/srv/irclogs.ubuntu.com/2012/03/26/#ubuntu-arm.txt

=== zumbi is now known as Guest18413
lilstevieinfinity, ok01:25
lilstevieI did notice it only happens on HF01:25
lilstevieprecise updated from oneiric doesn't give me the suspend warning01:26
lilsteviebut the hf precise image I have on my sdcard does01:26
michaelh1Hey, does anyone know who in Ubuntu creates the ARM UEC images http://uec-images.ubuntu.com/precise/current/ and the tool used?02:01
infinitylilstevie: That's probably less to do with el versus hf and more to do with upgrade versus fresh, I'd guess.02:10
infinitylilstevie: As in, maybe the old gnome-power-manager got it right, and its settings were carried over on upgrade (thanks to them being in your ~)02:11
infinitylilstevie: Could test that theory with your old ~ on a fresh precise/hf image, probably.02:11
infinitymichaelh1: Not sure exactly where those come from, you might try asking smoser.02:12
lilstevieinfinity, fair enough, oneiric never displayed the battery stats correctly anyway (stupid android driver) but the new power-manager or upower compensates02:13
lilstevieinfinity, I guess the other bug I have experienced is pulse audio crashing02:14
michaelh1infinity: ta, I'll send him an email02:14
lilsteviehad to remove pulse just for lightdm to not die02:15
twbmeh, that just reinforces my low opinion of pcmanfm's work02:16
lilstevietwb, ?02:20
twbnever mind02:30
twb(pcmanfm makes most of the LXDE bits)02:31
lilsteviethe issue isn't lightdm it is pulseaudio02:37
twbYeah well I like lennart even less :-)02:37
lilstevieany process that outputs sound with it will either crash, or go haywire02:37
infinitylilstevie: Reproduce, file bugs, etc.02:42
infinitylilstevie: Better still if you can reproduce on x86 too, but if it really is arm-specific, we should still deal with it. :P02:42
lilstevieinfinity, ok, I just don't know exactly how to get the bug, or reproduce bug, it is probably my kernel02:43
infinitylilstevie: Well, if it's your kernel, I might care less about a bug filing. :P02:43
lilstevieI can reliably reproduce on both tf101 and tf20102:43
lilsteviebut it would have been observed on the panda02:43
lilstevieor well, any device :p02:44
infinitySpeaking of the TF... Are we still in "no way in hell" land on trying to do a unified -tegra kernel?02:44
lilsteviewell, with mtype hacks we could02:44
lilsteviesomeone did something creative with boot.img creation that dynamically patched the mtype02:44
infinityThat sounds vile.02:45
lilstevieheh02:45
lilstevieyeah02:45
infinityWait, binary patched the kernel, or just mangled something else in the abootimg?02:45
lilstevieI have no idea02:45
lilstevieI wasn't too interested cause it sounded vile02:45
lilstevie:p02:45
infinityCause if we could get one KERNEL to work, but just needed different boot images, that's easily done.02:45
lilstevielet me check02:46
lilstevieyeah, if it just mangles abootimg I'm sure kernel-img.conf could have the device mtype02:46
infinityI'd happily ship two boot images (and let flash-kernel sort it for upgrades), if we could have a unified kernel.02:46
lilstevieyeah02:46
infinityflash-kernel, not kernel-img.conf, but yeah.02:46
lilsteviewell I am almost certain with a little work that tf101 and tf201 should be able to share a kernel02:47
lilsteviethat is tegra2/tegra302:47
infinityAnd AC100, and that Toshiba tablet that was similar, and...? ;)02:47
infinityWell, anything that's fastboot+tegra should, in theory, be doable.  One would hope.02:48
lilstevieI mean without hacks02:48
lilstevietf101 does not fastboot02:48
infinityuBoot tegra platforms (like the Trimslice) might be a bit more of a pipe dream for a unified kernel.02:48
infinityOh, tf101 doesn't?02:48
lilstevieno, well it uses the "fast boot" bootloader02:48
infinityabootimg kinda implies fastboot, doesn't it?02:48
