aboutsummaryrefslogtreecommitdiffstats
path: root/tools/perf/util/hist.c
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2012-10-01 13:28:49 -0400
committerLinus Torvalds <torvalds@linux-foundation.org>2012-10-01 13:28:49 -0400
commit7e92daaefa68e5ef1e1732e45231e73adbb724e7 (patch)
tree8e7f8ac9d82654df4c65939c6682f95510e22977 /tools/perf/util/hist.c
parent7a68294278ae714ce2632a54f0f46916dca64f56 (diff)
parent1d787d37c8ff6612b8151c6dff15bfa7347bcbdf (diff)
Merge branch 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull perf update from Ingo Molnar: "Lots of changes in this cycle as well, with hundreds of commits from over 30 contributors. Most of the activity was on the tooling side. Higher level changes: - New 'perf kvm' analysis tool, from Xiao Guangrong. - New 'perf trace' system-wide tracing tool - uprobes fixes + cleanups from Oleg Nesterov. - Lots of patches to make perf build on Android out of box, from Irina Tirdea - Extend ftrace function tracing utility to be more dynamic for its users. It allows for data passing to the callback functions, as well as reading regs as if a breakpoint were to trigger at function entry. The main goal of this patch series was to allow kprobes to use ftrace as an optimized probe point when a probe is placed on an ftrace nop. With lots of help from Masami Hiramatsu, and going through lots of iterations, we finally came up with a good solution. - Add cpumask for uncore pmu, use it in 'stat', from Yan, Zheng. - Various tracing updates from Steve Rostedt - Clean up and improve 'perf sched' performance by elliminating lots of needless calls to libtraceevent. - Event group parsing support, from Jiri Olsa - UI/gtk refactorings and improvements from Namhyung Kim - Add support for non-tracepoint events in perf script python, from Feng Tang - Add --symbols to 'script', similar to the one in 'report', from Feng Tang. Infrastructure enhancements and fixes: - Convert the trace builtins to use the growing evsel/evlist tracepoint infrastructure, removing several open coded constructs like switch like series of strcmp to dispatch events, etc. Basically what had already been showcased in 'perf sched'. - Add evsel constructor for tracepoints, that uses libtraceevent just to parse the /format events file, use it in a new 'perf test' to make sure the libtraceevent format parsing regressions can be more readily caught. - Some strange errors were happening in some builds, but not on the next, reported by several people, problem was some parser related files, generated during the build, didn't had proper make deps, fix from Eric Sandeen. - Introduce struct and cache information about the environment where a perf.data file was captured, from Namhyung Kim. - Fix handling of unresolved samples when --symbols is used in 'report', from Feng Tang. - Add union member access support to 'probe', from Hyeoncheol Lee. - Fixups to die() removal, from Namhyung Kim. - Render fixes for the TUI, from Namhyung Kim. - Don't enable annotation in non symbolic view, from Namhyung Kim. - Fix pipe mode in 'report', from Namhyung Kim. - Move related stats code from stat to util/, will be used by the 'stat' kvm tool, from Xiao Guangrong. - Remove die()/exit() calls from several tools. - Resolve vdso callchains, from Jiri Olsa - Don't pass const char pointers to basename, so that we can unconditionally use libgen.h and thus avoid ifdef BIONIC lines, from David Ahern - Refactor hist formatting so that it can be reused with the GTK browser, From Namhyung Kim - Fix build for another rbtree.c change, from Adrian Hunter. - Make 'perf diff' command work with evsel hists, from Jiri Olsa. - Use the only field_sep var that is set up: symbol_conf.field_sep, fix from Jiri Olsa. - .gitignore compiled python binaries, from Namhyung Kim. - Get rid of die() in more libtraceevent places, from Namhyung Kim. - Rename libtraceevent 'private' struct member to 'priv' so that it works in C++, from Steven Rostedt - Remove lots of exit()/die() calls from tools so that the main perf exit routine can take place, from David Ahern - Fix x86 build on x86-64, from David Ahern. - {int,str,rb}list fixes from Suzuki K Poulose - perf.data header fixes from Namhyung Kim - Allow user to indicate objdump path, needed in cross environments, from Maciek Borzecki - Fix hardware cache event name generation, fix from Jiri Olsa - Add round trip test for sw, hw and cache event names, catching the problem Jiri fixed, after Jiri's