**** BEGIN LOGGING AT Mon Jun 17 02:59:58 2019 Jun 17 06:48:28 good morning Jun 17 07:52:47 mckoan: morning :) Jun 17 08:13:16 hey alicef ;-) Jun 17 08:14:00 tutto bene ? :) Jun 17 09:10:05 for a given recipe, is it possible to get a list of layers that .bbappend it? right now i'm using `bitbake -e whatever` and grepping for bbappend but 1) that's still a lot of output 2) i don't know if i'm getting everything Jun 17 09:14:12 milloni: bitbake-layers show-appends Jun 17 09:15:04 nice, i had actually asked that question before, i'm hoping to remember the answer this time :D thanks Jun 17 09:52:43 hi Jun 17 09:52:59 I need to fetch zip file from artifactory using custrom header + api key Jun 17 09:53:18 is there some support for that or should I use do_fetch() to run curl with that? or even write new fetcher? Jun 17 10:12:19 can you not embed the key in a url parameter instead? Jun 17 10:16:10 rburton: hmm it needs als ospecial header like described here: curl -H "X-JFrog-Art-Api:ABcdEF" -X PUT ... Jun 17 10:17:49 write a new fetcher Jun 17 10:18:05 would be great if the http fetcher was redesigned a bit so you could trivially subclass it to add new headers Jun 17 10:18:10 maybe you can already, worth a look to see Jun 17 10:19:27 I checked wget fetcher and this support header for username:password Jun 17 10:19:37 ok I'll try to look what I can do :) Jun 17 10:19:41 thanks for info Jun 17 10:51:26 Hi! I'm getting "No rule to make target 'menuconfig'" and "No rule to make target 'Image'" for my kernel build with thud. Jun 17 10:51:30 The same operations work just fine if invoked from devshell instead of bitbake. Jun 17 10:51:34 Any ideas? Jun 17 10:53:05 frsc: so you are trying to do "bitbake menuconfig" or "bitbake image"? or what do you mean? Jun 17 10:56:18 LetoThe2nd: "bitbake virtual/kernel -c menuconfig" or just "bitbake virtual/kernel" Jun 17 10:56:47 And "make menuconfig" or just "make" from devshell work fine Jun 17 10:57:30 frsc: if i had to guess, i'd say that maybe something with your directories is off. but other than that, no idea right now. Jun 17 11:02:22 LetoThe2nd: Could be. I'm building from external source via devtool. After "devtool reset" it's now working again. Jun 17 11:08:02 LetoThe2nd: Hm, it doesn't seem to be related to devtool, but to the kernel source. The problem only occurs with Linux 5.2-rc5, but not with 5.1.10. Jun 17 11:30:02 hmm, I don't understand this. I'm adding "plymouth" to my initramfs. and the "plymouth" package is not apparently dependent on "dracut" Jun 17 11:30:47 but plymouth-initrd is. and I'm not including that package, neither is "plymouth" apparently dependent on "plymouth-initrd" Jun 17 11:31:32 however nevertheless, bitbake complains that my initramfs image recipe depends on "plymouth-initrd" and says "Missing or unbuildable dependency chain was: "['core-image-lmg6xx-initramfs', 'plymouth', 'dracut']" Jun 17 11:32:33 litb: maybe the dependency is there but just not apparent? Jun 17 11:32:34 why? the dependency declaration is this: RDEPENDS_${PN}-initrd = "bash dracut" . However, I'm merely *building* it, but I don't install it. How is it that bitbake wants to build dracut then? Jun 17 11:33:00 LetoThe2nd, I think that's the case, but it's not apparent to me yet Jun 17 11:33:35 can't I tell bitbake "I know that plymouth-initrd needs dracut at runtime. but I'm not installing plymouth-initrd, so please don't build dracut" Jun 17 11:34:10 Or is it that it just wants to have verified that dracut is *declared* somewhere? I.e i just need to add its layer? Jun 17 11:35:14 litb: you are building core-image-lmg6xx-initramfs, right? Jun 17 11:35:37 hah, I found a way to have this work without adding the respective layers. I can remove "initrd" from its "PACKAGECONFIG" list. because it creates that package "-initrd" conditionally Jun 17 11:35:43 LetoThe2nd, yeah Jun 17 11:35:49 litb: what happens if you are building plymouth directly? because you obviously can't build the full image if the rdepends in not fulfilled Jun 17 11:37:27 this in my distro config will do it I think PACKAGECONFIG_remove_pn-plymouth = "initrd" Jun 17 11:41:50 LetoThe2nd, hmm why can't I? Jun 17 11:42:09 after all, I wouldn't install the .deb that has the unfulfilled rdepends Jun 17 11:43:35 litb: if you tell bitbake to build an image for which it can't fulfill the complete dependency tree, be it runtime or buildtime, then i don't think it would go ahead and try, but abort and tell you so. Jun 17 11:43:43 (of course i