script: Notify embedder coalesced events to resolve potential WebDriver deadlock by yezhizhen · Pull Request #43202 · servo/servo

yezhizhen · 2026-03-12T04:54:45Z

Notify embedder the coalesced wheel/mousemove event.

We also stop cloning every single entry in pending_input_event.

Testing: This should unblock the potential WebDriver thread deadlock
when receiving simultaneous wheel or pointer inputs from multiple sources.
We added a new test that stably deadlock on 74c16ff.
Firefox/Safari/Chrome all pass the new test.
Fixes: #43136 (comment)

github-actions · 2026-03-12T04:55:14Z

🔨 Triggering try run (#22987221590) for Linux (WPT)

github-actions · 2026-03-12T05:23:02Z

Test results for linux-wpt from try job (#22987221590):

Flaky unexpected result (21)

CRASH [expected PASS] /_mozilla/shadow-dom/move-element-with-ua-shadow-tree-crash.html (#39473)
CRASH [expected OK] /_webgl/conformance/extensions/oes-texture-half-float-with-image.html
OK /beacon/beacon-basic.https.window.html (#41723)
- FAIL [expected PASS] subtest: Payload size restriction should be accumulated: type = arraybuffer
```
assert_false: expected false got true
```
OK /css/css-cascade/layer-cssom-order-reverse.html (#36094)
- PASS [expected FAIL] subtest: Delete layer invalidates @font-face
OK /css/css-fonts/generic-family-keywords-002.html (#40929)
- FAIL [expected PASS] subtest: font-family: -webkit-serif treated as <font-family>, not <generic-name>
```
assert_equals: expected 30 but got 50
```
- FAIL [expected PASS] subtest: font-family: -webkit-sans-serif treated as <font-family>, not <generic-name>
```
assert_equals: expected 30 but got 50
```
- FAIL [expected PASS] subtest: font-family: -webkit-cursive treated as <font-family>, not <generic-name>
```
assert_equals: expected 30 but got 50
```
- FAIL [expected PASS] subtest: font-family: -webkit-fantasy treated as <font-family>, not <generic-name>
```
assert_equals: expected 30 but got 50
```
- FAIL [expected PASS] subtest: font-family: -webkit-monospace treated as <font-family>, not <generic-name>
```
assert_equals: expected 30 but got 50
```
- FAIL [expected PASS] subtest: font-family: -webkit-system-ui treated as <font-family>, not <generic-name>
```
assert_equals: expected 30 but got 50
```
- FAIL [expected PASS] subtest: font-family: -webkit-math treated as <font-family>, not <generic-name>
```
assert_equals: expected 30 but got 50
```
- PASS [expected FAIL] subtest: font-family: -webkit-generic(fangsong) treated as <font-family>, not <generic-name>
- PASS [expected FAIL] subtest: font-family: -webkit-generic(kai) treated as <font-family>, not <generic-name>
- PASS [expected FAIL] subtest: font-family: -webkit-generic(khmer-mul) treated as <font-family>, not <generic-name>
- And 12 more unexpected results...
OK /fetch/metadata/generated/css-font-face.https.sub.tentative.html (#32732)
- PASS [expected FAIL] subtest: sec-fetch-mode
OK /html/browsers/browsing-the-web/navigating-across-documents/navigation-unload-cross-origin.sub.window.html (#29056)
- PASS [expected FAIL] subtest: Cross-origin navigation started from unload handler must be ignored
