**** BEGIN LOGGING AT Thu Apr 21 02:59:58 2016 Apr 21 09:51:46 Is there any support for profiling builds with BitBake, at task level? For example, I'd like to see how long each task takes, to start optimizing build times. Apr 21 09:57:06 mario-goulart: http://www.yoctoproject.org/docs/2.1/ref-manual/ref-manual.html#ref-classes-buildstats Apr 21 09:57:39 joshuagl: excellent. Thanks a lot. Apr 21 10:00:08 np Apr 21 10:11:52 joshuagl: one more question: do you know if there are tools to extract information out of the profiling data? Apr 21 10:27:14 mario-goulart: I think Toaster does that. It stores the info from buildstats in a db and shows it to you with a web ui Apr 21 10:27:24 mario-goulart: what do you mean by tools? that document references an analysis tool :-) Apr 21 10:39:15 joshuagl: sorry, I mean text-based based tools. Apr 21 10:40:58 mario-goulart: I'm afraid I'm not aware of any, but that doesn't mean they don't exist Apr 21 10:42:11 I see. Thanks anyway. :-) Also thanks for the tip about toaster, belen. Apr 21 10:53:51 Hello! Yday I have asked um Hilfe Apr 21 10:53:51 now I make MACHINE=qemumips bitbake vim Apr 21 10:53:51 -- Apr 21 10:53:51 Parsing of 898 .bb files complete (0 cached, 898 parsed). 1329 targets, 70 skipped, 0 masked, 0 errors. Apr 21 10:53:51 ERROR: Nothing PROVIDES 'vim' Apr 21 10:54:03 What can I do, please help :) Apr 21 10:58:42 Switch to emacs? :-) Apr 21 11:00:36 I can see vim in meta-openembedded Apr 21 11:00:37 Seriously, do you have meta-oe in your BBLAYERS? Apr 21 11:17:24 Seems no Apr 21 11:18:02 there another issue - qemumips is wrong arch, I need mipsel (mips32r1) Apr 21 11:19:25 can You please help how to start it from scratch, and compile whole system for mipsel, for my arch, and then add necessary packet manager etc Apr 21 11:19:25 I want learn more oe - looks like sharm, really power tool. It can do alles! Apr 21 11:21:59 hi all. quick question - where to take look if bitbake doesn't store build results of some recipe into sstate cache? I am building os based on freescale imx6 and everytime I remove local build (tmp), kernel is fetched and rebuild all the time Apr 21 11:22:03 thx Apr 21 11:27:53 msk, is the kernel recipe AUTO_REV? Apr 21 11:28:30 w8 Apr 21 11:32:21 ... parsing all includes ... Apr 21 11:35:23 seems not to be, grep nothing, bitbake -e nothing Apr 21 11:35:35 it's freescale kernel from meta-fsl-arm Apr 21 11:39:14 but the name is linux-fslc-imx-3.14-1.1.x+gitAUTOINC+327d5c9063-r4 huh, so yes, it uses git hash as version Apr 21 11:53:53 but SRCBRANCH = "3.14-1.1.x-imx" SRCREV = "327d5c9063b715c91a88655533d5e477a0afe218" so it is stabilized Apr 21 11:54:54 that was my best guess :( Apr 21 11:55:54 i compared stamps for 2 clean builds, only build and deploy are different Apr 21 11:56:27 all other stamps has the same hashes Apr 21 11:56:47 dunno what should be the problem Apr 21 11:57:17 I tried to set PV to some static value (without git autorev) - the same Apr 21 15:32:47 can you please say, where I need to set version for sources , which OE download automatically? Apr 21 15:48:34 Has anyone done a write up on using jenkins to do OE builds? Apr 21 15:53:22 depending on what you're looking to do, it's either trivial or very complex, depends on the requirements Apr 21 15:54:52 I figured that :) Apr 21 15:55:02 start with simple and build up from there Apr 21 16:02:35 Crofton|work: we use jenkins for all our builds... but really as kergoth said, it really depends on your setup/config.. and what you want to do. Apr 21 16:03:00 mostly, at the moment look for build fails for some setups Apr 21 16:08:19 Crofton|work: toaster world would be very interested in reading that write up Apr 21 16:09:06 belen1, seems like everyone does it, but no one wrote anything Apr 21 16:09:27 I need to reinstall newer linux on older server Apr 21 16:09:30 i'd like to see better jenkins/bitbake integration with regard to knowledge of the details of hte failure. ours just hands us a log file. tempted to fire up a yocto error reporting instance instead Apr 21 16:11:47 kergoth I agree.. the YP error reporting stuff has decent knowledge of what went wrong.. would be nice to get that into the build system tools Apr 21 16:12:33 what is the YP error reporting? Apr 