**** BEGIN LOGGING AT Thu Jul 12 02:59:58 2012 Jul 12 08:45:27 morning all Jul 12 08:48:00 bluelightning: afternoon here Jul 12 08:49:16 dlan: aha, but http://www.total-knowledge.com/~ilya/mips/ugt.html Jul 12 08:49:21 ;) Jul 12 08:51:25 bluelightning: oh, so, morning~ Jul 12 08:53:50 bluelightning: do you know who responsible for mips related patch? I'd like to get mipsel support in, but seems no one interest, or care?! Jul 12 08:54:25 the patch titled with "runqemu/mips: adjust runqemu script to support mipsel machine", which sent to the OE-core Jul 12 08:54:57 it's actually tivial, but I'd like get in, also if reject, I could carry it myself Jul 12 08:55:04 dlan: things are a bit tricky atm, we went through a period of very nasty breakage so RP_ is only taking obvious patches atm Jul 12 08:55:16 dlan: you could reply to your own patch with a ping Jul 12 08:56:10 bluelightning: actually I replied, but I don't who I should ping to.. Jul 12 08:56:33 yeah, bug fix only? no feature added, probably.. Jul 12 09:33:30 hi all Jul 12 09:46:34 dlan: I just had a look at your patch and replied Jul 12 09:47:45 bluelightning: yeah, I got your reply Jul 12 09:48:38 previously, i'm using "case ..esac" and then thought substitution maybe simplier Jul 12 09:53:46 bluelightning: how about this, if you're ok, I will re-send, http://pastebin.ca/2169918 Jul 12 09:55:27 dlan: yep, looks good to me Jul 12 09:55:45 bluelightning: ok, thanks Jul 12 10:50:21 morning all Jul 12 10:50:23 * pb_ stabs directfb Jul 12 10:51:00 their commit aaebf60b86acc666e4c680e39d0f72676b8298d5 is the most misguided fix for a "compiler warning" that I have seen for a long time. Jul 12 10:54:29 pb_, which repo? Jul 12 10:54:57 the main git.directfb.org one Jul 12 10:55:06 http://git.directfb.org/?p=core/DirectFB.git;a=commitdiff;h=aaebf60b86acc666e4c680e39d0f72676b8298d5 Jul 12 11:08:42 OE fails to build pseudo-native using the angstrom oebb.sh script. Jul 12 11:08:45 http://paste.frubar.net/15022 This is the log - Where could I start? Jul 12 11:08:54 *start to solve this problem. Jul 12 11:09:06 it likes like your machine is missing some headers Jul 12 11:10:48 /usr/include/gnu/stubs-64.h do you have this file? Jul 12 11:11:19 Is is MY machine or did OE miss to download them? as fas as I understand, OE should not use the local installed libs for crosscompilation? Jul 12 11:11:34 this is a native package Jul 12 11:11:42 so I believe it is using the host compiler Jul 12 11:11:56 your machine Jul 12 11:12:12 OE tries hard not to dpeend to much on your machine, but some things it needs Jul 12 11:12:55 ah, okay. The file is missing. So, I'll search for its dependencies. Jul 12 11:13:20 glibc-devel on Fedora Jul 12 11:13:37 gm Jul 12 11:13:40 OE should have checked before starting :) Jul 12 11:13:41 gm Jul 12 11:14:22 hmm, Debian did move to the eglibc some time ago afair ... I'll check Jul 12 11:14:36 ah yup Jul 12 11:14:44 florian: You seem to have worked with i.MX a lot. Aware of any issues with the i.MX28 FEC behaving erratically during init? Jul 12 11:15:53 okay, lets try again Jul 12 11:22:46 It was not the clib Jul 12 11:26:31 Hold on ... Shame on me Jul 12 11:26:36 heh Jul 12 11:26:42 you had me worried Jul 12 11:27:22 Debian moved to the eglibc, so I installed the eglibc dev package Jul 12 11:27:37 but this was obviously not the package it wanted Jul 12 11:28:21 /usr/bin/ld: i386 architecture of input file `/home/einball/oe/setup-scripts/build/tmp-angstrom_v2012_05-eglibc/sysroots/x86_64-linux/usr/lib/libsqlite3.a(sqlite3.o)' is incompatible with i386:x86-64 output Jul 12 11:28:42 I really ... hate this :) Jul 12 11:29:03 first it depends on the 64bit then it complains about not being able to link. Jul 12 11:36:04 likewise: hmm... not yet, but maybe i.MX28 is somewhat different