1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
|
From ce71348a09f6689dd01a68db64b172191d0182d8 Mon Sep 17 00:00:00 2001
From: Andrey Kosyakov <caseq@chromium.org>
Date: Thu, 21 Dec 2023 18:38:38 +0000
Subject: [PATCH] [bindings] Use v8::Array::Iterate for converting script
wrappables
This changes CreateIDLSequenceFromV8Array to use the new
v8::Array::Iterate() operation.
This speeds up the "execBundles" part of the microbenchmark
at crbug.com/dawn/1858 by around 3x.
This depends on crrev.com/c/4846594 landing (and rolling) first.
This is a slight re-work of https://crrev.com/c/4847447/3,
originally by jkummerow@chromium.org
Bug: v8:14218, dawn:1858, 1511239
Change-Id: Ia266556d05b4d53e6942e12609d1c08882b4ff0f
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5132129
Commit-Queue: Andrey Kosyakov <caseq@chromium.org>
Reviewed-by: Yuki Shiino <yukishiino@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1240236}
---
.../bindings/core/v8/native_value_traits.h | 6 ++
.../core/v8/native_value_traits_impl.h | 91 ++++++++++++++++++-
2 files changed, 95 insertions(+), 2 deletions(-)
diff --git a/third_party/blink/renderer/bindings/core/v8/native_value_traits.h b/third_party/blink/renderer/bindings/core/v8/native_value_traits.h
index 1e5a0790df6d..a5c28b37e945 100644
--- a/third_party/blink/renderer/bindings/core/v8/native_value_traits.h
+++ b/third_party/blink/renderer/bindings/core/v8/native_value_traits.h
@@ -84,6 +84,12 @@ struct NativeValueTraitsBase {
std::is_pointer_v<ImplType> ||
requires(ImplType value) { value.IsNull(); };
+ // This should only be true for certain subclasses of ScriptWrappable
+ // that satisfy the assumptions of CreateIDLSequenceFromV8ArraySlow() with
+ // regards to how NativeValue() is implemented for the underlying type.
+ static constexpr bool supports_scriptwrappable_specific_fast_array_iteration =
+ false;
+
template <typename... ExtraArgs>
static decltype(auto) ArgumentValue(v8::Isolate* isolate,
int argument_index,
diff --git a/third_party/blink/renderer/bindings/core/v8/native_value_traits_impl.h b/third_party/blink/renderer/bindings/core/v8/native_value_traits_impl.h
index 5011503dcf1c..f085b6e90516 100644
--- a/third_party/blink/renderer/bindings/core/v8/native_value_traits_impl.h
+++ b/third_party/blink/renderer/bindings/core/v8/native_value_traits_impl.h
@@ -1037,10 +1037,86 @@ CreateIDLSequenceFromV8ArraySlow(v8::Isolate* isolate,
return {};
}
- typename NativeValueTraits<IDLSequence<T>>::ImplType result;
+ using ResultType = typename NativeValueTraits<IDLSequence<T>>::ImplType;
+ ResultType result;
result.ReserveInitialCapacity(length);
v8::Local<v8::Context> current_context = isolate->GetCurrentContext();
v8::TryCatch try_block(isolate);
+
+ // Fast path -- we're creating a sequence of script wrappables, which can be
+ // done by directly getting underlying object as long as array types are
+ // homogeneous. With ScriptWrappables, we don't expect to enter JS during
+ // iteration, so we can rely on v8::Array::Iterate() which is much faster than
+ // iterating an array on the client side of the v8. Additionally, for most
+ // subsptyes of ScriptWrappables, we can speed up type checks (see more on
+ // that below next to supports_scriptwrappable_specific_fast_array_iteration
+ // check.
+ if constexpr (std::is_base_of_v<ScriptWrappable, T>) {
+ struct CallbackData {
+ STACK_ALLOCATED();
+
+ public:
+ v8::Isolate* isolate;
+ v8::TypecheckWitness witness;
+ ResultType& result;
+ ExceptionState& exception_state;
+ CallbackData(v8::Isolate* isolate,
+ ResultType& result,
+ ExceptionState& exception_state)
+ : isolate(isolate),
+ witness(isolate),
+ result(result),
+ exception_state(exception_state) {}
+ };
+
+ CallbackData callback_data(isolate, result, exception_state);
+ v8::Array::IterationCallback callback = [](uint32_t index,
+ v8::Local<v8::Value> v8_element,
+ void* data) {
+ CallbackData* callback_data = reinterpret_cast<CallbackData*>(data);
+ // 3.4. Initialize Si to the result of converting nextItem to an IDL value
+ // of type T.
