Skip to contentSkip to conversation

perf report/annotate: Support average IPC and IPC coverage for function

PATCH Superseded Jin Yao 3 participants 9 messages Nov 29, 2018 – Nov 30, 2018 perf events

v4 of 5 Superseded Latest is v5 →

lore
Share & tools
lore

Jin Yao

author
#0 lore raw

v4 cover perf report/annotate: Support average IPC and IPC coverage for function

Add supporting of displaying the average IPC and IPC coverage percentage per function.

For example,

$ perf record -b ...
$ perf report -s symbol or
  perf report -s symbol --stdio
Overhead  Symbol                           IPC   [IPC Coverage]
  39.60%  [.] __random                     2.30  [ 54.8%]
  18.02%  [.] main                         0.43  [ 54.3%]
  14.21%  [.] compute_flag                 2.29  [100.0%]
  14.16%  [.] rand                         0.36  [100.0%]
   7.06%  [.] __random_r                   2.57  [ 70.5%]
   6.85%  [.] rand@plt                     0.00  [  0.0%]

...

$ perf annotate --stdio2

Percent IPC Cycle (Average IPC: 2.30, IPC Coverage: 54.8%)

Disassembly of section .text:

000000000003aac0 <random@@GLIBC_2.2.5>:

  8.32  3.28              sub    $0x18,%rsp
        3.28              mov    $0x1,%esi
        3.28              xor    %eax,%eax
        3.28              cmpl   $0x0,argp_program_version_hook@@GLIBC_2.2.5+0x1e0

11.57 3.28 1 ↓ je 20
lock cmpxchg %esi,__abort_msg@@GLIBC_PRIVATE+0x8a0
↓ jne 29
↓ jmp 43
11.57 1.10 20: cmpxchg %esi,__abort_msg@@GLIBC_PRIVATE+0x8a0
...

v4:
---
Add a new patch "perf report: Documentation average IPC and IPC coverage" to explain the new columns "IPC" and "IPC Coverage".

Other patches are not changed.

v3:
---
Remove the sortkey "ipc" from command-line. The columns "IPC" and "[IPC Coverage]" are automatically enabled when "symbol" is specified.

Patch "perf report: Display average IPC and IPC coverage per symbol" is impacted.

v2:
---
1. Merge in Jiri's patch to support stdio mode

2. Add a new patch "perf annotate: Create a annotate2 flag in struct symbol" which records if the symbol has been annotated yet.

3. Minor update such as adding { } for multiline code in 'if' condition.

Jin Yao (4):
perf annotate: Compute average IPC and IPC coverage per symbol perf annotate: Create a annotate2 flag in struct symbol perf report: Display average IPC and IPC coverage per symbol perf report: Documentation average IPC and IPC coverage

 tools/perf/Documentation/perf-report.txt |  8 +++++
 tools/perf/builtin-report.c              | 26 ++++++++++++--
 tools/perf/util/annotate.c               | 42 ++++++++++++++++++++--
 tools/perf/util/annotate.h               |  5 +++
 tools/perf/util/hist.h                   |  1 +
 tools/perf/util/sort.c                   | 61 ++++++++++++++++++++++++++++++++
 tools/perf/util/sort.h                   |  2 ++
 tools/perf/util/symbol.h                 |  1 +
 8 files changed, 140 insertions(+), 6 deletions(-)
Signature · 1 line
2.7.4
author
#1 lore raw

PATCH v4 1/4 perf annotate: Compute average IPC and IPC coverage per symbol

Add support to perf report annotate view or perf annotate --stdio2 to aggregate the IPC derived from timed LBRs per symbol. We compute the average IPC and the IPC coverage percentage.

