**** BEGIN LOGGING AT Thu Nov 13 03:00:00 2014 Nov 13 09:25:09 good morning Nov 13 10:06:51 morning all Nov 13 10:37:06 <_manuele_> Hi All, I did a meta to deploy a image with a kernel patch. But when I populate sdk header patches are missing Nov 13 10:38:17 <_manuele_> to add patch I added in meta-myimage the file recipes-kernel/linux/linux-yocto_3.14.bbappend Nov 13 10:39:24 <_manuele_> what is wrong? Nov 13 10:41:58 hello I'm having trouble building a kernel in daisy, when i try "$bitbake linux-my-confl" i get "error my_kernel was skipped; incompatible with machine my_mach (not in COMPATIBLE_MACHINE)", i did set MACHINE_COMPATIBLE in both source/meta-me/recipe-kernel/linux/linux-my-conf.inc and .bb, what am I missing ? Nov 13 10:43:22 manuele, did you create the folder recipes-kernel/linux/linux-yocto_3.14/ and put your patch in ? Nov 13 11:03:46 where do i look to fix "not in COMPATIBLE_MACHINE" issue when trying to create recipe for a new kernel version ? Nov 13 11:08:22 Guest20990: that's mentioned in our kernel manual Nov 13 11:08:47 Guest20990: but you should be able to see in various BSPs how we add to COMPATIBLE_MACHINE Nov 13 11:24:16 Morning Nov 13 11:25:43 Does anyone understand the useradd functionality? I was looking at meta-skeleton/recipes-skeleton/useradd/useradd-example.bb and curious as it chown's home directories in do_install() but thats a chown on the development machine in the image dir so wouldnt that take the UID for the username on the dev system not the UID on the target? Nov 13 11:26:45 (the backstory is that I need to create a mount point on the target fs with specific user and perms but I cant use a pkg_postinst as it's a read-only rootfs) Nov 13 12:05:29 pev_: well, one thing to remember is that we rely on pseudo to get users & permissions right Nov 13 12:05:43 that mostly ensures that you get users from the target system being built rather than the build host Nov 13 12:08:55 bluelightning: Ah.... yes! That makes sense, thanks... Nov 13 12:09:01 so it should just work. Nov 13 13:04:06 pev_: indeed Nov 13 13:40:56 Is there some trick to creating an empty directory in a do_install_append() ?? Nov 13 13:55:37 pev_: install -d ${D}/yourpath/yourdirectory/ Nov 13 13:59:47 mckoan: that's what I did but for some reason never appears.... Hmmm. Nov 13 14:02:06 pev_: in the target filesystem of course Nov 13 14:05:28 mckoan: Yeah, it was - its something kooky in the recipe, i put some junk in my do_install_append() and it doesn't barf so looks like it isnt being used for some reason. Nov 13 14:06:00 The .bb is definitely being used as if i put junk in IMAGE_INSTALL it fails. Nov 13 14:06:15 I though a do_install() or do_install_append() would be unconditional? Nov 13 14:08:31 if I take the same and do in a different recipe it works?! Nov 13 14:28:58 Hi all. I use yocto from few days ago and I noticed that for each compilation, sdk is regenerated. But sdk compilation cause problems if the image is not completely generated. So, someone know how to disabling sdk generation for an image compilation ? Thank you in advance Nov 13 14:30:22 Because, for the moment, I relaunch the compilation after each error, and it is very boring Nov 13 14:31:04 how are you kicking off compilation? because a standardimage build should not generate the sdk Nov 13 14:33:51 bitbake my-devkit-image Nov 13 14:34:55 During the first compilations, bitbake did not generate the sdk Nov 13 14:35:20 dcyrille18: that's normal bitbake iot-devkit-image will only generate the image and whatever of the SDK it needs Nov 13 14:35:43 you probably want bitbake -c populate_sdk Nov 13 14:36:26 dcyrille18: hm, are you maybe confusing "sdk" and "toolchain the image itself needs"? Nov 13 14:36:57 Yes, I generated the sdk , there a few days , using this command in effect Nov 13 14:37:38 And since, bitbake generate image and sdk, for each recipe update Nov 13 14:38:21 dcyrille18: I dont really understand, you'll need to regenerate the SDK & image yes when you change a recipe (depends on the recipe of course) Nov 13 14:38:49 I want to just regenerate the image but not the sdk Nov 13 14:40:09 dcyrille18: well bitbake iot-devkit-image will let you do that Nov 13 14:40:21 I build that image all the time and rarely build an sdk Nov 13 15:04:20 Sorry for my absence. So, I said : Nov 13 15:04:22 In fact, my problem is that at start, bitbake generated me only the image. That's good. Then I generated the sdk for a colleague with the command " bitbake my- devkit-image -c populate_sdk". Since, for each update, bitbake recompile the image and the sdk while I need just the image. Nov 13 15:13:43 dcyrille18: oh that is weird Nov 13 15:14:45 dcyrille18: what targets are you using exactly? Nov 13 15:20:16 dcyrille18: in that case, don't run -c populate_sdk image, just bitbake image Nov 13 15:20:36 dcyrille18: bitbake assumes since dependencies of the SDK (in this case the image) have changed , then it needs rebuilding Nov 13 15:20:46 but if you don't try to build the SDK at all, it won't get rebuilt ;) Nov 13 15:22:43 ok thank you for your responses Nov 13 16:34:54 <_manuele_> How I can apply a configuration to kernel ? Nov 13 18:48:11 How can I get a recipe to re-download an archive? I do 'bitbake -c clean ', but it still uses a cached copy of the archive. Nov 13 18:52:37 ekw_: bitbake -c cleanall Nov 13 18:53:02 -c clean -> cleand the WORKDIR, but not sstate, not the downloads Nov 13 18:53:18 -c cleansstate -> clean WORKDIR and sstate Nov 13 18:58:17 ndec: cleansstate worked. thanks. Nov 13 18:58:48 ekw_: cleansstate doesn't force a re-download, but a rebuild without content from sstate cache. Nov 13 19:15:37 ndec: hmm ok. I'll try cleanall. Nov 13 19:23:10 ndec: ok cleanall works. I don't know what was happening with cleansstate. It also appears to re-download, but caused image build to behave strangely. Which is probably what you were trying to tell me, though I didn't quite understand what that meant. =) Nov 13 22:47:26 Is it possible to have a shell function call a python function with a variable (something not expanded by bitbake but provided by the shell function) as an argument? Nov 13 22:57:03 nope. shell can't call python at all, variable or otherwise **** BEGIN LOGGING AT Fri Nov 14 01:14:24 2014 **** ENDING LOGGING AT Fri Nov 14 03:00:00 2014