid()

This is certainly an option, but not as the default behaviour, because (1) there's no need to start as root in Linux and FreeBSD, and AFAIR in OpenBSD too; and (2) the ability to go with no root access is a useful feature in itself: for some node masters, buying their own VPS service may appear too expensive, while it seems to be relatively easy to gain access to someone else's server, but certainly without root privileges.

> Or write the program that does this as root, then calls the setuid() system call. After that, use `exec fedaserv ... -t fd`

Yes, this is exaclty what I planned for, but still didn't implement it, mainly because I can't test it on my own. Well, perhaps I should try to code something like that, and ask the public here to help me with the testing.

anonymous

From dragonflybsd cd-live (unverified) Sun Jul 5 16:37:00 2026 UTC in reply to this comment

pencil

Re: Re: Re: Re: sucessfull building on DragonFly

I have a live CD (dragonflybsd) to test this attempt.

anonymous

From vjkarbo (unverified) Sun Jul 5 17:41:28 2026 UTC in reply to this comment

pencil

Re: Re: Re: Re: sucessfull building on DragonFly

Many thanks to EVI and Andrey Stolyarov for their response. I tried the option from the first answer chown fedanet /dev/tun0 (and chown fedanet:fedanet) and yes, although the group (and user) changes, the interface still does not open. I haven't tried the option from the second answer with sysuid yet. To Andrey Stolyarov: I am ready to provide you with my VPS, and also spend as much time testing as necessary, I will write to you.

anonymous

From EVI (unverified) Mon Jul 6 05:30:00 2026 UTC in reply to this comment

pencil

Re: Re: Re: Re: Re: sucessfull building on DragonFly

"You have my sword. And you have my bow. And my VPS."

admin.jpg's userpic

From Andrey Stolyarov profile Mon Jul 6 10:18:55 2026 UTC in reply to this comment

pencil

Re: sucessfull building on DragonFly

LOL

admin.jpg's userpic

From Andrey Stolyarov profile Mon Jul 6 07:40:07 2026 UTC in reply to this comment

pencil

Re: sucessfull building on DragonFly

This sounds strange to me, I was sure they use the same kernel as in FreeBSD. Okay, perhaps I was wrong.

BTW, I replied to your email, so if you don't see the reply, check that "spam folder" plz :-)

anonymous

From EVI (unverified) Fri Jul 3 20:08:06 2026 UTC

pencil

NetBSD RPi 1B+ works

# ping6 feda:f22f:f544:8c95:1bb3:3897:100:feda
PING6(56=40+8+8 bytes) feda:f22f:f544:8c95:1bb3:3897:7700:feda --> feda:f22f:f544:8c95:1bb3:3897:100:feda
16 bytes from feda:f22f:f544:8c95:1bb3:3897:100:feda, icmp_seq=0 hlim=61 time=269.994 ms
16 bytes from feda:f22f:f544:8c95:1bb3:3897:100:feda, icmp_seq=1 hlim=61 time=269.329 ms
admin.jpg's userpic

From Andrey Stolyarov profile Sat Jul 4 19:20:32 2026 UTC in reply to this comment

pencil

Re: NetBSD RPi 1B+ works

Great news, thanks!

no userpic

From posyx profile Thu Jul 2 12:36:48 2026 UTC

pencil

Test

I modified and tested it myself, and it works. This version (0.0.72) also works.

Question for user Ilya: Did this fix the issue with duplicate packages in his “aarch64+debian” configuration?

admin.jpg's userpic

From Andrey Stolyarov profile Thu Jul 2 20:45:44 2026 UTC in reply to this comment

pencil

Re: Test

> This version (0.0.72) also works.

Thanks!

no userpic

From Ilya profile Thu Jul 23 22:34:08 2026 UTC in reply to this comment

pencil

Re: Test

Sorry for late response, I was a bit busy lately. I have updated my node to 0.0.72, I hope it works. Right now, I do not see a lot of messages on "debug" level logging, other than standard keepalive.

admin.jpg's userpic

From Andrey Stolyarov profile Fri Jul 24 15:56:50 2026 UTC in reply to this comment

pencil

Re: Re: Test

Yeah, from my side it looks like your node behaves correctly. Thanks!