linux-kernel - Re: [RESEND PATCH V3 3/8] perf/x86/intel: Support hardware TopDown metrics

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <9ec0a847-2b6b-3422-72bd-ac738f1c0d10@linux.intel.com>
Date:   Mon, 9 Sep 2019 09:40:26 -0400
From:   "Liang, Kan" <kan.liang@...ux.intel.com>
To:     Peter Zijlstra <peterz@...radead.org>,
        Andi Kleen <ak@...ux.intel.com>
Cc:     acme@...nel.org, mingo@...hat.com, linux-kernel@...r.kernel.org,
        tglx@...utronix.de, jolsa@...nel.org, eranian@...gle.com,
        alexander.shishkin@...ux.intel.com
Subject: Re: [RESEND PATCH V3 3/8] perf/x86/intel: Support hardware TopDown
 metrics

On 8/31/2019 5:19 AM, Peter Zijlstra wrote:
>>> Then there is no mucking about with that odd counter/metrics msr pair
>>> reset nonsense. Becuase that really stinks.
>> You have to write them to reset the internal counters.
> But not for ever read, only on METRIC_OVF.

The precision are lost if the counters were running much longer than the 
measured delta. We have to reset the counters after every read.
For example, the worst case is:
The previous reading was rounded up and the current reading is rounded 
down. The error of METRICS is 1/256 - 1/SLOTS, which is related to 
SLOTS. The error for each Topdown event is  (1/256 - 1/SLOTS) * SLOTS.
With the SLOTS increasing, the precision will get worse and worse.
Therefor, we have to reset the counters periodically.

Thanks,
Kan