| * ogra wonders what lool thinks about https://wiki.ubuntu.com/ARM/BuildEABIChroot :) | 13:35 | |
| Stskeeps | ogra: that's cool :) did you include debootstrap support for that or? | 13:39 |
|---|---|---|
| ogra | no, just a wrapper script that calls first and second stage separately and sets --arch armel | 13:39 |
| suihkulokki | A patch for qemu exists at http://qemu-arm-eabi.sourceforge.net/ to enable EABI support in qemu-arm | 13:40 |
| suihkulokki | <- thats so outdated :P | 13:40 |
| ogra | right | 13:40 |
| ogra | suihkulokki, ?? | 13:40 |
| ogra | is there a newer one ? | 13:40 |
| ogra | (that applies to 0.10.x) | 13:41 |
| suihkulokki | EABI is supported even before 0.10.x | 13:41 |
| ogra | weird | 13:41 |
| Stskeeps | ogra: ah, but it copies in the qemu-arm static binary into the target temporarily i presume | 13:42 |
| ogra | right | 13:42 |
| ogra | not temporarily, else you couldnt chroot :) | 13:42 |
| Stskeeps | ah yes | 13:42 |
| === cbrake_away is now known as cbrake | ||
| ogra | suihkulokki, i find it intresting that i could never use qemu-arm properly then ... given that we had 0.10 in jaunty already | 13:47 |
| suihkulokki | ogra: not really interesting.. just different (fixed in git head) issue | 14:09 |
| ogra | ah | 14:10 |
| suihkulokki | using git head highly recommended at the moment, as it will turn 0.11 any time now | 14:11 |
| lool | ogra: I think that it's nice but not good enough for package builds | 14:29 |
| ogra | lool, for home use its surely good | 14:29 |
| ogra | my mesa build finished in about 2h | 14:29 |
| ogra | on my laptop | 14:30 |
| lool | I just don't trust it | 14:30 |
| ogra | so if you want to do a pbuilder testbuild on armel before uploading etc thats surely a good way | 14:30 |
| ogra | i wouldnt use it as a buildd :) | 14:30 |
| lool | Right | 14:30 |
| ogra | and i dont know how it will behave with neon and v7 | 14:31 |
| === amitk_ is now known as amitk | ||
| lool | That's not an issue | 14:32 |
| ogra | its surely better than cross compiling the whole dep chain of gnome to testbuild a single gnome package in a cross compiler env | 14:32 |
| lool | syscalls don't map one to one on armel and x86; I wouldn't trust that they are all properly wrapped and you might end up building packages which assume support for stuff which isn't truly supported or vice-versa | 14:33 |
| ogra | yeah, its surely not for full production use | 14:33 |
| ogra | but its *faaaaast* :) | 14:34 |
| robert__ | hello! I'm trying to build an arm-rootfs using rootstock (tried the older script from wiki, too), but it just hangs after starting qemu saying "installing core packages" | 15:36 |
| ogra | it doesnt hang | 15:36 |
| ogra | its just very slow inside the VM | 15:36 |
| robert__ | really? how long does it do "nothing" (or at least saying nothing"? | 15:37 |
| robert__ | I just got curious after half an hour or so | 15:37 |
| ogra | give it time ... building the base system completely usually takes about 50min | 15:37 |
| ogra | the apt stage later is faster | 15:37 |
| robert__ | oh, okay. maybe on my host it will take some hours :) | 15:37 |
| robert__ | I will wait. thanks a lot | 15:38 |
| ogra | the 50min are on my laptop ... core2duo 1.8GHz | 15:38 |
| ogra | slower machines will indeed take a bit longer | 15:38 |
| robert__ | okay. well, another question. what's a minimal system just providing X? I am currently trying xubuntu-desktop | 15:39 |
| ogra | minimal system is really minimal ... | 15:40 |
| ogra | just enough to boot | 15:40 |
| ogra | if you only want boot and X add xorg | 15:41 |
| ogra | a lightweight desktop would be lxde | 15:41 |
