Expand description
AsyncUringStream - drives io_uring reads into GPU-visible memory
and advances the megakernel tail pointer on each completion.
The critical safety contract: every byte read lands in a
GpuMappedBuffer. Compatibility ingest uses registered
host-visible GPU mappings; canonical native ingest uses BAR1 peer memory
via GpuMappedBuffer::from_bar1_peer_with_owner plus NVMe passthrough.
The io_uring writer never targets an ordinary userspace bounce buffer.
Structs§
- Async
Uring Stream - Streaming reader that pushes chunked reads into an io_uring SQ and advances an atomic tail pointer the megakernel observes.
- GpuMapped
Buffer - GPU-visible memory region that io_uring is allowed to DMA into.
- Iovec
- Minimal
iovecstruct matching the Linux ABI forreadv.
Constants§
- IORING_
OP_ READV IORING_OP_READV- scatter-read into an array of iovecs.- IORING_
OP_ READ_ FIXED IORING_OP_READ_FIXED- read into a pre-registered buffer.- IORING_
OP_ URING_ CMD IORING_OP_URING_CMD- vendor-specific passthrough (NVMe). Kernel 6.0+.