New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 601184 link

Starred by 5 users

Issue metadata

Status: Duplicate
Merged: issue 537671
Owner:
Closed: Mar 2017
Cc:
EstimatedDays: ----
NextAction: ----
OS: Android
Pri: 2
Type: Bug

Blocked on:
issue 560446

Blocking:
issue 601023



Sign in to add a comment

Android may be susceptible to issue 560446 per requirement to replace the initial process priority binding before the initial navigation is submitted

Project Member Reported by gab@chromium.org, Apr 6 2016

Issue description

tl;dr; from  issue 560446 : renderers are backgrounded when they are neither playing audio nor have any visible widgets. This logic was recently found to cause problems on desktop as it would cause foreground renderers to be backgrounded until the initial navigation had committed (per having no visible widgets before then =S!)

We experimented with disabling the initial priority toggling on Windows (tracked in issue 579116) and with positive results are now moving on to making that workaround default baring a better solution (https://codereview.chromium.org/1769123002/#ps1).

On Android however this is not possible as the bindings explicitly have to be set on process creation FWIU (ref. email discussion in Sep2015 between sebsg/gab/ncarter/jaekyun/sievers/klobag)...

For now I will land https://codereview.chromium.org/1769123002 with a special case to not do the workaround on Android but we need to verify whether it is affected by that issue. If it is, finding a proper fix to that issue (a way to know the priority before the navigation completes) is even more critical.
 
Interesting. I can see the issue with moderate binding we recently introduced in  issue 485867 . Before that, the background tab has lower priority than the initial one. Now if a tab is backgrounded, but the browser hasn't been backgrounded yet, that tab has the same priority as the new process.

On Android, we can adjust the process priority. But we probably don't want to start with high before it is committed as it may get the current renderer process killed. 
See  crbug.com/537671 . Let's look into removing the determinedVisibility() hack and match the desktop initial state (visible?).

I think we mainly have to look at the 'open in new tab' use case again and make sure we don't regress.
Owner: gab@chromium.org
Status: Assigned (was: Untriaged)
Bug triage: gab@, you seems to be working on this, let me assign this bug to you.
Blocking: 601023
Project Member

Comment 6 by sheriffbot@chromium.org, Jun 1 2016

Labels: -M-52 M-53 MovedFrom-52
Moving this nonessential bug to the next milestone.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
Project Member

Comment 7 by sheriffbot@chromium.org, Jul 14 2016

Labels: -M-53 MovedFrom-53
This issue has been moved once and is lower than Pri-1. Removing the milestone.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Comment 8 by gab@chromium.org, Nov 2 2016

Status: Available (was: Assigned)

Comment 9 by boliu@chromium.org, Mar 15 2017

Mergedinto: 537671
Status: Duplicate (was: Available)

Comment 10 by boliu@chromium.org, Apr 17 2017

I'm looking into removing "determinedVisibility", ie the bug this got merged into. I see desktop explicitly chose to put a new process in the foreground, even if the tab is in the background. But android got left behind.

Can anyone dig up this thread and forward it to me:

> On Android however this is not possible as the bindings explicitly have to be set on process creation FWIU (ref. email discussion in Sep2015 between sebsg/gab/ncarter/jaekyun/sievers/klobag)...

Arguably, desktop is actually being intentionally incorrect here, which makes it super hard to reason about correctness. Also I can think of reasons android might not want to behave like desktop. But I'd rather read that thread first :p

Comment 11 by boliu@chromium.org, Apr 17 2017

imo this comment from nick is the more correct way to fix the desktop issue:
https://bugs.chromium.org/p/chromium/issues/detail?id=560446#c26

I don't see any replies on that bug, so not sure if that's already been discussed to be a bad solution or something?

Comment 12 by gab@chromium.org, Apr 18 2017

Re. "thread between sebsg/gab/ncarter/jaekyun/sievers/klobag": looks like retention policy ate it :(... don't know that it was bound to a google group discussion (that was a mistake...).

Responding to other comments on dupe bug.

Comment 13 by gab@chromium.org, Apr 18 2017

PS: a "mistake" you can blame me for... I've since learned to tie every meaningful discussion to a group for retention...

Sign in to add a comment