| bcx | doing cp /bin/cat \$DESTDIR/bin | 00:00 |
|---|---|---|
| penguin42 | gee - what do you reckon copy-cat does? | 00:00 |
| TJ- | !! that's one of your own scripts? | 00:00 |
| ubottu | TJ-: I am only a bot, please don't think I'm intelligent :) | 00:00 |
| bcx | yes | 00:00 |
| TJ- | bcx *slapz* | 00:01 |
| TJ- | bcx: but, the line you're showing isn't going to over-write bin/sh | 00:01 |
| bcx | not intended to :) | 00:02 |
| TJ- | bcx: my point is it isn't replacing bin/sh ... so not the explanation | 00:02 |
| bcx | you're right so where do you want i put set -x ? | 00:03 |
| TJ- | bcx: just show me the current complete update-initramfs log | 00:03 |
| bcx | should i revert set -x ? | 00:04 |
| TJ- | bcx: no, I don't mind that, I just want to check the order the hooks are called | 00:04 |
| penguin42 | TJ-: But if they're all hardlinked at that point what happens when the cp copies over cat? | 00:04 |
| TJ- | penguin42: not hardlinked, literally duplicates | 00:04 |
| TJ- | penguin42: as in, not identical inodes | 00:04 |
| penguin42 | hmm | 00:04 |
| bcx | https://paste.debian.net/315312 | 00:06 |
| TJ- | line 923 onwards. confirms its not the last hook | 00:07 |
| TJ- | bcx: you'd best show is the full text of both of your scripts | 00:07 |
| TJ- | s/is/us/ | 00:08 |
| TJ- | OH! there is no "Preferring busybox bin/sh over klibc" | 00:09 |
| bcx | https://paste.debian.net/315313 | 00:09 |
| bcx | sorry https://paste.debian.net/315313 | 00:10 |
| bcx | sorry https://paste.debian.net/315314 | 00:12 |
| bcx | BB_LIST ? | 00:13 |
| bcx | TJ-: https://paste.debian.net/315315 | 00:15 |
| bcx | TJ-: https://paste.debian.net/315316 set -x in zz-... | 00:18 |
| bcx | line 47 ? | 00:20 |
| TJ- | bcx: let's force mkinitramfs -k keep the generated file-system in /tmp/ so we can check what it has created without needing to open the cpio archive. Edit /usr/sbin/update-initramfs, and in generate_initramfs() function, add to OPTS="-o" the -k so you have OPTS="-o -k" | 00:21 |
| bcx | TJ-: no "preferring" because of line 104 | 00:23 |
| TJ- | bcx: yes, I wasn't pointing that out as an error, just a notice since it tells us zz-busybox.. isn't trying to replace an existing bin/sh put there by something else. Which also tells us the fault happens after zz-busybox... runs | 00:25 |
| bcx | ok | 00:25 |
| TJ- | bcx: in case I have different hooks installed here; can you show me "ls -latr /usr/share/initramfs-tools/hooks/" | 00:27 |
| bcx | since -k, update-initramfs fails https://paste.debian.net/315317 | 00:28 |
| bcx | https://paste.debian.net/315318 | 00:28 |
| TJ- | 315317 looks like it ran outside the chroot | 00:30 |
| bcx | is the patch correct ? https://paste.debian.net/315314 | 00:34 |
| bcx | hmmm ... ran from chroot | 00:34 |
| bcx | removing -k succeeds | 00:35 |
| TJ- | that is weird! | 00:35 |
| TJ- | /usr/sbin/mkinitramfs takes -k to set its 'keep=yes' | 00:36 |
| bcx | -k -o | 00:39 |
| bcx | TJ-: -o ARG | 00:40 |
| bcx | fixed | 00:40 |
| TJ- | yup, i figured the same out here :D | 00:41 |
| bcx | so now i get irfs uncompressed with cat instead of busybox | 00:42 |
| TJ- | bcx: now you can do "ls -l /var/tmp/mkinitramfs_*/bin/sh" | 00:42 |
| bcx | https://paste.debian.net/315320 | 00:43 |
| TJ- | right, so now we know for sure | 00:44 |
| TJ- | disable all your custom /etc/initramfs-tools/hooks/ scripts, re-run, check again | 00:45 |
| bcx | got busybox | 00:47 |
| bcx | so it's my scripts | 00:47 |
| TJ- | Yes, had to be! but why? any other scripts besides those you showed us? I can't see how "cp /bin/cat $DESTDIR/bin" would do that | 00:48 |
| bcx | it's copy-cat | 00:48 |
| TJ- | add "set -x" lets find out how | 00:48 |
| bcx | simple | 00:55 |
| bcx | it copies cat to the busybox's cat symlink | 00:56 |
| bcx | cp /bin/cat ..../bin | 00:57 |
| bcx | TJ-: ..../bin/cat already exists and is a link to busybox | 00:57 |
| bcx | so busybox get overwritten with cat | 00:57 |
| TJ- | bin/busybox isn't be overwritten though; bin/sh is | 00:58 |
