**** BEGIN LOGGING AT Thu Apr 12 02:59:58 2012 Apr 12 03:07:39 anyone here know how to manually install omap4 addons on a pandaboard? Apr 12 03:08:05 scratch that, accidentally solved it Apr 12 05:44:03 ubuntu on pandaboard is stuck in the install loop. does anyone know the command to break it out of tha tloop Apr 12 05:44:24 it keeps asking keyboard, location, username, password, then resets and repeats Apr 12 05:45:04 ubuntu 11.10 desktop omap4 image Apr 12 05:46:23 scratch that remembered oem-config-remove Apr 12 05:52:15 I'm surprised it doesn't remove oem on its own -- unles it doesn't have write access to the ramdisk? Apr 12 05:53:06 beats me Apr 12 05:53:24 i did it manually, and the train is now back on the tracks Apr 12 06:52:35 looks like ubuntu does not provide a "xf86-video-modesetting" driver Apr 12 06:56:01 marvin24: ubuntu's kernel does KMS, but probably not for relatively obscure arm GPUs Apr 12 06:57:13 well, I wouldn't name geforce/light an obscure arm GPU ... Apr 12 06:57:26 Shrug Apr 12 06:57:39 Compared to the vast numbers of x86-y intel/ati/nv GPUs, it is Apr 12 06:57:41 That's what I meant Apr 12 06:58:52 I just wanted to test it on my ac100 with the newly tegra drm posted yesterday Apr 12 06:59:16 but there's still other stuff to do and compiling it myself should be too difficult Apr 12 07:00:00 What I do know, is that on my tegra2 TF101, I have a native-resolution fbcon. Apr 12 07:00:33 I kinda assumed that meant it had KMS, but i haven't investigated. Oh- but that's with the android kernel, not the ubuntu kernel Apr 12 07:02:55 twb: I was talking about mainline code Apr 12 07:03:18 there is fbdev for tegra yet Apr 12 07:03:30 lilstevie told me the GPU support is best (for tf101, but they're mostly the same) in the Chrome OS kernel Apr 12 07:04:38 not at the moment it isn't :p Apr 12 07:05:29 I have native res in my fbcon and that's all I really care about Apr 12 07:05:38 heh Apr 12 07:05:40 yeah Apr 12 07:05:58 cause your odd and use xinit for everything Apr 12 07:06:05 well, android is on 2.6.39 (I guess) and chromeos on 3.0 ... Apr 12 07:06:21 marvin24: specific case is neither for those Apr 12 07:06:46 CrOS is at 2.6.38 and android at 2.6.36 Apr 12 07:06:50 It would be NICE if the animated gif didn't make xinit /usr/bin/midori http://www.bom.gov.au/products/IDR023.loop.shtml#skip visibly laggy, but not enough that ICBF maybe breaking my only working computer Apr 12 07:07:06 cause the 2.6.39 android kernel is silly Apr 12 07:07:42 mmh, I don't know which kernels are shipped with which version (of android or chromeos) Apr 12 07:07:55 but ac100 has a chromeos kernel with accel'ed X Apr 12 07:08:13 and it is version 3.0 (but still too old for precise) Apr 12 07:08:28 so we badly need a newer kernel sooner or later Apr 12 07:08:33 tf101 can have accelerated X with .39 but it is stupid Apr 12 07:08:39 there is some odd bug Apr 12 07:08:46 breaking fbdev Apr 12 07:08:47 no console? Apr 12 07:08:50 yep Apr 12 07:08:57 and f'd colours Apr 12 07:09:00 same here on the nv 3.1 kernel Apr 12 07:09:12 I posted something on the tegra forum, but no reply Apr 12 07:09:36 heh Apr 12 07:09:40 lilstevie: anything on xda-developers, you know of? Apr 12 07:09:45 nope Apr 12 07:09:59 fucked as in yuv thinking its rgb kinda thing? Apr 12 07:10:02 I am otherwise sidetracked with [redacted] Apr 12 07:10:10 twb: yes Apr 12 07:10:10 so let's hope the open source driver will make fast progress Apr 12 07:10:40 I didn't know there even was an accelerated open driver for it Apr 12 07:10:47 its new Apr 12 07:10:55 ah, cool Apr 12 07:11:48 this kind of new http://www.phoronix.com/scan.php?page=news_item&px=MTA4NjA Apr 12 07:14:15 there are a few other little whispers that I am hearing from nvidia, but they are no more than whispers Apr 12 07:16:26 we are having some good luck with getting 3.1 running on the tf201 though Apr 12 07:16:43 so far we can boot and mostly use android with it Apr 12 07:18:01 Do you get the impression that nvidia are slowly getting a clue about dealing with the FOSS community? Apr 12 07:18:21 The tf is the first product I've bought from them EVER since they have such a bad rep Apr 12 07:18:55 well the whispers I am hearing are that they are starting to embrace FOSS, at least in the arm sector Apr 12 07:19:21 Probably because it's not their core market Apr 12 07:19:50 Sounds like I should continue to boycott them on the desktop/server segment Apr 12 10:02:29 The arm-ubuntu and ubuntu core rootfs FAQs are not clear on who is and is not a developer. Apr 12 10:03:05 ? Apr 12 10:03:46 I've built stable kernel source for armv7 (920T core, WM8505/Via8505) on one of those generic 7" netbnooks Apr 12 10:03:53 Anyone who wants to write software is a developer, I imagine. Apr 12 10:03:57 