Auto merge of rust-lang#85828 - scottmcm:raw-eq, r=oli-obk

Stop generating `alloca`s & `memcmp` for simple short array equality

Example:
```rust
pub fn demo(x: [u16; 6], y: [u16; 6]) -> bool { x == y }
```

Before:
```llvm
define zeroext i1 `@_ZN10playground4demo17h48537f7eac23948fE(i96` %0, i96 %1) unnamed_addr #0 {
start:
  %y = alloca [6 x i16], align 8
  %x = alloca [6 x i16], align 8
  %.0..sroa_cast = bitcast [6 x i16]* %x to i96*
  store i96 %0, i96* %.0..sroa_cast, align 8
  %.0..sroa_cast3 = bitcast [6 x i16]* %y to i96*
  store i96 %1, i96* %.0..sroa_cast3, align 8
  %_11.i.i.i = bitcast [6 x i16]* %x to i8*
  %_14.i.i.i = bitcast [6 x i16]* %y to i8*
  %bcmp.i.i.i = call i32 `@bcmp(i8*` nonnull dereferenceable(12) %_11.i.i.i, i8* nonnull dereferenceable(12) %_14.i.i.i, i64 12) #2, !alias.scope !2
  %2 = icmp eq i32 %bcmp.i.i.i, 0
  ret i1 %2
}
```
```x86
playground::demo: # `@playground::demo`
	sub	rsp, 32
	mov	qword ptr [rsp], rdi
	mov	dword ptr [rsp + 8], esi
	mov	qword ptr [rsp + 16], rdx
	mov	dword ptr [rsp + 24], ecx
	xor	rdi, rdx
	xor	esi, ecx
	or	rsi, rdi
	sete	al
	add	rsp, 32
	ret
```

After:
```llvm
define zeroext i1 `@_ZN4mini4demo17h7a8994aaa314c981E(i96` %0, i96 %1) unnamed_addr #0 {
start:
  %2 = icmp eq i96 %0, %1
  ret i1 %2
}
```
```x86
_ZN4mini4demo17h7a8994aaa314c981E:
	xor	rcx, r8
	xor	edx, r9d
	or	rdx, rcx
	sete	al
	ret
```

