chore: add vendor directory

1 files changed, 1680 insertions, 0 deletions

diff --git a/vendor/bytes/src/bytes.rs b/vendor/bytes/src/bytes.rs
new file mode 100644
index 00000000..cdb6ea55
--- /dev/null
+++ b/vendor/bytes/src/bytes.rs
@@ -0,0 +1,1680 @@
+use core::iter::FromIterator;
+use core::mem::{self, ManuallyDrop};
+use core::ops::{Deref, RangeBounds};
+use core::ptr::NonNull;
+use core::{cmp, fmt, hash, ptr, slice, usize};
+
+use alloc::{
+    alloc::{dealloc, Layout},
+    borrow::Borrow,
+    boxed::Box,
+    string::String,
+    vec::Vec,
+};
+
+use crate::buf::IntoIter;
+#[allow(unused)]
+use crate::loom::sync::atomic::AtomicMut;
+use crate::loom::sync::atomic::{AtomicPtr, AtomicUsize, Ordering};
+use crate::{offset_from, Buf, BytesMut};
+
+/// A cheaply cloneable and sliceable chunk of contiguous memory.
+///
+/// `Bytes` is an efficient container for storing and operating on contiguous
+/// slices of memory. It is intended for use primarily in networking code, but
+/// could have applications elsewhere as well.
+///
+/// `Bytes` values facilitate zero-copy network programming by allowing multiple
+/// `Bytes` objects to point to the same underlying memory.
+///
+/// `Bytes` does not have a single implementation. It is an interface, whose
+/// exact behavior is implemented through dynamic dispatch in several underlying
+/// implementations of `Bytes`.
+///
+/// All `Bytes` implementations must fulfill the following requirements:
+/// - They are cheaply cloneable and thereby shareable between an unlimited amount
+///   of components, for example by modifying a reference count.
+/// - Instances can be sliced to refer to a subset of the original buffer.
+///
+/// ```
+/// use bytes::Bytes;
+///
+/// let mut mem = Bytes::from("Hello world");
+/// let a = mem.slice(0..5);
+///
+/// assert_eq!(a, "Hello");
+///
+/// let b = mem.split_to(6);
+///
+/// assert_eq!(mem, "world");
+/// assert_eq!(b, "Hello ");
+/// ```
+///
+/// # Memory layout
+///
+/// The `Bytes` struct itself is fairly small, limited to 4 `usize` fields used
+/// to track information about which segment of the underlying memory the
+/// `Bytes` handle has access to.
+///
+/// `Bytes` keeps both a pointer to the shared state containing the full memory
+/// slice and a pointer to the start of the region visible by the handle.
+/// `Bytes` also tracks the length of its view into the memory.
+///
+/// # Sharing
+///
+/// `Bytes` contains a vtable, which allows implementations of `Bytes` to define
+/// how sharing/cloning is implemented in detail.
+/// When `Bytes::clone()` is called, `Bytes` will call the vtable function for
+/// cloning the backing storage in order to share it behind multiple `Bytes`
+/// instances.
+///
+/// For `Bytes` implementations which refer to constant memory (e.g. created
+/// via `Bytes::from_static()`) the cloning implementation will be a no-op.
+///
+/// For `Bytes` implementations which point to a reference counted shared storage
+/// (e.g. an `Arc<[u8]>`), sharing will be implemented by increasing the
+/// reference count.
+///
+/// Due to this mechanism, multiple `Bytes` instances may point to the same
+/// shared memory region.
+/// Each `Bytes` instance can point to different sections within that
+/// memory region, and `Bytes` instances may or may not have overlapping views
+/// into the memory.
+///
+/// The following diagram visualizes a scenario where 2 `Bytes` instances make
+/// use of an `Arc`-based backing storage, and provide access to different views:
+///
+/// ```text
+///
+///    Arc ptrs                   ┌─────────┐
+///    ________________________ / │ Bytes 2 │
+///   /                           └─────────┘
+///  /          ┌───────────┐     |         |
+/// |_________/ │  Bytes 1  │     |         |
+/// |           └───────────┘     |         |
+/// |           |           | ___/ data     | tail
+/// |      data |      tail |/              |
+/// v           v           v               v
+/// ┌─────┬─────┬───────────┬───────────────┬─────┐
+/// │ Arc │     │           │               │     │
+/// └─────┴─────┴───────────┴───────────────┴─────┘
+/// ```
+pub struct Bytes {
+    ptr: *const u8,
+    len: usize,
+    // inlined "trait object"
+    data: AtomicPtr<()>,
+    vtable: &'static Vtable,
+}
+
+pub(crate) struct Vtable {
+    /// fn(data, ptr, len)
+    pub clone: unsafe fn(&AtomicPtr<()>, *const u8, usize) -> Bytes,
+    /// fn(data, ptr, len)
+    ///
+    /// takes `Bytes` to value
+    pub to_vec: unsafe fn(&AtomicPtr<()>, *const u8, usize) -> Vec<u8>,
+    pub to_mut: unsafe fn(&AtomicPtr<()>, *const u8, usize) -> BytesMut,
+    /// fn(data)
+    pub is_unique: unsafe fn(&AtomicPtr<()>) -> bool,
+    /// fn(data, ptr, len)
+    pub drop: unsafe fn(&mut AtomicPtr<()>, *const u8, usize),
+}
+
+impl Bytes {
+    /// Creates a new empty `Bytes`.
+    ///
+    /// This will not allocate and the returned `Bytes` handle will be empty.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::Bytes;
+    ///
+    /// let b = Bytes::new();
+    /// assert_eq!(&b[..], b"");
+    /// ```
+    #[inline]
+    #[cfg(not(all(loom, test)))]
+    pub const fn new() -> Self {
+        // Make it a named const to work around
+        // "unsizing casts are not allowed in const fn"
+        const EMPTY: &[u8] = &[];
+        Bytes::from_static(EMPTY)
+    }
+
+    /// Creates a new empty `Bytes`.
+    #[cfg(all(loom, test))]
+    pub fn new() -> Self {
+        const EMPTY: &[u8] = &[];
+        Bytes::from_static(EMPTY)
+    }
+
+    /// Creates a new `Bytes` from a static slice.
+    ///
+    /// The returned `Bytes` will point directly to the static slice. There is
+    /// no allocating or copying.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::Bytes;
+    ///
+    /// let b = Bytes::from_static(b"hello");
+    /// assert_eq!(&b[..], b"hello");
+    /// ```
+    #[inline]
+    #[cfg(not(all(loom, test)))]
+    pub const fn from_static(bytes: &'static [u8]) -> Self {
+        Bytes {
+            ptr: bytes.as_ptr(),
+            len: bytes.len(),
+            data: AtomicPtr::new(ptr::null_mut()),
+            vtable: &STATIC_VTABLE,
+        }
+    }
+
+    /// Creates a new `Bytes` from a static slice.
+    #[cfg(all(loom, test))]
+    pub fn from_static(bytes: &'static [u8]) -> Self {
+        Bytes {
+            ptr: bytes.as_ptr(),
+            len: bytes.len(),
+            data: AtomicPtr::new(ptr::null_mut()),
+            vtable: &STATIC_VTABLE,
+        }
+    }
+
+    /// Creates a new `Bytes` with length zero and the given pointer as the address.
