Introduce hashbrown fork

Cargo.toml

@@ -18,6 +18,3 @@ default-members = [
 [profile.bench]
 lto = false
 debug = true
-
-[patch.crates-io]
-hashbrown = { git = "https://github.com/udoprog/hashbrown", branch = "raw-infallible-context" }

crates/rune/Cargo.toml

@@ -42,7 +42,7 @@ serde = { version = "1.0.163", default-features = false, features = ["derive", "
 serde_bytes = { version = "0.11.9", default-features = false, features = ["alloc"] }
 smallvec = { version = "1.10.0", default-features = false, features = ["serde", "const_new"] }
 tracing =  { version = "0.1.37", default-features = false, features = ["attributes"] }
-hashbrown = { version = "0.14.0", features = ["raw", "serde"] }
+hashbrown = { version = "0.14.0", features = ["serde"] }
 musli = { version = "0.0.42", default-features = false, features = ["alloc"] }
 slab = { version = "0.4.8", default-features = false }
 

crates/rune/src/hashbrown.rs

@@ -1,3 +1,5 @@
 mod table;
+pub(crate) use self::fork::raw::RawIter;
 pub(crate) use self::table::{IterRef, Table};
-pub(crate) use ::hashbrown::raw::RawIter;
+
+mod fork;

crates/rune/src/hashbrown/fork.rs

@@ -0,0 +1,34 @@
+#![allow(unused)]
+
+// Copied and modified under the MIT license.
+// Copyright (c) 2016 Amanieu d'Antras
+//
+// Imported using import_hashbrown.ps1, the below section is the only part
+// copied by hand.
+//
+// After an import of the crate some sections might need to be modified.
+//
+// See: https://github.com/rust-lang/hashbrown
+// The relevant fork: https://github.com/udoprog/hashbrown/tree/raw-infallible-context
+// Relevant issue: https://github.com/rust-lang/hashbrown/issues/456
+
+#![allow(clippy::manual_map)]
+
+#[macro_use]
+mod macros;
+pub(crate) mod raw;
+mod scopeguard;
+
+/// The error type for `try_reserve` methods.
+#[derive(Clone, PartialEq, Eq, Debug)]
+pub enum TryReserveError {
+    /// Error due to the computed capacity exceeding the collection's maximum
+    /// (usually `isize::MAX` bytes).
+    CapacityOverflow,
+
+    /// The memory allocator returned an error
+    AllocError {
+        /// The layout of the allocation request that failed.
+        layout: alloc::alloc::Layout,
+    },
+}

crates/rune/src/hashbrown/fork/macros.rs

@@ -0,0 +1,70 @@
+// See the cfg-if crate.
+#[allow(unused_macro_rules)]
+macro_rules! cfg_if {
+    // match if/else chains with a final `else`
+    ($(
+        if #[cfg($($meta:meta),*)] { $($it:item)* }
+    ) else * else {
+        $($it2:item)*
+    }) => {
+        cfg_if! {
+            @__items
+            () ;
+            $( ( ($($meta),*) ($($it)*) ), )*
+            ( () ($($it2)*) ),
+        }
+    };
+
+    // match if/else chains lacking a final `else`
+    (
+        if #[cfg($($i_met:meta),*)] { $($i_it:item)* }
+        $(
+            else if #[cfg($($e_met:meta),*)] { $($e_it:item)* }
+        )*
+    ) => {
+        cfg_if! {
+            @__items
+            () ;
+            ( ($($i_met),*) ($($i_it)*) ),
+            $( ( ($($e_met),*) ($($e_it)*) ), )*
+            ( () () ),
+        }
+    };
+
+    // Internal and recursive macro to emit all the items
+    //
+    // Collects all the negated cfgs in a list at the beginning and after the
+    // semicolon is all the remaining items
+    (@__items ($($not:meta,)*) ; ) => {};
+    (@__items ($($not:meta,)*) ; ( ($($m:meta),*) ($($it:item)*) ), $($rest:tt)*) => {
+        // Emit all items within one block, applying an appropriate #[cfg]. The
+        // #[cfg] will require all `$m` matchers specified and must also negate
+        // all previous matchers.
+        cfg_if! { @__apply cfg(all($($m,)* not(any($($not),*)))), $($it)* }
+
+        // Recurse to emit all other items in `$rest`, and when we do so add all
+        // our `$m` matchers to the list of `$not` matchers as future emissions
+        // will have to negate everything we just matched as well.
+        cfg_if! { @__items ($($not,)* $($m,)*) ; $($rest)* }
+    };
+
+    // Internal macro to Apply a cfg attribute to a list of items
+    (@__apply $m:meta, $($it:item)*) => {
+        $(#[$m] $it)*
+    };
+}
+
+// Helper macro for specialization. This also helps avoid parse errors if the
+// default fn syntax for specialization changes in the future.
+#[cfg(feature = "nightly")]
+macro_rules! default_fn {
+	(#[$($a:tt)*] $($tt:tt)*) => {
+        #[$($a)*] default $($tt)*
+    }
+}
+#[cfg(not(feature = "nightly"))]
+macro_rules! default_fn {
+	($($tt:tt)*) => {
+        $($tt)*
+    }
+}

