**** BEGIN LOGGING AT Sat May 01 03:00:43 2021 May 01 04:06:15 jonesv: how powerful is the cpu ? May 01 08:04:51 I have a question about the release process of meta-openembedded. I saw that a recipe for yaml-cpp exists on the master branch, but not for dunfell or others. Are new recipes still ported back for dunfell or do you have to upgrade the build system to a newer release? May 01 12:31:30 it's an apq8009... do you think that it may be too slow to boot yocto? 🤔 May 01 12:33:43 The thing is that I replaced the `system` by the one I built with yocto, and I kept `boot` intact. So my understanding is that I kept the downstream kernel but just changed the rootfs. Now I'm not sure if `Starting udev` means that the kernel did recognize my rootfs at all or not... it's an ubi system and maybe I screwed up the setup. I just don't know what to do from here, I'd like to see more output from the kernel to May 01 12:33:43 understand maybe what is wrong May 01 15:17:25 jonesv: can you change kernel cmdline parameters ? May 01 15:17:43 if so you might add printk to it May 01 15:18:41 your CPU is powerful enough that it should finish udev quickly enough May 01 16:17:42 khem: I'll try, I'm now reading to see where the cmdline is stored (like does it go on the bootloader partition `aboot`, the kernel partition `boot` or the rootfs partition `system`? Because if I change it in the bootloader but only flash the kernel partition, that's useless :D May 01 16:18:09 I'm just scared of flashing anything at all on `aboot`, because if that fails, I won't be able to flash anything anymore May 01 17:15:09 khem: do you have enough knowledge of elf to tell if http://git.yoctoproject.org/cgit.cgi/poky-contrib/patch/?id=9b4812f8b51fc5ffd1c22981fcdd7426e035c556 is right? May 01 17:39:37 hi all -- what should I add in my recipe to have the (native) fallocate tool available? May 01 17:45:16 v2d: looks like it is part of util-linux May 01 17:49:51 RP: util-linux-native then? :s May 01 17:50:00 v2d: yes May 01 17:50:04 I meant DEPENDS = "util-linux-native" May 01 17:51:14 v2d: that is where I'd start... May 01 17:51:19 RP: I'm writing a simple recipe to generate a .vfat partition. Ideally we need a "vfat" IMAGE_FSTYPES imo. Would you be interested in that? May 01 17:57:53 v2d: raw partition copies often aren't as useful as you'd think as they tend to need to be combined into varying sized discs so maybe but I'd want to understand the way it could be used May 01 18:01:15 RP: for a concrete scenario, the RAUC update utility can upgrade redundant bootable MBR partition, by copying the raw partition and tweaking the MBR active flag atomically to the updated boot partition. May 01 18:02:00 So one could imagine doing a recipe like this: IMAGE_INSTALL = "virtual/bootloader" IMAGE_FSTYPES = "vfat" May 01 18:05:27 If a "vfat" image type is not acceptable, one can still write a recipe doing mkfs.vfat and such, or we can imagine the "bundle.bbclass" from meta-rauc to do it for us. May 01 18:06:57 v2d: if there is a user for it, we can probably add it, I doubt it costs us much to maintain May 01 18:08:39 WIC may use it for part --fstype vfat as well May 01 18:18:07 RP: you should get multiple SHT_NOTE one for each alignment e.g. 4. and 8 May 01 18:18:55 which is what I am seeing in comment in your patch May 01 18:20:32 so all .note.XXX sections which are aligned to 4 bytes will go to Second SHT_NOTE and 8bytes one go to first one May 01 18:23:34 RP: this is my recipes-core/partitions/boot.bb recipe: http://ix.io/3lvR (unfortunately I have no mean to abstract the barebox image names) May 01 18:25:04 RP: most probably patchelf should just ignore checking section to segment checks for notes May 01 18:28:20 Does adding a task as a dependency like do_foo[depends] += "recipe:do_bar" is enough to pull in "recipe", or does "recipe" need to be added to *DEPENDS as well? May 01 18:30:59 RP: Just looked into lld which does the section clubbing and I think your patch is right too, so either we can ignore or if we want to resync then your patch improves the situation, May 01 18:32:06 jonesv: I think rootfs should be able to boot ok unless your kernel depends on modules from rootfs and now you would have newer/different modules than the kernel in /boot so it might not be loading them May 01 20:15:05 khem: https://autobuilder.yoctoproject.org/typhoon/#/builders/89/builds/3373/steps/16/logs/stdio - different mingw issue with gcc 11 May 01 20:19:19 hmm I did build core-image-mingw-sdktest -cpopulate_sdk yesterday for qemux86 and SDKMACHINE = i686-mingw32 May 01 20:19:53 I see here you have SDKMACHINE = "x86_64-mingw32" May 01 20:20:06 who uses mingw stuff ? May 01 20:25:10 I have zero use of it and it always needs fixing, Feels like waste of time I wish there was a way to keep it broken and let interesting parties fix it May 01 20:27:19 I have seen such free labor goes unnoticed May 01 20:40:55 khem: JPEW does May 01 22:12:45 RP: sent a fix please try it out May 01 22:38:34 khem: thanks, will test May 01 22:39:00 khem: I think we need to rethink how we're handling meta-mingw and meta-gplv2 as they cause us both trouble :/ May 01 22:47:36 I still have "fallocate: not found" from my recipe while I DEPENDS = "util-linux-native"... Am I missing something? May 01 22:48:42 I also can see build/tmp/sysroots-components/x86_64/util-linux-native/usr/bin/fallocate, so what's the issue here? May 01 22:59:06 v2d: in the recipe that is missing it, is it in recipe-sysroot-native? May 01 23:01:22 RP: I am not sure to understand May 01 23:04:19 RP: I have no such thing like */my-boot-part-recipe/*/recipe-sysroot-native/usr/bin/fallocate May 01 23:04:29 but I do for other image recipes May 01 23:12:27 v2d: you added DEPENDS = "util-linux-native" to my-boot-part-recipe but you don't see fallocate there? May 01 23:12:47 RP: exact May 01 23:12:52 v2d: that seems strange and I'd check with bitbake -e that DEPENDS is set to what you think it should be May 01 23:13:31 ok May 01 23:14:02 It used to work until I added this recipe as a dependency from another (wic) image recipe. May 01 23:17:31 RP: I've added the recipe to EXTRA_IMAGEDEPENDS and do_image_wic[depends], do somehow it messes around with task ordering. I've changed the add_task statement to specify "after do_prepare_recipe_sysroot" and it works now. Normal? May 01 23:17:46 s/do/but/ May 02 01:17:11 Heyho, I'm on yocto dunfell and I'm seeing two (maybe one?) issues: 1) `devtool modify linux-yocto` results in a src tree in my workdir which has no patches applied. All devtool* branches are on the SRC_REV even though my recipe has patches. 2) If I `git am` a patch into that tree which adds a Kconfig option, this option is not honoured even if specified in the recipe along with others which are May 02 01:17:11 honoured May 02 01:18:03 Any tips/ideas on how to dig into this to sort out where the patches are being applied on devtool modify, and if this is related to the Kconfig being ignored stuff? **** ENDING LOGGING AT Sun May 02 02:59:56 2021