+    fn new_empty_with_ptr(ptr: *const u8) -> Self {
+        debug_assert!(!ptr.is_null());
+
+        // Detach this pointer's provenance from whichever allocation it came from, and reattach it
+        // to the provenance of the fake ZST [u8;0] at the same address.
+        let ptr = without_provenance(ptr as usize);
+
+        Bytes {
+            ptr,
+            len: 0,
+            data: AtomicPtr::new(ptr::null_mut()),
+            vtable: &STATIC_VTABLE,
+        }
+    }
+
+    /// Create [Bytes] with a buffer whose lifetime is controlled
+    /// via an explicit owner.
+    ///
+    /// A common use case is to zero-copy construct from mapped memory.
+    ///
+    /// ```
+    /// # struct File;
+    /// #
+    /// # impl File {
+    /// #     pub fn open(_: &str) -> Result<Self, ()> {
+    /// #         Ok(Self)
+    /// #     }
+    /// # }
+    /// #
+    /// # mod memmap2 {
+    /// #     pub struct Mmap;
+    /// #
+    /// #     impl Mmap {
+    /// #         pub unsafe fn map(_file: &super::File) -> Result<Self, ()> {
+    /// #             Ok(Self)
+    /// #         }
+    /// #     }
+    /// #
+    /// #     impl AsRef<[u8]> for Mmap {
+    /// #         fn as_ref(&self) -> &[u8] {
+    /// #             b"buf"
+    /// #         }
+    /// #     }
+    /// # }
+    /// use bytes::Bytes;
+    /// use memmap2::Mmap;
+    ///
+    /// # fn main() -> Result<(), ()> {
+    /// let file = File::open("upload_bundle.tar.gz")?;
+    /// let mmap = unsafe { Mmap::map(&file) }?;
+    /// let b = Bytes::from_owner(mmap);
+    /// # Ok(())
+    /// # }
+    /// ```
+    ///
+    /// The `owner` will be transferred to the constructed [Bytes] object, which
+    /// will ensure it is dropped once all remaining clones of the constructed
+    /// object are dropped. The owner will then be responsible for dropping the
+    /// specified region of memory as part of its [Drop] implementation.
+    ///
+    /// Note that converting [Bytes] constructed from an owner into a [BytesMut]
+    /// will always create a deep copy of the buffer into newly allocated memory.
+    pub fn from_owner<T>(owner: T) -> Self
+    where
+        T: AsRef<[u8]> + Send + 'static,
+    {
+        // Safety & Miri:
+        // The ownership of `owner` is first transferred to the `Owned` wrapper and `Bytes` object.
+        // This ensures that the owner is pinned in memory, allowing us to call `.as_ref()` safely
+        // since the lifetime of the owner is controlled by the lifetime of the new `Bytes` object,
+        // and the lifetime of the resulting borrowed `&[u8]` matches that of the owner.
+        // Note that this remains safe so long as we only call `.as_ref()` once.
+        //
+        // There are some additional special considerations here:
+        //   * We rely on Bytes's Drop impl to clean up memory should `.as_ref()` panic.
+        //   * Setting the `ptr` and `len` on the bytes object last (after moving the owner to
+        //     Bytes) allows Miri checks to pass since it avoids obtaining the `&[u8]` slice
+        //     from a stack-owned Box.
+        // More details on this: https://github.com/tokio-rs/bytes/pull/742/#discussion_r1813375863
+        //                  and: https://github.com/tokio-rs/bytes/pull/742/#discussion_r1813316032
+
+        let owned = Box::into_raw(Box::new(Owned {
+            lifetime: OwnedLifetime {
+                ref_cnt: AtomicUsize::new(1),
+                drop: owned_box_and_drop::<T>,
+            },
+            owner,
+        }));
+
+        let mut ret = Bytes {
+            ptr: NonNull::dangling().as_ptr(),
+            len: 0,
+            data: AtomicPtr::new(owned.cast()),
+            vtable: &OWNED_VTABLE,
+        };
+
+        let buf = unsafe { &*owned }.owner.as_ref();
+        ret.ptr = buf.as_ptr();
+        ret.len = buf.len();
+
+        ret
+    }
+
+    /// Returns the number of bytes contained in this `Bytes`.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::Bytes;
+    ///
+    /// let b = Bytes::from(&b"hello"[..]);
+    /// assert_eq!(b.len(), 5);
+    /// ```
+    #[inline]
+    pub const fn len(&self) -> usize {
+        self.len
+    }
+
+    /// Returns true if the `Bytes` has a length of 0.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::Bytes;
+    ///
+    /// let b = Bytes::new();
+    /// assert!(b.is_empty());
+    /// ```
+    #[inline]
+    pub const fn is_empty(&self) -> bool {
+        self.len == 0
+    }
+
+    /// Returns true if this is the only reference to the data and
+    /// `Into<BytesMut>` would avoid cloning the underlying buffer.
+    ///
+    /// Always returns false if the data is backed by a [static slice](Bytes::from_static),
+    /// or an [owner](Bytes::from_owner).
+    ///
+    /// The result of this method may be invalidated immediately if another
+    /// thread clones this value while this is being called. Ensure you have
+    /// unique access to this value (`&mut Bytes`) first if you need to be
+    /// certain the result is valid (i.e. for safety reasons).
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::Bytes;
+    ///
+    /// let a = Bytes::from(vec![1, 2, 3]);
+    /// assert!(a.is_unique());
+    /// let b = a.clone();
+    /// assert!(!a.is_unique());
+    /// ```
+    pub fn is_unique(&self) -> bool {
+        unsafe { (self.vtable.is_unique)(&self.data) }
+    }
+
+    /// Creates `Bytes` instance from slice, by copying it.
+    pub fn copy_from_slice(data: &[u8]) -> Self {
+        data.to_vec().into()
+    }
+
+    /// Returns a slice of self for the provided range.
+    ///
+    /// This will increment the reference count for the underlying memory and
+    /// return a new `Bytes` handle set to the slice.
+    ///
+    /// This operation is `O(1)`.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::Bytes;
+    ///
+    /// let a = Bytes::from(&b"hello world"[..]);
+    /// let b = a.slice(2..5);
+    ///
+    /// assert_eq!(&b[..], b"llo");
+    /// ```
+    ///
+    /// # Panics
+    ///
+    /// Requires that `begin <= end` and `end <= self.len()`, otherwise slicing
+    /// will panic.
