1257417 - [e10s] Pinned tabs aren't loaded when the content process crashes

Status: NEW

(Firefox :: General, defect)

Description

[Reuben Morais [:reuben]]

STR:

1) Content process crashes
2) I click to report the crash and restore the tab

Results:

Pinned tabs are unloaded.

Expected results:

Pinned tabs should reload when the content process is restored.

Comment 1

[:Gijs (he/him)]

Can you check bug 1104274 comment 5?

(In reply to Mike Conley (:mconley) - Needinfo me! from comment #5)
> That tells me that something screwed up with the takedown of those browser
> tabs when the content process crashed - so it's not surprising that your
> session got all screwed up. :/
> 
> If you see this again, can you please open the Browser Console to see if
> there's anything strange in there that might indicate what went wrong during
> the crash handling?

Comment 2

[:Gijs (he/him)]

Hmm, actually, this is like bug 1254569 instead - couldn't find that initially because it didn't have "e10s" in the summary.

What confuses me is that that was duped to a bug 1227630, which is about the favicons disappearing - but I think the point is that pinned tabs should autorestore, because they're pinned tabs, which might be a separate issue? Mike?

Comment 3

[arni2033]

> Expected results:
>
> Pinned tabs should reload when the content process is restored.

This is not the expected result. See bug 1256280 comment 9 (other comments as well)

Comment 4

[Reuben Morais [:reuben]]

(In reply to arni2033 from comment #3)
> > Expected results:
> >
> > Pinned tabs should reload when the content process is restored.
> 
> This is not the expected result. See bug 1256280 comment 9 (other comments
> as well)

Sure, the situation right now with a single content process is not optimal, but IMO not restoring them is a regression from the non-e10s behavior. My pinned tabs are pinned precisely because I want them to be loaded at all times.

Comment 5

[Mike Conley (:mconley) (:⚙️)]

(In reply to :Gijs Kruitbosch from comment #2)
> Hmm, actually, this is like bug 1254569 instead - couldn't find that
> initially because it didn't have "e10s" in the summary.
> 
> What confuses me is that that was duped to a bug 1227630, which is about the
> favicons disappearing - but I think the point is that pinned tabs should
> autorestore, because they're pinned tabs, which might be a separate issue?
> Mike?

Yeah agreed - I think we made a mistake while triaging and thought this was the favicon bug. I'll undupe it, redupe it to this bug, and get this one triaged.

Comment 7

[arni2033]

(In reply to Reuben Morais [:reuben] from comment #4)
> Sure, the situation right now with a single content process is not optimal, but IMO not
> restoring them is a regression from the non-e10s behavior. My pinned tabs are pinned precisely
> because I want them to be loaded at all times.
I thought that what you're requesting is automatic restoration of pinned tabs... My apologies :(

I only give a feedback about what's working wrong (IMO), and the obviously wrong things here are:
1) auto-reloading of pinned tabs _before_ user clicks "Restore Tab" (explained in bug 1256280)
2) reloading of pinned tabs even for users with prev "restore_pinned_tabs_on_demand" set to true