For example,

$ perf annotate --stdio2

Percent IPC Cycle (Average IPC: 2.30, IPC Coverage: 54.8%)

Disassembly of section .text:

000000000003aac0 <random@@GLIBC_2.2.5>:

  8.32  3.28              sub    $0x18,%rsp
        3.28              mov    $0x1,%esi
        3.28              xor    %eax,%eax
        3.28              cmpl   $0x0,argp_program_version_hook@@GLIBC_2.2.5+0x1e0
 11.57  3.28     1      ↓ je     20
                          lock   cmpxchg %esi,__abort_msg@@GLIBC_PRIVATE+0x8a0
                        ↓ jne    29
                        ↓ jmp    43
 11.57  1.10        20:   cmpxchg %esi,__abort_msg@@GLIBC_PRIVATE+0x8a0
  0.00  1.10     1      ↓ je     43
                    29:   lea    __abort_msg@@GLIBC_PRIVATE+0x8a0,%rdi
                          sub    $0x80,%rsp
                        → callq  __lll_lock_wait_private
                          add    $0x80,%rsp
  0.00  3.00        43:   lea    __ctype_b@GLIBC_2.2.5+0x38,%rdi
        3.00              lea    0xc(%rsp),%rsi
  8.49  3.00     1      → callq  __random_r
  7.91  1.94              cmpl   $0x0,argp_program_version_hook@@GLIBC_2.2.5+0x1e0
  0.00  1.94     1      ↓ je     68
                          lock   decl   __abort_msg@@GLIBC_PRIVATE+0x8a0
                        ↓ jne    70
                        ↓ jmp    8a
  0.00  2.00        68:   decl   __abort_msg@@GLIBC_PRIVATE+0x8a0
 21.56  2.00     1      ↓ je     8a
                    70:   lea    __abort_msg@@GLIBC_PRIVATE+0x8a0,%rdi
                          sub    $0x80,%rsp
                        → callq  __lll_unlock_wake_private
                          add    $0x80,%rsp
 21.56  2.90        8a:   movslq 0xc(%rsp),%rax
        2.90              add    $0x18,%rsp

9.03 2.90 1 ← retq

It shows for this symbol the average IPC is 2.30 and the IPC coverage is 54.8%.

---
 tools/perf/util/annotate.c | 41 ++++++++++++++++++++++++++++++++++++++---
 tools/perf/util/annotate.h |  5 +++++
 2 files changed, 43 insertions(+), 3 deletions(-)
tools/perf/util/annotate.c +38 -3
index 6936daf..4b2b1b0 100644
@@ -1000,6 +1000,7 @@ static unsigned annotation__count_insn(struct annotation *notes, u64 start, u64
 static void annotation__count_and_fill(struct annotation *notes, u64 start, u64 end, struct cyc_hist *ch)
 {
 	unsigned n_insn;
+	unsigned int cover_insn = 0;
 	u64 offset;
 
 	n_insn = annotation__count_insn(notes, start, end);
@@ -1013,21 +1014,34 @@ static void annotation__count_and_fill(struct annotation *notes, u64 start, u64
 		for (offset = start; offset <= end; offset++) {
 			struct annotation_line *al = notes->offsets[offset];
 
-			if (al)
+			if (al && al->ipc == 0.0) {
 				al->ipc = ipc;
+				cover_insn++;
+			}
+		}
+
+		if (cover_insn) {
+			notes->hit_cycles += ch->cycles;
+			notes->hit_insn += n_insn * ch->num;
+			notes->cover_insn += cover_insn;
 		}
 	}
 }
 
 void annotation__compute_ipc(struct annotation *notes, size_t size)
 {
-	u64 offset;
+	s64 offset;
 
 	if (!notes->src || !notes->src->cycles_hist)
 		return;
 
+	notes->total_insn = annotation__count_insn(notes, 0, size - 1);
+	notes->hit_cycles = 0;
+	notes->hit_insn = 0;
+	notes->cover_insn = 0;
+
 	pthread_mutex_lock(&notes->lock);
-	for (offset = 0; offset < size; ++offset) {
+	for (offset = size - 1; offset >= 0; --offset) {
 		struct cyc_hist *ch;
 
 		ch = &notes->src->cycles_hist[offset];
@@ -2563,6 +2577,22 @@ static void disasm_line__write(struct disasm_line *dl, struct annotation *notes,
 	disasm_line__scnprintf(dl, bf, size, !notes->options->use_offset);
 }
 
