[FFmpeg-devel] [PATCH] [libavformat] Avoid integer overflow on start_time with skip_samples.

[Dale Curtis]

This patch can be abandoned in favor of the other one which uses
av_sat_add64().

- dale

On Thu, Apr 30, 2020 at 3:20 PM Dale Curtis <dalecurtis at chromium.org> wrote:

> I've sent a follow up patch set implementing saturating operations if
> that's something folks are interested in.
>
> - dale
>
> On Thu, Apr 30, 2020 at 2:18 PM Dale Curtis <dalecurtis at chromium.org>
> wrote:
>
>> That said, instead of aborting the operation, perhaps it'd make more
>> sense for library functions to be av_saturated_add(), av_saturated_sub()
>> which saturate to INT64_MIN/MAX.
>>
>> - dale
>>
>> On Thu, Apr 30, 2020 at 1:26 PM Dale Curtis <dalecurtis at chromium.org>
>> wrote:
>>
>>> Aside: This overflow check is used in quite a few places now. I wonder
>>> if it's worth having a function like the following:
>>>
>>> int64_t av_no_overflow_add(int64_t a, int64_t b) {
>>>   return (a > 0 ? b <= INT64_MAX - a : b >= INT64_MIN - a) ? a + b : a;
>>> }
>>>
>>> Better name suggestions welcome... av_maybe_add_ts?
>>>
>>> On Thu, Apr 30, 2020 at 1:17 PM Dale Curtis <dalecurtis at chromium.org>
>>> wrote:
>>>
>>>> This applies the same workaround used elsewhere in the file for handling
>>>> overflow of addition.
>>>>
>>>> Signed-off-by: Dale Curtis <dalecurtis at chromium.org>
>>>> ---
>>>>  libavformat/utils.c | 15 +++++++++++----
>>>>  1 file changed, 11 insertions(+), 4 deletions(-)
>>>>
>>>