[FFmpeg-devel] lurking bugs in the mmx-related assembler code (?)

[u-h8zb at aetey.se]

On Sun, Oct 02, 2016 at 02:25:30AM +0200, Michael Niedermayer wrote:
> If all else fails we could add a emms call behind #if in
> av_malloc() and detect affected libcs but thats IMHO a ugly hack
> but better than declaring "non support"

Unfortunately I see an additional reason for avoiding such kind of a hack.

This could only work when the same party governs the whole life
cycle of the library and each related binary, both at library compile
time, binary link time and/or run time.

This can not be taken for granted.

At compile time we do not know which implementation will be behind the
ABI at run time (or at static link time, which also can be separate in
time and place from the compile).

An example: a binary might suddenly break when a system-wide C library
has been upgraded, bringing in slightly different implementation details.

As another example, musl has in fact some degree of binary compatibility
with glibc, as a shared library. People are sometimes using this
for running prebuild binaries. This would break all such binaries based
on ffmpeg.

Regards,
Rune