Re: [PATCH] perf report: Add option to collapse undesired parts of call graph

[Arnaldo Carvalho de Melo <acme@ghostprotocols.net>]

Em Fri, Dec 07, 2012 at 02:30:44AM -0500, Greg Price escreveu:
> If an application has an expensive function implemented with a large
> tree of calls to helper functions, the default call-graph presentation
> will be dominated by the many different call-chains within that
> function.  By treating the function as a black box, we can collect the
> call-chains leading into the function and compactly identify what to
> blame for expensive calls.

Looks like an interesting feature, will review this soon,

- Arnaldo
 
> For example, in this report the callers of garbage_collect() are
> scattered across the tree:
> $ perf report -d ruby 2>- | grep -m10 ^[^#]*[a-z]
>     22.03%     ruby  [.] gc_mark
>                --- gc_mark
>                   |--59.40%-- mark_keyvalue
>                   |          st_foreach
>                   |          gc_mark_children
>                   |          |--99.75%-- rb_gc_mark
>                   |          |          rb_vm_mark
>                   |          |          gc_mark_children
>                   |          |          gc_marks
>                   |          |          |--99.00%-- garbage_collect
> 
> If we make garbage_collect() a black box, its callers are coalesced:
> $ perf report --blackbox garbage_collect -d ruby 2>- | grep -m10 ^[^#]*[a-z]
>     72.92%     ruby  [.] garbage_collect
>                --- garbage_collect
>                    vm_xmalloc
>                   |--47.08%-- ruby_xmalloc
>                   |          st_insert2
>                   |          rb_hash_aset
>                   |          |--98.45%-- features_index_add
>                   |          |          rb_provide_feature
>                   |          |          rb_require_safe
>                   |          |          vm_call_method
> 
> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
> Cc: Paul Mackerras <paulus@samba.org>
> Cc: Ingo Molnar <mingo@redhat.com>
> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
> Cc: Jiri Olsa <jolsa@redhat.com>
> Cc: David Ahern <dsahern@gmail.com>
> Signed-off-by: Greg Price <price@mit.edu>
> ---
>  tools/perf/builtin-report.c | 17 +++++++++++++++--
>  tools/perf/builtin-top.c    |  3 +--
>  tools/perf/util/map.h       |  4 +++-
>  tools/perf/util/session.c   | 29 ++++++++++++++++++-----------
>  tools/perf/util/session.h   |  5 +++++
>  5 files changed, 42 insertions(+), 16 deletions(-)
> 
> diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
> index a61725d..3bbda35 100644
> --- a/tools/perf/builtin-report.c
> +++ b/tools/perf/builtin-report.c
> @@ -70,7 +70,7 @@ static int perf_report__add_branch_hist_entry(struct perf_tool *tool,
>  	if ((sort__has_parent || symbol_conf.use_callchain)
>  	    && sample->callchain) {
>  		err = machine__resolve_callchain(machine, evsel, al->thread,
> -						 sample, &parent);
> +						 sample, &parent, al);
>  		if (err)
>  			return err;
>  	}
> @@ -141,7 +141,7 @@ static int perf_evsel__add_hist_entry(struct perf_evsel *evsel,
>  
>  	if ((sort__has_parent || symbol_conf.use_callchain) && sample->callchain) {
>  		err = machine__resolve_callchain(machine, evsel, al->thread,
> -						 sample, &parent);
> +						 sample, &parent, al);
>  		if (err)
>  			return err;
>  	}
> @@ -607,6 +607,8 @@ int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
>  		     "Default: fractal,0.5,callee", &parse_callchain_opt, callchain_default_opt),
>  	OPT_BOOLEAN('G', "inverted", &report.inverted_callchain,
>  		    "alias for inverted call graph"),
> +	OPT_STRING(0, "blackbox", &blackbox_pattern, "regex",
> +		   "functions to treat as black boxes in call graphs, collapsing callees"),
>  	OPT_STRING('d', "dsos", &symbol_conf.dso_list_str, "dso[,dso...]",
>  		   "only consider symbols in these dsos"),
>  	OPT_STRING('c', "comms", &symbol_conf.comm_list_str, "comm[,comm...]",
> @@ -687,6 +689,17 @@ int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
>  
>  	}
>  
> +	if (blackbox_pattern) {
> +		int err = regcomp(&blackbox_regex, blackbox_pattern, REG_EXTENDED);
> +		if (err) {
> +			char buf[BUFSIZ];
> +			regerror(err, &blackbox_regex, buf, sizeof(buf));
> +			pr_err("Invalid blackbox regex: %s\n%s", blackbox_pattern, buf);
> +			goto error;
> +		}
> +		have_blackbox = 1;
> +	}
> +
>  	if (strcmp(report.input_name, "-") != 0)
>  		setup_browser(true);
>  	else {
> diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
> index ff6db80..ee969b5 100644