crates/rune/src/hashbrown/fork/raw/alloc.rs

@@ -0,0 +1,86 @@
+pub(crate) use self::inner::{do_alloc, Allocator, Global};
+
+// Nightly-case.
+// Use unstable `allocator_api` feature.
+// This is compatible with `allocator-api2` which can be enabled or not.
+// This is used when building for `std`.
+#[cfg(feature = "nightly")]
+mod inner {
+    use crate::alloc::alloc::Layout;
+    pub use crate::alloc::alloc::{Allocator, Global};
+    use core::ptr::NonNull;
+
+    #[allow(clippy::map_err_ignore)]
+    pub(crate) fn do_alloc<A: Allocator>(alloc: &A, layout: Layout) -> Result<NonNull<u8>, ()> {
+        match alloc.allocate(layout) {
+            Ok(ptr) => Ok(ptr.as_non_null_ptr()),
+            Err(_) => Err(()),
+        }
+    }
+}
+
+// Basic non-nightly case.
+// This uses `allocator-api2` enabled by default.
+// If any crate enables "nightly" in `allocator-api2`,
+// this will be equivalent to the nightly case,
+// since `allocator_api2::alloc::Allocator` would be re-export of
+// `core::alloc::Allocator`.
+#[cfg(all(not(feature = "nightly"), feature = "allocator-api2"))]
+mod inner {
+    use crate::alloc::alloc::Layout;
+    pub use allocator_api2::alloc::{Allocator, Global};
+    use core::ptr::NonNull;
+
+    #[allow(clippy::map_err_ignore)]
+    pub(crate) fn do_alloc<A: Allocator>(alloc: &A, layout: Layout) -> Result<NonNull<u8>, ()> {
+        match alloc.allocate(layout) {
+            Ok(ptr) => Ok(ptr.cast()),
+            Err(_) => Err(()),
+        }
+    }
+}
+
+// No-defaults case.
+// When building with default-features turned off and
+// neither `nightly` nor `allocator-api2` is enabled,
+// this will be used.
+// Making it impossible to use any custom allocator with collections defined
+// in this crate.
+// Any crate in build-tree can enable `allocator-api2`,
+// or `nightly` without disturbing users that don't want to use it.
+#[cfg(not(any(feature = "nightly", feature = "allocator-api2")))]
+mod inner {
+    use crate::alloc::alloc::{alloc, dealloc, Layout};
+    use core::ptr::NonNull;
+
+    #[allow(clippy::missing_safety_doc)] // not exposed outside of this crate
+    pub unsafe trait Allocator {
+        fn allocate(&self, layout: Layout) -> Result<NonNull<u8>, ()>;
+        unsafe fn deallocate(&self, ptr: NonNull<u8>, layout: Layout);
+    }
+
+    #[derive(Copy, Clone)]
+    pub struct Global;
+
+    unsafe impl Allocator for Global {
+        #[inline]
+        fn allocate(&self, layout: Layout) -> Result<NonNull<u8>, ()> {
+            unsafe { NonNull::new(alloc(layout)).ok_or(()) }
+        }
+        #[inline]
+        unsafe fn deallocate(&self, ptr: NonNull<u8>, layout: Layout) {
+            dealloc(ptr.as_ptr(), layout);
+        }
+    }
+
+    impl Default for Global {
+        #[inline]
+        fn default() -> Self {
+            Global
+        }
+    }
+
+    pub(crate) fn do_alloc<A: Allocator>(alloc: &A, layout: Layout) -> Result<NonNull<u8>, ()> {
+        alloc.allocate(layout)
+    }
+}

