Skip to contentSkip to conversation

perf top: Support --group-sort-idx to change the sort order

PATCH Superseded Jin Yao 2 participants 4 messages Mar 20, 2020 – Mar 22, 2020 perf events

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Jin Yao

author
#0 lore raw

PATCH 1/2 perf top: Support --group-sort-idx to change the sort order

perf report has supported the option --group-sort-idx, which sorts the output by the event at the index n in event group.

For example,
perf record -e cycles,instructions,cache-misses
perf report --group --group-sort-idx 2 --stdio

The perf-report output is sorted by cache-misses.

This patch supports --group-sort-idx in perf-top.

For example,
perf top --group -e cycles,instructions,cache-misses --group-sort-idx 2

The perf-top output is sorted by cache-misses.

---
 tools/perf/Documentation/perf-top.txt | 5 +++++
 tools/perf/builtin-top.c              | 4 ++++
 2 files changed, 9 insertions(+)
tools/perf/Documentation/perf-top.txt +5 -0
index 324b6b53c86b..0e97dcef794d 100644
@@ -53,6 +53,11 @@ Default is to monitor all CPUS.
 --group::
         Put the counters into a counter group.
 
+--group-sort-idx::
+	Sort the output by the event at the index n in group. If n is invalid,
+	sort by the first event. It can support multiple groups with different
+	amount of events. WARNING: This should be used on grouped events.
+
 -F <freq>::
 --freq=<freq>::
 	Profile at this frequency. Use 'max' to use the currently maximum
tools/perf/builtin-top.c +4 -0
index f6dd1a63f159..144043637cec 100644
@@ -1543,6 +1543,10 @@ int cmd_top(int argc, const char **argv)
 			"number of thread to run event synthesize"),
 	OPT_BOOLEAN(0, "namespaces", &opts->record_namespaces,
 		    "Record namespaces events"),
+	OPT_INTEGER(0, "group-sort-idx", &symbol_conf.group_sort_idx,
+		    "Sort the output by the event at the index n in group. "
+		    "If n is invalid, sort by the first event. "
+		    "WARNING: should be used on grouped events."),
 	OPTS_EVSWITCH(&top.evswitch),
 	OPT_END()
 	};
Signature · 1 line
2.17.1
author
#1 lore raw

PATCH 2/2 perf top: support hotkey to change sort order

It would be nice if we can use a hotkey in perf top browser to select a event for sorting.

For example,
perf top --group -e cycles,instructions,cache-misses

Samples

                Overhead  Shared Object             Symbol
  40.03%  45.71%   0.03%  div                       [.] main
  20.46%  14.67%   0.21%  libc-2.27.so              [.] __random_r
  20.01%  19.54%   0.02%  libc-2.27.so              [.] __random
   9.68%  10.68%   0.00%  div                       [.] compute_flag
   4.32%   4.70%   0.00%  libc-2.27.so              [.] rand
   3.84%   3.43%   0.00%  div                       [.] rand@plt
   0.05%   0.05%   2.33%  libc-2.27.so              [.] __strcmp_sse2_unaligned
   0.04%   0.08%   2.43%  perf                      [.] perf_hpp__is_dynamic_en
   0.04%   0.02%   6.64%  perf                      [.] rb_next
   0.04%   0.01%   3.87%  perf                      [.] dso__find_symbol
   0.04%   0.04%   1.77%  perf                      [.] sort__dso_cmp

When user press hotkey '2' (event index, starting from 0), it indicates to sort output by the third event in group (cache-misses).

Samples

                Overhead  Shared Object               Symbol
   4.07%   1.28%   6.68%  perf                        [.] rb_next
   3.57%   3.98%   4.11%  perf                        [.] __hists__insert_output
   3.67%  11.24%   3.60%  perf                        [.] perf_hpp__is_dynamic_e
   3.67%   3.20%   3.20%  perf                        [.] hpp__sort_overhead
   0.81%   0.06%   3.01%  perf                        [.] dso__find_symbol
   1.62%   5.47%   2.51%  perf                        [.] hists__match
   2.70%   1.86%   2.47%  libc-2.27.so                [.] _int_malloc
   0.19%   0.00%   2.29%  [kernel]                    [k] copy_page
   0.41%   0.32%   1.98%  perf                        [.] hists__decay_entries
   1.84%   3.67%   1.68%  perf                        [.] sort__dso_cmp
   0.16%   0.00%   1.63%  [kernel]                    [k] clear_page_erms

Now the output is sorted by cache-misses.

