Fix: add mutex protection in mcast_membership_report_on_query to prevent data race condition

[soopel]

No description provided.

[soopel]

Please check if there is no deadlock created by adding this mutex.

[soopel]

Please check if there is no deadlock created by adding this mutex.

Here is analysis by AI

Analysis Summary:

✅ No deadlock risk detected! Here's why:

Functions that lock group_mutex:

mcast_membership_report_on_query (our fix) - line 192
mt_mcast_join - line 572
mt_mcast_leave - line 677
Functions called while holding the lock in mcast_membership_report_on_query:

rte_pktmbuf_alloc - DPDK function, no mutex
mcast_fill_ipv4 - simple helper, no mutex usage
mcast_create_group_record_on_query - inline helper, no mutex usage
Only accesses data structures, no function calls that could acquire locks
Callers of mcast_membership_report_on_query:

mcast_membership_report_cb (timer callback) - no mutex usage
mt_mcast_restore - no mutex usage
mt_mcast_parse - no mutex usage
No recursive locking:

None of the callers hold group_mutex before calling the function
None of the callees try to acquire group_mutex
The mutex is not recursive (standard pthread mutex)
Conclusion: The mutex addition is safe and correct. There is no deadlock risk because:

No nested locking of the same mutex
No circular dependency with other locks
Callers don't hold the lock before calling
Functions called while holding the lock don't try to acquire it
The fix properly protects the shared data (group_num and group_list) without introducing deadlock risks.