On 05.01.2016 11:46, Michael Niedermayer wrote:
> On Tue, Jan 05, 2016 at 12:44:40AM +0100, Andreas Cadhalpun wrote:
>> Otherwise this can have some surprising effects (crashes), so let's
>> better not allow it.
>>
>> Signed-off-by: Andreas Cadhalpun <andreas.cadhal...@googlemail.com>
>> ---
>> libavcodec/parser.c | 11 +++++++++++
>> 1 file changed, 11 insertions(+)
>>
>> diff --git a/libavcodec/parser.c b/libavcodec/parser.c
>> index 2809158..1f38edb 100644
>> --- a/libavcodec/parser.c
>> +++ b/libavcodec/parser.c
>> @@ -141,6 +141,17 @@ int av_parser_parse2(AVCodecParserContext *s,
>> AVCodecContext *avctx,
>> int index, i;
>> uint8_t dummy_buf[AV_INPUT_BUFFER_PADDING_SIZE];
>>
>> + if (avctx->codec_id != s->parser->codec_ids[0] &&
>> + avctx->codec_id != s->parser->codec_ids[1] &&
>> + avctx->codec_id != s->parser->codec_ids[2] &&
>> + avctx->codec_id != s->parser->codec_ids[3] &&
>> + avctx->codec_id != s->parser->codec_ids[4]) {
>> + av_log(avctx, AV_LOG_ERROR,
>> + "The parser doesn't match the codec %s.\n",
>> + avcodec_get_name(avctx->codec_id));
>> + return buf_size;
>> + }
>
> does it also work to check if a parser is set when the codec id
> is changed ? as in below
That wouldn't work, as the codec id wasn't changed in force_codec_ids,
but in the API using program.
To reiterate, the problematic steps were:
* call avformat_find_stream_info, which detects a codec and initializes
a parser for it
* afterwards change the codec id in the API using program, so it
doesn't match with the parser
Thus I think the only reliable way to detect this is a check in
av_parser_parse2.
> (that would avoid doing 5 extra checks per packet)
Parsers aren't that speed critical code, I think.
Best regards,
Andreas
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
