| rinigus | @b100dian: don't think it matters now as we are in permissive, but we should look at it later. With BT, I think the problem is that it somehow tries to suspend and then stops it at the last moment due to wakelock. | 04:17 |
|---|---|---|
| rinigus | That leads to periodic reconnection of bt. | 04:17 |
| rinigus | What I don't like is that we also have android suspend service running. What is that doing, I don't know (on sfos) | 04:19 |
| T42 | <A_T_R> my hybris 17 sfos port was fine booting when i last tested 6 months back ,but now it is not booting sfos version 4.6.0.15 build recently ,i got ssh connection so what to debug now it is stuck at boot logo | 04:40 |
| T42 | <b100dian> rinugus: am playing with persist.vendor.service.bdroid.sibs off (s is for sleep) | 05:15 |
| T42 | <b100dian> rinigus: restarting (killing) bluetooth vendor service after setting that prop *seems* to not anymore warm the phone. | 05:20 |
| T42 | <b100dian> However no sleep either is happening(?) | 05:20 |
| T42 | <b100dian> But I can `echo deep > /sys/power/mem_sleep`, doesn't respond busy | 05:22 |
| rinigus | @b100dian: I haven't tried to echo to mem_sleep, so no idea if its busy :) . as for not getting warm without sleep - phone is rather cool even without suspend and can last for a while that way. looks like its mainly repetitive try-to-suspend/fail/oops-let's-resume cycle that gets it warm | 06:20 |
| T42 | <b100dian> Yes, that's why I'm testing disabling this https://github.com/Micronet-Ltd/android9_vendor/blob/master/qcom/proprietary/bluetooth/hidl_transport/bt/1.0/default/ibs_handler.cpp#L116 | 07:00 |
| rinigus | @b100dian: have you tried to reboot with that set off? instead of killing the vendor service? I wonder what will happen also if we disable android suspend service? don't think its patched on sfos... | 07:08 |
| T42 | <b100dian> Will try - how to do that quickly, overlay the vendor build.prop? | 07:10 |
| T42 | <b100dian> Or I can setprop in a .rc file right? | 07:11 |
| T42 | <b100dian> rinigus: nagara.rc? | 07:14 |
| T42 | <b100dian> Added om boot there. Kets see the next hours connected to the amazfish :) | 07:27 |
| T42 | <edp_17> mal: Looks like I do have the correct hybris version (12.1): https://paste.ubuntu.com/p/zCV8dJQpbh/ | 07:46 |
| T42 | <edp_17> So, the repo init was done correctly | 07:46 |
| rinigus | @b100dian: isn't it set if you setprop? or that's overwritten by .rc? we can overlay anything we want using /overlay subfolders accordingly. that's probably the fastest way to do it | 08:11 |
| T42 | <b100dian> I added "on boot\n setprop .." in nagara.rc and it is set, yes. | 08:15 |
| rinigus | OK, let's see. does it suspend? | 08:30 |
| T42 | <b100dian> Unfortunately no. But there is no failed sleep attempt either.. | 08:41 |
| T42 | <b100dian> # cat /sys/power/wake_lock | 08:43 |
| T42 | <b100dian> dsme_iphb_wakeup mce_display_on | 08:43 |
| T42 | <b100dian> The latter is normal since I'm using on screen terminal | 08:44 |
| T42 | <Mister_Magister> gotta turn off the display mate | 08:44 |
| T42 | <b100dian> @Mister_Magister yeah, hard to debug otherwise while on the go ;) | 08:47 |
| T42 | <b100dian> Ok, added a sleep 10 and no output, so no wakelocks | 08:52 |
| T42 | <b100dian> something is setting /sys/fs/autosleep to off | 08:58 |
| T42 | <b100dian> mce? | 08:59 |
| T42 | <elros34> @edp_17 naive question: do you have cloned correct repo with 12 branch in device/lge/hammerhead? | 09:23 |
| T42 | <edp_17> @elros34 , Yes, I think so. I haven't touched that for years. This is what I got in there: https://paste.ubuntu.com/p/8RVjzCRjg7/ | 09:26 |
