Re: [PATCH v2 02/10] x86/percpu: Clean up percpu_to_op()

[Nick Desaulniers]

On Fri, Jul 10, 2020 at 1:53 AM Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>
> On Fri, Jul 10, 2020 at 12:38:23AM -0400, Brian Gerst wrote:
> > On Thu, Jul 9, 2020 at 6:30 AM Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> > >
> > > On Sat, May 30, 2020 at 06:11:19PM -0400, Brian Gerst wrote:
> > > > +     if (0) {                                                        \
> > > > +             typeof(_var) pto_tmp__;                                 \
> > > > +             pto_tmp__ = (_val);                                     \
> > > > +             (void)pto_tmp__;                                        \
> > > > +     }                                                               \
> > >
> > > This is repeated at least once more; and it looks very similar to
> > > __typecheck() and typecheck() but is yet another variant afaict.
> >
> > The problem with typecheck() is that it will complain about a mismatch
> > between unsigned long and u64 (defined as unsigned long long) even
> > though both are 64-bits wide on x86-64.  Cleaning that mess up is
> > beyond the scope of this series, so I kept the existing checks.
>
> Fair enough; thanks for explaining.

I brought up the same point in v1, for more context:
https://lore.kernel.org/lkml/CAKwvOdnCcpS_9A2y9tMqeiAg2NfcVx=gNeA2V=+zHknit7wGkg@xxxxxxxxxxxxxx/
-- 
Thanks,
~Nick Desaulniers