+static void ipc_coverage_string(char *bf, int size, struct annotation *notes)
+{
+	double ipc = 0.0, coverage = 0.0;
+
+	if (notes->hit_cycles)
+		ipc = notes->hit_insn / ((double)notes->hit_cycles);
+
+	if (notes->total_insn) {
+		coverage = notes->cover_insn * 100.0 /
+			((double)notes->total_insn);
+	}
+
+	scnprintf(bf, size, "(Average IPC: %.2f, IPC Coverage: %.1f%%)",
+		  ipc, coverage);
+}
+
 static void __annotation_line__write(struct annotation_line *al, struct annotation *notes,
 				     bool first_line, bool current_entry, bool change_color, int width,
 				     void *obj, unsigned int percent_type,
@@ -2658,6 +2688,11 @@ static void __annotation_line__write(struct annotation_line *al, struct annotati
 					    ANNOTATION__MINMAX_CYCLES_WIDTH - 1,
 					    "Cycle(min/max)");
 		}
+
+		if (show_title && !*al->line) {
+			ipc_coverage_string(bf, sizeof(bf), notes);
+			obj__printf(obj, "%*s", ANNOTATION__AVG_IPC_WIDTH, bf);
+		}
 	}
 
 	obj__printf(obj, " ");
tools/perf/util/annotate.h +5 -0
index 5399ba2..fb64637 100644
@@ -64,6 +64,7 @@ bool ins__is_fused(struct arch *arch, const char *ins1, const char *ins2);
 #define ANNOTATION__IPC_WIDTH 6
 #define ANNOTATION__CYCLES_WIDTH 6
 #define ANNOTATION__MINMAX_CYCLES_WIDTH 19
+#define ANNOTATION__AVG_IPC_WIDTH 36
 
 struct annotation_options {
 	bool hide_src_code,
@@ -262,6 +263,10 @@ struct annotation {
 	pthread_mutex_t		lock;
 	u64			max_coverage;
 	u64			start;
+	u64			hit_cycles;
+	u64			hit_insn;
+	unsigned int		total_insn;
+	unsigned int		cover_insn;
 	struct annotation_options *options;
 	struct annotation_line	**offsets;
 	int			nr_events;
Signature · 1 line
2.7.4
author
#2 lore raw

PATCH v4 2/4 perf annotate: Create a annotate2 flag in struct symbol

In reply to Jin Yao

We often use the symbol__annotate2() to annotate a specified symbol. While annotating may take some time, so in order to avoid annotating the same symbol repeatedly, the patch creates a new flag to indicate the symbol has been annotated.

---
 tools/perf/util/annotate.c | 1 +
 tools/perf/util/symbol.h   | 1 +
 2 files changed, 2 insertions(+)
tools/perf/util/annotate.c +1 -0
index 4b2b1b0..f69d8e1 100644
@@ -2798,6 +2798,7 @@ int symbol__annotate2(struct symbol *sym, struct map *map, struct perf_evsel *ev
 	notes->nr_events = nr_pcnt;
 
 	annotation__update_column_widths(notes);
+	sym->annotate2 = true;
 
 	return 0;
 
tools/perf/util/symbol.h +1 -0
index d026d21..14d9d43 100644
@@ -63,6 +63,7 @@ struct symbol {
 	u8		ignore:1;
 	u8		inlined:1;
 	u8		arch_sym;
+	bool		annotate2;
 	char		name[0];
 };
 
Signature · 1 line
2.7.4
author
#3 lore raw

PATCH v4 3/4 perf report: Display average IPC and IPC coverage per symbol

In reply to Jin Yao

Support displaying the average IPC and IPC coverage per symbol in perf report TUI and stdio modes.

For example,

$ perf record -b ...
$ perf report -s symbol
Overhead  Symbol                           IPC   [IPC Coverage]
  39.60%  [.] __random                     2.30  [ 54.8%]
  18.02%  [.] main                         0.43  [ 54.3%]
  14.21%  [.] compute_flag                 2.29  [100.0%]
  14.16%  [.] rand                         0.36  [100.0%]
   7.06%  [.] __random_r                   2.57  [ 70.5%]
   6.85%  [.] rand@plt                     0.00  [  0.0%]

Jiri Olsa <[email protected]> provides the patch to support the stdio mode. I merge Jiri's code in this patch.

