Docs · P1

Large binaries & Ghidra startup deadlines

Why the first query dies with provider_timeout while the machine is still crunching.

What is happening

Headless Ghidra often imports and auto-analyzes before serving answers. On big binaries that work can exceed the startup deadline, so the client sees a timeout even though analysis continues in another process lifetime.

Mitigations builders use

Raise startup patience via env such as REA_GHIDRA_STARTUP_TIMEOUT_MS (confirm current name in upstream docs/README for your version).
Give the JVM enough heap as documented for Ghidra headless.
Prefer analyzing a smaller first slice when the product allows scoped imports.
Community feature asks: open an already-imported Ghidra project so import/analysis happen outside the interactive deadline — track upstream rather than inventing unsupported flags.

Debug loop

npx -y rea-agents@latest doctor --provider ghidra --json
# retry once after a cold JVM warm; then adjust timeout/heap

Also see TMPDIR pitfalls and Windows × Ghidra.