| === bivab_away is now known as bivab | ||
| sveinse | I'm running natty and every time I reboot, I need to run "sudo update-binfmts --enable qemu-arm" to be able to run armel binaries on intel. Is there a recent policy change disabling these binfmts at every boot? | 09:39 |
|---|---|---|
| Laney | Could someone spare 2 minutes to try a test build on an arm{el,hf} (shouldn't matter) porterbox for me? http://paste.debian.net/160402/ | 11:24 |
| shadeslayer | rsalveti: poke | 11:36 |
| shadeslayer | Laney: I have a armel pbuilder setup, would that work? | 11:37 |
| Laney | possibly | 11:37 |
| Laney | not sure if ghc will work there, mind | 11:37 |
| shadeslayer | ghc? I'm not quite familiar with that acronym | 11:37 |
| shadeslayer | Laney: uh, this is for debian? I have a precise armel pbuilder ... | 11:38 |
| Laney | ghc is a haskell compiler, and the target distro shouldn't matter | 11:38 |
| shadeslayer | ok | 11:38 |
| shadeslayer | Laney: uh, I didn't have haskell-devscripts, it's going to take a bit :) | 11:41 |
| shadeslayer | about 10 minutes or so | 11:41 |
| shadeslayer | ok so I have a problem where upstream stores double precision values in databases | 11:43 |
| shadeslayer | but that won't work on arm obviously | 11:43 |
| shadeslayer | so how do I work around databases getting tainted between arch's ? | 11:43 |
| shadeslayer | ( the data being stored is geolocation data, seems like it's important to store it in double ) | 11:44 |
| janimo` | shadeslayer, the ARMv7 processors that Ubuntu runs on have good double precision support | 11:46 |
| janimo` | even hardware registers for it :) | 11:46 |
| shadeslayer | janimo`: ok weird, then why does the build fail ? | 11:46 |
| shadeslayer | lemme get the build log | 11:46 |
| shadeslayer | https://launchpad.net/ubuntu/+source/digikam/4:2.5.0-1ubuntu1/+build/3215910 | 11:47 |
| shadeslayer | https://launchpad.net/ubuntu/+source/digikam/4:2.5.0-1ubuntu1/+build/3215911 | 11:47 |
| shadeslayer | /build/buildd/digikam-2.5.0/extra/kipi-plugins/photolayoutseditor/widgets/canvas/CropWidgetItem.cpp:343:62: error: no matching function for call to 'qMin(qreal, double)' | 11:47 |
| shadeslayer | and ever since I made that to qMin(qreal, qreal) everything works just fine | 11:48 |
| janimo` | shadeslayer, that is a Qt programming issue that only occurs on ARM and indeed has roots in ARM traditionally not being good at double precision | 11:49 |
| janimo` | qreal is defined as float on ARM and double elsewhere in Qt | 11:50 |
| shadeslayer | ok, so for ARM7 and above Qt should now define qreal as double? | 11:50 |
| shadeslayer | and this would make it a Qt issue | 11:50 |
| janimo` | when qreal and double are used interchangeably that is not an issue (x86) | 11:50 |
| janimo` | but when that code gets built on ARM float != double and gcc complain | 11:50 |
| shadeslayer | right | 11:50 |
| shadeslayer | janimo`: so essentially, qreal needs to be typedef'd to double on ARMv7 and above | 11:51 |
| janimo` | I am not sure it is possible anymore to change it back on ARMv7 in Qt, it likely breaks linking with binaries using Qt before the change | 11:51 |
| janimo` | so qreal stays float on ARM, but code needs to be careful and use qreal qhen qreal is meant and not use double liberally in its place | 11:52 |
| shadeslayer | hmm, yes, but I could report this issue upstream, might be fixed in time for Qt5 | 11:52 |
| janimo` | shadeslayer, that would be good indeed, Qt5 could be a place to fix this | 11:52 |
| janimo` | until then the kde apps using qMin must pass two qreals to that function | 11:52 |
| shadeslayer | gotcha | 11:52 |
| janimo` | so not identifiers declared as double | 11:53 |
| janimo` | lots of KDE or Qt based app have this issue still in Ubuntu and likely upstream | 11:53 |
| janimo` | Laney, will check your patch, took a while to get all build-deps sorted in a chroot | 11:54 |