$ perf report -s symbol --stdio
  # Overhead  Symbol                       IPC   [IPC Coverage]
  # ........  ...........................  ....................
  #
    39.60%  [.] __random                   2.30  [ 54.8%]
    18.02%  [.] main                       0.43  [ 54.3%]
    14.21%  [.] compute_flag               2.29  [100.0%]
    14.16%  [.] rand                       0.36  [100.0%]
     7.06%  [.] __random_r                 2.57  [ 70.5%]
     6.85%  [.] rand@plt                   0.00  [  0.0%]
     0.02%  [k] run_timer_softirq          1.60  [ 57.2%]

The columns "IPC" and "[IPC Coverage]" are automatically enabled when the sort-key "symbol" is specified. If the perf.data doesn't contain timed LBR information, columns are filled with "-".

For example,

  # Overhead  Symbol                       IPC   [IPC Coverage]
  # ........  ...........................  ....................
  #
      46.57%  [.] main                     -      -
      17.60%  [.] rand                     -      -
      15.84%  [.] __random_r               -      -
      11.90%  [.] __random                 -      -
       6.50%  [.] compute_flag             -      -
       1.59%  [.] rand@plt                 -      -
       0.00%  [.] _dl_relocate_object      -      -
       0.00%  [k] tlb_flush_mmu            -      -
       0.00%  [k] perf_event_mmap          -      -
       0.00%  [k] native_sched_clock       -      -
       0.00%  [k] intel_pmu_handle_irq_v4  -      -
       0.00%  [k] native_write_msr         -      -

v3:
---
Removed the sortkey 'ipc' from command-line. The columns "IPC" and "[IPC Coverage]" are automatically enabled when "symbol" is specified.

v2:
---
Merge in Jiri's patch to support stdio mode

---
 tools/perf/builtin-report.c | 26 ++++++++++++++++---
 tools/perf/util/hist.h      |  1 +
 tools/perf/util/sort.c      | 61 +++++++++++++++++++++++++++++++++++++++++++++
 tools/perf/util/sort.h      |  2 ++
 4 files changed, 87 insertions(+), 3 deletions(-)
tools/perf/builtin-report.c +23 -3
index 257c9c1..4958095 100644
@@ -85,6 +85,7 @@ struct report {
 	int			socket_filter;
 	DECLARE_BITMAP(cpu_bitmap, MAX_NR_CPUS);
 	struct branch_type_stat	brtype_stat;
+	bool			symbol_ipc;
 };
 
 static int report__config(const char *var, const char *value, void *cb)
@@ -129,7 +130,7 @@ static int hist_iter__report_callback(struct hist_entry_iter *iter,
 	struct mem_info *mi;
 	struct branch_info *bi;
 
-	if (!ui__has_annotation())
+	if (!ui__has_annotation() && !rep->symbol_ipc)
 		return 0;
 
 	hist__account_cycles(sample->branch_stack, al, sample,
@@ -174,7 +175,7 @@ static int hist_iter__branch_callback(struct hist_entry_iter *iter,
 	struct perf_evsel *evsel = iter->evsel;
 	int err;
 
-	if (!ui__has_annotation())
+	if (!ui__has_annotation() && !rep->symbol_ipc)
 		return 0;
 
 	hist__account_cycles(sample->branch_stack, al, sample,
@@ -1133,6 +1134,7 @@ int cmd_report(int argc, const char **argv)
 		.mode  = PERF_DATA_MODE_READ,
 	};
 	int ret = hists__init();
+	char sort_tmp[128];
 
 	if (ret < 0)
 		return ret;
@@ -1284,6 +1286,24 @@ int cmd_report(int argc, const char **argv)
 	else
 		use_browser = 0;
 
