MJPEG Tools Svn

Fall back correctly if gcc revision doesn't yet know about
-mcpu=athlon64

Fix-up the chroma X-tag fix-up...

The basic idea is:
o If feature_level == 0, leave any XYSCSS tags alone (since they
might be handled by the program itself).
o If feature_level > 0, process and remove a XYSCSS tag, but let
the real 'C' tag take precedence in case both tags exist (and
warn if there are both and they don't match).

Added the patch from Nicholas Redgrave

lavaddwave can now do:
Attached is a patch for lavaddwav that adds a parameter options facility.
There is only one option at present and that is to set the maximum size of
each output file in MB.

dualprime is working so remove mention of the artifacting bug.

One more fix, because we only want ot split AVI's and not quicktime movies

linux doesn't (apaprently) require that mutex attribute structures be
initialized before use. OS/X _does_. It is necessary to "create" or "initialize"
a mutex attribute with:
pthread_mutexattr_init(&mu_attr);
before setting flags or using the mutex.

Updated to reflect latest changes.

nasm is no longer needed (no .s files remaining).

The last of the .s files is gone (Thanks Gert!). No need for special rules
in mpeg2enc/Makefile.am or for strip_fPIC.sh to exist.

Complete the error checking and adding diagnostic message on mutex lock/unlock
failure. So far this has produced (on OS/X) the following when mpeg2enc
is run with other than -M 0:

INFO: [mpeg2enc] Creating worker thread
2 pthread_mutex_lock=22

That message comes from channel.hh. error 22 is EINVAL which seems to point
at the 'atomic' mutex not being initialized. The manpage says:
[EINVAL] The value specified by mutex is invalid.