[FFmpeg-devel] PATCH Fix SSE core-dumps when decoding AAC on optimized ffmpeg with shared libraries compiled under Mingw (win32)

Ramiro Polla ramiro.polla
Wed Oct 22 23:54:25 CEST 2008

On Wed, Oct 22, 2008 at 7:24 PM, Art Clarke <aclarke at vlideshow.com> wrote:
> On Wed, Oct 22, 2008 at 2:09 PM, Ramiro Polla <ramiro.polla at gmail.com> wrote:
>>> I think it's best if we just boost the pagerank of this message =). Or
>>> even better wait for confirmation on
>>> https://roundup.mplayerhq.hu/roundup/ffmpeg/issue688 and rename it
>>> (and then close it as invalid).
>> btw, feel free to add that information to http://ffmpeg.arrozcru.org/wiki
>> but not as an unconditional "use -fno-common!!!". The best is to
>> identify where exactly the bug comes from (gcc version + binutils
>> version), why it happens and on which systems are known to be
>> affected, why this fixes it, and point to the gcc bug report...
> I added a post to the forum there and referred it back here since I
> already answered all the questions you asked (including what the
> workaround fixed) in the original e-mail.

>From your message, it's still not clear if it's a bug that only
depends on gcc, or if it also depends on binutils. Also the gcc
versions affected aren't checked for. Did you test with the official
2.95.3, 3.4.5, 4.2.1, and 4.3.0 from MinGW? Your patch would affect
all of them, and future versions as well.

And "latest mingw runtime" is irrelevant. I assume "win32 library"
means w32api, in that case it's also irrelevant. Such explanations can
mislead someone looking for the exact cause.

> I don't feel like reposting
> it, and that wouldn't help increase the page rank anyway :)  At this
> point, I'm handing this off.  I was trying to actually help the
> community by giving something back in what I thought was the right
> way.  Sigh.

You did help. Many people will end up reading this thread when they
encounter that problem.

>> It is welcome there even if the list adds up...
> Sorry, I had thought patches were welcome here, even if they end up
> not being accepted.

They are welcome. And then they are reviewed. And alternatives are
suggested. Sometimes to the point where no patching is needed anymore.

Ramiro Polla

More information about the ffmpeg-devel mailing list