[Patch,fortran] PR 27740 Symbol versioning for libgfortran

[Brooks Moses]

Andrew Pinski wrote:
>>FAIL: gfortran.dg/csqrt_2.f  -O0  execution test
>>FAIL: gfortran.dg/csqrt_2.f  -O1  execution test
>>FAIL: gfortran.dg/csqrt_2.f  -O2  execution test
>>FAIL: gfortran.dg/csqrt_2.f  -O3 -fomit-frame-pointer  execution test
>>FAIL: gfortran.dg/csqrt_2.f  -O3 -fomit-frame-pointer -funroll-loops 
>>execution test
>>FAIL: gfortran.dg/csqrt_2.f  -O3 -fomit-frame-pointer -funroll-all-loops 
>>-finline-functions  execution test
>>FAIL: gfortran.dg/csqrt_2.f  -O3 -g  execution test
>>FAIL: gfortran.dg/csqrt_2.f  -Os  execution test
>>
>>(This is on Debian stable, with an unmodified trunk except for applying 
>>Richard Guenther's recent patch for fixincludes so it will bootstrap.)
> 
> This is a different problem in fact this was a bug in glibc and it was XFAIL'd
> on Linux until just recently when we (maybe not we, the gfortran developers) decided
> that glibc 2.4.0 has been out long enough to no longer xfail this testcase.

Oh, okay.  Right; I'd forgotten about that.

- Brooks