Allowing simd acceleration and added testcases

Author: Narfinger

mozjs/src/conversions.rs

@@ -526,6 +526,57 @@ impl FromJSValConvertible for f64 {
     }
 }
 
+#[cfg(all(
+    any(target_arch = "x86", target_arch = "x86_64"),
+    any(target_feature = "avx", target_feature = "sse2")
+))]
+/// Copies chars to the string using simd instructions
+pub unsafe fn fast_copy(chars: &[u8]) -> String {
+    use std::arch::x86_64;
+
+    let mut s = String::with_capacity(chars.len());
+    let mut count = 0;
+    let num_iter = chars.len() / 32;
+    let v = s.as_mut_vec();
+    (0..num_iter).for_each(|i| {
+        if cfg!(target_feature = "avx") {
+            let simd =
+                x86_64::_mm256_loadu_si256(chars.as_ptr().add(i * 32) as *const x86_64::__m256i);
+            x86_64::_mm256_storeu_si256(v.as_ptr().add(i * 32) as *mut x86_64::__m256i, simd);
+            count += 32;
+        } else {
+            let simd = x86_64::_mm_load_si128(chars.as_ptr().add(i * 16) as *const x86_64::__m128i);
+            x86_64::_mm_store_si128(v.as_ptr().add(i * 16) as *mut x86_64::__m128i, simd);
+            count += 16;
+        }
+    });
+
+    // bytes that do not fit into the simd instruction
+    for i in count..chars.len() {
+        chars.as_ptr().add(i).copy_to(v.as_mut_ptr().add(i), 1);
+        count += 1;
+    }
+    v.set_len(count);
+    s
+}
+
+#[cfg(not(any(
+    all(
+        any(target_arch = "x86", target_arch = "x86_64"),
+        any(target_feature = "avx", target_feature = "sse"),
+    ),
+    all(any(target_arch = "arm", target_arch = "aarch64"))
+)))]
+/// Copies chars to the string using a slower method instructions
+pub unsafe fn fast_copy(chars: &[u8]) -> String {
+    let mut v = Vec::with_capacity(chars.len() * 2);
+    v.set_len(chars.len() * 2);
+    let real_size = encoding_rs::mem::convert_latin1_to_utf8(chars, v.as_mut_slice());
+
+    v.truncate(real_size);
+    String::from_utf8_unchecked(v)
+}
+
 /// Converts a `JSString`, encoded in "Latin1" (i.e. U+0000-U+00FF encoded as 0x00-0xFF) into a
 /// `String`.
 pub unsafe fn latin1_to_string(cx: *mut JSContext, s: *mut JSString) -> String {
@@ -536,12 +587,7 @@ pub unsafe fn latin1_to_string(cx: *mut JSContext, s: *mut JSString) -> String {
     assert!(!chars.is_null());
 
     let chars = slice::from_raw_parts(chars, length as usize);
-    let mut v = Vec::with_capacity(length * 2);
-    v.set_len(length * 2);
-
-    let real_size = encoding_rs::mem::convert_latin1_to_utf8(chars, v.as_mut_slice());
-    v.truncate(real_size);
-    String::from_utf8_unchecked(v)
+    fast_copy(chars)
 }
 
 /// Converts a `JSString` into a `String`, regardless of used encoding.

mozjs/tests/external_string.rs

@@ -36,21 +36,13 @@ fn external_string() {
         ));
         let _ac = JSAutoRealm::new(context, global.get());
 
-        let latin1_base = "test latin-1";
-        let latin1_boxed = latin1_base.as_bytes().to_vec().into_boxed_slice();
-        let latin1_chars = Box::into_raw(latin1_boxed).cast::<u8>();
-
-        let callbacks = CreateJSExternalStringCallbacks(
-            &EXTERNAL_STRING_CALLBACKS_TRAPS,
-            latin1_base.len() as *mut c_void,
-        );
-        rooted!(in(context) let latin1_jsstr = JS_NewExternalStringLatin1(
-            context,
-            latin1_chars,
-            latin1_base.len(),
-            callbacks
-        ));
-        assert_eq!(jsstr_to_string(context, latin1_jsstr.get()), latin1_base);
+        test_latin1_string(context, "test latin1");
+        test_latin1_string(context, "abcdefghijklmnop"); // exactly 16 bytes
+        test_latin1_string(context, "abcdefghijklmnopq"); // 18 bytes
+        test_latin1_string(context, "abcdefghijklmno"); // 15 bytes
+        test_latin1_string(context, "abcdefghijklmnopqrstuvwxyzabcdef"); //32 bytes
+        test_latin1_string(context, "abcdefghijklmnopqrstuvwxyzabcde"); //31 bytes
+        test_latin1_string(context, "abcdefghijklmnopqrstuvwxyzabcdefg"); //33 bytes
 
         let utf16_base = "test utf-16 $€ \u{10437}\u{24B62}";
         let utf16_boxed = utf16_base
@@ -74,6 +66,24 @@ fn external_string() {
     }
 }
 
+#[cfg(test)]
+unsafe fn test_latin1_string(context: *mut mozjs::jsapi::JSContext, latin1_base: &str) {
+    let latin1_boxed = latin1_base.as_bytes().to_vec().into_boxed_slice();
+    let latin1_chars = Box::into_raw(latin1_boxed).cast::<u8>();
+
+    let callbacks = CreateJSExternalStringCallbacks(
+        &EXTERNAL_STRING_CALLBACKS_TRAPS,
+        latin1_base.len() as *mut c_void,
+    );
+    rooted!(in(context) let latin1_jsstr = JS_NewExternalStringLatin1(
+        context,
+        latin1_chars,
+        latin1_base.len(),
+        callbacks
+    ));
+    assert_eq!(jsstr_to_string(context, latin1_jsstr.get()), latin1_base);
+}
+
 static EXTERNAL_STRING_CALLBACKS_TRAPS: JSExternalStringCallbacksTraps =
     JSExternalStringCallbacksTraps {
         latin1Finalize: Some(latin1::finalize),