tracing: Don't use out-of-sync va_list in event printing

[ Upstream commit 2ef75e9bd2c998f1c6f6f23a3744136105ddefd5 ]

If trace_seq becomes full, trace_seq_vprintf() no longer consumes
arguments from va_list, making va_list out of sync with format
processing by trace_check_vprintf().

This causes va_arg() in trace_check_vprintf() to return wrong
positional argument, which results into a WARN_ON_ONCE() hit.

ftrace_stress_test from LTP triggers this situation.

Fix it by explicitly avoiding further use if va_list at the point
when it's consistency can no longer be guaranteed.

Link: https://lkml.kernel.org/r/20211118145516.13219-1-nikita.yushchenko@virtuozzo.com
Signed-off-by: Nikita Yushchenko <nikita.yushchenko@virtuozzo.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 5e452dd57af01700a25681e43f40d386ffbe085f..18db461f77cdf773f4f6ae4d0450b5ccda1c33fc 100644 (file)
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -3836,6 +3836,18 @@ void trace_check_vprintf(struct trace_iterator *iter, const char *fmt,
                iter->fmt[i] = '\0';
                trace_seq_vprintf(&iter->seq, iter->fmt, ap);
 
+               /*
+                * If iter->seq is full, the above call no longer guarantees
+                * that ap is in sync with fmt processing, and further calls
+                * to va_arg() can return wrong positional arguments.
+                *
+                * Ensure that ap is no longer used in this case.
+                */
+               if (iter->seq.full) {
+                       p = "";
+                       break;
+               }
+
                if (star)
                        len = va_arg(ap, int);