[00:14] persia, your dnybook, is that a 16G model ? [00:15] *dyna [00:16] ogra_: this won't give us working sound on ac100, would it? [00:16] NCommander, well [00:17] NCommander, we have a codec driver ... we have working devices in pulse ... we just dont have anyone who has a clue how to wire up the sound devices in the driver internally [00:17] ogra_: i.e., needs kernel hacking on an ALSA level, right? [00:17] yes [00:20] ogra_: what's partition 6? (I've already flashed my device several times) [00:20] NCommander, LNX [00:21] ogra_: handy, also, don't I have to reflash the partition map? [00:22] no, just part 6 [00:22] seems Daviey already ran into the first issue :( [00:23] ogra_: great. Last quesiton, your instructions seem to suggest that I need to use a USB key, or can I just use an SD card? [00:23] what do you want to achieve ? an internal install or one to external media ? [00:24] for internal it doesnt matter where you put the tarball, it will find it on SD as well as USB [00:25] for external you just need to make sure to put the tarball onto the media you dont want to install to :) [00:26] (and boot with both plugged in) [00:27] ah [00:56] Daviey: were you able to get your AC100 to work with the Android flashing tools? [01:18] ogra_: your installer crashed :-/ === tgall_foo is now known as Dr_Who [01:22] ogra_: Yeah. Mine is 16G === prpplague^2 is now known as prpplague [01:55] * NCommander has had no luck with ogra_'s installer [01:55] it appears /var doesn't get mounted [01:55] or something, as there are a lot of errors to that before it goes belly up [02:16] /var shouldn't be on a separate partition. [02:16] (unless this is the overlay hack) [02:17] Do you have a promt of any sort? === asac_ is now known as asac [03:45] I can't connect an app to the local xserver from ssh on natty, anyone know why? === zyga-afk is now known as zyga [11:28] persia, i'm looking for a /proc/partitions dump from such a dynabook [11:36] What is the most powerful android phone on the market? [11:36] probably something yu should ask in an android channel ? [11:36] (that uses an ARM cpu) [11:39] okay sorry wrong channel..., does 11.04 run on arm? as far as i can see, rootstock builds karmic or jaunty images? or do i have an old version? [11:40] use >= maverick [11:40] for recent htcs [11:43] ogra_: the touchscreen rotation problem was solved by cnd [11:45] awesome [11:46] Daviey, ha ! i found a hack to open a terminal in oem-config [11:46] not really helpful since it runs as oem user though [11:46] ogra_: it would be helpful to add on screen keyboard in oem-config [11:47] the same assistive technology menu that's also in gdm login screen [11:47] i think there is a bug open for that [11:47] yes but it doesn't receive any attention [11:47] all the tablet devices on the market but nobody cares [11:48] * ogra_ would first of all like to see florence integrated [11:50] we have onboard [11:50] I think it would be nicer to keep onboard and implement florence's auto show/hide features [11:51] https://bugs.launchpad.net/ubuntu/+source/oem-config/+bug/626055 oem-config [11:51] Ubuntu bug 626055 in smartphone "oem-config: make on-screen keyboard available" [Critical,Confirmed] [11:52] https://bugs.launchpad.net/smartphone/+bug/443986 [11:52] at-spi [11:52] Ubuntu bug 443986 in smartphone "RFE: Add option to automatically show and hide onboard" [Undecided,New] [11:53] ogra_: ahha! [11:54] ogra_: now all i need is a root exploit. [11:54] heh [11:55] does the samsung nexus s work with arm based ubuntu? [11:55] i can reproduce the bug here btw [11:55] ogra_: Oh? What do you think it is? [11:55] no idea yet, but i used my brandnew ac100 and for internal install it exposes the same issue [11:56] i will try an external install now so i can check logs [11:56] im thinking something like the htc desire z with a keyboard would be nice for an ubuntu install [11:56] interesting [11:56] i didnt have it on either installs to SD or USB [11:56] Daviey: yes [11:56] ogra_: If you want me to test anything, give me a shout. [11:56] damo22: yes, I don't see a problem [11:57] what about the keyboard mapping? [11:57] damo22: it is google nexus s though. not samsung nexus s [11:58] damo22: the keyboard drivers present to userspace as normal keyboards so you can remap just like you would remap any other keyboard inlinux [12:39] aha, i think i found the issue with the ac100 installer [12:41] ogra_: do you know why http://packages.ubuntu.com/natty/kde/kde-config-qt-graphicssystem is missing in natty arm ? [12:42] likely because it depends on some gles stuff that wasnt built [12:42] just a guess