patch, the test passes successfully. - Clean target should do clean for lib/traceevent too, fix from David Ahern - Check the right variable for allocation failure, fix from Namhyung Kim - Set up evsel->tp_format regardless of evsel->name being set already, fix from Namhyung Kim - Oprofile fixes from Robert Richter. - Remove perf_event_attr needless version inflation, from Jiri Olsa - Introduce libtraceevent strerror like error reporting facility, from Namhyung Kim - Add pmu mappings to perf.data header and use event names from cmd line, from Robert Richter - Fix include order for bison/flex-generated C files, from Ben Hutchings - Build fixes and documentation corrections from David Ahern - Assorted cleanups from Robert Richter - Let O= makes handle relative paths, from Steven Rostedt - perf script python fixes, from Feng Tang. - Initial bash completion support, from Frederic Weisbecker - Allow building without libelf, from Namhyung Kim. - Support DWARF CFI based unwind to have callchains when %bp based unwinding is not possible, from Jiri Olsa. - Symbol resolution fixes, while fixing support PPC64 files with an .opt ELF section was the end goal, several fixes for code that handles all architectures and cleanups are included, from Cody Schafer. - Assorted fixes for Documentation and build in 32 bit, from Robert Richter - Cache the libtraceevent event_format associated to each evsel early, so that we avoid relookups, i.e. calling pevent_find_event repeatedly when processing tracepoint events. [ This is to reduce the surface contact with libtraceevents and make clear what is that the perf tools needs from that lib: so far parsing the common and per event fields. ] - Don't stop the build if the audit libraries are not installed, fix from Namhyung Kim. - Fix bfd.h/libbfd detection with recent binutils, from Markus Trippelsdorf. - Improve warning message when libunwind devel packages not present, from Jiri Olsa" * 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (282 commits) perf trace: Add aliases for some syscalls perf probe: Print an enum type variable in "enum variable-name" format when showing accessible variables perf tools: Check libaudit availability for perf-trace builtin perf hists: Add missing period_* fields when collapsing a hist entry perf trace: New tool perf evsel: Export the event_format constructor perf evsel: Introduce rawptr() method perf tools: Use perf_evsel__newtp in the event parser perf evsel: The tracepoint constructor should store sys:name perf evlist: Introduce set_filter() method perf evlist: Renane set_filters method to apply_filters perf test: Add test to check we correctly parse and match syscall open parms perf evsel: Handle endianity in intval method perf evsel: Know if byte swap is needed perf tools: Allow handling a NULL cpu_map as meaning "all cpus" perf evsel: Improve tracepoint constructor setup tools lib traceevent: Fix error path on pevent_parse_event perf test: Fix build failure trace: Move trace event enable from fs_initcall to core_initcall tracing: Add an option for disabling markers ...
Diffstat (limited to 'tools/perf/util/hist.c')
-rw-r--r--tools/perf/util/hist.c721
1 files changed, 26 insertions, 695 deletions
diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
index f247ef2789a4..236bc9d98ff2 100644
--- a/tools/perf/util/hist.c
+++ b/tools/perf/util/hist.c
@@ -45,7 +45,7 @@ bool hists__new_col_len(struct hists *hists, enum hist_column col, u16 len)
45 return false; 45 return false;
46} 46}
47 47
48static void hists__reset_col_len(struct hists *hists) 48void hists__reset_col_len(struct hists *hists)
49{ 49{
50 enum hist_column col; 50 enum hist_column col;
51 51
@@ -63,7 +63,7 @@ static void hists__set_unres_dso_col_len(struct hists *hists, int dso)
63 hists__set_col_len(hists, dso, unresolved_col_width); 63 hists__set_col_len(hists, dso, unresolved_col_width);
64} 64}
65 65
66static void hists__calc_col_len(struct hists *hists, struct hist_entry *h) 66void hists__calc_col_len(struct hists *hists, struct hist_entry *h)
67{ 67{
68 const unsigned int unresolved_col_width = BITS_PER_LONG / 4; 68 const unsigned int unresolved_col_width = BITS_PER_LONG / 4;
69 u16 len; 69 u16 len;
@@ -114,6 +114,22 @@ static void hists__calc_col_len(struct hists *hists, struct hist_entry *h)
114 } 114 }
115} 115}
116 116
117void hists__output_recalc_col_len(struct hists *hists, int max_rows)
118{
119 struct rb_node *next = rb_first(&hists->entries);
120 struct hist_entry *n;
