Sure, but before we fire it up, let’s inventory the code: find all shared mutable state, lock usage, and any platform‑specific quirks. Static tools love to flag things that don’t actually race. Give me the tool’s version, its concurrency model, and a list of the modules we’re targeting. Then we can compare its findings against a fresh code‑review audit and see if it’s catching real problems or just blowing a fuse.

Running StaticRace 2.4.1 now. I’ll flag every hit that touches shared memory in user‑session, order‑processor, or inventory‑sync. Then we’ll line‑by‑line verify each claim against the actual lock graph and data‑flow. If the analyzer flags a mutex that’s never actually contended, we’ll ignore it—static analysis loves to be paranoid. Keep your eye on the false‑positive list; that’s usually where the real work is.

StaticRace 2.4.1 report – shared memory access candidates 1. Module: user‑session • File: session.c – Line 58: write to session->last_login without holding session_lock – Line 112: read of session->active_flag not protected by session_lock • File: auth.c – Line 47: update of global user_map without mutex 2. Module: order‑processor • File: process.c – Line 204: increment of order_counter without order_lock – Line 342: read of inventory_db while only holding read_lock (possible race if write occurs) 3. Module: inventory‑sync • File: sync.c – Line 89: write to shared inventory_cache without cache_lock – Line 150: read of inventory_cache after releasing cache_lock False positives flagged by the analyzer (ignore these): • File: user‑session.c, Line 30 – conditional lock acquisition that is never hit in practice • File: order‑processor.c, Line 298 – lock guard is static and never released in the execution path Next steps: - Open each file, locate the exact lines, and check whether the lock is actually held in all execution paths. - For lines 58 and 112 in session.c, verify that session_lock is acquired before access. If not, add it. - In auth.c line 47, confirm that user_map is a global that can be modified concurrently; if so, protect with a mutex. - In process.c line 204, ensure order_lock is acquired around the counter increment. - For inventory‑sync, confirm cache_lock usage around the cache reads/writes. Once you have the manual audit ready, compare each flagged line. If a lock is truly missing, mark it as a fix; if the analyzer missed a lock, that’s a new issue for the human review to catch. Good luck with the line‑by‑line scrubbing.