**** BEGIN LOGGING AT Tue Nov 04 03:00:00 2014 Nov 04 07:54:54 lpapp_: np Nov 04 08:26:05 good morning Nov 04 09:09:34 morning all Nov 04 09:38:50 Hello, may someone help me, I'm looking for an option to create a rpm of the kernel sources (for debugging on other machines) Nov 04 10:21:28 I'm trying to build a core-image-minimal image from the dizzy branch, but qemu-native fails to link: http://paste2.org/9PfFEZ9x Nov 04 10:21:36 Any ideas? Nov 04 10:23:09 I might add that I'm not using one of the officially supported/tested host distros, but Arch Linux Nov 04 10:28:47 simonl, well, g_type_check_instance_is_fundamentally_a seems to come from glib. Which version of glib do you have? Nov 04 10:37:13 AndersD: 2.42 Nov 04 10:45:11 That symbol should be in 2.42. http://paste2.org/9PfFEZ9x Nov 04 10:47:08 I think you'll have to double-check which libraries gets linked and from where. I'd have guessed that ld tried to link against an older glib... Nov 04 10:48:33 AndersD: Ok. That paste is mine btw, did you mean to post a different one? Nov 04 10:53:53 No, that was a mistake... ;) Nov 04 10:58:41 Any Ideas on how to build a kernel source rpm? Nov 04 11:30:45 Is it possible to restrict the archiver for some modules> Nov 04 11:45:30 m42e: yes, IIRC there is the ability to filter it per recipe and per license Nov 04 11:46:45 @bluelightning_ any suggestions to search for? I didn"t get any hint to that till now Nov 04 11:52:00 Morning... Nov 04 11:52:33 bluelightning: Do you know much about IMAGE_TYPE use? Nov 04 12:00:07 If I want to create a recipe that takes the image outputs of other recipes (i.e. u-boot, kernel, fs) to then combine them into a single archive are there any examples of this? Is doing something like one of the DEPLOYABLE_IMAGE_TYPES sensible? Nov 04 12:02:53 m42e: I'd have a look at the comments at the top of the meta/classes/archive*.bbclass files, if there's no useful info there I'll try to dig something out Nov 04 12:03:50 pev: I'm not an expert in that area I'm afraid Nov 04 12:04:55 pev: I know you can have dependencies between image types to ensure that you get one built before the other, and you can add task dependencies to ensure other recipes have done their do_deploy e.g. u-boot Nov 04 12:06:48 bluelightning: Yeah, I get the dependencies but wanting to put together what's effectively a combined firmware file but not sure how best to do it under Yocto. Nov 04 12:07:51 bluelightning: I can bodge it with a script in a recipe but the deployable image type mechanism looks better suited. I'm surprised there's nothing similar implemented, people can't always manually put together full firmware loads...? Nov 04 12:12:32 pev: it depends on how it has to be laid out Nov 04 12:13:49 we have support for generating initramfs images and embedding them into a kernel image, we have support for building x86-bootable images, and through wic we have support for building almost any kind of multi-partition disk image with a bootloader applied Nov 04 12:14:05 I took a look in the comments but the only way to filter is IMHO the license Nov 04 12:15:22 m42e: hmm, you may be right Nov 04 12:15:41 m42e: it would be worth opening up an enhancement bug in bugzilla for this Nov 04 12:16:01 in any case adding the capability should be a pretty straightforward modification to copyleft_filter.bbclass Nov 04 12:16:04 brb Nov 04 12:17:03 bluelightning: I have a firmware format that combines u-boot, DT, kernel, rootfs, FPGA images. At the moment I build the various recipes and manually run a script to generate the combined image but I'd ideally like to get it into a sensible recipe to do automatically instead. Nov 04 12:47:01 bluelightning: hmm, the file is part of poky, I'm not sure which product/component to choose for the issue Nov 04 13:10:47 m42e: it would go under oe-core Nov 04 13:21:32 hi.. i got some problem with yocto sdk when building gst-plugins-base, i got some libtool problem: http://paste.ubuntu.com/8818962/ Nov 04 13:21:40 bluelightning: done Nov 04 13:21:49 bluelightning: thanks a lot Nov 04 13:21:57 m42e: thanks Nov 04 13:23:09 pev: I have implemented my own imagetype class and added it to IMAGE_CLASSES: 'IMAGE_CLASSES += "myimagetype". myimagetype.bbclass inherits from image_types to do all the deafult stuff and then I have added any tasks I like to do more. Nov 04 13:24:07 pev: I need to run now, but I can answer questions later if you like. Nov 04 13:26:04 Hi, anyone an idea on why when building with INCOMPATIBLE_LICENSE = "GPLv3" bitbake selects readline_6.3.bb which is GPLv3+ instead of the readline_5.2.bb ? Nov 04 13:26:51 olani: Nice one