This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [3.4 patch] Fix PR16314, undefined reference to `__subdf3' for arm-softfloat


On Tue, Aug 31, 2004 at 02:12:16PM +0300, Vladimir Ivanov wrote:
> 
> On Tue, 31 Aug 2004, Richard Earnshaw wrote:
> 
> >On Tue, 2004-08-31 at 11:25, Vladimir Ivanov wrote:
> >>What about defining more targets, like "t-linux-crunch" and
> >>"t-linux-softfp"? This could solve some of the current ABI differences.
> >
> >Not really.  I suspect folks really want a single compiler that supports
> >both hard and soft float.  Separate t-* files only come into play with
> >separate configurations of the compiler.  Besides that it becomes a
> >maintenance nightmare.
> 
> Then we're back to the single "t-linux". Here's what I've been using:

> LIB1ASMSRC = arm/lib1funcs.asm
> LIB1ASMFUNCS = _udivsi3 _divsi3 _umodsi3 _modsi3 _dvmd_lnx \
> 	_negdf2 _addsubdf3 _muldivdf3 _cmpdf2 _unorddf2 _fixdfsi _fixunsdfsi 
> 	\
> 	_truncdfsf2 _negsf2 _addsubsf3 _muldivsf3 _cmpsf2 _unordsf2 \
> 	_fixsfsi _fixunssfsi

So, let's think about Richard's point.  Does the -mcpu=ep9312 version
_work_?  It probably doesn't - calls to divdf3 will always return using
the soft float convention and the compiler would expect them to return
using the Crunch registers.

Hopefully we can support using the soft-float ABI for libcalls soon.

-- 
Daniel Jacobowitz


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]