Project

General

Profile

Actions

Bug #1106

open

New intents sometimes aren't loaded on restart on Android 14 (WebView crash)

Added by Soren Stoutner 6 months ago. Updated 3 months ago.

Status:
In Progress
Priority:
3.x
Start date:
10/18/2023
Due date:
% Done:

0%

Estimated time:

Description

When restarting from a new intent, sometimes the old tabs are restored but the new URL isn't loaded.


Files

Actions #1

Updated by Soren Stoutner 6 months ago

So, it turns out that this bug is actually caused because Chromium is crashing on restart.

[10-27 22:19:01.173 2343: 2343 E/chromium ] [ERROR:aw_browser_terminator.cc(112)] Render process (14421) kill (OOM or update) wasn't handed by all associated webviews, killing application.

This causes the new intent to be lost and the app to be restored with the previous tabs.

There probably isn't anything I can do about it, but it will likely be fixed by a WebView update.

Actions #2

Updated by Soren Stoutner 5 months ago

  • Subject changed from New intents sometimes aren't loaded on restart on Android 14 to New intents sometimes aren't loaded on restart on Android 14 (WebView crash)
Actions #3

Updated by Soren Stoutner 4 months ago

Here is a fuller crash log:

[ 01-03 11:13:20.916  9026: 9060 W/cr_ChildProcessConn ]
onServiceDisconnected (crash or killed by oom): pid=17624 bindings:W   

[ 01-03 11:13:20.940  9026: 9060 W/cr_ChildProcessConn ]
onServiceDisconnected (crash or killed by oom): pid=14346 bindings:W  S

[ 01-03 11:13:20.943  9026: 9060 W/cr_ChildProcessConn ]
onServiceDisconnected (crash or killed by oom): pid=9188 bindings:W  S

[ 01-03 11:13:20.954  9026: 9026 E/chromium ]
[ERROR:aw_browser_terminator.cc(154)] Renderer process (17624) crash detected (code -1).

[ 01-03 11:13:20.957  9026: 9026 E/chromium ]
[ERROR:aw_browser_terminator.cc(154)] Renderer process (9627) crash detected (code -1).

[ 01-03 11:13:20.960  9026: 9060 W/cr_ChildProcessConn ]
onServiceDisconnected (crash or killed by oom): pid=18692 bindings:W  S

[ 01-03 11:13:20.966  9026: 9026 E/chromium ]
[ERROR:aw_browser_terminator.cc(154)] Renderer process (9070) crash detected (code -1).

[ 01-03 11:13:20.971  9026: 9026 E/chromium ]
[ERROR:aw_browser_terminator.cc(154)] Renderer process (9188) crash detected (code -1).

[ 01-03 11:13:20.971  9026: 9060 W/cr_ChildProcessConn ]
onServiceDisconnected (crash or killed by oom): pid=9627 bindings:W  S

[ 01-03 11:13:20.980  9026: 9026 E/chromium ]
[ERROR:aw_browser_terminator.cc(154)] Renderer process (14346) crash detected (code -1).

[ 01-03 11:13:20.984  9026: 9026 E/chromium ]
[ERROR:aw_browser_terminator.cc(154)] Renderer process (18692) crash detected (code -1).

[ 01-03 11:13:20.994  9026: 9026 E/chromium ]
[ERROR:aw_browser_terminator.cc(110)] Render process (9627) kill (OOM or update) wasn't handed by all associated webviews, killing application.
Actions #4

Updated by ask low 3 months ago

Yup strange issue, as this started occurring when after migrating from 13 to 14.

Actions #5

Updated by Soren Stoutner 3 months ago

Currently I am investigating if making changes to WebView DevTools is responsible. It appears to be so, but I do want to do a bit more testing to narrow down which one.

Actions #6

Updated by ask low 3 months ago

That could be a possibility too. Maybe site isolation itself ? Btw I can't reproduce this either. The crash seems to be random, only on heavier websites.

Actions #7

Updated by Soren Stoutner 3 months ago

In my testing, disabling the WebView DevTools settings that relate to site isolation and partitioning result in this problem happening less frequently, but it doesn't eliminate it. In other words, the root of the problem is something else, but site isolation and partitioning aggravates whatever it is.

I have updated the text at https://www.stoutner.com/webview-devtools/ with a warning.

Actions #8

Updated by Soren Stoutner 3 months ago

On my Pixel 5, WebView 121.0.6167.143 fixes this issue, but only if all the site isolation options in WebView DevTools are disabled.

I would assume that a future WebView update will fully resolve the problem.

Actions #9

Updated by ask low 3 months ago

Site isolation is crucial to me anyways. I'd rather have the bug instead of disabling it.

Actions #10

Updated by Soren Stoutner 3 months ago

As I have already explained in other places, Privacy Browser already does most of what other browsers call site isolation all the time anyways. Enabling the site isolation options in WebView DevTools adds a little, tiny bit of additional protection. but only if your device has already been fully compromised by malicious JavaScript.

Actions #11

Updated by ask low 3 months ago

That's true. But I did like the Webview DevTools flags concept & it not only isolates PB, but the rest of the apps that wrap weblinks around them that didn't follow it.

Actions #12

Updated by Soren Stoutner 3 months ago

That is a good point. Enabling site isolation in WebView DevTools does protect other apps besides Privacy Browser.

Of course, it also means this bug affects all other apps that use WebView. On my Pixel 5, with site isolation enabled and this current bug, it makes a lot of apps on my device crash frequently.

Actions

Also available in: Atom PDF