[PATCH] tracing: Show the preempt count of when the event was called
From: Steven Rostedt (Red Hat)
Date: Fri Jun 17 2016 - 17:40:58 EST
Because tracepoint callbacks are done with preemption enabled, the trace
events are always called with preempt disable due to the
rcu_read_lock_sched_notrace() in __DO_TRACE(). This causes the preempt count
shown in the recorded trace event to be inaccurate. It is always one more
that what the preempt_count was when the tracepoint was called.
If CONFIG_PREEMPT is enabled, subtract 1 from the preempt_count before
recording it in the trace buffer.
Link: http://lkml.kernel.org/r/20160525132537.GA10808@xxxxxxxxxxxxx
Reported-by: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
Signed-off-by: Steven Rostedt <rostedt@xxxxxxxxxxx>
---
kernel/trace/trace_events.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c
index fd449eb138cf..03c0a48c3ac4 100644
--- a/kernel/trace/trace_events.c
+++ b/kernel/trace/trace_events.c
@@ -261,6 +261,14 @@ void *trace_event_buffer_reserve(struct trace_event_buffer *fbuffer,
local_save_flags(fbuffer->flags);
fbuffer->pc = preempt_count();
+ /*
+ * If CONFIG_PREEMPT is enabled, then the tracepoint itself disables
+ * preemption (adding one to the preempt_count). Since we are
+ * interested in the preempt_count at the time the tracepoint was
+ * hit, we need to subtract one to offset the increment.
+ */
+ if (IS_ENABLED(CONFIG_PREEMPT))
+ fbuffer->pc--;
fbuffer->trace_file = trace_file;
fbuffer->event =
--
1.9.3