**** BEGIN LOGGING AT Thu Oct 03 02:59:58 2013 Oct 03 07:44:29 hi all i have used poky to create a filesystem and images for at91sam9x5ek, and my build system is ubuntu 12.04,but when i compile poky it stops at the following line http://pastebin.com/bPFc2MuU can you tell me what is that issues? Oct 03 08:30:39 Hi, I want to create a recipe under a EULA license, what should I put on LICENSE field? Oct 03 08:47:16 good morning Oct 03 08:50:51 hi all i have used poky to create a filesystem and images for at91sam9x5ek, and my build system is ubuntu 12.04,but when i compile poky it stops at the following line http://pastebin.com/bPFc2MuU can you tell me what is that issues? Oct 03 08:53:11 linu: possibly your /clancor disk filled up - do "bitbake -c cleanall linux-yocto-custom" and try again Oct 03 08:56:18 rburton, it shows 69G free space when using df -h command Oct 03 08:56:36 fair enough, download went wrong then Oct 03 08:56:45 do a cleanall to wipe the download away and try again Oct 03 08:59:15 rburton, actually i am fresher to yocto so i dont know exactly what to do,can you tell me how to clean downloads because i have little bit hesitation to clean all downloads because i have spent past 2 days on it,what should i do now Oct 03 09:01:56 rburton, yes i realized so now i need to clean linux-yocto-custom by execute bitbake -c cleanall linux-yocto-custom and do compilation again right? Oct 03 09:09:45 linu: yes Oct 03 09:10:17 the cleanall command will delete any build tress for you have that recipe, and any downloads for that recipe. Oct 03 09:11:34 morning all Oct 03 09:24:15 hi bluelightning Oct 03 09:35:06 hi RP Oct 03 09:47:09 rburton, i got this error http://pastebin.com/2G4kVQ4s while cleaning the custom kernel Oct 03 09:47:47 linu: looks like your SRC_URI is broken Oct 03 09:47:51 can you pastebin the recipe? Oct 03 09:48:52 rburton, sure Oct 03 09:54:30 rburton, please see http://pastebin.com/Wven7KtP Oct 03 09:59:22 Hi all. I'd like to compile Xenomai user-space application. Can I just add a line like this in my hellort.bb : Oct 03 09:59:25 LDFLAGS += "-L/usr/xenomai/lib -lxenomai -lpthread -lrt" Oct 03 10:00:12 i.e., will the libraries libpthread and librt be the one for target or for PC? I guess that bitbake takes care of this, right? Oct 03 10:00:14 rburton, is there any solution to solve Oct 03 10:02:40 linu: # Override SRC_URI in a bbappend file to point at a different source Oct 03 10:03:46 linu: you are supposed to customize the .bbappend and not linux-yocto-custom Oct 03 10:06:51 linu: the problem is that you used SRC_URI_append in your append, but append doesn't add a leading space. Oct 03 10:07:16 so do what ant_work said, and remember that when you use _append, you need to add whitespace yourself Oct 03 10:07:23 i.e. SRC_URI_append = " foo" Oct 03 10:08:46 oh, right.. as always the last line .... Oct 03 10:10:46 rburton, SRC_URI_append_at91sam9x5ek += " file://at91sam9x5ek/UBI_config.cfg" is it right? Oct 03 10:12:30 no Oct 03 10:12:46 well Oct 03 10:13:01 just use = instead of += Oct 03 10:15:29 rburton, SRC_URI_append_at91sam9x5ek = " file://at91sam9x5ek/UBI_config.cfg" ,some other changes? Oct 03 10:16:13 that's better Oct 03 10:16:32 and as ant_work said, don't edit linux-yocto-custom.bb directly, just set what you need in the bbappend Oct 03 10:18:54 Hi all. I have cross-compiled Xenomai, and now I am writing bb file for application that uses it's include during compiletion like this: Oct 03 10:18:56 