**** BEGIN LOGGING AT Tue May 07 02:59:59 2013 May 07 07:11:25 hrw: ping? can you point me to the patch/commit that makes the linaro toolchain switch on unaligned memory access by default? May 07 07:13:52 morning May 07 07:14:12 zecke: probably not - I do not track each of their commits May 07 07:15:00 hrw: do you know anything about that "feature" (i wonder if they honor the march/mcpu) May 07 07:15:24 zecke: no May 07 07:16:06 the standalone toolchain doesn't creating working OsmocomBB binaries and now we have users... ignoring our documentation, using the so great Linaro toolchain and wonder why stuff breaks May 07 07:16:23 zecke: toolchain-layer/recipes-devtools/gcc/gcc-4.6/linaro/gcc-4.6-linaro-r106827.patch May 07 07:17:02 zecke: iirc armv7a should handle unaligned memory access just fine May 07 07:17:43 kernel needs to be built with -mno-unaligned-access otherwise does not boot May 07 07:17:56 hrw: yes, armv6 as well May 07 07:18:13 zecke: and linaro toolchains are armv7a targetted May 07 07:18:45 https://github.com/shr-distribution/linux/commit/3fbd734f0fc2718235d7a7fcf82d9cb92161e28e May 07 07:21:20 JaMa: which branch of meta-oe do I need to see this patch? May 07 07:22:17 I guess it was in danny May 07 07:22:42 but maybe it got renamed later, check git log in that dir May 07 07:33:18 thanks May 07 07:37:32 morning all May 07 07:41:21 good morning May 07 08:18:32 morning all May 07 08:27:29 hi bluelightning, all May 07 08:27:40 hi mckoan May 07 09:15:10 in a recipe, can I find out the name of the recipe file? I want to add a bbclass that does some parsing of the actual recipe May 07 09:16:37 hi bluelightning, mckoan May 07 09:17:32 erbo: the FILE variable points to the file currently being parsed May 07 09:17:50 good morning all May 07 09:17:57 erbo: but what is it you're trying to achieve? May 07 09:18:02 hi apelete, silvio_ May 07 09:18:34 hi bluelightning , silvio_ May 07 09:20:12 hi apelete May 07 09:20:27 bluelightning: I want to create a "release-binary" bbclass that automatically generates a .tar.gz of the generated files and then creates a recipe that uses the .tar.gz instead of actually building the software.. So we can distribute closed source stuff useable in a yocto environment. May 07 09:22:17 erbo: interesting idea... probably worth posting an RFC about that to the OE-Core list May 07 09:25:02 speaking of ideas... May 07 09:25:10 readonly sstate-cache would be nice May 07 09:26:20 hrw: SSTATE_MIRRORS should accomplish that already, unless I misunderstand what you're asking for May 07 09:27:34 bluelightning: I have to do one build with patched gcc/eglibc. would like to reuse sstate-cache (from local hdd) as much as possible but not add anything to it. May 07 09:28:01 bluelightning: now I just did build with clean sstate but it took longer due to building everything May 07 09:28:27 hrw: SSTATE_MIRRORS will work for that, just use file:// to point to a different directory on your local machine May 07 09:29:19 bluelightning: thanks May 07 09:29:45 it can confirm that it works fine May 07 09:30:38 read-only PREMIRROR needs patch from ML May 07 09:40:14 hrw: don't forget to append "/PATH" to the end to ensure native sstate packages get picked up May 07 09:40:24 thx May 07 09:42:48 morning all May 07 09:45:35 hi pb_ May 07 09:45:46 hi bluelightning May 07 09:52:02 hi hrw, pb_ May 07 10:02:15 I have an image that give an error : "| ERROR: Function failed: do_rootfs", but in others doesn t appens, I can t find error in "log.do_rootfs" some suggest to find it? May 07 10:02:50 silvio_: can you pastebin the whole log.do_rootfs please? May 07 10:03:10 sure May 07 10:08:09 bluelightning, http://paste.ubuntu.com/5641014/ May 07 11:12:00 I got a small-medium sized server for building May 07 11:12:07 disk reads are quite good May 07 11:12:28 9052 MB/s for cached read May 07 11:12:35 457 MB/s for normal read May 07 11:17:52 silvio_: apparently you have no pkgdata recorded for a package called "kernel" May 07 11:17:59 silvio_: any idea why that would be the case? May 07 11:20:06 bluelightning, I have more than one kernel, could be this the trouble, maybe I broken something May 07 11:20:42 silvio_: well, it looks like license.bbclass has no way to handle there not being a package called "kernel" which it probably should May 07 11:44:32 bluelightning, now I m trying to resolve, but really I not clear what or better how it happens May 07 13:04:40 wow, compiled 'core-image-minimal' in 36 minutes May 07 13:05:01 big speed improvement for me May 07 13:13:08 eren: from scratch? May 07 13:14:06 mckoan: yeah May 07 13:14:06 with -j 8 May 07 13:14:30 eren: which cpu has your host? May 07 13:15:03 cpuinfo: http://paste.ubuntu.com/5641434/ May 07 13:15:10 meminfo: http://paste.ubuntu.com/5641435/ May 07 13:15:29 has 6GB of ram, but it looks enough for my project May 07 13:15:37 I intend to build only for 1 or 2 targets May 07 13:15:49 never exceeded 2GB while compiling May 07 13:15:59 and swap has not been used May 07 13:16:12 eren: sonds great! May 07 13:16:20 s/sonds/sounds May 07 13:17:28 the only problem is that I can only access the host when I'm in the university May 07 13:17:34 I'm waiting for VPN access May 07 13:21:10 bluelightning, did you pick up the gcc 4.8 fix already? May 07 13:21:34 Crofton|work: I did, yes, thanks for checking though May 07 13:21:52 thanks, so no one needs to send anything May 07 13:22:00 Crofton|work: right, it's taken care of May 07 13:22:04 thanks May 07 13:22:25 joshc, ^^ May 07 13:52:54 thanks, Crofton|work. May 07 13:52:58 and bluelightning :) May 07 14:22:47 bluelightning, I cannot resolve the issue, could u suggest me a way to proced? I m a bit confsed May 07 14:23:29 silvio_: license.bbclass doesn't support there not being a package called "kernel" May 07 14:23:39 silvio_: that would need to be fixed in order to get past this May 07 14:23:47 i.e. it should just fail gracefully May 07 15:07:58 bluelightning, solved.... bitbake -c clean virtual/kernel; bitbake virtual/kernel May 07 17:44:04 when creating "hddimage", does it include required boot code so that I can simply "dd" the image to compact flash? May 07 17:44:10 I guess it uses syslinux May 07 18:05:37 eren: that's the idea, afaik, yes May 07 19:40:33 seebs_: Added myself to the wheel group as secondary group, allowing the tests to complete.. May 07 19:43:14 Which is latest stable OE branch ? I dont get these "denzil" and "danny".. May 07 19:43:44 Dylan May 07 19:43:59 Dylan, Danny, Denzil.. Denzil is more or less unmaintained now (or will be soon).. May 07 19:44:29 fray, ok thanks, have I just missed where this is documented ? May 07 19:44:49 it was sent out to the mailing list.. I THOUGHT there was a release page on the oe twiki, but I may be wrong May 07 19:44:59 (there is on the YP wiki.. the code names are the same) May 07 19:48:35 anyone knows the reason for this "--libdir=${prefix}/lib/pseudo/lib" ? May 07 19:51:43 yes May 07 19:51:49 what do you need to kno? May 07 19:52:38 fray: are you talking to me or jkroon? May 07 19:52:44 you.. pseudo May 07 19:53:06 fray: I am doing silly things for fun May 07 19:53:15 fray: building on FreeBSD May 07 19:53:37 fray: I can build pseudo locally May 07 19:53:39 ok.. first thing to do is see if you can get pseudo working by itself on there.. it has to be able to intercept the calls... May 07 19:54:17 on linux, with multilibs.. it provides both a 32-bit and 64-bit version.. using the singular -native sysroot 'lib' dir.. the psuedo binary then sets the LD_PRELOAD to both versions.. allowing the ld.so to load the right one May 07 19:54:19 fray: I know, I hacked 3/4 of them I guess, so I was trying to give oe a shot May 07 19:54:45 fray: understood, I saw that.. May 07 19:54:49 some basic work was done for Darwin... but it never progressed beyond just making it 'go' May 07 19:55:03 fray: saw that too.. May 07 19:55:43 fray: actually (since I guess you are an author / contributor) it was not that hard too get it running May 07 19:55:55 fray: thanks for that... May 07 19:56:15 'seebs' is the author.. I just contribute where needed.. ;) May 07 19:56:27 but ya, it wasn't thought that it would be too hard to get it to work.. but someone just had to do it May 07 19:56:43 fray: I got a bit scarred intially, overloading syscalls... well it is good there aren't too many virus scanners on POSIX host I guess... ;) May 07 19:57:18 fray: back to the question... May 07 19:57:20 the thing is, this is the way Unix is designed to work.. giving you the ability to overload them