**** BEGIN LOGGING AT Wed May 22 03:02:00 2019 May 22 04:26:58 Morning! JaMa: nice! May 22 07:13:49 Morning! May 22 07:21:36 I'm quite busy this week, did I miss something about qemux86 ? May 22 07:26:16 Tofe: Not much, JaMa switched the compression ;) May 22 07:26:27 Will be quite a bit quicker to compress now ;) May 22 07:26:50 Using pigz May 22 07:27:30 Some time results here: https://github.com/webOS-ports/meta-webos-ports/commit/4c8ec6f8c317ecea7bf9e7b44ccf71328aebec74 May 22 07:28:01 So quite a difference :) May 22 07:29:22 wow, impressive May 22 07:29:30 Yes ;) May 22 07:29:38 On my quick builder it makes quite a difference ;) May 22 07:29:47 For the rest mainly costmetics here and there May 22 07:32:33 Is pigz available commonly on Windows too ? May 22 07:34:54 or maybe it's just gz format, but parallel? May 22 07:36:20 Seems to be exactly the same format. Good ! May 22 07:45:01 Yes, just a lot quicker ;) May 22 08:25:25 Tofe: I was looking into oFono a bit and it seems we could drop 1 patch being https://github.com/webOS-ports/meta-webos-ports/blob/warrior/meta-luneos/recipes-connectivity/ofono/ofono/0001-Enable-the-various-modem-plugins-we-support-again.patch by passing in an additional config --disable-rilmodem as per https://git.merproject.org/mer-core/ofono/blob/master/ofono/configure.ac#L175 I would May 22 08:25:26 say? May 22 08:32:24 let me see May 22 08:36:33 looks like it would work May 22 08:40:18 Tofe: make sure to return make update to refresh bblayers.conf in your local build May 22 08:40:31 there is new meta-luneos-backports-2.8 layer now May 22 08:40:55 JaMa: ok thanks, I'll be careful May 22 08:40:58 Tofe: Well the only device I could test it on is mone with a modem I guess.. Which is a TP4G May 22 08:41:31 JaMa: what is "2.8" for ? May 22 08:41:46 Tofe: 2.8 = Zeus May 22 08:41:56 So things that are updated in Zeus Yocto release already May 22 08:42:02 ok! May 22 08:42:38 https://layers.openembedded.org/layerindex/branch/warrior/layer/meta-luneos-backports-2-8/ May 22 08:43:29 It's quite a lot of things in there already May 22 08:44:40 ah no wait, the description seems to point to meta-webos-ports May 22 08:45:17 ah no, double wait, I didn't look carefully enough. May 22 10:15:56 JaMa/Tofe: Trying to bump oFono a bit with latest from Mer, however running into the following: https://bpaste.net/show/fb8779ba9127 May 22 10:16:25 I created a libglibutil recipe, it builds, I added libglibutil to DEPENDS for oFono, but it seems it doesn't find the header still. What am I doing wrong? May 22 10:17:33 libglibutil contain the required gutil_log.h as per https://git.merproject.org/mer-core/libglibutil/blob/master/include/gutil_log.h May 22 10:31:16 Herrie: for oFono, is it the only compilation failure?... because it's only the tests May 22 10:32:09 Tofe: Yeah well it seems so. I tried to disable the tests, but that doesn't seem to work May 22 10:35:55 ah, right, it's inconditionnally built May 22 10:36:54 you can check in sysroot where the header actually is; it's possible it's installed in usr/include/glibutil and pkgconfig doesn't add "glibutil" to the include paths May 22 10:39:53 Tofe: Ah seems I only have the .so not the headers in my ipk? May 22 10:40:42 This is my recipe for libglibutils https://bpaste.net/show/7b9116175791 May 22 10:40:58 Could be I forgot something trivial, like you know C(++) isn't my thing :P May 22 10:41:05 So it's mainly trial & error May 22 10:42:21 Herrie: try adding FILES_${PN}-dev += "${includedir}" or something like that May 22 10:42:52 but the header should land in the -dev package, not the main package May 22 10:43:16 dev package is empty it seems May 22 10:43:47 ok, that's probably the issue May 22 10:44:18 do you have the header in the image/ folder? and the .pc file ? May 22 10:46:36 In libglibutil (tmp-glibc/work/.../libglibutil/.../image right? May 22 10:47:39 ah, do make install-dev instead of make install May 22 10:48:15 it should solve everything, with a bit of luck May 22 10:48:20 Let's see :) May 22 10:49:42 Enabled GCC8 to try to fix the kernels, so now need to build toolchain as well, so will take a few more mins :P May 22 11:25:43 I have something in /usr/include/gutil now, but oFono still unhappy with the same error message May 22 11:25:52 So I guess it's looking in the wrong place somehow? May 22 11:33:45 Herrie: JaMa: this https://git.merproject.org/mer-core/libglibutil/blob/master/libglibutil.pc.in#L3 looks like it won't work in a sysroot prefixed build, right? May 22 11:34:55 this part should be fine May 22 11:34:59 Herrie: can you check the gcc command line that