**** BEGIN LOGGING AT Fri Apr 10 02:59:59 2015 Apr 10 06:55:45 I am building some recipes extra but not included in image Apr 10 06:56:41 But those packages are not updated in packages file at tmp/deploy/ipk/../Packages Apr 10 06:57:08 Any way to update the package information Apr 10 06:57:41 vickylinuxer: A bitback package-index should do it. Apr 10 07:02:36 Thanks alot Apr 10 07:02:36 It works Apr 10 07:27:50 good morning Apr 10 07:36:40 good morning Apr 10 07:36:47 anyone knows how to configure a specific tmpdir for a builder in yocto-autobuilder ? Apr 10 07:48:12 Is there a "easier" yocto alternative, where I don't need to compile anything? Something like docker, but where I get a image. Apr 10 07:48:24 Hey - anyone got a suggestion for an interesting shell script oddity? Ive got an md5 hash that looks a bit like "$1$2qyyasiDFa" etc... (i.e. not vars but a string) - I pass this to a script e.g. script '$1$2qyyasiDFa' Apr 10 07:49:17 but of course internally to pass on to another util in the script e.g. util "$HASH" it gets parsed as if shell var. If single quote passing, the variable doesn't get resolved. Catch-22 ... Apr 10 07:49:42 wCPO: That would be using pre-compiled binaries surely? Two different things! :-D Apr 10 07:51:13 wCPO: a lot of people provide binary images for a lot of boards - but usually thats what comes *out* of something like poky Apr 10 07:51:33 wCPO: its just another stage of deployment. Apr 10 07:51:46 pev: yocto seems a little bit too complex for my use case. I need it for a digital signage box. So I need bootsplash, X and firefox basically. It for a x86 box. Apr 10 07:52:14 wCPO: if you don't need it to be reproductible, install $STANDARDDISTRIBUTION and beat it into shape. Apr 10 07:53:17 LetoThe2nd: That is what we is doing now. Apr 10 07:55:15 wCPO: One of the knoppix type variants would be plenty maybe? Apr 10 07:55:22 wCPO: most distros also offer one way or another of automating the install, so just utilize that and be fine. its not really an 'easier yocto alternative', its basically a different use case. if it fits your needs, just use it. Apr 10 07:59:31 Anybody here who actually succeeded in installing Yocto to the BeagleBone Black? Apr 10 08:00:11 cart_man: yes, but its been so long since that this answer is basically useless to you. so whats your actual question/problem? Apr 10 08:03:01 Well im not quite sure what I am doing wrong but it wont boot...not even the little blue LEDs are flashing and my Armstrong is wiped off so that wont boot obviously... I partitioned my SD card into 2 partirions. One named BOOT(Fat32) and the other named ROOT(ext4) . I copied MLO and u-boot.img to the BOOT partition and unzipped the OS to the ROOT partition but it doen not seem to even want to boot even with pressing S2 on power up Apr 10 08:03:31 LetoThe2nd, Why wont my leds not even flah? Apr 10 08:04:20 LetoThe2nd, I also dont have a usb to serial converter so I wont be able to check the serial coms : / Apr 10 08:04:35 cart_man: i remember that back in the day i did that, i grabbed a known good sd card image. dd'ed it to the card, and then replaced the partition contents without changing the partitions or formatting them Apr 10 08:04:36 But that does not matter anyway since my LEDs dont even come on Apr 10 08:05:09 LetoThe2nd, Ok but which image because I cant find a .hddimg OR a .img in the Yocto prebuilt repo Apr 10 08:05:10 cart_man: maybe that can help you, but without a serial connection you'Re basically doomed, because you can'T see which boot stage actually fails (MLO, uboot, kernel) Apr 10 08:05:27 cart_man: a known good image from beagleboard.org Apr 10 08:05:36 dd that to the card, then start replacing. Apr 10 08:05:44 Ohh ok I see...good idea Apr 10 08:05:47 thanks man Apr 10 08:06:52 morning all Apr 10 08:10:25 bluelightning,Hi Apr 10 08:20:50 LetoThe2nd: Think I will try with debootstrap. Do you know any other good tools? Apr 10 08:21:23 wCPO: in what way is debootstrap related to what you asked earlier? Apr 10 08:22:49 LetoThe2nd: I can create a script, which "build" a minimalist version