lilsteviebut there is no fast boot access02:48
lilstevienvflash access is your only way of manipulating images02:49
infinityWell, sure.  That's true of the ac100 as well.02:49
lilsteviealso Trimslice isn't that pipedream02:49
lilstevieI recently saw on the u-boot mailing list a bootz patch02:50
lilstevieto boot zImage02:50
lilstevieand really02:50
infinityYeahp.02:50
infinityDon't know if we'll land that for precise.02:50
lilsteviethat would solve the problem02:50
infinityAnd even if we do, probably won't go an ditch all th uImages.02:50
lilsteviecertainly not for precise02:50
infinityBut it's a sane way forward.02:50
lilstevieprecise is only like what a week away02:50
lilstevie:p02:50
lilsteviewell a few weeks02:50
infinityA month.02:50
infinityBut yeah, "soon".02:50
infinityToo soon.02:51
infinityThese cycles always seem so long for non-LTS releases, and so damn short when I'm trying to make everything "just right" for an LTS. :/02:51
lilsteviehttps://github.com/astarasikov/mkbootimg-mtype-hacked02:52
lilstevietakes mtype as an arg02:52
lilstevielooks like it patches the zImage02:53
infinityYeah, I just got there.02:54
infinitySo much for that idea.02:54
lilstevieI guess the only other way is generic board that heuristically determines the platform02:54
infinityI wonder if one could parse mtype from the cmdline, or if we need it too early to be able to.02:55
lilsteviehm02:55
infinityWell, obviously, the real solution is being able to detect, yes.02:55
infinityBut I'm trying to think of quick hacks. ;)02:55
lilstevieheh02:55
infinitycmdline parsing happens pretty damned early, though.02:56
infinitySo, that should be doable.02:56
lilstevieyeah the question is if it happens quite early enough02:56
lilsteviethere are certain fuse structures that could be used for the tf10102:56
infinityWell, given that you need it early enough for nosmp to rewrite spinlocks, and for mem= to blacklist regions, it's got to be... Early.02:57
lilstevieheh actually that is a good point02:57
lilsteviewhen is memory blacklisting done?03:02
lilstevieactually I guess I should find the routine that parses the command line03:03
lilstevieinfinity, this may be viable03:20
lilstevieinfinity, it looks like the command line is parsed before mtype is processed03:21
lilstevieso using that you could add something like mtype= and patch it03:22
lilstevieinfinity, I guess the biggest question right now is how many tegra devices do we have04:00
lilstevieand how many mtypes04:00
lilsteviecause ac100 is harmony mtype, tf101 is ventana, and tf201 is cardhu04:01
student_needing_can someone help me get ubuntu to boot on my beaglebone? i make an sdcard with the premade 11.10 image and looking at screen i am stuck on "PHY 0:01 not found" and it won't go any farther04:15
infinitylilstevie: As long as they don't have any mutually exclusive drivers, then, this may be entirely viable.04:28
lilstevieinfinity, well thats where we could probably hack around a bit04:29
infinitystudent_needing_: You might want to try a 12.04 omap image.04:29
infinitystudent_needing_: Though I don't have a Beaglebone, so I can't be much help.04:29
lilstevietry to make it a little more modular04:29
student_needing_i can try out the new image, didn't know how stable it is though04:30
infinitystudent_needing_: In general, I'm happier with precise than with oneiric.  And this is a good time to avoid churn, as we're in a beta freeze, so the most recent dailies are fairly sane.04:31
student_needing_sounds good, downloading that image right now04:31
student_needing_one other question about ubuntu on the beaglebone, does it have access to the gpio pins like angstrom would, being able to read and set values on them?  is it the same process as angstrom?04:45
