Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
MIME-Version: 1.0
References: <20220527040407.4193232-1-irogers@google.com> <YpCeo6APNtXvrTPJ@hirez.programming.kicks-ass.net>
 <76dce5d3-6c69-e403-c4d9-f882967a4467@linux.intel.com> <YpIrVSGH314djkh9@worktop.programming.kicks-ass.net>
In-Reply-To: <YpIrVSGH314djkh9@worktop.programming.kicks-ass.net>
From:   Ian Rogers <irogers@google.com>
Date:   Sat, 28 May 2022 07:50:40 -0700
Message-ID: <CAP-5=fWwG+Gt+E7M8EYuDhgrvfUQLbP7ODn18r_axshqKxwuYA@mail.gmail.com>
Subject: Re: [PATCH v2] perf metrics: Add literal for system TSC frequency
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     Andi Kleen <ak@linux.intel.com>, perry.taylor@intel.com,
        caleb.biggers@intel.com, kshipra.bopardikar@intel.com,
        Kan Liang <kan.liang@linux.intel.com>,
        Zhengjun Xing <zhengjun.xing@linux.intel.com>,
        Ingo Molnar <mingo@redhat.com>,
        Arnaldo Carvalho de Melo <acme@kernel.org>,
        Mark Rutland <mark.rutland@arm.com>,
        Alexander Shishkin <alexander.shishkin@linux.intel.com>,
        Jiri Olsa <jolsa@redhat.com>,
        Namhyung Kim <namhyung@kernel.org>,
        Maxime Coquelin <mcoquelin.stm32@gmail.com>,
        Alexandre Torgue <alexandre.torgue@foss.st.com>,
        James Clark <james.clark@arm.com>,
        John Garry <john.garry@huawei.com>,
        linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org,
        Stephane Eranian <eranian@google.com>,
        Thomas Gleixner <tglx@linutronix.de>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Sat, May 28, 2022 at 7:02 AM Peter Zijlstra <peterz@infradead.org> wrote:
>
> On Fri, May 27, 2022 at 07:54:19AM -0700, Andi Kleen wrote:
>
> > > This all seems bonghits inspired... and perf actually does expose the
> > > tsc frequency. What do you think is in perf_event_mmap_page::time_* ?
> >
> >
> > That's not really available to perf stat, which is the primary metrics user.
>
> Why not? You can mmap any perf-fd (even software events) and these
> fields should be filled out.
>
> It should work on any x86 CPU that has a TSC. The only caveat is that
> the kernel must not have marked the TSC unstable.
>
> It could even work for virt -- all you need is for virt to use
> native_sched_clock() instead of the paravirt nonsense.

It will at least fail if inherit is enabled, no?

Thanks,
Ian