This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug target/37216] [cygming] Invalid alignment for SSE store to .comm data generated with -O3
- From: "brian at dessent dot net" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: 7 Oct 2008 12:01:37 -0000
- Subject: [Bug target/37216] [cygming] Invalid alignment for SSE store to .comm data generated with -O3
- References: <bug-37216-16627@http.gcc.gnu.org/bugzilla/>
- Reply-to: gcc-bugzilla at gcc dot gnu dot org
------- Comment #48 from brian at dessent dot net 2008-10-07 12:01 -------
Subject: Re: [cygming] Invalid alignment for SSE store to
.comm data generated with -O3
sherpya at netfarm dot it wrote:
> I'll test your patch for the first post of the bugreport, and I'll test also
> ffmpeg but I'm almost sure that this TARGET_SSE is not true in this specific
> case
You said you used -march=core2 to compile. This implicitly includes
-msse2 (and -msse) so yes it will be set.
> yes one can use -fno-common or the fix to avoid crashes when gcc vectorize
> but it does not cover all cases
What cases would it not cover? When placing items directly in bss or
data the required alignment can be unambiguously conveyed to the
assembler so I don't see how it would get lost.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37216