**** BEGIN LOGGING AT Thu Sep 04 02:59:58 2014 Sep 04 05:08:53 Hi yocto.. Sep 04 05:09:07 Is there any one. Sep 04 05:10:28 no Sep 04 06:07:59 mray-no-stay Sep 04 07:37:11 good morning Sep 04 07:42:32 good moaning ;) Sep 04 08:41:56 morning all Sep 04 08:44:05 Hello, Im having kernel panic issue when using runqemu script to boot from nfs, does anyone faced similar issue ? Sep 04 08:52:58 what's the correct way to cleaning up the work directory? i don't have dbus in my image, but stuff is getting built against tmp/work/arm1176jzfshf-vfp-poky-linux-gnueabi/dbus/1.6.18-r0/package/usr/lib/libdbus-1.so.3, when it shouldn't. i have done bitbake -c cleanall dbus, and it doesn't remove that file. manually removing the file fixes the problem, but isn't really a clean solution Sep 04 08:53:57 so, firstly, how do i clean up the work dir properly, and secondly, is there a way to remove all binaries created by recipes that my project no longer uses? Sep 04 09:03:25 BCMM: stuff is being linked against files under the package directory in the workdir? are you sure about that? Sep 04 09:03:50 normally they would need to be in the sysroot to be found by any other recipe Sep 04 09:04:08 bluelightning: hmm, i'll try putting the file back and rebuilding the recipe to confirm Sep 04 09:04:18 in case i've done some other step and forgotten it Sep 04 09:05:40 bluelightning: i found that file the stupid way, by just running `find` in tmp, for the so that was "missing" on the device Sep 04 09:10:26 bluelightning: uh, i guess i missed a step somewhere because it's working now. sorry Sep 04 09:10:48 nonetheless, is there a way to cleanly remove that stuff just to save disk space and be tidy? i am not going to need dbus again Sep 04 09:12:05 BCMM: bitbake -c clean dbus should get rid of that directory, if not it's likely it's an older version that no longer matches up with the dbus recipe that would currently be built Sep 04 09:13:20 there are things like rm_work and the cleanup-workdir script Sep 04 09:15:04 bluelightning: thanks, they sound useful - however, would they remove *everything*, forcing a full rebuild? or are they good for cleaning up obsolete stuff only? Sep 04 09:40:19 how to get bluez5 in core-image-weston Sep 04 09:40:22 ? Sep 04 09:46:33 as usual, add it to IMAGE_INSTALL if there is a recipe for it? Sep 04 09:48:20 for testing you can do that in local.conf, forreproductibility create your own image recipe based on core-image-weston that contains your additional wanted things. Sep 04 09:50:24 BCMM: the latter Sep 04 09:50:31 bluelightning: thanks! Sep 04 09:50:51 been running out of space lately and it helped a bit Sep 04 09:51:59 if you version manage your yocto stuff, it's a good idea to nuke your drive from time to time and do a clean checkout and clean build ;) Sep 04 09:52:18 sure takes forever but do it overnight Sep 04 09:52:44 also ensures that the stuff you version manage is buildable Sep 04 09:55:13 bluelightning: you are so patient with beginner's questions Sep 04 09:56:16 bluelightning: could I have one? ... layers do not what I want Sep 04 09:58:10 jaro123_: happy to help... what's the question? Sep 04 09:58:32 ERROR: Nothing PROVIDES 'jaro-image-dvbt-server' ... which is my recipe and normally it can be found Sep 04 09:59:17 if you run bitbake-layers show-layers, is the layer that your recipe appears in listed? Sep 04 09:59:24 unfortunately I added to my build/conf/bblayers.conf a new layer "${BSPDIR}/sources/meta-openpli \" Sep 04 10:00:09 eaxctly, the bitbake-layers show-layers does see my layer, but bitbake does not see my recipe in my layer Sep 04 10:00:53 the times when I had that error, the recipe name was misspelled Sep 04 10:01:04 my layer is empty... I just cloned my favourite recipe form fsl BSP meta-fsl-arm Sep 04 10:01:57 wah tis interresting, when I remove the meta-openpli from my build/conf/bblayers.conf the jaro-image-dvbt-server can be found! Sep 04 10:02:04 again Sep 04 10:04:26 jaro123_: maybe meta-openpli is doing something strange to BBFILES ? Sep 04 10:05:11 can you pastebin meta-openpli's conf/layer.conf ? Sep 04 10:05:34 yes Sep 04 10:08:19 BBPATH .= ":${LAYERDIR}" Sep 04 10:08:24 BBFILES += "${LAYERDIR}/recipes-*/*/*.bb ${LAYERDIR}/recipes-*/*/*.bbappend" Sep 04 10:08:28 BBFILE_COLLECTIONS += "openpli-layer" Sep 04 10:08:32 BBFILE_PATTERN_openpli-layer := "^${LAYERDIR}/" Sep 04 10:08:38 BBFILE_PRIORITY_openpli-layer = "7" Sep 04 10:08:43 OPENPLI_BASE := "${LAYERDIR}" Sep 04 10:08:47 PLISVNURL ?