and this is the one I have not done that much with personally Jul 12 11:45:16 florian: ok thanks Jul 12 12:17:25 einball: imx28 + recent oe (core + meta) + fedora 17 (x86_64) works like a charm here... Jul 12 12:18:15 however, fec init is tricky because reset must be applied manually by keeping modX pins high Jul 12 12:20:21 ensc|w: thanks for chiming in. what are the modX pins? Jul 12 12:21:53 3 pins to the external ethernet phy (afair, rx0/1 + rx_en for the reference lan87xx phy) Jul 12 13:45:46 ensc|w : tnx, need to check that. Jul 12 13:47:51 ah, i see i.mx28 support is in meta-fsl-arm now... I know at least one person who will be pleased by that Jul 12 13:47:59 :) Jul 12 13:51:01 bluelightning: referring to self? :) Jul 12 13:51:42 no, I don't have one of those devices, but I spoke to one person who was keen to move to OE-core but couldn't due to the lack of a bsp for that device Jul 12 13:52:21 There is tons of BSP's for the i.MX28 for OE over the last five years. (That might be the exact problem) Jul 12 14:01:20 likewise: any examples of such BSPs? Jul 12 14:01:21 Crofton|work: but please check by compiling hello.c instead of stat(${inludedir}/gnu/stubs.h) :D Jul 12 14:05:36 slapin: http://www.gitorious.org/schnitzeltony-oe-meta/meta-freescale Jul 12 14:06:13 slapin: and there were some OE classic "BSP";s floating around. Jul 12 14:06:39 slapin: different boards, different kernels, etc etc. Jul 12 14:07:59 ensc|w: I really don't understand, where the problem is and why nothing works on my machine :P Jul 12 14:09:34 I compiled PHP on my beagleboard .. because I needed it. But that took like forever :D Jul 12 14:12:12 any at91 BSPs, by the way? Jul 12 14:22:16 slapin: well, yes for OE Classic all popular embedded devices (until 2011) are covered AFAIK. Have you looked at OpenEmbedded Classic? Jul 12 14:22:33 slapin: I have used at91 /9200 and G45 series. Jul 12 14:23:34 ensc|w: can you show a kernel that pull those pins high explicitly from software (I assume temporarily config'ing these pins as GPO's? Jul 12 14:41:36 likewise: I try to migrate from OE-classic and too lazy to make my own layer hoping somebody already made that (even Atmel themselves) Jul 12 14:42:32 likewise: I'm mostly interested in 9260 and 9g20 Jul 12 14:43:47 and I beleive, that since interest in at91s is so low these days, an architecture is almost dead Jul 12 14:43:48 slapin: google: http://gitorious.org/oe-atmel/meta-atmel Jul 12 14:44:05 slapin: no, don't assume it to be dead. Jul 12 14:44:39 when i try to compile qt-embedded-gless i obtain error like this, some suggest? Jul 12 14:44:40 slapin: assuming it *is* dead, what SoC has taken it's place? Jul 12 14:44:54 ../../include/QtCore/../../src/corelib/kernel/qcoreevent.h:75:9: error: expected identifier before numeric constant Jul 12 14:44:54 | ../../include/QtCore/../../src/corelib/kernel/qcoreevent.h:75:9: error: expected '}' before numeric constant Jul 12 14:44:54 | ../../include/QtCore/../../src/corelib/kernel/qcoreevent.h:75:9: error: expected unqualified-id before numeric constant Jul 12 14:44:54 | ../../include/QtCore/../../src/corelib/kernel/qcoreevent.h:298:17: error: expected ')' before 'type' Jul 12 14:45:20 likewise: last commit is from 2011 in that repository Jul 12 14:45:36 its seams i need some includes, but it's not clear for me Jul 12 14:46:25 likewise: where I live, people replace at91s with low-end Samsung s3c24xx and i.MX23 Jul 12 14:46:37 slapin: yeah well, did the chips change afterwards? Maybe it's super stable :) Jul 12 14:46:48 slapin: where do you live? Jul 12 14:46:51 and ARM :o Jul 12 14:47:07 likewise: Russia Jul 12 14:47:14 einball: arm what? Jul 12 14:47:36 einball: we are only talking about ARM chips in this discussion, AFAIK Jul 12 14:48:21 likewise: but I don't see new projects