> --- a/tools/perf/builtin-top.c
> +++ b/tools/perf/builtin-top.c
> @@ -786,8 +786,7 @@ static void perf_event__process_sample(struct perf_tool *tool,
>  		    sample->callchain) {
>  			err = machine__resolve_callchain(machine, evsel,
>  							 al.thread, sample,
> -							 &parent);
> -
> +							 &parent, NULL);
>  			if (err)
>  				return;
>  		}
> diff --git a/tools/perf/util/map.h b/tools/perf/util/map.h
> index d2250fc..6d1b8e1 100644
> --- a/tools/perf/util/map.h
> +++ b/tools/perf/util/map.h
> @@ -23,6 +23,7 @@ struct ref_reloc_sym;
>  struct map_groups;
>  struct machine;
>  struct perf_evsel;
> +struct addr_location;
>  
>  struct map {
>  	union {
> @@ -163,7 +164,8 @@ int machine__resolve_callchain(struct machine *machine,
>  			       struct perf_evsel *evsel,
>  			       struct thread *thread,
>  			       struct perf_sample *sample,
> -			       struct symbol **parent);
> +			       struct symbol **parent,
> +			       struct addr_location *root_al);
>  int maps__set_kallsyms_ref_reloc_sym(struct map **maps, const char *symbol_name,
>  				     u64 addr);
>  
> diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
> index 8cdd232..9a8798c 100644
> --- a/tools/perf/util/session.c
> +++ b/tools/perf/util/session.c
> @@ -19,6 +19,10 @@
>  #include "unwind.h"
>  #include "vdso.h"
>  
> +regex_t blackbox_regex;
> +const char *blackbox_pattern;
> +int have_blackbox = 0;
> +
>  static int perf_session__open(struct perf_session *self, bool force)
>  {
>  	struct stat input_stat;
> @@ -226,11 +230,10 @@ void machine__remove_thread(struct machine *self, struct thread *th)
>  	list_add_tail(&th->node, &self->dead_threads);
>  }
>  
> -static bool symbol__match_parent_regex(struct symbol *sym)
> +static bool symbol__match_regex(struct symbol *sym, regex_t *regex)
>  {
> -	if (sym->name && !regexec(&parent_regex, sym->name, 0, NULL, 0))
> +	if (sym->name && !regexec(regex, sym->name, 0, NULL, 0))
>  		return 1;
> -
>  	return 0;
>  }
>  
> @@ -295,8 +298,8 @@ struct branch_info *machine__resolve_bstack(struct machine *self,
>  static int machine__resolve_callchain_sample(struct machine *machine,
>  					     struct thread *thread,
>  					     struct ip_callchain *chain,
> -					     struct symbol **parent)
> -
> +					     struct symbol **parent,
> +					     struct addr_location *root_al)
>  {
>  	u8 cpumode = PERF_RECORD_MISC_USER;
>  	unsigned int i;
> @@ -347,8 +350,13 @@ static int machine__resolve_callchain_sample(struct machine *machine,
>  					   MAP__FUNCTION, ip, &al, NULL);
>  		if (al.sym != NULL) {
>  			if (sort__has_parent && !*parent &&
> -			    symbol__match_parent_regex(al.sym))
> +			    symbol__match_regex(al.sym, &parent_regex))
>  				*parent = al.sym;
> +			else if (have_blackbox && root_al &&
> +			         symbol__match_regex(al.sym, &blackbox_regex)) {
> +				*root_al = al;
> +				callchain_cursor_reset(&callchain_cursor);
> +			}
>  			if (!symbol_conf.use_callchain)
>  				break;
>  		}
> @@ -373,15 +381,15 @@ int machine__resolve_callchain(struct machine *machine,
>  			       struct perf_evsel *evsel,
>  			       struct thread *thread,
>  			       struct perf_sample *sample,
> -			       struct symbol **parent)
> -
> +			       struct symbol **parent,
> +			       struct addr_location *root_al)
>  {
>  	int ret;
>  
>  	callchain_cursor_reset(&callchain_cursor);
>  
>  	ret = machine__resolve_callchain_sample(machine, thread,
> -						sample->callchain, parent);
> +	                                        sample->callchain, parent, root_al);
>  	if (ret)
>  		return ret;
>  
> @@ -1603,9 +1611,8 @@ void perf_evsel__print_ip(struct perf_evsel *evsel, union perf_event *event,
>  
>  	if (symbol_conf.use_callchain && sample->callchain) {
>  
> -
>  		if (machine__resolve_callchain(machine, evsel, al.thread,
> -					       sample, NULL) != 0) {
> +					       sample, NULL, NULL) != 0) {
>  			if (verbose)
>  				error("Failed to resolve callchain. Skipping\n");
>  			return;
> diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h
> index 0eae00a..6db3e55 100644
> --- a/tools/perf/util/session.h
> +++ b/tools/perf/util/session.h
> @@ -1,6 +1,7 @@
>  #ifndef __PERF_SESSION_H
>  #define __PERF_SESSION_H
>  
> +#include <regex.h>
>  #include "hist.h"
>  #include "event.h"
>  #include "header.h"
> @@ -9,6 +10,10 @@
>  #include <linux/rbtree.h>
>  #include <linux/perf_event.h>
>  
> +extern regex_t blackbox_regex;
> +extern const char *blackbox_pattern;
> +extern int have_blackbox;
> +
>  struct sample_queue;
>  struct ip_callchain;
>  struct thread;
> -- 
> 1.7.11.3