**** BEGIN LOGGING AT Thu Jul 02 02:59:59 2015 Jul 02 06:24:05 hmm. "running 20 hrs, 21 mins, 39 secs " Jul 02 06:24:13 I think I may have upset the autobuilder... Jul 02 06:33:52 <_4urele_> jmleo, if you are using "INHERIT += "buildhistory"" there is a file named installed-packages.txt Jul 02 07:43:53 hey.. how to print the DISTRO_FEATURES? Jul 02 07:47:38 `bitbake -e ` will dump the environment Jul 02 07:47:43 then apply grep to taste Jul 02 07:48:24 anyone work on hardening processors? Jul 02 07:50:12 my more direct question is, can the mmu be used to securely provision memory on a processor? Jul 02 07:55:27 redengin: care to elaborate? Jul 02 07:56:22 I'm looking at powerpc architecture and it looks like you could set the zpr and tlb at will Jul 02 07:56:49 i.e. you could browse through memory without faulting Jul 02 07:56:54 usually the OS has quite some freedom in manipulating these, yes Jul 02 07:57:04 the only way I see to stop that is an ISR that restores the TLB Jul 02 07:58:03 LetoThe2nd, would such things work on other architectures as well? Jul 02 07:58:34 well are you thinking userspace or kernelspace? FWIW such instructions usually require some kind of processor-superuser mode Jul 02 07:58:47 I thought I remembered some way on x86 to hide the tlb in private space Jul 02 07:59:13 its really not my field of expertise, i can only offer common knowledge Jul 02 07:59:47 LetoThe2nd, yes, I was pretty sure that x86 enforced a processor-level register protection Jul 02 08:00:28 so if the register is protected, what is the problem? Jul 02 08:00:58 I've not delved into cpu architectures enough to know which ones actually have such modes Jul 02 08:01:18 this was back in 2000 when I was looking at it Jul 02 08:01:34 x86 has rings, and ARM has supervisor mode Jul 02 08:02:04 any idea what the powerpc would have in this regard? Jul 02 08:02:04 i'm pretty sure any reasonably advanced architecture has something comparable Jul 02 08:02:40 never been a ppc man, so sorry no Jul 02 08:02:57 from my quick googling, I'm led to believe that apple dumped ppc for that reason Jul 02 08:03:39 ppc was dumped for the lack of availability of high-performance cpus and heat dissipation Jul 02 08:04:15 well possibly those as well :) Jul 02 08:04:27 and most likely primarily for those reasons Jul 02 08:05:06 i'm rather sure that register-level protection wa nowhere near that decision. and a really short google-fu from my side suggests that on ppc its called privileged mode and that this even goes back to the 68k days Jul 02 08:06:34 LetoThe2nd, can you pass on the link that mentions a ppc privelege mode? Jul 02 08:07:13 redengin: do you really expect me now to open google again, enter "powerpc privileged mode", copy the first link and paste it here for you? Jul 02 08:07:30 nm, I'm looking at it Jul 02 08:11:26 LetoThe2nd, thanks, I found what I was looking for Jul 02 08:11:33 :-) Jul 02 08:28:25 i want to switch from sysvinit to systemd, i have a simple recipe that inherits update-rc.d and i want to modify to inherit systemd, can someone provide a simple example? Jul 02 08:42:36 hmm, someone should really add a system recipe to meta-skeleton Jul 02 09:14:58 RP: halstead: I think I've broken the AB: " triggered trigger_main-build trigger_main-build_nowait ( running 23 hrs, 20 mins, 6 secs )" Jul 02 09:35:58 joshuagl: its because nightly-lsb-ppc is still running Jul 02 09:36:09 joshuagl: https://autobuilder.yoctoproject.org/main/builders/nightly-ppc-lsb/builds/361 Jul 02 09:37:52 joshuagl: and I can't seem to ssh into fedora22 Jul 