with at91 outsourced to me anymore, which also means something. Jul 12 14:48:38 slapin: there is a trend to just pick the cheapest parts, there is a lot of choice when you do not need features like SATA, H.264. Jul 12 14:49:10 slapin: what is the current trend than? i.MX? Jul 12 14:49:42 likewise: Cortex-A* CPU's Jul 12 14:49:49 likewise: but people tend to stick to the same things, why people change from at91 to mxs without any particular reason? that's costs money after all. Jul 12 14:50:17 at91 is ARM? Jul 12 14:50:38 einball: at91 contains ARM core Jul 12 14:50:44 oh, didn't know that Jul 12 14:51:01 Many people here in germany tend to use Cortex µC Jul 12 14:51:03 einball: most of these are arm926ej-s Jul 12 14:51:18 einball: at91s that is Jul 12 14:51:30 einball: with Linux or without? Jul 12 14:51:37 not or .. and! Jul 12 14:51:42 They use both Jul 12 14:51:45 einball: 'Cotrex' is too generic term Jul 12 14:52:20 I personally like to use cortex-A* Jul 12 14:52:39 A friend of mine says he plays better with the M series Jul 12 14:53:14 But I don't have enought knowledge to program them. Jul 12 14:53:15 einball: m series are very small mcus, a series are application processors or cpus. Jul 12 14:53:35 I know Jul 12 14:53:44 einball: read manuals at arm.com, they are quite open these days, and vendor's datasheet Jul 12 14:54:32 Depends on your application, but they develop gas stations Jul 12 14:54:33 einball: if you have proper dev board, it is usually no more than 2 weeks to start with new processor Jul 12 14:54:37 so they might go with both Jul 12 14:54:54 einball: depends on your applications, actually Jul 12 14:55:46 My knowledge stops at Arduino and Atmega's Jul 12 14:57:02 Iam more the hardware and signaling guy Jul 12 14:57:20 *signal Jul 12 15:02:13 einball: the knowledge on how to bring up these arm beasts won't harm even if you want to specialize, because finding bad soldering might be a huge task (last time it took a week and 3 broken boards for me). Jul 12 15:03:20 slapin: that's the problem. I study electrical engineering and basically learn nothing about manufacturing. Jul 12 15:04:05 So I'll have to learn on my own - very hard and frustrating if you have almost no time to do so because you will fall behind with your studies Jul 12 15:04:52 einball: buy parts and solder something simple yourself, that's great experience, very cheap, lots of fun and doesn't take too much time Jul 12 15:05:29 einball: some led blinker is a good start Jul 12 15:05:56 Currently I build a 2000 parts HF transceiver, so I know how to solder and I like it. But it's the rough "surface" of that bare metal that worries me Jul 12 15:06:21 *I currently Jul 12 15:06:38 (Programming part) Jul 12 15:07:57 likewise: as for at91, 'super-stable' approach is good for supporting legacy stuff, and is very bad for developing new product, so Atmel approach shows well, that they relly consider their architecture as legacy. Jul 12 15:08:33 likewise: while hardware doesn't change, software is continuousely evolving. Jul 12 15:11:10 einball: processor/mcu-based devices can be far simple than that, so you can take some cortex-m0, some power supplies, capacitors, crystal, ft2232 for jtag and uart, and have device under 100 parts which can do lots of things, and you learn new processor. Jul 12 15:12:20 einball: that's fun. because you solder it for 1-2 hours and have something working. Jul 12 15:14:04 I'll maybe start in half a year. Before I may do another µC I'll have to learn programming the 8051 in assembler :-/ Jul 12 15:14:30 bluelightning, the proto=file and path to local git repo does the trick, at least in classic Jul 12 15:14:34 (university stuff .. ) Jul 12 15:15:07 Crofton: ah right, that should also work in core Jul 12 15:16:26 einball: 8051 is really