This is the mail archive of the
ecos-discuss@sourceware.org
mailing list for the eCos project.
Re: gcc 4.1.2, some testcases fail when built with optimizations
On Tue, Jun 19, 2007 at 02:14:29PM +0200, Hans Rosenfeld wrote:
> On Mon, Jun 18, 2007 at 05:17:56PM -0600, Gary Thomas wrote:
> > If it's not fixed in 4.2, it looks like you have a well documented
> > test case to pass on to the GCC team (that's how bugs get fixed :-)
>
> Same behaviour in 4.2, I reported it and have been told its an aliasing
> violation (see http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32397).
>
> I'm not exactly sure about this whole thing, so I'll not argue with the
> gcc folks about this. Compiling libm with -fno-strict-aliasing makes it
> work, but I don't think that this is a solution at all.
>
> Note that there is at least another such "bug" in the libm code, but I
> have no idea where.
I just spent 2 minutes of reading the manual. I admit 2 minutes is
probably not enough......
Try playing with __attribute__((__may_alias__)) on
typedef Cyg_libm_ieee_double_shape_type ieee_double_shape_type;
in mathincl/fdlibm.h.
See
http://gcc.gnu.org/onlinedocs/gcc-4.2.0/gcc/Type-Attributes.html#Type-Attributes
Andrew
--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss