**** BEGIN LOGGING AT Tue Jul 23 02:59:57 2019 Jul 23 03:07:20 Evening, still trying to debug that yocto-on-intel-aero-drone problem. Jul 23 05:33:13 New news from stackoverflow: sdk can build my cmake project but recipe in bitbake can't Jul 23 06:21:10 * alessioigor waves all Jul 23 06:35:22 I'm trying to install a custom journald.conf but I face this error`file /etc/systemd/journald.conf conflicts between attempted installs of systemd-1:239` Jul 23 06:35:35 I'm doing something like this: http://git.yoctoproject.org/cgit/cgit.cgi/meta-intel-edison/tree/meta-intel-edison-distro/recipes-core/systemd/systemd_%25.bbappend **** BEGIN LOGGING AT Tue Jul 23 06:46:23 2019 Jul 23 07:03:28 New news from stackoverflow: Got an error "Checking for program pkg-config : not found" during execution "bitbake image_name" Jul 23 08:03:32 If I inherit systemd and set SYSTEMD_SERVICE_${PN}, do I still have to manually copy my .service file into the systemd directory? this feels redundant … Jul 23 08:08:41 jmiehe: yes, because typically the upstream *ships the service file* Jul 23 08:08:58 the class assumes it is installed Jul 23 08:10:42 where does that happen in the "example" connman recipe referenced here? https://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#ref-classes-systemd Jul 23 08:11:07 (= http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-connectivity/connman/connman.inc http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-connectivity/connman/connman_1.35.bb) Jul 23 09:00:47 how can I set the hostname in my image recipe? Jul 23 09:10:36 eg hostname_pn-base-files = "bob" in local.conf Jul 23 09:10:43 in an image recipe, you can't Jul 23 09:11:04 if you *need* to do it in an image recipe then write a rootfs postprocess hook to write a new hostname file Jul 23 09:12:32 rburton: ok, I tried the hostname_pn-base-files in the image recipe and it didn't work, thats's Jul 23 09:12:38 why I asked* Jul 23 09:13:18 the default hostname comes from the recipe called base-files, so you can't override that from another recipe Jul 23 09:13:40 jmiehe: things in one recipe cannot affect things in another recipe. thats why it didn't work in your case. Jul 23 09:13:55 jmiehe: recipe data is local, conf data is global. Jul 23 09:14:11 (i actually have any OE/YOP class that i teach say that out loud) Jul 23 09:14:23 important to understand Jul 23 09:14:26 but I can .bbappend to base-files, which is enough in this use case :) Jul 23 09:14:31 right Jul 23 09:14:47 rburton: is it. Jul 23 09:14:50 i just have that line in my local.conf so any machine for any arch i boot up is bob.local Jul 23 09:15:06 no alice?!? Jul 23 09:15:14 lol Jul 23 09:15:15 no Jul 23 09:15:21 there's no alice in my naming scheme Jul 23 09:15:55 scheme? so next ones are boob, booob, boooob …? Jul 23 09:16:19 melchett, flashheart, baldrick Jul 23 09:18:51 rburton: so your bob is kind of an alice, too. weird flex but ok. Jul 23 09:21:48 ix, caladan, junction, giediprime, arrakis :) Jul 23 09:34:03 I have a really simple recipe erroring out my build. Maybe I'm just dumb? :P https://pastebin.com/12zTT36p Jul 23 09:34:46 jmiehe: and how does it "error out"? Jul 23 09:36:02 with a 1581 line error log containing "Unable to install packages" and "Function failed: do_rootfs" Jul 23 09:36:52 so this is a pacakge of its own? Jul 23 09:38:14 yes, the idea was to quickly pull in some presets Jul 23 09:39:26 (and be able to switch them out as quickly) Jul 23 09:39:45 i can't put my finger on it but something is werid/wrong Jul 23 09:40:21 how is the recipe file called? and can you scroll up in the failing log, i kinda suspect some more information to be hidden in there. Jul 23 09:40:39 plus, what happens if you bitbake that package only? Jul 23 09:44:02 recipe is in mixin-static-ip_1.0.bb, with the files from SRC_URI in subdir "conf" Jul 23 09:44:38 bitbake mixin-static-ip ~ Tasks Summary: Attempted 454 tasks of which 440 didn't need to be rerun and all succeeded Jul 23 09:44:46 hm ok Jul 23 09:45:09 and inspecting the resulting package seems to be fine too? Jul 23 09:46:55 the work/$arch$/$pkg$/1.0-r0/image looks fine at least Jul 23 09:47:45 maybe it's because i'm trying to install files that are already