**** BEGIN LOGGING AT Wed Dec 03 02:59:59 2014 Dec 03 04:40:06 I was wondering if any one has ised how to resize root partition on first boot after dd. I created a script that I need to test but not sure how to run it before root / is mounted. What is the best way doing it? Dec 03 04:46:19 anyone had luck with x264 package? Dec 03 04:50:39 anyone had luck with x264 package? Dec 03 06:15:34 anyone had luck with x264 package? Dec 03 06:58:50 rburton: So far as I know, pseudo would look for a password file, and if it didn't find one in the system, it would fall back on host system passwd. But once it found a password file, it was done. It didn't check more than one file. Dec 03 07:22:47 morning! Is there a way to remove a file in a .bbappend that is added in the .bb ? Dec 03 07:35:16 hsychla: morning, yes. take a look at meta-fsl-arm/recipes-graphics/mesa/mesa_%.bbappend Dec 03 07:39:50 jabk, thanks! did not think of just deleting it in do_install_append. Dec 03 07:42:15 np Dec 03 07:46:05 anyone had x264 package working? Dec 03 07:58:24 good morning Dec 03 08:48:22 Hi ! I still have a compilation issue with hostapd, and can't ger rid of it... Here is the log: http://pastebin.com/93vhZ23T Dec 03 08:48:38 I tried RDEPENDS_${PN} += "libnl" in the recipe Dec 03 08:48:41 same issue Dec 03 08:48:57 and also INSANE_SKIP_${PN} += "compile-host-path" Dec 03 08:49:01 but still... Dec 03 08:50:40 morning all Dec 03 08:51:39 bluelightning: morning Dec 03 08:51:53 jmleo: hi... this is pretty straightforward, whatever you are building needs to be configured or patched so that it does not look in /usr/include/libnl3 Dec 03 08:53:16 bluelightning: the recipe is hostapd Dec 03 08:53:40 configuration will not be sufficient Dec 03 08:54:01 right, then it will need to be patched most likely Dec 03 08:57:29 looks like there are only two references to /usr/include/libnl3, in src/drivers/drivers.mk and src/drivers/drivers.mak, so it should be pretty easy to just patch those out Dec 03 08:58:22 although it may be necessary to replace those with the sysroot path, not usre Dec 03 08:58:32 bluelightning: the good question is : why is there references like these... Dec 03 08:58:41 yes, probably need to use sysroot Dec 03 09:06:11 jmleo: we may have become more strict about this sort of thing recently, some recipes might not have caught up Dec 03 09:07:12 bluelightning: ok, what is annoying is that the hostapd package is configured "in the marble" with this directory... patching it is ok, but the issue should be corrected on hostapd side... Dec 03 09:08:00 jmleo: possibly... you're more than welcome to take it up with hostapd upstream Dec 03 09:12:18 bluelightning: ping Dec 03 09:12:27 chankit: pong Dec 03 09:13:50 regarding bug report 5589, isomp4 wasn't built so it probably explains why qtdemux isn't there. Dec 03 09:13:58 I already posted that in bugzilla Dec 03 09:14:51 I have another question to throw though...regarding package x264, the build mysteriously wget something that didnt exist in the recipe specification Dec 03 09:15:30 so does yocto have a "hidden" mirror by default? Dec 03 09:21:02 chankit: we do have a default PREMIRRORS and MIRRORS value yes, but that just fetches a mirror of the upstream (and it would be expected that that is a tarball in the case of a git repository) Dec 03 09:24:43 bluelightning: the thing was the download failure on the undefined mirror on the recipe itself could have stalled the build process Dec 03 09:24:50 bluelightning: here's the fail log http://pastebin.com/HhBwEyvu Dec 03 09:25:39 chankit: is that the entire log or only part of it? Dec 03 09:26:09 bluelightning: partial log...the thing is that wget link wasn't defined in the recipe Dec 03 09:26:20 bluelightning: I'm referring to the x264 package btw Dec 03 09:26:23 the wget for the mirror would not be defined in the recipe, it wouldn't need to be Dec 