of debian, with all the package I need. Apr 10 08:24:38 wCPO: ... which is usually what you do *AFTER* you have a known good, booting environment, right? not if you're not even able to startup your stuff. Apr 10 08:25:02 LetoThe2nd: I do have a "known good, booting environmen" Apr 10 08:25:30 wCPO: plus, debootstrap probably doesn't play too well with custom kernel/module stuff as we have in the BBB. better read up on RCNs scripts that he provides Apr 10 08:26:29 LetoThe2nd: It is for a x86 Gigabyte Brix. I don't need any custom kernel/module. Apr 10 08:27:10 wCPO: ah sorry i messed you up with the other asker. well then go ahead. Apr 10 08:30:28 hmm, anyone know this by heart? I'm adding a new group, but I want to place root into it. How can I do that? Apr 10 08:31:13 gpasswd -a USER GROUP Apr 10 08:31:32 hmm I was hopeing this somehow integrated with GROUPADD_PARAM Apr 10 08:31:42 ah, you mean in a build process? Apr 10 08:31:45 yeah Apr 10 08:31:56 no idea then, but adding root to a group is rather pointless anyways ;) Apr 10 08:32:03 not really Apr 10 08:32:32 * rink_ is creating this nice group 'sshusers' :-) Apr 10 08:33:02 (yes I know I can use AllowRootLogin no and stuff ,but this is easier to tune) Apr 10 08:33:23 (goes, hides silent in his closet and weeps.) Apr 10 08:34:00 :-) if you can instruct my co-workers that root isn't the only account on a linux system, be my guest Apr 10 08:34:24 * rink_ pretty happy our applications do not run as root and that there is a privileges daemon which takes care of all root-thingies Apr 10 08:34:31 they're not my coworkers, so they're not my problem. my coworkers know how to use user accounts :) Apr 10 08:34:52 Pretty sure your coworkers don't come from Windows CE : Apr 10 08:35:00 where there is literally _0_ concept of security Apr 10 08:36:03 still i *personally* (so take it with a pinch of salt) think that giving in to their lazyness will cause more pain in the end than saying "NO ROOT ACCESS!" once now and forever and properly educating them Apr 10 08:36:41 oh, it's not enabled by default. Apr 10 08:37:15 * rink_ notes SSH isn't enabled by default either - and if you enable it, you must instruct the system to bind it to something else than 127.0.01 Apr 10 08:37:34 But let's just say I've tried that fight over and over... people are evil Apr 10 08:37:56 i know. but sorry, still can't help you there. never needed that. Apr 10 08:38:16 i know you can trick them, make an account call root which has uid != 0 :D Apr 10 08:39:14 LetoThe2nd: well fwiw, it seems you can use GROUPMEMS_PARAMS Apr 10 08:41:24 rink_: thanks for telling! *notes* Apr 10 08:41:26 * rink_ wishes there were more ppl who took embedded system security seriously Apr 10 08:42:25 embedded system security is easy, remove all the external interfaces no? Apr 10 08:42:51 nrossi: exactly Apr 10 08:43:09 and, despite all my wishes, our product _still_ does DHCP by default Apr 10 08:43:16 even though there is zero need for that Apr 10 08:43:28 but 'windows ce does it, so people expect it' Apr 10 08:45:47 LetoThe2nd: ugh, it seems GROUPMEMS_PARAMS is quite recent - my yocto doesn't have it Apr 10 08:46:13 * rink_ decides to add root group to sshd allowgroups - I'll do the crying when I get home Apr 10 09:19:20 good morning Apr 10 09:24:26 bluelightning: hmm, it seems that pseudo does not work after the migration. Apr 10 09:24:55 how have you determined that? Apr 10 09:26:59 bluelightning: I have determined that based on this bitbake output, https://paste.kde.org/petneznpc Apr 10 09:36:14 "If the build fails, install 32-bit developer packages." Apr 10 09:36:49 what packages exactly? why was it not needed before? Apr 10 09:37:28 change in behaviour after a much earlier pseudo upgrade I would guess Apr 10 09:38:01 git mv poky-dylan-9.0.1 src Apr 10 09:38:02 fatal: bad source, source=poky-dylan-9.0.1/meta/classes/archive-configured-source.bbclass, destination=src/meta/classes/archive-configured-source.bbclass Apr 10 09:38:06 it is another interesting thing. Apr 10 09:39:18 IIRC, this would be caused by your system having