21 16:12:59 http://www.yoctoproject.org/docs/2.0.1/dev-manual/dev-manual.html#using-the-error-reporting-tool Apr 21 16:13:29 basically in a failed build you can add the 'report-error' class.. and use the script 'error-report' to automatically send it up to a server.. Apr 21 16:14:16 the idea behind it is to share errors with others, so that you can see if someone else experienced the problem you did.. Apr 21 16:14:38 I would be willing to volunteer to speak to all of you about this and consolidate your thoughts. We might be able to use that as a starting point for that Jenkins write up. It would also help us in our toaster and tools-related research Apr 21 16:14:52 but the nice part (generally useful) is the error report has almost everything you need (in most cases) to figure out what the error was.. Apr 21 16:15:13 nice... Apr 21 16:15:38 the error server does not need access to the build dir or the recipes, or does it? Apr 21 16:15:51 no Apr 21 16:16:04 in the case of an error, a specific report is constructed.. Apr 21 16:16:12 the report is available as a file or can be uploaded ot the server.. Apr 21 16:16:45 hmm. interesting.. could we send our (linaro) error reports to errors.yocto.org? Apr 21 16:16:51 yes Apr 21 16:17:00 in fact, your encouraged to! Apr 21 16:17:16 that sounds interesting. it's looks like the kernelci.org thingy. Apr 21 16:18:41 so, our reports would be on something like http://errors.yoctoproject.org/Errors/Latest/Linaro, is that correct? Apr 21 16:26:08 wow! I did it! ./vim.vim: ELF 32-bit LSB executable, MIPS, MIPS-I version 1 (SYSV), dynamically linked, interpreter /lib/ld.so.1, for GNU/Linux 2.6.32, BuildID[sha1]=21da8cd7443c171d316ac10a202a1fb5a3ed751b, not stripped Apr 21 16:26:44 thank You for help and, sure, OE release! Apr 21 21:01:23 oe-classic; bitbake virtual/kernel didn't seem to build modules, and -c devshell with ../temp/run.do_compile_kernelmodules tells me there are none to build. Is there some other command to build the kernel modules? I know some wind up in my image (and they're not out of tree modules) Apr 21 21:01:56 bitbake virtual/kernel has always built modules, if they're enabled. Apr 21 21:02:10 kernel modules are packaged into separate binary packages. install 'kernel-modules' if you want them all Apr 21 21:02:19 weird, wonder what I'm doing wrong. Apr 21 21:03:01 find tmp/deploy/ -name kernel-module\*.ipk Apr 21 21:03:10 thx Apr 21 21:19:12 ah looks like rm_work was set, and we need to run.do_configure before we can run.do_compile_kernelmodules Apr 21 21:20:01 ahh, that explains it. yeah, you can't short-circuit the task dependencies and expect good results :) Apr 21 21:37:06 in 2016 aren't our hard drives big enough so that we don't need rm_work on all the time? Apr 21 21:39:43 will our hard drives ever be big enough that we don't need rm_work on all the time? :-) Apr 21 21:49:35 * kergoth can't remember the last time he ran rm_work, just wipe tmp on a regular basis, no large workdir content left when building from sstate Apr 21 21:52:07 joshuagl: I don't think I've ever turned it on permanently myself, I just delete tmp as kergoth said Apr 21 21:53:02 bluelightning: thoughts on using plugins for commands for bitbake-layers? I'm a bit torn, I like the idea of keeping it small and compact, but at the same time I'm not sure how long that'll be viable, and extensibility is useful Apr 21 21:54:59 kergoth: well, it's funny you ask, there is a looming question about what to do with yocto-layer Apr 21 21:55:14 it shouldn't really be in meta-yocto as it is currently, all it does is generate a layer from a template Apr 21 21:55:19 ah, if bitbake-alyers was extensible, that could be a plugin in the metadata.. Apr 21 21:55:21 but that's decidedly OE territory and not bitbake Apr 21 21:55:31 so extensibility would help for that at least Apr 21 21:55:46 yep Apr 21 21:55:56 so I think we do need to look at that Apr 21 21:56:03 fair enough, i might go ahead and prototype it then, since i'm working on the sort-layers/find-layers bits Apr 21 21:56:14 ok, great, thanks :) Apr 21 21:57:53 np, thanks for the input Apr 21 21:59:48 bluelightning: yeah, I jest. I've never used rm_work. I do delete tmp a lot too **** ENDING LOGGING AT Fri Apr 22 02:59:58 2016