[pkg-mad-maintainers] Bug#425112: Bug#425112: Bug#425112: madplay on sgi mips64 big endian

[Kurt Roeckx]

On Sat, May 19, 2007 at 10:10:28AM +0100, Sam Clegg wrote:
> On Sat, 2007-05-19 at 10:15 +0200, Gürkan Sengün wrote:
> > Package: madplay
> > Version: 0.15.2b-4
> > Severity: important
> > 
> > madplay somevalid.mp3 just plays noise on my
> > indigo2. mpg321 gets it right.
> > 
> > $ uname -a
> > Linux mips 2.6.20-1-r4k-ip22 #1 Tue Apr 24 23:33:09 UTC 2007 mips64 
> > GNU/Linux
> > 
> 
> Sounds like a potential endianess or wordsize issue.  Probably for
> former as I'm sure we would know about it if madplay didn't work on
> other 64-bit archs.

It works perfectly on amd64.

> Is your machine littlendian?  Has anyone else had experience using
> madplay in a littleendian archs?   

I think everybody has expierence using it on little endian hosts, like
i386 and amd64.

Someone blogged about using it on arm a few days ago, which is also
little endian.  He also said that things like mpg321, splay, aplay
and playsound didn't work.
See: http://www.hermann-uwe.de/blog/playing-audio-on-the-nslu2

The code in both libmad and madplay look at endianness, it even has
some assembler for mips, so it looks weird that it wouldn't work.  So
I'm guessing it's just a general endian problem.

It would be nice if we know that it's in libmad or madplay that we
should look.

I have no idea how to start to debug this without access to a mips
or big endian host.


Kurt