style: Allow to use ThinVec/nsTArray in the style crate

This allows to clean-up the previous patches by using a single ThinVec
(which stores length / capacity along with the allocation).

Differential Revision: https://phabricator.services.mozilla.com/D175029

9 files changed, 110 insertions, 48 deletions

diff --git a/Cargo.lock b/Cargo.lock
index e9073d5f982..05d71cf2a12 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -3396,6 +3396,7 @@ dependencies = [
  "smallbitvec",
  "smallvec",
  "string_cache",
+ "thin-vec",
  "time 0.1.45",
  "tokio",
  "url",
@@ -5711,6 +5712,7 @@ dependencies = [
  "string_cache",
  "style_derive",
  "style_traits",
+ "thin-vec",
  "time 0.1.45",
  "to_shmem",
  "to_shmem_derive",
@@ -5925,6 +5927,12 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "222a222a5bfe1bba4a77b45ec488a741b3cb8872e5e499451fd7d0129c9c7c3d"
 
 [[package]]
+name = "thin-vec"
+version = "0.2.12"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "aac81b6fd6beb5884b0cf3321b8117e6e5d47ecb6fc89f414cfdcca8b2fe2dd8"
+
+[[package]]
 name = "thiserror"
 version = "1.0.50"
 source = "registry+https://github.com/rust-lang/crates.io-index"
@@ -6060,6 +6068,7 @@ dependencies = [
  "smallbitvec",
  "smallvec",
  "string_cache",
+ "thin-vec",
 ]
 
 [[package]]
diff --git a/components/malloc_size_of/Cargo.toml b/components/malloc_size_of/Cargo.toml
index feb219f3c83..e1ffcf1c4dd 100644
--- a/components/malloc_size_of/Cargo.toml
+++ b/components/malloc_size_of/Cargo.toml
@@ -45,6 +45,7 @@ servo_arc = { path = "../servo_arc" }
 smallbitvec = { workspace = true }
 smallvec = { workspace = true }
 string_cache = { workspace = true, optional = true }
+thin-vec = { version = "0.2.1", features = ["gecko-ffi"] }
 time = { workspace = true, optional = true }
 tokio = { workspace = true }
 url = { workspace = true, optional = true }
diff --git a/components/malloc_size_of/lib.rs b/components/malloc_size_of/lib.rs
index d79375bf96b..2c064f1ebe2 100644
--- a/components/malloc_size_of/lib.rs
+++ b/components/malloc_size_of/lib.rs
@@ -396,6 +396,28 @@ where
     }
 }
 
