**** BEGIN LOGGING AT Tue Apr 05 02:59:58 2016 Apr 05 08:18:03 good morning Apr 05 08:34:43 mckoan: not sure about that yet Apr 05 08:53:37 LetoThe2nd: :-( I whish you the best! Apr 05 09:05:40 otavio, I use 4.1 fsl kernel, kernel-module-imx-gpu-viv 5.0.11.p7.4+fslc+git0+b7e8ef92be and imx-gpu-viv_5.0.11.p7.4-hfp. Module galcore is autoloading but i don't have /dev/galcore. Do you know what could be the problem? Apr 05 09:06:18 mckoan: no worries, just kidding you Apr 05 11:04:00 spaszPL: see dmesg Apr 05 11:04:15 otavio, nothing on dmesg Apr 05 11:04:43 spaszPL: so it does not seem to be loaded Apr 05 11:04:44 also tested with logFileSize option, manually insmod'ed. Apr 05 11:04:56 lsmod show that module exists Apr 05 11:05:03 but not message on dmesg. Apr 05 11:05:09 *no message Apr 05 11:06:42 i add my patch to this module which comments line " #include " because in linux-fslc 4.1 there is no busfreq-imx6 file. Apr 05 11:06:50 maybe this was the problem Apr 05 11:19:19 spaszPL: so it won't work Apr 05 11:19:31 spaszPL: yes, this is need Apr 05 11:19:58 spaszPL: I did not yet integrate the 4.1 so I did not test it myself Apr 05 12:09:01 otavio, 3.18 no busfreq-imx also Apr 05 12:09:25 so which kernel you use to compile kernel-module-imx-gpu-viv? Apr 05 12:09:36 linux-fsl is not working. Apr 05 12:10:09 Does this recipe work ever? Apr 05 12:13:32 hi there ! Apr 05 12:15:10 does someone now if it's possible to gprof a binary with (oe style) splitted debug symbols ? dunno if my binaries are ok, or if i am missing something from the command line (gprofs keeps telling me "gprof: file `/usr/bin/foo' has no symbols" even if i install foo-dbg) Apr 05 12:16:48 guessing i did not compile the binary properly Apr 05 12:25:42 fixed, binary finally has gprof instrumentation, same error... Apr 05 12:32:14 mckoan: no worries, just kidding you Apr 05 12:35:48 oops.... Apr 05 12:37:35 ok i need to call gprof on the .debug/foo... stoopid Apr 05 12:38:40 or not Apr 05 12:42:06 LetoThe2nd: double kidding today :-D Apr 05 12:42:39 mckoan: totally yes. or rather, hitting keys in ETERM and realizing too late. Apr 05 12:55:42 hi, all! Apr 05 12:55:51 I have a really stupid question Apr 05 12:56:36 to improve speed of my work I need to build kernel outside of OE but to be ABI-compatible with kernel modules which are on my rootfs Apr 05 12:57:44 I have the same config as kernel have in tmp, and I use compiler provided by meta-toolchain Apr 05 12:58:18 but still when I build kernel it refuses to load modules, while when I build kernel in tree, modules are loaded OK. Apr 05 12:58:28 what do I miss? Apr 05 14:29:23 Is there any bitbake documentation that describes the use of "bb.utils.contains()" Apr 05 14:38:23 probably, but the source is quite obvious Apr 05 14:38:25 def contains(variable, checkvalues, truevalue, falsevalue, d): Apr 05 14:39:07 whitespace split the value of variable and checkvalues, and if the latter is a subset of the former then return truevalue, otherwise falsevalue Apr 05 14:39:28 there's also contains_any() Apr 05 14:47:15 riz__: http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/commit/?h=ross/mut&id=bfd46addebf6d47baeb8bfec34f19b1605d95501 :) Apr 05 14:47:36 Awesome. Thanks Apr 05 15:11:51 is it somehow possible to instruct the file fetcher to fetch a file from an absoulte filepath? Apr 05 15:12:37 or what would you use to conditionally add files to a recipe that are located/generated outside of the OE build? Apr 05 15:14:20 Jin^eLD: yes, use an absolute file: uri Apr 05 15:15:20 rburton: file://tmp/pubring.gpg did not work Apr 05 15:15:38 that would be because that's a relative uri Apr 05 15:15:47 oh Apr 05 15:15:54 i miss another / I guess Apr 05 15:15:57 indeed Apr 05 15:16:37 thanks, that was easier than I expected :) Apr 05 15:41:46 is nested ${@evaluation} not possible?i.e. ${@base_contains("SETTING", "0", "${@d.getVar("OTHERSETTING")}", "xxx")} Apr 05 15:41:48 ? Apr 05 15:42:06 no need Apr 05 15:42:15 inside ${@ you're evaluating python Apr 05 15:46:49 hmm I fail to write that correctl i guess Apr 05 15:47:34 if i wanted to have a "file://" + d.getVar("SOMETHING") inside the base_contains, how would that look? Apr 05 15:48:30 right now I have ${@base_contains("GPG_SIGN_IMAGE", "0", "file://pubring.gpg file://secring.gpg file://trustdb.gpg", "file://" + d.getVar("GPG_HOME") + "/pubring.gpg", d)} which does not quite work Apr 05 15:48:52 or where exactly is python? inside "" ? Apr 05 15:51:19 that's likely more compex Apr 05 15:51:50 just write a function in the recipe? Apr 05 15:53:50 where should I hook it in if I want to modify the SRC_URI? Apr 05 15:56:06 python magic_src_uri { write python here to do what you want } Apr 05 15:56:19 SRC_URI = "${@magic_src_uri())}" Apr 05 15:56:53 thanks a lot one more time :) Apr 05 15:56:53 actually, def magic_src_uri: would be neater Apr 05 15:56:58 ok Apr 05 15:56:59 have a look at gnomebase.bbclass Apr 05 15:57:29 I don't see anything wrong with that section you qouted, except that