CFLAGS += "-I/usr/xenomai/include -D_GNU_SOURCE -D_REENTRANT -D__XENO__" Oct 03 10:19:15 How to instruct Bitbake where to look for -I/usr/xenomai/include ? Oct 03 10:19:19 rburton, if you don't mind can you tell me are the changes should i make in bbappend file Oct 03 10:24:41 linu: looks like you just need to set SRC_URI and COMPATIBLE_MACHINE. never used linux-yocto-custom myself. Oct 03 10:26:59 How to instruct Bitbake where to look for -I/usr/xenomai/include for the target? Oct 03 10:27:17 hi, is there a simple way to run a custom command every time after reboot? Oct 03 10:27:28 not touching the initscripts because we do not have time for that now. Oct 03 10:27:30 whan this include is in the build/tmp dir? Oct 03 10:27:33 is there a file which is captured all the time? Oct 03 10:27:43 lpapp: the easy answer is write an init script Oct 03 10:27:46 where we can put a command like "ifconfig A.B.C.D down" Oct 03 10:28:11 rburton: /etc/init.d/network does put all the connected interfaces up Oct 03 10:28:17 and we would need to get one up only. Oct 03 10:28:22 it would be better not to bring both up Oct 03 10:28:26 but I have 30 minutes to solve it. :D Oct 03 10:28:36 so I thought a captured file would be good enough to bring it down Oct 03 10:28:39 so is there such a file? Oct 03 10:28:44 what do you mean by captured? Oct 03 10:28:48 (I know there was one on Archlinux) Oct 03 10:28:55 rburton: sh file Oct 03 10:28:59 . ./file Oct 03 10:29:03 automatically by the initscript system Oct 03 10:29:19 i.e. customization interface for end users, running after all the "builtin". Oct 03 10:29:22 just write an init script, all you need to do is inherit update-rcd Oct 03 10:29:48 there is no update-rcd Oct 03 10:30:04 ls /etc/init.d/update-rcd Oct 03 10:30:04 ls: cannot access /etc/init.d/update-rcd: No such file or directory Oct 03 10:30:24 inherit update-rc.d in the recipe Oct 03 10:30:48 I do not wanna have any recipe Oct 03 10:30:59 I want to have make a quick workaround in 20 minutes for the board Oct 03 10:31:01 oh, you're hacking on the device? just edit the network init script then? Oct 03 10:31:02 before shipping. Oct 03 10:31:11 I do not think that is safe Oct 03 10:31:22 it would be safer like on Archlinux, to have a custom script for user stuff Oct 03 10:31:26 where users can put all their stuff Oct 03 10:31:34 and it is guaranteed to be run after all the builtin. Oct 03 10:31:39 Is someone aware of problems in meta-toolchain? Oct 03 10:31:41 http://privatepaste.com/6261b87ef9 Oct 03 10:31:43 what is the correct way to add include directory for compilation to bitbake recipe? Oct 03 10:31:44 so it cannot happen I bring the interface down, and then a builting brings it back up. Oct 03 10:32:28 rburton, i changed SRC_URI and COMPATIBLE_MACHINE in append instead in linux-yocto-custom_3.6.9.bb, and masked in linux-yocto-custom_3.6.9.bb. Oct 03 10:34:42 how do we give a path to include dir in Bitbake recipe to point to another package headers? Oct 03 11:40:34 drasko: you cannot point a recipe to another recipe WORKDIR. instead if A depends on B (to build), you need to have B export its .h files into the sysroot (do_install), then you need to have DEPENDS += "B" in A .bb file. Oct 03 12:03:17 ndec: yep, drasko and I sorted out the staging issue in #oe Oct 03 12:03:38 ah... Oct 03 12:11:51 hello Oct 03 12:12:14 has anyone an idea how can i fix this? Oct 03 12:12:18 ERROR: QA Issue: nativesdk-dbus: Files/directories were installed