**** BEGIN LOGGING AT Fri May 17 02:59:57 2019 May 17 04:48:47 It seems that my sstate cache behaves weirdly at times. May 17 04:48:53 There seems to be a "hidden" character, which makes do_rootfs() fail when installing binutils-lic package inside the image. May 17 04:49:00 https://pastebin.com/HR6XJH0R Anyone has helpful pointers (except for deleting sstate cache and building everything again)? May 17 04:49:06 I have tried "bitbake -c cleansstate binutils", however, that also doesn't solve the problem. May 17 05:13:36 has anyone else been having trouble with the list? May 17 05:13:46 apparently its rejecting my patches for some reason May 17 05:52:36 New news from stackoverflow: Custom eSDK build failure May 17 06:48:55 good morning May 17 06:50:05 howdy! i've got a strange / bug observation: a really simple meson-based recipe http://paste.ubuntu.com/p/CNVQJvNZnX/ works nicely with package_ipk, but fails with package_rpm. interestingly, this happens somewhen during rootfs creation, when dnf kicks in: https://paste.ubuntu.com/p/ZnjqfkHvnM/ May 17 06:50:35 mckoan: You too, Marco. May 17 07:03:31 alessioigor: ;-) May 17 07:35:20 LetoThe2nd: And no weird chars in recipe name? May 17 07:36:33 not at all May 17 07:36:38 "this-is_git.bb" May 17 07:37:13 and renaming to "thisis" does not change things too May 17 07:38:44 the fun fact is that it works nicely with package_ipk. so i suspect some (un-?)intended interaction between meson and package_rpm May 17 07:42:30 Can you see that the rpm got created ok in WORKDIR for this-is? May 17 07:42:52 So that it's not that the packaged ended up empty and wasn't created or something May 17 07:43:27 I don't know if the defaults for creating empty packages differes between ipk and rpm May 17 07:44:21 hum, deploy-rpms only has -dbg and -dev May 17 07:45:27 When I built the recipe locally no files are installed May 17 07:45:38 erbo: you mean the meson build file misses the install target and this breaks it? May 17 07:45:45 very well possible, indeed. May 17 07:46:03 yeah that seems to be the case May 17 07:46:13 erbo: let me check that, thanks a lot! May 17 07:53:00 New news from stackoverflow: An issue with postinst-intercept scripts May 17 08:00:53 erbo: that did the trick, thanks again! May 17 08:55:08 hopefully final version just sent.. has the skip login in it May 17 10:37:44 RP: the new rpm test skips itself if it can't find the needed rpm package, which I think is wrong May 17 10:38:05 RP: don't merge that please, until we get to the root cause of why the package isn't there sometimes May 17 10:38:39 run-postinsts should be there always unless the image is read-only May 17 11:02:49 Hi, May 17 11:02:49 I can't seem to compile util-linux-lsblk in warrior. Used to be ok. Is something wrong with http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=d11cbdf149cbfd273e76796a2444d4b9423c9ab0 that breaks the package? May 17 11:08:49 Error message when I bitbake util-linux-lsblk is May 17 11:08:49 ERROR: Nothing PROVIDES 'util-linux-lsblk'. Close matches: util-linux util-linux RPROVIDES util-linux-lsblk May 17 11:08:49 ZubairLK: well whats wrong with doing "bitbake util-linux" then May 17 11:08:49 I don't want the complete util-linux package installed in my image. Just lsblk. May 17 11:08:49 ZubairLK: well then install the package. but you cannot build it seperately anymore, it gets created as part of util-linux. thats basically what this commit does. **** BEGIN LOGGING AT Fri May 17 11:10:01 2019 May 17 11:10:36 ZubairLK, so you get no util-linux-lsblk* under tmp/deploy/* May 17 11:10:45 no package for it May 17 11:13:13 @kroon nope. `find . | grep util-linux-lsblk` returns nothing in tmp/deploy May 17 11:28:36 odd May 17 11:38:25 ZubairLK, does your log.do_compile reference lsblk.c ? May 17 11:41:48 I find these in do_package_write_ipk logs. May 17 11:41:49 NOTE: Not creating empty archive for util-linux-lslocks-2.32.1-r0 NOTE: Not creating empty archive for