Merge tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-nextgrafted

Pull networking updates from Jakub Kicinski:
 "Core:

   - Add dedicated kmem_cache for typical/small skb->head, avoid having
     to access struct page at kfree time, and improve memory use.

   - Introduce sysctl to set default RPS configuration for new netdevs.

   - Define Netlink protocol specification format which can be used to
     describe messages used by each family and auto-generate parsers.
     Add tools for generating kernel data structures and uAPI headers.

   - Expose all net/core sysctls inside netns.

   - Remove 4s sleep in netpoll if carrier is instantly detected on
     boot.

   - Add configurable limit of MDB entries per port, and port-vlan.

   - Continue populating drop reasons throughout the stack.

   - Retire a handful of legacy Qdiscs and classifiers.

  Protocols:

   - Support IPv4 big TCP (TSO frames larger than 64kB).

   - Add IP_LOCAL_PORT_RANGE socket option, to control local port range
     on socket by socket basis.

   - Track and report in procfs number of MPTCP sockets used.

   - Support mixing IPv4 and IPv6 flows in the in-kernel MPTCP path
     manager.

   - IPv6: don't check net.ipv6.route.max_size and rely on garbage
     collection to free memory (similarly to IPv4).

   - Support Penultimate Segment Pop (PSP) flavor in SRv6 (RFC8986).

   - ICMP: add per-rate limit counters.

   - Add support for user scanning requests in ieee802154.

   - Remove static WEP support.

   - Support minimal Wi-Fi 7 Extremely High Throughput (EHT) rate
     reporting.

   - WiFi 7 EHT channel puncturing support (client & AP).

  BPF:

   - Add a rbtree data structure following the "next-gen data structure"
     precedent set by recently added linked list, that is, by using
     kfunc + kptr instead of adding a new BPF map type.

   - Expose XDP hints via kfuncs with initial support for RX hash and
     timestamp metadata.

   - Add BPF_F_NO_TUNNEL_KEY extension to bpf_skb_set_tunnel_key to
     better support decap on GRE tunnel devices not operating in collect
     metadata.

   - Improve x86 JIT's codegen for PROBE_MEM runtime error checks.

   - Remove the need for trace_printk_lock for bpf_trace_printk and
     bpf_trace_vprintk helpers.

   - Extend libbpf's bpf_tracing.h support for tracing arguments of
     kprobes/uprobes and syscall as a special case.

   - Significantly reduce the search time for module symbols by
     livepatch and BPF.

   - Enable cpumasks to be used as kptrs, which is useful for tracing
     programs tracking which tasks end up running on which CPUs in
     different time intervals.

   - Add support for BPF trampoline on s390x and riscv64.

   - Add capability to export the XDP features supported by the NIC.

   - Add __bpf_kfunc tag for marking kernel functions as kfuncs.

   - Add cgroup.memory=nobpf kernel parameter option to disable BPF
     memory accounting for container environments.

  Netfilter:

   - Remove the CLUSTERIP target. It has been marked as obsolete for
     years, and we still have WARN splats wrt races of the out-of-band
     /proc interface installed by this target.

   - Add 'destroy' commands to nf_tables. They are identical to the
     existing 'delete' commands, but do not return an error if the
     referenced object (set, chain, rule...) did not exist.

  Driver API:

   - Improve cpumask_local_spread() locality to help NICs set the right
     IRQ affinity on AMD platforms.

   - Separate C22 and C45 MDIO bus transactions more clearly.

   - Introduce new DCB table to control DSCP rewrite on egress.

   - Support configuration of Physical Layer Collision Avoidance (PLCA)
     Reconciliation Sublayer (RS) (802.3cg-2019). Modern version of
     shared medium Ethernet.

   - Support for MAC Merge layer (IEEE 802.3-2018 clause 99). Allowing
     preemption of low priority frames by high priority frames.

   - Add support for controlling MACSec offload using netlink SET.

   - Rework devlink instance refcounts to allow registration and
     de-registration under the instance lock. Split the code into
     multiple files, drop some of the unnecessarily granular locks and
     factor out common parts of netlink operation handling.

   - Add TX frame aggregation parameters (for USB drivers).

   - Add a new attr TCA_EXT_WARN_MSG to report TC (offload) warning
     messages with notifications for debug.

   - Allow offloading of UDP NEW connections via act_ct.

   - Add support for per action HW stats in TC.

