[01:00] Where can I get a binary Linaro toolchain so that I can put it anywhere I want? [01:06] Check on #linaro. Someone there would know. [01:12] Grrr. A1 is a fail at this point. Kernel oops on second boot, shortly after mounting rootfs. http://paste.ubuntu.com/538505/ [01:12] ogra: rsalveti: ^^^ === rsalveti` is now known as rsalveti === amitk is now known as amitk-afk === amitk-afk is now known as amitk === hrw|gone is now known as hrw [07:54] morning [12:16] Hi guys. This is perhaps OT here, but I tried to cross compile Qt using the g++-arm-linux-gnueabi compiler from Maverick. The compilation failed, while the CodeSourcery gcc don't. The failure is "selected processor does not support Thumb mode `swp r4,r3,[r2]'". The compile options are set to "-march=armv7-a -mtune=cortex-a8 -mfloat-abi=softfp -mfpu=neon". I notice the CSL gcc is 4.4.1, while... [12:17] ...the Ubuntu gcc is 4.5.1. Does this ring any bells to anyone? [12:21] hrw: ^ [12:22] sveinse: can you report bug against gcc-4.5 and give me number? [12:24] hrw: In whos bug system? Gcc or ubutu? [12:24] ubuntu [12:25] hrw: I could try recompiling using the 4.4 version to see if its related to some difference between the CSL and Ubuntu or if its related to 4.4. vs 4.5 [12:27] tmzt_g2root: regarding the custom netbook-launcher-efl session [12:28] yes [12:28] rsalveti: maybe #linaro ? [12:29] tmzt_g2root: I found descriptions on how to build a custom session in some ubuntu wiki. I was searching for how to customize the gnome panel as it is locked [12:29] tmzt_g2root: the problems I see with netbook on small screen touchscreen devices are only few [12:29] dcordes: the instructions didn't work for me, the gconf stuff didn't change anything [12:30] yeah, it's pretty nice. installing packagekit fixed software-center [12:30] it's a bit off the screen but mostly usable [12:30] tmzt_g2root, hrw is in both channels luckily (#linaro and here) ;) [12:31] tmzt_g2root: 1) off-screen areas in some programs (related to maximus) 2) small buttons, scroll bars, etc => problems navigating [12:31] dcordes: let's work on here https://github.com/tmzt/native-netbook we should be able to share everything between hd2 and androix on wvga ws [12:32] dcordes, to change the locked gnome-panel properly you would have to change and rebuild ubuntu-netbook-efl-default-settings [12:32] tmzt_g2root: that would be nice. but to keep in touch with upstream we should document our work in launchpad [12:32] all panel settings we do there are mandatory [12:33] dcordes: right, I'm just trying to collect scripts and stuff and then when I'm ready to do a package like ogra says I can just migrate that stuff [12:33] we can put in on a ppa too, but I don't know if you have to be a developer for that [12:33] can't put raw scripts on ppa though [12:33] currently you have to for armel only packages [12:33] ubuntu-netbook-efl-default-settings is arch all though [12:34] so for that package any ppa will do [12:34] tmzt_g2root: ... forgot an important point 3) on screen keyboard not available everywhere [12:34] ogra_ac_: it will probably require changing launcher itself, we need kinetic scrolling and mtdev (finger width) support [12:34] dcordes: that might be an issue, it isn't for me with g2 though, but please solve it [12:34] I'm trying to get androix to send some events for soft keys on android [12:35] tmzt_g2root, well, then you have to wait until our armel PPA spec is implemented [12:35] I haven't decided how to handle them yet thouh [12:35] which wont happen before the alpha2 release i suspect [12:35] ogra_ac_: okay, I'm putting together a repo on androix.org for now [12:35] but it will enable armel on all PPAs [12:35] that would be good, especially if it could build for older architectures, but I don't expect that === ogra_ac_ is now known as ogra_ac [12:36] it will use the ubuntu defaults [12:36] tmzt_g2root: regarding on screen keyboard lack 3) https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/626055 [12:36] Launchpad bug 626055 in ubiquity (Ubuntu) "oem-config: make on-screen keyboard available (affects: 1) (heat: 40)" [Low,New] [12:36] maybe we could do a small panel thing that just does indicators, no clue if they would work with efl though [12:37] we could mix in illume and gnome, that would be confusing [12:37] have a look at ubiquity [12:37] it brings a minimal panel in maverick [12:37] i guess its not hard to rip the panel code out of it and build a standalone package [12:38] ogra_ac: I need to ask you about remixes and policies, I'm trying to build something to run Ubuntu on top of Android, for now I'm calling it Native Netbook [12:39] yeah, uquity works on 2d? [12:39] sure [12:39] cool, we'll look at that then [12:40] tmzt_g2root, mail trademarks@ubuntu.com, i dont know the exact policy [12:40] tmzt_g2root: small panel ? can you elaborate ? [12:40] dcordes: how do you find this stuff in lp? [12:40] might be that you cant use the logos etc [12:40] tmzt_g2root: I enter my lp URI :) [12:41] http://www.ubuntu.com/aboutus/trademarkpolicy [12:43] dcordes: I'm going to let you handle keyboard stuff :) but I'm sure I'll need it when people with nexus, etc. want