**** BEGIN LOGGING AT Wed Apr 13 02:59:58 2016 Apr 13 06:50:24 hi guys Apr 13 06:57:45 i'm wondering how i can change the IMAGE_TYPEDEP_iso to ext4 without editing the original bootimg.class? i tried to create a class in my layer that inherits from bootimg, but the variable is always overriden to ext3 Apr 13 08:29:31 Hello, since few days, we have some issue to fetch sources from Yocto git repositories, someone else does he have the same problem? Sometime, we obtain a timeout when we fetch sources throught git repo manifest. Apr 13 10:27:38 I have a package that installs a file used for a workaround. Now I need this file in another package regardless of the the first. However, both packages can't install the same file due to path conflicts. Apr 13 10:28:20 I guess I should put the workaround file in its own package and use dependencies to make sure it is installed. But how would I go about migrating to this package structure? Apr 13 12:14:58 hello Apr 13 12:26:37 belen, I got toaster started, but when I try to create a project, it complains about a version number Apr 13 12:26:41 sound familiar? Apr 13 12:46:01 With kergoth and rburton's help my build is past the libsdl problem. I am usng poky from git with the jethro patch. Apr 13 12:46:16 however I am now hittng the same problem which I encountered before. Apr 13 12:46:38 This is the same problem I was showing people at the yocto developer day Apr 13 12:46:53 u-boot is faiilng to find simple types. Apr 13 12:47:26 I've looked at the log and the environment using bitbake -e and I noticed it is using host platform paths for an arm target. Apr 13 12:47:56 I thought that wwas the problem but I've been told that uboot will use the host platform for some portions/tools. Apr 13 12:48:28 With that said, here is the console build failure log and the run.do_configure step Apr 13 12:48:49 any pointers to what I should be looking for to correct is much appreciated Apr 13 12:48:54 https://gist.github.com/netskink/90135d8774ac1548d3b14eeaed0e2514 Apr 13 12:59:27 Crofton|work: nope. What's the message? Apr 13 12:59:37 hang on Apr 13 13:00:12 oe-core checkout + bitbake at smae level (not nested) Apr 13 13:00:17 I made a bug btw Apr 13 13:00:29 click on new porject in upper right hand Apr 13 13:00:42 Crofton|work: right, I haven't even had a chance to look at my email yet Apr 13 13:00:44 enter gnuradio-test Apr 13 13:00:55 click create project Apr 13 13:01:18 reddish box appears with "Fields missing: projectversion" in it Apr 13 13:01:27 Crofton|work: found your bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=9455 Apr 13 13:01:28 Bug 9455: normal, Undecided, ---, toaster, NEW , When I try to make a new project toaster cries "Fields missing: projectversion" Apr 13 13:01:38 it was late when I made it :) Apr 13 13:02:36 it is possible the .templateconf is funny Apr 13 13:02:55 I need to get ready for a call , but nay advice would be handy Apr 13 13:02:59 Crofton|work: what branch/commit are you using? Apr 13 13:02:59 this is getting better Apr 13 13:03:05 master Apr 13 13:03:09 Crofton|work: commit? Apr 13 13:03:11 pulled yesterday Apr 13 13:03:34 b3afe7d9bd25a943e947de3ec064ea245173c5a8 Apr 13 13:03:45 I need khems from last fall to get this far :) Apr 13 13:03:51 bbiab Apr 13 13:04:39 if I do bitbake -c devshell u-boot-stream800 and then in the resultant shell do a ./configure is that the same as doing the configure step within bitbake? Apr 13 13:05:04 Crofton|work: ok. Will look into this: sounds like a real bug, I'm afraid Apr 13 13:06:26 hmm. actually there is not a configure script. it looks like it requires a make XXX_config step first. Apr 13 13:06:48 any idea which config file was used to do this in the bitbake configure step? Apr 13 13:07:14 