deps: V8: cherry-pick 64b36b441179

Dan Carney · aduh95 · commit f67161c90569 · 2026-07-21T15:40:49.000+02:00
Original commit message: optimize ascii fast path in WriteUtf8V2 Change-Id: If28168cb4395b953d0ec642ef4fc618ce963dbcd Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/7124103 Reviewed-by: Toon Verwaest <verwaest@chromium.org> Commit-Queue: Erik Corry <erikcorry@chromium.org> Reviewed-by: Erik Corry <erikcorry@chromium.org> Cr-Commit-Position: refs/heads/main@{#103542} Refs: v8/v8@64b36b4 PR-URL: #61712 Backport-PR-URL: #64630 Reviewed-By: Michaël Zasso <targos@protonmail.com>
diff --git a/common.gypi b/common.gypi
@@ -41,7 +41,7 @@
 
     # Reset this number to 0 on major V8 upgrades.
     # Increment by one for each non-official patch applied to deps/v8.
-    'v8_embedder_string': '-node.50',
+    'v8_embedder_string': '-node.51',
 
     ##### V8 defaults for Node.js #####
 
diff --git a/deps/v8/src/strings/unicode-inl.h b/deps/v8/src/strings/unicode-inl.h
@@ -10,6 +10,7 @@
 
 #include "src/base/logging.h"
 #include "src/utils/utils.h"
+#include "third_party/simdutf/simdutf.h"
 
 namespace unibrow {
 
@@ -219,6 +220,16 @@ bool Utf8::IsValidCharacter(uchar c) {
           c != kBadChar);
 }
 
+template <>
+bool Utf8::IsAsciiOneByteString<uint8_t>(const uint8_t* buffer, size_t size) {
+  return simdutf::validate_ascii(reinterpret_cast<const char*>(buffer), size);
+}
+
+template <>
+bool Utf8::IsAsciiOneByteString<uint16_t>(const uint16_t* buffer, size_t size) {
+  return false;
+}
+
 template <typename Char>
 Utf8::EncodingResult Utf8::Encode(v8::base::Vector<const Char> string,
                                   char* buffer, size_t capacity,
@@ -234,8 +245,17 @@ Utf8::EncodingResult Utf8::Encode(v8::base::Vector<const Char> string,
   const Char* characters = string.begin();
   size_t content_capacity = capacity - write_null;
   CHECK_LE(content_capacity, capacity);
-  uint16_t last = Utf16::kNoPreviousCharacter;
   size_t read_index = 0;
+  if (kSourceIsOneByte) {
+    size_t writeable = std::min(string.size(), content_capacity);
+    // Just memcpy when possible.
+    if (writeable > 0 && Utf8::IsAsciiOneByteString(characters, writeable)) {
+      memcpy(buffer, characters, writeable);
+      read_index = writeable;
+      write_index = writeable;
+    }
+  }
+  uint16_t last = Utf16::kNoPreviousCharacter;
   for (; read_index < string.size(); read_index++) {
     Char character = characters[read_index];
 
diff --git a/deps/v8/src/strings/unicode.h b/deps/v8/src/strings/unicode.h
@@ -212,6 +212,9 @@ class V8_EXPORT_PRIVATE Utf8 {
   // - valid code point range.
   static bool ValidateEncoding(const uint8_t* str, size_t length);
 
+  template <typename Char>
+  static bool IsAsciiOneByteString(const Char* buffer, size_t size);
+
   // Encode the given characters as Utf8 into the provided output buffer.
   struct EncodingResult {
     size_t bytes_written;
@@ -223,6 +226,14 @@ class V8_EXPORT_PRIVATE Utf8 {
                                bool replace_invalid_utf8);
 };
 
+template <>
+inline bool Utf8::IsAsciiOneByteString<uint8_t>(const uint8_t* buffer,
+                                                 size_t size);
+
+template <>
+inline bool Utf8::IsAsciiOneByteString<uint16_t>(const uint16_t* buffer,
+                                                  size_t size);
+
 #if V8_ENABLE_WEBASSEMBLY
 class V8_EXPORT_PRIVATE Wtf8 {
  public:
