| infinity | janimo: I'd assume it's because we moved the toolchain to ARMv5, but perhaps haven't told GHC about it. | 01:25 |
|---|---|---|
| infinity | janimo: If you don't have the time to poke it, I can. | 01:25 |
| infinity | janimo: My suspicion seems to be confirmed by the build log. Grabbing the source now to have a look. | 01:30 |
| twb | Don't forget you can ask #debian-haskell (OFTC) for advice re the ghc side of things | 01:32 |
| suihkulokki | ubuntu armel is armv5 again? | 01:33 |
| infinity | suihkulokki: SOrt of. | 01:38 |
| infinity | twb: I suspect it's actuall llvm that needs some love. Or maybe ghc just needs a violent re-bootstrap. | 01:38 |
| twb | Last time I looked (karmic?) GHC on arm went via C | 01:39 |
| twb | Didn't know they had transitioned to llvm | 01:39 |
| twb | Oh and (at least at the time) ghc didn't support arm, really, it was just the debian haskell team making it kinda-work | 01:40 |
| suihkulokki | infinity: you need to update the topic then :) | 01:56 |
| hrw | suihkulokki: toolchain is armv5 but that does not mean that ubuntu/armel is armv5 again | 02:11 |
| hrw | suihkulokki: no one rebuilt whole archive for it | 02:11 |
| twb | What does "toolchain is armv5" mean? That just those packages (presumably gcc, ld, etc) are compiled to run on armv5, though packages compiled WITH them target armv7? | 02:25 |
| hrw | twb: right | 02:40 |
| hrw | twb: ubuntu/armel was armv7 so most of packages still are. new ones will be armv5 | 02:41 |
| int_ua | Hi. Looks like we have a newer somewhat stable kernel for the Nokia N900. Can someone help with packaging it and publishing on some server, better in the official repository? | 02:44 |
| twb | hrw: why the backflip? Just because of raspberry? | 02:47 |
| === twb` is now known as twb | ||
| suihkulokki | twb: there is now armhf so having two armv7 ports doesn't make much sense | 02:49 |
| twb | Oh. | 02:52 |
| hrw | twb: who care about r/pi? | 02:52 |
| twb | hrw: exactly | 02:53 |
| twb | armhf branch isn't mentioned anywhere on https://wiki.ubuntu.com/ARM btw | 02:56 |
| jackh | rasberry is ARMv5 | 03:01 |
| infinity | jackh: It's v6. | 03:01 |
| jackh | twb: you mean ubunbu willl support that pi? | 03:02 |
| infinity | twb: Hrm, wiki/ARM seems pretty out of date in general. | 03:02 |
| jackh | infinity: ok, thanks for correct me | 03:02 |
| infinity | jackh: We're making no such commitment. | 03:02 |
| infinity | jackh: If we don't kill the armel port entirely, it might end up supporting the Pi, but that's still up in the air. | 03:02 |
| jackh | infinity: seems lot of work of packaging, testing, fixing? | 03:04 |
| infinity | jackh: Right now, we're just doing an organic/lazy rebuild. But, like I said, we may still drop the port entirely, so no promises. | 03:06 |
| jackh | infinity: seems not many devices and dev-boards are supported by ubuntu now | 03:06 |
| jackh | infinity: i plan to do some work to support some of them | 03:07 |
| infinity | omap, omap4, mx53, ac100, plus all the boards that Linaro ships kernels for... | 03:07 |
| infinity | We have no real interest in building 50 images for every possible target. | 03:07 |
| int_ua | jackh: what about N900? | 03:08 |
| jackh | infinity: as you know they are too expensive, not suitable for young guys to buy | 03:08 |
| jackh | infinity: i am try to do some on A8 systems, which is very cheap here | 03:08 |
| infinity | jackh: 200 USD is "expensive"? My, how times have changed. | 03:08 |
| jackh | infinity: Pi is 25bugs, you know that | 03:09 |
| infinity | Yes, but it's also not something you want to run a general prupose OS on. | 03:09 |
| infinity | An Ubuntu desktop wouldn't actually boot on it even if we were v6. | 03:10 |
| jackh | infinity: also hdmi is not pop here, i dont think they have hdmi tv | 03:10 |
| jackh | infinity: so i will do the things from the minimal rootfs and adds things on, if i meet any problem, i will disturb you:) | 03:12 |
| twb | infinity: these plans/changes will be set in stone as at the next release? | 03:16 |
| infinity | twb: It's more about if someone wants to sponsor the port. Canonical's not really keen on doing armel for free. | 03:17 |
| twb | Nod. | 03:18 |
| twb | I was just confused because I thought it had already been decided (like, two releases ago) that Ubuntu's armel was gonna be v7 and HF and the only arm port Ubuntu had | 03:18 |
| infinity | ..? | 03:19 |
| infinity | armel was never going to be HF. | 03:19 |
| infinity | Hence armhf. | 03:19 |
| twb | I guess I was misinformed | 03:19 |
| twb | Or rather, I misunderstood | 03:19 |
