chore: release v2.2.3

[github-actions]

🤖 New release

• rusty_alloc: 2.2.2 -> 2.2.3 (✓ API compatible changes)
• rusty_alloc-api: 2.2.2 -> 2.2.3

Changelog

rusty_alloc

2.2.3 - 2026-10-05

Fixed

• (segment) a failed re-commit is a failed span, not unbacked memory

Fixed

• A failed re-commit no longer hands out unbacked memory (Windows, purging
  on). With purge_delay >= 0 and purge_decommits on, a freed span is
  decommitted, and reuse re-commits it — but the result of that commit was
  discarded. Windows has no overcommit, so MEM_COMMIT fails once the
  system's commit is exhausted, and the span was carved anyway: the first
  store into it was an access violation in page_extend. One was seen in a
  shipping consumer (secure on, purge_delay = 0, the machine at 98.5 % of
  its commit limit). A failed re-commit is now a failed span: it stays on the
  free list still marked purged, the allocator moves on, and true exhaustion
  ends as an ordinary allocation failure. The default configuration
  (purge_delay = -1) never purges and was not exposed. Details:
  docs/plans/recommit-failure-ignored.md.
• The same for whole segments. segment_free and huge_free restore a
  purged or guarded segment's commit and access before recycling it, and
  dropped both results. A segment the OS will not restore is no longer
  recycled: it is released to the OS, or, inside an arena, retired in place.

Added

• stats::commit_failures(): a process-wide count of commits refused on
  those paths, also printed by stats::print_process as failed commits N.
  Non-zero after a fault in the page layer says the machine ran out of
  commit.

rusty_alloc-api

2.2.1 - 2026-09-25

Performance

• GlobalAlloc methods are #[inline], as the mimalloc crate's are, so
  rustc's __rust_alloc / __rust_dealloc shims carry the fast path instead
  of jumping to it, and dealloc is the free fast path with its null test
  folded away. Measured whole-program against 2.2.0 on two deterministic Rust
  workloads (bench/rust-globalalloc.sh): -21.1 % and -5.8 %.
• Layouts aligned up to two words (16 bytes on 64-bit) come from the natural
  size classes, which are already aligned that far, instead of the aligned
  path; every hashbrown table is such a layout. tests/natural_align.rs pins
  the alignment for every size class through alloc, alloc_zeroed and realloc.
• realloc keeps a block in place at any alignment when it fits instead
  of always allocating, copying and freeing above 8 bytes of alignment:
  -27.3 % on an over-aligned buffer workload.

---

This PR was generated with release-plz.