Re: CVS commit: src

[Maxime Villard <max%m00nbsd.net@localhost>]

Le 10/03/2017 à 19:01, Valery Ushakov a écrit :
> On Fri, Mar 10, 2017 at 13:09:11 +0000, Maxime Villard wrote:
>
> > Module Name:	src
> > Committed By:	maxv
> > Date:		Fri Mar 10 13:09:11 UTC 2017
> >
> > Modified Files:
> > 	src/sys/arch/i386/i386: pmc.c
> > 	src/sys/arch/x86/include: sysarch.h
> > 	src/usr.bin/pmc: pmc.1 pmc.c
> >
> > Log Message:
> > Switch to per-CPU PMC results, and completely rewrite the pmc(1) tool. Now
> > the PMCs are system-wide, fine-grained and more tunable by the user.
> >
> > We don't do application tracking, since it would require to store the PMC
> > values in mdproc and starting/stopping the counters on each context switch.
> > While this doesn't seem to be particularly difficult to achieve, I don't
> > think it is really interesting; and if someone really wants to measure
> > the performance of an application, they can simply schedctl it to a cpu
> > and look at the PMC results for this cpu.
>
> How does that affects the MI pmc(9) API?
>
> -uwe

It does not; x86 uses a MD sysarch which is intentionally opaque and
undocumented. PMCs are in many ways highly CPU-specific, and MD sysarches are
easier to maintain. I would also add that no one is interested in having a
common API for such features, as long as there is at least one tool that can
correctly use MD entries and display results.