Messages in this thread | | | Date | Fri, 12 Jul 2013 12:07:28 -0600 | From | David Ahern <> | Subject | Re: Yet more softlockups. |
| |
On 7/12/13 11:50 AM, Dave Jones wrote:
> Given you can run trinity long enough that you hit this however, makes me > think you won't be able to triger the bug I'm talking about. > Perhaps virtualised perf counters are somehow immune to this problem, because > on bare-metal, it literally takes seconds.
I can generate the message it just does not lock up:
... [ 131.837129] perf samples too long (35849 > 35714), lowering kernel.perf_event_max_sample_rate to 4000 ... [ 237.271170] INFO: NMI handler (perf_event_nmi_handler) took too long to run: 57900.001 msecs
And Dave Hansen: I think nmi.c has the same do_div problem as kernel/events/core.c that Stephane fixed. Your patch has:
whole_msecs = do_div(delta, (1000 * 1000)); decimal_msecs = do_div(delta, 1000) % 1000;
David
| |