Re: [C++ PATCH] Add -fconstexpr-ops-limit= option (PR c++/87481, take 2)

[Jason Merrill <jason at redhat dot com>]

On 3/22/19 8:53 AM, Jakub Jelinek wrote:
> On Fri, Mar 22, 2019 at 08:47:21AM -0400, Jason Merrill wrote:
> > On 3/21/19 7:45 PM, Jakub Jelinek wrote:
> > > On Thu, Mar 21, 2019 at 07:27:03PM -0400, Jason Merrill wrote:
> > > > On 3/15/19 4:07 PM, Jakub Jelinek wrote:
> > > > > +/* Number of cxx_eval_constant_expression calls (except skipped ones,
> > > > > +   on simple constants or location wrappers) encountered during current
> > > > > +   cxx_eval_outermost_constant_expr call.  */
> > > > > +static HOST_WIDE_INT constexpr_ops_count;
> > > >
> > > > Hmm, a global static variable is non-reentrant.  This may not be an issue in
> > > > practice because of instantiate_constexpr_fns, but still seems questionable.
> > >
> > > One option would be to add HOST_WIDE_INT *constexpr_ops_count; into
> > > constexpr_ctx structure (pointer, not the counter itself, because we
> > >     constexpr_ctx new_ctx = *ctx;
> > >     ... (&new_ctx, ...)
> > > ).
> >
> > That makes sense.
>
> I've in the meantime bootstrapped/regtested on x86_64-linux and i686-linux
> the following patch.  Ok for trunk?

OK.

> > I've also wondered about splitting out parts of the context that don't
> > change so frequently, to avoid adding an extra copy of 'strict' every time
> > we build a new ctx.
>
> I agree it is a good idea.  Can that be done incrementally though?

Yes.

Jason