**** BEGIN LOGGING AT Thu Feb 25 02:59:59 2016 Feb 25 07:53:34 Hi i'm getting /bin/sh: can't access tty; job control turned off after I boot the device using yocto image ...can anyone help? Feb 25 07:55:24 tknycc: 'the yocto image' is far too little information, unfortunately. which board, which image? something default, or did you add/remove something? did it work already, or never so far? Feb 25 07:57:31 I've added networkmanager and modemmanager to my image but they don't seem to autostart with the system, are there any init scripts for them? Feb 25 07:59:15 LetoThe2nd:never so far board is sabresd ...yocto daisy branch ful-multimedia-image and the complete log is pastebin.com/AHBrqfU9 Feb 25 08:00:22 tknycc: full-multimedia-image is not something that is in standard poky, so it must be something you added. Feb 25 08:00:45 even core-image-minimal has the same problem Feb 25 08:01:56 tknycc: basically i'd guess that something in your init system settings or kernel config gues wrong. Feb 25 08:02:57 tknycc: is there some known good starting point that you have? Feb 25 08:03:45 is that something tpardon? Feb 25 08:04:30 sorry, is that something to do with uboot params ? Feb 25 08:04:49 tknycc: maybe. depends on what you hand the kernel as bootparams Feb 25 08:05:31 my advice would be - reset evertyhing to the starting point, then repeat your steps to find where things break Feb 25 08:06:25 i've just cloned the community bsp and given bitbake core-image-minimal Feb 25 08:06:44 I did not add anything extra Feb 25 08:07:53 well i don't have a sabresd so i can't comment on that particular one. i just know that i tested the community layer against a wandboard and it worked fine Feb 25 08:08:34 ok Feb 25 08:09:20 maybe you accidentially did not carry out one of the steps properly. like i said - my suggestion is to start completely clean from scratch again. Feb 25 08:10:12 ok ...let me try that Feb 25 08:13:12 can you guess something with uboot params http://pastebin.com/q54g3Byt Feb 25 08:13:24 *env Feb 25 08:14:10 i won't dig through those scripts, sorry. little value in it, as i'm no sabresd user as i told you. and such things or often board specific Feb 25 08:14:25 tknycc: if you tinkered with the uboot env, reset it too. Feb 25 08:14:50 ok. Feb 25 12:30:12 I have a custom kernel where I'd like to build a few modules without including them as dependencies of the kernel-modules package. Feb 25 12:30:59 I could have sworn I knew a straight forward way of achieving this, but I can't for the life of me remember what I did the last time. Any suggestions? Feb 25 12:32:58 shush YoctoAutoBuilder Feb 25 13:27:52 hi, iam having a hard time compiling extensions of ruby gems in my build. The gems compiling itself works, but i get a QA Issue: Architecture did not match error. Feb 25 13:27:52 The compiled extensions also want to be installed in "extensions/x86_64-linux/" instead of "extensions/arm-linux-gnueabi. Feb 25 13:27:52 Is there anything i can do to let ruby compile the extensions for the right arch? Feb 25 13:29:41 you'll need to tell it to use the right compiler Feb 25 13:29:52 i suspect this is something that has to be done gem by gem Feb 25 13:37:00 hi all Feb 25 13:37:29 one problem i am facing while adding package as part of SDK Feb 25 13:37:41 say my package name is X, and ihave created x.bb Feb 25 13:38:07 after bitbake, it generates RPM like libx, libx-dev, libx-dbg Feb 25 13:38:27 what should I give in meta-toolchain.bb to add this as part of SDK ?? Feb 25 13:38:30 any hint ?? Feb 25 13:38:56 ruby somehow generates the makefiles that are used with a script extmk.rb and there is also a common.mk which has impact on the used compiler flags. in the ruby recipe (ruby.inc) there is a sed script that replaces compiler flags. i think i need to change something there Feb 25 13:39:14 everytime, i met with the error Feb 25 13:39:27 ERROR: nativesdk-X not found in the base feeds (x86_64-nativesdk noarch any all). Feb 25 13:40:26 because the generated RPM name is nativesdk-libX-*.rpm Feb 25 13:40:53 i have added nativesdk-X into meta-toolchain to make it part of SDK Feb 25 13:41:19 anyone came across like this scenario ?? Feb 25 13:49:09 use libX instead of X Feb 25 13:49:15 as the packages are not called X Feb 25 13:49:17 they're called libX Feb 25 13:49:38 remember recipes != packages Feb 25 14:02:33 any insight on my issue? (from ~90mins ago) Feb 25 14:04:40 hi Feb 25 14:04:50 