ahh ok Apr 12 10:03:58 I'm not sure what your statement means. Apr 12 10:04:06 (Which FAQ are you referring to?) Apr 12 10:04:35 infinity: https://wiki.ubuntu.com/Core Apr 12 10:04:39 this one. Apr 12 10:04:54 Also, the 920T isn't ARMv7... Apr 12 10:05:01 infinity: i was meaning that I will probably not contribute anything useful to the project Apr 12 10:05:33 but still want to hack on it Apr 12 10:05:46 andl that peice of info saves me a lot of effort to just get a crash at boot up doesn't it Apr 12 10:06:18 A 920T is probably v4t... Apr 12 10:06:22 * serishema didn't know and it's on the channel topic. Arm cores != ISAs how embarassing Apr 12 10:06:28 that explains why it is so hopelessly slow Apr 12 10:06:28 So, not even remotely supported by ubuntu-core. Apr 12 10:06:46 Debian may well support it. Apr 12 10:06:49 The actual chip is a via/wondermedia 8605 @ 600 Mhz Apr 12 10:07:02 * serishema is wrong about the 920t core part Apr 12 10:07:04 Sure, it's not about slow, it's about pure incompatiblity. Apr 12 10:07:23 * serishema realizes it would be an ISA / EABI issue Apr 12 10:07:42 so the wm8605 is definately an arm v4? and not v7? Apr 12 10:07:45 Oh, an 8605 sounds more like an ARMv7. ;) Apr 12 10:07:49 ahh Apr 12 10:07:50 good Apr 12 10:08:10 I guessed from the /board directory name in the wmt uboot code Apr 12 10:09:09 But yeah, slow it will be, relatively speaking. Apr 12 10:09:14 infinity: tl;dr basically I had been living under a rock (read: working for a boss who makes her run and develop for windows) for too long and wasted effort trying to write drivers for the on chip hardware when they actually exist Apr 12 10:09:26 * serishema is not expecting this to run X btw if that makes a difference Apr 12 10:09:34 It could run X, I'm sure. Apr 12 10:09:42 And many other shiny things. Apr 12 10:10:06 But yeah, it'll probably perform similarly to a beagleboard, or a Nokia n900, none of which are particularly speedy by today's standards. Apr 12 10:10:12 So, patience will be a virtue. ;) Apr 12 10:11:05 * infinity is surrounded by fast ARM kit, and keeps forgetting that's a relatively new invention. Apr 12 10:11:46 basically if it ran ssh, rdesktop and a browser (chrome or firefox,) vi, basic shell prompt tools.. could connect to the internet via my nokia 2730 classic and decode and play mp3/ogg vorbis, etc i'd be really pleased with that Apr 12 10:11:56 This is essentially intended as a mobile dumb terminal. Apr 12 10:12:05 All real processing happens on amazon EC2 anyway Apr 12 10:12:18 Well, any of that's doable. Apr 12 10:12:20 so it was a natural choice to use something so cheap it was virtually disposable rather than a real laptop :) Apr 12 10:12:47 We build out full userspace for ARM, so... It really comes down to kernel and video support for that SoC, which we obviously don't support out of the box. Apr 12 10:12:59 s/build out/build our/ Apr 12 10:13:27 (And, I'd recommend basing your work on precise/armhf ... Especially on a slower SoC, you'll be glad for the performance boost) Apr 12 10:13:39 Unless you're stuck using binary softfp drivers from a vendor. :/ Apr 12 10:13:43 Then armel it is. Apr 12 10:15:06 this is VFP? Apr 12 10:15:24 * serishema can't get her hands on any vendoer specific code. Apr 12 10:15:28 Well, it's all VFP, it's the ABI that changes. Apr 12 10:15:43 Welcome to the confusing world of ARM. Apr 12 10:16:25 time for a rebuild then. Apr 12 10:16:35 soft = soft floating point, softfp (which our armel port is) = vfp ops with soft calling conventions, hard = vfp ops with new calling convention to make use of extra registers, breaking ABI. Apr 12 10:17:34 ubuntu-armhf/debian-armhf = hard, ubuntu-armel = softfp, debian-armel = soft Apr 12 10:18:59 basically I got given a bricked one of these netbooks bricked. Apr 12 10:19:21 for free Apr 12 10:19:49 I've got as far as vfs: cannot mount root fs on uknown block (cos I haven't even made a root fs or initrd) Apr 12 10:20:00 * serishema is mostly just curious to see if there is a easier way than doing everything from scratch :) Apr 12 10:20:02 Well, at least you have a kernel. Apr 12 10:20:30 As for "from scratch", given that no installer will work for you, you get to put a tiny bit of effort in. Apr 12 10:20:39 But using ubuntu-core as your rootfs will work fine. Apr 12 10:21:09 Just need to maybe chroot in and add a user (or set a root password) before you try stuffing it on another device. :P Apr 12 10:22:57 nice. I most likely just have to recompile with the right eabi Apr 12 10:23:07 I am pretty sure I used armel Apr 12 10:23:10 which I don't want Apr 12 10:23:59 The kernel doesn't care. Apr 12 10:24:00 You're fine. Apr 12 10:24:17 Same kernel can run armel and armhf userspaces. Apr 12 10:24:34 oh. so the kernel will