Filip Roséen authored 8 years ago and Jean-Baptiste Kempf committed 8 years ago

Fix heap overflow due to invalid header data

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Better handles DVB extension for impaired and provides
better then description and language of the stream.

before -> qad (none)
after  -> lang (Descriptive audio for hearing impaired)

potential read overflow

fmt size is saved but first call is always 32x32

will mostly be overriden with this one

As a side note, refcounts are now initialized with 0, which is expected
so a "smart" pointer can increment the refcount itself upon acquiring
the pointer.

Salah-Eddin Shaban authored 8 years ago and Hugo Beauzée-Luyssen committed 8 years ago

Signed-off-by: Hugo Beauzée-Luyssen <hugo@beauzee.fr>

Filip Roséen authored 8 years ago and Jean-Baptiste Kempf committed 8 years ago

Given that the previous implementation stored the size of the current
chunk-payload in an uint32_t, it would potentially overflow when
adding the size of the chunk header and conditional padding.

These changes fixes the previously described by storing the
chunk-total size in a larger integer type, as well as making sure that
we do not pass a too big of a value to vlc_stream_Read (that would
cause problems on 32bit platforms).

Fixes #17562

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Filip Roséen authored 8 years ago and Jean-Baptiste Kempf committed 8 years ago

The previous implementation assumed that a 32-bit integer would fit
in an `int`, something which is not guaranteed and might cause an
integer overflow.

These changes changes the declared type of the relevant variables, by
also making a slight amount of clean-up on the affected paths, such
as:

  - merging declaration and initialization
  - fixing redundant if-conditions

fixes #17593

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

Limiting the size of the art-container does not make much sense, especially not
when the maximum height prior to these changes was set in pixels.

fixes #17749

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

See added documentation.

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

The functionality remains the same, the implementation should however
be easier to follow than what we had previously.

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

These changes moves logic from within InputDemuxNew, so that explicit
demux selection is in one place.

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

Given that the relevant function is only used from within
src/input/input.c, and that no internal functions of demux.c are
required within its implementation, these changes simply move the
function to a more accurate location.

It will also simplify future development in terms of MRL-extra data
handling.

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

Function used by stream-extractor modules to make it easy to create a
relative MRL for an entity within the data handled by the
stream-extractor.

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

The documentation associated with these changes should explain what
the function does in detail, and why it is necessary. In short the
function simply tries to attach a stream-extractor to a stream_t.

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

These changes introduces a new module type named stream_extractor. The
added documentation should explain when such module should be used,
but in short it allows for extraction of data within a stream,
effectively resulting in a new stream that refers to the extracted
data.

Interaction with the stream-extractor shall never happen directly,
instead the module-backend is written in a way so that it exposes a
stream_t to the public.

    [ access ] -> [ stream_t ] -> [ stream consumer ]
                   '- [ stream extractor ]

Future changes are necessary in order to make modules of this type
usable in practice, but has been split into individual commits so that
the changes are easier to follow.

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

There has never been any documentation related to what a MRL actually
is, these changes addresses that by (hopefully) describing the entity
in a way that makes it easier for future developers to treat them
correctly.

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

It is not always desirable to have documentation mixed with code.
This, as an example, applies to documentation about a certain feature
spanning several headers; or example implementations to aid future
developers.

These changes creates a designated place for such documentation.

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

Filip Roséen authored 8 years ago and Thomas Guillem committed 8 years ago

The function does not open MRLs, as correctly described in its
documentation, as such it is rather unfortunate that its name uses MRL
instead of URL (especially given that it cannot handle MRLs).

These changes are simply renaming all occurrences of the function, so
that the behavior of the function is properly reflected by its name.

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

NSNetServiceBrowser and NSNetService do not properly release the
delegates, see https://crbug.com/657495 and
https://openradar.appspot.com/28943305