to try it [12:44] tmzt_g2root: I don't have any android device, sorry. [12:44] dcordes: right, it's not an android issue though, it's a touchscreen issue [12:44] and you do have a leo [12:45] I would be using an osk in X, not androids [12:45] I am using onboard , the standard ubuntu osk [12:46] can't IME do what your bug 443986 says, pop up osk if the widget gets keyboard focus? [12:46] Launchpad bug 443986 in onboard "RFE: Add option to automatically show and hide onboard (affects: 5) (dups: 1) (heat: 18)" [Undecided,New] https://launchpad.net/bugs/443986 [12:47] I guess X really needs a flag to tell it if a physical keyboard is attaeched, or you would have to enumerate the Xi2 device list [12:48] tmzt_g2root: that would be nice [12:48] tmzt_g2root: what is IME ? [12:48] input methods [12:48] it would require a gtk/gdk patch too I assume [12:48] but at least there's a standard way to handle that [12:48] can you append it in the bug ? [12:48] it's also possible openmoko had this on the gtk version a few versions back [12:49] and figure out oauth again? :) [12:49] currently openmokoe shr uses some efl osk keyboard [12:49] how do you add a project or distribution? [12:49] currently, I'm talking about the gtk version they had [12:49] they also had the finger gtk theme [12:50] right I remember [12:50] I've ping #xorg-devel about the Xi question [12:50] dcordes: have you tried getting any of the new 3d stuff to compile against bionic so it can use libgles_cm? [12:50] many touchscreen device owners would love such a feature in Xorg [12:51] I have no clue about all the 3d stuff [12:51] it's one properties, PhysicalKeyboard or whatever [12:51] that's why I'm sticking with efl at the moment [12:52] dcordes: https://wiki.kubuntu.org/X/Blueprints/Touchscreen/UDS-M [12:52] that's not kubuntu specific, they just show up higher in google [12:53] ogra_ac: what happened to the mobile channel? where did all the netbook stuff get moved too? [12:54] tmzt_g2root: ok. we might as well discuss TS things in #ubuntu-touch [12:57] I should just see if starting metacity fixes it, I don't think maximus is supposed to work standalone [12:57] if you have a working panel config I can help you push it to my git or whatever you have [12:59] tmzt_g2root: in launchpad we might create a 'project' 'ubuntu on small screen (& touch screen devices)' [12:59] tmzt_g2root: then we can have bugs like off screen windows, lack of osk affect it [12:59] WVGA Touchscreen ? [12:59] that would exclude vga devices and exotic resolutions [13:00] like qvga? [13:00] you need this for your kaiser :) ? [13:00] WVGA is a difficult case, most of the bugs will be height related and also apply to VGA [13:01] QVGA screens are too small for any of this and the devices' other hw is too slow to run full ubuntu systems [13:01] in basement I have XGA 10" touchscreen device [13:01] I think small screen is good. but what about the approach ? [13:01] in general means [13:01] define small [13:01] ok max wvga [13:01] this isn't for XGA or WXGA or 1024x600 there's plenty of coverage for that [13:02] ah true [13:02] didn't think of that [13:02] we can do smartphone & pda then ? [13:02] it would cover phone functionalities [13:02] tmzt_g2root, netbook (unity) is fully handled by the desktop team now (since unity merges both desktops) [13:02] wvga+ like the motorola's I would say, but that's waht 854x496 or so [13:02] ogra_ac: is 2d/framebuffer still supported? [13:03] not atm [13:03] only with a std desktop as fallback [13:03] ogra_ac: I don't have 3d support yet, I should be able to get 3d working but it might just be redirected rendering/AIGLX stuff [13:04] tmzt_g2root: know what let's just go device specific (ubuntu on hd2 & g2) if others want to hop on we can generalize later [13:04] depends on your platform, there is no GLES support in unity yet [13:04] ogra_ac: so I'm starting with netbook-efl, most of these problems will be unrelated to the acutal desktop, bugs in programs etc. [13:04] ogra_ac: well, it's going to have to be GLES, so that's another reason to stick with 2d for now [13:05] dcordes: G2 won't have keyboard issues, the things we have in common are WVGA and Touchscreen [13:05] tmzt_g2root: ok I don't see a problem there [13:06] just note that we might stop using the efl launcher in ubuntu (so support will rather have to be community based) [13:06] ogra_ac: yeah, it will still be in natty or that's not known yet [13:07] we wont remove the package from the archive [13:08] but given that it will likely not be used by default anymore bugfixes and maintenance has to come from the community [13:08] I think the only thing we need to change in it is the scrolling issue for touchscreens [13:08] yeah, okay [13:09] does it have a maintainer or we need ubuntu developer for that? (not canoncial) [13:09] tmzt_g2root: how about "smallscreentouchscreen" project [13:09] well, you can indeed work through a sponsor [13:09] but for uploading to ubuntu you will need one [13:10] what is the project for? just tracking bugs against ubuntu that affect screenscreen touchscreens? I really think small is too vague [13:10] yes to track bugs that affect