Skip to contentSkip to conversation

perf trace: Fix noise and signed formatting of __probe_ip in bare dynamic probes

PATCH In review Aaron Tomlin 2 participants 3 messages Jun 12 – Jun 15 perf events

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 perf trace: Fix noise and signed formatting of __probe_ip in bare dynamic probes

When a dynamic probe is created without explicitly requested arguments via perf probe --add, the Ftrace subsystem automatically appends "__probe_ip" to the tracepoint format to record the instruction pointer.

Currently, perf trace prints this implicit field by default. Furthermore, because the formatting logic defaults to a standard signed integer representation, the kernel space memory address is erroneously displayed as a meaningless negative integer.

    ❯ sudo ./perf trace --event probe:proc_sys_open --max-events 1
         0.000 ps/1316543 probe:proc_sys_open(__probe_ip: -1406056956)

This patch addresses the user experience by combining two refinements:
1. "__probe_ip" is now hidden from the standard output, as its presence adds no contextual value for a bare probe.

2. If the user explicitly requests verbose output (--verbose), "__probe_ip" is intercepted and properly formatted as a hexadecimal kernel address, restoring its utility for debugging inline function hits.

❯ sudo ./perf trace --event probe:proc_sys_open --max-events 1
0.000 ps/1314074 probe:proc_sys_open()

❯ sudo ./perf trace --verbose --event probe:proc_sys_open --max-events 1 Using CPUID GenuineIntel-6-8E-C
mmap size 528384B
0.000 ps/1314366 probe:proc_sys_open(__probe_ip: 0xffffffffac314604)

---
Changes since v1:

- Fixed a bug where the hardcoded format string for "__probe_ip" bypassed the trace->show_arg_names configuration. The formatting logic is now properly decoupled to respect trace.show_arg_names when verbose mode is enabled.

- Linked to v1: https://lore.kernel.org/lkml/[email protected]/

---
 tools/perf/builtin-trace.c | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)
tools/perf/builtin-trace.c +21 -0
index 48615ddccd93..ac15b5588a27 100644
@@ -3234,6 +3234,27 @@ static size_t trace__fprintf_tp_fields(struct trace *trace, struct evsel *evsel,
 		if (val == 0 && !trace->show_zeros && !arg->show_zero && arg->strtoul != STUL_BTF_TYPE)
 			continue;
 
+		/*
+		 * __probe_ip is implicitly added to bare dynamic probes.
+		 * Suppress it by default to avoid cluttering the output.
+		 * If verbose mode is enabled, ensure it is formatted as a
+		 * hexadecimal memory address rather than a signed integer.
+		 */
+		if (!strcmp(field->name, "__probe_ip")) {
+			if (!verbose)
+				continue;
+
+			printed += scnprintf(bf + printed, size - printed,
+					     "%s", printed ? ", " : "");
+			if (trace->show_arg_names)
+				printed += scnprintf(bf + printed, size - printed,
+						     "%s: ", field->name);
+
+			printed += scnprintf(bf + printed, size - printed, "%#016llx",
+					     (unsigned long long)val);
+			continue;
+		}
+
 		printed += scnprintf(bf + printed, size - printed, "%s", printed ? ", " : "");
 
 		if (trace->show_arg_names)
Signature · 1 line
2.51.0
reviewer Reviewed-by on v2 1/1
#1 lore raw
Quoted 48 lines · Aaron Tomlin

When a dynamic probe is created without explicitly requested arguments via perf probe --add, the Ftrace subsystem automatically appends "__probe_ip" to the tracepoint format to record the instruction pointer.

Currently, perf trace prints this implicit field by default. Furthermore, because the formatting logic defaults to a standard signed integer representation, the kernel space memory address is erroneously displayed as a meaningless negative integer.

     ❯ sudo ./perf trace --event probe:proc_sys_open --max-events 1
          0.000 ps/1316543 probe:proc_sys_open(__probe_ip: -1406056956)

This patch addresses the user experience by combining two refinements:
1. "__probe_ip" is now hidden from the standard output, as its presence adds no contextual value for a bare probe.

2. If the user explicitly requests verbose output (--verbose), "__probe_ip" is intercepted and properly formatted as a hexadecimal kernel address, restoring its utility for debugging inline function hits.

❯ sudo ./perf trace --event probe:proc_sys_open --max-events 1
0.000 ps/1314074 probe:proc_sys_open()

❯ sudo ./perf trace --verbose --event probe:proc_sys_open --max-events 1 Using CPUID GenuineIntel-6-8E-C
mmap size 528384B
0.000 ps/1314366 probe:proc_sys_open(__probe_ip: 0xffffffffac314604)

---
Changes since v1:

- Fixed a bug where the hardcoded format string for "__probe_ip" bypassed the trace->show_arg_names configuration. The formatting logic is now properly decoupled to respect trace.show_arg_names when verbose mode is enabled.