| suihkulokki | I guess twb is confusing hf with softfloat abi vs hf abi | 03:19 |
| twb | I'll STFU now... | 03:19 |
| suihkulokki | it is confusing, no need to feel shame :P | 03:20 |
| jackh | hmm, infinity: suihkulokki: hf is hard floating, abi uses some mix of soft and hard floating, right? | 03:21 |
| twb | In my head "hf" means you must have a hardware FPU | 03:21 |
| infinity | Yeah, that's not what it means. | 03:22 |
| twb | Has someone written this stuff down somewhere? If so I'll just go read that | 03:22 |
| infinity | soft is software FP, softfp is hardware FP, but soft's calling conventions, hard is hardware FP, and using a new calling convention that uses the VFP registers. | 03:22 |
| infinity | soft ansd softfp are ABI-compatible, hard isn't. | 03:23 |
| twb | IOW softfp will work on either, and can use a hardware FPU, but will have minor overheads compared to hard? Whereas soft will simply ignore any hardware FPU? | 03:24 |
| twb | Or does "hard" actually work without a hardware FPU? | 03:24 |
| jackh | twb: iow, hf seems like a small intruction set | 03:26 |
| jackh | twb: the instructions will ask for and get result from some hardware units | 03:26 |
| suihkulokki | only softfloat works if you do not have VFP | 03:28 |
| twb | OK | 03:28 |
| twb | I guess all I really care about is making sure I pick the best build for whatever hardware is in front of me | 03:29 |
| twb | (Where "best" balances performance and effort - e.g. maybe it is worth recompiling the kernel, but not the entire distro, for my specific unit.) | 03:30 |
| jackh | twb: sorry, i still dont get what you want to do | 03:31 |
| twb | jackh: buy a computer, then make it go SUPER FAST | 03:31 |
| twb | Don't worry about it, I'm just talking shit | 03:31 |
| jackh | twb: i got it | 03:31 |
| jackh | twb: but that's x86 stuff, nothing to do with this channel | 03:32 |
| twb | Er, not on arm boxes | 03:32 |
| twb | Although my tf101 is still running oneiric and android's crappy kernel because I am actually too lazy to do anyhting but talk | 03:33 |
| recur | anyone know how to get Xfbdev to load fonts? My embedded device needs something nicer than the fixed font :D | 05:25 |
| twb | There are broadly two font subsystems; the "x font system", which deals with bitmap fonts, and xft, which deals with outline fonts. | 05:48 |
| twb | GTK2/3 and Qt3/4 are xft-based; almost everything is. xterm uses xft if you say e.g. "xterm -fa monospace" | 05:49 |
| twb | MOST apps that use xft, use a second library "fontconfig" to turn a logical font name like "Sans 12" into the path to a font file. | 05:50 |
| twb | Anyway, if it's an xft app I wouldn't expect anything special to be needed on the X server side, it's all X app side | 05:50 |
| recur | ohh cool | 05:52 |
| recur | thanks two, this is good info | 05:52 |
| recur | er twb, stupid spellcorrect ;) | 05:52 |
| twb | Sigh. | 05:54 |
| twb | You're about the fifth person in the last month to do that; it seems that everyone decided to turn on spell-checking at the same time... | 05:55 |
| twb | It's probably a new "feature" in pidgin or whatever the hell is popular these days... | 05:55 |
| furan | stupid question. I am working on desktop ubuntu with libavcodec-52, but when I installed libavcodec on my arm device I got -53. Is there any way to install -52 + dev? | 05:56 |
| recur | I'm using Textual, it's disabled now :D | 05:56 |
| twb | furan: if both boxes are running the same release, they should have the same version | 05:57 |
| furan | they apparently aren't | 05:58 |
| LetoThe2nd | maybe ont the desktop box there's medibuntu or such. | 05:58 |
| twb | furan: back- or forward-porting is not something you should attempt unless you are an experienced packager | 05:59 |
| twb | And that goes double for installing stuff directly (i.e. "make; make install" rather than "apt-get install") | 06:00 |
| furan | desktop:Linux version 2.6.32-41-generic (buildd@vernadsky) (gcc version 4.4.3 (Ubuntu 4.4.3-4ubuntu5.1) ) #89-Ubuntu SMP Fri Apr 27 22:22:09 UTC 2012 | 06:00 |
| furan | Linux version 3.2.0-psp7 (root@panda-a1-1gb) (gcc version 4.6.3 (Ubuntu/Linaro 4.6.3-1ubuntu4) ) #1 Fri Apr 13 04:55:05 UTC 2012 | 06:00 |
| furan | and that's the device | 06:00 |
| twb | furan: use lsb_release -a to check the relases, and look at "apt-cache policy" output to see if they have rogue package sources | 06:00 |
| twb | That sounds like lucid vs. precise | 06:00 |
| furan | hah, I guess I installed lucid instead of precise on my vmware instance | 06:01 |
| furan | thanks | 06:01 |
| furan | ugh now i've gotta change all my libavcodec stuff | 06:01 |