+    pub fn slice(&self, range: impl RangeBounds<usize>) -> Self {
+        use core::ops::Bound;
+
+        let len = self.len();
+
+        let begin = match range.start_bound() {
+            Bound::Included(&n) => n,
+            Bound::Excluded(&n) => n.checked_add(1).expect("out of range"),
+            Bound::Unbounded => 0,
+        };
+
+        let end = match range.end_bound() {
+            Bound::Included(&n) => n.checked_add(1).expect("out of range"),
+            Bound::Excluded(&n) => n,
+            Bound::Unbounded => len,
+        };
+
+        assert!(
+            begin <= end,
+            "range start must not be greater than end: {:?} <= {:?}",
+            begin,
+            end,
+        );
+        assert!(
+            end <= len,
+            "range end out of bounds: {:?} <= {:?}",
+            end,
+            len,
+        );
+
+        if end == begin {
+            return Bytes::new();
+        }
+
+        let mut ret = self.clone();
+
+        ret.len = end - begin;
+        ret.ptr = unsafe { ret.ptr.add(begin) };
+
+        ret
+    }
+
+    /// Returns a slice of self that is equivalent to the given `subset`.
+    ///
+    /// When processing a `Bytes` buffer with other tools, one often gets a
+    /// `&[u8]` which is in fact a slice of the `Bytes`, i.e. a subset of it.
+    /// This function turns that `&[u8]` into another `Bytes`, as if one had
+    /// called `self.slice()` with the offsets that correspond to `subset`.
+    ///
+    /// This operation is `O(1)`.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::Bytes;
+    ///
+    /// let bytes = Bytes::from(&b"012345678"[..]);
+    /// let as_slice = bytes.as_ref();
+    /// let subset = &as_slice[2..6];
+    /// let subslice = bytes.slice_ref(&subset);
+    /// assert_eq!(&subslice[..], b"2345");
+    /// ```
+    ///
+    /// # Panics
+    ///
+    /// Requires that the given `sub` slice is in fact contained within the
+    /// `Bytes` buffer; otherwise this function will panic.
+    pub fn slice_ref(&self, subset: &[u8]) -> Self {
+        // Empty slice and empty Bytes may have their pointers reset
+        // so explicitly allow empty slice to be a subslice of any slice.
+        if subset.is_empty() {
+            return Bytes::new();
+        }
+
+        let bytes_p = self.as_ptr() as usize;
+        let bytes_len = self.len();
+
+        let sub_p = subset.as_ptr() as usize;
+        let sub_len = subset.len();
+
+        assert!(
+            sub_p >= bytes_p,
+            "subset pointer ({:p}) is smaller than self pointer ({:p})",
+            subset.as_ptr(),
+            self.as_ptr(),
+        );
+        assert!(
+            sub_p + sub_len <= bytes_p + bytes_len,
+            "subset is out of bounds: self = ({:p}, {}), subset = ({:p}, {})",
+            self.as_ptr(),
+            bytes_len,
+            subset.as_ptr(),
+            sub_len,
+        );
+
+        let sub_offset = sub_p - bytes_p;
+
+        self.slice(sub_offset..(sub_offset + sub_len))
+    }
+
+    /// Splits the bytes into two at the given index.
+    ///
+    /// Afterwards `self` contains elements `[0, at)`, and the returned `Bytes`
+    /// contains elements `[at, len)`. It's guaranteed that the memory does not
+    /// move, that is, the address of `self` does not change, and the address of
+    /// the returned slice is `at` bytes after that.
+    ///
+    /// This is an `O(1)` operation that just increases the reference count and
+    /// sets a few indices.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::Bytes;
+    ///
+    /// let mut a = Bytes::from(&b"hello world"[..]);
+    /// let b = a.split_off(5);
+    ///
+    /// assert_eq!(&a[..], b"hello");
+    /// assert_eq!(&b[..], b" world");
+    /// ```
+    ///
+    /// # Panics
+    ///
+    /// Panics if `at > len`.
+    #[must_use = "consider Bytes::truncate if you don't need the other half"]
+    pub fn split_off(&mut self, at: usize) -> Self {
+        if at == self.len() {
+            return Bytes::new_empty_with_ptr(self.ptr.wrapping_add(at));
+        }
+
+        if at == 0 {
+            return mem::replace(self, Bytes::new_empty_with_ptr(self.ptr));
+        }
+
+        assert!(
+            at <= self.len(),
+            "split_off out of bounds: {:?} <= {:?}",
+            at,
+            self.len(),
+        );
+
+        let mut ret = self.clone();
+
+        self.len = at;
+
+        unsafe { ret.inc_start(at) };
+
+        ret
+    }
+
+    /// Splits the bytes into two at the given index.
+    ///
+    /// Afterwards `self` contains elements `[at, len)`, and the returned
+    /// `Bytes` contains elements `[0, at)`.
+    ///
+    /// This is an `O(1)` operation that just increases the reference count and
+    /// sets a few indices.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::Bytes;
+    ///
+    /// let mut a = Bytes::from(&b"hello world"[..]);
+    /// let b = a.split_to(5);
+    ///
+    /// assert_eq!(&a[..], b" world");
+    /// assert_eq!(&b[..], b"hello");
+    /// ```
+    ///
+    /// # Panics
+    ///
+    /// Panics if `at > len`.
+    #[must_use = "consider Bytes::advance if you don't need the other half"]
+    pub fn split_to(&mut self, at: usize) -> Self {
+        if at == self.len() {
+            let end_ptr = self.ptr.wrapping_add(at);
+            return mem::replace(self, Bytes::new_empty_with_ptr(end_ptr));
+        }
+
+        if at == 0 {
+            return Bytes::new_empty_with_ptr(self.ptr);
+        }
+
+        assert!(
+            at <= self.len(),
+            "split_to out of bounds: {:?} <= {:?}",
+            at,
+            self.len(),
+        );
+
+        let mut ret = self.clone();
+
+        unsafe { self.inc_start(at) };
+
+        ret.len = at;
+        ret
+    }
+
+    /// Shortens the buffer, keeping the first `len` bytes and dropping the
+    /// rest.
+    ///
+    /// If `len` is greater than the buffer's current length, this has no
+    /// effect.
+    ///
+    /// The [split_off](`Self::split_off()`) method can emulate `truncate`, but this causes the
+    /// excess bytes to be returned instead of dropped.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::Bytes;
+    ///
+    /// let mut buf = Bytes::from(&b"hello world"[..]);
+    /// buf.truncate(5);
+    /// assert_eq!(buf, b"hello"[..]);
+    /// ```
+    #[inline]
+    pub fn truncate(&mut self, len: usize) {
+        if len < self.len {
+            // The Vec "promotable" vtables do not store the capacity,
+            // so we cannot truncate while using this repr. We *have* to
+            // promote using `split_off` so the capacity can be stored.
+            if self.vtable as *const Vtable == &PROMOTABLE_EVEN_VTABLE
+                || self.vtable as *const Vtable == &PROMOTABLE_ODD_VTABLE
+            {
+                drop(self.split_off(len));
+            } else {
+                self.len = len;
+            }
+        }
+    }
+
+    /// Clears the buffer, removing all data.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::Bytes;
+    ///
+    /// let mut buf = Bytes::from(&b"hello world"[..]);
+    /// buf.clear();
+    /// assert!(buf.is_empty());
+    /// ```
+    #[inline]
+    pub fn clear(&mut self) {
+        self.truncate(0);
+    }
+
+    /// Try to convert self into `BytesMut`.
+    ///
+    /// If `self` is unique for the entire original buffer, this will succeed
+    /// and return a `BytesMut` with the contents of `self` without copying.