load any elf for init with the correct header and ISA type? Apr 12 10:24:44 and an eabi is a userspace dynamic linker thing? Apr 12 10:25:35 Basically, yeah. Apr 12 10:25:50 ahh this is a good thing Apr 12 10:26:19 I was going to just gcc -static everything to avoid this problem Apr 12 10:26:28 Ew. Apr 12 10:26:29 but that would be unrealistic on a board with 256mb of ram Apr 12 10:26:59 I assume running swap space on a SD card is a bad idea? Apr 12 10:27:04 Very. Apr 12 10:27:12 works, but is no fun at all Apr 12 10:27:14 Unless you like throwing out cards. Apr 12 10:27:27 (It's also just dead slow) Apr 12 10:27:42 * ogra hasnt managed to kill a card with that yet but its close to unusable if you start swapping Apr 12 10:27:46 Keeping your VM commit under your RAM size is the saner option, if you're building for a specific use case. Apr 12 10:27:58 and i have cards that are constantly in use for 2 years Apr 12 10:28:06 ogra: You're a very lucky man. Apr 12 10:28:16 ogra: Or your definition of "constantly" is different from mine. Apr 12 10:28:25 I just assumed that the write perf would suck too much to be useable. Apr 12 10:28:31 I didn't think about it wearing out the card Apr 12 10:28:37 well, in my testbuild machines i always use the same cards Apr 12 10:28:41 serishema: Write performance is awful, but wear is the real concern. Apr 12 10:28:43 across the releases Apr 12 10:28:57 serishema: Anal retentive SSD owners often swap to external rotary disks for wear reasons. Apr 12 10:29:17 (Or people like me just don't use swap) Apr 12 10:29:19 * serishema nods Apr 12 10:30:04 if I can't run ssh on a single VT with 256Mb of ram and no swap I have done something seriously wrong :) Apr 12 10:30:17 Indeed. Apr 12 10:30:23 it worked on my 486 with 8Mb Apr 12 10:30:27 10 years ago Apr 12 10:30:38 Well, times have changed a bit since kernel 2.0 Apr 12 10:30:45 we can't have all forgotten how to program because of java, etc that badly Apr 12 10:30:52 But, yeah, 256 is still "a lot" until you get into fancy desktop environments. Apr 12 10:31:32 and compilers are better so I assume any hand written arm ASM will be trash compared with gcc's output with -O2? Apr 12 10:31:46 which wasn't nessacarly true on 486 in 1999 Apr 12 10:31:48 oh, fun, compiz FTBFS Apr 12 10:31:57 ah, KDE issue Apr 12 10:32:07 phew ... wasnt my patch Apr 12 10:34:51 serishema: Depends on the quality of your assembly, or what you're trying to do. Apr 12 10:35:22 serishema: (Also depends on the quality of your C, as a good C programmer knows more or less what the compiler's going to optimise out anyway) Apr 12 10:35:32 serishema: But yeah, in general, GCC's not awful. Apr 12 10:35:55 if i'm writing asm it's because i'm doing eairly power up and can't do it in C. I hate ASM. Apr 12 10:36:04 ;) Apr 12 10:36:35 * serishema just stole the code WMT was forced to release for GPL violations for dram controller init Apr 12 10:36:37 I hate that stuff Apr 12 10:36:41 Board designers problem Apr 12 10:37:18 I have designed boards, but reverse engineering an unknown board is too hard for me :) Apr 12 10:37:42 I will definately need to get a newer uboot on it Apr 12 10:37:50 theirs is full of crappy bugs Apr 12 10:38:15 Meh. If it's good enough to jam some blobs at some offsets and jump, who cares? Apr 12 10:38:36 true. Apr 12 10:38:55 The only actual bug which matters is the power button GPIO Apr 12 10:39:05 which doesn't have a proper debounce, just a delay by a constant Apr 12 10:39:19 but the real delay it should wait varies with the battery voltage Apr 12 10:39:28 so you have to press the power button 10 times before it will turn on Apr 12 10:39:38 I'm unconvinced that anything related to power buttons is ever correct on any ARM device I own. Apr 12 10:39:47 From dev boards to phones to... Well, everything. Apr 12 10:39:51 that's normal? Apr 12 10:40:12 I noticed a // can't power off myself comment on a function which jumps into some bizare assembly Apr 12 10:40:19 Well, your specific issue isn't necessarily normal (though I have two phones that behave like that). Apr 12 10:40:32 which appears to among other things set the hibernate flag (WMT specific) Apr 12 10:40:33 But I have other boards that are equally but differently confused about what power buttons do. :) Apr 12 10:41:02 And a netbook that never actually powers off, just goes into a suspend state. Apr 12 10:41:19 that would explain my mysterious battery drain Apr 12 10:41:29 and the meaning of the // can't power off myself comment Apr 12 10:41:42 Potentially, yeah. Apr 12 10:42:04 ahh.. so is this software controllable at all? I can get my hands on a JTAG, etc. Apr 12 10:42:16 so reflashing the bootloader is OK Apr 12 10:42:16 Depends on the SoC. Apr 12 10:42:19 That's a big "maybe". Apr 