[00:00] Does it allow me to tweak the kernel conf ? [00:01] you pick a kernel deb you want to base it on, so if you need to tweak the kernel you need to compile it and package it [00:02] ok [00:03] and then it outputs a rootfs [00:03] topfs2, because its a uimage you can't just use make-kpkg kernel_image [00:03] ? [00:03] Well, I don't use uimage for now, or I'm not aware i'm using it [00:04] I just discovered that my tablet boots always on a rootfs.img, so i substituted the armel image or ubuntu 10.10 and it ran ok (appart for usb and touchscreen support) [00:04] Samae, which tablet? [00:05] up there ^ [00:05] Archos 101 it [00:05] ah, archos. [00:05] okahy [00:05] Ow, I meant I don't use uboot [00:06] but yeah it boots on zImage or uImage apparently [00:06] so next step, new rootfs and new kernel with proper modules [00:06] :) [00:07] I'll be back [00:07] gudnite [02:52] ogra: the binary is the same name, but different path [02:52] ogra: the new package for omap3 is x-loader-omap3-beagle [02:52] as we're generating images compatible with beagle by default [02:53] ogra: /usr/lib/x-loader/omap3530beagle/MLO === Amaranth_ is now known as Amaranth [03:50] rsalveti: still around. can you boot up latest 2.6.38-rc2 on panda? === Baybal_ is now known as BAYBAL === Amaranth_ is now known as Amaranth [07:52] ok got a different situation here, the board that i am building a rootfs image for only has 320mb for its rootfs, does ubuntu arm support me splitting in to /usr /home /var / et al [09:03] lilstevie: generate tarball with rootfs and then split it manually into parts [09:04] hrw, does the maverick image use a initramfs? === ka6sox is now known as ka6sox-away [09:08] yes, it does [09:09] hmmm..okay. I might have to do some playing then to get it to work. [09:09] hopefully it uses udev [09:31] hrw: then just make sure the other parts are mounted in the initramfs right? [09:33] probably [09:33] I use only / /home + /tmp in tmpfs on most of my machines [09:34] heh problem i have is that i only have 320mb in the available space for / [09:34] and the rest is split on SD cards [09:34] internal 16gb of which i can only flash 1.5GB of ext [09:42] i doubt loading off /var will gain you enough, you should offload /usr instead [09:43] as long as you use an initrd adding a line to fstab should be enough, no extra fiddling needed [09:48] ogra: but will that still give me enough space to be able to install stuff [09:58] well, thats a trial and r thing i guess [09:58] *error [09:59] of my installed 4G big system /usr eats 1.9G while /var occupies 350M [09:59] so sharing out /usr will definitely gain you more [10:00] 6.3GB /usr 8/2GB /var on my desktop [10:00] 6.3GB /usr 8.2GB /var on my desktop [10:01] hrw, apt-get clean ;) [10:01] ogra: 4GB is pbuilder [10:01] loading off /var surely gains you a lot too in the running system later [10:02] (logs, caches etc) [10:02] but yes, 800MB was apt cache [10:02] but for initial install if you are short on space /usr should be the first [10:03] and /tmp in tmpfs is something you should only do with a decent amount of ram [10:03] at least if you want to run apps like firefox [10:03] i wouldnt do that on 128M if you want to run any desktop bits [10:03] (for example) [10:07] I have 512mb ram usable [10:08] the SoIC has more ram but i dont know how to access it [10:09] and this is for arm :p does firefox have an arm port :p [10:12] port? [10:12] firefox nowadays just need to be compiled [10:12] heh fair enough [10:13] I remember days when anything mozilla based required strange patches to get it running [10:13] still trying to grasp how much will run on arm [10:13] lilstevie: 512MB ram? most will run [10:13] generally what is the best x manager to run in a touch environment [10:13] running firefox on arm is absolutely no problem. compiling locally still is a bit of a strain, though. [10:14] lilstevie: which arm cpu you have? [10:14] LetoThe2nd, depends on your hardware ;) [10:14] hrw: yeah it has 512mb ram, and some other ram that is really not known if it is usable [10:14] hrw: hummingbird [10:14] nice cpu [10:14] yeah :) [10:14] lilstevie, i use an ac100 here