03 09:26:51 the mirror is only a fallback (although it will try PREMIRRORS first) Dec 03 09:27:07 bluelightning: so may I unravel the mystery of wgetting the mirror? Dec 03 09:27:19 the issue here is probably that it could not fetch from the upstream git in your current network configuration Dec 03 09:27:27 bluelightning: so where are the mirrors refined? Dec 03 09:27:35 the fact that the mirror also failed is a secondary issue Dec 03 09:27:41 bluelightning: the thing is git cloning that repo worked for me Dec 03 09:27:53 bluelightning: so I Dec 03 09:27:54 meta-yocto/conf/distro/poky.conf Dec 03 09:28:02 'm confident I got my network config right Dec 03 09:28:22 it may be that git when executed within bitbake is unable to use whatever configuration you have set up outside of it Dec 03 09:28:40 brb, I have a meeting Dec 03 09:28:45 bluelightning: hmm...that can happen. Dec 03 09:28:46 sure Dec 03 09:28:57 see you like 2 hrs later perhaps Dec 03 09:33:20 where is a good spot to sit a script for boot time? Dec 03 10:59:41 jmleo: there? Dec 03 11:01:06 rburton: yes, but for 5 minutes Dec 03 11:01:13 I come back in ~2hrs Dec 03 11:02:05 jmleo: quick then: so my working theory is that the passwd your base-passwd is installing doesn't have a root entry in. are you modifying that? Dec 03 11:02:12 (re the passwd install root problem) Dec 03 11:02:32 rburton: I have wiped tmp and all sstate-cache files and this has solved the issue Dec 03 11:04:48 ah fair enough Dec 03 11:04:55 something must have got messed up Dec 03 11:31:26 where is rburton? :O Hi. Dec 03 11:33:52 I am just wondering why fixes on the first of December are not included in the pull request on the second of December? Is there some transient time for testing changes in the pull request? Dec 03 11:50:55 lpapp, rburton went offline about 50 mins ago. Dec 03 11:51:05 lpapp, jfyi Dec 03 11:55:20 lpapp, usually we'd like some time to comment on and review patches, thus there's usually a delay before they're included in a pull-request. Dec 03 11:55:52 Any specific patch you're thinking of? Dec 03 12:11:14 AndersD: http://permalink.gmane.org/gmane.comp.handhelds.openembedded.core/58181 Dec 03 12:11:17 hirata: thanks. Dec 03 12:12:22 lpapp, np! Dec 03 12:59:04 Is there a way to capture logs, or execute some task also when build/tasks fail? Or display logs from failed task? Dec 03 13:07:06 mcfrisk: we always capture output of every task to log.do_taskname under the temp directory of the workdir for the recipe Dec 03 13:07:14 (if that's what you're asking) Dec 03 13:26:30 bluelightning: ping Dec 03 13:26:37 chankit: pong# Dec 03 13:27:36 bluelightning: about the x264 mirror that we discussed before...how can I point the mirror to point to something that exists since wget-ing the mirror returns 404? Dec 03 13:28:04 chankit: just set PREMIRRORS Dec 03 13:28:12 chankit: but as I said earlier, that isn't the real problem here Dec 03 13:28:39 (well, to be fair, having your own local PREMIRRORS might be the proper solution anyway...) Dec 03 13:30:02 bluelightning: so you suggest that networks are the problem? Dec 03 13:30:37 bitbake-invoked git not being able to use your proxy is the main issue here yes Dec 03 13:31:58 bluelightning: I have already set the proxy inside local.conf so I'm not really sure what else I need to do to enforce the proxy inside the build environment Dec 03 13:32:32 chankit: maybe you have already seen it, but meta-yocto/conf/site.conf.sample should show how to configure the proxy correctly Dec 03 13:40:35 bluelightning: I actually glad that I have not seen that so I have another thing to fix. I'm wondering how my previous git related bakes got through.... Dec 03 13:47:53 bluelightning: btw, speaking of the packages_split thingy that you told me yesterday, I didn't see it inside the build directory (take python for example) Dec 03 13:48:02 where am i