feat: init Rust VM

boxing/Cargo.toml

@@ -0,0 +1,8 @@
+[package]
+name = "boxing"
+version = "0.1.3"
+edition = "2021"
+description = "NaN-boxing primitives (local fork with bool fix)"
+
+[dependencies]
+sptr = "0.3"

boxing/src/lib.rs

@@ -0,0 +1,2 @@
+pub mod nan;
+mod utils;

boxing/src/nan.rs

@@ -0,0 +1,7 @@
+pub mod raw;
+
+pub use raw::RawBox;
+
+const SIGN_MASK: u64 = 0x7FFF_FFFF_FFFF_FFFF;
+const QUIET_NAN: u64 = 0x7FF8_0000_0000_0000;
+const NEG_QUIET_NAN: u64 = 0xFFF8_0000_0000_0000;

boxing/src/nan/raw.rs

@@ -0,0 +1,471 @@
+use super::{NEG_QUIET_NAN, QUIET_NAN, SIGN_MASK};
+use crate::utils::ArrayExt;
+use sptr::Strict;
+use std::fmt;
+use std::mem::ManuallyDrop;
+use std::num::NonZeroU8;
+
+pub trait RawStore: Sized {
+    fn to_val(self, value: &mut Value);
+    fn from_val(value: &Value) -> Self;
+}
+
+impl RawStore for [u8; 6] {
+    #[inline]
+    fn to_val(self, value: &mut Value) {
+        value.set_data(self);
+    }
+
+    #[inline]
+    fn from_val(value: &Value) -> Self {
+        *value.data()
+    }
+}
+
+impl RawStore for bool {
+    #[inline]
+    fn to_val(self, value: &mut Value) {
+        value.set_data([u8::from(self)].truncate_to());
+    }
+
+    #[inline]
+    fn from_val(value: &Value) -> Self {
+        value.data()[0] == 1
+    }
+}
+
+macro_rules! int_store {
+    ($ty:ty) => {
+        impl RawStore for $ty {
+            #[inline]
+            fn to_val(self, value: &mut Value) {
+                let bytes = self.to_ne_bytes();
+                value.set_data(bytes.truncate_to());
+            }
+
+            #[inline]
+            fn from_val(value: &Value) -> Self {
+                <$ty>::from_ne_bytes(value.data().truncate_to())
+            }
+        }
+    };
+}
+
+int_store!(u8);
+int_store!(u16);
+int_store!(u32);
+
+int_store!(i8);
+int_store!(i16);
+int_store!(i32);
+
+fn store_ptr<P: Strict + Copy>(value: &mut Value, ptr: P) {
+    #[cfg(target_pointer_width = "64")]
+    {
+        assert!(
+            ptr.addr() <= 0x0000_FFFF_FFFF_FFFF,
+            "Pointer too large to store in NaN box"
+        );
+
+        let val = (unsafe { value.whole_mut() } as *mut [u8; 8]).cast::<P>();
+
+        let ptr = Strict::map_addr(ptr, |addr| {
+            addr | (usize::from(value.header().into_raw()) << 48)
+        });
+
+        unsafe { val.write(ptr) };
+    }
+
+    #[cfg(target_pointer_width = "32")]
+    {
+        let _ = (value, ptr);
+        unimplemented!("32-bit pointer storage not supported");
+    }
+}
+
+fn load_ptr<P: Strict>(value: &Value) -> P {
+    #[cfg(target_pointer_width = "64")]
+    {
+        let val = (unsafe { value.whole() } as *const [u8; 8]).cast::<P>();
+        let ptr = unsafe { val.read() };
+        Strict::map_addr(ptr, |addr| addr & 0x0000_FFFF_FFFF_FFFF)
+    }
+
+    #[cfg(target_pointer_width = "32")]
+    {
+        let _ = value;
+        unimplemented!("32-bit pointer storage not supported");
+    }
+}
+
+impl<T> RawStore for *const T {
+    fn to_val(self, value: &mut Value) {
+        store_ptr::<*const T>(value, self);
+    }
+
+    fn from_val(value: &Value) -> Self {
+        load_ptr::<*const T>(value)
+    }
+}
+
+impl<T> RawStore for *mut T {
+    fn to_val(self, value: &mut Value) {
+        store_ptr::<*mut T>(value, self);
+    }
+
+    fn from_val(value: &Value) -> Self {
+        load_ptr::<*mut T>(value)
+    }
+}
+
+#[derive(Copy, Clone, PartialEq, Eq)]
+enum TagVal {
+    _P1,
+    _P2,
+    _P3,
+    _P4,
+    _P5,
+    _P6,
+    _P7,
+
+    _N1,
+    _N2,
+    _N3,
+    _N4,
+    _N5,
+    _N6,
+    _N7,
+}
+
+#[derive(Copy, Clone, PartialEq, Eq)]
+pub struct RawTag(TagVal);
+
+impl RawTag {
+    #[inline]
+    #[must_use]
+    pub fn new(neg: bool, val: NonZeroU8) -> RawTag {
+        unsafe { Self::new_unchecked(neg, val.get() & 0x07) }
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn new_checked(neg: bool, val: u8) -> Option<RawTag> {
+        Some(RawTag(match (neg, val) {
+            (false, 1) => TagVal::_P1,
+            (false, 2) => TagVal::_P2,
+            (false, 3) => TagVal::_P3,
+            (false, 4) => TagVal::_P4,
+            (false, 5) => TagVal::_P5,
