**** BEGIN LOGGING AT Tue Jan 10 03:00:02 2017 Jan 10 09:22:48 I just started experimenting with IMAGE_FEATURES_append = " read-only-rootfs" and have a question regarding /var/lib which is mounted as tmpfs. Jan 10 09:23:17 I use opkg and it appears that the contents of the package database is stored in /var/lib Jan 10 09:24:25 deva: what would you suggest considering you're asking for a read only rootfs? Jan 10 09:24:26 First of all, I would like the database to not occupy tmpfs space since it will never be altered during runtime without the rootfs being remounted rw first Jan 10 09:24:26 what's the meaning of "out-of-tree" that is mentioned on the kernel dev manual Jan 10 09:25:12 aV_V: built objects in a different directory to the sources Jan 10 09:25:32 rburton: I have a small problem with galculator in the SDK tests. It needs intltool :/ Jan 10 09:25:54 rburton: thanks Jan 10 09:26:24 rburton, Perhaps changing /var/lib back to a regular ro mount and then have symlinks to specific packages that need to write into /var/volatile like it is done with eg. /var/log Jan 10 09:26:27 rburton: I can fix it by extending the intltool recipe to a nativesdk version and including that in the sdk Jan 10 09:27:18 maxin1, otavio: are you able to build icedtea-native without X11? Jan 10 09:27:51 RP: oh, was it using the host intltool all this time? Jan 10 09:28:08 mario-goulart: not yet, will do that. sorry for the delay .. Jan 10 09:28:26 deva: yes Jan 10 09:29:32 RP: intltool nativesdk should be in the sato sdk for sure. Jan 10 09:29:55 probably glib nativesdk too to be honest for those tools Jan 10 09:30:48 maxin1: oh, sorry. I don't intend to make any pressure. I just wanted to know if somebody actually has been able to build it without X11. As far as I can tell, that's currently not possible. Jan 10 09:33:51 I'm not sure if that's easily doable. Does anybody know how mainstream distros build openjdk? I know they have separate packages for the headless part, but I wonder if they actually build it headless or if they build it headful and just split packages afterwards. Jan 10 09:34:05 rburton: the intltool recipe doesn't even have the BBCLASSEXTEND Jan 10 09:35:24 rburton, Should I alter the bahviour using a post install script for opkg or how would you suggest I approach that task? Jan 10 09:36:47 I'm not really familliar with how the opkg package is integrated with the system, since it is included using PACKAGE_CLASSES ?= "package_ipk" Jan 10 09:37:22 sorry... it is IMAGE_FEATURES_append = " package-management" Jan 10 09:39:55 deva: have a look at meta/files/fs-perms.txt (look for link). Are you using systemd or sysvinit? Jan 10 09:40:37 RP, currently I'm using sysvinit, but I am planning on changing to systemd Jan 10 09:40:44 deva: initscripts/initscripts-1.0/volatiles for sysvinit or volatile-binds tecipe for systemd Jan 10 09:41:39 deva: I suspect you'd need to tweak the system layout Jan 10 09:42:57 RP, Perfect, thanks. That was just the information I needed :-) Jan 10 09:43:22 RP, Can I somehow overlay the fs-perms.txt file? Jan 10 09:45:28 deva: Its searched for though BBPATH iirc so putting a copy in your own layer should work if the layer ordering is right Jan 10 09:46:32 RP, Ok, so I just need a full copy of it. Do you know if it is possivble to use some kind of include statement? Jan 10 09:47:04 So I could include the fs-perms.txt in poky and then just add the lines I need to modify. Jan 10 09:47:08 deva: I suspect not, you'd need to look at the code that reads it to be sure Jan 10 09:47:33 deva: remember that your volatiles setup needs to match fs-perms Jan 10 09:49:03 I got a nice little email asking me to clean up my requests in bugzilla, though I have already submitted a solution to it. could you help me with what to do next? Jan 10 09:49:06 https://bugzilla.yoctoproject.org/show_bug.cgi?id=10669 Jan 10 09:49:07 Bug 10669: normal, Medium, 2.3 M1, linus.wallgren, IN PROGRESS REVIEW , apt package missing systemd files Jan 10 09:49:12 RP, I'll