belen, it makes it ewasier to get people using this stuff if it builds with distros other than poky Apr 13 13:08:53 Crofton|work: I know, I know, I know .. We really do try Apr 13 13:09:14 Crofton|work: but it's very hard to cover all the basis with the resources we have. Apr 13 13:09:26 Crofton|work: you have to have patience with us Apr 13 13:09:32 yep, just trying to give you feedback you can show to managers Apr 13 13:09:32 hi, what do I have to do to move the yocto build folder? Apr 13 13:09:45 Crofton|work: fair enough :) Apr 13 13:10:01 Crofton|work: we do want the feedback, and the bugs, don't get me wrong Apr 13 13:10:18 mwarning: i beliee you specify a variable. one sec I'll try to find it. Apr 13 13:10:30 that would be nice :) Apr 13 13:10:57 I do not like to build the whole environment anew Apr 13 13:10:59 look for YOCTO_WORK_DIR Apr 13 13:11:07 mine is in local-config Apr 13 13:11:23 * zeddii_home agrees. Crofton|work is good at breaking things :) Apr 13 13:13:31 I'm trying to find the command line used in the do_configure step for u-boot. Apr 13 13:13:41 Im in this dir ~/progs/tiotop/stream/unpack/packages/src/build/.build-yocto/tmp/work/anuvaaz-poky-linux-gnueabi/u-boot-stream800/2014.04-r20$ Apr 13 13:14:04 ant it has a bunch of patch files and then in a subdir uboot src Apr 13 13:14:46 I would like to find the config file it used when it configured the code prior to the compile Apr 13 13:14:47 Crofton|work: someone is looking into your bug. Hopefully we'll get it fixed Apr 13 13:15:10 how do I do this? I don't see a file in this dir which looks like a config file Apr 13 13:17:58 I'm a little worried about the .templateconf file orgin Apr 13 13:18:22 but then again nothing in it practically :) Apr 13 13:20:03 Crofton|work: we have a toaster call in 2 hours. Why don't you join us and explain to us what worries you? Plan B is sending an email to the toaster mailing list. Apr 13 13:20:08 davis: there is no YOCTO_WORK_DIR variable anywhere from what I can tell Apr 13 13:20:48 oh, that worry is nothing I read the file :) Apr 13 13:21:05 I just afraid my build isn't 100% clean Apr 13 13:22:15 mwarning: i'm using a twisted vendor build of yocto so my environment might not be same Apr 13 13:22:24 Crofton|work: up to you. Feel free to bring it up if it causes you trouble, or if you just want to troll us a bit ;) Apr 13 13:22:26 but examine you config dir Apr 13 13:22:37 davis: ok, thanks so far Apr 13 13:22:55 someplace in there, you might find a config file that has the environment variable you can tweak Apr 13 13:23:21 i'm guessing it will be in global_config for local_config Apr 13 13:23:29 belen, I'd like to get to an actual build failure before trolling hard :) Apr 13 13:23:57 Crofton|work: trolling "con fundamento" as we say in Spain Apr 13 13:25:32 exactly Apr 13 13:25:40 and yes, I had to use google translate Apr 13 13:32:47 belen, FYI: https://stackoverflow.com/questions/36521722/yocto-install-and-use-toaster Apr 13 13:45:52 belen: Good luck for the Toaster call!! :D Apr 13 13:47:18 Hope we will get some corrections soon. ;) Apr 13 13:47:39 so I have more info on this build failure with uboot Apr 13 13:47:45 https://gist.github.com/netskink/90135d8774ac1548d3b14eeaed0e2514 Apr 13 13:52:22 Crofton|work: thanks for the Stack Overflow link. I missed that one Apr 13 13:52:31 np Apr 13 13:58:13 is there any particular form required for adding patches to list of patch files or is it sufficient to add the diff -Naur contents? Apr 13 13:58:36 ii see stuff in the existing patches which look like they end in -- version number Apr 13 14:01:21 here is a more direct question. I've added a patch file, modified the .bb file to include it Apr 13 14:01:32 how do I restart the build using