- Linked to v1: https://lore.kernel.org/lkml/[email protected]/

---
  tools/perf/builtin-trace.c | 21 +++++++++++++++++++++
  1 file changed, 21 insertions(+)
tools/perf/builtin-trace.c +0 -0
index 48615ddccd93..ac15b5588a27 100644

tools/perf/builtin-trace.c static size_t trace__fprintf_tp_fields(struct trace *trace, struct evsel *evsel, · line 3243

Quoted 10 lines
tools/perf/builtin-trace.c +7 -0
@@ -3234,6 +3234,27 @@ static size_t trace__fprintf_tp_fields(struct trace *trace, struct evsel *evsel,
  		if (val == 0 && !trace->show_zeros && !arg->show_zero && arg->strtoul != STUL_BTF_TYPE)
  			continue;
 
+		/*
+		 * __probe_ip is implicitly added to bare dynamic probes.
+		 * Suppress it by default to avoid cluttering the output.
+		 * If verbose mode is enabled, ensure it is formatted as a
+		 * hexadecimal memory address rather than a signed integer.
+		 */
+		if (!strcmp(field->name, "__probe_ip")) {

I checked to see if it was possible to include the #define for this constant (FIELD_STRING_IP) in the tools folder, but there wasn't anything existing in include/ that fits nicely. Either way:

Quoted 17 lines
tools/perf/builtin-trace.c +14 -0
+			if (!verbose)
+				continue;
+
+			printed += scnprintf(bf + printed, size - printed,
+					     "%s", printed ? ", " : "");
+			if (trace->show_arg_names)
+				printed += scnprintf(bf + printed, size - printed,
+						     "%s: ", field->name);
+
+			printed += scnprintf(bf + printed, size - printed, "%#016llx",
+					     (unsigned long long)val);
+			continue;
+		}
+
  		printed += scnprintf(bf + printed, size - printed, "%s", printed ? ", " : "");
 
  		if (trace->show_arg_names)
#2 lore raw
Quoted 58 lines · James Clark
Quoted 44 lines

Currently, perf trace prints this implicit field by default. Furthermore, because the formatting logic defaults to a standard signed integer representation, the kernel space memory address is erroneously displayed as a meaningless negative integer.

     ❯ sudo ./perf trace --event probe:proc_sys_open --max-events 1
          0.000 ps/1316543 probe:proc_sys_open(__probe_ip: -1406056956)

This patch addresses the user experience by combining two refinements:
1. "__probe_ip" is now hidden from the standard output, as its presence adds no contextual value for a bare probe.

2. If the user explicitly requests verbose output (--verbose), "__probe_ip" is intercepted and properly formatted as a hexadecimal kernel address, restoring its utility for debugging inline function hits.

❯ sudo ./perf trace --event probe:proc_sys_open --max-events 1
0.000 ps/1314074 probe:proc_sys_open()

❯ sudo ./perf trace --verbose --event probe:proc_sys_open --max-events 1 Using CPUID GenuineIntel-6-8E-C
mmap size 528384B
0.000 ps/1314366 probe:proc_sys_open(__probe_ip: 0xffffffffac314604)

---
Changes since v1:

- Fixed a bug where the hardcoded format string for "__probe_ip" bypassed the trace->show_arg_names configuration. The formatting logic is now properly decoupled to respect trace.show_arg_names when verbose mode is enabled.

- Linked to v1: https://lore.kernel.org/lkml/[email protected]/

---
  tools/perf/builtin-trace.c | 21 +++++++++++++++++++++
  1 file changed, 21 insertions(+)
tools/perf/builtin-trace.c +0 -0
index 48615ddccd93..ac15b5588a27 100644

tools/perf/builtin-trace.c static size_t trace__fprintf_tp_fields(struct trace *trace, struct evsel *evsel, · line 3242

Quoted 9 lines
tools/perf/builtin-trace.c +7 -0
@@ -3234,6 +3234,27 @@ static size_t trace__fprintf_tp_fields(struct trace *trace, struct evsel *evsel,
  		if (val == 0 && !trace->show_zeros && !arg->show_zero && arg->strtoul != STUL_BTF_TYPE)
  			continue;
+		/*
+		 * __probe_ip is implicitly added to bare dynamic probes.
+		 * Suppress it by default to avoid cluttering the output.
+		 * If verbose mode is enabled, ensure it is formatted as a
+		 * hexadecimal memory address rather than a signed integer.
+		 */
+		if (!strcmp(field->name, "__probe_ip")) {

I checked to see if it was possible to include the #define for this constant (FIELD_STRING_IP) in the tools folder, but there wasn't anything existing in include/ that fits nicely. Either way:

Quoted 1 line

Hi James,

Yes, unfortunately.

Thank you for the review.

Kind regards,

Signature · 1 line
Aaron Tomlin