=== ericm-Zzz is now known as ericm|ubuntu [02:21] is this a good channel to ask about the ARM architecture? [02:21] If it's ubuntu-related, sure [02:22] If not, maaaybe [02:22] I'll take my chances then. I was wondering if the carry_out from the shifter is fed into the ALU? [02:23] in ARM v1-v5 (or so) [02:23] Way above my head [02:23] I can't find a diagram that would show details on this [02:24] ##workingset might know of an appropriate channel, also recommend you lurk here [02:25] aha, thx! === doko_ is now known as doko [09:19] infinity, can you see what is wrong with exo? [09:21] doko: Well, the log just looks like bad quoting in variable assignment... [09:22] FOO=bar baz ./quux ; bash: bar: No such file or directory [09:22] Err, "baz: No such"... But you get the idea. [09:22] Should be FOO="bar baz" ./quux [09:28] must have been blind ... [09:32] WOAH ! [09:32] 0 upgraded, 1021 newly installed, 0 to remove and 0 not upgraded. [09:32] Need to get 325 MB of archives. [09:32] After this operation, 955 MB of additional disk space will be used. [09:32] .... [09:32] thats from apt-get install ubuntu-desktop in my hf chroot ! [09:33] doko, infinity, you guys rock ! [09:33] janimo, can i have an armhf ac100 kernel ? :) [09:37] ppisati: Any progress on ti-omap4 kernels for armhf? [09:37] ogra_: 1021 newly installed? What was that the build-deps for? :) [09:37] build deps ? [09:38] thats just a plain apt-get install ubuntu-desktop [09:38] no build deps :) [09:38] on top of -core btw [09:41] abh, abootimg hasnt been built yet [09:41] *bah even [09:42] Ahh. [09:42] Can fix. [09:42] Anything else you need? [09:43] can you trigger a build without uploading a build1 version ? [09:43] else i'm happy to do that [09:43] There's nothing to trigger, I just had to score it up. [09:43] https://launchpad.net/ubuntu/+source/abootimg/0.6-1/+build/2960109 [09:43] oh, ok [09:43] hi, i have a ecafe slim netbook with a A8 freescale i.MX515 [09:44] Can I have it? [09:44] i'm triying to change OS to Ubuntu from this way https://wiki.ubuntu.com/ARM/MX5/QuickStart [09:44] infinity, stay with the qiockstart :) [09:44] but when i try to boot system dont boot from sd crad [09:44] *card [09:45] sefokuma: The uBoot on that card won't boot your system. [09:45] And the kernel may or may not. [09:45] unlikely that the kernel works, but you should be able to boot with the exiting kernel and bootloader the device ships with [09:45] you will need to do a bit of fiddling [09:46] infinity i select boot from external card with dip switch, i think that netbook must boot from it [09:46] if ubuntu armel for MX5 its prepared with uboot bootable... why dont work? [09:46] sefokuma, the bootloader on the SD wont support that hardware [09:46] ^ [09:46] ogra_ ok [09:47] you need to find a u-boot binary that does and replace the one on the SD [09:47] same goes for the kernel i guess [09:47] ogra_ aham ok [09:47] these images are specifically built for the quickstart board [09:47] oh! ok now i understand [09:47] userspace will work on your device [09:48] i will try install image on SD but overwrite uboot and ulinuximage with originals from guillemot [09:49] right [09:49] infinity: i'm starting it now [09:50] janimo: off topic [09:50] ppisati: Awesome. [09:50] ppisati: I uploaded meta that assumes 1401 will build on armhf. So, now I just need a kernel. ;) [09:50] janimo: is mosh micolay (whatever is the name/pronunciation) today? [09:52] janimo: Nicolai Mosh actually [09:52] ppisati: It should just be a question of "for i in `rgrep -l armel *`; do sed -i -e 's/armel/armel armhf/'; done && for i in find . -name \*armel\*; do (I can't be bothered to type a decent automated version of "cp foo-armel foo-armhf"); done [09:52] infinity: somehow :) [09:53] ppisati: I fact, I can do that right now, if you like, and you can work on newer ABIs and shiny feature merges with 3.2? ;) [09:54] infinity: i'm doing it manuallu right now [09:54] ppisati: I would have done it already, but I didn't want to step on your toes and make you grumpy if you were half done. [09:54] ppisati: Alright, cool. [09:54] ppisati: Just watch for both "rgrep armel *" and "find . -name \*armel\*"... The file copies was what bit apw when he did it. :P [09:54] (As in, he forgot) [09:55] ppisati, very true, you need the d-i bits as well which are in files by arch name [09:57] ogra_: https://launchpad.net/ubuntu/+source/abootimg/0.6-1/+build/2960109/+files/abootimg_0.6-1_armhf.deb [09:58] yay, thanks ! [09:58] now i just need a kernel [09:58] You have omap. Break out a Beagle. [09:58] * ogra_ would change the existing one himself but i really really dont want to touch git [09:58] But I can do ac100 on armhf, if you like. [09:59] well, there is a 3.0 branch ready for testing [09:59] I wasn't going to touch git, I was just going to mangle the packaging and let jani sort it. Not because I dislike git (I quite like it), but because I suspect I can't merge to whatever branch he's using anyway. [09:59] i would assume that janimo is already on it and we get armhf with the next upload [09:59] 3.0 would be shiny, but any working kernel is good for now. [09:59] Though it's unsupported anyway, blah blha. [09:59] So. [09:59] hmm, so you wouldnt touch git either [09:59] Wait for ppisati, and go armhf on your panda? :) [09:59] then i can do it as well [10:00] i want to upgrade my ac100 to precise anyway [10:00] so i can as well try the adventurous cross grade :) [10:02] It won't even remotely work. [10:02] So, like. Don't try. [10:02] pfft [10:02] libc-bin makes the world explode. [10:02] Well, I guess you can try. But be prepared to wipe and reinstall right after. :P [10:02] ogra_: Anyhow. I'm about to do the ac100 armhf thing. Unless you really want to. [10:03] if i unpack -core over my rootfs and then do a simple desktop install .... [10:03] no. do it if you feel like [10:04] Doing. [10:04] Build needed 45:44:37, 8327584k disk space [10:04] ^-- qt4-x11 on a babbage. [10:04] nice [10:04] We can't get rid of those things fast enough... [10:04] It takes, like, 13 hours on a Panda. [10:05] yeah [10:05] and 10min on a calxeda :) [10:05] Build needed 45:44:37, 8327584k disk space> hah, I remember that. Pretty painful [10:06] ogra_, but do you have one [10:06] 10min> nice [10:06] and indeed does anyone [10:06] * apw would like one to build kernels on [10:06] apw, sadly not yet [10:06] in a year from now though .... [10:06] *g* [10:07] then build time ->infinity right now on one ;/ [10:07] yesterday i started a armhf deboostrap... http://paste.ubuntu.com/761449/ for some reason it stalled and have now been stuck for 12h I: Configuring console-setup... [10:07] Sure 1 year + 10 mins ? [10:07] am i the only one hitting this? [10:07] xranby: I've not seen it. [10:08] xranby, did you mount /proc and /sys ? [10:08] i had it doing weird things to my host when it configured procps [10:08] The procps madness only happens on upgrade. [10:08] Not on debootstrap. [10:08] ogra_: no i simply crated the precise-hf foler and executed sudo debootstrap --verbose --arch armhf precise /media/dh0/chroot/precise-hf [10:08] xranby: Ctrl-C and try harder? :P [10:09] oh, right mine was an upgrade [10:09] xranby: Although, I haven't tried a full debootstrap yet. I'm only using --variant=buildd and --variant=minbase [10:09] xranby: (I'd recommend one of those) [10:09] ok i have now used ctrl-c [10:09] will chroot into it and see if it are usable [10:09] It won't be. [10:10] Just wipe it and use --variant=buildd. [10:10] heh ok [10:33] Would Ubuntu work on the new Eee Transformer Prime (including graphics support)? [10:39] soren: Graphics are a big "maybe, but not sure yet". The bigger issue right now will be locked bootloaders. [10:39] soren: So, uhh. We'll see. [10:39] soren: But I suspect a number of people will be buying them to make it happen. :P [10:41] infinity: What will I need to keep an eye on? Is there a tegra X.org driver where this is likely to land (for the graphics stuff, obviously)? [10:42] infinity: ...or is this all kernel-side these days, so keeping an eye on LKML will suffice? [10:43] soren: Oh, we'll probably have non-accelerate video out of the gate. It's binary drivers that will be iffy. [10:43] infinity: --variant=buildd = works for me [10:43] thanks [10:43] xranby: Figured it would. [10:43] soren: But, until we get past the bootloader, I make no promises about... Well... Anything. ;) [10:44] infinity: Oh, cool. Is there a generic graphics sort of thing for ARM as well (akin to VESA)? [10:44] There are a few framebuffer drivers. [10:44] For some large value of "few". [10:49] ogra_: Running a test-build of linux-ac100... [10:49] ogra_: If it's good, I'll upload before bed. [10:54] infinity: Using one of these framebuffer drivers instead of proper accelerated drivers, will that drain the battery with normal use, or only if I try to do a lot of OpenGL stuff, for instance? [10:55] soren: Who knows. The hardware's still a mystery to me until I have one. [10:55] infinity: Alright. I just thought that was the sort of thing you could answer based on past experience. No worries. === brendand_ is now known as brendand [10:56] soren: Well, on the Tegra2 netbook I use, I don't use binary drivers, and I get insanely good battery life. Not sure how useful that data point is, but I imagine Tegra3 stuff will play out similarly. [10:57] infinity: Wicked. Thanks. [11:09] infinity, awesome ! [11:09] * infinity taps foot. [11:10] My Panda needs some go-faster juice. [11:51] infinity: i want to do another thing to P/omap4 before i upload, but if you want i can give you an omap4 armhf kernel now [11:51] ppisati, we need it in the archive [11:51] to build images [12:19] hi again [12:19] i have now a ubuntu-core running on my arm [12:19] but.. i dont know default password of ubuntu-core fs lol [12:19] any idea? [12:19] there is none [12:19] and user¿ [12:19] nor is there a user or any networking config [12:19] you need to configure that by hand [12:20] -core is to build images on top of it, not for plain usage [12:21] yes i know [12:21] but i need first boot and login [12:21] to continue making system [12:21] if you want to use it you need to make a bunch of adjustments before using it (create a user, install the networking bits, configure the system etc [12:21] (before booting into it) [12:21] ok [12:21] i was following his steps [12:22] https://wiki.ubuntu.com/Core - Deploying Ubuntu Core [12:22] yeah, thats not complete yet [12:22] oh ok [12:22] we have a task to update the documentation for precise [12:25] sorry coniecction time out === sefokuma_ is now known as sefokuma [12:41] ok works [12:42] but im thinking on copy armel MX5 rootfs directly to SD cardroot filesystem [13:08] hey folks, does someone know whether someone still cares for the "mobile" packageset? [13:08] nobody does [13:08] same for netbook [13:09] though i'm not sure kubuntu-mobile didnt migrate to just be called mobile [13:09] better ask them :) [13:11] I'm not sure what mobile is anymore. [13:11] It has xfce stuff in it.. [13:12] xfce should build now, unless the desktop team hates xfce even more ... [13:12] is there a web ui to check packagesets contents and upload permissions? [13:12] I figure launchpadlib might expose it [13:12] lool: No idea. I'm pretty packageset ignorant. [13:13] lool: I know that creating/managing them is still an annying by-hand affair with some voodoo. [13:17] lool: edit_acl.py [13:17] Daviey: I'm not an archive admin [13:17] lool: neither am i, but you can use it to check who has upload access where [13:17] and packageset contents [13:18] Daviey: where is it? [13:19] oh wow, lp:ubuntu-archive-tools is public and I never used it [13:19] $ edit_acl.py -P mobile -S precise query [13:19] worked fine, thanks [13:20] groovy [13:20] hildon and xfce and other bits. [13:20] That's so painfully obsolete. [13:20] kill it ! [13:21] it seems we don't have a process for removing sources from packagesets [13:21] I mean binaries from packagesets when we remove sources [13:21] but you should be able to remove the whole packageset [13:21] which in this case would be proper [13:22] lool: Eh? Binaries aren't in packagesets. [13:22] lool: Package set define archive rights, it's source-level ganularity. [13:22] ah right [13:23] indeed these are used for source upload rights [13:27] Ok; going to lie down a bit, I feel too sick [13:56] top - 13:55:57 up 7:18, 0 users, load average: 12.08, 10.66, 9.29 [13:56] And they said the Panda wasn't server hardware... [14:02] ha ha [14:03] infinity: did you do mdad raid via partitions on a single sdcard? [14:03] redundancy++ [15:04] hi [15:05] can anyone help me to change some values for this tutorial https://wiki.ubuntu.com/ARM/BuildEABIChroot? because http://ports.ubuntu.com/ubuntu-ports/dists/karmic/Release doesnt exist [15:13] S0NiC, that howto i sreally really outdated [15:14] S0NiC, if you feel like, please delete it .... [15:15] (else we will delete it along woth the rootfs from scratch page etc) [15:15] *with [15:21] ogra_ can you tell me another howto? [15:22] not yet, we will rework all this with the switch to live-build [15:22] currently the existing howtos refer to obsolete tools adn the new tools arent properly documented yet [15:22] hmm shit i have to crosscompile a programm for amr7 [15:23] as for building a chroot, install qemu-user-static, run qemu-debootstrap [15:23] mom [15:23] that will create an arm chroot on an x86 machine [15:23] at least on systems newer than lucid [15:24] for lucid ask in #linaro for the lucid PPA of qemu [15:24] i am useing maverick [15:24] -.e [15:24] might be ok ... [15:24] have i to specifiy a target-arch= [15:25] http://nopaste.info/cae429e70f.html [15:25] because i get this error? [15:25] i think aemu-debootstrap defaults to armel ... if not, the same options debootstrap uses apply [15:25] *qemu [15:26] yeah, try --arch armel [15:26] beyond that, all other debootstrap options apply [15:28] ogra_ i tried... http://nopaste.info/4493dfff57.html but doesnt work ;( [15:29] well, it tells you what you are missing [15:29] just follow the suggestion the "usage" line gives you [15:30] suite is the release name i.e. oneiric, maverick etc ... [15:30] target is the dir relative to the dir you run the command in [15:33] * ogra_ hugs infinity [15:33] so tomorrow is the image day :) [15:34] ndec, before end of the week we should have armhf images for panda, beagle and ac100 ... [15:35] ogra_: hey! cool. [15:35] ndec, an armhf recompiled gles driver for panda would rock soooo hard :) [15:40] sorry i was away [15:43] ogra_ now it works. thanks [15:43] :) [15:44] S0NiC, does your project have heavy dependencies ? else you could just use the cross compiler [15:44] ogra_ i have a question for my understanding: if its finished i change to the directory and can compile there my program? [15:44] what do you mean by heavy depndencies? [15:44] you can run: sudo chroot [15:45] that will switch your filesystem root to the chroot [15:45] well, libraries etc [15:45] ogra_ iam new to this ;D thanks, i read this already [15:45] i try to compile a qtwebkit program [15:45] i think, thats heavy? ;) [15:45] ah, k, then the chroot approach is probably better [15:46] i did some tries but the conclusion was not soooooooooo good ;) [15:46] if it would have been a cmdline thing that only needs libc that would be lightweight enough to not use a chroot [15:47] e.g. with make i can compile my program on my target (pandaboard) and on my desktop machine. i tried it with g++ and it doesnt work (its a little c++ program) [15:47] i see thx [15:48] ogra_: are armel PPA building for armhf by default as well? [15:48] ndec, not yet [15:48] all that armhf stuff went way faster than expected (after massive probs in the bringup phase we're suddenly far ahead of the plan) [15:49] ndec, but whats there should be enough to re-roll the binary blob in preparation [15:49] i'll take care for PPAs soon [15:50] (i hope we can just make them dual build for both arm arches and be done) [15:50] ogra_ wow that needs a lot of time... to build the chroot [15:50] S0NiC, well, it downloads all packages from the archive [15:50] yes [15:50] then it unpacks them without using dpkg ... [15:51] and then it configures them by running the maintainer scripts ... in a qemu chroot each process is wrapped by a qemu call ... [15:51] so while thats a lot faster than running an actual qemu machine, its indeed a lot slower than if you would build an x86 chroot [15:51] sounds deep in the system ;D [15:52] it is deep in the system ... actually on kernel level (for the wrapping of the processes) [15:53] okay finished no i try to chroot [15:53] okay i chrooted an now i can try to compile my program ? [15:54] you will want to install some bits to actually be able to compile [15:54] ogra_: "dual build for both arches"? [15:54] a chroot like this is a virgin system [15:54] infinity, armel and armhf [15:54] in two runs [15:54] would that be very complex ? [15:54] ogra_: Yeah, I didn't quite get what you were driving at, though. [15:54] orga tahn apt-get updaten and so on? [15:54] ogra_: PPAs get armel and armhf build records. Just like every other arch. [15:54] S0NiC, apt-get update; apt-get install build-essential [15:55] that should get you the very basics [15:55] ah thanks [15:55] infinity, well, the armel ones are currently limited (artificailly i think) to build armel only [15:55] ogra_: Not entirely true. [15:55] S0NiC, and then indeed the -dev packages for the libs your app needs [15:55] ogra_: There's been plenty of armhf PPA activite. [15:55] infinity, ah, cool, i didnt know [15:56] orga thanks [15:56] ogra_: It's only the "virtual" ones that have that strange limitation (and they're not actuall virtual, obviously) [15:56] ogra_: It's... Messy. [15:56] infinity, well, we need to enable