**** BEGIN LOGGING AT Mon Aug 26 02:59:58 2013 Aug 26 08:07:33 Tofe: morning Aug 26 08:07:44 morphis: morning! Aug 26 08:07:53 morphis: I'm recompiling the kernel with dylan right now Aug 26 08:08:01 ah Aug 26 08:08:18 just to be sure that it is indeed the kernel that is problematic Aug 26 08:08:28 Tofe: see the last two commits of https://github.com/webOS-ports/luna-next/commits/morphis/work Aug 26 08:08:37 Tofe: ok, thats a good idea Aug 26 08:14:27 WebAppMgr is in sbin ? Aug 26 08:14:34 I thought it was bin Aug 26 08:15:27 it moved with the use of cmake-modules-webos Aug 26 08:15:33 we might want to revisit this later Aug 26 08:15:59 well, that's fine for me, it can make sense to be there Aug 26 08:19:25 for the rest, it looks fine, I just don't see where the management of windowHidden is done Aug 26 08:21:59 morphis: I've pushed llvm fix and newer oe-core, but needs newer meta-webos-ports which is only in master-next now Aug 26 08:22:16 morphis: will push it to master after another build is finished Aug 26 08:29:24 Tofe: thats one case I didn't looked at yet Aug 26 08:29:37 ok Aug 26 08:29:59 Tofe: I had a little bit trouble to understand the QML part as I didn't saw when I get a real CompositorWindow and when just the container Aug 26 08:30:07 JaMa: fine for me Aug 26 08:32:03 morphis: yes, the naming if not really perfect yet, sorry for that... I'll improve that. Aug 26 08:33:15 Tofe: no problem Aug 26 08:33:31 but yes, we have to improve it add some more documentation to make this easier understandable Aug 26 09:10:09 morphis: it reboots with the dylan kernel too, so it must be something else Aug 26 09:11:04 Tofe: interesting Aug 26 09:23:04 I guess "exec switch_root /rfs /sbin/init" does work with init being a symbolic link ? Aug 26 09:24:24 Tofe: yes Aug 26 09:24:45 ok, well, so it's not that. Aug 26 09:25:27 I'll insert a "touch" just before, just in case. Aug 26 09:26:37 Tofe: we can add adb to the initramfs Aug 26 09:27:03 and check for a specific kernel command line argument to start Aug 26 09:27:04 it Aug 26 09:27:38 I'll try the simple ways first Aug 26 09:28:40 ok Aug 26 09:40:52 oh, this time it did boot !?... Aug 26 09:42:16 I just inserted an "echo toto >> /rfs/titi" in the chroot... mmmh... Aug 26 09:43:08 but the system has a strange behavior Aug 26 09:43:31 chris@Ubuntu-OE:/media/Volatile/OpenWebOS/webos-ports-env/webos-ports$ adb shell Aug 26 09:43:31 / # ls Aug 26 09:43:31 Illegal instruction Aug 26 09:46:32 interesting Aug 26 09:46:38 can you get dmesg output? Aug 26 09:46:58 Tofe: btw. http://bpaste.net/show/f7vyLsyYmmpd3odKF0IM/ Aug 26 09:53:38 that could help, but it looks like the things are quite corrupted, either in memory, or on the fs Aug 26 09:55:08 Tofe: it's a tegra2 in the a500, right? Aug 26 09:55:13 right Aug 26 09:55:41 so no neon, and cortex a9 Aug 26 09:56:04 you checked wether those settings are still correct with the move to master? Aug 26 09:56:25 as "Illegal instruction" really suspects something like this Aug 26 09:57:07 right, that's a good hint Aug 26 09:57:48 Tofe: you should see the basic ones already in the bitbake output: TUNE_FEATURES = "armv7a vfp thumb neon cortexa8" Aug 26 10:02:30 TUNE_FEATURES = "armv7a vfp cortexa9" Aug 26 10:03:13 hm Aug 26 10:03:24 JaMa: any idea? Aug 26 10:03:52 maybe it has become a default for cortexa9 in gcc 4.8, I'll check Aug 26 10:04:32 you mean neon? Aug 26 10:04:38 yes Aug 26 10:07:12 Tofe: you could try to chec with objdump -d wether the build binaries contain any neon instruction Aug 26 10:07:24 good idea Aug 26 10:07:34 like Aug 26 10:07:34 arm-linux-gnueabihf-objdump -d tmp-eglibc/work/maguro-webos-linux-gnueabi/webos-ports-dev-image/1.0-r0/rootfs/usr/sbin/WebAppMgr | grep vmov Aug 26 10:07:35 effc: ec410b30 vmov d16, r0, r Aug 26 10:07:52 Tofe: you just need to find the right binary to check :) Aug 26 10:09:39 vmov is neon only ? Aug 26 10:09:57 I do have it in my WebAppMgr Aug 26 10:10:14 vmov d7, r0, r1 Aug 26 10:10:59 oh my Aug 26 10:11:18 I do have a commit like "qt4-webos: add no-neon option for a500 machine" Aug 26 10:11:24 but we are