see if I can make it work as intended. Thanks :-) Jan 10 09:50:02 ecksun: ask rburton if those patches are in mut ;-) Jan 10 09:50:14 okay, I dont know what mut is, but sure :D Jan 10 09:50:42 ecksun: its a testing branch where we queue/test patches before they make master Jan 10 09:51:42 okay, on your phrasing it shounds like there is no visibility into it? Jan 10 09:51:54 also, I just noticed I got an Acked-by from someone at intel, what does that mean? Jan 10 09:54:21 ecksun: its quite visible: http://git.yoctoproject.org/cgit.cgi/poky-contrib/log/?h=ross/mut Jan 10 09:54:37 ecksun: that means that the person reviewed and supported it Jan 10 09:54:44 alright, cool Jan 10 09:55:13 so just doing a git log --author me in that repo should tell me if its in there, right? Jan 10 09:56:54 but my PR was to poky? not poky-contrib Jan 10 09:57:03 ecksun: my grep in the web ui says those patches are not. I was suggesting you as rburton as he might add them if we prompt him Jan 10 09:57:24 alright Jan 10 09:57:29 ecksun: ecksun or he might know why they weren't included Jan 10 09:57:33 alright Jan 10 09:57:41 ecksun: with the holidays we're a bit behind with some patches :( Jan 10 09:57:46 cool, yes, he has been very helpful, I just didnt want to bother him unnescesarily Jan 10 09:58:07 RP, well, most of my patches and issues was submitted in october/november :P Jan 10 09:58:31 ecksun: I can only apologise then, we do our best :/ Jan 10 09:58:47 thanks, but no problem :) Jan 10 09:59:00 what I would have wished for is some more information about what the status of things are Jan 10 09:59:11 rburton, what is the status of https://bugzilla.yoctoproject.org/show_bug.cgi?id=10669 ? Jan 10 09:59:12 Bug 10669: normal, Medium, 2.3 M1, linus.wallgren, IN PROGRESS REVIEW , apt package missing systemd files Jan 10 09:59:29 ecksun: we do have work in progress on patchwork improvements to help with this Jan 10 09:59:42 alright, thats good :) Jan 10 10:00:02 if you are interested in feedback, another thing that took me a while to figure out was how to contribute, I still don't think I have gotten it right :D Jan 10 10:00:39 ecksun: the page on the website and README in the repo didn't make it clear? or you didn't find those? Jan 10 10:01:00 or something else? :) Jan 10 10:01:13 well, not clear enough anyway (since I don't know the proper procedure) Jan 10 10:02:13 I will try to find them again to give you an answer to that Jan 10 10:02:23 The rough procedure is to submit patches and then if they don't move, remind us as we do miss things :( Jan 10 10:02:39 patchwork should make it easier for everyone to find the status of patches in the future but we're not there yet Jan 10 10:02:51 yes, thats what I have been doing, but I have mostly been able to figure it out from all the helpful people here Jan 10 10:03:16 * RP notes the room for improvement Jan 10 10:03:59 in fact, Im trying to find some sort of contributers guide atm, but can't easily find it Jan 10 10:04:07 not on the homepage anyway Jan 10 10:04:44 ecksun: We are in the process of redoing the website. I think my personal view on the current one was "burn to the ground" :/ Jan 10 10:04:50 :D Jan 10 10:04:54 I can understand that Jan 10 10:05:08 found http://www.yoctoproject.org/docs/2.2/mega-manual/mega-manual.html#how-to-submit-a-change in the manual! :) Jan 10 10:06:23 ecksun: That is one of the things I was thinking of but reading it, it doesn't really answer some of your questions does it? :/ Jan 10 10:06:51 well, not yet anyway Jan 10 10:07:23 dunno if thats easy to fix, but just looking at the wall of text here: http://www.openembedded.org/wiki/Commit_Patch_Message_Guidelines makes me a bit hesitant to do a change Jan 10 10:07:51 in fact I noticed that a friend of mine failed with the commit message format on his first patch just recently Jan 10 10:07:54 ecksun: the manual is easy to change, if you open a bug entry with suggested changes we have a tech writer who works wonders on the manual Jan 10 10:08:06 