| === bivab_away is now known as bivab | |
| [09:39] <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? | |
| [11:24] <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/ | |
| [11:36] <shadeslayer> rsalveti: poke | |
| [11:37] <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 | |
| [11:38] <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 | |
| [11:41] <shadeslayer> Laney: uh, I didn't have haskell-devscripts, it's going to take a bit :) | |
| [11:41] <shadeslayer> about 10 minutes or so | |
| [11:43] <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 ? | |
| [11:44] <shadeslayer> ( the data being stored is geolocation data, seems like it's important to store it in double ) | |
| [11:46] <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 | |
| [11:47] <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)' | |
| [11:48] <shadeslayer> and ever since I made that to qMin(qreal, qreal) everything works just fine | |
| [11:49] <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 | |
| [11:50] <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 | |
| [11:51] <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 | |
| [11:52] <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 | |
| [11:53] <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 | |
| [11:54] <janimo`> Laney, will check your patch, took a while to get all build-deps sorted in a chroot | |
| [11:56] <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 | |
| [11:58] <janimo`> Laney, yes, finished successfully with that change. | |
| [11:59] <Laney> janimo`: great thanks, uploading to debian now then | |
| [11:59] <janimo`> Laney, cheers | |
| [12:00] <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 ;-) | |
| [12:01] <janimo`> Laney, sure np, just wanted to make sure you are not blocked on this when testing ARM changes ;) | |
| === chrisccoulson_ is now known as chrisccoulson | |
| [13:36] <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? | |
| [13:37] <jadahl> my purpose of doing so is cross compiling and running in QVM on the same computer | |
| [14:10] <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 | |
| [14:12] <dannf> infinity: do you know if we'll be rebasing on debian's version at any point? | |
| [14:13] <infinity> dannf: Not before precise. | |
| [14:13] <infinity> dannf: But yes, some day we'll switch. | |
| [14:13] <dannf> *nod* | |
| [14:18] <ogra_> :q | |
| [14:19] <infinity> E37: No write since last change (add ! to override) | |
| [14:20] <ogra_> hehe | |
| [14:21] <ogra_> silly unity, alt-tab never gets me where i expect to land | |
| [14:52] <jadahl> are the repository containing arm packages not under archive.ubuntu.com ? | |
| [14:54] <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 | |
| === bivab is now known as bivab_away | |
| [15:07] <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 | |
| [15:08] <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? | |
| [15:13] <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 | |
| [15:17] <GrueMaster> It would be super cool to have the drivers available for Beta 2. Is this the armhf version? | |
| [15:18] <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 | |
| [15:40] <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 | |
| [15:41] <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 | |
| [15:47] <ogra_> you mean jockey | |
| [15:47] <ogra_> that will likely require some changes | |
| [15:49] <rsalveti> ogra_: yes | |
| [15:49] <rsalveti> ogra_: something on my todo to check | |
| [15:50] <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 | |
| [15:53] <ogra_> yeah | |
| [15:56] <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 | |
| [15:57] <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 | |
| [16:01] <jadahl> yay. it's very possible to run arm binaries seamingly under x86_64 (via qemu) just as if they were normal binaries | |
| [16:02] <jadahl> and install third party libraries into /usr/arm-linux-gnueabi | |
| [16:03] <ppisati> rsalveti: you mean jesse barker? | |
| [16:04] <rsalveti> ppisati: http://www.linaro.org/about/meet-the-team/jaswinder-singh/ | |
| [16:05] <ppisati> rsalveti: k | |
| [16:10] <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. | |
| [16:15] <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? | |
| [16:16] <rsalveti> *fix that? | |
| [16:16] <ppisati> rsalveti: so, these patches fix the DKMS problem ok? i'll see what we can do | |
| [16:17] <rsalveti> ppisati: great, thanks :-) | |
| [16:17] <rsalveti> you'll make ti happy with that | |
| === bivab_away is now known as bivab | |
| [16:20] <ppisati> rsalveti: will they make my battery last longer in that case? :) | |
| [16:21] <rsalveti> ppisati: haha, don't think so ;-) | |
| [17:18] <hstimer> Anyone here using arm server under qemu? | |
| [17:37] <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? | |
| [17:38] <danboid> (during install) | |
| [17:39] <danboid> as I presume the desktop builds still don't support serial console output by default? | |
| [17:43] <GrueMaster> danboid: The desktop images will never support serial console by default, as that is not the nature of desktop images. | |
| [17:44] <GrueMaster> It has always supported HDMI video. | |
| [17:46] <danboid> GrueMaster, Hi! | |
| [17:47] <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 :) | |
| [17:48] <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 | |
| [17:49] <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? | |
| [17:50] <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. | |
| [17:51] <danboid> Yes, Q3 is the only other thing I know of | |
| [17:51] <danboid> (open arena) | |
| [17:52] <danboid> I'm hoping epsxe or that other psemu support GLES | |
| [17:52] <danboid> Thats something I'll be looking into soon | |
| [17:54] <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 :) | |
| [17:59] <danboid> is the blaze actually a working cell phone too? | |
| [18:02] <GrueMaster> I don't know. Might be. Not sure if it has 3G or what. | |
| === yofel_ is now known as yofel | |
| [19:33] <prpplague> rsalveti: ping | |
| [20:29] <rsalveti> prpplague: pong | |
| [20:30] <prpplague> rsalveti: just fyi, i wasn't planning to attend Connect | |
| [20:31] <prpplague> rsalveti: otherwise i would be happy to present if i were going | |
| [20:33] <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! | |
| [20:34] <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 | |
| === bivab is now known as bivab_away | |
| [23:46] <djshotglass> cheapest arm board i can buy today? | |
| [23:46] <djshotglass> only needs usb and ethernet or wifi | |
| [23:49] <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 | |
| [23:50] <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 | |
| [23:51] <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 | |
| [23:52] <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 | |
| [23:54] <rsalveti> you might be able to buy some used beagleboards as well on ebay | |
| [23:54] <rsalveti> or similar | |