**** BEGIN LOGGING AT Wed Apr 17 02:59:59 2013 Apr 17 03:20:30 so the ADT Installer uses a PMS repo to build a sysroot for the installer? Apr 17 03:20:37 er for the ADT/SDK* Apr 17 06:48:08 kergoth: Roughly speaking, the plan was to figure out exactly the information we needed, double check we had all the right info, then add in generic data driven bits if that made sense. I didn't what to end up adding broken interfaces or interfaces we didn't need Apr 17 06:48:28 kergoth: of course we never did get back to adding in the generic pieces but the intent is still there Apr 17 07:00:11 waynr: yes Apr 17 07:04:02 good morning Apr 17 07:09:15 halstead: ping Apr 17 07:10:53 morning mckoan Apr 17 07:13:17 AlexG: what's up? Apr 17 07:13:30 halstead: let Apr 17 07:13:36 s talk on private please :) Apr 17 07:15:17 AlexG: sure. Apr 17 07:29:27 priv msgs! Apr 17 08:13:58 morning all Apr 17 08:14:59 bluelightning: hey! Apr 17 08:15:05 hi mranostay Apr 17 08:18:57 hi bluelightning Apr 17 08:31:49 hi mattnie Apr 17 09:23:01 hi guys, is there a nice way to import an environment variable into a bitbake recipe? In this particular case, I need to access JAVA_HOME from inside a recipe Apr 17 09:26:35 Sput: you just need to add the variable name to BB_ENV_EXTRAWHITE Apr 17 09:26:45 that will allow it into the internal environment Apr 17 09:27:24 bluelightning: yeah, but it requires user configuration which I try to avoid Apr 17 09:27:35 unless I can do that inside the recipe? Apr 17 09:28:11 Sput: you can't no Apr 17 09:28:34 Sput: typically we don't rely on host tools that require passing in paths from outside Apr 17 09:28:44 so we don't hit this kind of problem often Apr 17 10:10:17 bluelightning: mmh. yeah, it's broken by design, but unfortunately I can't change that in this particular instance. I was playing around with BB_ORIGENV, but it's not working as expected Apr 17 10:12:00 Sput: I could only suggest supplying your own setup script which extends BB_ENV_EXTRAWHITE as appropriate Apr 17 10:14:59 I have a question about kernel configuration.. Is it possible to configure the kernel with a _defconfig file? Or how can i configure the kernel without using Kconfigs Apr 17 10:15:03 Sput: you should be able to query BB_ORIGENV and extract the data from there. How are you trying to use it? Apr 17 10:15:16 RP: failing hard at python :) Apr 17 10:15:29 mattnie: you can provide a defconfig for it to use, yes Apr 17 10:15:44 Sput: you mean you need an example? Apr 17 10:16:01 sec Apr 17 10:16:08 JAVA_HOME = "${@d.getVar("BB_ORIGENV", True).getVar("JAVA_HOME", False)}" Apr 17 10:16:12 is what I'm trying right now Apr 17 10:17:06 but of course I don't really know what I'm doing Apr 17 10:17:30 RP: So if i edit e.g. "tmp/work/imx6qsabresd-poky-linux-gnueabi/linux-imx-3.0.35-r32.10/git/arch/configs/imx6_defconfig" and build kernel new with "bitbake virtual/kernel -c compile -f; bitbake virtual/kernel" it uses the defconfig? Apr 17 10:18:03 Sput: "BB_ORIGENV", True -> "BB_ORIGENV", False Apr 17 10:18:05 RP: or do you have a short howto if i am wrong Apr 17 10:18:33 RP: what does this parameter mean btw? Apr 17 10:18:38 mattnie: you need to put the new defconfig in SRC_URI so it knows where to find it Apr 17 10:18:47 Sput: whether to try and expand the data or not Apr 17 10:19:04 aah. Apr 17 10:19:29 python () { Apr 17 10:19:29 origenv = d.getVar("BB_ORIGENV", False) Apr 17 10:19:29 abc = origenv.getVar("ABC", True) Apr 17 10:19:29 d.setVar("ABC", abc) Apr 17 10:19:29 } Apr 17 10:19:54 will set ABC from the environment. Your version is just the condensed one Apr 17 10:20:10 bluelightning: I guess I should mail scott this? Apr 17 10:21:10 RP: I guess so, the question would be where to put it Apr 17 10:21:29 