Skip to contentSkip to conversation

perf tools: Honor column width setting

PATCH In review Namhyung Kim 3 participants 23 messages Jul 9, 2014 – Jul 24, 2014 perf events

lore
Share & tools
lore

Namhyung Kim

author maintainer
#0 lore raw

cover perf tools: Honor column width setting

Hello,

This patchset is to control perf report/top output column width by -w/--column-widths option so that it can fit into the terminal size. The -w option is there for perf report but it ignored by recent output field changed due to some reason. This patchset fixes it and supports perf top also.

This is sometimes useful if your terminal is small and there's some C++ applications which have amazingly long symbol names. Without this patchset user might not see those symbols on TUI, since it maps left/right arrow keys to other functions.

The -w option sets column width starting from the first column (overhead or optional overhead_children column unless -F option is given). It doesn't make sense to limit those overhead columns so it's not a hard-limit for them. But it *is* a hard-limit for other columns such as comm, dso, symbol, and so on. One can use 0 not to limit/force a width for those columns.

For example:

  $ perf report --stdio -s comm,dso
  ...
  # Overhead  Command          Shared Object                                        
  # ........  ...............  .....................................................
  #
      71.08%  gnome-shell      perf-1497.map                                        
       9.23%  gnome-shell      swrast_dri.so                                        
       3.99%  Xorg             [unknown]                                            
       3.18%  Xorg             [kernel.kallsyms]                                    
       2.93%  gnome-shell      [kernel.kallsyms]                                    
       2.41%  swapper          [kernel.kallsyms]                                    
       1.55%  synergys         libpthread-2.15.so                                   
       1.08%  synergys         synergys                                             
       0.68%  systemd-journal  [kernel.kallsyms]                                    
       0.59%  synergys         libstdc++.so.6.0.17                                  
       0.58%  gnome-shell      libc-2.15.so                                         
       0.54%  kworker/0:2      [kernel.kallsyms]                                    
       0.20%  systemd-journal  [unknown]                                            
       0.19%  gnome-shell      libclutter-1.0.so.0.1000.8.#prelink#.1kh1we (deleted)
       0.18%  firefox          libxul.so (deleted)                                  
       0.17%  gnome-shell      libcogl.so.9.1.1.#prelink#.ZL3fn1 (deleted)          
       0.14%  firefox          [kernel.kallsyms]                                    
       0.14%  gnome-shell      libgobject-2.0.so.0.3200.4                           
       0.13%  gnome-shell      libpthread-2.15.so                                   
       0.11%  synergys         [kernel.kallsyms]                                    
       0.10%  perf             [kernel.kallsyms]                                    
  $ perf report --stdio -s comm,dso -w 0,10,20   # 0 means no limit
  ...
  # Overhead  Command     Shared Object       
  # ........  ..........  ....................
  #
      71.08%  gnome-shel  perf-1497.map       
       9.23%  gnome-shel  swrast_dri.so       
       3.99%  Xorg        [unknown]           
       3.18%  Xorg        [kernel.kallsyms]   
       2.93%  gnome-shel  [kernel.kallsyms]   
       2.41%  swapper     [kernel.kallsyms]   
       1.55%  synergys    libpthread-2.15.so  
       1.08%  synergys    synergys            
       0.68%  systemd-jo  [kernel.kallsyms]   
       0.59%  synergys    libstdc++.so.6.0.17 
       0.58%  gnome-shel  libc-2.15.so        
       0.54%  kworker/0:  [kernel.kallsyms]   
       0.20%  systemd-jo  [unknown]           
       0.19%  gnome-shel  libclutter-1.0.so.0.
       0.18%  firefox     libxul.so (deleted) 
       0.17%  gnome-shel  libcogl.so.9.1.1.#pr
       0.14%  firefox     [kernel.kallsyms]   
       0.14%  gnome-shel  libgobject-2.0.so.0.
       0.13%  gnome-shel  libpthread-2.15.so  
       0.11%  synergys    [kernel.kallsyms]   
       0.10%  perf        [kernel.kallsyms]   

You can also get this from 'perf/width-v1' branch on my tree

git://git.kernel.org/pub/scm/linux/kernel/git/namhyung/linux-perf.git

Thanks,
Namhyung

Namhyung Kim (6):
perf tools: Left-align output contents
perf tools: Make __hpp__fmt() receive an additional len argument perf tools: Save column length in perf_hpp_fmt
perf report: Honor column width setting
perf top: Add -w option for setting column width
perf tools: Add name field into perf_hpp_fmt

 tools/perf/Documentation/perf-report.txt |   2 +-
 tools/perf/Documentation/perf-top.txt    |   6 +
 tools/perf/builtin-top.c                 |   3 +
 tools/perf/ui/browsers/hists.c           |  31 ++--
 tools/perf/ui/gtk/hists.c                |  18 +-
 tools/perf/ui/hist.c                     | 275 +++++++++++++++++++------------
 tools/perf/ui/stdio/hist.c               |   4 +-
 tools/perf/util/color.c                  |  16 ++
 tools/perf/util/color.h                  |   1 +
 tools/perf/util/hist.h                   |  17 +-
 tools/perf/util/sort.c                   |  51 +++---
 11 files changed, 274 insertions(+), 150 deletions(-)
Signature · 1 line
2.0.0
author maintainer
#1 lore raw

PATCH 1/6 perf tools: Left-align output contents

Now perf left-aligns column headers but the contents does not. It should have same alignment.

---
 tools/perf/util/sort.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
tools/perf/util/sort.c +3 -3
index 14e5a039bc45..9a531743c8c2 100644
@@ -70,7 +70,7 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
 				       size_t size, unsigned int width)
 {
 	const char *comm = thread__comm_str(he->thread);
-	return repsep_snprintf(bf, size, "%*s:%5d", width - 6,
+	return repsep_snprintf(bf, size, "%-*s:%5d", width - 6,
 			       comm ?: "", he->thread->tid);
 }
 
@@ -106,7 +106,7 @@ sort__comm_sort(struct hist_entry *left, struct hist_entry *right)
 static int hist_entry__comm_snprintf(struct hist_entry *he, char *bf,
 				     size_t size, unsigned int width)
 {
-	return repsep_snprintf(bf, size, "%*s", width, comm__str(he->comm));
+	return repsep_snprintf(bf, size, "%-*s", width, comm__str(he->comm));
 }
 
 struct sort_entry sort_comm = {
@@ -305,7 +305,7 @@ static int hist_entry__srcline_snprintf(struct hist_entry *he, char *bf,
 					size_t size,
 					unsigned int width __maybe_unused)
 {
-	return repsep_snprintf(bf, size, "%s", he->srcline);
+	return repsep_snprintf(bf, size, "%-s", he->srcline);
 }
 
 struct sort_entry sort_srcline = {
Signature · 1 line
2.0.0
author maintainer
#2 lore raw

PATCH 2/6 perf tools: Make __hpp__fmt() receive an additional len argument

In reply to Namhyung Kim

So that it can properly handle alignment requirements later. To do that, add percent_color_len_snprintf() fucntion to help coloring of overhead columns.

---
 tools/perf/ui/browsers/hists.c | 14 ++++++++------
 tools/perf/ui/gtk/hists.c      |  8 +++++---
 tools/perf/ui/hist.c           | 43 +++++++++++++++++++++---------------------
 tools/perf/util/color.c        | 16 ++++++++++++++++
 tools/perf/util/color.h        |  1 +
 tools/perf/util/hist.h         |  4 ++--
 6 files changed, 54 insertions(+), 32 deletions(-)
tools/perf/ui/browsers/hists.c +8 -6
index a94b11fc5e00..02507ba944e3 100644
@@ -653,17 +653,18 @@ struct hpp_arg {
 static int __hpp__slsmg_color_printf(struct perf_hpp *hpp, const char *fmt, ...)
 {
 	struct hpp_arg *arg = hpp->ptr;
-	int ret;
+	int ret, len;
 	va_list args;
 	double percent;
 
 	va_start(args, fmt);
+	len = va_arg(args, int);
 	percent = va_arg(args, double);
 	va_end(args);
 
 	ui_browser__set_percent_color(arg->b, percent, arg->current_entry);
 
-	ret = scnprintf(hpp->buf, hpp->size, fmt, percent);
+	ret = scnprintf(hpp->buf, hpp->size, fmt, len, percent);
 	slsmg_printf("%s", hpp->buf);
 
 	advance_hpp(hpp, ret);
@@ -681,7 +682,7 @@ hist_browser__hpp_color_##_type(struct perf_hpp_fmt *fmt __maybe_unused,\
 				struct perf_hpp *hpp,			\
 				struct hist_entry *he)			\
 {									\
-	return __hpp__fmt(hpp, he, __hpp_get_##_field, " %6.2f%%",	\
+	return __hpp__fmt(hpp, he, __hpp_get_##_field, " %*.2f%%", 6,	\
 			  __hpp__slsmg_color_printf, true);		\
 }
 
@@ -697,13 +698,14 @@ hist_browser__hpp_color_##_type(struct perf_hpp_fmt *fmt __maybe_unused,\
 				struct hist_entry *he)			\
 {									\
 	if (!symbol_conf.cumulate_callchain) {				\
-		int ret = scnprintf(hpp->buf, hpp->size, "%8s", "N/A");	\
+		int ret = scnprintf(hpp->buf, hpp->size,		\
+				    "%*s", 8, "N/A");			\
 		slsmg_printf("%s", hpp->buf);				\
 									\
 		return ret;						\
 	}								\
-	return __hpp__fmt(hpp, he, __hpp_get_acc_##_field, " %6.2f%%",	\
-			  __hpp__slsmg_color_printf, true);		\
+	return __hpp__fmt(hpp, he, __hpp_get_acc_##_field, " %*.2f%%",	\
+			  6, __hpp__slsmg_color_printf, true);	\
 }
 
 __HPP_COLOR_PERCENT_FN(overhead, period)
tools/perf/ui/gtk/hists.c +5 -3
index 6ca60e482cdc..91f6cd7d2312 100644
@@ -11,6 +11,7 @@
 static int __percent_color_snprintf(struct perf_hpp *hpp, const char *fmt, ...)
 {
 	int ret = 0;
+	int len;
 	va_list args;
 	double percent;
 	const char *markup;
@@ -18,6 +19,7 @@ static int __percent_color_snprintf(struct perf_hpp *hpp, const char *fmt, ...)
 	size_t size = hpp->size;
 
 	va_start(args, fmt);
+	len = va_arg(args, int);
 	percent = va_arg(args, double);
 	va_end(args);
 
@@ -25,7 +27,7 @@ static int __percent_color_snprintf(struct perf_hpp *hpp, const char *fmt, ...)
 	if (markup)
 		ret += scnprintf(buf, size, markup);
 
-	ret += scnprintf(buf + ret, size - ret, fmt, percent);
+	ret += scnprintf(buf + ret, size - ret, fmt, len, percent);
 
 	if (markup)
 		ret += scnprintf(buf + ret, size - ret, "</span>");
@@ -43,7 +45,7 @@ static int perf_gtk__hpp_color_##_type(struct perf_hpp_fmt *fmt __maybe_unused,
 				       struct perf_hpp *hpp,			\
 				       struct hist_entry *he)			\
 {										\
-	return __hpp__fmt(hpp, he, he_get_##_field, " %6.2f%%",			\
+	return __hpp__fmt(hpp, he, he_get_##_field, " %*.2f%%", 6,		\
 			  __percent_color_snprintf, true);			\
 }
 
@@ -57,7 +59,7 @@ static int perf_gtk__hpp_color_##_type(struct perf_hpp_fmt *fmt __maybe_unused,
 				       struct perf_hpp *hpp,			\
 				       struct hist_entry *he)			\
 {										\
-	return __hpp__fmt_acc(hpp, he, he_get_acc_##_field, " %6.2f%%",		\
+	return __hpp__fmt_acc(hpp, he, he_get_acc_##_field, " %*.2f%%", 6, 	\
 			      __percent_color_snprintf, true);			\
 }
 
tools/perf/ui/hist.c +22 -21

130 lines not on this page · raw message

tools/perf/util/color.c +16 -0
index 87b8672eb413..f4654183d391 100644
@@ -335,3 +335,19 @@ int percent_color_snprintf(char *bf, size_t size, const char *fmt, ...)
 	va_end(args);
 	return value_color_snprintf(bf, size, fmt, percent);
 }
+
+int percent_color_len_snprintf(char *bf, size_t size, const char *fmt, ...)
+{
+	va_list args;
+	int len;
+	double percent;
+	const char *color;
+
+	va_start(args, fmt);
+	len = va_arg(args, int);
+	percent = va_arg(args, double);
+	va_end(args);
+
+	color = get_percent_color(percent);
+	return color_snprintf(bf, size, color, fmt, len, percent);
+}
tools/perf/util/color.h +1 -0
index 7ff30a62a132..0a594b8a0c26 100644
@@ -41,6 +41,7 @@ int color_fprintf_ln(FILE *fp, const char *color, const char *fmt, ...);
 int color_fwrite_lines(FILE *fp, const char *color, size_t count, const char *buf);
 int value_color_snprintf(char *bf, size_t size, const char *fmt, double value);
 int percent_color_snprintf(char *bf, size_t size, const char *fmt, ...);
+int percent_color_len_snprintf(char *bf, size_t size, const char *fmt, ...);
 int percent_color_fprintf(FILE *fp, const char *fmt, double percent);
 const char *get_percent_color(double percent);
 
tools/perf/util/hist.h +2 -2
index 742f49a85725..13d074db9ef1 100644
@@ -267,10 +267,10 @@ typedef int (*hpp_callback_fn)(struct perf_hpp *hpp, bool front);
 typedef int (*hpp_snprint_fn)(struct perf_hpp *hpp, const char *fmt, ...);
 
 int __hpp__fmt(struct perf_hpp *hpp, struct hist_entry *he,
-	       hpp_field_fn get_field, const char *fmt,
+	       hpp_field_fn get_field, const char *fmt, int len,
 	       hpp_snprint_fn print_fn, bool fmt_percent);
 int __hpp__fmt_acc(struct perf_hpp *hpp, struct hist_entry *he,
-		   hpp_field_fn get_field, const char *fmt,
+		   hpp_field_fn get_field, const char *fmt, int len,
 		   hpp_snprint_fn print_fn, bool fmt_percent);
 
 static inline void advance_hpp(struct perf_hpp *hpp, int inc)
Signature · 1 line
2.0.0
author maintainer
#3 lore raw

PATCH 3/6 perf tools: Save column length in perf_hpp_fmt

In reply to Namhyung Kim

Save column length in the hpp format and pass it to print functions. This is a preparation for users to control column width in the output.