   - Support hardware miss to TC action (continue processing in SW from
     a specific point in the action chain).

   - Warn if old Wireless Extension user space interface is used with
     modern cfg80211/mac80211 drivers. Do not support Wireless
     Extensions for Wi-Fi 7 devices at all. Everyone should switch to
     using nl80211 interface instead.

   - Improve the CAN bit timing configuration. Use extack to return
     error messages directly to user space, update the SJW handling,
     including the definition of a new default value that will benefit
     CAN-FD controllers, by increasing their oscillator tolerance.

  New hardware / drivers:

   - Ethernet:
      - nVidia BlueField-3 support (control traffic driver)
      - Ethernet support for imx93 SoCs
      - Motorcomm yt8531 gigabit Ethernet PHY
      - onsemi NCN26000 10BASE-T1S PHY (with support for PLCA)
      - Microchip LAN8841 PHY (incl. cable diagnostics and PTP)
      - Amlogic gxl MDIO mux

   - WiFi:
      - RealTek RTL8188EU (rtl8xxxu)
      - Qualcomm Wi-Fi 7 devices (ath12k)

   - CAN:
      - Renesas R-Car V4H

  Drivers:

   - Bluetooth:
      - Set Per Platform Antenna Gain (PPAG) for Intel controllers.

   - Ethernet NICs:
      - Intel (1G, igc):
         - support TSN / Qbv / packet scheduling features of i226 model
      - Intel (100G, ice):
         - use GNSS subsystem instead of TTY
         - multi-buffer XDP support
         - extend support for GPIO pins to E823 devices
      - nVidia/Mellanox:
         - update the shared buffer configuration on PFC commands
         - implement PTP adjphase function for HW offset control
         - TC support for Geneve and GRE with VF tunnel offload
         - more efficient crypto key management method
         - multi-port eswitch support
      - Netronome/Corigine:
         - add DCB IEEE support
         - support IPsec offloading for NFP3800
      - Freescale/NXP (enetc):
         - support XDP_REDIRECT for XDP non-linear buffers
         - improve reconfig, avoid link flap and waiting for idle
         - support MAC Merge layer
      - Other NICs:
         - sfc/ef100: add basic devlink support for ef100
         - ionic: rx_push mode operation (writing descriptors via MMIO)
         - bnxt: use the auxiliary bus abstraction for RDMA
         - r8169: disable ASPM and reset bus in case of tx timeout
         - cpsw: support QSGMII mode for J721e CPSW9G
         - cpts: support pulse-per-second output
         - ngbe: add an mdio bus driver
         - usbnet: optimize usbnet_bh() by avoiding unnecessary queuing
         - r8152: handle devices with FW with NCM support
         - amd-xgbe: support 10Mbps, 2.5GbE speeds and rx-adaptation
         - virtio-net: support multi buffer XDP
         - virtio/vsock: replace virtio_vsock_pkt with sk_buff
         - tsnep: XDP support

   - Ethernet high-speed switches:
      - nVidia/Mellanox (mlxsw):
         - add support for latency TLV (in FW control messages)
      - Microchip (sparx5):
         - separate explicit and implicit traffic forwarding rules, make
           the implicit rules always active
         - add support for egress DSCP rewrite
         - IS0 VCAP support (Ingress Classification)
         - IS2 VCAP filters (protos, L3 addrs, L4 ports, flags, ToS
           etc.)
         - ES2 VCAP support (Egress Access Control)
         - support for Per-Stream Filtering and Policing (802.1Q,
           8.6.5.1)

   - Ethernet embedded switches:
      - Marvell (mv88e6xxx):
         - add MAB (port auth) offload support
         - enable PTP receive for mv88e6390
      - NXP (ocelot):
         - support MAC Merge layer
         - support for the the vsc7512 internal copper phys
      - Microchip:
         - lan9303: convert to PHYLINK
         - lan966x: support TC flower filter statistics
         - lan937x: PTP support for KSZ9563/KSZ8563 and LAN937x
         - lan937x: support Credit Based Shaper configuration
         - ksz9477: support Energy Efficient Ethernet
      - other:
         - qca8k: convert to regmap read/write API, use bulk operations
         - rswitch: Improve TX timestamp accuracy

   - Intel WiFi (iwlwifi):
      - EHT (Wi-Fi 7) rate reporting
      - STEP equalizer support: transfer some STEP (connection to radio
        on platforms with integrated wifi) related parameters from the
        BIOS to the firmware.

