fix(hid): widen the Bolt per-slot probe budget for high-latency USB paths

[tagawa0525]

Summary

The per-slot Bolt probe cap (3 s) is sized for direct-radio latency. Behind a KVM's USB emulation a healthy HID++ 2.0 feature walk takes several seconds, so every slot hit the cap on every watcher tick and a newly paired device never acquired model info — it stayed permanently identity-only. This raises BOLT_SLOT_PROBE to 10 s. The cap is a ceiling, not a wait: healthy nodes still settle in a couple of seconds, and the concurrent slot walk plus the 25 s PROBE_BUDGET already cover the rest.

Changes

• hid: raise BOLT_SLOT_PROBE from 3 s to 10 s and reword the budget rustdoc to match the current concurrent-walk structure.

Testing

• cargo fmt --all -- --check / cargo clippy --workspace --all-targets -- -D warnings / cargo test --workspace — green on this branch.
• The failure mode was reproduced on real hardware: a KVM-routed Bolt receiver where the 3 s cap reproducibly starved both slots every tick. The widened cap has not yet been runtime-verified on that same path — to test, pair a device through a KVM-routed receiver and confirm model info and capabilities appear within a few ticks.

[greptile-apps]

Greptile Summary

The PR increases the Bolt per-slot HID++ feature-walk timeout from 3 seconds to 10 seconds to accommodate high-latency USB paths.

• Updates the timeout documentation to describe the sequential pairing-register pass and concurrent slot probing.
• Leaves the overall 25-second receiver probe budget unchanged.

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failure remains.

Important Files Changed

Filename	Overview
crates/openlogi-hid/src/inventory.rs	Widens the Bolt slot probe ceiling and updates its timing rationale; no eligible follow-up defect was established.

_{Reviews (4): Last reviewed commit: "docs(hid): attribute the KVM slot starva..." | Re-trigger Greptile}

[Copilot]

Pull request overview

This PR adjusts Bolt receiver probing behavior in openlogi-hid to better tolerate high-latency USB paths (e.g., KVM USB emulation) so paired devices can complete HID++ 2.0 feature discovery and surface full model/capability info.

Changes:

• Increased the Bolt per-slot HID++ 2.0 feature-walk timeout (BOLT_SLOT_PROBE) from 3s to 10s.
• Updated the probe budget rustdoc to reflect the current probing structure (including the sequential Bolt register phase + concurrent per-slot walks).

Suppressed comments (2)

crates/openlogi-hid/src/inventory.rs:74

• This rustdoc says the per-slot cap prevents “pinning its slot future forever”, but the outer node probe is already bounded by PROBE_BUDGET (so it’s not truly forever). Consider rewording to reflect that the cap prevents a single hung slot from stalling the receiver probe until the outer budget elapses.

/// Bounds a single device that stops answering its feature-walk reads (seen on
/// a recent macOS IOHID stack with a new MX Master 4) so it falls back to its
/// cached / identity-only data instead of pinning its slot future forever
/// (#218). Slots walk *concurrently* (mirroring the Unifying path), so this

crates/openlogi-hid/src/inventory.rs:81

• The comment references an “earlier 1 s cap”, but in this PR the cap is being changed from 3 s → 10 s. To keep the historical note consistent with the actual prior value (and the PR summary), this should likely refer to the earlier 3 s cap.

/// high-latency USB paths (a Bolt receiver behind a KVM's USB emulation) it
/// takes several seconds — the earlier 1 s cap starved every slot there, so a
/// newly paired device could never acquire model info at all. 10 s is generous

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[tagawa0525]

Fixed in e765148 — the PROBE_BUDGET rustdoc now scopes the sequential pairing-register pass to the Bolt path explicitly (Unifying's pairing-info registers use a different sub-register base and are not polled), and the BOLT_SLOT_PROBE doc names it as Bolt's.

[tagawa0525]

@greptileai review

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated no new comments.

Suppressed comments (1)

crates/openlogi-hid/src/inventory.rs:82

• The updated BOLT_SLOT_PROBE rustdoc explains the KVM/high-latency failure mode but attributes it to an “earlier 1 s cap”. In this PR the cap being widened is 3 s → 10 s (per PR description and prior code), so the doc should refer to the previous 3 s cap to avoid confusing readers about what change fixed the issue.

/// (the MX Master 4's 45 features take ~1–1.6 s over Bolt even awake), and on
/// high-latency USB paths (a Bolt receiver behind a KVM's USB emulation) it
/// takes several seconds — the earlier 1 s cap starved every slot there, so a
/// newly paired device could never acquire model info at all. 10 s is generous