**** BEGIN LOGGING AT Wed May 29 02:59:58 2013 May 29 04:24:38 how would I get the target sysroot from a native recipe? May 29 04:24:49 instead of native sysroot May 29 05:43:06 is there any documentation on the crossscripts mechanism? May 29 07:06:38 good morning May 29 07:16:27 moin May 29 07:24:34 morning May 29 07:24:45 strange issue with gdk-pixbuf trying to build a cache on the build host fs : http://pastebin.com/EE2Ytdpm (using dylan & oe-core) May 29 07:24:50 that's on a ubuntu server 12.04 LTS. Not reproduced on a Debian SID. May 29 07:30:55 anyone tried getting mesa llvmpipe to work? May 29 07:35:57 hi ericben, all May 29 08:25:17 morning all May 29 12:02:57 sgw1: how's the conference going? May 29 12:03:47 Net147: re your question about llvmpipe. no, but patches very welcome :) May 29 12:05:13 rburton: I have got it to compile but I had to manually modify a wrapper script in llvm and a generated .la file in mesa 9.1.2 May 29 12:05:22 ewww May 29 12:05:30 rburton: when I fix it properly I will post a patch May 29 12:05:36 epic May 29 12:06:37 llvm-config was using native sysroot, mesa .la was linking against target libstdc++ when it needs to link against native libstdc++ for glsl builtin_compiler May 29 12:07:30 swrast without llvmpipe in mesa is really slow on atom board May 29 12:08:29 after that I might try porting cedartrail driver to newer kernel May 29 12:11:15 I currently depend on linux 3.8+ for online resizing of MBR root partition (new ioctl) and specifying root by MBR disk identifier May 29 12:13:30 hi, I am trying to make a bsp layer in dylan 9.0. I get this error:" ERROR: Function failed: split_kernel_module_packages". I have no idea. May 29 12:15:25 in log this comes up: ERROR: System.map-None does not exist... May 29 12:18:15 rburton: remember this bug? ;) https://bugs.freedesktop.org/show_bug.cgi?id=44618 May 29 12:18:16 Bug 44618: was not found. May 29 12:49:37 hmm... May 29 13:57:56 Net147: just got back from dinner, conference is going well, made my presentation to about 50 people with good reception May 29 13:58:25 cool May 29 14:05:06 Net147: so my mesa 9.1.1 just failed as it was linking against host stdc++ May 29 14:05:17 Net147: 9.1.2 even. do you have a good patch for that? :) May 29 14:30:08 hi, has anyone an idea of how to really fix this problem : http://lists.openembedded.org/pipermail/openembedded-core/2013-May/079125.html ? May 29 14:40:36 ericben: I think the best thing would be to create a bug report in the yocto project bugzilla for this May 29 14:41:04 ericben: unfortunately Laurentiu (who worked on this) is away at the moment however so it may not get looked at immediately May 29 14:41:42 bluelightning: OK I though mailing list was useful to discuss this kind of problems before opening a bug. I'll wait for feedback, not a big issue at the moment for me May 29 14:42:14 ericben: the mailing list is fine, it's just that creating a bug means the issue is tracked and can get assigned to someone May 29 14:49:29 bluelightning: ok, which component seems fine for this ? there is nothing concerning classes or recipes-gnome. should I put it under graphics ? May 29 14:50:34 ericben: I think this should be probably go under core May 29 14:50:44 since it's to do with the postinstall code May 29 14:51:03 if you get the option to it would be worth assigning it to Laurentiu Palcu May 29 14:51:12 (or I can do it afterwards) May 29 14:51:58 ok for core, bug 4572 posted May 29 14:51:59 Bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=4572 major, Undecided, ---, saul.wold, NEW , update_icon_cache host access May 29 14:52:05 ;-) May 29 14:57:12 ericben: thanks May 29 15:09:22 ericben: I'd guess we need to set PATH properly in that environment May 29 15:10:02 oh, i meant to reply about that May 29 15:11:39 rburton: I managed to fix Mesa 9.1.2 cross compilation. I will test it on real hardware tomorrow if llvmpipe works with it. May 29 15:12:07 RP: ok, I don't have time to look at this now, will try later unless someone else find the fix before May 29 15:12:16 Net147: noticed there was a patch on the bug you linked too, appears to work here. not ideal though. May 29 15:12:41 hm, the path is absolute already May 29 15:12:48 oh, wrong one May 29 15:13:27 i wonder if we can just remove that line entirely May 29 15:13:53 rburton: my patch - http://pastebin.com/U08KfDDb May 29 15:14:59 rburton: feel free to test it out May 29 15:16:09 rburton: oh there is duplicate assignment there. will remove it... May 29 15:17:26 rburton: http://pastebin.com/bLK6PFCN May 29 15:18:11 Net147: like it May 29 15:18:27 rburton: thanks, I like it too =) May 29 15:18:31 rburton: compared to the other one May 29 15:19:49 rburton: I try to make it not affect the non cross compiling code path May 29 15:20:03 yeah May 29 15:20:17 Net147: you should put it in the bug May 29 15:20:35 rburton: ok, I will do that now May 29 15:23:09 rburton: done May 29 15:24:14 then I can fix the llvm-config script tomorrow May 29 15:25:01 I don't know if mesa in oe-core should have llvmpipe enabled, because llvm is in meta-openembedded May 29 15:25:13 unless we move llvm to oe-core May 29 15:25:18 Net147: personally, moving llvm is fine. May 29 15:25:38 rburton: default llvm on or off? May 29 15:26:03 that's a good question May 29 15:26:22 its such a big dependency its temping to bite the bullet and make mesa machine-specific. May 29 15:26:30 then certain machines can enable llvm easily May 29 15:27:08 I don't know if it works on architectures other than x86 and x86_64 May 29 15:27:21 afaik at the moment it doesn't May 29 15:27:38 but x86 machines that target stuff with modern gen hw don't need llvm pipe anyway May 29 15:28:14 I want it because there are no accelerated open source drivers for cedartrail that work on latest kernel May 29 15:28:25 and it's really slow at the moment May 29 15:29:41 though it may not be needed if I port the cedartrail driver to the new kernel May 29 15:34:57 LLVM may be used as a shader compiler for some graphics chipsets May 29 15:35:53 LLVM 3.3 + AMD Radeon R600 for example May 29 17:52:21 khem, ping May 29 20:38:42 zeddii: yes May 29 20:39:44 heading out. but I was just looking for a pastbin of your gcc 4.8 crash. Just to be sure I'm working on the same one :) **** BEGIN LOGGING AT Wed May 29 20:44:51 2013 May 29 20:46:33 zeddii: yes May 29 20:46:45 the crash is on x86_64/kvm May 29 20:46:47 enabled May 29 20:47:04 another issue I am seeing is that qemu itseld dies May 29 20:47:09 on qemumips64 May 29 20:47:45 keep in mind that qemu-native is compiled with gcc 4.8 May 29 20:47:50 too May 29 21:21:58 howdy khem May 29 21:32:44 scripts/postinst-intercepts is interesting ... but every native binary used in thos scripts creates a dependency for image.bbclass :-( May 29 21:55:23 denix!! **** ENDING LOGGING AT Thu May 30 03:00:00 2013