fix(deploy): probe-rs spawns via run_command_blocking — no console flash, 120s timeout

[zackees]

Problem

run_probe_rs_download / run_probe_rs_reset (crates/fbuild-deploy/src/probe_rs.rs) spawned probe-rs with raw std::process::Command::output(). The daemon is windowless on Windows, so every SWD flash allocated a fresh console for the console-subsystem probe-rs.exe — a visible console window flashed on the user's screen during each deploy. Observed live on an LPC845-BRK deploy today.

Two additional gaps from bypassing the blessed wrapper:

• No timeout: a wedged probe (the exact CMSIS-DAP v1.0.7 failure family this path exists to mitigate) would hold the daemon's spawn_blocking slot forever.
• No MSYS env stripping.

Fix

Route both wrappers through fbuild_core::subprocess::run_command_blocking — the house-standard spawn path (ban_raw_subprocess dylint) that already sets CREATE_NO_WINDOW (0x08000000) on Windows, strips MSYS env vars, and applies a timeout. New PROBE_RS_TIMEOUT = 120 s: a healthy LPC845-BRK flash completes in ~2 s; past 120 s the probe is wedged and the deploy fails with the captured stderr instead of hanging.

ToolOutput {stdout, stderr, exit_code} maps 1:1 onto ProbeRsRun, so callers are unchanged.

Why the dylint didn't catch this

ban_raw_subprocess covers exactly this pattern and probe_rs.rs is not allowlisted — but the Dylint CI workflow has been failing on main at the driver-build step (dylint::library_packages::build_library, toolchain-alias retry exhaustion) for at least the last 5 runs. Red gate = zero enforcement, and #935 merged through the gap. Filing the CI-infra issue separately; this PR just removes the violation.

Validation

• soldr cargo check -p fbuild-deploy — clean.
• Hardware: LPC845-BRK on COM10, Windows 10, LPC-Link2 CMSIS-DAP v1.0.7. fbuild deploy -e lpc845 via the FastLED autoresearch harness: flash completes (~2 s SWD via probe-rs), VCOM reopens immediately, freshly-flashed firmware answers JSON-RPC. Console-flash absence verified visually on a post-fix deploy.

Refs: #935, #936 (probe-rs SWD dispatch), #264 (ban_raw_subprocess).

🤖 Generated with Claude Code

Summary by CodeRabbit

• Bug Fixes
  • Improved reliability when running probe-rs actions by enforcing a timeout, helping prevent commands from hanging indefinitely.
  • Simplified command result handling so download and reset operations report exit status and output more consistently.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 86396255-8890-4cb4-955f-6d6b71d54f49

📥 Commits

Reviewing files that changed from the base of the PR and between 64f8950 and aa78800.

📒 Files selected for processing (1)

• crates/fbuild-deploy/src/probe_rs.rs

---

📝 Walkthrough

Walkthrough

Refactors probe_rs.rs to replace direct std::process::Command usage with the run_command_blocking helper, adding a 120-second PROBE_RS_TIMEOUT constant. Both run_probe_rs_download and run_probe_rs_reset now build ProbeRsRun directly from the helper's exit_code, stdout, and stderr fields.

Changes

Probe-rs subprocess timeout migration

Layer / File(s)	Summary
Timeout constant and helper import crates/fbuild-deploy/src/probe_rs.rs	Adds Duration and run_command_blocking imports, and defines PROBE_RS_TIMEOUT (120s) as a hard ceiling for probe-rs subprocess execution.
Apply timeout to download and reset flows crates/fbuild-deploy/src/probe_rs.rs	Updates run_probe_rs_download and run_probe_rs_reset to use run_command_blocking with PROBE_RS_TIMEOUT, convert argv to &str refs, and build ProbeRsRun from the helper's exit_code/stdout/stderr instead of manual Command output decoding.

Estimated code review effort: 2 (Simple) | ~10 minutes

Sequence Diagram(s)

sequenceDiagram
participant Caller
participant ProbeRsModule as run_probe_rs_download/run_probe_rs_reset
participant Subprocess as run_command_blocking
Caller->>ProbeRsModule: invoke probe-rs command
ProbeRsModule->>Subprocess: run_command_blocking(argv_refs, Some(PROBE_RS_TIMEOUT))
Subprocess-->>ProbeRsModule: exit_code, stdout, stderr
ProbeRsModule-->>Caller: ProbeRsRun

Loading

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the switch to run_command_blocking and the added 120-second timeout for probe-rs execution.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/probe-rs-no-window

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}