kilobyte | enyc: after Jun 26, there'll be no changes at all | 00:55 |
---|---|---|
enyc | kilobyte: huerm uerr what? | 01:08 |
golinux | I think he's talking about buster | 01:19 |
kilobyte | yeah | 01:21 |
kilobyte | I don't get the point of a delay where any kind of fixes are disallowed -- but it's nice for derivatives to sync release dates | 01:22 |
fsmithred | isn't there an old Russian custom of going back into the house for tea after you packed the carriage for a trip? Gives you time to remember what you forgot. | 01:48 |
onefang | Obvious. You forgot the tea. | 01:53 |
fsmithred | Centurion_Dan, are you around? I'm trying to create a label in gitlab, and it's not working. | 11:46 |
palinuro | hello, any progress on the grub2 efi bug? | 22:23 |
fsmithred | palinuro, did you see what jaromil wrote on the pad on wednesday? | 22:26 |
palinuro | nope, going to read it right now | 22:26 |
fsmithred | jrml- grub2 fix for uefi in beowulf (by a parrot os dev) https://git.devuan.org/palinuro/grub2/tree/debian/2.02+dfsg1-18devuan1 | 22:26 |
fsmithred | oh | 22:26 |
fsmithred | I guess you have seen that | 22:27 |
fsmithred | I don't see it in the repo yet | 22:27 |
palinuro | of course i saw it :) | 22:27 |
palinuro | i was just curious about the timeline to have it aboard | 22:27 |
fsmithred | is it working? | 22:28 |
fsmithred | does it build? | 22:28 |
fsmithred | wanna send me some packages to test? | 22:28 |
palinuro | yes it builds and i am testing it right now in `deb http://deb.parrotsec.org/parrot lts main` (dedicated devuan version) and `deb http://deb.parrotsec.org/parrot stable main` (parrot build) and both work properly respectively on a genuine devuan build and the standard parrot distribution | 22:31 |
palinuro | <fsmithred "wanna send me some packages to t"> https://deb.parrotsec.org/parrot/pool/main/g/grub2/ give the devuan1 packages a try | 22:32 |
fsmithred | yeah, and it occurred to me that the test I want to do requires that they be in a netinstall iso | 22:32 |
fsmithred | but not necessary | 22:33 |
fsmithred | this is for ceres/beowulf? | 22:35 |
palinuro | yes, compiled and tested on beowulf to fix an old buster bug | 22:35 |
fsmithred | ok | 22:36 |
fsmithred | you're talking about the name of the directory in the efi partition? | 22:36 |
palinuro | <fsmithred "yeah, and it occurred to me that"> i told jaromil i am going to rebuild d-i from my experimental beowulf clone, so the new grub2 udebs will be there | 22:37 |
palinuro | if i manage to play with it, i would love to contribute to the official d-i package in devuan as well | 22:37 |
fsmithred | :) | 22:38 |
palinuro | https://bugs.debian.org/908200 | 22:39 |
palinuro | <fsmithred "you're talking about the name of"> yes, it seems to be hardcoded in some way and no longer taken from the distro id | 22:39 |
palinuro | kali was the first distro affected by the bug, in fact the fix comes from raphael hertzog | 22:39 |
fsmithred | one way around it is to use 'grub-install --bootloader-id=' | 22:39 |
fsmithred | =something | 22:39 |
fsmithred | I ran into this a few months ago and I can't remember where it's coded. | 22:40 |
fsmithred | but I did see it | 22:40 |
fsmithred | palinuro, here's a discussion about it: https://dev1galaxy.org/viewtopic.php?id=2676 | 22:43 |
fsmithred | oh, with the signed package you would have to use '--bootloader-id=debian' | 22:45 |
palinuro | part of the "official" solution is to conflict against the -signed packages until we get proper signature support in devuan (or parrot in my case). the remaining part involves getting the proper distro id like in the previous versions | 22:47 |
fsmithred | Here's what I was looking for: https://lists.dyne.org/lurker/message/20190128.163234.70e714cb.en.html | 22:53 |
fsmithred | so what should I expect to happen when I install the -signed package? | 22:57 |
palinuro | not sure | 23:01 |
palinuro | but i think the second part of the hertzog's solution should fix it for -signed packages as well, but i don't know how to build such signed packages since rebuilding grub2 only builds template packages if no valid efi signatures are provided at build time | 23:02 |
palinuro | but i have not looked properly into it, so i may be wrong | 23:02 |
fsmithred | how about a postinstall script that runs 'mv efi/EFI/devuan efi/EFI/debian' | 23:05 |
fsmithred | I guess it would have to be part of a different package | 23:05 |
Centurion_Dan | fsmithred there is dbus ascii-proposed-security package about to drop in /devuan security | 23:13 |
fsmithred | cool, I can try that in a little while | 23:13 |
Centurion_Dan | * /devuan ascii-proposed-security. | 23:14 |
Centurion_Dan | It's missing armel and ppc64el... (the later I will fix for the stable-security and stable builds (Debian also pushed it to stable). | 23:15 |
fsmithred | palinuro, I booted a beowulf live-usb, installed grub-*-bin for amd64, ia32, and 386, installed grub-efi-amd64 and then installed the system to hard drive, let grub make the efi bootloader. | 23:49 |
fsmithred | bootloader dir is "devuan" | 23:49 |
fsmithred | and it boots. | 23:50 |
fsmithred | I didn't try the -signed package because, as you said, it's just the template. | 23:50 |
fsmithred | next test is to make sure I can make a uefi-bootable live-iso. | 23:50 |
palinuro | i think there is no need to rebuild the installer (at least by now) to test if the live/netinstall images are capable of installing a working system | 23:52 |
palinuro | i can confirm that such patches work properly on parrot (that follows debian buster) | 23:52 |
palinuro | we need to test it on beowulf | 23:52 |
fsmithred | the issue I ran into is that I automatically got the -signed package | 23:53 |
fsmithred | when I did the netinstall | 23:53 |
fsmithred | and then I couldn't boot | 23:54 |
palinuro | yes, this is the default behavior of the "broken" debian package | 23:58 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!