**** BEGIN LOGGING AT Wed Jan 05 02:59:56 2022 Jan 05 07:11:31 Morning! Jan 05 07:33:34 morning Jan 05 08:15:47 Morning! Jan 05 08:23:38 Tofe: FWIW seems that mainline status for Tissot (and Mido) is pretty good already, might be worth to have a go at it sometime Jan 05 08:24:23 https://wiki.postmarketos.org/wiki/Xiaomi_Mi_A1_(xiaomi-tissot) Jan 05 11:10:08 JaMa: I'm getting quite a bit of this lately when building different machines: https://bpa.st/IH2A Jan 05 11:13:14 If I do a "bb -c clean libjitterentropy" and rebuild all is OK, but I guess this behavior shouldn't be there? Jan 05 11:58:50 Tofe: FWIW I'm also not getting modems on Hammerhead somehow Jan 05 14:05:45 Herrie: always with libjitterentropy ? Jan 05 14:05:58 JaMa: Yes seems so Jan 05 14:06:30 Mainly when I switch between armv7 and aarch64 it seems Jan 05 14:06:50 At least that's how it seems Jan 05 14:08:05 I have never seen this one locally, will try to reproduce it Jan 05 14:18:42 This is with Honister Jan 05 18:00:02 Morning Jan 05 18:01:46 Tofe: morning! Jan 05 18:08:03 Tofe: If you want I can share my Hammerhead image Jan 05 18:08:21 Herrie: ah, yes, good idea Jan 05 18:09:29 It's on: luneos-dev-package-hammerhead-0-0-20220105.zip Jan 05 18:09:45 BTW with this one I bumped sensorfw as well, so those are broken, but that shouldn't affect oFono etc Jan 05 18:11:20 Tofe: The diff between our current sensorfw and the one in the build: https://github.com/Herrie82/sensorfw/pull/1 Jan 05 18:11:32 Probably something with the hybris bits that need reconfiguring in recipe somehow Jan 05 18:11:37 Just didn't spot it yet Jan 05 18:12:27 Tofe: BTW seems there are 2 PR's pending @ Mer's oFono atm to bring them to 1.26 and 1.27 at least Jan 05 18:44:56 I get similar results on hammerhead too Jan 05 19:04:39 I don't see anything obvious in logcat or journalctl here Jan 05 19:10:27 same here :/ Jan 05 19:12:26 It's probably some plugin or similar that needs explicit loading Jan 05 19:18:32 I don't seem to have /dev/socket/rilX here Jan 05 19:18:45 Not sure we should have that on Hammerhead or it would use binder too ? Jan 05 19:19:40 Ah OK should use binder it seems Jan 05 19:21:09 yes we migrated everything to binder Jan 05 19:21:27 I guess that came with the android 9 migration, though I don't really remember Jan 05 19:22:18 I have an old log from 3 years ago, where we see in the journal: "ofonod[2269]: Using /dev/socket/rild as socket for slot 0" Jan 05 19:27:00 Tofe: Well that was with RIL still Jan 05 19:27:06 With binder we should get different ones Jan 05 19:28:14 maybe it's time to investigate this "Could not find service hosting interface android.hardware.radio@1.0::IRadio/slot1" thing Jan 05 19:29:47 we do have "android.hardware.radio.config@1.0-service" in the .mk file, right ? Jan 05 19:29:50 I'll check Jan 05 19:32:22 mmh I don't see it in there https://github.com/Tofee/android_device_lge_hammerhead/blob/halium-9.0/device.mk Jan 05 19:33:39 we could try adding it and see if it solves our issue Jan 05 19:35:13 Tofe: well I don't have that error in journal for Hammerhead, only for Mido Jan 05 19:36:48 I guess we need this? https://github.com/Herrie82/android_device_xiaomi_tissot-1/blob/halium-9.0/device.mk#L95-L110 Jan 05 19:36:52 Or at least libril? Jan 05 19:37:26 some of that yes, not sure what exactly Jan 05 19:38:02 also not sure when "android.hardware.radio.config@1.0-service" in included in the target product Jan 05 19:38:34 Tofe: Yeah seems for Mido I have this (which obviously doesn't work, seeing the error msg...): https://github.com/Herrie82/android_device_xiaomi_mido/blob/halium-9.0-kernel-4.9-tissot-camera/device.mk#L67 Jan 05 19:45:51 Though it should for mido get some bits from the msm8953-common Jan 05 19:47:28 https://github.com/Herrie82/android_device_xiaomi_msm8953-common/blob/halium-9.0-extended/msm8953.mk#L294-L300 Jan 05 19:54:54 Seems we might still need to use RILD for Hammerhead somehow, not sure Jan 05 19:55:03 The libs in the device tree seem to suggest so Jan 05 19:56:15 https://github.com/Tofee/android_device_lge_hammerhead/blob/halium-9.0/proprietary-blobs.txt#L85 Jan 05 20:05:07 Tofe: Ah nope, when adding the old ril_subscription.conf it stil doesn't work Jan 05 20:52:03 Herrie: but we never had it working ? That would be surprising Jan 05 20:54:55 I kinda recall we had somehow Jan 05 20:54:58 Pretty sure I tested it Jan 05 20:55:02 At least I would get bars Jan 05 20:55:21 I have the /dev/binder /dev/hwbinder and /dev/vndbinder paths Jan 05 20:55:29 But not sure what else I can find based on those Jan 05 20:55:59 I guess we should have learned a bit more how the whole thing works :p Jan 05 20:56:18 Well it's weird that Tissot works and Mido/Hammerhead not Jan 05 20:56:25 Could be some binder defconfig option somehow Jan 05 20:56:57 I have also all the