OK /html/browsers/browsing-the-web/navigating-across-documents/navigation-unload-same-origin-fragment.html (#20768)
- FAIL [expected PASS] subtest: Tests that a fragment navigation in the unload handler will not block the initial navigation
```
assert_equals: expected "" but got "#fragment"
```
OK /html/browsers/history/the-history-interface/traverse_the_history_2.html (#21383)
- FAIL [expected PASS] subtest: Multiple history traversals, last would be aborted
```
assert_array_equals: Pages opened during history navigation expected property 1 to be 3 but got 2 (expected array [6, 3] got [6, 2])
```
CRASH [expected OK] /html/browsers/sandboxing/sandbox-initial-empty-document-toward-same-origin.html (#35948)
OK /html/semantics/document-metadata/the-meta-element/pragma-directives/attr-meta-http-equiv-refresh/allow-scripts-flag-changing-1.html (#39694)
- PASS [expected FAIL] subtest: Meta refresh is blocked by the allow-scripts sandbox flag at its creation time, not when refresh comes due
OK /html/semantics/document-metadata/the-meta-element/pragma-directives/attr-meta-http-equiv-refresh/allow-scripts-flag-changing-2.html (#39703)
- FAIL [expected PASS] subtest: Meta refresh of the original iframe is not blocked if moved into a sandboxed iframe
```
uncaught exception: Error: assert_unreached: The iframe into which the meta was moved must not refresh Reached unreachable code
```
TIMEOUT [expected OK] /html/semantics/embedded-content/the-iframe-element/iframe_sandbox_navigate_other_frame_popup.sub.html (#39702)
- TIMEOUT [expected FAIL] subtest: Sandboxed iframe can not navigate other frame's popup
```
Test timed out
```
TIMEOUT [expected OK] /html/user-activation/navigation-state-reset-sameorigin.html
- TIMEOUT [expected PASS] subtest: Post-navigation state reset.
```
Test timed out
```
OK /html/webappapis/dynamic-markup-insertion/document-write/module-tla-import.html (#42419)
- FAIL [expected PASS] subtest: document.write in an imported module
```
assert_true: onload must be called expected true got false
```
OK /visual-viewport/resize-event-order.html (#41981)
- PASS [expected FAIL] subtest: Popup: DOMWindow resize fired before VisualViewport.
OK /wasm/webapi/abort.any.html (#39966)
- FAIL [expected PASS] subtest: compileStreaming() asynchronously racing with abort should succeed or reject with AbortError
```
assert_equals: expected "AbortError" but got "CompileError"
```
CRASH [expected OK] /webaudio/the-audio-api/the-mediastreamaudiodestinationnode-interface/closed-audiocontext-construction.html
OK [expected TIMEOUT] /webstorage/localstorage-about-blank-3P-iframe-opens-3P-window.partitioned.html (#29053)
- PASS [expected TIMEOUT] subtest: StorageKey: test 3P about:blank window opened from a 3P iframe
OK /webxr/xrSession_features_deviceSupport.https.html (#24357)
- FAIL [expected PASS] subtest: Immersive XRSession requests with no supported device should reject
```
assert_unreached: Should have rejected: undefined Reached unreachable code
```
OK /xhr/send-redirect.htm (#32026)
- FAIL [expected PASS] subtest: XMLHttpRequest: send() - Redirects (basics) (302, GET, content.py)
```
assert_equals: expected (string) "GET" but got (object) null
```

