Re: [PATCH] [PR68097] Try to avoid recursing for floats in tree_*_nonnegative_warnv_p.

[Richard Biener <richard.guenther@gmail.com>]

On Tue, Nov 15, 2022 at 2:52 PM Aldy Hernandez <aldyh@redhat.com> wrote:
>
> On Mon, Nov 14, 2022 at 10:12 AM Richard Biener
> <richard.guenther@gmail.com> wrote:
> >
> > On Sat, Nov 12, 2022 at 7:30 PM Aldy Hernandez <aldyh@redhat.com> wrote:
> > >
> > > It irks me that a PR named "we should track ranges for floating-point
> > > hasn't been closed in this release.  This is an attempt to do just
> > > that.
> > >
> > > As mentioned in the PR, even though we track ranges for floats, it has
> > > been suggested that avoiding recursing through SSA defs in
> > > gimple_assign_nonnegative_warnv_p is also a goal.  We can do this with
> > > various ranger components without the need for a heavy handed approach
> > > (i.e. a full ranger).
> > >
> > > I have implemented two versions of known_float_sign_p() that answer
> > > the question whether we definitely know the sign for an operation or a
> > > tree expression.
> > >
> > > Both versions use get_global_range_query, which is a wrapper to query
> > > global ranges.  This means, that no caching or propagation is done.
> > > In the case of an SSA, we just return the global range for it (think
> > > SSA_NAME_RANGE_INFO).  In the case of a tree code with operands, we
> > > also use get_global_range_query to resolve the operands, and then call
> > > into range-ops, which is our lowest level component.  There is no
> > > ranger or gori involved.  All we're doing is resolving the operation
> > > with the ranges passed.
> > >
> > > This is enough to avoid recursing in the case where we definitely know
> > > the sign of a range.  Otherwise, we still recurse.
> > >
> > > Note that instead of get_global_range_query(), we could use
> > > get_range_query() which uses a ranger (if active in a pass), or
> > > get_global_range_query if not.  This would allow passes that have an
> > > active ranger (with enable_ranger) to use a full ranger.  These passes
> > > are currently, VRP, loop unswitching, DOM, loop versioning, etc.  If
> > > no ranger is active, get_range_query defaults to global ranges, so
> > > there's no additional penalty.
> > >
> > > Would this be acceptable, at least enough to close (or rename the PR ;-))?
> >
> > I think the checks would belong to the gimple_stmt_nonnegative_warnv_p function
> > only (that's the SSA name entry from the fold-const.cc ones)?
> >
> > I also notice the use of 'bool' for the "sign".  That's not really
> > descriptive.  We
> > have SIGNED and UNSIGNED (aka enum signop), not sure if that's the
> > perfect match vs. NEGATIVE and NONNEGATIVE.  Maybe the functions
> > name is just bad and they should be known_float_negative_p?
>
> Yeah, SIGNED and UNSIGNED doesn't seem to be much clearer than "bool signbit".
>
> For instance, we have the following in frange:
>
>   void set_nan (tree type, bool sign);
>   void update_nan (bool sign);
>   bool maybe_isnan (bool sign) const;
>   bool signbit_p (bool &signbit) const;
>
> I'm OK changing them to enum signop if you prefer.  I'm just not
> totally convinced it's more readable.
>
> ??

I think when talking about 'signbit' or 'sign' the old usual 'unsigned' type is
better than bool.    signbit_p feels a bit redundant (the _p).

We could have another enum like signop just I think the obvious
candidates { POSITIVE, NEGATIVE } or { NEGATIVE, NONNEGATIVE }
aren't too great.  The obvious alternative is to follow the existing uns_p
(unsigned?) parameters and not use bool sign but bool unsigned_p,
there 'true' and 'false' are obvious.  For 'signbit_p' it would then be
bool signbit_p (unsigned &signbit) being the true value of the bit
(and the return value indicates the UNKNOWN case).

Richard.

>
> Aldy
>