Received-SPF: pass (google.com: domain of linux-kernel+bounces-94645-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Date: Wed, 6 Mar 2024 19:21:57 -0300
From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: Namhyung Kim <namhyung@kernel.org>
Cc: Ian Rogers <irogers@google.com>, Jiri Olsa <jolsa@kernel.org>,
	Adrian Hunter <adrian.hunter@intel.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@kernel.org>, LKML <linux-kernel@vger.kernel.org>,
	linux-perf-users@vger.kernel.org, Andi Kleen <ak@linux.intel.com>
Subject: Re: [PATCH v2 0/4] perf annotate: Improve memory usage for symbol
 histogram
Message-ID: <ZejshUg8XiQz5YGa@x1>
References: <20240304230815.1440583-1-namhyung@kernel.org>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20240304230815.1440583-1-namhyung@kernel.org>

On Mon, Mar 04, 2024 at 03:08:11PM -0800, Namhyung Kim wrote:
> Hello,
> 
> This is another series of memory optimization in perf annotate.
> 
> v2 changes:
>  * fix a bug when offset is bigger than 16 bits
> 
> 
> When perf annotate (or perf report/top with TUI) processes samples, it
> needs to save the sample period (overhead) at instruction level.  For
> now, it allocates an array to do that for the whole symbol when it
> hits any new symbol.  This comes with a lot of waste since samples can
> be very few and instructions span to multiple bytes.
> 
> For example, when a sample hits symbol 'foo' that has size of 100 and
> that's the only sample falls into the symbol.  Then it needs to
> allocate a symbol histogram (sym_hist) and the its size would be
> 
>   16 (header) + 16 (sym_hist_entry) * 100 (symbol_size) = 1616
> 
> But actually it just needs 32 (header + sym_hist_entry) bytes.  Things
> get worse if the symbol size is bigger (and it doesn't have many
> samples in different places).  Also note that it needs a separate
> histogram for each event.
> 
> Let's split the sym_hist_entry and have it in a hash table so that it
> can allocate only necessary entries.
> 
> No functional change intended.
> 
> Thanks,
> Namhyung

No difference before/after on that 'perf annotate --stdio2' for all
binaries in a perf record of building perf using the default binutils
objdump disassembler, etc.

Reviewed-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>

- Arnaldo
 
> 
> Namhyung Kim (4):
>   perf annotate: Add a hashmap for symbol histogram
>   perf annotate: Calculate instruction overhead using hashmap
>   perf annotate: Remove sym_hist.addr[] array
>   perf annotate: Add comments in the data structures
> 
>  tools/perf/ui/gtk/annotate.c |  14 ++++-
>  tools/perf/util/annotate.c   | 116 ++++++++++++++++++++++-------------
>  tools/perf/util/annotate.h   |  86 +++++++++++++++++++++++---
>  3 files changed, 159 insertions(+), 57 deletions(-)
> 
> -- 
> 2.44.0.rc1.240.g4c46232300-goog