[00:55] hi [00:59] i tried to build a rootfs for qemu but everytime i get a Kernel panic - not syncing: Attempted to kill init! [01:02] How are you trying to build the rootfs? [01:03] like in https://wiki.ubuntu.com/ARM/RootfsFromScratch [01:04] sudo rootstock --fqdn ubuntu --login ubuntu --password ubuntu [01:05] and then mkdir /tmp/arm [01:05] sudo mount -o loop /tmp/arm.img /tmp/arm [01:05] cd /tmp/arm [01:05] sudo tar xzvf ../qemu-arm-rootfs.tgz [01:05] cd .. [01:05] sudo umount /tmp/arm [01:05] rm /tmp/arm [01:06] qemu with [01:06] sudo qemu-system-arm -M versatilepb -kernel /tmp/vmlinuz-2.6.28-versatile -cpu arm926 -hda /tmp/arm.img -m 256M -append "root=/dev/sda mem=256M" [01:08] That certainly looks right. [01:08] * persia tries [01:11] oh and qemu-img create arm.img 1G [01:11] sudo mkfs.ext2 -F arm.img [01:11] I seem to be bumping into a bug that prevents rootstock from running at all today. Sorry. [01:11] Someone else may have to help. [01:13] thanks anyway! [01:22] the rootfs from http://people.canonical.com/~ogra/arm/qemu/qemu-arm-rootfs.tgz works.... [01:35] i give up for today [01:35] bye [06:30] interesting... bootchart doesn't do anything on arm. [06:33] Nor does ureadahead. [06:33] Probably doesn't need to do anything, though. [06:50] wait, ureadahead should work! [06:50] it should be prepopulating the cache [06:51] Did we forget to add the patch into the arm branch of the kernel? [06:56] Then again, my ureadahead doesn't work on my host amd64 system, either... terminates with status 4. [06:56] =รพ [06:56] So maybe I'm just unlucky. [06:56] hmm, /var/lib/ureadahead/ has just debugfs; no "pack". [06:56] Also, my thing isn't using an initramfs, and update-initramfs does nothing. [06:58] http://pastebin.com/fc125c14 [06:59] that is: bash -x `which update-initramfs` -u -k all [07:00] Yeah, I'm only just starting to work with initramfs/initrd for arm server [07:00] it will be a while before I walk the entire chain [07:01] Plus, initramfs needs u-boot to know what to do with it, also. [07:01] As it is, the kernel has to come from elsewhere -- I went here: http://rcn-ee.net/deb/kernel/beagle/ [07:07] Are there any plans to have standard in-repo kernels for various boards? [07:10] no [07:10] at least, from what I saw, only the supported processors on their test platforms for now [07:10] Marvell Dove [07:10] etc [07:10] etc === jitesh_afk is now known as jiteshs === Meizirkki_ is now known as Meizirkki [13:01] does anyone have additional ideas what should be on the rootstock gui for its first iteration ? http://people.canonical.com/~ogra/rootstock-gui.png [13:01] asac, ^^^ ? [13:01] ogra: make the image size typable ;) [13:01] pfft [13:02] ogra: whats the "run tasksel" button [13:02] well, beyond that, any options you would like to see ? [13:02] if you click there it instantly runs tasksel? [13:02] it fires up the tasksel gui [13:02] so right when clicking? [13:03] http://people.canonical.com/~ogra/rootstock-tasksel.png [13:03] its a separate window [13:03] k [14:45] asac, http://people.canonical.com/~ogra/rootstock-gui.png more happy with that ? [14:46] ogra: can one type [14:46] (the file menu has a save profile/load profile as well) [14:46] indeed [14:46] well, you will, thats all mockup atm [15:04] ogra: seed url doesnt work? [15:04] no, thats not in the script yet either [15:04] kk [15:04] something for L+1 [15:05] image size -> implies we create a real image now? [15:05] no, should i call it rootfs size rather ? [15:05] rootfs max size [15:05] ? [15:05] why do we need that? cant we do an estimate? [15:06] its one point i would even like to automate, but that smells like a huge amount of code to compute it in advance [15:06] guess not ;) [15:06] right [15:06] also L+1 :) [15:06] but cant we say: 4GB by default? [15:06] no, people that do ubuntu-minimal will only neeed 500MB [15:06] but that space isnt wasted [15:06] i dont want to waste more diskspace than needed [15:06] its just temporary right= [15:07] ? [15:07] its used during build [15:07] right [15:07] so your