yeah, thats good :D Jan 10 10:08:39 I just don't know how to improve things :D (easy to critizise, hard to improve) Jan 10 10:09:29 ecksun: I actually submitted a whole load of manual changes on Friday for some issues similar but different to this Jan 10 10:09:35 okay Jan 10 10:09:47 ecksun: trouble is we do this day in day out and we're too close to it to see where the problems at Jan 10 10:09:48 are Jan 10 10:09:54 what do you think I should do? open an issue saying its difficult to find things? Jan 10 10:10:00 yes, thats why Im trying to give you my problems :D Jan 10 10:10:08 but its rather vague also Jan 10 10:10:19 ecksun: yes, an open bug would be a start but we do need to be a little more specific Jan 10 10:10:35 yeah Jan 10 10:11:18 how about an issue about finding it difficult to find out the status of my patch? Jan 10 10:12:33 ecksun: Open a bug and I'll try and give some feedback about what we need to document Jan 10 10:16:06 okay Jan 10 10:16:39 ecksun: I think I'm going to propose changes to that manual section Jan 10 10:17:02 alright Jan 10 10:17:12 now I just need to figure out how to create a bug Jan 10 10:18:00 if you find it useful, I can always give feedback on things Jan 10 10:18:44 here is another thing, and I remember having some difficulties with this before as well, I dont know what product to file a bug under Jan 10 10:19:00 https://bugzilla.yoctoproject.org/enter_bug.cgi?classification=Documentation Jan 10 10:25:12 I created this: https://bugzilla.yoctoproject.org/show_bug.cgi?id=10905 its very brief, do you have any input? Jan 10 10:25:13 Bug 10905: enhancement, Undecided, ---, srifenbark, NEW , Difficult to find status of patch Jan 10 10:31:27 * rburton is back Jan 10 10:32:53 rburton, RP: I remember vaguely that there is a way to set a variable in a way that the value still gets used despite other modifications - is that correct? How? Jan 10 10:33:12 yeah there's an override you can use Jan 10 10:33:18 _forcevariable or something? Jan 10 10:33:26 should have called it _youareevil Jan 10 10:33:47 ecksun: I've added something to that bug Jan 10 10:34:20 rburton: indeed, _forcevariable it is. Jan 10 10:34:27 pohly: that is the one Jan 10 10:34:59 I can't wait to need _reallyforcevariable and _reallyreallyforcevariable :) Jan 10 10:35:13 a higher priority layer means it applies earlier? Jan 10 10:35:27 RP cool, I got an email with a wall of text :D will check it out in a bit Jan 10 10:35:33 aV_V: I can never remember... Jan 10 10:36:12 ecksun: I've tried to break it down into sections with the key pieces first, then details. Whether I got the key pieces right or not, less sure Jan 10 10:42:05 ecksun: so those apt fixes got incorporated into a 30-long series involving apt that broke something badly, and i think i was waiting for a revision. Jan 10 10:43:14 okay Jan 10 10:43:38 what does that mean for me? should I do some changes? Jan 10 10:48:54 just digging trhough the patches now Jan 10 10:49:22 I have a couple of other ones as well that I dont know the status of, but I don't dare ask too much :P Jan 10 10:49:29 go for it, please Jan 10 10:49:46 patchworks links are best, but subject lines work Jan 10 10:50:29 ah theres you patch Jan 10 10:50:44 sorry, wrong thread - yours is standalone and i missed the v2 Jan 10 10:50:51 merged ;) Jan 10 10:50:56 (to my staging branch) Jan 10 10:53:31 In order to boot the "ovmf-shell-image" with runqemu, I would have to rename the image recipe to "ovmf-image-shell", because runqemu only treats parameters which contain "-image-" as rootfs. Jan 10 10:54:47 IMHO that can and should be extended to accept also -image at the end of the parameter. I think that "ovmf-shell-image" sounds more natural. Jan 10 10:54:50 yes Jan 10 10:55:34 I'm appending my own defconfig but it's not the applied :( Jan 10 10:55:41 Okay, will include that in my UEFI patch series. Jan 10 11:01:53 http://pastebin.com/mTQGvjcP Jan 10 11:02:14 the vendor's layer priority is 4, mine is 3 Jan 10 11:02:33 why is not