   - Qualcomm 802.11ax WiFi (ath11k):
      - IPQ5018 support
      - Fine Timing Measurement (FTM) responder role support
      - channel 177 support

   - MediaTek WiFi (mt76):
      - per-PHY LED support
      - mt7996: EHT (Wi-Fi 7) support
      - Wireless Ethernet Dispatch (WED) reset support
      - switch to using page pool allocator

   - RealTek WiFi (rtw89):
      - support new version of Bluetooth co-existance

   - Mobile:
      - rmnet: support TX aggregation"

* tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (1872 commits)
  page_pool: add a comment explaining the fragment counter usage
  net: ethtool: fix __ethtool_dev_mm_supported() implementation
  ethtool: pse-pd: Fix double word in comments
  xsk: add linux/vmalloc.h to xsk.c
  sefltests: netdevsim: wait for devlink instance after netns removal
  selftest: fib_tests: Always cleanup before exit
  net/mlx5e: Align IPsec ASO result memory to be as required by hardware
  net/mlx5e: TC, Set CT miss to the specific ct action instance
  net/mlx5e: Rename CHAIN_TO_REG to MAPPED_OBJ_TO_REG
  net/mlx5: Refactor tc miss handling to a single function
  net/mlx5: Kconfig: Make tc offload depend on tc skb extension
  net/sched: flower: Support hardware miss to tc action
  net/sched: flower: Move filter handle initialization earlier
  net/sched: cls_api: Support hardware miss to tc action
  net/sched: Rename user cookie and act cookie
  sfc: fix builds without CONFIG_RTC_LIB
  sfc: clean up some inconsistent indentings
  net/mlx4_en: Introduce flexible array to silence overflow warning
  net: lan966x: Fix possible deadlock inside PTP
  net/ulp: Remove redundant ->clone() test in inet_clone_ulp().
  ...