thanks! I'll take you up on that... Nov 04 13:35:45 pev, I've got something similar, where I'm adding a custom image type which creates am UBI-image for a number of partitions in it. Nov 04 13:41:52 AndersD: Does it include other images as well? I think that's the trickier bit? Nov 04 13:44:25 That exact image, no, but give me a minute and I should be able to find a variant of that image that we developed for a customer... Nov 04 13:44:43 That exact image, no, but give me a minute and I should be able to find a variant of that image that we developed for a customer... Nov 04 13:45:36 Hm, wouldn't it be nice to have a couple of standard compound firmware file formats that are flexible and a configurable firmware update mechanism to match... Always re-inventing the wheel...! Nov 04 13:51:02 Well, I've got quite some hope for wic, though I haven't had time to play around with it yet. In worst case, it should only be to develop a few plugins for wic. Nov 04 13:52:33 Nevertheless, the image_types bbclass we used for the customer did only reference the second file system image in a way similar to http://paste2.org/pC1pts3W Nov 04 13:52:49 The trick to get it to automatically build the second-image was: Nov 04 13:53:53 Adding this to our main recipe: do_rootfs[depends] += "second-image:do_rootfs" Nov 04 13:54:37 And possibly do_deploy[depends] += "u-boot:do_deploy" (I'm not sure if this was needed, or if it was added anyway). Nov 04 13:55:16 Adding the do_rootfs[depends] above, force bitbake to build second-image when you're building your main imaeg Nov 04 14:16:49 pev, AndersD: I combine our bootloader with a custom partition table and a set of partitions into a firmware image and then deploy it. I added a task for each discrete part that should go into the final image and one task to concatenate the parts. Nov 04 14:17:55 I've looked into replacing the whole mess with wic, but had to focus on other tasks. Nov 04 14:34:48 hi. i need a hint. i need to enable/disable a service (ntpd) from python. what is the best way to do that? i'm using traditional sysvinit, not systemd ... Nov 04 14:37:48 should i use remove and symlink from /etc/init.d to /etc/rcX.d, or is there a better tool for that? Nov 04 14:44:52 ionte: where is this python code running? Nov 04 14:45:25 ionte: typically you'd use the update-rc.d script to manage those symlinks Nov 04 14:47:05 void-dev: that shouldn't happen, it should pick the 5.2 version under those circumstances Nov 04 14:47:20 void-dev: which version of the build system are you using? Nov 04 14:50:43 AndersD, olani: Intersting, I've not heard of wic before... Nov 04 14:56:44 bluelightning: dizzy Nov 04 14:58:07 bluelightning: BB_VERSION = "1.24.0" Nov 04 15:00:03 void-dev: ok, that sounds like a bug in that case - is it producing a warning / error or just going ahead and building it? Nov 04 15:00:23 bluelightning: it is just going ahead Nov 04 15:00:40 bluelightning: just saw it by looking at the license.manifest Nov 04 15:00:52 ok, that is bad... we need to fix that Nov 04 15:01:09 void-dev: would you mind filing a bug in bugzilla? Nov 04 15:02:21 pev: The problem with wic is that you probably have to write new plugins for anything that is not grub or uefi (I forget exactly what is supported) Nov 04 15:02:39 bluelightning: the python code is running on the device where i want to enable/disable services. yocto dist. Nov 04 15:02:58 ionte: ok, in that case definitely use update-rc.d; you should find it's already installed Nov 04 15:03:14 bluelightning: i have no update-rc.d script unfortunately. perhaps it's in another package... Nov 04 15:03:37 bluelightning: will do once I can reproduce using core-image-minimal build (instead of a meta-ivi based build) Nov 04 15:03:54 ionte: ok, AFAIK the package is called update-rc.d as well Nov 04 15:05:51 bluelightning: ok, thanks! Nov 04 15:09:26 ionte: FWIW, packages that install initscripts should get update-rc.d added to their RRECOMMENDS, although I think the package might end up getting uninstalled if you don't have package management enabled Nov 04 15:14:23 bluelightning: Nov 04 15:14:46 bluelightning: strange. i added update-rc.d but i still do not have the command available... Nov 04 15:15:31 ionte: do you have package management enabled or disabled in the image? Nov 04 15:16:38 so after a "slight" interruption, I'm back to trying to build dizzy... Nov 04 15:17:23 bluelightning: hm, how do i see that? i have not added it myself, and my image inherits from "image" Nov 04 15:17:35 building qemu fails, and mucking about a little in the devshell shows that (at least when building