it's also one of those limited use things, I suspect some people could abuse it :/ Apr 17 10:22:02 RP: I'd say enter a doc bug report, then it won't get lost Apr 17 10:22:16 RP: hmm, that seems to do something, but it now fails later. will investigate, thanks for the pointers for now :) Apr 17 10:24:28 bluelightning: right, its one of those double edged swords Apr 17 10:24:40 bluelightning: I've mailed Scott, cc'd you Apr 17 10:24:59 bluelightning: we can bugzila it if necessary Apr 17 10:26:58 Sput: bitbake -e | grep JAVA_HOME is a good debugging tool for this Apr 17 10:27:22 oh nice Apr 17 10:28:22 yeah, it seems to get the correct value now, but somehow the recipe fails where it didn't yesterday. but it also fails when I just hardcode the path, so something else is amiss Apr 17 10:32:11 what happened to beautiful bitbake-env utility? Apr 17 10:32:13 is it still developed? Apr 17 10:34:59 eren: it hasn't moved; I'd like to look at how we can integrate it in the coming development cycle Apr 17 10:35:16 (assuming bitbake-env is kergoth's utility) Apr 17 10:38:48 bluelightning: yeah, it's kergoth's Apr 17 10:39:19 It's nice to hear that it would be integrated in the future Apr 17 10:55:33 RP: can you help me? Do I have to add SRC_URI to conf/local.conf? If i am right I have found an example SRC_URI = "git://git.kernel.org/pub/scm/utils/kernel/kmod/kmod.git;protocol=git;tag=v${PV}" but i am t stupid to transfer it for my use.. Apr 17 11:02:01 mattnie: you would set that in a recipe Apr 17 11:02:57 specifically, the recipe for the piece of software you want to build (if it's something new you would create a new one) Apr 17 11:05:52 mattnie: http://www.yoctoproject.org/docs/latest/kernel-dev/kernel-dev.html#changing-the-configuration Apr 17 11:06:01 mattnie: I'd suggest reading that manual Apr 17 11:15:09 RP & bluelightning: thx I will read and try it :) Apr 17 11:31:14 can I build a 32-bit sdk on a 64-bit hosts? Apr 17 11:35:20 erbo: yes Apr 17 11:36:58 zecke: do you know how I do it? By default bitbake -c populate_sdk myimage just builds a 64-bit version Apr 17 11:37:32 erbo: I think the local.conf has documentation for that. just study it Apr 17 11:40:15 zecke: will do, thanks Apr 17 12:08:03 hi! what is the package/class responsible for creating initial files in /dev? I mean those that are statically present in an image's rootfs Apr 17 12:25:46 gizero: initscripts Apr 17 12:26:49 mckoan: hi! thanks... investigating on a missing /dev/ttyS2... Apr 17 12:36:00 gizero: udev didn't set it? Apr 17 12:38:16 strangely enough ttyS0 is there while ttyS2 is not... even if the kernel is correctly using ttyS2 as its console Apr 17 13:09:47 mckoan: noticed that my image was even missing the udev init script. Hence udevd was not running at all. Due to wrong mixture of meta-layers here Apr 17 13:09:55 mckoan: fixed now Apr 17 13:18:46 hm. I keep getting this backtrace, does that look familiar to anyone? http://dpaste.com/1061884/ Apr 17 13:19:29 I have no idea why it would want to launch any UI Apr 17 13:19:37 never seen anything like that before Apr 17 13:19:50 you need a UI to see what is happening.. the knotty UI is the text console Apr 17 13:20:03 and I'm pretty sure it worked before, I already cleaned my build dir and state cache Apr 17 13:20:14 how would I debug this? Apr 17 13:20:29 I've never encountered a bug like this, so I'm not sure.. Apr 17 13:20:51 it looks to me like something is failing in bitbake itself... but I have no idea how or why.. Apr 17 13:21:03 are you using master or a release? Apr 17 13:21:08 latest dylan Apr 17 13:21:27 but it also fails with 3b2713684e which is a bit older Apr 17 13:21:32 * fray goes to look to see if there have been any recent bitbake changes