perf callchain: Allow disabling call graphs per event

This patch introduce "call-graph=no" to disable per-event callgraph.

Here is an example.

perf record -e 'cpu/cpu-cycles,call-graph=fp/,cpu/instructions,call-graph=no/' sleep 1

perf report --stdio

# To display the perf.data header info, please use
--header/--header-only options.
#
#
# Total Lost Samples: 0
#
# Samples: 6 of event 'cpu/cpu-cycles,call-graph=fp/'
# Event count (approx.): 774218
#
# Children Self Command Shared Object Symbol
# ........ ........ ....... ................ ........................................
#
61.94% 0.00% sleep [kernel.vmlinux] [k] entry_SYSCALL_64_fastpath
|
---entry_SYSCALL_64_fastpath
|
|--97.30%-- __brk
|
--2.70%-- mmap64
_dl_check_map_versions
_dl_check_all_versions

61.94% 0.00% sleep [kernel.vmlinux] [k] perf_event_mmap
|
---perf_event_mmap
|
|--97.30%-- do_brk
| sys_brk
| entry_SYSCALL_64_fastpath
| __brk
|
--2.70%-- mmap_region
do_mmap_pgoff
vm_mmap_pgoff
sys_mmap_pgoff
sys_mmap
entry_SYSCALL_64_fastpath
mmap64
_dl_check_map_versions
_dl_check_all_versions
......

# Samples: 6 of event 'cpu/instructions,call-graph=no/'
# Event count (approx.): 359692
#
# Children Self Command Shared Object Symbol
# ........ ........ ....... ................ .................................
#
89.03% 0.00% sleep [unknown] [.] 0xffff6598ffff6598
89.03% 0.00% sleep ld-2.17.so [.] _dl_resolve_conflicts
89.03% 0.00% sleep [kernel.vmlinux] [k] page_fault

Signed-off-by: Kan Liang <[e-mail]>
Tested-by: Arnaldo Carvalho de Melo <[e-mail]>
Cc: Andi Kleen <[e-mail]>
Cc: Jiri Olsa <[e-mail]>
Cc: Namhyung Kim <[e-mail]>
Link: http://lkml.kernel.org/r/[e-mail]
Signed-off-by: Arnaldo Carvalho de Melo <[e-mail]>

tools/perf/Documentation/perf-record.txt

···
 		    enabling time stamping. 0 for disabling time stamping.
 		    The default is 1.
 	  - 'call-graph': Disable/enable callgraph. Acceptable str are "fp" for
-			 FP mode, "dwarf" for DWARF mode, "lbr" for LBR mode.
+			 FP mode, "dwarf" for DWARF mode, "lbr" for LBR mode and
+			 "no" for disable callgraph.
 	  - 'stack-size': user stack size for dwarf mode
 	  Note: If user explicitly sets options which conflict with the params,
 	  the value set by the params will be overridden.

tools/perf/builtin-annotate.c

···
 		if (nr_samples > 0) {
 			total_nr_samples += nr_samples;
 			hists__collapse_resort(hists, NULL);
+			/* Don't sort callchain */
+			perf_evsel__reset_sample_bit(pos, CALLCHAIN);
 			hists__output_resort(hists, NULL);
 
 			if (symbol_conf.event_group &&

tools/perf/builtin-diff.c

···
 		if (verbose || data__files_cnt > 2)
 			data__fprintf();
 
+		/* Don't sort callchain for perf diff */
+		perf_evsel__reset_sample_bit(evsel_base, CALLCHAIN);
+
 		hists__process(hists_base);
 	}
 }

tools/perf/tests/hists_cumulate.c

···
 
 	symbol_conf.use_callchain = false;
 	symbol_conf.cumulate_callchain = false;
+	perf_evsel__reset_sample_bit(evsel, CALLCHAIN);
 
 	setup_sorting();
 	callchain_register_param(&callchain_param);
···
 
 	symbol_conf.use_callchain = true;
 	symbol_conf.cumulate_callchain = false;
+	perf_evsel__set_sample_bit(evsel, CALLCHAIN);
 
 	setup_sorting();
 	callchain_register_param(&callchain_param);
···
 
 	symbol_conf.use_callchain = false;
 	symbol_conf.cumulate_callchain = true;
+	perf_evsel__reset_sample_bit(evsel, CALLCHAIN);
 
 	setup_sorting();
 	callchain_register_param(&callchain_param);
···
 
 	symbol_conf.use_callchain = true;
 	symbol_conf.cumulate_callchain = true;
+	perf_evsel__set_sample_bit(evsel, CALLCHAIN);
 
 	setup_sorting();
 	callchain_register_param(&callchain_param);

tools/perf/util/evsel.c

···
 
 		/* parse callgraph parameters */
 		if (callgraph_buf != NULL) {
-			param.enabled = true;
-			if (parse_callchain_record(callgraph_buf, &param)) {
-				pr_err("per-event callgraph setting for %s failed. "
-				       "Apply callgraph global setting for it\n",
-				       evsel->name);
-				return;
+			if (!strcmp(callgraph_buf, "no")) {
+				param.enabled = false;
+				param.record_mode = CALLCHAIN_NONE;
+			} else {
+				param.enabled = true;
+				if (parse_callchain_record(callgraph_buf, &param)) {
+					pr_err("per-event callgraph setting for %s failed. "
+					       "Apply callgraph global setting for it\n",
+					       evsel->name);
+					return;
+				}
 			}
 		}
 		if (dump_size > 0) {

tools/perf/util/hist.c

···
 
 static void __hists__insert_output_entry(struct rb_root *entries,
 					 struct hist_entry *he,
-					 u64 min_callchain_hits)
+					 u64 min_callchain_hits,
+					 bool use_callchain)
 {
 	struct rb_node **p = &entries->rb_node;
 	struct rb_node *parent = NULL;
 	struct hist_entry *iter;
 
-	if (symbol_conf.use_callchain)
+	if (use_callchain)
 		callchain_param.sort(&he->sorted_chain, he->callchain,
 				      min_callchain_hits, &callchain_param);
 
···
 	struct rb_node *next;
 	struct hist_entry *n;
 	u64 min_callchain_hits;
+	struct perf_evsel *evsel = hists_to_evsel(hists);
+	bool use_callchain = evsel ? (evsel->attr.sample_type & PERF_SAMPLE_CALLCHAIN) : symbol_conf.use_callchain;
 
 	min_callchain_hits = hists->stats.total_period * (callchain_param.min_percent / 100);
 
···
 		n = rb_entry(next, struct hist_entry, rb_node_in);
 		next = rb_next(&n->rb_node_in);
 
-		__hists__insert_output_entry(&hists->entries, n, min_callchain_hits);
+		__hists__insert_output_entry(&hists->entries, n, min_callchain_hits, use_callchain);
 		hists__inc_stats(hists, n);
 
 		if (!n->filtered)