but not shipped Oct 03 12:12:18 /run Oct 03 12:12:18 /run/dbus Oct 03 12:12:18 ERROR: QA run found fatal errors. Please consider fixing them. Oct 03 12:12:18 ERROR: Function failed: do_package_qa Oct 03 12:12:20 ERROR: Logfile of failure stored in: /build/tmp/work/x86_64-nativesdk-pokysdk-linux/nativesdk-dbus/1.6.10-r6.0/temp/log.do_package.27032 Oct 03 12:12:22 ERROR: Task 1663 (virtual:nativesdk: /meta/recipes-core/dbus/dbus_1.6.10.bb, do_package) failed with exit code '1' Oct 03 12:22:26 wesu: http://permalink.gmane.org/gmane.comp.handhelds.openembedded/58530 Oct 03 12:23:03 wesu: although that's dbus so it should already be handled... what version of the build system are you using? Oct 03 12:25:10 fsl-community-bsp-platform master branch, i think it use the master branch of yocto Oct 03 12:48:47 Krz-: did you get your QA error issue sorted out the other day btw? Oct 03 12:53:01 bluelightning, wesu, got the same issu, I tried to up all my git with git pull origin, but still. Oct 03 12:53:23 I'm building nativesdk-dbus now to see if I can reproduce it Oct 03 12:53:53 I tryied to bitbake meta-toolchain-qt Oct 03 12:54:28 i'm also Oct 03 12:55:26 right, but the apparent issue is in nativesdk-dbus Oct 03 13:04:17 ok, reproduced here Oct 03 13:26:24 TuTizz: wesu: I just sent a patch: http://patchwork.openembedded.org/patch/59217/ Oct 03 13:33:59 bluelightning, testing... Oct 03 13:36:38 bluelightning: great, thx Oct 03 13:55:08 gm all Oct 03 13:55:14 I am seeing this Oct 03 13:55:16 Computing transaction...error: Can't install nativesdk-qemu-helper-1.0-r9.0@x86_64_nativesdk: no package provides libc6 >= 2.18 Oct 03 13:55:28 when building meta-toolchain for x86_64 Oct 03 13:55:38 anyone seeing same on dora? Oct 03 14:24:09 zeddii: around ? Oct 03 14:32:43 here. at least partially, on the phone. what's up? Oct 03 14:33:18 zeddii: | /bin/sh: scripts/mod/modpost: No such file or directory Oct 03 14:33:18 | make[5]: *** [__modpost] Error 127 Oct 03 14:33:18 | make[4]: *** [modules] Error 2 Oct 03 14:33:28 I see this often Oct 03 14:33:44 interesting. just on a straight up kernel build >? Oct 03 14:33:45 it seems kernel is not staging this host tool in sysroot Oct 03 14:33:59 no its an external module Oct 03 14:34:20 it should be here poky/build/tmp-eglibc/sysroots/re-64b/usr/src/kernel/scripts/mod/modpost Oct 03 14:34:24 but doesnt show up Oct 03 14:34:42 and then module decides to build it in place and that does not work either Oct 03 14:34:50 does the module depend on virtual/kernel, or d_populate_sysroot Oct 03 14:35:00 virtual/kernel Oct 03 14:35:42 well it just inherits module Oct 03 14:35:54 and hopefully that should add all deps Oct 03 14:36:33 yah, and that should be right. but I do know that in the past, I have had to explicitly add :do_populate_sysroot for some packages that need more than the source in the staging dir. Oct 03 14:37:49 but I tried it after running do_populate_sysroot manually for kernel Oct 03 14:37:59 this file does not appear Oct 03 14:39:37 oe_runmake -C $kerneldir CC="${KERNEL_CC}" LD="${KERNEL_LD}" clean Oct 03 14:39:37 make -C $kerneldir _mrproper_scripts Oct 03 14:39:37 find $kerneldir -path $kerneldir/lib -prune -o -path $kerneldir/tools -prune -o -path $kerneldir/scripts -prune -o -name "*.