12 10:42:28 when i connect the battery does it actually jump to __start or is that board specific? Apr 12 10:42:29 And I don't know anything about the one you've got. Apr 12 10:42:46 That's also board specific. Apr 12 10:42:52 ahh ok. Apr 12 10:43:06 Some reset on power, and jump somewhere sane, some just start garbled and wait for a keypress event. Apr 12 10:43:16 (ie: a soft power button that shouldn't even exist) Apr 12 10:43:25 urgh Apr 12 10:43:46 so just wiring a hardware power switch onto the battery is a bad idea Apr 12 10:43:51 since it probably wouldn't work Apr 12 10:43:59 Unless it would. Apr 12 10:44:10 It would work on a Pandaboard. Apr 12 10:44:16 (Maybe even all OMAP SoCs) Apr 12 10:44:29 OMAP SoCs are the only ones i've seen which I don't think are insane Apr 12 10:44:32 It wouldn't work for beans on an iMX53, which insists on powering on to confusion. Apr 12 10:44:33 unfortunately they are too expensive Apr 12 10:45:06 I get my arm stuff for free because people give me their bricked devices Apr 12 10:45:35 usually they have not even overwritten the boot loader, so a board/soc specific micro SD card can load a kernel onto them Apr 12 10:46:14 otherwise.... it's down to the university electronics lab for some unauthorized use of university resources :) Apr 12 10:48:06 anyway it doesn't matter. I will just put the 26 amp hour deep cycle battery which is just sitting on my floor on float charge in my car as a 2nd battery Apr 12 10:48:19 and then it can waste as much power with soft off as it likes :) Apr 12 10:49:15 I occasionally use it to power flourecent lights during power cuts in winter, but that's about it Apr 12 10:49:37 * serishema lives in the wop wops so every time the wind gets up the power and xDSL go out Apr 12 10:50:00 anyway enough blathering. I will download this and put it in qemu-system arm and see if I can get it to go Apr 12 10:51:03 infinity: and as for the orriginal question. I will just subscribe to both mailing lists and lurk Apr 12 10:51:20 and should I happen to accidentally do something useful I will post the patch :) Apr 12 10:51:27 Heh. Sure. Apr 12 10:51:38 #debian-arm on oftc is another place worth lurking. Apr 12 10:51:47 oftc? Apr 12 10:51:57 irc.oftc.net Apr 12 10:52:08 ahh. another fs/oss related net? Apr 12 10:52:13 debian hides there :) Apr 12 10:52:13 Yeah Apr 12 10:52:20 irc.debian.org points to oftc. Apr 12 10:52:20 I wonder where #debian went Apr 12 10:52:35 So, that would be your answer. ;) Apr 12 10:52:41 Cool Apr 12 10:52:47 We only moved, like, a decade ago. Apr 12 10:52:53 *cough* Apr 12 10:52:57 hhyaha Apr 12 10:53:09 * serishema disappeared sometime around kernel 2.6.35 Apr 12 10:53:10 not true, i remember still seeing debian channels around at warty days Apr 12 10:53:13 and has just come back Apr 12 10:53:21 ogra: Yeah, the move was slow. Apr 12 10:53:22 but close to a decade already :) Apr 12 10:53:31 naturally it is taking forever to get back up to speed on everything Apr 12 10:53:39 there have been so many changes in 3.x Apr 12 10:53:54 but I just grabbed the source and broke stuff, rather than trying to find answers on the internet Apr 12 10:53:57 hopefully that wasn't stupid Apr 12 10:54:10 That's the best way to do it, IMO. :P Apr 12 10:54:19 Even if you end up duplicating effort, you learn. Apr 12 10:54:45 And then you can submit patches and make the various subystem maintainers cry. Apr 12 10:55:00 haha. I'm not that evil Apr 12 10:55:02 I derive so much pleasure every day from reading davidm's rants about idiot submissions. Apr 12 10:55:08 link? Apr 12 10:55:16 oh .. I shall have to join that mailing list too? Apr 12 10:55:46 I wouldn't subscribe to LKML, if I were you, unless you're a masochist. Apr 12 10:55:56 oh LKML? Apr 12 10:56:08 I've heard the horror stories Apr 12 10:56:13 didn't realize Apr 12 10:56:24 * serishema wouldn't even attempt to get a patch into mainline Apr 12 10:56:34 Well, Dave's rants tend to be on places like his G+ account, and in private with friends, but. LKML's educational, if you're bored. Apr 12 10:56:54 I tend to stay on the other side of the syscall layer, but a disturbing number of my friends are kernel developers. Apr 12 10:57:09 the right side generally Apr 12 10:57:21 Both sides suck. :P Apr 12 10:57:29 It's just insanity to try to work with both. Apr 12 10:57:37 but I had the misfortune of actually being able to make sense of the wrong side of the syscall layer of that certain popular proprietary OS Apr 12 10:57:42 haha snap Apr 12 10:57:46 So, I let other people do kernel crap, and I'm sitting here with 5 parallel glibc builds going right now. Apr 12 10:58:14 So before the recession I got paid as much as $80US an hour to do consulting with user mode/ kernel mode interface crap on windows