**** BEGIN LOGGING AT Mon Jul 31 03:00:03 2017 Jul 31 06:29:40 nnn Jul 31 06:51:46 khem: build mostly looked ok but oe-selftest went mental. Not sure why as yet Jul 31 06:53:06 is it possible to run `devtool modify ` but skipping do_patch? Jul 31 06:53:59 bboozzoo: not sure but you could just pop the patches back off? Jul 31 06:55:55 RP: i was hoping there was some hidden no-patch switch Jul 31 06:56:34 devtool extract will also try patch, right? Jul 31 07:09:37 bboozzoo: bluelightning would probably know offhand Jul 31 08:04:55 Hi guys! Trying to build Yocto pyro, both from tarball or the latest git. After running "bitbake core-image-sato", perl-native do_configure step fails, saying "Cannot build with both -Duserelocatableinc and -Duseshrplib". How can I resolve it? Jul 31 08:18:35 RP: bboozzoo: what's the question? Jul 31 08:19:21 bluelightning: i was wondering if there's a way to skip do_path when doing `devtool modify` (or devtool extract for that matter) Jul 31 08:20:53 bboozzoo: hmm, I thought there was but it seems not Jul 31 08:21:11 bboozzoo: are you wanting to skip it because it's failing or because you just want the unpatched source? Jul 31 08:21:55 i'm integrating with another layer, and the patch fails to apply Jul 31 08:22:33 so i figured, why not try devtool and apply the patches manually, fixing the ones that fail Jul 31 08:24:14 which i'm doing just now, just that i had to temporarly skip all patches from that other to get the source extracted and set up to build with Jul 31 08:24:32 bboozzoo: hmm, I don't think we've designed it with that use case in mind unfortunately, but we could certainly look into it Jul 31 08:25:15 not a big problem, just something that might be nice from ux perspective Jul 31 10:50:56 Hi! I have a recipe to build macchina.io that was working with morty but after updating to pyro i get the error "/lib/ld-linux-armhf.so.3: No such file or directory". It's looking fot the file in the wrong place. I already read the release notes and can't anyting that could have created this problem. How can I fix this? Jul 31 13:18:18 rburton: good luck! Jul 31 13:18:43 * rburton wonders what for Jul 31 13:18:55 rburton: mr. icecc Jul 31 13:19:01 ah Jul 31 13:19:02 lol Jul 31 13:19:14 rburton: i tried, failed, and decided to > /dev/null Jul 31 14:09:17 what is the difference between morty, krogoth, jethro, etc? Jul 31 14:09:35 see the documentation and the change notes Jul 31 14:09:58 are these different releases of yocto proper? Jul 31 14:10:08 yes Jul 31 14:12:25 i hope every question i ask is not answered with "see the documentation" ... Jul 31 14:12:47 when it comes to version differences most likely it will be Jul 31 14:14:44 I struggle to remember which name is which version number, so I'd likely do the same thing. it helps to know the recent releases are in alphabetical order, however. Jul 31 14:15:07 careful about your presumptions. as you can see already, my question was a "meta" question - i didn't even know these were various versions of yocto. i am brand new to yocto. Jul 31 14:15:11 most of the people in this channel work on the in-development version or one of the last two stable versions.. outside of that you are likely on your own or need a more commercial support solution where they can answer those questions Jul 31 14:15:26 i think i'll be working with morty Jul 31 14:16:11 shauno: that's good to know. Jul 31 14:16:17 morty is 2.2, which will be EOL (from a community perspective) this fall.. Jul 31 14:16:42 from a commercial perspective, there are more then one company using morty and expecting to continue to support it for many more years.. (my employer being on of them) Jul 31 14:16:52 2.3 is pyro and will be EOL next spring.. Jul 31 14:16:56 2.4 is in development Jul 31 14:17:05 (will be named Rocko unless something changes) Jul 31 14:17:23 seebs: https://wiki.yoctoproject.org/wiki/Releases Jul 31 14:18:06 seebs -> 'see' Jul 31 14:18:24 what is the "4.1.15" version i see here: http://variwiki.com/index.php?title=DART-6UL Jul 31 14:18:37 is that the varascite version of that project? Jul 31 14:18:51 that would be there specific code... Jul 31 14:18:55 nothing to do with the YP. Jul 31 14:19:16 cool. thanks. Jul 31 14:19:22 yates: I think it is related to the linux kernel version (the freescale patched one) Jul 31 14:19:29 ah. Jul 31 14:19:31 it's an odd code too.. since it