System hanged due to deadlock between CPUs.

Issue

• Our customer encountered a problem that NMI interrupt was received due to BMC watchdog timeout.
• Analyzing vmcore, it was seen that it was in deadlock status between CPU1and CPU2, and all CPUs were in hardware interrupt disabled status.
• From this situation, fork() on CPU0 also waited indefinitely for lock acquirement and then WDTThread running on CPU0 could not work and accordingly went into watchdog timeout.