121 int row = 0;
122
123 hists__reset_col_len(hists);
124
125 while (next && row++ < max_rows) {
126 n = rb_entry(next, struct hist_entry, rb_node);
127 if (!n->filtered)
128 hists__calc_col_len(hists, n);
129 next = rb_next(&n->rb_node);
130 }
131}
132
117static void hist_entry__add_cpumode_period(struct hist_entry *he, 133static void hist_entry__add_cpumode_period(struct hist_entry *he,
118 unsigned int cpumode, u64 period) 134 unsigned int cpumode, u64 period)
119{ 135{
@@ -378,7 +394,7 @@ void hist_entry__free(struct hist_entry *he)
378 * collapse the histogram 394 * collapse the histogram
379 */ 395 */
380 396
381static bool hists__collapse_insert_entry(struct hists *hists __used, 397static bool hists__collapse_insert_entry(struct hists *hists __maybe_unused,
382 struct rb_root *root, 398 struct rb_root *root,
383 struct hist_entry *he) 399 struct hist_entry *he)
384{ 400{
@@ -394,8 +410,13 @@ static bool hists__collapse_insert_entry(struct hists *hists __used,
394 cmp = hist_entry__collapse(iter, he); 410 cmp = hist_entry__collapse(iter, he);
395 411
396 if (!cmp) { 412 if (!cmp) {
397 iter->period += he->period; 413 iter->period += he->period;
398 iter->nr_events += he->nr_events; 414 iter->period_sys += he->period_sys;
415 iter->period_us += he->period_us;
416 iter->period_guest_sys += he->period_guest_sys;
417 iter->period_guest_us += he->period_guest_us;
418 iter->nr_events += he->nr_events;
419
399 if (symbol_conf.use_callchain) { 420 if (symbol_conf.use_callchain) {
400 callchain_cursor_reset(&callchain_cursor); 421 callchain_cursor_reset(&callchain_cursor);
401 callchain_merge(&callchain_cursor, 422 callchain_merge(&callchain_cursor,
@@ -547,674 +568,6 @@ void hists__output_resort_threaded(struct hists *hists)
547 return __hists__output_resort(hists, true); 568 return __hists__output_resort(hists, true);
548} 569}
549 570
550static size_t callchain__fprintf_left_margin(FILE *fp, int left_margin)
551{
552 int i;
553 int ret = fprintf(fp, " ");
554
555 for (i = 0; i < left_margin; i++)
556 ret += fprintf(fp, " ");
557
558 return ret;
559}
560
561static size_t ipchain__fprintf_graph_line(FILE *fp, int depth, int depth_mask,
562 int left_margin)
563{
564 int i;
565 size_t ret = callchain__fprintf_left_margin(fp, left_margin);
566
567 for (i = 0; i < depth; i++)
568 if (depth_mask & (1 << i))
569 ret += fprintf(fp, "| ");
570 else
571 ret += fprintf(fp, " ");
572
573 ret += fprintf(fp, "\n");
574
575 return ret;
576}
577
578static size_t ipchain__fprintf_graph(FILE *fp, struct callchain_list *chain,
579 int depth, int depth_mask, int period,
580 u64 total_samples, u64 hits,
581 int left_margin)
582{
583 int i;
584 size_t ret = 0;
585
586 ret += callchain__fprintf_left_margin(fp, left_margin);
587 for (i = 0; i < depth; i++) {
588 if (depth_mask & (1 << i))
589 ret += fprintf(fp, "|");
590 else
591 ret += fprintf(fp, " ");
592 if (!period && i == depth - 1) {
593 double percent;
594
595 percent = hits * 100.0 / total_samples;
596 ret += percent_color_fprintf(fp, "--%2.2f%%-- ", percent);
597 } else
598 ret += fprintf(fp, "%s", " ");
599 }
600 if (chain->ms.sym)
601 ret += fprintf(fp, "%s\n", chain->ms.sym->name);
602 else
603 ret += fprintf(fp, "0x%0" PRIx64 "\n", chain->ip);
604
605 return ret;
606}
607
608static struct symbol *rem_sq_bracket;
609static struct callchain_list rem_hits;
610
611static void init_rem_hits(void)
612{
613 rem_sq_bracket = malloc(sizeof(*rem_sq_bracket) + 6);
614 if (!rem_sq_bracket) {
615 fprintf(stderr, "Not enough memory to display remaining hits\n");
616 return;
617 }
618
619 strcpy(rem_sq_bracket->name, "[...]");
620 rem_hits.ms.sym = rem_sq_bracket;
621}
622
623static size_t __callchain__fprintf_graph(FILE *fp, struct rb_root *root,
624 u64 total_samples, int depth,
625 int depth_mask, int left_margin)
626{
627 struct rb_node *node, *next;
628 struct callchain_node *child;
629 struct callchain_list *chain;
630 int new_depth_mask = depth_mask;
631 u64 remaining;
632 size_t ret = 0;
633 int i;
634 uint entries_printed = 0;
635
636 remaining = total_samples;
637
638 node = rb_first(root);
639 while (node) {
640 u64 new_total;
641 u64 cumul;
642
643 child = rb_entry(node, struct callchain_node, rb_node);
644 cumul = callchain_cumul_hits(child);
645 remaining -= cumul;
646
647 /*
648 * The depth mask manages the output of pipes that show
649 * the depth. We don't want to keep the pipes of the current
650 * level for the last child of this depth.
