**** BEGIN LOGGING AT Thu Jul 19 03:00:02 2018 Jul 19 07:42:18 Hi, I'd like to change the do_deploy_append() function in layers/poky/meta/classes/kernel-devicetree.bbclass. I think the right way is to reimplement the function in layers/meta-muos/recipes-kernel/kernel-devicetree/kernel-devicetree.bbappend Jul 19 07:43:03 But I get an "ERROR: No recipes available for..." What's the right way to extend a class in yocto Jul 19 07:43:32 Maybe inherit from the class? But without appending how do my layers see the changes? Jul 19 09:21:08 Hello. Jul 19 09:21:28 New news from stackoverflow: yocto linux systemd daemon Jul 19 09:29:08 I'm using meta-swupdate-boards wich have a .bbappend file for recipe-extended/images/update-image .the recipe use a file sw-description .I wish to use a sw-description from my layer instead. I tried making a .bbappend with a new SRC_URI but when building the recipe try to use the old and the new SRC_URI ... can somebody help please ? here is my small .bbappend : http://dpaste.com/11VZ90Z Jul 19 11:25:34 Hi, I'm using BB_DONT_CACHE variable set to "1" in one of my recipes, and this makes bitbake to force parsing this recipe multiple times in the same bitbake session, I think before each Task execution. Jul 19 11:27:02 sorry I'll write again, I'm using BB_DONT_CACHE variable set to "1" in one of my recipes, and this makes bitbake to force parsing this recipe multiple times in the same bitbake session, I think before each Task execution. My question, is there a way to force parsing this recipe only one time at the begining of each bitbake session, but not anymore during the execution? I only thought about stamp files! Jul 19 12:10:10 hello, trying to produce a packages for each file found somewhere in source... looking at (openembedded's) python recipe, it has block of python() { ... } that does just that (but no task is specified). when is this block run ? Jul 19 12:31:58 ak77: after parse, before build Jul 19 13:36:42 sorry I got a connection error! I was asking that I use BB_DONT_CACHE in one of my recipes, and this causes multiple recipe parsing during a single bitbake session before each TASK execution!! Is there a way to force bitbake to keep parsing this recipe at the beginning of each bitbake session but not anymore? so no multiple parsing in one session? I only thought about file stamps! Jul 19 14:25:13 I do no see anyone on the zoom meeting... (( Jul 19 14:28:16 Well, the zoom is going to begin shortly. The guys are in. Thanks. Jul 19 14:35:12 sjolley: can I get hte link for the meeting? Jul 19 14:35:24 I have the old invite Jul 19 14:36:42 sjolley: nvm I got it Jul 19 14:37:59 god it feels like the a-team in here now Jul 19 14:39:05 https://zoom.us/j/454367603 Jul 19 14:39:20 This is the Zoom live (currently) meeting. Jul 19 15:04:08 rburton: haha Jul 19 15:04:41 aehs29: paul is in europe briefly so he joined triage for the first time in years too Jul 19 15:05:51 rburton what middle of the night triage is no fun for him? slacker Jul 19 15:13:29 rburton: haha I didnt know, I was miraculously able to wake up on time today Jul 19 15:16:22 fray: i know right?, no shame Jul 19 15:54:05 Does anyone have experience using the useradd bbclass in Rocko? I have USERADD_PACKAGES, etc. all set and my recipe seems to run fine. The prepare sysroot task indicates it ran useradd without error. However the user doesn't exist on the target image. Jul 19 15:54:26 (I followed the meta-skeleton example) Jul 19 15:54:49 did you include the package produced with the useradd into your image? Jul 19 15:54:57 Yes Jul 19 15:54:58 prepare sysroot is only for the build side.. Jul 19 15:55:27 if you used rpm packages, you can simply query the package and check if the useradd was in place.. go to your tmp/deploy/rpms ... rpm -qp --scripts Jul 19 15:55:59 if it's deb or opkg you can use tar and ar to break them