Subregs and objects with holes (new patch)

[Rask Ingemann Lambertsen]

On Fri, Nov 24, 2006 at 01:50:08PM +0000, Joseph S. Myers wrote:
> On Fri, 24 Nov 2006, Rask Ingemann Lambertsen wrote:
> 
> >    AFAIK, i686-pc-linux-gnu defaults to -m96bit-long-double. How did you
> > test it with -m128bit-long-double?
> 
> There was no reduced testcase given with the patch introducing the special 
> handling of objects with holes (neither added to the testsuite, nor given 
> inline in the message) and so I couldn't test that.  I'd welcome the 
> addition of a -m128bit-long-double testcase showing the original problem 
> to the testsuite.

   My best bet is to add it as a multilib during testing:

Index: gcc/config/i386/t-pmm_malloc
===================================================================
--- gcc/config/i386/t-pmm_malloc	(revision 119099)
+++ gcc/config/i386/t-pmm_malloc	(working copy)
@@ -4,3 +4,6 @@
 mm_malloc.h: $(srcdir)/config/i386/pmm_malloc.h
 	rm -f $@
 	cat $^ > $@
+
+MULTILIB_OPTIONS=m128bit-long-double
+MULTILIB_DIRNAMES=ld128bit

   Additionally, to test my modifications to subreg_regno_offset() and
subreg_offset_representable(), I'm using wrappers around the two functions
to compare the results of the new and old ones and abort if they aren't
identical.

-- 
Rask Ingemann Lambertsen