[csS]" -exec rm '{}' \; Oct 03 14:39:47 yah. Oct 03 14:39:55 I am sure _mrproper_scripts is shaving it off ? Oct 03 14:40:14 I was just about to say, we remove scripts so they won't be packaged in the kernel-dev package, but in the sysroot, they should be fine. Oct 03 14:40:52 I think its not even making into sysroot Oct 03 14:41:37 right. I'm looking at the kernel.bbclass as well, and if that's done before the sysroot stage, that'll do it. we can either go restore them in the sysroot explicitly, or make sure we remove after it goes to the sysroot. Oct 03 14:42:46 sysroot_stage_all_append() { sysroot_stage_dir ${D}${KERNEL_SRC_PATH} ${SYSROOT_DESTDIR}${KERNEL_SRC_PATH} Oct 03 14:42:49 } Oct 03 14:42:50 this is 1.4 btw. Oct 03 14:43:03 heh. Oct 03 14:43:13 it's the same in master, and I had *just* landed on that same line. Oct 03 14:43:29 since I've been doing a lot of work with the kernel SDK lately, I'm way to famliar with it. Oct 03 14:43:48 OK so it seems Oct 03 14:44:05 darknighte_znc: what is znc ? Oct 03 14:44:22 zeddii: so in do_install we first clean it all and then do make -C $kerneldir _mrproper_scripts Oct 03 14:44:33 and my guess is thats not making modpost Oct 03 14:44:35 so yes, that'll have it removed, before it goes to the sysroot. the simplest thing to do might just be: cd ${SYSROOT_DESTDIR}${KERNEL_SRC_PATH}/scripts; make scripts Oct 03 14:44:52 in that stage_append. Oct 03 14:49:05 hmmm ok Oct 03 14:53:59 zeddii: that seems to have fixed it Oct 03 14:54:14 may be we should patch dora too ? Oct 03 14:54:52 that would be reasonable, as far as I'm concerned. did you want to fire off a patch, I can sign off on it as well. or I can do the patch. like I said, I've been all through this code a lot lately. Oct 03 14:55:17 if you do it all the better :) Oct 03 14:55:39 * khem is always in time shortage and today with a runny nose too Oct 03 14:55:40 I'll do it, I can test quickly and fire it off for master and dora, or are they still the same. Oct 03 14:56:04 * zeddii admits to not tracking the branching that closely. Oct 03 14:58:00 very little delta, last time i looked Oct 03 15:19:14 Hi guys. I need to build ffmpeg on Yocto. I've seen there's a recipe in openembedded git: https://github.com/openembedded/openembedded/tree/master/recipes/ffmpeg Oct 03 15:19:47 is that repository Yocto compatible or only some old cruft (seeing the commit dates)? Oct 03 15:20:15 that's oe classic, incompatible with oe-core based setups like yocto(poky) Oct 03 15:20:21 panda84kde: we now build libav rather than ffmpeg Oct 03 15:20:56 kergoth: thanks. Oct 03 15:21:32 bluelightning: thanks. In which layer is it? Oct 03 15:22:00 oh, it's in poky :) Oct 03 15:22:08 panda84kde: in dylan and earlier it was in meta-oe; in dora/master it's in OE-Core (and therefore Poky also) Oct 03 15:24:30 bluelightning: thanks a lot. Oct 03 15:24:37 np Oct 03 15:46:24 Hey, I'm having real issues with the postinst rules in my build.... It doesn't look like they get run on the target, however I can manually run "sh -x /var/lib/opkg/info/mypackage.postinst" so I know the scripts right. I'm also checking environment via "if [ x"$D" = "x" " as other scripts / the reference manual... Any suggestions as to how I can debug further? Oct 03 15:46:57 pev: double-check your test, as it maybe running (and succeeding) on the build host. Oct 03 15:47:03 if that happens, it won't run on the target Oct 03 15:47:37 I thought that's what the "exit 1" in the host case protects against? Oct 03 15:47:38 if you *need* to run on the host, check $D and exit 1 Oct 03 15:51:28 pev: master or dylan/other branch? Oct 03 15:56:43 Hm, it's an odd codebase I've been given without full history