564861 - [OOPP] NPN_IntFromIdentifier returns INT_MIN always?

Status: RESOLVED FIXED

(Core Graveyard :: Plug-ins, defect)

Description

[Benjamin Smedberg]

From Ben Pryor @ Microsoft:

> We've found an issue with the latest Firefox builds and Silverlight. The 
> issue only repros when OOPP is enabled for the Silverlight plugin.
> 
> The issue concerns array-style access to a scriptable object (a 
> Silverlight object exposed to Javascript through Firefox's NPRuntime). 
> In Javascript, if I use array-style syntax against such an object, i.e.:
> 
> myobject[0]
> 
> then the following occurs:
> 
> - the browser calls our object's NPGetPropertyFunctionPtr
> 
> - we call NPN_IdentifierIsString on the NPIdentifier, and it returns false
> 
> - we call NPN_IntFromIdentifier on the NPIdentifier, and it returns 
> -2147483648 (i.e. INT_MIN for a 32-bit signed int)
> 
> NPN_IntFromIdentifier seems to return the same value regardless of the 
> Javascript syntax, i.e. myobject[0], myobject[1], etc. all give the same 
> behavior.
> 
> If OOPP is disabled for the Silverlight plugin, then 
> NPN_IntFromIdentifier returns 0, 1, etc. as expected.

Comment 1

[Benjamin Smedberg]

Attachment: Fix conditional and stop using -1 sentinel incorrectly, rev. 1

Comment 2

[Benjamin Smedberg]

Attachment: Test

Comment 3

[LegNeato]

Comment on attachment 444463 [details] [diff] [review]
Fix conditional and stop using -1 sentinel incorrectly, rev. 1

a=LegNeato for 1.9.2.4

Comment 4

[LegNeato]

Can we get a trybuild we can point MS at to see if this completely fixes the issue?

Comment 5

[LegNeato]

trybuild is at https://build.mozilla.org/tryserver-builds/[e-mail]-bug564861/

Comment 6

[Benjamin Smedberg]

http://hg.mozilla.org/mozilla-central/rev/44347d8a48bc
http://hg.mozilla.org/mozilla-central/rev/4980a5f7467f

(in today's m-c nightly)

Comment 8

[Benjamin Smedberg]

http://hg.mozilla.org/releases/mozilla-1.9.2/rev/d1e526c08e25
http://hg.mozilla.org/releases/mozilla-1.9.2/rev/6a7c8bc32810
http://hg.mozilla.org/releases/mozilla-1.9.2/rev/e68bf78d90f0
http://hg.mozilla.org/releases/mozilla-1.9.2/rev/205a9f77988e

Comment 9

[Al Billings [:abillings - ex-MoCo]]

Mochitest passes on 1.9.2 post-checkin. Is this enough to verify this issue since there is no manual STR for the bug?

Comment 10

[Benjamin Smedberg]

You should verify the dupped bug 559760 also, but the bug is fairly straightforward and the test covers it.