1 files changed, 558 insertions, 0 deletions

diff --git a/rust/alloc/raw_vec.rs b/rust/alloc/raw_vec.rs
new file mode 100644
index 000000000..eb77db5de
--- /dev/null
+++ b/rust/alloc/raw_vec.rs
@@ -0,0 +1,558 @@
+// SPDX-License-Identifier: Apache-2.0 OR MIT
+
+#![unstable(feature = "raw_vec_internals", reason = "unstable const warnings", issue = "none")]
+
+use core::alloc::LayoutError;
+use core::cmp;
+use core::intrinsics;
+use core::mem::{self, ManuallyDrop, MaybeUninit};
+use core::ops::Drop;
+use core::ptr::{self, NonNull, Unique};
+use core::slice;
+
+#[cfg(not(no_global_oom_handling))]
+use crate::alloc::handle_alloc_error;
+use crate::alloc::{Allocator, Global, Layout};
+use crate::boxed::Box;
+use crate::collections::TryReserveError;
+use crate::collections::TryReserveErrorKind::*;
+
+#[cfg(test)]
+mod tests;
+
+enum AllocInit {
+    /// The contents of the new memory are uninitialized.
+    Uninitialized,
+    /// The new memory is guaranteed to be zeroed.
+    #[allow(dead_code)]
+    Zeroed,
+}
+
+/// A low-level utility for more ergonomically allocating, reallocating, and deallocating
+/// a buffer of memory on the heap without having to worry about all the corner cases
+/// involved. This type is excellent for building your own data structures like Vec and VecDeque.
+/// In particular:
+///
+/// * Produces `Unique::dangling()` on zero-sized types.
+/// * Produces `Unique::dangling()` on zero-length allocations.
+/// * Avoids freeing `Unique::dangling()`.
+/// * Catches all overflows in capacity computations (promotes them to "capacity overflow" panics).
+/// * Guards against 32-bit systems allocating more than isize::MAX bytes.
+/// * Guards against overflowing your length.
+/// * Calls `handle_alloc_error` for fallible allocations.
+/// * Contains a `ptr::Unique` and thus endows the user with all related benefits.
+/// * Uses the excess returned from the allocator to use the largest available capacity.
+///
+/// This type does not in anyway inspect the memory that it manages. When dropped it *will*
+/// free its memory, but it *won't* try to drop its contents. It is up to the user of `RawVec`
+/// to handle the actual things *stored* inside of a `RawVec`.
+///
+/// Note that the excess of a zero-sized types is always infinite, so `capacity()` always returns
+/// `usize::MAX`. This means that you need to be careful when round-tripping this type with a
+/// `Box<[T]>`, since `capacity()` won't yield the length.
+#[allow(missing_debug_implementations)]
+pub(crate) struct RawVec<T, A: Allocator = Global> {
+    ptr: Unique<T>,
+    cap: usize,
+    alloc: A,
+}
+
+impl<T> RawVec<T, Global> {
+    /// HACK(Centril): This exists because stable `const fn` can only call stable `const fn`, so
+    /// they cannot call `Self::new()`.
+    ///
+    /// If you change `RawVec<T>::new` or dependencies, please take care to not introduce anything
+    /// that would truly const-call something unstable.
+    pub const NEW: Self = Self::new();
+
+    /// Creates the biggest possible `RawVec` (on the system heap)
+    /// without allocating. If `T` has positive size, then this makes a
+    /// `RawVec` with capacity `0`. If `T` is zero-sized, then it makes a
+    /// `RawVec` with capacity `usize::MAX`. Useful for implementing
+    /// delayed allocation.
+    #[must_use]
+    pub const fn new() -> Self {
+        Self::new_in(Global)
+    }
+
+    /// Creates a `RawVec` (on the system heap) with exactly the
+    /// capacity and alignment requirements for a `[T; capacity]`. This is
+    /// equivalent to calling `RawVec::new` when `capacity` is `0` or `T` is
+    /// zero-sized. Note that if `T` is zero-sized this means you will
+    /// *not* get a `RawVec` with the requested capacity.
+    ///
+    /// # Panics
+    ///
+    /// Panics if the requested capacity exceeds `isize::MAX` bytes.
+    ///
+    /// # Aborts
+    ///
+    /// Aborts on OOM.
+    #[cfg(not(any(no_global_oom_handling, test)))]
+    #[must_use]
+    #[inline]
+    pub fn with_capacity(capacity: usize) -> Self {
+        Self::with_capacity_in(capacity, Global)
+    }
+
+    /// Like `with_capacity`, but guarantees the buffer is zeroed.
+    #[cfg(not(any(no_global_oom_handling, test)))]
+    #[must_use]
+    #[inline]
+    pub fn with_capacity_zeroed(capacity: usize) -> Self {
+        Self::with_capacity_zeroed_in(capacity, Global)
+    }
+}
+
+impl<T, A: Allocator> RawVec<T, A> {
+    // Tiny Vecs are dumb. Skip to:
+    // - 8 if the element size is 1, because any heap allocators is likely
+    //   to round up a request of less than 8 bytes to at least 8 bytes.