now qt5... Aug 26 10:13:03 what is the equivalent of qt4-webos for qt5 ? Aug 26 10:13:51 :) Aug 26 10:13:54 qtbase Aug 26 10:14:38 and some more modules Aug 26 10:16:10 morphis: there's no central place to add just a "-no-neon" option to the equivalent of QT5_MACHINE_CONFIG_FLAGS ? Aug 26 10:17:00 there is qt5.inc but JaMa should know the position place to put it Aug 26 10:17:34 Tofe: but generally there QT_CONFIG_FLAGS in qtbase.inc you can append to Aug 26 10:17:50 ok, yes I saw this one, I'll try it there Aug 26 10:18:48 Tofe: but you should check any further parts too if even a ls causes a Illegal Instruction warning Aug 26 10:19:38 well, it's ls through adb, knowing that luna-next has been started... I think the whole system was already corrupted Aug 26 10:20:35 adb itself had some difficulties to connect to the tablet, so busybox was already a bit out of service Aug 26 10:20:48 but yes, that's still strange. Aug 26 10:23:23 ok Aug 26 10:30:18 plenty of vmov in busybox binary Aug 26 10:31:29 but it could be a non-neon instruction Aug 26 10:37:52 Tofe: yes QT_CONFIG_FLAGS is best place for it, only problem is that it should be disabled or enabled for all machines with given TUNE_PKGARCH Aug 26 10:38:28 Tofe: so for local build you can use some override, but for upstreamable change I need to think a bit more about it :) Aug 26 10:38:30 JaMa: yes, I'll first test if it helps, and them I'll try to find a good way to selectively add that option for a500 Aug 26 10:47:46 Tofe: yes vmov is a neon instruction Aug 26 10:48:51 morphis: but it could also be a simple vfp instruction, isn't it ? Aug 26 10:49:11 at least the ARM doc seemed to say so Aug 26 10:49:34 Tofe: http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dui0489c/Cjahjdeb.html Aug 26 10:49:43 VMOV (between an ARM register and a NEON scalar) Aug 26 10:49:43 Transfer contents between an ARM register and a NEON scalar. Aug 26 10:50:44 yes, when it is used with a neon scalar, otherwise it's fine... did I miss a point ? Aug 26 10:53:54 ah I missed the chapter title Aug 26 10:54:06 "Instructions shared by NEON and VFP" Aug 26 10:54:34 then you should search for another pure neon instruction Aug 26 10:59:04 yes, I tried some, but didn't find any in WAM nor busybox yet Aug 26 11:05:44 ok Aug 26 11:06:01 we currently have some stability issues with qtwebkit Aug 26 11:50:53 ah, great, lsm-common doesn't configure now... it doesn't find /usr/bin/qt5/uic... Aug 26 12:10:01 hmm. stability issues in qtwebkit. this doesn't sound abnormal Aug 26 12:17:26 EricBlade: :) Aug 26 12:17:42 Tofe: I have a patch for that Aug 26 12:18:27 Tofe: I hoped JaMa would have applied it already to upstrea meta-qt5 Aug 26 12:18:43 Tofe: otherwise take from here: http://bpaste.net/show/OR1zI7jNcCfhgxbCtuN2/ Aug 26 12:29:53 cmake one? already applied and layers.txt updated Aug 26 12:38:17 JaMa: ok, great Aug 26 12:39:59 ok, just have to update then, thanks Aug 26 12:41:04 * JaMa will push needed meta-webos-ports update in a minute Aug 26 12:41:13 NOTE: Running task 5325 of 5652 Aug 26 12:41:39 unfortunatelly SRCREV changes were postponed so not yet in :/ Aug 26 13:08:06 JaMa: yeah saw that Aug 26 13:09:55 morphis: btw is SRCREV bump missing in luna-webappmanager? Aug 26 13:10:17 morphis: looking on git diff master master-next I see LIC_FILES_CHKSUM update but the same SRCREV Aug 26 13:10:39 -SRCREV = "14c06fb2fb1a26236a5c71e1c6f493d0e5de681f" Aug 26 13:10:39 +SRCREV = "5203bb70978db27bacec4dc906941221f7b1e69e" Aug 26 13:10:45 ^ is in morphis/work Aug 26 13:14:11 No recipes available for: ..../meta-webos-ports/recipes-support/curl/curl_7.31.0.bbappend ? Aug 26 13:14:49 Tofe: that's why I have to push updated meta-webos-ports soon Aug 26 13:14:55 :) ok Aug 26 13:23:01 JaMa: let me bump some of the components again than you can cherry pick them all together Aug 26 13:23:16 wait a sec if you can Aug 26 13:23:29 I'll push updated master for you to base it on it Aug 26 13:23:36 JaMa: hm, I forgot to push an updated morphis/work Aug 26 13:23:41 it doesn't have any further bumps Aug 26 13:23:49 so master should be fine