**** BEGIN LOGGING AT Wed Apr 28 02:59:56 2021 Apr 28 08:19:59 Morning! Apr 28 08:39:43 Morning! Apr 28 09:53:57 Tofe: I tried with reverted initramfs patch, still stuck on LOS boot logo without ADB :S Apr 28 09:57:47 Herrie: do you still have the dmesg log? Apr 28 09:58:04 to check that the device is booted in normal mode, not android mode Apr 28 09:58:17 ah but no we saw systemd in the logs, so can't be android Apr 28 10:01:40 Tofe: Well I have pmsg, but that's garbled Apr 28 10:01:47 The dmsg I have seems from another boot Apr 28 10:01:50 4 days ago Apr 28 10:03:50 Herrie: I propose you boot into recovery, modify /data/luneos/lib/system/systemd/android-system.service, comment out the two "ExecStart", and add a new one "ExecStart=/bin/true" Apr 28 10:08:59 Tofe: Ah that's a good idea Apr 28 10:21:30 Tofe: https://paste.ubuntu.com/p/767TBTQ8wr/ Apr 28 10:22:31 a lot of binder stuff Apr 28 10:22:40 if mido 32bit or 64bit ? Apr 28 10:22:43 is* Apr 28 10:23:23 Tofe: Ehm seems I might have forgotten to update udev rules :P Apr 28 10:23:32 Last commit 3 years ago :P Apr 28 10:23:35 That's not helping Apr 28 10:23:40 :) Apr 28 10:23:49 not sure that the real cause though Apr 28 10:23:59 Well it gives funny warnings to start with Apr 28 10:24:19 where is your defconfig, currently? Apr 28 10:29:45 Tofe: https://github.com/Herrie82/android_kernel_xiaomi_msm8953-1/commit/c38574eefa43bbf2abf9218701e5400c930687b6 Apr 28 10:29:51 Let me fix those udev rules first Apr 28 10:29:56 Seems I'm missing quite some including binder :p Apr 28 10:33:48 Tofe: This looks slightly better: https://paste.ubuntu.com/p/3ZJFMC4KH5/ Apr 28 10:33:59 Let me try to run with enabled android-service again Apr 28 10:35:03 Herrie: no, just boot, and in adb do lxc-start -n android Apr 28 10:40:40 Tofe: When I do that it start with the LOS boot logo again and ADB is gone Apr 28 10:42:05 then the problem is still there Apr 28 10:42:18 but I though the halium patches do disable the boot animation ? Apr 28 10:42:24 thought* Apr 28 10:44:02 Unless it doesn't get patched somehow which would be weird Apr 28 10:44:05 Let me check my script Apr 28 11:11:24 can you check that frameworks/native/0004-hybris-Allow-surfaceflinger-to-be-started-from-Sailf.patch is applied ? especially services/surfaceflinger/StartPropertySetThread.cpp line 35 Apr 28 12:18:50 Tofe: It seems all commented out by the patch Apr 28 12:19:15 I only have "property_set(kTimestampProperty, mTimestampPropertyValue ? "1" : "0");" and "return false" remaining Apr 28 12:21:16 I will build Halium once more, just to be sure Apr 28 12:41:32 Seems something is off with the patches being applied correctly Apr 28 12:41:34 Will debug a bit Apr 28 14:13:49 Seems apply-patches.sh isn't always very great in terms of accepting not 100% correctly formatted patches Apr 28 14:13:58 Seems our Mako/Tenderloin patch was acting up a bit Apr 28 14:14:04 After I removed it, other patches got applied Apr 28 14:16:14 So rebuilding Mido's Halium with your ueventd patch Apr 28 14:16:16 Let's see Apr 28 14:54:09 Tofe: Ooh booting UI! Apr 28 14:54:43 Sound, rotation working Apr 28 14:55:08 Wifi not, but I guess