legacy. AVR and ARM are much much simpler, and less brain damaged... Jul 12 15:16:36 someone use gles with nvidia tegra? Jul 12 15:16:47 qt-embedded-gles Jul 12 15:16:50 I know ... And I hate my professor for using this old stuff. Jul 12 15:27:16 silvio__: curiosity: which tegra chip and board are you using? Jul 12 15:28:13 hehe tegra Jul 12 15:29:58 woglinde: don't you have a tegra-based netbook? Jul 12 15:30:39 bluelightning yes I have tegra2 Jul 12 15:34:15 toradex t20 Jul 12 15:34:29 woglinde: could you please tell the model of tegra2 netbook? Jul 12 15:34:29 *g* Jul 12 15:34:33 likewise: toradex t20 Jul 12 15:34:36 slpain ac100 Jul 12 15:34:41 args slapin Jul 12 15:34:50 and I bought the tablet too Jul 12 15:34:55 woglinde: is it tegra2? Jul 12 15:34:58 yes Jul 12 15:35:01 both Jul 12 15:35:24 tablet is folio100 Jul 12 15:35:30 still running android Jul 12 15:35:32 wow, than I have tegra2 too, I thought it was tegra1 Jul 12 15:35:39 no Jul 12 15:35:57 there wasnt any real customer tegra1 device so far Jul 12 15:36:29 woglinde: I run ubuntu 11 on ac100, it somewhat works, no drivers, though. Jul 12 15:36:44 woglinde: that's cool to know Jul 12 15:36:51 upgrade to 12.04 and kernel 3.1 Jul 12 15:37:17 btw. because of tegra I have one commit in linux ;) Jul 12 15:37:25 toradex t20 was the red cross sponsoring action at EW2012 Jul 12 15:37:41 woglinde: I think it won't help with graphics much. Jul 12 15:37:41 almost bought one, but it would end up on the pile here Jul 12 15:38:02 I didnt finish the tegra-layer so far Jul 12 15:38:53 I'd prefer to install OE on that netbook, to reduce memory footprint and make it somewhat faster. Jul 12 15:39:02 uhm Jul 12 15:39:16 install blackbox or some else wm Jul 12 15:39:23 you dont get much ram with oe Jul 12 15:39:28 I have a recipe question... Currently I have a recipe which takes a pristine source and applies a patch to it. When I update the patch, I have to update the PR to ensure that the system picks up the changed patch when building the recipe. Jul 12 15:40:02 you have to update the PR so your package repos know there is an update Jul 12 15:40:30 I'd like to generate the patch every time the recipe is run. I can use svn diff to do this, but I'm not sure how to force OE to *always* rebuild the package since I dont' think I want to update the PR every single time the build is done Jul 12 15:40:41 woglinde: I'd prefer to cross-compile stuff for it, et al. Jul 12 15:40:55 slapin do what you want ;) Jul 12 15:41:01 its opensource Jul 12 15:42:52 likewise: so what u use? Jul 12 15:43:00 woglinde: does the PR have to be ever-increasing, or does it just have to be different from the last build? Jul 12 15:43:12 tzanger o.O Jul 12 15:43:37 when you update you either increase the pr or put it pack to 0 if it is a major update Jul 12 15:43:45 ups back Jul 12 15:43:57 hi kgilmer Jul 12 15:44:07 silvio__: Currently I have AVR32, PowerPC 2020, i.MX53 projects Jul 12 15:44:45 woglinde: can I make hte PR the md5sum of the patchfile? that way the PR will be different if the patch changed from last time, and remain the same if not Jul 12 15:44:46 silvio__: and i.MX28 Jul 12 15:44:55 er.. wait. PR is checked before it decides to build. that won't work at all Jul 12 15:45:58 tzanger what is your problem? Jul 12 15:46:10 new version -> PR change Jul 12 15:47:32 woglinde: the patch is generated upon package build. I don't want to manually create the patch and manually update PR when the patch changed Jul 12 15:47:53 o.O Jul 12 15:48:01 basically if I ask bitbake to build the package, I want to generate the patch and build only if the patch is different from last time Jul 12 15:48:02 (or there was no last time) Jul 12 15:48:32 tzanger I dont think bitbake support this Jul 12 15:48:43 and I dont know whats so hard on PR