Stable unexpected results that are known to be intermittent (15)

TIMEOUT /FileAPI/url/url-in-tags-revoke.window.html (#19978)
- TIMEOUT [expected PASS] subtest: Fetching a blob URL immediately before revoking it works in <script> tags.
```
Test timed out
```
OK /FileAPI/url/url-with-fetch.any.worker.html (#21517)
- FAIL [expected PASS] subtest: Revoke blob URL after calling fetch, fetch should succeed
```
promise_test: Unhandled rejection with value: object "TypeError: Network error: Blob URL store error: InvalidFileID"
```
FAIL [expected PASS] /_mozilla/mozilla/sslfail.html (#10760)
TIMEOUT [expected OK] /_mozilla/mozilla/window_resize_event.html (#36741)
- TIMEOUT [expected PASS] subtest: Popup onresize event fires after resizeTo
```
Test timed out
```
OK /content-security-policy/frame-ancestors/frame-ancestors-path-ignored.window.html (#36468)
- FAIL [expected PASS] subtest: A 'frame-ancestors' CSP directive with a URL that includes a path should be ignored.
```
assert_unreached: The IFrame should have been blocked (or cross-origin). It wasn't. Reached unreachable code
```
OK /css/css-fonts/generic-family-keywords-001.html (#37467)
- FAIL [expected PASS] subtest: @font-face matching for quoted and unquoted generic(kai)
```
assert_equals: quoted generic(kai) matches  @font-face rule expected 50 but got 30
```
OK /css/css-fonts/generic-family-keywords-003.html (#38994)
- PASS [expected FAIL] subtest: @font-face matching for quoted and unquoted cursive (drawing text in a canvas)
- PASS [expected FAIL] subtest: @font-face matching for quoted and unquoted monospace (drawing text in a canvas)
- PASS [expected FAIL] subtest: @font-face matching for quoted and unquoted system-ui (drawing text in a canvas)
- PASS [expected FAIL] subtest: @font-face matching for quoted and unquoted generic(fangsong) (drawing text in a canvas)
- PASS [expected FAIL] subtest: @font-face matching for quoted and unquoted generic(khmer-mul) (drawing text in a canvas)
- PASS [expected FAIL] subtest: @font-face matching for quoted and unquoted ui-serif (drawing text in a canvas)
OK [expected TIMEOUT] /html/anonymous-iframe/indexeddb.tentative.https.window.html (#39254)
- FAIL [expected TIMEOUT] subtest: indexeddb
```
assert_equals: expected (undefined) undefined but got (string) "ef3fff34-7b29-4496-a774-e9a9ec31ac1a"
```
OK /html/browsers/browsing-the-web/navigating-across-documents/replace-before-load/a-click.html (#28697)
- PASS [expected FAIL] subtest: aElement.click() before the load event must NOT replace
OK /html/browsers/history/the-history-interface/traverse_the_history_5.html (#21383)
- FAIL [expected PASS] subtest: Multiple history traversals, last would be aborted
```
assert_array_equals: Pages opened during history navigation expected property 1 to be 5 but got 3 (expected array [6, 5] got [6, 3])
```
TIMEOUT /html/interaction/focus/the-autofocus-attribute/supported-elements.html (#24145)
- TIMEOUT [expected PASS] subtest: Non-HTMLElement should not support autofocus
```
Test timed out
```
- NOTRUN [expected TIMEOUT] subtest: Host element with delegatesFocus should support autofocus
TIMEOUT /html/semantics/embedded-content/media-elements/preserves-pitch.html (#40352)
- PASS [expected TIMEOUT] subtest: Speed-ups should not change the pitch when preservesPitch=true
- PASS [expected NOTRUN] subtest: Slow-downs should not change the pitch when preservesPitch=true
- TIMEOUT [expected NOTRUN] subtest: Speed-ups should change the pitch when preservesPitch=false
```
Test timed out
```
OK /navigation-timing/test-navigation-type-reload.html (#33334)
- FAIL [expected PASS] subtest: Reload domComplete > Original domComplete
```
assert_true: Reload domComplete &gt; Original domComplete expected true got false
```
- FAIL [expected PASS] subtest: Reload domInteractive > Original domInteractive
```
assert_true: Reload domInteractive &gt; Original domInteractive expected true got false
```
- FAIL [expected PASS] subtest: Reload fetchStart > Original fetchStart
```
assert_true: Reload fetchStart &gt; Original fetchStart expected true got false
```
- FAIL [expected PASS] subtest: Reload loadEventEnd > Original loadEventEnd
```
assert_true: Reload loadEventEnd &gt; Original loadEventEnd expected true got false
```
- FAIL [expected PASS] subtest: Reload loadEventStart > Original loadEventStart
```
assert_true: Reload loadEventStart &gt; Original loadEventStart expected true got false
```
OK /resource-timing/buffer-full-add-then-clear.html (#40819)
- FAIL [expected PASS] subtest: Test that if the buffer is cleared after entries were added to the secondary buffer, those entries make it into the primary one
```
assert_equals: Number of entries does not match the expected value. expected 3 but got 0
```

OK /webaudio/the-audio-api/the-audiobuffersourcenode-interface/sub-sample-buffer-stitching.html (#22849)

FAIL [expected PASS] subtest: buffer-stitching-1

assert_approx_equals: Stitched sine‑wave buffers at sample rate 44100 sample[16770] |2.8768211046036067e-24 - 0.9055536389350891| = 0.9055536389350891 &gt; 0.000090957 expected 0.9055536389350891 +/- 0.000090957 but got 2.8768211046036067e-24

github-actions · 2026-03-12T05:23:09Z

✨ Try run (#22987221590) succeeded.

yezhizhen · 2026-03-12T06:08:22Z

Testing: ?. This should unblock the potential WebDriver thread deadlock,
when we have two sources of wheel/pointer input.