drivers Apr 12 10:58:27 mostly for fixing STOP IRQL_LESS_THAN_EQUAL Apr 12 10:58:33 I don't remember the code Apr 12 10:58:35 I don't want tio Apr 12 10:58:43 Everyone's favourite bluescreen. Apr 12 10:58:57 Yeah. Apr 12 10:59:05 Since it's ALWAYS caused by the dev not reading the docs :) Apr 12 10:59:14 unfortunately this includes MS internal devs Apr 12 10:59:18 on wincore who should know better Apr 12 10:59:27 I never did learn enough to know what it meant. Based on the constant name, I'm assuming integer over/underflows and a lost pointer to an IRQ or something? Apr 12 10:59:34 Nah Apr 12 10:59:40 IRQL is an abstraction Apr 12 10:59:42 a moronic one Apr 12 10:59:47 Of course. Apr 12 10:59:59 it represents IRQ mask registers on the board/CPU specific interrupt controller Apr 12 11:00:45 basically if you don't know that certain things require eg the timer interrupt to be masked off Apr 12 11:00:49 you shouldn't be writing drivers Apr 12 11:01:02 But MS came up with an abstraction to make it easier which actually makes it harder Apr 12 11:01:08 Ignorance has never prevented driver development in the past. Apr 12 11:01:19 NT kernel = bastardized version of VMS Apr 12 11:01:55 * serishema doesn't even have windows installed Apr 12 11:01:56 I used VMS in high school, but we never wrote to the hardware. Apr 12 11:02:06 i only read about VMS Apr 12 11:02:08 for class Apr 12 11:02:10 before my time Apr 12 11:02:29 For some reason, they didn't think that doing kernel development classes that would let us crash the entire school by offlining the VAX would be a good idea. Apr 12 11:03:08 they were probably right Apr 12 11:03:16 Probably. ;) Apr 12 11:03:20 I didn't learn how to do windows kernel at university Apr 12 11:03:28 basically I was at a school which forced us to use Windows Apr 12 11:03:40 so at that age I had the naevity to try and make it not suck Apr 12 11:04:06 My friend's doing a CS degree right now, and I'm moderately impressed that most of their classes are on Linux systems. Apr 12 11:04:09 I became the bane of the network admins for all my custom stuff which some how ran when I logged in even though they had supposedly disabled user login scripts Apr 12 11:04:19 Most of his profs are idiots. But at least they're learning on open systems. Apr 12 11:04:29 Yeah I transfered Apr 12 11:04:56 Being forced to use windows was the lesser evil than being the only girl in a class of 185 with an openly mysogenistic bigot for a professor Apr 12 11:05:47 getting my boobs grabbed and being asked if they were real and responding with a ju jisu wristlocking was a day in the life Apr 12 11:05:53 But their course was also 10 years out of date Apr 12 11:06:04 You got groped? Seriously? Apr 12 11:06:04 In terms of material Apr 12 11:06:08 What the fuck is wrong with people? Apr 12 11:06:21 infinity: it was a get wasted, party and get laid type of school Apr 12 11:06:24 I missed the memo Apr 12 11:06:28 I mean, I'm used to geeks being socially inept, and moreso with women, but... Wow. Apr 12 11:06:32 people who actuially wanted to work went else where Apr 12 11:06:53 by other students only.. but yeah. Apr 12 11:07:13 Sounds lovely. :/ Apr 12 11:07:50 if a staff member did that the cops would hear about it. But they quickly learnt that I was not to be got on the wrong side of Apr 12 11:08:12 * serishema lived in ghettos .. built computers out of rubbish and learnt C instead of participating in drugs/gangs, etc Apr 12 11:08:22 Right, remind me not to invite you to the Ubuntu-ARM frat kegger. Apr 12 11:08:39 depends. Apr 12 11:09:20 infinity: I won't comment since freenode: "there are no girls on the internets" will be in effect.. but let's just say my boyfriend has another girlfriend and I am fine with this. Apr 12 11:09:29 * serishema isn't your average blond Apr 12 11:09:35 anyway this is getting far too OT Apr 12 11:09:36 PM? Apr 12 11:13:04 serishema: you sound kickass :-D Apr 12 11:13:44 backlog was one of the more enjoyable return to work after lunch reads Ive had for a while Apr 12 11:14:10 hi XorA Apr 12 11:14:45 VMS is a long and distant memory Apr 12 11:14:58 XorA: when it comes to VMS I actually don't know what I'm talking about Apr 12 11:15:08 i'm 27 so therefore too young to have seen a real live VMS system Apr 12 11:16:44 ah right, at univerity the student email/blah system was VMS Apr 12 11:16:56 infinity: Why try Debian? Apr 12 11:17:23 I was already used to posix style from Amiga so all the [blah]bleh; stuff used to confuse me until I got the hang of it Apr 12 11:17:28 tomtiger11: I suspect he was answering my questions when I thought my chip was something older and ubuntu core wouldn't go . Apr 12 11:17:29 tomtiger11: Why provide context with your question? Apr 12 11:17:29 I've seen a VMS