Adding "long double" SPARC-gcc intrinsics to uclibc

[Austin Foxley]

On 11/18/2010 04:29 AM, Konrad Eisele wrote:
> Hi,
> I have a patch that adds the "long double" SPARC-gcc intrinsics to uclibc,
> the (_Q_xxx functions). They where currently empty defined in
> libc/sysdeps/linux/sparc/qp_ops.c. The routines are taken from glibc:
> glibc-2.9/sysdeps/sparc/sparc32/soft-fp/*
> 
> Because the patch is 200k I think it will not go through the list filter,
> so I first send it as a link and then as a attachment in another mail:
> 
> ftp://gaisler.com/gaisler.com/tmp/uClibc-0.9.31-quad-routines.patch

Thanks for this. I had been previously solving this issue with uClibc on
leon with a patch to GCC based on the glibc soft-fp. You can find it
here: http://gcc.gnu.org/ml/gcc-patches/2009-04/msg01137.html if you are
interested. (no long double support though)

That said, I have no problem committing this into uClibc, as my patch is
unlikely to ever make it into gcc.

It didn't apply perfectly to master, but I fixed it up and committed it.
Please test and patch against the master branch in the future.

On an unrelated note, I've done quite a bit of work porting the original
gaisler patches forward to current uClibc and nptl. Would you mind
testing out master branch, and sending any further patches you have?

Thanks,
Austin


> 
> -- Greetings Konrad
> 
> Possible Changelog entry:
> 
> ---------------
> Subject: Adding quad float gcc intrinsic implementation for SPARC
> 
> Sparc-Gcc generates q_xxx intrinsic calls for quad float ("long double")
> code.  The routines from glibc's glibc-2.9/sysdeps/sparc/sparc32/soft-fp/*
> where taken and ported to uclibc.
> 
> Signed-off-by: Konrad Eisele <konrad at gaisler.com>
> 
> _______________________________________________
> uClibc mailing list
> uClibc at uclibc.org
> http://lists.busybox.net/mailman/listinfo/uclibc