| bcx | they all get overwritten | 00:59 |
| TJ- | and cp /bin/cat is copying from an absolute path to ${DESTDIR}/bin which should only result in ${DESTDIR}/bin/cat | 00:59 |
| TJ- | bcx, oh! because they're hard linked? | 00:59 |
| bcx | hmmmm | 01:00 |
| bcx | shouldn't | 01:00 |
| bcx | i will add ls -l before | 01:00 |
| TJ- | I thought earlier we tested and bin/busybox --version showed busybox but bin/sh --version showed cat | 01:00 |
| bcx | no it was system's busybox that showed version | 01:01 |
| TJ- | if all the busyboc hardlinks are replaced then they'd all be 52000 bytes long, not 331792 | 01:01 |
| TJ- | bcx: The commands I was giving you were meant to be run against the files in the initrd, not the system root | 01:02 |
| bcx | is that hard links ? | 01:02 |
| bcx | https://paste.debian.net/315322 | 01:03 |
| penguin42 | hard links | 01:03 |
| penguin42 | see the '91' in the link count | 01:03 |
| TJ- | bcx: you'd need to do "ls -il" to see the inode numbers | 01:04 |
| TJ- | oh, you did | 01:04 |
| TJ- | no, you didn't :D | 01:04 |
| TJ- | I'm SO tired I can't think | 01:04 |
| bcx | https://paste.debian.net/315323 | 01:05 |
| penguin42 | yeh, bed is calling | 01:05 |
| bcx | thanks a lot | 01:05 |
| bcx | so i just have to rm before cp and it is ok | 01:05 |
| TJ- | bcx: OK, you can see those are all hardlinks to the bin/busybox... now re-enable your copy-cat script and test again | 01:05 |
| bcx | anyway these hardlinks did not exist with trusty & vivid | 01:05 |
| TJ- | bcx: "cp --remove-destination SOURCE DEST" will work, if this is the cause. Test it first to prove this is it | 01:06 |
| TJ- | I'd like to see proof that the cp is causing this | 01:07 |
| TJ- | in which case you'd need to use that copy_exec() hook function - possibly why its there? | 01:07 |
| bcx | it does | 01:07 |
| bcx | https://paste.debian.net/315324 | 01:08 |
| bcx | --remove-destination should be default in scripts :) | 01:08 |
| bcx | thanks a lot TJ- | 01:09 |
| bcx | very instructive journey | 01:09 |
| TJ- | It's a bit 'unexpected' that 'cp' only moves data into the destination rather than replacing the name+inode | 01:10 |
| bcx | in some cases we want that | 01:13 |
| bcx | here not but i will remeber about using cp --remove-destination in scripts | 01:14 |
| bcx | now my installer supports wily :) good as it is now my main target | 01:15 |
| bcx | user test | 01:15 |
| bcx | thanks again & respect | 01:16 |
| bcx | love to see what open source communities can do | 01:17 |
| bcx | TJ-: btw i would love you to review the installer before public release, will notice you in a few days | 01:18 |
| lordievader | Good morning. | 08:27 |
| BluesKaj | 'Morning all | 10:17 |
| hvoigt | Hi, I am trying to install 15.10 but just found that the usb stick I made does not boot without UEFI mode. Is this a known problem? Where do I find the bugtracker? Google did not really help for 15.10 | 12:26 |
| hvoigt | It boots and works all fine with UEFI enabled (but since my Windows is installed without UEFI I can not install it in UEFI mode) | 12:29 |
| BluesKaj | hvoigt, it won't boot in legacy mode with secure boot disabled ? | 12:30 |
| === ztane_ is now known as ztane | ||
| hvoigt | BluesKaj: Yes I got an error there, but I just tried 15.04 and the first time it did boot but after a restart it failed. So maybe it might be a problem with my stick will check now and report back | 12:45 |
| BluesKaj | hvoigt, ok good | 12:46 |
| hvoigt | ok I tested a second time with a sd-card. Got the same problem: "Does not boot *without* UEFI mode". I get this directly after selecting the non-UEFI option from the boot menu of my board: http://postimg.org/image/yv5ulsycv/ | 13:34 |
| hvoigt | With UEFI mode all seems fine | 13:34 |
| hvoigt | This should be really easy to reproduce since its happens immediately after boot and every time. | 13:36 |
| penguin42 | 10 out of 10 for helpful error messages | 13:37 |
| hvoigt | penguin42: for me? Thanks! | 13:46 |