651 * Except if we have remaining filtered hits. They will
652 * supersede the last child
653 */
654 next = rb_next(node);
655 if (!next && (callchain_param.mode != CHAIN_GRAPH_REL || !remaining))
656 new_depth_mask &= ~(1 << (depth - 1));
657
658 /*
659 * But we keep the older depth mask for the line separator
660 * to keep the level link until we reach the last child
661 */
662 ret += ipchain__fprintf_graph_line(fp, depth, depth_mask,
663 left_margin);
664 i = 0;
665 list_for_each_entry(chain, &child->val, list) {
666 ret += ipchain__fprintf_graph(fp, chain, depth,
667 new_depth_mask, i++,
668 total_samples,
669 cumul,
670 left_margin);
671 }
672
673 if (callchain_param.mode == CHAIN_GRAPH_REL)
674 new_total = child->children_hit;
675 else
676 new_total = total_samples;
677
678 ret += __callchain__fprintf_graph(fp, &child->rb_root, new_total,
679 depth + 1,
680 new_depth_mask | (1 << depth),
681 left_margin);
682 node = next;
683 if (++entries_printed == callchain_param.print_limit)
684 break;
685 }
686
687 if (callchain_param.mode == CHAIN_GRAPH_REL &&
688 remaining && remaining != total_samples) {
689
690 if (!rem_sq_bracket)
691 return ret;
692
693 new_depth_mask &= ~(1 << (depth - 1));
694 ret += ipchain__fprintf_graph(fp, &rem_hits, depth,
695 new_depth_mask, 0, total_samples,
696 remaining, left_margin);
697 }
698
699 return ret;
700}
701
702static size_t callchain__fprintf_graph(FILE *fp, struct rb_root *root,
703 u64 total_samples, int left_margin)
704{
705 struct callchain_node *cnode;
706 struct callchain_list *chain;
707 u32 entries_printed = 0;
708 bool printed = false;
709 struct rb_node *node;
710 int i = 0;
711 int ret = 0;
712
713 /*
714 * If have one single callchain root, don't bother printing
715 * its percentage (100 % in fractal mode and the same percentage
716 * than the hist in graph mode). This also avoid one level of column.
717 */
718 node = rb_first(root);
719 if (node && !rb_next(node)) {
720 cnode = rb_entry(node, struct callchain_node, rb_node);
721 list_for_each_entry(chain, &cnode->val, list) {
722 /*
723 * If we sort by symbol, the first entry is the same than
724 * the symbol. No need to print it otherwise it appears as
725 * displayed twice.