+	if (sort_order && strstr(sort_order, "ipc")) {
+		parse_options_usage(report_usage, options, "s", 1);
+		goto error;
+	}
+
+	if (sort_order && strstr(sort_order, "symbol")) {
+		if (sort__mode == SORT_MODE__BRANCH) {
+			snprintf(sort_tmp, sizeof(sort_tmp), "%s,%s",
+				 sort_order, "ipc_lbr");
+			report.symbol_ipc = true;
+		} else {
+			snprintf(sort_tmp, sizeof(sort_tmp), "%s,%s",
+				 sort_order, "ipc_null");
+		}
+
+		sort_order = sort_tmp;
+	}
+
 	if (setup_sorting(session->evlist) < 0) {
 		if (sort_order)
 			parse_options_usage(report_usage, options, "s", 1);
@@ -1311,7 +1331,7 @@ int cmd_report(int argc, const char **argv)
 	 * so don't allocate extra space that won't be used in the stdio
 	 * implementation.
 	 */
-	if (ui__has_annotation()) {
+	if (ui__has_annotation() || report.symbol_ipc) {
 		ret = symbol__annotation_init();
 		if (ret < 0)
 			goto error;
tools/perf/util/hist.h +1 -0
index 3badd7f..664b5ed 100644
@@ -62,6 +62,7 @@ enum hist_column {
 	HISTC_TRACE,
 	HISTC_SYM_SIZE,
 	HISTC_DSO_SIZE,
+	HISTC_SYMBOL_IPC,
 	HISTC_NR_COLS, /* Last entry */
 };
 
tools/perf/util/sort.c +61 -0

85 lines not on this page · raw message

tools/perf/util/sort.h +2 -0
index a97cf8e..130fe37 100644
@@ -229,6 +229,7 @@ enum sort_type {
 	SORT_SYM_SIZE,
 	SORT_DSO_SIZE,
 	SORT_CGROUP_ID,
+	SORT_SYM_IPC_NULL,
 
 	/* branch stack specific sort keys */
 	__SORT_BRANCH_STACK,
@@ -242,6 +243,7 @@ enum sort_type {
 	SORT_CYCLES,
 	SORT_SRCLINE_FROM,
 	SORT_SRCLINE_TO,
+	SORT_SYM_IPC,
 
 	/* memory mode specific sort keys */
 	__SORT_MEMORY_MODE,
Signature · 1 line
2.7.4
author
#4 lore raw

PATCH v4 4/4 perf report: Documentation average IPC and IPC coverage

In reply to Jin Yao

Add explanations for new columns "IPC" and "IPC coverage" in perf documentation.

---
 tools/perf/Documentation/perf-report.txt | 8 ++++++++
 1 file changed, 8 insertions(+)
tools/perf/Documentation/perf-report.txt +8 -0
index 474a494..e5a32f3 100644
@@ -126,6 +126,14 @@ OPTIONS
 	And default sort keys are changed to comm, dso_from, symbol_from, dso_to
 	and symbol_to, see '--branch-stack'.
 
+	When the sort key symbol is specified, columns "IPC" and "IPC Coverage"
+	are enabled automatically. Column "IPC" reports the average IPC per function
+	and column "IPC coverage" reports the percentage of instructions with
+	sampled IPC in this function. IPC means Instruction Per Cycle. If it's low,
+	it indicates there may be performance bottleneck when the function is
+	executed, such as, memory access bottleneck. If a function has high overhead
+	and low IPC, it's worth further analysis for performance optimization.
+
 	If the --mem-mode option is used, the following sort keys are also available
 	(incompatible with --branch-stack):
 	symbol_daddr, dso_daddr, locked, tlb, mem, snoop, dcacheline.
Signature · 1 line
2.7.4
Reviewed-by on v4 4/4
#5 lore raw

Re: [PATCH v4 4/4] perf report: Documentation average IPC and IPC coverage

Quoted 13 lines · Jin Yao

Add explanations for new columns "IPC" and "IPC coverage" in perf documentation.

