**** BEGIN LOGGING AT Thu Mar 14 02:59:57 2019 Mar 14 03:08:42 devtool modify cannot find a recipe inside sources Mar 14 03:09:21 i put sl.bb inside sources/sl/ and doing devtool modify sl.bb sl/ Mar 14 04:19:24 What language are the recipes written with? Mar 14 04:22:24 User_: you are asking to me ? Mar 14 04:22:58 User_: they are .bb recipes manly just shell scrpits Mar 14 04:25:20 alicef, asking to channel Mar 14 04:25:39 User_: you can write it in python and bash Mar 14 04:25:41 alicef, the shell must be any particular shell? Mar 14 04:25:41 why Mar 14 04:26:01 i think is some tagged version of bash Mar 14 04:28:50 User_: https://www.yoctoproject.org/docs/2.5/dev-manual/dev-manual.html#recipe-syntax Mar 14 04:36:19 alicef, so the script language is a custom one for yocto and the regular python? Mar 14 04:36:47 eh ? Mar 14 04:37:11 if is a general question about yocto yes Mar 14 04:37:37 if is a question about my previous question, I don't understand the connection. Mar 14 04:39:11 alicef, I mean the language used to write the recipe. There is no mention that it has to conform to bash or other shell languages Mar 14 04:40:27 learningc: https://www.yoctoproject.org/docs/2.5/dev-manual/dev-manual.html#recipe-syntax Mar 14 04:46:50 Does the yocto docs exist in pdf format? Mar 14 04:58:23 learningc: http://lmgtfy.com/?s=d&q=yocto+manual+pdf Mar 14 05:03:06 :D Mar 14 05:47:41 python3-argparse has disappeared between krogoth -> thud, it's now part of python3-core. How do I write a recipe that uses arg-parse that works for both krogoth and thud? Mar 14 10:32:48 hello :) Mar 14 10:33:06 I am trying to make a i.MX6 boot from sdcard but I cannot manage to :( (imx6qdlsabresd) Mar 14 10:33:23 what must I do with my .wic files to make it work ? Mar 14 10:33:34 I have exactly the same case than https://stackoverflow.com/questions/49738198/how-do-i-boot-a-yocto-generated-image-on-the-mx6-sabre-devices Mar 14 10:33:59 I see the freescale logo show up, but in the UART I have nothing after "Starting kernel ..." Mar 14 10:35:51 jcelerier: well that sounds pretty much like it happily boots, but you just don't have any boot messages or tty set up Mar 14 10:37:22 or at least u-boot comes up Mar 14 10:42:26 yes, u-boot comes up Mar 14 10:42:51 I tried both the core-image-minimal and fsl-image-machine-test Mar 14 10:43:05 I should at least get a tty, right ? Mar 14 10:45:16 "should" hum, no idea. don't know that machine setup Mar 14 13:04:00 zeddii: So shouldn't the symlink here shown here https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/recipes-kernel/linux/kernel-devsrc.bb#n51 be pointing to ${D}${KERNEL_BUILD_ROOT} ? Or am I mistaken ? Mar 14 13:11:56 henriknj: pretty sure zeddii is off for the week Mar 14 13:12:10 marka: alright, it was worth a try Mar 14 13:12:15 maybe you can answer my question? Mar 14 13:13:13 * marka looks Mar 14 13:14:58 My issue is that the symlink, as it is right now, ends up point to /lib/modules/${KERNEL_VERSION}/source, but should really have been relative to the image-rootfs it is part of Mar 14 13:17:29 the link is only valid on the target Mar 14 13:17:34 in which case it is correct Mar 14 13:18:10 yes, but not valid for a sdk Mar 14 13:18:56 agreed, for the sdk it would be a problem Mar 14 13:19:50 so yes, you would potentially have an argument it could be useful for on target and sdk if it was a relative link Mar 14 13:20:41 but changing it to add ${D} would not be the correct answer Mar 14 13:20:47 it used to work in a sdk context before zeddii made some major changes to kernel-devsrc Mar 14 13:20:58 lnr will make a relative link for you given two absolute paths Mar 14 13:21:36 henriknj: unfortunately I didn't follow along with his changes, I was too busy with other distractions Mar 14 13:23:20 anyway, see if you can sort it out with a relative link and send it to the list or directly to zeddii to evaluate Mar 14 13:23:53 marka: ill try that Mar 14 13:25:38 anyone else notice lately that bitbake doesn't complete when there is a parse error. I have to double ctrl-c and then forcefully kill bitbake processes Mar 14 13:34:49 