that patch? Apr 13 14:02:33 bitbake [recipename] Apr 13 14:02:44 bitbake -c liststasks u-boot-stream800 shows Apr 13 14:03:11 wow my build system is lagged Apr 13 14:03:25 do_patch Apr 13 14:04:02 so I could do bitbake -c do_patch u-boot800-stream800 and it would redo the patch step, right? Apr 13 14:09:24 davis: diff or a git format-patch both work Apr 13 14:12:45 CTtpollard: many thanks Apr 13 14:13:04 davis: many patches are generated via 'gitdiff'. In that case, the number at at the bottom is the version of git used to create the patch. Apr 13 14:13:10 i did the do_patch task. I was afraid it would not actually execute since thiss step had been done already Apr 13 14:13:23 however, it did a un_pack and then patch Apr 13 14:13:35 i'm afraid it erased my patch, lets see Apr 13 14:14:22 hmm. my patch is still in tmp/work/... dir Apr 13 14:14:24 so far so good Apr 13 14:15:34 and my patch is applied in the ubtoot dir Apr 13 14:16:28 so far so good. .ets see if bitbake u-boot-stream8000 works now Apr 13 14:31:47 rock and roll. it works. That dude marex in #uboot rocks the roll. Apr 13 14:32:37 bbiab Apr 13 14:34:29 I see in a number of recipes when someone wants to installd a systemd service they inherit the systemd.bbclass, but it seems if you want to install a system level systemd configuration they just manually install it Apr 13 14:34:42 is that just legacy stuff in recipes or the best practice? Apr 13 14:34:57 inherit systemd is best practise Apr 13 14:34:59 e.g. https://github.com/openembedded/openembedded-core/blob/master/meta/recipes-core/dropbear/dropbear.inc#L80 Apr 13 14:35:11 dropbear needs fixing Apr 13 14:35:42 rburton: okay, followup question: systemd.bbclass doesn't seem to mention installing to the "system" unit files directory Apr 13 14:36:08 looks again Apr 13 14:36:10 dropbear is fine Apr 13 14:36:17 you install the files however you need to Apr 13 14:36:29 upstream may install them, or in dropbears case they need to be installed manually Apr 13 14:36:42 then systemd class handles the postinst scripts Apr 13 14:36:46 afaict the systemd.bbclass generates some extra packaging related tasks Apr 13 14:37:16 oh I see it does handle the system units as well Apr 13 15:56:16 zeddii_home: around ? Apr 13 15:57:29 khem`: yep. Apr 13 16:10:11 hm this is strange, if I inherit systemd, but don't actually use any systemd variables (I only need to install the systemd system unit file) none of the bbclass behaviors are activated Apr 13 16:10:22 I guess I need to explicitly do a SYSTEMD_PACKAGES = "${PN Apr 13 16:10:25 }" ? Apr 13 16:12:10 so i have a couple of kernels in this vendor tree. If I do bitbake -c listtasks virtual/kernel it shows the tasks but not which particular kernel or layers are used. Is there a way to do that? Apr 13 16:14:15 what do you mean? the layers used are the layers you added to your build. the kernel being built is the recipe that resolves to virtual/kernel. bitbake -e virtual/kernel | grep '^FILE=' Apr 13 16:14:36 ahh, bitbake -v virtual/kernel shows a line selecting linux-stream800 to satisy virtual/kernel Apr 13 16:14:39 that works Apr 13 16:15:42 SYSTEMD_PACKAGES = "${PN}" doesn't change anything it seems, hrm Apr 13 16:15:45 kergoth: i'm trying to decipher this vendors setup for which of the kernels they provided. its not clear. Apr 13 16:20:35 this is odd. when I do my builds, it says MACHINE=anuvaaz but in the linux-stream800_4.1.bb it has a list for COMPATIBLE_MACHINE and anuvaaz is not one of them. Apr 13 16:30:14 hmm. I am trying to figure out why my kernel .config mods are vanishing. I move to the reference build for poky which uses x86qemu. ie. not this vendor version. Apr 13 16:30:37 in the stock, I can do bitbake -c menuconfig virtual/kernel