+    /// If `self` is not unique for the entire original buffer, this will fail
+    /// and return self.
+    ///
+    /// This will also always fail if the buffer was constructed via either
+    /// [from_owner](Bytes::from_owner) or [from_static](Bytes::from_static).
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::{Bytes, BytesMut};
+    ///
+    /// let bytes = Bytes::from(b"hello".to_vec());
+    /// assert_eq!(bytes.try_into_mut(), Ok(BytesMut::from(&b"hello"[..])));
+    /// ```
+    pub fn try_into_mut(self) -> Result<BytesMut, Bytes> {
+        if self.is_unique() {
+            Ok(self.into())
+        } else {
+            Err(self)
+        }
+    }
+
+    #[inline]
+    pub(crate) unsafe fn with_vtable(
+        ptr: *const u8,
+        len: usize,
+        data: AtomicPtr<()>,
+        vtable: &'static Vtable,
+    ) -> Bytes {
+        Bytes {
+            ptr,
+            len,
+            data,
+            vtable,
+        }
+    }
+
+    // private
+
+    #[inline]
+    fn as_slice(&self) -> &[u8] {
+        unsafe { slice::from_raw_parts(self.ptr, self.len) }
+    }
+
+    #[inline]
+    unsafe fn inc_start(&mut self, by: usize) {
+        // should already be asserted, but debug assert for tests
+        debug_assert!(self.len >= by, "internal: inc_start out of bounds");
+        self.len -= by;
+        self.ptr = self.ptr.add(by);
+    }
+}
+
+// Vtable must enforce this behavior
+unsafe impl Send for Bytes {}
+unsafe impl Sync for Bytes {}
+
+impl Drop for Bytes {
+    #[inline]
+    fn drop(&mut self) {
+        unsafe { (self.vtable.drop)(&mut self.data, self.ptr, self.len) }
+    }
+}
+
+impl Clone for Bytes {
+    #[inline]
+    fn clone(&self) -> Bytes {
+        unsafe { (self.vtable.clone)(&self.data, self.ptr, self.len) }
+    }
+}
+
+impl Buf for Bytes {
+    #[inline]
+    fn remaining(&self) -> usize {
+        self.len()
+    }
+
+    #[inline]
+    fn chunk(&self) -> &[u8] {
+        self.as_slice()
+    }
+
+    #[inline]
+    fn advance(&mut self, cnt: usize) {
+        assert!(
+            cnt <= self.len(),
+            "cannot advance past `remaining`: {:?} <= {:?}",
+            cnt,
+            self.len(),
+        );
+
+        unsafe {
+            self.inc_start(cnt);
+        }
+    }
+
+    fn copy_to_bytes(&mut self, len: usize) -> Self {
+        self.split_to(len)
+    }
+}
+
+impl Deref for Bytes {
+    type Target = [u8];
+
+    #[inline]
+    fn deref(&self) -> &[u8] {
+        self.as_slice()
+    }
+}
+
+impl AsRef<[u8]> for Bytes {
+    #[inline]
+    fn as_ref(&self) -> &[u8] {
+        self.as_slice()
+    }
+}
+
+impl hash::Hash for Bytes {
+    fn hash<H>(&self, state: &mut H)
+    where
+        H: hash::Hasher,
+    {
+        self.as_slice().hash(state);
+    }
+}
+
+impl Borrow<[u8]> for Bytes {
+    fn borrow(&self) -> &[u8] {
+        self.as_slice()
+    }
+}
+
+impl IntoIterator for Bytes {
+    type Item = u8;
+    type IntoIter = IntoIter<Bytes>;
+
+    fn into_iter(self) -> Self::IntoIter {
+        IntoIter::new(self)
+    }
+}
+
+impl<'a> IntoIterator for &'a Bytes {
+    type Item = &'a u8;
+    type IntoIter = core::slice::Iter<'a, u8>;
+
+    fn into_iter(self) -> Self::IntoIter {
+        self.as_slice().iter()
+    }
+}
+
+impl FromIterator<u8> for Bytes {
+    fn from_iter<T: IntoIterator<Item = u8>>(into_iter: T) -> Self {
+        Vec::from_iter(into_iter).into()
+    }
+}
+
+// impl Eq
+
+impl PartialEq for Bytes {
+    fn eq(&self, other: &Bytes) -> bool {
+        self.as_slice() == other.as_slice()
+    }
+}
+
+impl PartialOrd for Bytes {
+    fn partial_cmp(&self, other: &Bytes) -> Option<cmp::Ordering> {
+        self.as_slice().partial_cmp(other.as_slice())
+    }
+}
+
+impl Ord for Bytes {
+    fn cmp(&self, other: &Bytes) -> cmp::Ordering {
+        self.as_slice().cmp(other.as_slice())
+    }
+}
+
+impl Eq for Bytes {}
+
+impl PartialEq<[u8]> for Bytes {
+    fn eq(&self, other: &[u8]) -> bool {
+        self.as_slice() == other
+    }
+}
+
+impl PartialOrd<[u8]> for Bytes {
+    fn partial_cmp(&self, other: &[u8]) -> Option<cmp::Ordering> {
+        self.as_slice().partial_cmp(other)
+    }
+}
+
+impl PartialEq<Bytes> for [u8] {
+    fn eq(&self, other: &Bytes) -> bool {
+        *other == *self
+    }
+}
+
+impl PartialOrd<Bytes> for [u8] {
+    fn partial_cmp(&self, other: &Bytes) -> Option<cmp::Ordering> {
+        <[u8] as PartialOrd<[u8]>>::partial_cmp(self, other)
+    }
+}
+
+impl PartialEq<str> for Bytes {
+    fn eq(&self, other: &str) -> bool {
+        self.as_slice() == other.as_bytes()
+    }
+}
+
+impl PartialOrd<str> for Bytes {
+    fn partial_cmp(&self, other: &str) -> Option<cmp::Ordering> {
+        self.as_slice().partial_cmp(other.as_bytes())
+    }
+}
+
+impl PartialEq<Bytes> for str {
+    fn eq(&self, other: &Bytes) -> bool {
+        *other == *self
+    }
+}
+
+impl PartialOrd<Bytes> for str {
+    fn partial_cmp(&self, other: &Bytes) -> Option<cmp::Ordering> {
+        <[u8] as PartialOrd<[u8]>>::partial_cmp(self.as_bytes(), other)
+    }
+}
+
+impl PartialEq<Vec<u8>> for Bytes {
+    fn eq(&self, other: &Vec<u8>) -> bool {
+        *self == other[..]
+    }
+}
+
+impl PartialOrd<Vec<u8>> for Bytes {
+    fn partial_cmp(&self, other: &Vec<u8>) -> Option<cmp::Ordering> {
+        self.as_slice().partial_cmp(&other[..])
+    }
+}
+
+impl PartialEq<Bytes> for Vec<u8> {
+    fn eq(&self, other: &Bytes) -> bool {
+        *other == *self
+    }
+}
+
+impl PartialOrd<Bytes> for Vec<u8> {
+    fn partial_cmp(&self, other: &Bytes) -> Option<cmp::Ordering> {
+        <[u8] as PartialOrd<[u8]>>::partial_cmp(self, other)
+    }
+}
+
+impl PartialEq<String> for Bytes {
+    fn eq(&self, other: &String) -> bool {
+        *self == other[..]
