Re: FAIL: gcc.c-torture/execute/950628-1.c execution, -O1

[Richard Henderson <rth at redhat dot com>]

On Fri, Jan 26, 2001 at 03:25:00PM -0500, Richard Kenner wrote:
> I think so too, but I'm not totally sure.  Do we gain anything other than
> aliasing by knowing that something is unchanging?

I don't think so.  It *does* happen to be a more accessible
form of aliasing information, so that rtx_varies_p and the
like can use it.

From the places which we actually use RTL_UNCHANGING_P, as
opposed to copy it copy it from one MEM to another, we use
it as an indication that the value is function invariant.
See rtlanal.c and local-alloc.c for example.

(Which might indicate an opportunity to have a pass construct
/u bits based on observed invariance rather than from features
in the source language.)

As for how much we might lose based on reuse, it depends on
what the lifetime of the object is.  The object has to be live
for a large enough span that whether or not the value is
function invariant matters at all.  I suspect, however, that
we don't ask this question enough to matter much.


r~