make a mod to append a version string with JFD Apr 13 16:30:52 build it, run it and uname -a shows my mod. Apr 13 16:31:44 but when I do a bitbake -c devshell virtual/kernel and then make menuconfig I see my version string still there. However when I do find . -type f | xargs grep JFD I dont see my .config show up. Apr 13 16:32:11 where is the devshell storing the kernel config I am modifying in the stock poky setup? Apr 13 16:33:34 inside make menuconfig, when I click save its not showing a path. its showing the .config as default Apr 13 16:34:16 if I save it even says .config saved, but its not in the dir after I exit. Apr 13 16:38:57 :1 Apr 13 16:48:45 hi, looking at the wpa_supplicant bb file in meta-intel-edison (here: http://git.yoctoproject.org/cgit/cgit.cgi/meta-intel-edison/tree/meta-intel-edison-distro/recipes-connectivity/wpa_supplicant/wpa-supplicant_android-4.4.4_r2.0.1.bb) Apr 13 16:49:00 it includes a "require recipes-connectivity/wpa-supplicant/wpa-supplicant.inc" Apr 13 16:49:16 but that wpa-supplicant.inc is not in that meta layer at all. Apr 13 16:49:33 how can I find out where it is pulling it from (or whether this is a bug...) ? Apr 13 16:50:00 short of git cloning all the meta repos and hunting for it which is the only way I see to find it Apr 13 16:50:29 does the layer readme not list exactly what layers itd epends on? Apr 13 16:50:37 all layers are supposed to provide that Apr 13 16:51:49 in this case it's likely mismatched layer versions which are the problem. wpa-supplicant.inc hasn't existed in quite some time, so presumably they'd need your other layers to be from a previous release, not master Apr 13 16:51:57 yes, it lists "bitbake, openembedded-core and xxxx" :) Apr 13 16:55:08 kergoth, okay - then maybe I'm better off starting from a branch instead of master - I'm just looking for a working wpa_supplicant recipe to pull into my yocto build. Apr 13 16:55:48 they should indicate the needed branch in the readme, if not that's certainly a bug in the layer Apr 13 16:56:19 that said, the 'fido' branch of oe-core along with version 1.26 of bitbake should likely do Apr 13 16:56:47 wpa-supplicant.inc was removed in jethro (2.0), but still existed in fido (1.8) Apr 13 16:56:56 from some checking of git history in the oe-core repo Apr 13 17:01:15 actually I'll probably just switching to using the openembedded wpa-supplicant recipe as my starting point. Apr 13 17:01:40 thanks for the pointer. Apr 13 17:01:51 if the layer is intendedd to work with a different version, you'll likely hit numerous pain points. building mismatched branches is unlikely to go anywhere Apr 13 17:03:59 Can confirm. Apr 13 17:04:53 I'm building based on jethro. I'll try master from openembedded (which includes supplicant 2.5). If that fails I'll fallback to jethro branch (which has wpa_supplicant 2.4). Once I get one of them building I can probably get things going Apr 13 17:05:38 if you need that layer, and it depends on fido, trying to get jethro or krogoth to work is likely a losing proposition. but good luck with it Apr 13 17:05:56 depends on how much work you intend to do and what your requirements are, though Apr 13 17:06:48 kergoth, sorry - I wasn't clear. I'm dropping the wpa_supplicant from meta-edison altogether and just switching to using the one from openembedded directly. I have no preference for the edison one - it was just the first one I came across googling for a wpa_supplicant recipe Apr 13 17:08:59 you shouldnt' be googling for random recipes an expecting them to work Apr 13 17:09:03 use the layer index if you have to use anything Apr 13 17:10:17 * csd hopes wisdom of yocto will come but is still a grasshopper in this realm Apr 13 17:20:37 csd, we all start out as absolute