of where its coming from but it's basically 10.4 + extras AFAICT Oct 03 15:57:33 OK, double checked and I put some echos in the host part and I can see them coming out in the build correctly so that's right... Oct 03 16:02:12 Oh, I see - it sets up another postinst script "S98run-postinsts" which in turn runs "opkg-cl configure" Oct 03 16:02:18 which seems to be failing Oct 03 16:02:38 pev: oh, there was a patch recently for something like that Oct 03 16:02:52 So that's not great either as that means that on first boot it runs the postinst stuff after everything else boots which is too late...! Oct 03 16:02:55 Hm... Oct 03 16:02:58 pev: systemd or sysvinit? Oct 03 16:03:03 sysvinit Oct 03 16:03:34 and opkg-cl configure seems to fail anyway :-/ Oct 03 16:03:37 Meh... Oct 03 16:07:11 bluelightning_: weird, if I bitbake libav I get the packages "libswscale2, libswscale-dev and libswscale-staticdev" but then if I add "libswscale" or "libswscale2" in my IMAGE_INSTALL of my recipe I get: "ERROR: Nothing RPROVIDES 'libswscale2'" Oct 03 16:07:40 the other pieces "libavcodec" "libavdevice" etc work though Oct 03 16:08:11 panda84kde: this is probably a difference in the debian-renamed package name vs. the name understood by the build system Oct 03 16:08:54 bluelightning_: how can I dig that deeper? Oct 03 16:09:41 oh wait Oct 03 16:09:56 * panda84kde freezes Oct 03 16:10:31 panda84kde: so, libav breaks out its packages dynamically Oct 03 16:10:51 so we don't know for sure what packages will be produced until do_package Oct 03 16:11:12 however, the recipe does give a PACKAGES_DYNAMIC value, but libswscale is missing there; probably just need to add it Oct 03 16:11:40 PACKAGES_DYNAMIC tells the build system to expect certain packages to be provided Oct 03 16:12:21 panda84kde: btw though, do you really need to mention libraries individually? Oct 03 16:13:00 bluelightning_: do they all get installed if I put "libav"? Oct 03 16:13:59 panda84kde: they may not, however unless whatever you are using that uses libav libraries uses dlopen() to use the libs, they should be brought in via shared library dependencies automatically Oct 03 16:14:25 bluelightning_, wesu said that the patch was working on his build, it is not for mine. Oct 03 16:14:39 http://patchwork.openembedded.org/patch/59217/ Oct 03 16:15:00 I did a bitbake -c cleanall nativesdk-dbus Oct 03 16:15:06 then bitbake nativesdk-dbus Oct 03 16:15:23 do not know what to do Oct 03 16:15:32 I'm not sure how it could not work Oct 03 16:15:36 bluelightning: I see. At the moment however what I'm building is outside Yocto builsystem. I'm building an SDK image. Oct 03 16:15:51 TuTizz: you're sure the patch has been applied? Oct 03 16:16:08 panda84kde: ok, I see Oct 03 16:16:27 panda84kde: probably you just need to add swscale to PACKAGES_DYNAMIC in the libav recipe Oct 03 16:16:48 bluelightning: yeah, trying now Oct 03 16:17:31 bluelightning: seems to work. I'll send a patch after the image is generated and tested ;) Oct 03 16:17:39 panda84kde: great, thanks Oct 03 16:17:46 cat meta/recipes-core/dbus/dbus.inc | grep 'rm -rf ${D}${localstatedir}/run' Oct 03 16:17:46 rm -rf ${D}${localstatedir}/run Oct 03 16:17:46 rm -rf ${D}${localstatedir}/run Oct 03 16:18:21 in do_install_class-nativesdk() { Oct 03 16:22:25 TuTizz: do bitbake -e nativesdk-dbus | grep ^D= and look in that directory, do you have var/run under there? Oct 03 16:24:51 there is opt only Oct 03 16:25:30 er Oct 03 16:25:35 opt? Oct 