= "svn://svn.code.sf.net/p/openpli/code" Sep 04 10:08:52 PLISVNPROTO ?= "http" Sep 04 10:08:56 PLISVNBRANCH ?= "trunk" Sep 04 10:09:03 BBMASK .= "|meta-openpli/recipes-bsp/dreambox" Sep 04 10:09:54 ... and enumerates a recepies in the meta-openpli here Sep 04 10:11:23 it could be a BBPATH ? Sep 04 10:12:18 no Sep 04 10:13:59 maybe it looks like BBMASK Sep 04 10:16:22 BBMASK would prevent recipes from being picked up yes, though the expression there shouldn't cause that - one way to find out would be to comment it out Sep 04 10:16:47 ok Sep 04 10:16:48 (btw when I said pastebin, what I meant was a service such as http://pastebin.com/ ) Sep 04 10:17:06 oh better Sep 04 10:17:18 sorry for messing it here Sep 04 10:17:27 no worries Sep 04 10:20:43 commenting BBMASK lines in the meta-openpli/conf/layer.conf caused a python stack trace as bitbake result... so better I will analyse it Sep 04 10:21:02 thank you bluelighting for guiding toward a good direction Sep 04 10:21:53 really? what kind of backtrace? that doesn't sound right... Sep 04 10:26:09 this is probably a side effect Sep 04 10:26:12 I will most probably extract interresting recipes from it ... and skip the problem for now Sep 04 10:26:47 it might be worth alerting the maintainer that that layer is causing problems Sep 04 10:26:58 I'm a bit concerned that the BBMASK would somehow remove your recipe though Sep 04 10:27:13 that shouldn't happen AFAICT Sep 04 10:27:40 maybe I am wrong because I am poky beginer so I do have no idea about the build system itself Sep 04 10:29:39 but the whole build system looks like very fragile and resulting messages not helpfull a much Sep 04 10:30:35 I wouldn't say fragile, but its flexibility does mean it can be complicated at times Sep 04 10:30:40 so I am very carefull to judge something Sep 04 10:30:51 the trouble with BBMASK is that it is a massive hammer... it basically says just ignore this regex when searching for files Sep 04 10:31:05 so bitbake is completely blind to files masked out in that way Sep 04 10:31:37 maintainers really should not be setting it in their layer.conf, that is poor practice Sep 04 10:31:44 how can I trace the parsing process ? make the bitbake verbose do not help Sep 04 10:32:05 there's bitbake -DDD but that produces *a lot* of output Sep 04 10:32:24 can you please just pastebin the backtrace? Sep 04 10:32:25 and the strace does not seem as the best tool Sep 04 10:32:31 yes Sep 04 10:36:00 http://pastebin.com/VvWhRDzN Sep 04 10:40:01 #BBMASK .= "|meta-openpli/recipes-bsp/vuplus" Sep 04 10:40:09 prevents probably this error Sep 04 10:41:09 meta-openpli/conf/layer.conf http://pastebin.com/fvMqymvy Sep 04 10:43:26 all BBMASK in the meta-openpli/conf/layer.conf must be wrong because if uncomment any of them then the bitbake can not see my recipe Sep 04 10:50:37 but the BBMASK should not filter my recipe sources/meta-jaro.new/recipes-dvbt-server/images/jaro-image-dvbt-server.bb Sep 04 11:04:44 hi all Sep 04 11:05:02 what's wrong if i get "ERROR: packagegroup-base-extended not found in the base feeds" ? Sep 04 11:28:22 by Sep 04 11:28:25 bye Sep 04 14:14:37 censor: just saw you comment form yesterday. Did you manage to delete your Toaster builds? Sep 04 14:21:36 belen1: unfortunately not, but it wasn't that urgent Sep 04 14:24:58 censor: I ran a quick test and was able to delete builds, but I believe you need to run the command from the build directory. Are you using the daisy release? Sep 04 14:26:10 yes, i did Sep 04 14:26:33 hmm, it just tried again with an ID higher than 1, it looks like it's doing something ;) Sep 04 14:30:49 thanks for having me try again, looks like it works for other builds Sep 04 14:31:23 censor: but not for build 1? Funny stuff. I'll make sure to try that Sep 04 14:35:24 has anyone in here succesfully used chromium from meta-browser? Sep 04 14:35:50 it builds, it starts, but i always has that blue screen that something went wrong Sep 04 14:36:31 i get this error: ERROR:component_loader.cc(138)] Failed to parse extension manifest. Sep 04 14:36:43 and three of these: ERROR:sandbox_linux.cc(297)] InitializeSandbox() called with multiple threads in process gpu-process Sep 04 14:37:30 --no-sandbox in the mean time solves? Sep 04 14:39:09 no, same issue =( Sep 04 14:41:19 censor: hmm okay. difficult to tell then. only debugging, sorry Sep 04 14:41:54 okay, i'll try with v37 