damn, silence isn't good. I suppose I might have picked up some transient issue based on when I pulled. Mar 14 13:35:44 Maybe an odd question :) Mar 14 13:35:52 Is it possible to configure OE/Yocto Mar 14 13:36:03 (and relatively easy) Mar 14 13:36:20 to use toolchain 6.x to compile a single recipe Mar 14 13:36:28 and the rest of the system would use 4.9 ? Mar 14 13:38:46 I'm thinking of adding this toolchain to poky recipes Mar 14 13:39:00 and then set GCCVERSION to 6.4 Mar 14 13:39:14 but this would change the setting for the whole build Mar 14 13:39:53 The question is if anybody tested setting PREFERRED_VERSION_${PN} = 6.4 for a single recipe Mar 14 13:46:11 lukma: Sadly that won't work as changes in a single recipe can only really affect that recipe itself, gcc-native will have already been built Mar 14 13:46:34 don't suppose anybody has an example of using multiple kernel build configurations in one build , with one having an initramfs containing kexec for "chain loading" the fully-fledged kernel? Mar 14 13:48:31 paulbarker: Ok, so you confirmed my understanding Mar 14 13:48:48 It may be then best to forward update gcc to 6.4 Mar 14 13:48:59 and rebuild everything ...... Mar 14 13:50:26 paulbarker: Thanks for clarification Mar 14 14:09:23 tz: the current recipes and such around initramfs should be close to what you need Mar 14 14:10:10 essentially you can create a second kernel recipe Mar 14 14:10:16 Hello again: Is it possible for a recipe in a "multiconfig" build dir to depend on a recipe that is not in a "multiconfig" (e.g. just part of the normal local.conf, not multiconfig/mymulticonfig.conf)? Mar 14 14:10:38 Also, what is the canonical way for a recipe to "get the output of" one of the recipes it depends on? Mar 14 14:12:37 yep, thanks for that... seems like it might be feasible to adapt what I need Mar 14 14:16:04 tz: for the case where the initramfs will be embedded in the kernel and the initramfs image includes kernel modules we essentially build two kernels Mar 14 14:16:22 anyway, a starting point Mar 14 14:17:03 been a long time since I have touched that code, there were lots of gotchyas in there Mar 14 14:22:37 fair enough, it's confusing me as well, so it's good to know it's not just me ;) Mar 14 15:05:05 I'm trying to create a new directory and populate some files in it from a IMAGE_POSTPROCESS_COMMAND -- I am able to see the files show up in my tmp/work/.../rootfs directory but they don't appear in the resulting image. Anyone know why? Mar 14 15:05:40 jtrimbl3: you're doing it after the *image* is built (i.e. .ext4), too late Mar 14 15:05:51 you want ROOTFS_POSTPROCESS_COMMAND, which is after rootfs population but before *image* creation Mar 14 15:11:42 kergoth: Aha that makes perfect sense. Mar 14 15:13:17 there are also preprocess commands, but not needed very often Mar 14 15:17:34 To my other question before: If recipe A depends on recipe B, is there a canonical way for A to "find" B's outputs in tmp/work ? Mar 14 15:26:26 So in latest_versionstring (within bitbake/lib/bb/fetch2/wget.py) the 3rd argument is "d". When I var_dump "d" I get that it's a DataSmart, but no other runtime info. Mar 14 15:26:40 Kind of hard to search documentation for the letter "d"... Mar 14 15:27:12 Striking7: d is also a bb.DataSmart instance, see bb/lib/data for the api Mar 14 15:27:19 Anyone know how I can tell what sort of goodies I can find within "D"? The debugger in Wing IDE seems not to want to tell me anything about it Mar 14 15:27:20 s/also/always/ Mar 14 15:27:24 Thanks rburton Mar 14 15:37:21 New news from stackoverflow: Why can't a Yocto SDK build a Yocto SDK? Mar 14 16:03:11 is Armin Kuster on this channel? It's regarding the lttng-* update patch for the stable release. Mar 14 16:38:26 hey! is there a way to modify packageconfig for for a package from a recipe for another package? e.g. set config from recipe_a for recipe_b, which is a dependency to recipe_a. Mar 14 16:38:55 sashko: no Mar 14 16:39:30 rburton: thank you Mar 14 17:47:09 I have inherited a distro/image where there are a lot of inter-layer require statements. When I then attempt