02 09:38:18 RP: right, that build has been running for >12hrs Jul 02 09:38:35 joshuagl: congrats, you do appear to have broken that AB slave Jul 02 09:38:45 joshuagl: 22 hours Jul 02 09:38:54 :-( Jul 02 09:38:58 * joshuagl feels bad Jul 02 10:11:16 <_4urele_> Hi all Jul 02 10:12:49 <_4urele_> I just generated a sdk from a custom distro and image, it seems something is missing, when i try to use this sdk with an autotools based project, automake complains about LIBTOOL and LT_INIT Jul 02 10:13:06 <_4urele_> (complains that there are missing Jul 02 10:17:44 <_4urele_> is there something to do to specify what we would like to be built within the sdk? Jul 02 10:18:57 joshuagl: erm your fido-next has been running for 24 hours Jul 02 10:19:31 rburton: yes, I already confessed that I appear to have broken the AB Jul 02 10:19:35 I'm not sure how Jul 02 10:20:23 i have a git clone thats now 12 hours old Jul 02 10:20:28 so i'll go around killing things Jul 02 10:21:01 hi folks, i'm new here but trying to help clarify how automotive-grade-linux should best organise its layers/repos... would this be a good place to ask about that, or would it be better in #oe or somewhere else? Jul 02 10:22:48 paulsherwood: here, or #oe, or one of the various mailing lists Jul 02 10:23:06 paulsherwood: you might have a better discussion on a mail unless you have some simple questions to start Jul 02 10:23:34 fwiw, the yocto compliance rules say to split your BSP layers from your recipe layers from your distro policy layers Jul 02 10:24:33 rburton: ok great, thanks. maybe my first question (i've not done much with yocto or oe) is what's the best way to think of demarcation between oe and yocto? Jul 02 10:25:16 rburton: the background to this is agl is working on a distro now, wants to get good re-use out of existing layers, including oe layers and (for example) meta-ivi Jul 02 10:25:37 paulsherwood: yocto is an umbrella project, oe-core is one of the pieces Jul 02 10:25:46 you can't download yocto Jul 02 10:25:48 i'm interested to help mimimise forking and duplication Jul 02 10:26:42 rburton: ack - so yocto is collections of layers and recipes from other places, plus some additional layers and recipes originated at yocto? Jul 02 10:26:54 s/originated/upstream/ Jul 02 10:28:13 sort of Jul 02 10:28:33 so for example, should agl take from oe layers from yocto, or direct from oe? Jul 02 10:28:46 what would the difference be? Jul 02 10:28:56 oe layers are maintained at openembedded.org Jul 02 10:29:18 or github, or git.yoctoproject.org, depending on the layer Jul 02 10:29:19 i don't know, tbh. does yocto change oe layers, or add things to them? Jul 02 10:29:20 there's no forks Jul 02 10:29:28 ok, excellent Jul 02 10:30:24 i'm assuming/hoping agl won't fork either... that woudl be dumb. but there's at least some discussion around forking wrt meta-ivi, which would not be ideal Jul 02 10:34:39 forking would be dumb, yes. co-operating with the people who are working on meta-ivi makes a lot more sense, especially as i expect there's a lot of the same names involved. Jul 02 10:37:13 joshuagl: beth fixed the ab. i've re-kicked my mut which shouldnt take long. feel free to add a fido-next to the queue if it makes sense Jul 02 10:38:27 rburton: hmm, OK - any idea what went wrong? Jul 02 10:38:40 rburton: is another fido-next likely to cause the same issue? Jul 02 10:40:11 not sure, only one way to find out ;) Jul 02 10:41:16 hehe Jul 02 10:41:44 JaMa: could your bitbake world builds have a shortlog of the difference between oe-core master and jansa/master