+impl<T> MallocShallowSizeOf for thin_vec::ThinVec<T> {
+    fn shallow_size_of(&self, ops: &mut MallocSizeOfOps) -> usize {
+        if self.capacity() == 0 {
+            // If it's the singleton we might not be a heap pointer.
+            return 0;
+        }
+
+        assert_eq!(std::mem::size_of::<Self>(), std::mem::size_of::<*const ()>());
+        unsafe { ops.malloc_size_of(*(self as *const Self as *const *const ())) }
+    }
+}
+
+impl<T: MallocSizeOf> MallocSizeOf for thin_vec::ThinVec<T> {
+    fn size_of(&self, ops: &mut MallocSizeOfOps) -> usize {
+        let mut n = self.shallow_size_of(ops);
+        for elem in self.iter() {
+            n += elem.size_of(ops);
+        }
+        n
+    }
+}
+
 macro_rules! malloc_size_of_hash_set {
     ($ty:ty) => {
         impl<T, S> MallocShallowSizeOf for $ty
diff --git a/components/style/Cargo.toml b/components/style/Cargo.toml
index 802aefb988d..3a6ac07539b 100644
--- a/components/style/Cargo.toml
+++ b/components/style/Cargo.toml
@@ -77,6 +77,7 @@ string_cache = { version = "0.8", optional = true }
 style_derive = { path = "../style_derive" }
 style_traits = { workspace = true }
 time = "0.1"
+thin-vec = { version = "0.2.1", features = ["gecko-ffi"] }
 to_shmem = { path = "../to_shmem" }
 to_shmem_derive = { path = "../to_shmem_derive" }
 uluru = "3.0"
diff --git a/components/style/gecko/pseudo_element_definition.mako.rs b/components/style/gecko/pseudo_element_definition.mako.rs
index b7f8cca409c..73e7893c998 100644
--- a/components/style/gecko/pseudo_element_definition.mako.rs
+++ b/components/style/gecko/pseudo_element_definition.mako.rs
@@ -11,7 +11,7 @@ pub enum PseudoElement {
     % for pseudo in PSEUDOS:
         /// ${pseudo.value}
         % if pseudo.is_tree_pseudo_element():
-        ${pseudo.capitalized_pseudo()}(Box<Box<[Atom]>>),
+        ${pseudo.capitalized_pseudo()}(thin_vec::ThinVec<Atom>),
         % elif pseudo.pseudo_ident == "highlight":
         ${pseudo.capitalized_pseudo()}(AtomIdent),
         % else:
@@ -210,7 +210,7 @@ impl PseudoElement {
             },
             _ => {
                 if starts_with_ignore_ascii_case(name, "-moz-tree-") {
-                    return PseudoElement::tree_pseudo_element(name, Box::new([]))
+                    return PseudoElement::tree_pseudo_element(name, Default::default())
                 }
                 const WEBKIT_PREFIX: &str = "-webkit-";
                 if allow_unkown_webkit && starts_with_ignore_ascii_case(name, WEBKIT_PREFIX) {
@@ -228,12 +228,12 @@ impl PseudoElement {
     ///
     /// Returns `None` if the pseudo-element is not recognized.
     #[inline]
-    pub fn tree_pseudo_element(name: &str, args: Box<[Atom]>) -> Option<Self> {
+    pub fn tree_pseudo_element(name: &str, args: thin_vec::ThinVec<Atom>) -> Option<Self> {
         debug_assert!(starts_with_ignore_ascii_case(name, "-moz-tree-"));
         let tree_part = &name[10..];
         % for pseudo in TREE_PSEUDOS:
             if tree_part.eq_ignore_ascii_case("${pseudo.value[11:]}") {
-                return Some(${pseudo_element_variant(pseudo, "args.into()")});
+                return Some(${pseudo_element_variant(pseudo, "args")});
             }
         % endfor
         None
diff --git a/components/style/gecko/selector_parser.rs b/components/style/gecko/selector_parser.rs
index 2beaaa238d2..db1fda493c4 100644
--- a/components/style/gecko/selector_parser.rs
+++ b/components/style/gecko/selector_parser.rs
@@ -19,7 +19,8 @@ use dom::{DocumentState, ElementState};
 use selectors::parser::SelectorParseErrorKind;
 use selectors::SelectorList;
 use std::fmt;
-use style_traits::{Comma, CssWriter, OneOrMoreSeparated, ParseError, StyleParseErrorKind, ToCss as ToCss_};
+use style_traits::{CssWriter, ParseError, StyleParseErrorKind, ToCss as ToCss_};
+use thin_vec::ThinVec;
 
 pub use crate::gecko::pseudo_element::{
     PseudoElement, EAGER_PSEUDOS, EAGER_PSEUDO_COUNT, PSEUDO_COUNT,
@@ -38,13 +39,9 @@ bitflags! {
 }
 
 /// The type used to store the language argument to the `:lang` pseudo-class.
-#[derive(Clone, Debug, Eq, MallocSizeOf, PartialEq, ToShmem)]
-pub enum Lang {
-    /// A single language code.
-    Single(AtomIdent),
-    /// A list of language codes.
-    List(Box<Vec<AtomIdent>>),
-}
+#[derive(Clone, Debug, Eq, MallocSizeOf, PartialEq, ToCss, ToShmem)]
+#[css(comma)]
+pub struct Lang(#[css(iterable)] pub ThinVec<AtomIdent>);
 
 macro_rules! pseudo_class_name {
     ([$(($css:expr, $name:ident, $state:tt, $flags:tt),)*]) => {
@@ -66,10 +63,6 @@ macro_rules! pseudo_class_name {
 }
 apply_non_ts_list!(pseudo_class_name);
 