---
 tools/perf/Documentation/perf-report.txt | 8 ++++++++
 1 file changed, 8 insertions(+)
tools/perf/Documentation/perf-report.txt +0 -0
index 474a494..e5a32f3 100644

tools/perf/Documentation/perf-report.txt OPTIONS · line 135

Quoted 10 lines
tools/perf/Documentation/perf-report.txt +7 -0
@@ -126,6 +126,14 @@ OPTIONS
 	And default sort keys are changed to comm, dso_from, symbol_from, dso_to
 	and symbol_to, see '--branch-stack'.
 
+	When the sort key symbol is specified, columns "IPC" and "IPC Coverage"
+	are enabled automatically. Column "IPC" reports the average IPC per function
+	and column "IPC coverage" reports the percentage of instructions with
+	sampled IPC in this function. IPC means Instruction Per Cycle. If it's low,
+	it indicates there may be performance bottleneck when the function is
+	executed, such as, memory access bottleneck. If a function has high overhead
+	and low IPC, it's worth further analysis for performance optimization.

Thank you for adding this!

Just a few small nits:

s/may be performance bottleneck
/may be a performance bottleneck

s/such as, memory access bottleneck
/such as a memory access bottleneck

s/it's worth further analysis for performance optimization.
/it's worth further analyzing it to optimize its performance.

?

Other than that:

Ingo

author
#6 lore raw

Re: [PATCH v4 4/4] perf report: Documentation average IPC and IPC coverage

Quoted 40 lines · Ingo Molnar
Quoted 13 lines · Jin Yao

Add explanations for new columns "IPC" and "IPC coverage" in perf documentation.

---
  tools/perf/Documentation/perf-report.txt | 8 ++++++++
  1 file changed, 8 insertions(+)
tools/perf/Documentation/perf-report.txt +0 -0
index 474a494..e5a32f3 100644

tools/perf/Documentation/perf-report.txt OPTIONS · line 135

Quoted 10 lines
tools/perf/Documentation/perf-report.txt +7 -0
@@ -126,6 +126,14 @@ OPTIONS
  	And default sort keys are changed to comm, dso_from, symbol_from, dso_to
  	and symbol_to, see '--branch-stack'.
 
+	When the sort key symbol is specified, columns "IPC" and "IPC Coverage"
+	are enabled automatically. Column "IPC" reports the average IPC per function
+	and column "IPC coverage" reports the percentage of instructions with
+	sampled IPC in this function. IPC means Instruction Per Cycle. If it's low,
+	it indicates there may be performance bottleneck when the function is
+	executed, such as, memory access bottleneck. If a function has high overhead
+	and low IPC, it's worth further analysis for performance optimization.

Thank you for adding this!

Just a few small nits:

s/may be performance bottleneck
/may be a performance bottleneck

s/such as, memory access bottleneck
/such as a memory access bottleneck

s/it's worth further analysis for performance optimization.
/it's worth further analyzing it to optimize its performance.

Quoted 8 lines

?

Other than that:

Ingo

Thanks Ingo!

I will add these fixes in v5.

Thanks
Jin Yao

Reviewed-by on the v4 series
#7 lore raw
In reply to Jin Yao
Quoted 60 lines · Jin Yao

Add supporting of displaying the average IPC and IPC coverage percentage per function.

For example,

$ perf record -b ...
$ perf report -s symbol or
  perf report -s symbol --stdio
Overhead  Symbol                           IPC   [IPC Coverage]
  39.60%  [.] __random                     2.30  [ 54.8%]
  18.02%  [.] main                         0.43  [ 54.3%]
  14.21%  [.] compute_flag                 2.29  [100.0%]
  14.16%  [.] rand                         0.36  [100.0%]
   7.06%  [.] __random_r                   2.57  [ 70.5%]
   6.85%  [.] rand@plt                     0.00  [  0.0%]

...