our devices [13:10] as far as aspect, we really aren't going to support vga either because that would require massive chnages from what every laptop/netbook is now [13:11] and maybe to upload non-sub-project-specific scripts/code later [13:12] well, the packages, but those are all config stuff, I wouldn't be putting any system stuff there at all [13:12] yes config etc [13:12] I mean like launcher defaults, maybe unity places stuff if we can switch to unity [13:16] * ogra_ac is off for a while [13:17] tmzt_g2root: smartphonebuntu ? I think it is not neccessary for the project name to refelct the exact aims [13:17] no, read the trademark policy [13:18] which ogra linked to [13:18] argh [13:18] smartphone-remix [13:18] possibly [13:18] you like it? any other idea ? [13:19] it [13:19] I'm looking for developer to create new on-screen keyboard with Input Method Editor. Similar to those available on mobile devices. Currently I'm using matchbox-keyboard but for my purpose I guess it's worth start from scratch. [13:19] oops [13:19] it seems to be what I'm trying to do [13:41] tmzt_g2root: https://launchpad.net/smartphone [13:52] can you duplicate the ~lg bugs there? [13:52] link them [14:04] GrueMaster, where is the bug for that and does cooloney already know (and work on it) ? [14:09] tmzt_g2root: I did already. check https://bugs.launchpad.net/smartphone [14:10] dcordes: I discussed the xinput problem, I think I can just add a property to the device by patching evdev, or in my case, in the server init [14:10] so it would show up when you do xinput list-props [14:11] there's already a "Device Enabled" property [14:12] tmzt_g2root: to which device ? [14:12] then the IME just has to walk that and see if any are Physical && Enabled, and if so, not show the osk [14:12] to the keyboard device [14:12] ok [14:12] then we patch evdev to handle the SW event that's created when you slide the keyboard [14:13] for you, when you plug your usb keyboard the same thing happens [14:13] and for me, I get the value from android and set the priv myself [14:14] is this a global keyboard presence detection mechanism ? [14:14] yeah [14:14] patching evdev - is this specific to the device (driver) ? [14:14] but onscreen keyboards and x2x won't register that property [14:15] no, evdev already detects if it it's a keyboard, it will be wrong for uinput keyboards so we may have to put it in udev or something else [14:15] like hal did, but it didn't propogate it [14:15] are there evdev / Xorg bugtrackers ? [14:16] there are, but I'll propose it to the list later, we can just put it on my github for now [14:16] youz should publish the approach and gather interested people's attention [14:16] I already have half of xorg cloned [14:16] ok bugtracker is nice [14:16] * ogra_ac would recommend asking in #ubuntu-x [14:16] ogra_ac: is that supposed to be public? I was having an issue with it [14:16] sure thats public [14:16] never mind [14:16] weird [14:17] ogra_ac: as tmzt_g2root's approach sounds to me it will be better to have it a global X thing and not discuss it distro internal [14:17] sure [14:17] ogra_ac: I pinged daniels in #xorg-devel he agreed with the property thing, and reminded me of Device Enabled [14:17] then again there might be many ubuntu developers interested in supporting it [14:17] still you should talk to the ubuntu X force [14:18] I'm sure acpi devices already have something like that [14:18] right [14:18] touchpads I mean [14:18] ogra_ac: soon as it is on Xorg list we should refer to it there [14:19] I have to run catch you guys around [14:53] sigh [14:53] so i think the kernel panic on boot is run-init failing [14:54] /init: exec: line 331: run-init: Unknown error 17718852 [14:54] * ogra wonders if that toolchain related [14:58] GrueMaster, rsalveti, bug 683683 [14:58] Launchpad bug 683683 in klibc (Ubuntu) "run-init on omap4 in natty dies with "run-init: Unknown error 17718852" (affects: 1) (heat: 8)" [High,New] https://launchpad.net/bugs/683683 === zyga is now known as zyga-food [15:12] NCommander, so apparently klibc wasnt rebuilt yet in natty [15:13] GrueMaster, btw, the error is easily visible if you dont tinker with serial consoles, it gets clearly printed on the screen [15:13] We are basing our PickleJar Linux on Ubuntu Maverick. :D [15:14] PickleJar Linux is the distro for our Pico Node project. [15:16] RobotGuy, great to hear [15:17] We have an expansion board for BeagleBoard-xM in the design process now and hope to have the first few boards soon. Find out more at http://www.picklejar.org [15:18] Our kernel version is 2.6.35.4 [15:19] hrm, run-init is 93 LOC ... of which 40 are license text [15:20] * ogra doesnt see what could be wrong in there [15:21] ogra: so no changes rebuild needed maybe? [15:22] not sure [15:22] look at the code, probably you see something obvious [15:22] i surely dont [15:22] hmm runinitlib.c seems to define some glibc stuff at the top [15:29] ogra: something is going hidiously wrongin run_init() [15:29] why would it ? [15:30] it didnt change [15:30] ogra: I'm saying that's what happening, I don't know why, but the error message being printed out suggests that's