change Jul 12 15:48:48 I'm trying to determine how bitbake decides whether to build something Jul 12 15:49:10 woglinde: automated daily builds with a development team that is changing the branch I'm generating a patch from Jul 12 15:49:21 I guess I can script it to automatically mangle the recipe but that sounds messy Jul 12 15:49:37 PR change isn't hard... unless you have to generate the patch and bump PR every night Jul 12 15:50:06 choose another development process Jul 12 15:51:11 what I can do is set up a subversion commit hook that generates the patch and updates the PR, It hink Jul 12 15:51:12 er think Jul 12 16:04:22 woglinde: Do I have to update PR if my URI_SRC is a repository? Will OE always pull from the repo and detect changes when asked to build? Jul 12 16:04:43 tzanger: use OE-Core Jul 12 16:04:46 i.e. the recipe does not change, but the patch file in the SRC_URI svn:// repo changed Jul 12 16:04:51 and you will get what you want Jul 12 16:04:59 I've never heard of OE-Core, googling Jul 12 16:05:09 its OE nextgen Jul 12 16:05:11 ah Jul 12 16:05:27 woglinde: howdy Jul 12 16:05:28 there are plans to move to that but I can't move everything over from Angstrom for this Jul 12 16:05:41 woglinde: did systemd/uclibc work for you ? Jul 12 16:06:03 tzanger: angstrom as of today uses OE-Core too Jul 12 16:06:17 so if you use latest angstrom you should get it as well Jul 12 16:07:16 hi khem Jul 12 16:07:23 yes but that's the same issue... updating the entire repo and fixing whateve breakage is found Jul 12 16:07:43 we'll update in time but updating *today* or for this software cycle isn't possible Jul 12 16:07:59 khem sorry didnot test latest state Jul 12 16:07:59 then live with what you have is only option for you Jul 12 16:08:08 woglinde: please try it out Jul 12 16:08:18 woglinde: I didnt have enought cycles Jul 12 16:08:25 khem: I am, I'm asking questions about what OE does Jul 12 16:08:32 khem uclibc-git? Jul 12 16:08:36 woglinde: yes Jul 12 16:08:41 okay Jul 12 16:08:50 hm baby is angry Jul 12 16:08:59 woglinde: I want angstrom to work with uclibc Jul 12 16:09:05 woglinde: you got a second one ? Jul 12 16:09:10 if I have a svn:// URI in SRC_URI and bitbake is asked to build the recipe, does it check ot see if the svn repo changed from the last time the recipe was built? Jul 12 16:09:15 khem yes Jul 12 16:09:48 tzanger: if you use AUTOREV for SRCREV then it will Jul 12 16:10:10 woglinde: both boys ? or mix or girls Jul 12 16:10:37 khem: aha... let me look those up, thank you Jul 12 16:10:56 yes both boys Jul 12 16:11:20 woglinde: how old is the baby? I just had a son on the 3rd Jul 12 16:11:43 woglinde: 3rd July, ? Jul 12 16:19:05 bb Jul 12 16:19:18 quit Jul 12 16:23:29 khem: hm, ok, if I'm reading this correctly then in my local.conf I would add "SRCREV_my-recipe-here ?= ${AUTOREV}" to local.conf. That would instruct OE to always pull from the repo for that package, is that correct? Jul 12 16:24:13 you would do it in recipe Jul 12 16:24:18 my local.conf has no lines relating to packages right now, which is why I'm unsure that I understand the OE user manual Jul 12 16:25:28 hm ok, the documentation has a bug then. Jul 12 16:25:29 it states "If you really absolutely have to follow the latest commits, you can do that by adding ’SRCREV_pn-linux-davinci ?