Cases rec->counter == {0, 1} are checked already.

While x * (x - 1) * 1000 = 0 have many solutions greater than 1 for both modulo 2^32 and 2^64, that is not the case for x * (x - 1) = 0, so split division into two.

It is not scary in practice because mod 2^64 solutions are huge and minimal mod 2^32 solution is 30-bit number.

Cc: stable@vger.kernel.org Fixes: e31f7939c1c27 ("ftrace: Avoid potential division by zero in function profiler") Signed-off-by: Nikolay Kuratov kniv@yandex-team.ru --- kernel/trace/ftrace.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 728ecda6e8d4..e1c05c4c29c2 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -570,12 +570,12 @@ static int function_stat_show(struct seq_file *m, void *v) stddev = rec->counter * rec->time_squared - rec->time * rec->time;

+ stddev = div64_ul(stddev, rec->counter * (rec->counter - 1)); /* * Divide only 1000 for ns^2 -> us^2 conversion. * trace_print_graph_duration will divide 1000 again. */ - stddev = div64_ul(stddev, - rec->counter * (rec->counter - 1) * 1000); + stddev = div64_ul(stddev, 1000); }

trace_seq_init(&s);