| smp4488 | im trying to install the omap3 sgx drivers on 12.04, here is the output http://paste.ubuntu.com/875606/ | 05:05 |
|---|---|---|
| smp4488 | obsoleted or only available from another source? | 05:05 |
| * micahg hugs janimo`, maybe we'll get chromium and ghc on armhf on the ssame day | 09:48 | |
| janimo` | would be nice. But until I see them both successfully built I am not celebrating :) | 09:49 |
| * janimo` hugs micahg back | 09:49 | |
| === amitk is now known as amitk-afk | ||
| brendand | i guess it's not normal that to get the 'correct' resolution on a pandaboard with a hdmi -> dvi cable you need to boot it with the cable connected to the panda's dvi socket and then switch over to hdmi to get the actual picture | 10:57 |
| brendand | if i boot with it connected to the panda's hdmi socket then i get a 'squashed' image | 10:58 |
| brendand | and why is it just shutting off after a few minutes? | 11:09 |
| ogra_ | brendand, how do you power the board ? | 11:24 |
| ogra_ | should at least be a 3A (better 4A) 5V supply | 11:25 |
| brendand | ogra_, 5v straight to the wall | 11:25 |
| brendand | ogra_, has always been working fine | 11:25 |
| brendand | maybe it's too much? should i switch back to usb power from my laptop? | 11:26 |
| ogra_ | that doesnt mean much if power consumption changed due to driver changes :) | 11:26 |
| xranby | brendand: do your monitor support "p" resolutions? | 11:26 |
| brendand | xranby, meaning? | 11:26 |
| xranby | brendand: it might be that your monitor only support interlaced resolutions | 11:26 |
| ogra_ | how many amps does your PSU have ? | 11:26 |
| brendand | ogra_, well, it's still a USB cable, but it goes into a wall socket | 11:27 |
| xranby | brendand: which of these resolutions do your monitor support? 720p, 1080i, or 1080p | 11:27 |
| brendand | xranby, none of those? | 11:29 |
| brendand | according to xrandr on my laptop | 11:29 |
| brendand | but that's probably not true | 11:29 |
| ogra_ | then the squashed image might be normal, depending on what exactly the EDID of the monitor supplies to the driver | 11:29 |
| brendand | that's just the vga outputs | 11:30 |
| ogra_ | and if the driver can support that | 11:30 |
| brendand | ogra_, oh, the squashed image is secondary - i can get over that | 11:30 |
| brendand | ogra_, it's the power off that's bothering me | 11:30 |
| ogra_ | well, you should find out how many amps your PSu actually offers to the board | 11:30 |
| brendand | happens a minute or so after login | 11:30 |
| ogra_ | below 3A you will get probs if you have any USB devices attached | 11:31 |
| ogra_ | usually a non Y USb cable doesnt provide enough | 11:31 |
| brendand | ogra_, i don't have an ammeter unfortunately | 11:31 |
| ogra_ | (i think its 500-750mA per USB line ... | 11:32 |
| ogra_ | ) | 11:32 |
| brendand | ogra_, the wall plug is definitely 5v | 11:33 |
| ogra_ | sure, else you wouldnt be able to plug USB directly into it | 11:33 |
| ogra_ | but whats important is the power | 11:33 |
| ogra_ | if you power the panda through plain USB i wouldnt even bother to try running any desktop image on it | 11:34 |
| ogra_ | and even headless will get issues once the kernel powered up all devices | 11:34 |
| brendand | ogra_, so here's the full deal. i have a usb cable with one end USB and one end 5v connector | 11:39 |
| brendand | ogra_, i also have a wall adapter from a kindle | 11:39 |
| brendand | ogra_, which is 5v | 11:39 |
| ogra_ | the voltage is ireelevant | 11:40 |
| brendand | ogra_, so i have two options, one is to plug the usb cable into my laptop, the other to plug it into the wall adapter | 11:40 |
| brendand | ogra_, if i plug it into the laptop it stays on | 11:40 |
