Re: perf_event sampling a multithreaded process - ioctl with PERF_EVENT_IOC_SET_OUTPUT fails

[Daniel John FitzGerald <daniel.j.fitzgerald@gmail.com>]

*Sighs*

I found the problem.  I sent the wrong index counter as the
perf_event_open CPU... they were all being opened on different CPUs.

Thanks for the help :-)

Regards,
Dan FitzGerald

An enlightenment painter would paint a grand house on a lawn;
A romantic painter would paint it on fire.

On 04/19/2017 12:37 PM, Daniel John FitzGerald wrote:
> I'm confused as to why either would be the case; I'm specifying the same PID and same CPU number.  Is there any documentation out there on all of the restrictions to PERF_EVENT_IOC_SET_OUTPUT?
>
> For reference, attached is the source for my function that calls perf_event_open(), mmaps the ring buffer, and tries to do the ioctl.
>
> Regards,
> Dan FitzGerald
>
> An enlightenment painter would paint a grand house on a lawn;
> A romantic painter would paint it on fire.
>
> On 04/19/2017 11:46 AM, Andi Kleen wrote:
>> Daniel John FitzGerald <daniel.j.fitzgerald@gmail.com> writes:
>>> The problem that I am running into is that whenever I attempt to call
>>> ioctl() with PERF_EVENT_IOC_SET_OUTPUT, I get a -1 response with
>>> errno==EINVAL.
>> There are lots of restrictions to PERF_EVENT_IOC_SET_OUTPUT.
>>
>> Likely you're violating that it has to be on the same CPU or same task.
>>
>> -Andi