present? Jul 23 09:47:45 no, i really mean the resulting ipk or whatever package manager you use Jul 23 09:49:03 actually i have no idea how to inspect those Jul 23 09:50:53 somewhere in tmp/deploy/$MACHINE/$PACKAGEMANAGER/... Jul 23 09:51:14 so if its ipk, there should be something like mixin-static-ip_1.0.ipk Jul 23 09:51:18 yes Jul 23 09:51:24 and you can inspect that with dpkg :) Jul 23 09:54:17 yes looks good Jul 23 09:55:48 then the next thing is probably really a deeper inspection of the error log. Jul 23 09:56:45 During installation, ${systemd_unitdir}/network/10-eth0.network is already present. Should that be the problem? Jul 23 09:56:51 bingo. Jul 23 10:00:22 if two packages provide the same file, then creating the rootfs fails. how should bitbake know which one you want? Jul 23 10:28:26 understood. Jul 23 10:28:57 Is systemd_unitdir provided by those python functions in systemd.bbclass? http://cgit.openembedded.org/openembedded-core/tree/meta/classes/systemd.bbclass?h=sumo Jul 23 10:55:08 Hello. Jul 23 10:56:04 Is it possible to run commands that require root rights (e.g. kpartx) in a class? If I run the command straight away in the class, I get a permission denied error. Jul 23 11:00:34 evaluate: well this would totally sabotage being able to build said thing as a user Jul 23 11:01:15 evaluate: so just guessing from the name, it rather sounds like something that should be done in wic or such Jul 23 11:02:16 LetoThe2nd, what I'm actually trying to do is mount the image to run some commands in it. Not sure if that can be done in the wic... Jul 23 11:02:38 evaluate: ROOTFS_POSTPROCESS_COMMAND Jul 23 11:02:52 or ROOTFS_POSTPROCESS_CMD, something along that Jul 23 11:03:10 LetoThe2nd, Ok, let me look that up. Jul 23 11:19:10 LetoThe2nd, I see that the ROOTFS_POSTPROCESS_COMMAND also takes a function, and I don't see a mention of any special rights being present there. If I run e.g. kpartx in that function, will it run elevated? Jul 23 11:20:55 no Jul 23 11:21:13 it will have root rights for the content of that image, but not for the machine. Jul 23 11:21:28 my advice is: try to describe what you are *ACTUALLY* trying to do Jul 23 11:21:51 kpartx sounds totally like the Y-part of a XY-question (feel free to look that up) Jul 23 11:22:56 LetoThe2nd, indeed. What I am actually trying to do is install grub-pc as the bootloader. Jul 23 11:23:37 I tried replicating what grub-install does, to write the data directly to the disk image, but was unsuccessful, so now I am trying to mount the disk and run grub-install in it. Jul 23 11:28:42 sounds like something wic should do. Jul 23 11:31:11 here's the grub-efi example: https://git.yoctoproject.org/cgit.cgi/poky/tree/scripts/lib/wic/plugins/source/bootimg-efi.py Jul 23 11:32:12 and the corresponding wks file https://git.yoctoproject.org/cgit.cgi/poky/tree/scripts/lib/wic/canned-wks/mkefidisk.wks Jul 23 11:34:19 New news from stackoverflow: How to change the gcc version in Yocto Sumo? Jul 23 11:36:04 LetoThe2nd, yeah, I already saw the efi one, but the thing is that grub-pc requires some weird stuff (an extra partition and also it needs to write into the empty sector in the MBR and) so it's completely different from what the efi script does. Jul 23 11:36:59 It's more similar to the syslinux one (which writes the data using dd), but unfortunately there's not a lot of documentation on how to write grub to the disk using dd, since there is the grub-install utility which does everything for yo. Jul 23 11:37:03 you* Jul 23 11:37:53 evaluate: then you'll have to dig through the syslinux and efi examples and distill out your needed bits and pieces. going the kpartx route is definitely wrong. Jul 23 11:39:04 maybe then grub-install utility can by itself do its magic on some binary instead of raw blocks, i don't know. Jul 23 11:39:23 but anything that requires root on the build host is a total red flag. Jul 23 11:40:45 LetoThe2nd, the information gap for this is not in yocto, it's in grub. Like I said, I theoretically know what needs to be done (write the stage1 into MBR and stage2 into a separate BIOS boot partition) but there is some more stuff that's needed (e.g. stage1 needs to contain the pointer to stage2). Jul 23 