Re: [PATCH 1/2] AArch64 glibc port

["Joseph S. Myers" <joseph@codesourcery.com>]

On Wed, 7 Nov 2012, Marcus Shawcroft wrote:

> This resolves the majority of the outstanding issues.   The solution
> above is not quite sufficient because given a true 0, FP_ROUND may set
> working bits resulting in the underflow test failing.  I resolved this
> by adjusting FP_ROUND such that it does not attempt to round a true
> zero, is that sane? I've attached my current patch. The patch is

I think it would be better to put all the rounding logic inside the 
existing "if (_FP_FRAC_LOW_##wc(X) & 7)" conditional that determines 
whether to set the INEXACT exception - if none of the low three bits is 
set, there's no point in doing anything for rounding.

soft-fp patches should go to libc-alpha.  (Once checked in, the patch to 
update all the soft-fp files in libgcc from their soft-fp versions goes to 
gcc-patches, of course.)

> against glibc/soft-fp but I've actually be using it applied to
> libgcc/soft-fp. What is the recommend way to test patches to soft-fp
> in glibc?

You could ask the architecture maintainer of an architecture that uses 
soft-fp in glibc (e.g. SPARC) to test it, though I think testing in libgcc 
is just as useful.

-- 
Joseph S. Myers
joseph@codesourcery.com