is executed ? It should have a "-I/usr/include/gutil" somewhere May 22 11:36:46 JaMa: ok thanks, I didn't remember exactly if pkgconfig itself could prefix its paths May 22 11:37:59 Tofe: yes pkgconf should do it, but it's possible that ofono e.g. calls hosts pkg-config May 22 11:41:57 In my do_configure doesn't mention any libglibutil: https://bpaste.net/show/0d5c86a6cd65 May 22 11:43:53 damn May 22 11:44:26 When I check ofono/.../recipe-sysroot/usr/include I have a gutil in there with the file May 22 11:44:30 Herrie: https://git.merproject.org/mer-core/ofono/blob/master/ofono/configure.ac#L189 May 22 11:45:14 Ah it seems that they only check for the package when you configured enable_sailfish_rilmodem May 22 11:45:16 Herrie: do you have either --enable-sailfish-pushforwarder or --enable-sailfish-rilmodem in the recipe? May 22 11:45:25 Which is wrong I think ;) May 22 11:45:58 Or they shouldn't the test either without these ;) May 22 11:46:09 + build May 22 11:46:35 or you can just add PACKAGECONFIG for test and leave that disabled :) May 22 11:49:51 JaMa: The root cause seems to be https://git.merproject.org/Herrie/ofono/blob/master/ofono/Makefile.am#L1112 which is not conditional May 22 11:50:45 While the pkgconfig is May 22 11:51:58 yes, one of the two cases must be fixed May 22 11:52:33 (and then we'll upstream it) May 22 12:05:21 OK trying a few things May 22 12:06:27 They did quite a mess with their SAILFISH flags all over the code May 22 12:06:35 Yeah May 22 12:06:51 I'm not really sure we need their oFono, but I guess it's tweaked for VoiceCall May 22 12:07:11 probably... at some point we should check :) May 22 12:07:22 There was a reason we switched to theirs ;) May 22 12:09:13 Ok getting a step further now ;) May 22 12:09:27 "unit/test-dbus-queue.c:69: error: undefined reference to 'dbus_message_append_args' May 22 12:09:38 Which is probably another missing dependency somewhere May 22 12:10:49 too/too new dbus on our side? May 22 12:10:55 +old May 22 12:11:11 I don't think so, let me see what they did May 22 12:12:57 I suspect https://git.merproject.org/Herrie/ofono/commit/2ec6fc749d8e53dc7f5af8f47c84539dc1b866f3 May 22 12:13:01 Their changes to configure.ac May 22 12:14:17 you sure? it doesn't affect the problematic file here it seems May 22 12:14:43 Let me try, we anyway have recipe for dbus-glib already May 22 12:14:54 Just add it to the depends and update the configure.ac to test ;) May 22 12:16:09 Easy enough to test;) May 22 12:16:35 the .c doesn't include "dbus-message.h", only "dbus-queue.h"; this is already bad practice May 22 12:25:00 Herrie: one more issue detected when upstreaming some of meta-webos-ports recipes, ERROR: geoclue-2.5.2-r0 do_configure: QA Issue: geoclue: configure was passed unrecognised options: docs [unknown-configure-option] May 22 12:25:36 JaMa: Let me check May 22 12:26:34 Seems 2.5.3 is out as well ;) May 22 12:26:40 Let me tackle them both May 22 12:29:58 JaMa: Is it OK to use .tar.gz from GitLab or similar issues as GitHub and better to use SRCREV ? May 22 12:30:57 Herrie: for me I'd say SRCREV May 22 12:31:14 Herrie: I've already fixed it when preparing the commit for meta-oe May 22 12:31:25 please don't change it in meta-webos-ports anymore May 22 12:31:48 JaMa: Link? May 22 12:31:59 I don't know if gitlab suffers from the same, I've just checked that 2.5 really isn't uploaded to https://www.freedesktop.org/software/geoclue/releases/ anymore May 22 12:32:24 Yeah they switched to GitLab now May 22 12:32:27 http://git.openembedded.org/meta-openembedded-contrib/commit/?h=jansa/master May 22 12:32:36 But I guess SRCREV would be safer in the end? May 22 12:33:12 I know they switched to gitlab, but that doesn't imply that they won't be providing developer uploaded tarballs like some projects on github do May 22 12:33:26 depends on how gitlab implements the archives May 22 12:33:35 I guess time will tell :P May 22 12:33:36 if it's generated on demand then yes May 22 12:33:47 We can always swtich in the future May 22 12:34:05 there are 2 recipes with gitlab archives in meta-oe arealdy May 22 12:43:22 JaMa: 2.5.3 update: https://github.com/webOS-ports/meta-webos-ports/commit/9c137cc86223776d8a7219f57ea71f1aef799cd9 May 22 12:50:49 Herrie: it would be great if you could fork meta-openembedded on github and provide such updates there, so I can really just git send-email them May 22 13:15:01 JaMa: OK :) May 22 13:15:42 Seems you picked it already? May 22 13:16:54 Nice to see you also put the pidgin plugins :) May 22 13:16:58 Makes sense :) May 22 13:18:09 Tofe: Their oFono is really a mess May 