This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] Fix PR37216 [cygming] Invalid alignment for SSE store to .comm data generated with -O3
- From: Dave Korn <dave dot korn dot cygwin at googlemail dot com>
- To: Danny Smith <dansmister at gmail dot com>
- Cc: Dave Korn <dave dot korn dot cygwin at googlemail dot com>, GCC Patches <gcc-patches at gcc dot gnu dot org>
- Date: Sun, 24 May 2009 12:46:38 +0100
- Subject: Re: [PATCH] Fix PR37216 [cygming] Invalid alignment for SSE store to .comm data generated with -O3
- References: <4A187898.firstname.lastname@example.org> <email@example.com>
Danny Smith wrote:
> It would also be nice to test the effect of this on
> testsuite/gcc.dg/compat/struct-layout-1.exp tests after
> removing -fno-common switch for mingw and cygwin in
> struct-layout-1_generate.c, line 49.
At the moment, this failed for the same reason as 20030811-1.c; in the light
of the comment from varasm.c I've just discovered(*), I need to respin the
patch to allow for size being zero and allocate (at least?) one byte (even if
it does turn out to be invalid to have a zero-length array at file scope).
I'll add yours and Uros' suggestions while I'm doing it and repost shortly;
(*) - http://gcc.gnu.org/ml/gcc/2009-05/msg00593.html