| brendand | ogra_, if i plug it into the wall adapter it powers off | 11:40 |
| brendand | so i have got my workarounds now, but things have definitely changed | 11:41 |
| ogra_ | right, both will not provide enough power, one just stays a bit more stable | 11:41 |
| ogra_ | seriously, get a proper PSU | 11:41 |
| ogra_ | powering through USB will have unpredictable sideefects and you wont be able if the issues you see have to do with SW or power | 11:42 |
| ogra_ | a single USB cable *cant* provide enough power for a panda | 11:42 |
| ogra_ | s/abel/able to tell/ | 11:43 |
| brendand | ogra_, you're probably going to smack me down for this, but it is possible the cable itself is not a USB cable but just has a USB connector on one end? | 11:46 |
| ogra_ | unlikely | 11:48 |
| brendand | although you've made it quite clear that the optimal setup is to have a proper 5v PSU, with no hint of usb-ness about it | 11:48 |
| ogra_ | you can run the board headless with an Y-USB cable, that will just provide enough power to run a headless system without any attached USB devices | 11:49 |
| ogra_ | if you run a desktop and use a USB kbd and mouse as well as the builtin USB NIC, you *need* more than 2A | 11:50 |
| brendand | ogra_, ok, so this is the cusp of the problem. you'll be shocked to know i don't have much electrical training | 11:51 |
| brendand | so the wall plug is actually only 0.35 Amps | 11:51 |
| ogra_ | you dont need any electric training | 11:51 |
| ogra_ | just make sure your HW fulfills the needed specs | 11:52 |
| brendand | ok, i'll rephrase it. i haven't a f*ing clue | 11:52 |
| ogra_ | let me give you a clue then: buy a >2A 5V power supply | 11:53 |
| ogra_ | (and note they are expensive, unlikely you get one for less than $15-20) | 11:53 |
| ogra_ | even your way to power from the laptop wont gain you a stable platform | 11:54 |
| ogra_ | (even though it doesnt shut off immediately) | 11:54 |
| brendand | no, but for now it's working (mostly) until i get a proper PSU | 11:54 |
| brendand | which i will | 11:55 |
| ogra_ | ringht, just dont ask about bugs until you have stable HW | 11:55 |
| ogra_ | since the low power makes it totally unpredictable if there is actually a SW bug | 11:55 |
| brendand | i wasn't aware of the required ampage, so now i know what might cause any strangeness | 11:56 |
| ogra_ | right :) | 11:56 |
| sledges | zyga, I ran the built from sources kernel on imx53 finally | 12:03 |
| zyga | sledges, and? | 12:04 |
| sledges | s/ran/successfully ran/ :) | 12:04 |
| sledges | you need to use identical toolchain | 12:04 |
| sledges | to that of the working binary | 12:04 |
| sledges | there's also a patch to work with gcc4.6.2 (failed in my research so far, but there are still things to try), 1 moment... | 12:06 |
| sledges | https://build.pub.meego.com/package/view_file?file=000-kcflag-mno-unaligned-access.patch&package=kernel-adaptation-n900&project=CE%3AAdaptation%3AN900&rev=9cdc38022e14fe78051a2e799587a375 | 12:06 |
| xranby | hmm why are this patch not part of the ti omap4 kernel tree? http://www.arm.linux.org.uk/developer/patches/viewpatch.php?id=7178/1 | 12:15 |
| zyga | sledges, does sata work for you? | 12:17 |
| sledges | yes | 12:17 |
| zyga | sledges, I suspect my board is simply broken | 12:17 |
| zyga | not soldered right or something | 12:18 |
| sledges | I had sata working fine with that kernel - did you try that kernel?: http://cdimage.ubuntu.com/releases/oneiric/release/ubuntu-11.10-preinstalled-desktop-armel+mx5.img.gz | 12:18 |
| sledges | break in uboot and override bootargs with rootfs-on-sata | 12:19 |
| sledges | i tried other ~5 kernels from different sources (linaro, ltib, 2.6.35/38), none of them worked (or halting after decompress or not enumerating SATA) | 12:21 |
