**** BEGIN LOGGING AT Wed Jan 20 02:59:59 2016 Jan 20 08:10:28 good morning Jan 20 10:06:14 bluelightning: i was more interesting in seeing if anyone is actually using it Jan 20 11:05:50 rburton: using what now? Jan 20 11:06:42 eudev I guess, he seems obsessed Jan 20 11:10:04 prompted by sgw discovering that udev 182 doesn't work with kernel 4.4 Jan 20 11:10:47 better go to FOSDEM and watch Lennart's talk, since systemd is the future Jan 20 11:11:25 * Crofton|work needs to add the patchworl talk to related OE talks Jan 20 11:11:43 https://fosdem.org/2016/schedule/event/systemd/ Jan 20 11:12:25 hmmm, did it get accepted? Jan 20 11:13:05 https://fosdem.org/2016/schedule/event/patchwork/ Jan 20 11:13:10 annoying, lightening talk Jan 20 11:13:36 still gets the project some eyeballs Jan 20 13:22:55 eudev, now is that the gentoo fork or one of the others? Jan 20 13:26:15 the gentoo one Jan 20 13:26:23 also used by LFS and Arch now, if you don't want systemd Jan 20 13:27:29 arch is going back to sanity land? Jan 20 13:27:43 rburton: and Alpine Linux, which doesn't have systemd at all Jan 20 13:28:07 (which I discovered through Khem taking musl-related patches from there) Jan 20 14:02:59 belen, is trying to do a toaster build so she ha something to show at FOSDEM that is close to the demo systems I ahve Jan 20 14:03:11 but with meta-sdr etc she sees this with jthro Jan 20 14:03:12 Nothing PROVIDES 'libsdl-native' (but Jan 20 14:03:13 virtual:native:/home/yocto/master/_toaster_clones/_git___git.yoctoproject.o Jan 20 14:03:13 rg_poky_jethro/meta/recipes-devtools/qemu/qemu_2.4.0.bb DEPENDS on or Jan 20 14:03:13 otherwise requires it) Jan 20 14:04:34 possibly related to the removal of sdl-native from ASSUME_PROVIDED ? Jan 20 14:05:53 using a local.conf from master with the jethro branch, maybe? Jan 20 14:08:23 belen, ^^^^ Jan 20 14:09:41 Crofton|work, joshuagl: mmm, I see. If that was the problem, shouldn't I see it in any build I attempt with the same Toater set up? Jan 20 14:10:32 I do not thin I use anything related to sdl/qemu unless by weird accident Jan 20 14:15:35 Crofton|work: I'll build something from the poky layers and see what happens, just in case it's a toaster bug Jan 20 14:23:26 belen: yes, any build for a qemu machine at least (as that's what adds qemu-native to the dependency chain) Jan 20 14:26:22 joshuagl: and indeed this happens with any jethro build you start from toaster. What a bug-finding day :/ Jan 20 14:29:01 qemu-native gets into deps chains very easily, ie using fontconfig means image construction needs qemu-native Jan 20 14:35:35 oh, of course. So it's almost certainly going to be in a shr image then Jan 20 14:57:17 Crofton|work: mystery solved https://bugzilla.yoctoproject.org/show_bug.cgi?id=8958 Jan 20 14:57:23 thanks for asking Jan 20 14:57:46 thanks belen Jan 20 14:58:17 need to beat toaster so it works for everyone :) Jan 20 14:58:38 at this point, having some builds where we can show depends flowing out from gqrx Jan 20 14:58:51 should be good enough for people we see at FOSDEM Jan 20 14:59:12 I'm also curious how people react to two things running same thing Jan 20 15:42:25 i am using powertop and it shows nan/wakeups/sec Jan 20 15:43:48 I'm still surprised toaster has so many assumptions about how people do things, considering our flexibility is such a core part of the project Jan 20 16:02:36 kergoth, ++ Jan 20 16:02:51 but we do not to help them identify these assumptions :) Jan 20 16:02:57 we do need to Jan 20 17:14:10 Crofton|work: who says Toaster doesn't