This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: -malign-double behavior on W2K
- To: N8TM at aol dot com
- Subject: Re: -malign-double behavior on W2K
- From: craig at jcb-sc dot com
- Date: 29 May 1999 17:39:34 -0000
- Cc: egcs-bugs at egcs dot cygnus dot com
- Cc: craig at jcb-sc dot com
- References: <e66ab653.24818754@aol.com>
>Checking the results of various compiler options shows that -malign-double
>does take effect on Craig Burley's align0.F running on W2K. Maybe I was
>over-optimistic in thinking that register spills might be expected to be
>aligned when using -malign-double, even though that happens frequently now
>when not running on W2K.
I think 64-bit spills *should* be aligned, but I don't know how
alignment was fixed to guess why they might not be. Maybe it's
just a bug, perhaps a very hard-to-fix one. Certainly there are
no ABI requirements for spills that require them to appear in
certain locations!?
(AFAIK, -malign-double isn't needed anymore except for "extreme"
alignment...the default is, now, to align except when doing so would
violate the ABI. At least, that's basically how I've documented it.)
tq vm, (burley)