Re: [PATCH] tree-optimization/114074 - CHREC multiplication and undefined overflow

[Jakub Jelinek <jakub@redhat.com>]

On Mon, Feb 26, 2024 at 04:51:08PM +0100, Michael Matz wrote:
> On Mon, 26 Feb 2024, Jakub Jelinek wrote:
> 
> > > Will update the patch, I think any improvement should be done
> > > to get_range_pos_neg (it's a bit odd in behavior for unsigned
> > > but I have only signed things incoming).
> > 
> > For unsigned if it always returned 1, it would be quite useless, there would
> > be no point for the caller to call it in that case.
> 
> Which seems to make sense for a function called ...pos_neg on unsigned 
> types.  I would expect calling it to be useless and always return "yep, 
> non-negative, why did you ask?".

The callers heavily rely on it doing for unsigned types what it does now
and it matches what the function comment says.
If you have a suggestion for a different name that wouldn't be much longer
than the current one, it can be renamed.
get_msb_range?
Except it doesn't return a range for the most significant bit, but a bitmask
which values are possible.

	Jakub