= ${AUTOREV}’ to your local.conf, for example. In this case, you’d build against the most recent and unstable source for the pn-linux-davinci package." Jul 12 16:25:52 khem 16.5 Jul 12 16:38:47 woglinde: congrats Jul 12 16:38:59 woglinde: my kids are 4 and 6 Jul 12 16:39:11 little bigger troubles than you Jul 12 16:39:16 but still troubles :) Jul 12 16:47:31 khem: wait until they hit their teens, or in the case of daughters, about 8 :-) Jul 12 16:50:45 I've got them from 17 down to two weeks. little ones are easier and fun, but around 7-12 for boys is really nice Jul 12 17:14:29 hi Jul 12 17:14:45 meta-kirkwood is the only kirkwood support OE has nowadays? Jul 12 17:27:37 hrw: AFAIK yes. Jul 12 17:27:53 hrw: and it's probably not (much) newer than classic. Jul 12 17:28:05 khem what do I need in local.conf for master of meta-angstroem? Jul 12 17:28:13 hrw: haven't touched it though for long Jul 12 17:28:31 all I tried so far it did not find the right version Jul 12 17:28:32 likewise: I unbricked my sheevaplug and it runs fine with HEAD uboot/kernel Jul 12 17:28:33 hrw: I will open that can of worms, when Marvell gives me Armada XP Jul 12 17:28:39 likewise: :D Jul 12 17:28:52 armadaxp... I played a little with their devboard Jul 12 17:28:54 hrw: so far, they cannot push Reply on their email :) Jul 12 17:29:04 hrw: drool, I need the PCIe on it Jul 12 17:29:16 hrw: 2x PCIe x4 right Jul 12 17:29:39 hrw: How can I get one? Should I apply at Canonical? Jul 12 17:30:16 likewise: I do not know who at Canonical has them even Jul 12 17:30:27 likewise: I spend time at Linaro rather then at Canonical Jul 12 17:30:38 and the one I played with was not Canonical either Jul 12 17:31:15 hrw: Marvell is hard to approach, even their dist is not happy about small customers Jul 12 17:33:57 hm okay found it Jul 12 17:42:19 Crofton: no i compiled that version Jul 12 17:44:36 hm or not Jul 12 17:44:38 damn Jul 12 17:45:38 arsg typo Jul 12 17:46:54 lovely ;) Jul 12 17:47:33 I flashed fresh uboot, booted 5MB kernel (with busybox initramfs) and have network and angstrom repositories available so can setup ubifs etc Jul 12 17:48:41 hrw cool Jul 12 17:49:20 hrw how is oe being used in linaro now i wonder Jul 12 17:51:03 onoffon: toolchain WG uses oe for some gcc testing Jul 12 17:51:08 cbrake are you here? Jul 12 17:51:21 likewise howdy Jul 12 17:51:44 you mentioned some armv8 using it Jul 12 17:52:12 * onoffon is khem on phone btw Jul 12 17:53:01 onoffon: we (I) will add armv8 support to get something bootable quickly so we can start bootstrapping ubuntu and test fastmodels Jul 12 17:53:35 nice Jul 12 17:54:10 i would wait for your patches Jul 12 17:54:11 onoffon: howdy to you too :) Jul 12 17:54:36 (Re: [19:51] onoffon likewise howdy) Jul 12 17:55:07 likewise do u use avr32 Jul 12 17:56:51 onoffon: yes w/o linux in a current project, with linux in a VERY OLD project Jul 12 17:58:04 ok Jul 12 17:58:15 what libc Jul 12 18:00:40 hrw it will be 64 bit right Jul 12 18:00:50 onoffon: yes, 64bit Jul 12 18:01:02 onoffon: 32bit support is optional on aarch64 Jul 12 18:01:58 cool Jul 12 18:02:05 also qemu Jul 12 18:03:51 no, qemu is not planned yet Jul 12 18:04:24 there is armv8 fastmodel but I do not remember when it will be publically available (if at all) Jul 12 18:04:27 hmm Jul 12 18:05:06 so what will u use for target Jul 12 18:05:18 vexpress-v8 Jul 12 18:05:28 or how it will be named Jul 12 18:05:37 k Jul 12 18:05:46 at beginning probably linux-dummy only from OE side Jul 12 18:05:53 no details yet Jul 12 18:06:14 arm wont send free to devs :) Jul 12 18:06:43 onoffon: you know that 'legal' part of ARM is >50% of company? Jul 12 18:07:17 yeah linux dummy is a breather when doing new arch port Jul 12 18:07:36 where the heck preferred version 0.9.32+0.9.33% comes from Jul 12 18:07:45 so iys a lawyer company Jul 12 18:08:13 i used it for mips 64 Jul 12 18:08:52 args found it Jul 12 18:10:19 woglinde angstrom? Jul 12 18:12:15 yeah I had it in site.conf Jul 12 18:13:58 leds on sheeva are fun... kernel tells: "there is red and green" when I see green and blue. At the end there is no red one and green(kernel) is blue(device) Jul 12 18:18:13 your kernel is color blind Jul 12 18:26:12 i compiled using bitbake, how do i get the toolchain? Jul 12 18:26:19 to cross