| ogra | there are some examples on the RootfsFromScratch wikipage | 15:42 |
| robert__ | hm, I think I'll stick with xubuntu-desktop first, and maybe remove unneeded things later | 15:44 |
| robert__ | the created image should be bootable on an OMAP-board, right? | 15:48 |
| ogra | its just a tgz | 15:51 |
| ogra | yes, the packages should all run on OMAP | 15:51 |
| robert__ | okay, I'll try it later when the system is fully configured | 15:52 |
| robert__ | ogra: The script did not take too long, it booted the system a few minutes ago and is unpacking the packages now | 16:46 |
| ogra | ah, good | 16:46 |
| robert__ | ogra: Do you know what compiler was used to build the armel packages? | 16:52 |
| ogra | gcc | 16:53 |
| ogra | depending on which release you use 4.3 or 4.4 | 16:53 |
| robert__ | okay, thanks. that should fit with the other applications I plan to use | 16:54 |
| robert__ | ogra: "--seed build-essential,openssh-server" took about 15 minutes to build the full rootfs on an core2@2,66GHz | 19:10 |
| ogra | sounds about right | 19:11 |
| robert__ | but the system does not boot, last messages is: " * Starting OpenBSD Secure Shell server sshd [ OK ] | 19:12 |
| robert__ | chown: failed to get attributes of `/var/log/dmesg': Stale NFS file handle | 19:12 |
| robert__ | chmod: failed to get attributes of `/var/log/dmesg': Stale NFS file handle | 19:12 |
| robert__ | /etc/rc2.d/S70bootlogs.sh: 61: cannot create /var/log/dmesg: Stale NFS file handle | 19:12 |
| robert__ | chgrp: cannot access `/var/log/dmesg': Stale NFS file handle | 19:12 |
| robert__ | mv: accessing `/var/log/udev': Stale NFS file handle | 19:12 |
| robert__ | " | 19:13 |
| robert__ | I don't understand why it seems to look for stuff on NFS at all... | 19:13 |
| * ogra points to https://bugs.launchpad.net/ubuntu/+source/glibc/+bug/391094 | 19:13 | |
| ubot4 | Launchpad bug 391094 in glibc "the "stale NFS file handle" error should be reworded" [Undecided,New] | 19:13 |
| ogra | when did you check out the script ? | 19:13 |
| robert__ | an hour ago or so | 19:14 |
| robert__ | rev17 I think | 19:14 |
| ogra | there was a bug where it was not properly unmounting the VM image | 19:14 |
| ogra | you could do an fsck ... but its worrying that you still see it | 19:15 |
| ogra | did you build for jaunty or karmic ? | 19:16 |
| robert__ | I built jaunty | 19:16 |
| ogra | hrm, i checked that, thats bad | 19:17 |
| robert__ | the build was absolutely fresh, first boot | 19:17 |
| ogra | yeah | 19:17 |
| robert__ | and I unmounted the image to where I unpacked rootfs correctly | 19:17 |
| ogra | it didnt unmount the VM cleanly and then rolled a tarball that contained uncleared inodes | 19:17 |
| ogra | you will see stuff being mentioned in the logfile i guess | 19:18 |
| ogra | the best you can do is an fsck | 19:18 |
| ogra | it will clean up the dangling inodes | 19:19 |
| ogra | i will look deeper into that on the weekend | 19:19 |
| ogra | i'm 100% sure its fixed in karmic, but i only tested one run under jaunty | 19:19 |
| robert__ | the image had wrong inodes, strange | 19:21 |
| ogra | not if you think about it :) | 19:21 |
| robert__ | because qemu was killed while the fs was mounted? | 19:22 |
| ogra | the image is built in a vm ... the vm isnt shut down properly ... then the image is mounted again and a tarball is created from the content | 19:22 |
| robert__ | but it still hangs with " * Starting OpenBSD Secure Shell server sshd [ OK ] | 19:22 |
| robert__ | chown: failed to get attributes of `/var/log/dmesg': No such file or directory | 19:22 |
| robert__ | chmod: failed to get attributes of `/var/log/dmesg': No such file or directory | 19:22 |
