[Bug target/57341] [4.8/4.9 Regression] wrong code on x86_64-linux at -O3 in 32-bit mode

rguenther at suse dot de gcc-bugzilla@gcc.gnu.org
Wed May 22 17:49:00 GMT 2013


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57341

--- Comment #8 from rguenther at suse dot de <rguenther at suse dot de> ---
"jakub at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org> wrote:

>http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57341
>
>--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
>(In reply to Michael Matz from comment #5)
>> Nope.  Our memory model allows this, the write will dynamically
>change
>> the type of the written memory cell.
>
>But why is that?  Just to handle placement new?  I'd say it would be
>better to
>make that explicit in the IL.

And all abominations of it. You repeatedly say this and I always respond with
that this is not possible. We have had that at some point and it did not work.

>(In reply to Richard Biener from comment #6)
>> /* { dg-options "-msse" { target { i?86-*-* x86_64-*-* } } } */
>
>looks wrong, that will just crash on pre-SSE CPUs (I know, rare).
>I think better would be:
>
>/* { dg-additional-options "-msse" { target sse2_runtime } } */
>
>(we don't have sse_runtime target, but I guess it doesn't matter if it
>won't be
>tested on pre-SSE2 CPUs, most of them have SSE2 these days).

I'll fix that.



More information about the Gcc-bugs mailing list