beginners. hang in there man, you'll get it. Apr 13 17:21:44 csd: Try the jethro-next branch in meta-openembedded before you slam it all the way over to master. Apr 13 17:40:42 hmm. i found work-shared/qemux86/kernel-build-artifacts/.config and a similar one for my other build system for anuvaaz Apr 13 17:41:20 both have my mods, but x86 shows results in uname -a and /proc/config.gz but not in anuvaaz Apr 13 17:47:19 hmm. edits on the anuvaaz yocto build do not stick with changes to this file Apr 13 17:50:00 here is the recipe, i'm confused as to how it works Apr 13 17:50:05 https://gist.github.com/netskink/7700c036eec034281bab009a7f68bc33 Apr 13 17:50:18 on line 29 it sasys file://defconfig Apr 13 17:50:42 which defconfig is that? I have more than one. Apr 13 17:51:00 csd: nevermind. There's no difference between jethro and jethro-next for wpa-supplicant in meta-openembedded. Apr 13 17:51:40 And that was a very buzzword-heavy sentence, right there. Apr 13 18:03:32 oh noes, edited a `do_install` section and it caused a whole recompile of rethinkdb.. goodbye 1.5 hours on this tiny test build box Apr 13 18:20:29 is the reason, I can do bitbake -c menuconfig virtual/kernel and then bitbake virtual/kernel and my settings stick because the reference kernel for x86 uses linux-yocto Apr 13 18:20:52 nope Apr 13 18:20:56 and my vendor uses a kernel.org kernel recipe so it does nto handle these capabilitiies? Apr 13 18:21:30 it really depends on how the recipe is writing .config from the defconfig. there's no standard mechanism for it Apr 13 18:21:34 heh, i just switched tabs to see your otheer reply. kergoth, i need to buy you a beer. Apr 13 18:22:03 varies from one recipe to the next. if they don't implement the logic in the recipe itself, then do_configure will write .config from ${WORKDIR}/defconfig, assuming that there was a file://defconfig to put into workdir Apr 13 18:22:27 i'd thought it only wrote that if .config didn't already exist, but it's possible it writes it every time, if that's the case it'd explain your behavior Apr 13 18:22:33 can you discern something from that gist i pasted earlier? I have a file://defconfig line Apr 13 18:22:34 your best bet is really to properly persist your changes Apr 13 18:22:51 but I have a few different defconfigs in the file system Apr 13 18:23:12 i was expeciting to see just one, or at least one in the immediate subdir Apr 13 18:23:30 for your convience here it is again Apr 13 18:23:34 https://gist.github.com/netskink/7700c036eec034281bab009a7f68bc33 Apr 13 18:23:34 'in the file system' is rather vague, that doesn't tell us much of anything Apr 13 18:23:54 that paste shows where in the fs, 'm doiing the find Apr 13 18:24:17 right, so the file://defconfig is pulling defconfig from FILESPATH, and they are't doing anything with it in the recipe, so it's using the default defconfig -> .config logic that's in do_configure in kernel.bbclass Apr 13 18:25:22 im looking for a classes dir Apr 13 18:25:35 it's in oe-core/meta/ or poky/meta/ Apr 13 18:26:47 Here's my suggestion: do your -c menuconfig, copy the new .config out of tmp, yocto-layer create local 1; bitbake-layers add-layer meta-local; recipetool appendsrcfile -W meta-local virtual/kernel /path/to/the/new/defconfig defconfig Apr 13 18:27:34 erm, actually that 1 isn't ideal, either change that to 10 or modify LAYER_PRIORITY in meta-local/conf/layer.conf after the fact -- we want it higher priority than your bsp layer so it can override its files reliably Apr 13 18:28:21 recipetool appendsrcfile makes it easy to override source files for a recipe, either files in the source tree or workdir. in this case we're overriding the defconfig Apr 13 18:28:57 appendfile is also useful, it lets you override target