system. It was even running. It just wasn't doing anything. Apr 12 11:17:41 infinity: *steal* Apr 12 11:18:09 infinity: Seriously, Is ubuntu-arm incompatible with Raspberry Pi? Apr 12 11:18:14 tomtiger11: Yes. Apr 12 11:18:23 tomtiger11: yes, Raspberry Pi is old technology Apr 12 11:18:33 infinity: Is it the RAM? Apr 12 11:18:34 tomtiger11: The Pi is ARMv6, we build for ARMv7. Apr 12 11:18:38 Oh Apr 12 11:18:57 you have to be nimble to keep up with the ubuntu boys Apr 12 11:19:05 /girls Apr 12 11:21:08 ogra: do you think we still need our biweekly call? Apr 12 11:25:00 * serishema much prefers "get a real computer" or "bitty box" and leaving it at that :) Apr 12 11:25:16 since being a bitty box is actually the whole point with arm :) Apr 12 12:00:44 tomtiger11: bring the question to the next UDS, to lower the optimization level for the community maintained armel to get back to debian level now when ubuntu-arm have switched to use armhf Apr 12 12:05:20 xranby: It's already being discussed internally. Apr 12 12:05:47 xranby: And, honestly, while it would be a community port, it needs to be driven by Canonical for infrastructure reasons, so you may as well let us beat the horse dead a bit first. Apr 12 12:05:52 xranby: (But it's being beaten) Apr 12 12:07:32 i want to build the package for Ubuntu ARM Apr 12 12:07:39 how do i do ?? Apr 12 12:08:27 pnphi: you have some ptions 1. setup a complete build environment on a ubuntu arm machine and simply build it using dpkg-buildpackage Apr 12 12:08:40 2. use xdeb to cross compile a package Apr 12 12:10:01 from source of package ubuntu ?? Apr 12 12:10:27 pnphi: yes from source Apr 12 12:11:16 by the way Apr 12 12:11:41 in the Chroot, can i build the package ? Apr 12 12:11:59 from source Apr 12 12:12:23 yes, as long as you have installed all compilers inside the chroot Apr 12 12:13:00 what link will i add source list ?? Apr 12 12:13:27 that i can get source Apr 12 12:13:35 pnphi: deb-src http://archive.ubuntu.com/ubuntu/ precise main universe restricted Apr 12 12:13:46 ok Apr 12 12:13:49 its the same source archive for all architectures Apr 12 12:14:04 for all ? ok i do Apr 12 12:15:21 ok thank you so much Apr 12 12:15:41 pnphi: youre welcome Apr 12 12:16:07 infinity: ok i take canonical wants to try a neon build .. what other optimizations are currently untested= Apr 12 12:16:08 ? Apr 12 12:16:36 xranby: Err, we do? Apr 12 12:17:03 let me rephrase the question.. what is it left to flog on the horse? :) Apr 12 12:17:42 No, no. The other direction. Apr 12 12:17:48 As in, we also want to drop the baseline for armel. Apr 12 12:18:01 But we're discussing in-house how much effort that will be, and if we really want to do it. Apr 12 12:18:35 ok, thank you for the heads up Apr 12 12:23:14 where is the tutorial for xdeb ??? i don't search The xdeb Apr 12 12:35:53 pnphi: https://wiki.linaro.org/Platform/DevPlatform/CrossCompile/UsingXdeb Apr 12 12:36:50 pnphi: wookey are the person to ask about current xdeb status Apr 12 12:38:20 ndec, not really i think, in ubuntu we just swallow what linaro gives us nowadays, but for ppisati it is important to get the kernel issues across, i wonder if we could take that bit to mail or so Apr 12 12:39:24 ogra_: ok. i see. we just need to make sure ppisati, sebjan , me and andy discuss sometimes... Apr 12 12:39:52 right, i dont know how important the call is for ppisati ... i guess thats something he needs to answer Apr 12 12:44:45 ok thanks Apr 12 12:55:08 i install the package "g++-arm-linux-gnueab" but E: Unable to locate package g++-arm-linux-gnueabi E: Couldn't find any package by regex 'g++-arm-linux-gnueabi' Apr 12 12:58:42 apt-get install gcc-arm-linux-gnueabihf Apr 12 12:58:53 try that one Apr 12 13:40:24 hmm, so doing a dpkg --get/set-selections between two ac100 precise installs (one armhf the other armel) makesd me end up with linux-image....-mx5 installed on the target system Apr 12 13:40:40 * ogra_ wonders why Apr 12 13:55:02 hmpf, actually i end up with *only* the linaro lt-mx5 image installed Apr 12 13:55:04 :( Apr 12 14:00:06 install the package "g++-arm-linux-gnueab" but E: Unable to locate package g++-arm-linux-gnueabi E: Couldn't find any package by regex 'g++-arm-linux-gnueabi' Apr 12 14:03:37 apt-get install gcc-arm-linux-gnueabihf Apr 12 14:09:34 joined Apr 12 14:09:36 install the package "g++-arm-linux-gnueab" but E: Unable to locate package g++-arm-linux-gnueabi E: Couldn't find any package by regex 'g++-arm-linux-gnueabi' Apr 12 14:09:53 what do i add source ? Apr 12 14:12:27 apt-get install gcc-arm-linux-gnueabihf Apr 12 14:21:53 E: Unable to locate package gcc-arm-linux-gnueabihf Apr 12 14:26:18 how use Xdeb to build package ?? Apr 12 15:34:27 ogra_: I don't know if you saw it yesterday, but it'd need to