On 27/10/2007, Pav Lucistnik <p...@freebsd.org> wrote:
> It have troubles compiling assembly on amd64 in h264.c:
Could you do me a favour and have a quick look into this? I do not
have an amd64 machine running FreeBSD currently. There were a few
postings to ffmpeg mailing list in August with this exact problem, but
nobody came up with a solution:
http://lists.mplayerhq.hu/pipermail/ffmpeg-user/2007-August/010874.html
http://lists.mplayerhq.hu/pipermail/ffmpeg-user/2007-August/011084.html
However, since the ffmpeg mailing list is not filling up with messages
like this right now, maybe a more recent version compiles.
I'd suggest to replace h264.c and/or the included headers by
up-to-date versions from svn.
Thanks in advance,
Riggs
> here is an updated tarball that hopefully fixes the dvd and livemedia
> problems that occurred with the last preview:
http://www.rrr.de/~riggs/mplayer/m20071111.tar.bz2
This one should fix the bt848 problem. (At least it compiles with the
bt848 module now.)
Dear bt848 owners, please confirm.
Please don't hesitate to send reports and patches.
TIA,
Riggs
That assembly comes from cabac.h file, which is identical in SVN to the
version in port. I also tried replacing h264.c and h264.h, with no
visible change.
--
Pav Lucistnik <p...@oook.cz>
<p...@FreeBSD.org>
I have a quantum car. Every time I look at the speedometer I get lost...
> FYI, I built it on amd64 -CURRENT with attached patches. Inline
> assembly in cabac.h generates invalid instructions for amd64. The
> Makefile patch disables it. Also, sysdep/pci_x86.c works fine on
> amd64 and linker fails without it.
Groove, the libavcodec-Makefile fixed build on my amd64/6-stable too.
> Compiler did not align stack variables. Libavcodec has been miscompiled
> and may be very slow or crash. This is not a bug in libavcodec,
> but in the compiler. You may try recompiling using gcc >= 4.2.
> Do not report crashes to FFmpeg developers.
Scary :)
Fixes I noticed since rc1:
- DTS audio on certain HD DVD content now decoded properly
- WMV/VC-1 video problems fixed
Can't test win32 codecs audio sync problem here (amd64 - no wine)
Generally, I like it.
--
Pav Lucistnik <p...@oook.cz>
<p...@FreeBSD.org>
I want to earn the right to be obnoxious before I'm too bitter to really
enjoy it. -- Able
> FYI, I built it on amd64 -CURRENT with attached patches. Inline
> assembly in cabac.h generates invalid instructions for amd64. The
> Makefile patch disables it. Also, sysdep/pci_x86.c works fine on
> amd64 and linker fails without it.
This is great! If you don't have objections, I am going to include
these fixes and prepare a new tarball for testing on the weekend.
Riggs
FYI, I built it on amd64 -CURRENT with attached patches. Inline
assembly in cabac.h generates invalid instructions for amd64. The
Makefile patch disables it. Also, sysdep/pci_x86.c works fine on
amd64 and linker fails without it.
Jung-uk Kim
http://www.rrr.de/~riggs/mplayer/m20071118.tar.bz2
includes all the suggestions and fixes I have received so far,
including Jung-uk Kim's solution to enable amd64 builds again.
Please test and comment.
TIA,
Riggs
Hi,
Works fine on -current/amd64.
CFLAGS=-DNDEBUG -O2 -pipe -fno-strict-aliasing
WITHOUT_DEBUG=yes
WITHOUT_RTCPU=yes
WITHOUT_MENCODER=yes
Best Regards,
- --
Hyogeol Lee
hyoge...@gmail.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.4 (FreeBSD)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFHQFcQ1D7/GiH6QSERCpQ+AJ0WN2dnOtIvpoxAHS6guEM3TvLkTgCfToau
OUe4+9j38KTc+OTaDjUxWwM=
=tGyV
-----END PGP SIGNATURE-----
> http://www.rrr.de/~riggs/mplayer/m20071118.tar.bz2
>
> includes all the suggestions and fixes I have received so far,
> including Jung-uk Kim's solution to enable amd64 builds again.
Looks good here.
--
Pav Lucistnik <p...@oook.cz>
<p...@FreeBSD.org>
KDE is for the techies who feel they can't be productive without being
able to control the exact amount of bevel in their window frames in 2%
increments.
http://www.rrr.de/~riggs/mplayer/m20071119.tar.bz2
is mostly identical to yesterday's version, but contains Eugene's
patch to enable AMR audio codec support.
I believe we can send-pr this version unless nobody discovers showstoppers.
Riggs