where we are blowing up [15:30] well, but take a look at the code, probably you see something i dont [15:30] /* If run_init returns, something went wrong */ [15:30] fprintf(stderr, "%s: %s: %s\n", program, error, strerror(errno)); [15:31] no, we are blowing up in line 88 already [15:31] in run_init(realroot, console, init, initargs); [15:31] which seems to come from runinitlib.c [15:32] through run-init.h [15:32] right, run_init() is returning when it shouldn't e [15:32] yes [15:33] the execv() call is failing [15:34] only place it can fail in this code and return a string of "Unknown error" I think [15:34] well, it could fail because it gets wrong args [15:34] and the linaro bug says the errno is random [15:34] so nothing to grab [15:34] which linaro bug? [15:35] did you read my bug ? [15:35] yeah, I did [15:35] I don't see anything from linaro [15:35] oh [15:35] reload ? [15:35] :) [15:35] * NCommander just hit refresh [15:35] oh, i did too [15:35] there is a debian bug linked [15:35] Debug #334917 [15:35] er [15:36] Debian bug #334917 [15:36] Debian bug 334917 in klibc "klibc barfs on m68k syscall interface" [Important,Open] http://bugs.debian.org/334917 [15:36] nice bug title [15:36] Debug seems appropriate, somehow. :D :D [15:38] ogra: looking at the debian bug, I think its unrelated simply because execve is a syscall and the first one is called directly. the ARM EABI syscall interface in klibc is correct else this would have blown up ages ago. So similar error, but I think unrelated causes [15:38] ogra: Ok I'm semi-awake. What do you mean: GrueMaster, btw, the error is easily visible if you dont tinker with serial consoles, it gets clearly printed on the screen [15:38] It is not clearly visible on my screen. [15:39] GrueMaster, you could have found it with the image from the 19th if you hadnt set up a serial console and just dropped splash from cmdline [15:39] thats how i got it at least === zyga-food is now known as zyga [15:39] it shows up right before the kernel panic (which it does because there is no rootfs) [15:40] ogra: I'm looking at the linaro branch right now to see if anything is clear from that [15:40] branch ? [15:40] do they have a separate klibc branch from ubuntu ? [15:41] So then why doesn't it show up with serial console? [15:41] That doesn't make sense. [15:41] ogra: they said in their bug that they have a branch to fix it [15:42] NCommander, yes, for the linaro build tools issue that shows up too [15:42] NCommander, (OSError: [Errno 2] No such file or directory from remove_binary_dir.py from _run_code from _run_module_as_main). [15:42] not related to to booting [15:47] * ogra reboots with break=bottom [15:47] lets take a look at the environment [15:49] ok [15:50] exec run-init /root /sbin/init gets me the same error === dmart__ is now known as dmart [16:04] ogra: any chance you can get gdbserver into the initramfs environment? [16:04] * NCommander finally got enough coffee into his bloodstream to be lucid enough to think [16:04] try it ? [16:05] ogra: your ahead of me on having things to test on :-). I'm still reading emails while drinking GrueBrew Coffee [16:05] i dont have any things to test on [16:06] i just added break=bottom to my cdmline [16:06] you need to write a hook and rebuild the initrd for getting your gdbserver included [16:06] or for getting any other debug tool [16:08] ogra: just copy it in somehow, its not that difficult [16:08] but I'll get on it [16:08] ?? [16:08] "copy it in somehow" ? [16:09] ogra: gdbserver is a tiny little stub binary [16:09] then attach to it with a cross-debugger [16:10] you still need to re-roll the initrd [16:16] so init=/bin/bash doesnt work either [16:17] * ogra drops all bootarchs apart from root= [16:18] argh [16:18] bad idea ... [16:18] * ogra beats jasper over the head [16:22] So it's a jasper issue? I'm still reading emails and backscrolls. [16:22] hey any idea how to get sound working on the beagle xm rev b [16:22] there is a rev b? sigh. [16:23] yeah :( [16:23] lol [16:24] i dont think they fixed anything really [16:25] no, but it is possible that something changed requiring an updated kernel...again. [16:25] Are you running the stock maverick release image? [16:26] GrueMaster, ?? [16:27] its a klibc issue [16:27] nothing to do with jasper [16:27] and I am not talking to you atm ogra. Read the current thread. [16:27] mellis: ??? [16:38] sorry just went to get food [16:38] yeah i got the minimal image and installed ubuntu-desktop package over it [16:39] no problem. [16:39] minimal image? [16:41] also called the demo image [16:41] i couldint get s-video to work on the preinstalled image [16:43] prpplague, have a look at: http://dmtechtalk.wordpress.com/2010/12/01/progress-on-an-arm-cluster-server-box/ [16:43] yeah i got the demo image then installed ubuntu-desktop and the omap driveres on top [16:43] prpplague, it will fit nicely into a standard 19" rack cabinet [16:44] If this image is attached to the ubuntu repositories, try enabling maverick-proposed and updating. I think there was a fix for audio in alsa-utils, but it may be only omap4 specific. [16:45] ok i will try [16:47] * prpplague looks [16:47] davidm: right, i'm actually doing some with them on that [16:49] humm my webcam also wont work so could it be a kernal problem? [16:51] mellis: does it work with x86 box? [16:52] the webcam yes [17:00] hrw: I confirm problems cross compiling Qt for both gcc 4.4 and 4.5, so its related to some difference between the ubuntu cross compiler and the CSL. I'll file a bug. [17:00] and i cant find anything missing from my setup [17:02] sveinse: mention package versions for gcc 4.4/4.5 and version of Qt. if Qt is from nokia then give url please [17:02] * hrw -> out === hrw is now known as hrw|gone [17:34] hmm, that dist-upgrade of my chroot will take a while [18:04] GrueMaster: lmbench has the wrong bin path, if you noticed [18:05] currently the package delivers armv5tel-linux-gnu instead of armv7l-linux-gnu [18:05] that's because the package path is decided during build time, and detected by a script during runtime [18:05] hadn't gotten that far yet. Any suggestions? [18:05] and this package was built in an armv5 machine, while in karmic hehe :-) [18:06] copying the files should be enough, but I also recreated the package, let me post you the link [18:06] dcordes: how do you switch the default gnome-session back to GNOME ? [18:06] without using gdm [18:07] tmzt_g2root, /usr/lib/gdm/gdm-set-default-session [18:09] as the user? [18:09] I don't seem to have that [18:09] oh, I'm not using gdm at all [18:09] no, its system wide, not as a user [18:10] just starting ck-session-launch dbus-launch --exit-with-session gnome-session [18:10] right, but I'm not getting a panel following these instructions [18:10] https://help.ubuntu.com/community/UbuntuNetbookEdition/ConvertGnomeSession [18:12] GrueMaster: http://people.canonical.com/~rsalveti/lmbench/ for now [18:13] ok [18:13] we should have a bug, will see if we got it already otherwise need to fill it [18:15] hey i just installed the omap kernal package how do i get it to boot [18:15] tmzt_g2root: that is the wiki page I was talking about ! [18:15] dcordes: I know, I saw it yesterday, but it doesn't work for me [18:16] /apps/maximus/exclude_class [Empathy,Totem,Gwibber,Gnome-language-selector,Gtk-recordMyDesktop,Onboard,Vlc,Seahorse-agent,Gnome-keyring-prompt] [18:16] tmzt_g2root: there was some error in the 2d instruction part [18:17] I'll have to add dialog or something to that, or just gedit's or libgtk open [18:18] tmzt_g2root: do you want me to look up anything in my rootfs ? [18:18] what script do you use to start the gnome-session? [18:20] sudo ln -s /etc/xdg/xdg-une/autostart/maximus-autostart.desktop /etc/xdg/autostart/ [18:20] sudo ln -s /etc/xdg/xdg-une-efl/autostart/netbook-launcher-efl.desktop /etc/xdg/autostart/ [18:20] I have both of those, I don't get a panel when starting gnome-session [18:21] the first time I started gnome-session it was a normal desktop, before symlinking that stuff [18:21] the first line.. I think it should be [18:21] /etc/xdg/xdg-une-efl/autostart/netbook-launcher-efl.desktop /etc/xdg/autostart/ [18:21] /desktop/gnome/session/required_components/windowmanager mutter [18:21] /desktop/gnome/session/required_components/panel '' [18:21] oh [18:22] can you try that ? [18:26] still no panel [18:27] nevermind, it was just slow [18:28] NCommander, so building klibc brings intresting results in the buildlog [18:28] vte doesn't work, I have never seen that before [18:28] NCommander, it has -march=armv4 -mtune=strongarm hardcoded [18:29] (but given that has always been the casei wouldnt think that has any influence on our bug) [18:30] ogra_ac: yeah, was looking at that, but I think its irrevelent [18:30] klibc didn't help, nor did upstart [18:30] you rebuilt it ? [18:30] I tried downgrading the kernel to maverick release on a hunch [18:30] ogra_ac: yeah [18:30] what would upstart have to do with it ? [18:30] we are way before upstart [18:31] and as i said in the bug init=/bin/bash didnt work either [18:31] ogra_ac: sorry, meant to say that I don't htink upstart helps [18:31] ogra_ac: or part of it [18:31] no, upstart is out of scope here [18:31] right [18:31] Very very odd [18:31] i still think its the args that are messed up [18:31] try to add some schos to /init [18:31] ogra_ac: yeah, but what changed to kill the args [18:31] *echos [18:32] echo $@ [18:32] echo ${rootmnt} [18:32] and [18:33] echo ${init} [18:33] see what that shows [18:33] also note that break= drops you into a subshell, so you cant run run-init manually [18:33] ogra_ac: how close are you to doing so? (I can rapidly respin the initrams ATM on my ac100, but if your already doing it ...) [18:33] i'm not even near my panda atm [18:34] just building klibc remotely with different options [18:34] so i would appreciate if you could test [18:34] ogra_ac: what script calls run-init? (or do I need to smack the code) [18:34] /init [18:35] its a shell script [18:35] just add some echos [18:35] lives in /usr/share/initramfs-tools/init [18:35] ogra_ac: thanks. I thought you called run-init directly thoguh within a console with proper args and still got a crash [18:35] in the normal system [18:35] i tried that [18:35] ogra_ac: disclaimer: I don't get the run-init