might be wrong, just as usual) Jun 17 11:44:41 LetoThe2nd, hm, but in my case, I'm fulfilling the dependencies in the image, as far as I can see Jun 17 11:45:20 because the image doesn't contain plymouth-initrd, so there's no dependency, neither build nor runtime, of any package in the image, to dracut Jun 17 11:45:40 I think i'm just think-blocking today Jun 17 12:02:18 So the kernel build problem described above is caused by the following patchset in v5.2: Jun 17 12:02:19 https://patchwork.kernel.org/patch/10878447/ Jun 17 12:03:29 Now, I just need to find out if Yocto/OE needs a fix to work with the new kernel, or if the kernel needs a fix to avoid breaking the build. Jun 17 12:14:57 frsc: I’m building 5.2 constantly with linux-yocto-dev, and all is fine in master Jun 17 12:17:00 zeddii: Ok, strange. Which Yocto release are you using? Jun 17 12:17:20 Ah master Jun 17 12:17:29 Sorry missed it Jun 17 12:18:45 zeddii: Did you already try devtool with out-of-tree build? Jun 17 12:20:54 naw. I never use devtool. Jun 17 12:36:16 zeddii: I tried without devtool and the problem still persists Jun 17 12:36:38 I just built and booted 5.2-rc5 :D Jun 17 12:37:18 zeddii: I guess I have to try master and see if there's some change in there that I miss in thud Jun 17 12:37:32 qemux86-64 login: root Jun 17 12:37:33 root@qemux86-64:~# uname -a Jun 17 12:37:33 Linux qemux86-64 5.2.0-rc5-yoctodev-standard #1 SMP PREEMPT Mon Jun 17 12:29:47 UTC 2019 x86_64 GNU/Linux Jun 17 12:38:34 generally speaking, the newest kernels do not always build in the older branches. I’ve had to tweak a few things in master since I’ve been working through the new kernels on the way to the version for the fall release. But nothing around the make options not properly passing. Jun 17 12:45:33 Hi guys. I need to add the 'gperf-native' dependency to DEPENDS variable to build the chromium package... From the meta-browser repo I see that there are a ready patch: https://patchwork.openembedded.org/patch/137122/ . But, I'm can't checkout to this commit, because then bitbake say that "no recipes available for chromium-wayland_48', that is true because before that commit the wayland version was incremented to v53. So, how to add the 'gperf-native' to Jun 17 12:45:33 DEPENDS without of modification of original 'chromium' recipe? Jun 17 12:46:12 How to do it correctly? Jun 17 13:09:31 what's the correct way to modify a file installed by a packet? an .bbappend file with do_install_append() { /* modify the file here */ } ? Jun 17 13:17:14 zeddii: I just tried master and warrior. Both seem to work fine. So I'll switch to warrior... Jun 17 13:21:38 yah. if you have that choice, that’s probably easier than digging out what changed and trying to backport, ec. Jun 17 13:21:46 s/ec/etc/ Jun 17 13:30:50 zeddii: Yes, for now I still have that choice as so far this is still in development stage and I just avoided the upgrade to warrior till now. But now I have a good reason to switch. Jun 17 14:42:42 Hi everyone. I have a little question. Is it possible to remove package from packagegroup in image recipe ? Thanks for help Jun 17 14:43:14 No Jun 17 14:44:03 If you don't want the package in the packagegroup, remove it from the packagegroup. If you don't want to modify the packagegroup, don't install it in your image recipe. Jun 17 14:45:27 yes, but i have a packagegroup which includes or not package dependings on machine type. Example : ${@oe.utils.conditional('MACHINE', 'qemux86-64', '', 'my_package', d)} Jun 17 14:45:30 it works Jun 17 14:46:37 but we decide to backup sstatecache on server, for all targets we have, and it seems that sstate and packagegroup is not very well managed Jun 17 14:47:28 no version is set in package recipe ... maybe the problem comes from here Jun 17 14:50:34 this may be related to https://bugzilla.yoctoproject.org/show_bug.cgi?id=5970 or https://bugzilla.yoctoproject.org/show_bug.cgi?id=7298 Jun 17 14:50:37 Bug 5970: enhancement, Medium, 1.9, richard.purdie, VERIFIED FIXED, sstate signature generator issues Jun 17 14:50:38 Bug 7298: normal, Medium, 2.99, richard.purdie, NEW , packagegroups are not rebuilt when rdependency pkg name changes Jun 17 14:50:43 although I thought that should work by now. Jun 17 14:52:21 I'm not sure if sharing sstate cache between different $MACHINE targets is supported, though. Jun 17 14:57:29 I'm using sumo (2.5.2). Not sure it is present in this version Jun 17 14:59:05 thanks for your reply Jun 17 15:02:28 i will move packages which need