[RFC][PATCH 4/4] perf/events: Use helper functions in event assignment to shrink macro size

[Steven Rostedt <rostedt@goodmis.org>]

[-- Attachment #1: 0004-perf-events-Use-helper-functions-in-event-assignment.patch --]
[-- Type: text/plain, Size: 5305 bytes --]

From: Steven Rostedt <srostedt@redhat.com>

The functions that assign the contents for the perf software events are
defined by the TRACE_EVENT() macros. Each event has its own unique
way to assign data to its buffer. When you have over 700 events,
that means there's 700 functions assigning data uniquely for each
event.

By making helper functions in the core kernel to do the work
instead, we can shrink the size of the kernel down a bit.

With a kernel configured with 707 events, the change in size was:

   text    data     bss     dec     hex filename
19966937        2594648 1945600 24507185        175f331 vmlinux-before
19924761        2594584 1945600 24464945        1754e31 vmlinux-after

That's a total of 42240 bytes, which comes down to 59 bytes per event.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
 include/linux/ftrace_event.h    |   13 +++++++++++++
 include/trace/ftrace.h          |   31 ++++++++-----------------------
 kernel/trace/trace_event_perf.c |   26 ++++++++++++++++++++++++++
 3 files changed, 47 insertions(+), 23 deletions(-)

diff --git a/include/linux/ftrace_event.h b/include/linux/ftrace_event.h
index b9f39d3..06c6d04 100644
--- a/include/linux/ftrace_event.h
+++ b/include/linux/ftrace_event.h
@@ -313,6 +313,19 @@ struct perf_event;
 
 DECLARE_PER_CPU(struct pt_regs, perf_trace_regs);
 
+struct perf_trace_event {
+	struct pt_regs		regs;
+	u64			addr;
+	u64			count;
+	int			entry_size;
+	int			rctx;
+};
+
+extern void *perf_trace_event_setup(struct ftrace_event_call *event_call, 
+				    struct perf_trace_event *pe);
+extern void perf_trace_event_submit(void *raw_data, struct ftrace_event_call *event_call,
+				    struct perf_trace_event *pe);
+
 extern int  perf_trace_init(struct perf_event *event);
 extern void perf_trace_destroy(struct perf_event *event);
 extern int  perf_trace_add(struct perf_event *event, int flags);
diff --git a/include/trace/ftrace.h b/include/trace/ftrace.h
index 8118451..a02d48d 100644
--- a/include/trace/ftrace.h
+++ b/include/trace/ftrace.h
@@ -677,10 +677,10 @@ __attribute__((section("_ftrace_events"))) *__event_##call = &event_##call
 #define __get_str(field) (char *)__get_dynamic_array(field)
 
 #undef __perf_addr
-#define __perf_addr(a) __addr = (a)
+#define __perf_addr(a) __pe.addr = (a)
 
 #undef __perf_count
-#define __perf_count(c) __count = (c)
+#define __perf_count(c) __pe.count = (c)
 
 #undef TP_perf_assign
 #define TP_perf_assign(args...) args
@@ -693,36 +693,21 @@ perf_trace_##call(void *__data, proto)					\
 	struct ftrace_event_call *event_call = __data;			\
 	struct ftrace_data_offsets_##call __maybe_unused __data_offsets;\
 	struct ftrace_raw_##call *entry;				\
-	struct pt_regs __regs;						\
-	u64 __addr = 0, __count = 1;					\
-	struct hlist_head *head;					\
-	int __entry_size;						\
+	struct perf_trace_event __pe;					\
 	int __data_size;						\
-	int rctx;							\
-									\
-	perf_fetch_caller_regs(&__regs);				\
 									\
 	__data_size = ftrace_get_offsets_##call(&__data_offsets, args); \
-	__entry_size = ALIGN(__data_size + sizeof(*entry) + sizeof(u32),\
-			     sizeof(u64));				\
-	__entry_size -= sizeof(u32);					\
-									\
-	if (WARN_ONCE(__entry_size > PERF_MAX_TRACE_SIZE,		\
-		      "profile buffer not large enough"))		\
-		return;							\
+	__pe.entry_size = __data_size + sizeof(*entry);			\
+	__pe.addr = 0;							\
+	__pe.count = 1;							\
 									\
-	entry = (struct ftrace_raw_##call *)perf_trace_buf_prepare(	\
-		__entry_size, event_call->event.type, &__regs, &rctx);	\
-	if (!entry)							\
-		return;							\
+	entry = perf_trace_event_setup(event_call, &__pe);		\
 									\
 	tstruct								\
 									\
 	{ assign; }							\
 									\
-	head = this_cpu_ptr(event_call->perf_events);			\
-	perf_trace_buf_submit(entry, __entry_size, rctx, __addr,	\
-		__count, &__regs, head);				\
+	perf_trace_event_submit(entry, event_call, &__pe);		\
 }
 
 /*
diff --git a/kernel/trace/trace_event_perf.c b/kernel/trace/trace_event_perf.c
index 9824419..75484aa 100644
--- a/kernel/trace/trace_event_perf.c
+++ b/kernel/trace/trace_event_perf.c
@@ -21,6 +21,32 @@ typedef typeof(unsigned long [PERF_MAX_TRACE_SIZE / sizeof(unsigned long)])
 /* Count the events in use (per event id, not per instance) */
 static int	total_ref_count;
 
+void *perf_trace_event_setup(struct ftrace_event_call *event_call, 
+			     struct perf_trace_event *pe)
+{
+	perf_fetch_caller_regs(&pe->regs);
+
+	pe->entry_size = ALIGN(pe->entry_size + sizeof(u32), sizeof(u64));
+	pe->entry_size -= sizeof(u32);
+
+	if (WARN_ONCE(pe->entry_size > PERF_MAX_TRACE_SIZE,
+		      "profile buffer not large enough"))
+		return NULL;
+
+	return perf_trace_buf_prepare(pe->entry_size,
+			event_call->event.type, &pe->regs, &pe->rctx);
+}
+
+void perf_trace_event_submit(void *raw_data, struct ftrace_event_call *event_call,
+			     struct perf_trace_event *pe)
+{
+	struct hlist_head *head;
+
+	head = this_cpu_ptr(event_call->perf_events);
+	perf_trace_buf_submit(raw_data, pe->entry_size, pe->rctx, pe->addr,
+			      pe->count, &pe->regs, head);
+}
+
 static int perf_trace_event_perm(struct ftrace_event_call *tp_event,
 				 struct perf_event *p_event)
 {
-- 
1.7.10.4