files rather than source files Apr 13 18:29:34 this is easier with linux-yocto kernels or other kernels that support configuration fragments, as then you could use bitbake -c diffconfig to create a .cfg which holds just your changes after the menuconfig Apr 13 18:29:49 then you could use appendsrcfile to add that to SRC_URI rather than overriding all of defcnfig Apr 13 18:29:54 but not all kernels support fragments yet Apr 13 18:30:14 primarily because as mentioned before theere's no standard mechanism by which a defconfig becomes .config, so there's no standard place to inject the logic Apr 13 18:30:26 has to be done on a case by case basis Apr 13 18:30:44 i tried using kernel fragments at ELC, mark showed me how to do that, but it does not work for this kernel. I saved those notes off for future reference when we use a yocto kernel Apr 13 18:30:48 * kergoth nods Apr 13 18:31:24 id' start with this, but we can go over how to add fragments support to that kernel later if you like Apr 13 18:31:38 i've done so for certain bsp kernels for mentor's yocto-based product Apr 13 18:31:43 that is a lot good info, i can run with that for a while. many thanks. Apr 13 18:34:17 when running "bitbake custom-image -c populate_sdk" I received the build error "satisfy_dependencies_for: Cannot satisfy the following dependencies for packagegroup-qt5-toolchain-target: qtx11extras-mkspecs". Does anyone know how to solve this? Apr 13 18:35:37 in the mentor graphics distro layer we add new recipetool commands: kernel_set_defconfig, kernel_add_fragments, kernel_set_configs, and kernel_add_dts. all are just light wrappers around appendsrcfile, though :) Apr 13 18:36:51 cool. Let me work with this info for a bit. i'll be back. Apr 13 18:37:02 thank you for the info. Apr 13 18:37:12 np Apr 13 18:40:29 is there any documentation on generating persisent net udev rules for a minimal image? Apr 13 18:40:52 I've got some wacky if names I'd like to wrangle Apr 13 18:42:54 oh I see this is systemd specific huh Apr 13 18:43:46 mbroadst: maybe you can do some udev rules to rewrite the interface names Apr 13 18:43:55 might be simpler Apr 13 18:44:33 looks like this works fine: ln -s /dev/null /etc/udev/rules.d/80-net-setup-link.rules Apr 13 18:44:36 should have just rtfm :) Apr 13 19:28:06 * armpit wonders how we will handle Trade mark license in sources Apr 13 19:36:34 kergoth: this process seems to be working. your suggestion was the most precise i've ever seen in an irc channel. particulary noteworthy since i was expecting errors to abound. Apr 13 19:36:54 i've never heard of recipetool before Apr 13 19:46:28 pfft, after booting the new build, uname -a does not show my name and CONFIG_USB_ETH is not set. /cry Apr 13 19:47:45 bblayers has a meta-JFD-KERNEL at the bottom. Apr 13 20:21:03 davis: do a bitbake -c unpack virtual/kernel, then diff the defconfig in the recipe's WORKDIR (see bitbake -e virtual/kernel|grep WORKDIR=) against hte one in your layer. see if it matches up or not. if not, check the layer priority of the bsp layer vs your layer, yours needs to be higher Apr 13 21:45:28 If I have executable files that I want to load into my targets file system (e.g. under /usr/bin), how do I do that in my custom recipe so that it is automatically done when I build my image? Apr 14 02:27:28 hello everyone Apr 14 02:46:27 hi respawn1 Apr 14 02:56:55 I had a question, I'm not sure I entirely understand just what Yocto is. I am not a linux noob, but an average user, not a programmer or developer. I am interested in building my 'own' Linux system. I was considering "Linux from Scratch" before I came across the Yocto Project. Is Yocto a tool that can help me in this regard? thank you **** ENDING LOGGING AT Thu Apr 14 02:59:58 2016