I need a defconfig tweak there Apr 28 14:55:16 I remember I disabled it probably Apr 28 14:57:53 Logcat: https://paste.ubuntu.com/p/ZM735pPFwR/ Apr 28 15:07:29 And journal: https://paste.ubuntu.com/p/bMq5wy3yhN/ Apr 28 15:07:35 Let's try to fix wifi Apr 28 15:10:50 And BT Apr 28 15:12:05 Herrie: ooh great Apr 28 15:12:30 so the patches was the right lead after all Apr 28 15:12:47 Tofe: Well not sure I needed the ueventd one in the end Apr 28 15:12:51 I guess I coudl test without it Apr 28 15:12:59 wifi should be easy Apr 28 15:13:17 Just seems that apply-patches.sh was stopping because of an incorrectly formatted patch and didn't continue patching others Apr 28 15:13:22 BT as well Apr 28 15:13:27 Didn't do the bluebinder stuff in defconfig Apr 28 15:13:41 /vendor/bin/gx_fpd you can disable it, it's the fingerprint Apr 28 15:13:51 ah, yes, VHCI is needed Apr 28 15:14:22 for mm-qcamera-daemon you might need a nasty hack, if you want to try having the camera working Apr 28 15:14:39 Tofe: Well I think I patched tissot camera stuff into my device repo for 4.9 Apr 28 15:14:44 So yeah it's anyway hacky Apr 28 15:14:54 Another thing I noticed is that vibrator is not working Apr 28 15:17:04 I don't recall the status on other xiaomi ports Apr 28 15:17:31 because vibrator nyx-hybris doesn't handle hwbinder, iirc Apr 28 15:20:19 Tofe: Well I guess I need this for mido as well: https://github.com/webOS-ports/meta-webos-ports/commit/4eea3eea96421d37b845c75077abb6b2915ecb28 Apr 28 15:29:44 that is, if you have access to the vibrator using /sys APIs Apr 28 15:29:59 but that could be Apr 28 15:34:14 Let me just try Apr 28 15:36:26 Will flash a new image with all changes Apr 28 15:36:32 If everything is behaving better, will PR it Apr 28 15:36:40 Should be in good enough state to migrate as well then Apr 28 15:36:51 If BT, WiFi and Vibration work as well Apr 28 15:36:57 Messwerk behaves nicely Apr 28 15:48:53 Wifi works now, vibration half, BT not yet it seems Apr 28 15:48:57 Any thoughts on how to debug that? Apr 28 15:50:29 Log looks pretty OK for BT I'd say: https://paste.ubuntu.com/p/ZW22bktGwx/ Apr 28 15:50:53 I see these 2 though: Dec 16 18:25:43 mido bluetoothd[2864]: ../bluez-5.55/profiles/network/bnep.c:bnep_init() kernel lacks bnep-protocol support Apr 28 15:51:01 And Dec 16 18:25:43 mido bluetoothd[2864]: ../bluez-5.55/src/plugin.c:plugin_init() System does not support network plugin Apr 28 15:51:06 Not sure that's "normal" or not Apr 28 15:56:03 Herrie: not complaint from bluebinder ? Apr 28 15:57:11 Herrie: I don't see the bluebinder service in your log Apr 28 15:57:24 Tofe: Ehm would help if I add it in mido RDEPENDS :p Apr 28 15:57:29 hehe Apr 28 15:57:48 also, I think the hciattach stuff is now useless, TBC Apr 28 15:58:21 Yeah probably Apr 28 15:58:25 I didn't remove it yet, but should Apr 28 16:00:28 OK removed in local build as well Apr 28 16:00:34 Let's see now Apr 28 16:23:52 Hmmz Apr 28 16:24:00 Still no BT it seems, but Bluebinder works now Apr 28 16:24:19 https://paste.ubuntu.com/p/tvzJ5j2yzv/ Apr 28 16:24:25 And wifi also doesn't work now Apr 28 16:24:26 Weird Apr 28 16:24:27 Rebooting Apr 