**** BEGIN LOGGING AT Fri Jul 04 02:59:58 2014 Jul 04 04:05:09 * moto-timo is going to bed Jul 04 07:20:51 what does people in here use to deploy OE to devices in a production facility? Jul 04 07:50:29 good morning Jul 04 07:59:28 hi mckoan Jul 04 08:46:05 anarsoul: gm Jul 04 08:46:34 tried to test the calibration but not yet there... Jul 04 08:46:47 OOM (udevadm) Jul 04 08:47:18 good morning, ant_work Jul 04 08:47:59 ant_work: 32mb ram? Jul 04 08:48:08 no, 64 Jul 04 08:48:13 hmmm Jul 04 08:48:37 make sure that CONFIG_LEGACY_PTYS is disabled in kernel Jul 04 08:48:57 I have two suspects: 1) meta-oe may be adding smthg bad Jul 04 08:49:18 2) recent rework of udev recipes / compatibilty for systemd and sysvinit Jul 04 08:49:19 2month-old build works ok on 64mb Jul 04 08:49:38 core-* images built with oe-core only did boot ok Jul 04 08:49:51 couple of weeks ago Jul 04 08:50:13 anyway, check for CONFIG_LEGACY_PTYS Jul 04 08:50:29 it's funny how we battle wit the damned udev cache since years :/ Jul 04 08:50:55 anarsoul: kernel has not changed, even I've removed stuff recentl Jul 04 08:51:21 ant_work: just check, those legacy PTYs eat too much space in tmpfs Jul 04 08:51:40 with this config enabled I'm not able to boot oe rootfs on rx1950 (32mb ram) Jul 04 08:53:27 anarsoul: changes in oe-core 9 days ago...hmmm Jul 04 08:56:43 ant_work: I still suggest you to check if CONFIG_LEGACY_PTYS is disabled. Jul 04 08:57:03 it is not explicitely enabled Jul 04 08:57:34 could you share your rootfs? Jul 04 08:58:01 yes but later, is on another machine Jul 04 08:58:16 for armv4/collie you mean? Jul 04 08:58:53 I'm looking now at the recent commits playing withudevadm/udev_cache Jul 04 09:11:45 anarsoul: heh..like in 2011 http://infobot.rikers.org/%23oe/20110518.html.gz Jul 04 09:12:10 I remember having battled with it couple of times Jul 04 09:13:00 :) Jul 04 09:14:52 Jul 04 09:15:01 really, we could live with devtmpfs Jul 04 09:15:36 now we have all the CGROUPS stuff just because of udev Jul 04 09:16:05 yeah, we could Jul 04 09:17:56 udev on a device like this is just a waste of memory/cpu time Jul 04 10:03:02 Hello there. Does anyone now which package contains weston demos (like weston-gears etc.)? The basic wayland/weston setup just gives me weston, weston-info, weston-launch and weston-terminal. Jul 04 12:02:00 I hate holidays! Jul 04 12:03:55 Phil, shut up and enjoy Jul 04 12:04:00 ;) Jul 04 12:04:11 :) Jul 04 12:04:32 don't miss the Android vs. armv7 SWP thread Jul 04 12:04:37 if you have free time Jul 04 12:04:47 lakml? Jul 04 12:04:48 it's already 49 posts and it's 4th july Jul 04 12:04:51 yes Jul 04 12:05:38 http://lists.infradead.org/pipermail/linux-arm-kernel/2014-July/268504.html Jul 04 12:11:51 ant_work: Relatively few of the people there are in the US! Jul 04 12:17:52 ^_^ Jul 04 12:18:35 broonie: really one single patch over vanilla and Android boots? wow Jul 04 12:18:43 unexpected Jul 04 12:19:59 ant_work: Most of the trick is device support. Jul 04 12:26:04 Crofton|work: one of last RMK posts is even summoning Linus T. Jul 04 12:26:11 that's th enuclear strike Jul 04 12:26:12 omg Jul 04 12:26:26 I saw ones where they suggested that would be like summoning Sauron Jul 04 12:26:34 he he Jul 04 12:27:00 mru is arguing with rmk again Jul 04 12:28:12 yea, uncomfortable Jul 04 12:33:51 someone from .us should complain to rmk about arguing on a .us holilday Jul 04 12:35:50 Crofton|work: LOL Jul 04 12:36:12 as I recall someone mentioned he complained about people emailing on a weekend :) Jul 04 12:39:55 heh Jul 04 12:39:57 SWP Jul 04 12:40:11 that was discussed at lenght during connect macau as well Jul 04 12:40:18 Crofton|work: I've got to know about his ways a bit 'rough' when I was working with StrongArm in 2002 ;-) Jul 04 12:41:54 arguing with him is a problem because he's mostly right ;) Jul 04 12:43:09 "Asking them all to Jul 04 12:43:10 recompile is not feasible" Jul 04 12:43:12 hahahaha Jul 04 12:43:26 android sucks, news at 11 Jul 04 12:44:22 I figure the handset manufacturers want you to replace your handset every year anyway Jul 04 12:49:46 and you still get a 2 year old android version with it Jul 04 12:50:55 I don't own any A. device atm but I guess when you upgrade the 'firmware' you upgrade both kernel + userspace Jul 04 12:51:11 or is it perverswely done? Jul 04 12:51:51 and for the apps in the store, well, how many stores? you can count on one hand Jul 04 12:52:19 there are a lot of app stores Jul 04 