Author: bors

core/src/array/equality.rs

@@ -0,0 +1,160 @@
+#[stable(feature = "rust1", since = "1.0.0")]
+impl<A, B, const N: usize> PartialEq<[B; N]> for [A; N]
+where
+    A: PartialEq<B>,
+{
+    #[inline]
+    fn eq(&self, other: &[B; N]) -> bool {
+        SpecArrayEq::spec_eq(self, other)
+    }
+    #[inline]
+    fn ne(&self, other: &[B; N]) -> bool {
+        SpecArrayEq::spec_ne(self, other)
+    }
+}
+
+#[stable(feature = "rust1", since = "1.0.0")]
+impl<A, B, const N: usize> PartialEq<[B]> for [A; N]
+where
+    A: PartialEq<B>,
+{
+    #[inline]
+    fn eq(&self, other: &[B]) -> bool {
+        self[..] == other[..]
+    }
+    #[inline]
+    fn ne(&self, other: &[B]) -> bool {
+        self[..] != other[..]
+    }
+}
+
+#[stable(feature = "rust1", since = "1.0.0")]
+impl<A, B, const N: usize> PartialEq<[A; N]> for [B]
+where
+    B: PartialEq<A>,
+{
+    #[inline]
+    fn eq(&self, other: &[A; N]) -> bool {
+        self[..] == other[..]
+    }
+    #[inline]
+    fn ne(&self, other: &[A; N]) -> bool {
+        self[..] != other[..]
+    }
+}
+
+#[stable(feature = "rust1", since = "1.0.0")]
+impl<A, B, const N: usize> PartialEq<&[B]> for [A; N]
+where
+    A: PartialEq<B>,
+{
+    #[inline]
+    fn eq(&self, other: &&[B]) -> bool {
+        self[..] == other[..]
+    }
+    #[inline]
+    fn ne(&self, other: &&[B]) -> bool {
+        self[..] != other[..]
+    }
+}
+
+#[stable(feature = "rust1", since = "1.0.0")]
+impl<A, B, const N: usize> PartialEq<[A; N]> for &[B]
+where
+    B: PartialEq<A>,
+{
+    #[inline]
+    fn eq(&self, other: &[A; N]) -> bool {
+        self[..] == other[..]
+    }
+    #[inline]
+    fn ne(&self, other: &[A; N]) -> bool {
+        self[..] != other[..]
+    }
+}
+
+#[stable(feature = "rust1", since = "1.0.0")]
+impl<A, B, const N: usize> PartialEq<&mut [B]> for [A; N]
+where
+    A: PartialEq<B>,
+{
+    #[inline]
+    fn eq(&self, other: &&mut [B]) -> bool {
+        self[..] == other[..]
+    }
+    #[inline]
+    fn ne(&self, other: &&mut [B]) -> bool {
+        self[..] != other[..]
+    }
+}
+
+#[stable(feature = "rust1", since = "1.0.0")]
+impl<A, B, const N: usize> PartialEq<[A; N]> for &mut [B]
+where
+    B: PartialEq<A>,
+{
+    #[inline]
+    fn eq(&self, other: &[A; N]) -> bool {
+        self[..] == other[..]
+    }
+    #[inline]
+    fn ne(&self, other: &[A; N]) -> bool {
+        self[..] != other[..]
+    }
+}
+
+// NOTE: some less important impls are omitted to reduce code bloat
+// __impl_slice_eq2! { [A; $N], &'b [B; $N] }
+// __impl_slice_eq2! { [A; $N], &'b mut [B; $N] }
+
+#[stable(feature = "rust1", since = "1.0.0")]
+impl<T: Eq, const N: usize> Eq for [T; N] {}
+
+trait SpecArrayEq<Other, const N: usize>: Sized {
+    fn spec_eq(a: &[Self; N], b: &[Other; N]) -> bool;
+    fn spec_ne(a: &[Self; N], b: &[Other; N]) -> bool;
+}
+
+impl<T: PartialEq<Other>, Other, const N: usize> SpecArrayEq<Other, N> for T {
+    default fn spec_eq(a: &[Self; N], b: &[Other; N]) -> bool {
+        a[..] == b[..]
+    }
+    default fn spec_ne(a: &[Self; N], b: &[Other; N]) -> bool {
+        a[..] != b[..]
+    }
+}
+
+impl<T: PartialEq<U> + IsRawEqComparable<U>, U, const N: usize> SpecArrayEq<U, N> for T {
+    #[cfg(bootstrap)]
+    fn spec_eq(a: &[T; N], b: &[U; N]) -> bool {
+        a[..] == b[..]
+    }
+    #[cfg(not(bootstrap))]
+    fn spec_eq(a: &[T; N], b: &[U; N]) -> bool {
+        // SAFETY: This is why `IsRawEqComparable` is an `unsafe trait`.
+        unsafe {
+            let b = &*b.as_ptr().cast::<[T; N]>();
+            crate::intrinsics::raw_eq(a, b)
+        }
+    }
+    fn spec_ne(a: &[T; N], b: &[U; N]) -> bool {
+        !Self::spec_eq(a, b)
+    }
+}
+
+/// `U` exists on here mostly because `min_specialization` didn't let me
+/// repeat the `T` type parameter in the above specialization, so instead
+/// the `T == U` constraint comes from the impls on this.
+/// # Safety
+/// - Neither `Self` nor `U` has any padding.
+/// - `Self` and `U` have the same layout.
+/// - `Self: PartialEq<U>` is byte-wise (this means no floats, among other things)
+#[rustc_specialization_trait]
+unsafe trait IsRawEqComparable<U> {}
+
+macro_rules! is_raw_comparable {
+    ($($t:ty),+) => {$(
+        unsafe impl IsRawEqComparable<$t> for $t {}
+    )+};
+}
+is_raw_comparable!(bool, char, u8, u16, u32, u64, u128, usize, i8, i16, i32, i64, i128, isize);

core/src/array/mod.rs

@@ -14,6 +14,7 @@ use crate::mem::{self, MaybeUninit};
 use crate::ops::{Index, IndexMut};
 use crate::slice::{Iter, IterMut};
 
+mod equality;
 mod iter;
 
 #[stable(feature = "array_value_iter", since = "1.51.0")]
@@ -230,118 +231,6 @@ where
     }
 }
 