+    }
+}
+
+impl PartialOrd<String> for Bytes {
+    fn partial_cmp(&self, other: &String) -> Option<cmp::Ordering> {
+        self.as_slice().partial_cmp(other.as_bytes())
+    }
+}
+
+impl PartialEq<Bytes> for String {
+    fn eq(&self, other: &Bytes) -> bool {
+        *other == *self
+    }
+}
+
+impl PartialOrd<Bytes> for String {
+    fn partial_cmp(&self, other: &Bytes) -> Option<cmp::Ordering> {
+        <[u8] as PartialOrd<[u8]>>::partial_cmp(self.as_bytes(), other)
+    }
+}
+
+impl PartialEq<Bytes> for &[u8] {
+    fn eq(&self, other: &Bytes) -> bool {
+        *other == *self
+    }
+}
+
+impl PartialOrd<Bytes> for &[u8] {
+    fn partial_cmp(&self, other: &Bytes) -> Option<cmp::Ordering> {
+        <[u8] as PartialOrd<[u8]>>::partial_cmp(self, other)
+    }
+}
+
+impl PartialEq<Bytes> for &str {
+    fn eq(&self, other: &Bytes) -> bool {
+        *other == *self
+    }
+}
+
+impl PartialOrd<Bytes> for &str {
+    fn partial_cmp(&self, other: &Bytes) -> Option<cmp::Ordering> {
+        <[u8] as PartialOrd<[u8]>>::partial_cmp(self.as_bytes(), other)
+    }
+}
+
+impl<'a, T: ?Sized> PartialEq<&'a T> for Bytes
+where
+    Bytes: PartialEq<T>,
+{
+    fn eq(&self, other: &&'a T) -> bool {
+        *self == **other
+    }
+}
+
+impl<'a, T: ?Sized> PartialOrd<&'a T> for Bytes
+where
+    Bytes: PartialOrd<T>,
+{
+    fn partial_cmp(&self, other: &&'a T) -> Option<cmp::Ordering> {
+        self.partial_cmp(&**other)
+    }
+}
+
+// impl From
+
+impl Default for Bytes {
+    #[inline]
+    fn default() -> Bytes {
+        Bytes::new()
+    }
+}
+
+impl From<&'static [u8]> for Bytes {
+    fn from(slice: &'static [u8]) -> Bytes {
+        Bytes::from_static(slice)
+    }
+}
+
+impl From<&'static str> for Bytes {
+    fn from(slice: &'static str) -> Bytes {
+        Bytes::from_static(slice.as_bytes())
+    }
+}
+
+impl From<Vec<u8>> for Bytes {
+    fn from(vec: Vec<u8>) -> Bytes {
+        let mut vec = ManuallyDrop::new(vec);
+        let ptr = vec.as_mut_ptr();
+        let len = vec.len();
+        let cap = vec.capacity();
+
+        // Avoid an extra allocation if possible.
+        if len == cap {
+            let vec = ManuallyDrop::into_inner(vec);
+            return Bytes::from(vec.into_boxed_slice());
+        }
+
+        let shared = Box::new(Shared {
+            buf: ptr,
+            cap,
+            ref_cnt: AtomicUsize::new(1),
+        });
+
+        let shared = Box::into_raw(shared);
+        // The pointer should be aligned, so this assert should
+        // always succeed.
+        debug_assert!(
+            0 == (shared as usize & KIND_MASK),
+            "internal: Box<Shared> should have an aligned pointer",
+        );
+        Bytes {
+            ptr,
+            len,
+            data: AtomicPtr::new(shared as _),
+            vtable: &SHARED_VTABLE,
+        }
+    }
+}
+
+impl From<Box<[u8]>> for Bytes {
+    fn from(slice: Box<[u8]>) -> Bytes {
+        // Box<[u8]> doesn't contain a heap allocation for empty slices,
+        // so the pointer isn't aligned enough for the KIND_VEC stashing to
+        // work.
+        if slice.is_empty() {
+            return Bytes::new();
+        }
+
+        let len = slice.len();
+        let ptr = Box::into_raw(slice) as *mut u8;
+
+        if ptr as usize & 0x1 == 0 {
+            let data = ptr_map(ptr, |addr| addr | KIND_VEC);
+            Bytes {
+                ptr,
+                len,
+                data: AtomicPtr::new(data.cast()),
+                vtable: &PROMOTABLE_EVEN_VTABLE,
+            }
+        } else {
+            Bytes {
+                ptr,
+                len,
+                data: AtomicPtr::new(ptr.cast()),
+                vtable: &PROMOTABLE_ODD_VTABLE,
+            }
+        }
+    }
+}
+
+impl From<Bytes> for BytesMut {
+    /// Convert self into `BytesMut`.
+    ///
+    /// If `bytes` is unique for the entire original buffer, this will return a
+    /// `BytesMut` with the contents of `bytes` without copying.
+    /// If `bytes` is not unique for the entire original buffer, this will make
+    /// a copy of `bytes` subset of the original buffer in a new `BytesMut`.