726 */
727 if (!i++ && sort__first_dimension == SORT_SYM)
728 continue;
729 if (!printed) {
730 ret += callchain__fprintf_left_margin(fp, left_margin);
731 ret += fprintf(fp, "|\n");
732 ret += callchain__fprintf_left_margin(fp, left_margin);
733 ret += fprintf(fp, "---");
734 left_margin += 3;
735 printed = true;
736 } else
737 ret += callchain__fprintf_left_margin(fp, left_margin);
738
739 if (chain->ms.sym)
740 ret += fprintf(fp, " %s\n", chain->ms.sym->name);
741 else
742 ret += fprintf(fp, " %p\n", (void *)(long)chain->ip);
743
744 if (++entries_printed == callchain_param.print_limit)
745 break;
746 }
747 root = &cnode->rb_root;
748 }
749
750 ret += __callchain__fprintf_graph(fp, root, total_samples,
751 1, 1, left_margin);
752 ret += fprintf(fp, "\n");
753
754 return ret;
755}
756
757static size_t __callchain__fprintf_flat(FILE *fp,
758 struct callchain_node *self,
759 u64 total_samples)
760{
761 struct callchain_list *chain;
762 size_t ret = 0;
763
764 if (!self)
765 return 0;
766
767 ret += __callchain__fprintf_flat(fp, self->parent, total_samples);
768
769
770 list_for_each_entry(chain, &self->val, list) {
771 if (chain->ip >= PERF_CONTEXT_MAX)
772 continue;
773 if (chain->ms.sym)
774 ret += fprintf(fp, " %s\n", chain->ms.sym->name);
775 else
776 ret += fprintf(fp, " %p\n",
777 (void *)(long)chain->ip);
778 }
779
780 return ret;
781}
782
783static size_t callchain__fprintf_flat(FILE *fp, struct rb_root *self,
784 u64 total_samples)
785{
786 size_t ret = 0;
787 u32 entries_printed = 0;
788 struct rb_node *rb_node;
789 struct callchain_node *chain;
790
791 rb_node = rb_first(self);
792 while (rb_node) {
793 double percent;
794
795 chain = rb_entry(rb_node, struct callchain_node, rb_node);
796 percent = chain->hit * 100.0 / total_samples;
797
798 ret = percent_color_fprintf(fp, " %6.2f%%\n", percent);
799 ret += __callchain__fprintf_flat(fp, chain, total_samples);
800 ret += fprintf(fp, "\n");
801 if (++entries_printed == callchain_param.print_limit)
802 break;
803
804 rb_node = rb_next(rb_node);
805 }
806
807 return ret;
808}
809
810static size_t hist_entry_callchain__fprintf(struct hist_entry *he,
811 u64 total_samples, int left_margin,
812 FILE *fp)
813{
814 switch (callchain_param.mode) {
815 case CHAIN_GRAPH_REL:
816 return callchain__fprintf_graph(fp, &he->sorted_chain, he->period,
817 left_margin);
818 break;
819 case CHAIN_GRAPH_ABS:
820 return callchain__fprintf_graph(fp, &he->sorted_chain, total_samples,
821 left_margin);
822 break;
823 case CHAIN_FLAT:
824 return callchain__fprintf_flat(fp, &he->sorted_chain, total_samples);
825 break;
826 case CHAIN_NONE:
827 break;
828 default:
829 pr_err("Bad callchain mode\n");
830 }
831
832 return 0;
833}
834
835void hists__output_recalc_col_len(struct hists *hists, int max_rows)
836{
837 struct rb_node *next = rb_first(&hists->entries);
838 struct hist_entry *n;
839 int row = 0;
840
841 hists__reset_col_len(hists);
842
843 while (next && row++ < max_rows) {
844 n = rb_entry(next, struct hist_entry, rb_node);
845 if (!n->filtered)
846 hists__calc_col_len(hists, n);
847 next = rb_next(&n->rb_node);
848 }
849}
850
851static int hist_entry__pcnt_snprintf(struct hist_entry *he, char *s,
852 size_t size, struct hists *pair_hists,
853 bool show_displacement, long displacement,
854 bool color, u64 total_period)
855{
856 u64 period, total, period_sys, period_us, period_guest_sys, period_guest_us;
857 u64 nr_events;
858 const char *sep = symbol_conf.field_sep;
859 int ret;
860
861 if (symbol_conf.exclude_other && !he->parent)
