perf_counter: optionally provide the pid/tid of the sampled task

Allow cpu wide counters to profile userspace by providing what process
the sample belongs to.

This raises the first issue with the output type, lots of these
options: group, tid, callchain, etc.. are non-exclusive and could be
combined, suggesting a bitfield.

However, things like the mmap() data stream doesn't fit in that.

How to split the type field...

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Arjan van de Ven <arjan@infradead.org>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Orig-LKML-Reference: <20090325113317.013775235@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

include/linux/perf_counter.h

@@ -127,8 +127,9 @@ struct perf_counter_hw_event {
 				exclude_kernel :  1, /* ditto kernel          */
 				exclude_hv     :  1, /* ditto hypervisor      */
 				exclude_idle   :  1, /* don't count when idle */
+				include_tid    :  1, /* include the tid */
 
-				__reserved_1   : 55;
+				__reserved_1   : 54;
 
 	__u32			extra_config_len;
 	__u32			__reserved_4;
@@ -164,6 +165,8 @@ struct perf_event_header {
 enum perf_event_type {
 	PERF_EVENT_IP		= 0,
 	PERF_EVENT_GROUP	= 1,
+
+	__PERF_EVENT_TID	= 0x100,
 };
 
 #ifdef __KERNEL__

kernel/perf_counter.c

@@ -1528,16 +1528,30 @@ static int perf_output_write(struct perf_counter *counter, int nmi,
 static void perf_output_simple(struct perf_counter *counter,
 			       int nmi, struct pt_regs *regs)
 {
+	unsigned int size;
 	struct {
 		struct perf_event_header header;
 		u64 ip;
+		u32 pid, tid;
 	} event;
 
 	event.header.type = PERF_EVENT_IP;
-	event.header.size = sizeof(event);
 	event.ip = instruction_pointer(regs);
 
-	perf_output_write(counter, nmi, &event, sizeof(event));
+	size = sizeof(event);
+
+	if (counter->hw_event.include_tid) {
+		/* namespace issues */
+		event.pid = current->group_leader->pid;
+		event.tid = current->pid;
+
+		event.header.type |= __PERF_EVENT_TID;
+	} else
+		size -= sizeof(u64);
+
+	event.header.size = size;
+
+	perf_output_write(counter, nmi, &event, size);
 }
 
 static void perf_output_group(struct perf_counter *counter, int nmi)