**** BEGIN LOGGING AT Wed Nov 04 02:59:57 2020 Nov 04 07:36:07 Morning! Nov 04 07:40:37 Morning! Nov 04 08:00:03 Morning! Nov 04 08:47:52 Tofe: Seems some new things are popping up on OSE again ;) https://github.com/webosose/com.webos.service.mediaindexer Nov 04 10:05:39 morning Nov 04 10:24:18 howdy, how goes the testing? Nov 04 10:24:20 :) Nov 04 11:04:03 16-bb-threads didn't fit in tmpfs (will increase and re-run later) Nov 04 11:04:18 8-bb-threads: Nov 04 11:04:20 60.83user 12.42system 7:58:20elapsed 0%CPU (0avgtext+0avgdata 28924maxresident)k Nov 04 11:04:23 152inputs+136outputs (1major+13722minor)pagefaults 0swaps Nov 04 11:04:42 now running 2-bb-threads, it will be a while to finish, at least 7 more hours Nov 04 11:05:10 it's significantly faster, but still far from raven on HDD Nov 04 11:10:12 :) Nov 04 11:10:38 hopefully its better enough for the project. Nov 04 11:11:07 eventually you might not need your own server but can use the pool for being able to build multiple images simultaneously. Nov 04 11:12:25 I think it's good enough for LuneOS, me and herie can build images locally much faster and for some "release builds" we really don't need to wait for them Nov 04 11:12:45 if it takes few days to build them all, then it's completely fine Nov 04 11:14:40 but if nobody else is using this server now, then it would be interesting at least as a test, to run this benchmark in dom0 to see the HVM overhead Nov 04 11:16:37 I guess you have some OOB access, maybe even booting ubuntu in ramdisk and running the benchmark there, but don't know how difficult it would be to arrange for you Nov 04 11:29:59 Tofe: https://github.com/mer-hybris/qt5-qpa-hwcomposer-plugin/pull/80 :D Nov 04 11:53:51 that merge was quick :D Nov 04 12:12:25 :) Nov 04 12:12:51 the tissot kernel issue with yylex is caused by ubuntu 20.10 with gcc-10 without -fcommon Nov 04 12:46:13 JaMa, I'll talk to my cohort about running in the Dom0 Nov 04 12:47:13 Tofe: Well it's done, so means we can drop a ton of patches at least there Nov 04 13:23:40 Herrie: yes, that's good news Nov 04 14:36:05 not sure how to debug tissot kernel modules issues, unlike on pinephone the modules don't have neither .plt nor .text.ftrace_tramplines sections https://paste.ubuntu.com/p/rC3NxGRXQ6/ Nov 04 14:38:34 JaMa: you mean it's identical? Nov 04 14:39:08 not completely identical, but no section has executable bit set (which was the reason for pinephone kernel rejecting to load them) Nov 04 14:39:42 this is readelf -S diff from /lib/modules/3.18.31/kernel/drivers/staging/prima/wlan.ko between dunfell and gatesgarth build Nov 04 14:42:53 ok Nov 04 14:42:58 or more precisely there are 4 sections with AX flags but in both dunfell and gatesgarth (.text, .text.unlikely, .exit.text, .init.text) Nov 04 14:43:15 I haven't looked into dmesg yet, though, maybe there's somethin Nov 04 14:45:26 ah: "[ 35.275291] module wlan: overflow in relocation type 261 val ffffffbffc173c14" Nov 04 14:46:25 one of those I think: https://github.com/Tofee/tissot/blob/tissot-o-oss/arch/arm64/kernel/module.c#L205 Nov 04 14:46:26 other modules do work? Nov 04 14:47:18 nope, none work Nov 04 14:47:19 well we build only one more overlay Nov 04 14:47:45 and memnotify Nov 04 14:48:04 though those two don't raise any error in dmesg Nov 04 14:48:52 ah, right memnotify is built by separate recipe, I was looking only in kernel WORKDIR Nov 04 14:49:12 don't raise any error, but also show error in insmod? Nov 04 14:49:22 yes Nov 04 14:49:22 any error in dmesg Nov 04 14:51:28 b24a557527f97ad88619d5bd4c8017c635056d69 from upstream does apply cleanly, but no idea if it helps :) Nov 04 14:54:23 probably worth a try; but I don't have a tissot build nearby, so it'll take time to check Nov 04 14:54:39 f930896967fa3 arm64: module: avoid undefined shift behavior in reloc_data() Nov 04 14:54:49 is more likely related Nov 04 14:54:57 as found in http://lkml.iu.edu/hypermail/linux/kernel/1605.3/04962.html Nov 04 14:58:29 I'm building it now, can share the image Nov 04 14:58:45 do you want just those modules? Nov 04 15:01:07 Tofe: https://drive.google.com/file/d/1u3hXXO1japFANq8L2eD3gyoPt7fE6JzI/view?usp=sharing wlan.ko Nov 04 15:01:16 I'm building it too, just the kernel should be fast enough Nov 04 15:01:34 with both patches backported? Nov 04 15:01:38 not yet Nov 04 15:02:53 https://drive.google.com/file/d/1-BFReHN-62kH4Ot7dW8TrpNlpq_1AOpP/view?usp=sharing overlay.ko Nov 04 15:05:55 same overflow message for wlan.ko in dmesg Nov 04 15:06:42 insmod: ERROR: could not insert module /tmp/overlay.ko: Invalid module format Nov 04 15:06:47 so, same Nov 04 15:06:58 hmm :/ Nov 04 15:07:34 I guess I should buy wife new phone and try it here as well :) Nov 04 15:09:39 I was planing to buy her Mi A4 as it's good value, but as it's canceled, I'll need to search somewhere else Nov 04 15:47:56 JaMa: I just realized, the fixes probably affect the kernel itself, not the modules Nov 04 16:07:08 should I share the kernel image or patch to build it on your side? Nov 04 16:09:26 https://github.com/shr-distribution/meta-smartphone/commit/b398834fc5b4392a1e322b5b61358b3ac11155ac Nov 04 16:09:52 I'll build it on my side, thanks Nov 04 16:47:03 JaMa: yes, it looks like it works now Nov 04 16:47:19 tissot doesn't boot, but maybe that's something else Nov 04 16:57:25 lxc isn't installed on tissot's image ? It's a bit surprising Nov 04 17:37:54 I know who broke it Nov 04 17:38:04 let him fix that :) Nov 04 17:43:37 should be fixed in meta-webos-ports now Nov 04 17:51:20 :) ok Nov 04 19:47:05 JaMa, how goes the testing on Bonaire? Nov 04 19:47:06 :) Nov 05 00:14:10 73.87user 15.30system 12:04:32elapsed 0%CPU (0avgtext+0avgdata 28516maxresident)k Nov 05 00:14:13 8inputs+136outputs (0major+13756minor)pagefaults 0swaps Nov 05 00:14:34 for 7-build-2-bb-threads finished recently, now restarting the failed build **** ENDING LOGGING AT Thu Nov 05 02:59:57 2020