**** BEGIN LOGGING AT Wed Apr 13 02:59:58 2016 Apr 13 11:23:40 So whats the deal with u-boot DEPENDS, should they be on the .bb recipe or in u-boot.inc? Apr 13 11:30:15 i ask, cause i noticed that u-boot depends on 'bc-native' but I am not sure if that should be added to the versioned recipe or as a common DEPENDS in u-boot.inc? Apr 13 12:15:26 hello Apr 13 12:16:14 Does uboot for arm require includes and libs in the host computer? Apr 13 12:18:47 davis: it should compile fine with the standalone toolchain Apr 13 12:19:14 hmm. many thanks. Apr 13 12:19:27 mind, i said "should", not "must" :-) Apr 13 12:19:47 i've got a problem with a yocto vendor build which uses uboot. Apr 13 12:20:31 thats unfortunate for you, but without additional details extremely hard to help. Apr 13 12:20:32 Its failing to compile and when I examine the environment I see that it uses host libs and includes so I thought iit was a mistake Apr 13 12:21:01 its of course possible it uses some helper things. for example to prepare a logo or such Apr 13 12:21:04 but if it uses the host normally, this is not the problem. Apr 13 12:38:26 here is the intial triage of this problem. If someone could help point out what I should look at, I would be extremely happy. Apr 13 12:38:43 https://gist.github.com/netskink/90135d8774ac1548d3b14eeaed0e2514 Apr 13 13:18:20 lol... "They are looking for a very specialist resource that is a OpenEmbedded 2.0 Build (Software) Engineer. I believe this may also be known as Yocto? ". i am curious what koen might think -) Apr 13 13:20:00 this email is making the rounds Apr 13 13:20:10 * Crofton|work is sad he doesn't have a copy Apr 13 13:20:58 ndec: people have been forwarding that email to me as well Apr 13 13:21:02 hehe Apr 13 13:33:43 i don't have a copy either! Apr 13 13:33:46 * rburton sobs Apr 13 13:34:03 * rburton does the bike-almost-back-together-again dance Apr 13 13:35:02 Surely you have spare bikes Apr 13 13:35:31 rburton: bike or motorcycle? Apr 13 13:35:42 Crofton|work: spare bike is a boring hybrid for the shops Apr 13 13:35:47 davis: cycle bike Apr 13 13:35:55 now avec le shiny! Apr 13 13:36:19 cool. i enjoy two wheels as well. Apr 13 13:38:51 rburton: disclaimer: I haven't checked master yet, only jethro Apr 13 13:39:02 rburton: what are your thought on enabling dri3 by default? Apr 13 13:39:09 +s somewhere Apr 13 13:39:57 getting kodi to run well required way too much bbappends, both for the intel and amd gpu case Apr 13 13:40:10 nouveau was less work Apr 13 13:41:48 koen: last i heard it was so broken it was safer to disable it just in case it got used :) Apr 13 13:41:49 ndec: even better, I received a second mail from that guy saying he heard I was "a nice guy" Apr 13 13:42:06 clearly not informed :) Apr 13 13:42:48 rburton: well, I couldn't get it to build due to missing and/or mismatched PACKAGECONFIGs, but the interwebs suggest dri3 is awesome Apr 13 13:58:52 koen: not sure i'd trust them. patches to make it work trivially would be welcome, but i hear enough moaning from our graphics team to think that enabling dri3 for xserver is a bad idea right now Apr 13 13:59:33 ok Apr 13 14:00:08 iirc we do seem to defaulting to SNA Apr 13 14:00:36 which the interwebs hate, but seems to Just Work (TM) for me Apr 13 14:00:42 (sample size == 2) Apr 13 14:01:33 the hardware on my desk - a fairly old NUC - doesn't even support SNA Apr 13 14:56:40 * Crofton|work curses Apr 13 14:56:41 [balister@thuvia openembedded-core (master)]$ . ./oe-init-build-env ../build ../bitbake/ Apr 13 14:56:41 bash: /home/balister/src/toaster/bitbake/bin/../../scripts/oe-buildenv-internal: No such file or directory Apr 13 14:57:58 fscking assumes bitbake checked out inside oe-core I think Apr 13 15:02:52 hmm Apr 13 15:03:11 why did it work the second time Apr 13 15:03:45 and toaster spews errors Apr 13 15:03:47 grrr Apr 13 15:03:53 start over and file more bugs Apr 13 15:05:13 RP, I'm a little worried that something in bitbake/oe-core broke having bitbake outside oe-core Apr 13 15:05:33 I need to re-check my work before filing a bug, but might not get back to this for a few