Displaying 3 results from an estimated 3 matches for "cur_page_gp".
2017 Dec 07
2
[PATCH 0/2] libopusfile int64 overflows
On Tue, Nov 28, 2017 at 3:22 PM, James Zern <jzern at google.com> wrote:
> On Mon, Nov 20, 2017 at 1:07 PM, James Zern <jzern at google.com> wrote:
>> Just an attempt to avoid overflows with an explicit check, I don't know if
>> there's a better way to identify corrupt input here.
>>
>> James Zern (2):
>> op_pcm_seek: fix int64 overflow
2017 Dec 07
1
[PATCH 0/2] libopusfile int64 overflows
...;> /*If we have samples to trim...*/
>> if(diff>0){
>> /*If we trimmed the entire packet, stop (the spec says
>> encoders
>
>
> For this one, the only way that durations[pi] - diff can overflow on the
> positive side is if cur_page_gp < prev_packet_gp (in fact, it must be very
> much less). So the question becomes what behavior we would like in that
> case. In the case where cur_page_gp < prev_packet_gp but diff -
> durations[pi] does not overflow, then we will have diff > durations[pi]
> after the subtracti...
2017 Dec 07
0
[PATCH 0/2] libopusfile int64 overflows
...} else diff=durations[pi]-diff;
> /*If we have samples to trim...*/
> if(diff>0){
> /*If we trimmed the entire packet, stop (the spec says encoders
For this one, the only way that durations[pi] - diff can overflow on the
positive side is if cur_page_gp < prev_packet_gp (in fact, it must be
very much less). So the question becomes what behavior we would like in
that case. In the case where cur_page_gp < prev_packet_gp but diff -
durations[pi] does not overflow, then we will have diff > durations[pi]
after the subtraction and we will t...