to run the distro on another machine things fail due to missing recipes. However, this happens even if these are recipes which is not being used. What is a good approach for dealing with this problem? Mar 14 21:23:17 paulbarker: you build the netns package in meta-virtualization layer since you submitted your changes back in Jan? Mar 14 21:24:35 marka: Yes, worked for me on thud & master last I checked. What's the issue? Mar 14 21:24:51 I am seeing a build fail Mar 14 21:25:17 I haven't tried to peel off commits to 'reverse time' and find when it busted Mar 14 21:25:35 On master or thud? Mar 14 21:25:39 master Mar 14 21:26:24 I want to get it working before I start doing some go related cleanup across all the meta-virtualization GO recipes Mar 14 21:28:18 Could you pastebin the initial output from bitbake (showing the layer hashes, target MACHINE, etc) plus the error you're seeing? Mar 14 21:29:22 paulbarker: sure. You aren't on the hook for fixing this, I was going to take care of it. I was mostly curious if you had seen the error and possibly had a fix in the works. Mar 14 21:29:44 I just want to make sure it doesn't bite me! Mar 14 21:31:58 At the least I can see if I spot anything different from my setup where I know it's working Mar 14 21:32:44 the error is here https://pastebin.com/aP8neERj Mar 14 21:33:23 and the configuration is here https://pastebin.com/Gx5uLbzF Mar 14 21:35:30 Ah ok. I've not currently got any x86_64 targets in our nightly builds Mar 14 21:36:41 paulbarker: we are a good pair then Mar 14 21:37:06 I usually build too much x86_64 and in the process of doing more ARM builds Mar 14 21:37:15 I have x86, arm & arm64 but not x86_64 Mar 14 21:37:58 that is why I am doing some GO cleanup, the two patches i have out to oe-core and a series I want ready for meta-virt when Bruce gets back from vacation Mar 14 21:38:46 anyway, I will sort through this, thanks for the responses Mar 14 21:39:08 and if I do manage to break anything along the way be sure to ping me Mar 14 21:39:16 Ah excellent. netns is due an upgrade but v0.5.0+ seem to have different behaviour which breaks my use case. I've not had time to look into it due to other commitments Mar 14 21:40:11 ya, I will hold off on the uprev, my goal was just to have it build before I made the global changes affecting all the go recipes Mar 14 21:40:19 If you get a patch you want me to test then just ping me Mar 14 21:40:23 can't start without a good foundation Mar 14 21:40:32 I can throw it through our build agents Mar 14 21:40:43 for sure, I can fire it off to you before the list Mar 14 21:40:48 I should have something tomorrow Mar 14 21:40:54 just about to wrap up for the day Mar 14 21:41:01 Please do. It's a core part of our container setup for Oryx Mar 14 21:41:56 since you care about ARM and go feel free to take a look at the two patches I have out on oe-core Mar 14 21:42:19 you might want to give them a spin before they are merged Mar 14 21:43:27 * marka flees Mar 14 21:45:19 `go.bbclass: Export GOARM` and `goarch.bbclass: set TARGET_GOARM as '7' for cortexa7`? Mar 14 21:49:09 Ah sorry, there's a v2 of that pair Mar 14 21:52:45 halstead: I'm getting 502 Bad Gateway from https://git.openembedded.org/ Mar 14 21:53:08 paulbarker, Let's see. Mar 14 21:55:12 Hi Does anyone know if there is a reason why the devtool build command doesnt run 'deploy' task? Mar 14 21:55:12 Right now its only running build task, populate sysroot, packagedata Mar 14 21:55:34 paulbarker, It's hit swap and is having some trouble. I'll get it fixed. Mar 14 21:56:04 halstead: Thanks :) No rush for me Mar 14 22:13:39 paulbarker, git.openembedded.org is available again. Mar 14 22:16:40 halstead: Thanks! Mar 14 22:18:00 paulbarker, Thanks for the report. It seems it should have reported the beginning of the issue about 4 hours ago but monitoring didn't report. Mar 14 22:23:50 I've recreated the monitoring rule. We should be set now. Mar 14 22:31:25 halstead, for the AB ? Mar 14 22:31:51 thhhaanks ?? Mar 14 22:31:59 armpit, This was for https://git.openembedded.org/ **** ENDING LOGGING AT Fri Mar 15 02:59:57 2019