[for-next][PATCH 29/29] tracing: Comment why cond_snapshot is checked outside of max_lock protection
From: Steven Rostedt
Date: Wed Feb 20 2019 - 13:38:37 EST
From: "Steven Rostedt (VMware)" <rostedt@xxxxxxxxxxx>
Before setting tr->cond_snapshot, it must be NULL before it can be updated.
It can go to NULL when a trace event hist trigger is created or removed, and
can only be modified under the max_lock spin lock. But because it can only
be set to something other than NULL under both the max_lock spin lock as
well as the trace_types_lock, we can perform the check if it is not NULL
only under the trace_types_lock and fail out without having to grab the
max_lock spin lock.
This is very subtle, and deserves a comment.
Signed-off-by: Steven Rostedt (VMware) <rostedt@xxxxxxxxxxx>
---
kernel/trace/trace.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 0460cc0f28fd..2cf3c747a357 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -1116,6 +1116,14 @@ int tracing_snapshot_cond_enable(struct trace_array *tr, void *cond_data,
goto fail_unlock;
}
+ /*
+ * The cond_snapshot can only change to NULL without the
+ * trace_types_lock. We don't care if we race with it going
+ * to NULL, but we want to make sure that it's not set to
+ * something other than NULL when we get here, which we can
+ * do safely with only holding the trace_types_lock and not
+ * having to take the max_lock.
+ */
if (tr->cond_snapshot) {
ret = -EBUSY;
goto fail_unlock;
--
2.20.1