[FFmpeg-devel] [PATCH] avformat/avcodec: Add DTS-UHD demuxer and parser, movenc support.

[Paul B Mahol]

On Wed, Jun 14, 2023 at 8:01 AM Paul B Mahol <onemda at gmail.com> wrote:

>
>
> On Wed, Jun 14, 2023 at 7:37 AM Paul B Mahol <onemda at gmail.com> wrote:
>
>>
>>
>> On Wed, Jun 14, 2023 at 2:00 AM Roy Funderburk <royffmpeg at funderburk.us>
>> wrote:
>>
>>>
>>> On 6/13/23 11:35 AM, Paul B Mahol wrote:
>>> > Doing allocation in probe?
>>> > Probing should be very fast.
>>>
>>> In line 143 of the avformat patch, memory allocation is removed from the
>>> probe
>>>
>>> >>+int dtsuhd_frame(DTSUHD *h, const uint8_t *data, size_t data_bytes,
>>> >>+                DTSUHDFrameInfo *fi, DTSUHDDescriptorInfo *di)
>>> >>+{
>>> >>+    gb = &h->gb;
>>> >>+    init_get_bits(gb, data, data_bytes * 8);
>>> > init_get_bits8, and check return code.
>>>
>>> In line 986 of the avcodec patch, changed to using init_get_bits8 and
>>> added return code check.
>>>
>>> >>+        fi->sync = h->is_sync_frame;
>>> >>+        fi->frame_bytes = h->frame_bytes;
>>> >>+        fi->sample_rate = h->sample_rate;
>>> >>+        fi->sample_count = (h->frame_duration * fi->sample_rate)
>>> /(h->clock_rate * fraction);
>>> >>+        fi->duration = (double)fi->sample_count / fi->sample_rate;
>>> >Please  no double  type.
>>> >Also make use of av_rescale.
>>>
>>> Around line 1108 of the avcodec patch, I found the duration was not
>>> needed and removed it.
>>>
>>>
>>> On 6/13/23 12:04 PM, Anton Khirnov wrote:
>>> > Also do note that sharing structs across libraries opens you to various
>>> > compatibility questions [2]. It might be easier to sidestep them by
>>> > having a function in libavcodec that accepts AVCodecParameters and
>>> fills
>>> > them according to the data, rather than pass codec-specific structs
>>> > between libavformat and libavcodec.
>>>
>>> In line 1061 of the avcodec patch, changed the function to:
>>>   int av_dtsuhd_frame(DTSUHD *h, const uint8_t *data, size_t data_bytes,
>>>         AVCodecParameters *codecpar, uint8_t **udts, int *udts_size)
>>>
>>>
>>>
>> Usually libavcodec code does use get_vlc2() for variable length codes,
>> instead of
>> usage of show_bits/skip_bits/get_bits.
>>
>>
> Also there is no reason to use int for elements in tables when max value
> can be lower.
> Current table reading/handling code should be completely rewritten to use
> get_vlc2().
> And tables split so length of codes use uint8_t type.
>

In parse_stream_params() function, there is skip_bits(gb, 36*get_bits1(gb))

but skip_bits can read usually max 25 bits.
Use skip_bits_long().



>
>>
>>
>>
>>>
>>> Thank you for reviewing this,
>>> -Roy_______________________________________________
>>> ffmpeg-devel mailing list
>>> ffmpeg-devel at ffmpeg.org
>>> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>>>
>>> To unsubscribe, visit link above, or email
>>> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".
>>>
>>