util-linux-lscpu-2.32.1-r0 NOTE: Not creating empty archive for util-linux-look-2.32.1-r0 May 17 11:42:21 yes lsblk.c is referenced in the do_compile log May 17 11:49:49 ZubairLK: I know bitbake couldn't find out which recipe provide which package when PACKAGES_DYNAMIC was used a while back. Maybe that's still the case, would explain your issue. May 17 11:50:41 ZubairLK: do you get the util-linux-lsblk rpm in tmp/deploy/ if you bitbake util-linux? May 17 11:51:00 s/rpm/ipk May 17 11:52:18 No. no util-linux-lsblk package when I do a bitbake util-linux May 17 11:52:41 probably because of "temp/log.do_package_write_ipk.10434:128:NOTE: Not creating empty archive for util-linux-lsblk-2.32.1-r0" May 17 11:53:04 temp/log.do_package.21524:1070:DEBUG: util-linux-lsblk: Dependency libblkid.so.1 requires package util-linux-libblkid (used by files: /home/zubairlk/resin/yocto/resin-intel/build/tmp/work/core2-64-poky-linux/util-linux/2.32.1-r0/packages-split/util-linux-lsblk/usr/bin/lsblk) May 17 11:53:10 temp/log.do_package.21524:1071:DEBUG: util-linux-lsblk: Dependency libsmartcols.so.1 requires package util-linux-libsmartcols (used by files: /home/zubairlk/resin/yocto/resin-intel/build/tmp/work/core2-64-poky-linux/util-linux/2.32.1-r0/packages-split/util-linux-lsblk/usr/bin/lsblk) May 17 11:53:27 those don't look nice. May 17 11:53:39 a couple of other dependencies like that as well May 17 11:54:19 Hmm, since it put "/usr/bin/lsblk" in "packages-split/util-linux-lsblk" I wonder why it is skipped due to being empty? May 17 11:55:05 temp/log.do_package.21524:521:DEBUG: LIBNAMES: pkg util-linux-lsblk libs 0 bins 1 sonames [] May 17 11:55:10 indeed. it does say 1 binary May 17 11:56:09 actually. packages-split/util-linux-lsblk/ folder is empty. May 17 11:56:25 in fact many packages-split/util-linux-*/ folders are empty May 17 11:57:49 Weird, other logs analyzed which file the binary linked to, and then it must have existed in packages-split/util-linux-lsblk/, right? May 17 12:01:59 ZubairLK: can you check in image/usr/bin/ if util-linux-blkid is a symlink? May 17 12:02:53 That seems to get some extra handling that will put it into the package of the symlink target, if I read the recipe correctly May 17 12:05:58 -rwxr-xr-x 1 root root 341544 May 15 21:29 image/usr/bin/lsblk May 17 12:06:11 nope. do you mean blkid or lsblk? lsblk is not a symlink May 17 12:06:47 blkid in sbin is also not a symlink -rwxr-xr-x 1 root root 426400 May 15 21:29 image/sbin/blkid May 17 12:06:49 -rwxr-xr-x 1 root root 426400 May 15 21:29 image/sbin/blkid May 17 12:09:04 oh, I meant lsblk :) May 17 12:13:10 -rwxr-xr-x 1 root root 341544 May 15 21:29 image/usr/bin/lsblk May 17 12:13:15 nope. not symlink May 17 12:42:26 ZubairLK: I did a quick test locally using warrior, and I do get a util-linux-lsblk rpm atleast May 17 13:24:00 New news from stackoverflow: Yocto - Create and populate a separate /home partition May 17 13:43:21 strange. i'll double check May 17 13:51:32 #windows May 17 13:59:20 seems like something in my bbappend (a few alternative_links. not lsblk) might have broken this. because i removed those and I see the package now May 17 14:24:14 New news from stackoverflow: global yocto AM_CFLAGS to pass to the gcc compiler May 17 19:22:22 not sure if I asked this already but is there a way for adding a dependency on a PACKAGECONFIG options for the ptest packages of a recipe (i.e lttng-tools and the lttng-ust packageconfig option). I would like to force the usage of lttng-ust when the ptest package is installed. May 17 23:08:42 kanavin_: agreed May 17 23:26:49 RP: so I noticed that now the default multiconfig does not have a name May 17 23:27:05 RP: vs before it was called default May 17 23:27:19 RP: Im not necessarily against it, just wondering on the reasoning behind that May 17 23:27:32 RP: my multiconf depends looks a little weird now May 17 23:29:38 do_image[mcdepends] = "multiconfig::arm64:do_rootfs" **** ENDING LOGGING AT Sat May 18 03:00:11 2019