---
 tools/perf/ui/browsers/hists.c |   2 +-
 tools/perf/ui/hist.c           | 135 +++++++++++++++++++++++++++--------------
 tools/perf/util/hist.h         |   2 +
 tools/perf/util/sort.c         |   3 +-
 4 files changed, 94 insertions(+), 48 deletions(-)
tools/perf/ui/browsers/hists.c +1 -1
index 02507ba944e3..c1d8d3925fe1 100644
@@ -849,7 +849,7 @@ static int hists__scnprintf_headers(char *buf, size_t size, struct hists *hists)
 		if (perf_hpp__should_skip(fmt))
 			continue;
 
-		/* We need to add the length of the columns header. */
+		/* We need to ensure length of the columns header. */
 		perf_hpp__reset_width(fmt, hists);
 
 		ret = fmt->header(fmt, &dummy_hpp, hists_to_evsel(hists));
tools/perf/ui/hist.c +89 -46

226 lines not on this page · raw message

tools/perf/util/hist.h +2 -0
index 13d074db9ef1..a7ae890f4c71 100644
@@ -207,6 +207,7 @@ struct perf_hpp_fmt {
 	struct list_head list;
 	struct list_head sort_list;
 	bool elide;
+	int len;
 };
 
 extern struct list_head perf_hpp__list;
@@ -261,6 +262,7 @@ static inline bool perf_hpp__should_skip(struct perf_hpp_fmt *format)
 }
 
 void perf_hpp__reset_width(struct perf_hpp_fmt *fmt, struct hists *hists);
+void perf_hpp__reset_sort_width(struct perf_hpp_fmt *fmt, struct hists *hists);
 
 typedef u64 (*hpp_field_fn)(struct hist_entry *he);
 typedef int (*hpp_callback_fn)(struct perf_hpp *hpp, bool front);
tools/perf/util/sort.c +2 -1
index 9a531743c8c2..396b0094943d 100644
@@ -1194,7 +1194,7 @@ bool perf_hpp__same_sort_entry(struct perf_hpp_fmt *a, struct perf_hpp_fmt *b)
 	return hse_a->se == hse_b->se;
 }
 
-void perf_hpp__reset_width(struct perf_hpp_fmt *fmt, struct hists *hists)
+void perf_hpp__reset_sort_width(struct perf_hpp_fmt *fmt, struct hists *hists)
 {
 	struct hpp_sort_entry *hse;
 
@@ -1265,6 +1265,7 @@ __sort_dimension__alloc_hpp(struct sort_dimension *sd)
 	INIT_LIST_HEAD(&hse->hpp.list);
 	INIT_LIST_HEAD(&hse->hpp.sort_list);
 	hse->hpp.elide = false;
+	hse->hpp.len = 0;
 
 	return hse;
 }
Signature · 1 line
2.0.0
#4 lore raw

Re: [PATCH 3/6] perf tools: Save column length in perf_hpp_fmt

Quoted 2 lines · Namhyung Kim

Save column length in the hpp format and pass it to print functions. This is a preparation for users to control column width in the output.

SNIP

Quoted 18 lines
tools/perf/ui/hist.c +18 -0
+
+void perf_hpp__reset_width(struct perf_hpp_fmt *fmt, struct hists *hists)
+{
+	int idx;
+
+	if (perf_hpp__is_sort_entry(fmt))
+		return perf_hpp__reset_sort_width(fmt, hists);
+
+	for (idx = 0; idx < PERF_HPP__MAX_INDEX; idx++) {
+		if (fmt == &perf_hpp__format[idx])
+			break;
+	}
+
+	if (idx == PERF_HPP__MAX_INDEX)
+		return;
+
+	switch (idx) {
+	case PERF_HPP__OVERHEAD:

tools/perf/ui/hist.c unsigned int hists__sort_list_width(struct hists *hists) · line 658

Quoted 10 lines
tools/perf/ui/hist.c +10 -0
+	case PERF_HPP__OVERHEAD_SYS:
+	case PERF_HPP__OVERHEAD_US:
+	case PERF_HPP__OVERHEAD_ACC:
+		fmt->len = 8;
+		break;
+
+	case PERF_HPP__OVERHEAD_GUEST_SYS:
+	case PERF_HPP__OVERHEAD_GUEST_US:
+		fmt->len = 9;
+		break;

just curious.. why is it 9 for guest %, while we use 8 for host? I understand that was the current state

jirka

author maintainer
#5 lore raw

Re: [PATCH 3/6] perf tools: Save column length in perf_hpp_fmt

Quoted 29 lines · Jiri Olsa
Quoted 2 lines · Namhyung Kim

Save column length in the hpp format and pass it to print functions. This is a preparation for users to control column width in the output.

SNIP

Quoted 23 lines
tools/perf/ui/hist.c +23 -0
+
+void perf_hpp__reset_width(struct perf_hpp_fmt *fmt, struct hists *hists)
+{
+     int idx;
+
+     if (perf_hpp__is_sort_entry(fmt))
+             return perf_hpp__reset_sort_width(fmt, hists);
+
+     for (idx = 0; idx < PERF_HPP__MAX_INDEX; idx++) {
+             if (fmt == &perf_hpp__format[idx])
+                     break;
+     }
+
+     if (idx == PERF_HPP__MAX_INDEX)
+             return;
+
+     switch (idx) {
+     case PERF_HPP__OVERHEAD:
+     case PERF_HPP__OVERHEAD_SYS:
+     case PERF_HPP__OVERHEAD_US:
+     case PERF_HPP__OVERHEAD_ACC:
+             fmt->len = 8;
+             break;
Quoted 8 lines

tools/perf/ui/hist.c unsigned int hists__sort_list_width(struct hists *hists) · line 658

Quoted 5 lines
tools/perf/ui/hist.c +5 -0
+
+     case PERF_HPP__OVERHEAD_GUEST_SYS:
+     case PERF_HPP__OVERHEAD_GUEST_US:
+             fmt->len = 9;
+             break;

just curious.. why is it 9 for guest %, while we use 8 for host? I understand that was the current state

It's the strlen of header - "guest sys". :)

Thanks,
Namhyung

#6 lore raw

Re: [PATCH 3/6] perf tools: Save column length in perf_hpp_fmt

Quoted 39 lines · Namhyung Kim
Quoted 29 lines · Jiri Olsa
Quoted 2 lines · Namhyung Kim

Save column length in the hpp format and pass it to print functions. This is a preparation for users to control column width in the output.

SNIP

Quoted 23 lines
tools/perf/ui/hist.c +23 -0
+
+void perf_hpp__reset_width(struct perf_hpp_fmt *fmt, struct hists *hists)
+{
+     int idx;
+
+     if (perf_hpp__is_sort_entry(fmt))
+             return perf_hpp__reset_sort_width(fmt, hists);
+
+     for (idx = 0; idx < PERF_HPP__MAX_INDEX; idx++) {
+             if (fmt == &perf_hpp__format[idx])
+                     break;
+     }
+
+     if (idx == PERF_HPP__MAX_INDEX)
+             return;
+
+     switch (idx) {
+     case PERF_HPP__OVERHEAD:
+     case PERF_HPP__OVERHEAD_SYS:
+     case PERF_HPP__OVERHEAD_US:
+     case PERF_HPP__OVERHEAD_ACC:
+             fmt->len = 8;
+             break;
Quoted 8 lines

tools/perf/ui/hist.c unsigned int hists__sort_list_width(struct hists *hists) · line 658

Quoted 5 lines
tools/perf/ui/hist.c +5 -0
+
+     case PERF_HPP__OVERHEAD_GUEST_SYS:
+     case PERF_HPP__OVERHEAD_GUEST_US:
+             fmt->len = 9;
+             break;

just curious.. why is it 9 for guest %, while we use 8 for host? I understand that was the current state

Quoted 1 line

It's the strlen of header - "guest sys". :)

argh.. I did not think of header, just value ;-)

thanks,
jirka

author maintainer
#7 lore raw

PATCH 4/6 perf report: Honor column width setting

In reply to Namhyung Kim

Set column width and do not change it if user gives -w/--column-widths option. It'll truncate longer symbols than the width if exists.