though [12:43] https://bugs.launchpad.net/unity-2d/+bug/782326 [12:43] Ubuntu bug 782326 in unity-2d "unity-2d: flickers with xf86-video-fbdev on some machines" [Undecided,New] [12:43] heh [12:43] * ogra_ never used unity-2d on x86 [12:43] : [12:51] ok, issue verified [12:53] * ogra_ doesnt get why the error pops up during kbd config, its totally unrelated [12:54] NCommander, it was just the initramfs getting to big, nothing with /var ... i'll roll a fixed image later today [12:55] this 4MB restriction for the initrd is really hard to handle :( [12:55] ogra_: could I just get a headless tarball? [12:55] :-) [12:55] i can roll one, sure, but not right now [12:55] oneiric will have both :) [12:55] <_persia> NCommander: How badly do you want one? I've an old one (but it has all the bugs ogra fixed in the past while) [12:56] since half the server team has ac100s now [12:56] <_persia> (not headless, but doesn't even try to perform an install) [12:56] so we need to give them server like images for their netbooks :) [12:56] <_persia> Ugh. No. [12:57] * ogra_ wishes he could find a way to use a bigger initrd than 4M somehow [12:57] <_persia> We need to integrate LXC and libvirt so that folks can play. [12:57] _persia, already happening ;) [12:57] * ogra_ already had requests to add all the needed config bits to the ac100 kernel ... [12:58] -2-ac100 should be fully lxc capable [12:59] <_persia> ogra_: There are plenty of non-kernel issues to make it seamless still :) [13:00] * lilstevie is looking forward to seeing what oneiric will bring [13:00] _persia, sure, but the platform support is there [13:01] That helps. [13:11] ogra_: how would I go about further investigating lack of pacakge [13:11] ogra_: is there a way to recommend a package being added in the arm ? [13:12] ubuntu ARM :) [13:12] <_persia> dcordes: Is the package missing entirely in Ubuntu, or just for armel? [13:13] For the most part we try *not* to do anything architecture-specific (this is not always avoidable) [13:13] persia: it is missing just for armel. [13:14] persia: http://packages.ubuntu.com/natty/kde/kde-config-qt-graphicssystem [13:15] dcordes, Take a look at the output of `apt-cache showsrc ${PACKAGE} | grep ^Architecture` [13:15] persia: ok [13:15] In this case, it's "any", which means that LP should attempt to build it for every architecture. [13:16] Next, check https://launchpad.net/ubuntu/+source/${SRCPACKAGE}/ and dig into the latest upload for your release [13:18] kcm-qt-graphicssystem (The binary 'kde-config-qt-graphicssystem' is part of the kcm-qt-graphicssystem package) [13:19] Right. Now, if you look at the details for 1.2-0ubuntu1, you'll see the list of "Package files". The ones ending "_armel.deb" indicate that armel binaries were generated. [13:20] You can verify they have been published by checking https://ports.ubuntu.com/ubuntu-ports/pool/universe/k/kcm-qt-graphicssystem/ [13:22] You can check to make sure the "universe" component is enabled in /etc/apt.sources.list (if you need to enable it, be sure to update your apt cache (`apt-get update` is one way to do this)) [13:22] Alternately, if you're using a mirror, check to make sure the package appears on the mirror. [13:22] (some mirrors only mirror subsets of Ubuntu, which is fine) [13:24] persia: ok [13:25] Oh, and if you're concerned about the output at packages.ubuntu.com : the system that uses to collect information is completely unaware of anything other than i386 and amd64, so it can never be right. [13:26] ogra, rsalveti are out initrd images lz compressed now? [13:27] janimo, not that i know of [13:27] the uInitrd are marked as gzip but inside it seems is is lz [13:27] just try file [13:27] persia: it would be nice to have a nicer web interface to the armel packages. [13:27] file says :data [13:27] (on the initrd.img) [13:27] After this operation, 150 MB of additional disk space will be used. [13:27] I am not sure if it's so good to use so much extra space just for https://bugs.launchpad.net/unity-2d/+bug/782326 [13:27] Ubuntu bug 782326 in unity-2d "unity-2d: flickers with xf86-video-fbdev on some machines" [Undecided,New] [13:28] dcordes, The number of packages that differ between architectures is well less than 1% of the archive. I agree it would be nice to be able to export unified dists information, but that's pending a stack of other things, and will probably be some time. [13:29] persia: how about simple preliminary search feature ? [13:29] Hrm? What do you mean? [13:29] something more comfortable than https://ports.ubuntu.com/ubuntu-ports/pool [13:30] What are you trying to accomplish? [13:30] I want