From: "Liang, Kan" <kan.liang@linux.intel.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: acme@kernel.org, mingo@redhat.com, linux-kernel@vger.kernel.org,
tglx@linutronix.de, jolsa@kernel.org, eranian@google.com,
alexander.shishkin@linux.intel.com, ak@linux.intel.com
Subject: Re: [PATCH 4/9] perf/x86/intel: Support hardware TopDown metrics
Date: Tue, 28 May 2019 14:23:09 -0400 [thread overview]
Message-ID: <287c2c84-25cf-fdce-a3c3-49a6ee93ae4c@linux.intel.com> (raw)
In-Reply-To: <20190528124349.GU2606@hirez.programming.kicks-ass.net>
On 5/28/2019 8:43 AM, Peter Zijlstra wrote:
> On Tue, May 21, 2019 at 02:40:50PM -0700, kan.liang@linux.intel.com wrote:
>> The 8bit metrics ratio values lose precision when the measurement period
>> gets longer.
>>
>> To avoid this we always reset the metric value when reading, as we
>> already accumulate the count in the perf count value.
>>
>> For a long period read, low precision is acceptable.
>> For a short period read, the register will be reset often enough that it
>> is not a problem.
>
>> The PERF_METRICS may report wrong value if its delta was less than 1/255
>> of SLOTS (Fixed counter 3).
>>
>> To avoid this, the PERF_METRICS and SLOTS registers have to be reset
>> simultaneously. The slots value has to be cached as well.
>
> That doesn't sound like it is NMI-safe.
> >
>
>> RDPMC
>> =========
>> The TopDown can be collected per thread/process. To use TopDown
>> through RDPMC in applications on Icelake, the metrics and slots values
>> have to be saved/restored during context switching.
>>
>> Add specific set_period() to specially handle the slots and metrics
>> event. Because,
>> - The initial value must be 0.
>> - Only need to restore the value in context switch. For other cases,
>> the counters have been cleared after read.
>
> So the above claims to explain RDPMC, but doesn't mention that magic
> value below at all. In fact, I don't see how the above relates to RDPMC
> at all.
Current perf only support per-core Topdown RDPMC. On Icelake, it can be
extended to per-thread Topdown RDPMC.
It tries to explain the extra work for per-thread topdown RDPMC, e.g.
save/restore slots and metrics value in context switch.
>
>> @@ -2141,7 +2157,9 @@ static int x86_pmu_event_idx(struct perf_event *event)
>> if (!(event->hw.flags & PERF_X86_EVENT_RDPMC_ALLOWED))
>> return 0;
>>
>> - if (x86_pmu.num_counters_fixed && idx >= INTEL_PMC_IDX_FIXED) {
>> + if (is_metric_idx(idx))
>> + idx = 1 << 29;
>
> I can't find this in the SDM RDPMC description. What does it return?
It will return the value of PERF_METRICS. I will add it in the changelog.
Thanks,
Kan
>
>> + else if (x86_pmu.num_counters_fixed && idx >= INTEL_PMC_IDX_FIXED) {
>> idx -= INTEL_PMC_IDX_FIXED;
>> idx |= 1 << 30;
>> }
next prev parent reply other threads:[~2019-05-28 18:23 UTC|newest]
Thread overview: 44+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-05-21 21:40 [PATCH 0/9] TopDown metrics support for Icelake kan.liang
2019-05-21 21:40 ` [PATCH 1/9] perf/core: Support a REMOVE transaction kan.liang
2019-05-21 21:40 ` [PATCH 2/9] perf/x86/intel: Basic support for metrics counters kan.liang
2019-05-28 12:05 ` Peter Zijlstra
2019-05-28 18:20 ` Liang, Kan
2019-05-28 12:15 ` Peter Zijlstra
2019-05-28 18:21 ` Liang, Kan
2019-05-29 7:28 ` Peter Zijlstra
2019-05-29 14:40 ` Liang, Kan
2019-05-29 16:46 ` Peter Zijlstra
2019-05-29 8:14 ` Peter Zijlstra
2019-05-21 21:40 ` [PATCH 3/9] perf/x86/intel: Support overflows on SLOTS kan.liang
2019-05-28 12:20 ` Peter Zijlstra
2019-05-28 18:22 ` Liang, Kan
2019-05-21 21:40 ` [PATCH 4/9] perf/x86/intel: Support hardware TopDown metrics kan.liang
2019-05-28 12:43 ` Peter Zijlstra
2019-05-28 18:23 ` Liang, Kan [this message]
2019-05-29 7:30 ` Peter Zijlstra
2019-05-28 12:53 ` Peter Zijlstra
2019-05-28 12:56 ` Peter Zijlstra
2019-05-28 13:32 ` Peter Zijlstra
2019-05-28 13:30 ` Peter Zijlstra
2019-05-28 18:24 ` Liang, Kan
2019-05-29 7:34 ` Peter Zijlstra
2019-05-29 14:41 ` Liang, Kan
2019-05-28 13:43 ` Peter Zijlstra
2019-05-28 18:24 ` Liang, Kan
2019-05-29 7:54 ` Peter Zijlstra
2019-05-29 14:42 ` Liang, Kan
2019-05-29 16:58 ` Peter Zijlstra
2019-06-04 20:39 ` Liang, Kan
2019-05-28 13:48 ` Peter Zijlstra
2019-05-28 18:24 ` Liang, Kan
2019-05-29 7:57 ` Peter Zijlstra
2019-05-29 14:42 ` Liang, Kan
2019-05-21 21:40 ` [PATCH 5/9] perf/x86/intel: Set correct weight for TopDown metrics events kan.liang
2019-05-28 13:50 ` Peter Zijlstra
2019-05-21 21:40 ` [PATCH 6/9] perf/x86/intel: Export new TopDown metrics events for Icelake kan.liang
2019-05-21 21:40 ` [PATCH 7/9] perf/x86/intel: Disable sampling read slots and topdown kan.liang
2019-05-28 13:52 ` Peter Zijlstra
2019-05-28 18:25 ` Liang, Kan
2019-05-29 7:58 ` Peter Zijlstra
2019-05-21 21:40 ` [PATCH 8/9] perf, tools, stat: Support new per thread TopDown metrics kan.liang
2019-05-21 21:40 ` [PATCH 9/9] perf, tools: Add documentation for topdown metrics kan.liang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=287c2c84-25cf-fdce-a3c3-49a6ee93ae4c@linux.intel.com \
--to=kan.liang@linux.intel.com \
--cc=acme@kernel.org \
--cc=ak@linux.intel.com \
--cc=alexander.shishkin@linux.intel.com \
--cc=eranian@google.com \
--cc=jolsa@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=tglx@linutronix.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).