| sledges | so don't give up :) and give it a last shot, zyga | 12:21 |
| zyga | sledges, I cannot try right now, that device is production | 12:22 |
| zyga | sledges, I'd need a new IMX from linaro | 12:22 |
| zyga | this one is personal | 12:22 |
| sledges | production? IMX from linaro - you mean HW? I thought Linaro provide only SW BSPs | 12:23 |
| zyga | sledges, I mean I'm using this imx to do stuff 24/7 and I cannot tinker with it | 12:23 |
| zyga | sledges, linaro as my host/employer so that I can use it for my linaro-work | 12:24 |
| sledges | zyga, understood ;) | 12:24 |
| sledges | well, good luck and keep in touch! | 12:24 |
| zyga | sledges, I need panda es, origen, and imx53 from the last gen lineup | 12:24 |
| sledges | wow, smoking! :) | 12:24 |
| === amitk-afk is now known as amitk | ||
| xranby | jamespage: ping i will try if that patch improves the https://bugs.launchpad.net/ubuntu/+source/openjdk-6/+bug/845158 situation | 12:32 |
| ubot2` | Launchpad bug 845158 in openjdk-6 "Frequent java task hang on ARM server" [Undecided,Confirmed] | 12:32 |
| xranby | jamespage: this patch applies cleanly http://www.arm.linux.org.uk/developer/patches/viewpatch.php?id=7178/1 on the ti omap4 kernel tree | 12:32 |
| xranby | i am compiling my own kernel using it | 12:33 |
| marvin24 | any idea if /sys/class/power/battery/current_now is allowed to return a negative value (to show the battery is discharging)? | 12:40 |
| marvin24 | upower seems to apply abs() to it, but other power daemons seem to assume only positive currents | 12:41 |
| marvin24 | I failed to find any documentation about the required values (beside that the field is signed int) | 12:41 |
| danboid | I've installed 11.10 on my pandaboard but what I think is gdm isn't letting me log into anything but Unity even though XFCE, fluxbox are listed and supposedly chosen | 15:46 |
| danboid | I'm choosing 'Other' on the login screen | 15:46 |
| danboid | then I pick say XFCE, login - Unity! | 15:47 |
| danboid | Have I go to replace gdm with kdm or another display manager? | 15:47 |
| ogra_ | gdm wasnt used in a while in ubuntu | 15:48 |
| ogra_ | so that might be a gdm specific bug | 15:48 |
| danboid | ogra_, I'm just using the default display manager in 11.10 - I was just presuming it was gdm | 15:49 |
| ogra_ | no, thats lightdm | 15:49 |
| danboid | ogra_, So this looks like a lightdm bug then | 15:49 |
| ogra_ | and it should just give you the right login (it surely does here) | 15:49 |
| ogra_ | is your ~/.dmrc not writable or some such ? | 15:50 |
| danboid | Its writable | 15:51 |
| danboid | I've tried logging into Unity 2D, XFCD | 15:51 |
| danboid | XFCE and fluxbox with no luck | 15:51 |
| danboid | I'm new to lightdm so | 15:52 |
| ogra_ | its not different from gdm in behavior | 15:52 |
| ogra_ | just smaller :) | 15:52 |
| ogra_ | it checks the available sessions, updtaes dmrc and runs the session from there | 15:53 |
| danboid | Maybe I can fix it by editing .dmrc? | 15:53 |
| danboid | or how do I force it to update .dmrc? It doesn't list all the desktop/wms in there at, | 15:54 |
| danboid | atm | 15:54 |
| ogra_ | do both (flux and xfce) have sessions in /usr/share/xsessions ? | 15:54 |
| ogra_ | iirc thats what gdm and lightdem check for nowadays | 15:54 |
| danboid | Yeah - I've got 6 files in there now | 15:55 |
| danboid | inc ones for XFCE and Flux | 15:55 |
| ogra_ | check if they have proper Exec lines | 15:55 |
| ogra_ | and if tehse executables actually exist | 15:56 |
| ogra_ | you can edit dmrc if you want | 15:57 |
| ogra_ | Session= needs the name of the .desktop file in /usr/share/xsessions you want to use i think | 15:57 |