in the mail? Jul 02 10:42:57 * joshuagl queues Jul 02 10:43:37 rburton: i believe the issue with meta-ivi is that agl may have different needs, which would lead to (say) separating some parts out, leading to (say) meta-ivi-common, meta-genivi, meta-agl or similar Jul 02 10:44:25 and iiuc the employer of the maintainers is reluctant to go that route (but i may be wrong) Jul 02 10:45:07 in any case i've raised the question of whether additional engineers could participate in maintaining meta-ivi Jul 02 10:45:49 also, i'm trying to work out what agl should want to host in its own repos, vs at yocto or oe. Jul 02 10:46:04 anyone ever tried systemd with yocto > 1.7? it boots and looks good, but i do not get a login: Jul 02 10:46:07 [ TIME ] Timed out waiting for device dev-ttymxc0.device. Jul 02 10:46:10 [DEPEND] Dependency failed for Serial Getty on ttymxc0. Jul 02 10:46:43 ericbutters, enable CONFIG_FHANDLE Jul 02 10:47:00 shoragan: okay Jul 02 10:48:01 joshuagl: "util-linux-2.25.2/text-utils/pg.c:286:3: warning: implicit declaration of function 'vidputs' [-Wimplicit-function-declaration]" <— in master!! Jul 02 10:48:13 i blame you josh Jul 02 10:48:25 rburton: yes, current log: https://bpaste.net/show/e4d9dcb51e8c Jul 02 10:54:25 rburton: !!! Jul 02 10:54:48 rburton: master is most definitely not my fault, I have ~3 recent patches in there Jul 02 10:59:34 shoragan: thanks! that solved it Jul 02 11:00:11 ericbutters, nice Jul 02 12:48:53 <_4urele_> hello Jul 02 12:49:14 <_4urele_> I have a problem with libtool Jul 02 12:49:50 <_4urele_> with yocto 1.7 while generating the sdk I had files in "sysroots/i686-castelsdk-linux/usr/share/aclocal/" Jul 02 12:50:05 <_4urele_> now this folder is almost empty Jul 02 13:01:10 <_4urele_> is it possible to add packages to sdk? Jul 02 13:41:58 Hi guys ! Jul 02 13:43:06 I'm having a small issue. I tried to compile an imx6 image using an external armhf compiler, and linux boots, but the image is missing ldconfig (and probably a few other things). Jul 02 13:43:20 Any thoughts on what could cause that ? Jul 02 13:51:53 hey guys Jul 02 13:52:20 how can I compile kerenel with one toolchain but rootfs with another one? Jul 02 13:52:46 just appnd EXTERNAL_TOOLCHAIN to the kerenel recipe? Jul 02 13:59:35 Ox4: Maybe by finding a way to set the CROSS_COMPILE variable ? Or by setting the appropriate compiler prefix in the menuconfig ? Just throwing ideas :) Jul 02 14:02:36 * _4urele_ is currious about Ox4 idea... Jul 02 14:04:45 Ox4: But, yes ... why would you want to do that ? Jul 02 14:11:31 I know it is not good approach, but the reason is in the following: Jul 02 14:12:29 I have a kernel witch boots if it is compiled only by old toolchain (gcc-4.5.1) Jul 02 14:12:55 and I have a bunch of application which should be compiled with -std=c++11 flag Jul 02 14:13:08 but 4.5.1 doesn't support such flag Jul 02 14:13:36 and if I compile the kernel with the newest one toolchain it just freezes after unpacking Jul 02 14:14:11 that's why I decided to try such approach: compile the kernel by 4.5.1 toolchain and applications by 4.9 Jul 02 14:14:41 sorry for my english btw, it is in alpha version :) Jul 02 14:15:14 Ox4: if you build your kernel with 4.9 does it hang just after the 'Starting linux' banner? Jul 02 14:26:56 tf_: yes, it does Jul 02 14:27:33 tf_: the kernel version 2.6.37 Jul 02 14:28:16 Ox4: you are missing the memset patches, in a recent kernel look for 'fix memset-related crashes' in the commit log; there are two patches, apply that to your kernel Jul 02 14:29:47 tf_: can I apply these