you forgot the argument to expand GPG_HOME on th egetVar Apr 05 15:57:49 that said, it'd be simpler to juse use normal variable expansion. ${@base_contains("GPG_SIGN_IMAGE", "0", "file://pubring.gpg file://secring.gpg file://trustdb.gpg", "file://${GPG_HOME}/pubring.gpg", d)} Apr 05 15:57:54 (gnome_verdir used to be more complex until oe.utils.trim_version was invented, but we kept the function name for compatibility) Apr 05 15:58:00 yeah that works too Apr 05 15:58:10 oh, I did not need getVar there Apr 05 15:58:11 i guess i should read the question properly Apr 05 15:58:35 I was thinking too compliced I guess Apr 05 15:58:37 thanks Apr 05 16:04:12 thanks kergoth :) Apr 05 16:04:27 :) np Apr 05 16:39:32 has anyone tried the win10 linux subsystem to build oe yet? Apr 05 16:39:50 heh, I jokingly asked 2 days ago the same question Apr 05 16:40:31 jokingly? i'm deadly serious. i presume it fails drastically. Apr 05 16:40:48 i wonder what uname says Apr 05 16:42:28 seriousness was below 33%, so I didn't mention that. Sure it would be interesting to see /how/ it fails Apr 05 18:13:29 When you create a project not based of poky, which just includes oe-core + bitbake + meta-intel, do you have to create a custom distro since there is no reference distro such as poky provided? Apr 05 18:14:50 it's possible to build without a distro Apr 05 18:15:04 if you want to create a distro because thte default behavior isn't what you want, then go for it Apr 05 18:15:12 but by default DISTRO=nodistro and you can build without poky Apr 05 18:17:14 has anyone run into a case where bitbake says no bbfiles match any of the BBFILE_PATTERNs? Apr 05 18:17:19 it's acting like i have no recipes, even though i do Apr 05 18:17:20 I read that, but I am just confused I guess Apr 05 18:17:27 how so? Apr 05 18:17:39 a distro is a just a config file that's sourced by bitbake.conf, just as the machine .conf and local.conf are Apr 05 18:18:03 With nodistro, does that mean you are just using the default linux kernel settings? Apr 05 18:18:29 i don't understand the question Apr 05 18:18:36 what exactly do you thin kthe distro does to the kernel? Apr 05 18:18:44 the machine controls kernela nd bootloader selection, not the distro Apr 05 18:18:45 That is my confusion Apr 05 18:18:52 as i said yesterday Apr 05 18:19:12 the distro defines overarching policy, what package manager is used, etc. has absolutely nothing to do with the kernel Apr 05 18:24:36 OK. I see Apr 05 18:25:08 I was just examining poky and saw that it was choosing the linux version Apr 05 18:25:18 e.g. 4.1% Apr 05 18:25:27 Unless that is not the kernel version Apr 05 18:28:56 that's pretty unusual, generally it's up to the machine to make calls like that. no idea what that's about Apr 05 18:29:58 Yea, for yocto they both specify it Apr 05 18:30:02 redundant Apr 05 18:30:52 That redundancy is what makes it hard to draw the lines on which layer governs what Apr 05 18:31:43 did you read the link i shared the other day about the orthogonality of distro, machine, and image? that's how we run things. there are always exceptions to any rule, and not all layers do what they should, but that's the goal Apr 05 18:31:57 Yup. HAve it right in front of me Apr 05 18:32:20 It was a great help. Just trying to relate that to what I am actually seeing in the project files Apr 05 18:32:43 Yeah. Understood Apr 05 18:36:04 in any case, your distro should only define what it needs, same for the machine. there's a lot of blind copying/pasting of files beween layers that happens, but it's not ideal. it's okay as a starting point, but folks need to remember to go back and make sure they have things they actually want and need Apr 05 18:36:21 heh Apr 05 18:37:00 i'm sure there was a good reason for poky to set that version at one time, likely due tot hefact that linux-yocto has multiple versions and is used by multiple machines/bsps, so somsone made the call that the distro could decide the version and let the machine decide the recipe Apr 05 18:37:02 * kergoth shrugs Apr 05 18:37:34 i've said it before and i'll say it again, our flexibility is both our greatest strength and our greatest weakness since it leads to confusion about how things should be done :) trade-offs.. Apr 05 18:37:38 Right Apr 05 18:37:50 If you use no-distro, where do you put the DISTRO_FEATURES? Apr 05 18:38:03 if you arne't happy with the default distro features, then you'd create a distro Apr 05 18:38:22 see meta/conf/distro/include/default-distrovars.inc Apr 05 18:38:32 Hahah, I was just opening it Apr 05 18:38:37 that's parsed after the distro and whatnot to set all the defaults Apr 05 18:38:40 :) Apr 05 18:38:49 OK. I think I am getting the hang of this Apr 05 18:42:06 riz__: have you read pidge's chapter of architecture of open source systems about yocto/bitbake? Apr 05 18:42:11 might be interesting background, if not Apr 05 18:42:28 http://www.aosabook.org/en/yocto.html Apr 05 18:42:55 No, but I will now. Thanks! **** ENDING LOGGING AT Wed Apr 06 02:59:58 2016