[JSC] Improve the performance of maps in VariableEnvironments by ast-hugger · Pull Request #60797 · WebKit/WebKit

ast-hugger · 2026-03-17T18:48:10Z

`577296d`

[JSC] Improve the performance of maps in VariableEnvironments
https://bugs.webkit.org/show_bug.cgi?id=310108
rdar://172755753

Reviewed by Yusuke Suzuki.

This patch improves the performance of JavaScript parser.
Key changes:

    * Source/WTF/wtf/InlineMap.h: Added.
    * Tools/TestWebKitAPI/Tests/WTF/InlineMap.cpp: Added.

Introduces a new map implementation, largely compatible with the existing
UncheckedKeyHashMap. The map stores entries inline using a flat array with linear lookup
when the number of entries is at or below the InlineCapacity parameter value.
This is implemented as a new class rather than a change to the existing
UncheckedKeyHashMap / HashMap / HashTable class group to keep the change localized.
Changing the base HashTable to allow linear inline storage might be an interesting
exercise, but it would be touching many use cases and is better done as a separate
project.

    * Source/JavaScriptCore/parser/VariableEnvironment.h:
    * Source/JavaScriptCore/runtime/CachedTypes.cpp:

Class VariableEnvironment is changed to use the new map for its bindings.

Testing: The new class comes with unit tests. The parser is covered by existing tests.
The patch changes 3 of them:

    * JSTests/stress/can-declare-global-var-invoked-before-any-binding-is-created-eval.js:
    * JSTests/stress/can-declare-global-var-invoked-before-any-binding-is-created-global.js:
    * JSTests/stress/eval-func-decl-in-global-of-eval.js:

These tests were relying on the specific iteration order of the original hash map by
assuming which of the two problematic bindings was visited and detected first. The purpose
of the tests is to verify that a problematic binding prevents valid bindings in the same
eval unit from taking effect.

In this context, it actually makes sense to only have one problematic binding in the test.
This both avoids the non-determinism of which binding is detected first, and reduces the
chances of a false negative (having the valid bindings not take effect simply because the
invalid one was processed first). The patch removes the second problematic definition in
each test.

Canonical link: https://commits.webkit.org/309506@main

092cd7b

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	✅ 🧪 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
	✅ 🛠 watch
	✅ 🛠 watch-sim

webkit-early-warning-system · 2026-03-17T18:48:17Z

EWS run on previous version of this PR (hash b47e68f)

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	~~🧪 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
	✅ 🛠 watch
	~~🛠 watch-sim~~

webkit-early-warning-system · 2026-03-17T20:07:25Z

EWS run on previous version of this PR (hash a34ae33)

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	~~🧪 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~~
	~~🛠 watch~~
	~~🛠 watch-sim~~

Constellation · 2026-03-17T18:54:50Z

Source/WTF/wtf/InlineMap.h

+                    --m_size;
+                    if (i != m_size) {
+                        // Move last entry to fill the gap
+                        new (&entryStorage[i]) Entry(WTF::move(entryStorage[m_size]));


Using std::construct_at if we use std::destroy_at?

Fixed, this one and the "ditto"s

Constellation · 2026-03-17T20:05:31Z

Source/WTF/wtf/InlineMap.h

+        // Hashed mode - mark as deleted
+        Entry* slot = findKeyOrEmpty(key);
+        if (isEmptyOrDeletedEntry(*slot))
+            return false;
+
+        std::destroy_at(slot);
+        constructDeletedEntry(*slot);
+        --m_size;
+        return true;


Remove needs to do some rehashing based on load status because we may have a hashtable which is filled with Deleted entries.

Good point. Let's address this as a follow-up polish PR.

Constellation · 2026-03-17T20:05:57Z

Source/WTF/wtf/InlineMap.h

+            if (wasInline || !isEmptyOrDeletedEntry(oldEntries[i])) {
+                Entry* slot = findKeyOrEmptyInStorage(oldEntries[i].key, newEntries, newCapacity);
+                ASSERT(isEmptyEntry(*slot));
+                new (slot) Entry(WTF::move(oldEntries[i]));


Constellation · 2026-03-17T20:06:24Z

Source/WTF/wtf/InlineMap.h

+            for (unsigned i = 0; i < m_size; ++i)
+                new (&inlineStorage()[i]) Entry(WTF::move(other.inlineStorage()[i]));


Constellation · 2026-03-17T20:06:34Z

Source/WTF/wtf/InlineMap.h

+
+            if (size < m_capacity) {
+                Entry* slot = &entryStorage[size];
+                new (slot) Entry(std::forward<K>(key), std::forward<V>(value));


Constellation · 2026-03-17T20:06:51Z

Source/WTF/wtf/InlineMap.h

+            if (isDeletedEntry(srcEntries[i]))
+                constructDeletedEntry(destEntries[i]);
+            else
+                new (&destEntries[i]) Entry(srcEntries[i]);


Constellation · 2026-03-17T20:07:09Z

Source/WTF/wtf/InlineMap.h

+    {
+        if (this != &other) {
+            std::destroy_at(this);
+            new (this) InlineMap(WTF::move(other));


Constellation · 2026-03-17T20:07:17Z

Source/WTF/wtf/InlineMap.h

+            return { iterator { slot, end }, false };
+
+        // Slot is either empty or deleted - we can insert here
+        new (slot) Entry(std::forward<K>(key), std::forward<V>(value));


Constellation · 2026-03-17T20:07:26Z

Source/WTF/wtf/InlineMap.h

+
+            // Construct the Entry directly with the deleted key and empty value
+            // This avoids constructing an empty Entry first and then overwriting it
+            new (NotNull, std::addressof(entry)) Entry(WTF::move(deletedKey), MappedTraits::emptyValue());


