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: emulating tls for systems without


 Jakub Jelinek
> On Thu, Oct 05, 2006 at 02:12:39PM +1300, Danny Smith wrote:
> > Yes, mingw32 is one of those.  There are a few small patches to get 
> > libgomp to now work for mingw32 as well, but I was holding 
> off on the 
> > small stuff until
> > stage 1, assuming that the "great stuff" would wait till then. 
> > 
> > Should I still wait?
> 
> AFAIK none of them were dependent on the emutls stuff (they 
> just make libgomp usable on your target and if you don't use 
> threadprivate, you don't need native TLS nor emutls), all of 
> them were approved by Richard and IMHO they certainly should 
> go into 4.2. Sure, it can't be considered a regression bug 
> fix, since OpenMP support has only been added in 4.2, but all 
> the patches were small, non-intrusive that I'd say it would 
> be a pitty not to include them in 4.2.
> 

OK, I have commited the following:

gcc/
	* config/i386/mingw32.h (GOMP_SELF_SPECS): Add -mthreads for
openmp.

libgomp/
	* testsuite/libgomp.c/barrier-1.c: Change timestamp tests from
	'<' to '<='.

	* acinclude.m4 (HAVE_ATTRIBUTE_ALIAS): Remove
__USER_LABEL_PREFIX__ from
	test.
	* configure: Regenerate.
	* fortran.c  (ialias_redirect): Add __USER_LABEL_PREFIX__ to
alias.

libgfortran/
	* acinclude.m4 (HAVE_ATTRIBUTE_ALIAS): Remove
__USER_LABEL_PREFIX__ from test.
	* configure: Regenerate.

Danny


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