atm, should all work fine [10:15] just hoping to free it from android hell :p [10:15] ogra: adequte relation to HW given, of course. running ff on a pentium133 is also rather pointless. [10:15] lilstevie: so galaxy s/tab/nexuss? [10:15] for a tablet like device is not nice [10:15] tab [10:16] and i wasnt dumb enough to upgrade to one of the dev releases that has locked bootloaders so it just happily runs unsigned kernel/initramfs combos [10:16] has to be the most open device i have ever had lol [10:16] ogra: on an openrd here ff worked quite good. :-) [10:16] bb in ~15minutes [10:21] heh.. someone remind about someone before leaving... [10:22] * hrw uses hummingbird to play youtube videos now. [10:22] lol [10:22] but on nexus s/android [10:22] the tab is nice hardware [10:22] lilstevie: tab has 1024x600? [10:22] yep [10:23] 16GB storage, 512MB ram, netboot resolution... looks like good target for ubuntu [10:23] yeah thats what i though [10:23] but I do not know would I switch [10:24] only thing I am going to struggle with is the initramfs [10:25] make sure you have your kernel and modules properly installed in the rootfs [10:25] then just run update-initramfs -c -v [10:25] lilstevie: on arm devices I just avoid initramfs and boot directly into rootfs [10:26] hrw: that is a tad difficult on the tab [10:26] as the kernel gets the initramfs compiled in [10:26] as hw is always same, rootfs is same device etc [10:26] you lose some functionality without initrd [10:27] (filesystem check, encrypted filesystems etc) [10:29] ogra: update-initramfs from inside qemu or just in the rootfs [10:30] either will work, it creates a /boot/initrd.img- [10:34] ll /mnt/` [10:34] oups :) [10:35] SGX535 has full accel on ubuntu yes? [10:39] ogra: since when fsck needs initramfs? [10:39] ogra: fsck needs rootfs/ro [10:39] it happens before / is mounted rw [10:39] ogra: and for this initramfs is not required [10:39] well, i dont have fsck on systems where i dont use initrd [10:40] ogra: maybe thats because ubuntu forces initramfs to be used to fsck [10:40] might be, i never checked ... since we use initrd by default on all official images anyway [10:50] ogra: How can I submit patches to rootstock? [10:51] sveinse, create your own branch on LP and file merge requests [10:59] speaking of merging.... [10:59] ogra: does the kernel need to be in the rootfs [10:59] flash-kernel? [10:59] lilstevie, the modules do [10:59] i'm not sure update-initramfs checks for a vmlinuz file [10:59] worst case you can just touch it though [11:00] ok good cause my kernel isnt compiled until after i have the initramfs :p [11:00] hrw, yeah yeah ... did drop off my TODO, there are just other more pressing things atm [11:00] *didnt (indeed) [11:04] I saw some comments above regarding initrd: What specifically is done in initrd? Because my (rootfs based) system does not use initrd at all, but it seems to work as intended [11:04] ...I meant roostock based system [11:07] sveinse, make sure your users never ever install any filesystem encryption bits, raid or device manager ... all these need initrd [11:08] beyond that initrd indeed draws the splash and runs plymouth (whithout which fsck will not work since mountall uses libplymouth for communication) [11:11] so ogra let me get this right, best way to do it will be set up fstab, move modules in to place chroot to the rootfs then run update-initramfs [11:11] ogra, ok thanks. Quite frankly it good to be without (except for fsck). I have a strict customer requirement of short bootup time. [11:12] I see the PID counter is what around 3000, so it does alot at boot even without initrd === nslu2-log_ is now known as nslu2-log === ka6sox is now known as ka6sox-away === ka6sox-away is now known as ka6sox [11:54] hi all ... does anyone know how to get verbose log output during bootup, preferably over serial? I don't really know to to achieve this with upstart+plymouth [11:55] dmart: remove quiet from cmdline and add nosplash? [11:55] might not be enough though [11:55] no need for nosplash on serial though [11:55] dropping quiet should be enough (and indeed the right console= args) [11:55] dmart: For serial, it's a matter of listing your serial console last if you want it to be the system console IIRC [11:56] That is, console=ttyS2, console=tty0 will output some kernel messages on both, but /dev/console will only go to last arg, or something like that [11:56] just make it the only console [11:56] no need for tty0 if you want serial boot output [11:57] egh is there anthing that stops 32bit linux binaries executing on 10.10 64bit? [11:58] for some reason it keeps telling me file not found for the executable when it is clearly there [12:04] lilstevie: What's the exact message? [12:05] No Such File or Directory [12:05] it clearly exists [12:05] lilstevie: The full message [12:05] lilstevie: Usually, it's missing ld-linux for 32-bits; it's in a separate package [12:06] lilstevie: You want to install libc6-i386 [12:06] lilstevie: You can strace your binary to see which actual file is missing [12:06] /opt/toolchains/arm-2009q3/bin/arm-none-linux-gnueabi-gcc: no such file or directoy [12:07] directory* [12:07] lilstevie: Ok; that sounds like missing 32-bits ld-linux; run strace -etrace=file /opt/toolchains/arm-2009q3/bin/arm-none-linux-gnueabi-gcc [12:09] http://pastie.org/1492398 [12:10] lool: I don't think we're using quiet or splash... this is just a BSP bringup problem. [12:11] lool: installing libc6-i386 fixed :) [12:43] hm, ogra i get an error off that initramfs telling me it is the wrong datatype [12:44] cause it is aparently not cpio [12:55] its gzipped cpio [12:57] Hmm. It seems getting rootstock to use more than "maverick" (e.g. maverick-updates and maverick-security) isn't as trivial as I hoped it would be. It seems debootstrap doesn't do a full "apt-get" resolution of the latest package. It seems to be locked against maverick/main. Objections? [12:58] yeah i figured it out [12:58] i ended up checking it with file [12:58] discovered that it was gzip [12:58] was a bit of a facepalm moment :/ [12:59] So perhaps it's better to rootstock/debootstrap and then do a apt-get upgrade after installation to get the maverick-security and/or maverick-updates [13:03] or properly patch rootstock to generate a proper sources.list [13:04] i.e. by installing and using python-apt [13:08] ogra, I've added a --sources option to rootstock where you can provide your own sources.list [13:08] However, debootstrap internals doesn't consider anything else than maverick/main [13:13] sveinse, you could prot rootstock to multistrap ;) [13:28] hi [13:30] In http://cdimage.ubuntu.com/ubuntu-netbook/daily-preinstalled/20110123/natty-preinstalled-netbook-armel+omap4.img.gz I get 'No valid session found' after gdm startup. Is it a known problem ? [13:30] /etc/gdm/custom.conf has une-efl [13:44] dcordes: yup,algo got this with today's image [13:44] seems that I'm unable to log-in in any session [13:45] thats weird [13:45] but i think the whole session handling in gnome changed [13:45] once unity-2d is the default it should work again [13:48] yup, probably because of the gnome update [13:52] rsalveti: any workaround ? [13:53] well, teeh gnome desktop session should work [13:53] *the [13:53] rsalveti, so about that xloader prob on omap3 [13:54] the current build script looks for a package named x-loader-omap [13:54] I'm now trying to at least log-in at the gnome classic [13:54] and doesnt find it [13:54] so i assumed the binary package name changed [13:54] rsalveti: ok hope it works [13:55] ogra: the package name changed and the mlo path [13:55] but if you didn't change, it should still work [13:55] right [13:55] as we didn't remove the old package yet [13:55] hmm [13:55] weird [13:55] it doesnt find it [13:55] hm, ok, we should remove just the omap 4 one [13:55] the version changed, but it kind of replaced the old x-loader package [13:55] used by omap 3 [13:56] that could be the reason [13:56] dcordes: gnome classic seems to work [13:56] http://paste.ubuntu.com/557665/ [13:56] rsalveti: nice. is it 'gnome-desktop' in custom.conf ? [13:56] thats what i get in the buildlogs [13:57] ogra: yup, saw that when you pointed me the logs yesterday [13:57] dcordes: just select ubuntu classic desktop while