**** BEGIN LOGGING AT Wed Jul 12 03:00:05 2017 Jul 12 06:27:09 Why does util-linux.inc check for DISTRO_FEATURES here: Jul 12 06:27:09 PACKAGES =+ "${@bb.utils.contains('DISTRO_FEATURES', 'pam', 'util-linux-runuser', '', d)}" Jul 12 06:27:47 DISTRO_FEATURES is merely the default for pam in PACKAGECONFIG: Jul 12 06:27:47 PACKAGECONFIG_class-target ?= "${@bb.utils.filter('DISTRO_FEATURES', 'pam', d)}" Jul 12 06:28:04 But one could have pam enabled in DISTRO_FEATURES and disable in PACKAGECONFIG. Jul 12 06:28:31 That's not wrong per se (the package probably ends up being empty), but very odd. Jul 12 09:52:59 maxin: pkgconf works \o/ Jul 12 09:53:07 going to throw it at the AB to see what explodes Jul 12 11:05:25 rburton_: what's the idea with mesa having so many -dev packages? Jul 12 11:12:30 I'm asking since e.g. khrplatform.h goes to libegl-mesa-dev which is not correct (it's also needed by gles) Jul 12 11:19:56 Hi, though kernel config for "CONFIG_HW_RANDOM" "CONFIG_HW_RANDOM_INTEL" are set to "y" Jul 12 11:19:57 I am getting .ko files in rootfs image Jul 12 11:20:00 any clues ? Jul 12 11:21:27 not getting* Jul 12 11:27:58 rburton_: good to hear that :) Jul 12 11:34:40 arpallet: "y" builds the driver in-kernel right? so no .ko files Jul 12 11:35:09 arpallet: as jku said, shouldn't that be "m" instead of "y" ? Jul 12 11:36:25 Thanks jku maxin, ya missed that. Will update to "m" Jul 12 11:50:28 jku: tbh probably seemed like a good idea at the time Jul 12 11:55:51 rburton_: good ideas? I could add yet another package for that one generic khronos header ... Jul 12 11:56:03 mesa-dev itself currently has dri.pc (?) and vulkan driver headers Jul 12 11:56:08 or merge the dev stuff all in one Jul 12 11:56:21 yeah, just wondering if a lot of things will break Jul 12 11:56:38 maybe I could try that Jul 12 11:56:53 well with sufficient provides you'll keep the migration path. magic deps *are* broken sadly for situations like that. Jul 12 11:57:01 but with provides that might just work Jul 12 11:57:10 right Jul 12 11:58:01 I'll send the rpi build fix first then, this might be a bit trickier Jul 12 12:35:15 hello guys, for systemd unit files there is the "systemd_system_unitdir" variable Jul 12 12:35:33 is there any variable that points to /lib/systemd/network ? Jul 12 12:37:00 no. those variables can be found in bitbake.conf Jul 12 15:11:04 Hi, I tried submitting a patch to oe-devel on Monday afternoon but I have not seen my message turn up in the archives or anywhere else yet. I tried contacting the mailing list admin yesterday but have not seen a response yet. Can someone check if my emails are going straight to a junk folder? Jul 12 15:11:30 I successfully subscribed to the list on Monday before sending my patch. Jul 12 15:13:05 TafThorne: i see mails from you on the yocto and oe-core lists, but nothing on oe-devel Jul 12 15:14:38 kergoth: Thank you for checking. Maybe I got confused about where I sent the patch and I have been looking the wrong place. Let me check that. Jul 12 15:16:52 On Monday at 16:04 (current UK times) I sent an email titled "[meta-networking][krogoth][PATCH] Update ser2net To Version 3.1" to openembedded-devel@lists.openembedded.org. I was CCed on this message. Jul 12 15:17:28 I sent a follow up message to listmaster Tuesday at 11:28. Jul 12 15:22:11 kergoth: unless I got the address wrong, I did attempt to send the patch to oe-devel. I sent a separate email at noon today to the yocto list about something else. I've noticed that list can take a few hours to publish so I was not too worried about seeing that in the archive yet. Jul 12 15:26:04 Anyone here have experience enabling gfortran in core-image-sato-sdk? Jul 12 16:00:08 nas_: define FORTRAN = ",fortran" in local.conf Jul 12 16:03:44 Any additional required steps? Jul 12 16:04:05 The directions I followed (which failed, hence me being here) had more things involved, but perhaps the additional steps caused all the trouble? Jul 12 16:24:16 kergoth: see what you think of http://git.yoctoproject.org/cgit.cgi/poky-contrib/commit/?h=rpurdie/wip-rss2&id=d723084440752aedd9585993c568aafe5e956911 if you have a minute. Still need to clean up a few bits of it but its getting there and works Jul 12 16:26:08 RP: conceptually i'm fully on board, i'll take a look at the code a bit later today when i have some time Jul 12 16:27:51 kergoth: thanks. I think