Apr 17 13:21:32 Sput: how are you invoking bitbake? Apr 17 13:21:44 just bitbake Apr 17 13:22:02 after setting up the usual build environment, of course Apr 17 13:22:26 ya, I don't see anything that would suggest this kind of failure.. sorry Apr 17 13:22:26 Sput: I'm just wondering how I would go about reproducing the issue Apr 17 13:23:09 still trying to determine if it really always fails with the same recipe (which worked properly yesterday) Apr 17 13:23:35 even if you have a faulty recipe, you should never get a failure like this Apr 17 13:23:37 unfortunately, I can't share our layer Apr 17 13:23:57 sure, but if you can boil it down to a simple test case that would really help a lot Apr 17 13:24:00 yep Apr 17 13:24:11 I definitely agree with fray, no backtraces should ever be printed Apr 17 13:24:35 I was hoping you'd seen something like this before and could point me in some direction for further investigation Apr 17 13:24:59 unfortunately no... not sure if RP or kergoth have ( http://dpaste.com/1061884/ for reference ) Apr 17 13:25:31 Sput: enough diskspace? Apr 17 13:25:47 yep Apr 17 13:25:54 75G should be sufficient :) Apr 17 13:25:57 Sput: It the use of BB_ORIGENV Apr 17 13:26:02 it's caught me out before :( Apr 17 13:26:12 Sput: the variable is not pickleable and you're upseting the signature generation code Apr 17 13:26:23 RP: well, mayybe it started with me playing around with the python stuff, but even if I remove the line it still fails Apr 17 13:26:25 hmm Apr 17 13:26:52 unfortunately I don't even know what "pickleable" means :) Apr 17 13:26:56 Sput: you might have to manually force the dependencies or something Apr 17 13:27:27 RP, is there any way we can catch that condition and print a better error? Apr 17 13:27:42 fray: yes Apr 17 13:27:54 we should file a bug (with a way to reproduce it) and we should fix it Apr 17 13:28:15 right now, the rc2 release build is due in hours and I have a more pressing issue I need to look at though :/ Apr 17 13:28:39 RP: so what would be a proper workaround/fix in my recipe? which dependency I need to add? Apr 17 13:28:48 sorry that I'm so incompetent here :) Apr 17 13:29:20 Sput: Off the top of my head I don't know, I'd need to write up the example I deleted earlier, then try some things Apr 17 13:29:31 ok Apr 17 13:29:32 Sput: I need to look at something else right now :( Apr 17 13:29:45 understandably :) Apr 17 13:30:42 could it happen that a python snippet I commented out with # in the recipe still gets evaluated somewhere? Apr 17 13:30:48 that might explain why it still fails Apr 17 13:31:04 Sput: commented out it shouldn't be the issue Apr 17 13:31:23 ok. I'll try out some more stuff once the prerequisited have finished building Apr 17 13:31:30 *prerequisites Apr 17 14:54:35 test Apr 17 15:30:56 RP: so apparently, this python stuff is actually evaluated even in a commented line. after removing it completely, I can at least build without a crash (although of course that doesn't solve the actual problem that I'd like to be able to import a variable from the environment) Apr 17 15:31:43 I wish that had occured to me before wiping my build directory twice :) Apr 17 15:32:04 Sput: you mean a commented line in something like a shell function don't you? :/ Apr 17 15:32:19 yes Apr 17 15:32:27 Sput: you didn't say that earlier Apr 17 15:32:52 Sput: bitbake expands the python in shell functions before it tries to understand the shell Apr 17 15:33:00 RP: oh, I didn't know that Apr 17 15:33:04 today I learned :) Apr 17 15:33:29 actually. I wonder if I can somehow set that in global scope then Apr 17 15:33:48 Sput: it will still try and parse it if you use it somewhere Apr 17 15:34:01 bitbake's variable handling is so confusing :/ Apr 17 15:34:17 RP: hm okay Apr 17 15:35:14 well, then maybe all that's left is whitelisting in my local conf... although I think there should be some official and working way of importing random env vars into a recipe Apr 17 15:35:40 the main source of confusion in this regard is the fact that bitbake's variables and shell variables use the same syntax. ${}. Apr 17 15:40:21 Sput: are you using danny? Apr 17 15:41:03 hm, where is the right place to report a bug for the meta-raspberrypi layer? Apr 17 15:41:19 gkiagia: what does the layer README say? Apr 17 15:42:21 well, it doesn't say Apr 17 15:43:20 gkiagia: it really should. I'm email the listed maintainer and suggest they add that information Apr 17 15:43:48 RP: dylan Apr 17 15:44:06 Sput: hmm, I didn't get the error here with the anonymous python Apr 17 15:46:27 JAVA_HOME="${@d.getVar("BB_ORIGENV", False).getVar("JAVA_HOME", True)}" Apr 17 15:46:27 is the exact phrase I'm using, maybe I have some syntax error in there? Apr 17 15:46:39 it works in the sense that the variable has the correct value afterwards Apr 17 15:48:23 isn't BB_ORIGENV a dict, rather than a DataSmart object, or am I remembering wrong? Apr 17 15:48:54 kergoth: its a datasmart Apr 17 15:49:03 ah, okay, cool Apr 17 15:49:23 Sput: do JAVA_HOME[vardepvalue] = "${JAVA_HOME}" Apr 17 15:49:39 Sput: that should make it happier, I could reproduce here and it sorted it Apr 17 15:49:54 Sput: if you'd tried the anonymous python I showed you, that didn't have this problem Apr 17 15:50:56 can I just use this snippet in a shell function? Apr 17 15:52:11 Sput: do this outside the shell function, then use the variable in the shell function Apr 17 15:52:17 ok, lemme try Apr 17 15:52:27 this shell/python mix is also I still need to fully understand :) Apr 17 15:52:47 Sput: i.e. both the JAVA_HOME assignment and the vardepvalue Apr 17 15:53:05 Sput: then in the shell just do JAVA_HOME=${JAVA_HOME} Apr 17 15:53:22 I know this looks ugly but it should work Apr 17 15:56:00 don't care about the looks :) Apr 17 15:56:43 I would like to mention that I really love how helpful people in here are. bitbake can be pretty confusing for a relative noob, and I've had my problems solved by you guys more than once already Apr 17 15:57:10 Sput, that's part of the goal of the YP.. a community where you can get answers to issues like this Apr 17 15:58:20 and now I got it to work indeed \o/ Apr 17 15:58:23 thanks a lot! Apr 17 15:58:25 interesting, latest rebuild from sstate-cache was failing with "Fatal error: invalid -march= option: `armv5te'" in every target image, rebuilding toolchain fixed that Apr 17 15:58:57 s/image/recipe/ Apr 17 15:59:22 Sput: bitbake is pretty confusing for a grizzled expert too Apr 17 15:59:33 ;) Apr 17 15:59:40 ls Apr 17 15:59:42 oops Apr 17 16:01:23 fray, of course, I must point out the community pre-dates the yocto project :) Apr 17 16:01:32 yes Apr 17 16:02:48 back in the good old days you got mocked for not understanding bitbake! ;) Apr 17 16:02:58 ha Apr 17 16:03:49 maybe this particular issue would make a good addition for utils.bbclass? import_env_var() or something? Apr 17 16:03:57 and depending who is on, you might get mocked again :) Apr 17 16:04:41 zeddii, ping Apr 17 16:05:36 is there a wuick answer to this Apr 17 16:05:38 http://pastebin.com/hBYNrz2A Apr 17 16:06:00 gah, I need to check one more thing Apr 17 16:11:17 I needed to move my yocto directory to another hard disk and bitbake now tells me I need to rebuild. But how to do this? Apr 17 16:12:10 a simple bitbake -c clean does not resolve my situation Apr 17 16:12:33 also, do not forget to specify the defconfig if you are using linux-yocto-custom without the kconf stuff Apr 17 16:12:34 you move