Sep 04 15:01:20 zeddii, http://pastebin.com/0v5yGfH7 Sep 04 15:01:47 that looks like something I would blame you for Sep 04 15:07:04 hmm. no changes in configme that should trigger that ..but never say never. smells more like a recent toolchain change to me. I can try and reproduce it here. Sep 04 15:15:41 hi zeddii - Crofton pointed me your direction for a kernel recipe issue that has me stumped... Sep 04 15:15:54 I made a custom kernel recipe that pulls from an external repo (looks very similar to linux-yocto-custom) and it does not build cleanly under certain circumstances. Sep 04 15:15:58 The facts: Sep 04 15:15:59 The problem occurs when I have not already explicitly built the gcc recipe. Sep 04 15:15:59 I narrowed the problem down to a failure in kernel_configme, I found that during some config operations a build step fails to find crtbegin.o Sep 04 15:16:01 This does _not_ cause kernel_configme task to die however. I found the error in a file in the recipe's work dir: linux/.meta/cfg/standard/x64/merge_log.txt Sep 04 15:16:04 Example recipe and error details here: http://pastebin.com/0v5yGfH7 Sep 04 15:20:14 zeddii, jhernstr is the guy with th eissue, he just joined here Sep 04 15:21:48 Forgot to mention that the configure task is where the build dies, although it appears that things start to go wonky in kernel_configme Sep 04 15:26:43 kernel_configme is simply calling lkc, which is of course uses the kernel's host tools .. and they must be built. Your compiler is busted on that box, so that implies that we need to depend on gcc-native. the kernel.bbclass only depends on the target-gcc as the moment. Sep 04 15:26:53 what sort of development box is this ? Sep 04 15:31:18 I'm running Linux Mint 13, but the build environment is generated by an additional layer of build tools we use in-house. I have been told that we setup our OE builds in a very standard, very vanilla fashion. Sep 04 15:31:45 An interesting point is that I can build the linux-yocto recipe (tweaked only to add my target machine) without encountering this error Sep 04 15:32:10 something is different, otherwise the builders and nightly tests would be screaming red. Sep 04 15:32:20 linux-yocto-custom .. is no different than linux-yocto. Sep 04 15:32:30 only the SRC_URI differs for the most part. Sep 04 15:34:40 Oh I agree it's almost definitely a problem on my end, I'm looking for some suggestions on how to fish better at this point =\ Sep 04 15:36:46 I updated http://pastebin.com/cMrN2gyW with some more details on what specific version of crtbegin.o gets created after I build gcc. I'm curious if the makeup of the sysroot folder before and after suggest anything obvious to you that I might have overlooked... Sep 04 15:41:28 I'll have a look and see if something jumps out at me. Sep 04 15:41:44 my builder(s) are churning on the 3.17 yocto recipes, but I'll try and sneak in a test or two. Sep 04 16:07:15 success! i finally got a working wandboard image with sound and lights and music... or rather: working chrome ;-) Sep 04 16:08:08 but i only noticed just now that it doesn't shut down properly. i can see that the SPDIF port's light goes out, but the splash screen still stays - how can i fix that? Sep 04 16:09:17 censor: so what was the trick now? Sep 04 16:10:42 using chrome v37 instead of v35 Sep 04 16:10:52 that one worked right off the bat Sep 04 16:11:20 even with hw acceleration right away, very nice Sep 04 16:17:33 ah, ok Sep 04 16:59:11 if i wanted to set a packageconfig variable for my chromium package, could i just put 'PACKAGECONFIG[chromium]="disable-api-keys-info-bar"' into my image recipe? Sep 04 16:59:28 or should that go into the local.conf? Sep 04 18:02:47 i'm a bit at loss... i want to disable that Google API Keys bar in chromium Sep 04 18:03:15 i've tried 'PACKAGECONFIG[chromium] = "disable-api-keys-info-bar"' in the image recipe's .bb Sep 04 18:04:15 i created meta-test/recipes/meta-browser/chromium/chromium_37.0.2062.0.bbappend that contains 'PACKAGECONFIG_append = "disable-api-keys-info-bar"' Sep 04 18:04:24 both didn't show any effect =( Sep 04 18:04:44 what's the righ way to add that value to chromium's packageconfig? Sep 04 20:40:02 * maxtothemax__ 's workstation says: save my life I'm goin down for the last time Sep 04 22:00:41 How do i generate an sdcard image? Adding IMAGE_FSTYPES += "sdcard" to local.conf didn't do anything. **** ENDING LOGGING AT Fri Sep 05 02:59:59 2014