compile Jul 12 18:26:32 bye all Jul 12 18:26:35 when i downloaded a narcissus image o got the toolchain for free, how to build it myself ? Jul 12 18:26:44 Desert: bitbake meta-toolchain(-sdk) Jul 12 18:26:57 hrw cool! Jul 12 18:26:59 thanks Jul 12 18:31:29 whats the difference between meta-toolchain and meta-toolchain-sdk ? Jul 12 18:33:52 how do i install aditional packages to a toolchain ? Jul 12 18:33:57 i need opencv Jul 12 18:34:50 or maybe im asking the wrong question, im developing in my (x86) laptop, but i cross compile, so i need opencv for ARM installed along my dev system Jul 12 18:34:55 how can i achieve this ? Jul 12 18:39:06 Desert: in that case, you should build it for your target Jul 12 18:39:25 and thenenter it as dependency to those recipes where you need opencv Jul 12 18:44:34 so, i can modify meta-toolchain and enter opencv requirements? Jul 12 18:44:48 and it will create a toolchain with opencv baked in ? Jul 12 18:51:44 not sure what you mean with baked in ..? Jul 12 18:51:51 opencv is just a bunch of libraries, right? Jul 12 18:54:30 uhm, yes, i was thinking about a tar.gz with everything packed, to distribute to colleagues with everything set up Jul 12 19:15:45 Desert yes than you need to extend to opencv recipe for sdk stuff Jul 12 19:16:25 woglinde: can you point to a tutorial on how to do this, im a newbie Jul 12 19:18:28 you could try to add BBCLASSEXTEND = "native nativesdk" Jul 12 19:18:33 in the recipe Jul 12 19:18:41 or nativesdk only Jul 12 19:18:57 but maybee all requirements of openvc need thats too Jul 12 19:19:05 but it depends on the recipe too Jul 12 20:06:49 hm coreutils complains about LIC_FILES_CHKSUM Jul 12 20:38:25 Desert: there are two variables you need to add to when it comes to SDKs TOOLCHAIN_TARGET_TASK for any target packages and TOOLCHAIN_HOST_TASK for natisdk packages Jul 12 20:38:33 just do TOOLCHAIN_TARGET_TASK += " Jul 12 20:38:49 TOOLCHAIN_TARGET_TASK += "opencv-dev" and you must be set Jul 12 20:39:15 khem hm ah Jul 12 20:39:44 hi khem Jul 12 20:39:56 hi woglinde Jul 12 20:40:07 hi gnutoo Jul 12 20:40:14 GNUtoo-desktop: hello Jul 12 20:40:43 khem, firefox is still broken, I wonder if I update the recipe, would it have a chance to be magically fixed? Jul 12 20:41:32 (chromium is also broken but I care less about it) Jul 12 20:51:37 GNUtoo-desktop: I doubt Jul 12 20:51:54 ok Jul 12 20:51:58 GNUtoo-desktop: I think we have to try with binutils bleeding edge Jul 12 20:52:05 ok Jul 12 20:52:10 I think its fixed in top of tree Jul 12 20:52:17 ok nice Jul 12 20:53:18 how do I try bleeding edge binutils? Jul 12 20:53:36 good question. Jul 12 20:53:44 I guess I will create recipes and let you know Jul 12 20:53:51 wont happen today Jul 12 20:53:56 but I made a note Jul 12 20:54:13 ok Jul 12 20:54:17 thanks a lot!!!! Jul 12 21:44:53 hi pwgen Jul 12 21:45:07 hi .. Jul 12 21:45:44 installing debian with debian installer on an external usb disk (:-) Jul 12 21:45:59 and? Jul 12 21:46:10 where is oe in that? Jul 12 21:46:11 it seems i do nat have a problem with ext4dev in the kexecboot, its more lvm related .. Jul 12 21:46:47 ... FG.. that is the image from where i started the debian installer .. Jul 12 21:49:03 i have build serveral oe images , systemd-image is booting very fast ... Jul 12 21:50:10 anyone ever see: https://gist.github.com/3101315 Jul 12 21:51:33 I can run the task standalone and it works Jul 12 21:51:53 just under bitbake does it fail Jul 12 21:52:09 msm could be qemu which segfaults Jul 12 21:52:14 check your dmesg output Jul 12 21:53:00 rm[24817]: segfault at 0000000000000000 rip 0000003eb26784a0 rsp 00007fff4b7dec68 error 4 Jul 12 21:53:06 im not in qemu though? Jul 12 21:53:17 I never saw rm segfaulting Jul 12 21:53:55 i wonder if i need to reboot Jul 12 21:54:01 i did upgrade a lot