Implement the Origin-Agent-Cluster header by annevk · Pull Request #66162 · WebKit/WebKit

annevk · 2026-06-01T08:47:17Z

`b4c4069`

Implement the Origin-Agent-Cluster header
https://bugs.webkit.org/show_bug.cgi?id=216618
rdar://69452369

Reviewed by Alex Christensen.

This implements origin-keyed agent clusters and gates it behind
OriginAgentClusterEnabled (set to preview). It is standardized here:

https://html.spec.whatwg.org/multipage/browsers.html#origin-isolation

This new keying does not impact process allocation, although we could
decide to make it impact that in the future if we wanted to. It impacts
document.domain, serialization and deserialization of certain objects,
and the new window.originAgentCluster getter.

We add a new test that is upstreamed at
web-platform-tests/wpt#60311 that catches an
issue with our existing agent cluster implementation for opaque
origins and we also fix that issue.

As the test infrastructure continues to use the same browsing context
group for each test, we add a way to reset the browsing context group
state so it appears as if you are getting a fresh browsing context
group when you start a new test.

The subtest failures of regression-1399759.https.sub.html are due to
about:blank inheriting from the parent instead of the initiator of the
navigation. That's a distinct issue from this new feature.

going-back.sub.https.html is skipped because it times out on some bots
at the history.back() step, while the feature-related assertions pass.
I suspect it's related to the bfcache. I could not reproduce it locally
(1000 iterations of the test as well as 10 full-directory runs).

Canonical link: https://commits.webkit.org/314346@main

b06eacb

Misc	iOS, visionOS, tvOS & watchOS	macOS	Linux	Windows	Apple Internal
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 win	⏳ 🛠 ios-apple
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2	~~🧪 win-tests~~	⏳ 🛠 mac-apple
✅ 🧪 webkitperl	~~🧪 ios-wk2~~	✅ 🧪 api-mac	✅ 🧪 api-wpe		⏳ 🛠 vision-apple
	~~🧪 ios-wk2-wpt~~	✅ 🧪 api-mac-debug	✅ 🛠 gtk3-libwebrtc
✅ 🛠 🧪 jsc-x86-64	~~🧪 api-ios~~	✅ 🧪 mac-wk1	✅ 🛠 gtk
✅ 🛠 🧪 jsc-debug-arm64	✅ 🛠 ios-safer-cpp	✅ 🧪 mac-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 vision	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 api-gtk
✅ 🛠 🧪 merge	✅ 🛠 vision-sim	✅ 🧪 mac-wk2-stress	✅ 🛠 playstation
	✅ 🧪 vision-wk2	✅ 🧪 mac-intel-wk2	~~🛠 jsc-armv7~~
	✅ 🛠 tv	✅ 🛠 mac-safer-cpp	~~🧪 jsc-armv7-tests~~
	✅ 🛠 tv-sim	✅ 🧪 mac-site-isolation
	✅ 🛠 watch
	✅ 🛠 watch-sim

webkit-early-warning-system · 2026-06-01T08:48:54Z

EWS run on previous version of this PR (hash f5dd211)

Details

Misc	iOS, visionOS, tvOS & watchOS	macOS	Linux	Windows	Apple Internal
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 win	✅ 🛠 ios-apple
~~🧪 bindings~~	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	❌ 🧪 wpe-wk2	~~🧪 win-tests~~	✅ 🛠 mac-apple
✅ 🧪 webkitperl	~~🧪 ios-wk2~~	✅ 🧪 api-mac	✅ 🧪 api-wpe		✅ 🛠 vision-apple
	~~🧪 ios-wk2-wpt~~	✅ 🧪 api-mac-debug	✅ 🛠 gtk3-libwebrtc
✅ 🛠 🧪 jsc-x86-64	🧪 api-ios	✅ 🧪 mac-wk1	✅ 🛠 gtk
✅ 🛠 🧪 jsc-debug-arm64	❌ 🛠 ios-safer-cpp	❌ 🧪 mac-wk2	❌ 🧪 gtk-wk2
	✅ 🛠 vision	~~🧪 mac-AS-debug-wk2~~	✅ 🧪 api-gtk
	✅ 🛠 vision-sim	❌ 🧪 mac-wk2-stress	✅ 🛠 playstation
	✅ 🧪 vision-wk2	❌ 🧪 mac-intel-wk2	~~🛠 jsc-armv7~~
	✅ 🛠 tv	❌ 🛠 mac-safer-cpp	~~🧪 jsc-armv7-tests~~
	✅ 🛠 tv-sim	❌ 🧪 mac-site-isolation
	✅ 🛠 watch
	✅ 🛠 watch-sim

webkit-ews-buildbot · 2026-06-01T10:26:21Z

macOS Safer C++ Build #111505 (`f5dd211`)

