perf trace: avoid garbage when not printing a trace event's arguments
authorBenjamin Peterson <benjamin@engflow.com>
Sun, 3 Nov 2024 20:48:16 +0000 (20:48 +0000)
committerNamhyung Kim <namhyung@kernel.org>
Wed, 6 Nov 2024 07:27:17 +0000 (23:27 -0800)
trace__fprintf_tp_fields may not print any tracepoint arguments. E.g., if the
argument values are all zero. Previously, this would result in a totally
uninitialized buffer being passed to fprintf, which could lead to garbage on the
console. Fix the problem by passing the number of initialized bytes fprintf.

Fixes: f11b2803bb88 ("perf trace: Allow choosing how to augment the tracepoint arguments")
Signed-off-by: Benjamin Peterson <benjamin@engflow.com>
Tested-by: Howard Chu <howardchu95@gmail.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Link: https://lore.kernel.org/r/20241103204816.7834-1-benjamin@engflow.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
tools/perf/builtin-trace.c

index c1e9e14..a661fbd 100644 (file)
@@ -3087,7 +3087,7 @@ static size_t trace__fprintf_tp_fields(struct trace *trace, struct evsel *evsel,
                printed += syscall_arg_fmt__scnprintf_val(arg, bf + printed, size - printed, &syscall_arg, val);
        }
 
-       return printed + fprintf(trace->output, "%s", bf);
+       return printed + fprintf(trace->output, "%.*s", (int)printed, bf);
 }
 
 static int trace__event_handler(struct trace *trace, struct evsel *evsel,