IRC channel logs
2026-06-07.log
back to list of logs
<rsmarples>sam_: i have a pfinet init script that starts with the 0.0.0.0 address, dhcpcd starts afterwards and then configures the address on it using fsysopts <rsmarples>now, the question is, do we want the pfinet script to create the /dev/eth0 node if it doesn't exist? <sam_>I feel like yes, at least for now, it's not clear to me what else would create it <sam_>others might disagree though <rsmarples>MAKEDEV eth0 reports `settrans: eth0: (ipc/mig) bad request message` <rsmarples>MAKEDEV eth0 reports `settrans: eth0: (ipc/mig) bad request message ID` <rsmarples>so i managed to mess something up as runttys doesn't finish <rsmarples>ok, is runttys supposed to be a daemon or not? <rsmarples>ok, we now boot cleanly and sshd and the rest of the services now autostart <rsmarples>which means i can start the vm and remotely login without needing to dick on the console <rsmarples>so in summary ... pfinet creates /dev/eth0 and starts on the 0.0.0.0 address, dhcpcd then negotiates DHCP and configures pfinet with the real address and sshd sees this and just works. magic <sam_>this maks a lot of other things so much easier <sam_>being able to ssh in w/ no prior interaction is great <rsmarples>helps a lot as the # key won't work over VNC which hampers writing any code. <rsmarples>that's somethig hurd side not recoginisng my apple keyboard i think <rsmarples>sam_: I'll try and ship you my init.d changes for pfinet and runttys tomorrow <rsmarples>well, termbin is not working for me on port 9999. so /sigh <sam_>thanks, look forward t oit <rsmarples>then we just need to add the network provision to the dhcpcd use in it's init.d script <rsmarples>alr, reduced my dhcpcd diffs to just headers now <sam_>rsmarples: thanks. working on some other bits today but will get to it later or tomorrow, remind me if i forget <rsmarples>if not done by tuesday I'll have to send round "The Boyz" <rsmarples>ok, dhcpcd now fully works in a chroot on hurd as well, so nice and secure now <rsmarples>sam_: let me know when you finish the that glibc bug for me as that's now my show stopper for a release <rsmarples>the header dance with if_ether i may have to live with sadly <bjc>rsmarples: do you have those patches available publically yet? <rsmarples>bjc: aside from the hurd specific parts, it's all in my public github repo <rsmarples>once sam has fixed some defines i can reduce a lot of crud and finally commit something hurdish i think <diegonc>I guess the chroot needs internet access, have you linked the pfinet files ( /servers/socket/2 )? <diegonc>I had to do something like this for openntpd <diegonc>settrans -cp /var/run/openntpd/servers/socket/2 /hurd/firmlink /servers/socket/2 <diegonc>settrans -cp /var/run/openntpd/servers/socket/26 /hurd/firmlink /servers/socket/26 <rsmarples>nah, I'll just use my ipc to contact the non chroot process to work it out <rsmarples>does hurd support IP_PKTINFO socket options? <rsmarples>guessing that might be a lwip vs pfinet thing <rsmarples>hmmm, i see code in hurd github for it in pfinet <youpi>Pellescours: I can't parse "the libc read hang pflocal call return (do not hang)" <youpi>do you mean that the read() libc function returns, while it should not? <youpi>(when the socket is shut down, read() should indeed return 0, not hang) <gnucode>hmmm, elpa-magit isn't working, but install magit from inside emacs does work.