working as I want to? Jan 10 11:03:18 While testing on Ubuntu 14.04.5, I found that devshell or menuconfig is actually resulting like this : http://imgur.com/a/JDWt8 Jan 10 11:04:16 rburton, cool :D Jan 10 11:04:26 Just wondering, am I the only one who is having this issue? Jan 10 11:04:36 rburton, here is another one: http://lists.openembedded.org/pipermail/openembedded-core/2016-October/128106.html Jan 10 11:05:02 and here another: http://lists.openembedded.org/pipermail/openembedded-core/2016-November/128746.html Jan 10 11:10:47 Is there a reason why meta-intel doesn't write .qemuboot.conf files for MACHINE = intel-corei7-64 and friends? I understand that the goal is to run on real hardware, but I'd assume that the same images can also be made to run on a suitably configured qemu, which would be useful for testing. Jan 10 11:10:53 hi! can someone help me plz with my bb? http://pastebin.com/qnW70u4C it fails cause the git repo isnt checkouted into the {WORKDIR}/git and cant find the COPYING file (LIC_FILES_CHSUM points to in invalid file) Jan 10 11:11:14 ecksun: the reload one was merged november 23rd Jan 10 11:11:49 ecksun: so was the package_manager deb args one Jan 10 11:12:42 (my recommended method for tracking stuff submitted upstream is a feature branch per series based on oe-core master, then you can rebase periodically to see what has been merged, what hasn't, and what won't apply any more) Jan 10 11:15:22 on the kernel recipe there is a COMPATIBLE_MACHINE variable, could be that? Jan 10 11:17:26 rburton, merged as in its in master? Jan 10 11:17:58 if so, thats awesome :) (how am I supposed to know that? :D) Jan 10 11:20:55 either by noticing that its in master (ie you rebase your submitted branch), or tracking the status in https://patchwork.openembedded.org/patch/133316/ (which is actually working for that patch) Jan 10 11:21:15 ideally when we have patchworks working well enough it can send a "your patch was merged" mail for us Jan 10 11:21:40 ah, so the answer is patchworks :D Jan 10 11:21:50 oh, so that is patchwork Jan 10 11:21:54 yes, apart from when patchwork get confused Jan 10 11:21:59 :D Jan 10 11:22:14 another solution would be that RP runs a script that mails everyone when he pushes to the git server Jan 10 11:23:16 well, a manual reply would be nice also, but it sounds like your workflow doesn't really allow for that easily Jan 10 11:24:30 anyway, thanks for both your help :) Jan 10 11:29:23 ecksun: manual mails are not going to happen as that would be writing about 100 a week Jan 10 11:30:54 ecksun: I once did do that. I was quickly going to go insane Jan 10 11:31:08 well, more insane. I think its too late for that :/ Jan 10 11:31:26 RP: surely there exists a script to mail patch authors in pre-push or whatever the hook is called Jan 10 11:31:41 greg kh certainly has something similar for the kernel Jan 10 11:32:13 rburton: I meant replying individually on the list to the patch series Jan 10 11:32:24 rburton: I'm sure patchwork could do it and yes, I'm sure there are others Jan 10 11:32:48 ah, https://github.com/git-multimail/git-multimail Jan 10 11:33:10 also shipped as part of git, so that's pretty official Jan 10 11:34:06 i'd be happy to see mails directly sent to the author on push to master Jan 10 11:34:34 in fact we have a commits list already don't we? can that not just cc the author? Jan 10 11:35:17 rburton: just go make it happen ;-) Jan 10 12:13:39 rburton, tbh that doesnt sound too bad, but I understand that :) Jan 10 12:21:58 maxin1: so did sqlite upstream revert the change we found caused a regression, or fix it? Jan 10 12:23:22 rburton: as per the communication with the maintainer, fix is already present in 3.16.0. Jan 10 12:24:18 cool Jan 10 12:24:39 we best run that through the performance cluster to verify Jan 10 12:26:05 sujith_h: no, you're not the only one :) https://bugzilla.yoctoproject.org/show_bug.cgi?id=10312 Jan 10 12:26:06 Bug 10312: normal, Medium, 2.3, randy.macleod, IN PROGRESS REVIEW , Failed