Messages in this thread | | | Date | Fri, 20 May 2011 13:53:07 +0200 | From | Ingo Molnar <> | Subject | Re: [RFC] ACPI, APEI, Generic Hardware Error Source (GHES) injecting support |
| |
* Borislav Petkov <bp@alien8.de> wrote:
> On Tue, May 17, 2011 at 09:18:03PM +0200, Ingo Molnar wrote: > > So for example [sufficienty privileged] user-space could inject *any* > > perf event - for example a PERF_COUNT_HW_CACHE_MISSES event (for test > > purposes) and any tooling that runs could not tell apart this injected > > event from a real event. > > Yeah about that, I was recently speculating how that would work. So do we do > > $ perf record ... > > in the one xterm, and, in the other, > > $ perf inject > > so that while recording, we can inject some events from userspace? Or > do we inject it, it gets buffered somewhere in the meantime and then > the next perf record session sees it along with the remaining injection > events?
Well, for persistent events there would be interim buffering even if there's no observation going on anywhere. I.e. there's always an 'observer' of events.
For non-persistent events, if they are injected, then they are like trace events for which nobody is interested in: they are lost.
Thanks,
Ingo
| |