| robert__ | " | 19:22 |
| ogra | you can create that file it shouldnt stop you from booting | 19:23 |
| ogra | -rw-r----- 1 root adm 50862 2009-07-17 13:44 /var/log/dmesg | 19:23 |
| ogra | should look like that | 19:24 |
| robert__ | well, but that should not stop the boot process, right? | 19:24 |
| ogra | right | 19:24 |
| * ogra has to go soon | 19:24 | |
| Sarvatt_ | hmm 15 minutes? are you using amd64 or i386 on it robert__? its still around 3.5 hours for a build-essential,openssh-server karmic on a 2.4ghz turion x2 herex2 for some reason | 19:25 |
| robert__ | Sarvatt_: maybe 20 minutes, but not more | 19:26 |
| Sarvatt_ | debian is 20 minutes though | 19:26 |
| ogra | amd64's qemu-syste-arm might run slower | 19:26 |
| robert__ | it's i386 | 19:26 |
| Sarvatt_ | yeah thats what imthinking | 19:26 |
| Sarvatt_ | wow, cant type today :) | 19:26 |
| ogra | anyway, i really have to go now ... i'll look into that tomorrow | 19:27 |
| robert__ | ogra: strange, /var/log/dmesg exists in the image | 19:27 |
| Sarvatt_ | thats a pretty crazy difference, i'll try it on i386 | 19:27 |
| robert__ | all right, I'll try to get it running | 19:27 |
| ogra | if you cant get it fixed, worst case use build-arm-rootfs for jaunty | 19:28 |
| robert__ | I will try that, too. just 15 minutes ;) | 19:28 |
| robert__ | ah, I tried jaunty. you mean k... (forgot the name :) ) | 19:29 |
| robert__ | can karmic only be run with cortex-a8 as cpu? I need to use is for an 926 | 19:32 |
| Sarvatt_ | you can use it for that now but its going to change soon to be armv6+,, would need to change rootstock around a bit because i think ogra moved it over to cortex in there | 19:35 |
| robert__ | yes, just a line to change. I was just wondering if the binary format changed for karmic | 19:39 |
| Sarvatt_ | not yet | 19:39 |
| Sarvatt_ | still armv5t arch default for gcc in karmic | 19:40 |
| robert__ | by the way, what do you use ubuntu/arm for? are there tablet pcs strong enough to run ubuntu on yet? | 19:42 |
| Sarvatt_ | i wish! | 19:43 |
| Sarvatt_ | i'm just using it for a build environment for a future system and to check for problems in pixman from git here | 19:44 |
| robert__ | so it's basically a "proof of concept" project at the moment? | 19:44 |
| Sarvatt_ | qemu is running so slow on my amd64 machines that i've resorted to using karmic in a chroot on my htc dream phone for all of that :D | 19:46 |
| robert__ | I'm planning to profile some applications on real hardware (OMAP developer board) | 19:47 |
| robert__ | hm, I have an htc himalaya laying around here, but never managed to run linux on it | 19:47 |
| robert__ | just managed to boot a kernel | 19:47 |
| Sarvatt_ | debian might be a better fit for both of those, jaunty and karmic arm are using arch armv5t with cortex-a8 tune for everything and i'm just guessing that would make things slower on the older hardware | 19:52 |
| robert__ | is there a list of CFLAGS used? | 19:54 |
| robert__ | ogra: karmic rootfs also hangs after "* Starting OpenBSD Secure Shell server sshd [ OK ]" | 20:11 |
| === bizkut-miau is now known as bizkut | ||
| robert__ | I tried removing the symlinkgs in /etc/rc2.d which started services after sshd, without success | 20:18 |
| robert__ | I'll continue tomorrow. see you! | 20:18 |
| Sarvatt_ | ogra: holy crap, build-arm-eabi-chroot is great! thanks for packaging that! | 20:43 |
| === Omegamoon is now known as Omegamoon|kraftw | ||
| === Omegamoon|kraftw is now known as Omegamoon | ||
| === cbrake is now known as cbrake_away | ||
| === Sarvatt_ is now known as Sarvatt | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!