to execute child process "oe-gnome-terminal-phonehome" (No such file or directory) Jan 10 12:31:32 rburton: that will be better. Jan 10 12:42:42 ed2: have you generated a debugfs with wic enabled? Jan 10 12:42:53 ed2: (it breaks here) Jan 10 12:43:07 | DEBUG: Executing shell function do_image_debugfs_wic Jan 10 12:43:07 | Error: exec_cmd: cp /data/poky-master/tmp-glibc/work/intel_corei7_64-poky-linux/core-image-minimal/1.0-r0/core-image-minimal-1.0/hddimg/EFI/BOOT/* /data/poky-master/tmp-glibc/work/intel_corei7_64-poky-linux/core-image-minimal/1.0-r0/deploy-core-image-minimal-image-complete/core-image-minimal-intel-corei7-64-20170110123043-dbg/build/hdd/boot/EFI/BOOT returned '1' instead of 0 Jan 10 12:43:12 because hddimg/ is empty Jan 10 12:47:35 rburton: heh, that's why I want CI for wic codebase :( Jan 10 12:48:20 rburton: no idea why enabling debugfs makes hddimg empty. Jan 10 12:49:25 i'm guessing bitbake is wiping out the old directories Jan 10 12:49:39 rburton: thanks for pointing out to this. I'll have to take care of it anyway: https://bugzilla.yoctoproject.org/show_bug.cgi?id=10835 Jan 10 12:49:40 Bug 10835: normal, Medium, 2.3 M3, eduard.bartosh, ACCEPTED , WIC should not rely on hddimg creation for finding all needed artifacts Jan 10 12:49:49 cool Jan 10 12:52:24 I like this yocti bot :) What else it can do? Jan 10 12:52:54 ~yocti slap ed2 Jan 10 12:53:00 hmmm Jan 10 12:53:56 yocti, praise rburton Jan 10 12:53:57 Crofton|work: Error: There are no praises in my database for #yocto. Jan 10 12:54:18 yocti, seen ed2 Jan 10 12:54:18 Crofton|work: ed2 was last seen in #yocto 1 minute and 53 seconds ago: I like this yocti bot :) What else it can do? Jan 10 12:54:39 ed2, /msg yocti help Jan 10 12:56:08 Crofton: it didn't help much: (help [] []) -- This command gives a useful description of what does. is only necessary if the command is in more than one plugin. Jan 10 12:56:51 Admin, Bugzilla, Channel, ChannelLogger, Config, Karma, Misc, Owner, Praise, Seen, Services, and User Jan 10 12:56:52 Jan 10 12:56:57 is what I must have got earlier Jan 10 12:57:05 Crofton: do you happen to know where can I see its sources? Jan 10 12:57:14 halstead, ^^^ Jan 10 12:57:27 yocti, config Jan 10 12:57:27 Crofton|work: (config []) -- If is given, sets the value of to . Otherwise, returns the current value of . You may omit the leading "supybot." in the name if you so choose. Jan 10 12:57:36 ah supybot :) Jan 10 12:57:49 that should give you a start looking for source Jan 10 12:58:07 Crofton|work: thank you! Jan 10 12:59:07 clearly we need to be able to control builds via an irc bot Jan 10 13:00:41 Getting info about commits from mailman urls would be good too. Jan 10 13:02:48 good point Jan 10 13:03:04 pretty sure the bot can handle pluging Jan 10 13:03:09 plugins Jan 10 13:17:32 can u tell me the best way to include the headers of db-dev (package) from db(recipe) into my build? db-dev is installed, but i missing the headers Jan 10 13:19:18 tb_, I wonder if the recipe has a problem? Jan 10 13:20:35 db-dev is certainly where the headers are Jan 10 13:22:41 i added db to CORE_IMAGE_EXTRA_INSTALL and db + libdb-dev to IMAGE_INSTALL, but cant find the headers? :/ Jan 10 13:23:49 also tried with only db-dev instead of libdb-dev too Jan 10 13:24:38 tbh what you actually want to do is add dev-pkgs to IMAGE_FEATURES - then you get all headers that you'll need Jan 10 13:26:20 Error: Transaction check error: Jan 10 13:26:20 package base-files-3.0.14-r89.qemux86 is intended for a different architecture Jan 10 13:26:33 would it be super-bad to disable architecture check in rpm4 altogether? Jan 10 13:27:26 the remaining (fingers crossed) failures in making an image with dnf are this, and the need to 'virtualize' post-in rpm scriptslets Jan 10 13:58:03 are there other ways to append a custom defconfig aside of adding it to the FILEPATH? Jan 10 14:46:50 hi there Jan 10 14:47:05 is possible reference a patch inside a tar.gz in SRC_URI Jan 10 14:47:06 ? Jan 10 14:49:33 