work? ;) http://imgur.com/VuZjOIa Jan 20 17:14:34 excellent Jan 20 17:15:02 Crofton|work: should it work for Minnow? Jan 20 17:15:07 it should Jan 20 17:15:10 run demo image Jan 20 17:15:19 takes a bit longre, but has app demo systemns ahve Jan 20 17:15:42 Crofton|work: will try that next then. Any other hardware you want me to build it for? Jan 20 17:16:00 try ettus-e300 or ettus-e3xx-sg1 Jan 20 17:16:06 whichevver is in jethro :) Jan 20 17:16:42 Crofton|work: will do all that and let you know if I run into any trouble Jan 20 17:16:51 thanks Jan 20 17:45:43 rburton: http://imgur.com/031PV9z thats XFCE on musl Jan 20 18:00:15 khem: nice! Jan 20 18:00:27 khem: fwiw your musl branch passed my local tests, just needs to pass a full AB run Jan 20 18:35:35 Is anyone using OE to build within Docker? If so any tips ? Jan 20 18:36:12 I.e. to isolate the host machine away as much as possible. Jan 20 18:37:10 Or perhaps Rockit containers ? Jan 20 18:37:19 https://coreos.com/blog/rocket/ Jan 20 19:16:18 rburton: ok Jan 20 19:16:33 I am going to start running some runtime benchmarks now Jan 20 19:34:22 morning all Jan 20 19:36:40 gm bluelightning Jan 20 19:38:01 hi khem Jan 20 20:49:18 denix: like rpi do we have sdcard image type for beagle ? Jan 20 20:49:38 denix: in angstrom days it was done kind of externally IIRC Jan 20 20:53:52 khem: we've been using wic for beagle, the sdcard bootpart wks likely works as is, we use variants of it for different sd card sizes Jan 20 20:54:10 particularly nice alongside the wic image fstype Jan 20 20:56:54 kergoth: do you have some instructions Jan 20 20:59:57 khem: https://github.com/MentorEmbedded/meta-mentor/blob/9144315b454f4d53b6ed6fba804a518f2842ba38/meta-mel/conf/include/wic-common.inc + WKS_FILE = "sdimage-bootpart.wks" Jan 20 21:02:32 then for boneblack specifically, IMAGE_BOOT_FILES ?= "${UBOOT_BINARY} MLO uEnv.txt uboot.env", IMAGE_DEPENDS_wic_append = " virtual/bootloader" where UBOOT_BINARY is likely u-boot.img Jan 20 21:02:37 that'll get the right files into teh boot partition Jan 20 21:03:08 i think i bbappended u-boot to inject uEnv.txt & uboot.env just to get them into the boot partition, since something has to provide them Jan 20 21:03:41 yeah, i had a do_deploy_append(){} to put them in ${DEPLOYDIR}/ Jan 20 21:03:48 that's what works for us, anyway Jan 20 21:03:57 not sure if that's ideal Jan 20 21:04:00 * kergoth shrugs Jan 20 21:04:21 works nicely once you have it set up Jan 20 21:06:14 https://github.com/MentorEmbedded/meta-mentor/tree/master/meta-mel/scripts/lib/wic/canned-wks has the fixed size sdcard wks files Jan 20 21:06:51 not perfect, since afaict wic has no way to specify the max image size and let the rootfs grow to it, so the rootfs size is hardcoded in the wks for each, and even then the exact available size on the card varies from card to card Jan 20 21:06:52 but works Jan 20 21:06:58 at least for the cards we've tested Jan 20 21:07:20 a better route is to grow the partition to the card size at runtime, that's what the boneblack debian image includes Jan 20 21:07:28 need to get around to adding a script for that Jan 20 21:07:35 yes thats more useful Jan 20 21:07:48 and simply addresses all sizes Jan 20 21:08:07 yeah. still useful to have fixed size files available for read-only-rootfs support, but that's about it Jan 20 21:08:40 I think meta-yocto should have this as a sample for bb Jan 20 21:08:57 that's a good idea Jan 20 21:09:16 may be you should contribute it Jan 20 21:09:21 now that wic is more usable it'd be nice to see