This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] PR libgomp/28296: Bootstrap failure on Tru64
- From: Roger Sayle <roger at eyesopen dot com>
- To: Andrew Pinski <pinskia at physics dot uc dot edu>
- Cc: gcc-patches at gcc dot gnu dot org, Richard Henderson <rth at redhat dot com>
- Date: Mon, 11 Sep 2006 09:38:57 -0600 (MDT)
- Subject: Re: [PATCH] PR libgomp/28296: Bootstrap failure on Tru64
Hi Andrew,
On Mon, 11 Sep 2006, Andrew Pinski wrote:
> On Mon, 2006-09-11 at 09:21 -0600, Roger Sayle wrote:
> > The following patch resolves PR libgomp/28296 which a boostrap failure
> > on alphaev67-dec-osf5.1 triggered by Eric Christopher's 2006-07-05
> > change to libgomp's configure.ac to avoid the use of -pthread on darwin.
> > http://gcc.gnu.org/viewcvs?view=rev&revision=115205
> >
> > Unfortunately, that change contained two mistakes: the first is that
> > it removed -pthread from $CFLAGS on all platforms (it's necessary on
> > those that require _REENTRANT to be defined to get the pthreads APIs),
> > and the second was that it used "default)" in a case statement instead
> > of "*)" thereby disabling the setting of XPCFLAGS on all non-darwin
> > platforms.
> >
> > The change below has been tested on alphaev67-dec-osf5.1 where it
> > allows a default bootstrap, except for java, to complete without
> > problems.
>
> Can't we just get darwin to ignore -pthread? There are already bugs
> asking for -pthread to become more target independent.
I agree that would be the sensible change. We've already added -pthread
to gcc on Solaris and IRIX purely to support libgomp's configury.
However I'm not sure we're quite ready to make that a requirement of a
port. Making libgomp's configury more flexible seems like a good
intermediate/compromise step.
We've also got PR's for enabling libgomp to use of system thread libraries
other than POSIX pthreads. :-)
Roger
--