change vram from 32 to 40 at panda Apr 12 15:34:34 to make SGX to work properly, with flip chain and such Apr 12 15:34:52 rsalveti, yup, i did, i forgot to ask infinity if he had changed it already Apr 12 15:35:06 wasnt there also the option to drop the mem= args too ? Apr 12 15:35:06 didn't yet open a bug for it, as I know we have tons of places with that thing hardcoded Apr 12 15:35:17 since we dont have ducati support in this kernel ? Apr 12 15:35:17 not with current kernel I guess Apr 12 15:35:22 oh, that's true Apr 12 15:35:31 the next support will be with the newer kernel Apr 12 15:35:38 * ogra_ will make sure that changes before release Apr 12 15:35:39 so we can safely remove the hole then Apr 12 15:35:46 k Apr 12 15:36:02 ogra_: thanks Apr 12 18:24:04 anyone know anything about a gstreamer0.10-plugins-bad package and how to get around this show stoper Apr 12 18:25:34 looks like that package is corrupt Apr 12 18:29:42 and reflash and retry :/ Apr 12 18:37:37 MrCurious: you mean at the TI PPA? Apr 12 18:37:58 not sure what you mean? Apr 12 18:38:04 the ti licence accepting? Apr 12 18:38:28 i installed ubuntu 11.10 desktop omap4 Apr 12 18:38:43 then did apt-get update upgrade and dist-upgrade Apr 12 18:38:58 then tried to install ubuntu-omap4-extras Apr 12 18:39:09 then found gstreamer bad was failing Apr 12 18:42:30 was following these instructions http://blog.sarine.nl/2011/12/07/installing-ubuntu-11-10-on-pandaboard/ Apr 12 19:08:20 yeah, that one Apr 12 19:08:29 the broken package was caused by an issue with the builder Apr 12 19:08:40 there's one way to get it fixed, that's to skip the test checking Apr 12 19:08:44 let me email TI about it Apr 12 19:09:57 how do i skip the test checking? Apr 12 19:16:01 needs to rebuild the package Apr 12 19:16:12 if you build it locally, you'll not face that issue Apr 12 19:16:20 the problem only happens at the launchpad builder Apr 12 19:17:46 guess that means my install is on hold until they fix it Apr 12 19:22:27 MrCurious: you can build it locally and install it by hand Apr 12 19:22:48 I'll send the proposed fix now by email, I believe they will be able to fix it tomorrow Apr 12 19:23:08 awesome Apr 12 19:23:33 i am not sure where to get the sources for that, and i have never worked out how to tell apt i did sumething manually Apr 12 19:23:45 but i think i can wait until tomorrow Apr 12 19:27:22 oh wow! the install didnt loop around kbd, username, .... this time Apr 12 19:28:55 rsalveti: Running the testsuite generates a broken package? That seems a bit wrong. Apr 12 19:29:47 infinity: when running the test suite under qemu it basically explodes :-) Apr 12 19:29:49 that's the problem Apr 12 19:30:02 that's why skipping the test run fixes the build problem, because the build is fine Apr 12 19:30:14 and goes fine at qemu, just when running the tests that everything breaks Apr 12 19:31:06 Oh, I see. As in, it's actually put qemu in a broken state. Apr 12 19:31:19 And people still think qemu is a sane and reasonable solution... Apr 12 19:32:00 i am seeing that ppa.launchpad.net oneiric release is missing a public key, and cant be verified Apr 12 20:56:44 anyone know a good ARM SoC with CAN? Apr 12 20:57:02 maybe i.MX28 ? Apr 12 20:57:17 http://www.freescale.com/webapp/sps/site/taxonomy.jsp?code=IMX28_FAMILY Apr 12 20:57:32 wait, thats ARM9, aka like v5, so old Apr 12 20:58:19 http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=i.MX357 Apr 12 21:04:02 scientes, how about the ti am335x (2 can ports) beaglebone... Apr 12 21:04:14 scientes: Expand CAN? Apr 12 21:04:22 infinity, Controller Area Network Apr 12 21:04:36 its a broadcast-only (!) bus heavily used in automotive Apr 12 21:06:09 can: controller area network core (rev 20090105 abi 8) Apr 12 21:06:16 ^-- Should I assume that means I have one? Apr 12 21:06:24 infinity, yeah Apr 12 21:06:28 (That's dmesg on an i.MX53 QuickStart) Apr 12 21:06:32 check if you have a can0 with ip link Apr 12 21:06:47 (i.e. if there is a SocketCAN driver) Apr 12 21:07:00 I have no can0, but that could just be a module not loaded... Apr 12 21:07:15 Or not built in. Apr 12 21:07:26 This kernel's somewhat sketchy... Apr 12 21:08:59 scientes: The i.MX53 Data Sheet, at least, claims "Two controller area network (FlexCAN) interfaces, 1 Mbps each" Apr 12 21:10:26 infinity, try modprobe flexcan mpc5xxx_can Apr 12 21:11:02 scientes: That would work better if I had either of those modules. ;) Apr 12 21:11:12 scientes: Did I mention my sketchy kernel on this board? Apr 12 21:11:17 well, they are in drivers/net/can Apr 12 21:11:26 so do a kernel rebuild! Apr 12 21:11:33 is it mainlined? Apr 12 21:11:57 I could just install a different kernel. The machine's my mirror right now, and I don't want to offline it. Apr 12 21:12:00 