some 32-bit packages installed (glibc?), enough to trigger pseudo wanting to build its 32-bit support - but not the headers, so it fails Apr 10 09:39:36 I really don't know much about the archiver class(es) Apr 10 09:48:16 bluelightning: ok, thanks; I cannot seem to find a pseudo 32 package on my Debian. Apr 10 09:48:26 is the package mentioned in the dependency section? Apr 10 09:49:37 http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#migration-1.5-host-dependency-changes Apr 10 09:49:44 probably there ought to be a section like that for 1.6, too Apr 10 09:49:54 it does not affect everyone, I understand, but it affects some. Apr 10 09:50:20 you wouldn't be looking for pseudo, look for glibc and other core libraries Apr 10 09:50:50 ok, well, then "If the build fails, install 32-bit developer packages." was confusing to me. Apr 10 09:51:17 since it was a pseudo issue, I thought I would need to install a 32 bit version of pseudo. Apr 10 09:51:55 by all means file a bug against pseudo for an error message improvement Apr 10 09:56:00 hmm, libc6-i386 is already installed. Apr 10 09:56:07 apparently, that is not enough. Apr 10 10:01:31 hmmm can anyone help me getting scponly to build with yocto ? Apr 10 10:01:35 * rink_ is getting configure errors Apr 10 10:01:55 /home/rink/fsl-community-bsp/build/tmp/work/cortexa9hf-vfp-neon-poky-linux-gnueabi/scponly/4.8-r1/scponly-4.8/configure: line 4007: syntax error near unexpected token `PROG_SFTP_SERVER,' Apr 10 10:02:19 that line in question is SCPONLY_PATH_PROG_DEFINE(PROG_SFTP_SERVER, sftp-server, ) Apr 10 10:02:19 lpapp: and the headers? Apr 10 10:02:37 i.e. the corresponding dev/devel package(s)? Apr 10 10:03:49 * rink_ wonders if it processes the configure.in file correctly; I'm just inherit-ing autotools Apr 10 10:04:50 * rink_ doesn't see the macro being defined anywhere Apr 10 10:08:45 unfortunately some software packages ship broken scripts that can't be autoreconf'd Apr 10 10:08:56 ugh Apr 10 10:09:00 * rink_ gives up, tries rssh instead Apr 10 10:09:05 it's usually worse for older packages Apr 10 10:09:16 well the latest stable version of scponly is from 2008 Apr 10 10:10:19 usually I look around for patches shipped in distros (debian, fedora) and if that fails, override do_configure to oe_runconf so that the upstream configure script is used (but that usually has its own pitfalls) Apr 10 10:10:39 hmm let's see Apr 10 10:12:38 * rink_ wonders how to find patches in fedora repo Apr 10 10:12:52 it seems that what debian did is just replace the entire configure file Apr 10 10:19:13 bluelightning: | /home/rink/fsl-community-bsp/build/tmp/sysroots/x86_64-linux/usr/libexec/cortexa9hf-vfp-neon-poky-linux-gnueabi/gcc/arm-poky-linux-gnueabi/4.8.2/ld: rssh_chroot_helper.o: relocation R_ARM_MOVW_ABS_NC against `username' can not be used when making a shared object; recompile with -fPIC Apr 10 10:19:19 any idea what that means ? Apr 10 10:21:37 rink_: sure, rssh_chroot_helper needs to compiled with -fPIC Apr 10 10:21:51 so make sure its configure scripts does set it. Apr 10 10:22:19 well, I am unsure _why_ it needs -fPIC Apr 10 10:22:23 as neither the config scripts sets it Apr 10 10:22:35 and appearantly it is not needed on other platforms/archs Apr 10 10:24:09 rink_: i've seen the same when building a shared lib linking against qt Apr 10 10:24:45 and maybe its not needed on other builds if they link static only Apr 10 10:30:04 * rink_ notes it seems to try to detect to build static Apr 10 10:30:09 but even bluntly adding -fPIC does not help Apr 10 10:30:11 * rink_ sighs Apr 10 10:30:41 bluelightning: thanks, that was it! Apr 10 10:30:48 bluelightning: seems the kernel logic also changed? It fails in here. Apr 10 10:30:56 which logic? Apr 10 10:36:01 bluelightning: https://paste.kde.org/pk40grntc Apr 10 10:38:13 lpapp: is that revision actually in master of the specified repo, or another branch? Apr 10 10:39:35 bluelightning: it is the hash of the 3.2.1 tag Apr 10 10:40:04 again... is that on master? or another branch? or no branch? Apr 10 10:40:35 no idea Apr 10 10:41:00 well, you'll probably need to find out Apr 10 10:41:05 but since KBRANCH ?