refactor(fspy): simplify shm_io to a plain frame arena

[wan9chi]

Motivation

The gated API reads shared memory only when the gate is closed and nothing was in flight, which proves every claimed frame was written in full; the gate guard is released after the frame drops. The three-state frame header (0 / +size / -size), the atomic header stores and the release fences in shm_io existed so a reader could tolerate crashed or in-flight writers. Those states are unreachable through the gated API.

This PR deletes them. The arena keeps one atomic, its end-offset word. A frame becomes | size: u32 | content | padding |, written once at claim time with a plain store: the claim made the range exclusive, and the gate release publishes the content. Crash exclusion lives in the gate, where a leaked count means the run is not cached. The arena stops being a public surface and becomes an internal of the gated module. Mapping::as_slice goes away with its last caller, which leaves fspy_shm without unsafe API.

write_encoded enters the gate before it sizes the value, so a post-close call reports Claim(Closed) for any value, including one that encodes to zero bytes.

No public API change: GatedShmWriter, GatedShmReceiver, GatedShmReader, FrameMut, ClaimError and StillWriting keep their signatures. The channel layer, the fspy crate and both preload clients are untouched.

🤖 Generated with Claude Code

[wan9chi]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• refactor(fspy): simplify shm_io to a plain frame arena #578 👈 (View in Graphite)
• fix(fspy): replace the IPC file lock with an in-mapping close gate #577
• fix(ipc): use named FIFOs for task communication on Unix #569
• fix(fspy): unify shared memory on a sparse temp file across all platforms #576
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[github-actions]

fspy benchmark

linux

dynamic/launch             change  +0.66%  [ -7.65% ..  +9.35%]  overhead  +281.48%
dynamic/access             change  -3.46%  [-18.94% .. +22.92%]  overhead   +62.44%
static/launch              change  +0.10%  [ -9.48% ..  +9.74%]  overhead  +672.91%
static/access              change  +2.01%  [-10.66% .. +18.96%]  overhead  +859.68%

macos

dynamic/launch             change  +0.13%  [ -2.26% ..  +2.28%]  overhead  +218.89%
dynamic/access             change  -3.06%  [ -6.05% ..  -0.59%]  overhead   +25.30%

windows

dynamic/launch             change  -0.90%  [ -7.63% ..  +5.81%]  overhead   +26.09%
dynamic/access             change  +0.00%  [ -1.50% ..  +1.18%]  overhead    +9.66%

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: b03e50b979

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".