---
 tools/perf/ui/browsers/hists.c | 23 ++++++++----
 tools/perf/ui/gtk/hists.c      | 10 ++---
 tools/perf/ui/hist.c           | 84 +++++++++++++++++++++++++++++-------------
 tools/perf/ui/stdio/hist.c     |  4 +-
 tools/perf/util/hist.h         | 14 ++++---
 tools/perf/util/sort.c         | 44 +++++++++++++---------
 6 files changed, 116 insertions(+), 63 deletions(-)
tools/perf/ui/browsers/hists.c +15 -8
index c1d8d3925fe1..f0ef031558dd 100644
@@ -678,12 +678,12 @@ static u64 __hpp_get_##_field(struct hist_entry *he)			\
 }									\
 									\
 static int								\
-hist_browser__hpp_color_##_type(struct perf_hpp_fmt *fmt __maybe_unused,\
+hist_browser__hpp_color_##_type(struct perf_hpp_fmt *fmt,		\
 				struct perf_hpp *hpp,			\
 				struct hist_entry *he)			\
 {									\
-	return __hpp__fmt(hpp, he, __hpp_get_##_field, " %*.2f%%", 6,	\
-			  __hpp__slsmg_color_printf, true);		\
+	return hpp__fmt(fmt, hpp, he, __hpp_get_##_field, " %*.2f%%",	\
+			__hpp__slsmg_color_printf, true);		\
 }
 
 #define __HPP_COLOR_ACC_PERCENT_FN(_type, _field)			\
@@ -693,19 +693,20 @@ static u64 __hpp_get_acc_##_field(struct hist_entry *he)		\
 }									\
 									\
 static int								\
-hist_browser__hpp_color_##_type(struct perf_hpp_fmt *fmt __maybe_unused,\
+hist_browser__hpp_color_##_type(struct perf_hpp_fmt *fmt,		\
 				struct perf_hpp *hpp,			\
 				struct hist_entry *he)			\
 {									\
 	if (!symbol_conf.cumulate_callchain) {				\
+		int len = fmt->user_len ?: fmt->len;			\
 		int ret = scnprintf(hpp->buf, hpp->size,		\
-				    "%*s", 8, "N/A");			\
+				    "%*s", len, "N/A");			\
 		slsmg_printf("%s", hpp->buf);				\
 									\
 		return ret;						\
 	}								\
-	return __hpp__fmt(hpp, he, __hpp_get_acc_##_field, " %*.2f%%",	\
-			  6, __hpp__slsmg_color_printf, true);	\
+	return hpp__fmt(fmt, hpp, he, __hpp_get_acc_##_field,		\
+			" %*.2f%%", __hpp__slsmg_color_printf, true);	\
 }
 
 __HPP_COLOR_PERCENT_FN(overhead, period)
@@ -797,8 +798,11 @@ static int hist_browser__show_entry(struct hist_browser *browser,
 			if (fmt->color) {
 				width -= fmt->color(fmt, &hpp, entry);
 			} else {
-				width -= fmt->entry(fmt, &hpp, entry);
+				int ret = fmt->entry(fmt, &hpp, entry);
+				s[ret] = '\0';
 				slsmg_printf("%s", s);
+
+				width -= ret;
 			}
 		}
 
@@ -1549,6 +1553,9 @@ static int perf_evsel__hists_browse(struct perf_evsel *evsel, int nr_events,
 
 	memset(options, 0, sizeof(options));
 
+	if (symbol_conf.col_width_list_str)
+		perf_hpp__set_user_width(symbol_conf.col_width_list_str);
+
 	while (1) {
 		const struct thread *thread = NULL;
 		const struct dso *dso = NULL;
tools/perf/ui/gtk/hists.c +5 -5
index 91f6cd7d2312..897b2e140428 100644
@@ -41,12 +41,12 @@ static u64 he_get_##_field(struct hist_entry *he)				\
 	return he->stat._field;							\
 }										\
 										\
-static int perf_gtk__hpp_color_##_type(struct perf_hpp_fmt *fmt __maybe_unused,	\
+static int perf_gtk__hpp_color_##_type(struct perf_hpp_fmt *fmt,		\
 				       struct perf_hpp *hpp,			\
 				       struct hist_entry *he)			\
 {										\
-	return __hpp__fmt(hpp, he, he_get_##_field, " %*.2f%%", 6,		\
-			  __percent_color_snprintf, true);			\
+	return hpp__fmt(fmt, hpp, he, he_get_##_field, " %*.2f%%",		\
+			__percent_color_snprintf, true);			\
 }
 
 #define __HPP_COLOR_ACC_PERCENT_FN(_type, _field)				\
@@ -59,8 +59,8 @@ static int perf_gtk__hpp_color_##_type(struct perf_hpp_fmt *fmt __maybe_unused,
 				       struct perf_hpp *hpp,			\
 				       struct hist_entry *he)			\
 {										\
-	return __hpp__fmt_acc(hpp, he, he_get_acc_##_field, " %*.2f%%", 6, 	\
-			      __percent_color_snprintf, true);			\
+	return hpp__fmt_acc(fmt, hpp, he, he_get_acc_##_field, " %*.2f%%", 	\
+			    __percent_color_snprintf, true);			\
 }
 
 __HPP_COLOR_PERCENT_FN(overhead, period)
tools/perf/ui/hist.c +58 -26

141 lines not on this page · raw message

tools/perf/ui/stdio/hist.c +3 -1
index 90122abd3721..59da427c11bb 100644
@@ -395,10 +395,12 @@ size_t hists__fprintf(struct hists *hists, bool show_header, int max_rows,
 
 	init_rem_hits();
 
-
 	perf_hpp__for_each_format(fmt)
 		perf_hpp__reset_width(fmt, hists);
 
+	if (symbol_conf.col_width_list_str)
+		perf_hpp__set_user_width(symbol_conf.col_width_list_str);
+
 	if (!show_header)
 		goto print_entries;
 
tools/perf/util/hist.h +8 -6
index a7ae890f4c71..2e70003fcd00 100644
@@ -208,6 +208,7 @@ struct perf_hpp_fmt {
 	struct list_head sort_list;
 	bool elide;
 	int len;
+	int user_len;
 };
 
 extern struct list_head perf_hpp__list;
@@ -263,17 +264,18 @@ static inline bool perf_hpp__should_skip(struct perf_hpp_fmt *format)
 
 void perf_hpp__reset_width(struct perf_hpp_fmt *fmt, struct hists *hists);
 void perf_hpp__reset_sort_width(struct perf_hpp_fmt *fmt, struct hists *hists);
+void perf_hpp__set_user_width(const char *width_list_str);
 
 typedef u64 (*hpp_field_fn)(struct hist_entry *he);
 typedef int (*hpp_callback_fn)(struct perf_hpp *hpp, bool front);
 typedef int (*hpp_snprint_fn)(struct perf_hpp *hpp, const char *fmt, ...);
 
-int __hpp__fmt(struct perf_hpp *hpp, struct hist_entry *he,
-	       hpp_field_fn get_field, const char *fmt, int len,
-	       hpp_snprint_fn print_fn, bool fmt_percent);
-int __hpp__fmt_acc(struct perf_hpp *hpp, struct hist_entry *he,
-		   hpp_field_fn get_field, const char *fmt, int len,
-		   hpp_snprint_fn print_fn, bool fmt_percent);
+int hpp__fmt(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
+	     struct hist_entry *he, hpp_field_fn get_field,
+	     const char *fmtstr, hpp_snprint_fn print_fn, bool fmt_percent);
+int hpp__fmt_acc(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
+		 struct hist_entry *he, hpp_field_fn get_field,
+		 const char *fmtstr, hpp_snprint_fn print_fn, bool fmt_percent);
 
 static inline void advance_hpp(struct perf_hpp *hpp, int inc)
 {
tools/perf/util/sort.c +27 -17

127 lines not on this page · raw message

Signature · 1 line
2.0.0
#8 lore raw

Re: [PATCH 4/6] perf report: Honor column width setting

Quoted 2 lines · Namhyung Kim

Set column width and do not change it if user gives -w/--column-widths option. It'll truncate longer symbols than the width if exists.