862 return 0;
863
864 if (pair_hists) {
865 period = he->pair ? he->pair->period : 0;
866 nr_events = he->pair ? he->pair->nr_events : 0;
867 total = pair_hists->stats.total_period;
868 period_sys = he->pair ? he->pair->period_sys : 0;
869 period_us = he->pair ? he->pair->period_us : 0;
870 period_guest_sys = he->pair ? he->pair->period_guest_sys : 0;
871 period_guest_us = he->pair ? he->pair->period_guest_us : 0;
872 } else {
873 period = he->period;
874 nr_events = he->nr_events;
875 total = total_period;
876 period_sys = he->period_sys;
877 period_us = he->period_us;
878 period_guest_sys = he->period_guest_sys;
879 period_guest_us = he->period_guest_us;
880 }
881
882 if (total) {
883 if (color)
884 ret = percent_color_snprintf(s, size,
885 sep ? "%.2f" : " %6.2f%%",
886 (period * 100.0) / total);
887 else
888 ret = scnprintf(s, size, sep ? "%.2f" : " %6.2f%%",
889 (period * 100.0) / total);
890 if (symbol_conf.show_cpu_utilization) {
891 ret += percent_color_snprintf(s + ret, size - ret,
892 sep ? "%.2f" : " %6.2f%%",
893 (period_sys * 100.0) / total);
894 ret += percent_color_snprintf(s + ret, size - ret,
895 sep ? "%.2f" : " %6.2f%%",
896 (period_us * 100.0) / total);
897 if (perf_guest) {
898 ret += percent_color_snprintf(s + ret,
899 size - ret,
900 sep ? "%.2f" : " %6.2f%%",
901 (period_guest_sys * 100.0) /
902 total);
903 ret += percent_color_snprintf(s + ret,
904 size - ret,
905 sep ? "%.2f" : " %6.2f%%",
906 (period_guest_us * 100.0) /
907 total);
908 }
909 }
910 } else
911 ret = scnprintf(s, size, sep ? "%" PRIu64 : "%12" PRIu64 " ", period);
912
913 if (symbol_conf.show_nr_samples) {
914 if (sep)
915 ret += scnprintf(s + ret, size - ret, "%c%" PRIu64, *sep, nr_events);
916 else
917 ret += scnprintf(s + ret, size - ret, "%11" PRIu64, nr_events);
918 }
919
920 if (symbol_conf.show_total_period) {
921 if (sep)
922 ret += scnprintf(s + ret, size - ret, "%c%" PRIu64, *sep, period);
923 else
924 ret += scnprintf(s + ret, size - ret, " %12" PRIu64, period);
925 }
926
927 if (pair_hists) {
928 char bf[32];
929 double old_percent = 0, new_percent = 0, diff;
930
931 if (total > 0)
932 old_percent = (period * 100.0) / total;
933 if (total_period > 0)
934 new_percent = (he->period * 100.0) / total_period;
935
936 diff = new_percent - old_percent;
937
938 if (fabs(diff) >= 0.01)
939 scnprintf(bf, sizeof(bf), "%+4.2F%%", diff);
940 else
941 scnprintf(bf, sizeof(bf), " ");
942
943 if (sep)
944 ret += scnprintf(s + ret, size - ret, "%c%s", *sep, bf);
945 else
946 ret += scnprintf(s + ret, size - ret, "%11.11s", bf);
947
948 if (show_displacement) {
949 if (displacement)
950 scnprintf(bf, sizeof(bf), "%+4ld", displacement);
951 else
952 scnprintf(bf, sizeof(bf), " ");
953
954 if (sep)
955 ret += scnprintf(s + ret, size - ret, "%c%s", *sep, bf);
956 else
957 ret += scnprintf(s + ret, size - ret, "%6.6s", bf);
958 }
959 }
960
961 return ret;
962}
963
964int hist_entry__snprintf(struct hist_entry *he, char *s, size_t size,
965 struct hists *hists)
966{
967 const char *sep = symbol_conf.field_sep;
968 struct sort_entry *se;
969 int ret = 0;
970
971 list_for_each_entry(se, &hist_entry__sort_list, list) {
972 if (se->elide)
973 continue;
974
975 ret += scnprintf(s + ret, size - ret, "%s", sep ?: " ");
976 ret += se->se_snprintf(he, s + ret, size - ret,
977 hists__col_len(hists, se->se_width_idx));
978 }
979
980 return ret;
981}
982
983static int hist_entry__fprintf(struct hist_entry *he, size_t size,
984 struct hists *hists, struct hists *pair_hists,
