**** BEGIN LOGGING AT Sun Mar 03 02:59:58 2013 **** BEGIN LOGGING AT Sun Mar 03 05:37:01 2013 Mar 03 16:28:18 does yocto support musl libc ? ( http://www.musl-libc.org ) Mar 03 16:53:07 do images with oe-core's systemd boot? I've built new one for qemux86-64 and it fails to mount couple of dirs and then cycles trying emergency boot Mar 03 16:53:40 I was expecting is because of broken upgrade paths but after reflashing new image does not boot too Mar 03 16:54:41 JaMa: i have not tried for two weeks Mar 03 16:54:44 http://paste.debian.net/239562/ Mar 03 16:55:22 JaMa: moron question.. do you have tmpfs in your kernel? Mar 03 16:55:26 I'll probably return systemd recipe and bbclass to meta-systemd to get old working behavior back Mar 03 16:55:48 zecke: yes and same image was working fine with systemd from meta-systemd Mar 03 16:56:06 zecke: this was just to try oe-core's copy Mar 03 16:56:37 JaMa: I will re-base my work-in-progress sysmoBTS image this week and can comment then. Mar 03 16:57:14 zecke: ok, please comment in "[OE-core] RFE: make the init manager an image feature (again)" Mar 03 16:58:10 thanks for pursuing this guys, we need more testing by a diverse group on master Mar 03 16:58:36 I'm only able to test simple aimges for baord bring up Mar 03 16:58:44 so can't test alternate inits atm Mar 03 17:29:44 JaMa: I am building a qemu image now Mar 03 17:30:06 JaMa: I am rebasing over 55d1757..82cb88e on poky Mar 03 17:37:39 JaMa: you are the author of the meta-qt5 repo on github? Mar 03 17:38:13 yes one of authors Mar 03 17:39:09 I am tweaking this one to meet my requirements, which are: (1) no X (only EGL & opengl ES 2), (2) using an i.MX6 as target, which means I have to pass "-device imx6" to ./configure (3) webkit-qml added Mar 03 17:39:21 don't use it yet, I've fixes for qt5-native and qtjsbackend-native recipes, but need to finish adaptations to target recipes to build correctly Mar 03 17:39:23 now I wonder if there are plans on your side for any of these Mar 03 17:39:53 no plan for i.MX6 for sure Mar 03 17:40:15 no X should work (we're using it without X) Mar 03 17:40:57 well the virtual/libgl dependency can be a problem. I replaced it with virtual/libgles2 and virtual/egl Mar 03 17:42:10 and you are using it without X. on what platform? Mar 03 17:42:32 we need it working in qemu/vbox so swrast from mesa needs to work too Mar 03 17:42:48 on armv7a device Mar 03 17:42:50 because, the EGL initialization is platform specific Mar 03 17:43:15 we have own qpa Mar 03 17:43:22 with a SGX graphics chip, the defaults work. with for example an iMX6, it wouldnt work without the -device imx6 flag Mar 03 17:43:38 because then, qtbase is built with imx6-specific EGL initialization Mar 03 17:44:41 I guess just adding "-device " to EXTRA_OECONF in the platform recipes is enough. still, its a good idea to keep that in mind. Mar 03 19:48:45 JaMa: I built on my laptop so it took a bit. poky HEAD boots my sysmocom-nitb-image with systemd all the way to the login Mar 03 20:44:21 zecke: thanks Mar 03 20:52:32 anyone do much with vmdk images? Just tried booting my first vmdk image and got stuck in an infinite loop looking for initrd. Mar 03 21:18:36 um JaMa, is there a reason why you dont set EXTRA_OECONF in qtbase.inc ? you do it in qt5-native.inc, after all Mar 03 21:21:00 dv_: see jansa/test branch in meta-qt5, I plan to change qtbase Mar 03 21:21:09 right now it's just broken wrt sstate Mar 03 21:21:17 hmm okay Mar 03 22:07:34 JaMa: is there anything you want to know about my systemd image? Mar 03 22:13:08 you have systemd working with yocto? Mar 03 22:14:47 well, the work in progress in OE-core Mar 03 22:15:16 just curious: does it improve boot speed any? Mar 03 22:16:20 I didn't measure that Mar 03 22:16:49 for an embedded device, figured boot time is one of the important factors. Mar 03 22:18:04 hw watchdog and monitoring of tasks is my main reason to look at it Mar 03 22:26:43 hw watchdog better supported would be great. Mar 03 22:29:50 zecke: not really, I'm working on something else now and I have working images created with meta-systemd to compare with the one created with systemd from oe-core Mar 03 22:30:35 zecke: in the end I'll proably just fix parsing of meta-systemd, because I care about upgrade paths and scenarios reported in that email thread Mar 03 22:52:56 okay Mar 03 22:53:05 * zecke heads to bed Mar 03 23:23:09 blloyd_, an embedded device typically only loads a handful of services - in that scenario systemd will do more harm than good Mar 03 23:28:22 also, embedded devices typically have slow internal I/O Mar 03 23:28:47 so parallelizing service startups might actually slow down things **** ENDING LOGGING AT Mon Mar 04 02:59:58 2013