03 16:26:18 ah right Oct 03 16:26:54 TuTizz: then the directories are gone, in which case how could it be complaining about them? Oct 03 16:27:10 TuTizz: the error you are getting now, is it definitely still nativesdk-dbus? Oct 03 16:27:46 first the opt's tree : http://pastebin.com/gntKBFbQ Oct 03 16:27:53 second the issue : Oct 03 16:28:53 http://pastebin.com/08Pir6cu Oct 03 16:31:31 TuTizz: what about bitbake -e nativesdk-dbus | grep "^export localstatedir=" Oct 03 16:32:47 export localstatedir="/opt/poky/1.5/sysroots/i686-pokysdk-linux/var" Oct 03 16:33:18 ls /opt/poky/1.5/sysroots/i686-pokysdk-linux/var no file or directory Oct 03 16:33:43 the rm command is deleting ${D}${localstatedir}/run Oct 03 16:33:57 which unless I've missed something should be the offending directory Oct 03 16:34:46 are you looking for a quick fix or just verifying the patch? Oct 03 16:35:35 i tried doing that yesterday manually and it wasn't working for me either :( ended up having to use INSANE_SKIP because i couldn't figure it out Oct 03 16:35:41 well I'd like to find out how the patch did not fix the problem Oct 03 16:35:51 because if not we do need a proper fix Oct 03 16:37:21 ok my mistake bluelightning Oct 03 16:37:28 you don't want my /opt Oct 03 16:37:39 you want /sources/poky/build/tmp/work/i686-nativesdk-pokysdk-linux/nativesdk-dbus/1.6.10-r6.0/image/opt/poky/1.5/sysroots/i686-pokysdk-linux Oct 03 16:37:43 i agree. I tried rm-ing /run and /run/dbus yesterday in do_install_class-nativesdk() and it kept giving me the same QA error Oct 03 16:37:51 moin Oct 03 16:38:08 ls : Oct 03 16:38:14 etc usr var Oct 03 16:38:15 danbeard: the error is confusing , the offending files are actually in /var/run Oct 03 16:38:19 he is still here Oct 03 16:38:44 TuTizz: var has run under it? Oct 03 16:38:56 yes lib and run Oct 03 16:39:11 bluelightning: aha! cool let me see what happens if I try and rm those Oct 03 16:41:09 TuTizz: maybe add this to the do_install_class-nativesdk: echo ${D}${localstatedir} Oct 03 16:41:21 TuTizz: then look at log.do_package to see what it is printing Oct 03 16:41:30 TuTizz: er, log.do_install that is Oct 03 16:41:44 ok Oct 03 16:52:31 bluelightning, can I put echo "azerty" before echo ${D}${localstatedir} for easely grep it? Oct 03 16:52:46 TuTizz: certainly Oct 03 16:53:14 this is a correct sentence? (with double quote?) Oct 03 16:54:35 quotes are optional in this case but yes Oct 03 16:54:46 ok Oct 03 16:55:06 khem: znc is an IRC bouncer. keeps my session connected even when I'm afk. Oct 03 16:55:58 hmm what does it stand for :) Oct 03 16:56:13 zero network connect or somesuch Oct 03 16:56:44 khem: AFAIK, it doesn't mean anything, but you may be right... Oct 03 16:56:46 * darknighte digs Oct 03 16:58:14 khem: nope. I don't have a clue, so your version seems just about as likely as anyone elses. Oct 03 17:02:07 ok Oct 03 17:04:10 bluelightning, I put some echo in do_install(), do_install_class-native() and do_install_class-nativesdk(), then bitbake -c clean nativesdk-dbus and bitbake nativesdk-dbus Oct 03 17:04:23 got nothing when i grep my file Oct 03 17:04:31 i686-nativesdk-pokysdk-linux/nativesdk-dbus/1.6.10-r6.0/temp> Oct 03 17:04:49 cat * | grep azerty Oct 03 17:04:50 nothing Oct 03 17:05:44 khem, good morning have you had a chance to look at bug #5133 Oct 03 17:05:45 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=5133 normal, Medium+, 1.5 M5, laurentiu.palcu, ACCEPTED , [Test Case 207] sudoku-savant project