Received-SPF: softfail (google.com: domain of transitioning linux-kernel-owner@vger.kernel.org does not designate 23.128.96.19 as permitted sender) client-ip=23.128.96.19;
Message-ID: <52903e58-e74c-5ea0-36b4-277ea3610af4@intel.com>
Date:   Tue, 15 Mar 2022 22:11:54 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Firefox/91.0 Thunderbird/91.5.0
Subject: Re: [PATCH] perf intel-pt: Synthesize cycle events
Content-Language: en-US
To:     "Steinar H. Gunderson" <sesse@google.com>
Cc:     Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@redhat.com>,
        Arnaldo Carvalho de Melo <acme@kernel.org>,
        Alexander Shishkin <alexander.shishkin@linux.intel.com>,
        Jiri Olsa <jolsa@redhat.com>,
        Namhyung Kim <namhyung@kernel.org>,
        linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org
References: <20220310093844.982656-1-sesse@google.com>
 <586de5fc-858b-2693-1986-5c77e8c0e3d0@intel.com>
 <YiuKAk7SaXP7B7Ee@google.com>
 <ba2c49da-22c5-06ea-e953-82211b953ca8@intel.com>
 <YjBnet2813sEGjZY@google.com>
 <c50bb87d-9fee-c4f9-a350-8729e503e43a@intel.com>
 <YjDUO6bbyfGw/u0C@google.com>
From:   Adrian Hunter <adrian.hunter@intel.com>
Organization: Intel Finland Oy, Registered Address: PL 281, 00181 Helsinki,
 Business Identity Code: 0357606 - 4, Domiciled in Helsinki
In-Reply-To: <YjDUO6bbyfGw/u0C@google.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

On 15.3.2022 20.00, Steinar H. Gunderson wrote:
> On Tue, Mar 15, 2022 at 01:32:38PM +0200, Adrian Hunter wrote:
>>> I think the structure looks good, but I'm not sure about updating
>>> e.g. ptq->last_cy_insn_cnt in both functions? Does that make sense?
>> It should only be updated in the new intel_pt_synth_cycle_sample().
>> intel_pt_synth_instruction_sample() should be unchanged.
> 
> Hm, OK. But something definitely changed between my original patch and
> your change. (The first patch; I didn't try the last one yet.) With my
> patch, I got (on a specific trace, synthing cycles only with perf report
> --itrace=y0nse):
> 
> Samples: 4M of event 'cycles:uH', Event count (approx.): 4844309
> 
> With yours on the same file:
> 
> Samples: 2M of event 'cycles:uH', Event count (approx.): 77622449
> 
> The relative times between functions are also pretty different (although
> none of them are obviously crazy), so one of them has to be wrong.
> Is this to be expected, ie., would you expect your change to fix some
> bad bug on cycle-only synth? For reference, “perf script --itrace=i0ns
> -F +ipc | grep -c IPC:” (a quick proxy for the number of CYC packets :-) )
> yields 4836782, so I'm a bit surprised why there are only 2M events
> being emitted from that.

Sorry, my first suggestion has issues, but the second is better.