ndec's PPA for hf then so he can build us the shiny for panda :) [15:57] Which team/ppa is that again? [15:57] that has a physical builder attached [15:57] TI [15:57] ogra_ sorry for interrupt your conversation but i get an E: Unable to locate package build-essential [15:57] ogra_: That was remarkably unhelpful. [15:57] ti-omap-dev is teh team iirc [15:57] 404. [15:57] I'm thinking not. ;) [15:57] S0NiC, hmm, did you apt-get update ? [15:58] * infinity asks his panda. [15:58] yes [15:58] nopaste.info/e869c03242.html [16:00] ogra_: Oh, I can't tell from looking at it what kind of PPA it is. But it might just magically get armhf build records on the next upload. Dunno. [16:02] ogra_: Oh, I can't tell from looking at it what kind of PPA it is. But it might just magically get armhf build records on the next upload. Dunno. [16:03] infinity, https://launchpad.net/~tiomap-dev/+archive/release [16:03] * ogra_ had a crash here [16:03] chromium actung up usually leads to OOM at some point [16:04] i'll better buy a bucket to stop the leaking [16:04] *acting [16:04] Or open fewer tabs. [16:04] ogra_ i have to leave now, it possible to talk later to you=? [16:04] luckily i dont use a laptop ... so it doesnt leak on my lap ... [16:04] a netbook must leak to the net, right ? [16:06] cu guys and thanks for the help [16:06] ciao [16:50] infinity, can you apply the mmap patch to the ac100 image too? [16:50] doko, i guess thats rather a task for janimo [16:51] (he maintains the git tree) [16:51] ok [16:51] doko: Yeah, I was just enabling armhf in the packaging, jani's working on 3.x source mangling. [16:51] and he doesnt seem around today [16:51] (probably a holiday in romania) [16:51] I believe someone said it was, yeah. [16:52] Anyhow.. [16:52] ogra_: You have an ac100 kernel. [16:52] done already ? [16:52] geez, these pandas are so fast [16:52] ogra_: Can't make images with it yet, cause I can't publish it until the librarian is recovered. :P [16:52] ogra_: But you can download it from LP and install it locally and such. [16:52] right, i plan to roll manual builds tomorrow [16:53] and will then test the full image [16:53] Builds should Just Work. [16:53] Unless I screwed up. ;) [16:53] sure [16:53] But what are the odds, right? [16:53] else i'll fix them [16:54] if they all work fine i wonder how to make them autobuild ... i fear we are getting to a point where a full set of images will take 24h :) [16:54] i guess we should just do some selected areches for the start [16:54] ac100 and omap4 or so [16:55] Well, it's on a different machine. [16:55] sure sure [16:55] And post-processing doesn't take long at all. [16:55] How do I mount the preinstalled image again to replace the kernel, i cant seem to find it on the wiki. [16:55] Xase, just re-plug the card after writing it [16:55] I bet all 4 flavours on armhf will build in about the same time as they do on armel over two buildds. [16:55] right [16:56] Oh lol.. [16:56] thats 4h per build and flavour [16:56] Never thought about tbhat [16:56] it's not even dd'd yet [16:56] klol [16:56] brb [16:56] so 8h for two flavours of the same subarch [16:56] ogra_: Yes, but arches build in parallel. [16:56] 16 for 4 etc etc [16:56] ogra_: My point is that armhf should be about the same speed as armel, so adding it slows nothing down. [16:56] it fuills the few time gaps we have öeft [16:57] *left [16:57] ...? [16:57] What? [16:57] Parallel. [16:57] Same time. [16:57] hi, i'm sure it's good place to ask here. i have issue with hdmi audio working on pandabaord with ubuntu 11.10. analog audio works. for hdmi i get 0,5 sec proper sound then it cut and nothing and only some low clicks. worked fine on 10.10 [16:57] you can only build two in paralell [16:57] What are you talking about? :) [16:57] omap4 is always built on the same builder [16:57] armel+omap4 is. [16:57] we build at least four flavours of omap4 images atm [16:57] armhf+omap4 is another machine. [16:58] we have new live builders ?!? [16:58] Adding armhf slows nothing. [16:58] Yes... [16:58] is there know issue for hdmi audio, i know for some it works fine [16:58] OH ! [16:58] Lamont and I have been debugging the armhf builder for the last two days. :P [16:58] * ogra_ humps infinity's leg [16:58] i DIDNT KNOW ! [16:58] whoops [16:59] i would say disregard the caps but they somehow feel appropriate :) [17:00] haha [17:05] Erm, ogra_, infinity. Kernel panic with the omap armhf netinstall kernel. [17:06] :( [17:06]