We can add a test for this that stably deadlocks.

servo-wpt-sync · 2026-03-12T06:30:03Z

🤖 Opened new upstream WPT pull request (web-platform-tests/wpt#58445) with upstreamable changes.

servo-wpt-sync · 2026-03-12T06:46:59Z

✍ Updated existing upstream WPT pull request (web-platform-tests/wpt#58445) title and body.

servo-wpt-sync · 2026-03-12T07:50:34Z

📝 Transplanted new upstreamable changes to existing upstream WPT pull request (web-platform-tests/wpt#58445).

components/shared/embedder/input_events.rs

servo-wpt-sync · 2026-03-12T09:21:21Z

📝 Transplanted new upstreamable changes to existing upstream WPT pull request (web-platform-tests/wpt#58445).

Signed-off-by: Euclid Ye <[email protected]>

servo-wpt-sync · 2026-03-12T09:43:48Z

📝 Transplanted new upstreamable changes to existing upstream WPT pull request (web-platform-tests/wpt#58445).

servo-wpt-sync · 2026-03-12T09:45:05Z

✍ Updated existing upstream WPT pull request (web-platform-tests/wpt#58445) title and body.

components/script/dom/document_event_handler.rs

Signed-off-by: Euclid Ye <[email protected]>

servo-wpt-sync · 2026-03-12T10:31:14Z

📝 Transplanted new upstreamable changes to existing upstream WPT pull request (web-platform-tests/wpt#58445).

servo-wpt-sync · 2026-03-12T10:32:32Z

✍ Updated existing upstream WPT pull request (web-platform-tests/wpt#58445) title and body.

mrobinson · 2026-03-12T10:36:13Z

components/script/dom/document_event_handler.rs

+    coalesced_move_event_ids: DomRefCell<Vec<InputEventId>>,
    /// The index of the last wheel event in the pending input events queue.
+    /// This is non-standard behaviour.
+    /// According to <https://www.w3.org/TR/pointerevents/#dfn-coalesced-events>,


Oh, I had no idea this was in the specification. I guess we are going to have to expose these events somehow.

Yeah. These are supposed to be a list of PointerEvents, accessed through getCoalescedEvents() in the fired PointerEvent.
We need to store extra events tho.

Personally, I hope we don't rush this for performance consideration :|

I guess it is needed for things like shooting game/drawing.. Which should not be our current concern.

Signed-off-by: Euclid Ye <[email protected]>

servo-wpt-sync · 2026-03-12T12:30:38Z

📝 Transplanted new upstreamable changes to existing upstream WPT pull request (web-platform-tests/wpt#58445).

This is a hot path. This way we free the unused memory earlier. Didn't realize the difference in #43202. Testing: This is a memory optimization that does not change visible behaviour. Signed-off-by: Euclid Ye <[email protected]>

…43262) This is now possible since #43202. We remove the workaround introduced long ago. Testing: Before #43202, this would deadlock in many tests, and is therefore covered by existing tests. [Try](https://github.com/servo/servo/actions/runs/23080857516) --------- Signed-off-by: Euclid Ye <[email protected]>

yezhizhen requested a review from xiaochengh March 12, 2026 04:54

yezhizhen requested a review from gterzian as a code owner March 12, 2026 04:54

servo-highfive added the S-awaiting-review There is new code that needs to be reviewed. label Mar 12, 2026

yezhizhen added the T-linux-wpt Do a try run of the WPT label Mar 12, 2026

github-actions bot removed the T-linux-wpt Do a try run of the WPT label Mar 12, 2026

yezhizhen marked this pull request as draft March 12, 2026 06:08

yezhizhen marked this pull request as ready for review March 12, 2026 06:21

yezhizhen closed this Mar 12, 2026

yezhizhen reopened this Mar 12, 2026

yezhizhen closed this Mar 12, 2026

yezhizhen reopened this Mar 12, 2026

servo-wpt-sync mentioned this pull request Mar 12, 2026

script: Notify embedder coalesced events to resolve potential WebDriver deadlock web-platform-tests/wpt#58445