| hvoigt | So is there anything left I can do? | 13:47 |
| hvoigt | Maybe a workaround so I can proceed with the installation and continue testing? | 13:48 |
| penguin42 | hvoigt: I'm not sure since it doesn't give you much more info | 13:49 |
| hvoigt | BTW, starting 15.04 from the same stick without UEFI works now. Maybe I broke the filesystem before by pulling the stick out to early sometime. | 13:49 |
| hvoigt | penguin42: and unfortunately I do not have any experience with syslinux ... is there a bugtracker or something where this should go to? Or do you just collect bugs internally? | 13:52 |
| hvoigt | would also be interested for something I can watch so I can test once its fixed .. | 13:53 |
| penguin42 | hvoigt: I think you need to file a bug on ubuntu's launchpad - if you've got it booted (via the EFI boot) then you could run ubuntu-bug | 13:53 |
| penguin42 | hvoigt: You'll need an account launchpad.net | 13:53 |
| hvoigt | penguin42: so I should run ubuntu-bug from the livesystem rather than from my current ubuntu installation? I think I have a launchpad account... | 13:55 |
| penguin42 | hvoigt: Yeh, I think best from the livesystem, https://wiki.ubuntu.com/Bugs/FindRightPackage#During_boot suggests it's best to file boot bugs against the kernel | 13:56 |
| BluesKaj | hvoigt, I don't like repeating, but did you disable secure boot when trying to boot in legacy mode? | 13:56 |
| hvoigt | BluesKaj: there is secure boot in legacy mode? I thought that is a uefi thing? I can double check. | 13:58 |
| BluesKaj | yeah, guess it depends on your uefi version | 14:02 |
| BluesKaj | my laptop has a mode called "backflash" that emulates a regular bios | 14:05 |
| penguin42 | never underestimate how broken bioses can be | 14:11 |
| hvoigt | BluesKaj: ah ok | 14:17 |
| hvoigt | I think I found a bugreport about this: https://bugs.launchpad.net/ubuntu/+source/usb-creator/+bug/1499746 | 14:18 |
| ubottu | Launchpad bug 1499746 in usb-creator (Ubuntu Wily) "Early Syslinux Boot Error in pendrive made with the Ubuntu Startup Disk Creator" [Critical,Triaged] | 14:18 |
| BluesKaj | hvoigt, recommend you dd to copy the iso to the usb stick, usb creator was iffy when i tried it | 14:28 |
| BluesKaj | !dd | 14:28 |
| BluesKaj | !info dd | 14:29 |
| ubottu | Package dd does not exist in wily | 14:29 |
| BluesKaj | gawd, what a useless infobot | 14:29 |
| BluesKaj | hvoigt, http://digitalsanctum.com/2007/01/26/how-to-make-an-iso-image-using-dd/ | 14:30 |
| OerHeks | dd is part of coreutils | 15:03 |
| BluesKaj | OerHeks, no matter, the bot should be informed of that and link to it | 15:05 |
| hvoigt | BluesKaj: Thanks for the tip, I ended up creating an usb stick with the creator and using syslinux from 15.04 (by booting the live system) to fix the created usb stick (syslinux --install /dev/sdXX) | 17:09 |
| hvoigt | got that tip from the bugtraq entry saying that one should unpack the syslinux binary from 15.04 to fix it. | 17:09 |
| BluesKaj | well. I suggest you try dd the next time, it hasn't failed me yet, hvoigt | 17:11 |
| hvoigt | Just to be complete I will try this now, since my impression is that it has something to do with the image itself and not the "startup disk creator" | 17:14 |
| hvoigt | BluesKaj: You were right. It works with dd out of the box. | 17:34 |
| hvoigt | Why is the startup media creator always popping up on google? Will just stick with dd for the future. | 17:34 |
| BluesKaj | good thought, dd is scary to some users dunno why, altho it can blank a hdd if misused | 17:38 |
| BluesKaj | google isn't as relaible for linux searches as it should be | 17:39 |
| hvoigt | BluesKaj: yeah I would not suggest dd to many endusers because of the risk to destroy everything if you switch target and source but if you are the think before you type kind of guy, I think its perfectly safe | 18:32 |
| BluesKaj | hvoigt, yeah, it's a pretty simple command, hard to screw up IMO | 18:34 |
| penguin42 | hvoigt: The problem is what else to suggest that actually works | 18:34 |