| Laney | janimo`: thanks. I think it's straightforward - skein.cabal just doesn't set the define for unknown arches and skein_port.h errors with #error if it isn't set | 11:56 |
| janimo` | Laney, I just added that debian/rules change and now waiting for it to finish one way or another | 11:56 |
| janimo` | Laney, yes, finished successfully with that change. | 11:58 |
| Laney | janimo`: great thanks, uploading to debian now then | 11:59 |
| janimo` | Laney, cheers | 11:59 |
| janimo` | Laney, no public ARM porter boxes for you to play with when needed? | 12:00 |
| Laney | there are, but you have to wait a while for BDs to be installed | 12:00 |
| Laney | thought it might be quicker to ask here ;-) | 12:00 |
| janimo` | Laney, sure np, just wanted to make sure you are not blocked on this when testing ARM changes ;) | 12:01 |
| === chrisccoulson_ is now known as chrisccoulson | ||
| jadahl | am I supposed to be able to easily install ARM versions of libraries alongside the others on my x86_64 system in Ubuntu 12.04? | 13:36 |
| jadahl | my purpose of doing so is cross compiling and running in QVM on the same computer | 13:37 |
| infinity | dannf: Your highbank branch ( https://code.launchpad.net/~dannf/flash-kernel/highbank-support ) doesn't include updates to debian/f-k-i.postinst | 14:10 |
| dannf | infinity: ok - i'll take a look. | 14:10 |
| infinity | dannf: Don't ask why all that code is duplicated, it is what it is. :P | 14:10 |
| dannf | infinity: do you know if we'll be rebasing on debian's version at any point? | 14:12 |
| infinity | dannf: Not before precise. | 14:13 |
| infinity | dannf: But yes, some day we'll switch. | 14:13 |
| dannf | *nod* | 14:13 |
| ogra_ | :q | 14:18 |
| infinity | E37: No write since last change (add ! to override) | 14:19 |
| ogra_ | hehe | 14:20 |
| ogra_ | silly unity, alt-tab never gets me where i expect to land | 14:21 |
| jadahl | are the repository containing arm packages not under archive.ubuntu.com ? | 14:52 |
| ogra_ | jadahl, nope, they never were | 14:54 |
| infinity | jadahl: ports.ubuntu.com/ubuntu-ports | 14:54 |
| jadahl | ah | 14:54 |
| ogra_ | armhf (and the old armel) are on ports.ubuntu.com | 14:54 |
| infinity | (and powerpc) | 14:54 |
| jadahl | nice, thanks | 14:54 |
| infinity | More easily stated as everything !x86. :P | 14:54 |
| === bivab is now known as bivab_away | ||
| rsalveti | GrueMaster: lightdm seems to be quite slow even on panda | 15:07 |
| rsalveti | at least while moving the cursor the cpu gets easily consumed | 15:07 |
| GrueMaster | rsalveti: I haven't been paying attention. Working on arm server workload test development. | 15:07 |
| rsalveti | after launching unity-2d it works fine, but back to lightdm makes it slow again | 15:07 |
| rsalveti | GrueMaster: ok, will open a bug for that | 15:08 |
| rsalveti | would like to have perf available to check, but that is another bug | 15:08 |
| rsalveti | ppisati: were you able to check the broken perf bug? | 15:08 |
| rsalveti | ogra_: infinity: attached a debdiff at bug 959928 with the fix | 15:13 |
| ubot2` | Launchpad bug 959928 in xorg-server "Driver fallback for ARM loads the vendor driver twice and gets invalidated with the error" [High,In progress] https://launchpad.net/bugs/959928 | 15:13 |
| rsalveti | so we can have the pvr driver loaded without a xorg.conf file | 15:13 |
| rsalveti | another fix needed to have the working pvr driver at the archive | 15:13 |
| GrueMaster | It would be super cool to have the drivers available for Beta 2. Is this the armhf version? | 15:17 |
| rsalveti | GrueMaster: yup | 15:18 |
| rsalveti | GrueMaster: I got the drivers already in place, updating the bug in a few | 15:18 |
| GrueMaster | awesome. | 15:18 |
| rsalveti | would need a review and someone to sponsor | 15:18 |
| rsalveti | and this xorg fix to be included as well | 15:18 |
| rsalveti | GrueMaster: infinity: ogra_: bug 959924 properly updated | 15:40 |