compile in sdk image Oct 03 17:07:46 TuTizz: this kind of suggests maybe you're not building the same recipe that you're editing? Oct 03 17:09:19 I already tryied to find ./meta* -name dbus* in my source repertory but he did not found any bbappend Oct 03 17:13:05 TuTizz: does this point to where you're editing dbus.inc? bitbake -e nativesdk-dbus | grep ^FILE= Oct 03 17:13:49 you right Oct 03 17:13:50 oO Oct 03 17:13:57 openembedded-core Oct 03 17:14:09 i forgot openembedded-core Oct 03 17:14:13 in my path Oct 03 17:14:55 and of course this is not a meta* folder Oct 03 17:15:03 that why I cannot found it Oct 03 17:15:25 in my case your patch had to be apply to poky/openembedded-core/meta/recipes-core/dbus/dbus.inc Oct 03 17:17:11 * TuTizz fluching Oct 03 17:17:33 is Oct 03 17:17:53 thanks a lot bluelightning Oct 03 17:18:26 TuTizz: no worries, I'm just glad we solved the mystery... :) Oct 03 17:27:14 have a good evening, bye Oct 03 17:53:51 bluelightning: sorry for late reply, had to go to meeting: ok I had the same issue yesterday as TuTizz, was modifying the wrong dbus.inc.. with patch applied to right dbus.inc, it works. mystery solved :) Oct 03 17:55:51 danbeard: ok, great :) Oct 03 17:56:02 Hm, fixed the postinst - in the function I was doing a grep that failed (checking for missing conf line to patch) and that managed to cause the function to exit prematurely causing opkg-cl configure to fail... Oct 03 17:56:19 cutting and pasting it into a standalone script and calling that works fine instead...! Oct 03 17:56:30 Makes me want to headbut a wall for a while... Oct 03 17:57:57 pev: if it wasn't clear what was causing the problem maybe we should look to make an improvement there Oct 03 17:58:51 However, the postinst running at S98 meanst that it doesnt patch things in time for processes starting earlier to pick up on the changes... Surely everyone else has that problem? I guess the quick fix is to change the S98run-postinsts to force a reboot afterwards? Oct 03 17:59:50 Or get my earlier services to check whether the changes have occurred and take care of it themselves instead... Not very pretty at any rate! Oct 03 18:01:00 pev: the ideal fix is to make it so your postinstall can run on the host Oct 03 18:01:45 bluelightning: Yeah, I'm not sure what to suggest - it's one of those bad luck situations. I think the best thing would be to modify S98run-postinsts so that of opkg-cl configure fails it dumps out an error message. A BIG problem is that after calling opkg-cl, S98run-postinsts deletes itself whether it has succeeded or not... Oct 03 18:01:54 hiding the evidence... Oct 03 18:02:14 bluelightning: Yeah, I guess so. It's not as tidy if I did that but such is life...! Oct 03 18:03:20 that certainly doesn't sound ideal Oct 03 18:03:22 we'd certainly take bug reports/patches to make specific improvements Oct 03 18:04:39 Is there a bug reporting form somewhere? I dont have time to test a patch at the mo but would be quick & easy to submit a report. Oct 03 18:05:33 pev: there is... http://bugzilla.yoctoproject.org Oct 03 18:06:11 the postinstall stuff has been tweaked in 1.5 but I think that this particular area may still work in the same way Oct 03 18:06:49 Great thanks... Oct 03 18:07:08 bluelightning: It still is same - I'd checked git... Oct 03 18:07:17 ok Oct 03 18:14:33 Hm, with postinst functions, do you not need to start with a hashbang if executing on the host? The recipes seem inconsistent! Oct 03 18:29:52 Also - what's