12:52:28 but only one official google store Jul 04 12:52:36 really? didn't know Jul 04 12:54:12 heh, it's expected that some old apps aren't work on newer device and/or firmware Jul 04 12:54:33 (for Android, of course) Jul 04 12:54:52 Apple is more drastic with that Jul 04 12:55:02 and the devices are much more expensive Jul 04 12:55:11 (in the past at least) Jul 04 12:56:18 people wanting to sell to owners of ios6 or ios7 had to recompile the older apps Jul 04 12:56:31 end Jul 04 12:56:45 what I want to say is that there won't be armv8 android for already released versions of android Jul 04 12:56:48 i.e. for 4.4 Jul 04 12:57:13 http://releases.linaro.org/14.06/android/images/armv8-android-juno-lsk Jul 04 12:57:23 that's android64 4.4.3 Jul 04 12:57:26 they'll probably release 5.0 for armv8, so it's ok to say to devs "you have to recompile your apps to make them work with android v5" Jul 04 12:57:53 koen: are there any android devices in production with armv8 cpu? Jul 04 12:58:46 none for sale Jul 04 12:58:57 the only armv8 device you can buy is the iphone 5s Jul 04 12:59:20 the server type stuff is still "call us, we'll give you a quote" Jul 04 12:59:41 maybe in a few weeks you can buy armv8 phones in china Jul 04 13:02:54 koen: so it's only a matter of bumping sdk version or whatever they have Jul 04 13:03:27 instead of messing with kernel and introducing emulation of not-anymore-supported instructions Jul 04 14:44:09 anarsoul: btw "If 64MB or less, then we need to start udev daemon with parameter --children-max=1, which will effectively disable all parallel processing" Jul 04 14:45:31 that seems for systemd, though Jul 04 15:37:02 ant_work: oh, ok Jul 04 21:04:17 ant__: so, did my patch fix ts calibration for your device? Jul 04 21:04:26 hi Jul 04 21:04:31 still battling with OOM Jul 04 21:04:49 single-threaded udev didn't help? Jul 04 21:05:03 I've tried first disabling udev-cache Jul 04 21:05:10 in not enough :/ Jul 04 21:06:47 so, basically rootfs built by oe-core isn't bootable on machines with 64mb or less ram? Jul 04 21:08:20 http://pastebin.com/qRAYstC5 Jul 04 21:08:47 see, using oe-core alone and core-image-* I didn't hit that issues Jul 04 21:08:59 besides, there is a new one Jul 04 21:09:15 Cannot open `/mnt/boot/zImage': No such file or directory Jul 04 21:09:28 lrwxrwxrwx 1 root root 34 lug 4 00:31 zImage -> /boot/zImage-3.14.5-yocto-standard Jul 04 21:09:36 -rw-r--r-- 1 root root 2108008 lug 3 23:26 zImage-3.14.5-yocto-standard Jul 04 21:09:40 this doesn't boot Jul 04 21:09:57 plian symlink like this is ok Jul 04 21:09:57 lrwxrwxrwx 1 root root 28 lug 4 22:50 zImage -> zImage-3.14.5-yocto-standard Jul 04 21:10:17 I guess there is some toxic change in oe-core wrt kernel symlinks Jul 04 21:10:52 14mb for qpe? Jul 04 21:11:16 has not yet started Jul 04 21:13:23 what's the difference in init stuff between core-image-* and opie-image? Jul 04 21:13:47 default in oe-core is still sysvinit Jul 04 21:14:14 maybe some systemd contamination is coming from meta-oe? Jul 04 21:14:49 I don't use that layer for my normal builds Jul 04 21:15:23 ant__: http://pastebin.com/9UYux7D5 Jul 04 21:41:02 anarsoul: here I've tried --max-childs=1 to udevadm control Jul 04 21:41:04 http://pastebin.com/3pFRkrp3 Jul 04 21:41:52 I have to blacklist mtd probing Jul 04 21:41:53 mtd_probe invoked oom-killer: Jul 04 23:53:15 Hi everybody Jul 04 23:53:23 I need some help Jul 04 23:53:56 I am trying to build a simple console image for angstrom distro and I am getting an error on bluez5 recipe Jul 04 23:54:25 but I don't need this package. I would like to configure my settings to tell oe not to build it Jul 04 23:54:31 but I don't know how to do that Jul 05 00:00:11 Slyde: which image are you building? Jul 05 00:04:11 console-image Jul 05 00:05:49 maybe I figure it out... Jul 05 00:06:18 I added the meta-eca layer to have the connman-init-systemd recipe and this requires bluez5 Jul 05 00:06:54 oe was using the recipe from the meta-eca layer but now I set the preferred version to 5.15 which is the version from the meta layer Jul 05 00:07:00 and now it is compiling Jul 05 00:13:15 excellent. good luck! Jul 05 00:15:24 0: console-image-1.0-r0 do_rootfs (pid 30271) Jul 05 00:15:35 good Jul 05 00:16:10 sometimes, we just need to explain the issue Jul 05 00:16:17 thanks for reading Jul 05 00:24:25 it's always good to have a sounding board. Just saying it outloud makes the problem suddenly appear. **** ENDING LOGGING AT Sat Jul 05 02:59:59 2014