Re: [RFC PATCH 3/3] perf report: add --spe options for arm-spe

From: James Clark <James.Clark@arm.com>
To: Tan Xiaojun <tanxiaojun@huawei.com>,
	"peterz@infradead.org" <peterz@infradead.org>,
	"mingo@redhat.com" <mingo@redhat.com>,
	"acme@kernel.org" <acme@kernel.org>,
	"alexander.shishkin@linux.intel.com" 
	<alexander.shishkin@linux.intel.com>,
	"jolsa@redhat.com" <jolsa@redhat.com>,
	"namhyung@kernel.org" <namhyung@kernel.org>,
	"ak@linux.intel.com" <ak@linux.intel.com>,
	"adrian.hunter@intel.com" <adrian.hunter@intel.com>,
	"yao.jin@linux.intel.com" <yao.jin@linux.intel.com>,
	"tmricht@linux.ibm.com" <tmricht@linux.ibm.com>,
	"brueckner@linux.ibm.com" <brueckner@linux.ibm.com>,
	"songliubraving@fb.com" <songliubraving@fb.com>,
	"gregkh@linuxfoundation.org" <gregkh@linuxfoundation.org>,
	Kim Phillips <Kim.Phillips@arm.com>
Cc: "gengdongjiu@huawei.com" <gengdongjiu@huawei.com>,
	"wxf.wang@hisilicon.com" <wxf.wang@hisilicon.com>,
	"liwei391@huawei.com" <liwei391@huawei.com>,
	"huawei.libin@huawei.com" <huawei.libin@huawei.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	Jeremy Linton <Jeremy.Linton@arm.com>,
	"linux-perf-users@vger.kernel.org"
	<linux-perf-users@vger.kernel.org>, nd <nd@arm.com>
Subject: Re: [RFC PATCH 3/3] perf report: add --spe options for arm-spe
Date: Wed, 21 Aug 2019 12:38:10 +0000	[thread overview]
Message-ID: <8e8deead-6d59-9203-a01a-fe63362ebdf0@arm.com> (raw)
In-Reply-To: <1564738813-10944-4-git-send-email-tanxiaojun@huawei.com>

Hi,

I also had a look at this and had a question about the --spe option.
It seems that whatever options I give it, the output is the same:

	perf report 
And
	perf report --spe=t

Both give the same result:

	# Samples: 4  of event 'llc-miss'
	# Event count (approx.): 4
	#
	# Children      Self  Command  Shared Object      Symbol                    
	# ........  ........  .......  .................  ..........................
	#
	...
	# Samples: 0  of event 'tlb-miss'
	# Event count (approx.): 0
	#
	# Children      Self  Command  Shared Object  Symbol
	# ........  ........  .......  .............  ......
	#

	# Samples: 83  of event 'branch-miss'
	# Event count (approx.): 83
	#
	# Children      Self  Command  Shared Object      Symbol                   
	# ........  ........  .......  .................  .........................
	#
	...

I would have expected it to not include the branch and LLC sections for the second
command with --spe=t.

And that leads me to another point. Does it make sense to have this option as a post
processing step? SPE already has support for filtering events at collection time with
the PMSFCR_EL1 register.

Should we try to make the interface more like PEBS, where you specify which events you
are interested in doing precise tracing on like this?

	perf record -e branch-misses:pp

And then perf could use the modifier to configure SPE so that it only records branch
misses? The benefits of this would be keeping the user interface for precise tracing
similar between platforms.

Thanks
James

On 02/08/2019 10:40, Tan Xiaojun wrote:
> The previous patch added support in "perf report" for some arm-spe
> events(llc-miss, tlb-miss, branch-miss). This patch adds their help
> instructions.
> 
> Signed-off-by: Tan Xiaojun <tanxiaojun@huawei.com>
> ---
>  tools/perf/Documentation/perf-report.txt | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/tools/perf/Documentation/perf-report.txt b/tools/perf/Documentation/perf-report.txt
> index 987261d..d998d4b 100644
> --- a/tools/perf/Documentation/perf-report.txt
> +++ b/tools/perf/Documentation/perf-report.txt
> @@ -445,6 +445,15 @@ include::itrace.txt[]
>  
>  	To disable decoding entirely, use --no-itrace.
>  
> +--spe::
> +	Options for decoding arm-spe tracing data. The options are:
> +
> +		l	synthesize llc miss events
> +		t	synthesize tlb miss events
> +		b	synthesize branch miss events
> +
> +	The default is all events i.e. the same as --spe=ltb
> +
>  --full-source-path::
>  	Show the full path for source files for srcline output.
>  
> 