up and look as well.. (sorry don't know an automated way to do that) Jul 19 15:56:10 point is, look to see if the recipe you think is doing this work, actually has it in there... Jul 19 15:56:23 once you verify it IS, then move onto the image and verify that recipe was actually installed.. Jul 19 15:56:52 assuming it was, you may have to play around and add your own install scripts to try to instrument it Jul 19 15:57:11 (but usually it's one of the first two issues.. either not loaded into the recipe or not installed) Jul 19 16:02:37 That's the thing. I'm not sure the useradd class actually puts anything in the package since it's editing the /etc/passwd and /etc/group in the recipe-sysroot. I don't see where that class then moves/delivers those changes. Jul 19 16:02:55 the recipe-sysroot is only edited for the build steps.. Jul 19 16:03:08 the class -will- change the packages that are produced.. if it does not, something is not configured prooperly Jul 19 16:03:31 purpose of the useradd package stuff is to make the addition of users dynamic on package installation Jul 19 16:03:53 thats why the first thing to check is the package itself, if the scripts are not inserted, then the recipe is not configured properly Jul 19 16:04:03 (even if it worked for the sysroot) Jul 19 16:04:55 (sysroot generation doesn't know which packages will be created, so if the pattern is valid it'll create the users.. but if the package is never generated, then there is no script inserted) Jul 19 16:09:09 I see the script it generated in the preinst of the control tarball. Jul 19 16:09:54 good.. then it's an issue of making sure that package is installed.. if it is.. I'm not sure what to suggest next.. (it quickly gets into some very deep debugging to verify everything is actually running) Jul 19 16:09:58 And the package is in the image's manifest Jul 19 16:17:57 Looks like I had to add a file to the package since it's otherwise empty. I used a directory for kicks and used install -m 700 followed by changing the ownership to that user, but the installed directory was 755... Jul 19 16:28:28 fray: thanks for the sanity check. Now for the mystery of why "install -m 700" isn't being honored for the directory (user's home directorY) Jul 19 16:44:38 the system will automatically 'correct' certain directories.. Jul 19 16:45:31 look at meta/files/fs-perms.txt Jul 19 16:48:03 Interesting. Well, that's too bad that it forces all of home to be 755 and only sets /home/root to 700. Jul 19 16:49:06 you can (and should) provide your own override version of that file.. Jul 19 16:49:11 Right Jul 19 16:49:17 it doesn't have to be the WHOLE file, just the one line you want changed.. Jul 19 16:49:20 I was noticing that in bitbake.conf, FILESYSTEM_PERMS_TABLES. Jul 19 16:49:34 then update your system to load yours last.. last in wins Jul 19 16:53:48 I'm not sure i follow your "WHOLE file" comment. It looks like the package bbclass only pulls in one file, so if mine is a one-liner then none of the others from the original will work. Jul 19 16:54:19 no, it reads from a list of files to load.. FILESYSTEM_PERMS_TABLES.. Jul 19 16:54:26 append to the list you variant.. Jul 19 16:54:38 Ah, overlooked that. Jul 19 16:54:39 it reads the files in order and applies the changes to an in memory table Jul 19 16:54:42 the last in wins Jul 19 16:54:54 thus you never have to copy the whole thing Jul 19 17:25:36 fray: appending my own didn't work. I'm also confused why I would need to though. In re-reading that fs-perms.txt file, it is set not to walk the /home directory. Shouldn't the subdirectory my package installs be unaffected then? Jul 19 17:26:02 ya, then it should be fine.. Jul 19 17:26:08 it's the walk setting you'll care about Jul 19 18:10:50 when building a kernel, linux-xyz, does yocto/bitbake look for linux-xyz_%.bbappend files in all subdirectories of all