= "master" says master, I assume so. Apr 10 10:41:11 http://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#migration-1.6-matching-branch-requirement-for-git-fetching Apr 10 10:41:46 "If you do not specify a branch, BitBake looks in the default "master" branch." Apr 10 10:41:58 master is not the default in many repositories though Apr 10 10:42:10 perhaps it should check HEAD instead? Apr 10 10:42:15 where it points to? Apr 10 10:49:45 bluelightning: yeah, the only problem is that it takes me an hour to clone that repository. Apr 10 10:50:16 surely you have a clone of it already somewhere... Apr 10 10:50:31 not really, no. Apr 10 10:50:38 and none of your colleagues? Apr 10 10:50:48 I am alone in here with this :-) Apr 10 10:51:03 well, I certainly can't tell you which branch it's on... Apr 10 10:51:07 but let me see if sstate-cache can get me to somewhere... Apr 10 10:51:31 you'd be much more likely to find it in git2/ under a previous downloads directory Apr 10 10:52:41 yeah, that is what I need sstate-cache for Apr 10 10:52:47 to get to that point. Apr 10 10:52:55 or well, the download thingamajig Apr 10 10:53:06 I certainly have no git2. Apr 10 10:59:58 the sstate-cache does not contain downloads Apr 10 11:00:17 do_fetch is not an sstate-acceleratable task Apr 10 11:06:20 ok, I cloned the linux repository finally. Apr 10 11:06:44 it is good that it did not take an hour this time, still it was some. Apr 10 11:09:16 hmm, I need to learn git Apr 10 11:09:26 I do not know how to get the "branch" for a "tag". Apr 10 11:10:17 perhaps: git branch -r --contains tagname Apr 10 11:10:28 yes, I figured out Apr 10 11:10:32 from this: http://stackoverflow.com/questions/15806448/git-how-to-find-out-on-which-branch-tag-is Apr 10 11:10:49 although it just resally returns the output of git branch Apr 10 11:11:35 git branch -r --contains v3.2.1 origin/linux-3.2.y Apr 10 11:11:42 seems that it is not master? Apr 10 11:23:36 bluelightning: shall I change KBRANCH to that? Apr 10 11:42:13 bluelightning: that did the trick; oh the license file checksum has also changed? Apr 10 11:46:32 5.5.10 Apr 10 12:14:03 * rink_ shrugs ... avoid autofools problems by just adding a patch that supplies a config.h and Makefile feels... wrong Apr 10 13:38:35 anyone seen issue where we strip only one from many hardlinks for the same file, so that in the end we have image directory with 5 hardlinks for the same file and packages-split has 4 hardlinks to old unstripped version and one stripped binary? Apr 10 13:39:56 well visible with mke2fs+mkfs.ext* or e2fsck+fsck.ext* files from e2fsprogs Apr 10 13:40:03 RP: ^ rings a bell? Apr 10 13:43:49 nice e2fsprogs configure has option to use symlinks Apr 10 13:51:08 JaMa: there is an open bug for it Apr 10 13:51:31 JaMa: https://bugzilla.yoctoproject.org/show_bug.cgi?id=7586 Apr 10 13:51:32 Bug 7586: normal, Medium+, 1.9 M1, eduard.bartosh, NEW , Hardlinks are not preserved and files are not stripped Apr 10 14:09:24 JaMa: I've just started to work on that bug. Do you have any thoughts/ideas on how to fix it? Apr 10 14:11:47 ed_: I've just sent work around for e2fsprogs Apr 10 14:12:01 because that was blocking me in dizzy to fit our image to recover partition Apr 10 14:15:22 JaMa: can you comment in bugzilla about your workaround? Is it about using symlinks instead of hardlinks by e2fsprogs? Apr 10 14:37:11 ed_: I did Apr 10 15:57:51 bluelightning: seems that our application does not build with daisy Apr 10 15:57:58 it built with dylan, however. Apr 10 15:58:23 https://paste.kde.org/pxmcg5knh Apr 10 15:59:01 hmm, I'm not sure what that error means Apr 10 16:00:08 me neither :/ Apr 10 16:02:05 khem`: don't suppose you've seen that one before? ^ Apr 10 17:43:22 bluelightning: the same recipe worked with dylan. Apr 10 17:43:32 so daisy must have changed the situation somehow. Apr 10 17:45:00 only in that it's a different version of gcc, with different bugs Apr 10 17:45:05 bbl Apr 10 19:35:59 HI .. quick question I am trying to build a core-image-minimal