to fix https://bugs.launchpad.net/unity-2d/+bug/782326 [13:30] Ubuntu bug 782326 in unity-2d "unity-2d: flickers with xf86-video-fbdev on some machines" [Undecided,New] [13:31] I need to run all QT programs with this paramter: "-graphicssystem native" [13:31] No. You need Qt to properly select "native" as the graphicssystem to use when presented with that hardware. [13:33] Ah, there seems to be a blueprint about that, but it appears to be "not started". jazh appears to have been working on it. [13:35] apachelogger reported that Qt supports the QT_GRAPHICSSYSTEM environment variable, which might be a lighter-weight workaround in the meantime (see http://apachelog.wordpress.com/2010/09/05/qt-graphics-system-kcm/ ) [13:37] (note that you'll need to configure the environment variable so that gnome-session sets it, as startkde is unlikely to run in a unity environment) [13:41] ogra, I am looking at the uInitrd files in the preinstalled 11.04 and oneiric images and they seem to be lzma [13:41] where are they generated? [13:42] janimo: as part of livecd-rootfs for preinstalls [13:43] I am trying to debug the SD alignment issue so I need to change jasper scripts in initrd [13:43] i dont think they stay lzma after install [13:43] ogra_: with the ac100, the 'install' step completes fine, oem-config explodes into confetti [13:43] NCommander, yeah, thats caused by a to big initrd [13:43] ah [13:44] * NCommander vaguely wonders how difficult porting u-boot to the ac100 would be. */pipedream* [13:44] i messed around with a diversion of plymouth in initrd, somehow that didnt work, so plymouth ends up in the initrd [13:44] * Daviey vaguely wonders if NCommander will ever respond to his mails. [13:44] NCommander, ah indeed. INITRD_COMPRESSOR=lzma there [13:44] NCommander, already ported, the prob is that kbd and display dont work [13:44] ogra_, probably not after install [13:44] janimo, right, but the installed initramfs-tools default to gzip [13:45] so with the first local regeneration they switch [13:45] ogra_: I take it someone found the pads for the serial port [13:45] ogra_, which reminds me. Why does the VFAT need to be rewritten when resizing on first boot? Isn't only the 2nd partition expanding [13:45] ? [13:45] NCommander, yes [13:45] Daviey: er, what email? [13:46] janimo, to make sure MLO lands in the first sector again ... if that bug is gone (which i heard should be with next x-loader) we can drop that crap [13:46] NCommander: Subject: Ubuntu ARM Server release for 11.10 (server-o-arm-server) [13:46] apparently the whole CHS stuff is fixed [13:47] ogra_, but does the resize not only entail increasing the number of cylinders? the first partition parameters should stay the same [13:47] Daviey: ack, sorry, I missed it the first time around [13:47] I also read that bug is gone for a while already [13:47] * NCommander sees your follow up [13:48] * janimo tries to modify existing uInitrd and convince panda to take it without saying [13:48] Daviey: davidm is working to schelude a call on a regular basis so we can sync wr.t. to our tasks and progress on said tasks [13:48] wrong image format for source command === prpplague is now known as prpplague^2 [13:51] NCommander: super! Thanks. [14:09] persia: ah I remember I tried that and failed automatically setting this (profile.d) [14:10] Isn't profile.d sourced by shells, rather than the session managers? [14:11] ah :) [14:16] dcordes, If you're creating a spice package, stuff something in /etc/X11/Xsession.d : If you want to do something for your user, put something in $HOME/.xsessionrc [14:17] There are very likely to be display-manager- and session-manager-specific ways to do this, but I suspect something general is more likely to meet your needs in a flexible manner. [14:17] Xsession.d is fine I already have a script there for screen rotation [14:18] and to remove the 'overlay scrollbars' [14:18] export LIBOVERLAY_SCROLLBAR=0 [14:19] I will use that script as a global Xsession.d env script [14:20] hello all [14:20] hey Vaati [14:20] so, is this also for general arm dev ? [14:21] Daviey, NCommander, new installer image is up [14:21] should work now [14:21] ogra_: those are famous last works [14:21] and words too :) [14:22] lol [14:22] but, am I allowed to ask non-specific questions about arm development? [14:22] well, the installer is fine, i'm not yet sure my diversion hackery survives a dist-upgrade though [14:22] 'cos ##arm is rather silent... [14:22] * ogra_ is just testing that [14:24] how can I test if Xorg parsed my environment variables ? [14:24] Vaati, We're mostly focused on integration of stuff and