Android Apps Using System Webview Crash

Jolla Phone 2026. I am noticing that anything that relies on Android’s system webview (e.g. Aurora non-microG sign-in, some apps) are VERY prone to crash. Seems maybe to involve typing in text boxes. I can reproduce this pretty much 100% of the time with Aurora Store when the microG sign-in option is disabled. I don’t see a thing in system journal or user journal. Maybe AAS has its own logging?

OK, going deeper into this, trying to make some kind of reproduction. So far no success. However, Privacy Browser from F-Droid uses the system webview. Turn on JavaScript, DOM storage or whatever else, basically everything needed to make the modern web work. Then go to Wikipedia, Youtube, or any JS-heavy site, and it will crash… at least on my phone.

Alright, after quite a bit of iteration, I have managed to reproduce this more or less on demand (at least on my phone). Attached to this post is a python script that, when run as a web server and visited in a browser, visits Wikpedia in an iframe and loads new pages every X seconds. It can also bounce up and down on scroll. One and/or both of these will eventually crash the system webview. Sometimes it takes under 5 seconds, other times it can take a few minutes.

I have removed FUTO, Patch Manager, wiped AppSupport once to start it fresh, and under all circumstances the crash happens. It would be very useful if some other people can run this script using a system webview browser from F-Droid.

I used Privacy Browser. Make sure JavaScript, Cookies, DOM storage are on. Probably only JS matters, but might as well turn everything on.

Then you can run the server from your phone (or anywhere the phone can reach, I guess), visit http://127.0.0.1:8100/. The browser should crash somewhere between seconds and minutes. There’s also some querystring parameters you can give it to change exactly what the driver does. But just visiting the URL alone should be sufficient. I just ran it and it crashed in like 3 seconds max.

wv-repro.targz (4.6 KB)


Works fine here

Thanks for the test. How long did you run it for? And also Jolla Phone 2026?

Edit: see 205 seconds lol.

Yeah J2 5.2.0.17 (twenty chars)

How much RAM, storage space? I am dredging the bottom of the barrel to find any kind of differentiator between my phone and your phone, basically….

12/256, more likely culprit is some ‘dirty hack’/tweak, did you use any of those (like the ones trying to fix pin detection/store installation source in android apps?). You could try to uninstall apps (and packages installed from cli, Install history from chum will help listing those) one by one, but if some AIslop is behind it, who knows if it will uninstall cleanly, but might be worth a try before full reflash

Have installed a few patches, but currently removed, and it was only the ones that affected the status bar. Never installed the pin lock thing. But yes I am basically digging through the filesystem now to see if there is some library or dep that got messed up by another app installation.

Did you try the full reinstall of AAS, just deleting in store and installing again leaves a lot of stuff 4.5 android app support restart itself in a loop - #2 by tuplasuhveli

1 Like

I actually nuked (well, renamed) the entire installation of AppSupport, restarted it and let it set up a new one, and still had the same issue. So I didn’t do the exact clean way, but it was a clean startup of AAS fresh, insofar as I could tell.

The brutal way will also remove all android apps, so make backups if you go that way, but should still save time vs full reflash

Removed it the brutal way, all 3 directories gone. The crash remains.

Is your system webview version 149.0.7827.114 (782711401)?

Also if possible, can you send the log file you had of running for 205 seconds? PM is fine, if you want.


https://files.catbox.moe/9zsu80.gz

Thanks. I will see how this diverges from my own log.

Look also through ‘systemctl’ output and see if there are any weird services there (containing harbour or just an appname), AI seems to love those and has problems properly cleaning after itself on uninstall. Btw that’s a crazy amount of repos you have active, if someone starts experimenting with a system lib and uploads experimental ofono/libc or other system lib, you might accidentally pull it, for your own sake disable the repos once you installed the app you want and reenable when you need an update, much saner approach than 30+ repos pulling whatever was cobbled together today, the update cadence with AI almost guarantees there is hardly any testing

1 Like

One other thing that would be good to know: are you using only the stock keyboard?

Yup (2000000000 chars)

Me too (20000 chars)

OK, so I finally broke down and factory reset.

And the end result is…

Still crashes. So at this point I think I can safely say it is NOT an issue with anything I did to the device…