---
 tools/perf/builtin-top.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)
tools/perf/builtin-top.c +8 -2
index 144043637cec..b39f6ffb874e 100644
@@ -616,6 +616,7 @@ static void *display_thread_tui(void *arg)
 		.arg		= top,
 		.refresh	= top->delay_secs,
 	};
+	int ret;
 
 	/* In order to read symbols from other namespaces perf to  needs to call
 	 * setns(2).  This isn't permitted if the struct_fs has multiple users.
@@ -626,6 +627,7 @@ static void *display_thread_tui(void *arg)
 
 	prctl(PR_SET_NAME, "perf-top-UI", 0, 0, 0);
 
+repeat:
 	perf_top__sort_new_samples(top);
 
 	/*
@@ -638,13 +640,17 @@ static void *display_thread_tui(void *arg)
 		hists->uid_filter_str = top->record_opts.target.uid_str;
 	}
 
-	perf_evlist__tui_browse_hists(top->evlist, help, &hbt,
+	ret = perf_evlist__tui_browse_hists(top->evlist, help, &hbt,
 				      top->min_percent,
 				      &top->session->header.env,
 				      !top->record_opts.overwrite,
 				      &top->annotation_opts);
 
-	stop_top();
+	if (ret == K_RELOAD)
+		goto repeat;
+	else
+		stop_top();
+
 	return NULL;
 }
 
Signature · 1 line
2.17.1
#2 lore raw

Re: [PATCH 2/2] perf top: support hotkey to change sort order

Em Fri, Mar 20, 2020 at 03:24:14PM +0800, Jin Yao escreveu:

Quoted 71 lines

It would be nice if we can use a hotkey in perf top browser to select a event for sorting.

For example,
perf top --group -e cycles,instructions,cache-misses

Samples

                Overhead  Shared Object             Symbol
  40.03%  45.71%   0.03%  div                       [.] main
  20.46%  14.67%   0.21%  libc-2.27.so              [.] __random_r
  20.01%  19.54%   0.02%  libc-2.27.so              [.] __random
   9.68%  10.68%   0.00%  div                       [.] compute_flag
   4.32%   4.70%   0.00%  libc-2.27.so              [.] rand
   3.84%   3.43%   0.00%  div                       [.] rand@plt
   0.05%   0.05%   2.33%  libc-2.27.so              [.] __strcmp_sse2_unaligned
   0.04%   0.08%   2.43%  perf                      [.] perf_hpp__is_dynamic_en
   0.04%   0.02%   6.64%  perf                      [.] rb_next
   0.04%   0.01%   3.87%  perf                      [.] dso__find_symbol
   0.04%   0.04%   1.77%  perf                      [.] sort__dso_cmp

When user press hotkey '2' (event index, starting from 0), it indicates to sort output by the third event in group (cache-misses).

Samples

                Overhead  Shared Object               Symbol
   4.07%   1.28%   6.68%  perf                        [.] rb_next
   3.57%   3.98%   4.11%  perf                        [.] __hists__insert_output
   3.67%  11.24%   3.60%  perf                        [.] perf_hpp__is_dynamic_e
   3.67%   3.20%   3.20%  perf                        [.] hpp__sort_overhead
   0.81%   0.06%   3.01%  perf                        [.] dso__find_symbol
   1.62%   5.47%   2.51%  perf                        [.] hists__match
   2.70%   1.86%   2.47%  libc-2.27.so                [.] _int_malloc
   0.19%   0.00%   2.29%  [kernel]                    [k] copy_page
   0.41%   0.32%   1.98%  perf                        [.] hists__decay_entries
   1.84%   3.67%   1.68%  perf                        [.] sort__dso_cmp
   0.16%   0.00%   1.63%  [kernel]                    [k] clear_page_erms

Now the output is sorted by cache-misses.

---
 tools/perf/builtin-top.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)
tools/perf/builtin-top.c +3 -1
index 144043637cec..b39f6ffb874e 100644
@@ -616,6 +616,7 @@ static void *display_thread_tui(void *arg)
 		.arg		= top,
 		.refresh	= top->delay_secs,
 	};
+	int ret;
 
 	/* In order to read symbols from other namespaces perf to  needs to call
 	 * setns(2).  This isn't permitted if the struct_fs has multiple users.
@@ -626,6 +627,7 @@ static void *display_thread_tui(void *arg)
 
 	prctl(PR_SET_NAME, "perf-top-UI", 0, 0, 0);
 
+repeat:
 	perf_top__sort_new_samples(top);
 
 	/*
@@ -638,13 +640,17 @@ static void *display_thread_tui(void *arg)
 		hists->uid_filter_str = top->record_opts.target.uid_str;
 	}
 
-	perf_evlist__tui_browse_hists(top->evlist, help, &hbt,
+	ret = perf_evlist__tui_browse_hists(top->evlist, help, &hbt,
 				      top->min_percent,

tools/perf/builtin-top.c static void *display_thread_tui(void *arg) · line 653

Quoted 10 lines
tools/perf/builtin-top.c +5 -1
 				      &top->session->header.env,
 				      !top->record_opts.overwrite,
 				      &top->annotation_opts);
 
-	stop_top();
+	if (ret == K_RELOAD)
+		goto repeat;
+	else
+		stop_top();
+

That is really nice and small, but shouldn't we flush all the histograms that were in place, sorted by the previous key? I think we have a 'z' for zeroing samples that may be what we need, take a look, please,

- Arnaldo

Quoted 6 lines
tools/perf/builtin-top.c +0 -0
 	return NULL;
 }
 
Signature · 1 line
2.17.1
Signature · 1 line
- Arnaldo
author
#3 lore raw

Re: [PATCH 2/2] perf top: support hotkey to change sort order

Quoted 87 lines · Arnaldo Carvalho de Melo

Em Fri, Mar 20, 2020 at 03:24:14PM +0800, Jin Yao escreveu:

Quoted 71 lines

It would be nice if we can use a hotkey in perf top browser to select a event for sorting.