of packages on this machine Jul 12 21:54:28 o.O Jul 12 21:54:29 centos 5.6 -> 5.8 Jul 12 21:54:48 maybee Jul 12 21:55:01 woglinde: pwgan said it was a pain to kexecboot linux on that tegra3 Android tablets ;) Jul 12 21:55:13 uhm Jul 12 21:55:22 I wonder that kexecboot works at all Jul 12 21:55:38 kexec needs sort of hardboot patch Jul 12 21:56:19 there is a strange KEXEC_HARDBOOT option for a 3.1.10 kernel that allows to use a patched kexec Jul 12 21:56:29 then they still use ext4dev so this doesn't match with fs-detection Jul 12 21:56:53 anymore Jul 12 21:57:00 and with this any linux (arm based )can run on the tf201 Jul 12 21:57:02 pwgen the 3.1.10 kernel from nvidia git? Jul 12 21:58:02 woglinde: http://eric-weiss.de/oe/oe-core/meta-erics/recipes-kernel/linux/linux-tf201-stock_git.bb Jul 12 21:58:52 but still fighting with varoius kernel configurations . Jul 12 22:00:25 ah yes they are pulling over the nvidia git tree Jul 12 22:00:27 but i could manage to get the Angstrom-Gnome image build and run. but not touchpad and touchscreen Jul 12 22:00:36 yupp Jul 12 22:01:13 hm Jul 12 22:01:20 xorg problem? Jul 12 22:01:55 xorg-driver-input-mtouch missing . thats what they are using on ubuntu, and i could get it work under debian Jul 12 22:02:33 but ubuntu did a lot of patchwork to detect the atmel-maxtouch touchscreen proper Jul 12 22:02:54 pwgen, we have a multitouch driver.... Jul 12 22:02:57 let me look Jul 12 22:03:02 it's used at least for crespo Jul 12 22:03:10 (nexus S) Jul 12 22:03:20 hm I wonder if that hard kexec works on tegra2 too Jul 12 22:04:04 ah Jul 12 22:04:06 +#elif defined(CONFIG_ARCH_TEGRA_2x_SOC) || defined(CONFIG_ARCH_TEGRA_3x_SOC) Jul 12 22:04:08 looks like Jul 12 22:04:10 xf86-input-mtev_git.bb Jul 12 22:04:22 it's not mtouch tough Jul 12 22:04:23 after i can now kexecboot i will try to get as most stuff ass possible run under debian , and then add it to my oe-core images . Jul 12 22:05:53 woglinde: ask RayMan in the asus-transformer channel. i think he knows also a lot about tegra2 ( your AC100 ) Jul 12 22:06:57 pwgen can you give me a short breefing how do you call kexec stuff? Jul 12 22:08:39 woglinde: there are 2 small patches for the kexec-tools-2.0.2 required. and the patches for KEXEC_HARDBOOT Jul 12 22:10:19 then you can call kexec -l --load-hardboot --mem-min=0x98000000 --mem-max=0xc0000000 --append=init=/dev/zero(:) --initrd=something /boot/zImage Jul 12 22:10:44 init dev/zero Jul 12 22:10:45 lol Jul 12 22:10:53 *FG* Jul 12 22:11:06 or init=/bin/emacs ... Jul 12 22:11:21 okay kexec than put all stuff into ram Jul 12 22:11:31 makes a reset and calls the kernel from there Jul 12 22:12:55 the kexec -e will coldstart the system, it will load the asus loader, and then start the defined kernel, but the target kernel needs to be somehow ( assume KEXEC_HARDBOOT enabled ) patched , for this zou cannot boot the origianl andoid image Jul 12 22:13:58 and the bootloader must be "rooted", they have something with bootloader encryption Jul 12 22:17:21 q Jul 12 22:18:29 http://pastebin.com/Xc0NdXJz Jul 12 22:20:40 and http://pastebin.com/P7vsyqQy are required to patch kexectools-2.0.2 Jul 12 22:21:14 good nite Jul 12 22:21:23 CU Jul 12 22:33:41 khem: i guess i have to add that TOOLCHAIN_TARGET_TASK += "opencv-dev" to some .bb, my guess is meta-toolchain_****.bb Jul 12 22:33:52 but where is that -bb file, im targeting beagleboard Jul 12 22:34:31 i did a locate, but i found several meta-toolchain.bb Jul 12 22:34:46 opie, qte, efl, gnuradio Jul 12 22:34:49 and a bare Jul 12 22:34:59 setup-scripts/sources/openembedded-core/meta/recipes-core/meta/meta-toolchain.bb Jul 12 22:36:03 which is mostly empty Jul 12 22:36:07 how to know which one? Jul 12 22:36:20 http://pastebin.com/jEgiYB8G **** ENDING LOGGING AT Fri Jul 13 02:59:58 2012