looks like it could be date derived, but it's not Jul 31 14:19:48 zero_note, ahh so they're not using the morty kernel but their own.. ok.. that would make more sense Jul 31 14:20:28 so each YP version is tied to a specific kernel version? Jul 31 14:20:43 fray: here http://variwiki.com/index.php?title=DART-6UL_Yocto&release=RELEASE_MORTY_V1.0_DART-6UL you can find "Release version: morty-fslc-4.1.15-mx6ul-v1.0" Jul 31 14:20:58 zero_note: did you mean me? Jul 31 14:21:04 The YP includes a reference kernel source (usually more then one) that includes tooling, patches and other things to help make it easier to maintain, configure and build the kernels.. Jul 31 14:21:07 which gives you info about yocto version, kernel version and machine type Jul 31 14:21:14 some vendors produce their own kernels (some continue to use the tools, some do not).. Jul 31 14:21:23 yates: indeed, sorry :) Jul 31 14:21:37 not using the Yocto kernel can cause issues as you are more or less on your own when dealing with the kernel.. the community members won't know how to help you.. Jul 31 14:21:57 hi all. i want to set PR programatically and I use python __anonymous to write timestamp to /tmp/timestamp and set PR from this. but how can i remove the file after the package was built? is there a way to get the task name in __anonymous function or how to addtask after last task to remove the timestamp file? Jul 31 14:22:43 (The YP kernel tooling is designed to manage multiple patch sets, configurations and boards all in one configuration. Some object to this added complexity because they are only targeting one board... but if you are working in an environment with multiple configurations, especially multiple processor types it is -really- useful) Jul 31 14:23:20 what is "poky"? Jul 31 14:23:26 you would have to define a task to run -after- the build. The 'rm_work' task can be used as a model for that Jul 31 14:23:38 poky is the default distribution configuration from the Yocto Project Jul 31 14:23:59 you don't know how much answers to these basic questions help. thanks ! Jul 31 14:24:47 bee_ BTW instead of using an __anonymous chunk for PR values, it would be better to write your own PR service Jul 31 14:25:03 hi anyone got ruby sinatra working on yoctow? Jul 31 14:25:03 the PR service interface is -very- simple.. and likely have everything you need in it to generate what you want Jul 31 14:25:26 (it's a simple XMLRPC interface to a remote server that includes the hash and version information for various things...) Jul 31 14:26:57 the code fo rthe exisitng server is located in: bitbake/bin/bitbake-prserv and bitbake/lib/prserv Jul 31 14:27:07 you could easily make your own version Jul 31 14:27:21 fray: thank you very much! Jul 31 14:27:23 then no changes to the core system will be required, you would just run your custom server and point your conf/local.conf to it Jul 31 14:28:03 ...at one point I had documented the interfaces.. but I'm not finding it right now.. Jul 31 14:28:23 it would be in the advanced YP Dev-Day slides from probably Berlin, and Dublin Jul 31 14:29:49 nope.. not berlin Jul 31 14:30:20 Ahh was in San Deigo 2016 slides Jul 31 14:30:22 https://www.yoctoproject.org/sites/default/files/yocto_devday_advanced_class_sandiego_1.pdf Jul 31 14:30:39 start on page 46 for an explanation of how the PR server works Jul 31 14:31:24 fray: thank you again :-) Jul 31 14:31:40 did freescale tie their morty release to a specific linux version because of updated/added drivers to support their i.MX6? (and are slow/unwilling to propagate these to subsequent linux releases?) Jul 31 14:32:16 if you find that you want/need more information that is normally passed to the PR service, please start a discussion on either openembedded-core@lists.openembedded.org or bitbake-devel@lists.openembedded.org Jul 31 14:32:26 explain what information you think is missing and why you need it.. Jul 31 14:32:52 we designed the system to be expanded with custom PR servers... but so far that is the only one I know that has been written (outside of some playing around I've done) Jul 31 14:33:12 yates, often that is the reason.. they have a specific driver that they don't want to port around.. Jul 31 14:33:35 however, sometimes it's either 'higher versions are better' (i.e. I want a newer version number then someone else, even though there is no real technical reason) Jul 31 14:34:03 