**** BEGIN LOGGING AT Mon Mar 04 02:59:59 2013 Mar 04 05:48:35 Hi, when I build yocto by armhf (a9+vfp+neon), which value should I set as DEFAULTTUNE? Mar 04 05:48:56 When I specify cortexa9 hf-neon as this, building of a kernel was built fail. Mar 04 06:01:53 iwamatsu: apply http://patchwork.openembedded.org/patch/44143/ Mar 04 06:03:51 khem: thanks. if I want to build by by armhf (a9+vfp+neon), cortexa9 hf-neon is correct? Mar 04 06:10:01 iwamatsu: yes Mar 04 06:10:34 khem: thank you ! Mar 04 06:10:55 np Mar 04 08:29:11 What happens if multiple packages try to provide same files in a sysroot? I still keep missing usr/include/EGL/egl.h with meta-fsl-arm Mar 04 08:32:56 mcfrisk: do you DEPEND on the graphics drivers -dev package, e.g. gpu-viv-bin-mx6q-dev ? Mar 04 08:33:34 I guess it really should be via a virtual package though Mar 04 08:33:54 yes, I depend on virtual/libegl Mar 04 08:34:27 If I rebuild gpu-viv the issues is fixed, but a clean build fails. Mar 04 08:36:43 ah, maybe gpu-viv should also provide virtual/libegl, now it only has virtual/egl Mar 04 08:38:41 mcfrisk: the mesa package also provides virtual/egl but not virtual/libegl.. Mar 04 08:39:09 is there supposed to be a difference between elg / libegl? Mar 04 08:40:39 erbo: not on denzil, there mesa-common.inc provides virtual/libegl. I had to hack meta-fsl-arm danny to a denzil project. Mar 04 08:42:10 oh, ok.. On master it's PROVIDES = "virtual/dri virtual/libgl virtual/libgles1 virtual/libgles2 virtual/egl" Mar 04 08:42:27 so maybe they renamed it Mar 04 08:53:58 I'll try providing virtual/egl as well, but how to manage "ERROR: multiple bb files are to be build which each provide virtual/libgles1", so both mesa and gpu-viv provide these I'd need to choose somehow. Mar 04 08:56:00 good morning Mar 04 08:59:07 ah, danny meta-fsl-arm is also missing all mx6 PREFERRED_PROVIDERS_virtual/*gl* Mar 04 10:50:27 Is EXTRA_RDEPENDS_${PN}_append_${MACHINE} a valid dependency line? Trying to separate HW specific dependencies and common ones. Mar 04 10:51:35 mcfrisk: you might want RDEPENDS_${PN}_append_${MACHINE} Mar 04 10:51:57 mcfrisk: depends if the EXTRA_xxx means something in the context you're in Mar 04 10:52:14 but its valid Mar 04 10:53:35 RP, thanks. Where are these syntaxes defined? Can't find them from poky-ref-manual. Mar 04 10:53:53 mcfrisk: they're called overrides Mar 04 10:54:19 mcfrisk: two applied together in this case (_append and _${MACHINE}) Mar 04 10:54:40 mcfrisk: bitbake manual as something about overrides, not sure about anything else Mar 04 12:41:59 RP: kergoth: no bitbake hangs since reverting 2 pool patches Mar 04 12:51:27 JaMa: hmm. Which python version? Mar 04 12:51:39 JaMa: we need to narrow this down somehow :/ Mar 04 12:53:17 2.7.3 Mar 04 12:55:10 last time it was with separate process for PR service, isn't pool similar to it? Mar 04 12:55:31 the 'non-x11' support distro has been broken Mar 04 12:58:38 ERROR: Nothing PROVIDES 'directfb-native' (but virtual:native:/home/otavio/hacking/fsl-community-bsp-mx6/sources/poky/meta/recipes-graphics/cairo/cairo_1.12.14.bb DEPENDS on or otherwise requires it) Mar 04 12:58:42 NOTE: Runtime target 'consolekit' is unbuildable, removing... Mar 04 12:58:45 Missing or unbuildable dependency chain was: ['consolekit', 'gtk+-native', 'cairo-native', 'directfb-native'] Mar 04 13:00:18 otavio: there is patch from ross on ML, probably fixing this Mar 04 13:00:45 well not consolekit, but gtk+-native dependency Mar 04 13:00:57 hum, let me check Mar 04 13:03:19 JaMa: I applied the 4 patches; let's see how it goes Mar 04 13:04:22 otavio: I'll try to fix meta-systemd parsing as it is now, so we can keep our own bbclass until oe-core version is working for us Mar 04 13:04:40 JaMa: it'd be great Mar 04 13:04:53 JaMa: you'll preserve the package splitting right? Mar 04 13:05:29 yes Mar 04 13:05:40 JaMa: In current oe-core way I have 4 products which won't go to 1.4 and stay in danny Mar 04 13:05:44 JaMa: awesome Mar 04 13:06:04 JaMa: \m/ Mar 04 13:06:14 if we need to keep meta-systemd for upgrade path, then it's the same work to keep it working for us properly Mar 04 13:06:26 JaMa: yes; I agree Mar 04 13:07:00 JaMa: I bet many people will be happy with this working Mar 04 13:07:01 lol Mar 04 13:07:09 but I'll finish some builds first.. load average: 19.75, 22.40, 17.92 Mar 04 13:08:11 yey Mar 04 13:13:33 hm, i thought that directfb thing was sorted Mar 04 13:14:23 otavio: you should mail the list/bug/something :) Mar 04 13:30:29 otavio: it parses fine again with 2 patches (one for oe-core one for meta-oe) Mar 04 13:45:58 Hi Yocto! Mar 04 13:47:26 I've just started looking into yocto+openembedded for cross platform development and have a quick