[FFmpeg-devel] [PATCH 2/2] avformat/rpl: Fix check for negative values

[Rémi Denis-Courmont]

Hi,

Le 18 novembre 2024 05:09:11 GMT+02:00, Michael Niedermayer <michael at niedermayer.cc> a écrit :
>Fixes: signed integer overflow: 10 * -1923267925333400000 cannot be represented in type 'int64_t' (aka 'long')
>Fixes: 378891963/clusterfuzz-testcase-minimized-fuzzer_loadfile_direct-5714338935013376
>Found-by: ossfuzz
>Reported-by: Kacper Michajlow <kasper93 at gmail.com>
>Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
>---
> libavformat/rpl.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
>diff --git a/libavformat/rpl.c b/libavformat/rpl.c
>index e971d0588f3..b30d769efb2 100644
>--- a/libavformat/rpl.c
>+++ b/libavformat/rpl.c
>@@ -102,7 +102,7 @@ static AVRational read_fps(const char* line, int* error)
>         line++;
>     for (; *line>='0' && *line<='9'; line++) {
>         // Truncate any numerator too large to fit into an int64_t
>-        if (num > (INT64_MAX - 9) / 10 || den > INT64_MAX / 10)
>+        if (num > (INT64_MAX - 9) / 10ULL || den > INT64_MAX / 10ULL)
>             break;
>         num  = 10 * num + (*line - '0');
>         den *= 10;

We have ckd_mul() for this, AFAIK. Easier to read, easier to write and also easier for the compiler to optimise correctly.