making sure all of Ubuntu is available on ARM (and as many devices as possible are supported by Ubuntu). [14:24] ok [14:24] If you have questions related to how things work, we can try, but we may not be the ideal resource. [14:25] And if you're writing arch-specific code that isn't kernel or driver related, some folk will encourage you to be portable :) [14:25] ok, well I'm trying to set up a gcc toolchain for the STM32 vl discovery board (arm7 cortex m3) under ubuntu [14:25] i.e.; I want to use ubuntu to program it [14:26] and I've found the stm32flash software, but apparently it requires a usart, so I cant use the on-board jtag... however, someone created a bootloader called versaloon -- anyone familar with it? [14:28] Heh. That's an area we probably ought to investigate: using ARM as a host for embedded development. [14:28] aww yeeaaah [14:29] We usually focus on running Ubuntu on the ARM chip (but I don't believe the m3 implements all the instructions used in Ubuntu) [14:29] ok [14:29] persia: I put "export QT_GRAPHICSSYSTEM=native" in /etc/X11/Xsession.d/54setupenv [14:29] well, has someone ported puppy linux to the m3 yet? [14:29] or some other form of *nix [14:30] ogra_: oooo [14:31] ogra_: just boot-installer-2.6.37-1-ac100.img, same tgz? [14:31] yep [14:31] tgz didnt change [14:31] super! [14:31] thanks for your attention to this ogra_ [14:31] To the m3? I'm not sure it *can* run Linux. If it can, the Yocto folk are most likely to have a working environment. [14:31] persia: in fact even if I run "QT_GRAPHICSSYSTEM=native DISPLAY=:0 unity-2d-launcher" from a remote terminal, it still doesn't work [14:31] yocto? [14:32] Daviey, well, i want to have proper oneiric images so the natty ones are my testbed [14:32] ah, found a port [14:32] all the ac100 stuff will go to the archive soon [14:32] http://www.linux-arm.org/LinuxKernel/LinuxM3 [14:32] if you were wondering [14:33] Vaati, You'll probably need ulibc with that: there's no MMU [14:33] ogra_: i know i might be a PITA but we could get like a minimal install without unity so just like a cli and ssh? [14:34] zul, read my conversation with persia above ;) [14:34] on my TODO [14:34] ogra_: cool thanks [14:34] persia: maybe unity-2d was compiled in a way it will not parse the QT_GRAPHICSSYTEM env var ? [14:35] That would be odd, but perhaps. [14:35] omg [14:35] I think I stumbled across a solution to a toolchain... [14:35] without even looking for it [14:36] sigh ... updates to gnome-user-guide always make me think my system is dead [14:36] at any rate -- what are the minimum specs for running ubuntu on an arm? [14:36] i wonder whats its doing all that time [14:37] ogra_, There's something annoying at the FS layer that needs sorting. It's the same reason cp is slow to install. [14:37] well, all other packages arent that slow [14:37] Vaati, support for the ARMv7a ISA [14:37] it must do something special [14:37] ogra_, They don't have anything *near* the file count. [14:38] (unless I misremember g-u-g) [14:38] ah, now it moved [14:47] ok, i take everything i said about g-u-g back, ubuntu-docs is worse [14:53] we have ubuntu docs? [14:54] Yes. [14:54] A single source generates a package to populate yelp *and* the majority of the contents of help.ubuntu.com [14:57] Bug 791852 [14:57] Launchpad bug 791852 in unity-2d "unity-2d: does not parse QT_GRAPHICSSYSTEM env var" [Undecided,New] https://launchpad.net/bugs/791852 [14:59] Except that's not something unity-2d is *supposed* to do: that should happen in Qt. [15:00] persia: I don't understand. could you elaborate ? [15:02] Qt clients specifically should never care which graphicssystem is in use. They should just call into Qt abstractions. Qt then should turn those abstractions into something that actually does something on the system (frequently through additional abstraction layers). [15:03] persia: so at what place must the env var be present for the clients to get the correct graphicssystem ? [15:05] At the process level. So, `QT_GRAPHICSSYSTEM=native unity-2d-launcher` ought do the right thing. But this is because Qt notices the environment variable, not because unity-2d does. [15:05] this is what I do but it does not work. so I can assume something with Qt is fishy ? [15:10] I believe so. I believe that the bug has the wrong task. I'm not sure precisely *what* is responsible for handling QT_GRAPHICSSYSTEM though, but I suspect the bug would get attention from the right folk if retargeted to the right place (mind you, finding this probably requires more investigation) [15:11] ok I will leave it as is, affecting unity-2d, and hope for the unity