This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: SSE types and structures
> On Thu, Jun 13, 2002 at 07:17:48PM +0200, Jan Hubicka wrote:
> > ! #define ADJUST_FIELD_ALIGN(FIELD, COMPUTED) \
> > ! (!TARGET_64BIT && !TARGET_ALIGN_DOUBLE && !DECL_USER_ALIGN (FIELD) \
> > ! && TYPE_MODE (TREE_CODE (TREE_TYPE (FIELD)) == ARRAY_TYPE \
> > ! ? get_inner_array_type (FIELD) \
> > ! : TREE_TYPE (FIELD)) == DFmode \
> > ! ? MIN ((COMPUTED), 32) : COMPUTED)
>
> You don't care for XFmode/TFmode.
I think we don't need XFmode, as it's natural alignment is 4 anyway.
I duno about TFmode, I think no ABI rely on it at the moment, so I don't
need to lower the alignemnt.
What I definitly missed is DImode.
>
> You should probably look at TREE_CODE == REAL_TYPE and
> COMPLEX_TYPE instead of TYPE_MODE.
Maybe, assuming that we want to give up the alignment for TFmodes, that
looks like shame to me.
>
> What sort of compatibility testing has this received?
I've rebuilt gcc with it and byte compared, but still I did missed the
DImode stuff. I am just working on that.
DO you have better idea about the testing?
Will testsuite files cover the layout issues enought.
Honza
>
>
> r~