| penguin42 | BluesKaj: Oh come on, swap an if and an of? Get the sd? one character wrong... | 18:35 |
| hvoigt | I just thought since ubuntu has this startup media creator it is the "more correct" way of doing things, since using dd always like a kind of hack. But IMO a reliable hack is better than a sometimes working tool (especially when just want it to work). | 18:35 |
| penguin42 | hvoigt: Yes but the user has a problem with the thing that it created | 18:35 |
| hvoigt | penguin42: you mean since the stick is unreadable by windows after using dd ? | 18:36 |
| hvoigt | I guess thats true since on Windows its not as easy to manage partitiontables as on linux | 18:37 |
| penguin42 | hvoigt: No, they have boot problems with the stick created using the bootable device creator | 18:38 |
| penguin42 | hvoigt: Oh it's you anyway (checks scrollback) | 18:38 |
| penguin42 | hvoigt: dd is the right thing to use after finding that the normal stuff didn't work | 18:38 |
| hvoigt | penguin42: yeah its a good fallback/workaround definitely | 18:39 |
| BluesKaj | and unetbootin only worked 50% of the tries here | 18:42 |
| penguin42 | and seems to have been getting worse | 18:44 |
| BluesKaj | haven't use it for a couple of yrs | 18:45 |
| BluesKaj | since I discovered dd | 18:45 |
| hvoigt | I only heard that some systems might not recognize the stick since there is no real partition table on them but a cd/dvd filesystem. | 18:52 |
| penguin42 | hvoigt: The headers on them tend to be pretty complex these days, with partitions, filesystems, floppy boot images (often multiple) all to work around different broken firmwares | 18:54 |
| furkan | does anybody else get regular segfaults 15.10 beta? | 21:04 |
| furkan | pidgin, chrome, texworks... | 21:05 |
| furkan | i haven't had any with firefox | 21:05 |
| fhf | i havent got any so far on 15.10 with unity8&mir | 21:05 |
| furkan | you mean unity8/mir are actually in a usable state now? | 21:07 |
| fhf | nope, you have to install normal willy and add | 21:08 |
| fhf | unity8-desktop-session-mir package | 21:09 |
| fhf | and select Unity 8 at LightDM | 21:09 |
| furkan | i see | 21:10 |
| calamari | I'm having trouble with systemd. multiple issues, but the first is network manager. it says dependency failed. how can I figure how what it wants? | 21:55 |
| Faux | systemctl status network-manager.service # might have the details. | 21:57 |
| calamari | specifically, "Unit NetworkManager-wait-online.service has failed" | 21:58 |
| calamari | it just tells me the same thing, but it doesn't tell me what dependency is needed | 21:59 |
| hades08 | anyone could help me with ubuntu-core / snappy and apparmor ? | 21:59 |
| calamari | "Dependency failed for Network Manager Wait Online." | 21:59 |
| hades08 | i want to put apparmor in complain for some lxd container | 22:00 |
| hades08 | and im having issues :s | 22:00 |
| calamari | I guess network manager just doesn't work in wily | 22:19 |
| calamari | I wonder if I can put vivid back on | 22:19 |
| calamari | okay got past the networkmanager problem | 22:54 |
| calamari | tried apt-get --reinstall install systemd and got 5 Error getting authority: Error initializing authority: Could not connect: No such file or directory (g-io-error-quark, 1) | 22:55 |
| calamari | then it hangs | 22:55 |
| calamari | oh good after ctrl-z and kill -9'ing the last one, at least ctrl-c works now | 22:56 |
| calamari | can anyone help? I'm totally screwed now, can't even install packages due to this systemd garbage | 23:09 |
| calamari | it hangs every time | 23:09 |
| fhf | calamari: systemd is quite basic now so i dont think it can be reinstalled right away, you can always override apt moanings be removing some files which store dep | 23:15 |
| fhf | i gtg now so mby others can help, gn folks | 23:15 |
| calamari | here is a screenshot http://s24.postimg.org/bv8fodwvp/1444605589763.jpg | 23:21 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