the 'sstate-cache' directory to the new machine, along with your downloads.. Apr 17 16:12:36 then start a build.. Apr 17 16:12:49 it will re-use as much as it can.. which SHOULD be around 95% or more fo the build Apr 17 16:13:01 (ohh and of course the configuration needs to be the same..) Apr 17 16:13:10 if the host type has changed, the reuse will be closer to about 80% in my experience Apr 17 16:13:23 the key though is the sstate-cache holds all of the reusable components Apr 17 16:15:02 ok thanks for your fast reply Apr 17 16:15:17 i am currently testing it Apr 17 16:17:19 how robust is the sstate-cache for sharing between multiple simultaneous builds (in different build directories of course), possibly for different targets too? Apr 17 16:18:04 it should 'just work' Apr 17 16:21:22 Sput: tried and tested in that scenario Apr 17 16:21:33 that's good to know :) Apr 17 16:24:26 for integrating a Yocto setup in a CI system that supports various build configurations that should save lots and lots of rebuild time Apr 17 17:02:32 gm all Apr 17 17:03:49 Jefro: I'm listening in on mute. Apr 17 19:21:02 I have a kernel config fragment that is not making it to the .config; How do I figure out why? Apr 17 19:27:11 ok, I have a funny do_configure_prepend included by my kernel recipe.... Apr 17 21:26:04 Crofton|work: This mailing list problem is now seriously hurting the release of the project Apr 17 21:26:27 Crofton|work: I'm not seeing things from people and am having to ask them to send me the messages off list Apr 17 21:26:28 people are supposed to be working on this Apr 17 21:26:45 Crofton|work: I just want to make it clear this is killing things for me :( Apr 17 21:27:03 can't work effectively with this lag, its stupid Apr 17 21:27:22 to be fair, I've also heard issues with yp lists and lag at times Apr 17 21:27:40 let me check the last email I received on the subject Apr 17 21:27:41 Crofton|work: which got investigated and fixed within hour s afaik Apr 17 21:27:41 yes it has been a problem there lately too Apr 17 21:28:01 all the people who need to be working together know they need to proceed Apr 17 21:32:11 * mranostay waves from Collab Apr 17 21:35:34 hi Crofton|work Apr 17 21:35:41 why are you not here? Apr 17 21:37:10 save some time Apr 17 22:23:51 hmmm, I wonder why this build always wants to rebuild the kernel every time I run bitbake? Apr 17 22:29:50 thaytan: does the kernel version include DATETIME by any chance? Apr 17 22:34:48 bluelightning: no - linux-ti81xx-psp-2.6.37-r1 Apr 17 22:54:44 thaytan: hmm not sure, I would suggest asking on the meta-ti list Apr 17 22:55:34 thaytan: kernel.bbclass does use DATETIME but it is supposed to be set up so that doesn't cause a rebuild when that value is different Apr 17 22:55:52 thaytan: ti? presumably this isn't work, right? :) Apr 17 23:00:26 bluelightning: yeah, it seems weird that it's just that recipe. I'll keep poking Apr 17 23:01:02 rburton: evaluating some hardware Apr 17 23:01:20 thaytan: I'd suggest using bitbake-diffsigs, but that may not be easy to use with denzil (you are using denzil right?) Apr 17 23:03:11 bluelightning: no, this is danny Apr 17 23:04:23 thaytan: I can't see a linux-ti81xx-psp recipe in meta-ti on danny, but I guess it's possible you're not pulling that from meta-ti I guess Apr 17 23:04:46 bluelightning: it was deleted about a month ago Apr 17 23:05:14 I see Apr 17 23:05:40 anyway, I'll either figure it out or tolerate the rebuilds :) Apr 17 23:06:37 well bitbake-diffsigs ought to be able to tell you what changed that altered the signature and triggered the rebuild Apr 17 23:06:55 thaytan: do you have TARGET_ARCH in the machine config, by any chance? **** ENDING LOGGING AT Thu Apr 18 02:59:58 2013