perf_counter: add PERF_RECORD_CPU
Allow recording the CPU number the event was generated on.
RFC: this leaves a u32 as reserved, should we fill in the
node_id() there, or leave this open for future extention,
as userspace can already easily do the cpu->node mapping
if needed.
[ Impact: extend perfcounter output record format ]
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
LKML-Reference: <20090508170029.008627711@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h
index 0e6303d..614f921 100644
--- a/include/linux/perf_counter.h
+++ b/include/linux/perf_counter.h
@@ -105,6 +105,7 @@
PERF_RECORD_GROUP = 1U << 4,
PERF_RECORD_CALLCHAIN = 1U << 5,
PERF_RECORD_CONFIG = 1U << 6,
+ PERF_RECORD_CPU = 1U << 7,
};
/*
@@ -260,6 +261,7 @@
* { u64 time; } && PERF_RECORD_TIME
* { u64 addr; } && PERF_RECORD_ADDR
* { u64 config; } && PERF_RECORD_CONFIG
+ * { u32 cpu, res; } && PERF_RECORD_CPU
*
* { u64 nr;
* { u64 event, val; } cnt[nr]; } && PERF_RECORD_GROUP
diff --git a/kernel/perf_counter.c b/kernel/perf_counter.c
index c615f52..d850a1f 100644
--- a/kernel/perf_counter.c
+++ b/kernel/perf_counter.c
@@ -1956,6 +1956,9 @@
struct perf_callchain_entry *callchain = NULL;
int callchain_size = 0;
u64 time;
+ struct {
+ u32 cpu, reserved;
+ } cpu_entry;
header.type = 0;
header.size = sizeof(header);
@@ -1999,6 +2002,13 @@
header.size += sizeof(u64);
}
+ if (record_type & PERF_RECORD_CPU) {
+ header.type |= PERF_RECORD_CPU;
+ header.size += sizeof(cpu_entry);
+
+ cpu_entry.cpu = raw_smp_processor_id();
+ }
+
if (record_type & PERF_RECORD_GROUP) {
header.type |= PERF_RECORD_GROUP;
header.size += sizeof(u64) +
@@ -2037,6 +2047,9 @@
if (record_type & PERF_RECORD_CONFIG)
perf_output_put(&handle, counter->hw_event.config);
+ if (record_type & PERF_RECORD_CPU)
+ perf_output_put(&handle, cpu_entry);
+
/*
* XXX PERF_RECORD_GROUP vs inherited counters seems difficult.
*/