22 13:18:23 I guess we should try the upstream proper one again sometime May 22 13:19:24 Only real issue might be Voicecall May 22 13:19:36 webos-telephonyd shouldn't be a problem May 22 13:19:48 phonesim & mmsd also not May 22 13:21:48 JaMa: Similar to the pidgin plugins, this one might be useful for Hunspell: https://github.com/webOS-ports/meta-webos-ports/tree/warrior/meta-luneos/recipes-upstreamable/hunspell-dictionaries May 22 13:22:00 Might need some tweaking though to the recipe May 22 13:22:45 Will provide a bunch of dictionaries to Hunspell to make it useful ;) May 22 13:23:16 Herrie: why did you inherit webos_filesystem_paths in it? doesn't seem to be used May 22 13:25:14 JaMa: Seems that can go indeed, just like the ;protocol=git;branch=master in SRC_URI and we need to add PV May 22 13:25:20 You want me to clean that up? May 22 13:27:28 Is https://github.com/wooorm/dictionaries/blob/master/package.json#L4 this an issue? The 0.0.0 ? May 22 13:31:29 already doing that May 22 13:31:53 Herrie: not an issue, but better than default 1.0 May 22 13:39:12 JaMa: Sorry some of these were hacked together and I left them there once they worked... Could use some TLC at some places as you noticed ;) May 22 13:41:03 Herrie: yeah, no problem May 22 13:41:47 now when we're closer to upstream we should better work with it, even if we do the work in upstream it will be less work for us in long run May 22 13:43:07 JaMa: Yes :) May 22 14:45:26 Herrie: ofono PR seems fine to merge? May 22 14:46:13 no .bb change -> merged May 22 14:50:32 Tofe: Yeah just cleanup for now May 22 14:50:36 I think we should try upstream May 22 14:50:44 Mer's one is quite horrible code wise May 22 15:50:13 Tofe: the patches for linux-lg-hammerhead didn't apply cleanly, so I've uploaded all into https://github.com/shr-distribution/linux/tree/hammerhead/4.16/postmarketos-linux-qcom May 22 15:54:04 https://github.com/shr-distribution/meta-smartphone/commit/e3c00976446ad08209c2c5f0591a864443b31671 May 22 17:19:09 JaMa: I think I'll drop 0002-Add-ramconsole.patch anyway, I never got it to work... May 22 17:22:40 4-way Zen3 would be handy now load average: 136.82, 112.85, 98.32 May 22 17:33:00 JaMa: Where do you have the gcc8 patches for 3.18 kernel? Is this in shr Linux repo? May 22 17:36:48 no, in internal LG repo :/ May 22 17:39:38 Anywhere you could put them so I could peek? You can anonimize them to hide target names and other info that's not supposed to be out there. Simple pastes of the changed lines/files would already help. May 22 17:50:13 https://paste.ubuntu.com/p/v64kpYR3fW/ May 22 17:50:55 all except last 2 are backports from vanilla kernel May 22 17:51:33 and it was on heavily patched 3.16 kernel May 22 17:56:12 JaMa: ok May 22 17:56:35 I get some .err asm issues on mdm_common.c and fork.c May 22 17:56:53 Not much more info in the compile log or I don't know how to read it May 22 17:57:01 Anyway to make it more verbose? May 22 17:57:39 yes, that's the same issue, but it gets triggered from pretty much every .o, fork.c is just one of first to fail May 22 17:59:08 see http://lists.openembedded.org/pipermail/openembedded-devel/2018-May/194274.html May 22 18:00:23 you can use -S to compile only and investigate the code, but I'm pretty sure it's the same __put_user as above May 22 18:02:21 this is getting better and better load average: 206.09, 167.89, 151.84 May 22 18:07:27 -S ? I never used that one May 22 18:08:01 I usually use "MACHINE=mako bb -f linux-lg-mako -c compile" for example May 22 18:15:12 gcc -S May 22 18:17:33 And how do I add that easily ? May 22 18:23:48 I didn't say it is easy May 22 18:24:49 you can remove all exit lines from run.do_compile script then source it in shell, then call $CC with the same params as Makefile does, but manually and then add -S (and modify -o to some other file) May 22 18:25:19 you can also use devshell, but I usually do it like above May 22 18:26:12 OK thnx for pointers May 22 18:26:17 to see exact commands used by Makefile, some Makefiles accept V=1, sometimes full command is shown when calling make with -n (and then grepping the failing filenames from the output May 22 19:20:07 Ok May 22 20:14:50 Seems I got 3.4 to build with only 2 small commits :P May 22 20:14:59 No test on device yet though May 22 20:15:37 https://github.com/Herrie82/android_kernel_google_msm/commits/ubp-5.1 May 22 20:15:50 Both found with some Google Fu ;) May 22 20:21:55 cool a year ago it wasn't there yet when I was searching :) May 22 20:27:52 I need to test with others, but seem to fix stuff ;) **** ENDING LOGGING AT Thu May 23 03:00:36 2019