suppose to see it? Dec 03 13:48:57 chankit: the workdir for the recipe Dec 03 13:49:18 chankit: if you're unsure where to find that, you can get bitbake to tell you: bitbake -e recipename | grep ^WORKDIR= Dec 03 13:49:36 so e.g. bitbake -e python | grep ^WORKDIR= Dec 03 13:55:37 bluelightning: roger that...will check that once I got python finish baking at my home machine Dec 03 13:57:43 so will bitbake -e recipename | grep ^PACKAGES= also tell us the packages that a recipe splits? Dec 03 14:01:04 chankit: that's not complete list that will work for every recipe though - it doesn't take into account packages that are added dynamically during do_package (e.g. the plugin splitting that happens in gstreamer plugins recipes) Dec 03 14:01:37 the guaranteed way to get a full list is to build the recipe (at least up to do_package) Dec 03 14:13:48 bluelightning: interesting.....thanks again..have so much to learn. I'm afraid I will come out with more troubles though I bet so I will keep this channel updated Dec 03 14:14:25 sure, no problem Dec 03 14:14:37 how long did it take to create yocto building architechure? Dec 03 14:15:00 excuse my wrong spelling of arch..hure Dec 03 14:15:29 well, years... it's based on the good work of the OpenEmbedded project which started back in ~2005 Dec 03 14:16:22 and that adopted ideas from other systems including Gentoo (and Debian) Dec 03 14:16:23 to be honest, I really like this for its good error reporting mechanism which I don't see too frequently I'm afraid Dec 03 14:16:40 I can see the recipes mechanism coming from gentoo Dec 03 14:17:25 I'm glad you think our error reporting is good... I always think we can do better, but maybe that's the curse of being so familiar with it :) Dec 03 14:18:21 bluelightning: not to say that it's perfect but it's one of the best I have ever seen (or perhaps I'm not old enough to see things) Dec 03 14:18:42 bluelightning: are you one of the maintainers? Dec 03 14:19:34 chankit: yes, it's my job primarily to work on bitbake and the core parts of the system Dec 03 14:19:38 the only bad thing I'm nitpicking now would be there's not much people during my working hours (roughly day time for GMT+8) Dec 03 14:20:29 I do wish sometimes we had a dedicated support team, but then perhaps we would be less of an open source project... Dec 03 14:21:00 not that I mind helping people, far from it, I just know not everyone gets their questions answered Dec 03 14:22:01 I totally understand..because it's free world out here so when a person helps another for free, we can't expect much Dec 03 14:22:43 do you keep your IRC after work then? Dec 03 14:22:58 sometimes yes Dec 03 14:23:30 so how many people are paid to maintain yocto right now? Dec 03 14:23:35 10+? Dec 03 14:23:40 though I'm in GMT, so before ~9:30AM GMT I'm either asleep or getting ready for work Dec 03 14:24:02 probably more, if you count all of the various bits of the system Dec 03 14:24:12 and all of the companies contributing to it Dec 03 14:27:35 so comparing OpenWRT and yocto, can Yocto squeeze itself into router as well as OpenWRT? Dec 03 14:29:46 in theory, though I think OpenWRT has a lot more router-specific software and configuration than we do Dec 03 14:30:11 it's more that you could use our tools to build a router-suitable Linux distro (and I'm fairly sure that people have) Dec 03 14:48:07 bluelightning: true enough...are you under Linux Foundation then? Dec 03 14:49:25 chankit: no, I work for Intel Dec 03 14:57:15 bluelightning: owh okay...I thought yocto is under Linux Foundation hence my assumption Dec 03 14:57:55 chankit: it is, but various member companies of the project contribute engineering resources Dec 03 14:58:26 (and a number of non-member companies do so as well) Dec 03 14:59:52 bluelightning: sounds like serious politics Dec 03 15:00:56 not too much fortunately, at least not that affects