**** BEGIN LOGGING AT Fri Apr 20 03:00:01 2018 Apr 20 06:30:15 New news from stackoverflow: Removing busybox completely from a Yocto generated image Apr 20 07:11:25 hello, are we getting a new Yocto release this month or is it getting delayed? Apr 20 07:47:32 Hello ! Apr 20 07:48:37 Helhi Apr 20 08:41:02 fray: you don't need to re-execute do_install to trigger the issue, it's something which happens in do_package Apr 20 08:42:23 I got it reproduced 9 times from 126 iterations over night Apr 20 09:18:28 hello guys.. bitbake tells me that WARNING: libpam-1.3.0-r5 do_pam_sanity: Building libpam but 'pam' isn't in DISTRO_FEATURES, PAM won't work correctly Apr 20 09:19:11 so in conf/local.con I have added : DISTRO_FEATURES_append = " systemd pam" Apr 20 09:20:03 but if i run bitbake li Apr 20 09:20:20 bitbake pam: Error: Nothing PEOVIDES pam Apr 20 09:22:37 fberg: bitbake libpam is what you want Apr 20 09:22:51 jaMa: thanks Apr 20 10:01:02 New news from stackoverflow: How can I add or delete the recipe in Yocto image by toaster Apr 20 11:28:14 i have a question about shared-state.. is it a good idea to share the same SSTATE_DIR between different users that are using different versions of Yocto (like krogoth and thud) or even with users that are differen openembedded based distros like AGL .. ? Apr 20 12:08:26 clopez: should be fine, in principle. sstate cache is pretty robust. it won't likely cause problems, except possibly a performance decline from having really large directories. Apr 20 12:08:38 but you shouldn't have actual clashes where you pick up the wrong package or anything. Apr 20 12:35:30 I want to add a module to my yocto kernel Apr 20 12:35:39 where can i find the .config file? Apr 20 12:50:53 franco_: in tmp/work/ and deeper until you get to your linux kernel work directory Apr 20 12:51:26 franco_: usually it's named defconfig, you can find build location with bitbake -e virtual/kernel | grep '^B=' Apr 20 12:51:42 more or less Apr 20 12:52:38 https://www.yoctoproject.org/docs/current/mega-manual/mega-manual.html#kernel-dev-common Apr 20 12:55:19 franco_: mine is in tmp/work/-gnueabi//4.9.11-r0/build Apr 20 12:56:07 this is inside the build directory hierarchy, of course Apr 20 12:56:28 seebs: ok.. thanks for the info Apr 20 12:57:37 khem: https://autobuilder.yocto.io/builders/nightly-musl/builds/975/steps/BuildImages/logs/stdio Apr 20 13:01:38 New news from stackoverflow: ath10k Firmware : IBSS (Ad-Hoc) - WiFi : Mode missing Apr 20 13:04:24 I had modified the kernel source, now what have i to do in order to include the "new kernel" in my next image? Apr 20 13:19:16 franco_: you can take a look at this: https://www.yoctoproject.org/docs/2.4.2/kernel-dev/kernel-dev.html#changing-the-configuration Apr 20 13:21:00 franco_: oh wait, you mean you are not using a new configuration, but have actually edited the source itself? Apr 20 13:22:08 I have edited the .config file in order to include a new module Apr 20 13:22:43 franco_: then the documented method should work for you Apr 20 13:23:14 ok thx Apr 20 13:23:28 If i would like to change the sources, what is the best way? Apr 20 13:23:34 franco_: I recommend you do the configuration changes in your own separate layer if you are working with a 3rd party BSP Apr 20 13:23:36 Creating a .patch file? Apr 20 13:24:26 franco_: for source changes, I find it convenient to have a separate kernel fork git repository and write my own kernel recipe Apr 20 13:24:58 then include the kernel recipe name in my machine configuration definition Apr 20 13:25:32 ok Apr 20 13:25:48 not necessary if you are merely doing small changes, of course Apr 20 13:26:02 i.e. if you only have a few small patches Apr 20 13:26:18 Regarding the point 1. After I changed the .config file using make menuconfig. Have i to run make? Apr 20 13:26:33 but in my case there are a few custom drivers and devicetrees Apr 20 13:26:50 or is sufficient to run bitbake core-image-minimal to obtain an image with the new kernel? Apr 20 13:27:47 franco_: bitbake -c menuconfig virtual/kernel Apr 20 13:28:11 bitbake -C compile virtual/kernel; Apr 20 13:30:12 what is the difference between those two commands? Apr 20 13:41:22 franco_: https://www.yoctoproject.org/docs/2.4.2/ref-manual/ref-manual.html#usingpoky-debugging-taskrunning Apr 20 13:43:23 for the difference between -c and -C Apr 20 13:47:42 first command