| === zyga is now known as zyga-afk | ||
| === zyga-afk is now known as zyga | ||
| shadeslayer | hey guys, just wondering, how do you test build your ARM packages? | 18:09 |
| ogra_ | like we test build our x86 packages | 18:12 |
| shadeslayer | ogra_: pbuilder? But how do you specify you want a ARM pbuilder? | 18:16 |
| shadeslayer | I can make x86 and x86_64 pbuilders, and I was trying out qemubuilder last night, but that just refused to work | 18:17 |
| xranby_ac100 | jamespage: ping | 18:17 |
| GrueMaster | shadeslayer: We build our packages natively on arm platforms. | 18:17 |
| jamespage | xranby_ac100, hey! | 18:17 |
| xranby_ac100 | jamespage: hi | 18:17 |
| shadeslayer | GrueMaster: right, but what if I don't have one of those? :) | 18:17 |
| micahg | janimo`: \o/ chromium on armhf | 18:17 |
| xranby_ac100 | jamespage: i might have found a fix for the server stalls under heavy load | 18:17 |
| ogra_ | shadeslayer, i dont use pbuilder but there are knobs and switches iirc to make it run with qemu-arm-static | 18:18 |
| GrueMaster | shadeslayer: Not sure what to tell you, other than maybe buy one? They are fairly cheap. | 18:18 |
| jamespage | xranby_ac100, great! I saw you note earlier - how did that work out? | 18:18 |
| xranby_ac100 | jamespage: from the limited testing i have done so fair... looks promising.. but i need to run more tests to be sure.. i am using a patched kernel for my ac100 here | 18:19 |
| shadeslayer | ogra_: there's a qemu-system-arm | 18:19 |
| ogra_ | shadeslayer, you want qe,u-user-static | 18:19 |
| shadeslayer | oh | 18:19 |
| xranby_ac100 | jamespage: is there some quick way you can cross compile a kernel using xdeb?? | 18:19 |
| shadeslayer | ogra_: what's wrong with the former? | 18:19 |
| ogra_ | that runs a full vm | 18:19 |
| xranby_ac100 | jamespage: i was really hoping to check if you had had any chance to run any tests | 18:20 |
| ogra_ | anywayx, i only know there is an implementation in pbuilder that uses it, i have no idea at all how that works or how you use it | 18:20 |
| shadeslayer | oh cool | 18:20 |
| shadeslayer | I'll look into it | 18:20 |
| shadeslayer | thanks ogra_ | 18:21 |
| ogra_ | welcome :) | 18:21 |
| GrueMaster | xranby_ac100: What tests are you running that are causing stalls? I'd like to run them here, as I am working on workload testing on arm server. | 18:22 |
| xranby_ac100 | GrueMaster: my test are to run the jogamp jogl unittests | 18:23 |
| ogra_ | cp 1TB file from mmcblk0p1 to mmcblk1p1 :) | 18:23 |
| xranby_ac100 | GrueMaster: let me give you a link to how to build and run them | 18:23 |
| shadeslayer | would probably have been easier if my Transformer had a proper SBK version ... *grumble* | 18:24 |
| GrueMaster | That would be great. | 18:25 |
| xranby_ac100 | GrueMaster: http://www.trimslice.com/forum/viewtopic.php?p=1804&sid=f05c252a82821fdd7d3924941c1bfb78#p1804 | 18:26 |
| xranby_ac100 | when gluegen and jogl are build then run sh scripts/make.jogl.all.linux-armv7.sh junit.run | 18:26 |
| xranby_ac100 | to start the unittests | 18:26 |
| xranby_ac100 | when i test on my pandaboard running oneiric i can trigger a stall within 10min | 18:27 |
| xranby_ac100 | a stall means that the process gets stuck in some kernel lock and ps ax also stall when trying to list processes | 18:28 |
| xranby_ac100 | GrueMaster: my panda dmesg looks like this http://paste.ubuntu.com/875835/ | 18:28 |
| xranby_ac100 | what i hope are that this patch http://www.arm.linux.org.uk/developer/patches/viewpatch.php?id=7178/1 | 18:29 |
| xranby_ac100 | might fix it.. both the ti omap kernel trees and the ac100 kernel tree are missing this patch | 18:29 |
| xranby_ac100 | its merged into linus git tree upstream | 18:29 |