+    // - 4 if elements are moderate-sized (<= 1 KiB).
+    // - 1 otherwise, to avoid wasting too much space for very short Vecs.
+    pub(crate) const MIN_NON_ZERO_CAP: usize = if mem::size_of::<T>() == 1 {
+        8
+    } else if mem::size_of::<T>() <= 1024 {
+        4
+    } else {
+        1
+    };
+
+    /// Like `new`, but parameterized over the choice of allocator for
+    /// the returned `RawVec`.
+    pub const fn new_in(alloc: A) -> Self {
+        // `cap: 0` means "unallocated". zero-sized types are ignored.
+        Self { ptr: Unique::dangling(), cap: 0, alloc }
+    }
+
+    /// Like `with_capacity`, but parameterized over the choice of
+    /// allocator for the returned `RawVec`.
+    #[cfg(not(no_global_oom_handling))]
+    #[inline]
+    pub fn with_capacity_in(capacity: usize, alloc: A) -> Self {
+        Self::allocate_in(capacity, AllocInit::Uninitialized, alloc)
+    }
+
+    /// Like `try_with_capacity`, but parameterized over the choice of
+    /// allocator for the returned `RawVec`.
+    #[inline]
+    pub fn try_with_capacity_in(capacity: usize, alloc: A) -> Result<Self, TryReserveError> {
+        Self::try_allocate_in(capacity, AllocInit::Uninitialized, alloc)
+    }
+
+    /// Like `with_capacity_zeroed`, but parameterized over the choice
+    /// of allocator for the returned `RawVec`.
+    #[cfg(not(no_global_oom_handling))]
+    #[inline]
+    pub fn with_capacity_zeroed_in(capacity: usize, alloc: A) -> Self {
+        Self::allocate_in(capacity, AllocInit::Zeroed, alloc)
+    }
+
+    /// Converts the entire buffer into `Box<[MaybeUninit<T>]>` with the specified `len`.
+    ///
+    /// Note that this will correctly reconstitute any `cap` changes
+    /// that may have been performed. (See description of type for details.)
+    ///
+    /// # Safety
+    ///
+    /// * `len` must be greater than or equal to the most recently requested capacity, and
+    /// * `len` must be less than or equal to `self.capacity()`.
+    ///
+    /// Note, that the requested capacity and `self.capacity()` could differ, as
+    /// an allocator could overallocate and return a greater memory block than requested.
+    pub unsafe fn into_box(self, len: usize) -> Box<[MaybeUninit<T>], A> {
+        // Sanity-check one half of the safety requirement (we cannot check the other half).
+        debug_assert!(
+            len <= self.capacity(),
+            "`len` must be smaller than or equal to `self.capacity()`"
+        );
+
+        let me = ManuallyDrop::new(self);
+        unsafe {
+            let slice = slice::from_raw_parts_mut(me.ptr() as *mut MaybeUninit<T>, len);
+            Box::from_raw_in(slice, ptr::read(&me.alloc))
+        }
+    }
+
+    #[cfg(not(no_global_oom_handling))]
+    fn allocate_in(capacity: usize, init: AllocInit, alloc: A) -> Self {
+        // Don't allocate here because `Drop` will not deallocate when `capacity` is 0.
+        if mem::size_of::<T>() == 0 || capacity == 0 {
+            Self::new_in(alloc)
+        } else {
+            // We avoid `unwrap_or_else` here because it bloats the amount of
+            // LLVM IR generated.
+            let layout = match Layout::array::<T>(capacity) {
+                Ok(layout) => layout,
+                Err(_) => capacity_overflow(),
+            };
+            match alloc_guard(layout.size()) {
+                Ok(_) => {}
+                Err(_) => capacity_overflow(),
+            }
+            let result = match init {
+                AllocInit::Uninitialized => alloc.allocate(layout),
+                AllocInit::Zeroed => alloc.allocate_zeroed(layout),
+            };
+            let ptr = match result {
+                Ok(ptr) => ptr,
+                Err(_) => handle_alloc_error(layout),
+            };
+
+            // Allocators currently return a `NonNull<[u8]>` whose length
+            // matches the size requested. If that ever changes, the capacity
+            // here should change to `ptr.len() / mem::size_of::<T>()`.
+            Self {
+                ptr: unsafe { Unique::new_unchecked(ptr.cast().as_ptr()) },
+                cap: capacity,
+                alloc,
+            }
+        }
+    }
+
+    fn try_allocate_in(capacity: usize, init: AllocInit, alloc: A) -> Result<Self, TryReserveError> {
+        // Don't allocate here because `Drop` will not deallocate when `capacity` is 0.
+        if mem::size_of::<T>() == 0 || capacity == 0 {
+            return Ok(Self::new_in(alloc));
+        }
+
+        let layout = Layout::array::<T>(capacity).map_err(|_| CapacityOverflow)?;