defined by nayfe lets you run menuconfig and configure your kernel, second one actually compiles it Apr 20 13:48:42 however, working in such a way, it might eventually be easy to lose the changes made with menuconfig Apr 20 13:49:20 I suggest backing up the .config after the menuconfig step Apr 20 13:50:01 but honestly, me and my colleague prefer to play with kernel configuration outside of Yocto Apr 20 15:53:30 How can I check what package pulls in another? I'm getting glibc-dbg on target, but I cannot immediately find which package causes it to be pulled in. Apr 20 15:53:54 if using RPM, on the target rpm -q --whatrequires glibc-dbg Apr 20 15:54:32 otherwise you'll have to manually scan the package dependencies and create your own map to figure it out.. most likely it's a RRECOMMEND that is doing it, from another -dbg and/or -dev package Apr 20 15:54:34 fray: ah, alas, I have ipk. And no package-manager. so it'll ned to be host-side Apr 20 15:55:18 ya, so you will have to map it manually.. ipk's are nothing more then an ar archive.. just pull out the control and you can interpret it pretty easily and create a mapping. Apr 20 15:56:02 Yeah, I have a dumpfile of all ipks that I can search. Works well for most cases, but refering to libc6-dbg is one of the most occurring pacakges :D Apr 20 15:59:29 you need to use the list of things installed (or asked to be installed) onto the target.... Apr 20 15:59:37 then start there and work outward for dependencies.. Apr 20 15:59:58 fray: I probably need to write a py script for doing that parsing Apr 20 16:00:37 sveinse: Have you enabled buildhistory? Apr 20 16:00:48 paulbarker: yes Apr 20 16:01:19 I've used 'dot maps' in the past for some of that... but if the image is complex, it's too difficult to read Apr 20 16:06:03 If you look in the buildhistory directory you should see some dependency graphs in .dot format and a directory full of package info Apr 20 16:09:02 oh damn, many uppercase sadness Apr 20 16:09:15 Found it. What I did was: find build/tmp/deploy/ipk -name \*.ipk -exec less '{}' ';' | cat >ipk_files; grep -e '\(Package\|libc6-dbg\)' ipk_files.txt |grep -B1 libc6-dbg |grep 'Package:' |grep -v -e'Package:.*-dbg' Apr 20 16:09:17 I spent a whole day trying to understand why python3 was failing during build Apr 20 16:09:31 and it was because the $PWD contained upper case letters Apr 20 16:09:33 :( Apr 20 16:10:18 Handful isn't it :D In English: Find all packages that depends or reccomends libc6-dbg but doesn't have -dbg in its own name :D Apr 20 16:10:26 sveinse: That is some good grep-fu Apr 20 16:12:41 paulbarker: indeed :D -- in the longer run, I think perhaps we should have a simple tool to query dependencies, aka what rpm and dpkg can do on target Apr 20 16:16:10 we do Apr 20 16:16:19 oe-depends-dot Apr 20 16:16:42 ah rburton thanks stupid mistake in packaging Apr 20 16:24:37 oe-depends-dot: copmmand not found Apr 20 16:27:26 and thats what you get for not tracking master like a real engineer Apr 20 16:28:42 rburton: :D We're on a product release streak, so not an option right now... Apr 20 16:28:48 pah ;) Apr 20 16:29:00 easy to rip out of cgit if you want it, doesn't depend on anything else Apr 20 16:29:29 rburton: But anyhow, oe-depends-dot seems to be producing gigantic dot schemas, right? Not just "give-me-what-pulls-me-in" kinda thing, right? Apr 20 16:29:37 * sveinse guessing Apr 20 16:29:52 sveinse: no, oe-depends-dot parses the .dot files emitted by bitbake -g Apr 20 16:29:58 which includes what you told it to include, and their deps Apr 20 16:30:28 kergoth: ah right, cool Apr 20 16:31:22 thats all buildtime and known-at-build-time depends, still need to extend pkgdatautil to dig around the dependnecies of the produced binarie Apr 20 16:31:33 poky-contrib:ross/pkg has the start of a tool to do that, fwiw Apr 20 16:34:23 rburton: sent the whole series fixed Apr 20 17:03:25 I'm trying to set up a virtualbox build and seeing bitbake: Unable to connect or start server errors, ubuntu 16.04 on Mac host Apr 20 17:04:42 I've followed the quickstart guide, read the bitbake manual, tried verbose without any additional info Apr 20 17:05:38 I'm trying on a Linux hosted 16.04 instance now Apr 20 17:05:44 anyone seen this before? Apr 20 17:06:28 I originally saw this with OpenBMC, but I am now also seeing the error with poky Apr 20 17:07:59 rburton, trust