SNIP

Quoted 8 lines

static inline void advance_hpp(struct perf_hpp *hpp, int inc) {

tools/perf/util/sort.c +0 -0
index 396b0094943d..f89714329c0f 100644

tools/perf/util/sort.c static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf, · line 76

Quoted 8 lines
tools/perf/util/sort.c +3 -1
@@ -70,7 +70,9 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
 				       size_t size, unsigned int width)
 {
 	const char *comm = thread__comm_str(he->thread);
-	return repsep_snprintf(bf, size, "%-*s:%5d", width - 6,
+
+	width = max(7U, width) - 6;
+	return repsep_snprintf(bf, size, "%-*.*s:%5d", width, width,
 			       comm ?: "", he->thread->tid);

I couldn't found what your format string:
"%-*.*s:%5d", width, width

is supposed to do... it gives me the same result as for:
"%-*s:%5d", width

you use it on more places so I guess it's not a typo ;-)

thanks,
jirka

author maintainer
#9 lore raw

Re: [PATCH 4/6] perf report: Honor column width setting

Quoted 26 lines · Jiri Olsa
Quoted 2 lines · Namhyung Kim

Set column width and do not change it if user gives -w/--column-widths option. It'll truncate longer symbols than the width if exists.

SNIP

Quoted 8 lines

static inline void advance_hpp(struct perf_hpp *hpp, int inc) {

tools/perf/util/sort.c +0 -0
index 396b0094943d..f89714329c0f 100644

tools/perf/util/sort.c static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf, · line 76

Quoted 8 lines
tools/perf/util/sort.c +3 -1
@@ -70,7 +70,9 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
                                     size_t size, unsigned int width)
 {
      const char *comm = thread__comm_str(he->thread);
-     return repsep_snprintf(bf, size, "%-*s:%5d", width - 6,
+
+     width = max(7U, width) - 6;
+     return repsep_snprintf(bf, size, "%-*.*s:%5d", width, width,
                             comm ?: "", he->thread->tid);

I couldn't found what your format string:
"%-*.*s:%5d", width, width

Quoted 4 lines

is supposed to do... it gives me the same result as for:
"%-*s:%5d", width

you use it on more places so I guess it's not a typo ;-)

Yep, it's NOT a typo. The second * is necessary for limiting maximum length of printed string - without it we cannot control column alignment of long symbols.

Thanks,
Namhyung

#10 lore raw

Re: [PATCH 4/6] perf report: Honor column width setting

In reply to Namhyung Kim
Quoted 2 lines · Namhyung Kim

Set column width and do not change it if user gives -w/--column-widths option. It'll truncate longer symbols than the width if exists.

SNIP

tools/perf/util/sort.c static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf, · line 77

Quoted 12 lines
tools/perf/util/sort.c +3 -1
@@ -70,7 +70,9 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
 				       size_t size, unsigned int width)
 {
 	const char *comm = thread__comm_str(he->thread);
-	return repsep_snprintf(bf, size, "%-*s:%5d", width - 6,
+
+	width = max(7U, width) - 6;
+	return repsep_snprintf(bf, size, "%-*.*s:%5d", width, width,
 			       comm ?: "", he->thread->tid);
 }

also we now print out all sort headers left alligned, which is wrong for sort_thread header: "Command: Pid' that mirrors value strings with ':' being aligned

maybe something like in attached patch..?

thanks,
jirka

---

tools/perf/util/sort.c +6 -1
index f89714329c0f..fbbe86b7f63c 100644
@@ -78,6 +78,7 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
 
 struct sort_entry sort_thread = {
 	.se_header	= "Command:  Pid",
+	.se_header_right= true,
 	.se_cmp		= sort__thread_cmp,
 	.se_snprintf	= hist_entry__thread_snprintf,
 	.se_width_idx	= HISTC_THREAD,
@@ -1212,13 +1213,17 @@ static int __sort__hpp_header(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
 {
 	struct hpp_sort_entry *hse;
 	size_t len = fmt->user_len;
+	const char *fmt_str = "%-*.*s";
 
 	hse = container_of(fmt, struct hpp_sort_entry, hpp);
 
 	if (!len)
 		len = hists__col_len(&evsel->hists, hse->se->se_width_idx);
 
-	return scnprintf(hpp->buf, hpp->size, "%-*.*s", len, len, hse->se->se_header);
+	if (hse->se->se_header_right)
+		fmt_str = "%*.*s";
+
+	return scnprintf(hpp->buf, hpp->size, fmt_str, len, len, hse->se->se_header);
 }
 
 static int __sort__hpp_width(struct perf_hpp_fmt *fmt,
tools/perf/util/sort.h +1 -0
index 041f0c9cea2b..16a07b9b16c2 100644
@@ -197,6 +197,7 @@ struct sort_entry {
 	struct list_head list;
 
 	const char *se_header;
+	bool se_header_right;
 
 	int64_t (*se_cmp)(struct hist_entry *, struct hist_entry *);
 	int64_t (*se_collapse)(struct hist_entry *, struct hist_entry *);
#11 lore raw

Re: [PATCH 4/6] perf report: Honor column width setting

Em Thu, Jul 24, 2014 at 02:57:51PM +0200, Jiri Olsa escreveu:

Quoted 35 lines · Namhyung Kim
Quoted 2 lines

Set column width and do not change it if user gives -w/--column-widths option. It'll truncate longer symbols than the width if exists.

SNIP

tools/perf/util/sort.c static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf, · line 77

Quoted 12 lines
tools/perf/util/sort.c +3 -1
@@ -70,7 +70,9 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
 				       size_t size, unsigned int width)
 {
 	const char *comm = thread__comm_str(he->thread);
-	return repsep_snprintf(bf, size, "%-*s:%5d", width - 6,
+
+	width = max(7U, width) - 6;
+	return repsep_snprintf(bf, size, "%-*.*s:%5d", width, width,
 			       comm ?: "", he->thread->tid);
 }

also we now print out all sort headers left alligned, which is wrong for sort_thread header: "Command: Pid' that mirrors value strings with ':' being aligned

maybe something like in attached patch..?

thanks,
jirka

---

tools/perf/util/sort.c +0 -0
index f89714329c0f..fbbe86b7f63c 100644

tools/perf/util/sort.c static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf, · line 81

Quoted 4 lines
tools/perf/util/sort.c +1 -0
@@ -78,6 +78,7 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
 
 struct sort_entry sort_thread = {
 	.se_header	= "Command:  Pid",
+	.se_header_right= true,

Is this just for the header? Perhaps its more clearly named .se_right_justify?

