**** BEGIN LOGGING AT Wed Jun 22 02:59:58 2016 Jun 22 09:13:59 Hello everybody Jun 22 10:00:21 hello Jun 22 10:00:42 how can i configure bitbake to put the results OUTSIDE tmp/ ? Jun 22 10:28:14 c0rnel: change DEPLOY_DIR if you mean what ends up in tmp/deploy Jun 22 11:37:38 rburton, thank you very much Jun 22 11:41:39 ok, I think I have solved my problem of splitting a yocto build into multiple partitions and then reassembling them at build type Jun 22 11:41:57 so I want my final image to have a root partition and a data partition, Jun 22 11:42:06 both root and data are populated by buildroot Jun 22 11:42:16 so I made two new image types (not images) Jun 22 11:42:35 one of them creates a .tar of rootfs, excluding the /data directory Jun 22 11:42:47 the second creates a .tar containing only the /data directory Jun 22 11:42:58 since those are image types, we don't populate the rootfs twice Jun 22 11:43:16 and then I have a wic image, depending on both, that assembles the whole thing into a final image Jun 22 11:43:20 tada Jun 22 11:49:04 \o/ Jun 22 12:29:11 \o/ Jun 22 13:18:09 MORE SUCCESS Jun 22 13:24:27 :) Jun 22 14:08:41 i want to have date/time stamp in the deploy dirname Jun 22 14:08:58 but my attempts to configure this in localconf failed so far Jun 22 14:09:18 if this is NOT configured in local.conf, where it is configured? Jun 22 14:10:24 i tried something like this: RELTIME = "time.strftime("%Y%m%d-%H%M%S")" Jun 22 14:10:24 DEPLOY_DIR = "${TOPDIR}/deploy/${RELTIME}" Jun 22 14:10:54 i 'believe' it does not like RELTIME = "time.strftime("%Y%m%d-%H%M%S")" Jun 22 14:23:00 c0rnel: at no point do you say "interpret this string as python" Jun 22 14:23:17 also, you'll want to use := so you don't get a new deploydir every time it gets expanded Jun 22 14:35:39 rburton, pff, how do i tell it to interpret this string as python? Jun 22 14:35:55 sorry, but i can't seem to find anything useful in the internet :( Jun 22 14:35:55 ${@time…..} Jun 22 14:36:04 oh, ok Jun 22 14:36:06 let me try it Jun 22 14:36:25 and := for immediate evaluation otherwise every time you use the variable the python runs and you get new times Jun 22 14:38:13 rburton, would this still create a new directory everytime i create a build? Jun 22 14:38:38 * c0rnel thinks so Jun 22 14:40:03 yes, Jun 22 14:43:00 rburton, thank you very much Jun 22 14:47:41 hey everyone, how do I submit something I think is a bug? Jun 22 14:47:59 jaskij: probably on the mailing list as a patch Jun 22 14:48:39 boucman_work thanks, will do that if I can confirm it works as intended Jun 22 14:49:54 jaskij: there's a bugzilla for bug reports and the mailing list for discussion too Jun 22 15:22:57 rburton seems I was right... there is a hardwired dependency on polkit for NetworkManager Jun 22 15:24:27 and it seems to run just fine without it Jun 22 16:09:12 c0rnel: also we already have a datetime variable. ${DATETIME} Jun 22 16:09:15 * kergoth yawns Jun 22 17:11:55 when using the yocto kernel, i'm curious how the linkage between the build and a specific set of configuration data gets done, by MACHINE? Jun 22 17:12:41 for example, building for the galileo board (intel-quark) the kernel configuration data can be found at http://git.yoctoproject.org/cgit/cgit.cgi/yocto-kernel-cache/tree/bsp/intel-common?h=yocto-4.4 Jun 22 17:13:24 but how does the build associate what you're building for, with the configuration data for the quark kernel? Jun 22 17:18:42 afaik te recipe/append sets the correct KBRANCH and KERNEL_FEATURES, and the kern-tools tooling uses the sccs files and whatnot in the meta repo to govern the configuration. read the yocto kernel development manual and optionally linux-yocto_*.bb and kernel-yocto.bbclass Jun 22 17:39:04 hello guys Jun 22 17:40:04 there is a way to build an installer on yocto, like populate_sdk, but with a custom script that contains a tar.gz inside? Jun 22 17:40:47 can be just a recipe that does not creates a rpm, but a script? Jun 22 17:55:23 is it not possible to add text to the kernel cmdline by adding 'APPEND += "..."' statements to conf/local.conf? Jun 22 17:56:15 might need _append, depending on how and where it's defined. beyond that, it's only used in the bbclasses for live/hddimg/etc, for example. it won't do anything if you arent using those to construct disk images Jun 22 17:56:29 wic doesn't obeyi t at all at this point, since the cmdline is hardcoded in the wks Jun 22 17:57:09 wks? Jun 22 17:59:23 oh... wic config file Jun 22 17:59:45 (used wic for the first time yesterday) Jun 22 18:02:24 oh... so i couldn't even bbappend it either Jun 22 18:03:13 the .wks hardcodes the kernel cmdline in the arguments passed to its commands Jun 22 18:03:37 i have a patch series on the list which adds support for .wks.in templates which expand variables, so you can use ${APPEND} in the .wks.in file Jun 22 18:04:10 