https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119399
--- Comment #5 from Richard Sandiford <rsandifo at gcc dot gnu.org> ---
(In reply to rguent...@suse.de from comment #4)
> >, so for a 4-element
> > vector, the only problem cases are p==q+4, p==q+8 and p==q+12. That's
> > equivalent to testing whether the unsigned value p-(q+4) <= 8.
>
> So the reasoning is that for p == q + 12 iff q + 12 wraps then
> it has to be outside of the object, thus n is low enough that
> we don't reach that point?
Yeah.
> w/o alignment knowledge problematic cases would include p == q + 1
> to p == q + 15, but that would still be no problem I think, the
> same wrapping argument can be used to simplify the
> p == q + 15 compare to p - (q + 4) <= 11? What was the reason
> to not do p - q <= 15 (or 12)?
p - q <= 15 would include p == q, but we should use the vector loop for that.
FWIW, the current condition for unaligned pointers is:
_46 = q_11(D) + 4;
_15 = p_12(D) - _46;
_54 = (sizetype) _15;
_32 = _54 + 3;
if (_32 > 14)
goto <bb 5>; [89.44%]
else
goto <bb 11>; [10.56%]
[https://godbolt.org/z/6ra5o4Wj9] but the forthcoming patch will make it:
_46 = (sizetype) p_12(D);
_15 = (sizetype) q_11(D);
_34 = _46 + 18446744073709551615;
_35 = _34 - _15;
if (_35 > 14)
goto <bb 5>; [89.44%]
else
goto <bb 11>; [10.56%]