day-to-day engineering anyway :) Dec 03 15:04:21 bluelightning: my log question is more about capturing all detailed logs from a build, wether the build passed or failed. Now I need to find and tar all logs after bitbake call, and have a special case if the build failed. I'd rather move this to bitbake somehow. Dec 03 15:06:08 mcfrisk: I don't think we have anything for that (other than what we're doing with Toaster) Dec 03 15:07:33 bluelightning: ok, so I guess wrapper script around bitbake is the only option atm then. Dec 03 15:07:53 bluelightning: speaking of toaster, I tried it but when I tried to access it by localhost:8200, I got error 501 Dec 03 15:08:03 has it got to do with proxies...again? Dec 03 15:09:50 chankit: no, I suspect it's a confusing message on startup, 8200 is the wrong port - I believe it's 8000 Dec 03 15:21:28 bluelightning: don't get you at confusing message on startup...was 8200 a typo then? Dec 03 15:24:56 chankit: I believe 8200 is the control port for bitbake, not the port you're meant to access with your web browser to see the interface Dec 03 15:39:01 bluelightning: okay that's interesting...didnt see port 8000 though Dec 03 15:39:19 but that's not urgent i would say...just wanted to try out Dec 03 16:20:56 Any good bitbake example recipe for building an application written in C using Makefile? Dec 03 16:24:43 stiandre_: have you check the user guide? there are examples. Dec 03 16:25:39 ndec: yes, i have, but does not found any good (suitable for my needs).. Dec 03 16:25:58 ndec: but thanks for answering Dec 03 16:26:27 stiandre_: what's missing from http://www.yoctoproject.org/docs/1.7/dev-manual/dev-manual.html#new-recipe-testing-examples? Dec 03 16:28:41 ndec: thanks, but do you know if there is possible to just execute make (oe_runmake) to use an already exiting Makefile, or do I need to put the Makefile content into do_compile? Dec 03 16:29:30 you can definitely have your own Makefile. Dec 03 16:29:51 see section 5.3.18.3 Dec 03 16:31:32 ndec: thank you.. haha, sorry for asking the obvious :) Dec 03 16:31:42 np. Dec 03 19:03:55 I'm writing a recipe for a package that requires ncurses, but even though I've added ncurses to DEPENDS, the configure step complains about curses.h not being present :s Dec 03 19:04:17 configure is probably looking in the wrong location or has an invalid test then Dec 03 19:04:39 I've seen poorly written configure checks for litteral '/usr/include/curses.h' which won't work when cross compiling Dec 03 19:06:10 Perhaps. I tried installing ncurses5-dev on my build machine, and it would pass then. Dec 03 19:06:35 ya.. definitely a bug in the recipe you are creating then.. it's looking at the host and should be querying the compiler instead Dec 03 19:06:40 Then it got stuck later on a similar thing, but just installing stuff locally kinda defeats the purpose of using Yocto Dec 03 19:06:42 configure has standard include tests.. thatswhat it should use Dec 03 19:06:57 you'll have to fix the poorly written configure Dec 03 19:07:07 fray: thanks, I'll have a look and see if I can get the configure script to check the right stuff. Dec 03 19:53:37 is there a linux tool to properly read a ring buffer - that would have the same effect as running " tail -f /var/log/something" Dec 03 19:54:15 I'm using remoteproc and announce RSC_TRACE so the remote proc can write some logs Dec 03 20:12:05 or does it exist a way to clear the trace buffer Dec 03 20:36:15 tomz1, how does the --size paramter work in a wks file? Dec 03 20:53:42 nm, I am an idiot Dec 04 00:21:44 bluelightning: ping Dec 04 02:05:15 I need to share some patch files and variables defined within a layer with other .bb files (some in this layer, some in another). If I want to refer to the patch locations, are their some appropriate var to use? Dec 04 02:05:21 s/are/is/ **** ENDING LOGGING AT Thu Dec 04 02:59:58 2014