and sometimes it is just laziness, as I mentioned before not wanting to deal with the overhead of enabling easier configuration and related Jul 31 14:35:22 isn't there a process for submitting such driver changes to the linux kernel developers and get them included in future versions automatically? Jul 31 14:37:07 i'm asking because, if we're going to change yocto versions (e.g., pyro) and kernels, now would be the ideal time, but i wouldn't want to if the drivers are an issue. Jul 31 14:40:08 Hi! I have a recipe to build macchina.io that was working with morty but after updating to pyro i get the error "/lib/ld-linux-armhf.so.3: No such file or directory". It's looking fot the file in the wrong place. I already read the release notes and can't anyting that could have created this problem. How can I fix this? Jul 31 14:49:31 yates, yes.. preferred is working w/kernel.org to get things upstream, and then once they do they naturally flow into the YP approach.. Jul 31 14:49:45 in the case where that does not happen.. the changes can also be directly submitted to the YP kernel... Jul 31 14:49:59 and where that does not happen, you can manage your own drivers, but still use the YP kernel fairly easily.. Jul 31 14:50:14 so there really isn't a good excuse (long term) for this behavior.. Jul 31 14:51:40 i see Jul 31 15:28:58 khem: | ../mesa-17.1.5/configure: line 22355: test: : integer expression expected <-- sorry Jul 31 15:29:38 khem: with the latest patches in master Jul 31 15:33:35 morning jefro Jul 31 16:32:57 Hello! I am trying to resize a partition in Arago. I had a 1.8GB partition of ext3 type which I pulled the journals off of , then deleted and recreated larger in fdisk. then I attempted to use resizepart /dev/mmcblk1 2 3500000000 Jul 31 16:34:16 this was supposed to tell Linux/Arago that it has a new root file system partition that is now 3.5GB... it did not complain, but it did not update either, I still see 1.8GB when I df -h at the command line Jul 31 16:34:28 you can't just resize the partition, you have to resize the filesystem Jul 31 16:34:38 google for how to resize ext3 Jul 31 16:34:41 resize2fs, iirc Jul 31 16:36:18 if I remember right, Arago did not have resize2fs in it... which I why I thought resizepart would perform the same function. Was this an incorrect assumption? Jul 31 16:36:58 resizepart is, again, for resizing the partition, or more specifically telling the kernel about the resized partition Jul 31 16:37:06 that has nothing to do with the filesystem that's actually on the partition Jul 31 16:37:10 Yeah, there's no resize2fs binary in Arago Jul 31 16:37:13 right now you just have a lot of empty space after the end ofthe filesystem Jul 31 16:37:17 then you'll need to install e2fsprogs Jul 31 16:37:49 I see: e2freefrag e2fsck e2image e2label e2undo Jul 31 16:40:17 does that mean a recompile of the system to get the e2fsprogs installed... or are these the correct components to the package listed above? Jul 31 16:40:30 that's a small subset of e2fsprogs Jul 31 16:40:42 obviously, as i just told you about a tool that's not installed, but comes from e2fsprogs :) Jul 31 16:40:56 doesn't arago have a feed you can use to install the bits from? Jul 31 16:41:14 that's what i was going to suggest, whether you have to re-run bitbake to get it depends on what arago provides, not very familiar with that distro Jul 31 16:41:24 if they have an irc channel, that may be a better bet Jul 31 16:41:38 gibzaz51A! Jul 31 16:41:55 stupid thing.. Jul 31 16:42:06 * paulg_ logs into fray 's bank Jul 31 16:42:07 Thanks! Jul 31 16:42:14 luckily not my bank password.. Jul 31 16:42:19 not even my WR password.. ;) Jul 31 16:43:03 Yeah that is part of the issue with Arago- you have to bitbake everything into it because the feeds are only intended for bake/time... not run time modifications :( Jul 31 16:43:13 he he Jul 31 16:44:10 Thanks to all Jul 31 16:46:53 * fray notes if this damned thing didn't require a password refresh after 5 minutes of inactivity --- and the xterms didn't periodically just decide to take focus.. it wouldn't be an issue.. Jul 31 16:46:55 ohh well.. Jul 31 16:47:45 (which is why I don't use a common password on that machine -- too damned likely to type it into the wrong place) :P Jul 31 16:55:56 rburton: you have to help with some debugging, I cant see this error on 3 different linux flavors I have Jul 