and change behaviors as necessary May 07 19:58:42 fray: understood, you have to hack a box first, before being able to set LD_PRELOAD, so there is no real issue anyway... May 07 20:00:17 anyway, I added SRC_URI = "git:///home/jeroen/software.ext/pseudo;branch=FreeBSD;protocol=file" to pseudo_git.bb and bumped its priority.. May 07 20:02:42 this complains since I have sqllite in /opt/poky/bpp3/tmp-eglibc/sysroots/i386-freebsd/usr/lib/libsqlite3.a and not at /opt/poky/bpp3/tmp-eglibc/sysroots/i386-freebsd/usr/lib/pseudo/lib/libsqlite3.a May 07 20:03:40 of course I can adjust it to work, but there is likely an idea behind it, which I missed.... May 07 20:04:06 Okay, long story there. May 07 20:04:14 you should be able to pass the location of the SQLite directly.. May 07 20:04:23 the libdir isn't where to find other libs, it's where pseudo should put it's libs.. May 07 20:04:31 behavior got overloaded a while back May 07 20:04:31 It is desirable that pseudo's library directory NOT be anyone else's library directory, so it does /pseudo/*. May 07 20:04:37 Then it has pseudo/lib and pseudo/lib64. May 07 20:04:54 But sqlite should probably not be in the pseudo/lib directory, it's not pseudo-specific. May 07 20:11:24 fray, seebs: thanks, I am aware I can pass the dirs, but it is the bitbake recipe using these paths, hence I assumed there was some reasoning behind it.. May 07 20:11:50 when bitbake builds, it uses a static version of libsqlite and it's passed directly.. May 07 20:11:55 (or well it was unless its been change) May 07 20:11:55 fray, seebs: I will adjust the paths to see how much trouble is ahead of me... ;) May 07 20:12:26 thats the --with-sqlite-lib= value May 07 20:12:48 remember the "--libdir=" is NOT the directory to find other libraries.. it's the place to install pseudo May 07 20:13:21 look at the pseudo.inc for --with-sqlite-lib, --with-sqlite and --enable-static-sqlite May 07 20:13:24 those are the magic values May 07 20:18:16 kergoth? May 07 20:29:53 So, danny goes with bitbake 1.16 ? May 07 20:33:23 y May 07 21:18:09 find oe-core wr*/ meta*/ -type f | xargs grep find_cfgs May 07 21:18:12 oops May 07 21:29:18 JaMa, thanks May 07 22:04:25 JaMa: ping May 07 22:05:07 pong May 07 22:05:13 hi there May 07 22:05:18 hi May 07 22:06:00 please put on hold the meta-initramfs patches a bit more May 07 22:06:22 it looks like x86/x86_64 build is fixed May 07 22:06:37 so no need fo that masking May 07 22:08:10 while there, I'm updating to the latest kexec-tools-klibc_2.0.4 May 07 22:11:28 meta-initramfs patches are waiting for bluelightning, so ping him ^^ May 07 22:11:39 how so? May 07 22:12:00 JaMa: ant_home is also a maintainer, his changes don't need my sign-off :) May 07 22:12:38 :) May 07 22:12:59 true, but I'm keeping patches for layers with dedicated maintainers for them :) May 07 22:13:16 well as long as at least one of them has R/W :) May 07 22:16:40 anyway, I've lured GNUtoo and he's testing on x86 May 07 22:16:52 ;) May 07 22:17:32 ant_home: well let me know when you want the patches applied then May 07 22:17:49 sure, thx May 07 22:19:45 ERROR: The function busybox_cfg defined in busybox.inc was already declared in busybox-oe-min.inc. BitBake has a global python function namespace so shared functions should be declared in a common include file rather than being duplicated, or if the functions are different, please use different function names. May 07 22:19:47 oops May 07 22:51:48 maybe this is a weird question, but is there a way to disable the archive integrity check for a single package? May 07 22:53:36 mr_science: you wouldn't be trying to build from a git snapshot archive by any chance? May 07 22:57:01 no, i have plenty of git recipes May 07 22:57:41 this is a "special" one to build a package which installs the result of an android build May 07 22:58:28 so i have a jenkins build job the builds an .apk and a text file, but of which are additional SRC_URIs in a recipe May 07 22:58:37 s/the/that/ May 07 22:58:49 s/but/both/ May 07 22:59:00 * mr_science is rife with typos today... May 07 22:59:23 i just found what looks like what i was looking for May 07 22:59:51 base.bbclass checks for "OE_ALLOW_INSECURE_DOWNLOADS" so that looks like it **** ENDING LOGGING AT Wed May 08 02:59:58 2013