folks to know what is the correct project affected :) [15:20] zumbi_, hrw: I think the conclusion was that eglibc is misdetecting the new triplet as being non-eabi [15:20] wrong channel [15:20] * dmart sighs === transitlogger is now known as apachelogger === apachelogger is now known as randalogger [17:05] ogra_ is man of the match, your latest boot img worked greated [17:06] awesome ! [17:08] GrueMaster: Linux omap 2.6.39-3-omap #10 Thu Jun 2 17:41:26 CEST 2011 armv7l GNU/Linux [17:08] GrueMaster: i have usb [17:08] GrueMaster: beagle xm rev a [17:08] GrueMaster: i just ssh-ed in [17:08] Is this the same kernel that is in the current images? [17:09] GrueMaster: uhm no, latest oneiric [17:09] That's what I mean. [17:10] GrueMaster: ok, but then the bug will be fixed with the next release [17:10] The kernel I have is 2.6.39.3.4 [17:10] Ah. [17:10] flag@omap:~$ cat /proc/version_signature [17:10] Ubuntu 2.6.39-3.10-omap 2.6.39 [17:10] ok [17:10] GrueMaster: can you double check with latest oneiric? [17:10] and wait [17:11] one more check [17:11] Texas Instruments X-Loader 1.5.0 (Apr 11 2011 - 09:47:05) [17:11] U-Boot 2011.03 (Apr 20 2011 - 07:19:53) [17:11] do they correspond? [17:11] Latest image is 20110601, which I already tested. [17:11] * ppisati wonders if correspond is the correct word at all... [17:11] x-loader & u-boot should be the same. [17:12] GrueMaster: please check [17:12] GrueMaster: i did a lot experimentation [17:12] Give me a sec to setup. [17:13] no hurry [17:14] x-loader & u-boot are the same on the 20110601 image. [17:14] ok [17:14] /proc/version_signature? [17:16] That should be in the bug report. [17:16] right [17:16] but is the same as mine... [17:16] uhmmm... [17:17] ok, i'll install that image then [17:19] I recommend the headless image, otherwise you won't get very far. [17:21] GrueMaster: k [17:26] * ppisati -> out for a walk... [18:24] ogra_: Is there a way to restore factory default Toshiba image after installing your image? [18:27] GrueMaster, yes, you can restore the full backup you took before changing anything, vnflash should always work [18:28] *nvflash [18:28] Good to know. [18:28] indeed that requires that you actually backed up everything first :) [18:32] I haven't even started messing with it beyond plugging into power. [18:32] oh, you got one too now ? [18:32] (not michaels) [18:32] Got it yesterday. [18:32] wow [18:32] its like an epidemic thing how these things pop up around me :) [18:38] lol [18:53] ogra_: Is this expected, nvec nvec.0: new transaction during send trying to retransmit! ? [18:53] Every minute or so in dmesg [18:53] Daviey, yes, thats a workaround in the kernel that makes your input devices not die [18:54] in the former kernel revision the complete input system locked up every time you see this message ... and only a shutdown, making the device 30sec powerless and reboot could make it work again [18:54] ah [18:55] we somehow get event storms on the nvidia event controller, nobody knows why yet [18:55] if you see that message in demsg it actually means nvec eas restarted [18:55] *was [19:47] what is the easiest way to preseed a few values on the preinstalled headless image? I'd like to skip most installer questions as I run fresh images frequently and that slows the process down [19:47] janimo, well, currently only setting them on kernel cmdline is supported [19:48] but that works well [19:48] can I set cmdline from uboot or only from boot.scr? [19:49] you could do it manually from u-boot shell but would have to manually type in the whole boot.scr content indeed [19:52] ok. On the cmdline I pass a preseed file path or even the actual preseeded value list? [19:52] only values [19:52] and keys [19:52] as i said above, files arent supported yet, only setting preseed options directly on cmdline [19:53] for oneiric i plan to add file support though [20:05] Can't you set it to pull a preseed from http? [20:05] I've not dealt with preseeds before. [20:06] GrueMaster, the prob is that you need code to process it ... where it comes from doesnt matter much [20:06] I thought oem-config could handle it. [20:06] oem-config *consumes* preseeding [20:07] and the code processing preseed that i added last release simply only reads /proc/cmdline atm [20:07] hmmm. [20:07] right, what persia said [20:07] the code *setting* the options from file or cmdline is whats matters [20:08] that should be trivial to add though, will have it in place before A2 [20:08] preseed files are a requirement for server images === Sarvatt_ is now known as Sarvatt [23:24] Hey everyone! [23:24] grmpf [23:26] * ogra fights with oeniric