lavf: fix signed overflow in avformat_find_stream_info()
authorMans Rullgard <mans@mansr.com>
Sat, 8 Oct 2011 01:06:26 +0000 (02:06 +0100)
committerMichael Niedermayer <michaelni@gmx.at>
Fri, 4 Nov 2011 00:00:18 +0000 (01:00 +0100)
On the first iteration through this code, last_dts is always
INT64_MIN (AV_NOPTS_VALUE) and the subtraction overflows in
an invalid manner.  Although the result is only used if the
input values are valid, performing the subtraction is still
not allowed in a strict environment.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit a31e9f68a426f634e002282885c6c2eb1bfbea44)

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
libavformat/utils.c

index b15582a..400d968 100644 (file)
@@ -2396,9 +2396,9 @@ int av_find_stream_info(AVFormatContext *ic)
         }
         {
             int64_t last = st->info->last_dts;
-            int64_t duration= pkt->dts - last;
 
-            if(pkt->dts != AV_NOPTS_VALUE && last != AV_NOPTS_VALUE && duration>0){
+            if(pkt->dts != AV_NOPTS_VALUE && last != AV_NOPTS_VALUE && pkt->dts > last){
+                int64_t duration= pkt->dts - last;
                 double dur= duration * av_q2d(st->time_base);
 
 //                if(st->codec->codec_type == AVMEDIA_TYPE_VIDEO)