IRC channel logs
2026-07-22.log
back to list of logs
<etno>Hello, I am trying to setup a tun0 device with pfinet, and can't seem to flow packets. Here are the commands I use : <etno>setup the pfinet translator: settrans -a /home/ovalenti/2 hurd/pfinet/pfinet -i /home/ovalenti/tun0 -6 /home/ovalenti/26 -a 192.168.88.1 -p 192.168.88.2 <etno>then run a bash shell in the new stack context: remap /servers/socket/2 /home/ovalenti/2 /servers/socket/26 /home/ovalenti/26 -- /bin/bash <etno>Then I "cat tun0" (tried with a small C program calling read() in a loop) <etno>...while I "ping 192.168.88.2", but nothing comes out of the "cat" command <youpi>etno: ping is also within the remap? <etno>ifconfig shows the tunnel device <youpi>I'd say put prints in pfinet to make sure what really happens <youpi>notably check if tunnel_xmit gest the packet <youpi>to determine whether the issue is within tunnel.c or in the rest of pfinet <youpi>(always start with checking *where* the bug happens) <rrq>hmm where/how does it provide debug logging? <etno>Thanks youpi, it helps already knowing that my commands actually make sense :) <rrq>hmm is there some commands similar to "ip" on hurd? (using ifconfig just kills eth0) <youpi>if it kills eth0, that's a bug to fix ;) <rrq>yeah, I do "ifconfig /dev/tap2 10.3.3.1 nemask 255.255.255.0 up" and get the respons .. long line: <rrq>ifconfig: Could not make pfinet /servers/socket/2 drive /dev/tap2 with '-i /dev/tap2 /hurd/pfinet --interface=/dev/eth0 --address=192.168.240.2 --netmask=255.255.255.0 --gateway=192.168.240.1 --address6=fe80::5254:12:3456/10 --address6=fe80::5054:ff:fe12:3456/10 --gateway6=::': (ipc/mig) server died <Alicia>missing a t in netmask (in case it was also missing in the actual command) <rrq>yes.. my misspelling here, not in my command <rrq>that 192.168.240.2 address is the one /dev/eth0 has/had ... but I don't understand why it should get involved at all <rrq>I have a pfinet tied to /dev/tap2 already, but not using /servers/socket/2 (why should I?) <youpi>rrq: ifconfig uses fsysopts to show the user the current pfinet daemon configuration <rrq>I have 2 pfinet processes <youpi>did you tell ifconfig which one to use? <rrq>you mean ifconfig looks for a particular pathname to find a pfinet translator? <rrq>I guess "/servers/socket/2? then? <youpi>yes, like glibc's network access etc. <rrq>new attempt: remap /servers/socket/2 /servers/tap2 -- ifconfig <rrq>/dev/tap2 10.3.3.1 netmask 255.255.255.0 up <rrq>... hmm how do I convince it to use my new pfinet variant, presently living at /root/hurd/build/pfinet/pfinet ? <youpi>by settrans-ing it on /servers/tap2 then remaping to it? <rrq>tyring agin; first attempt died because the "tap client" was already attached, so replacing with a new failed.. then following that, ifconfig wants to use /hurd/pfinet I guess ... maybe I need another remap <youpi>ifconfig doesn't use /hurd/pfinet <youpi>it just accesses /servers/socket/2 <youpi>which you can remap to whatever place you have settrans'ed your own pfinet <rrq>mm I get: /root/hurd/build/pfinet/pfinet: tap2: Device or resource busy <rrq>/usr/bin/ifconfig: Could not make pfinet /servers/socket/2 drive /dev/tap2 with '-i /dev/tap2 /root/hurd/build/pfinet/pfinet --interface=tap2 --address=0.0.0.0 --netmask=255.0.0.0': (ipc/mig) server died <rrq>/bin/settrans: fsys_goaway: (ipc/mig) server died <youpi>what you have pasted includes various stuff <youpi>if you don't tell us what you are doing, we can't tell what you are doing wrong <youpi>(and we can't interpret the messages you get) <rrq>sorry it was 3 error lines, though irssi ate the first line <rrq>exactly I did the following 2: <rrq>settrans -cga /servers/tap2 /root/hurd/build/pfine <rrq>t/pfinet -i /dev/tap2 <youpi>did you add support for recognizing "tap" ? <rrq>hmm all as one line, then <rrq>remap /servers/socket/2 /servers/tap2 -- ifconfig /dev/tap2 10.3.3.1 netmask 255.255.255.0 up <rrq>which gave 3 error lines: <rrq> /root/hurd/build/pfinet/pfinet: tap2: Device or resource busy <rrq> /usr/bin/ifconfig: Could not make pfinet /servers/socket/2 drive /dev/tap2 with '-i /dev/tap2 /root/hurd/build/pfinet/pfinet --interface=tap2 --address=0.0.0.0 --netmask=255.0.0.0': (ipc/mig) server died <rrq> /bin/settrans: fsys_goaway: (ipc/mig) server died <youpi>ifconfig show the interface as tap2, doesn't it? <youpi>it'd then be ifconfig tap2 10.3.3.1 netmask 255.255.255.0 up <youpi>so it knows it should just change the address and not change the interface list <youpi>it works for tun for instance <youpi>(that's why I advised first making sure you know how to make tun running, before adding code which you don't