This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] Crossjumping corrupts alias info
- From: Richard Henderson <rth at redhat dot com>
- To: Andreas Krebbel <Andreas dot Krebbel at web dot de>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Thu, 4 Mar 2004 10:57:24 -0800
- Subject: Re: [patch] Crossjumping corrupts alias info
- References: <200402201750.05271.krebbel1@de.ibm.com> <20040220230123.GE29641@redhat.com> <200403041452.55274.Andreas.Krebbel@web.de>
On Thu, Mar 04, 2004 at 02:52:55PM +0100, Andreas Krebbel wrote:
> So for different expression
> with equal offsets the offsets should be cleared as well.
That wouldn't hurt, true. I don't think the offset part would
actually be used without the expression.
> I understand this for the alignment data, but is this right for the size?
Hmm. The size is the size of *this* access. The only time they
wouldn't already be identical is if this is a BLKmode copy. Can
we manage to unify two known-but-different-size memcpy's? Maybe,
if the constraint on the immediate length operand forces it into
a register.
So, yes, I think you're right, max for size.
r~