1127230 - Change the NPAPI sandbox prefs to integers to indicate the level of sandboxing.

Status: RESOLVED FIXED

(Core :: Security: Process Sandboxing, defect)

Description

[Bob Owen (:bobowen)]

As cpeterson mentioned in Bug 1126402 Comment 7, the different levels of sandboxing of the NPAPI processes, would be better being controlled by integer prefs.

Comment 1

[Bob Owen (:bobowen)]

Attachment: Change the NPAPI sandbox prefs to integers to indicate the level of sandboxing.

This changes the bool prefs into integer prefs as suggested.

It means I can look into adding a third level, with some of the suggestions from bbondy for tightening the sandbox.
This might prove useful when we get some sort of test suite for Flash.

(In reply to Chris Peterson [:cpeterson] from comment #7)

> That would simplify some of the C++ checks below and make testing easier.
> For example, you could ask people to run some test comparing (say) sandbox
> levels 2 and 3.

Chris, did you just mean not having to deal with two bool prefs or is there something that I've missed?

Comment 2

[Chris Peterson [:cpeterson]]

Comment on attachment 8557191 [details] [diff] [review]
Change the NPAPI sandbox prefs to integers to indicate the level of sandboxing.

Review of attachment 8557191 [details] [diff] [review]:
-----------------------------------------------------------------

(In reply to Bob Owen (:bobowen) from comment #1)
> (In reply to Chris Peterson [:cpeterson] from comment #7)
> 
> > That would simplify some of the C++ checks below and make testing easier.
> > For example, you could ask people to run some test comparing (say) sandbox
> > levels 2 and 3.
> 
> Chris, did you just mean not having to deal with two bool prefs or is there
> something that I've missed?

I just meant the two bool prefs. The integer levels are just one pref and don't have the confusion around invalid combinations of bool states. :)

Comment 3

[Benjamin Smedberg]

Comment on attachment 8557191 [details] [diff] [review]
Change the NPAPI sandbox prefs to integers to indicate the level of sandboxing.

I don't think we should re-use the same pref names when they have a different type. How about dom.ipc.plugins.sandbox-level.* ?

r=me otherwise

Comment 4

[Bob Owen (:bobowen)]

Thanks Ben.

Just realised I hadn't pushed this to try at all.
https://treeherder.mozilla.org/#/jobs?repo=try&revision=8915e17a2100

Comment 5

[Bob Owen (:bobowen)]

remote:   https://hg.mozilla.org/integration/mozilla-inbound/rev/9c95e28087ca

Comment 6

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/9c95e28087ca