patches for 2.6.37 kernel? Jul 02 14:31:02 tf_: btw, could you plese send me a link? Jul 02 14:49:46 hi guys, how do we set autologin for a specific user in yocto? Jul 02 14:56:36 Guys, how is it that armhf isn't supported on the i.mx6 boards ? Jul 02 16:14:06 hey, I'm trying to install an debian package with yocto, the process succeeds but when I'm trying to start the program I'm installed i get sh ./no such file or directory, Jul 02 16:14:11 archtiercture should be fine Jul 02 16:14:19 i think its more than an linking error Jul 02 16:14:20 http://pastebin.com/cBag3ZAc Jul 02 16:14:50 line 24 Jul 02 16:14:53 or 336 maybe? Jul 02 16:24:20 wrock_: you're on your own doing something like that. the odds of it working at all are slim Jul 02 16:25:17 what do you mean? Jul 02 16:25:44 installing a package from one distro into another distro is always an iffy proposition, whether in embedded or on the desktop Jul 02 16:25:51 create a proper recipe Jul 02 16:26:24 yeah thought so.. Jul 02 16:26:55 might be possible to get it to work, but it'll never be officially supported Jul 02 16:26:57 * kergoth looks at the pastebin Jul 02 16:27:43 I don't see anything wrong in that pastebin Jul 02 16:28:37 no such file or directory when you run a binary tends to suggest that its trying to find the loader and the path isn't what it expects Jul 02 16:28:53 ldd [binary] will let you verify this Jul 02 16:34:27 k, let me check Jul 02 16:40:04 rburton: not a dynamic excecutable Jul 02 16:44:27 i guess it's some linking problem, while creating/isntalling the package Jul 02 16:44:37 i'll go for the cmake recipe, to compile it seperately Jul 02 16:44:44 might be better on the long run... Jul 02 17:36:06 I'm trying to compile a base yocto image with Mono. I'm using an Ubuntu 14.04 64 bit system to compile the image. I followed the steps on the yocto documentation. But when I run bitbake I get error compiling sqlite3. The error seems to be related to multilib:lib32. I would appretiate any help. I'm very new to yocto. Jul 02 17:36:06 The error log is as follows: Jul 02 17:36:06 | i586-lib32-linux-libtool: install: warning: `libsqlite3.la' has not been installed in `/usr/lib' Jul 02 17:36:06 | i586-lib32-linux-libtool: install: /usr/bin/install -c .libs/sqlite3 /home/ghost/yocto/build-changeling-appsec-vm/tmp/work/x86-lib32-linux/lib32-sqlite3/3_3.8.10.0-r0/image/usr/bin/sqlite3 Jul 02 17:36:06 | make[1]: Leaving directory `/home/ghost/yocto/build-changeling-appsec-vm/tmp/work/x86-lib32-linux/lib32-sqlite3/3_3.8.10.0-r0/sqlite-autoconf-3081000' Jul 02 17:36:07 | /home/ghost/yocto/build-changeling-appsec-vm/tmp/work/x86-lib32-linux/lib32-sqlite3/3_3.8.10.0-r0/temp/run.do_install.24772: 120: /home/ghost/yocto/build-changeling-appsec-vm/tmp/work/x86-lib32-linux/lib32-sqlite3/3_3.8.10.0-r0/temp/run.do_install.24772: pushd: not found Jul 02 17:36:07 | WARNING: exit code 127 from a shell command. Jul 02 17:36:08 | ERROR: Function failed: do_install (log file is located at /home/ghost/yocto/build-changeling-appsec-vm/tmp/work/x86-lib32-linux/lib32-sqlite3/3_3.8.10.0-r0/temp/log.do_install.24772) Jul 02 17:36:08 ERROR: Task 548 (virtual:multilib:lib32:/home/ghost/yocto/poky/meta/recipes-support/sqlite/sqlite3_3.8.10.0.bb, do_install) failed with exit code '1' Jul 02 17:36:25 please don't spam the channel, use a pastebin Jul 02 17:37:13 sorry.. will do that Jul 02 17:38:27 http://pastebin.com/beVLGJem Jul 02 17:39:22 looks like the sqlite recipe should be fixed to not use pushd. Jul 02 17:41:00 I downloaded the recipe from yocto project Jul 02 17:43:00 