Re: [PATCH] perf/core: generate overflow signal when samples are dropped (WAS: Re: [REGRESSION] perf/core: PMU interrupts dropped if we entered the kernel in the "skid" region)

From: Ingo Molnar
Date: Wed Jul 12 2017 - 03:57:26 EST

* Jin, Yao <yao.jin@xxxxxxxxxxxxxxx> wrote:

> Could we provide 2 options in user space when enabling the event sampling?
> One option is for the use case like rr debugger which only cares the PMI
> interrupt but doesn't care the skid. The skid samples doesn't need to be
> dropped.

Since it's an information leak, this is not something we want to expose as an
interface. It's also a very ugly interface: why not just *clear* the sample,
instead of dropping it?

The hardware messed up and gave us something we specifically did not permit
user-space to collect. We have to fix the bad effects to the best extent we can,
and not based on some knob.