Quoted 34 lines
tools/perf/util/sort.c +5 -1
 	.se_cmp		= sort__thread_cmp,
 	.se_snprintf	= hist_entry__thread_snprintf,
 	.se_width_idx	= HISTC_THREAD,
@@ -1212,13 +1213,17 @@ static int __sort__hpp_header(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
 {
 	struct hpp_sort_entry *hse;
 	size_t len = fmt->user_len;
+	const char *fmt_str = "%-*.*s";
 
 	hse = container_of(fmt, struct hpp_sort_entry, hpp);
 
 	if (!len)
 		len = hists__col_len(&evsel->hists, hse->se->se_width_idx);
 
-	return scnprintf(hpp->buf, hpp->size, "%-*.*s", len, len, hse->se->se_header);
+	if (hse->se->se_header_right)
+		fmt_str = "%*.*s";
+
+	return scnprintf(hpp->buf, hpp->size, fmt_str, len, len, hse->se->se_header);
 }
 
 static int __sort__hpp_width(struct perf_hpp_fmt *fmt,
tools/perf/util/sort.h +1 -0
index 041f0c9cea2b..16a07b9b16c2 100644
@@ -197,6 +197,7 @@ struct sort_entry {
 	struct list_head list;
 
 	const char *se_header;
+	bool se_header_right;
 
 	int64_t (*se_cmp)(struct hist_entry *, struct hist_entry *);
 	int64_t (*se_collapse)(struct hist_entry *, struct hist_entry *);
#12 lore raw

Re: [PATCH 4/6] perf report: Honor column width setting

Quoted 37 lines · Arnaldo Carvalho de Melo

Em Thu, Jul 24, 2014 at 02:57:51PM +0200, Jiri Olsa escreveu:

Quoted 36 lines · Namhyung Kim
Quoted 2 lines

Set column width and do not change it if user gives -w/--column-widths option. It'll truncate longer symbols than the width if exists.

SNIP

tools/perf/util/sort.c static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf, · line 77

Quoted 12 lines
tools/perf/util/sort.c +3 -1
@@ -70,7 +70,9 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
 				       size_t size, unsigned int width)
 {
 	const char *comm = thread__comm_str(he->thread);
-	return repsep_snprintf(bf, size, "%-*s:%5d", width - 6,
+
+	width = max(7U, width) - 6;
+	return repsep_snprintf(bf, size, "%-*.*s:%5d", width, width,
 			       comm ?: "", he->thread->tid);
 }

also we now print out all sort headers left alligned, which is wrong for sort_thread header: "Command: Pid' that mirrors value strings with ':' being aligned

maybe something like in attached patch..?

thanks,
jirka

---

tools/perf/util/sort.c +0 -0
index f89714329c0f..fbbe86b7f63c 100644
@@ -78,6 +78,7 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
 
Quoted 5 lines

tools/perf/util/sort.c static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf, · line 81

Quoted 3 lines
tools/perf/util/sort.c +1 -0
 struct sort_entry sort_thread = {
 	.se_header	= "Command:  Pid",
+	.se_header_right= true,

Is this just for the header? Perhaps its more clearly named .se_right_justify?

yep, it's just for header.. let's see Namhyung might have better idea how to handle this problem

jirka

author maintainer
#13 lore raw

Re: [PATCH 4/6] perf report: Honor column width setting

Hi guys,

Quoted 38 lines · Jiri Olsa
Quoted 37 lines · Arnaldo Carvalho de Melo

Em Thu, Jul 24, 2014 at 02:57:51PM +0200, Jiri Olsa escreveu:

Quoted 36 lines · Namhyung Kim
Quoted 2 lines

Set column width and do not change it if user gives -w/--column-widths option. It'll truncate longer symbols than the width if exists.

SNIP

tools/perf/util/sort.c static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf, · line 77

Quoted 12 lines
tools/perf/util/sort.c +3 -1
@@ -70,7 +70,9 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
                                 size_t size, unsigned int width)
 {
  const char *comm = thread__comm_str(he->thread);
- return repsep_snprintf(bf, size, "%-*s:%5d", width - 6,
+
+ width = max(7U, width) - 6;
+ return repsep_snprintf(bf, size, "%-*.*s:%5d", width, width,
                         comm ?: "", he->thread->tid);
 }

also we now print out all sort headers left alligned, which is wrong for sort_thread header: "Command: Pid' that mirrors value strings with ':' being aligned

maybe something like in attached patch..?

thanks,
jirka

---

tools/perf/util/sort.c +0 -0
index f89714329c0f..fbbe86b7f63c 100644
@@ -78,6 +78,7 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
 
Quoted 8 lines
Quoted 5 lines

tools/perf/util/sort.c static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf, · line 81

Quoted 3 lines
tools/perf/util/sort.c +1 -0
 struct sort_entry sort_thread = {
    .se_header      = "Command:  Pid",
+   .se_header_right= true,

Is this just for the header? Perhaps its more clearly named .se_right_justify?

yep, it's just for header.. let's see Namhyung might have better idea how to handle this problem

I think it still needs to be left-aligned. So I'd rather change the order - i.e. "Pid: Command". Is it okay for you?

Thanks,
Namhyung

#14 lore raw

Re: [PATCH 4/6] perf report: Honor column width setting

Quoted 47 lines · Namhyung Kim

Hi guys,

Quoted 44 lines · Jiri Olsa
Quoted 42 lines · Arnaldo Carvalho de Melo

Em Thu, Jul 24, 2014 at 02:57:51PM +0200, Jiri Olsa escreveu:

Quoted 35 lines · Namhyung Kim
Quoted 2 lines

Set column width and do not change it if user gives -w/--column-widths option. It'll truncate longer symbols than the width if exists.

SNIP

tools/perf/util/sort.c static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf, · line 77

Quoted 12 lines
tools/perf/util/sort.c +3 -1
@@ -70,7 +70,9 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
                                 size_t size, unsigned int width)
 {
  const char *comm = thread__comm_str(he->thread);
- return repsep_snprintf(bf, size, "%-*s:%5d", width - 6,
+
+ width = max(7U, width) - 6;
+ return repsep_snprintf(bf, size, "%-*.*s:%5d", width, width,
                         comm ?: "", he->thread->tid);
 }

also we now print out all sort headers left alligned, which is wrong for sort_thread header: "Command: Pid' that mirrors value strings with ':' being aligned

maybe something like in attached patch..?

thanks,
jirka

---

tools/perf/util/sort.c +0 -0
index f89714329c0f..fbbe86b7f63c 100644

tools/perf/util/sort.c static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf, · line 81

Quoted 4 lines
tools/perf/util/sort.c +1 -0
@@ -78,6 +78,7 @@ static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
 
 struct sort_entry sort_thread = {
    .se_header      = "Command:  Pid",
+   .se_header_right= true,

Is this just for the header? Perhaps its more clearly named .se_right_justify?

Quoted 5 lines
Quoted 2 lines

yep, it's just for header.. let's see Namhyung might have better idea how to handle this problem

I think it still needs to be left-aligned. So I'd rather change the order - i.e. "Pid: Command". Is it okay for you?

yes, jirka

#15 lore raw

Re: [PATCH 4/6] perf report: Honor column width setting

Em Thu, Jul 24, 2014 at 05:58:44PM +0200, Jiri Olsa escreveu:

Quoted 8 lines · Namhyung Kim
Quoted 6 lines · Jiri Olsa
Quoted 5 lines · Arnaldo Carvalho de Melo
Quoted 4 lines

Em Thu, Jul 24, 2014 at 02:57:51PM +0200, Jiri Olsa escreveu:

Quoted 3 lines

struct sort_entry sort_thread = {
.se_header = "Command: Pid",
+ .se_header_right= true,

Quoted 6 lines
Quoted 6 lines
Quoted 4 lines
Quoted 1 line

Is this just for the header? Perhaps its more clearly named .se_right_justify?

yep, it's just for header.. let's see Namhyung might have better idea how to handle this problem

I think it still needs to be left-aligned. So I'd rather change the

Good, I like simpler fixes :)

Quoted 3 lines
Quoted 1 line

order - i.e. "Pid: Command". Is it okay for you?

yes, jirka

Yes, its is more usual to have pid+command anyway, just run 'ps' :-)

- Arnaldo

author maintainer
#16 lore raw

PATCH 5/6 perf top: Add -w option for setting column width

In reply to Namhyung Kim

Add -w/--column-widths option like perf report does so that users are able to see symbols even with some very long C++ library/functions. It can be a list separated by comma for each column.