+    ///
+    /// # Examples
+    ///
+    /// ```
+    /// use bytes::{Bytes, BytesMut};
+    ///
+    /// let bytes = Bytes::from(b"hello".to_vec());
+    /// assert_eq!(BytesMut::from(bytes), BytesMut::from(&b"hello"[..]));
+    /// ```
+    fn from(bytes: Bytes) -> Self {
+        let bytes = ManuallyDrop::new(bytes);
+        unsafe { (bytes.vtable.to_mut)(&bytes.data, bytes.ptr, bytes.len) }
+    }
+}
+
+impl From<String> for Bytes {
+    fn from(s: String) -> Bytes {
+        Bytes::from(s.into_bytes())
+    }
+}
+
+impl From<Bytes> for Vec<u8> {
+    fn from(bytes: Bytes) -> Vec<u8> {
+        let bytes = ManuallyDrop::new(bytes);
+        unsafe { (bytes.vtable.to_vec)(&bytes.data, bytes.ptr, bytes.len) }
+    }
+}
+
+// ===== impl Vtable =====
+
+impl fmt::Debug for Vtable {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.debug_struct("Vtable")
+            .field("clone", &(self.clone as *const ()))
+            .field("drop", &(self.drop as *const ()))
+            .finish()
+    }
+}
+
+// ===== impl StaticVtable =====
+
+const STATIC_VTABLE: Vtable = Vtable {
+    clone: static_clone,
+    to_vec: static_to_vec,
+    to_mut: static_to_mut,
+    is_unique: static_is_unique,
+    drop: static_drop,
+};
+
+unsafe fn static_clone(_: &AtomicPtr<()>, ptr: *const u8, len: usize) -> Bytes {
+    let slice = slice::from_raw_parts(ptr, len);
+    Bytes::from_static(slice)
+}
+
+unsafe fn static_to_vec(_: &AtomicPtr<()>, ptr: *const u8, len: usize) -> Vec<u8> {
+    let slice = slice::from_raw_parts(ptr, len);
+    slice.to_vec()
+}
+
+unsafe fn static_to_mut(_: &AtomicPtr<()>, ptr: *const u8, len: usize) -> BytesMut {
+    let slice = slice::from_raw_parts(ptr, len);
+    BytesMut::from(slice)
+}
+
+fn static_is_unique(_: &AtomicPtr<()>) -> bool {
+    false
+}
+
+unsafe fn static_drop(_: &mut AtomicPtr<()>, _: *const u8, _: usize) {
+    // nothing to drop for &'static [u8]
+}
+
+// ===== impl OwnedVtable =====
+
+#[repr(C)]
+struct OwnedLifetime {
+    ref_cnt: AtomicUsize,
+    drop: unsafe fn(*mut ()),
+}
+
+#[repr(C)]
+struct Owned<T> {
+    lifetime: OwnedLifetime,
+    owner: T,
+}
+
+unsafe fn owned_box_and_drop<T>(ptr: *mut ()) {
+    let b: Box<Owned<T>> = Box::from_raw(ptr as _);
+    drop(b);
+}
+
+unsafe fn owned_clone(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> Bytes {
+    let owned = data.load(Ordering::Relaxed);
+    let ref_cnt = &(*owned.cast::<OwnedLifetime>()).ref_cnt;
+    let old_cnt = ref_cnt.fetch_add(1, Ordering::Relaxed);
+    if old_cnt > usize::MAX >> 1 {
+        crate::abort()
+    }
+
+    Bytes {
+        ptr,
+        len,
+        data: AtomicPtr::new(owned as _),
+        vtable: &OWNED_VTABLE,
+    }
+}
+
+unsafe fn owned_to_vec(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> Vec<u8> {
+    let slice = slice::from_raw_parts(ptr, len);
+    let vec = slice.to_vec();
+    owned_drop_impl(data.load(Ordering::Relaxed));
+    vec
+}
+
+unsafe fn owned_to_mut(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> BytesMut {
+    BytesMut::from_vec(owned_to_vec(data, ptr, len))
+}
+
+unsafe fn owned_is_unique(_data: &AtomicPtr<()>) -> bool {
+    false
+}
+
+unsafe fn owned_drop_impl(owned: *mut ()) {
+    let lifetime = owned.cast::<OwnedLifetime>();
+    let ref_cnt = &(*lifetime).ref_cnt;
+
+    let old_cnt = ref_cnt.fetch_sub(1, Ordering::Release);
+    debug_assert!(
+        old_cnt > 0 && old_cnt <= usize::MAX >> 1,
+        "expected non-zero refcount and no underflow"
+    );
+    if old_cnt != 1 {
+        return;
+    }
+    ref_cnt.load(Ordering::Acquire);
+
+    let drop_fn = &(*lifetime).drop;
+    drop_fn(owned)
+}
+
+unsafe fn owned_drop(data: &mut AtomicPtr<()>, _ptr: *const u8, _len: usize) {
+    let owned = data.load(Ordering::Relaxed);
+    owned_drop_impl(owned);
+}
+
+static OWNED_VTABLE: Vtable = Vtable {
+    clone: owned_clone,
+    to_vec: owned_to_vec,
+    to_mut: owned_to_mut,
+    is_unique: owned_is_unique,
+    drop: owned_drop,
+};
+
+// ===== impl PromotableVtable =====
+
+static PROMOTABLE_EVEN_VTABLE: Vtable = Vtable {
+    clone: promotable_even_clone,
+    to_vec: promotable_even_to_vec,
+    to_mut: promotable_even_to_mut,
+    is_unique: promotable_is_unique,
+    drop: promotable_even_drop,
+};
+
+static PROMOTABLE_ODD_VTABLE: Vtable = Vtable {
+    clone: promotable_odd_clone,
+    to_vec: promotable_odd_to_vec,
+    to_mut: promotable_odd_to_mut,
+    is_unique: promotable_is_unique,
+    drop: promotable_odd_drop,
+};
+
+unsafe fn promotable_even_clone(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> Bytes {
+    let shared = data.load(Ordering::Acquire);
+    let kind = shared as usize & KIND_MASK;
+
+    if kind == KIND_ARC {
+        shallow_clone_arc(shared.cast(), ptr, len)
+    } else {
+        debug_assert_eq!(kind, KIND_VEC);
+        let buf = ptr_map(shared.cast(), |addr| addr & !KIND_MASK);
+        shallow_clone_vec(data, shared, buf, ptr, len)
+    }
+}
+
+unsafe fn promotable_to_vec(
+    data: &AtomicPtr<()>,
+    ptr: *const u8,
+    len: usize,
+    f: fn(*mut ()) -> *mut u8,
+) -> Vec<u8> {
+    let shared = data.load(Ordering::Acquire);
+    let kind = shared as usize & KIND_MASK;
+
+    if kind == KIND_ARC {
+        shared_to_vec_impl(shared.cast(), ptr, len)
+    } else {
+        // If Bytes holds a Vec, then the offset must be 0.
+        debug_assert_eq!(kind, KIND_VEC);
+
+        let buf = f(shared);
+
+        let cap = offset_from(ptr, buf) + len;
+
+        // Copy back buffer
+        ptr::copy(ptr, buf, len);
+
+        Vec::from_raw_parts(buf, len, cap)
+    }
+}
+
+unsafe fn promotable_to_mut(
+    data: &AtomicPtr<()>,
+    ptr: *const u8,
+    len: usize,
+    f: fn(*mut ()) -> *mut u8,
+) -> BytesMut {
+    let shared = data.load(Ordering::Acquire);
+    let kind = shared as usize & KIND_MASK;
+
+    if kind == KIND_ARC {
+        shared_to_mut_impl(shared.cast(), ptr, len)
+    } else {
+        // KIND_VEC is a view of an underlying buffer at a certain offset.
+        // The ptr + len always represents the end of that buffer.
+        // Before truncating it, it is first promoted to KIND_ARC.
+        // Thus, we can safely reconstruct a Vec from it without leaking memory.
