| Unit193 | Hrm, any idea when riscv will be available for PPAs? | 00:46 |
|---|---|---|
| sarnold | Unit193: I've got a checkbox on one of my ppas, though maybe I have that due to magic powers | 00:56 |
| Unit193 | My box is all grayed out. | 00:56 |
| sarnold | oh :( | 00:56 |
| Unit193 | Which normally I don't care about riscv at all, but something is failing only there, and for some silly reason it's not migrating. | 00:57 |
| sarnold | Unit193: probably the way to find out if you can have it flipped is to file a question on the launchpad questions section | 00:57 |
| sarnold | Unit193: I think some of the biletto ppas have it enabled too, would that be any help? | 00:57 |
| Unit193 | sarnold: Well...I don't really want special treatment if it's not really ready yet. :3 | 00:57 |
| sarnold | (I've never tried these) | 00:57 |
| Unit193 | I have no idea how to use those either. :D | 00:57 |
| sarnold | hehe | 00:57 |
| sarnold | Unit193: heh, was it failing package tests? https://bugs.launchpad.net/ubuntu/+source/dpkg/+bug/1891686 | 01:24 |
| ubottu | Launchpad bug 1891686 in dpkg (Ubuntu) "Please default to nocheck on riscv64" [Undecided,Fix released] | 01:24 |
| Trevinho | mwhudson: if it helps, this is the diff of what has been installed in gjs (+ proposed) vs the failing one: https://paste.ubuntu.com/p/VDTQpzVWQT/ | 01:40 |
| Trevinho | mwhudson: gobject-introspection amd64 1.64.1-1build2 looks to be a good candidate | 01:40 |
| mwhudson | Trevinho: yeah, sounds plausible | 01:42 |
| Trevinho | mwhudson: you can probably do a build1 with that, or I can but tomorrow | 01:43 |
| mwhudson | Trevinho: well it's also part of the libffi transition so they will migrate together | 01:43 |
| Trevinho | yeah, sure, I mean if we need to have a specific b-d for britney happyness | 01:43 |
| mwhudson | i don't think so | 01:44 |
| mwhudson | thanks for checking though! | 01:44 |
| Unit193 | sarnold: FTBFS. | 02:23 |
| rs2009 | Hi, I noticed that the filesystem.squashfs in the Groovy Gorilla daily builds have /boot/vmlinuz and /boot/initrd.img as broken symlinks. They point to initrd.img-* and vmlinuz-*, which do not exist. I was building a custom Ubuntu ISO from an existing Ubuntu 20.10 daily ISO using the chroot method. I copied the initramfs from the /casper/ directory | 07:04 |
| rs2009 | in the custom disk and installed the linux-image in the chroot, as I needed to change the plymouth logo. I then ran update-initramfs -u -k all inside the chroot. I copied /boot/initrd.img to /casper/ and built the ISO. When I booted the ISO in a VM, it showed the plymouth boot screen for one second, after which it dropped to the BusyBox initramfs | 07:04 |
| rs2009 | prompt. Did something change in 20.10? I'm not aware of any change apart from the switch to GRUB for both BIOS and UEFI. | 07:04 |
| === cpaelzer__ is now known as cpaelzer | ||
| === didrocks999 is now known as didrocks | ||
| ddstreet | Laney i think the autopkgtest armhf lxd-worker needs to be restarted to pick up a worker.conf change, to make haveged a long_test; it is still running it with the default timeout, though haveged was added to long_tests a week or so ago | 12:08 |
| ddstreet | could you check it when you have a chance? | 12:08 |
| Laney | ddstreet: it does, who deployed it? | 12:08 |
| Laney | they might need some education | 12:08 |
| * Laney brings out the education stick | 12:08 | |
| ddstreet | Laney you mean who deployed autopkgtest? this is on the official builders https://autopkgtest.ubuntu.com/packages/h/haveged/groovy/armhf | 12:08 |
| Laney | who merged the config change | 12:08 |
| ddstreet | oh, i think vorlon | 12:09 |
| ddstreet | https://git.launchpad.net/autopkgtest-cloud/commit/?id=5763aed3a3a1c1eb3839bf74263f885ae979527e | 12:09 |
| Laney | right, missed being done on armhf | 12:10 |
| Laney | done now | 12:10 |
| ddstreet | thanks! | 12:10 |
| xnox | rs2009: is this desktop or server? | 15:02 |
| xnox | rs2009: did you also specify environmental variable to trigger casper initrd to be generated with uuid checks? | 15:03 |
| rs2009 | xnox: This is the desktop variant. I wasn't building a daily image, rather was chrooting into the ISO and making some changes, and then rebuilding the ISO. | 15:18 |
| === smoser1 is now known as smoser | ||
