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