Kernel version: 3.10, loaded with only the smallest kernel and file system
Phenomenon: 1 frequent sleep scheduling, occasionally system first caton
Positioning process:
1, appear the phenomenon, the top view CPU normal
2, check the interrupt, appear the phenomenon, hpet interrupt also gets stuck, without counting statistics continue to
Scheduling after normal, normal statistics,
3, through Linux tool tools perf, only irq related statistics, consulting AE normal
4, card immediately, check the CPU call stack
Continue to follow up, reply is found in the resume scheduling leaf function gets stuck,
5, through the EJTAG tool view, only the CPU interrupt when relevant statistics, the following
Some experts met?