Merged

yezhizhen force-pushed the feedback-to-coalesced branch from a419db4 to 2b8a008 Compare March 12, 2026 07:46

mrobinson reviewed Mar 12, 2026

View reviewed changes

components/shared/embedder/input_events.rs Outdated Show resolved Hide resolved

yezhizhen requested a review from mrobinson March 12, 2026 09:17

yezhizhen added 2 commits March 12, 2026 17:39

.

b02a0d4

Signed-off-by: Euclid Ye <[email protected]>

Simultaneous wheel events in different directions should not deadlock

31cb2d6

Signed-off-by: Euclid Ye <[email protected]>

yezhizhen force-pushed the feedback-to-coalesced branch from 0c34f92 to ba5b4a6 Compare March 12, 2026 09:39

mrobinson reviewed Mar 12, 2026

View reviewed changes

components/script/dom/document_event_handler.rs Outdated Show resolved Hide resolved

Simplify code: same result

f729e2e

Signed-off-by: Euclid Ye <[email protected]>

yezhizhen force-pushed the feedback-to-coalesced branch from ba5b4a6 to f729e2e Compare March 12, 2026 10:27

mrobinson approved these changes Mar 12, 2026

View reviewed changes

servo-highfive removed the S-awaiting-review There is new code that needs to be reviewed. label Mar 12, 2026

yezhizhen added this pull request to the merge queue Mar 12, 2026

servo-highfive added the S-awaiting-merge The PR is in the process of compiling and running tests on the automated CI. label Mar 12, 2026

yezhizhen removed this pull request from the merge queue due to a manual request Mar 12, 2026

servo-highfive removed the S-awaiting-merge The PR is in the process of compiling and running tests on the automated CI. label Mar 12, 2026

Remove redundant parameter

8ed102f

Signed-off-by: Euclid Ye <[email protected]>

servo-highfive added the S-awaiting-review There is new code that needs to be reviewed. label Mar 12, 2026

yezhizhen enabled auto-merge March 12, 2026 12:39

yezhizhen added this pull request to the merge queue Mar 12, 2026

servo-highfive added the S-awaiting-merge The PR is in the process of compiling and running tests on the automated CI. label Mar 12, 2026

Merged via the queue into servo:main with commit e88c5dd Mar 12, 2026
33 checks passed

yezhizhen deleted the feedback-to-coalesced branch March 12, 2026 14:07

servo-highfive removed the S-awaiting-merge The PR is in the process of compiling and running tests on the automated CI. label Mar 12, 2026

yezhizhen mentioned this pull request Mar 13, 2026

script: Free memory earlier in hot path of input handling #43226

Merged

yezhizhen mentioned this pull request Mar 14, 2026

webdriver: Always wait for mousemove/wheel in a blocking manner #43262

Merged

Uh oh!

Conversation

yezhizhen commented Mar 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions bot commented Mar 12, 2026

Uh oh!

github-actions bot commented Mar 12, 2026

Uh oh!

github-actions bot commented Mar 12, 2026

Uh oh!

yezhizhen commented Mar 12, 2026

Uh oh!

servo-wpt-sync commented Mar 12, 2026

Uh oh!

servo-wpt-sync commented Mar 12, 2026

Uh oh!

servo-wpt-sync commented Mar 12, 2026

Uh oh!

Uh oh!

servo-wpt-sync commented Mar 12, 2026

Uh oh!

servo-wpt-sync commented Mar 12, 2026

Uh oh!

servo-wpt-sync commented Mar 12, 2026

Uh oh!

Uh oh!

servo-wpt-sync commented Mar 12, 2026

Uh oh!

servo-wpt-sync commented Mar 12, 2026

Uh oh!

mrobinson Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

yezhizhen Mar 12, 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!

yezhizhen Mar 12, 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!

servo-wpt-sync commented Mar 12, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

yezhizhen commented Mar 12, 2026 •

edited

Loading

yezhizhen Mar 12, 2026 •

edited

Loading

yezhizhen Mar 12, 2026 •

edited

Loading