| ubot2` | Launchpad bug 959924 in ubuntu "[needs-packaging] pvr sgx driver and kernel module for Pandaboard" [Wishlist,In progress] https://launchpad.net/bugs/959924 | 15:40 |
| rsalveti | needs a review and someone to sponsor the packages as well | 15:40 |
| rsalveti | just tested and it worked fine here on both 4430 and 4460 | 15:41 |
| rsalveti | for ARMHF | 15:41 |
| ogra_ | thx | 15:41 |
| rsalveti | once that lands we can also enable the notification for users to install the driver once they install the images | 15:41 |
| rsalveti | like what we have for radeon or nvidia | 15:41 |
| ogra_ | you mean jockey | 15:47 |
| ogra_ | that will likely require some changes | 15:47 |
| rsalveti | ogra_: yes | 15:49 |
| rsalveti | ogra_: something on my todo to check | 15:49 |
| rsalveti | would be great to have that notification | 15:50 |
| rsalveti | and with just one click, boom, sgx working :-) | 15:50 |
| rsalveti | and then people can install TI's PPA for a newer driver, with the newer kernel and such | 15:50 |
| rsalveti | but the default is already enough to get it to work with opengles applications | 15:50 |
| ogra_ | yeah | 15:53 |
| ppisati | ops | 15:56 |
| ppisati | rsalveti: checking it | 15:56 |
| rsalveti | ppisati: thanks | 15:56 |
| ppisati | rsalveti: it seems it's in the code ($somewhere) | 15:56 |
| ppisati | rsalveti: it's not a config issue, and upstream is ok | 15:56 |
| rsalveti | ppisati: would be useful to check with jessi and andy, I remember we had a similar issue in the past with 3.1 | 15:57 |
| rsalveti | might be missing one patch or two | 15:57 |
| ppisati | rsalveti: i'll shot them an email then, thanks | 15:57 |
| rsalveti | ppisati: great | 15:57 |
| jadahl | yay. it's very possible to run arm binaries seamingly under x86_64 (via qemu) just as if they were normal binaries | 16:01 |
| jadahl | and install third party libraries into /usr/arm-linux-gnueabi | 16:02 |
| ppisati | rsalveti: you mean jesse barker? | 16:03 |
| rsalveti | ppisati: http://www.linaro.org/about/meet-the-team/jaswinder-singh/ | 16:04 |
| ppisati | rsalveti: k | 16:05 |
| GrueMaster | Wishlist? rsalveti, your pvr sgx packaging bug has been marked as a wishlist item. I would think it is higher importance than that. | 16:10 |
| rsalveti | GrueMaster: that's the default for needs-packaging | 16:10 |
| rsalveti | afaik | 16:10 |
| GrueMaster | ah, ok. | 16:10 |
| rsalveti | ppisati: also, can you check why the patches for cross compiling to make it compatible with dkms got reverted? | 16:15 |
| rsalveti | ppisati: bug 666267 | 16:15 |
| ubot2` | Launchpad bug 666267 in linux-ti-omap4 "Cross compiled headers package breaks DKMS compilation" [Medium,Confirmed] https://launchpad.net/bugs/666267 | 16:15 |
| rsalveti | that's useful for TI, as they want to be able to cross compile the kernel and still be able to use dkms-based packages | 16:15 |
| ppisati | rsalveti: becaseu they broke cross-cmpiling in a chroot | 16:15 |
| ppisati | iirc | 16:15 |
| rsalveti | ppisati: can't we just fix? | 16:15 |
| rsalveti | *fix that? | 16:16 |
| ppisati | rsalveti: so, these patches fix the DKMS problem ok? i'll see what we can do | 16:16 |
| rsalveti | ppisati: great, thanks :-) | 16:17 |
| rsalveti | you'll make ti happy with that | 16:17 |
| === bivab_away is now known as bivab | ||
| ppisati | rsalveti: will they make my battery last longer in that case? :) | 16:20 |
| rsalveti | ppisati: haha, don't think so ;-) | 16:21 |
| hstimer | Anyone here using arm server under qemu? | 17:18 |
| danboid | If I download the latest precise desktop build can I expect to get a picture out of HDMI on my panda or should I just stick with the server image? | 17:37 |
| danboid | (during install) | 17:38 |
| danboid | as I presume the desktop builds still don't support serial console output by default? | 17:39 |
| GrueMaster | danboid: The desktop images will never support serial console by default, as that is not the nature of desktop images. | 17:43 |
| GrueMaster | It has always supported HDMI video. | 17:44 |