this one is a merge and then deal with any fallout type of patch... Jul 12 16:28:12 * RP will try and get it some wider testing Jul 12 16:28:15 i'd agree with that, we need real world testing to hit any corner cases Jul 12 16:32:43 RP: i was about to fire a mut, shall i pull it in for giggles? :) Jul 12 16:32:55 what could posssibly go wrong Jul 12 16:38:18 rburton: yeah apply muprhy's law Jul 12 16:39:07 RP: whats your take on turning hardening on by default for poky Jul 12 16:41:14 zeddii: do we enable ASLR in default configs for yocto kernels ? Jul 12 16:43:19 RP: i love the mess of your wip branches, starting with how its named wip-rss2 and ending with commits like revert: revert: revert: psplash: test. Jul 12 16:44:54 RP: can't just pick that commit sadly Jul 12 16:46:28 khem. which variant ? i.e. RANDOMIZE_BASE, or the sysctl enablement ? Jul 12 16:55:16 rburton: I'm curious Jul 12 16:55:30 rburton: just ignore the single reject from the cherry-pick Jul 12 16:55:54 zeddii: pie compiler flags I think Jul 12 16:56:05 khem: it may be time, there is a performance cost though :/ Jul 12 17:13:01 randomize_base needs the relocatable flags, so while arch dependent, they are normally on. there are different levels of knobs that can be turned after that. Jul 12 17:33:47 Hi guys! I wonder how I can check which packages are build for 'bitbake meta-ide-support' and can I add some. Can you help me? Jul 12 17:58:16 RP: I think with iot world of today security is important concern Jul 12 17:58:37 RP: we see all major desktop distro switch in 2017 Jul 12 18:00:53 RP: more and more embedded systems are now connected devices, it will be a welcome change if we made that as default Jul 12 18:55:56 khem: the method you described earlier for enabling fortran did not work. It results in an error relaing to backtrace-supporting.h not being found Jul 12 18:58:46 nas_: it might be a different problem thats latent and only shows up when fortran is enabled in gcc Jul 12 19:01:03 I tried to ensure that libbacktrace is included in the build, but it seems the gfortran recipe cannot find it Jul 12 19:01:09 (or it failed to build for some reason) Jul 12 19:03:37 Ok so the file "backtrace-supported.h" does exist but the libgfortran build process cannot locate it. What is the process for fixing this? Jul 12 19:04:24 nas_: which release are you on ? Jul 12 19:04:32 2.3 Jul 12 19:05:30 anyone ever see issues like this running docker on mac osx? https://gist.github.com/mattsm/c23d67400ec9ba52e4a5860fadc972ac Jul 12 19:05:58 awesome error Jul 12 19:06:05 presumably some disaster involving fake file systems Jul 12 19:13:03 rburton e.g. it will be a pita to debug Jul 12 19:13:41 i'd suggest wiping tmp and trying again to see if that helps Jul 12 19:13:59 it's fuse.osxfs too, neat Jul 12 19:15:28 ah. bets on that upstream having a file name and directory name that only differ by case? Jul 12 19:16:19 yes :) Jul 12 19:16:39 it has a PATCHES file in the tarball, but we drop the patches into a patches/ directory Jul 12 19:17:12 probably, related Jul 12 19:17:12 https://docs.docker.com/docker-for-mac/osxfs/ Jul 12 19:17:13 you can work around it by writing a do_patch prefunc that deletes ${S}/PATCHES Jul 12 19:17:34 yes, hfs+ is case insensitive by default Jul 12 19:17:41 because thanks apple Jul 12 19:17:47 what would be the upstream way to fix this? Jul 12 19:18:06 to abort if we detect a case-insensitive file system Jul 12 19:18:11 writing the sanity test now Jul 12 19:18:18 this is just the first failure, you'll find more Jul 12 19:18:20 a mac osx fixup meta layer? Jul 12 19:18:22 i know Jul 12 19:18:29 just curious, if i get really interested in going all the way Jul 12 19:19:08 i recommend building in a case-sensitive loopback or something, but a fixup here would be a do_unpack postfunc or do_patch prefunc to delete ${S}/PATCHES Jul 12 19:20:45 nas_: I think there were issues with gcc6 and fortran builds we might be prone to those Jul 12 19:21:04 It does seem that way. I am trying a workaround as we speak Jul 12 19:21:15 rburton probably better to do the switch, good to figure the underlying issue though Jul 12 19:27:25 mattsm: just switch to a more capable filesystem. case-sensitive hfs+ disk image, etc Jul 12 19:29:24 khem: failing this workaround, is it possible (I imagine it is) to tell gcc to build an earlier version in order to avoid this