28 16:28:20 https://paste.ubuntu.com/p/4Qb9Nm9gv7/ Apr 28 16:28:39 Dec 16 18:32:50 mido kernel: VOS ASSERT in vos_start Line 926 Apr 28 16:28:49 Dec 16 18:32:50 mido modprobe[3416]: modprobe: ERROR: could not insert 'wlan': Operation not permitted Apr 28 16:48:13 that's unexpected... Apr 28 16:50:11 Herrie: can you pinpoint vos_start source code in your kernel in the wlan module ? maybe it's a firmware loading issue or something ? Apr 28 16:53:33 It seems that this happened after I updated udev rules :S Apr 28 17:13:58 Tofe: VOSS bits are in here: https://github.com/Herrie82/android_kernel_xiaomi_msm8953-1/tree/halium-9.0-mido-4.9/drivers/staging/prima/CORE/VOSS/src Apr 28 17:24:33 https://github.com/Herrie82/android_kernel_xiaomi_msm8953-1/blob/halium-9.0-mido-4.9/drivers/staging/prima/CORE/VOSS/src/vos_api.c#L926 so here; it seems related to the NVDownload stuff, which could point to a firmware download Apr 28 17:48:23 I'll check logcat Apr 28 17:49:53 Hmmz 12-16 17:30:34.673 102 102 E wcnss_service: source file do not exist /data/misc/wifi/nvbin/msm8953_293_0x00_0x01_0x00_nv.bin Apr 28 17:50:23 12-16 17:30:37.272 35 332 E vendor.qti.bluetooth@1.0-mct_transport: InitTransport: Cannot open /dev/smd3: No such device Apr 28 17:50:52 12-16 17:30:41.362 35 473 E vendor.qti.bluetooth@1.0-async_fd_watcher: ThreadRoutine unable to set SCHED_FIFO for pid 35, tid 473, error Operation not permitted Apr 28 17:51:34 Full logcat for reference: https://paste.ubuntu.com/p/BDTWYQKNWx/ Apr 28 17:56:03 Herrie: what's this /dev/smd3 ? also, verify that /data/misc/wifi/nvbin/msm8953_293_0x00_0x01_0x00_nv.bin does exist, it looks for it for wifi, might be important Apr 28 17:58:12 and also, one other modification you did was to remove the hciattach stuff: could this be related ? Apr 28 18:21:56 Seems I don't have this .bin Apr 28 18:22:08 https://paste.ubuntu.com/p/CMDQDGh4nN/ Apr 28 18:31:24 Also not in my Halium build dir it seems Apr 28 19:37:40 After reflash, wifi works Apr 28 19:37:49 So something funky somewhere Apr 28 19:38:39 ok, well, at least it works after a clean flash Apr 28 19:39:01 BT also works now Apr 28 19:48:35 great ! let's PR that, then Apr 28 19:53:26 On my side I'm trying to upgrade pinephone's kernel to 5.12, but I get into issues with booting the kernel itself Apr 28 19:57:30 Seems I did put hciattach back Apr 28 19:57:42 But that shouldn't do much right? Apr 28 19:58:15 Dec 16 18:25:31 mido hciattach.sh[2594]: could not set property Apr 28 19:58:24 Dec 16 18:25:31 mido hciattach.sh[2588]: /usr/bin/hciattach.sh: line 15: /sys/module/hci_smd/parameters/hcismd_set: No such file or directory Apr 28 20:06:11 let me check, maybe it does several things, one of them being useful Apr 28 20:07:25 it does set several properties related to BT Apr 28 20:07:43 Yes Apr 28 20:07:48 It seems the first one doesn't work Apr 28 20:08:02 setprop ro.qualcomm.bt.hci_transport smd Apr 28 20:08:11 This one also not: echo 1 > /sys/module/hci_smd/parameters/hcismd_set Apr 28 20:08:17 But the others seem to work Apr 28 20:08:26 When I execute them, I see stuff happening in BT app UI Apr 28 20:08:29 rfkill can also be usefull