error, just the panic on the serial console [18:35] just to be told that wont work because i'm in a subshell [18:36] NCommander, you will get it on the monitor if you boot without splash [18:36] you will also get it if you boot with break=bottom and ctrl-d out of it to continue the boot [18:37] run-init will only work if called by pid 1 [18:37] so you cant call it manually [18:40] ogra_ac: I don't have a monitor I can use in my "office" [18:40] then just break=bottom [18:41] ogra_ac: will do [18:41] * NCommander kicks his ac100 [18:41] and hit ctrl-d [18:41] did you finally manage to wear out your emmc ? [18:41] :) [18:41] ogra_ac: no, the USB controller is having mini-seizures [18:42] ah [18:42] and every once in awhile, it and my USB HDD decide to cause a process to become a zombie [18:42] very annoying when thats dpkg [18:44] just fix the kernel ;) [18:44] ogra_ac: not paid to do that ATM :-/ [18:44] * ogra_ac will spend a good part of his holiday improving the ac100 [18:45] ogra_ac: tegra? [18:45] tmzt_g2root, yep [18:46] ogra_ac: \o/ [18:46] NCommander, dont party to early, i will concentrate on moving PM and buttom management into the .29 kernel only [18:47] ugh [18:47] /o\ [18:48] for never kernel talk to marvin24 [18:48] ;) [18:48] he has something based on .36, just misses a regulator specialist [18:49] regulator? [18:49] to power on the LCD i think [18:49] it boots to a serial USB console [18:49] clean up the nvidia oem stuff so it stops asking a binary layer to describe the hardware? [18:49] but has no device drivers yet [18:49] no, i will just revert the hacks tochiba put in [18:50] then get nouveu?? ported [18:50] and see that i get some sane defaults into the kernel [18:50] nah [18:50] i dont care about 3D drivers [18:50] i just need power management [18:50] the 2d is crtc only [18:50] lcdc [18:51] i need backlight on/off on lid close and want to have ondemand scaling working [18:51] extra points for dimming support [18:51] and a proper ubuntu initrd [18:51] thats my current focus for the vacation [18:52] you don't have an android phones do you? [18:52] nope, i have a sane phone [18:54] smart mount -t devpts devpts $MOUNTPT/dev [18:55] NCommander, want to have my klibc built for armv7 ? [18:55] or do you have one already ? [18:56] ogra_ac: haven't needed it yet [18:57] k [18:57] ogra_ac: got debugging data [18:57] ogra_ac: rootmnt is "/root" [18:57] init is "/sbin/init" [18:57] arguments: "fixrtc" [18:57] That was amazingly unuseful [18:57] yeah [18:57] * NCommander throws it into the bug for completeness sake [18:58] i wonder if the bindmounted /dev is ok [18:58] i.e. if we get the right console [18:58] ogra_ac: easy way to test that [18:59] but run-init has a sanity check for that it should fail with "unable to open consoel" or something like that [18:59] and its the right major/minor for dev console [19:00] in /dev [19:00] i'm talking about $rootmount/dev [19:01] udev should move mount /dev to /root/dev, but i wonder what happens if that doesnt work/happen [19:01] ogra_ac: system would fallover [19:01] yeah, likely even before run-init [19:02] ogra_ac: I think you might be onto something [19:02] /root/dev is populated [19:03] we have a newer upstream udev [19:03] and udev sometimes breaks backwards compatilbility with older kernels [19:03] see /usr/share/initramfs-tools/scripts/init-bottom/udev [19:03] and mainline kernel is on 37 [19:03] hrm [19:03] probably add some ls in there [19:03] give me a sec, I want to check something else first [19:03] before the move ls /dev [19:04] after the move ls /root/dev [19:04] and compare [19:04] ogra_ac: I'm trying to downgrade udev [19:04] to maverick [19:05] just rule it out completely [19:05] k [19:06] maverick has 162 [19:06] natty 164 [19:07] * ogra_ac checks changelogs [19:09] looks okayish [19:15] ogra_ac: no love there, udev downgrade and still a hang [19:15] I'm runng out of things to downgrade [19:16] so want my klibc package ? [19:16] ogra_ac: sure althoguh at this point I don't think it will help [19:17] i dont think it either [19:19] ogra_ac: doing the ls devs doesn't reveal anytthing interesting [19:19] I'm running out of ideas [19:19] http://people.canonical.com/~ogra/libklibc_1.5.20-1_armel.deb and http://people.canonical.com/~ogra/klibc-utils_1.5.20-1_armel.deb [19:22] her where would i go to report / check a bug in the omap kernel [19:22] mellis: downgrading the kenrel didn't fix it [19:22] Don't think this is a kernel problem, probably something with userland [19:23] i upgraded my kernal and it fixed my sound problem but my s-video went dead [19:24] ogra_ac: I'm wonder ... [19:24] * NCommander is starting to think run-init == redherring [19:24] yes [19:24] thats why i looked at the args [19:24] what do you suspect ? [19:25] Anyone there knows the correct gcc options for TI OMAP3? -march=armv7-a -mtune=cortex-a8 [19:25] ogra_ac: well, ... I'm not so sure now. I just tried replacing /sbin/init on my panda with /bin/bash, and letting it boot and it just fell over regardless [19:26] ?? [19:26] you mean in the rootfs ? [19:26] yeah [19:26] well, we dont get to the rootfs [19:26] We're getting as far as the