**** BEGIN LOGGING AT Mon Oct 09 03:00:01 2017 Oct 09 07:24:29 Hi Oct 09 07:25:13 I did check release table for poky and for rocko it says fall this year, do we have some mo specific date (I would like to bump my current morty to rocko). thx Oct 09 07:34:03 open-nandra: usually end of oct, begin of nov, around ELCE and YPDD Oct 09 07:34:12 give or take some days/weeks Oct 09 07:34:51 LetoThe2nd: ok thanks a lot. So i'll use current rocko branch and wait for official release :+1" Oct 09 07:34:53 LetoThe2nd: ok thanks a lot. So i'll use current rocko branch and wait for official release :+1: Oct 09 07:36:13 open-nandra: https://wiki.yoctoproject.org/wiki/Yocto_2.4_Schedule#Yocto_Project_2.4_release_schedule_.28release_date:_Oct._20.2C_2017.29 Oct 09 07:38:23 LetoThe2nd: oh great thanks a lot Oct 09 08:01:43 Hi all, do you know how you can load a kernel module except using : /etc/modprob.d , udev , KERNEL_MODULE_AUTOLOAD. It's because i have a kernel module that is loaded automatically and i would like to know how. Thanks ! Oct 09 08:38:10 hi all.. i have a special request to create packages dynamical depending on a variable. so, if variable is set to SOLUTION-A or SOLUTION-B i want to deliver package PN-$SOLUTION. Right now i use eventhandler to set PACKAGES variable, please see here: https://paste.ubuntu.com/25706105/ Oct 09 08:38:33 can someone please review and tell me if this approach is okay, or if there is a better solution for this? Oct 09 08:38:39 thanks! Oct 09 08:43:16 btooth: you can use python inside assignments by doing ${@some python here} Oct 09 08:43:56 or, just put both packages in the PACKAGES variable, and only actually put files in them depending on the variable Oct 09 08:44:03 empty packages don't get generated by default Oct 09 08:57:56 rburton: thanks. but do you think it is okay to go the way using my eventhandler? are there any doubts? Oct 09 09:16:15 RP: was just looking at the BuildLog but I see that the patches from master-next were all merged? Oct 09 09:16:20 s/see/think? Oct 09 09:18:30 joshuagl: yes, I just did that Oct 09 09:18:46 joshuagl: I doubt AB failures were due to them Oct 09 09:18:53 btooth: the event handler way is massively overcomplicated Oct 09 09:20:37 RP: agreed. The only one that worries me is the "oe-repo has incorrect checksum" on nightly-x86-64 Oct 09 09:21:18 joshuagl: we have seen that once before a long time ago. I could never reproduce Oct 09 09:21:27 it did have a bug entry Oct 09 09:21:34 RP: ah, OK. Good to know. I'll try and dig up the bug. Thanks! Oct 09 09:32:33 rburton: how to use multiple python lines inside assignments Oct 09 09:32:48 btooth: write a function, call it. Oct 09 09:32:55 thanks Oct 09 09:33:07 if you've a variable you just want to look at then you don't need multiple lines Oct 09 09:33:23 ie see all the use of bb.utils.contains() Oct 09 09:45:31 hello, is git.freescale.com having a sad week? Oct 09 12:03:06 Hi all, anyone using Asterisk here? I'm having trouble with https://github.com/ctrob67/meta-telephony/blob/master/recipes-asterisk/asterisk/asterisk_13.16.0.bb and bundled pjproject Oct 09 12:04:50 nayfe : no but what is your problem? Oct 09 12:06:09 Im trying to build a program that depends on libdrm, but xf86.h in libdrm includes a file which exists but is in /usr/include/libdrm/drm.h, while gcc only looks for drm.h in /usr/include - why is this the case and how can I fix this Oct 09 12:07:30 ChrysD: in configure step, asterisk main ./configure tries to call ./configure from third-party bundled pjproject, and it fails with "aconfigure: error: cannot run C compiled programs If you meant to cross compile, use `--host' Oct 09 12:18:35 wouterstreamit: afaict the libdrm pkgconfig file sets the include paths, so it sounds like you're not doing that. gcc won't look in a subdirectory unless told otherwise. Oct 09 12:28:16 nayfe : can't help sorry. Be sure you have everything compatible. Oct 09 12:31:29 ChrysD: np, i'll try to fix it, or take the external pjproject solution ... Oct 09 12:41:50 patchtest doesn't like me. Is "Upstream-Status" supposed to be case-sensitive? Oct 09 12:48:36 rburton: how would I make yocto use that pkgconfig file then? Oct 09 12:52:46 paulbarker: I fear it is Oct 09 12:54:33 joshuagl: Is that intentional though? If so I'll re-submit my patch. If not I can send a change for patchtest-oe to make it case insenstive Oct 09 12:57:56 paulbarker: I believe it's intentional, it has definitely come up on the list before Oct 09 12:58:59 ok cool. I wonder if we can make patchtest-oe more helpful here then. Could it issue a warning if it spots "Upstream-status" with instruction to change the case? I've not really looked at what's possible with the framework Oct 09 12:59:18 If it's possible I don't mind taking a go at the change later in the week Oct 09 13:19:51 wouterstreamit: its not yocto, its what you're building. yocto is just running make or whatever. Oct 09 13:20:10 paulbarker: the messages could definitely be better. Oct 09 13:21:23 hello, could someone recommend a build server specification for a company that wants to do Yocto project CI? Oct 09 13:21:54 is anyone out there using Xeon-based rack servers for their builds? Oct 09 13:22:29 or are people most