Re: [BUG] NULL pointer dereference probably caused by kernel/trace/ring_buffer.c

[Steven Rostedt]

On Tue, 1 Nov 2022 23:01:38 +0100
Roland Ruckerbauer <roland.rucky@xxxxxxxxx> wrote:

> Just figured out something, the poll() errors starting at fd 29 (which
> is CPU 24).
> But I only have 24 cores. Looked at
> /sys/kernel/debug/tracing/instances/rasdaemon/per_cpu/, and voila, it
> lists 32 cpus not 24.

Ah, and this explains why I never triggered it. There's some machines
that list nr_cpu_ids as more than what exists (I guess it is "possible
CPUs" and not "actual CPUs").

OK, this is a great clue. I'll dig deeper.

Thanks a lot for telling me this.

Takashi, can you check if the problem machines have more CPUs listed in
/sys/kernel/tracing/per_cpu/cpu* than is on the machine?

-- Steve