istarilucky: i don't think so Jan 10 14:50:24 neither me, but i need to come and ask :) Jan 10 14:50:37 thanks, alimon Jan 10 14:58:50 does Alejandro del Castillo hang in here? Jan 10 15:02:57 rburton: I can poke him if you want Jan 10 15:03:51 aratiu: if he's around that would be great, no rush if its early Jan 10 15:06:02 he's not in yet Jan 10 15:06:10 I'll tell him to come here Jan 10 16:20:42 Anyone have a link to a custom kernel recipe that I can use as a template? Jan 10 16:28:20 hey everyone Jan 10 16:28:35 how do I set a static IP for my image? Jan 10 16:36:04 bernarrrrrrrrrrr, you could install a .network file for systemd Jan 10 16:36:42 Cubi_: how do I add that to my recipe? Jan 10 16:37:48 hello Jan 10 16:38:11 I have a weird problem Jan 10 16:38:53 I have two disks in my yocto build computer. One is a motherboard mounted ssd and the other is a mechanical drive. Jan 10 16:39:49 i installed the mechanical drive which is large for doing nightly builds and archiving yocto builds. ie. disk images, sdk's etc. Jan 10 16:40:04 I never have a problem with the ssd builds Jan 10 16:40:30 when I installed the mechnanical drive it was used and had about 8K hours of usage on it Jan 10 16:41:07 periodically i get coredumps while doing the build on the mechanical drive Jan 10 16:41:41 its always in the same programs but its variable wether all errors will occur Jan 10 16:42:21 the programs are bison-native conftest, m4-native conftest and likewise for the target versions. Jan 10 16:42:30 you meant the ssd is msata or m.2 nvme? Jan 10 16:42:55 might be useful to share the stack and anything in dmesg Jan 10 16:42:57 & is the mechanical drive just regular sata? Jan 10 16:43:05 i believe so Jan 10 16:43:22 so trying to debug it, I played with the fstab settings Jan 10 16:43:25 usually, dodgy ram is to blame for random build failures Jan 10 16:44:16 i ran the smart tools and the drive worked ok for the non destructive long tests, there were a few errors in one of the critical tests, 3 Jan 10 16:44:50 so I asked for a new drive, perhaps the mechanical disk was on the edge of failure Jan 10 16:45:13 i got a new similar mechanical disk, mounted both and did simulatanous builds Jan 10 16:45:57 the disk which had the orig errors had the same problem. coredumps. the second mechnaical brand new disk did not Jan 10 16:47:10 dmesg will probably tell you more, but its possible that operations were failing as the disk was failing, so the binary aborts Jan 10 16:47:17 I also changed the sata connector I used to connect the disks to the motherboard, a colleague noted that the motherboard manual said if you are using the sata express ports will be unavailable when using the M.2 sata interface module Jan 10 16:47:58 so with one drive working and one drive failing, booted to the seagate usb smart tests tool Jan 10 16:48:19 i ran the destructive test on the mechanical disk which has been failing Jan 10 16:48:35 sure enough on the erase test it fails. Jan 10 16:49:00 I'm like well that is the problem, its that the disk with 8K hours of usage its gone bad. Jan 10 16:49:43 i remove the disk, and now only have the ssd m2 disk for my / and the mechanical sata for my /opt Jan 10 16:50:26 boot up and start my build on /opt and what worked fine the day before fails Jan 10 16:53:18 dmesg does not show any thing which looks like an error Jan 10 17:00:48 how do I add connman to my image? Jan 10 17:01:18 the same way you'd add any other package Jan 10 17:05:13 bernarrrrrrrrrrr: look for your images recipe and add it there. Jan 10 17:06:40 or image install append in local.conf Jan 10 17:07:13 davis: I'm trying to edit my conf/bblayers.conf Jan 10 17:07:42 bblayers is not where you add packages Jan 10 17:07:56 how should I do it? Jan 10 17:08:37 I've already tried adding IMAGE_INSTALL_append += "connman" to my conf/local.conf Jan 10 17:08:41 with no success Jan 10 17:09:39 _append doesn't add whitespace, so you're looking for IMAGE_INSTALL_append = " connman" Jan 10 17:09:48 this is