[Márk Mihalik | SiliconLabs (silabs-MarkM)]

Issue

• Address and Thread Sanitizer findings #265
• JIRA: ZWAVE_SDK-2741

Closes: None

Change

TX complete and RX report handlers raced on session status and expiry. Copy or update session fields under the lock and re-arm the timer after release.

Checklist

• Pull request template has been filled out correctly
• ./ci/scripts/clang-tidy.sh check has been run locally (see CONTRIBUTING.md)
• Documentation has been updated if necessary
• A Contribution License Agreement has been established between Silicon Labs (@SiliconLabs) and author's company (matching email domain)

[siliconlabsdevops-svc-devops]

GitHub PR artifact is not available. Re-run CI, then apply the CTT label again.

[Márk Mihalik | SiliconLabs (silabs-MarkM)]

bugbot run

[cursor]

PR Summary

Medium Risk
Changes concurrency and timeout behavior on the Z-Wave Supervision TX/RX path; incorrect locking or callback ordering could mis-report command outcomes, but scope is limited to the supervision command class.

Overview
Fixes races between TX completion and incoming Supervision Reports by guarding the session map with a recursive mutex and routing all mutations through new locked helpers.

Session lifecycle: zwave_command_class_supervision_copy_session_id, zwave_command_class_supervision_handle_send_data_complete, and zwave_command_class_supervision_handle_incoming_report update status, expiry, and tx_info under supervision_sessions_lock, then invoke user callbacks and re-arm the supervision timer after releasing the lock. Report parsing in command_class_supervision.cpp and the TX complete path in zwave_command_class_supervision.c now delegate to these APIs instead of touching session pointers directly.

Timer behavior: Expiry scanning batches failed sessions, fires callbacks outside the lock, and zwave_command_class_supervision_restart_timer arms the next deadline while holding the same mutex so concurrent TX/RX updates cannot overwrite a newer expiry. Wire session IDs use an atomic next_session_id; send path reads the ID via copy_session_id rather than dereferencing a session after create.

^{Reviewed by Cursor Bugbot for commit a412098. Configure here.}

[cursor]

Stale Bugbot comment from a previous run.

[Márk Mihalik | SiliconLabs (silabs-MarkM)]

bugbot run

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit a412098. Configure here.}

[siliconlabsdevops-svc-devops]

CTT Jenkins job: $zpc_ctt_jenkins_url/job/zw-ctt-automation-multibranch/job/main/4157/

[Peter Balogh (silabs-peterb)]

CTT Jenkins job: $zpc_ctt_jenkins_url/job/zw-ctt-automation-multibranch/job/main/4157/

No regression was introduced 👌