log-in [13:58] ogra: you should now get x-loader-omap3-beagle [13:58] rsalveti: ok. gdm doesn't use custom.conf any longer ? [13:58] ogra, I have to admit I know nothing of multistrapping... :( [13:59] rsalveti, is there a transitional package ? [13:59] ogra: nops, as this package wasn't installed at any image [13:59] (though i'm not sure NCommander's way of extracting it would manage this, iirc he extracts the package directly) [14:00] and x-loader-omap doesn't make much sense now, as we also have overo [14:00] can somebody recommend a known working qemu kernel for the current (above) natty builds ? I tried some yetserday and ran into illegal instruction problems which I think were NEON related [14:00] and soon have igepv2 [14:00] that are all omap [14:00] indeed [14:05] in regards of the igepv2: Are there any efforts on using the DSP from linux? [14:06] or more specifically: ubuntu [14:13] sveinse, it's possible with natty's 2.6.37 kernel, (dspbridge), just need to package 2 git project (dsp-gst, dsp-tools) then package the ti codec's.. [14:36] rsalveti, http://ports.ubuntu.com/ubuntu-ports/pool/main/x/x-loader/ has not a single natty build, did anyone care for getting the newly natty binaries to main yet ? [14:36] ogra: hm, lp says it's published [14:37] or this is something that lp is not yet tracking? [14:37] published != main promoted [14:37] publish in main [14:37] that only means it went into the archive [14:37] *published [14:37] and the source is in main [14:37] but the binaries werent transitioned [14:37] http://ports.ubuntu.com/ubuntu-ports/pool/universe/x/x-loader/ [14:37] all of them sit in universe [14:38] hm, ok, sources at main but debs still in universe [14:38] right [14:38] an archive admin needs to promote the beagle and panda binaries to main [14:39] do we need to find a bug? or just ask someone? [14:39] yup, at least beagle and panda [14:40] don't know if it's ok for linaro to have overo at universe [14:40] but then they can just ask it if it's a problem [14:40] pitti promotes for us [14:41] ogra: a lot easier when you ask for it :-) [14:41] heh [14:41] ogra: thanks anyway, and sorry for the confusion [14:41] well, all fine now [14:42] next publisher run we should be fine, i'll adjust the build scripts for the new names and file locations then [14:43] cool [14:46] ok, scripts adjusted [14:47] ogra: nice, are you firing up another image for it? or just wait until tomorrow? [14:49] we will have to wait until the promotion is done, i'll trigger an imagebuild later [14:49] ok [15:01] ogra, you about? [15:01] davidm, indeed i am [15:02] sorry, the canonical server kicked me outr [15:02] Ah was wondering what happened [15:02] cool [16:16] ogra, can bug #683683 be closed out? [16:16] Launchpad bug 683683 in busybox "run-init on omap3, omap4 in natty dies if busybox is built with -marm" [High,Fix released] https://launchpad.net/bugs/683683 [16:16] jcrigby, no idea, ask the toolchain guys [16:16] or is it still a bug but there is a workaround [16:16] its a toolchain issue [16:16] ogra, ok [16:16] the other tasks are fix released iirc [16:20] jcrigby, iirc davidgiluk assigned it to gcc (and doko moved it to gcc4.5 or so) [16:20] ogra, yes I see that. [16:21] the prob is that compiling with -marm breaks under 4.5 apparently [16:21] at least for busybox [16:21] ogra, right [16:50] does anyone have a 101 for lp/bzr? i *think* I have created a branch within rootstock, but I'm uncertain how to branch from the main branch... [17:41] * sveinse finally managed to make his own rootstock branch and commit his patch. First time of lp+bzr has quite a learning curve [17:42] * rsalveti brb, late lunch [17:52] sveinse, thanks, i like that change, i'll be able to cleanup some of my roostock for debian tweaks.. ;) [17:53] yeah, I'm working on a custom HW board, and we use rootstock all the time to generate the rootfs images [17:53] so I can't use the official released images [17:57] same here. ;) [18:00] rcn-ee_at_work: running ubuntu or debian on target? [18:00] ..presume ubuntu since you're here ;) [18:01] all the above.. ;) i'm really itching for debian wheezy with