Crate rustfs_uring 

Cancel-safe async io_uring read backend for RustFS (rustfs/backlog#894, hardened per the #1048/#1051 audit).

This crate proves and enforces the ownership model any production io_uring integration in RustFS must follow:

• The read buffer and the file handle are owned by the driver’s pending (orphan) table from SQE submission until the CQE arrives. The kernel may write into the buffer at any point in that window, so nothing else is allowed to free or move its heap allocation.
• Dropping the caller-side future only abandons the result. It never touches the buffer. Optionally it submits IORING_OP_ASYNC_CANCEL to accelerate the CQE; reclamation still happens only at the CQE.
• Driver shutdown cancels all in-flight ops and drains the ring to in_flight == 0 (with a bounded escape hatch) before the ring is unmapped.

Status: read path only, Linux only. The driver supports positioned buffered reads, O_DIRECT reads with internal alignment, sharded rings, async backpressure, eventfd-driven reaping, graceful restricted-environment detection, and bounded shutdown drain. The write path is intentionally out of scope for 0.1.0; see the design notes for the invariant details.

Structs

StatsSnapshot

Point-in-time copy of the driver counters.

UringDriver

Enums

ProbeFailure

Why the probe refused to start the io_uring driver.