❌ Found 3 failing files with 5 issues. Please address these issues before landing. See WebKit Guidelines for Safer C++ Programming.
(cc @rniwa)

webkit-ews-buildbot · 2026-06-01T10:41:55Z

iOS Safer C++ Build #30153 (`f5dd211`)

❌ Found 3 failing files with 4 issues. Please address these issues before landing. See WebKit Guidelines for Safer C++ Programming.
(cc @rniwa)

webkit-early-warning-system · 2026-06-01T11:10:51Z

EWS run on previous version of this PR (hash a910c5e)

Details

Misc	iOS, visionOS, tvOS & watchOS	macOS	Linux	Windows	Apple Internal
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 win	✅ 🛠 ios-apple
~~🧪 bindings~~	~~🛠 ios-sim~~	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2	~~🧪 win-tests~~	✅ 🛠 mac-apple
✅ 🧪 webkitperl	❌ 🧪 ios-wk2	🧪 api-mac	✅ 🧪 api-wpe		✅ 🛠 vision-apple
	~~🧪 ios-wk2-wpt~~	✅ 🧪 api-mac-debug	✅ 🛠 gtk3-libwebrtc
✅ 🛠 🧪 jsc-x86-64	✅ 🧪 api-ios	✅ 🧪 mac-wk1	✅ 🛠 gtk
✅ 🛠 🧪 jsc-debug-arm64	✅ 🛠 ios-safer-cpp	✅ 🧪 mac-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 vision	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 api-gtk
	✅ 🛠 vision-sim	✅ 🧪 mac-wk2-stress	✅ 🛠 playstation
	✅ 🧪 vision-wk2	❌ 🧪 mac-intel-wk2	~~🛠 jsc-armv7~~
	✅ 🛠 tv	✅ 🛠 mac-safer-cpp	~~🧪 jsc-armv7-tests~~
	✅ 🛠 tv-sim	✅ 🧪 mac-site-isolation
	✅ 🛠 watch
	✅ 🛠 watch-sim

achristensen07 · 2026-06-01T14:34:01Z

@@ -101,6 +111,8 @@ class BrowsingContextGroup : public RefCountedAndCanMakeWeakPtr<BrowsingContextG
    HashMap<WebCore::Site, WeakPtr<FrameProcess>> m_processMap;
    WeakListHashSet<WebPageProxy> m_pages;
    WeakHashMap<WebPageProxy, HashSet<Ref<RemotePageProxy>>> m_remotePages;
+
+    HashMap<WebCore::SecurityOriginData, WebCore::OriginKeyed> m_historicalAgentClusterKeyMap;


Could this just be a HashSet of origins that are origin keyed?
Are origins ever removed, or does this just grow for the lifetime of the BrowsingContextGroup?

It grows for the lifetime of the BCG.

And we can't use a set because we need to track:

Not seen at all.

Seen and origin-keyed.

Seen and not origin-keyed.

Because we don't want to flip on the prior keying.

achristensen07 · 2026-06-01T14:36:24Z

+    void clearBrowsingContextGroupForTesting()
+    {
+        m_identifier = WebCore::BrowsingContextGroupIdentifier::generate();
+        m_historicalAgentClusterKeyMap.clear();


If we put this in BrowsingContextGroup.cpp we can probably remove the additional headers included from this header and use forward declarations instead.

