[Bug target/36502] i386/darwin generates unnecessary stack ops in every function

["howarth at nitro dot med dot uc dot edu" <gcc-bugzilla at gcc dot gnu dot org>]

------- Comment #45 from howarth at nitro dot med dot uc dot edu  2010-09-06 13:57 -------
(In reply to comment #44)
> Created an attachment (id=21709)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=21709&action=view) [edit]
> retain redefinition of MAIN_STACK_BOUNDARY as required
> 

Testsuite results for the PR36502v9.patch are shown in
http://gcc.gnu.org/ml/gcc-testresults/2010-09/msg00537.html. The compiler ICEs
shown for gfortran.dg/backspace_1.f, gfortran.dg/record_marker_2.f,
gfortran.dg/graphite/pr42393-1.f90 libgomp.fortran/appendix-a/a.16.1.f90,
libgomp.fortran/omp_atomic2.f90, libgomp.graphite/force-parallel-3.c,
libgomp.graphite/force-parallel-9.c and 25_algorithms/heap/moveable.cc
shouldn't be due to my patch as the identical patch (except for test cases
corrections) was tested in
http://gcc.gnu.org/ml/gcc-testresults/2010-09/msg00168.html and didn't show
them. While I don't see these in other reported i386-apple-darwin10
testresults, those aren't using --enable-checking=yes. I will rebuild gcc trunk
without PR36502v9.patch, reconfirm those ICEs and file PRs against them
separately.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36502