+        debug_assert_eq!(kind, KIND_VEC);
+
+        let buf = f(shared);
+        let off = offset_from(ptr, buf);
+        let cap = off + len;
+        let v = Vec::from_raw_parts(buf, cap, cap);
+
+        let mut b = BytesMut::from_vec(v);
+        b.advance_unchecked(off);
+        b
+    }
+}
+
+unsafe fn promotable_even_to_vec(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> Vec<u8> {
+    promotable_to_vec(data, ptr, len, |shared| {
+        ptr_map(shared.cast(), |addr| addr & !KIND_MASK)
+    })
+}
+
+unsafe fn promotable_even_to_mut(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> BytesMut {
+    promotable_to_mut(data, ptr, len, |shared| {
+        ptr_map(shared.cast(), |addr| addr & !KIND_MASK)
+    })
+}
+
+unsafe fn promotable_even_drop(data: &mut AtomicPtr<()>, ptr: *const u8, len: usize) {
+    data.with_mut(|shared| {
+        let shared = *shared;
+        let kind = shared as usize & KIND_MASK;
+
+        if kind == KIND_ARC {
+            release_shared(shared.cast());
+        } else {
+            debug_assert_eq!(kind, KIND_VEC);
+            let buf = ptr_map(shared.cast(), |addr| addr & !KIND_MASK);
+            free_boxed_slice(buf, ptr, len);
+        }
+    });
+}
+
+unsafe fn promotable_odd_clone(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> Bytes {
+    let shared = data.load(Ordering::Acquire);
+    let kind = shared as usize & KIND_MASK;
+
+    if kind == KIND_ARC {
+        shallow_clone_arc(shared as _, ptr, len)
+    } else {
+        debug_assert_eq!(kind, KIND_VEC);
+        shallow_clone_vec(data, shared, shared.cast(), ptr, len)
+    }
+}
+
+unsafe fn promotable_odd_to_vec(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> Vec<u8> {
+    promotable_to_vec(data, ptr, len, |shared| shared.cast())
+}
+
+unsafe fn promotable_odd_to_mut(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> BytesMut {
+    promotable_to_mut(data, ptr, len, |shared| shared.cast())
+}
+
+unsafe fn promotable_odd_drop(data: &mut AtomicPtr<()>, ptr: *const u8, len: usize) {
+    data.with_mut(|shared| {
+        let shared = *shared;
+        let kind = shared as usize & KIND_MASK;
+
+        if kind == KIND_ARC {
+            release_shared(shared.cast());
+        } else {
+            debug_assert_eq!(kind, KIND_VEC);
+
+            free_boxed_slice(shared.cast(), ptr, len);
+        }
+    });
+}
+
+unsafe fn promotable_is_unique(data: &AtomicPtr<()>) -> bool {
+    let shared = data.load(Ordering::Acquire);
+    let kind = shared as usize & KIND_MASK;
+
+    if kind == KIND_ARC {
+        let ref_cnt = (*shared.cast::<Shared>()).ref_cnt.load(Ordering::Relaxed);
+        ref_cnt == 1
+    } else {
+        true
+    }
+}
+
+unsafe fn free_boxed_slice(buf: *mut u8, offset: *const u8, len: usize) {
+    let cap = offset_from(offset, buf) + len;
+    dealloc(buf, Layout::from_size_align(cap, 1).unwrap())
+}
+
+// ===== impl SharedVtable =====
+
+struct Shared {
+    // Holds arguments to dealloc upon Drop, but otherwise doesn't use them
+    buf: *mut u8,
+    cap: usize,
+    ref_cnt: AtomicUsize,
+}
+
+impl Drop for Shared {
+    fn drop(&mut self) {
+        unsafe { dealloc(self.buf, Layout::from_size_align(self.cap, 1).unwrap()) }
+    }
+}
+
+// Assert that the alignment of `Shared` is divisible by 2.
+// This is a necessary invariant since we depend on allocating `Shared` a
+// shared object to implicitly carry the `KIND_ARC` flag in its pointer.
+// This flag is set when the LSB is 0.
+const _: [(); 0 - mem::align_of::<Shared>() % 2] = []; // Assert that the alignment of `Shared` is divisible by 2.
+
+static SHARED_VTABLE: Vtable = Vtable {
+    clone: shared_clone,
+    to_vec: shared_to_vec,
+    to_mut: shared_to_mut,
+    is_unique: shared_is_unique,
+    drop: shared_drop,
+};
+
+const KIND_ARC: usize = 0b0;
+const KIND_VEC: usize = 0b1;
+const KIND_MASK: usize = 0b1;
+
+unsafe fn shared_clone(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> Bytes {
+    let shared = data.load(Ordering::Relaxed);
+    shallow_clone_arc(shared as _, ptr, len)
+}
+
+unsafe fn shared_to_vec_impl(shared: *mut Shared, ptr: *const u8, len: usize) -> Vec<u8> {
+    // Check that the ref_cnt is 1 (unique).
+    //
+    // If it is unique, then it is set to 0 with AcqRel fence for the same
+    // reason in release_shared.
+    //
+    // Otherwise, we take the other branch and call release_shared.
+    if (*shared)
+        .ref_cnt
+        .compare_exchange(1, 0, Ordering::AcqRel, Ordering::Relaxed)
+        .is_ok()
+    {
+        // Deallocate the `Shared` instance without running its destructor.
+        let shared = *Box::from_raw(shared);
+        let shared = ManuallyDrop::new(shared);
+        let buf = shared.buf;
+        let cap = shared.cap;
+
+        // Copy back buffer
+        ptr::copy(ptr, buf, len);
+
+        Vec::from_raw_parts(buf, len, cap)
+    } else {
+        let v = slice::from_raw_parts(ptr, len).to_vec();
+        release_shared(shared);
+        v
+    }
+}
+
+unsafe fn shared_to_vec(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> Vec<u8> {
+    shared_to_vec_impl(data.load(Ordering::Relaxed).cast(), ptr, len)
+}
+
+unsafe fn shared_to_mut_impl(shared: *mut Shared, ptr: *const u8, len: usize) -> BytesMut {
+    // The goal is to check if the current handle is the only handle
+    // that currently has access to the buffer. This is done by
+    // checking if the `ref_cnt` is currently 1.
+    //
+    // The `Acquire` ordering synchronizes with the `Release` as
+    // part of the `fetch_sub` in `release_shared`. The `fetch_sub`
+    // operation guarantees that any mutations done in other threads
+    // are ordered before the `ref_cnt` is decremented. As such,
+    // this `Acquire` will guarantee that those mutations are
+    // visible to the current thread.
+    //
+    // Otherwise, we take the other branch, copy the data and call `release_shared`.
+    if (*shared).ref_cnt.load(Ordering::Acquire) == 1 {
+        // Deallocate the `Shared` instance without running its destructor.
+        let shared = *Box::from_raw(shared);
+        let shared = ManuallyDrop::new(shared);
+        let buf = shared.buf;
+        let cap = shared.cap;
+
+        // Rebuild Vec
+        let off = offset_from(ptr, buf);
+        let v = Vec::from_raw_parts(buf, len + off, cap);
+
+        let mut b = BytesMut::from_vec(v);
+        b.advance_unchecked(off);
+        b
+    } else {
+        // Copy the data from Shared in a new Vec, then release it
+        let v = slice::from_raw_parts(ptr, len).to_vec();
+        release_shared(shared);
+        BytesMut::from_vec(v)
+    }
+}
+
+unsafe fn shared_to_mut(data: &AtomicPtr<()>, ptr: *const u8, len: usize) -> BytesMut {
+    shared_to_mut_impl(data.load(Ordering::Relaxed).cast(), ptr, len)
+}
+
+pub(crate) unsafe fn shared_is_unique(data: &AtomicPtr<()>) -> bool {
+    let shared = data.load(Ordering::Acquire);
+    let ref_cnt = (*shared.cast::<Shared>()).ref_cnt.load(Ordering::Relaxed);
+    ref_cnt == 1
+}
+
+unsafe fn shared_drop(data: &mut AtomicPtr<()>, _ptr: *const u8, _len: usize) {
+    data.with_mut(|shared| {
+        release_shared(shared.cast());
+    });
+}
+
+unsafe fn shallow_clone_arc(shared: *mut Shared, ptr: *const u8, len: usize) -> Bytes {
+    let old_size = (*shared).ref_cnt.fetch_add(1, Ordering::Relaxed);
+
+    if old_size > usize::MAX >> 1 {
+        crate::abort();
+    }
+
+    Bytes {
+        ptr,
+        len,
+        data: AtomicPtr::new(shared as _),
+        vtable: &SHARED_VTABLE,
+    }
+}
+
+#[cold]
+unsafe fn shallow_clone_vec(
+    atom: &AtomicPtr<()>,
+    ptr: *const (),
+    buf: *mut u8,
+    offset: *const u8,
+    len: usize,
+) -> Bytes {
+    // If the buffer is still tracked in a `Vec<u8>`. It is time to
+    // promote the vec to an `Arc`. This could potentially be called
+    // concurrently, so some care must be taken.