| danboid | GrueMaster, Hi! | 17:46 |
| danboid | GrueMaster, Last time I tried a daily for precise desktop armhf I got no picture via hdmi - not even a fb console | 17:47 |
| danboid | GrueMaster, but that was before it was supposed to work | 17:47 |
| GrueMaster | That's odd. I'm running the daily from yesterday, works just fine here. | 17:47 |
| danboid | GrueMaster, about 2 weeks ago :) | 17:47 |
| GrueMaster | Are you using the hdmi port closest to the usb/nic plugs? | 17:48 |
| danboid | yep | 17:48 |
| danboid | I'll try the desktop one then | 17:48 |
| danboid | it ,ay be fine now | 17:48 |
| GrueMaster | Beta 1 also worked. | 17:48 |
| GrueMaster | How are you powering your panda? | 17:48 |
| danboid | GrueMaster, OK, thats handy to know | 17:48 |
| danboid | GrueMaster, 5V 4A PSU | 17:49 |
| GrueMaster | ok. Sometimes I see people having issues when powering over min-usb from a laptop. | 17:49 |
| danboid | GrueMaster, Do you know of any interesting apps that support GLES except KDE? What about epsxe? | 17:49 |
| danboid | GrueMaster, or any good games of course! :) | 17:50 |
| GrueMaster | Not off hand. I haven't done much testing on that because before now, video drivers were post-release. | 17:50 |
| GrueMaster | You could try pts or Open Arena. | 17:50 |
| danboid | Yes, Q3 is the only other thing I know of | 17:51 |
| danboid | (open arena) | 17:51 |
| danboid | I'm hoping epsxe or that other psemu support GLES | 17:52 |
| danboid | Thats something I'll be looking into soon | 17:52 |
| danboid | I just updated the main ubuntu kernel compilation guide to cater for us arm (Linaro) users btw | 17:54 |
| danboid | ie how to know which config file to edit :) | 17:54 |
| danboid | is the blaze actually a working cell phone too? | 17:59 |
| GrueMaster | I don't know. Might be. Not sure if it has 3G or what. | 18:02 |
| === yofel_ is now known as yofel | ||
| prpplague | rsalveti: ping | 19:33 |
| rsalveti | prpplague: pong | 20:29 |
| prpplague | rsalveti: just fyi, i wasn't planning to attend Connect | 20:30 |
| prpplague | rsalveti: otherwise i would be happy to present if i were going | 20:31 |
| rsalveti | prpplague: yeah, would be nice, but it's a quite long trip | 20:33 |
| rsalveti | guess we'll be able to present something at least | 20:33 |
| rsalveti | with help of jcrigby | 20:33 |
| prpplague | rsalveti: hehe, don't mind the length of the trip, but the $$ , hehe | 20:33 |
| prpplague | rsalveti: dandy! | 20:33 |
| rsalveti | prpplague: yeah, must be quite expensive, didn't yet check | 20:34 |
| rsalveti | prpplague: hopefully we'll be able to show the support for the main boards we're maintaining at linaro | 20:34 |
| === bivab is now known as bivab_away | ||
| djshotglass | cheapest arm board i can buy today? | 23:46 |
| djshotglass | only needs usb and ethernet or wifi | 23:46 |
| rsalveti | djshotglass: | 23:49 |
| rsalveti | djshotglass: I'd recommend a beaglebone | 23:49 |
| djshotglass | thats almost $100 :/ | 23:49 |
| djshotglass | i was looking for something like $25 | 23:49 |
| djshotglass | because i need 4 of them | 23:50 |
| rsalveti | then you'd need to grab a raspberry pi | 23:50 |
| djshotglass | cant get a rasp for months | 23:50 |
| djshotglass | pogoplugs on ebay for $25ish | 23:50 |
| djshotglass | hmm | 23:50 |
| rsalveti | that might work | 23:51 |
| GrueMaster | Just remember that neither pogoplug or raspberry pi will work with ubuntu. | 23:51 |
| djshotglass | what why | 23:51 |
| rsalveti | yup, you'd need to use debian | 23:51 |
| rsalveti | because they are not armv7 | 23:52 |
| GrueMaster | Ubuntu==armv7. | 23:52 |
| djshotglass | meh i can live with that | 23:52 |
| GrueMaster | Cheapest armv7 (that I know of) is the beaglebone. http://beagleboard.org | 23:52 |
| rsalveti | you might be able to buy some used beagleboards as well on ebay | 23:54 |
| rsalveti | or similar | 23:54 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!