Aug 26 13:26:20 pushed Aug 26 13:28:40 JaMa: hm, master should have all needed things to build fine (even for the updates LIC_FILES_CHKSUM of luna-webappmanager) Aug 26 13:32:27 luna-next was failing here with: Aug 26 13:32:28 | CMake Error at /OE/webos/owpb/webos-ports/tmp-eglibc/sysroots/x86_64-linux/usr/share/cmake-2.8/Modules/webOS/webOS.cmake:158 (message): Aug 26 13:32:31 | Builds must be done "out-of-source". Aug 26 13:33:15 ah, I know this one :) Aug 26 13:33:29 and luna-webappmanager was: Aug 26 13:33:30 ERROR: luna-webappmanager: md5 data is not matching for file://COPYING;md5=d32239bcb673463ab874e80d47fae504 Aug 26 13:33:33 ERROR: luna-webappmanager: The new md5 checksum is 0f93d2cf04b94ac3f04a789a1fb11ead Aug 26 13:33:51 JaMa: let me check and correct this Aug 26 13:34:11 well this checksum is also in SRCREV 5203bb70978db27bacec4dc906941221f7b1e69e Aug 26 13:45:01 JaMa: pushed two more commits to morphis/work which should update luna-next/luna-webappmanager to build fine Aug 26 13:45:10 not tested yet so they are in morphis/work Aug 26 13:47:12 morphis: OK, building with top 2 cherry-picked Aug 26 13:48:01 ok Aug 26 14:49:20 morphis: with these 2 cherry-picked I still get: Aug 26 14:49:21 | CMake Error at /OE/webos/owpb/webos-ports/tmp-eglibc/sysroots/x86_64-linux/usr/share/cmake-2.8/Modules/webOS/webOS.cmake:398 (message): Aug 26 14:49:24 | Component version from build system () != configured version (0.1.0) Aug 26 14:50:01 JaMa: narf. I forgot the webos specific version number Aug 26 14:50:13 yes it's only in luna-webappmanager Aug 26 14:50:30 one minute Aug 26 14:51:02 morphis: btw when updating SRCREVs, should I take care of parsing these values from PV? Aug 26 14:51:13 JaMa: pushed updated morphis/work Aug 26 14:51:47 JaMa: you mean updating SRCREVs of upstream components? Aug 26 14:51:49 morphis: it should probably work just fine if you just set PV and inherit webos_enhanced_submissions Aug 26 14:52:09 no I meant rebasing meta-webos-ports on meta-webos with new SRCREV stuff Aug 26 14:52:35 that would be good Aug 26 14:53:02 ok I'll keep that in mind when testing it Aug 26 14:53:36 I was thinking about rebasing our meta-webos-ports already on that srcrev-in-p_v pull request from github, but I guess i can wait for another week for final version Aug 26 14:54:40 JaMa: ok Aug 26 15:10:50 mmh I did rebased nearly two hours ago, did I rush in too much ? Aug 26 15:28:30 Tofe: depends Aug 26 18:23:07 morphis: I got the error with "LIC_FILES_CHKSUM does not match," Aug 26 18:23:42 Tofe: you need to cherry-pick to changes for bumping luna-next/luna-webappmaanger from my morphis/work branch Aug 26 18:23:49 oki Aug 26 18:43:49 morphis: hm I took the two last commits, which seemed to be the right ones, but I still get the error Aug 26 18:44:23 about md5 checksums Aug 26 18:44:38 I did cleansstate both recipes, but still it didn't help Aug 26 18:48:52 Tofe: for which component? Aug 26 18:51:04 both, luna-next and luna-webappmanager Aug 26 18:51:34 can you give me the build output? Aug 26 18:52:02 http://pastebin.com/s27izrUR Aug 26 18:53:54 narf Aug 26 18:54:02 I wrote md4 instead of md5 in the recipe Aug 26 18:54:07 :) Aug 26 18:54:31 sorry, I should really stop working on anything until I got my master thesis out Aug 26 18:54:52 :p I understand Aug 26 18:54:55 pushed updated morphis/work Aug 26 18:55:03 ok ! thx Aug 26 18:55:06 next week everything will be better :) Aug 26 19:12:17 Tofe: does it work now better for you? Aug 26 19:25:30 morphis: I had the same in my build (with LIC_FILES_CHKSUM) so thanks++ Aug 26 19:25:54 JaMa, Tofe: sorry for this :( Aug 26 19:30:11 no problem, it was my fault for pushing new revs too soon Aug 26 19:31:42 :) Aug 26 21:07:20 Tofe: you came further with your neon issue? Aug 26 21:10:48 oh, not yet, I had to make a pause this evening to bring a bed through the streets of paris Aug 26 21:11:04 well, anyway, nevermind :) Aug 26 23:05:00 JaMa, Tofe|Away: finally everything in master should build now Aug 27 00:05:53 * dkirker stumbles around, half awake **** ENDING LOGGING AT Tue Aug 27 02:59:58 2013