Apr 28 20:09:06 so let's keep it; maybe we could check for [ -e /sys/module/hci_smd/parameters/hcismd_set ] before the echo Apr 28 20:11:53 We might want to check @ sfos what they're doing? Apr 28 20:12:26 not sure, as they don't start android in a lxc container... Apr 28 20:16:09 Yeah don't see much in their systemd bits Apr 28 20:46:24 Tofe: Also seems android_input group is gone now: Dec 16 18:25:16 mido systemd-udevd[1110]: /etc/udev/rules.d/65-android.rules:33 Unknown group 'android_input', ignoring Apr 28 20:49:03 Herrie: you mean we need to add it Apr 28 20:53:57 Tofe: I'm not 100% sure yet, just I noticed this Apr 28 20:55:08 Seems we're missing 1004, 1005, 1009 here: https://github.com/shr-distribution/meta-smartphone/blob/master/meta-android/recipes-core/android-system/android-system_1.0.bb Apr 28 20:55:14 If I check: https://android.googlesource.com/platform/system/core/+/refs/heads/android11-gsi/libcutils/include/private/android_filesystem_config.h Apr 28 20:55:22 But could be they're elsewhere Apr 28 20:56:00 1004 and 1005 should be in here? https://github.com/shr-distribution/meta-smartphone/blob/master/meta-android/recipes-core/base-passwd/base-passwd/change-android-group-ids.patch Apr 28 20:56:43 Herrie: only for the ones that are common with a classic distro Apr 28 20:57:57 1009 is missing Apr 28 20:58:31 android_input should be 1004... Apr 28 20:58:40 Or maybe we should just call it "input" instead of "android_input" ? Apr 28 20:58:51 1004 is in /etc/group Apr 28 20:58:56 yes, it should be "input" Apr 28 20:59:09 where does "android_" come from? Apr 28 20:59:58 https://github.com/shr-distribution/meta-smartphone/blob/master/meta-android/recipes-core/udev/udev-extraconf/65-android.rules#L33 ah ok Apr 28 21:00:23 Testing now Apr 28 21:00:47 looks like a very old mismatch between the patch for /etc/groups and the rules morphis wrote Apr 28 21:01:01 Well even UBPorts still uses this it seems Apr 28 21:01:07 At least in Erfan's repo I found it Apr 28 21:01:50 Yup that did the trick Apr 28 21:02:04 Still have issues with oem_2901, oem_2902 and oem_2905 but well Apr 28 21:02:39 these ones are probably newer groups; and also we probably don't have to care Apr 28 21:03:09 "The range 2900-2999 is reserved for the vendor partition" Apr 28 21:03:26 Yeah, but we might need to Apr 28 21:03:29 I'll check a bit Apr 28 21:04:00 we have vendor_qti_diag and so on Apr 28 21:04:14 just not the right names... Apr 28 21:04:53 Let me grep in my halium tree Apr 28 21:19:53 Tofe: Seems we might need to add these for our MSM8953 targets? https://github.com/LineageOS/android_device_xiaomi_msm8953-common/blob/lineage-16.0/config.fs#L1-L17 Apr 28 23:09:41 Herrie: mmh maybe, I never looked at that kind of fle Apr 28 23:09:43 file* Apr 28 23:20:02 Tofe: Ah seems we simply need to change https://github.com/shr-distribution/meta-smartphone/blob/master/meta-android/recipes-core/android-system/android-system_1.0.bb#L123-L128 to oem_2901 to oem_2906 and errors are gone in my logs Apr 28 23:20:16 Anyway will do some PR in the morning Apr 28 23:20:20 gn8 Apr 28 23:40:37 gn8 **** ENDING LOGGING AT Thu Apr 29 02:59:56 2021