+        alloc_guard(layout.size())?;
+        let result = match init {
+            AllocInit::Uninitialized => alloc.allocate(layout),
+            AllocInit::Zeroed => alloc.allocate_zeroed(layout),
+        };
+        let ptr = result.map_err(|_| AllocError { layout, non_exhaustive: () })?;
+
+        // Allocators currently return a `NonNull<[u8]>` whose length
+        // matches the size requested. If that ever changes, the capacity
+        // here should change to `ptr.len() / mem::size_of::<T>()`.
+        Ok(Self {
+            ptr: unsafe { Unique::new_unchecked(ptr.cast().as_ptr()) },
+            cap: capacity,
+            alloc,
+        })
+    }
+
+    /// Reconstitutes a `RawVec` from a pointer, capacity, and allocator.
+    ///
+    /// # Safety
+    ///
+    /// The `ptr` must be allocated (via the given allocator `alloc`), and with the given
+    /// `capacity`.
+    /// The `capacity` cannot exceed `isize::MAX` for sized types. (only a concern on 32-bit
+    /// systems). ZST vectors may have a capacity up to `usize::MAX`.
+    /// If the `ptr` and `capacity` come from a `RawVec` created via `alloc`, then this is
+    /// guaranteed.
+    #[inline]
+    pub unsafe fn from_raw_parts_in(ptr: *mut T, capacity: usize, alloc: A) -> Self {
+        Self { ptr: unsafe { Unique::new_unchecked(ptr) }, cap: capacity, alloc }
+    }
+
+    /// Gets a raw pointer to the start of the allocation. Note that this is
+    /// `Unique::dangling()` if `capacity == 0` or `T` is zero-sized. In the former case, you must
+    /// be careful.
+    #[inline]
+    pub fn ptr(&self) -> *mut T {
+        self.ptr.as_ptr()
+    }
+
+    /// Gets the capacity of the allocation.
+    ///
+    /// This will always be `usize::MAX` if `T` is zero-sized.
+    #[inline(always)]
+    pub fn capacity(&self) -> usize {
+        if mem::size_of::<T>() == 0 { usize::MAX } else { self.cap }
+    }
+
+    /// Returns a shared reference to the allocator backing this `RawVec`.
+    pub fn allocator(&self) -> &A {
+        &self.alloc
+    }
+
+    fn current_memory(&self) -> Option<(NonNull<u8>, Layout)> {
+        if mem::size_of::<T>() == 0 || self.cap == 0 {
+            None
+        } else {
+            // We have an allocated chunk of memory, so we can bypass runtime
+            // checks to get our current layout.
+            unsafe {
+                let layout = Layout::array::<T>(self.cap).unwrap_unchecked();
+                Some((self.ptr.cast().into(), layout))
+            }
+        }
+    }
+
+    /// Ensures that the buffer contains at least enough space to hold `len +
+    /// additional` elements. If it doesn't already have enough capacity, will
+    /// reallocate enough space plus comfortable slack space to get amortized
+    /// *O*(1) behavior. Will limit this behavior if it would needlessly cause
+    /// itself to panic.
+    ///
+    /// If `len` exceeds `self.capacity()`, this may fail to actually allocate
+    /// the requested space. This is not really unsafe, but the unsafe
+    /// code *you* write that relies on the behavior of this function may break.
+    ///
+    /// This is ideal for implementing a bulk-push operation like `extend`.
+    ///
+    /// # Panics
+    ///
+    /// Panics if the new capacity exceeds `isize::MAX` bytes.
+    ///
+    /// # Aborts
+    ///
+    /// Aborts on OOM.
+    #[cfg(not(no_global_oom_handling))]
+    #[inline]
+    pub fn reserve(&mut self, len: usize, additional: usize) {
+        // Callers expect this function to be very cheap when there is already sufficient capacity.
+        // Therefore, we move all the resizing and error-handling logic from grow_amortized and
+        // handle_reserve behind a call, while making sure that this function is likely to be
+        // inlined as just a comparison and a call if the comparison fails.
+        #[cold]
+        fn do_reserve_and_handle<T, A: Allocator>(
+            slf: &mut RawVec<T, A>,
+            len: usize,
+            additional: usize,
+        ) {
+            handle_reserve(slf.grow_amortized(len, additional));
+        }
+
+        if self.needs_to_grow(len, additional) {
+            do_reserve_and_handle(self, len, additional);
+        }
+    }
+
+    /// A specialized version of `reserve()` used only by the hot and
+    /// oft-instantiated `Vec::push()`, which does its own capacity check.
+    #[cfg(not(no_global_oom_handling))]
+    #[inline(never)]
+    pub fn reserve_for_push(&mut self, len: usize) {
+        handle_reserve(self.grow_amortized(len, 1));
+    }
+
