The bitstream buffers are now private and freed by libavcodec. For
backward compatibility, the hold bitstream buffer pointer is left NULL
(applications were supposed to av_freep() it).
Signed-off-by: NAnton Khirnov <anton@khirnov.net>

Signed-off-by: NAnton Khirnov <anton@khirnov.net>

Signed-off-by: NAnton Khirnov <anton@khirnov.net>

Signed-off-by: NAnton Khirnov <anton@khirnov.net>

Signed-off-by: NAnton Khirnov <anton@khirnov.net>

It limits the duration of the data read from a given input.

Insert (a)trim filters on the corresponding inputs, so the extra frames
are decoded and discarded.

Using -ss 0 to drop frames with negative timestamps is a perfectly valid
use case.

This will allow using the same code when one filter can be inserted both
on input and output.

If any samples are still buffered when request_frame returns EOF, they
won't be returned currently.

Fixes an invalid free.

Bug-id: 448

Signed-off-by: NAnton Khirnov <anton@khirnov.net>

Incidentally also remove a warning.

Some code paths can call it with invalid length.

CC: libav-stable@libav.org

'filter' is too generic, since the functions in question deal with
several filters.

Also add const qualifiers to those variables.

This function should never modify the filter.

The actual usage of the pragma is within functions and older versions
of gcc only support the pragma outside of functions.

Set output_picture_number in H.264 parser.
Signed-off-by: NDiego Biurrun <diego@biurrun.de>

Signed-off-by: NMartin Storsjö <martin@martin.st>

Signed-off-by: NDiego Biurrun <diego@biurrun.de>

Signed-off-by: NDiego Biurrun <diego@biurrun.de>

Signed-off-by: NMartin Storsjö <martin@martin.st>

Signed-off-by: NAnton Khirnov <anton@khirnov.net>

A separate rtcp port can already be set when opening the rtp
protocol normally, but when doing port setup as in RTSP (where
we first need to open the local ports and pass them to the peer,
and only then receive the remote peer port numbers), we didn't
check the same url parameter as in the normal open routine.
Signed-off-by: NMartin Storsjö <martin@martin.st>

I doubt that anyone ever would try to send a 1 byte packet
via the RTP protocol, but check just in case - it shouldn't
crash at least.
Signed-off-by: NMartin Storsjö <martin@martin.st>

Signed-off-by: NMartin Storsjö <martin@martin.st>

This seems to have been missed in 636ced8e.
Signed-off-by: NMartin Storsjö <martin@martin.st>

AVIOContext has got an av_class member that only gets set if
opening the context using avio_open2, but not if allocating a
custom IO context. A caller that wants to read AVOptions from
an AVIOContext (recursively using AV_OPT_SEARCH_CHILDREN) may
not know if the AVIOContext actually has got a class set or not.
Signed-off-by: NMartin Storsjö <martin@martin.st>

Mixing these with the rtp depacketizer functions in rtpdec.h is
no good.
Signed-off-by: NMartin Storsjö <martin@martin.st>

This avoids having duplicate code where only the fd parameter
differs.
Signed-off-by: NMartin Storsjö <martin@martin.st>