985 bool show_displacement, long displacement,
986 u64 total_period, FILE *fp)
987{
988 char bf[512];
989 int ret;
990
991 if (size == 0 || size > sizeof(bf))
992 size = sizeof(bf);
993
994 ret = hist_entry__pcnt_snprintf(he, bf, size, pair_hists,
995 show_displacement, displacement,
996 true, total_period);
997 hist_entry__snprintf(he, bf + ret, size - ret, hists);
998 return fprintf(fp, "%s\n", bf);
999}
1000
1001static size_t hist_entry__fprintf_callchain(struct hist_entry *he,
1002 struct hists *hists,
1003 u64 total_period, FILE *fp)
1004{
1005 int left_margin = 0;
1006
1007 if (sort__first_dimension == SORT_COMM) {
1008 struct sort_entry *se = list_first_entry(&hist_entry__sort_list,
1009 typeof(*se), list);
1010 left_margin = hists__col_len(hists, se->se_width_idx);
1011 left_margin -= thread__comm_len(he->thread);
1012 }
1013
1014 return hist_entry_callchain__fprintf(he, total_period, left_margin, fp);
1015}
1016
1017size_t hists__fprintf(struct hists *hists, struct hists *pair,
1018 bool show_displacement, bool show_header, int max_rows,
1019 int max_cols, FILE *fp)
1020{
1021 struct sort_entry *se;
1022 struct rb_node *nd;
1023 size_t ret = 0;
1024 u64 total_period;
1025 unsigned long position = 1;
1026 long displacement = 0;
1027 unsigned int width;
1028 const char *sep = symbol_conf.field_sep;
1029 const char *col_width = symbol_conf.col_width_list_str;
1030 int nr_rows = 0;
1031
1032 init_rem_hits();
1033
1034 if (!show_header)
1035 goto print_entries;
1036
1037 fprintf(fp, "# %s", pair ? "Baseline" : "Overhead");
1038
1039 if (symbol_conf.show_cpu_utilization) {
1040 if (sep) {
1041 ret += fprintf(fp, "%csys", *sep);
1042 ret += fprintf(fp, "%cus", *sep);
1043 if (perf_guest) {
1044 ret += fprintf(fp, "%cguest sys", *sep);
1045 ret += fprintf(fp, "%cguest us", *sep);
1046 }
1047 } else {
1048 ret += fprintf(fp, " sys ");
1049 ret += fprintf(fp, " us ");
1050 if (perf_guest) {
1051 ret += fprintf(fp, " guest sys ");
1052 ret += fprintf(fp, " guest us ");
1053 }
1054 }
1055 }
1056
1057 if (symbol_conf.show_nr_samples) {
1058 if (sep)
1059 fprintf(fp, "%cSamples", *sep);
1060 else
1061 fputs(" Samples ", fp);
1062 }
1063
1064 if (symbol_conf.show_total_period) {
1065 if (sep)
1066 ret += fprintf(fp, "%cPeriod", *sep);
1067 else
1068 ret += fprintf(fp, " Period ");
1069 }
1070
1071 if (pair) {
1072 if (sep)
1073 ret += fprintf(fp, "%cDelta", *sep);
1074 else
1075 ret += fprintf(fp, " Delta ");
1076
1077 if (show_displacement) {
1078 if (sep)
1079 ret += fprintf(fp, "%cDisplacement", *sep);
1080 else
1081 ret += fprintf(fp, " Displ");
1082 }
1083 }
1084
1085 list_for_each_entry(se, &hist_entry__sort_list, list) {
1086 if (se->elide)
1087 continue;
1088 if (sep) {
1089 fprintf(fp, "%c%s", *sep, se->se_header);
1090 continue;
1091 }
1092 width = strlen(se->se_header);
1093 if (symbol_conf.col_width_list_str) {
1094 if (col_width) {
1095 hists__set_col_len(hists, se->se_width_idx,
1096 atoi(col_width));
1097 col_width = strchr(col_width, ',');
1098 if (col_width)
1099 ++col_width;
1100 }
1101 }
1102 if (!hists__new_col_len(hists, se->se_width_idx, width))
1103 width = hists__col_len(hists, se->se_width_idx);
1104 fprintf(fp, " %*s", width, se->se_header);
1105 }
1106
1107 fprintf(fp, "\n");
1108 if (max_rows && ++nr_rows >= max_rows)
1109 goto out;
1110
1111 if (sep)
1112 goto print_entries;
1113
1114 fprintf(fp, "# ........");
1115 if (symbol_conf.show_cpu_utilization)
1116 fprintf(fp, " ....... .......");
1117 if (symbol_conf.show_nr_samples)