michaelh1student_needing_: here's my notes on the BeagleBone: http://juju.net.nz/michaelh/notes/bone.html04:58
=== michaelh1 is now known as michaelh1|away
twbI have a PHP weenie who is currently running PHP lucene and OOMing the resources he has been allocated.05:36
twbHe wants to migrate to "solr", but AFAICT that needs a full JVM *and* tomcat on top of that, so I'm not convinced it'll be much of a win.05:36
twbOops, this isn't #ubuntu-server05:36
lilstevielol'05:36
need_helpso im still trying to boot my beaglebone into ubuntu and im running into a lot of trouble05:52
need_helpright now on /dev/ttyUSB0 when i press a key it tries to start up ubuntu on /dev/ttyUSB1 (i'm using screen into both) but always fails, gets stuck at "PHY 0:01 not found"05:53
need_helpnow my /dev/ttyUSB1 has a login prompt for Ubuntu precise but doesn't respond to keystrokes, only /dev/ttyUSB0 responds to keystrokes05:54
need_helpi also just saw on /dev/ttyUSB1 that i get this error a little before the other error - "init: ureadahead main process (176) terminated with status 5"05:56
=== Guest14119 is now known as panda|x201
=== rsalveti` is now known as rsalveti
hrwhttp://marcin.juszkiewicz.com.pl/2012/03/26/ubuntu-12-04-precise-and-cross-compilation-of-arm-kernels/ - refreshed instructions for precise10:32
LetoThe2ndhrw: ah nice10:36
LetoThe2ndhrw: does dpkg-buildpackage in this context accept a local .config, or does it have to be added to some patchset before?10:37
=== arun is now known as arun__
ogra_hrw, you should add a step "make your changes here" or some such, i doubt anyone will just rtebuild the kernel package for fun without making any changes10:39
LetoThe2ndogra_: thats the consequence ;)10:39
hrwogra_: that part is left for reader. I do not remember when last time I understood how ubuntu kernels are maintained10:39
hrwI prefer 'make uImage modules -j44'10:40
ogra_LetoThe2nd, fakeroot debian/rules editconfigs; dpkg-builpackage -b -aarmhf10:40
LetoThe2ndogra_: *note*10:40
hrwogra_: feel free to add comment ;)10:40
ogra_(that brings up menuconfig with the configs from the debian dir)10:40
* suihkulokki finds it annoying how debian/ubuntu kernels make a simple kernel build process very complicated10:41
suihkulokkialthough I understand why it ends up being so complex10:41
ogra_suihkulokki, how is that complicated ?10:41
hrwogra_: providing own kernel config needs more work then it is worth?10:42
LetoThe2ndsame here, given the addition from ogra i see no more complexity than in a manual biuld.10:42
ogra_fakeroot debian/rules editconfigs; dpkg-builpackage -b -aarmhf .... vs. .... make menuconfig; make ... i dont see much difference apart from the convenience of having a package with the first one (which also brongs the proper postinst handling for flash-kernel etc)10:42
ogra_*brings10:43
hrwflash-kernel... I hope that it will die ugly death sooner the later10:43
ogra_it never will (sadly)10:43
ogra_but it will get better in 12.1010:43
ogra_well, fsvo never ... once we have arm support in grub it wont be flash-kernel but update-grub ;)10:44
hrwfinally rewrite?10:44
ogra_yeah10:44
ogra_its already in debian testing ... but needs some massive changes for ubuntu10:44
hrwogra_: for me it can be ubmaumba even as long as it works the same on each machine10:44
ogra_thats why i say grub ;)10:44
hrwand that on mx53 it cares of uImage/uInitrd but not boot.scr which it cares of on omap3 with saying 'not supported' on pandaboard at same time10:45
ogra_well, get the venros on UEFI and you wont have these probs ;)10:45
ogra_*vendors10:45
suihkulokkithen we'll have all the problems we can read about in matthew garrets blog10:46
suihkulokkiya!10:46
ogra_indeed, a new system introduces new issues10:46
ogra_but it wont differ that much from x86 anymore, will be easier to adapt for people10:47