adding nativesdk-libX doesnot help Feb 25 14:04:54 getting the same error Feb 25 15:38:05 Anyone know if there are any plans to support the wic image fstype with live/hddimg? Feb 25 15:44:15 is it save to build for different machines in the same build directory? Feb 25 15:44:58 yep Feb 25 15:46:09 thanks :) Feb 25 16:10:26 How do I recreate the package index? Feb 25 16:10:44 I built a new package, and would like install it on my target using smart. Feb 25 16:11:15 Apparently the package index of my feed needs to be manually recreated. Feb 25 16:13:12 bitbake package-index Feb 25 16:17:32 simple as that. Perfect! Thanks. Feb 25 16:35:37 ok something weird and dissapointing happened Feb 25 16:35:51 I paused the bitbake last night and reopened it today Feb 25 16:36:04 I had about 2500 packets compiled Feb 25 16:36:13 now it started back from 1200 Feb 25 16:36:35 this is frustrating because the same hdd space is occupied and I'm compiling under virtualbox... Feb 25 16:36:55 and I'm at 1200 now with 11 gb remaining out of 60 gb Feb 25 16:37:41 I patched the autotools.bbclass because it wasn't compiling openjdk-8-jre Feb 25 16:37:47 the whole bitbake process is very broken and not newbie friendly Feb 25 16:37:59 and after that this happened Feb 25 16:38:10 :( Feb 25 16:40:07 sixsmith: out of interest, why are you building in virtualbox? Feb 25 16:41:45 I don't have the means to otherwise Feb 25 16:42:32 will it be ok if I'm hitting the hdd limit to copy the folder somehow to anotehr machine? ( because increasing the hdd space of a fixed virtual drive image seems impossible) Feb 25 16:42:44 and then start the bitbake process there? Feb 25 16:43:37 sixsmith: the task list it shows is out of the total tasks to be run this time Feb 25 16:43:40 not the total ever run Feb 25 16:43:50 so yes, itll show a smaller amount, since it's picking up where it left off Feb 25 16:43:54 you're just misunderstanding what it's telling you Feb 25 16:44:28 well yesterday it said somerhin like "2500 out of 5000" Feb 25 16:44:30 now it says Feb 25 16:44:41 "1300 out of 5000" Feb 25 17:03:38 Any wic experts around? Feb 25 17:04:45 kergoth: edbart if he's around Feb 25 17:04:53 its 7pm for him so maybe not Feb 25 17:08:22 aha, looks like it's just a bug. wic prepares swap partitions, but then doesn't write that partition to the disk Feb 25 17:08:35 so the swap partition ends up not having had mkswap run on it in the final image Feb 25 17:08:48 is there anything write? Feb 25 17:08:50 assemble() only operates on partitions with a source, even though partitions without a source are still lprepared Feb 25 17:09:12 the parted bits are run, but no data is written to that partition on diks Feb 25 17:09:14 * kergoth digs Feb 25 17:09:28 my point is does there need to be anything written (i.e. mkswap)? Feb 25 17:09:39 I know mkswap writes out a few bytes of an ID string.. but is it really needed? Feb 25 17:10:01 with sparse files, the 'hole' left should just be all zero's anyway.. Feb 25 17:10:27 without it, swapon fails Feb 25 17:10:43 ahh ok Feb 25 17:10:49 so the magic is needed Feb 25 17:10:53 which, depending on the configuration, can cause failures/errors. i.e. "Unable to find swap-space signature" Feb 25 17:10:56 seems so Feb 25 17:11:15 also, what's the point in constructing a partition image on disk and then throwing it away? either don't create the empty partitions / swapon partitions at all, or use them :P Feb 25 17:12:38 and it's not just swap here. i.e. an empty squashfs partition is created, running mksquashfs, and then thrown away, so that partition won't be mountable on target Feb 25 17:12:43 afaict anyway Feb 25 17:13:29 * kergoth tries to fix Feb 25 17:14:02 ah, i see Feb 25 17:14:13 i was incorrect, it' sjust swap that's bugged Feb 25 17:14:16 * kergoth tests fix Feb 25 17:14:57 one-liner, figures :) Feb 25 17:32:24 damnit rpath errors in db Feb 25 17:32:39 was it Crofton|work or kergoth who was talking about those Feb 25 17:32:47 me Feb 25 17:33:12 Hmm, wonder what happened to the patches to add ovmf to oe-core Feb 25 17:33:16 I expect that explains the patch you asked about :) Feb 25 17:33:18 would be nice for testing efi images in qemu Feb 25 17:33:33 Crofton|work: so what does hardcode_into_libs do? Feb 25 17:33:47 you are lucky I recall what I did :) Feb 25 17:34:01 I suspect you will find it contrl rpath flags though Feb 25 17:37:07 ah, i see, found the bug in bugzilla Feb 25 18:22:34 rburton: I haven't personally debugged code that forks in gdb, but I seem to recall that 7.10 has code for doing that? Feb 25 18:31:20 morning all Feb 25 18:31:33 7.10 has most of the code to do it.. ;) Feb 25 18:31:46 there are a few bugs apparently.. 