+            (false, 6) => TagVal::_P6,
+            (false, 7) => TagVal::_P7,
+
+            (true, 1) => TagVal::_N1,
+            (true, 2) => TagVal::_N2,
+            (true, 3) => TagVal::_N3,
+            (true, 4) => TagVal::_N4,
+            (true, 5) => TagVal::_N5,
+            (true, 6) => TagVal::_N6,
+            (true, 7) => TagVal::_N7,
+
+            _ => return None,
+        }))
+    }
+
+    /// # Safety
+    ///
+    /// `val` must be in the range `1..8`
+    #[inline]
+    #[must_use]
+    pub unsafe fn new_unchecked(neg: bool, val: u8) -> RawTag {
+        RawTag(match (neg, val) {
+            (false, 1) => TagVal::_P1,
+            (false, 2) => TagVal::_P2,
+            (false, 3) => TagVal::_P3,
+            (false, 4) => TagVal::_P4,
+            (false, 5) => TagVal::_P5,
+            (false, 6) => TagVal::_P6,
+            (false, 7) => TagVal::_P7,
+
+            (true, 1) => TagVal::_N1,
+            (true, 2) => TagVal::_N2,
+            (true, 3) => TagVal::_N3,
+            (true, 4) => TagVal::_N4,
+            (true, 5) => TagVal::_N5,
+            (true, 6) => TagVal::_N6,
+            (true, 7) => TagVal::_N7,
+
+            _ => unsafe { core::hint::unreachable_unchecked() },
+        })
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn is_neg(self) -> bool {
+        matches!(self.0, |TagVal::_N1| TagVal::_N2
+            | TagVal::_N3
+            | TagVal::_N4
+            | TagVal::_N5
+            | TagVal::_N6
+            | TagVal::_N7)
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn val(self) -> NonZeroU8 {
+        match self.0 {
+            TagVal::_P1 | TagVal::_N1 => NonZeroU8::MIN,
+            TagVal::_P2 | TagVal::_N2 => NonZeroU8::MIN.saturating_add(1),
+            TagVal::_P3 | TagVal::_N3 => NonZeroU8::MIN.saturating_add(2),
+            TagVal::_P4 | TagVal::_N4 => NonZeroU8::MIN.saturating_add(3),
+            TagVal::_P5 | TagVal::_N5 => NonZeroU8::MIN.saturating_add(4),
+            TagVal::_P6 | TagVal::_N6 => NonZeroU8::MIN.saturating_add(5),
+            TagVal::_P7 | TagVal::_N7 => NonZeroU8::MIN.saturating_add(6),
+        }
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn neg_val(self) -> (bool, u8) {
+        match self.0 {
+            TagVal::_P1 => (false, 1),
+            TagVal::_P2 => (false, 2),
+            TagVal::_P3 => (false, 3),
+            TagVal::_P4 => (false, 4),
+            TagVal::_P5 => (false, 5),
+            TagVal::_P6 => (false, 6),
+            TagVal::_P7 => (false, 7),
+            TagVal::_N1 => (true, 1),
+            TagVal::_N2 => (true, 2),
+            TagVal::_N3 => (true, 3),
+            TagVal::_N4 => (true, 4),
+            TagVal::_N5 => (true, 5),
+            TagVal::_N6 => (true, 6),
+            TagVal::_N7 => (true, 7),
+        }
+    }
+}
+
+impl fmt::Debug for RawTag {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.debug_struct("RawTag")
+            .field("neg", &self.is_neg())
+            .field("val", &self.val())
+            .finish()
+    }
+}
+
+#[derive(Copy, Clone, Debug, PartialEq)]
+#[repr(transparent)]
+struct Header(u16);
+
+impl Header {
+    #[inline]
+    fn new(tag: RawTag) -> Header {
+        let (neg, val) = tag.neg_val();
+        Header(0x7FF8 | (u16::from(neg) << 15) | u16::from(val))
+    }
+
+    #[inline]
+    fn tag(self) -> RawTag {
+        unsafe { RawTag::new_unchecked(self.get_sign(), self.get_tag()) }
+    }
+
+    #[inline]
+    fn get_sign(self) -> bool {
+        self.0 & 0x8000 != 0
+    }
+
+    #[inline]
+    fn get_tag(self) -> u8 {
+        (self.0 & 0x0007) as u8
+    }
+
+    #[inline]
+    fn into_raw(self) -> u16 {
+        self.0
+    }
+}
+
+#[derive(Clone, Debug, PartialEq)]
+#[repr(C, align(8))]
+pub struct Value {
+    #[cfg(target_endian = "big")]
+    header: Header,
+    data: [u8; 6],
+    #[cfg(target_endian = "little")]
+    header: Header,
+}
+
+impl Value {
+    #[inline]
+    pub fn new(tag: RawTag, data: [u8; 6]) -> Value {
+        Value {
+            header: Header::new(tag),
+            data,
+        }
+    }
+
+    #[inline]
+    pub fn empty(tag: RawTag) -> Value {
+        Value::new(tag, [0; 6])