31 16:56:52 fray: 2-factor authentication helps ;) Jul 31 16:57:47 khem, any alternative armv5 toolchain to test? Jul 31 16:57:59 I couldn't find any at linaro Jul 31 16:58:07 only gcc6 Jul 31 17:03:49 ant_work: try free electron one Jul 31 17:04:38 http://free-electrons.com/blog/free-and-ready-to-use-cross-compilation-toolchains/ Jul 31 17:09:52 khem, thanks Jul 31 17:11:33 khem: unfortunately no gcc7 for armv5-eabi (not surprising) Jul 31 17:12:32 btw, have you read about these gcc7 userspace issues reported in the answer to my msg on LAKML? Jul 31 17:13:33 I'd say it does not happen with our toolchain, I only have issues with kernel afais Jul 31 17:14:48 I plan to learn arm asm oneday but it will take some time :) Jul 31 17:16:25 bbl Jul 31 17:31:19 rburton, kergoth: we don't publish binary feeds due to some legal concerns... Jul 31 17:31:51 * kergoth nods, been there Jul 31 17:33:04 not sure I understand this statement - Yeah that is part of the issue with Arago- you have to bitbake everything into it because the feeds are only intended for bake/time... not run time modifications :( Jul 31 17:35:11 if you built your images yourself, you do have a feed with packages, from where you can install e2fsprogs-resize2fs package... Jul 31 18:06:31 is there a separate bitbake channel, or is this good? Jul 31 18:08:33 is there a prescriptive form of the bitbake manual? the one here seems descriptive, i.e., a long series of "if you want to ... do this ..." Jul 31 18:08:47 http://www.yoctoproject.org/docs/2.3.1/bitbake-user-manual/bitbake-user-manual.html Jul 31 18:09:27 for example, if i have a file mybb.bb, is mybb the "packagename"? Jul 31 18:09:49 otherwise i don't understand the "bitbake foo" example in 1.5.2.2 Jul 31 18:12:00 how are tasks specified inside a .bb? Jul 31 18:12:06 what is the syntax? Jul 31 18:12:34 the manual states that the default task is "build", but there is no string "build" in a sample .bb file i'm examining. Jul 31 18:12:37 the bitbake user manual explains the task syntax Jul 31 18:12:47 there doesn't need to be, because base.bbclass and bitbake.conf define the defaults Jul 31 18:16:51 if you were squirreled down in a subdirectory, how would bitbake know where the base.bbclass and bitbake.conf files were? Jul 31 18:17:15 current path doesn't matter Jul 31 18:17:52 classes and config files are found via BBPATH, which is etiher defined directly in the environment, or more often nowadays bitbake searches up from $PWD to locate conf/bblayers.conf, which defines BBLAYERS, and parses conf/layer.conf in each specified layer, which adds itself to BBPATH Jul 31 18:18:21 aha. Jul 31 18:50:22 are "recipe" and "target" synonymous in bitbake terminology? (i.e., the name of a PROVIDES)? Jul 31 18:50:36 yeah, that's accurate Jul 31 18:51:06 "package" tends to be ambiguous. generally we try to stick to using it to refer tot he binary packages emitted by the recipes, but there coudl be remnants Jul 31 18:51:56 that helps unconfuse me.. thx Jul 31 18:52:42 technically the recipe is the .bb file, which may provide any number of things, including itself Jul 31 18:52:52 "target" as the name of something from PROVIDES or PN is likely best Jul 31 19:16:09 kergoth: what other thing would a recipe.bb file provide other than itself (for example)? Jul 31 19:19:01 khem: will replicate tomorrow. have you tried with /bin/sh->dash? Jul 31 19:24:14 rburton: I haven't dont want to mess with this machine Jul 31 19:24:22 wuss ;) Jul 31 19:24:27 i'll look tomorrow :) Jul 31 19:24:57 rburton: probably a ubuntu VM should be able to reproduce it as well. Jul 31 19:25:02 yeah Jul 31 19:25:07 let me check Jul 31 19:25:08 kergoth: i guess you mean other targets? (other PROVIDES)? Jul 31 19:25:08 sh->dash is standard debian Jul 31 19:25:12 what I have on this machien Jul 31 19:25:57 rburton: are you having soething like http://git.openembedded.org/openembedded-core-contrib/commit/?h=kraj/master&id=083f35114041011c2699d5da6848b19d90247de8 Jul 31 19:26:05 yes Jul 31 19:26:14 then we have same stuff Jul 31 19:27:22 my ubuntu box has sh->bash Jul 31 19:28:06 but its a shared box so I dont want to mess others up Jul 31 20:08:05 khem: LLVM_VERSION_MAJOR isn't set in that block Jul 31 20:08:58 guessing it doesn't get set properly in llvm_set_envronment_variables? Jul 31 20:09:04 but now, time to logoff **** ENDING LOGGING AT Tue Aug 01 03:00:00 2017