yes, i realize that, and i'm saying the recipe seems to have a bug in it. you can work around it by running sudo dpkg-reconfigure dash and telling it to use bash rather than dash as /bin/sh Jul 02 17:43:11 it's not supposed to be necessary anymore, but that'd work around it until the recipe is fixed Jul 02 17:43:18 i'd suggest opening a bug in the yocto bug tracking system Jul 02 17:43:39 oh ok Jul 02 17:43:44 I will try that.. Thanks Jul 02 17:44:06 np Jul 02 18:50:04 should a package dependency on virtual/kernel force sstate repackaging in multi-machine build? Jul 02 19:32:10 SUCCESS \o/ Jul 02 19:35:18 Crofton|work: see i can make the AB complete a build Jul 02 19:37:22 rofl Jul 02 19:40:36 otavio: do you use ubi filesystems? Jul 02 20:06:35 Hi guys! I have an image for the BBB and the meta-virtualization layer setup (it boots up fine), but I'm not sure where to start to get OVS up and running on my device since I can't find the binaries or kernel modules for it. Anyone could point to me in the right direction? Jul 02 21:59:57 when disabling GPLv3 gcc5 fails now in gcc-runtime packaging Jul 02 22:00:15 home/kraj/work/rdk/build-b2120-h412-a9/tmp/work/cortexa9hf-vfp-neon-rdk-linux-gnueabi/gcc-runtime/5.1.0-r0/temp/run.buildhistory_list_pkg_files.12960: 124: /home/kraj/work/rdk/build-b2120-h412-a9/tmp/work/cortexa9hf-vfp-neon-rdk-linux-gnueabi/gcc-runtime/5.1.0-r0/temp/run.buildhistory_list_pkg_files.12960: cannot create /home/kraj/work/rdk/build-b2120-h412-a9/buildhistory/packages/cortexa9hf-vfp-neon-rdk-linux-gnueabi/gcc-runtime/gcc-runtime-dbg/files-in-gcc- Jul 02 22:00:15 runtime-dbg.txt: Directory nonexistent Jul 02 22:00:38 anyone has some quick pointers to what might have changed before I go on hunt Jul 02 22:02:35 a patch just hit the list for that Jul 02 22:03:03 it wasn't an ideal patch, however. see the thread '[OE-core] [PATCH] buildhistory.bbclass: Check output folder is present before creating files-in-$pkg.txt' Jul 02 22:03:21 i see Jul 02 22:03:55 obviously related to the merged '[OE-core] [PATCH] buildhistory.bbclass: Create a files-in-.txt containg recipe's pkg files' Jul 02 22:17:17 yeah, sometimes things go through cracks Jul 02 22:19:00 I wonder why we dont have a GPL3-less option as a gate Jul 02 22:19:10 its too commonly used option in embedded space Jul 02 22:20:16 does any know if I can make an "IF [ DISTRO_FEATURES = "systemd" ]", inside do_configure ? Jul 02 22:20:54 khem`: we do, obviously theres a difference between your no-v3 and the AB's no-v3 Jul 02 22:21:17 khem`: https://autobuilder.yoctoproject.org/main/builders/nightly-non-gpl3 Jul 02 22:21:34 hmmm Jul 02 22:21:35 INCOMPATIBLE_LICENSE = "GPLv3" Jul 02 22:21:43 that's missing A and L Jul 02 22:21:46 the docs are right, the ab is wrong Jul 02 22:22:49 (fixing) Jul 02 22:23:17 well there you go Jul 02 22:23:39 realBigfoot: you need to use some bitbake magic Jul 02 22:24:20 khem`, inside a python (), you mean ? Jul 02 22:24:45 no Jul 02 22:24:46 if [ "${@bb.utils.contains('DISTRO_FEATURES', 'systemd', 'systemd', '', d)}" = "systemd" ]; then Jul 02 22:24:57 ah Jul 02 22:24:57 do something Jul 02 22:24:58 got it Jul 02 22:25:14 perfect! Jul 02 22:25:21 thank you Jul 02 22:26:11 * khem` reverts e013fe79cdf67429ff986d8ee6af510b34b31c9d Jul 02 22:26:38 thanks kergoth Jul 02 22:39:52 np Jul 02 22:40:23 another similar option is if "${@bb.utils.contains('DISTRO_FEATURES', 'systemd', 'true', 'false', d)}"; then :; fi Jul 02 22:40:38 i should probably test gplv3-less builds more often myself **** ENDING LOGGING AT Fri Jul 03 02:59:58 2015