1118 fprintf(fp, " ..........");
1119 if (symbol_conf.show_total_period)
1120 fprintf(fp, " ............");
1121 if (pair) {
1122 fprintf(fp, " ..........");
1123 if (show_displacement)
1124 fprintf(fp, " .....");
1125 }
1126 list_for_each_entry(se, &hist_entry__sort_list, list) {
1127 unsigned int i;
1128
1129 if (se->elide)
1130 continue;
1131
1132 fprintf(fp, " ");
1133 width = hists__col_len(hists, se->se_width_idx);
1134 if (width == 0)
1135 width = strlen(se->se_header);
1136 for (i = 0; i < width; i++)
1137 fprintf(fp, ".");
1138 }
1139
1140 fprintf(fp, "\n");
1141 if (max_rows && ++nr_rows >= max_rows)
1142 goto out;
1143
1144 fprintf(fp, "#\n");
1145 if (max_rows && ++nr_rows >= max_rows)
1146 goto out;
1147
1148print_entries:
1149 total_period = hists->stats.total_period;
1150
1151 for (nd = rb_first(&hists->entries); nd; nd = rb_next(nd)) {
1152 struct hist_entry *h = rb_entry(nd, struct hist_entry, rb_node);
1153
1154 if (h->filtered)
1155 continue;
1156
1157 if (show_displacement) {
1158 if (h->pair != NULL)
1159 displacement = ((long)h->pair->position -
1160 (long)position);
1161 else
1162 displacement = 0;
1163 ++position;
1164 }
1165 ret += hist_entry__fprintf(h, max_cols, hists, pair, show_displacement,
1166 displacement, total_period, fp);
1167
1168 if (symbol_conf.use_callchain)
1169 ret += hist_entry__fprintf_callchain(h, hists, total_period, fp);
1170 if (max_rows && ++nr_rows >= max_rows)
1171 goto out;
1172
1173 if (h->ms.map == NULL && verbose > 1) {
1174 __map_groups__fprintf_maps(&h->thread->mg,
1175 MAP__FUNCTION, verbose, fp);
1176 fprintf(fp, "%.10s end\n", graph_dotted_line);
1177 }
1178 }
1179out:
1180 free(rem_sq_bracket);
1181
1182 return ret;
1183}
1184
1185/*
1186 * See hists__fprintf to match the column widths
1187 */
1188unsigned int hists__sort_list_width(struct hists *hists)
1189{
1190 struct sort_entry *se;
1191 int ret = 9; /* total % */
1192
1193 if (symbol_conf.show_cpu_utilization) {
1194 ret += 7; /* count_sys % */
1195 ret += 6; /* count_us % */
1196 if (perf_guest) {
1197 ret += 13; /* count_guest_sys % */
1198 ret += 12; /* count_guest_us % */
1199 }
1200 }
1201
1202 if (symbol_conf.show_nr_samples)
1203 ret += 11;
1204
1205 if (symbol_conf.show_total_period)
1206 ret += 13;
1207
1208 list_for_each_entry(se, &hist_entry__sort_list, list)
1209 if (!se->elide)
1210 ret += 2 + hists__col_len(hists, se->se_width_idx);
1211
1212 if (verbose) /* Addr + origin */
1213 ret += 3 + BITS_PER_LONG / 4;
1214
1215 return ret;
1216}
1217
1218static void hists__remove_entry_filter(struct hists *hists, struct hist_entry *h, 571static void hists__remove_entry_filter(struct hists *hists, struct hist_entry *h,
1219 enum hist_filter filter) 572 enum hist_filter filter)
1220{ 573{
@@ -1342,25 +695,3 @@ void hists__inc_nr_events(struct hists *hists, u32 type)
1342 ++hists->stats.nr_events[0]; 695 ++hists->stats.nr_events[0];
1343 ++hists->stats.nr_events[type]; 696 ++hists->stats.nr_events[type];
1344} 697}
1345
1346size_t hists__fprintf_nr_events(struct hists *hists, FILE *fp)
1347{
1348 int i;
1349 size_t ret = 0;
1350
1351 for (i = 0; i < PERF_RECORD_HEADER_MAX; ++i) {
1352 const char *name;
1353
1354 if (hists->stats.nr_events[i] == 0)
1355 continue;
1356
1357 name = perf_event__name(i);
1358 if (!strcmp(name, "UNKNOWN"))
1359 continue;
1360
1361 ret += fprintf(fp, "%16s events: %10d\n", name,
1362 hists->stats.nr_events[i]);
1363 }
1364
1365 return ret;
1366}
generated by cgit v1.2.2 (git 2.25.0) at 2025-12-17 18:24:58 -0500