814022 - Fix instanceof for new DOM bindings

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Peter Van der Beken [:peterv]]

Attachment: v1

There are a couple of problems:
- We don't unwrap security wrappers around new binding objects in GetWrappedNativeOfJSObject (called by the hasInstance hook for interface objects)
- We don't unwrap security wrappers around new binding objects in nsJSIID::HasInstance

A number of tests use instanceof with a DOM object and a DOM interface object from a different scope. I've fixed a bunch of them, even though the fixes above make it unnecessary. I still think we should fix these tests. Ideally we'd stop supporting this altogether by getting rid of the hasInstance hook on interface objects, but we'd need to fix all the chrome code that does it (and probably provide a chrome-only alternative). I tried restricting the QI in hasInstance to chrome code, but that doesn't work for now because we have EventTargets for which the first interface on the primary prototype chain does not inherit from nsIDOMEventTarget :-/.

Comment 1

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Comment on attachment 684038 [details] [diff] [review]
v1

Review of attachment 684038 [details] [diff] [review]:
-----------------------------------------------------------------

::: js/xpconnect/src/XPCJSID.cpp
@@ +482,5 @@
>                  return NS_ERROR_FAILURE;
> +        } else {
> +            JSObject* unsafeObj =
> +                XPCWrapper::Unwrap(cx, obj, /* stopAtOuter = */ false);
> +            JSObject* cur = unsafeObj ? unsafeObj : obj; 

Trailing whitespace

::: js/xpconnect/src/nsXPConnect.cpp
@@ +1384,5 @@
>          return (nsISupports*)xpc_GetJSPrivate(obj2);
>  
> +    JSObject* unsafeObj =
> +        XPCWrapper::Unwrap(aJSContext, aJSObj, /* stopAtOuter = */ false);
> +    JSObject* cur = unsafeObj ? unsafeObj : aJSObj; 

Again

Comment 2

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 684038 [details] [diff] [review]
v1

r=me

Comment 3

[Peter Van der Beken [:peterv]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/3e9a567ce55c

Comment 4

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/3e9a567ce55c