http://git.yoctoproject.org/cgit/cgit.cgi/meta-intel/tree/scripts/lib/wic/canned-wks/mkgalileodisk.wks Jun 22 18:04:35 ...and then define APPEND in local.conf? Jun 22 18:04:54 or wherever, i guess Jun 22 18:05:51 since wic is a separate utility from bitbake, does it consult things like local.conf? Jun 22 18:06:13 where would someone define the APPEND variable so the .wks.in could use it? Jun 22 18:08:15 no, it doesn't, which is why i implemented .wks.in the image class. Jun 22 18:08:33 generates a .wks with the bitbake variables expanded and passes the generated .wks to wic Jun 22 18:10:35 ah, thanks :-) Jun 22 18:10:44 https://gist.github.com/kergoth/ae0f426b8b6f426b6b4f7a70a024b5b8 - i intend to add something like this to my distro as a sane default, and add the lines about how to add kgdboc to it to the local.conf template, at least for the machines/bsps that obey APPEND, since of course many do not. even getting wic to obey APPEND will only help for machines using bootloaders which wic generates the config for -- grub-efi, syslinux, etc. to get APPEND obeyed for Jun 22 18:10:44 other bsps we mgiht have to write out e.g. a u-boot environment file to the sd partition for u-boots that load it, etc Jun 22 18:11:18 I haven't had time to dig further, but my goal would be for every machine to obey it, to rename it to something more appropriate, and give it a sane default and examples of useful bits to append to it in local.conf Jun 22 18:14:00 the tooling to create a good kernel cmdline looks like it's going down the same path as the tooling for the kernel configuration itself! Jun 22 18:25:56 halstead: pls check email when you get a chance wrt new AB cluster Jun 22 20:20:58 bluelightning oddball question - how would you sum up the esdk in one bullet point? Jun 22 20:22:13 Jefro: perhaps "SDK providing toolchain plus additional tools to add, modify and test software on the target machine" ? Jun 22 20:23:46 awesome, thanks, that could work well Jun 22 20:58:33 I have some precompiled (closed source) packages that I need to include in my image. But i'm getting errors. Jun 22 20:59:17 QA Issue: Architecture did not match, QA Issue: No GNU_HASH in the elf binary, QA Issue: non -staticdev package contains static .a library Jun 22 21:00:15 also a ton of warnings: ..is owned by uid 1000, which is the same as the user running myrecipe Jun 22 21:04:30 Is what I'm doing not recommended? I would like to end up with a complete image when I'm done with all the packages and files needed already included. Jun 22 21:13:11 the uid 1000 thing is easily fixed with a chown to root in your do_install. it's becuase you're using cp -a to copy the files, it's keeping the ownership of the user you're doing the build as Jun 22 21:13:15 which wont' exist on target Jun 22 21:13:25 but any of the checks can be easily disabled with INSANE_SKIP Jun 22 21:13:43 that said, the architecture did not match sounds particularly problematic. putting an x86 binary on your arm machine isn't going to be very useful :) Jun 22 21:14:05 I'm fairly certain the architecture is correct Jun 22 21:14:13 "linux-imx6yocto-armle-fbdev-obj" Jun 22 21:17:10 The INSANE_SKIP look promissing. Thanks kergoth Jun 22 21:18:14 And what you said about chmod makes sense too Jun 22 21:18:44 i mean chown Jun 22 21:54:20 hi there, I have one patch for linux-yocto-4.4 Jun 22 21:54:29 it's a backport that applies cleanly Jun 22 21:54:41 where do I send it in order to be applied? Jun 22 22:14:58 also, are you guys aware of CVE-2016-1409 fix? I wonder if or when will it be available in Yocto Jun 22 22:20:31 Xz: is that the correct number? when I search for that all I find is references to Cisco IOS Jun 22 22:26:58 bluelightning: This vulnerability is not Cisco specific: any IPv6 processing unit not capable of dropping such packets early in the processing path or in hardware is affected by this vulnerability. Jun 22 22:27:09 bluelightning: it's about ipv6 in general Jun 22 22:54:27 Xz: ok, that's interesting... at least Debian has already dismissed it as being Cisco-specific... is there an upstream fix for this? Jun 22 23:03:08 bluelightning: to be honest I don't know. I got notified of this problem as a security breach that might affect Galileo. After reading description I assumed Galileo is affected as it supports IPv6. Let me google more on that. **** BEGIN LOGGING AT Wed Jun 22 23:20:46 2016 Jun 22 23:24:41 bluelightning: you are right, it affects Cisco IOS only Jun 22 23:29:22 Xz: ok, that's fortunate **** BEGIN LOGGING AT Thu Jun 23 00:23:33 2016 Jun 23 00:49:01 RP: all i just changed was a packageconfig for systemtap and bitbake is re-running about 500 setscene tasks for recipes i didn't touch. i know you're offline, but if you see this and have any ideas.. haven't isolated a test case, yet, just noticed setscene re-executions happening in a number of builds recently **** ENDING LOGGING AT Thu Jun 23 02:59:57 2016