+ v8::TypecheckWitness& witness = callback_data->witness;
+ // We can speed up type check by taking advantage of V8's type witness,
+ // provided traits' NativeValue implementation doesn't have additional
+ // logic beyond checking the type and calling ToScriptWrappable().
+ if constexpr (
+ NativeValueTraits<
+ T>::supports_scriptwrappable_specific_fast_array_iteration) {
+ if (witness.Matches(v8_element)) {
+ auto&& value = ToScriptWrappable(v8_element.As<v8::Object>())
+ ->template ToImpl<T>();
+ callback_data->result.push_back(std::move(value));
+ return v8::Array::CallbackResult::kContinue;
+ }
+ }
+ auto&& element = NativeValueTraits<T>::NativeValue(
+ callback_data->isolate, v8_element, callback_data->exception_state);
+ if (callback_data->exception_state.HadException()) {
+ // It doesn't matter whether we return `kException` or `kBreak` here,
+ // as that only affects the return value of `v8_array->Iterate()`,
+ // which we are ignoring.
+ return v8::Array::CallbackResult::kException;
+ }
+ if constexpr (
+ NativeValueTraits<
+ T>::supports_scriptwrappable_specific_fast_array_iteration) {
+ witness.Update(v8_element);
+ }
+ callback_data->result.push_back(std::move(element));
+ return v8::Array::CallbackResult::kContinue;
+ };
+ if (!v8_array->Iterate(current_context, callback, &callback_data)
+ .IsJust()) {
+ if (try_block.HasCaught()) {
+ exception_state.RethrowV8Exception(try_block.Exception());
+ }
+ DCHECK(exception_state.HadException());
+ return {};
+ }
+ return result;
+ }
+
// Array length may change if array is mutated during iteration.
for (uint32_t i = 0; i < v8_array->Length(); ++i) {
v8::Local<v8::Value> v8_element;
@@ -1056,6 +1132,7 @@ CreateIDLSequenceFromV8ArraySlow(v8::Isolate* isolate,
return {};
result.push_back(std::move(element));
}
+
// 3.2. If next is false, then return an IDL sequence value of type
// sequence<T> of length i, where the value of the element at index j is Sj.
return result;
@@ -1398,6 +1475,7 @@ struct NativeValueTraits<T> : public NativeValueTraitsBase<T*> {
}
};
+// Interface types
template <typename T>
requires std::derived_from<T, CallbackInterfaceBase>
struct NativeValueTraits<IDLNullable<T>>
@@ -1470,12 +1548,21 @@ struct NativeValueTraits<T> : public NativeValueTraitsBase<T> {
template <typename T>
requires std::derived_from<T, ScriptWrappable>
struct NativeValueTraits<T> : public NativeValueTraitsBase<T*> {
+ // This signifies that CreateIDLSequenceFromV8ArraySlow() may apply
+ // certain optimization based on assumptions about `NativeValue()`
+ // implementation below. For subclasses of ScriptWrappable that have
+ // different implementation of NativeValue(), this should remain false.
+ static constexpr bool supports_scriptwrappable_specific_fast_array_iteration =
+ true;
+
static inline T* NativeValue(v8::Isolate* isolate,
v8::Local<v8::Value> value,
ExceptionState& exception_state) {
const WrapperTypeInfo* wrapper_type_info = T::GetStaticWrapperTypeInfo();
- if (V8PerIsolateData::From(isolate)->HasInstance(wrapper_type_info, value))
+ if (V8PerIsolateData::From(isolate)->HasInstance(wrapper_type_info,
+ value)) {
return ToScriptWrappable(value.As<v8::Object>())->template ToImpl<T>();
+ }
bindings::NativeValueTraitsInterfaceNotOfType(wrapper_type_info,
exception_state);
|