Messages in this thread | | | Date | Mon, 20 Apr 2015 17:41:48 +0200 | From | Rabin Vincent <> | Subject | Re: qemu:arm test failure due to commit 8053871d0f7f (smp: Fix smp_call_function_single_async() locking) |
| |
On Sun, Apr 19, 2015 at 08:01:40PM +0200, Ingo Molnar wrote: > That's all fine and good, but why is an IPI sent to a non-existent > CPU? It's not like we don't know which CPU is up and down.
The perf events code is trying to call smp_call_function_single() on the non-existent CPU in perf_event_exit_cpu_context() while handling the CPU_UP_CANCELED notification. perf_cpu_notify() handles CPU_UP_CANCELED and CPU_DOWN_PREPARE in the same way.
(cpu_up() is tried for the non-existing CPUs because in this case what is specified in the device tree does not match reality.)
| |