| furan | oh well | 06:01 |
| twb | it's better in the long run iMO | 06:01 |
| janimo | infinity, good to know you're on ghc then :) | 06:08 |
| janimo | also good to hear we try an armv5 armel flavour even if just experimenting | 06:08 |
| === zumbi_ is now known as Guest88650 | ||
| ogra_ | hmpf, i guess that got swallowed by my reconnect | 11:26 |
| ogra_ | janimo, any news about the new ac100 kernel ? | 11:26 |
| ogra_ | (would be nice to have something for A1 the binary drivers work with ... which means at least 3.1) | 11:26 |
| janimo | ogra_, I will have a look, would be nice indeed. I did no check marvin24's tree in almost a month I think | 11:26 |
| ogra_ | well, i'm using a self rolled 3.1 since the new binary driver came out | 11:28 |
| ogra_ | despite the driver bugs it seems ot work just fine | 11:28 |
| janimo | ogra_, I still hope all tegra2 stuff gets mainlined for 3.6 and maybe catches 12.10 :) | 11:28 |
| ogra_ | but you wanted to switch to 3.2 you said in the past, that will require somce signficant changes (devicetree) | 11:29 |
| ogra_ | 3.6 would be nice but i wouldnt count on it | 11:29 |
| janimo | I wanted to switch to whichever kernel is least hassle to maitain :) | 11:30 |
| janimo | if possible as new as possible to be in line with ubuntu | 11:30 |
| ogra_ | sure | 11:30 |
| lilstevie | ogra_: you don't have any issues with 3.1 | 11:31 |
| ogra_ | it would be nice to have working images for the quantal milestones though, since we have a policy that images have to work nowadays | 11:31 |
| ogra_ | i dont, no | 11:31 |
| lilstevie | I still can't figure out what is causing my hang | 11:31 |
| ogra_ | seems as stable as 3.0 was | 11:31 |
| lilstevie | as long as I break it is fine | 11:32 |
| ogra_ | lilstevie, try removing plymouth | 11:32 |
| ogra_ | (by removing the plymouth hooks and scripts from /usr/share/initramfs-tools and rebuilding the initrd) | 11:33 |
| lilstevie | when I ask it to remove plymouth it wants to remove everything | 11:33 |
| lilstevie | ah | 11:33 |
| ogra_ | yeah, mountall needs libplymouth iirc | 11:34 |
| lilstevie | well waiting for bonnie++ to finish then I will try without plymouth | 11:35 |
| lilstevie | also gpower is being a little trigger happy with the dual battery | 11:38 |
| janimo | ogra_, btw IIRC you packaged the tegra armhf drivers right? | 11:43 |
| ogra_ | http://people.canonical.com/~ogra/nvidia-tegra/ | 11:43 |
| ogra_ | waiting for a newer kernel before i upload to the archive | 11:43 |
| janimo | ogra_, ah they completely fail with old kernel, ok | 11:44 |
| janimo | both armel and armhf in there? | 11:44 |
| ogra_ | nope, only armhf | 11:44 |
| janimo | or we drop armel completely for quantal | 11:44 |
| ogra_ | still waiting for mgmt decision ... as it seems we will keep it around in teh oinfrastructure but not touch it at all | 11:45 |
| janimo | lilstevie, do you know what the status of mainlining tegra2 support is? | 11:45 |
| lilstevie | janimo: not good afaik | 11:46 |
| lilstevie | ogra_: removed the plymouth hook altogether from initramfs and it is still doing it | 11:53 |
| ogra_ | try also disabling the upstart plymouth jobs in /etc/init | 11:53 |
| lilstevie | ok | 11:54 |
| ogra_ | or make plymouth non executable or move it to .orig and macke /usr/bin/plymouth a link to /bin/true | 11:55 |
| ogra_ | oh, and indeed, drop "splash" from the cmdline :) | 11:56 |
| lilstevie | I don't have splash on the cmdline | 11:56 |
| lilstevie | ogra_: symlinking plymouth and plymouthd with /bin/true worked | 12:05 |
| ogra_ | must be some issue with tegrafb or so | 12:06 |
| lilstevie | probably | 12:06 |
| ogra_ | i saw that too, but only on tegra devices with 3.1 and newer kernels | 12:06 |
| lilstevie | this is with 3.1 | 12:06 |
| ogra_ | ah, yeah | 12:07 |
| ogra_ | i was told 3.2 would be better with that | 12:07 |
| lilstevie | hm | 12:07 |
| janimo | lilstevie, do you know if android and ubuntu can dual-boot on a tf101? | 15:54 |
| janimo | adn whether there are other than 3g hw differences between tf101 and tf101g? | 15:55 |
| === int_ua_ is now known as int_ua | ||
| lilstevie | janimo: depends how you mean by dual-boot | 22:33 |
| lilstevie | janimo: I do it at the cost of recovery, but we have recently started getting kexec and kexecboot working | 22:34 |
| lilstevie | janimo: as for the difference between the tf101 and tf101g; they are simple, 3G modem attached to mini pci-e port, and different sbk set in the cpu, the rest of the hardware is 100% identical | 22:47 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!