there) some libraries are linked from a poky sysroot, and some from the host system Nov 04 15:17:58 ionte: ok so I assume you don't (you would need to have "package-management" in IMAGE_FEATURES, and that's not the default) Nov 04 15:18:16 simonl: qemu-native you mean? Nov 04 15:18:24 I'm using an unsupported host distro (Arch), but somehow this doesn't feel like something that should depend on the host distro Nov 04 15:18:26 bluelightning: yes Nov 04 15:18:46 simonl: native recipes build for the build host, so they will link to libraries on the host Nov 04 15:19:41 bluelightning: except this one uses libgobject from the poky sysroot which is missing symbols the host libs need Nov 04 15:20:23 simonl: which host lib(s)? Nov 04 15:20:36 at least that's what I think is happening, I'm not super experienced with linking problems :P Nov 04 15:23:48 bluelightning: I think gtk-x11-2.0 is the one (there is an error related to that one at least) Nov 04 15:23:56 http://paste2.org/F1d7pfO0 Nov 04 15:24:15 make LDFLAGS="-v -t" run in a devshell Nov 04 15:25:26 (incomplete) build log: http://paste2.org/9PfFEZ9x Nov 04 15:26:22 bluelightning: you seem to be right. update-rc.d is removed if packaging is not enabled. it is removed by "image.bbclass" ... Nov 04 15:30:00 ionte: unfortunately there is a baked in assumption that if you don't intend to install packaging that you won't need to add and remove services Nov 04 15:30:20 if you don't intend to install and remove packages, that is Nov 04 15:33:08 yes. very strange assumption. Nov 04 15:33:29 i need to enable/disable ntpd depending on preference of the customer. Nov 04 15:42:47 my CFLAGS, LDFLAGS, PKG_CONFIG_PATH, etc in the qemu-native devshell contain various references to $BUILDDIR/tmp/sysroots. is this expected? Nov 04 15:43:57 bluelightning: bug filed Nov 04 15:54:31 void-dev: thanks, I've confirmed the issue and updated the bug Nov 04 15:55:29 YPTM: armin on Nov 04 15:56:39 YPTM: Ready-Access Number: 8007302996/9139049836 Access Code: 2705751 Nov 04 15:56:47 YPTM: Stephen Joined Nov 04 15:58:29 YPTM: Alex Vaduva here! Nov 04 15:58:51 YPTM: Saul is on Nov 04 15:59:12 YPTM: tom z on Nov 04 15:59:20 armpit: you still OK doing the stable for 1.7 Nov 04 15:59:29 yep Nov 04 15:59:35 armpit: thanks Nov 04 15:59:54 Sona is on Nov 04 16:00:15 YPTM: Cristian joined Nov 04 16:01:50 sona: I think I forgot to mention we have some tools that use the upstream cvechecker, is that something you are thinking about for Security related or not? Nov 04 16:01:56 YPTM: Richard joined Nov 04 16:02:13 YPTM: tom z on Nov 04 16:02:46 YPTM: what tool? Nov 04 16:02:49 YPTM: Alin join Nov 04 16:03:03 YPTM: belen joined Nov 04 16:04:08 YPTM: I have a conflict with another meeting FYI Nov 04 16:04:09 yeah, about 1.7... I am still waiting for my cold (free) beer :-P Nov 04 16:05:06 YPTM: Michael is attending IRC only today. I can't make the call. Nov 04 16:05:39 Sona: see a branch from Mulhern (an intern we had last year) http://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=mulhern/cvechecker-final2 Nov 04 16:08:06 cristianiorga: it'll be good to see it :) Nov 04 16:09:38 YPTM: thanx sgw (is this Saul?) Nov 04 16:11:09 YPTM: I would like to update “https://wiki.yoctoproject.org/wiki/Security” page, some old info there, I would like to add some new info  Nov 04 16:12:16 Sona: yes sgw == Sau! Nov 04 16:12:21 Sona: please do, all of the content there can be replaced Nov 04 16:12:31 (this is Paul speaking) Nov 04 16:12:38 * zeddii has joined the meeting late Nov 04 16:12:41 well, typing... Nov 04 16:12:58 halstead: can you check if Sona can have an account on the Wiki Nov 04 16:13:12 sgw_, Yes. One moment. Nov 04 16:14:16 woo! Nov 04 16:14:28 YPTM is done. Nov 04 16:14:39 * zeddii barely snuck in to hear anything Nov 04 16:15:12 armpit: which email do you want people to CC you for patches? I see a mvista and gmail that you have used Nov 04 16:19:53 sgw_, Sona, Your wiki account SonaSarmadi is approved. Nov 04 16:20:27 halstead: thanks! Nov 04 16:24:04 Thanx "halstead" :) Nov 04 16:28:24 halstead: shall I go to " request one" in the wiki and reqest an account or just use this and login? what passwd to use in this case? Nov 04 16:29:24 Sona, it should have e-mailed you a login link. Nov 04 16:30:18 sgw_: gmail Nov 04 16:31:44 halstead: hmmm I haven't got any yet :) Nov 04 16:35:54 Sona, I'll try to make it re-send your password. Nov 04 16:36:16 ******** Nov 04 16:36:53 halstead: maybe email address was not correct? please use sona@enea.se or sona.sarmadi@enea.com Nov 04 16:37:49 Sona, That's the problem. The e-mail is set to