11:42:04 evaluate: i see what you mean, really. but it doesn't change the context and restrictions of using root. Jul 23 11:43:49 If I want to .bbappend say nginx, does it make any difference if i do that in recipes-httpd/nginx/nginx.bbappend or in recipes-foo/bar/nginx.bbappend? also, what's the difference between nginx.bbappend and nginx_%.bbappend? Jul 23 11:52:32 [btw] I fixed my double file installation by installing into the systemd override directory at /etc/systemd Jul 23 12:21:38 If is need a custom wks, am I obliged to create a `machine` conf or there is a more simple way? Jul 23 12:21:47 Like setting a variable? Jul 23 12:37:41 Is there any way to inherit machine from other layers changing a couple of vars? Jul 23 13:18:27 a machine config can happily include another machine conf if it just wants to modify it slightly Jul 23 13:21:44 hi Jul 23 13:23:37 rburton: I've managed to simple set the wks file for wic img, and it works (since machine conf uses ?=), though I can't figure it out how to set partitioning for hddimg as well. Jul 23 13:24:08 Should I set wks for hddimg as well somehow? Jul 23 13:30:37 Ah, I see, it's in classes/image-live Jul 23 14:07:49 I know it is possible to apply a custom patch in a BitBake recipe,,, now can I simply revert a commit instead of making patches ? Jul 23 14:18:26 RP: Posted a patch with some SQL optimizations for the hash server... I don't think multithreading the server should be necessary. Jul 23 14:19:34 JPEW: ok, cool. Will have to take a look Jul 23 14:20:54 JPEW: patch looks good, thanks Jul 23 14:21:30 Hopefully that will eliminate the linear search and drastically cut down the millions of execute ops :) Jul 23 14:23:42 JPEW: I've a few other things I need to sort then I'll get back to figuring out how it managed that crazy number Jul 23 14:24:27 (the ops seem to be in cooker rather than the hashserv but fixing that will help the connection issues) Jul 23 14:25:52 Hmm, ya. That could be the lookup in the cached persistDB then... but that would be strange because it also has an index on the key. Jul 23 14:26:40 JPEW: its persistDB. The best thing I can do is remove it Jul 23 14:27:06 True Jul 23 14:59:39 my google calendar only has invalid meeting IDs :-( Jul 23 14:59:44 (zoom meetings) Jul 23 15:00:44 yes, what it the meeting number? Jul 23 15:01:15 990 892 712 Jul 23 15:02:09 https://zoom.us/j/990892712 Jul 23 15:05:28 can anyone else hear RP? Jul 23 15:17:09 tlwoerner: people seem to be able to Jul 23 15:33:07 How to set the partition table for the default live installer? Jul 23 15:57:52 Ah, it's hardcoded in install script Jul 23 17:39:17 I'm looking at switching our internal bsp to use yocto style kernel meta data. However, I'd like to use an in-tree defconfig as the base. Setting KBUILD_DEFCONFIG appends the defconfig to the end of the merge list instead of the beginning, any suggestions? Jul 23 17:55:22 zeddii: ^ Jul 23 17:58:34 More generally, is there a way to reference in-tree defconfig/fragments from an .scc file. KBUILD_DEFCONFIG was what I found to try based on the mega manual. Jul 23 18:05:21 New news from stackoverflow: Overwriting Yocto Classes through meta-layer Jul 23 18:17:43 Good afternoon. Anyone familiar with the innards of iwlwifi? I'm trying to debug an error with an intel aero drone. iwlwifi keeps crashing. I'm not looking for a fix so much as a better idea of what is going wrong, for the inevitable arguments with intel customersupport. Jul 23 19:52:51 what is the analogue of cgroupfs-mount in poky? Jul 23 20:19:04 JPEW: found the problem, a stupid/lazy bit of code. Brute forcing the taskhash updates was a bad idea, we need to do it carefully Jul 23 20:19:21 RP: In the runqueue? Jul 23 20:19:43 JPEW: yes. 41k calls to get_unihash rather than 1.7 million now :) Jul 23 20:19:56 Ah, ya. That will do it :) Jul 23 20:20:46 JPEW: fast enough I can't time it now :) Jul 23 20:21:33 profile says a few seconds Jul 23 20:35:42 New news from stackoverflow: Find Codeaurora oe installation manifest Jul 23 21:03:40 JPEW: http://git.yoctoproject.org/cgit.cgi/poky-contrib/log/?h=rpurdie/t222 top four patches are the work in progress which now works FWIW **** ENDING LOGGING AT Wed Jul 24 02:59:57 2019