7.11 or 7.10.2 or soemthing like that fixes the ones I know of Feb 25 18:32:03 (most bugs BTW are in cross-debugging over a fork.. local on-target debug over fork should just work) Feb 25 18:32:20 I do think dbeugging over a fork though requires a switch to be enabled in GDB, the default is to ignore fork Feb 25 18:33:43 huh, i didn't know it supported that. cool Feb 25 18:34:28 Hmm, I can't decide if I think https://bugzilla.yoctoproject.org/show_bug.cgi?id=9171 is worth pursuing or not. leaning toward not, just as the bang for the buck is low Feb 25 18:34:30 Bug 9171: normal, Undecided, ---, ross.burton, NEW , Consider limiting the scope of depchains to runtime deps Feb 25 18:35:14 * kergoth ponders Feb 25 18:35:22 debug over fork (even on the target) is 'relatively new'.. within the past couple of years.. Feb 25 18:35:32 the debug over fork (remote) is -very- new and within the last couple months Feb 25 18:35:50 ah, nice Feb 25 18:40:56 and now... back to staring at pseudo's startup code very carefully and with a great deal of distrust. Feb 25 18:45:27 Irgendwie nervt es langsam immer mehr dass ich auf meinem grossen starken Rechner daheim nur OS X habe anstatt Linux. Vielleicht sollte ich dualboot versuchen Feb 25 18:45:39 ich brauch OS X wirklich nur fuer Krieeeg Feb 25 18:45:50 oh wrong channel Feb 25 18:45:52 sorry! Feb 25 19:00:16 Jeena: interesting nonetheless ;-) Feb 25 19:00:26 hehe Feb 25 19:38:08 <_william_> hi all Feb 25 21:21:17 so we diagnosed a couple of oddities with the qemu hang... Feb 25 22:03:26 1) hardened x86 VM "just worked" and somehow qemu-native qemu-arm binary ended up with a proper PT_PAX header and was marked ex Feb 25 22:03:50 *no host side qemu installed, just libsdl Feb 25 22:05:06 2) hardened x86_64 machine (not a vm) had the do_rootfs problem, and somehow qemu-native qemu-arm binary ended up *without* a proper PT_PAX header Feb 25 22:09:22 host side qemu didn't work either until we marked it exm Feb 25 22:12:05 so it looks like there's an issue with (gentoo host) qemu package on hardened (might need an extra flag set on install) but i have no idea how/why the qemu recipe built differently inside bitbake Feb 25 22:12:24 "Things that make you go Hmmm..." Feb 25 22:13:34 But... non-hardened gentoo x86/_64 works Fine for a few years Feb 25 22:13:53 how can we get rid of the Big Fat Warning? Feb 25 22:14:58 do i need to be the Official Supporter? (no jock-strap jokes please Crofton thnx) Feb 25 22:15:42 Official supporters get the key to working qemu Feb 25 22:16:56 my guess is since the PaX options are different between x86/x86_64 that might be it... Feb 25 22:19:19 x86_64 multili-strict, check... straight x86 and x86_64, check and check Feb 25 22:19:41 VMs and real hardware, check and double-check Feb 25 22:33:10 fray: around? Feb 25 22:34:01 ya Feb 25 22:34:59 fray: rpm5 just blew up, i just replied on the list Feb 25 22:35:39 non-intel target right? Feb 25 22:35:42 I just founda rpbolem Feb 25 22:36:28 intersting, which target is that (BSP)? Feb 25 22:36:44 that is the same thing I'm working on, but a different problem.. I'm wondering if it's the same issue or somethinge lse Feb 25 22:37:05 (SSE2 usage was enabled by default in this module, and that obviously doesn't work on targets w/o SSE) Feb 25 22:42:33 yeah this was qemux86 Feb 25 22:42:53 luckily, i usually use corei7-64 :) Feb 25 22:43:22 Hmm.. I didn't hit it there.. maybe I was using x86-64, I'll re-check when my aarch64 build completes Feb 25 22:43:37 sounds like the same thing though, so that's okay Feb 25 22:47:56 I'm building qemux86 right now Feb 26 00:01:54 Today's accomplishment: Went through the pseudo startup process repeatedly trying to figure out where it may or may not have race conditions, and I think maybe made a plan on how to make a more robust startup process that would be better at indicating errors through an exit code. Feb 26 00:02:11 I am pleased, because I didn't even have to duct tape my hands away from the keyboard to keep from typing before I understood the problem. I think. Feb 26 02:08:33 * nerdboy hands seebs a muffin Feb 26 02:08:50 doesn't need duct tape... **** ENDING LOGGING AT Fri Feb 26 02:59:58 2016