hours. Apr 13 15:05:42 watnted to give you a heads up this close to release Apr 13 15:07:12 Crofton|work: hmm, I appreciate the heads up but I don't think anything much there has changed Apr 13 15:07:21 yeah, I need to recheck Apr 13 15:07:34 trying to dupe a toaster issue that is related Apr 13 15:07:45 need to get ready to go outside and see the dentist Apr 13 15:08:02 Crofton|work: I did talk with Brian about how to fix some of the toaster/bitbake location issues Apr 13 15:08:30 checout just now is choking hard :( Apr 13 15:08:35 was closer last night :) Apr 13 15:09:32 I need to start a new shell and make sure mine isn't contaminated from earlier efforts Apr 13 17:52:49 i have a vendor supplied yocto build which has a kernel in it. In the default poky x86build, I can make changes to the kernel config and see the results in the resultant run of the build. Apr 13 17:53:25 in this vendor version, I have found the kernel, but I can not make any changes to the config and have it show up in the boot Apr 13 17:54:21 i suspect its overwriting the config each time I build it. Is there any way to determine what's happending? Apr 13 17:54:36 I've created a gist of the kernel recipe which I think is being used here Apr 13 17:54:50 https://gist.github.com/netskink/7700c036eec034281bab009a7f68bc33 Apr 13 18:20:26 davis: any change to the workdir will go away as soon as the recipe gets cleaned next. it's by definition temporamry, unless you've persisted your change into a layer. and if you use devshell rather htan -c menuconfig to make the change, bitbake will have no knowledge of the config having been changed, so future tasks won't be re-run to include it (i.e. compile, install) Apr 13 21:47:04 If I have executable files that I want to load into my targets file system (e.g. under /usr/bin), how do I do that in my custom recipe so that it is automatically done when I build my image? Apr 13 21:56:11 riz___: so you'd have a recipe that installs those files to ${D}${bindir} and from there they'd be packaged automatically; then in the image recipe, add the name of the package (which will be the name of the recipe) to IMAGE_INSTALL Apr 13 21:56:19 simple as that Apr 13 21:56:39 That is quite simple lol Apr 13 21:56:45 What id D? Apr 13 21:57:16 in the context of do_install it's the temporary directory where installed files should go Apr 13 21:57:29 Also I am unfamiliar with the syntax found in do_install, such as -m and -d Apr 13 21:57:37 Where can I learn about all of that? Apr 13 21:57:41 man install Apr 13 21:58:03 -m specifies the permissions, -d specifies to create the directory if it doesn't exist Apr 13 21:58:04 ok, thanks! Apr 13 21:58:41 Lets say I want to add a line to my my rc.local in /etc. How would that be accomplished Apr 13 21:58:46 ? Apr 13 22:00:02 there are a few ways to accomplish that depending on what the line is there for Apr 13 22:00:22 you could do it from a postinstall script for a package if the line is associated with a particular package Apr 13 22:00:28 It is basically there to run one of those executables I installed upon boot Apr 13 22:00:46 the entire line is /etc/bin/ Apr 13 22:00:50 if it's more general for the way the image is set up then you could do it from a function called from ROOTFS_POSTPROCESS_COMMAND within the image Apr 13 22:01:40 ok, that's probably a postinstall script - however, if that's what it is perhaps you should consider using a proper script in /etc/init.d and inherit update-rc.d to have it enabled Apr 13 22:02:41 Hmm. I am unfamiliar with that but I will do some research. Why do you suggest that as opposed to the postinstall script? Apr 13 22:06:20 well, apart from being best practice and slightly tidier, you then have a bit more control over when the service starts and you can write the script such that it can be stopped and started manually as well as automatically which can be useful for debugging (assuming it is actually a running process rather than a one-shot execution) Apr 13 22:07:06 this may be of interest though it's a bit sparse: http://www.yoctoproject.org/docs/current/dev-manual/dev-manual.html#new-recipe-enabling-system-services