-impl OneOrMoreSeparated for AtomIdent {
-    type S = Comma;
-}
-
 impl ToCss for NonTSPseudoClass {
     fn to_css<W>(&self, dest: &mut W) -> fmt::Result
     where
@@ -79,12 +72,9 @@ impl ToCss for NonTSPseudoClass {
             ([$(($css:expr, $name:ident, $state:tt, $flags:tt),)*]) => {
                 match *self {
                     $(NonTSPseudoClass::$name => concat!(":", $css),)*
-                    NonTSPseudoClass::Lang(ref s) => {
+                    NonTSPseudoClass::Lang(ref lang) => {
                         dest.write_str(":lang(")?;
-                        match &s {
-                            Lang::Single(lang) => cssparser::ToCss::to_css(lang, dest)?,
-                            Lang::List(list) => list.to_css(&mut CssWriter::new(dest))?,
-                        }
+                        lang.to_css(&mut CssWriter::new(dest))?;
                         return dest.write_char(')');
                     },
                     NonTSPseudoClass::MozLocaleDir(ref dir) => {
@@ -394,11 +384,7 @@ impl<'a, 'i> ::selectors::Parser<'i> for SelectorParser<'a> {
                 if result.is_empty() {
                     return Err(parser.new_custom_error(StyleParseErrorKind::UnspecifiedError));
                 }
-                if result.len() == 1 {
-                    NonTSPseudoClass::Lang(Lang::Single(result[0].clone()))
-                } else {
-                    NonTSPseudoClass::Lang(Lang::List(Box::new(result)))
-                }
+                NonTSPseudoClass::Lang(Lang(result.into()))
             },
             "-moz-locale-dir" => {
                 NonTSPseudoClass::MozLocaleDir(Direction::parse(parser)?)
@@ -448,7 +434,7 @@ impl<'a, 'i> ::selectors::Parser<'i> for SelectorParser<'a> {
         if starts_with_ignore_ascii_case(&name, "-moz-tree-") {
             // Tree pseudo-elements can have zero or more arguments, separated
             // by either comma or space.
-            let mut args = Vec::new();
+            let mut args = ThinVec::new();
             loop {
                 let location = parser.current_source_location();
                 match parser.next() {
@@ -462,7 +448,6 @@ impl<'a, 'i> ::selectors::Parser<'i> for SelectorParser<'a> {
                     _ => unreachable!("Parser::next() shouldn't return any other error"),
                 }
             }
-            let args = args.into_boxed_slice();
             if let Some(pseudo) = PseudoElement::tree_pseudo_element(&name, args) {
                 if self.is_pseudo_element_enabled(&pseudo) {
                     return Ok(pseudo);
diff --git a/components/style/gecko/wrapper.rs b/components/style/gecko/wrapper.rs
index 762ff037416..de2bd4c5f4c 100644
--- a/components/style/gecko/wrapper.rs
+++ b/components/style/gecko/wrapper.rs
@@ -1566,26 +1566,14 @@ impl<'le> TElement for GeckoElement<'le> {
             Some(Some(ref atom)) => atom.as_ptr(),
             _ => ptr::null_mut(),
         };
-        match value {
-            Lang::Single(lang) => unsafe {
-                Gecko_MatchLang(
-                    self.0,
-                    override_lang_ptr,
-                    override_lang.is_some(),
-                    lang.as_slice().as_ptr(),
-                )
-            },
-            Lang::List(list) => {
-                list.iter().any(|lang| unsafe {
-                    Gecko_MatchLang(
-                        self.0,
-                        override_lang_ptr,
-                        override_lang.is_some(),
-                        lang.as_slice().as_ptr(),
-                    )
-                })
-            },
-        }
+        value.0.iter().any(|lang| unsafe {
+            Gecko_MatchLang(
+                self.0,
+                override_lang_ptr,
+                override_lang.is_some(),
+                lang.as_slice().as_ptr(),
+            )
+        })
     }
 