binder sockets Jan 05 20:57:23 Weird thing is that Mer's kernel check script doesn't have any binder values in there Jan 05 20:57:45 I would expect those to be there, no? https://github.com/mer-hybris/mer-kernel-check/blob/master/mer_verify_kernel_config Jan 05 20:58:45 Well they don't have android in a container, so maybe they just don't need hwbinder... not sure really Jan 05 20:58:57 I would think they do use it though Jan 05 21:02:42 Herrie: maybe there's a way of tracing things with /sys/kernel/debug/binder Jan 05 21:05:52 I don't see much readable info in there to be honest Jan 05 21:05:58 Seems like a bunch of references just Jan 05 21:07:16 well from what I can see, ofono doesn't even interact with it Jan 05 21:07:52 I'll cross-check with tissot Jan 05 21:09:28 yes, confirmed Jan 05 21:09:58 in proc/ subfolder we see the pid of the processes which have a binder session Jan 05 21:10:07 on tissot we see the pid of ofono Jan 05 21:10:11 not on hammerhead Jan 05 21:11:22 Tofe: OK Jan 05 21:11:38 Seems there's a flag called --enable-sailfish-debuglog Jan 05 21:11:44 Might provide some more logging? Jan 05 21:12:06 well it's already possible to start manually ofono with more verbose debug Jan 05 21:12:14 and it's quite verbose :p Jan 05 21:12:27 let me just find what is the line to start Jan 05 21:13:24 OFONO_RIL_DEVICE=ril /usr/sbin/ofonod -n --nobacktrace -d --noplugin=hfp_bluez5 Jan 05 21:15:54 Herrie: https://paste2.org/hfV5pUEI Jan 05 21:16:56 so it seems ofono does its job, but the android side doesn't Jan 05 21:18:08 in logcat it's clearly related to "Could not find service hosting interface android.hardware.radio@1.2::IRadio/slot1" Jan 05 21:19:27 Well i don't seem to have that in logcat Jan 05 21:19:30 Let me check that again Jan 05 21:21:54 No mention of RIL or radio in my logcat on Hammerhead Jan 05 21:22:25 Just a few mentions of modem Jan 05 21:23:25 I get similar output for debug oFono though Jan 05 21:24:07 Herrie: to trigger the message in logcat, run logcat, and in another window do a "systemctl restart ofono" Jan 05 21:24:26 ah, sorry, also Jan 05 21:24:31 "logcat -b all" Jan 05 21:28:41 OK I get the same now Jan 05 21:28:48 I guess I needed the -b all for logcat Jan 05 21:33:03 ok, good we have the same thing Jan 05 21:33:28 so it would seem we're missing an RIL piece on android side Jan 05 21:33:36 like we thought before Jan 05 21:35:38 on tissot, I think /vendor/bin/hw/qcrild is taking charge of that Jan 05 21:37:03 Well for Mido seems the switch to qcrild is only made in LOS 18.1: https://github.com/zeelog/OTA/releases Jan 05 21:38:21 https://github.com/zeelog/android_device_xiaomi_mido/commit/bad4f83957a8f083b6e57f9ee5786bee8c5be225 Jan 05 21:41:16 I'd say it mainly adds QMI interface and IWLAN (I don't know what that is), and newer hwbinder radio interface compared to rild Jan 05 21:42:18 Tofe: Well seems for Hammerhead we might need to add something like this to manifest.xml: https://github.com/Herrie82/android_device_xiaomi_msm8953-common/blob/halium-9.0/manifest.xml#L186-L209 Jan 05 21:43:50 I agree, something like that Jan 05 21:43:56 Just question is what kind of values we'd need in there... Jan 05 21:44:21 Yeah, totally a beginner here Jan 05 21:45:03 Ah we already have something: https://github.com/Tofee/android_device_lge_hammerhead/blob/halium-9.0/manifest.xml#L147-L166 Jan 05 21:47:23 ok so maybe we're not that far off Jan 05 21:48:24 So in our device.mk we should make sure we ship those both I guess similar to: https://github.com/Herrie82/android_device_xiaomi_mido/commit/63d45fa6153cc36bf1f6950ea75fc6e260d6d4ad#diff-247ee86229a709a6e2eedfc1d3c4a557825aee073e20b0112ab76f4ca8e4bc4e ? Jan 05 21:50:32 yes Jan 05 21:59:44 Seems we might be missing a few more from that XML, I'll cross check them Jan 05 22:03:58 I don't think we need them all though Jan 05 22:07:06 Seems most are there Jan 05 22:08:44 the "PRODUCT_PACKAGES += ..." idea is probably our best lead Jan 05 22:12:32 Yeah I guess so Jan 05 22:12:38 Just not really sure what to add Jan 05 22:12:45 There's -impl and -service suffix Jan 05 22:13:18 Do we need libril, libqcril ? Jan 05 22:15:27 I guess trial & error a bit Jan 05 22:19:16 Should have a new system.img shortly hopefully Jan 05 22:20:34 I'd put -impl, -service, and libril, for a start Jan 05 22:21:32 no need to flash anything in a first time, just check if you get "android.hardware.radio*" files under system/ or under vendor/ Jan 05 22:22:56 anyway I'll go off to bed soon Jan 05 22:23:30 Same here ;) Jan 05 22:23:44 Just wanted to try to build something Jan 05 22:33:57 Gn8 Jan 05 22:50:19 Seems there are some files in there now Jan 05 22:50:23 Let me test it shortly Jan 05 23:05:14 New log for tomorrow: https://paste.ofcode.org/M2MCpAMpKyk669qVtjCjrS **** ENDING LOGGING AT Thu Jan 06 02:59:57 2022