crates/rune/src/hashbrown/fork/raw/bitmask.rs

@@ -0,0 +1,133 @@
+use super::imp::{
+    BitMaskWord, NonZeroBitMaskWord, BITMASK_ITER_MASK, BITMASK_MASK, BITMASK_STRIDE,
+};
+
+/// A bit mask which contains the result of a `Match` operation on a `Group` and
+/// allows iterating through them.
+///
+/// The bit mask is arranged so that low-order bits represent lower memory
+/// addresses for group match results.
+///
+/// For implementation reasons, the bits in the set may be sparsely packed with
+/// groups of 8 bits representing one element. If any of these bits are non-zero
+/// then this element is considered to true in the mask. If this is the
+/// case, `BITMASK_STRIDE` will be 8 to indicate a divide-by-8 should be
+/// performed on counts/indices to normalize this difference. `BITMASK_MASK` is
+/// similarly a mask of all the actually-used bits.
+///
+/// To iterate over a bit mask, it must be converted to a form where only 1 bit
+/// is set per element. This is done by applying `BITMASK_ITER_MASK` on the
+/// mask bits.
+#[derive(Copy, Clone)]
+pub(crate) struct BitMask(pub(crate) BitMaskWord);
+
+#[allow(clippy::use_self)]
+impl BitMask {
+    /// Returns a new `BitMask` with all bits inverted.
+    #[inline]
+    #[must_use]
+    #[allow(dead_code)]
+    pub(crate) fn invert(self) -> Self {
+        BitMask(self.0 ^ BITMASK_MASK)
+    }
+
+    /// Returns a new `BitMask` with the lowest bit removed.
+    #[inline]
+    #[must_use]
+    fn remove_lowest_bit(self) -> Self {
+        BitMask(self.0 & (self.0 - 1))
+    }
+
+    /// Returns whether the `BitMask` has at least one set bit.
+    #[inline]
+    pub(crate) fn any_bit_set(self) -> bool {
+        self.0 != 0
+    }
+
+    /// Returns the first set bit in the `BitMask`, if there is one.
+    #[inline]
+    pub(crate) fn lowest_set_bit(self) -> Option<usize> {
+        if let Some(nonzero) = NonZeroBitMaskWord::new(self.0) {
+            Some(Self::nonzero_trailing_zeros(nonzero))
+        } else {
+            None
+        }
+    }
+
+    /// Returns the number of trailing zeroes in the `BitMask`.
+    #[inline]
+    pub(crate) fn trailing_zeros(self) -> usize {
+        // ARM doesn't have a trailing_zeroes instruction, and instead uses
+        // reverse_bits (RBIT) + leading_zeroes (CLZ). However older ARM
+        // versions (pre-ARMv7) don't have RBIT and need to emulate it
+        // instead. Since we only have 1 bit set in each byte on ARM, we can
+        // use swap_bytes (REV) + leading_zeroes instead.
+        if cfg!(target_arch = "arm") && BITMASK_STRIDE % 8 == 0 {
+            self.0.swap_bytes().leading_zeros() as usize / BITMASK_STRIDE
+        } else {
+            self.0.trailing_zeros() as usize / BITMASK_STRIDE
+        }
+    }
+
+    /// Same as above but takes a `NonZeroBitMaskWord`.
+    #[inline]
+    fn nonzero_trailing_zeros(nonzero: NonZeroBitMaskWord) -> usize {
+        if cfg!(target_arch = "arm") && BITMASK_STRIDE % 8 == 0 {
+            // SAFETY: A byte-swapped non-zero value is still non-zero.
+            let swapped = unsafe { NonZeroBitMaskWord::new_unchecked(nonzero.get().swap_bytes()) };
+            swapped.leading_zeros() as usize / BITMASK_STRIDE
+        } else {
+            nonzero.trailing_zeros() as usize / BITMASK_STRIDE
+        }
+    }
+
+    /// Returns the number of leading zeroes in the `BitMask`.
+    #[inline]
+    pub(crate) fn leading_zeros(self) -> usize {
+        self.0.leading_zeros() as usize / BITMASK_STRIDE
+    }
+}
+
+impl IntoIterator for BitMask {
+    type Item = usize;
+    type IntoIter = BitMaskIter;
+
+    #[inline]
+    fn into_iter(self) -> BitMaskIter {
+        // A BitMask only requires each element (group of bits) to be non-zero.
+        // However for iteration we need each element to only contain 1 bit.
+        BitMaskIter(BitMask(self.0 & BITMASK_ITER_MASK))
+    }
+}
+
+/// Iterator over the contents of a `BitMask`, returning the indices of set
+/// bits.
+#[derive(Copy, Clone)]
+pub(crate) struct BitMaskIter(pub(crate) BitMask);
+
+impl BitMaskIter {
+    /// Flip the bit in the mask for the entry at the given index.
+    ///
+    /// Returns the bit's previous state.
+    #[inline]
+    #[allow(clippy::cast_ptr_alignment)]
+    #[cfg(feature = "raw")]
+    pub(crate) unsafe fn flip(&mut self, index: usize) -> bool {
+        // NOTE: The + BITMASK_STRIDE - 1 is to set the high bit.
+        let mask = 1 << (index * BITMASK_STRIDE + BITMASK_STRIDE - 1);
+        self.0 .0 ^= mask;
+        // The bit was set if the bit is now 0.
+        self.0 .0 & mask == 0
+    }
+}
+
+impl Iterator for BitMaskIter {
+    type Item = usize;
+
+    #[inline]
+    fn next(&mut self) -> Option<usize> {
+        let bit = self.0.lowest_set_bit()?;
+        self.0 = self.0.remove_lowest_bit();
+        Some(bit)
+    }
+}