host definately needs that amount [15:07] we can take as much as the file system has ;) [15:07] no, that would strictly bind to the filesystem [15:07] run a check before build if /tmp partition has more than 4GB ... if so acquire it, other pop up a dialog and ask user to select a smaller guess ;) [15:07] i want to preserve the funstionallity that you can use the script without a FS [15:08] without a fS? [15:08] FS? [15:08] without a target partition [15:08] how can a rootfs work without a filesystem? [15:08] the script should still go on to produce tarballs [15:08] i dont think i understand you. keep the option ;) [15:08] as well as the gui ... [15:08] i know [15:09] only if you fill in the bottom option to write to target partition we actually have a size [15:09] i am saying. if your local /tmp partition has more than 4GB size, use that as image size [15:09] otherwise ask :) [15:09] but i see that its not accurate [15:09] maybe "grow-on-demand"? [15:09] guess thats harder [15:10] yes, thats quite hard the stages are to disconnected [15:10] i would go with a hard 4G for the gui ... [15:11] given that when you run the gui you really should make sure to have enough build space [15:11] anyway. i think we should set the image size to a reasonable default and move it from the top [15:11] i think i just care so much because i would expect the top most option to be quite important [15:11] the prob is what happens if a user is insane enough to pick kubuntu-debsktop, ubuntu-dekstop, xubuntu-desktop at the same time [15:11] right [15:11] lets move it down [15:12] and use a reasonable default (e.g. 4 or 2 GB) [15:12] 4 then [15:12] thats definately safe for most operations [15:12] also name it (build space) or something [15:12] with a minimal size coded in the spinbutton [15:12] i.e. 0.5 G [15:13] i dont care about minimal size ;) [15:13] well, the actual build space is a bit more than the image size [15:13] doesnt matter ;) ... name it: rootfs build area [15:13] (kernel Packages file etc... but that should be in the megabytes [15:13] ok [15:14] what is in datei ? [15:14] in the menu? [15:14] load profile/save profile and quit [15:15] ok [15:15] to load/save the set of options you picked [15:28] asac, reload :) [15:31] ogra: good ... maybe in (TEMP) in brackets [15:31] ;) [15:31] but its ok [15:32] yeah, thats easy to add when i code that stuff ... its currently all glade to get a grasp of the layout i want [15:33] for mockups there is nothing better than glade :) [15:33] oh wasnt there the idea to also allow adding .deb files? [15:34] yes, thats for later though [15:34] k [15:34] only one additional mirror? [15:34] or a list of? [15:35] one for now [15:37] the mirror stuff can definately need some extra love later (i'll turn that into post FF bugs) currently it needs to be a properly set up ubuntu archive [15:37] i'd like to make it handle PPAs as well but that needs some extra gpg fiddling in the code thats not there yet [15:47] i think its ok ;) [15:47] just wondered if a syntax like in pbuilder is allowed [15:48] that allows you to specify more than one mirror in one line iirc [15:49] well, that can surely be added... just a check for comma in the argument and split accordingly [15:49] heh, i just wanted to point jamie to commanet #3 on http://ograblog.wordpress.com/2010/02/15/the-shiniest-and-fastest-arms-ever-in-ubuntu/ [15:49] i guess he knows why he just disconnected :P [16:36] Hi guys, can you remember who was looking at the Thumb-2 issues in gmp? Was it dyfet? [16:36] * ogra thinks so [16:37] but you might not catch him today (national holiday in the US) [16:40] dmart: According to https://bugs.launchpad.net/bugs/513732 he was assigned shortly [16:40] Launchpad bug 513732 in gmp (Ubuntu Lucid) (and 1 other project) "[arm] needs porting to thumb2 (affects: 1)" [High,Triaged] [16:41] thanks [16:48] * jmc93739653 finds himself in the role of a sterotypical American, as he didn't know it _was_ a holiday in the US today. [16:50] bank holiday though [16:50] most businesses are open