yeah, that $149 board is alot better than the $100 usb dongles some are selling Apr 12 21:12:35 > 720p encode Apr 12 21:12:36 woah Apr 12 21:13:24 and SATA nice Apr 12 21:14:05 "VMware player to bring up the Linux image on a Windows PC" Apr 12 21:14:11 never knew VMware did emulation Apr 12 21:14:32 It's a good board. And the SoC clearly supports what you want. What I have no idea of is if the board provides ports/pinouts for what you need. ;) Apr 12 21:16:59 wow, the picture of it, its just so packed with ports Apr 12 21:20:29 scientes: rcn-ee also pointed out the beaglebone. You can find it on http://beagleboard.org. Apr 12 21:24:28 GrueMaster, but the iMX53 is clearly a much more powerful board Apr 12 21:25:11 Define "powerful". It really depends on your usage model/budget. Apr 12 21:25:47 (and I was just pointing out another option). Apr 12 21:26:22 yeah thxc Apr 12 21:26:26 i did see that Apr 12 21:26:34 * scientes would really like a 4X core Apr 12 21:26:58 i think we all are waiting for that day. ;) Apr 12 21:27:32 i'm actually using a E-350 x86 board Apr 12 21:27:58 but i'd like CAN, and with the prices of usb ones (that are just integrated ARM chips with a gadget port) it makes sense to use one of these Apr 12 21:28:03 if i want CAN at all Apr 12 21:28:19 Not to brag, but I have tested one. Server side only though. Apr 12 21:28:51 4x Core system, that is. Apr 12 21:29:21 I've played with i.MX6s too. Sadly, they don't actually exist. Apr 12 21:29:34 The ones I used were, I was told, figments of my imagination. Apr 12 21:29:48 Heh. Apr 12 21:30:00 so did you take it home for yourself, and just tell them it never existed.. Apr 12 21:30:07 * GrueMaster wants armv8. Apr 12 21:31:45 64-bit baby Apr 12 21:34:03 anyone awake? Apr 12 21:34:48 I tried to install Ubuntu 11.10 with a preinstalled omap4 desktop image on my brand new Pandaboard ES Apr 12 21:35:15 It worked fine, until third reboot(after it resized my partition, and then checked the fs) it just greeted me with a terminal login Apr 12 21:35:30 I checked the /etc/passwd and the /home folder, no users except root Apr 12 21:35:47 got image from http://cdimage.ubuntu.com/releases/11.10/release/ Apr 12 21:37:55 well, there were plenty of users in etc/passwd but no regular one, and no folders in homedir Apr 12 21:38:58 are there some cheaper LVDS touchscreens? Apr 12 21:47:52 scientes: tincantools will have some lvds boards and kits available at the end of next week Apr 12 21:48:05 scientes: first models will be without touchscreen Apr 12 21:48:31 scientes: works with panda and beagle - http://tincantools.com/product.php?productid=16158&cat=0&page=1&featured Apr 12 21:49:25 scientes: is there something specific you are looking for? Apr 12 21:51:22 no really prpplague Apr 12 21:51:28 i have a mimo usb touchscreen Apr 12 21:51:33 that is working with multiseat Apr 12 21:53:03 anyone have a snowball board? Apr 12 21:55:02 whats that? Apr 12 21:57:24 XorA: hehe Apr 12 21:59:50 what? Apr 12 21:59:54 do I miss something? Apr 12 22:10:14 XorA: hehe, no just points out that it isn't well known Apr 12 22:10:29 I had to goggle it Apr 12 22:18:26 woo hoo, got 11.10 on my pandaboard, and got the omap extras installed Apr 12 22:53:39 > omap extras Apr 12 22:53:45 *binary blobs Apr 12 23:37:53 pandaboard ES and 12.04, after latest update there is no desktop, just the login screen over and over. Apr 13 00:46:39 and goes fine at qemu, just when running the tests that everything breaks Apr 13 00:46:44 argh Apr 13 00:47:01 jimerickson: do you have the pvr driver installed? Apr 13 00:47:13 did you try to login at ubuntu 2d? Apr 13 00:50:14 rsalveti: Did you break something? Apr 13 00:51:18 infinity: nops, just wanted to know if the issue is related with unity 3d Apr 13 00:52:13 but I noticed today that nux also got updated to a newer version Apr 13 00:52:22 seems to be still at proposed Apr 13 00:52:54 I think that's all waiting on some co-ordinated unity release. Apr 13 00:53:03 I've lost track. Apr 13 00:53:40 I'm scared that this could break the desktop again, we'll see Apr 13 00:53:50 at beta 2 I couldn't even login at unity 3d Apr 13 02:45:30 rsalveti: yes i had the pvr driver installed. yes tried login at 2D. Apr 13 02:45:43 jimerickson: even 2d failed? Apr 13 02:45:58 yes Apr 13 02:46:09 I'm updating my rootfs, will check in a few minutes Apr 13 02:47:11 i am re-imaging my sd card with yesterdays image. will try again without pvr driver installed. Apr 13 02:49:00 sgx shouldn't interfere at unity-2d Apr 13 02:53:02 hm, ubuntu-desktop is not installable anymore Apr 13 02:53:13 the update removed a few packages as well Apr 13 02:54:17 probably archive out-of-sync issues Apr 13 02:59:16 should i wait to update? **** ENDING LOGGING AT Fri Apr 13 02:59:58 2012