me, its me behind 12703 Apr 20 17:08:05 what we did was: default MACHINE=raspberrypi3, added meta-openembedded layer and meta-raspberrypi Apr 20 17:08:15 build default image, and breaks with python3 recipe Apr 20 17:08:35 after two days of debugging, I tried a mv of recipes into the home, instead of home/Riccardo/Prova/whatever Apr 20 17:08:37 and this worked Apr 20 17:09:06 LocutusOfBorg: i've had my build tree in a uppercase directory for four years now Apr 20 17:09:07 now I did a build with a "Prova" in $PWD and worked, I'm doing a build with Documents, to find which keyword is throwking the error Apr 20 17:09:23 rburton, my wild guess is that the combo, or some "keyword" is the clue Apr 20 17:09:29 yeah my hunch is a bad sed somewhere Apr 20 17:09:40 for sure, after 5 years of dailywork in yocto, I had seen lots of upper cases Apr 20 17:10:03 and BTW the worst error in my life I have ever seen, was a PWD containing "TIERRA" (upper case) Apr 20 17:10:30 and yocto throwing out an error like "hey, too many errors in this image build log" Apr 20 17:10:38 because of some grep ERR |wc -l and if > 50 bail out Apr 20 17:10:47 (and the PATH was the issue here :p Apr 20 17:11:15 I don't want to give up right now on that bug, so I'll reopen in case to further discuss or at least close with an useful explanation for people having the same issue Apr 20 17:11:57 but for some reasons icc is detected and gcc can't understand its flags Apr 20 17:12:30 (this is also why I asked Riccardo, to provide the path in the bug report, so people can reproduce it) Apr 20 17:14:56 LocutusOfBorg: icc is something you need to turn on, according to configure.ac Apr 20 17:15:26 yep, I wild guess some sed too Apr 20 17:15:31 probably a bug in python3 Apr 20 17:15:34 TBH Apr 20 17:21:01 update on virtualbox/bitbake errors: it works perfectly on a virtualbox linux hosted image configured exactly the same way Apr 20 18:01:08 krtaylor: seems virtualbox has issues on mac then Apr 20 18:02:51 krtaylor: are you using a directory on the host to build in or do you have a disk image Apr 20 18:03:48 LocutusOfBorg: I have seen rpm qripe about uppercase in version parts of packages but otherwise I use it all the time Apr 20 18:04:00 especially in directory names Apr 20 18:04:18 khem: we sanity check that and abort now Apr 20 18:05:27 you mean the version part ? Apr 20 18:08:51 fedora seems to be ok with it see https://fedoraproject.org/wiki/Package_Versioning_Examples Apr 20 18:14:19 khem, yes, but I have heard others building fine in VB on a mac, must be config Apr 20 18:14:52 rburton, I am using a shared folder on the mac, hm, I'll try a new image w/o sharing Apr 20 18:42:35 krtaylor: hfs is a bit rubbish, so yeah that won't work Apr 20 19:47:11 rburton, thanks, this weekend, I'll try a mac ubuntu build w/o shared folder and report back the results Apr 20 20:03:40 khem otavio: got a new go bug https://autobuilder.yocto.io/builders/nightly-world-lsb/builds/941/steps/BuildImages/logs/stdio looks like its using host cflags on CC_FOR_BUILD builds? Apr 20 20:03:48 erm, target cflags on host builds Apr 20 20:03:51 if you see what i mean Apr 20 20:17:43 rburton: it looks like -fstack-protector-strong is being used on host part of the build of go-runtime Apr 20 22:05:26 where's rburton when you need him :) Apr 20 22:12:03 otavio: look, a challenging u-boot bug, you wanna fix it ? Apr 20 22:41:26 zeddii_home: the 4.4 preempt-rt kernel hangs on some hardware. I have a patch for it but the correct solution is to update the preempt-rt patchset I think Apr 20 22:43:13 zeddii_home: this fixes it and is part of the new rt patch: https://pastebin.com/VYk5TKjQ Apr 20 22:59:40 zeddii_home: I haven't tested it, but I believe the 4.9 kernel should have the same issue and fix Apr 20 23:53:30 Update on my virtualbox on Mac build problem, I have verified that it is the shared directory that was the problem - exact same image w/o shared working directory builds fine Apr 20 23:53:57 Just FYI in case someone else has bitbake connection issues on a Mac Apr 21 00:01:08 clsulliv: cool. thanks. I have some updates queued, and will double check. I was sitting on things while the release finalized. Apr 21 00:20:05 * armpit zeddii_home sounds like the easter bunny Apr 21 01:15:46 he’s making a list and checking twiiiice! **** ENDING LOGGING AT Sat Apr 21 03:00:05 2018