in the quick start for what its worth Jan 10 17:09:58 i dont know if hob works anymore, but perhaps you can try to use that tool to get you started. Jan 10 17:10:23 hob is no longer in master. all scrubbed away. Jan 10 17:10:29 bummer Jan 10 17:10:39 what is the name of the new tool which runs in a webpage? Jan 10 17:10:42 Toaster Jan 10 17:11:05 that's it. perhaps bernarrrrrrrrrrr could use that to a feel for how to work with yocto Jan 10 17:11:08 toaster can make custom images and let you add packages by clicking Jan 10 17:11:58 im using this for my fstab entry for /opt on ubuntu Jan 10 17:12:02 UUID=e94d9289-5049-4035-a4c1-48085f5c44cc /opt ext4 defaults 0 2 Jan 10 17:12:34 yeah ive read about toaster.. i might give it a try later.. sounds nice Jan 10 17:13:14 ive tried to play with the defaults portion but I erased my comments. When I thought the disk was working Jan 10 17:14:08 i also asked the guy who built this machine for me, if he played with the overclock settings in the bios and he said no Jan 10 17:14:58 it doesn't make sense that the builds complete on the ssd but coredump on a brand new disk. the used one did make sense. Jan 10 17:16:53 davis: try a memory test Jan 10 17:17:53 davis: have you checked the disk on another machine? best to rule out physical disk failure too Jan 10 17:18:20 rburton: what is a good memory test to try? Jan 10 17:18:27 davis: memtest Jan 10 17:19:21 nrossi: so the orig disk I had, actually came from a different machine which had problems. When it was given to me and I asked about the history of the disk, I was like what? Jan 10 17:20:13 but the new disk I opened from shrink wrap and smart reported 3 hours of usage which is spot on since it took me that long to rsync the failing disk contents to it. Jan 10 17:20:34 i rsync'd over everything except the lostandfound dir Jan 10 17:20:43 rburton: many thanks i'll try that. Jan 10 17:21:54 rburton: on ubuntu there is memtest86+ and memtester, i assume you mean memtest86+ Jan 10 17:21:59 yeah thats it Jan 10 17:22:05 leave it running overnight to rule out ram Jan 10 17:22:22 ok many thanks Jan 10 17:24:13 davis: just note that it is possible for a brand new disk to be faulty. If you don't find any issues with the memory, try the new disk in another machine Jan 10 17:25:03 what's odd is that both mechanical disks fail in the exact four yocto steps Jan 10 17:25:10 never on the ssd builds though Jan 10 17:26:30 actually looking at dmesg etc to determine *why* they're failing would be a good start Jan 10 17:29:45 I've built a dhcpd recipe with a configuration for my desired target, but despite dhcpd.service being in /usr/lib/systemd/network, the DHCP server service doesn't start on start-up. Any ideas what I'm doing wrong? Jan 10 17:32:28 rburton: i did as you suggested. i did not see any errors in dmesg Jan 10 17:32:58 i only get this popup from ubuntu saying a failure occured and that the offending binaryt has been removed from the system. Jan 10 17:33:07 how useful Jan 10 17:33:16 an actual stack trace would be helpful Jan 10 17:35:20 there was a stack trace in the details pane of the popup Jan 10 17:35:29 i'll save it off next time. Jan 10 18:00:50 how do I add ssh-server to core-image-minimal? Jan 10 18:01:10 I added EXTRA_IMAGE_FEATURES ?= "ssh-server-openssh" to my conf/local.conf Jan 10 18:01:41 when I ping my beaglebone it seems to be up Jan 10 18:02:09 but when I try to ssh into it from my host I get a connection refused Jan 10 18:04:56 use bitbake -e and bitbake -e core-image-minimal to make sure EXTRA_IMAGE_FEATURES and IMAGE_FEATURES are what you expect them to be Jan 10 18:41:43 thanks a lot guys! Jan 10 19:41:48 what is the difference of do_configure_prepend_class-target() to do_configure_prepend() in a recipe? When I use one and when I use other? Jan 10 19:43:43 _class-target only applies to target recipes Jan 10 19:43:50 that is, not native, cross, nativesdk, or crosssdk Jan 10 19:43:59 it's only needed if you set BBCLASSEXTEND to create those