For example,
perf top --group -e cycles,instructions,cache-misses

Samples

                 Overhead  Shared Object             Symbol
   40.03%  45.71%   0.03%  div                       [.] main
   20.46%  14.67%   0.21%  libc-2.27.so              [.] __random_r
   20.01%  19.54%   0.02%  libc-2.27.so              [.] __random
    9.68%  10.68%   0.00%  div                       [.] compute_flag
    4.32%   4.70%   0.00%  libc-2.27.so              [.] rand
    3.84%   3.43%   0.00%  div                       [.] rand@plt
    0.05%   0.05%   2.33%  libc-2.27.so              [.] __strcmp_sse2_unaligned
    0.04%   0.08%   2.43%  perf                      [.] perf_hpp__is_dynamic_en
    0.04%   0.02%   6.64%  perf                      [.] rb_next
    0.04%   0.01%   3.87%  perf                      [.] dso__find_symbol
    0.04%   0.04%   1.77%  perf                      [.] sort__dso_cmp

When user press hotkey '2' (event index, starting from 0), it indicates to sort output by the third event in group (cache-misses).

Samples

                 Overhead  Shared Object               Symbol
    4.07%   1.28%   6.68%  perf                        [.] rb_next
    3.57%   3.98%   4.11%  perf                        [.] __hists__insert_output
    3.67%  11.24%   3.60%  perf                        [.] perf_hpp__is_dynamic_e
    3.67%   3.20%   3.20%  perf                        [.] hpp__sort_overhead
    0.81%   0.06%   3.01%  perf                        [.] dso__find_symbol
    1.62%   5.47%   2.51%  perf                        [.] hists__match
    2.70%   1.86%   2.47%  libc-2.27.so                [.] _int_malloc
    0.19%   0.00%   2.29%  [kernel]                    [k] copy_page
    0.41%   0.32%   1.98%  perf                        [.] hists__decay_entries
    1.84%   3.67%   1.68%  perf                        [.] sort__dso_cmp
    0.16%   0.00%   1.63%  [kernel]                    [k] clear_page_erms

Now the output is sorted by cache-misses.

---
  tools/perf/builtin-top.c | 10 ++++++++--
  1 file changed, 8 insertions(+), 2 deletions(-)
tools/perf/builtin-top.c +3 -1
index 144043637cec..b39f6ffb874e 100644
@@ -616,6 +616,7 @@ static void *display_thread_tui(void *arg)
  		.arg		= top,
  		.refresh	= top->delay_secs,
  	};
+	int ret;
 
  	/* In order to read symbols from other namespaces perf to  needs to call
  	 * setns(2).  This isn't permitted if the struct_fs has multiple users.
@@ -626,6 +627,7 @@ static void *display_thread_tui(void *arg)
 
  	prctl(PR_SET_NAME, "perf-top-UI", 0, 0, 0);
 
+repeat:
  	perf_top__sort_new_samples(top);
 
  	/*
@@ -638,13 +640,17 @@ static void *display_thread_tui(void *arg)
  		hists->uid_filter_str = top->record_opts.target.uid_str;
  	}
 
-	perf_evlist__tui_browse_hists(top->evlist, help, &hbt,
+	ret = perf_evlist__tui_browse_hists(top->evlist, help, &hbt,
  				      top->min_percent,

tools/perf/builtin-top.c static void *display_thread_tui(void *arg) · line 653

Quoted 10 lines
tools/perf/builtin-top.c +5 -1
  				      &top->session->header.env,
  				      !top->record_opts.overwrite,
  				      &top->annotation_opts);
 
-	stop_top();
+	if (ret == K_RELOAD)
+		goto repeat;
+	else
+		stop_top();
+

That is really nice and small, but shouldn't we flush all the histograms that were in place, sorted by the previous key? I think we have a 'z' for zeroing samples that may be what we need, take a look, please,

Quoted 2 lines

- Arnaldo

Set top->zero to true if return key is K_RELOAD. So that in perf_top__resort_hists(), it will delete hists entries by calling hists__delete_entries(hists).

+repeat:

        perf_top__sort_new_samples(top);

        /*
@@ -638,13 +640,18 @@ static void *display_thread_tui(void *arg)
                hists->uid_filter_str = top->record_opts.target.uid_str;
        }

- perf_evlist__tui_browse_hists(top->evlist, help, &hbt,
+ ret = perf_evlist__tui_browse_hists(top->evlist, help, &hbt,

                                      top->min_percent,
                                      &top->session->header.env,
                                      !top->record_opts.overwrite,
                                      &top->annotation_opts);

- stop_top();
+ if (ret == K_RELOAD) {
+ top->zero = true;
+ goto repeat;
+ } else
+ stop_top();
+

Is this OK?

Thanks
Jin Yao

Quoted 7 lines
Quoted 6 lines
tools/perf/builtin-top.c +0 -0
  	return NULL;
  }
 
Signature · 1 line
2.17.1