perf: Support grouped event lists

perf: Support grouped event lists

[Andi Kleen]

As requested, this patchkit supports grouped event lists, by using
the stem as the section header. In addition it adds CPU override
support, which is useful for testing.

The updated Intel event lists with redundant entries removed are available in
git://git.kernel.org/pub/scm/linux/kernel/git/ak/linux-misc.git perf/intel-json-files-2

% perf list
...
icache:
  icache.hit                                        
       [Number of Instruction Cache, Streaming Buffer and Victim Cache Reads. both cacheable and noncacheable, including UC fetches]
  icache.ifetch_stall                               
       [Cycles where a code-fetch stalled due to L1 instruction-cache miss or an iTLB miss]
...
idq:
  idq.all_dsb_cycles_4_uops                         
       [Cycles Decode Stream Buffer (DSB) is delivering 4 Uops]
  idq.all_dsb_cycles_any_uops                       
       [Cycles Decode Stream Buffer (DSB) is delivering any Uop]
  idq.all_mite_cycles_4_uops                        
       [Cycles MITE is delivering 4 Uops]

[PATCH 1/2] perf, tools: Group alias perf list by section

[Andi Kleen]

From: Andi Kleen <ak@linux.intel.com>

The first first element in an hardware event name defines the general
area, usually the part of the micro architecture it is refering to.
Group the perf list output by these sections by adding section headers.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
 tools/perf/util/pmu.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c
index 578a537..ba50300 100644
--- a/tools/perf/util/pmu.c
+++ b/tools/perf/util/pmu.c
@@ -982,6 +982,8 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag)
 	struct pair *aliases;
 	int numdesc = 0;
 	int columns = pager_get_columns();
+	const char *last_section;
+	int last_section_len;
 
 	pmu = NULL;
 	len = 0;
@@ -1029,6 +1031,8 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag)
 		}
 	}
 	len = j;
+	last_section = NULL;
+	last_section_len = 0;
 	qsort(aliases, len, sizeof(struct pair), cmp_pair);
 	for (j = 0; j < len; j++) {
 		if (name_only) {
@@ -1036,8 +1040,21 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag)
 			continue;
 		}
 		if (aliases[j].desc && !quiet_flag) {
+			char *dot;
+			char *name = aliases[j].name;
+
 			if (numdesc++ == 0)
 				printf("\n");
+			dot = strchr(name, '.');
+			if (dot &&
+				(!last_section ||
+				 strncmp(last_section, name, last_section_len))) {
+				last_section_len = dot - name;
+				printf("%s%.*s:\n",
+						last_section ? "\n" : "",
+						last_section_len, name);
+				last_section = name;
+			}
 			printf("  %-50s\n", aliases[j].name);
 			printf("%*s", 8, "[");
 			wordwrap(aliases[j].desc, 8, columns, 0);
-- 
2.1.0

[PATCH 2/2] perf, tools: Add override support for event list CPUID

[Andi Kleen]

From: Andi Kleen <ak@linux.intel.com>

Add a PERF_CPUID variable to override the CPUID of the current CPU.
This is useful for testing, so that all event lists can be tested on a single
system.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
 tools/perf/util/pmu.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c
index ba50300..5500126 100644
--- a/tools/perf/util/pmu.c
+++ b/tools/perf/util/pmu.c
@@ -474,7 +474,9 @@ static int pmu_add_cpu_aliases(void *data)
 	struct pmu_event *pe;
 	char *cpuid;
 
-	cpuid = get_cpuid_str();
+	cpuid = getenv("PERF_CPUID");
+	if (!cpuid)
+		cpuid = get_cpuid_str();
 	if (!cpuid)
 		return 0;
 
-- 
2.1.0