| GrueMaster | Interesting. Do we have a bug against the kernel for this? it should be possible to get it in for precise kernel. | 18:30 |
| GrueMaster | (if it isn't there already). | 18:31 |
| jamespage | xranby_ac100, I can run some tests that load up things quite well - yes | 18:32 |
| xranby_ac100 | GrueMaster: we have an old bug not yet pinned to the kernel https://bugs.launchpad.net/ubuntu/+source/openjdk-6/+bug/845158 | 18:34 |
| ubot2` | Launchpad bug 845158 in openjdk-6 "Frequent java task hang on ARM server" [Undecided,Confirmed] | 18:34 |
| xranby_ac100 | GrueMaster: i have been looking high and low for a better testcase | 18:35 |
| * prpplague grumbles about how buggy his ubuntu desktop has been since it has been updated | 18:35 | |
| xranby_ac100 | GrueMaster: this issue poped up adain frequently when i was testing the opengl-es bindings with the jogamp community | 18:38 |
| xranby_ac100 | the best test i have are to run the jogamp unittests on a panda | 18:38 |
| xranby_ac100 | you can also pick one test and run it 100 times in a row | 18:39 |
| GrueMaster | xranby_ac100: Looking at our 3.2 kernel git tree, I am not seeing this patch applied. Oddly, the link says it is in 3.2-rc3 (unless it is the x86 version of the patch only). | 18:39 |
| xranby_ac100 | GrueMaster: when i checked today its not applied in any ofthe panda 3.1 and 3.2 tree | 18:39 |
| xranby_ac100 | s | 18:39 |
| GrueMaster | I'm always looking for new and interesting tests I can automate. I'll see if I can get this test scripted and automated. I'll also file a kernel bug for this patch. | 18:40 |
| xranby_ac100 | GrueMaster: it are found upstream here http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=history;f=arch/arm/mm/fault.c;h=bb7eac381a8e60f619591e7c80c4ad30cc972d62;hb=HEAD | 18:42 |
| GrueMaster | Excellent, thanks. I'll update my local ubuntu kernel git tree and double check, then if it isn't there, I'll file a bug and add this link. | 18:43 |
| GrueMaster | There. bug 951043 filed. | 18:56 |
| ubot2` | Launchpad bug 951043 in linux-ti-omap4 "Port OOM changes into do_page_fault for arm" [Medium,New] https://launchpad.net/bugs/951043 | 18:56 |
| GrueMaster | I didn't add the ac100 or mx5 kernels. | 18:57 |
| xranby_ac100 | GrueMaster: thanks a lot, poke me if you for some reason fail to build jogamp | 19:51 |
| GrueMaster | Will do. | 19:54 |
| pbuckley | syntax error: unknown user 'puppet' in statoverride file | 20:30 |
| pbuckley | i always forget what causes this | 20:30 |
| pbuckley | (dpkg) | 20:31 |
| === zyga is now known as zyga-weekend | ||
| infinity | pbuckley: Pretty much exactly as it sounds. Someone's done a dpkg-statoverride to make something owned by "puppet", but the user doesn't exist. | 20:33 |
| pbuckley | probably from the puppetlabs package for puppet.. and apt-get purge didnt clean it | 20:34 |
| pbuckley | how do i fix it? | 20:34 |
| pbuckley | Can't exec "/tmp/lightdm.config.24521": Permission denied at /usr/share/perl/5.14/IPC/Open3.pm line 186. | 20:35 |
| pbuckley | also been seeing this | 20:35 |
| pbuckley | maybe its time i start over from a fresh flash | 20:36 |
| infinity | Simplest method would just be to create a puppet user before trying to remove those packages that seem grumpy, and then delete it. Otherwise, you can fix the actual statoverride in question. | 20:36 |
| pbuckley | ah good idea | 20:36 |
| pbuckley | that seemed to fix the override | 20:38 |
| pbuckley | so im assuming that exec issue comes from /tmp being mount noexec | 20:38 |
| pbuckley | which i did because it was complaining about mount -o remount,exec | 20:38 |
| pbuckley | (and at the time /tmp wasn't a seperate mount) | 20:39 |
| infinity | We don't support noexec tmp by default. | 20:39 |