often doing the builds on their desktop workstations? Oct 09 13:22:39 eduardas_m: as much as you can afford Oct 09 13:23:15 eduardas_m: the autobuilder farm is lots of Intel(R) Xeon(R) CPU E5-2697 v3 @ 2.60GHz Oct 09 13:23:31 rburton: that is not very specific...lets say I can afford up to 4000 USD... how far can that take me? Oct 09 13:23:46 go to dell.com, spec a system Oct 09 13:24:18 you've a budget so build up to that. Oct 09 13:24:40 rburton: for example: I do not even know what is better: two socket xeon system or one xeon with same amount of cores? Oct 09 13:25:33 and what is better to invest in: a more powerful processor, SSDs or lots of RAM? Oct 09 13:26:11 all of them ;0 Oct 09 13:26:32 I have no prior experience with server or workstation grade hardware Oct 09 13:26:52 compilation is CPU intensive, benefits from lots of memory for caching, and you'll want lots of disk space if you intend to keep build artifacts for a long time Oct 09 13:27:12 personally go for more ram so you can do builds in tmpfs and only persist what you want to keep on a real disk Oct 09 13:27:42 my workstation has 64G of ram, half of which is the build tmpfs Oct 09 13:28:19 twin socket will give better peformance as you can't run all the cores at full speed on a single processor Oct 09 13:30:34 rburton: I have checked the size of my temporary directory (tmp) and its size is 76 GB... does that mean I have to aim for 128 GB RAM? Oct 09 13:30:51 no, you use rm_work to remove the bits that are already built Oct 09 13:30:57 I build a lot of Qt related stuff, etc. Oct 09 13:31:58 or just use a real disk. swings/roundabouts. tmpfs builds are a little bit faster but not incrediby. Oct 09 13:32:39 doesn't rm_work mean I won't be able to inspect the intermediate build files... checked out and configured, patched source for example? Oct 09 13:34:00 for example if my build stops at an error? Oct 09 13:35:49 it only deletes when the build is successful Oct 09 13:36:52 rburton: that is good to know... thank you for the useful information Oct 09 13:50:19 rburton: thank you, I have found out was was wrong Oct 09 14:52:12 rburton: I would like to ask you what processor or processors is your workstation using? Is it a consumer processor or a dual-Xeon workstation? how long does it take for you to compile core-image-sato for x86-64 from scratch when using tmpfs? Oct 09 15:03:02 eduardas_m: i have a Xeon E5-2690 system, sato from scratch takes about 40 minutes or so? haven't timed it for a while. Oct 09 15:09:28 rburton: thank you for answering Oct 09 15:27:21 ChrysD: For the record, it was a known bug of asterisk version, after updating it, it works like a charm. Oct 09 15:32:18 khem: can you resend gcc: Backport fix for a segfault on riscv with the missing SOB fixed? Oct 09 15:32:38 I do like 'SOB' as the acronym.. ;) Oct 09 15:36:43 I have two configs: conf.armA/conf/local.conf and conf.armB/conf/local.conf. How can I specify the same TMPDIR and SSTATE_DIR without absolute path? TOPDIR is reffering to conf.arm/... Oct 09 15:40:32 use a relative path such as ../../build/? Oct 09 15:40:47 or use absolute paths. this file isn't going into git or anything right? Oct 09 15:46:40 rburton, no, it's not going to git. But relative paths are too ugly. Oct 09 15:46:58 ../ looks strange also. Oct 09 15:46:59 so use an absolute path Oct 09 15:47:30 maybe there is something like OEROOT exported ? Oct 09 16:27:00 problem: 'libsdl-native is set to be ASSUME_PROVIDED but sdl-config can't be found in PATH. Please either install it, or configure qemu not to require sdl" Oct 09 16:27:11 but : [dv@dvpc conf.tpp (pyro)]$ sdl-config Usage: sdl-config [--prefix[=DIR]] [--exec-prefix[=DIR]] [--version] [--cflags] [--libs] Oct 09 16:32:01 I'm trying to use swupdate for my BSP. For now, I'm only trying to boot the swupdate-image generated ramdisk. Someone's using it ? should I load kernel/DTB/ramdisk separatly, or put those files into one using mkimage -T multi ? Oct 09 16:38:43 none of that has anything to do with swupdate. swupdate is a way to deliver content to your device and your configuration stored in the swupdate compound image controls hwat's done with each piece of content. what you want where is up to you, and depends on how your hardware boots Oct 09 16:45:43 kergoth: indead, i was giving context, and you're right, there is no standard here, everyone's doing it's own way ;) Oct 09 16:54:48 Dvorkin: basically, don't set libsdl-native in ASSUME_PROVIDED. it probably won't work anyway. Oct 09 19:59:00 rburton: done Oct 09 20:14:57 khem, hi there Oct 09 20:15:37 khem, having solved most other kernel issues I would like to sort out the gcc7 vs aarmv5 miscompilation issue Oct 09 20:15:59 do you have any other report? Oct 09 20:44:11 khem, it looks in some ways related to the current thread Oct 09 20:44:12 [PATCH] ARM: compressed: discard ksym/kcrctab input section Oct 09 22:33:28 ant_home: hmm have you tried this fix ? Oct 09 22:34:03 not yet, read today Oct 09 22:34:26 anyone seeing ssh connections being refused to a rpi3 running master/rocko opnssh server ? dropbear works ok Oct 09 22:34:46 ssh to itself works ok too its just that other machines cant ssh **** ENDING LOGGING AT Tue Oct 10 03:00:00 2017