all the various sdcard scripts and bbclasses to go away eventually.. Jan 20 21:09:26 true, i'll add to the todo Jan 20 21:09:39 working on submitting as much as possible for the 2.1 timeframe now anyway Jan 20 21:09:54 yeah I have tonnes to upstream here as well Jan 20 21:10:01 but mostly to meta-oe Jan 20 21:10:14 OE core with musl is good for me Jan 20 21:10:23 nice, i have yet to find the time to play with it :( Jan 20 21:10:59 kergoth, ++ Jan 20 21:11:00 its neat you mentor should think about it for mel-lite or mel-tiny :) Jan 20 21:11:20 I'd like to see all BSP's come with clear instructions how to make borads boot Jan 20 21:11:29 infact, I even see it to replace glibc in most of non LSB things Jan 20 21:11:42 I have XFCE desktop build using it Jan 20 21:12:03 agreed Jan 20 21:16:04 I got my sdr image running on a minnow easily, only hard part was finding instructions for making bootable media Jan 20 21:23:52 minnow has really good performance I must say Jan 20 21:24:37 I'm just happy to have the same thing on two very different boards Jan 20 21:26:18 true Jan 20 21:26:54 and toaster data to go with it :) Jan 20 21:36:35 kergoth: uboot.env and uEnv.txt files are not part of standard build for u-boot right ? Jan 20 21:48:44 khem: there used to be sdcard image class in meta-ti, but it wasn't supported and got dropped. we build sd card images outside of bitbake for our products - I was hoping to eventually use wic Jan 20 21:51:19 Hello there, is anyone using Docker or Rocket as a build image machine layer for OE builds? If so I'd love to know more... Jan 20 21:52:57 Im just about to give this a try so it would be great to know the pro's/con's before hand. Jan 20 22:06:51 denix: it would be great if we could determine what the barriers are to using wic and have bugs filed to address them - is that something you could help with? Jan 20 22:09:31 seems this is the change that broke xterm that I'd mentioned earlier... Jan 20 22:09:35 paul@yow-cube1:~/poky/meta-openembedded$ git show ad5d2ff6c782a3319aedc25cc641574e39328460|grep ^[-+]FILES_ Jan 20 22:09:35 -FILES_${PN} += " /usr/lib/X11" Jan 20 22:09:35 +FILES_${PN} += "${libdir}/X11" Jan 20 22:09:36 paul@yow-cube1:~/poky/meta-openembedded$ Jan 21 00:53:36 bluelightning: SD image format is too machine specific at the moment Jan 21 00:53:54 bluelightning: wic is in right direction Jan 21 00:54:15 khem: right, but we need to know what the concrete changes required are Jan 21 00:54:17 but we need to let machines specify some piece of it Jan 21 00:54:34 of course Jan 21 00:54:35 from highlevel Jan 21 00:54:50 all a machine should be concerned about is Jan 21 00:54:58 1. boot partition 2. rootfs partiion Jan 21 00:55:06 everything else is distro choice Jan 21 00:55:19 if they want to have another logical partition Jan 21 00:55:21 or so Jan 21 00:55:58 so if wic can let machines choose. 1. and 2. with its formats they support Jan 21 00:56:02 and content of boot partition Jan 21 00:56:07 we should be good Jan 21 00:56:15 I think rootfs can be generic as well Jan 21 00:56:31 its just content of /boot Jan 21 00:56:40 and kmods may be Jan 21 00:56:54 but format wise they are standard locations Jan 21 00:57:29 IMO machines should provide a small fragment into larger wic config Jan 21 00:57:39 not otherway around Jan 21 00:57:56 then wic should also impose its requirements on machines Jan 21 00:58:00 it can not be one way Jan 21 00:58:31 something like in order to support wic machine should support x, y, z **** ENDING LOGGING AT Thu Jan 21 02:59:58 2016