| voidanix[m] | Thaodan: if CONFIG_PM_AUTOSLEEP is disabled then there is no sleep right? is there anything in SFOS that does/can do this in userspace? | 02:52 |
|---|---|---|
| rinigus | voidanix: to my understanding, we rely on autosleep in sfos. Which could be an issue as I don't think it is used anymore in Android (kernel config was disabled at least on Tama) | 05:07 |
| voidanix[m] | it isn't indeed, option was marked deprecated | 05:08 |
| voidanix[m] | was looking at my laptop's and librem5's config as well and PM_AUTOSLEEP is off on both | 05:08 |
| rinigus | For Tama, I emailed it and it works | 05:09 |
| voidanix[m] | wdym you emailed it | 05:11 |
| rinigus | Enabled it :) | 05:17 |
| rinigus | voidanix: ^ | 05:18 |
| voidanix[m] | oh, yea that's expected | 05:19 |
| Thaodan | voidanix[m]: from what I understood is that it is only as deprecated as Android doesn't need it anymore | 06:38 |
| Thaodan | on the desktop it's not needed since sleep is triggered explicitly, I don't know how the librem5 does it | 06:40 |
| rinigus | Thaodan: problem is that Android vendors don't test with it anymore. as a result, we get weird issues with suspend, such as happened to tama. namely, tama was resuming and suspending in very fast loop - something I had to break by ensuring that resume time has some minimum duration | 06:42 |
| T42 | <Verevka86> Does anyone have video recording in the camera based on hybris17.1-18.1? | 13:24 |
| T42 | <edp_17> Even camera doesn't work on my port with android 17.1 base. 😠| 13:30 |
| T42 | <edp_17> The camera app doesn't crash only shows black screen for both picture and video. | 13:31 |
| T42 | <Verevka86> Photo work for me (17.1,18) (re @edp_17: Even camera doesn't ...) | 13:31 |
| Thaodan | Do you have rinigus miniaudiopolicy still running? | 13:31 |
| T42 | <edp_17> @Verevka86 : Did you have to do some magic or it worked out of the box? | 13:32 |
| T42 | <Verevka86> and with and without it with the latest droidmedia, video recording does not work (re @SailfishFreenodeIRCBridgeBot: <Thaodan>Do you have...) | 13:33 |
| T42 | <Verevka86> only symlinks to apex libraries (re @edp_17: @Verevka86 : Did you...) | 13:34 |
| T42 | <edp_17> I have droidmedia 0.20211101.0-1. | 13:34 |
| T42 | <edp_17> What apex libraries have you symlinked? | 13:34 |
| T42 | <Verevka86> https://github.com/SailfishOS-sagit/droid-config-sagit/tree/hybris-17.1/sparse/odm (re @edp_17: What apex libraries ...) | 13:35 |
| T42 | <edp_17> Thanks. | 13:35 |
| T42 | <edp_17> I only have these four in there : libandroidicu.so, libicui18n.so, libicuuc.so and libnativehelper.so | 13:37 |
| T42 | <Verevka86> It's enough (re @edp_17: I only have these fo...) | 13:38 |
| T42 | <Verevka86> does the torch work? | 13:38 |
| T42 | <edp_17> Should I symlink /vendor/lib/egl directory into /odm/lib/ ? | 13:39 |
| T42 | <edp_17> No, torch doesn't work either. When I toggles in Top menu, the icon is showing as it is on / off but led doesn't lit up. | 13:39 |
| Thaodan | rinigus: Did you try to get your changes to the autosuspend delay to aosp? | 13:42 |
| Thaodan | It might make sense to push them to sony aosp if they make sense for us and aosp doesn't use autosleep | 13:42 |
| T42 | <nephros> lbt: fyi, build.sailfishos.org timing out again. | 16:43 |
| Thaodan | @nephos: Something is wrong, webhooks just have been fixed now this. | 16:45 |
| rinigus | Thaodan: the commit I am talking about is https://github.com/sailfishos-sony-tama/android_kernel_sony_msm/commit/65ad0ca39046bb32fa286612ead75692242c35ee . Which is going to impact all SFOS/Sony devices | 16:51 |
| rinigus | No, I haven't submitted it. | 16:51 |
| Thaodan | rinigus: That was the commit I was talking about too | 16:55 |
| rinigus | Thaodan: I see. did you need it for some other device as well? | 16:56 |
| Thaodan | rinigus: I haven't check but as suspend is fixed on kumano now I will see if it helps there too but I think it should help on all devices. | 17:11 |
| rinigus | Thaodan: it would make sense to submit it if it helps many. looking forward to your test. | 17:49 |
| Thaodan | rinigus: Will see, how did you test? | 17:49 |
| Thaodan | I looked at mainline autosleep, not much changes there. | 17:50 |
| rinigus | Thaodan: I used systemdatascope + collectd and checked % time in suspend, duration of single suspend and how many suspend attempts were successful vs failed | 17:52 |
| voidanix[m] | i don't think the kernel should be modified to respect processes like this, it feels like a time hack | 18:46 |
| voidanix[m] | userspace should manage that... | 18:46 |
| rinigus | voidanix: that hack was as kernel is responsible for autosleep as well. but sure, if it can be done in userspace and works better... for example, that 0.25s is just based on my experiments with tama. | 18:48 |
| rinigus | probably something better can be done | 18:48 |
| Thaodan | voidanix[m]: Well autosleep is a a kernel based solution. | 19:49 |
| Thaodan | And improving autosleep is still good. | 19:49 |
| piggz | the smallest bugs take the longest :D | 20:56 |
| piggz | i may just go with option 2, the hacky script :) | 20:56 |
| T42 | <Verevka86> video recording only works in armv7hl hybris17.1, does not work in aarch64 | 22:00 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!