+    }
+
+    pub fn store<T: RawStore>(tag: RawTag, val: T) -> Value {
+        let mut v = Value::new(tag, [0; 6]);
+        T::to_val(val, &mut v);
+        v
+    }
+
+    pub fn load<T: RawStore>(self) -> T {
+        T::from_val(&self)
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn tag(&self) -> RawTag {
+        self.header.tag()
+    }
+
+    #[inline]
+    fn header(&self) -> &Header {
+        &self.header
+    }
+
+    #[inline]
+    pub fn set_data(&mut self, val: [u8; 6]) {
+        self.data = val;
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn data(&self) -> &[u8; 6] {
+        &self.data
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn data_mut(&mut self) -> &mut [u8; 6] {
+        &mut self.data
+    }
+
+    #[inline]
+    #[must_use]
+    pub unsafe fn whole(&self) -> &[u8; 8] {
+        let ptr = (self as *const Value).cast::<[u8; 8]>();
+        unsafe { &*ptr }
+    }
+
+    #[inline]
+    #[must_use]
+    pub unsafe fn whole_mut(&mut self) -> &mut [u8; 8] {
+        let ptr = (self as *mut Value).cast::<[u8; 8]>();
+        unsafe { &mut *ptr }
+    }
+}
+
+#[repr(C)]
+pub union RawBox {
+    float: f64,
+    value: ManuallyDrop<Value>,
+    bits: u64,
+    #[cfg(target_pointer_width = "64")]
+    ptr: *const (),
+    #[cfg(target_pointer_width = "32")]
+    ptr: (u32, *const ()),
+}
+
+impl RawBox {
+    #[inline]
+    #[must_use]
+    pub fn from_float(val: f64) -> RawBox {
+        match (val.is_nan(), val.is_sign_positive()) {
+            (true, true) => RawBox {
+                float: f64::from_bits(QUIET_NAN),
+            },
+            (true, false) => RawBox {
+                float: f64::from_bits(NEG_QUIET_NAN),
+            },
+            (false, _) => RawBox { float: val },
+        }
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn from_value(value: Value) -> RawBox {
+        RawBox {
+            value: ManuallyDrop::new(value),
+        }
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn tag(&self) -> Option<RawTag> {
+        if self.is_value() {
+            Some(unsafe { self.value.tag() })
+        } else {
+            None
+        }
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn is_float(&self) -> bool {
+        (unsafe { !self.float.is_nan() } || unsafe { self.bits & SIGN_MASK == QUIET_NAN })
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn is_value(&self) -> bool {
+        (unsafe { self.float.is_nan() } && unsafe { self.bits & SIGN_MASK != QUIET_NAN })
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn float(&self) -> Option<&f64> {
+        if self.is_float() {
+            Some(unsafe { &self.float })
+        } else {
+            None
+        }
+    }
+
+    #[inline]
+    #[must_use]
+    pub fn value(&self) -> Option<&Value> {
+        if self.is_value() {
+            Some(unsafe { &self.value })
+        } else {
+            None
+        }
+    }
+
+    #[inline]
+    pub fn into_float_unchecked(self) -> f64 {
+        unsafe { self.float }
+    }
+}
+
+impl Clone for RawBox {
+    #[inline]
+    fn clone(&self) -> Self {
+        RawBox {
+            ptr: unsafe { self.ptr },
+        }
+    }
+}
+
+impl fmt::Debug for RawBox {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        match self.float() {
+            Some(val) => f.debug_tuple("RawBox::Float").field(val).finish(),
+            None => {
+                let val = self.value().expect("RawBox is neither float nor value");
+
+                f.debug_struct("RawBox::Data")
+                    .field("tag", &val.tag())
+                    .field("data", val.data())
+                    .finish()
+            }
+        }
+    }
+}

boxing/src/utils.rs

@@ -0,0 +1,16 @@
+pub trait ArrayExt<const LEN: usize> {
+    type Elem;
+
+    fn truncate_to<const M: usize>(self) -> [Self::Elem; M];
+}
+
+impl<T: Default + Copy, const N: usize> ArrayExt<N> for [T; N] {
+    type Elem = T;
+
+    fn truncate_to<const M: usize>(self) -> [Self::Elem; M] {
+        let copy_len = usize::min(N, M);
+        let mut out = [T::default(); M];
+        out[0..copy_len].copy_from_slice(&self[0..copy_len]);
+        out
+    }
+}