**** BEGIN LOGGING AT Tue Mar 15 02:59:58 2016 Mar 15 07:26:51 good morning Mar 15 10:02:40 hi, all! Mar 15 10:03:14 how can I build external kernel module using OE? 'inherit module' doesn't help Mar 15 10:04:26 the module Makefile rule is standard: $(MAKE) -C $(KDIR) M=$(CURDIR) modules Mar 15 10:05:22 the parameters from 'inherit module' are quite strange. Is there any standard way? Mar 15 10:08:44 slapin: what about http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#working-with-out-of-tree-modules Mar 15 10:08:47 ? Mar 15 10:16:13 LetoThe2nd: well, for this to work I need to rewrite the Makefile, which is strange Mar 15 10:16:34 LetoThe2nd: thanks a lot for help! Mar 15 10:17:40 slapin: patching makefiles for use in OE is not that uncommon, unfortunately. good luck! Mar 15 10:19:37 LetoThe2nd: I thought there was some standard solution for standard case, but it seems there is no, I just asked hoping I miss something, thanks for explanation Mar 15 12:17:51 hi, with recent environment (bitbake master, fedora 23), I see bitbake iterating above a huge number of filedescriptors: Mar 15 12:17:54 [pid 15945] close(666502) = -1 EBADF (Bad file descriptor) Mar 15 12:18:13 is this a known issue? Mar 15 12:30:58 seems to by an issue in python2's subprocess.Popen() Mar 15 12:30:59 strace -f -eclose python2 -c 'import subprocess; subprocess.Popen("true", close_fds = True, stdout=subprocess.PIPE, shell=False).communicate()' Mar 15 12:31:46 [pid 16228] close(1048575) = -1 EBADF (Bad file descriptor) Mar 15 12:31:46 [pid 16227] close(4) = 0 Mar 15 12:32:07 python3 seems to be ok Mar 15 12:41:34 ensc|w: but bitbake still isn't compatible with python3, right? Mar 15 12:41:59 no; unfortunately not Mar 15 12:42:30 I will have to call ulimit -n to reduce number of fds Mar 15 12:43:03 docker seems to set them to 1Mi Mar 15 13:40:45 I got warnings and errors from log_check in rootfs.py when i create a rootfs with dbus package. How can i disable this error or how can i fix it? logfile: http://pastebin.com/eU5d3f4u Mar 15 13:42:35 are you doing a build with lots of debugging enabled? Mar 15 13:44:02 Matthias_: looks like you've got something enabled that's spewing all the packagedata to the logs, and it's not in oe-core master Mar 15 13:44:47 (which is triggering the log checker) Mar 15 13:45:05 specifically nothing in oe-core master/jethro/fido produces "Packagedata" in the lgos Mar 15 13:49:12 its a good input i will check this. thanks Mar 15 13:52:10 rburton: good hint. that was the issue. Mar 15 14:27:30 Can a recipe use DEPENDS on a package group? (mine is acting like I haven't specified anything which is why I ask) Mar 15 14:28:25 eengie: it can, but it doesn't make any sense Mar 15 14:28:59 DEPENDS are build time dependencies, packagegroups list bunch of packages as runtime dependencies Mar 15 14:29:11 are you confusing DEPENDS with RDEPENDS? Mar 15 14:29:14 so having build time dependency on packagegroup doesn't bring any dependency Mar 15 14:29:32 Got it... thanks JaMa. Mar 15 14:29:55 (I was trying to avoid listing 5 packages that _should_ get installed together usually by just giving them a big common name) Mar 15 16:30:22 Is there a special way to specify a configuration script with the update-rc.d class that will ensure it gets a "stop" call when the device is shutting down? Mar 15 16:34:43 Never mind... I see the "exiting" message in my shutdown log, so it's definitely getting the kill. Trouble is it seems to be getting killed after the network interfaces on which it depends to announce its departure. Mar 15 16:35:37 use systemd for better control over ordering Mar 15 20:36:08 bluelightning, did you get around to try? Mar 15 20:36:35 fischerm: I did and it installed fine here... I can only assume there's something I'm not replicating from your setup Mar 15 20:36:44 hrm Mar 15 20:40:14 fischerm: so, in the partially installed SDK directory there's a conf/local.conf - does that have a UNINATIVE_CHECKSUM line in it? Mar 15 20:42:26 bluelightning, INHERIT += "uninative" Mar 15 20:42:35 no checksum Mar 15 20:43:54 bluelightning, http://pastebin.com/JHLVZzyS Mar 15 20:45:25 fischerm: which oe-core or poky revision did you build this from? Mar 15 20:46:05 bluelightning, reasonably new master Mar 15 20:46:34 bluelightning, meta: master:380ee36811939d9 Mar 15 20:48:19 fischerm: ah, so you unfortunately went inbetween an update to uninative.bbclass and the fix in the extensible SDK for that update Mar 15 20:49:39 fischerm: update to at least be177739b46d3296c0c0c01310b499ffc6782d43 and that issue should be fixed Mar 15 20:50:52 current master is a RC for 2.1M3 by the way, so shouldn't be that broken :) Mar 15 20:53:00 bluelightning, hah, thx Mar 15 20:55:12 rburton, thx Mar 15 21:18:12 bluelightning: do you know where the logic to make rpm save postinsts into /etc/rpm-postinsts is? trying to find where the logic that handles the error codes from the postinst is Mar 15 21:19:53 rburton: grep for rpm-postinsts in meta/lib/oe/package_manager.py Mar 15 21:21:16 oh the cross-scriptlet wrapper is put around every postinst by rpm? Mar 15 21:21:30 now, ditto for dpkg and opkg :) Mar 15 22:55:05 bluelightning, that worked, with the constraint that the ../../downloads I had set in my local.conf made it into the sdk, don't know if that's on purpose Mar 15 22:55:34 fischerm: it only automatically removes variables whose value starts with / Mar 15 22:56:11 fischerm: if you want to you can append DL_DIR to SDK_LOCAL_CONF_BLACKLIST to ensure it doesn't go through Mar 15 22:56:57 I have to say I wouldn't recommend using a relative path like that in any case Mar 15 22:57:16 k Mar 15 22:57:30 thanks Mar 15 22:58:18 ok, I cried success to early Mar 15 22:58:19 :-/ **** ENDING LOGGING AT Wed Mar 16 02:59:58 2016