1067513 - Incorporate base::LazyInstance fixes from upstream

Status: RESOLVED FIXED

(Core :: IPC, defect)

Description

[(inactive) Jim Chen [:jchen] [:darchons]]

The Chromium class base::LazyInstance has several known issues that were fixed upstream. We should incorporate these fixes. E.g.,

http://src.chromium.org/viewvc/chrome?view=revision&revision=41434
http://src.chromium.org/viewvc/chrome?view=revision&revision=69107
http://src.chromium.org/viewvc/chrome?view=revision&revision=85487
http://src.chromium.org/viewvc/chrome?view=revision&revision=106763

Comment 1

[(inactive) Jim Chen [:jchen] [:darchons]]

Attachment: Import updated base::LazyInstance from upstream (v1)

This patch imports the following files from upstream, and adapts them to the
in-tree fork at ipc/chromium,

* src/base/lazy_instance.cc
* src/base/lazy_instance.h

The updated code fixes several issues with LazyInstance [1-4], which we suspect
are causing crashes on Android (such as bug 1060009). LazyInstance usages in
other files are updated to reflect changes in the class interface.

Try: https://tbpl.mozilla.org/?tree=Try&rev=9f043c9fe425

[1] https://code.google.com/p/chromium/issues/detail?id=28749
[2] https://code.google.com/p/chromium/issues/detail?id=64930
[3] https://code.google.com/p/chromium/issues/detail?id=65298
[4] https://code.google.com/p/chromium/issues/detail?id=82122

Comment 2

[Benjamin Smedberg]

Comment on attachment 8490163 [details] [diff] [review]
Import updated base::LazyInstance from upstream (v1)

Do I need to actually-review this, or is this a straight import from upstream?

Does the other fork at /security/sandbox/chromium/base/lazy_instance.h also need this?

Comment 3

[(inactive) Jim Chen [:jchen] [:darchons]]

(In reply to Benjamin Smedberg  [:bsmedberg] from comment #2)
> Comment on attachment 8490163 [details] [diff] [review]
> Import updated base::LazyInstance from upstream (v1)
> 
> Do I need to actually-review this, or is this a straight import from
> upstream?

Straight import, except I had to change LazyInstance::private_buf_ to using mozilla::AlignedStorage2 instead of the Chromium version (it's a straightforward change though).

> Does the other fork at /security/sandbox/chromium/base/lazy_instance.h also
> need this?

That version is already up-to-date.

Comment 4

[(inactive) Jim Chen [:jchen] [:darchons]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/92ffb3d125f6

Comment 5

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/92ffb3d125f6

Comment 6

[(inactive) Jim Chen [:jchen] [:darchons]]

Comment on attachment 8490163 [details] [diff] [review]
Import updated base::LazyInstance from upstream (v1)

Approval Request Comment

[Feature/regressing bug #]: N/A

[User impact if declined]: We suspect this patch may fix the unidentified crash in bug 1060009 that's happening on all channels.

[Describe test coverage new/current, TBPL]: m-c; locally

[Risks and why]: Small risk because this introduces new code, even though the code comes from existing Chromium upstream source; however, we do want to see if this fixes the crashes in bug 1060009, which we can only find out through the beta channel.

[String/UUID change made/needed]: None

Comment 7

[Sylvestre Ledru [:Sylvestre]]

Comment on attachment 8490163 [details] [diff] [review]
Import updated base::LazyInstance from upstream (v1)

I am going to take it in aurora, if it does not cause issue, I will merge it in beta (probably beta 6 or 7)

Comment 8

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/0a5628977fd9

Comment 9

[Sylvestre Ledru [:Sylvestre]]

Comment on attachment 8490163 [details] [diff] [review]
Import updated base::LazyInstance from upstream (v1)

Taking it for beta 6

Comment 10

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-beta/rev/d6aa05e710f2