to check the ${MACHINE} variable to my image recipe and let only package compatible with all architecture Jun 17 15:02:33 regards Jun 17 15:23:42 neverpanic: it is supported Jun 17 15:35:33 hello folks Jun 17 15:36:22 how does yocto know, when I execute "bitbake ", that it has to execute wic? If i say "bitbake ", it won't (hopefully) execute wic on it Jun 17 15:36:43 IMAGE_FSTYPES controls image construction from the rootfs. initramfs recipes generally override it Jun 17 15:36:45 * kergoth yawns Jun 17 15:36:58 lol, must be a common question Jun 17 15:37:02 thanks Jun 17 15:37:35 ahh i remember. there you can say "ext3" and "wic" and such Jun 17 15:38:49 exactly Jun 17 15:40:06 hm. we have bootimg-partition that copies certain files from the deploy-image area to /boot. but if you use bootimg-pcbios or other plugins, you can't make use of it. that's sad Jun 17 15:40:40 i'm about to copy the code in bootimg-partition into my own wic plugin, which is bootimg-pcbios copied, and extended with that code that supports copying files from the deploy area. Jun 17 15:40:48 I guess there's a way better way tho Jun 17 15:41:09 that's probably your best bet, unless the plugin had more flexibility, which i doubt. there's a fair bit of hardcoding in those plugins at the moment.. Jun 17 15:41:32 best would probably be to prep a patch to allow it to support that functionality and submit that to oe-core, and your own plugin isn't a bad way to go even if you pursue that route Jun 17 15:43:52 Why is perl under the "perl-sanity" directory? Jun 17 15:48:15 JPEW: it was done when perl was being cleaned up, it should really move backj Jun 17 15:49:19 JPEW: thanks for the reminder, patch in next to remind me Jun 17 15:49:28 New news from stackoverflow: How to I deploy Azure IoT Edge on Yocto on Raspberry Pi? Jun 17 16:49:37 New news from stackoverflow: Yocto Toaster only shows first rootfs built Jun 17 18:36:30 Hi I am trying to build a clang cross toolchain for raspberrypi2 target. I had used yocto for the build, created an SDK successfully. I am currently in the process of building libc++abi for clang Jun 17 18:36:53 But I am not able to use the SDK with -DCMAKE_CXX_COMPILER=/sdk/toolchain/bin/clang++ Jun 17 18:37:16 https://pastebin.com/Prm8uThU Jun 17 18:38:15 I am not sure how to use clang with yocto generated SDK Jun 17 18:38:45 I have set -DCMAKE_TOOLCHAIN_FILE=$OE_CMAKE_TOOLCHAIN_FILE Jun 17 20:50:12 New news from stackoverflow: Tracking variable assignments in bitbake Jun 17 21:53:15 What's the status of git submodule fetching in thud? I'm still getting intermittent unpack failures with the message "No up to date source found: clone directory not available or not up to date" Jun 17 22:28:18 paulbarker: I backported the fetcher fixes that seemed to make AGL happy, but I believe there's been some more rework since those Jun 17 22:36:07 can someone be more specific? is ref commit ? Jun 17 22:42:56 * armpit no backport requests Jun 17 22:43:36 armpit: since I backported mhatle's earlier fixes, there's been fd27ab6 and 30fe86d, it's not clear if those would be related to paulbarker's issue Jun 17 22:44:21 smurray, what repo are those hashes for? Jun 17 22:44:25 armpit: in bitbake master, I mean, sorry Jun 17 22:44:45 k, let me look at it Jun 17 22:44:54 thanks Jun 17 22:46:42 armpit: they seem to cherry-pick to 1.40 cleanly, at least Jun 17 22:47:55 thanks Jun 17 22:48:24 I've had no reports of failures with the full set of git submodle fetcher updates.. Jun 17 22:48:37 and yes, they SHOULD all cherry pick cleanly... they were designed that way Jun 17 22:51:32 fray: thanks. AGL's usage of gitsm has been reduced quite a lot in the last few months, so it's likely not stressing that code as much as it used to Jun 17 23:28:00 Is there a reason why in Warrior a devshell doesn't have the exact same env (CC/CFLAGS, ext) as during do_compile? Trying to troubleshoot a recipe and CC in devshell doesn't have thing's like '-mpu' set, which cause very different errors? Jun 17 23:38:16 stangerm2, so are you saying master behaves differently? Jun 17 23:38:53 2.2 does Jun 17 23:38:55 stangerm2, does temp/run.do_configure/complile set things correctly ? Jun 17 23:41:03 The recipe doesn't have any custom appends or rules to either do_compile or do_configure Jun 17 23:42:02 it does `inherit module` I'm not sure if that matters Jun 17 23:50:34 I'm take it to the ML, thanks Jun 17 23:54:20 k Jun 17 23:54:42 as i don't have clue **** ENDING LOGGING AT Tue Jun 18 02:59:57 2019