with mpich . When I build mpich byitself is working fine with a warning ( no package provides /usr/bin/bash) Apr 10 19:36:06 but latter that warning became a bug Apr 10 19:36:18 when I try to build the core-image-minimal Apr 10 19:36:35 does anybody has been with this problem :) Apr 10 19:38:49 sounds like you have a shell script that needs bash, but your recipe doesn't declare that runtime dependency. Apr 10 19:38:58 either fix the script to not need bash, or fix the rdepends to indicate that it needs it Apr 10 19:39:45 thanks kergoth .. I will test the rdepends Apr 10 19:46:12 kergoth: actually the receipe has the RDEPENDS_${PN} += "bash perl libxml2" Apr 10 19:46:31 ah, i didn't notice the /usr/bin/bash Apr 10 19:46:38 bash isn't installed under that path Apr 10 19:46:40 it's in /bin/bash Apr 10 19:46:45 so find the script pointing at the wrong bash path :P Apr 10 19:46:51 specifically, grep for #!/usr/bin/bash Apr 10 19:49:38 kergoth: ok .. thanks Apr 10 20:00:13 kergoth: the grep inside mpich shows that they are using #! /bin/bash. I realize the full log say more ingo when I build with bitbake Apr 10 20:00:21 ERROR: Unable to install packages. Command '/home/vrodri3/poky/build/tmp/sysroots/x86_64-linux/usr/bin/smart --quiet --data-dir=/home/vrodri3/poky/build/tmp/work/qemux86-poky-linux/core-image-minimal/1.0-r0/rootfs/var/lib/smart install -y mpich@i586 gcc@i586 run-postinsts@all packagegroup-core-boot@qemux86 bash@i586' returned 1: Apr 10 20:00:21 error: Can't install mpich-3.1.1-r0@i586: no package provides /usr/bin/bash Apr 10 20:00:37 did you grep the *output* of the build? Apr 10 20:00:49 bitbake the recipe, then examine ${WORKDIR}/image/ Apr 10 20:01:54 hum ok .. I was greping the src code of mpich Apr 10 20:01:57 my mistake Apr 10 20:01:59 tahnsk Apr 10 20:02:01 *thanks Apr 10 20:05:11 victor_rodriguez: were you able to figure it out? Apr 10 20:06:01 aehs29: not yet .. debuging Apr 10 20:06:06 aehs29: have you ? Apr 10 20:06:28 victor_rodriguez: I just read this Apr 10 20:06:37 :D Apr 10 20:15:11 aehs29: kergoth : Requires: /usr/bin/bash Apr 10 20:15:23 that is in tmp/work/i586-poky-linux/mpich/3.1.1-r0/mpich.spec Apr 10 20:15:40 victor_rodriguez: if you mess around with meta/classes/insane.bbclasss you'd get the offending file Apr 10 20:15:41 hmmm hidden symbol `pthread_atfork' Apr 10 20:16:14 victor_rodriguez: k let me check Apr 10 20:16:31 it seems this package is implementing some sort of fork ? Apr 10 20:16:45 and hihacks the glibc version Apr 10 20:16:56 maybe Apr 10 20:17:02 linux-gpib Apr 10 20:17:29 humm Apr 10 20:17:57 not sure .. I am not that expert on glib Apr 10 20:18:54 victor_rodriguez: I only see /bin/bash on mpich.spec Apr 10 20:20:22 aehs29: I think I got it Apr 10 20:20:31 on the Makefile of mpich they have this Apr 10 20:20:41 9100 BASH_SHELL = @BASH_SHELL@ Apr 10 20:20:59 in some way they got the bash you have on your build system Apr 10 20:21:11 when I type which bash I got /usr/bin/bash Apr 10 20:21:20 what do you have aehs29? Apr 10 20:30:11 sorry aehs29 I got disconnected Apr 10 20:30:28 aehs29: cna you build mpich on the core image minimal ? Apr 10 20:33:24 victor_rodriguez: I got bin/bash Apr 10 20:33:29 victor_rodriguez: its building Apr 10 20:36:08 nice Apr 10 20:36:15 if is a prblem with my ... dev machine Apr 10 20:36:21 I wil lbe more relax Apr 10 21:19:30 aehs29: :D Apr 10 21:19:43 make my day please :D Apr 10 21:55:56 * paulg_ wonders why everyone else uses /etc/default/grub/ but we (yocto) use /etc/grub.d/ instead.... Apr 11 02:14:59 Poky (Yocto Project Reference Distro) 1.6.2 imx6sxsabresd /dev/ttymxc0 Apr 11 02:15:35 imx6sxsabresd login: Apr 11 02:22:34 Hi I am trying to build custom kerenl for raspberrypi (3.18). I have set EXTERNALSRC_pn-linux-raspberrypi = <> in local.conf file located in build dir. I getting error when trying to compile kernel. Can someone help me to fix this issue ? Here is the error log: http://paste.ubuntu.com/10796839/ Apr 11 02:23:50 I am inherting externalsrc like this : INHERIT += "externalsrc" **** ENDING LOGGING AT Sat Apr 11 02:59:58 2015