execve() call [19:26] no matter what you use [19:26] ? [19:27] oh [19:27] yes [19:27] i tired init=/bin/bash already [19:27] on a different note, I did some preliminary testing of the linaro kernel on my BeagleXM, and it seems ok. Rythmnbox has issues, but I am able to get test sounds to play (R & L are reversed). [19:27] what the hell could cause an exec() call to crap itself? [19:27] GrueMaster, awesome, can you let apw know ? [19:27] then we have fulfilled our workitems [19:27] for the kernel team [19:30] ogra_ac: execve is a straight systemcall, but we're not getting an OOPS [19:30] something bloody odd is going on [19:30] why would you get an OOPS ? [19:30] ogra_ac: er, right, a bad syscall won't cause that [19:30] ugh [19:30] */brain fart* [19:30] * NCommander rebuilds klibc with debugging info [19:30] yea, I can smell it in here. :P [19:30] the OOPS you see has not much to do with run-init failing [19:31] ogra_ac: yeah, I got that [19:31] the OOPS you see is simply the kernel not finiding init [19:31] right [19:31] hrm [19:31] which you also get on older releases if init isnt executable or available [19:32] ignore the kernel in this context its all run-init [19:32] ogra_ac: I updated the kernel whiteboard for that task. [19:32] ogra_ac: right, so I'm rebuilding it now with -O2 -g, and will see if I can attach a debugger to it [19:33] GrueMaster, merci [19:33] (knowing our luck, that will mysteriously fix it) [19:33] NCommander, did you test with my v7 build from above ? [19:34] http://people.canonical.com/~ogra/libklibc_1.5.20-1_armel.deb and http://people.canonical.com/~ogra/klibc-utils_1.5.20-1_armel.deb [19:34] NCommander, ^^^ [19:34] what if you can't run any executable on rootfs because of some weird ld.so.blah mismatch [19:35] hmm [19:35] with klibc? [19:35] yeah, someone should try booting with a maverick initrd [19:37] tmzt_g2root, klibc is only used in initrd [19:37] not paying close enough attention [19:37] I've got crazy stuff like this but not to the point of err values getting overwritten [19:38] but, you could try init=/lib/ld-linux.so.3 [19:38] and see if it prints anything [19:42] dcordes: you had efl and panel at the same time? I can't get that to work [19:43] why wont the omap kernels work with s-video [19:44] It isn't an easy interface to test. [19:44] mellis: missing i2c controller driver? [19:44] Not sure if it is just a config flag or it requires more kernel work. [19:44] no idea [19:45] sorry encoder [19:45] i have one kernel that works [19:46] The only thing I have available for testing that output is my 42" LCD in the livingroom, but it is very difficult to get setup for svideo. [19:47] yeah i only have a s-video tv so ima bit stuck [20:03] hrw|gone: Here is bug #683832 as instructed. [20:04] Launchpad bug 683832 in gcc-4.4-armel-cross (Ubuntu) "gcc fails to cross compile Qt (affects: 1) (heat: 6)" [Undecided,New] https://launchpad.net/bugs/683832 [20:11] I reported the bug in gcc-4.4-armel-cross, but it also applies to gcc-4.5-armel-cross. Do I need to add a new bug, or can I link them for both packages? [20:32] ah haw [20:32] might figured out what broke [20:33] tell me [20:34] the suspense is killing us [20:34] ogra_ac: or not [20:34] but if I set init to /bin/dash [20:34] it works [20:34] the plot thickens [20:35] er [20:35] maybe not [20:35] strange [20:35] as soon as run-init runs, it breaks the enviornment [20:36] no, /init unsets most of it [20:36] ogra_ac: I'm referring to running run-init directly [20:36] before it runs run-init [20:36] if I runit through strace, it makes some very pretty output and works [20:36] (sorta) [20:37] you cant run it from a pid other than 1 [20:37] I get the same sorta errors though [20:37] (you cant run it from the initramfs shell) [20:38] if you want to strace or gdb it you have to do that inside /init [20:38] you can never do it manually [20:46] strace isn't very useful [20:46] kernel panics before it finishs printing debug info [20:47] and gdb will be painful to get commands in [20:50] ogra_ac: indeed [20:50] I'm nearly out of ideas here on how to debug [20:50] I just did a diff on a good initrd vs a bad one and other than binaries the only diff is this: [20:51] diff -ru goodtree/init badtree/init [20:51] --- goodtree/init 2010-12-01 13:41:26.376116002 -0700 [20:51] +++ badtree/init 2010-12-01 13:41:26.796116002 -0700 [20:51] @@ -53,6 +53,11 @@ [20:51] export resume= [20:51] export resume_offset= [20:51] [20:51] +# mdadm needs hostname to be set. This has to be done before the udev rules are called! [20:51] +if [ -f "/etc/hostname" ]; then [20:51] + /bin/hostname -b -F /etc/hostname 2>&1 1>/dev/null [20:51] +fi [20:51] + [20:51] yeah, i checked that already [20:51] # Bring in the main config [20:51] jcrigby: PASTEBIN PLEASE [20:51] . /conf/initramfs.conf [20:51] for conf in conf/conf.d/*; do [20:51] NCommander, I figured it was short enough [20:51] sorry [20:51] and we dont have /etc/hostname inside initrd [20:51] so thats a noop [20:52] jcrigby: can you do a diff to tell us what binaries changed between last known good and current? [20:52] but we also have a new udev and i see some systemd code in it, but