  $ perf top -w 0,20,30

The value of 0 means there's no limit.

---
 tools/perf/Documentation/perf-report.txt | 2 +-
 tools/perf/Documentation/perf-top.txt    | 6 ++++++
 tools/perf/builtin-top.c                 | 3 +++
 3 files changed, 10 insertions(+), 1 deletion(-)
tools/perf/Documentation/perf-report.txt +1 -1
index d2b59af62bc0..d561e0214f52 100644
@@ -147,7 +147,7 @@ OPTIONS
 -w::
 --column-widths=<width[,width...]>::
 	Force each column width to the provided list, for large terminal
-	readability.
+	readability.  0 means no limit (default behavior).
 
 -t::
 --field-separator=::
tools/perf/Documentation/perf-top.txt +6 -0
index 180ae02137a5..28fdee394880 100644
@@ -193,6 +193,12 @@ Default is to monitor all CPUS.
 	sum of shown entries will be always 100%. "absolute" means it retains
 	the original value before and after the filter is applied.
 
+-w::
+--column-widths=<width[,width...]>::
+	Force each column width to the provided list, for large terminal
+	readability.  0 means no limit (default behavior).
+
+
 INTERACTIVE PROMPTING KEYS
 --------------------------
 
tools/perf/builtin-top.c +3 -0
index 377971dc89a3..bde216b2071c 100644
@@ -1131,6 +1131,9 @@ int cmd_top(int argc, const char **argv, const char *prefix __maybe_unused)
 		     "Don't show entries under that percent", parse_percent_limit),
 	OPT_CALLBACK(0, "percentage", NULL, "relative|absolute",
 		     "How to display percentage of filtered entries", parse_filter_percentage),
+	OPT_STRING('w', "column-widths", &symbol_conf.col_width_list_str,
+		   "width[,width...]",
+		   "don't try to adjust column width, use these fixed values"),
 	OPT_END()
 	};
 	const char * const top_usage[] = {
Signature · 1 line
2.0.0
author maintainer
#17 lore raw

PATCH 6/6 perf tools: Add name field into perf_hpp_fmt

In reply to Namhyung Kim

It makes the code a bit simpler and easier to debug IMHO.

I guess it can also remove similar code in perf diff, but let's keep it for a future work. :)

---
 tools/perf/ui/gtk/hists.c |   4 +-
 tools/perf/ui/hist.c      | 109 ++++++++++++++++++++++------------------------
 tools/perf/util/hist.h    |   1 +
 tools/perf/util/sort.c    |   6 +--
 4 files changed, 57 insertions(+), 63 deletions(-)
tools/perf/ui/gtk/hists.c +1 -3
index 897b2e140428..f3fa4258b256 100644
@@ -207,10 +207,8 @@ static void perf_gtk__show_hists(GtkWidget *window, struct hists *hists,
 		if (perf_hpp__is_sort_entry(fmt))
 			sym_col = col_idx;
 
-		fmt->header(fmt, &hpp, hists_to_evsel(hists));
-
 		gtk_tree_view_insert_column_with_attributes(GTK_TREE_VIEW(view),
-							    -1, ltrim(s),
+							    -1, fmt->name,
 							    renderer, "markup",
 							    col_idx++, NULL);
 	}
tools/perf/ui/hist.c +52 -57

182 lines not on this page · raw message

tools/perf/util/hist.h +1 -0
index 2e70003fcd00..95405a8fbd95 100644
@@ -192,6 +192,7 @@ struct perf_hpp {
 };
 
 struct perf_hpp_fmt {
+	const char *name;
 	int (*header)(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
 		      struct perf_evsel *evsel);
 	int (*width)(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
tools/perf/util/sort.c +3 -3
index f89714329c0f..7a1944f8d001 100644
@@ -1203,8 +1203,7 @@ void perf_hpp__reset_sort_width(struct perf_hpp_fmt *fmt, struct hists *hists)
 		return;
 
 	hse = container_of(fmt, struct hpp_sort_entry, hpp);
-	hists__new_col_len(hists, hse->se->se_width_idx,
-			   strlen(hse->se->se_header));
+	hists__new_col_len(hists, hse->se->se_width_idx, strlen(fmt->name));
 }
 
 static int __sort__hpp_header(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
@@ -1218,7 +1217,7 @@ static int __sort__hpp_header(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
 	if (!len)
 		len = hists__col_len(&evsel->hists, hse->se->se_width_idx);
 
-	return scnprintf(hpp->buf, hpp->size, "%-*.*s", len, len, hse->se->se_header);
+	return scnprintf(hpp->buf, hpp->size, "%-*.*s", len, len, fmt->name);
 }
 
 static int __sort__hpp_width(struct perf_hpp_fmt *fmt,
@@ -1262,6 +1261,7 @@ __sort_dimension__alloc_hpp(struct sort_dimension *sd)
 	}
 
 	hse->se = sd->entry;
+	hse->hpp.name = sd->entry->se_header;
 	hse->hpp.header = __sort__hpp_header;
 	hse->hpp.width = __sort__hpp_width;
 	hse->hpp.entry = __sort__hpp_entry;
Signature · 1 line
2.0.0
#18 lore raw
In reply to Namhyung Kim
Quoted 13 lines · Namhyung Kim

Hello,

This patchset is to control perf report/top output column width by -w/--column-widths option so that it can fit into the terminal size. The -w option is there for perf report but it ignored by recent output field changed due to some reason. This patchset fixes it and supports perf top also.

This is sometimes useful if your terminal is small and there's some C++ applications which have amazingly long symbol names. Without this patchset user might not see those symbols on TUI, since it maps left/right arrow keys to other functions.

Quoted 6 lines

The -w option sets column width starting from the first column (overhead or optional overhead_children column unless -F option is given). It doesn't make sense to limit those overhead columns so it's not a hard-limit for them. But it *is* a hard-limit for other columns such as comm, dso, symbol, and so on. One can use 0 not to limit/force a width for those columns.

hi,
I've got broken TUI output for "perf report --group"

Samples: 17 of event 'anon group { cycles, instructions }', Event count (approx.): 9145256
56.44% 31.59% ls libc-2.17.so [.] __strcoll_l
39.94% 0.00% ls ld-2.17.so [.] _dl_new_object

 3.48% 0.00%  ls  [kernel.kallsyms]  [k] setup_arg_pages
 0.14% 0.33%  ls  [kernel.kallsyms]  [k] native_write_msr_safe
 0.00% 30.17%  ls  [kernel.kallsyms]  [k] security_inode_permission
 0.00% 29.78%  ls  ls                 [.] indent

0.00% 8.12% ls [kernel.kallsyms] [k] __slab_alloc

I have 'show-headers' set to false in ~/.perfconfig and the output got fixed after displaying headers by pressing 'H'

jirka

author maintainer
#19 lore raw

Hi Jiri,

Quoted 27 lines · Jiri Olsa
Quoted 13 lines · Namhyung Kim

Hello,

This patchset is to control perf report/top output column width by -w/--column-widths option so that it can fit into the terminal size. The -w option is there for perf report but it ignored by recent output field changed due to some reason. This patchset fixes it and supports perf top also.