$ perf annotate --stdio2

Percent IPC Cycle (Average IPC: 2.30, IPC Coverage: 54.8%)

Disassembly of section .text:

000000000003aac0 <random@@GLIBC_2.2.5>:

  8.32  3.28              sub    $0x18,%rsp
        3.28              mov    $0x1,%esi
        3.28              xor    %eax,%eax
        3.28              cmpl   $0x0,argp_program_version_hook@@GLIBC_2.2.5+0x1e0

11.57 3.28 1 ↓ je 20
lock cmpxchg %esi,__abort_msg@@GLIBC_PRIVATE+0x8a0
↓ jne 29
↓ jmp 43
11.57 1.10 20: cmpxchg %esi,__abort_msg@@GLIBC_PRIVATE+0x8a0
...

v4:
---
Add a new patch "perf report: Documentation average IPC and IPC coverage" to explain the new columns "IPC" and "IPC Coverage".

Other patches are not changed.

v3:
---
Remove the sortkey "ipc" from command-line. The columns "IPC" and "[IPC Coverage]" are automatically enabled when "symbol" is specified.

Patch "perf report: Display average IPC and IPC coverage per symbol" is impacted.

v2:
---
1. Merge in Jiri's patch to support stdio mode

2. Add a new patch "perf annotate: Create a annotate2 flag in struct symbol" which records if the symbol has been annotated yet.

Quoted 8 lines

3. Minor update such as adding { } for multiline code in 'if' condition.

Jin Yao (4):
perf annotate: Compute average IPC and IPC coverage per symbol perf annotate: Create a annotate2 flag in struct symbol perf report: Display average IPC and IPC coverage per symbol perf report: Documentation average IPC and IPC coverage

thanks,
jirka

author
#8 lore raw
Quoted 70 lines · Jiri Olsa
Quoted 60 lines · Jin Yao

Add supporting of displaying the average IPC and IPC coverage percentage per function.

For example,

$ perf record -b ...
$ perf report -s symbol or
   perf report -s symbol --stdio

Overhead Symbol IPC [IPC Coverage]

   39.60%  [.] __random                     2.30  [ 54.8%]
   18.02%  [.] main                         0.43  [ 54.3%]
   14.21%  [.] compute_flag                 2.29  [100.0%]
   14.16%  [.] rand                         0.36  [100.0%]
    7.06%  [.] __random_r                   2.57  [ 70.5%]
    6.85%  [.] rand@plt                     0.00  [  0.0%]

...

$ perf annotate --stdio2

Percent IPC Cycle (Average IPC: 2.30, IPC Coverage: 54.8%)

Disassembly of section .text:

000000000003aac0 <random@@GLIBC_2.2.5>:

   8.32  3.28              sub    $0x18,%rsp
         3.28              mov    $0x1,%esi
         3.28              xor    %eax,%eax
         3.28              cmpl   $0x0,argp_program_version_hook@@GLIBC_2.2.5+0x1e0

11.57 3.28 1 ↓ je 20
lock cmpxchg %esi,__abort_msg@@GLIBC_PRIVATE+0x8a0
↓ jne 29
↓ jmp 43
11.57 1.10 20: cmpxchg %esi,__abort_msg@@GLIBC_PRIVATE+0x8a0
...

v4:
---
Add a new patch "perf report: Documentation average IPC and IPC coverage" to explain the new columns "IPC" and "IPC Coverage".

Other patches are not changed.

v3:
---
Remove the sortkey "ipc" from command-line. The columns "IPC" and "[IPC Coverage]" are automatically enabled when "symbol" is specified.

Patch "perf report: Display average IPC and IPC coverage per symbol" is impacted.

v2:
---
1. Merge in Jiri's patch to support stdio mode

2. Add a new patch "perf annotate: Create a annotate2 flag in struct symbol" which records if the symbol has been annotated yet.

Quoted 8 lines

3. Minor update such as adding { } for multiline code in 'if' condition.

Jin Yao (4):
perf annotate: Compute average IPC and IPC coverage per symbol perf annotate: Create a annotate2 flag in struct symbol perf report: Display average IPC and IPC coverage per symbol perf report: Documentation average IPC and IPC coverage

Quoted 5 lines

thanks,
jirka

Thanks so much for reviewing the patches.

I will post v5 which includes some minor updates for documentation.

Thanks
Jin Yao