[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20190220001116.3b3yyndjjmqs3hmm@studium.uni-erlangen.de>
Date: Wed, 20 Feb 2019 01:11:16 +0100
From: Jonas Rabenstein <jonas.rabenstein@...dium.uni-erlangen.de>
To: linux-perf-users@...r.kernel.org
Cc: Peter Zijlstra <peterz@...radead.org>,
Ingo Molnar <mingo@...hat.com>,
Milian Wolff <milian.wolff@...b.com>,
Arnaldo Carvalho de Melo <acme@...nel.org>,
Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
Jiri Olsa <jolsa@...hat.com>,
Namhyung Kim <namhyung@...nel.org>,
Andi Kleen <ak@...ux.intel.com>,
Thomas Richter <tmricht@...ux.ibm.com>,
Stephane Eranian <eranian@...gle.com>,
Jonas Rabenstein <jonas.rabenstein@...dium.uni-erlangen.de>,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/2] perf evsel: add inline functions to sample callchain
output
On Tue, Feb 19, 2019 at 07:38:10PM +0100, Jonas Rabenstein wrote:
> Whenever a callchain shall be printed search for each address whether
> inline information is available and add those symbols to the output
> if symbol_conf.inline_name is enabled.
>
> Signed-off-by: Jonas Rabenstein <jonas.rabenstein@...dium.uni-erlangen.de>
> ---
> tools/perf/util/evsel_fprintf.c | 46 +++++++++++++++++++++++++++++++++
> 1 file changed, 46 insertions(+)
>
> diff --git a/tools/perf/util/evsel_fprintf.c b/tools/perf/util/evsel_fprintf.c
> index c710f687ddf4..9f9ece453dc4 100644
> --- a/tools/perf/util/evsel_fprintf.c
> +++ b/tools/perf/util/evsel_fprintf.c
> @@ -168,6 +168,49 @@ static int __fprintf_callchain_link(u64 ip, struct map *map, struct symbol *symb
> return printed;
> }
>
> +static int __fprintf_callchain_inlines(struct callchain_cursor_node *node,
> + bool first, int left_alignment,
> + unsigned int print_opts, FILE *fp)
> +{
> + int printed = 0;
> + struct inline_node *inline_node;
> + struct inline_list *ilist;
> + u64 addr;
> +
> + if (!symbol_conf.inline_name)
> + return 0;
> +
> + if (!node->map || !node->map->dso)
> + return 0;
> +
> + addr = node->map->map_ip(node->map, node->ip);
I stumbled across a problem with the usage of map_ip here. Using the map_ip
on dynamic binaries does resolve the inline symbols but it fails with static
ones. Using node->ip directly works for static binaries but fails for
static ones...
Is there a way to decide whether node->map->dso is a static or dynamic
object file? Furthermore, is it even correct for a static dso to have
map->map_ip set to map__map_ip instead of identity__map_ip?
> +
> + inline_node = inlines__tree_find(&node->map->dso->inlined_nodes,
> + addr);
> + if (!inline_node) {
> + inline_node = dso__parse_addr_inlines(node->map->dso,
> + addr, node->sym);
> + if (!inline_node)
> + return 0;
> + inlines__tree_insert(&node->map->dso->inlined_nodes,
> + inline_node);
> + }
> +
> + list_for_each_entry(ilist, &inline_node->val, list) {
> + if (ilist->symbol == node->sym)
> + break;
> +
> + printed += __fprintf_callchain_link(node->ip, node->map,
> + ilist->symbol,
> + ilist->srcline,
> + first, left_alignment,
> + print_opts, fp);
> + first = (first && printed == 0);
> + }
> +
> + return printed;
> +}
> +
> int sample__fprintf_callchain(struct perf_sample *sample, int left_alignment,
> unsigned int print_opts, struct callchain_cursor *cursor,
> FILE *fp)
> @@ -183,6 +226,9 @@ int sample__fprintf_callchain(struct perf_sample *sample, int left_alignment,
> if (!node)
> break;
>
> + printed += __fprintf_callchain_inlines(node, (printed == 0),
> + left_alignment,
> + print_opts, fp);
>
> printed += __fprintf_callchain_link(node->ip, node->map,
> node->sym, NULL,
> --
> 2.19.2
>
Powered by blists - more mailing lists