mirror of
https://gitlab.freedesktop.org/gstreamer/gstreamer.git
synced 2024-09-08 13:18:52 +00:00
80 lines
4.1 KiB
Text
80 lines
4.1 KiB
Text
OUTDATED
|
|
--------
|
|
|
|
The point of the metadata is to provide some context for each buffer. In
|
|
the case of audio data, for instance, it would provide the samplerate, bit
|
|
depth, and channel count.
|
|
|
|
The trick is that there may be multiple types of metadata ganged onto a
|
|
single buffer. This is why they're going to be a GList. This does mean
|
|
extra overhead in all cases, but I think it's minimal. The GList type
|
|
uses a chunk allocater so we're not wasting too much memory or time when
|
|
adding to the list.
|
|
|
|
The trick is dealing with these structs as they pass through a pipeline,
|
|
since they have potentially different mutability properties. For
|
|
instance, if you've got a mp3 decoder connected to a tee, which sends the
|
|
buffers off to both the decoder and a spectrum analyzer (and then a
|
|
visualization element). The preferred setup would be where every time a
|
|
audio/raw metadata comes down the pipe (indicating a potential change in
|
|
audio format), the audiosink and spectrum would just save off pointers.
|
|
|
|
So when exactly does this metadata go away (deallocated)? Well, that
|
|
means metadata has to be refcounted. But that gets rather hairy. OK, in
|
|
the simple case you create a metadata struct, it comes with refcount set
|
|
to 1. You pass it through, it stays one, eventually someone drops the
|
|
last reference on the buffer it's tied to, you free the metadata too.
|
|
Easy. What if you tee? You could go through and for every metadata in
|
|
the buffer, increment the refcount by the same as the buffer. So in the
|
|
above case (tee'd), the audiosink and spectrum would get the buffer with a
|
|
refcount of 2, and it'd have a metadata with refcount 2. Do they ref it
|
|
each themselves, then unref the buffer? Or do they remove the metadata?
|
|
Removing the metadata would require a buffer CoW, which would suck, so
|
|
yes, they'd just ref the metadata.
|
|
|
|
But.... what if they're all in different threads? Then we're off into
|
|
the magical world of mutexes. Everything with a refcount in a threaded
|
|
world must be mutexed, else you can do atomic increment and atomic
|
|
dec&test. Can this be done from C easily? Perhaps it needs to be found
|
|
from kernel includes via autoconf?
|
|
|
|
|
|
|
|
|
|
The goal in designing the way metadata will be defined and used is to keep
|
|
it as simple as possible. The basis for accomplishing this is the fact
|
|
that in order to actually use (rather than just pass) the metadata, you
|
|
have to know what the fields are, which means you have to have compiled in
|
|
support for that metadata at build time. Therefore, if you're using
|
|
metadata, you must have build-time access to the necessary include file
|
|
that defines it.
|
|
|
|
So, given that you've got an include file, it would be nice if the whole
|
|
thing could be contained there. This would limit the need to be linked
|
|
against something, or have load-time requirements as to that has to be
|
|
loaded before you are.
|
|
|
|
Given that really all metadata is is a region of memory of a given size
|
|
with a certain signature, this isn't all that hard. First you lay out the
|
|
struct that defines the metadata. Then you set up #defines that expand to
|
|
the size of the struct in question, as well as the four-cc code that
|
|
defines the type.
|
|
|
|
The work is done by a few #defines, a la the #defines used in all Gtk
|
|
objects. The first is a NEW() method that allocates the memory for the
|
|
metadata and fills in all the normal fields (type, size, utility
|
|
functions). Because of the way it's defined (as a #define, no less),
|
|
you'll have to invoke it as META_NEW(meta), since it can't return()
|
|
anything.
|
|
|
|
Another #define will check to make sure a meta is indeed that type by
|
|
verifying the type code and size. Theoretically, meta types can overlap
|
|
with the same fourcc code, as long as they have different sizes. But I
|
|
probably ought to have a global public registry so people writing things
|
|
don't conflict. MSFT got that right, at least.
|
|
|
|
So, a hairy problem is what to do when there are utility functions
|
|
associated with one of these things. One option is to not bother with
|
|
them. This is very likely a possible solution, since metadata is supposed
|
|
to be flat memory of a given size. Not much to do to either free or copy
|
|
it, is there?
|