layers, or are the directory names relevent? Jul 19 18:13:39 e.g., if i had meta-mylayer/a-funky-linux-kernel-subdire/linux-xyz_%.bbappend file would it get picked up and applied? Jul 19 18:17:03 so after reading the man i think it's the former Jul 19 18:17:32 right>? Jul 19 18:18:21 I think it depends on the layer... see BBFILES in layer.conf Jul 19 18:27:44 did you mean in bblayers.conf? Jul 19 18:27:57 build/conf/bblayers.conf? Jul 19 18:29:34 i have BBFILES in both build/conf/bblayers.conf and in meta-mylayer/conf/layer.conf Jul 19 18:29:42 so this? BBFILES += "${LAYERDIR}/recipes*/*/*.bb ${LAYERDIR}/recipes*/*/*.bbappend" Jul 19 18:30:13 would pickup ann directories in any subdirectory beginning with "recipes"? Jul 19 18:30:25 s/ann/all/ Jul 19 18:44:32 Is there a way to have wic generate an empty partition? Jul 19 18:45:35 yates: Doesn't really matter. .bbappends will only be found if in BBFILES, regardless of where that comes from, at least AFAIK Jul 19 18:46:52 yates: To your later question, yes the example you gave would only find .bbappends in recipes* directory. I think thats pretty standard methodology in layers Jul 19 18:56:27 true. devtool's workspace differs, but most don't Jul 19 19:08:41 JPEW: ack. thanks. Jul 19 19:15:27 is there a way to see all the files that were put onto the rootfs? Jul 19 19:16:32 a manifest or somesuch? Jul 19 19:16:52 Is TPM easily supported in Yocto? Curious, have others used it? Jul 19 20:52:35 again i'm totally confused. Jul 19 20:53:21 i have a layer (meta-swupdate) that provides some components which can be instsalled into an image/rootfs Jul 19 20:54:03 how do i instruct my image to install those? Jul 19 20:55:07 it apparently isn't as easy as creating a bbappend for my image: https://paste.fedoraproject.org/paste/yPNyDBGP7AaYT7E-NQR9Vw Jul 19 21:01:28 is IMAGE_INSTALL the correct list to append? Jul 19 21:04:39 is says here that recipes set IMAGE_INSTALL in an image's .bbclass: https://www.yoctoproject.org/docs/1.8/ref-manual/ref-manual.html#var-IMAGE_INSTALL Jul 19 21:05:02 then why is this guide saying to put it into a .bbappend: http://sbabic.github.io/swupdate/swupdate.html#features Jul 19 21:24:13 yates: I'm a bit rusty on yocto stuff, but my money is on an order of operations issue. bbappend should work Jul 19 21:24:58 += operator happens immediately, whereas _append happens late. if you have a += followed by an =, it will be overwritten Jul 19 21:26:04 bitbake -e | less should tell you whether it truly is an order of operations issue Jul 19 21:42:51 clsulliv1: thanks Jul 19 21:44:37 here is part of the -e output: https://paste.fedoraproject.org/paste/yEEhOc0e66tIwF0aqDjv5A Jul 19 21:45:40 ..and I don't see swupdate in there... Jul 19 21:47:00 i also do not see the bbappend there that should be appending the IMAGE_INSTALL Jul 19 21:50:02 my .bbappend is in the layer sources/meta-ebtron/recipes-kernel/linux-swupdate, and linux-swupdate is nowhere in the -e output Jul 19 21:50:10 why isn't it picking it up? Jul 19 21:51:31 ?? Jul 19 21:53:16 https://paste.fedoraproject.org/paste/ChUzb3Ahlq0ZuoEAlsVCpQ Jul 19 21:54:02 but the wildcards in BBFILES do cover it.. Jul 19 21:56:16 i tried an unusual thing here. i have two linux-variscite_%.bbappend files in my layer, one under meta-ebtron/recipes-kernel/linux and one under meta-ebtron/recipes-kernel/linux-swupdate. will that work? Jul 19 21:56:44 i was hoping they would both be applied Jul 19 22:13:32 yates: are you trying to bbappend the kernel recipe or image recipe? IMAGE_INSTALL is specific to images, so it should be appended to your hw-test-image Jul 19 22:30:40 doh! Jul 19 22:32:57 clsulliv1: thank you thank you thank you. Jul 19 22:51:02 that took a special kinda' stupid.. Jul 20 01:24:38 New news from stackoverflow: Bitbake QA requires deletion of {bindir} **** ENDING LOGGING AT Fri Jul 20 03:00:01 2018