know if it is correct or not, always start from a known-working situation) <youpi>(the last warning about fsys_goaway is due to remap itself) <rrq>hmm, plain: remap /servers/socket/2 /servers/tap2 -- ifconfig <rrq>tells about tap2, and plain: ifconfig tells about /dev/eth0 <youpi>so use tap2, not /dev/tap2, when configuring it <youpi>I guess somehow pfinet strips the /dev/ prefix when dealing with tun/tap <youpi>which we should probably fix for less surprises such as seen here <rrq>put it also needs remap <youpi>the remap is only for the access to /servers/socket/2 <rrq>well if I forget that, ifconfig kills eth0 for me <youpi>you mean if you run ifconfig -i /dev/tap0 for an existing pfinet (without talking about any change)? <youpi>yes, as I mentioned at the very beginning, that's probably just a bug to fix <youpi>we are not out of bugs to fix, we are out of people actually taking time to fix them <rrq>yes. I'm not always sure myself whether a cerain thing is a bug or just due to my ignorance <youpi>servers crashing is never user's fault <rrq>.. I guess also it points at the other difficulty, the get things documented enough <rrq>ACTION is trying to work out a reasnable exuse ... <rrq>anyhow, my tap client is reasonably happy, but I don't think the device setup is correct yet <rrq>mmm about bugs, the hurd "make" stops at make[1]: *** No rule to make target '../libstore/libstore.so', needed by 'libdiskfs.so.0.3'. Stop. <rrq>but "make pfinet" is fine, after setting llink to ../libtrivfs in pfinet <GRFJX4>Somebody told me, hurd does 64 bit nowadays. But all the documentation I read (wiki and the like) said it’s still 32 bit. Is the documentation outdated? <rrq>yes, documentation is slightly behind; e.g. I'm using hurd in qemu x86_64 (debian hurd) <rrq>it's still smp 1 though <etno>When I try to attach gdb to a live translator (pfinet) on UP+amd64, latest debian gdb packages, it fails with : "warning: Can't wait for pid 1291: No child processes" . Is this a known issue ? <etno>The initial attach works as expected, it reads symbols, and I can set breakpoints. It is when I do "continue" that the above message is displayed, then no breakpoint ever triggers. <etno>In the meantime, I will start to troubleshoot <yang3>Hello is Lenovo t430 a supported laptop? <jab>yang3: If you already own one, then try flashing a Hurd image to an SSD and then boot the thinkpad T430. Let me know what happens! <yang3>ok, i dont own one yet, thinking about the purchase from a friend <yang3>Also my T60 boot seems to be somewhat stuck, I need to check one day, what is bugering it <yang3>I think i can log in, but internet doesnt work <jab>I haven't bought one yet. That should be one that I should try though. <jab>try out the T430 and we'll know <jab>so apparently I have tried it before, and it did not work for me. <jab>maybe I am wrong...I don't think I have tried the T430. <yang3>regarding T60, i have a port base station <yang3>do I need to plug the ethernet into the base or into the laptop? <yang3>error in boot is : task /hurd/netdde(579) looked up a bogus port 82 for 2982 :most probably a bug <yang3>"ifconfig" doesnt seem to work, and "ip a l" neither <jab>yang3: the ethernet should work when you put it into the laptop directly. <jab>try this command (I think this is right) <jab>sudo service restart networking <jab>sudo service networking restart <yang3>it just keeps respawning that error <yang3>I have an entry inside /etc/nertwork/intetrfaces <yang3>I have an entry inside /etc/network/interfaces <jab>yang3: you could try rumpnet <yang3>jab is the last command # settrans -fgap /servers/socket/2 /hurd/pfinet -i /dev/wm0 <jab>-c is to create the filesystem node, but /servers/socket/2 most likely already exists <jab>the last command on that rumpnet page should probably be "sudo service networking restart" <yang3>after search_interface : cannot find wm0 <yang3>./hurd/pfinet : device open on /dev/wm0 (os/device) no such device <yang3>settrans /hurd/pfinet: translator died <jab>maybe rumpnet doesn't support the T60 at the moment. :( <jab>it only supports some ethernet nicks, and netdde should support the T60... <yang3>i think it used to work before, but now idle 2 months and it doesnt <yang3>I'll take it oout of the dock <jab>send an email to bug-hurd <jab>DonjuanPlatinum: nice to meet you too! <yang3>#%$"#&$&#\\\\\\\@@@@ nice, it was a faulty ethernet cable....because I remember it used to work <yang3>so glad that i got it back now <yang3>is there a command that shows battery capacity? <jab>yang3: let me know if you find that command... <yang3>jab going for a huuuuuge upgrade, 350 packages, I hope it won't break