question: how do I make Yocto/OE/bitbake use my own ARMv7 toolchain? Mar 04 13:48:17 The closest documentation I could find is http://www.yoctoproject.org/docs/1.3/adt-manual/adt-manual.html#using-an-existing-toolchain-tarball Mar 04 13:48:58 why do you want to use your own? Mar 04 13:51:32 because my development board contains an A15 processor and is based on a v7 architecture Mar 04 13:54:11 Is it complicated to use a custom toolchain? Mar 04 13:55:35 denix, did a talk on it at ELC a couple of weeks ago Mar 04 13:55:59 but what is wrong with the existing toolchain options? Mar 04 13:56:13 do we need new tune settings for A15? Mar 04 13:59:11 Before I try to answer that, do you know what architecture http://downloads.yoctoproject.org/releases/yocto/yocto-1.3/toolchain/x86_64/poky-eglibc-x86_64-arm-toolchain-gmae-1.3.sh compiles for? Mar 04 13:59:31 On the tin it just says its an arm compiler Mar 04 13:59:49 JaMa: the idea is that the "pool" code and the original code in the cooker should be functionally equivalent Mar 04 14:00:20 inside that .sh script it extracts to environment-setup-armv5te-poky-linux-gnueabi Mar 04 14:00:21 JaMa: The PR server used multithreading which conflicted with multiprocessing, the pool code doesn't use multithreading Mar 04 14:00:30 so I'm guessing its v5 Mar 04 14:01:18 bdholt1: The standard toolchain in the project supports arm v7 just fine Mar 04 14:02:01 bdholt1: that one is likely an armv5te sysroot with an gcc 4.7 compiler which would support v7 Mar 04 14:02:21 ok Mar 04 14:02:51 I thought I'd ask before I started downloading and compiling the world Mar 04 15:03:26 I'm trying to add PV to some local recipes that haven't had it before with buildhistory turned on. I'm now getting slews of "Package version...went backwards which would break package feeds" errors. Is there any way to avoid that without turning off buildhistory? Mar 04 15:05:00 whoa, a bit excessive announcements methinks... Mar 04 15:07:42 Crofton|work: I did submit A15 tunes some time ago - nobody was interested. they are in meta-ti now Mar 04 15:09:33 Bagder, auto-builder should not fail Mar 04 15:10:17 of course not, but that's still a very large blurb for telling that Mar 04 15:11:00 Bagder, I am sure pidge will tune the settings/messages Mar 04 15:11:08 denix, that concerns me Mar 04 15:13:56 JaMa: great Mar 04 15:14:08 rburton: I am looking at it so I may end fixing it Mar 04 15:59:24 denix: wasn't the problem that some changes to the tune stuff happened around the same time and you never resent with the updates after the changes? Mar 04 16:01:35 RP: there were changes pending from JaMa, but blocked due to 1.3 release, they never went in, last time I talked to JaMa at ELC. I'll resend them Mar 04 16:13:30 denix: thanks, things sometimes do get missed, not intentionally Mar 04 16:27:30 pidge: care to explain those? Mar 04 17:03:35 RP: looking Mar 04 17:04:19 RP: my guess is someone killed all. Mar 04 17:08:25 doesn't that require authentication? Mar 04 17:10:18 zibri: yes. trying to find out who did Mar 04 17:12:00 RP: ab coming down for a moment. Mar 04 17:33:06 apologies for the bot spam. I've shut off exceptions, so this shouldn't happen again Mar 04 17:40:05 is this known: WARNING: perf: The compile log indicates that host include and/or library paths were used. Mar 04 17:53:25 anybody else seeing lists.yoctoproject.org mail being stuck on the server and not delivered in the past several hours? mailman archives are updated though... Mar 04 17:54:09 last emails I've gotten were from about 6 hours ago Mar 04 19:29:23 hello everybody. what's the preferred way to remove a feature from the DISTRO_FEATURES list? Mar 04 19:32:13 produce your own distribution file, and then set it to whatever is appropriate Mar 04 19:32:22 you can start with a copy of the poky.conf for instance Mar 04 19:33:16 fray: so there's no way to pop an item from the list? Mar 04 19:33:28 not ones that are easy or suggested... Mar 04 19:33:41 when you create your own custom distribution you have the ultimate freedom to change whatever settings.. Mar 04 19:33:52 poky.conf is a good example of one distribution configuration Mar 04 19:34:07 fray: ok, I'll do it that way. thanks a lot Mar 04 19:34:27 np Mar 04 19:34:41 if you have any questions of the fields I may be able to help you.. but they are fairly well labeled... (I think) Mar 04 19:35:02 my suggestion is to create your own layer and put your configuration file(s) there.. that way you can re-use it easier when you upgrade or want to share it Mar 04 19:38:00 fray: yeah, that's a good point. I'll begin with a custom layer Mar 05 02:00:52 otavio: I kicked your danny-next for meta-fsl **** ENDING LOGGING AT Tue Mar 05 02:59:58 2013