     fn is_html_document_body_element(&self) -> bool {
diff --git a/components/to_shmem/Cargo.toml b/components/to_shmem/Cargo.toml
index 9ef2cefe112..6de19e47df5 100644
--- a/components/to_shmem/Cargo.toml
+++ b/components/to_shmem/Cargo.toml
@@ -20,3 +20,4 @@ servo_arc = { path = "../servo_arc" }
 smallbitvec = { workspace = true }
 smallvec = { workspace = true }
 string_cache = { workspace = true, optional = true }
+thin-vec = { version = "0.2.1", features = ["gecko-ffi"] }
diff --git a/components/to_shmem/lib.rs b/components/to_shmem/lib.rs
index e01c144cae0..5a968c1283b 100644
--- a/components/to_shmem/lib.rs
+++ b/components/to_shmem/lib.rs
@@ -12,6 +12,8 @@
 #![crate_name = "to_shmem"]
 #![crate_type = "rlib"]
 
+extern crate thin_vec;
+
 use std::alloc::Layout;
 #[cfg(debug_assertions)]
 use std::any::TypeId;
@@ -31,6 +33,7 @@ use std::{isize, slice, str};
 use servo_arc::{Arc, ThinArc};
 use smallbitvec::{InternalStorage, SmallBitVec};
 use smallvec::{Array, SmallVec};
+use thin_vec::ThinVec;
 
 /// Result type for ToShmem::to_shmem.
 ///
@@ -498,6 +501,58 @@ impl<H: 'static + ToShmem, T: 'static + ToShmem> ToShmem for ThinArc<H, T> {
     }
 }
 
+impl<T: ToShmem> ToShmem for ThinVec<T> {
+    fn to_shmem(&self, builder: &mut SharedMemoryBuilder) -> Result<Self> {
+        let len = self.len();
+        if len == 0 {
+            return Ok(ManuallyDrop::new(Self::new()))
+        }
+
+        assert_eq!(mem::size_of::<Self>(), mem::size_of::<*const ()>());
+
+        // nsTArrayHeader size.
+        // FIXME: Would be nice not to hard-code this, but in practice thin-vec crate also relies
+        // on this.
+        let header_size = 2 * mem::size_of::<u32>();
+        let header_align = mem::size_of::<u32>();
+
+        let item_size = mem::size_of::<T>();
+        let item_align = mem::align_of::<T>();
+
+        // We don't need to support underalignment for now, this could be supported if needed.
+        assert!(item_align >= header_align);
+
+        // This is explicitly unsupported by ThinVec, see:
+        // https://searchfox.org/mozilla-central/rev/ad732108b073742d7324f998c085f459674a6846/third_party/rust/thin-vec/src/lib.rs#375-386
+        assert!(item_align <= header_size);
+        let header_padding = 0;
+
+        let layout = Layout::from_size_align(header_size + header_padding + padded_size(item_size, item_align) * len, item_align).unwrap();
+
+        let shmem_header_ptr = builder.alloc::<u8>(layout);
+        let shmem_data_ptr = unsafe { shmem_header_ptr.add(header_size + header_padding) };
+
+        let data_ptr = self.as_ptr() as *const T as *const u8;
+        let header_ptr = unsafe { data_ptr.sub(header_size + header_padding) };
+
+        unsafe {
+            // Copy the header. Note this might copy a wrong capacity, but it doesn't matter,
+            // because shared memory ptrs are immutable anyways, and we can't relocate.
+            ptr::copy(header_ptr, shmem_header_ptr, header_size);
+            // ToShmem + copy the contents into the shared buffer.
+            to_shmem_slice_ptr(self.iter(), shmem_data_ptr as *mut T, builder)?;
+            // Return the new ThinVec, which is just a pointer to the shared memory buffer.
+            let shmem_thinvec: Self = mem::transmute(shmem_header_ptr);
+
+            // Sanity-check that the ptr and length match.
+            debug_assert_eq!(shmem_thinvec.as_ptr(), shmem_data_ptr as *const T);
+            debug_assert_eq!(shmem_thinvec.len(), len);
+
+            Ok(ManuallyDrop::new(shmem_thinvec))
+        }
+    }
+}
+
 impl ToShmem for SmallBitVec {
     fn to_shmem(&self, builder: &mut SharedMemoryBuilder) -> Result<Self> {
         let storage = match self.clone().into_storage() {