I will stop inlining it, but I can't forward declare anything new unless we wrap the HashMap in a unique_ptr which I suspect we don't want to do.

webkit-early-warning-system · 2026-06-01T15:47:41Z

EWS run on previous version of this PR (hash 3cfe28c)

Details

Misc	iOS, visionOS, tvOS & watchOS	macOS	Linux	Windows	Apple Internal
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 win	✅ 🛠 ios-apple
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2	❌ 🧪 win-tests	✅ 🛠 mac-apple
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe		✅ 🛠 vision-apple
	✅ 🧪 ios-wk2-wpt	✅ 🧪 api-mac-debug	✅ 🛠 gtk3-libwebrtc
✅ 🛠 🧪 jsc-x86-64	✅ 🧪 api-ios	✅ 🧪 mac-wk1	✅ 🛠 gtk
✅ 🛠 🧪 jsc-debug-arm64	✅ 🛠 ios-safer-cpp	✅ 🧪 mac-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 vision	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 api-gtk
	✅ 🛠 vision-sim	✅ 🧪 mac-wk2-stress	✅ 🛠 playstation
	✅ 🧪 vision-wk2	✅ 🧪 mac-intel-wk2	~~🛠 jsc-armv7~~
	✅ 🛠 tv	✅ 🛠 mac-safer-cpp	~~🧪 jsc-armv7-tests~~
	✅ 🛠 tv-sim	✅ 🧪 mac-site-isolation
	✅ 🛠 watch
	✅ 🛠 watch-sim

achristensen07 · 2026-06-02T00:42:30Z

    });
 }

+void WKPageClearBrowsingContextGroupForTesting(WKPageRef pageRef)


I wonder if this would be better just added as part of WKPageResetStateBetweenTests

achristensen07 · 2026-06-02T00:43:39Z

+    OriginAgentClusterPolicy isolatedCopy() const & { return { isOriginKeyed }; }
+    OriginAgentClusterPolicy isolatedCopy() && { return { isOriginKeyed }; }


There doesn't seem to be a good reason to have a separate function for if it's an l-value or an r-value if they both are identical. Maybe we're planning to add strings here, I'm not sure.

Yeah, I realized we can remove the entire struct for now so I went ahead and did that. Perhaps we need something more involved when we flip the default, but maybe not.

webkit-early-warning-system · 2026-06-02T06:01:36Z

EWS run on current version of this PR (hash b06eacb)

Details

Misc	iOS, visionOS, tvOS & watchOS	macOS	Linux	Windows	Apple Internal
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 win	⏳ 🛠 ios-apple
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2	~~🧪 win-tests~~	⏳ 🛠 mac-apple
✅ 🧪 webkitperl	~~🧪 ios-wk2~~	✅ 🧪 api-mac	✅ 🧪 api-wpe		⏳ 🛠 vision-apple
	~~🧪 ios-wk2-wpt~~	✅ 🧪 api-mac-debug	✅ 🛠 gtk3-libwebrtc
✅ 🛠 🧪 jsc-x86-64	~~🧪 api-ios~~	✅ 🧪 mac-wk1	✅ 🛠 gtk
✅ 🛠 🧪 jsc-debug-arm64	✅ 🛠 ios-safer-cpp	✅ 🧪 mac-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 vision	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 api-gtk
✅ 🛠 🧪 merge	✅ 🛠 vision-sim	✅ 🧪 mac-wk2-stress	✅ 🛠 playstation
	✅ 🧪 vision-wk2	✅ 🧪 mac-intel-wk2	~~🛠 jsc-armv7~~
	✅ 🛠 tv	✅ 🛠 mac-safer-cpp	~~🧪 jsc-armv7-tests~~
	✅ 🛠 tv-sim	✅ 🧪 mac-site-isolation
	✅ 🛠 watch
	✅ 🛠 watch-sim

https://bugs.webkit.org/show_bug.cgi?id=216618 rdar://69452369 Reviewed by Alex Christensen. This implements origin-keyed agent clusters and gates it behind OriginAgentClusterEnabled (set to preview). It is standardized here: https://html.spec.whatwg.org/multipage/browsers.html#origin-isolation This new keying does not impact process allocation, although we could decide to make it impact that in the future if we wanted to. It impacts document.domain, serialization and deserialization of certain objects, and the new window.originAgentCluster getter. We add a new test that is upstreamed at web-platform-tests/wpt#60311 that catches an issue with our existing agent cluster implementation for opaque origins and we also fix that issue. As the test infrastructure continues to use the same browsing context group for each test, we add a way to reset the browsing context group state so it appears as if you are getting a fresh browsing context group when you start a new test. The subtest failures of regression-1399759.https.sub.html are due to about:blank inheriting from the parent instead of the initiator of the navigation. That's a distinct issue from this new feature. going-back.sub.https.html is skipped because it times out on some bots at the history.back() step, while the feature-related assertions pass. I suspect it's related to the bfcache. I could not reproduce it locally (1000 iterations of the test as well as 10 full-directory runs). Canonical link: https://commits.webkit.org/314346@main

webkit-commit-queue · 2026-06-02T07:41:21Z

Committed 314346@main (b4c4069): https://commits.webkit.org/314346@main

Reviewed commits have been landed. Closing PR #66162 and removing active labels.