issue? Jul 12 19:32:07 ERROR: OE-core's config sanity checker detected a potential misconfiguration. Jul 12 19:32:07 Either fix the cause of this error or at your own risk disable the checker (see sanity.conf). Jul 12 19:32:08 Following is the list of potential problems / advisories: Jul 12 19:32:08 The TMPDIR (/data/poky-tmp/master) can't be on a case-insensitive file system. Jul 12 19:32:10 fixed :) Jul 12 19:34:13 kergoth rburton and done... ;) Jul 12 19:37:17 rburton: nice Jul 12 19:38:12 presumbly the new apple fs is case-insensitive too, would be nice if they switched that to something sensible Jul 12 19:38:35 could be pretty invasive for long time users, though. i doubt it'll happen Jul 12 19:39:18 Presumably they will keep the option, as they have in the past Jul 12 19:39:56 yeah Jul 12 19:40:21 So if you don't want to run Adobe products, you can make your /Application case-sensitive :-P Jul 12 19:41:10 "APFS, like HFS+, is case-sensitive on iOS and is available in case-sensitive and case-insensitive variants on macOS, with case-insensitive being the default." Jul 12 19:41:10 https://developer.apple.com/library/content/documentation/FileManagement/Conceptual/APFS_Guide/FAQ/FAQ.html Jul 12 19:41:31 Hey all - I'm working on a derivative of the sabre imx7d. It can boot into linux just fine on the mmc, but for some reason the update and setupdate commands aren't available in u-boot and if I connect to it with the mfgtools it doesn't let me flash Jul 12 19:41:57 anyone know what could cause that? I haven't changed the u-boot, zimage, or device tree on the development PC that's flashing the imx7 Jul 12 19:42:26 I haven't changed ucl2.xml either Jul 12 19:43:05 I even switched iMX7 boards to make sure it wasn't something on-board causing the behavior Jul 12 19:44:14 apfs case-sensitive is still an option, but still non-default because Adobe still haven't quite got the hang of OSX - http://i.imgur.com/ODs5GLr.png - but a nice touch is now that volumes can share space within the same container, you don't have to worry about sizing. Jul 12 20:21:15 RP: mut fired with your patch in Jul 12 20:27:23 bah, royalpurple left, you can jump to the hab failsafe code from uboot Jul 12 20:29:51 https://github.com/reMarkableAS/uboot/blob/zero-gravitas/board/reMarkable/zero-gravitas/zero-gravitas.c#L1087 is the address in case he returns Jul 12 20:41:23 kergoth rburton not i hit this issue https://github.com/wrpseudo/pseudo/issues/5 Jul 12 20:41:48 anyways, was just trying to see if it would be easy... seems not super easy Jul 12 20:42:13 ah fun, couldn't bind listening socket: File name too long Jul 12 20:42:25 that might be an easy fix in pseudo Jul 12 20:42:33 give the docker a proper file system... Jul 12 20:43:48 rburton i gave it a case-sensitive one... Jul 12 20:43:55 ext4 ;) Jul 12 20:44:17 if you care, spending ten minutes chasing where that error comes from and if the path is just something hfs doesn't like shouldn't be too difficult Jul 12 20:44:26 we build pseudo from source anyway, so you've got it to hand Jul 12 20:45:51 rburton was at least going to try the latest version Jul 12 20:46:33 (of psuedo) Jul 12 21:56:50 hmm, do we have a common bit of code/logic to adjust a defconfig or enable/disable fragments based on PACKAGECONFIG? Jul 12 21:56:53 khem? Jul 12 21:56:56 hmm Jul 12 22:16:01 zeddii: http://errors.yoctoproject.org/Errors/Details/147287/ just happened. i think its a rare race Jul 12 22:18:13 hmm, update_font_cache hit a segfault building for qemux86-64.. there's something very wrong with that Jul 12 22:19:45 qemu: uncaught target signal 11 (Segmentation fault) - core dumped Jul 12 22:19:47 hrmph Jul 12 22:25:49 anyone else run into that? Jul 12 22:25:52 before i dive in Jul 12 22:27:22 master? Jul 12 22:27:29 reproducable on demand? Jul 12 22:27:49 there was a change to the qemu machine flags recently, which shouldn't have but may have broken emulation again Jul 12 22:42:50 kergoth: I dont think we have common bit to match packageconfig to kconfig Jul 12 22:43:12 k, figured as much. i'm guessing a copule random recipes probably have their own version for one-offs Jul 12 22:43:14 kergoth: I have see that segfault myself sometimes Jul 12 22:43:27 kergoth: yeah Jul 12 22:43:36 busbox would be prime candidate Jul 12 22:45:29 busbox? Jul 12 22:59:02 busybox **** ENDING LOGGING AT Thu Jul 13 03:00:02 2017