-#[stable(feature = "rust1", since = "1.0.0")]
-impl<A, B, const N: usize> PartialEq<[B; N]> for [A; N]
-where
-    A: PartialEq<B>,
-{
-    #[inline]
-    fn eq(&self, other: &[B; N]) -> bool {
-        self[..] == other[..]
-    }
-    #[inline]
-    fn ne(&self, other: &[B; N]) -> bool {
-        self[..] != other[..]
-    }
-}
-
-#[stable(feature = "rust1", since = "1.0.0")]
-impl<A, B, const N: usize> PartialEq<[B]> for [A; N]
-where
-    A: PartialEq<B>,
-{
-    #[inline]
-    fn eq(&self, other: &[B]) -> bool {
-        self[..] == other[..]
-    }
-    #[inline]
-    fn ne(&self, other: &[B]) -> bool {
-        self[..] != other[..]
-    }
-}
-
-#[stable(feature = "rust1", since = "1.0.0")]
-impl<A, B, const N: usize> PartialEq<[A; N]> for [B]
-where
-    B: PartialEq<A>,
-{
-    #[inline]
-    fn eq(&self, other: &[A; N]) -> bool {
-        self[..] == other[..]
-    }
-    #[inline]
-    fn ne(&self, other: &[A; N]) -> bool {
-        self[..] != other[..]
-    }
-}
-
-#[stable(feature = "rust1", since = "1.0.0")]
-impl<A, B, const N: usize> PartialEq<&[B]> for [A; N]
-where
-    A: PartialEq<B>,
-{
-    #[inline]
-    fn eq(&self, other: &&[B]) -> bool {
-        self[..] == other[..]
-    }
-    #[inline]
-    fn ne(&self, other: &&[B]) -> bool {
-        self[..] != other[..]
-    }
-}
-
-#[stable(feature = "rust1", since = "1.0.0")]
-impl<A, B, const N: usize> PartialEq<[A; N]> for &[B]
-where
-    B: PartialEq<A>,
-{
-    #[inline]
-    fn eq(&self, other: &[A; N]) -> bool {
-        self[..] == other[..]
-    }
-    #[inline]
-    fn ne(&self, other: &[A; N]) -> bool {
-        self[..] != other[..]
-    }
-}
-
-#[stable(feature = "rust1", since = "1.0.0")]
-impl<A, B, const N: usize> PartialEq<&mut [B]> for [A; N]
-where
-    A: PartialEq<B>,
-{
-    #[inline]
-    fn eq(&self, other: &&mut [B]) -> bool {
-        self[..] == other[..]
-    }
-    #[inline]
-    fn ne(&self, other: &&mut [B]) -> bool {
-        self[..] != other[..]
-    }
-}
-
-#[stable(feature = "rust1", since = "1.0.0")]
-impl<A, B, const N: usize> PartialEq<[A; N]> for &mut [B]
-where
-    B: PartialEq<A>,
-{
-    #[inline]
-    fn eq(&self, other: &[A; N]) -> bool {
-        self[..] == other[..]
-    }
-    #[inline]
-    fn ne(&self, other: &[A; N]) -> bool {
-        self[..] != other[..]
-    }
-}
-
-// NOTE: some less important impls are omitted to reduce code bloat
-// __impl_slice_eq2! { [A; $N], &'b [B; $N] }
-// __impl_slice_eq2! { [A; $N], &'b mut [B; $N] }
-
-#[stable(feature = "rust1", since = "1.0.0")]
-impl<T: Eq, const N: usize> Eq for [T; N] {}
-
 #[stable(feature = "rust1", since = "1.0.0")]
 impl<T: PartialOrd, const N: usize> PartialOrd for [T; N] {
     #[inline]

core/src/intrinsics.rs

@@ -1913,6 +1913,26 @@ extern "rust-intrinsic" {
     /// Allocate at compile time. Should not be called at runtime.
     #[rustc_const_unstable(feature = "const_heap", issue = "79597")]
     pub fn const_allocate(size: usize, align: usize) -> *mut u8;
+
+    /// Determines whether the raw bytes of the two values are equal.
+    ///
+    /// The is particularly handy for arrays, since it allows things like just
+    /// comparing `i96`s instead of forcing `alloca`s for `[6 x i16]`.
+    ///
+    /// Above some backend-decided threshold this will emit calls to `memcmp`,
+    /// like slice equality does, instead of causing massive code size.
+    ///
+    /// # Safety
+    ///
+    /// It's UB to call this if any of the *bytes* in `*a` or `*b` are uninitialized.
+    /// Note that this is a stricter criterion than just the *values* being
+    /// fully-initialized: if `T` has padding, it's UB to call this intrinsic.
+    ///
+    /// (The implementation is allowed to branch on the results of comparisons,
+    /// which is UB if any of their inputs are `undef`.)
+    #[cfg(not(bootstrap))]
+    #[rustc_const_unstable(feature = "const_intrinsic_raw_eq", issue = "none")]
+    pub fn raw_eq<T>(a: &T, b: &T) -> bool;
 }
 
 // Some functions are defined here because they accidentally got made