+    /// The same as `reserve`, but returns on errors instead of panicking or aborting.
+    pub fn try_reserve(&mut self, len: usize, additional: usize) -> Result<(), TryReserveError> {
+        if self.needs_to_grow(len, additional) {
+            self.grow_amortized(len, additional)
+        } else {
+            Ok(())
+        }
+    }
+
+    /// The same as `reserve_for_push`, but returns on errors instead of panicking or aborting.
+    #[inline(never)]
+    pub fn try_reserve_for_push(&mut self, len: usize) -> Result<(), TryReserveError> {
+        self.grow_amortized(len, 1)
+    }
+
+    /// Ensures that the buffer contains at least enough space to hold `len +
+    /// additional` elements. If it doesn't already, will reallocate the
+    /// minimum possible amount of memory necessary. Generally this will be
+    /// exactly the amount of memory necessary, but in principle the allocator
+    /// is free to give back more than we asked for.
+    ///
+    /// If `len` exceeds `self.capacity()`, this may fail to actually allocate
+    /// the requested space. This is not really unsafe, but the unsafe code
+    /// *you* write that relies on the behavior of this function may break.
+    ///
+    /// # Panics
+    ///
+    /// Panics if the new capacity exceeds `isize::MAX` bytes.
+    ///
+    /// # Aborts
+    ///
+    /// Aborts on OOM.
+    #[cfg(not(no_global_oom_handling))]
+    pub fn reserve_exact(&mut self, len: usize, additional: usize) {
+        handle_reserve(self.try_reserve_exact(len, additional));
+    }
+
+    /// The same as `reserve_exact`, but returns on errors instead of panicking or aborting.
+    pub fn try_reserve_exact(
+        &mut self,
+        len: usize,
+        additional: usize,
+    ) -> Result<(), TryReserveError> {
+        if self.needs_to_grow(len, additional) { self.grow_exact(len, additional) } else { Ok(()) }
+    }
+
+    /// Shrinks the buffer down to the specified capacity. If the given amount
+    /// is 0, actually completely deallocates.
+    ///
+    /// # Panics
+    ///
+    /// Panics if the given amount is *larger* than the current capacity.
+    ///
+    /// # Aborts
+    ///
+    /// Aborts on OOM.
+    #[cfg(not(no_global_oom_handling))]
+    pub fn shrink_to_fit(&mut self, cap: usize) {
+        handle_reserve(self.shrink(cap));
+    }
+}
+
+impl<T, A: Allocator> RawVec<T, A> {
+    /// Returns if the buffer needs to grow to fulfill the needed extra capacity.
+    /// Mainly used to make inlining reserve-calls possible without inlining `grow`.
+    fn needs_to_grow(&self, len: usize, additional: usize) -> bool {
+        additional > self.capacity().wrapping_sub(len)
+    }
+
+    fn set_ptr_and_cap(&mut self, ptr: NonNull<[u8]>, cap: usize) {
+        // Allocators currently return a `NonNull<[u8]>` whose length matches
+        // the size requested. If that ever changes, the capacity here should
+        // change to `ptr.len() / mem::size_of::<T>()`.
+        self.ptr = unsafe { Unique::new_unchecked(ptr.cast().as_ptr()) };
+        self.cap = cap;
+    }
+
+    // This method is usually instantiated many times. So we want it to be as
+    // small as possible, to improve compile times. But we also want as much of
+    // its contents to be statically computable as possible, to make the
+    // generated code run faster. Therefore, this method is carefully written
+    // so that all of the code that depends on `T` is within it, while as much
+    // of the code that doesn't depend on `T` as possible is in functions that
+    // are non-generic over `T`.
+    fn grow_amortized(&mut self, len: usize, additional: usize) -> Result<(), TryReserveError> {
+        // This is ensured by the calling contexts.
+        debug_assert!(additional > 0);
+
+        if mem::size_of::<T>() == 0 {
+            // Since we return a capacity of `usize::MAX` when `elem_size` is
+            // 0, getting to here necessarily means the `RawVec` is overfull.
+            return Err(CapacityOverflow.into());
+        }
+
+        // Nothing we can really do about these checks, sadly.
+        let required_cap = len.checked_add(additional).ok_or(CapacityOverflow)?;
+
+        // This guarantees exponential growth. The doubling cannot overflow
+        // because `cap <= isize::MAX` and the type of `cap` is `usize`.
+        let cap = cmp::max(self.cap * 2, required_cap);
+        let cap = cmp::max(Self::MIN_NON_ZERO_CAP, cap);
+
+        let new_layout = Layout::array::<T>(cap);
+
+        // `finish_grow` is non-generic over `T`.
+        let ptr = finish_grow(new_layout, self.current_memory(), &mut self.alloc)?;
