**** BEGIN LOGGING AT Thu Dec 11 02:59:58 2014 Dec 11 04:32:24 * nerdboy waves Dec 11 04:33:45 are there any analysis tools for the sources used in an image? Dec 11 04:34:56 tools/hooks/other? Dec 11 07:32:07 hmmm Dec 11 07:32:24 using a custom kernel for an x86 board is very broken Dec 11 07:32:36 unless you use a version newer than linux-yocto Dec 11 07:32:50 since the intel-foo-common packages will be valid for all x86 boards Dec 11 08:04:42 good morning Dec 11 08:11:47 * nerdboy checks toaster on a fresh build Dec 11 11:02:27 morning all Dec 11 11:09:10 hi bluelightning Dec 11 11:09:26 hi woglinde Dec 11 11:09:34 bit late today? Dec 11 11:09:56 yes, had to run some errands Dec 11 11:10:09 I was here earlier as well, I just didn't say good morning :) Dec 11 11:10:43 woglinde: btw I think we are almost ready to go on the meta-java move, we just need Mario's pubkey Dec 11 11:11:05 we could probably go ahead and make the move right now if you're happy for me to do so Dec 11 11:17:53 ok he just pinged me, have sent his key to Michael Dec 11 11:18:00 so I reckon we should be pretty much ready Dec 11 11:26:02 hm okay Dec 11 13:12:13 Hello, is it possible to make an image dependency with 2 core-images ? Dec 11 13:25:14 ? Dec 11 13:28:50 let's say I have inird.bb and rootfs.bb, there are images. I want that doing "bitbake rootfs" triggers the initrd target Dec 11 13:30:45 I can't find a way to make the DEPENDS working with image recipes Dec 11 13:33:05 there is no way Dec 11 13:34:04 OK, thanks for the info Dec 11 13:34:10 well, that's not quite true Dec 11 13:34:28 we do have several mechanisms for bringing initramfs images into a main image Dec 11 13:34:43 e.g. INITRD_IMAGE and INITRAMFS_IMAGE Dec 11 13:35:04 I'm not 100% familiar with that area but I do know it is something we at least try to support Dec 11 13:37:32 thx, gonna take a deeper look at this Dec 11 13:46:29 I do know we could document it a lot better than we do now - as far as I know all we have is what's mentioned in the variable reference section for those variables in the YP reference manual Dec 11 13:52:28 I think what I'm looking for is close to do_bundle_initramfs Dec 11 13:52:40 do_bundle_initramfs[doc] = "Combines an initial ramdisk image and kernel together to form a single image" Dec 11 16:00:53 Something smells **really** fishy with bitbake task scheduling on dizzy and master. do_compiles in DEPENDS recipes are getting run before do_configure. Dec 11 16:01:33 In particular, this appears to be the cause of simonl's archlinux breakage building qemu-native: on my machine, qemu-native attempts to start building way, way before glib-2.0-native finishes building. Dec 11 16:02:59 Apparently others on my team (ksharp) have been noticing this for a while, and have been working around it by adding workarounds that AFAIK should never have been necessary, like: do_install[depends] += "niacctbase:do_populate_sysroot" Dec 11 16:03:16 where "niacctbase" was already present in DEPENDS Dec 11 16:03:37 and `bitbake -g` looks OK Dec 11 16:04:16 did DEPENDS behavior change recently so that it no longer means "these packages must populate sysroot before building this recipe"? Dec 11 16:11:15 no, of course not. it'd be useless otherwise Dec 11 16:11:47 ok that's good Dec 11 16:12:58 quickly grepping back in IRC logs, I saw some discussion between kroon and rburton in july about DEPENDS not necessarily doing the "right" thing, involving either -native or PACKAGECONFIG'd packages (haven't figured that out yet from the context). But this has been happening for packages which use neither of those mechanism.s Dec 11 16:14:59 hmmmm Dec 11 16:15:08 when did 'bitbake foo-1.2' stop working? Dec 11 16:15:54 ? `bitbake -c clean glib-2.0` works for me, but that's because the recipe is literally called glib-2.0 Dec 11 16:17:02 I added linux-libc-headers version 3.18 Dec 11 16:17:19 and 'bitbake linux-libc-headers-3.18' didn't work Dec 11 16:17:49 neither did using _ or : as seperator Dec 11 16:18:16 was PV ever an override? Dec 11 16:18:53 yes Dec 11 16:19:10 I know it worked in oe-classic Dec 11 16:20:33 for some value of "worked". it never worked all that well, and there was always hte danger of ending upw ith multiple versions built in the same build dir if the specified version wasn't the preferred one, but yeah, it used to build Dec 11 16:22:12 koen: see 493e8b46fdad9a4a0363ab9e614de122d4ea140d in poky, a87c205bb6cefd5e1a41b8e7ef02b5bfa380e3b6 in oe-core, from some quick git-fu Dec 11 16:30:43 ah Dec 11 16:34:19 moin Dec 11 16:35:08 last day of computer science for kids... Dec 11 16:44:19 koen: some time ago IIRC Dec 11 16:44:31 ah I see kergoth already mentioned Dec 11 16:45:54 rtollert: it would be worth inspecting the task graph output of bitbake -g to see if anything is missing in terms of what should be added through DEPENDS Dec 11 16:50:23 "qemu-native.do_configure" -> "glib-2.0-native.do_populate_sysroot" Dec 11 16:50:23 "qemu-native.do_compile" -> "qemu-native.do_configure" Dec 11 16:50:30 bluelightning: that ought to do me, right? Dec 11 16:51:14 and yet, the build fails in qemu-native.do_compile because libgobject doesn't exist yet in the sysroot, and after the task fails, I see: Dec 11 16:51:19 NOTE: recipe glib-2.0-1_2.40.0-r0: task do_compile: Succeeded Dec 11 17:00:14 hmm, that is not glib-2.0-native Dec 11 17:07:16 Hello, does anyone use syslog ? My program become blocked with two concurrent syslog calls (second one from IT)... Dec 11 17:11:24 rtollert: it sounds like something very wrong is happening there yes Dec 11 17:12:45 bluelightning_: ok, I might be an idiot, glib-2.0-native was in fact populating sysroot beforehand, glib-2.0 was totally unrelated and scheduled correctly. Going back right now and revisiting IRC logs for previous discussion on that matter... and also double-checking on the other funniness we've been seeing Dec 11 17:26:08 vincenet what program? Dec 11 17:28:45 woglinde : personal program and also in a small program just for testing. It sounds like reentrancy case however, first call is well finished. Dec 11 17:30:42 vincenet okay now have a look at the channel topic Dec 11 17:31:34 I am heading home till later maybe Dec 11 17:32:20 it is running under dylan. is it inappropriate question here Dec 11 17:32:57 have you tried it under a desktop linux? Dec 11 17:33:51 no, good idea Dec 11 18:05:47 rtollert : syslog lead to blocking under desktop program test. **** ENDING LOGGING AT Fri Dec 12 02:59:59 2014