+
+    // First, allocate a new `Shared` instance containing the
+    // `Vec` fields. It's important to note that `ptr`, `len`,
+    // and `cap` cannot be mutated without having `&mut self`.
+    // This means that these fields will not be concurrently
+    // updated and since the buffer hasn't been promoted to an
+    // `Arc`, those three fields still are the components of the
+    // vector.
+    let shared = Box::new(Shared {
+        buf,
+        cap: offset_from(offset, buf) + len,
+        // Initialize refcount to 2. One for this reference, and one
+        // for the new clone that will be returned from
+        // `shallow_clone`.
+        ref_cnt: AtomicUsize::new(2),
+    });
+
+    let shared = Box::into_raw(shared);
+
+    // The pointer should be aligned, so this assert should
+    // always succeed.
+    debug_assert!(
+        0 == (shared as usize & KIND_MASK),
+        "internal: Box<Shared> should have an aligned pointer",
+    );
+
+    // Try compare & swapping the pointer into the `arc` field.
+    // `Release` is used synchronize with other threads that
+    // will load the `arc` field.
+    //
+    // If the `compare_exchange` fails, then the thread lost the
+    // race to promote the buffer to shared. The `Acquire`
+    // ordering will synchronize with the `compare_exchange`
+    // that happened in the other thread and the `Shared`
+    // pointed to by `actual` will be visible.
+    match atom.compare_exchange(ptr as _, shared as _, Ordering::AcqRel, Ordering::Acquire) {
+        Ok(actual) => {
+            debug_assert!(actual as usize == ptr as usize);
+            // The upgrade was successful, the new handle can be
+            // returned.
+            Bytes {
+                ptr: offset,
+                len,
+                data: AtomicPtr::new(shared as _),
+                vtable: &SHARED_VTABLE,
+            }
+        }
+        Err(actual) => {
+            // The upgrade failed, a concurrent clone happened. Release
+            // the allocation that was made in this thread, it will not
+            // be needed.
+            let shared = Box::from_raw(shared);
+            mem::forget(*shared);
+
+            // Buffer already promoted to shared storage, so increment ref
+            // count.
+            shallow_clone_arc(actual as _, offset, len)
+        }
+    }
+}
+
+unsafe fn release_shared(ptr: *mut Shared) {
+    // `Shared` storage... follow the drop steps from Arc.
+    if (*ptr).ref_cnt.fetch_sub(1, Ordering::Release) != 1 {
+        return;
+    }
+
+    // This fence is needed to prevent reordering of use of the data and
+    // deletion of the data.  Because it is marked `Release`, the decreasing
+    // of the reference count synchronizes with this `Acquire` fence. This
+    // means that use of the data happens before decreasing the reference
+    // count, which happens before this fence, which happens before the
+    // deletion of the data.
+    //
+    // As explained in the [Boost documentation][1],
+    //
+    // > It is important to enforce any possible access to the object in one
+    // > thread (through an existing reference) to *happen before* deleting
+    // > the object in a different thread. This is achieved by a "release"
+    // > operation after dropping a reference (any access to the object
+    // > through this reference must obviously happened before), and an
+    // > "acquire" operation before deleting the object.
+    //
+    // [1]: (www.boost.org/doc/libs/1_55_0/doc/html/atomic/usage_examples.html)
+    //
+    // Thread sanitizer does not support atomic fences. Use an atomic load
+    // instead.
+    (*ptr).ref_cnt.load(Ordering::Acquire);
+
+    // Drop the data
+    drop(Box::from_raw(ptr));
+}
+
+// Ideally we would always use this version of `ptr_map` since it is strict
+// provenance compatible, but it results in worse codegen. We will however still
+// use it on miri because it gives better diagnostics for people who test bytes
+// code with miri.
+//
+// See https://github.com/tokio-rs/bytes/pull/545 for more info.
+#[cfg(miri)]
+fn ptr_map<F>(ptr: *mut u8, f: F) -> *mut u8
+where
+    F: FnOnce(usize) -> usize,
+{
+    let old_addr = ptr as usize;
+    let new_addr = f(old_addr);
+    let diff = new_addr.wrapping_sub(old_addr);
+    ptr.wrapping_add(diff)
+}
+
+#[cfg(not(miri))]
+fn ptr_map<F>(ptr: *mut u8, f: F) -> *mut u8
+where
+    F: FnOnce(usize) -> usize,
+{
+    let old_addr = ptr as usize;
+    let new_addr = f(old_addr);
+    new_addr as *mut u8
+}
+
+fn without_provenance(ptr: usize) -> *const u8 {
+    core::ptr::null::<u8>().wrapping_add(ptr)
+}
+
+// compile-fails
+
+/// ```compile_fail
+/// use bytes::Bytes;
+/// #[deny(unused_must_use)]
+/// {
+///     let mut b1 = Bytes::from("hello world");
+///     b1.split_to(6);
+/// }
+/// ```
+fn _split_to_must_use() {}
+
+/// ```compile_fail
+/// use bytes::Bytes;
+/// #[deny(unused_must_use)]
+/// {
+///     let mut b1 = Bytes::from("hello world");
+///     b1.split_off(6);
+/// }
+/// ```
+fn _split_off_must_use() {}
+
+// fuzz tests
+#[cfg(all(test, loom))]
+mod fuzz {
+    use loom::sync::Arc;
+    use loom::thread;
+
+    use super::Bytes;
+    #[test]
+    fn bytes_cloning_vec() {
+        loom::model(|| {
+            let a = Bytes::from(b"abcdefgh".to_vec());
+            let addr = a.as_ptr() as usize;
+
+            // test the Bytes::clone is Sync by putting it in an Arc
+            let a1 = Arc::new(a);
+            let a2 = a1.clone();
+
+            let t1 = thread::spawn(move || {
+                let b: Bytes = (*a1).clone();
+                assert_eq!(b.as_ptr() as usize, addr);
+            });
+
+            let t2 = thread::spawn(move || {
+                let b: Bytes = (*a2).clone();
+                assert_eq!(b.as_ptr() as usize, addr);
+            });
+
+            t1.join().unwrap();
+            t2.join().unwrap();
+        });
+    }
+}