+        self.set_ptr_and_cap(ptr, cap);
+        Ok(())
+    }
+
+    // The constraints on this method are much the same as those on
+    // `grow_amortized`, but this method is usually instantiated less often so
+    // it's less critical.
+    fn grow_exact(&mut self, len: usize, additional: usize) -> Result<(), TryReserveError> {
+        if mem::size_of::<T>() == 0 {
+            // Since we return a capacity of `usize::MAX` when the type size is
+            // 0, getting to here necessarily means the `RawVec` is overfull.
+            return Err(CapacityOverflow.into());
+        }
+
+        let cap = len.checked_add(additional).ok_or(CapacityOverflow)?;
+        let new_layout = Layout::array::<T>(cap);
+
+        // `finish_grow` is non-generic over `T`.
+        let ptr = finish_grow(new_layout, self.current_memory(), &mut self.alloc)?;
+        self.set_ptr_and_cap(ptr, cap);
+        Ok(())
+    }
+
+    #[allow(dead_code)]
+    fn shrink(&mut self, cap: usize) -> Result<(), TryReserveError> {
+        assert!(cap <= self.capacity(), "Tried to shrink to a larger capacity");
+
+        let (ptr, layout) = if let Some(mem) = self.current_memory() { mem } else { return Ok(()) };
+
+        let ptr = unsafe {
+            // `Layout::array` cannot overflow here because it would have
+            // overflowed earlier when capacity was larger.
+            let new_layout = Layout::array::<T>(cap).unwrap_unchecked();
+            self.alloc
+                .shrink(ptr, layout, new_layout)
+                .map_err(|_| AllocError { layout: new_layout, non_exhaustive: () })?
+        };
+        self.set_ptr_and_cap(ptr, cap);
+        Ok(())
+    }
+}
+
+// This function is outside `RawVec` to minimize compile times. See the comment
+// above `RawVec::grow_amortized` for details. (The `A` parameter isn't
+// significant, because the number of different `A` types seen in practice is
+// much smaller than the number of `T` types.)
+#[inline(never)]
+fn finish_grow<A>(
+    new_layout: Result<Layout, LayoutError>,
+    current_memory: Option<(NonNull<u8>, Layout)>,
+    alloc: &mut A,
+) -> Result<NonNull<[u8]>, TryReserveError>
+where
+    A: Allocator,
+{
+    // Check for the error here to minimize the size of `RawVec::grow_*`.
+    let new_layout = new_layout.map_err(|_| CapacityOverflow)?;
+
+    alloc_guard(new_layout.size())?;
+
+    let memory = if let Some((ptr, old_layout)) = current_memory {
+        debug_assert_eq!(old_layout.align(), new_layout.align());
+        unsafe {
+            // The allocator checks for alignment equality
+            intrinsics::assume(old_layout.align() == new_layout.align());
+            alloc.grow(ptr, old_layout, new_layout)
+        }
+    } else {
+        alloc.allocate(new_layout)
+    };
+
+    memory.map_err(|_| AllocError { layout: new_layout, non_exhaustive: () }.into())
+}
+
+unsafe impl<#[may_dangle] T, A: Allocator> Drop for RawVec<T, A> {
+    /// Frees the memory owned by the `RawVec` *without* trying to drop its contents.
+    fn drop(&mut self) {
+        if let Some((ptr, layout)) = self.current_memory() {
+            unsafe { self.alloc.deallocate(ptr, layout) }
+        }
+    }
+}
+
+// Central function for reserve error handling.
+#[cfg(not(no_global_oom_handling))]
+#[inline]
+fn handle_reserve(result: Result<(), TryReserveError>) {
+    match result.map_err(|e| e.kind()) {
+        Err(CapacityOverflow) => capacity_overflow(),
+        Err(AllocError { layout, .. }) => handle_alloc_error(layout),
+        Ok(()) => { /* yay */ }
+    }
+}
+
+// We need to guarantee the following:
+// * We don't ever allocate `> isize::MAX` byte-size objects.
+// * We don't overflow `usize::MAX` and actually allocate too little.
+//
+// On 64-bit we just need to check for overflow since trying to allocate
+// `> isize::MAX` bytes will surely fail. On 32-bit and 16-bit we need to add
+// an extra guard for this in case we're running on a platform which can use
+// all 4GB in user-space, e.g., PAE or x32.
+
+#[inline]
+fn alloc_guard(alloc_size: usize) -> Result<(), TryReserveError> {
+    if usize::BITS < 64 && alloc_size > isize::MAX as usize {
+        Err(CapacityOverflow.into())
+    } else {
+        Ok(())
+    }
+}
+
+// One central function responsible for reporting capacity overflows. This'll
+// ensure that the code generation related to these panics is minimal as there's
+// only one location which panics rather than a bunch throughout the module.
+#[cfg(not(no_global_oom_handling))]
+fn capacity_overflow() -> ! {
+    panic!("capacity overflow");
+}