Constellation · 2026-03-17T20:07:32Z

Source/WTF/wtf/InlineMap.h

+        if constexpr (EntryTraits::emptyValueIsZero)
+            zeroBytes(entry);
+        else
+            new (NotNull, std::addressof(entry)) Entry(KeyTraits::emptyValue(), MappedTraits::emptyValue());


webkit-early-warning-system · 2026-03-17T20:22:30Z

EWS run on previous version of this PR (hash 03a722d)

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	❌ 🧪 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~~
	~~🛠 watch~~
	~~🛠 watch-sim~~

webkit-early-warning-system · 2026-03-17T20:53:25Z

EWS run on current version of this PR (hash 092cd7b)

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	✅ 🧪 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
	✅ 🛠 watch
	✅ 🛠 watch-sim

https://bugs.webkit.org/show_bug.cgi?id=310108 rdar://172755753 Reviewed by Yusuke Suzuki. This patch improves the performance of JavaScript parser. Key changes: * Source/WTF/wtf/InlineMap.h: Added. * Tools/TestWebKitAPI/Tests/WTF/InlineMap.cpp: Added. Introduces a new map implementation, largely compatible with the existing UncheckedKeyHashMap. The map stores entries inline using a flat array with linear lookup when the number of entries is at or below the InlineCapacity parameter value. This is implemented as a new class rather than a change to the existing UncheckedKeyHashMap / HashMap / HashTable class group to keep the change localized. Changing the base HashTable to allow linear inline storage might be an interesting exercise, but it would be touching many use cases and is better done as a separate project. * Source/JavaScriptCore/parser/VariableEnvironment.h: * Source/JavaScriptCore/runtime/CachedTypes.cpp: Class VariableEnvironment is changed to use the new map for its bindings. Testing: The new class comes with unit tests. The parser is covered by existing tests. The patch changes 3 of them: * JSTests/stress/can-declare-global-var-invoked-before-any-binding-is-created-eval.js: * JSTests/stress/can-declare-global-var-invoked-before-any-binding-is-created-global.js: * JSTests/stress/eval-func-decl-in-global-of-eval.js: These tests were relying on the specific iteration order of the original hash map by assuming which of the two problematic bindings was visited and detected first. The purpose of the tests is to verify that a problematic binding prevents valid bindings in the same eval unit from taking effect. In this context, it actually makes sense to only have one problematic binding in the test. This both avoids the non-determinism of which binding is detected first, and reduces the chances of a false negative (having the valid bindings not take effect simply because the invalid one was processed first). The patch removes the second problematic definition in each test. Canonical link: https://commits.webkit.org/309506@main

webkit-commit-queue · 2026-03-18T21:56:22Z

Committed 309506@main (577296d): https://commits.webkit.org/309506@main

Reviewed commits have been landed. Closing PR #60797 and removing active labels.

ast-hugger requested a review from a team as a code owner March 17, 2026 18:48

ast-hugger self-assigned this Mar 17, 2026

ast-hugger added the JavaScriptCore For bugs in JavaScriptCore, the JS engine used by WebKit, other than kxmlcore issues. label Mar 17, 2026

webkit-ews-buildbot added the merging-blocked Applied to prevent a change from being merged label Mar 17, 2026

ast-hugger removed the merging-blocked Applied to prevent a change from being merged label Mar 17, 2026

ast-hugger force-pushed the perf-inlinemap-squashed branch from b47e68f to a34ae33 Compare March 17, 2026 20:07

Constellation reviewed Mar 17, 2026

View reviewed changes

webkit-ews-buildbot added the merging-blocked Applied to prevent a change from being merged label Mar 17, 2026

ast-hugger removed the merging-blocked Applied to prevent a change from being merged label Mar 17, 2026

ast-hugger force-pushed the perf-inlinemap-squashed branch from a34ae33 to 03a722d Compare March 17, 2026 20:22

webkit-ews-buildbot added the merging-blocked Applied to prevent a change from being merged label Mar 17, 2026

ast-hugger removed the merging-blocked Applied to prevent a change from being merged label Mar 17, 2026

webkit-ews-buildbot added the merging-blocked Applied to prevent a change from being merged label Mar 17, 2026

ast-hugger removed the merging-blocked Applied to prevent a change from being merged label Mar 17, 2026

ast-hugger force-pushed the perf-inlinemap-squashed branch from 03a722d to 092cd7b Compare March 17, 2026 20:53

Constellation approved these changes Mar 18, 2026

View reviewed changes

ast-hugger added the merge-queue Applied to send a pull request to merge-queue label Mar 18, 2026

webkit-commit-queue force-pushed the perf-inlinemap-squashed branch from 092cd7b to 577296d Compare March 18, 2026 21:56

webkit-commit-queue merged commit 577296d into WebKit:main Mar 18, 2026

webkit-commit-queue removed the merge-queue Applied to send a pull request to merge-queue label Mar 18, 2026

		for (unsigned i = 0; i < m_size; ++i)
		new (&inlineStorage()[i]) Entry(WTF::move(other.inlineStorage()[i]));

Conversation

ast-hugger commented Mar 17, 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 Mar 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

webkit-early-warning-system commented Mar 17, 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!

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!

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!

webkit-early-warning-system commented Mar 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

webkit-early-warning-system commented Mar 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

webkit-commit-queue commented Mar 18, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

ast-hugger commented Mar 17, 2026 •

edited by webkit-early-warning-system

Loading

webkit-early-warning-system commented Mar 17, 2026 •

edited

Loading

webkit-early-warning-system commented Mar 17, 2026 •

edited

Loading

webkit-early-warning-system commented Mar 17, 2026 •

edited

Loading

webkit-early-warning-system commented Mar 17, 2026 •

edited

Loading