Apr 13 22:07:10 Oh yeah, before I forget. For my first question, does the executable go in a directory "files" alongside my recipe. And then in my recipe do I include " SRC_URI = "file:// " so that I can use it in the do_install? Apr 13 22:07:24 that will work yes Apr 13 22:07:45 OK. I will read that. thanks Apr 13 22:08:05 Wow, so it was really simple. The entire recipe will be a single do_install i guess Apr 13 22:09:54 if you get a warning about the binary already being stripped of debugging symbols you'll probably need to set INSANE_SKIP_${PN} += "already-stripped" in the recipe Apr 13 22:11:00 OK. gotcha Apr 13 22:11:20 Thanks! Apr 13 22:17:50 One last question. My custom recipe is throwing an error that there is no license provided. Since it is a custo license, what do I give it? Apr 13 22:18:29 good question Apr 13 22:18:54 you can probably legitimately set LICENSE = "CLOSED" if it doesn't incorporate open source code Apr 13 22:19:09 that turns off the LIC_FILES_CHKSUM check Apr 13 22:19:56 ok cool Apr 13 22:31:46 bluelightning: Since I already have the executable files do I just do a "cp" in my do_install or do I still have to do "install"? Apr 13 22:32:12 riz___: I would suggest install -m 0755 Apr 13 22:32:35 That way we are sure the permissions are correcrt? Apr 13 22:32:42 yes, as well as the ownership Apr 13 22:32:58 excellent Apr 13 22:33:51 and when specifiying something from SRC_URI such as file://executable I can just say "executable"? Nothing has to prepend that to let it know it is in "file"? Apr 13 22:35:10 no, the "files" directory is already in FILESPATH so it'll be searched Apr 13 22:35:41 That is what I thought. I am seeig a few examples that have ${B}. Not sure what B is Apr 13 22:37:17 ${B} is the build directory, it's where things like object files will go to when compiling Apr 13 22:37:34 ok. understrood Apr 13 22:37:35 it wouldn't be appropriate in SRC_URI Apr 13 22:37:43 thanks again Apr 13 22:37:48 np Apr 13 22:47:39 bluelightning: I got errors saying that it could not fetch the URL from any source. Apr 13 22:48:10 Function failed: Fetcher failure for URL: 'file://eGTouchG'. Unable to fetch URL from any source. Apr 13 22:49:16 and that is definitely the exact name of the file, right? Apr 13 22:49:39 you can check log.do_fetch for the exact list of directories searched Apr 13 22:52:36 OK. I was under the impression the "files" floder had to be one directory up from where the recipe is Apr 13 22:52:43 I put it alongside the recipe and it found it Apr 13 22:56:05 hmm. failed Apr 13 22:56:14 Failed on an install of an ini file Apr 13 22:56:40 IT said no such file or directory for the .ini Apr 13 22:56:43 Clearly there Apr 13 22:57:08 Would those be treated differently. I mean as far as using install -m 0755 Apr 13 22:58:08 Actually I commented that out. It isnt finding anything Apr 13 22:58:13 Not even the executables Apr 13 22:58:57 install: cannot stat 'eGTouchG': No such file or directory Apr 13 23:00:59 bluelightning: http://pastebin.com/jgETJBUW this is what I have Apr 13 23:02:15 the SRC_URI says eTouchD not eTouchG ... ? Apr 13 23:02:39 also, directories must be created before installing files into them... ${D} starts out empty Apr 13 23:03:00 er, eGTouch* I meant Apr 13 23:04:04 Sorry I fixed that D but same story Apr 13 23:04:56 bluelightning: I am confused with ${D} Apr 13 23:05:05 What do I set that as? Apr 13 23:05:12 I thought it was something generated during build Apr 13 23:05:48 actually you probably need to prefix the source path for the install -m lines with ${WORKDIR}/ Apr 13 23:06:02 also 0755 isn't correct for the ini file, that should probably be 0644 Apr 13 23:06:24 you don't set D, it's set to a reasonable value already Apr 13 23:07:45 OK. That WORKDIR addition worked Apr 13 23:08:01 the directory specified ${D} gets created automatically, but its contents are entirely up to your recipe (and possibly classes it inherits in some instances) Apr 13 23:08:03 A bunch of QA rdepends warnings but it worked Apr 13 23:08:14 ok **** ENDING LOGGING AT Thu Apr 14 02:59:58 2016