armhf.. ;) [18:01] oh? how come? [18:01] We've selected ubuntu as target OS since this had the best armv7 support when we had to choose [18:03] yeah and it's pretty quick.. (lucid->natty)... but it's still using soft floating point... so it'll be interesting if ubuntu enables hardfp for natty++ (i'm hoping) [18:03] *that* I can put my vote to! [18:05] is natty particularly faster then maverick? BTW I'm running on omap3 (3530) [18:06] i wouldn't run it yet, as it's not even alpha-2, but it's a more patched gcc with the same compiler settings.. [18:07] wheezy will be targeted for armv7? [18:08] I seem to be asking google the wrong questions... [18:09] sveinse, it looks like wheezy will have the older "armel" and a new "armhf" for armv7 stuff .. http://wiki.debian.org/ArmHardFloatPort [18:09] ah hf=hard float... thanks [18:13] natty boots on the pandaboard nicely [18:13] not so much on the Versatile Express .. yet [18:19] Are there any known issues getting a graphic boot on the pandaboard at the moment? [19:02] rcn-ee_at_work, sveinse, ubuntu will have hf in about two releases [19:04] thanks, ogra cool.. [19:04] probably before wheezy :) [19:05] ;) for those that wait for the 'release'.. ;) [19:05] indeed [19:05] probally about 2 years just like squeeze... [19:05] the current debian hf port uses the ubuntu toolchain i think [19:06] yeah it's that linaro/ubuntu gcc-4.5. [19:06] right [19:37] rsalveti: tried out the gnome desktop session and it works [19:37] rsalveti: somehow metacity didn't show up once [19:38] dcordes: check bug 707014 [19:38] Launchpad bug 707014 in netbook-launcher-efl "Ubuntu Netbook Edition 2D fails to launch from gdm by giving "No valid session found"." [Undecided,New] https://launchpad.net/bugs/707014 [19:38] Is this possibly due to a change in jasper-initramfs? [19:38] nops [19:38] gnome-session [19:38] (I'm just now booting today's image). [19:39] Hmmm [19:41] rsalveti: cool thanks [19:43] -Exec=gnome-session [19:43] +Exec=gnome-session --session=une-efl [19:44] and a new file [19:44] ok [19:44] une-efl.session at /usr/share/gnome-session/sessions [19:45] it would be nice to be able to change the gnome-panel in nbl efl === ka6sox is now known as ka6sox-away [20:06] rsalveti: is there a way to do this except for using a converted gnome session? I had bad experience with that with nbl not coming up [20:07] dcordes: why you don't want to use a gnome-session for that? [20:08] I'm not sure about how to start netbook-launcher-efl correctly when doing it manually [20:08] just putt it in the session startup before but it didn't come up sometimes [20:08] even with large delay [20:10] that was in maverick release images === ka6sox-away is now known as ka6sox [20:30] just giving rootstock a little try. when i follow the guide in the wiki and fire up the image in qemu, it fails with segmentation fault. [20:30] LetoThe2nd: https://wiki.ubuntu.com/ARM/RootStock/KnownIssues [20:31] I think we have the fix for at least 2 bugs, but not yet packaged [20:31] rsalveti: so what to do? checkout trunk and use that? [20:32] bug is not related with rootstock, it's probably qemu [20:33] rsalveti: just built it from scratch... might it be fixed in the ubuntu repos? probably not, i'd guess... [20:33] rsalveti, before you add that file to /usr/share/gnome-session/sessions talk to didrocks, there is a certain naming scheme that has to be used i was told [20:34] (great that there is a fix !) [20:37] ogra: oh, ok [20:37] i think the distro name needs to be in the filename [20:37] i had to make unity-2d 2d-ubuntu [20:38] could be, let me check with him [20:39] That will suck. Not good if people want to try different launchers. [20:39] ?? [20:40] people will never see that file [20:40] its just a naming convention for the .session files that gnome put in place [20:41] wont matter if you call it efl-ubuntu or foo-ubuntu or whatever, you can still select the efl launcher in gdm [20:42] well, just got applied by janimo, but will talk to didrocks once he's on-line [20:42] thanks [20:42] with luck i'll kick out efl on wed. :) [20:42] :-) [20:43] though there might be a MIR blocker ... [20:43] -2d is