Re: [PATCH] Make is_signed_type() simpler

[Steven Rostedt]

On Sun, 29 Sep 2019 23:44:24 +0300
Alexey Dobriyan <adobriyan@xxxxxxxxx> wrote:

> On Sun, Sep 29, 2019 at 04:15:31PM -0400, Steven Rostedt wrote:
> > On Sun, 29 Sep 2019 23:06:19 +0300
> > Alexey Dobriyan <adobriyan@xxxxxxxxx> wrote:
> >   
> > > * Simply compare -1 with 0,
> > > * Drop unnecessary parenthesis sets
> > > 
> > > New macro leaves pointer as "unsigned type" but gives a warning,
> > > which should be fine because asking whether a pointer is signed is
> > > strange question.
> > > 
> > > I'm not sure what's going on in the i915 driver, it is shipping kernel
> > > pointers to userspace.  
> > 
> > This tells us what the patch does, not why.  
> 
> Check the subject line.

I don't see how it's simpler.

-#define is_signed_type(type)	(((type)(-1)) < (type)1)
+#define is_signed_type(type)	((type)-1 < 0)
 

Requires more rational that "make it simpler". Rewriting futex or tty
layer code would be something I would love to see, but just replacing
"(type)1" with "0" isn't worth the churn.

-- Steve