Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 14 Aug 2015 18:51:04 -0700
From:      Stanislav Sedov <stas@freebsd.org>
To:        Sai Prajeeth <csprajeeth@gmail.com>
Cc:        freebsd-performance@freebsd.org
Subject:   Re: pmcstat profiles kernel code?
Message-ID:  <A6EA4336-907B-423B-AD6A-D99A8288DD21@freebsd.org>
In-Reply-To: <CAFJamtP7bA6NpfPb0VOSVAoeAjbGcA4OHgxG66Dwz0H3OetSfQ@mail.gmail.com>
References:  <CAFJamtP7bA6NpfPb0VOSVAoeAjbGcA4OHgxG66Dwz0H3OetSfQ@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help

> On Jun 12, 2015, at 3:58 AM, Sai Prajeeth <csprajeeth@gmail.com> wrote:
> 
> Hi,
> 
> I am not sure if this is the list that addresses this question but here it
> goes. I would like to count hardware events such as instructions retired /
> unhalted-core-cycles for all the systems for BOTH OS and APPLICATION. I am
> aware that pmcstat has a system wide monitoring mode that counts events
> across all the CPUs for all processes. I want to know whether the
> statistics for kernel code are collected as well ? (which I dont think so).
> 

Yes, system wide pmcstat collects counters for the entire system, both kernel
and userland processes.  Generally, you should see kernel functions show up
in pmcstat running in a system wide top mode.

--
Stanislav Sedov
ST4096-RIPE





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?A6EA4336-907B-423B-AD6A-D99A8288DD21>