This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc 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: [PATCH 3/3] sparc32/sparcv9: add a VIS3 version of fdim


On 2016-08-26 09:34, Paul E. Murphy wrote:
> 
> 
> On 08/26/2016 09:25 AM, Paul E. Murphy wrote:
> > diff --git a/sysdeps/sparc/sparc32/sparcv9/fpu/multiarch/s_fdim.c b/sysdeps/sparc/sparc32/sparcv9/fpu/multiarch/s_fdim.c
> > index c9c7704..9666741 100644
> > --- a/sysdeps/sparc/sparc32/sparcv9/fpu/multiarch/s_fdim.c
> > +++ b/sysdeps/sparc/sparc32/sparcv9/fpu/multiarch/s_fdim.c
> > @@ -29,4 +29,4 @@ weak_alias (__fdim, fdim)
> >  # define __fdim __fdim_generic
> >  #endif
> > 
> > -#include <math/s_fdim.c>
> > +#include <ldbl-opt/s_fdim.c>
> 
> Note, Joseph mentioned in BZ this is used for 64b builds,
> which does not use ldbl-opt. So, this is not a correct fix.

I actually think your fix is the correct one. I have found an old build
log and this is the file used when building with --disable-multiarch. I
will run a build to confirm that.

I do wonder if simply using "#include_next <s_fdim.c>" would prevent
this kind of error. It's a GNU extension, but it's seems to be used in
many places already, so it's probably acceptable.

Aurelien

-- 
Aurelien Jarno                          GPG: 4096R/1DDD8C9B
aurelien@aurel32.net                 http://www.aurel32.net


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