Hi Paul,

FYI, I got the below call trace dump when rebooting kernel inside KVM:

 * Asking all remaining processes to terminate...       
[  102.397875] Task dump for CPU 1:
[  102.398126] swapper/1       R running      0     0      1 0x00000000
[  102.398645]  810c9ad7 810ef4bb af31848b 00000598 50ce7b76 ffffffff 8e53b380 00000017
[  102.399320]  8e98cd45 00000000 8e537ec0 81d47ccc 8c45ff4c 81716e65 8e537640 8c45ff70
[  102.400078]  00000000 8ee9e400 0000d248 8c45ff44 810c59db 00000000 81d47ccc 0000d248
[  102.400100] Call Trace:
[  102.400100]  [<810c9ad7>] ? sched_clock_cpu+0x247/0x290
[  102.400100]  [<810ef4bb>] ? ktime_get+0xbb/0x1a0
[  102.400100]  [<81716e65>] ? cpuidle_play_dead+0x25/0x170
[  102.400100]  [<810c59db>] ? idle_task_exit+0x10b/0x220
[  102.400100]  [<81045bb4>] ? native_play_dead+0x2c4/0x340
[  102.400100]  [<81013faf>] ? cpu_idle+0x18f/0x220
[  102.400100]  [<818533f3>] ? start_secondary+0x4ca/0x4de

and it's bisected down to

[b637a328bd4f43a0e146d1eef0142b650ba0d644] rcu: Print remote CPU's stacks in stall warnings

Thanks,
Fengguang