lilstevieis there any new documentation on building packages for kernel10:56
=== Guest18413 is now known as zumbi
=== arun_ is now known as arun
=== yofel_ is now known as yofel
GrueMasterjanimo`: Question on mongodb.  There is no arm build, and I don't see in launchpad where arm is even enabled for it.  Can you check that?15:11
janimo`GrueMaster, I have been working on mongo/arm on and off for the past weeks15:17
janimo`trying to make and send patches upstream15:17
GrueMasterok.15:17
janimo`at the moment there's no ARM pkg enabled - although I consider adding my WIP to Precise to get the arm binaries rolling at least, and maybe the finishing touches as SRU15:18
janimo`GrueMaster, has it come up in server talks?15:18
GrueMasterYes.  Some of the tests we are working on run on mongodb apparently.15:18
janimo`ok, I'll upload my changes these days - but mongo on ARM will FTBFS due to failed tests15:19
janimo`anyway good to know there is interest I'll boost the prio15:19
GrueMasterYea, I just double-checked.  There is interest, but no rush.15:20
=== ndec_ is now known as ndec
steevhow... how do you change font settings in 12.04?15:34
steevspecifically, need to see if subpixel AA is on, and if so, disable it.15:34
ogra_steev, i dont think you easily can switch that around anymore (gnome upstream dropped it), you can switch the fontsize in the acceesibility settings though15:56
steevogra_: hm, okay16:11
steevogra_: well in gnome3, you can use gnome-tweak-tool, would the same work in unity?16:12
ogra_no idea, try it :)16:12
steevogra_: nope, gnome-tweak-tool requires gnome-shell 3.3.2 which won't be installed17:47
steevblargh18:02
steevno gnome-tweak-tool, and no dconf-editor18:02
steevogra_: okay, looks like i can do it manually with gsettings, would be nice to have dconf-editor or gnome-tweak-tool for a pretty gui though18:17
ogra_well, nothing stops you from installing it :)18:18
=== arun__ is now known as arun_
=== calculu5 is now known as calculus
=== shadeslayer_ is now known as shadeslayer
=== prpplague^2 is now known as prpplague
steevogra_: gnome-tweak-tool requires gnome-shell but gnome shell can't be installed.  dconf-editor doesn't seem to exist19:50
=== michaelh1|away is now known as michaelh1
micahgsteev: wfm, is your mirror and apt-cache up to date?20:25
steevmicahg: which one works?20:40
micahgoh, arm, sorry :)20:41
micahgyeah, it's broke ATM20:41
steevmicahg: no, not arch, i meant which? g-t-t or d-e20:48
micahggnome-tweak-tool, but I"m on amd64 (this and the ubuntu+1 channel are right next to each other in my IRC client :))20:48
steevi'm on precise on the arm box20:51
micahgsteev: you want dconf-tools20:53
steevmicahg: ah, okay20:53
steevcommand-not-found didn't suggest it20:53
steevand apt-cache search dconf-editor didn't show anything either20:54
GrueMastersteev: I see the package on launchpadlibrarian https://launchpad.net/ubuntu/+source/d-conf/0.11.7-0ubuntu1/+build/3300937  Not sure why it isn't in the pool.21:37
GrueMastersteev: I just ran apt-get update && apt-cache search dconf-tools and it shows up now.21:40
steev    GrueMaster right, i searched for dconf-editor21:53
steevi should have just searched for dconf21:53
GrueMasterAh.21:53
GrueMasterThat's ok.  We'll probably rename it in a couple of cycles, just to keep users guessing.  :P21:54
steevheh21:54
* ogra_ votes for calling it "frank"21:57
GrueMasterConsidering it was called gconf-editor last cycle...22:00
ogra_no, that was another app22:01
ogra_(still exists)22:01
GrueMasterDifferent app, same function (configuring unity crap).22:02
* GrueMaster needs a break.22:02
steevGrueMaster: true, but they serve different purposes (upstream at least, dconf is for stuff that's done converted to the new dconf, gconf is for legacy ish)22:12

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!