Bug 225997 – [JSC] Call custom accessors / values with their holder's global object

• Component: JavaScriptCore
• Status: RESOLVED FIXED

[Alexey Shvayka]

[JSC] Call custom accessors / values with their holder's global object

[Alexey Shvayka]

Created attachment 429120 [details]
Patch

[Alexey Shvayka]

Created attachment 429495 [details]
Patch

Fix global object for non-reified properties, use CustomAccessor for JS built-ins, and add tests.

[EWS Watchlist]

This patch modifies the imported WPT tests. Please ensure that any changes on the tests (not coming from a WPT import) are exported to WPT. Please see https://trac.webkit.org/wiki/WPTExportProcess

[Radar WebKit Bug Importer]

<rdar://problem/78541835>

[Alexey Shvayka]

Created attachment 430897 [details]
Patch

Rebase on top of relanded put() revamp.

[Yusuke Suzuki]

That's really awesome.

[Alexey Shvayka]

Created attachment 431994 [details]
WIP

Introduce CallFrame::callerLexicalGlobalObject() and [CustomNativeFunctionAccessor] extended attribute to accomodate JSLocation setters, which need lexical global object to set document.referrer.

[Alexey Shvayka]

Created attachment 431995 [details]
WIP

Add missing Source/WebCore/bindings/js/JSLocationCustom.h file.

[Alexey Shvayka]

Created attachment 432455 [details]
Patch

Call CustomValue setters with correct realm as well, rework callerGlobalObject() to use vm.topCallFrame for custom accessors, revert changes to create_hash_table, and add tests.

[Alexey Shvayka]

Created attachment 432710 [details]
Patch

Fix --debug build, smsimplify callerGlobalObject(), add test for JSCustomSetterFunction passing correct referrer, and improve ChangeLog.

[Alexey Shvayka]

Created attachment 432711 [details]
Patch

svn add LayoutTests/imported

[Alexey Shvayka]

(In reply to Yusuke Suzuki from comment #6)
> That's really awesome.

Thanks! I would appreciate a review.

Once we land this, implementing RegExpConstructor legacy features (https://bugs.webkit.org/show_bug.cgi?id=220233) would be so much easier.

[Darin Adler]

Looks like a real regression in this test:

    http/tests/security/block-top-level-navigations-by-untrusted-first-party-iframes.html

We should probably hold off on review until that is resolved.

[Alexey Shvayka]

Created attachment 432794 [details]
Patch

Attempt to retrieve VMEntryScope's global object even if vm.topCallFrame is nullptr, fixing the test regression, and make JSCustomSetterFunction test more sophisticated.

[Yusuke Suzuki]

Comment on attachment 432794 [details]
Patch

r=me

[Alexey Shvayka]

Created attachment 434056 [details]
Patch for landing

[Alexey Shvayka]

Created attachment 434099 [details]
Patch for landing

svn add <missing LayoutTest files>.

[Alexey Shvayka]

Comment on attachment 434099 [details]
Patch for landing

I'm putting r+ on my own patch (that was already reviewed by Yusuke (thanks!)) so the GitHub's WPT export bot mark my PR (https://github.com/web-platform-tests/wpt/pull/29767) as reviewed for me to land it.

[Alexey Shvayka]

Created attachment 434110 [details]
Patch for landing

Remove trailing whitespace from wpt/WebIDL/ecmascript-binding/global-object-implicit-this-value-cross-realm.html.

[EWS]

Committed r280256 (239923@main): <https://commits.webkit.org/239923@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 434110 [details].