| xnox | rs2009: when rebuilding initrd you must set CASPER_GENERATE_UUID environment variable. | 15:19 |
| xnox | rs2009: or have grub.cfg to have boot=casper | 15:19 |
| xnox | in kernel commandline. | 15:19 |
| rs2009 | xnox: I already have boot=casper in the kernel commandline. | 15:20 |
| xnox | rs2009: by default we build initrd for the iso with CASPER_GENERATE_UUID=y set, matching uuid committed on the iso in .disk/, and with clean kernel commandline. | 15:20 |
| xnox | rs2009: in that case i don't know why your boot failed. and you do not give enough information for me to suggest antyhing to you. | 15:20 |
| xnox | rs2009: it sounds like you have missbuilt your kernel/initrd pair, and it doesn't match the rest of the iso. | 15:21 |
| rs2009 | xnox: I was building a custom Ubuntu ISO from an existing Ubuntu 20.10 daily ISO (using the chroot method) when I saw that /boot/initrd.img and /boot/vmlinuz are broken symlinks in filesystem.squashfs. I copied the initramfs from the /casper/ directory in the custom disk and installed the linux-image in the chroot, as I needed to change the | 15:23 |
| rs2009 | plymouth logo. I then ran update-initramfs -u -k all inside the chroot. I copied /boot/initrd.img to /casper/ and built the ISO. When I booted the ISO in a VM, it showed the plymouth boot screen for one second, after which it dropped to the BusyBox initramfs prompt. | 15:23 |
| rs2009 | xnox: the daily ISO has 5.4.0-42-generic (checked in /lib/modules) and the initrd kernel version corresponds to it. | 15:28 |
| rs2009 | xnox: This issue does not happen with the default initramfs, rather with an initramfs built with update-initramfs | 15:33 |
| Laney | rbalint: https://launchpadlibrarian.net/495890903/buildlog_ubuntu-groovy-armhf.mozjs78_78.2.0-1~fakesync1_BUILDING.txt.gz does this interest you? | 15:33 |
| Laney | also, I saw your ping on that autopkgtest-cloud MR, will review this week | 15:34 |
| xnox | rs2009: you should have no need to install linux-image.... as that one might be different from the rest of iso. | 15:34 |
| xnox | rs2009: and you should restore initrd&vmlinuz from casper into where the broken links point to. | 15:34 |
| xnox | rs2009: and you should generate new initrd for that same kernel version, not any other different one. | 15:35 |
| xnox | rs2009: calling update-initramfs without environment variables set, and without correct kernel version as installed in the chroot, will not generate a complete nor valid initrd usable for cd boot. | 15:35 |
| rs2009 | xnox: The issue happens without linux-image as well. I had restored the initrd and vmlinuz to where the broken links point to. I was building the initramfs for the kernel in the ISO only. I did not know that it was mandatory to build with CASPER_GENERATE_UUID=y set. I thought it wasn't required if boot=casper was being added to the command line. | 15:37 |
| xnox | rs2009: and how did you rebuild the initrd? which kernel version did you specify? | 15:39 |
| xnox | rs2009: after you build with CASPER_GENERATE_UUID=y you also need to copy out the same uuid into the iso. otherwise there will be uuid-missmatch and you will fail to boot. | 15:40 |
| rs2009 | xnox: I used update-initramfs and specified 5.4.0-42-generic which I found in /lib/modules/. | 15:41 |
| xnox | rs2009: yeap that's the right version number, so that should have worked. | 15:42 |
| rs2009 | xnox: I'm trying out CASPER_GENERATE_UUID=y right now | 15:42 |
| xnox | generate initrd with UUID, copy it out and place it on the iso in .disk/ dir, remaster the iso and hopefully things will start working | 15:42 |
| xnox | if they don't, edit kernel cmdline and append ignore_uuid to ignore uuid missmatch, to doulbe check if that's what's causing the failure to boot | 15:43 |
| rs2009 | xnox: ignore_uuid works, so it looks to be an issue with the UUID. I couldn't find the UUID in the output of update-initramfs. | 15:59 |
| xnox | rs2009: it's inside the initrd.... one has to unpack it to copy it into the .disk/uuid | 16:33 |
| rs2009 | xnox: Yes, I was doing that :) Thank you for your help! | 16:34 |
| rbalint | Laney, re: MR, thanks! re: mozjs, no, sorry, i'd like to get glibc rounds done first | 19:54 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!