sone.sarmadi@enea.com Nov 04 16:41:27 RP: Regarding your commit 6bd2d9d395 (base.bbclass: Enable using 'make clean' for rebuilds), wouldn't it be a good idea to allow the part that checks for the existence of a Makefile and runs make clean to be overridable somehow? Nov 04 16:42:49 RP: The reason I ask is because we have a recipe that broke because of that commit. It has its own do_compile which does make in two subfolders. However, there is a Makefile in the topfolder, but it has no clean target... In this case I would have liked to be able to specify that the make clean should take place in the two subfolders... Nov 04 16:43:08 Saur: you can though, just by defining your own empty do_configure Nov 04 16:43:35 if we could check to see if the clean target is valid that would be an improvement, but I don't know if that's practical Nov 04 16:46:21 bluelightning: Well, true, I can override the do_configure, but then I don't get the expected benefit (unless I re-implement all of base_do_configure). Nov 04 16:46:59 sounds like we need a new shell function to do a build-level clean Nov 04 16:47:27 bluelightning: Something like a do_configure_clean that I could override to specify how the clean should be done... Nov 04 16:48:22 kergoth: Yes, exactly. Nov 04 16:48:53 right, understood Nov 04 16:48:56 kergoth: Though I guess it can easily be confused with bitbake -c clean ... Nov 04 17:01:57 Saur: yes, I think we need to improve this Nov 04 17:02:51 RP: Good. At least I have a simple workaround for my build problem now until it is improved. Nov 04 17:03:18 Saur: I can't tempt you to send a patch to improve it? Nov 04 17:03:28 :) Nov 04 17:07:26 RP: You can, but I am not sure what would be a good way to do it. I mean, making the part that does the cleaning in do_configure overridable is probably not too hard. However, I can imagine making a build-level clean a separate BitBake target would make it more generic. And how that would affect other BitBake targets I do not know... Nov 04 17:08:08 just add a new shell function/task and run it from do_configure. should be straightforward Nov 04 17:09:46 kergoth, RP: And what to call it? do_clean() is obviously already taken... Nov 04 17:10:13 i'd probably call it makeclean or buildclean or something, but i suck at names, presumably someone else will have a better idea Nov 04 17:10:35 start a thread on the list, i'd say Nov 04 17:30:24 RP, bluelightning & al: With the change of making overlapping files an error (5cc591748e), how are you supposed to handle renamed recipes? Wiping tmp just because someone renamed a recipe seems somewhat excessive... Nov 04 17:32:11 Sona, Were you able to log in? Nov 04 17:37:15 halstead: I haven't got any email :( Nov 04 17:57:17 Sona, I sent you the credentials directly in IRC. I'll send them via e-mail again. Nov 04 18:28:43 Saur: that's probably a question for RP... Nov 04 18:28:54 Saur: I know that he's been trying to figure out how to properly solve that entire issue for a while Nov 04 18:29:06 bluelightning: Ok Nov 04 18:29:07 the change in policy was in response to several bugs being reported that were a result of junk being left in the sysroot Nov 04 18:29:29 Saur: this is an area we have a problem, and an open bug :/ Nov 04 18:29:35 it might be that the renamed recipe issue specifically could be handled somewhat automatically, it's the general case that's hard Nov 04 18:29:39 Saur: it turns out to be a hard problem Nov 04 18:30:16 a workaround for now is to clean the recipe before doing the rename (only works if you know about the rename, of course) Nov 04 18:30:18 Is there some easy way to remove a manifest and all files mentioned in it? Nov 04 18:31:16 Saur: there are functions for that in sstate.bbclass but they're not easily accessible externally Nov 04 18:31:17 bluelightning: Unfortunately you typically don't... Nov 04 18:31:26 Saur: right :( Nov 04 18:32:20 bbl Nov 04 18:32:36 A small improvement would be if the manifest file was mentioned with an absolute path so I wouldn't have to go hunting for it... Nov 04 18:32:54 (locate was my friend) Nov 04 19:10:01 I created a recipe to build qt-creator 3.2.1 for the target. Should i send a patch to the mailing list for meta-qt5 or should it be reviewed in github before sending the first patch? Nov 04 23:34:29 Who should I ping about 'Downloads' page issues? Nov 04 23:55:53 otavio: halstead Nov 05 00:07:26 bluelightning: thx Nov 05 00:14:48 bluelightning: The layer index needs dizzy addition Nov 05 00:15:05 otavio: ah yes I forgot about that Nov 05 00:15:37 bluelightning: thx **** ENDING LOGGING AT Wed Nov 05 03:00:00 2014