| pbuckley | hrmm.. seemed to want it that way at one point | 20:40 |
| infinity | (dpkg, as you note above, unpacks various scripts and runs them from tmp) | 20:40 |
| pbuckley | ill set it back | 20:40 |
| infinity | There are ways to change that, and have dpkg use other directories, I believe. | 20:40 |
| infinity | Well, I suppose you could just run apt/dpkg with a different TMPDIR set. | 20:40 |
| pbuckley | Updating software catalog...this may take a moment. | 20:41 |
| pbuckley | WARNING:softwarecenter.db.update:The file: '/usr/share/app-install/desktop/kde-telepathy-send-file:kde4__ktp-send-file.desktop' could not be read correctly. The application associated with this file will not be included in the software catalog. Please consider raising a bug report for this issue with the maintainer of that application | 20:41 |
| pbuckley | all sorts of new errors this morning | 20:41 |
| pbuckley | heh | 20:41 |
| pbuckley | yeh removing the /tmp mount fixed the issue | 20:41 |
| pbuckley | now its just a directory | 20:41 |
| infinity | The software center thing is a bug, not your problem. ;) | 20:42 |
| pbuckley | \o/ | 20:42 |
| pbuckley | i like those | 20:43 |
| electroglue | can someone tell me if ubuntu-omap4-extras is available for pangolin? | 21:17 |
| GrueMaster | electroglue: I don't think TI has pushed them up yet. | 21:20 |
| infinity | Most of the bits are in their experimental PPA, but it requires the kernel from the oneiric PPA to work, if I recall. | 21:22 |
| infinity | electroglue: ^ | 21:22 |
| janimo` | micahg and ghc too \o/ | 21:26 |
| infinity | janimo`: You've had a fruitful week. ;) | 21:27 |
| janimo` | infinity, yes, in hindsight it could have been a fruitful single day had I paid more attention and panda had not locked up | 21:28 |
| infinity | janimo`: Meh, hindsight is like that. | 21:28 |
| infinity | janimo`: Did you end up going the "mangle ghc" route, or "make llvm have sane defaults" route? | 21:29 |
| janimo` | infinity, the sed-patch route. Without the clean rule :D | 21:29 |
| infinity | You sick man. | 21:29 |
| janimo` | since it is temporary :) | 21:29 |
| janimo` | well. If I had touched m4 code I'd just be sicker now, so it was the right choice | 21:30 |
| infinity | m4 isn't that painful | 21:30 |
| GrueMaster | Neither is pulling nostril hairs with vise grips, but why try. | 21:31 |
| infinity | ... I've done that. | 21:31 |
| * GrueMaster is the lease bit surprised. | 21:31 | |
| GrueMaster | *least | 21:31 |
| infinity | And s/is/isn't/ ? | 21:32 |
| infinity | I think Friday afternoon has taken hold of your fingers. ;) | 21:32 |
| GrueMaster | Exactly. I hear my neighbor mowing his lawn, and see blue sky outside my basement window. Thoughts lie elsewhere. | 21:33 |
| infinity | Swimming in an pool filled with beer? | 21:33 |
| electroglue | infinity: so using an oneric kernel really means that the packages aren't available for pangolin correct? I'm looking for hardfp and omap4 extras to be available with the kernel provided by pangolin | 21:42 |
| infinity | electroglue: Yeah, as far as I know, that's "soon, but not yet". | 21:42 |
| pbuckley | i feel your pain GrueMaster, it is beautiful outside.. blue skys.. i see the ocean outside my window | 21:43 |
| pbuckley | really hard to focus on work | 21:43 |
| === yofel_ is now known as yofel | ||
| pbuckley | stupid question.. what generates /var/run/motd? | 22:47 |
| GrueMaster | Probably motd? | 22:54 |
| GrueMaster | (just a wild guess). | 22:54 |
| infinity | pbuckley: See /etc/update-motd.d/* | 22:54 |
| infinity | pbuckley: Mashed together by pam_motd | 22:55 |
| pbuckley | awesome thank you | 22:55 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!