annevk self-assigned this Jun 1, 2026

annevk added the WebCore Misc. For miscellaneous bugs in the WebCore framework (and not JavaScriptCore or WebKit). label Jun 1, 2026

webkit-ews-buildbot added the merging-blocked Applied to prevent a change from being merged label Jun 1, 2026

annevk removed the merging-blocked Applied to prevent a change from being merged label Jun 1, 2026

annevk force-pushed the eng/Implement-the-Origin-Agent-Cluster-header branch from f5dd211 to a910c5e Compare June 1, 2026 11:09

annevk marked this pull request as ready for review June 1, 2026 12:39

annevk requested review from JonWBedard, cdumez and rniwa as code owners June 1, 2026 12:39

webkit-ews-buildbot added the merging-blocked Applied to prevent a change from being merged label Jun 1, 2026

annevk removed the merging-blocked Applied to prevent a change from being merged label Jun 1, 2026

achristensen07 reviewed Jun 1, 2026

View reviewed changes

webkit-ews-buildbot added the merging-blocked Applied to prevent a change from being merged label Jun 1, 2026

annevk removed the merging-blocked Applied to prevent a change from being merged label Jun 1, 2026

annevk force-pushed the eng/Implement-the-Origin-Agent-Cluster-header branch from a910c5e to 3cfe28c Compare June 1, 2026 15:46

webkit-ews-buildbot added the merging-blocked Applied to prevent a change from being merged label Jun 1, 2026

achristensen07 approved these changes Jun 2, 2026

View reviewed changes

annevk removed the merging-blocked Applied to prevent a change from being merged label Jun 2, 2026

annevk force-pushed the eng/Implement-the-Origin-Agent-Cluster-header branch from 3cfe28c to b06eacb Compare June 2, 2026 06:01

annevk added the merge-queue Applied to send a pull request to merge-queue label Jun 2, 2026

webkit-commit-queue force-pushed the eng/Implement-the-Origin-Agent-Cluster-header branch from b06eacb to b4c4069 Compare June 2, 2026 07:41

webkit-commit-queue merged commit b4c4069 into WebKit:main Jun 2, 2026

webkit-commit-queue removed the merge-queue Applied to send a pull request to merge-queue label Jun 2, 2026

annevk deleted the eng/Implement-the-Origin-Agent-Cluster-header branch June 2, 2026 07:41

		OriginAgentClusterPolicy isolatedCopy() const & { return { isOriginKeyed }; }
		OriginAgentClusterPolicy isolatedCopy() && { return { isOriginKeyed }; }

Conversation

annevk commented Jun 1, 2026 • edited by webkit-early-warning-system Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

webkit-early-warning-system commented Jun 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

webkit-ews-buildbot commented Jun 1, 2026

macOS Safer C++ Build #111505 (f5dd211)

Uh oh!

webkit-ews-buildbot commented Jun 1, 2026

iOS Safer C++ Build #30153 (f5dd211)

Uh oh!

webkit-early-warning-system commented Jun 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

annevk Jun 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

webkit-early-warning-system commented Jun 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

webkit-early-warning-system commented Jun 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

webkit-commit-queue commented Jun 2, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

annevk commented Jun 1, 2026 •

edited by webkit-early-warning-system

Loading

webkit-early-warning-system commented Jun 1, 2026 •

edited

Loading

macOS Safer C++ Build #111505 (`f5dd211`)

iOS Safer C++ Build #30153 (`f5dd211`)

webkit-early-warning-system commented Jun 1, 2026 •

edited

Loading

annevk Jun 1, 2026 •

edited

Loading

webkit-early-warning-system commented Jun 1, 2026 •

edited

Loading

webkit-early-warning-system commented Jun 2, 2026 •

edited

Loading