then why wouldnt that also fail on other ubuntu images [20:52] ogra_ac: downgrading udev didn't help [20:52] frustatingly enough [20:52] how did you downgrade udev ? [20:53] ogra_ac: apt-get install udev=162-2 && update-initramfs -u [20:53] that wont clean up anything [20:53] just overwrite [20:53] ? [20:53] you could as well do dpkg -x udev*.deb / [20:54] if there are new binaries in the newer udev they will still persist [20:54] ogra_ac: well, do you have any better ideas? [20:54] no [20:54] ogra: why would the new binaries persist? [20:54] because there is nothing removing them [20:55] the only sane way to test that would be to take maverick and step by step upgrade [20:55] ogra_ac: dpkg will remove files that only exist in the old package as part of a downgrade [20:55] but thats painful and will take lots of time [20:55] ogra_ac: I'm out of ideas TBH [20:55] and I don't know a good way to debug run-init [20:56] i dont think its run-init itself [20:57] a good test to see if its actually userspace would be to: boot maverick with init=/foo break=bottom ... then ctrl-d as you do it in natty and see if you get the same error [20:57] or make /sbin/init unexecutable or some such instead of init=/foo [20:57] ok full diff here http://pastebin.ubuntu.com/538798/ [20:57] ogra_ac: exec() is failingin run-init [20:58] exec of the subprocess for /sbin/init, no ? [20:58] so the differences are udev, the kernel, busybox, console-setup, and initramfs-tools [20:58] and some support libs [20:59] busybox ... hmm [20:59] ignore initramfs-tools [20:59] thats only the hostname stuff [21:00] wait-for-root can be ignored, we have it mounted properl at that point [21:00] don't think its libuuid [21:00] libc had an upgrade though [21:00] i think all of the stuff at the bottom thats in sbin can be ignored actually [21:01] libuuid is fine [21:01] I'm pretty sure its not udev [21:01] else you wouldnt have /root mounted [21:01] cause we would have seen breakage during mounting [21:01] libc didnt change between maverick and natty [21:01] i guess libgcc did [21:02] but i wouldnt know what effec that should have [21:02] the best candidate is busybox [21:03] the only big difference we have is the kernel vs x86 [21:03] ogra_ac: glibc had an upgrade [21:03] and linaro as well is on .35 ( jcrigby might correct me) [21:04] NCommander, not in natty [21:04] ogra_ac: have we had any working natty images? [21:04] not that I know of. [21:04] nope [21:05] * ogra_ac doesnt see any eglibc uploads [21:05] nor any syncs that wuld have been processed [21:06] NCommander, the only image that was tested before yesterdays and todays was the one from the 19th and GrueMaster reported the same issue with that one [21:07] Sorry I couldn't debug it further. I found it late that Friday and spend the working part of the following week testing maverick-proposed and doing bug triage. [21:07] dont worry [21:08] NCommander, https://launchpad.net/ubuntu/+source/eglibc 2.12.1-0ubuntu9 in maverick and natty [21:10] tmzt_g2root: yes. that works for me in the ootb une-efl session as well as in the hacked gnome efl session [21:10] tmzt_g2root: (panel and efl) [21:11] tmzt_g2root: did you post to the X ml yet ? [21:11] ogra_ac: I have no idea ATM, it possibly could be udev but that seems unliekly [21:11] GrueMaster: do we know if maverick->natty upgrades work on armel? [21:11] why shouldnt they ? [21:12] at this state of the relase they are manual work anyway [21:12] hasn't been tested yet. [21:13] ogra_ac: because someone wanted to try it, or do dev work? [21:13] well, as i said, it will be a lot of manual work anyway atm [21:13] dmesg [21:13] wrong window (again) [21:13] NCommander, so what we could try is to do piecemeal updates of the bits that end up in intramfs [21:14] and rebuild the initrd after each updated package [21:14] then we should see where it breaks [21:15] ogra_ac: thinking about it right now. I think we're going to miss A1 though [21:15] NCommander, if you dont feel like doing that i will do it tomorrow, its past 10pm here and i dont count on us making A1 anymore so i will soon stop working [21:16] I'm just mentally thinking [21:16] but i think piecemeal upgrade step by step is the best way to find the bad guy [21:21] GrueMaster, btw, are you sure your monitor issue isnt something similar ? i.e. is the board still alive even without screen ? [21:22] * ogra_ac just sees that eglibc in maverick release isnt actually the one in natty [21:23] the one in natty is the same as in -updates and one up from -security [21:29] I reimaged on a faster SD card and haven't been able to reproduce that issue. [21:29] The Sd card I had initial issues with was a class 2 4G (all I had available at the time). [21:29] I'm currently running maverick with the latest proposed updates and other than getting lost during screensaver, it is working fine now. [21:29] Unfortunately, my es2.1 is currently tied up with lmbench testing and NCommander has the other panda. [21:29] I can try updating my XM image to see if I can get it to break. [21:30] that would help too, given that linaro sees the issue with beagles [21:32] if you find anything, dump it on the bug