variants Jan 10 19:44:20 hum Jan 10 19:44:27 thanks Jan 10 20:47:49 so I've got the leasefile for the dhcpd daemon, which indicates my systemd service may actually be running, but may have something wrong with its command-line. Can anyone take a look? Jan 10 20:47:51 http://pastebin.com/wgrQye4H Jan 10 20:55:21 khem: hi - could you resend your patch for morty? https://patchwork.openembedded.org/patch/133002/ Jan 10 21:44:36 zeddii_home: zeddii: have you tried a core-image-sato-sdk build with 4.9? I'm getting: Computing transaction...error: Can't install kernel-devsrc-1.0-r0@intel_corei7_64: no package provides /bin/awk Jan 10 21:45:10 nope. haven’t tried (and hence seen that one) .. clearly a new script that needs awk. Jan 10 21:46:08 * paulg assumes meta-intel was in use given the i7 reference in the pkg name. Jan 10 21:47:38 paulg: indeed, working on adding 4.9 to meta-intel Jan 10 21:48:39 I want to put a Warning or Info message in that pops up when someone uses the container image type but doesn't set the kernel provider to linux-dummy. If I do the check in the bbclass, it shows up multiple times, essentially for all things that inherit image.bbclass. Is there a preferred way to get that type of warning to only show up once? Jan 10 21:48:41 might be worth a test with just the basic layers ; I know I've inadvertently laid traps for myself with forgotten bbappends in my/our layers that caused issues nobody else was seeing. Jan 10 21:49:42 * paulg feeds core-image-sato-sdk to a builder to see what happens Jan 10 21:53:09 fetching new gcc ; results might be avail in an hour or two.... :-/ Jan 10 21:53:40 :) Jan 10 22:50:50 clsulliv, I'm up to ----- 0: kernel-devsrc-1.0-r0 do_package - 664s (pid 15913) Jan 10 22:51:04 what build step did you get the awk issue at? Jan 10 22:53:40 I'm using linux-yocto-dev (so 4.9) but not using meta-intel Jan 10 22:53:55 using the core2 default x86-64 tuning. Jan 10 22:55:50 paulg: didn't hit it until do_rootfs unfortunately Jan 10 22:57:12 ok, will let it run to the end.... Jan 10 22:58:55 may have to resume this discussion tomorrow. :) Jan 10 23:00:19 up to ---- 0: kernel-devsrc-1.0-r0 do_package_write_rpm - 265s (pid 11413) Jan 10 23:01:01 paulg: appreciate you trying to reproduce this! Jan 10 23:02:00 at least it isn't an -rt issue I'm looking at for once. :) Jan 10 23:16:22 clsulliv, seems like you can rule out meta-intel.... Jan 10 23:16:30 Computing transaction...error: Can't install kernel-devsrc-1.0-r0@genericx86_64: no package provides /bin/awk Jan 10 23:16:46 look familiar? :) Jan 10 23:19:55 anyway, will look at it more tomorrow. Jan 11 00:37:48 dl9pf: you can reply to thread and CC morty maintainer ( Armin ) with request to backport Jan 11 01:14:59 Is it possible to change the build options for native packages? Jan 11 01:16:00 I want to debug my native packages before deploying to the target in gdb but they seem to be stripped. Jan 11 01:17:34 catch22: https://www.yoctoproject.org/docs/current/ref-manual/ref-manual.html#var-BUILD_CFLAGS Jan 11 01:18:43 rewitt: excellent, thanks Jan 11 01:39:32 rewitt: static linking should be no problem with musl and its well supported, add -static to CFLAGS and LDFLAGS globally Jan 11 01:39:53 rewitt: although if you have many apps then it might not give you best results Jan 11 01:40:01 in terms for size opts Jan 11 01:40:19 but say if you just want busybox + your app Jan 11 01:40:42 which is a typical embedded scenario in some cases, this might be helpful Jan 11 01:41:04 musl based poky-tiny images are around 600K anyway Jan 11 01:41:41 full musl size is approx 500K that includes all libraries e.g. libm/librt included Jan 11 01:52:47 rewitt : adding BUILD_CPPFLAGS caues a rebuild of everything, I've found adding OECMAKE_C_FLAGS_append_class-native = " -O0 -ggdb " to the specific recipe appears to gives me what I want just for those recipes. Do you know if you can adjust build_cflags for specific packages from local.conf? **** ENDING LOGGING AT Wed Jan 11 03:00:01 2017