the bitbake command to force a postinst rule to occur? Is it part of creating the rootfs? Oct 03 18:46:31 I think it depends on where you want it. I think I've used SSTATEPOSTINSTFUNCS recently, which runs the functions even when using a package from sstate instead of actually building it. Oct 03 19:18:13 OVERRIDES_append = ":bitbake-come-on-just-work-with-me-here" Oct 03 22:20:25 hi Oct 04 00:07:35 Argh... Getting really annoyed - I've got a pkg_postinst function in my recipe but I can't ever see it being called on the development host... Any ideas how to work out why not? Been looking at it a few hours and failing Oct 04 00:25:09 pev: check the opkg postinst file and see what's in it Oct 04 00:25:20 unless you're not creating ipks... Oct 04 00:26:09 if your build includes the package manager stuff, then opkg will install all those info files under /usr/lib/opk/info i believe Oct 04 00:27:05 mr_science: i believe it's /var/lib/opkg/info Oct 04 00:27:22 thought so Oct 04 00:27:32 the other one is classic... Oct 04 00:27:47 ah, good to know Oct 04 00:28:03 anyway, there should be a .postinst file Oct 04 00:28:36 might be specific to arago/oe-classic Oct 04 00:29:16 but the *.postinst will chow what pev is looking for Oct 04 00:29:35 now it's time to go home... Oct 04 01:03:34 khem: ping? Oct 04 01:42:17 Thanks guys... Oct 04 01:49:43 Anyone looked at doing a bbappend of init-ifupdown? Im looking at the implementation here : http://git.yoctoproject.org/cgit/cgit.cgi/meta-intel/tree/meta-sys940x/recipes-core/init-ifupdown/init-ifupdown_1.0.bbappend?h=master and wondering where the dirname used comes from? Oct 04 01:50:02 ie the dirnames here : http://git.yoctoproject.org/cgit/cgit.cgi/meta-intel/tree/meta-sys940x/recipes-core/init-ifupdown/files Oct 04 01:52:27 pev: When you refer to dirnames are you talking about specifically? because the machine name gets auto expanded out to the FILESPATH (which is how the interface file gets included from those subdirs) Oct 04 02:02:31 nrossi: Ah, I din't know there was a machine name - where does that come from? Oct 04 02:03:49 pev: for each entry in FILESEXTRAPATH it gets expanded with all the values in OVERRIDE, if you want to know the specifcs run bitbake -e init-ifupdown and look at the FILESPATH variable :) Oct 04 02:05:50 prev: Sorry i mean the 'FILESOVERRIDES' variable Oct 04 02:10:38 Cheers... Oct 04 02:11:51 also, do you know if you can override INITSCRIPT_PARAMS to *not* create the rc.d links? Oct 04 02:14:01 i usually comment out that stuff Oct 04 02:14:28 did you get your package-postinst stuff sorted out? Oct 04 02:15:55 you can leave in the inherit update-rc.d Oct 04 02:16:19 i think... Oct 04 02:18:42 Nah, I wasted several hours with trying to get it to work then gave up and went for a different less tidy workaround :-( Oct 04 02:30:55 it actually does work, in both classic and poky master Oct 04 02:31:19 what kinds of packages do you have enabled? Oct 04 02:32:23 i had some issues with debs when i started on poky so i went back to the ipk default Oct 04 02:32:40 haven't tried deb/rpm since then Oct 04 02:32:57 Yeah, I'm stuck with opkg from someone elses decision... Oct 04 02:33:03 lemme paste something... Oct 04 02:49:32 http://paste2.org/2Gdt4PcG Oct 04 02:53:24 you have to specify a package name, so just "pkg_postinst() {" won't work Oct 04 02:57:01 what does your pkg_postinst_foo look like, and what did you end up with? **** ENDING LOGGING AT Fri Oct 04 02:59:58 2013