[mythtv-users] Compiling MythTV for PPC without Altivec?

[Eric Gilbert]

On 3/2/07, Henry Leinhos <henry-list at leinhos.com> wrote:
>
> Hi,
> I'm trying to compile mythtv (0.20 as well as svn) for my G3/PPC-750 desktop
> machine.  The standard configure options (i.e. --enable-altivec) gives me an
> Illegal Instruction error at startup.  If I configure with
>  --disable-altivec, I get the following assembler errors:
>
> ...
> make[2]: Entering directory `/home/henry/src/mythtv-svn/libs/libmythmpeg2'
> gcc -c -pipe -w -O3 -Wall -Wno-switch -DPIC -fPIC -fno-common -D_REENTRANT
>  -DUSING_FREEBOX -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -DPREFIX=\"/usr\"
>  -DLIBDIR=\"/usr/lib\" -DQT_NO_DEBUG -DQT_THREAD_SUPPORT -DQT_SHARED
>  -I/usr/lib/qt-3.3/mkspecs/default -I. -I/usr/include -I../../libs/libavcodec
>  -I/usr/lib/qt-3.3//include -o cpu_accel.o cpu_accel.c
> {standard input}: Assembler messages:
> {standard input}:59: Error: unsupported relocation against v0
> {standard input}:59: Error: unsupported relocation against v0
> {standard input}:59: Error: unsupported relocation against v0
> make[2]: *** [cpu_accel.o] Error 1
> make[2]: Leaving directory `/home/henry/src/mythtv-svn/libs/libmythmpeg2'
> make[1]: *** [sub-libmythmpeg2] Error 2
> make[1]: Leaving directory `/home/henry/src/mythtv-svn/libs'
>
>
> Are there any additional CFLAGS I need to add (-m options) to get this to
> compile without altivec?
>
> Thanks,
> Henry
>

I have Kuro/HG-WR that has a PowerPC chip without altivec and ran into
a similar compile issue with the .20-fixes branch and svn. I used
--disable-altivec for configure but I don't know of any specific
CFLAGS that you need since they are taken care of for me by the debian
package source I am using. Take a look at my partial writeup on the
Kuro Wiki http://www.kurobox.com/mwiki/index.php/Mythtv to get a patch
for the .20-fixes branch.  I based that patch on feedback from the
Kuro forum that pointed me to changeset 11203 here
http://cvs.mythtv.org/trac/changeset/11203. If you need to patch svn
you probably only need to apply the last hunk of the patch related to
cpu_state.c

Good luck,
Eric