This is sometimes useful if your terminal is small and there's some C++ applications which have amazingly long symbol names. Without this patchset user might not see those symbols on TUI, since it maps left/right arrow keys to other functions.

Quoted 6 lines

The -w option sets column width starting from the first column (overhead or optional overhead_children column unless -F option is given). It doesn't make sense to limit those overhead columns so it's not a hard-limit for them. But it *is* a hard-limit for other columns such as comm, dso, symbol, and so on. One can use 0 not to limit/force a width for those columns.

hi,
I've got broken TUI output for "perf report --group"

Samples: 17 of event 'anon group { cycles, instructions }', Event count (approx.): 9145256
56.44% 31.59% ls libc-2.17.so [.] __strcoll_l
39.94% 0.00% ls ld-2.17.so [.] _dl_new_object

Quoted 8 lines
 3.48% 0.00%  ls  [kernel.kallsyms]  [k] setup_arg_pages
 0.14% 0.33%  ls  [kernel.kallsyms]  [k] native_write_msr_safe
 0.00% 30.17%  ls  [kernel.kallsyms]  [k] security_inode_permission
 0.00% 29.78%  ls  ls                 [.] indent

0.00% 8.12% ls [kernel.kallsyms] [k] __slab_alloc

I have 'show-headers' set to false in ~/.perfconfig and the output got fixed after displaying headers by pressing 'H'

Argh, you're right. It should be fixed with this patch:

Thanks,
Namhyung

Quoted 1 line

From a8fffc7802402d5ddd03e02956304199fa20fae9 Mon Sep 17 00:00:00 2001

From: Namhyung Kim <[email protected]>
Date: Wed, 23 Jul 2014 16:20:45 +0900
Subject: [PATCH] perf tools: Fix column alignment when headers aren't shown on
TUI

If user sets ui.show-headers config option to false, it didn't calculate default column width so it broke the alignment. This is because it does the calculation just before showing headers.

Move it to the beginning of the hist browser so that it can be called regardless of the config option.

---
 tools/perf/ui/browsers/hists.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
tools/perf/ui/browsers/hists.c +4 -3
index f0ef031558dd..ab9ca221fc2c 100644
@@ -853,9 +853,6 @@ static int hists__scnprintf_headers(char *buf, size_t size, struct hists *hists)
 		if (perf_hpp__should_skip(fmt))
 			continue;
 
-		/* We need to ensure length of the columns header. */
-		perf_hpp__reset_width(fmt, hists);
-
 		ret = fmt->header(fmt, &dummy_hpp, hists_to_evsel(hists));
 		if (advance_hpp_check(&dummy_hpp, ret))
 			break;
@@ -1504,6 +1501,7 @@ static int perf_evsel__hists_browse(struct perf_evsel *evsel, int nr_events,
 	char buf[64];
 	char script_opt[64];
 	int delay_secs = hbt ? hbt->refresh : 0;
+	struct perf_hpp_fmt *fmt;
 
 #define HIST_BROWSER_HELP_COMMON					\
 	"h/?/F1        Show this window\n"				\
@@ -1553,6 +1551,9 @@ static int perf_evsel__hists_browse(struct perf_evsel *evsel, int nr_events,
 
 	memset(options, 0, sizeof(options));
 
+	perf_hpp__for_each_format(fmt)
+		perf_hpp__reset_width(fmt, hists);
+
 	if (symbol_conf.col_width_list_str)
 		perf_hpp__set_user_width(symbol_conf.col_width_list_str);
 
Signature · 1 line
2.0.0
#20 lore raw
Quoted 36 lines · Namhyung Kim

Hi Jiri,

Quoted 33 lines · Jiri Olsa
Quoted 13 lines · Namhyung Kim

Hello,

This patchset is to control perf report/top output column width by -w/--column-widths option so that it can fit into the terminal size. The -w option is there for perf report but it ignored by recent output field changed due to some reason. This patchset fixes it and supports perf top also.

This is sometimes useful if your terminal is small and there's some C++ applications which have amazingly long symbol names. Without this patchset user might not see those symbols on TUI, since it maps left/right arrow keys to other functions.

Quoted 6 lines

The -w option sets column width starting from the first column (overhead or optional overhead_children column unless -F option is given). It doesn't make sense to limit those overhead columns so it's not a hard-limit for them. But it *is* a hard-limit for other columns such as comm, dso, symbol, and so on. One can use 0 not to limit/force a width for those columns.

hi,
I've got broken TUI output for "perf report --group"

Samples: 17 of event 'anon group { cycles, instructions }', Event count (approx.): 9145256
56.44% 31.59% ls libc-2.17.so [.] __strcoll_l
39.94% 0.00% ls ld-2.17.so [.] _dl_new_object

 3.48% 0.00%  ls  [kernel.kallsyms]  [k] setup_arg_pages
 0.14% 0.33%  ls  [kernel.kallsyms]  [k] native_write_msr_safe
 0.00% 30.17%  ls  [kernel.kallsyms]  [k] security_inode_permission
 0.00% 29.78%  ls  ls                 [.] indent

0.00% 8.12% ls [kernel.kallsyms] [k] __slab_alloc

Quoted 4 lines
Quoted 2 lines

I have 'show-headers' set to false in ~/.perfconfig and the output got fixed after displaying headers by pressing 'H'

Argh, you're right. It should be fixed with this patch:

yep, that fixed that for me

thanks,
jirka

#21 lore raw
In reply to Namhyung Kim

Em Wed, Jul 09, 2014 at 02:28:08PM +0900, Namhyung Kim escreveu:

Quoted 13 lines

Hello,

This patchset is to control perf report/top output column width by -w/--column-widths option so that it can fit into the terminal size. The -w option is there for perf report but it ignored by recent output field changed due to some reason. This patchset fixes it and supports perf top also.

This is sometimes useful if your terminal is small and there's some C++ applications which have amazingly long symbol names. Without this patchset user might not see those symbols on TUI, since it maps left/right arrow keys to other functions.

Quoted 6 lines

The -w option sets column width starting from the first column (overhead or optional overhead_children column unless -F option is given). It doesn't make sense to limit those overhead columns so it's not a hard-limit for them. But it *is* a hard-limit for other columns such as comm, dso, symbol, and so on. One can use 0 not to limit/force a width for those columns.

Due to the issues pointed out, will wait for a v2 patchset, ok?

- Arnaldo

author maintainer
#22 lore raw
Quoted 14 lines · Arnaldo Carvalho de Melo

Em Wed, Jul 09, 2014 at 02:28:08PM +0900, Namhyung Kim escreveu:

Quoted 13 lines

Hello,

This patchset is to control perf report/top output column width by -w/--column-widths option so that it can fit into the terminal size. The -w option is there for perf report but it ignored by recent output field changed due to some reason. This patchset fixes it and supports perf top also.

This is sometimes useful if your terminal is small and there's some C++ applications which have amazingly long symbol names. Without this patchset user might not see those symbols on TUI, since it maps left/right arrow keys to other functions.

Quoted 8 lines
Quoted 6 lines

The -w option sets column width starting from the first column (overhead or optional overhead_children column unless -F option is given). It doesn't make sense to limit those overhead columns so it's not a hard-limit for them. But it *is* a hard-limit for other columns such as comm, dso, symbol, and so on. One can use 0 not to limit/force a width for those columns.

Due to the issues pointed out, will wait for a v2 patchset, ok?

Okay. Will send v2 tomorrow.

Thanks,
Namhyung