| T42 | <edp_17> That's why I am puzzled about that error message. | 09:27 |
| T42 | <elros34> same for vendor/lge and device/qcom/common? | 09:32 |
| rinigus | @b100dian: how do you monitor suspend? I would switch screen off, wait 1-2 minutes and then check with mcetool. (actually, I am using collectd for it as it plots sleep stats too, including failed attempts) | 10:07 |
| T42 | <b100dian> Sysmon plots the same data i guess | 10:08 |
| T42 | <b100dian> I'll look into mcetool | 10:10 |
| Mister_Magister | @b100dian don't bother sysmon is doing the same hting | 10:14 |
| Mister_Magister | sysmon is usling colelctd | 10:14 |
| rinigus | sysmon (if that monitor from openrepos) is not using collectd. collectd is used by systemdatascope | 10:19 |
| rinigus | the main thing is to have screen off :) | 10:20 |
| rinigus | @b100dian: also check if you switch off BT whether it enters suspend. as it could just be that the suspend doesn't work with bt | 10:21 |
| T42 | <b100dian> It is working w/o Bt and it used to _try_ to sleep on Bt but fail. And now with that software IBS disabled it doesnt't even try to sleep | 10:48 |
| T42 | <TheVancedGamer> I have functioning sleep with BT on FP5 fwiw | 11:05 |
| T42 | <TheVancedGamer> I spent about 3 weeks on it :D | 11:05 |
| T42 | <TheVancedGamer> wanted my pebble to stay properly connected and have my phone sleep | 11:05 |
| rinigus | @TheVancedGamer: and, apart from 3 weeks, what was the solution for you? :) | 11:11 |
| mal | @TheVancedGamer was it some kernel change? any luck with the wlan issue we had? | 12:06 |
| T42 | <b100dian> @TheVancedGamer You mean the 5.15 backport here ? https://gitlab.com/ubports/porting/reference-device-ports/android11/fairphone-5/kernel-fairphone-qcm6490/-/commit/7afe6555f34aac8a2300a5c7ef7dfd9cdb6faf26 | 14:42 |
| T42 | <edp_17> mal, @elros34 : I am still struggling to make backports as 10.4 recommends in hadk-faq. The error is "external/backports-bluetooth/Android.mk:62: error: writing to readonly directory: "backports" " : https://paste.ubuntu.com/p/FyTgzmhcXt/ | 14:56 |
| T42 | <edp_17> Please note I outputted some variables at the top between line 27 and 39. I wanted to see whether they are correct. | 14:57 |
| T42 | <elros34> I would start with this "warning: * Using prebuilt kernel binary instead of source " | 15:01 |
| T42 | <elros34> did you clone kernel? | 15:01 |
| T42 | <elros34> and build hybris-hal (kernel)? | 15:07 |
| T42 | <edp_17> Yes, I did clone it from Lineage. Yes I did built it then deleted the whole dev environment and re-created it from local manifest. | 15:17 |
| T42 | <edp_17> Okay, I backout the backport-bluetooth and try to build hybris-hal (kernel) then when it builds, I'll try again. | 15:25 |
| T42 | <elros34> dont forget about patches and kernel configs according to faq | 15:27 |
| T42 | <edp_17> I won't thanks. | 15:29 |
| T42 | <b100dian> everyone here doing bluetooth backports ;) | 15:32 |
| mal | there have been cases where manual changes have been needed, depends on what kind of patches there are in the kernel | 15:44 |
| T42 | <edp_17> @b100dian , Yeah, I just realized both the hammerhead and jfltexx have 3.4 kerbel, so, I thought gave them a shot if I could build sfos 5 for them. | 20:28 |
| T42 | <edp_17> The jfltexx will be challenging because its touch screen is broken and mouse cursor not visible. | 20:30 |
| T42 | <b100dian> I mean it was a coincidence us talking about the same subject but for different kernels (and reasons) | 21:20 |
| *** Mister_Magister_ is now known as Mister_Magister | 21:25 | |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!