currently untranslatable [20:43] (there is no gettext support yet) [20:46] hm [20:48] not sure i can get it past the MIR team that way, there are plans for gettext support, but no code or anything [20:48] bottom line - is qemu always segfaults with "qemu: fatal: VS[LR]I.64 not implemented", is there any way to get started with rootstock, or is this just a showstopper for now? [20:48] LetoThe2nd: what qemu version are you using? [20:48] and what ubuntu version [20:48] jd@tabr:~/rootstock$ qemu-system-arm --version [20:48] QEMU PC emulator version 0.12.5 (qemu-kvm-0.12.5), Copyright (c) 2003-2008 Fabrice Bellard [20:49] ubuntu 10.10, amd64. [20:49] LetoThe2nd, are you trying to build a 'natty' arm rootfs? [20:49] rcn-ee_at_work: just trying to get started with https://wiki.ubuntu.com/ARM/RootfsFromScratch [20:50] LetoThe2nd, just for reference, can you pastebin your full 'rootstock' command.. [20:51] rcn-ee_at_work: of course. http://paste.pocoo.org/show/326221/ [20:53] yeah, your aren't giving qemu any chance with that... unless you run it on an armel machine, xubuntu-desktop is going to kill qemu... [20:56] rcn-ee_at_work: ok, trying it without seed now. [20:56] rcn-ee_at_work: is there any particular reason why xubuntu kills qemu? or what other packages will have the same effect? [20:56] LetoThe2nd, well just something less entensive for the inital install.. [20:58] LetoThe2nd, it varies on release, some *.deb packages cause issues when installed thru qemu... Ever since lucid it's been a pain in the rear to track down.. [21:00] rcn-ee_at_work: i see.... strange. the message would indicate something missing in qemu's instruction set, so i'd guess it could happen with any package if the compiler only decided to put in that one instruction. [21:04] rcn-ee_at_work: taking away the seed, the emulated system does not segfault, but just stop with a blinking cursor after leaving the kernels boot. :-/ [21:10] are you dumping any info to the serial console? [21:10] rcn-ee_at_work: no, just stating with the exact example command given on the wiki page. [21:12] LetoThe2nd, in your previous message "the emulated system does not segfault, but just stop with a blinking cursor" do you have a serial console for debugging [21:14] rcn-ee_at_work: ok, probably stupid question - where? [21:14] qemu-system-arm -M versatilepb -cpu cortex-a8 -kernel ./vmlinuz -hda qemu-armel-201101242154.img -m 256 -append "root=/dev/sda mem=256M devtmpfs.mount=0 rw" [21:14] qemu-system-arm -M versatilepb -cpu cortex-a8 -kernel ./vmlinuz -hda qemu-armel-201101242154.img -m 256 -append "root=/dev/sda mem=256M devtmpfs.mount=0 rw" [21:14] ohsry for doubleposting. [21:16] ah, booting the image with qemu... sorry haven't done that for atleast a year... [21:17] rcn-ee_at_work: for the protocol, i try to start following this: https://wiki.ubuntu.com/ARM/RootfsFromScratch ... hoping that the "demo" commands provide a working starting point. [21:22] LetoThe2nd, based on the age of that wiki document, i bet it was with karmic's default, so you need to add "--dist karmic" but that's pretty old.. [21:37] rsalveti: around? [21:37] TheUni: yup [21:38] rsalveti: i've just merged our massive code refactor, i'd like to begin working with you on an evaluation when you get a chance [21:38] (xbmc) [21:39] TheUni: awesome, I can try to check this out this week and focus more on next week [21:39] as we moved it to alpha 3 [21:39] yep, i saw [21:40] TheUni: did you create any release for it? or basically current trunk? [21:40] rsalveti: current master. release is still a ways out [21:42] np, will try to check that out [21:42] rsalveti: as i've mentioned, you might not target for natty, but we would greatly appreciate some ideas when it comes to packaging. what we've got currently is a monstrosity :) [21:42] haha, np [21:42] let's see what we can at least get for natty [21:56] rcn-ee_at_work: for the protocol: it works with dist karmic. [21:57] yeah, karmic's easier for qemu... compile for armv6 vs armv7 + thumb for lucid/maverick/natty... [21:58] rcn-ee_at_work: gotta read up a bit on qemus changes, maybe a more recent trunk might do :-)